Role for the Third Intracellular Loop in Cell Surface Stabilization of the α2A-Adrenergic Receptor*

Previous studies have shown that α2A-adrenergic receptor (α2A-AR) retention at the basolateral surface of polarized MDCKII cells involves its third intracellular (3i loop). The present studies examining mutant α2A-ARs possessing short deletions of the 3i loop indicate that no single region can completely account for the accelerated surface turnover of the Δ3iα2A-AR, suggesting that the entire 3i loop is involved in basolateral retention. Both wild-type and Δ3i loop α2A-ARs are extracted from polarized Madin-Darby canine kidney (MDCK) cells with 0.2% Triton X-100 and with a similar concentration/response profile, suggesting that Triton X-100-resistant interactions of the α2A-AR with cytoskeletal proteins are not involved in receptor retention on the basolateral surface. The indistinguishable basolateral t 1 2 for either the wild-type or nonsense 3i loop α2A-AR suggests that the stabilizing properties of the α2A-AR 3i loop are not uniquely dependent on a specific sequence of amino acids. The accelerated turnover of Δ3i α2A-AR cannot be attributed to alteration in agonist-elicited α2A-AR redistribution, because α2A-ARs are not down-regulated in response to agonist. Taken together, the present studies show that stabilization of the α2A-AR on the basolateral surface of MDCKII cells involves multiple mechanisms, with the third intracellular loop playing a central role in regulating these processes.

The ␣ 2 -adrenergic receptor (␣ 2 -AR) 1 is a member of a large family of G protein-coupled receptors that are predicted to have seven transmembrane-spanning regions (1,2). Three subtypes of ␣ 2 -ARs exist and couple to members of the G i and G o class of G-proteins to mediate a variety of physiological responses (3,4).
Receptor localization and stabilization on the cell surface of target cells are two critical contributors to the sensitivity and extent of signaling by G protein-coupled receptors. There is a growing body of evidence that discrete localization of G proteincoupled receptors may play a role in specificity of signaling by these receptors (5)(6)(7)(8). A precedent already exists for the microcompartmentation of signaling molecules such as protein kinase C (9), cAMP-dependent protein kinase (10), Ca 2ϩ /calmodulin-dependent protein kinase II (11), kinases involved in the yeast mating response (12), and NO synthase (13,14) by interaction of these effector molecules with signaling "scaffold" proteins.
In polarized cells, receptor localization is essential for vectorial information transfer, as occurs for ␣ 2 -AR regulation of Na ϩ and H 2 O transport in renal (15) and intestinal (16,17) epithelial cells. Madin-Darby canine kidney (MDCKII) cells cultured in Transwell culture dishes have provided an excellent model system for polarized renal epithelial cells. The localization of the ␣ 2A -AR subtype on the basolateral surface of these cells (18) recapitulates the basolateral localization of this receptor in vivo, based on physiological (19) and pharmacological (20) data.
Examination of molecular regions of the three ␣ 2 -AR subtypes in polarized MDCKII cells indicates that basolateral targeting of these receptors involves sequences in or near the membrane bilayer (18,21,22). In contrast, the large third intracellular loop of the receptor appears to play a role in stabilizing the ␣ 2A -AR on the plasma membrane, because mutant ␣ 2A -ARs that lack 119 amino acids from the large third intracellular loop (⌬3i loop ␣ 2A -AR) have a cell surface half-life (t1 ⁄2 ) of 4.5 h compared with a t1 ⁄2 of 10 h for the wild-type ␣ 2A -AR (21). The present studies have explored the structural features of the ␣ 2A -AR 3i loop responsible for stabilizing the ␣ 2A -AR on the plasma membrane of MDCKII cells.
