Constitutive internalization of constitutively active agiotensin II AT(1A) receptor mutants is blocked by inverse agonists.

As constitutively active mutants (CAMs) mimic an active conformation, they can be used to characterize the process of G protein-coupled receptor activation. Here, we used CAMs to study the link between activation and internalization of the angiotensin II AT(1A) receptor. The cellular localization of fluorescently tagged N111A, I245T, and L305Q mutants was determined by confocal microscopy. In the absence of ligand, CAMs were mostly located in intracellular vesicles, whereas the wild-type AT(1A) was found at the cell surface. After 2 h incubation with inverse agonist, losartan, CAMs were translocated to the plasma membrane. Similar observations were made in H295, a human adrenocortical cell line which expresses physiologically the AT(1) receptor. This phenomenon, which was not dependent on protein synthesis and the pharmacology and kinetics of which were similar to the recycling of the wild-type receptor, was called "externalization". After externalization and losartan removal, the L305Q CAM underwent rapid ligand-independent endocytosis, with the same kinetics and temperature sensitivity as the angiotensin II-induced internalization of the wild-type AT(1A). Moreover, the addition of a second mutation known to block internalization (Delta 329 truncation) prevented intracellular localization of the CAM. These data show that AT(1A) CAMs are constitutively and permanently internalized and recycled. This mechanism is different from the down-regulation observed for CAMs of other G protein-coupled receptors and thus defines a new paradigm for the cellular regulation of CAMs.

G protein-coupled receptors (GPCR) 1 form one of the largest protein families, with several hundred members in humans (1). Despite the wide variety of ligands and physiological roles, these receptors are all structurally characterized by seventransmembrane domains and most of them are thought to share common activation and desensitization mechanisms. GPCRs are supposed to isomerize spontaneously between an inactive (R) and an active state (R*), the latter being responsible for G protein coupling and subsequent intracellular signaling. This two-state model is probably oversimplified but is helpful for the interpretation of mutagenesis and pharmacological data. It is supported by the existence of constitutively active mutants (CAMs) in the GPCR family. These mutants mimic the active state and therefore present permanent ligandindependent signaling. Ligands able to block this constitutive activity are called inverse agonists. In the two-state model, the inverse agonists have preferential affinity for the inactive state (R). Conversely, "regular" agonists preferentially bind the active state (R*).
Many GPCRs are desensitized after G protein activation, i.e. they become insensitive to agonists. The binding of arrestins to the receptor is known to play a major role in this process and is favored by the phosphorylation of the receptor by specific GPCR kinases. The arrestins prevent further interaction with the G proteins. They also promote internalization via clathrincoated pit-dependent endocytosis, which results in the disappearance of the receptor from the plasma membrane (2). Finally, they participate in the recycling of the receptor, which can take a few minutes to a few hours, depending on the type of GPCR. In some cases, some of the receptors can also be down-regulated, leading to long-term desensitization. Activation is thus a physiological prerequisite for receptor internalization and these two processes are probably highly dependent on each other. As CAMs mimic the active conformation of the receptor, they should help to elucidate the link between activation and internalization.
In this study, we used the angiotensin II (AngII) type 1 receptor (AT 1 ) as a model to address this question. The AT 1 receptor regulates the contraction and hypertrophy of vascular smooth muscle cell contraction and the secretion of aldosterone. Thus it plays a critical role in the control of blood pressure and sodium homeostasis. This pivotal physiological role makes the AT 1 receptor an important therapeutic target and numerous antihypertensive and cardioprotector agents have been developed. As a consequence, it is also one of the most studied GPCRs. Its cDNA was first cloned in 1991 (3). The AT 1 receptor (AT 1A and AT 1B in rodents) leads to the G␣ q/11 -mediated activation of phospholipase C-␤, which generates diacylglycerol and inositol (1,4,5)-trisphosphate. Following the phosphorylation of the intracellular sequences (4,5), the AT 1 receptor is rapidly internalized (t1 ⁄2 ϳ 5 min) (6 -10) in clathrin-coated pits after interaction with ␤-arrestin 1 and dynamin 1 and 2 (11). The receptor is then slowly and partly recycled at the plasma membrane (7,12,13). We were able to directly visualize the internalization process of an EGFP-tagged AT 1 receptor using confocal microscopy (10).
Only one CAM was identified by site-directed mutagenesis (N111A) (14,15), but we used a random mutagenesis approach to identify several other CAMs of the AT 1A receptor (16). The ligand-independent activation of the AT 1 signaling pathway induced by these mutants is abolished by the inverse agonists, losartan (14,16) and irbesartan. 2 To study the link between activation and internalization/recycling, we analyzed the cellular trafficking of three EGFP-tagged CAMs of the AT 1A receptor (N111A, I245T, and L305Q).

