Regulated Interactions of the (cid:1) 2A Adrenergic Receptor with Spinophilin, 14-3-3 (cid:2) , and Arrestin 3*

The present studies demonstrate that no single stretch of sequence in the third intracellular (3i) loop of the (cid:1) 2A adrenergic receptor ( (cid:1) 2A -AR) can fully account for its previously described interactions with spinophilin (Richman, J. G., Brady, A. E., J. L., and Limbird, L. Biol. Chem. 14-3-3 (cid:2) (Prezeau, L., Biol. Chem. and arrestin 3 (Wu, and Lanier, S. J. Biol. Chem. 272, 17836–17842), suggesting that a three-dimensional surface, rather than a linear sequence, provides the basis for these interactions as proposed for 3i loop tethering of the (cid:1) 2A -AR to the basolateral surface of Madin-Darby canine kidney cells (Edwards, S. W., and Limbird, L. E. J. Biol. Chem. 274, 16331–16336). Sequences at the extreme N-terminal and C-terminal ends of the 3i loop are critical for interaction with spinophilin but not for interaction with 14-3-3 (cid:2)

The ␣ 2 -ARs often mediate their responses in highly polarized cells, such as epithelial cells from the kidney and intestine or neurons of the peripheral and central nervous system (9). We have studied the polarization of the ␣ 2 -AR subtypes extensively in Madin-Darby canine kidney (MDCKII) cells as a model system. In these studies, we demonstrated that basolateral targeting of the ␣ 2A -AR required sequences embedded in or near the bilayer, whereas retention on that surface involved the third intracellular (3i) loop (10). These findings suggested the possibility that the 3i loop may interact with proteins underlying the basolateral surface to extend its half-life there.
In studies intended to identify 3i loop-interacting proteins, we demonstrated that the 3i loop interacts with 14-3-3 (11) as well as with spinophilin (12). Receptor interactions with spinophilin, in particular, are enhanced by agonist occupancy of the receptor (12). Morphological studies established that both 14-3-3 2 and spinophilin (12)(13)(14) are enriched at the basolateral surface of MDCKII cells, although these proteins also are expressed widely throughout the cytoplasm.
To further understand the role of receptor interactions with spinophilin and/or 14-3-3 proteins, we sought to identify the particular regions of the ␣ 2A -AR 3i loop sequence involved in the protein-protein contacts. The present studies examine the basis for ␣ 2A -AR 3i loop interactions with spinophilin, 14-3-3 proteins, and arrestin 3 (also known as ␤-arrestin 2), another molecule known to interact with the ␣ 2A -AR 3i loop (15,16). We also explored the functional relevance of these interactions by assessing the ability of spinophilin or arrestin to compete for receptor interactions upon agonist occupancy. The data obtained are consistent with a regulatory cycle where agonistregulated interactions of these proteins occur, perhaps relevant for ␣ 2A -AR-mediated functions and/or localization.
Synthesis of Radio-labeled Wild Type and Mutant ␣ 2A -AR 3i Loops-Four methionines were inserted via PCR into the N terminus of the porcine ␣ 2A -AR 3i loop (amino acids 218 -377 (12)), and the resulting product was subcloned into the modified pGEMEX2 vector in which the sequence encoding viral coat protein gene 10 was deleted. The resulting construct, referred to as (Met) 4 -␣ 2A 3i (structure WT in Fig. 3 Wild type and mutant 3i loops were transcribed, translated, and 35 S-labeled using the Promega transcription and translation-coupled (TNT) rabbit reticulocyte lysate kit as described previously (11). Following each synthesis, products were analyzed and quantitated by 12% SDS-PAGE and autoradiography. The band representing each probe was cut out of the dried gel and counted in scintillation mixture to determine cpm/aliquot of product attributable to 35 S-labeled 3i loops. GST pull-down assays were performed such that each incubation contained an equivalent amount of 35 S-labeled wild type or mutant 3i loops.