Cell Culture-Madin-Darby canine kidney cells (MDCKII) were obtained from Enrique Rodriguez-Boulan (Cornell Univ. New York, NY) and maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Sigma), 100 units/ml penicillin, and 100 mg/ml streptomycin (referred to as complete Dulbecco's modified Eagle's medium) at 37°C, 5% CO 2 . For polarity experiments, MDCK II cells were seeded at a density of 1 ϫ 10 6 cells/24.5-mm polycarbonate membrane filter (Transwell chambers, 0.4-m pore size, Costar, Cambridge, MA) and cultured for 5-8 days with a change of medium every 1-2 days. Before each experiment, the integrity of the monolayer was assessed by adding [ 3 H]methoxyinulin to the apical medium and monitoring its leak after a 1-h incubation at 37°C from the apical to the basolateral compartment by sampling and counting the basolateral medium in a scintillation counter (Packard Tricarb). Chambers with greater than 5% leak/h were not evaluated. been previously described (21,24). Briefly, site-directed mutagenesis was used to create a novel NotI restriction enzyme site in the region of the ␣ 2A -AR cDNA encoding the C-terminal end of the putative third intracellular (3i) loop. Cleavage of this mutant ␣ 2A -AR cDNA with NotI restriction enzymes removes the DNA fragment encoding amino acids 240 -359. This ⌬3i loop mutant receptor has 36 amino acids linking transmembrane domains 5 and 6 of the ␣ 2A -AR as shown in Fig. 1A.
Oligo-directed mutagenesis in M13 phage was utilized to create incremental deletions of the predicted third intracellular loop of the ␣ 2A -AR. Single oligos were designed against sequences flanking those encoding the amino acids selected for each deletion. The deletion mutations were confirmed by dideoxynucleotide sequencing and then subcloned into the pCMV4 mammalian expression vector. Deletions corresponding to DNA encoding the following amino acids were made in this manner: ⌬aa252-267, ⌬aa268 -285, ⌬aa286 -303, ⌬aa315-326, and ⌬aa327-340. Fig. 1 provides a schematic diagram of the regions encoded by these deleted amino acids.
The nonsense loop was designed by taking advantage of the method used for making the original ⌬aa240 -359 (⌬3i loop ␣ 2A -AR). Because two NotI enzyme sites were used to remove the 3i loop, it was possible to subclone this segment of the gene back into the receptor in two orientations. The correct orientation produced a receptor that corresponded to the wild-type ␣ 2A -AR sequence except for two point mutations at the site of the engineered NotI site (K359A, S360A). The opposite orientation also produces an open reading frame with a 3i loop of the same amino acid length but with very little sequence homology to the wild-type receptor (See Fig. 4). We refer to this mutant ␣ 2A -AR as the nonsense 3i loop ␣ 2A -AR.
All mutations were verified using dideoxy-DNA sequencing (Sequenase kit, U. S. Biochemical Corp.) of the single-stranded DNA utilizing T7 DNA polymerase with ␣-35 S-dATP. Once verified, the mutant inserts were subcloned from M13 into the pCMV4-TAG-␣ 2A -AR expression vector (18) containing an N-terminal hemagglutinin epitope (YPYDVP-DYA) to which antibodies are available commercially (Berkeley Antibody Co.). These plasmid constructs were verified by double-stranded DNA sequencing through the region of the mutation. COS M6 cells were transiently transfected with the plasmid DNA encoding the wild-type and mutant ␣ 2A -ARs, and membranes from the transient transfectants were assayed for [ 3 H]yohimbine binding before developing permanent MDCK cell lines expressing these mutant ␣ 2A -AR. MDCK cell lines permanently expressing the wild-type or mutant ␣ 2A -AR were created as described previously (18) (Table I).