Construction of the EGFP-tagged CAM and WT Receptors-The
EGFP-AT 1A receptor was constructed in three steps: 1) the nucleotide sequence of the insulin receptor signal peptide (sp) was amplified from the pET vector (17) with the following primers: 5Ј (ACCGGTCGCCAC-CATGGGCACCGGGGG) and 3Ј (ACCGGTAGGTGGCCCGCGGCGC), which inserted an AgeI site at both ends of the sp. The PCR fragment was digested with AgeI and inserted into the AgeI site of pEGFP-C3 (CLONTECH), this construct was called ps-EGFP. 2) A linker, consisting of six copies of the myc epitope with the following amino acid sequence: Glu-Gln-Lys-Leu-Ile-Ser-Glu-Glu-Asp-Leu-Gly-Arg-Phe (Glu-Gln-Lys-Leu-Ile-Ser-Glu-Glu-Asp-Leu) 5 , was inserted into the EagI site of pEAT 1A ⌺ (18). This construct was called 6mycs-AT 1A .
3) The 6mycs-AT 1A vector was digested with HindIII and XmaI and inserted into the HindIII and XmaI site of ps-EGFP. The construct was called ps-EGFP-6mycs-AT 1A and the corresponding receptor was called EGFP-AT 1A .
For transient H295 cell transfection, cells were plated in polyallylamine-treated 6-or 24-well plates at 50 -70% confluence the day before transfection. Cells were then transfected with 0.35 g/well for 24-well plates or 0.7 g/well for 6-well plates of AT 1A -EGFP or L305Q-EGFP plasmids, using a liposomal transfection reagent (Lipo-fectAMINE Plus, Invitrogen). FACS analysis indicated that ϳ15% of the cells were transfected.
Pharmacological and Signaling Properties of the EGFP-AT 1A , EGFP-N111A, EGFP-I245T, EGFP-L305Q, EGFP-⌬329, and EGFP-L305Q-⌬329 Receptors in HEK-293 Cells or in H295 Cells-1) Binding experiments with [ 125 I]labeled Ang II were performed on intact cells, as previously described (21) except that the incubations with [ 125 I]AngII were performed for 3 h at 4°C. Binding data were analyzed by linear regression using the Microsoft Excel 5 program.
2) To measure inositol phosphate (IP) production, cells were transfected with 0.5 g/200,000 cells of G␣ q cDNA (a gift from B. Conklin, Departments of Medicine and Pharmacology, University of California, San Francisco, CA). The cells were metabolically labeled with myo-[ 3 H]inositol as previously described (21) and then the IP content was determined by methanol extraction and separation on a Dowex AG1-X8 (Bio-Rad) column. To determine the functional consequences of losartan-induced externalization, cells were pretreated with LiCl at 16°C instead of 37°C.
3) The aequorin assay was performed as previously described (16) on stable HEK-EGFP-AT 1A or HEK-EGFP-L305Q cell lines, transiently transfected with the plasmid encoding the bioluminescent calciumsensitive protein, aequorin, in 96-well plates. 48 h after transfection, the cells were incubated for 2 h at 37°C in the presence or absence of 1 M losartan in medium supplemented with 0.5 M coelenterazine and 1% FCS. The cells were subsequently washed twice with aequorin buffer and incubated for 30 min at 16°C in 50 l of aequorin buffer (16). Cells were stimulated by adding 50 l of increasing concentrations of AngII and the luminescence was measured 30 s later in a TopCount counter (Packard).
Fluorescence-activated Cell Sorting (FACS)-Cells expressing EGFPtagged receptors were prepared and analyzed by FACS as previously described (22).
Protein Metabolic Labeling and Immunoprecipitation-Stable HEK-EGFP-AT 1A or HEK-EGFP-L305Q cells were starved for 1 h in methionine-and cysteine-free Ham's F-12 (Invitrogen). The starved cells were labeled for 5 min with 50 Ci/ml of [ 35 S]Redivue Promix (Amersham Pharmacia Biotech) in the same medium. Cells were then rinsed twice in PBS and incubated for various time in Dulbecco's modified Eagle's medium supplemented with 7.5% FCS at 37°C. Cells were then processed for immunoprecipitation as previously described (10), except that 1 g of a monoclonal anti-myc antibody (9E10) was used for the immunoprecipitation (Santa Cruz Biotechnology). Some immunoprecipitated samples were treated with Endo H and PNGase F enzymes (both from Biolabs) according to the manufacturers recommendations.