Preparation of GST-␣ 2A -AR 3i Loops-A DNA fragment encoding the ␣ 2A -AR 3i loop amino acids 239 -370, representing the NotI/Eco47III fragment from the cDNA encoding the porcine ␣ 2A -AR, was blunted and then subcloned into the pGEX2T SmaI site in frame with the DNA encoding GST; the resulting construct is referred to as GST-␣ 2A 3i short (GST-␣ 2A 3is, structure WT S in Fig. 4 Fig. 4). The N-terminal half of the 3i loop (aa 218 -294) was fused with GST by subcloning the EcoRI/ HindIII (blunt) fragment from structure 9 into pGEX2T/EcoRI (blunt) vector; the resulting structure is referred to as structure 28. In all studies where GST fusion proteins were employed, we stained the same SDS-PAGE gel as evaluated with autoradiography with Coomassie Blue to confirm that equivalent amounts were present in all incubations.
Preparation of GST-Spinophilin-(151-444) Fusion Protein-GSTspinophilin fusion protein was generated with spinophilin amino acid regions 151-444 (referred to as GST-Sp151-444) and expressed in DH5␣ or BL21 cells as described previously (12). Briefly, bacteria were grown at 37°C to an A 600 ϭ 0.6. GST or GST fusion protein expression was initiated with the addition of 1 mM isopropyl-␤-D-thiogalactopyranoside and allowed to proceed for 2-6 h at 37°C. Bacteria were collected and lysed by probe sonication for three 30-s bursts on ice in TT ϩ buffer (50 mM Tris-HCl, pH 7.4, 0.5% Triton X-100, 1 mg/ml lysozyme, 200 mM NaCl, 100 M PMSF, 1 g/ml soybean trypsin inhibitor, 1 g/ml leupeptin, 10 units/ml aprotinin). The lysate was centrifuged for 10 min at 13,000 ϫ g at 4°C. The supernatant (15 ml) of this centrifugation was incubated with GSH-agarose (1 ml of a 1:1 slurry equilibrated in TT ϩ buffer) for 1 h at 4°C via inversion. This solution was transferred to a 0.8 ϫ 4-cm Poly-Prep column (Bio-Rad) and washed sequentially with 12 ml of TT ϩ buffer, 3 ml of 333 mM NaCl in TT ϩ buffer, and then with 6 ml of TT ϩ buffer. GST or GST fusion protein was eluted from the GSH-agarose by adding 3 ml of 10 mM GSH (free acid) in TT ϩ buffer, pH 7.5. Eluted protein was concentrated and exchanged into PBS buffer using an Amicon stirred cell.
In Vitro Synthesis of Spinophilin-(151-444), 14-3-3, and Arrestin 3-Four methionines were inserted via PCR into the N terminus of a spinophilin fragment, spinophilin amino acids 151-444, and the resulting product was subcloned into the modified pGEMEX2 vector in which the sequence-encoding viral coat protein Gene 10 was deleted. The resulting construct pGEMEX2-(Met) 4 -Sp151-444 was verified by DNA sequencing. pCDNA3-HA14-3-3 was a generous gift from Dr. Haian Fu (Emory University). Arrestin 3 cDNA, graciously provided by Dr. Jef-fery Benovic (Thomas Jefferson University), was PCR-amplified and subcloned into the pCDNA3-HA vector. Resulting pCDNA3-HA-arrestin 3 was verified by DNA sequencing. Sp151-444, HA-14-3-3, and HA-arrestin 3 were transcribed and translated with or without [ 35 S]Met using TNT rabbit reticulocyte lysate kit as described previously (12). Following each synthesis, radiolabeled products were analyzed by 10% SDS-PAGE and autoradiography and quantitated by counting the representative band on dried gel in scintillation mixture. The concentrations of unlabeled proteins synthesized in parallel were estimated accordingly, assuming that in vitro transcription/translation from the same template with or without [ 35 S]Met would proceed at comparable efficiency.