Determination of the Half-life of Wild-type or Mutant ␣ 2A -AR on the Basolateral Membrane-To determine ␣ 2A -AR half-life on the basolateral surface, a metabolic labeling strategy was employed. MDCK cells expressing wild-type or mutant ␣ 2A -AR were incubated with [ 35 S]Cys/ Met ("pulse") and then incubated for various periods of time ("chase") before isolation of basolateral ␣ 2A -AR using sequential biotinylation, extraction, immunoisolation, and streptavidin-agarose chromatogra-phy. The procedures utilized have been previously described (21) except for the modifications outlined as follows. Specifically, MDCK cells grown in Transwell culture were metabolically labeled with 1 Ci/l 35 S-Express protein labeling mix for 60 min in Cys/Met-free Dulbecco's modified Eagle's medium (18). After labeling, the cells were washed once with Dulbecco's phosphate-buffered saline (dPBS) and incubated for various periods of time (generally 0, 3, 6, and 18 h) at 37°C, 5% CO 2 in chase medium (complete Dulbecco's modified Eagle's medium supplemented with 1 mM cysteine and 1 mM methionine). At the conclusion of each chase period, ␣ 2A -ARs residing on the basolateral surface of these cells were isolated by biotinylating the basolateral cell surface with biotin hydrazide or sulfo-NHS biotin and subjecting detergent extracts of membranes prepared from these cells to sequential immunoprecipitation with the 12CA5 anti-hemagglutinin epitope antibody followed by streptavidin-agarose chromatography (21). The streptavidin-agarose eluates were separated by SDS-polyacrylamide gel electrophoresis, and the amount of radiolabeled ␣ 2A -AR remaining on the basolateral surface after various durations of chase was determined. The gels were exposed to film, and the gel area corresponding to the radiolabeled ␣ 2A -AR was removed and counted (in 10 ml of scintillation Fluor) in a Packard beta counter. Similar-sized gel slices that did not correspond to any radiolabeled protein band were excised and counted to quantify the background 35  To assess the effects of agonist on ␣ 2A -AR redistribution, MDCKII cells expressing either wild-type or ⌬3i loop ␣ 2A -AR were treated for 24 h with 100 M epinephrine and 100 M ascorbic acid to prevent oxidation of the epinephrine, as described previously (25). Control cells were treated with 100 M ascorbic acid alone. Receptor density was determined by [ 3 H]yohimbine saturation binding.
Determining the Triton X-100 Extractability of Wild-type and ⌬3i Loop ␣ 2A -AR-MDCK cells permanently transfected with either wildtype or ⌬3i loop ␣ 2A -AR were grown on Transwells for 7 days. Transepithelial leak of [ 3 H]methoxyinulin was determined as described earlier for confirmation of functional polarization. To compare the Triton X-100 extractability of wild-type and ⌬3i loop ␣ 2A -AR, the ␣ 2A -ARs in intact cells were covalently modified via a radioiodinated photoaffinity label and then extracted with increasing concentrations of Triton X-100 (%v/v). Briefly, cells were washed with dPBS supplemented with 0.5 mM CaCl 2 and 1.0 mM MgCl 2 (dPBS-CM), and the Transwells were inverted and incubated 1 h at 22°C in the dark with 150 l of dPBS-CM containing 0.2 Ci/well (0.9 nM) 125 I-Rau-Az-Pec. After 1 h, the wells were washed with dPBS containing 1 mM glutathione, suspended in a Rayonet UV photoilluminator, and photolysed for 3 min with 300-nm light. The ␣ 2A -ARs expressed on the basolateral surface were identified by biotinylation with Sulfo-NHS-biotin in triethanolamine buffer for 2 20-min incubations as described above.