Biochemical Measurement of Internalization and Recycling-Cells were either pretreated with 10 M monensin (Sigma) for 1 h at 37°C or left untreated. These cells were incubated with 100 nM AngII for 30 min at 4°C and then incubated at 37°C for 30 min to allow internalization. At this point an acid wash (0.2 M acetic acid, 0.5 M NaCl in binding buffer, 5 min at 4°C) was performed to remove surface bound AngII. For the recycling studies, cells were additionally incubated in Dulbecco's modified Eagle's medium supplemented with 1% FCS in the presence or absence of 10 M monensin for various periods of time at 37°C. To determine the rates of internalization and recycling, binding experiments with [ 125 I]labeled AngII were carried out on intact cells for 3 h at 4°C as previously described (21).
Measurement of Internalization, Externalization, and Re-internalization by Confocal Microscopy-Confocal microscopy was also used to analyze receptor trafficking as described previously (10). Cells (50,000 cells/well) were seeded on polyallylamine-treated chambered coverglass 8 wells (Nunc) and treated for 1 h at 37°C with 70 M cycloheximide (Sigma). Cells were then incubated for 30 min at 4°C with various ligands in Earle's buffer (10). Internalization was promoted by incubating the cells in Earle's complete buffer (10) at 37 or at 16°C for various periods of time: 30 min with 100 nM (or 10 nM in Fig. 5C) AngII, 100 nM [Sar 1 -Ile 8 ]AngII, 10 M L162,313, or for 2 h with 1 M losartan and 10 M irbesartan. For the re-internalization study, after losartan treatment, cells were washed for 30 min at 4°C in Earle's complete buffer and then incubated at 37°C or 16°C. When required cells were treated with 10 M monensin for the entire assay. After the incubation periods, the cells were rinsed in ice-cold Earle's buffer and fixed by incubating them for 10 min in 100% methanol at 4°C.
Cells were examined with a Leica TCS NT confocal laser scanning microscope configured with a Leica DM IRBE inverted microscope equipped with an argon/helium/neon laser. EGFP fluorescence was detected following 100% excitation at 488 nm by use of a spectrophotometer set with a window between 530 and 600 nm. Images of individual cells (1024 ϫ 1024 pixels) were obtained by use of a ϫ63 oilimmersion objective. Each image was done on a cross-section through the cells.
Quantification of the Subcellular Distribution of the Fluorescence-Digital image analysis using a specific macro software (10, 22) derived from the public domain NIH Image software (developed by the U.S. National Institutes of Health and available on the Internet at rsb.info. nih.gov/nih-image/) was used to measure the subcellular distribution of 2 C. Parnot, unpublished results. the EGFP-tagged receptors. This software allowed to measure the S, C, and N values, which correspond to the mean density of the surface, cytoplasm, and nucleus fluorescence. The background fluorescence (N) was subtracted from the S and C values to give the SЈ and CЈ values. The SЈ/CЈ ratio provides reliable information about the level of cell surface expression and the internalization state of the fluorescent receptor.
Statistics-Results are expressed as mean Ϯ S.E. Statistical significance was assessed by the Student's t test.

Characterization of the EGFP-tagged CAMs of the AngII AT 1A
Receptor-The CAM N111A, I245T, L305Q, and the wildtype (WT) AT 1A receptors were tagged at the N terminus with EGFP to determine their subcellular localization and trafficking. A signal peptide was fused to the N terminus of EGFP to allow the correct exportation of the chimeric protein to the plasma membrane. A spacer, consisting of a tandem repeat of the myc epitope (63 amino acids), was inserted between EGFP and the AT 1A receptor to prevent steric encumbrance of the AngII-binding site.
The EGFP-tagged CAM and WT receptors were stably transfected in HEK-293 cells and their functional properties were analyzed (Table I). The K d of [ 125 I]AngII for the EGFP-AT 1A and the EGFP-L305Q were both similar to the known K d of the non-tagged WT receptor (K d ϭ 0.61 nM for the AT 1A receptor (19)). As evaluated by binding of labeled AngII, the EGFP-L305Q receptor presents a lower plasma membrane expression compared with the WT EGFP-AT 1A receptor (Table I). We checked the constitutive activity of the EGFP-L305Q receptor by measuring the agonist-independent production of IP in the stable cell line, after transient transfection of the G␣ q cDNA to increase the sensitivity of the assay (16,23). The results were normalized with respect to the expression levels of the receptor to allow accurate comparison (Table I). The basal activity of the cell line expressing the EGFP-L305Q receptor was three times as high as that of the cells expressing the WT EGFP-AT 1A , showing that the mutant fused to EGFP retained its constitutive activity. Moreover, the production of IP was increased by AngII treatment (Table I). The EGFP-N111A and EGFP-I245T receptors have similar K d (0.76 Ϯ 0.13 and 0.95 Ϯ 0.19 nM, respectively) to that of the WT receptor and the same basal constitutive activity as the EGFP-L305Q receptor (EGFP-N111A, 770 Ϯ 107; EGFP-I245T, 203 Ϯ 16). Thus the EGFPtagged receptors were fully functional in terms of ligand binding and second messenger production.