Binding of 3i loops to GST-Spinophilin-Increasing amounts of GST or GST-Sp151-444 fusion protein (0.00034 -34 g, representing an estimated 2.34 ϫ 10 Ϫ11 M to 1.77 ϫ 10 Ϫ6 M of the fusion protein in the incubation) were incubated with 50 l of GSH-agarose (1:1 slurry equilibrated with TTB buffer (50 mM Tris-HCl, pH 7.4, 0.05% Triton X-100, 10% glycerol, 0.01% bovine serum albumin, 100 M PMSF) in a total reaction volume at 235 l for the first 5 points (0.00034 -3.4 g) and 315 l for the last point (34 g) for 2 h at 4°C. Then 15 l of 35 S-labeled 3i loop (31,000 cpm, estimated as 2.3 ϫ 10 Ϫ11 M) was added to each incubation and rotated for another 2 h at 4°C via inversion. After collection by centrifugation, the resin was washed three times with 1 ml of TTB. Interaction with GST-Sp151-444 versus GST (control) was determined by elution of the 3i loop into 1ϫ Laemmli buffer (400 mM Tris, pH 6.8, 700 mM ␤-mercaptoethanol, 1% SDS, 10% glycerol) and separation of the eluates by 12% SDS-PAGE. The degree of interaction was quantitated by cutting and counting the bands from the dried gel corresponding to the 3i loop (determined via autoradiography) in scintillation mixture.
To assess the relative affinity of the 3i loop for spinophilin, arrestin 3, or 14-3-3, 5 g of GST or GST-Sp151-444 was incubated with 20 l of GSH-agarose slurry for 2 h at 4°C. Increasing amounts (0 -16 l) of in vitro translated Sp151-444, 14-3-3, or arrestin 3 were mixed with rabbit reticulocyte lysate to reach a total volume of 16 l and then added to each incubation together with 6000 cpm (estimated as 9.2 ϫ 10 Ϫ12 M) 35 S-labeled 3i loop with a total reaction volume at 120 l. After 2 h of incubation at 4°C, resins were washed and eluted. 35 S-labeled 3i loop in the eluate was separated and quantitated as described above.
Interactions of mutant 35 S-labeled 3i loops with GST-Sp151-444 were assessed as follows: in each 250-l reaction, 25 g of GST or GST-Sp151-444 fusion protein was incubated with 50 l of GSHagarose (1:1 slurry equilibrated with TTB) for 2 h at 4°C. Equal amounts of 35 S-labeled wild type or mutant 3i loops were added to each incubation and rotated for another 2 h at 4°C. After collection by centrifugation, the resin was washed three times with 1 ml of TTB. Bound wild type or mutant 3i loops were eluted, resolved, and quantitated as above.
Binding of Spinophilin-(151-444), 14-3-3, and Arrestin 3 to GST-␣ 2A -AR 3i Loops-GST or GST fusion proteins expressed from a 25-ml culture of DH5␣ bacteria transformed with wild type or mutant GST-3i loop constructs (estimated as 25 g per culture) were incubated with 100 l of GSH agarose (1:1 slurry equilibrated with TTB buffer) for 2 h at 4°C (as described above). After three washes with TTB (333 mM NaCl was included in the second wash), either 120,000 cpm of 35 Slabeled 14-3-3 or Sp151-444 or 11,000 cpm of 35 S-labeled arrestin 3 was added to each incubation with a reaction volume of 200 l in TTB and rotated for another 2 h at 4°C. The resin was then collected and washed three times with 1 ml of TTB. Interactions of wild type or mutant GST-3i loops with Sp151-444, 14-3-3, and arrestin 3 were determined by elution of the Sp151-444, 14-3-3, or arrestin 3 in 1ϫ Laemmli buffer and separation of the eluates by 12% (for 14-3-3) or 10% (for Sp151-444 and arrestin 3) SDS-PAGE, respectively. To quantitate the degree of interaction, the bands corresponding to Sp151-444, 14-3-3, or arrestin 3 were cut from the dried gel (determined via autoradiography) and counted in a scintillation mixture.