To determine the Triton X-100 extractability of ␣ 2A -ARs in polarized cells, Transwells with the photolabeled ␣ 2A -AR were washed with dPBS and extracted with increasing concentrations of Triton X-100 using a modification of a previously described protocol (26). The polycarbonate filters were first removed from the Transwell support and placed into a 12-well dish with each 24-mm well containing 190 l of a Triton X-100 extraction buffer (15 mM Tris, pH 8, 120 mM NaCl, 25 mM KCl, 0.1 mM EGTA, 0.5 mM EDTA) with no Triton X-100. The cells on polycarbonate filters were rocked gently for 5 min at 4°C. The filters were then transferred to a well with extraction buffer containing 0.05% Triton X-100 and rocked 5 min. This procedure was repeated with increasing concentrations of Triton X-100 (0.1, 0.2, 0.5, and 1.0%). After exposure to 1% Triton X-100, the residual cellular material, operationally defined as "Triton shells," was scraped into 200 l of RIPA buffer. A set of control Transwells were subjected to the same procedure, except that each successive buffer contained no Triton X-100. The final RIPA extraction buffer from each well was transferred to a 0.6-ml Eppendorf tube, and the wells were washed with 200 l of RIPA buffer. All extracts were brought up to a final volume of 500 l with RIPA buffer, and biotinylated proteins were isolated using streptavidin-agarose chromatography. After an overnight incubation, the streptavidin beads were eluted with 1ϫ SDS sample buffer at 90°C for 30 min. This elution was repeated, and the combined eluates were loaded onto a 10% SDSpolyacrylamide gel. The dried gels were exposed to preflashed Kodak film for 3-5 days. The receptor was identified as a radioactive band migrating at a position characteristic of the ␣ 2A -AR and whose photoaffinity-labeling was blocked in the presence of 10 M phentolamine in separate control studies.

No Small Region in the ␣ 2A -AR Third Intracellular Loop Contains All of the Necessary Information for Stabilization of the Receptor on the Cell Surface-
We observed previously that deletion of 119 amino acids from the 3i loop of the ␣ 2A -AR (amino acids 240 -359) generates a structure (⌬3i ␣ 2A -AR) that has a basolateral t1 ⁄2 of ϳ4.5 h compared with 10 -12 h for the wild-type ␣ 2A -AR in polarized MDCKII cells. By analogy with the ability of a 21-amino acid insert into the short (D2S) dopamine receptor to create the long dopamine receptor isoform (D2L) and dramatically slow the rate of sequestration (27), the 3i loop of the ␣ 2A -AR was examined to determine whether a single small amino acid sequence could account for the stabilization of the receptor.
Five ϳ20 amino acids deletions were made within the ␣ 2A -AR 3i loop, as shown schematically in Fig. 1A. Demarcation of the regions selected for individual deletions was based on secondary structural predictions of Chou and Fasman analysis (52); for example, ⌬aa286 -303 and ⌬aa315-326 are predicted by this analysis to form amphipathic ␣ helices. In addition, the ⌬aa286 -303 removes the LEESSSS sequence recognized for phosphorylation by G protein-coupled receptor kinases (28,29). This was of interest because G protein-coupled receptor kinasemediated phosphorylation of these receptors promotes association with arrestins that have been shown to act as adaptors and recruit some G protein-coupled receptors into clathrincoated pits (30 -32).
The surface stability for each of the ␣ 2A -AR structures examined (Fig. 1A) was determined by pulse/chase metabolic labeling strategies and isolation of ␣ 2A -AR on the basolateral surface by sequential biotinylation and streptavidin-agarose isolation of detergent-solubilized receptor (see "Experimental Procedures"). As shown in Fig. 1B, ⌬aa286 - Direct Cytoskeletal Interactions Do Not Appear to Account for Stabilization of the ␣ 2A -AR via Its 3i Loop-One mechanism that might account for stabilization of the ␣ 2A -AR on the basolateral surface of polarized renal epithelial cells could be direct interaction of the receptor with the underlying cytoskeleton. In this case, accelerated turnover of the ⌬3i loop ␣ 2A -AR could result from loss of these direct cytoskeletal interactions. We utilized differential sensitivity to extraction by Triton X-100 as an indicator of direct and stable association with the cytoskeleton (33)(34)(35). This approach has been informative in revealing the association of the polytopic Na ϩ -K ϩ -ATPase (26,36) and of the single transmembrane-spanning CD44 protein (37) with the cadherin-dependent ankyrin-fodrin matrix underlying the basolateral surface of polarized MDCK cells.