Cellular Localization of the EGFP-tagged WT and CAM AT 1A Receptors-The EGFP tag enabled the cellular localization of the WT and CAM receptors to be determined by confocal microscopy. Interestingly, whereas the WT EGFP-AT 1A receptors were localized at the cell surface, the CAM receptors (N111A, L305Q, and I245T) were mainly located in intracellular vesicles in the cytoplasm and some cells expressed low amount on their plasma membrane (Fig. 1A).
This constitutive intracellular localization of the CAM receptors was quantified by calculating the ratio of surface and cytoplasmic fluorescence densities, denoted here as SЈ/CЈ (see "Materials and Methods"). For the WT EGFP-AT 1A receptor, the SЈ/CЈ ratio was typically 1.5-2.0 in the absence of the ligand and decreased to 0.5 after AngII-induced internalization (Table I, (Table I and Fig. 2B). The SЈ/CЈ ratios for the WT and L305Q receptors were further reduced by incubation with AngII, showing that a nonnegligible fraction of the receptors was still at the plasma membrane and was internalized in the presence of AngII (Fig. 2C). The presence of the EGFP-L305Q receptor at the cell surface was confirmed by binding experiments (Table I).
We assessed whether these differences in basal SЈ/CЈ ratios of WT and CAM receptors were due to differences in the total  number of receptors. We used FACS to quantify the total fluorescence per cell on 10,000 cells stably expressing the EGFP-L305Q and the WT receptors (Table I). The two other CAMs presented comparable total fluorescence (data not shown). We also excluded the possibility that plasma membrane targeting was altered due to the presence of EGFP at the N terminus, because the WT or L305Q receptors with a C-terminal EGFP tag presented the same cellular distribution as their N-terminal-tagged counterparts (data not shown).
Metabolic labeling experiments were performed to identify differences in the maturation of the WT and L305Q receptors. The WT EGFP-AT 1A and the EGFP-L305Q receptors both corresponded to ϳ80-kDa bands and the maximal intensity was reached after a 5-min pulse and progressively decreasing after a 60-min chase (Fig. 1B). The maturation of the proteins was studied by their sensitivity to Endo H and PNGase F (Fig. 1C). After a 5-min pulse without chase, both of the receptor types were sensitive to both enzymes, as indicated by the increase in their electrophoretic mobility. Surprisingly, after a 60-min chase both receptors were sensitive to Endo H. This migration profile, i.e. sensitivity to both enzymes, was also observed on a total cell extract after deglycosylation. In addition to the ϳ80-kDa band, the WT EGFP-AT 1A receptor was also represented by a 65-kDa band, which is not the consequence of EGFP-tag cleavage (data not shown). In conclusion, the WT EGFP-AT 1A and the EGFP-L305Q receptors present similar maturation/ degradation and glycosylation profiles and the intracellular localization of the CAMs is therefore an intrinsic property of the constitutive activity and not due to the intracellular accumulation of the receptor during biosynthesis.
These results provide strong evidence for the constitutive intracellular localization of three different AT A CAMs (N111A, I245T, and L305Q). The CAM receptors remaining at the plasma membrane were fully functional, displaying AngII binding, as well as AngII-induced signaling and internalization comparable to the WT receptor.
Effect of an Inverse Agonist, Losartan, on the Cellular Localization of the AT 1A CAMs-The inverse agonist, losartan, is known to inhibit the constitutive signaling activity of the AT 1A CAMs (14,16). We hypothesized that this compound would also affect the cellular localization of the CAMs. Our results ( Fig.  2A) confirmed previous findings (8, 10) that losartan does not modify the membrane localization of the WT EGFP-AT 1A receptor. The fluorescence of untreated CAM cells was mainly intracellular, whereas the treatment with losartan resulted in the appearance of plasma membrane fluorescence. This was clearly visible on confocal images ( Fig. 2A) and resulted in a 50 -100% increase in the basal SЈ/CЈ ratio of these receptors (Fig. 2B). This plasma membrane translocation resulted in a 29% increase in the number of cell surface-binding sites on the L305Q mutant, as measured by [ 125 I]AngII binding after losartan wash-away (data not shown). We called this phenomenon "externalization." We used the L305Q mutant to determine whether CAM externalization was specific to inverse agonists or could be observed with other pharmacological molecules. We tested the peptide antagonist, [Sar 1 -Ile 8 ]AngII, the non-peptide agonist, L162,313, and the non-peptide antagonist, irbesartan, which has similar inverse agonist properties to losartan (data not shown). [Sar 1 -Ile 8 ]AngII induced similar levels of internalization of the WT and L305Q receptors as AngII. L162,313 induced a slight internalization of both WT and L305Q receptors (Fig. 2C). Interestingly, irbesartan had a very similar effect to losartan. It did not modify the localization of the WT receptor, whereas it induced a membrane translocation of the EGFP-L305Q receptors as shown by a ϳ100% increase in the SЈ/CЈ ratio (Fig. 2C). Therefore, only the inverse agonists induced the externalization of the EGFP-L305Q receptor.