Co-immunoisolation of Spinophilin with the ␣ 2A -AR-The ability of agonist occupancy of the ␣ 2A -AR to increase association of a particular protein with the ␣ 2A -AR or an ␣ 2A -AR-enriched complex was assessed using co-immunoisolation strategies. COS-M6 cells transiently transfected with cDNA encoding HA-␣ 2A -AR (17), GRK2 (a generous gift from Dr. Robert Lefkowitz at Duke University), Myc-tagged spinophilin (12), and/or GFP-arrestin 3 (a generous gift from Dr. Mark Caron at Duke University) protein (0.2, 0.6, 6, and 6 g per 10-cm plate, respectively), were serum-starved overnight and treated with 100 M epinephrine (agonist-treated) or 1 M yohimbine (antagonist-treated) for 5 min at 37°C in the presence of 1 M propranolol to block epinephrine effects on trace endogenous ␤-adrenergic receptors. The incubation was terminated by placing the culture dishes on ice, aspirating the incubation medium, and washing it once with ice-cold PBS containing 1 mM MgCl 2 and 0.5 mM CaCl 2 . Cells were then scraped into ice-cold lysis buffer containing 15 mM Hepes, pH 7.6, 5 mM EGTA, and 5 mM EDTA with 10 mM sodium pyrophosphate freshly added and passaged through a 19-gauge needle five times up/down followed by 15 min of centrifugation at 18,000 ϫ g at 4°C. Pellets were solubilized in 4 mg/ml D␤M, 0.8 mg/ml CHS lysis buffer containing 20 mM Hepes, pH 7.6, 25 mM glycylglycine, and 5 mM EGTA with the following protease inhibitors: 100 M PMSF, 10 g/ml leupeptin, 5 g/ml aprotinin, and 1 g/ml soybean trypsin inhibitor, by 10 passage through a 25-gauge needle on a 1-ml syringe. The extract was centrifuged at 100,000 ϫ g for 30 min at 4°C. The supernatant was defined as the solubilized membrane preparation.
Typically, 0.4 ml of detergent extract was precleared with 25 l of a 1:1 protein G-agarose/D␤M/CHS lysis buffer slurry and incubated with a 1:50 dilution of rat anti-HA antibody overnight at 4°C before addition of 25 l of a 1:1 slurry of protein G-agarose for 2 h. The protein G-agarose was pre-equilibrated in D␤M/CHS lysis buffer (see above) also containing 0.25% bovine serum albumin to block nonspecific binding of protein. The protein G-agarose-rat anti-HA-␣ 2A -AR complex was then washed three times with 1 ml of D␤M/CHS wash buffer (0.5 mg/ml D␤M, 0.1 mg/ml CHS, 20 mM Hepes, 25 mM glycylglycine, 5 mM EDTA). Bound protein was eluted in 1ϫ Laemmli buffer by a 15-s water bath sonication and 5 min of incubation at 70°C and separated on 10% SDS-PAGE. Western blot analyses were performed following ECL Western blotting protocols (Amersham Biosciences).
Intact Cell Receptor Phosphorylation-COS-M6 cells grown in 12.5cm 2 flasks were transiently transfected with cDNA encoding wild type HA-␣ 2A -AR or a mutant HA-␣ 2A -AR ⌬LEESSSS (10). Cells were labeled with 0.1 mCi/ml [ 32 P]orthophosphate for 1 h following incubation in phosphate-free, serum-free Dulbecco's modified Eagle's medium for 2 h at 37°C. Cells were then treated with 100 M epinephrine (agonisttreated) or 1 M yohimbine (antagonist-treated) for 5 min at 37°C in the presence of 1 M propranolol to block epinephrine effects on trace endogenous ␤-adrenergic receptors. The incubation was terminated by placing the culture dishes on ice, aspirating the incubation medium, and washing the dishes twice with ice-cold PBS containing 2 mM Na 3 VO 4 . Cells were then scraped into 400 l of PBS lysis buffer containing 1% Triton X-100, 0.05% SDS, 1 mM EDTA, 1 mM EGTA, 10 mM NaF, 10 mM sodium pyrophosphate, and protease inhibitors, passaged ten times through a 20-gauge needle on a 1-ml syringe, and rotated end over end for 30 min at 4°C. The lysate was centrifuged at 100,000 ϫ g for 30 min at 4°C; the resulting supernatant was subjected to immunoprecipitation with rat anti-HA antibody and 15 l of a 1:1 slurry of protein G-agarose/PBS lysis buffer. Immunoisolates were washed three times with 1 ml of PBS lysis buffer, eluted in 1ϫ Laemmli buffer by a 15-s water bath sonication and 5 min of incubation at 70°C, separated on 10% SDS-PAGE, and analyzed by autoradiography.