As shown in Fig. 2A, both the wild-type ␣ 2A -AR and the ⌬3i ␣ 2A -AR are released from polarized MDCKII cells when exposed to 0.2% Triton X-100 in the presence of 2 mM EDTA and 2 mM EGTA (0 mM [Ca 2ϩ ] o and 0 mM [Mg 2ϩ ] o ). For comparison, proteins directly associated with the cytoskeleton, such as the Na ϩ -K ϩ -ATPase, are not completely extracted by 0.5% Triton X-100 under similar, but slightly more stringent, divalent cation-free extracellular conditions, whereas the ␣ 2A -AR is completely extracted (26,36), 2 suggesting that the ␣ 2A -AR does not interact directly or stably with the cytoskeleton. When Triton X-100 extractions were performed in the presence of 0.5 mM Ca 2ϩ and 1 mM Mg 2ϩ , the amount of Triton X-100 required for extraction of Ն60% of the photoaffinity-labeled surface receptors was increased from 0.2% ( Fig. 2A) to 0.5% (Fig. 2B) but with no difference in extraction efficiency between wild-type and mutant ␣ 2A -AR. These data suggest that ␣ 2A -AR stability on the basolateral surface is influenced by protein-protein interactions that involve a Ca 2ϩ (likely cadherin)-organized substratum, but that these interactions cannot explain the difference in cell surface stability of the wild-type and ⌬3i loop structures, as they are extracted in a comparable manner even in the presence of Ca 2ϩ .
Specific Amino Acid Sequences Within the Third Intracellular Loop Do Not Appear to Be Required for Stabilization of the ␣ 2A -AR on the Cell Surface-If bulk size of the 3i loop is sufficient for stabilization of the ␣ 2A -AR, then a loop containing the same number of amino acids as the wild-type ␣ 2A -AR should manifest the same surface t1 ⁄2 regardless of the primary sequence within the 3i loop. To test this hypothesis, we constructed a receptor that contains a nonsense 3i loop corresponding to the exact number of amino acids as in the wild-type ␣ 2A -AR 3i loop but with very little sequence homology. The sequences of the wild-type and nonsense ␣ 2A -AR 3i loops are compared in Fig. 3. In four experiments using two different clonal cell lines expressing the ␣ 2A -AR 3i nonsense loop, the half-life of this structure was indistinguishable from that characteristic of the wild-type receptor as shown in Fig. 3.
These findings are consistent with a mechanism where the size of the 3i loop structure plays an important role in stability of the ␣ 2A -AR on the cell surface. There are examples of membrane proteins localized in surface microdomains by virtue of so-called "corrals," often established by the cytoskeletal proteins underlying the cell surface (38 -40). Consequently, ␣ 2A -AR surface stability might arise by steric principles, dictated by the size of the 3i loop (Fig. 3). If corrals partitioned the ␣ 2A -AR, and lack of corralling were responsible for accelerated turnover of the ⌬ 3i loop ␣ 2A -AR, then we should expect a more rapid lateral diffusion coefficient and a significantly greater mobile fraction for the ⌬3i loop ␣ 2A -AR. However, Uhlén et al. (41) have previously reported that the difference in surface half-life between ␣ 2A -AR and ⌬3i loop ␣ 2A -AR is not paralleled by a difference in lateral diffusion coefficients for each receptor structure, estimated at ϳ2.2 ϫ 10 Ϫ10 cm 2 /s using fluorescence recovery after photobleaching. Thus, the mechanistic significance of the retention of the ␣ 2A -AR 3i nonsense loop on the basolateral surface for a duration comparable with the wildtype receptor is unexplained at present.