Cellular Localization of the WT and L305Q Receptors in H295 Cells-H295 is a human adrenocortical carcinoma cell line (24) which is a model for AngII-responsive aldosterone secretion via activation of endogenous AT 1 receptors (25). In this physiological model of AngII action, the cellular distribution of transiently transfected EGFP-tagged WT AT 1A and L305Q receptors was studied in basal conditions or after treatment with losartan. According to efficiency of transfection and binding experiments, the transfected H295 cells expressed an equivalent number or a maximum of twice the amount of recombinant (WT or mutated) AT 1 receptors as compared with the amount of endogenous receptors. This allowed to follow the cellular localization of the WT and mutated AT 1 receptor at a physiological level of expression and in a physiologically relevant cell.
In these cells, AngII induced internalization of both EGFPtagged receptors (data not shown). Whereas the WT receptor was almost exclusively localized at the plasma membrane, the L305Q receptor presented a plasma membrane and a diffuse cytoplasmic localization (Fig. 3A). The SЈ/CЈ ratio for the WT receptor was 8.90 Ϯ 1.24 and was dramatically reduced in the case of the L305Q receptor (2.30 Ϯ 0.26) (Fig. 3B). Treatment with losartan had little effect on the cellular localization and the SЈ/CЈ ratio (13.03 Ϯ 3.17) of the WT receptor, whereas it induced a disappearance of the cytoplasmic fluorescence and a major increase in the SЈ/CЈ ratio (9.80 Ϯ 2.07) for the L305Q receptor (Fig. 3B). Thus, the intracellular localization and losartan-induced externalization of the L305Q receptor were also observed in a different and more physiological cellular model (H295 cells).
The Externalization Induced by Inverse Agonists Is Dependent on Recycling Mechanisms-Externalization experiments were performed on cells treated with cycloheximide prior to and during the experiment. This treatment prevented de novo biosynthesis, thus removing all receptors from the biosynthesis/ secretion pathway, as indicated by the disappearance of fluorescence in the corresponding structures. Thus, externalization could not be explained by the arrival of newly synthesized receptors at the plasma membrane. We hypothesized that externalization is due to a recycling mechanism.
The classical AngII-induced internalization/recycling mechanisms were first characterized by a biochemical procedure on EGFP-AT 1A and EGFP-L305Q receptors. The WT EGFP-AT 1A and EGFP-L305Q receptors were quickly internalized (t1 ⁄2 ϳ 5 min) after the addition of 100 nM AngII and this process is maximum at 30 min (Ref. 8, and data not shown). At this time, acid washing of the surface AngII allowed to follow receptor recycling by [ 125 I]AngII binding. The recycling of the EGFP-L305Q receptor was maximal 2 h after maximal internalization (Fig. 4A). The EGFP-AT 1A receptors presented the same recycling properties as the EGFP-L305Q receptors (data not shown). This phenomenon was blocked when the cells were treated with 10 M monensin, which is known to inhibit recycling (Fig. 4B). These results show that: 1) the EGFP-tagged receptors recycle equally as well as the untagged receptors, and 2) that the L305Q CAM retains normal recycling properties.
Interestingly, the time dependence of losartan-induced externalization of the L305Q CAM resembled the kinetics of the recycling process. Indeed, after treatment of the HEK-EGFP-L305Q cells with losartan, the SЈ/CЈ ratio progressively increased until it reached a plateau at 2 h (Fig. 4C). In addition, when EGFP-L305Q cells were pretreated with monensin, losartan was no longer able to promote the externalization of the receptor, as the SЈ/CЈ ratio did not increase compared with untreated cells and even decreased slightly (Fig. 4D). Thus, the externalization process observed upon incubation with losartan was comparable to a recycling mechanism, recycled receptors being blocked at the cell surface by losartan.

Cytoplasmic Translocation of the EGFP-L305Q Receptor after the Withdrawal of Losartan Is Dependent on Internalization
Mechanisms-We questioned what would happen to the externalized receptors after the removal of losartan. Would they permanently remain at the plasma membrane, suggesting that the distribution of CAM receptors is static? Or would they go back into the cells, as would be expected if the observed distri-bution of CAM receptors is the result of a dynamic cycling process?