RESULTS AND DISCUSSION
Interaction of the ␣ 2A -AR 3i Loop with Spinophilin-Spinophilin (neurabin II) is a multidomain protein with an Nterminal actin-binding domain, a receptor-interaction domain (aa 151-444 (Refs. 12, 14)), a phosphatase 1-binding and regulatory domain (18), a PDZ domain, and a C terminus composed of three coiled-coil domains (13,19) as shown in Fig. 1A. The interaction of the ␣ 2A -AR with spinophilin has been demonstrated to occur under a number of experimental conditions, including agonist-regulated immunoisolation from intact cells, interaction of the 35 S-labeled ␣ 2A -AR 3i loop with GST-spinophilin, and binding of 35 S-labeled spinophilin to affinityabsorbed intact ␣ 2A -AR (12). In this study, we focused on characterizing interaction of the ␣ 2A -AR 3i loop with spinophilin-(151-444), the receptor-interaction domain (12,14). As shown in Fig. 1B,  We have shown previously that not only spinophilin (12), but also 14-3-3 (11) interacts with the ␣ 2A -AR 3i loop. Others have demonstrated the interaction of arrestin 3 (also known as ␤-arrestin 2) with the 3i loop of the ␣ 2A -AR (15,16). To evaluate the relative affinity of the 3i loop for spinophilin versus 14-3-3 or arrestin 3, we examined the ability of in vitro translated but unlabeled 14-3-3, spinophilin, or arrestin 3 to "usurp" the 35 S-labeled 3i loop ligand available for binding to GST-Sp151-444 ( Fig. 2A). We utilized this approach because, unlike for GST-Sp151-444, we were unable to synthesize large quantities of either GST-14-3-3 or GST-arrestin 3, and thus could not perform competition binding experiments in the traditional fashion. As shown in Fig. 2B, the presence of in vitro translated, unlabeled (Met) 4 -spinophilin-(151-444), 14-3-3, and arrestin 3 decreases availability of the 35 S-labeld 3i loop for GST-Sp151-444 binding. Spinophilin-(151-444) and arrestin 3 have a relatively equal affinity in doing so, which is 10-fold more sensitive than inhibition of 35 S-labeled 3i loop binding by 14-3-3 (Fig. 2C). Under the conditions of this experiment, the 3i loop is not phosphorylated, which has implications for the interpretations of findings to be discussed in Fig. 6, A and C. It is also interesting to note that spinophilin competes for 80% of 35  with spinophilin-(151-444), we constructed a variety of truncations and deletions within the 3i loop, shown schematically in Fig. 3A. The structure designated WT represents the entire ␣ 2A -AR 3i loop (aa 218 -377) as reported previously (20). Structures 1 through 8 illustrate the deleted regions throughout the ␣ 2A -AR 3i loop (defined in detail under "Experimental Procedures"), many of which represent computer-predicted regions of defined secondary structure, such as the predicted amphipathic helices encoded by regions deleted in structures 3 and 6. As demonstrated in Fig. 3B, however, each of these eight different 3i loop structures are as capable as the WT loop in interacting with the GST-Sp151-444 fusion protein, indicating that none of the deleted sequences are required for the 3i loop-spinophilin interaction. Similarly, deletion of either the entire amino-half (structure 10) or the carboxyl-half (structure 9) of the 3i loop also does not eliminate its interaction with spinophilin (Fig.  3B, lower left panel). These data suggest that no single stretch of sequence can account for ␣ 2A -AR 3i loop interaction with spinophilin; we interpret these observations to suggest that a folded surface formed by multiple non-contiguous regions of the loop interacts with spinophilin. Interestingly, we have similarly found that multiple, non-contiguous regions of the ␣ 2A -AR-3i loop are involved in stabilizing the ␣ 2A -AR on the basolateral surface (21).