Sustained Agonist Exposure in MDCKII Cells Does Not Decrease Receptor Density for Either Wild-type or ⌬3i loop ␣ 2A -AR-One mechanism that might account for accelerated sur-face turnover of the ⌬3i loop ␣ 2A -AR would be enhanced agonist-elicited redistribution and subsequent down-regulation of this mutant receptor compared with the wild-type receptor. Consequently, we examined the effect of prolonged agonist exposure on steady-state ␣ 2A -AR density in MDCKII cells expressing wild-type or ⌬3i loop ␣ 2A -AR. As shown in Fig. 4, treatment of MDCKII cells with 100 M epinephrine for 24 h results in no detectable down-regulation of either the wild-type or the ⌬3i loop ␣ 2A -AR. In fact, there is even a slight increase in receptor density following agonist incubation, perhaps because of ligand-dependent receptor stabilization (42). 3 These findings are consistent with previous reports that the ␣ 2A -AR subtype does not undergo agonist-induced down-regulation in MDCKII cells (43) and Chinese hamster fibroblast cells (28), although this subtype has been reported to down-regulate in Chinese hamster ovary cells (25,44,45). In addition, the ␣ 2A -AR subtype, in contrast to the ␣ 2B -AR subtype, does not manifest agonist-elicited redistribution (46), 4 nor is there any evidence for intracellular localization of the ⌬3i loop ␣ 2A -AR in MDCKII cells either by immunocytochemistry (48) or cell surface biotinylation. 5 In contrast to our findings and the lack of effect of removal of the 3i loop on agonist-elicited ␣ 2A -AR redistribution, several muscarinic receptor subtypes have been shown to undergo agonist-elicited sequestration and down-regulation in a manner influenced by the 3i loop; mutations within the 3i loop reduce sequestration (47, 49 -51) and deletion of the 3i loop slows the rate of down-regulation for the m2 receptor (47). . After exposure to 1% Triton X-100, the residual cellular material, defined as Triton shells, was solubilized with RIPA buffer (Non Ext.). Biotinylated proteins were isolated using streptavidin-agarose chromatography. The ␣ 2A -AR in the eluates was resolved on a 10% SDS-polyacrylamide gel. Control experiments indicated that the radioactive band shown corresponds to 125 I-Rau-AzPec photoaffinity-labeled ␣ 2A -AR, based on its relative migration on 10% gels and the blockade of its labeling by ␣ 2A -AR antagonists. The results shown compare wildtype ␣ 2A -AR (Tag3 clone at 25 pmol/mg of protein) and ⌬3i loop ␣ 2A -AR (T3 at 3.4 pmol/mg of protein (A) or T66B at 2.8 pmol/mg of protein (B)). These data are representative of at least three separate experiments. This extraction profile is not dependent on receptor density because two cell lines with nearly 10-fold different levels of wild-type ␣ 2A -AR expression (Tag3 clone at 25 pmol/mg of protein versus T24 clone at 3.4 pmol/mg of protein) were examined with the same results. MDCKII cells expressing either wild-type or ⌬3i loop ␣ 2A -AR were treated for 24 h with 100 M epinephrine (ϩ100 M ascorbic acid to prevent epinephrine oxidation). Control cells were treated with 100 M ascorbic acid alone. Receptor density was determined by [ 3 H]yohimbine binding as described under "Experimental Procedures." Results are the mean ϮS.E. from six separate experiments. Results shown are from MDCKII cells grown to confluence on 60-mm cell culture dishes. However, analysis of MDCKII cells polarized in Transwells gave the same results. No decrease in receptor density was observed; in fact, an increase was noted by analogy with findings in other cultured cell systems 3 suggesting that ligand occupancy stabilizes receptor density and affirming that epinephrine remains viable during the course of the incubation. specific amino acid sequences are not necessarily required for basolateral retention and suggesting that the bulk of the 3i loop may be sufficient to stabilize the ␣ 2A -AR on the basolateral surface.