To answer these questions, we took advantage of the fact that losartan can be rapidly dissociated from the receptor by rinsing at 4°C and the sensitivity of the internalization process to temperature. After maximal externalization of the L305Q receptor with losartan, the inverse agonist was removed by rinsing at 4°C and the kinetic of reinternalization process was then observed at 37°C, by the microscopy re-internalization assay (see "Materials and Methods"). The plasma membrane fluorescence disappeared from HEK-EGFP-L305Q cells after 2.5 min at 37°C and the effect was maximal at 5 min (Fig. 5A). This was comparable to the kinetics of the WT receptor internalization (t1 ⁄2 ϭ 2.7 min (10)). In addition, the reinternalization of the L305Q receptor was abolished when the assay was performed at 16°C (instead of 37°C) (Fig. 5B), a temperature that also blocks the AngII-induced internalization of the WT receptor (Fig. 5C) and which blocks the clathrin-dependent internalization of other receptors (26).
These results show that losartan-induced externalization is a highly transient process, dependent on the presence of inverse agonists. Moreover, the removal of losartan results in the cytoplasmic translocation of the EGFP-L305Q receptor dependent on internalization mechanisms. The Double Mutant, EGFP-L305Q/⌬329, Is Localized at the Plasma Membrane-To confirm the role of internalization in the cellular localization of the CAM AT 1A receptors, we used a AT 1A receptor mutant truncated at residue 329, which presents a default of internalization (19). Both EGFP-⌬329 and EGFP-L305Q/⌬329 mutants present the same binding and signaling properties as the corresponding untagged receptors (data not shown). The basal IP production of EGFP-L305Q/⌬329 was four times higher than that of the WT and EGFP-⌬329 receptors (data not shown), showing that its constitutive activity had been conserved.
Both cells transfected with EGFP-⌬329 and with EGFP-L305Q/⌬329 show a fluorescence localized at the plasma membrane (Fig. 6A) and a similar SЈ/CЈ ratio to the WT receptor (Fig. 6B) in basal condition. In the double mutant, EGFP-L305Q/⌬329, the phenotypic trait of EGFP-L305Q (i.e. constitutive intracellular localization) was abolished by the ⌬329 truncation. These data strongly suggest that the CAMs of the AngII AT 1A receptor are constitutively internalized.
Functional Consequences of Constitutive Internalization and Externalization of CAM AT 1A Receptors-As the physiological response is quantitatively dependent on the number of binding sites at the cell surface, we questioned whether the constitutive internalization and losartan-induced externalization of the CAM receptors modulates their signaling properties. Instantaneous AngII-induced Ca 2ϩ mobilization was measured using an aequorin test on HEK-EGFP-AT 1A and HEK-EGFP-L305Q cells that had either been pretreated with losartan or left untreated. Losartan pretreatment induced a decrease of 30% in the calcium response to AngII for both the EGFP-AT 1A and the EGFP-L305Q receptors (Fig. 7A). Analysis of the accumulation of IP over a 30-min period in the same experimental conditions show that losartan pretreatment did not have a major effect on the EGFP-AT 1A and the EGFP-L305Q receptors (Fig. 7B). These data suggest a dual effect of the inverse agonist, losartan, which increases the number of CAM receptors at the cell surface and also stabilizes them in an inactive state. DISCUSSION In the present study, we investigated the links between activation and the cellular localization of the EGFP-tagged CAMs (N111A, I245T, and L305Q) of the AngII AT 1A receptor. We observed that the CAMs were mostly localized in intracellular compartments and that inverse agonists, but not agonists or neutral antagonists, induced their translocation to the plasma membrane. A cytoplasmic localization and a losartaninduced externalization of the mutant L305Q receptor was also observed in another more physiological cellular model (H295), which expresses endogenous AT 1 receptors. Its intracellular localization is less pronounced than that observed in transfected HEK-293 cells, probably due to differences in the kinetics and efficiency of internalization and recycling of the AT 1 receptor between the two cell types. These observations are not due to the impaired folding of the CAMs, which would prevent them from exiting the biosynthetic pathway. We showed indeed that the biosynthesis of the L305Q CAM and WT AT 1A are identical. Imidazole-like inverse agonists cannot cross the plasma membrane and therefore cannot stabilize the unfolded receptor intracellularly, as shown for a misfolded V2 vasopressin mutant (27). Several pieces of evidence suggest that it is more likely that the CAMs of the AT 1A receptor are permanently internalized and recycled: first, the plasma membrane translocation of the L305Q CAM, or externalization, induced by inverse agonist has the same kinetics and pharmacology as the recycling mechanism. Second, after externalization by the inverse agonists, L305Q CAM is quickly and spontaneously readdressed to the cytoplasm upon ligand removal by a mechanism comparable to internalization. Third, the double mutant, EGFP-L305Q/⌬329, which is constitutively active but lacks a domain required for internalization, is localized at the plasma membrane. Altogether, these data indicate that the CAMs of the AngII AT 1A receptor are constitutively and permanently internalized and recycled. Inverse agonists block the CAM in an inactive state when the recycled receptor reaches the plasma membrane, thus preventing the rapid re-internalization and resulting in the accumulation of the receptor at the membrane (Fig. 8).