Since the 3i loops of all three ␣ 2 -AR subtypes interact with spinophilin (12) and amino acid sequences among these loops are very divergent, we performed a sequence homology search among the 3i loops of the three ␣ 2 -AR subtypes. Five homologous regions were revealed (cf.  as sequences a, b, and c) and two at the C terminus (denoted as sequences d and e) of the ␣ 2A -AR 3i loop. 3 We investigated the importance of these five homologous sequences for interaction of the N-or C-terminal loop with spinophilin (Fig. 3). As shown in Fig. 3, deletion of  sequences a, b, or c in the amino-half (structures 11, 12, and 13) eliminates binding of the 3i loop to spinophilin in the absence of the C-terminal half of the 3i loop. Similarly, if the distal end of the 3i loop (aa 350 -377) containing sequences d and e is removed from the C-terminal half (structure 14), in vitro binding of this mutant loop to spinophilin is significantly diminished (Fig. 3B). These data suggest that homologous sequences within the 3i loops of all three ␣ 2 -AR subtypes are critical for interaction with spinophilin. To further test this interpretation in the context of the entire 3i loop, we generated structures 15 and 16. Structure 15 is virtually a combination of structures 13 and 14. Interestingly, although neither structure 13 nor 14 interacts with spinophilin, structure 15 binds to spinophilin just as well as the wild type ␣ 2A -AR 3i loop. Significantly, structure 16 (containing aa 232-359), which has the extreme end sequences (mainly containing sequences a and e) of the 3i loop deleted (Fig. 3A), has virtually no detectable interaction with GST-Sp151-444 (less than 4% of that by the wild type loop, Fig. 3B), indicating that sequences a and e are essential for 3i loop interaction with spinophilin. Taken together, the data indicate that multiple non-contiguous regions are involved in interaction of the ␣ 2A -AR 3i loop with spinophilin and sequences a and e are particularly critical for forming the interacting interface.
Differential Interactions of the ␣ 2A -AR 3i Loop with Spinophilin versus 14-3-3 and Arrestin 3-We also evaluated the regions of the 3i loop critical for interactions of 14-3-3 and arrestin 3. The wild type 3i loop containing aa 239 -371 (WT S ) and loops with incremental deletions were fused with GST (Fig. 4A) and examined for the ability to interact with radiolabeled 14-3-3 and arrestin 3. As shown in Fig. 4, B and C, incremental deletions across the 3i loop do not perturb the binding of either 14-3-3 or arrestin 3, indicating that interactions with these two proteins, as with spinophilin, also involve multiple, non-contiguous regions. Of interest is the observation that structure 25 (aa 239 -359, Fig. 4A), which is an analogue of structure 16 (aa 232-359) in Fig. 3, interacts with both 14-3-3 and arrestin 3 very well (Fig. 4, B and C) even though structure 16 has virtually no detectable binding to spinophilin (Fig. 3B).
We had exploited subtly different experimental strategies to measure binding of the ␣ 2A -AR 3i loop to spinophilin (Fig. 3) versus binding to 14-3-3 or arrestin. This is because 35 Slabeled Sp151-444 manifests some binding to GST alone, confounding studies using radiolabeled spinophilin to GST-3i loop structures. Nonetheless, as shown in Fig. 4E, we can directly compare the regions. To rule out the possibility that different experimental approaches led to our observed differences in requirements for binding of the 3i loop to 14-3-3 or arrestin 3 compared with spinophilin, we tested the interaction of structure 25 with radiolabeled Sp151-444. As shown in Fig. 4E, structure 25 has virtually no binding to spinophilin. We also noticed that structure 26 (aa 239 -294), analogous to structure 11 in Fig. 3, does not interact with either 14-3-3 or arrestin 3 (Fig. 4, B and C), just as structure 11 does not bind to spinophilin (Fig. 3B). We further asked if the entire N-terminal half of the 3i loop (aa 218 -294, structure 28 in Fig. 4D, analogous to structure 9 in Fig. 3) could interact with 14-3-3 and arrestin 3. Addition of sequences from aa 218 -238 of the 3i loop only partially restores the loop interaction with 14-3-3 and arrestin 3 (Fig. 4D), which is different from the ability of spinophilin to bind to structure 9 (an analogue of structure 28) almost as well as the WT structure containing the entire loop (see left lanes in bottom panel of Fig. 3B). We further confirmed the interaction of structure 28 with radiolabeled Sp151-444 in Fig. 4E, which is comparable to that of structure WT L with spinophilin.