Although the spontaneous internalization of the CAMs of the GPCR family has long been considered as a likely possibility, no direct evidence is available yet. Several CAMs are present within intracellular compartments, such as mutants of the PTH receptor (28), ␣ 1B adrenergic receptor (AR) (29), and the yeast pheromone receptors, Ste2p and Ste3p (30). Other studies have shown that CAMs can be spontaneously phosphorylated, desensitized, and/or down-regulated, but they have not provided evidence for constitutive internalization (31). Basal phosphorylation is enhanced for two CAMs of the human LH receptor (32,33) and for the CAMs of the ␣ 1B (34) and ␤ 2 -AR (35). The ␤ 2 -AR CAM is also constitutively desensitized and constitutively down-regulated and both phenomena are reversed by overnight treatment with the inverse agonist, betaxolol (35). Constitutive down-regulation reversed by inverse agonists was later observed for several other CAMs, including mutants of the TRH receptor (36) and the ␣ 1B AR (29,37), and also for the WT histamine receptor H2, which exhibits natural constitutive activity (38). In all cases, the effects of the inverse agonists are only observed after a long period of treatment, consistent with the stabilization of newly synthesized receptors. For example, the levels of CAM ␤ 2 -AR are increased 4 -7-fold after 24 h treatment, through a protein synthesis-dependent process that does not change the subcellular distribution of the receptor (39). This is thus believed to result from the constitutive addressing of the mutants to degradation pathways. In some cases, inherent instability and/or folding defects of the CAMs are involved in this down-regulation process. This is clearly the case for the CAMs of the yeast receptors, Ste2p and Ste3p, which remain stuck in the biosynthesis pathway because of impaired folding (30). Stabilizing effects explain how both inverse agonists and agonists can up-regulate the levels of Step2 and Step3 CAMs, but also of an ␣ 2A CAM (40) and a ␤ 2 CAM in Sf9 cells (41) or even in vivo (42).
Conversely, the phenomenon observed here is not linked to the down-regulation or instability of the CAMs of the AT 1A receptor. It is independent of protein synthesis, as all assays were done on cells treated with cycloheximide. It does not result from the instability of the CAMs, as their metabolism was similar to the WT receptor. It does not involve protein stabilization as other peptide or non-peptide ligands, which all differ from inverse agonists by their ability to induce internalization, were unable to relocalize the receptor to the plasma membrane. Our results suggest a very different mechanism and strongly suggest that AT 1A CAMs are constitutively and permanently internalized and are then recycled. It is not clear whether the same behavior occurs for other CAMs in the GPCR family, but there are several reasons why this phenomenon has never been reported before. First, due to the difficulty in obtaining antibodies and the recent introduction of epitope fluorescent tagging, the cellular distribution of GPCR has only been studied morphologically in a limited number of GPCR. Second, the internalization/ recycling kinetics of the AT 1A receptor differ from those of other classical GPCR because it is rapidly internalized (within minutes) and slowly recycled (within hours), whereas other GPCR are either rapidly internalized and recycled (␤ 2 -AR) or not recycled but degraded (LH receptor). The peculiar kinetic of the AT 1A receptor may favor the intracellular accumulation of the receptor.
Some examples in the literature are partially reminiscent of our observations, suggesting that the phenomenon described here may be relevant to other CAMs. Although the kinetics are different, two WT GPCRs have been shown to be constitutively internalized: the thrombin receptor (43) and the cholecystokinin receptor type A (44). The WT ␣ 1D AR is naturally constitutively active and mostly localized in intracellular compartments, whereas 24 h in the presence of prazosin causes redistribution of the receptor from intracellular sites to cellular periphery (45). Finally, other examples include a deletion mutant of the -opioid receptor (46), which is constitutively internalized and recycled, and an inactive mutant of the human vasopressin receptor, which is constitutively sequestered in arrestin-associated intracellular vesicles (47). These mutants are not constitutively active, but demonstrate that GPCRs can be constitutively desensitized by mechanisms distinct from constitutive down-regulation.
Although they do not rule out the permanent cycling of CAMs, recent results on the phosphorylation of the N111A and N111G mutants of the AT 1A receptor raise the question of the molecular mechanisms regulating this constitutive internalization. Unexpectedly, the phosphorylation of N111A and N111G mutants was not elevated in basal conditions and, unlike the WT receptor, it was not increased by AngII treatment (48). However, the CAMs are normally internalized in response to AngII (Fig. 2C and Ref. 48). The phosphorylation status of the two other mutants studied here (I245T and L305Q) is not known, but the study by Thomas et al. (48) suggests that the AngII-induced phosphorylation of the AT 1A receptor is not mandatory for internalization and more generally, that phosphorylation and internalization can be dissociated. This is also in agreement with the fact that phosphorylation is not mandatory for arrestin binding (49,50).
Another major difference between our study and previous reports on the regulation of CAMs is the functional consequence of treatment with inverse agonists. In the case of CAMs of the ␣ 1B AR (37), the TRH receptor (36), and the ␤ 2 AR (35), the up-regulation of receptor levels induced by inverse agonists was accompanied by highly enhanced signaling responses. Conversely, after losartan-induced externalization of the AT 1A CAM, we observed a reduced AngII-induced calcium mobilization and no significant change in IP turnover. Unfortunately, pretreatment with losartan does not only block the receptor at the membrane, but also partly desensitizes it, as shown by the calcium signaling of the WT receptor. Therefore, the unchanged or even reduced signaling efficiency of the CAM after losartan treatment is probably due to the small positive effect In the basal state the wild-type AT 1A receptor was mostly localized at the cell surface. Inverse agonists (such as losartan and irbesartan) had no effect on the cellular localization, whereas AngII induced the rapid internalization (i) of the receptor in intracellular vesicles. The receptor was then slowly and partly recycled (r) at the plasma membrane. In the basal state the CAMs of the AT 1A receptor were mostly localized in intracellular vesicles. Losartan induced the plasma membrane translocation of the CAMs. In the basal state or after losartan-induced externalization, AngII induced the internalization of the receptor in intracellular vesicles. of the increased number of receptors at the cell surface, which is counterbalanced by partial inactivation. Another possible explanation is that, as previously discussed by Milligan and Bond (51), the signaling efficiency is dependent on the levels of expression of G proteins and second messenger generating enzymes. For example, treatment with inverse agonists did not enhance signaling for the ␤ 2 CAM when the mutant was expressed in NG108-15 cells (52).
The initial purpose of this study was to gain insight into the link between the activation and internalization of the AT 1A receptor. We wanted to determine whether the receptor is inexorably targeted for internalization by the cell machinery when the receptor is in the active conformation and whether activation necessarily results in internalization. The C-terminal deletion mutant, ⌬329, couples to the G protein but does not become internalized (19,53), suggesting that activation can be dissociated from internalization. However, this is due to the absence of a large domain necessary for the interaction with internalization machinery rather than a difference in the overall conformation of the receptor. More interestingly, the peptide ligand [Sar 1 ,Ile 4 ,Ile 8 ]AngII has been reported to activate the N111A and N111G mutants without inducing their internalization (data not shown in Ref. 48). In contrast, the results presented here suggest that the active conformation of the AT 1A receptor is an "internalization-sensitive" conformation. The three CAMs studied carry mutations at different positions in the transmembrane domains but present similar patterns of cellular distribution. This suggests that the active conformation of the WT receptor cannot avoid internalization.
Conversely, the AT 1A receptor can quite easily adopt conformations that do not activate signaling but are recognized for internalization. Non-signaling mutants of the AT 1A receptor (54,55) are internalized in response to AngII to the same degree as the WT receptor (8,55). AngII peptide antagonists are able to induce internalization of the receptor (8,10). This activation-independent internalization also takes place for other GPCRs, and is also supported by the fact that mutants of the -opioid and vasopressin receptors are constitutively internalized without being constitutively active (46,47). These results suggest that the activation of the signaling pathways by a GPCR requires a much more specific conformation that the conformation required to trigger internalization. As arrestin binding is mandatory and is the first step for the internalization of these GPCRs, arrestin probably recognizes a broader spectrum of conformations that the G proteins.
In conclusion, this study shows that the CAMs of the AngII AT 1A receptor are constitutively and permanently internalized and recycled. The externalization phenomenon described here should define a new paradigm for agonist-independent CAM regulation, distinct from the strong down-regulation observed for a number of other CAMs in the GPCR family and first exemplified for the ␤ 2 -AR (35). Furthermore, this study provides important insights on the molecular determinants of activation and internalization.