Thus, although multiple noncontiguous regions of the loop are needed for binding to these three regulatory proteins, differential interactions of the 3i loop are evident with spinophilin versus 14-3-3 and arrestin 3. Spinophilin interacts with the N-(aa 218 -294) and C-terminal half (aa 294 -377) of the loop with relatively similar effectiveness, just slightly lower than the binding detectable with the entire loop (aa 218 -377, Figs. 3B and 4E). However, the C-terminal half of the loop interacts with 14-3-3 and arrestin 3 as well as the entire loop, and both are about twice as effective in binding 14-3-3 and arrestin 3 as the N-terminal half of the loop (Fig. 4, B-D). Moreover, deletion of the extreme end sequences of the loop containing sequences a and e does not affect the loop interactions with 14-3-3 and arrestin 3 (see structure 25 in Fig. 4, B and C), but eliminates interaction with spinophilin to less than 4% of control values (see structure 16 in Fig. 3B and structure 25 in Fig. 4E). An important implication of these findings (summarized in Fig.  4F) is that spinophilin, via N-terminal sequences of an appropriately folded loop, and arrestin (or 14-3-3), via regions in the C terminus of the loop, may be able to simultaneously interact with the ␣ 2A -AR. This interpretation also is consistent with findings in Fig. 2C, where arrestin 3 and 14-3-3 compete for only 50% of 35 S-labeled 3i loop binding to GST-spinophilin.
Regulated Interactions of ␣ 2A -AR with Spinophilin and Arrestin 3-Our data suggest that the ␣ 2A -AR interacts with spinophilin, arrestin 3, and 14-3-3 through multiple regions of the 3i loop (Figs. 3 and 4) and that arrestin 3 and 14-3-3 compete for spinophilin interaction with the loop in vitro (Fig.  2). It was therefore of interest to determine in the context of intact cells whether these interactions occur independently of one another, e.g. in different target cells or in different subdomains of a single cell, or whether they occur as part of a regulatory cycle. Extant data suggest that a regulatory cycle depicted schematically in Fig. 5 is possible. Based on existing data, it is reasonable to postulate that the ␣ 2A -AR interacts with 14-3-3 when the receptor is in an inactive state. Interactions of ␣ 2A -AR with 14-3-3 are competed for by a phosphorylated Raf peptide but not by its non-phosphorylated counterpart (11), and agonist activation of ␣ 2A -AR has been demonstrated to activate the Ras/Raf cascade in a variety of target cells (11,22,23). Agonist occupancy may favor at least a transient interaction or more stable association with spinophilin since ag-3 D. Mochly-Rosen and Q. Wang, unpublished findings.  (24). In addition, we have shown in vitro that the interactions of the ␣ 2A -AR 3i loop with spinophilin, 14-3-3, and arrestin 3 are capable of competing with each other (Fig. 2). To test the possibility that, in fact, these interactions are competitive with one another in the context of a cell, rather than parallel and functionally nonrelated interactions, we examined the ability of arrestin 3 and spinophilin to mutually influence each other's interactions with the ␣ 2A -AR. We examined the ability of arrestin 3 to influence spinophilin interaction with the ␣ 2A -AR in COS cells, taking advantage of the relatively low level of expression of endogenous arrestins in this cell line. As shown in Fig. 6A, agonist occupancy of the ␣ 2A -AR indeed enriches the amount of exogenously expressed Myc-tagged spinophilin in the HA-␣ 2A -AR immunoisolated complex by 1.95-fold (Fig. 6A, lane 2 versus lane 1, p Ͻ 0.05), as expected from our previous finding (12). However, this increased association of spinophilin is prevented, or masked, when arrestin 3 is exogenously overexpressed in COS-M6 cells (Fig. 6A, lane 4 versus lane 3). Even the basal interaction of Myc-tagged spinophilin with HA-␣ 2A -AR before agonist treatment is decreased to 78% of the control values (Fig. 6A, lane 3  versus lane 1, p Ͻ 0.05). These data are consistent with the interpretation that spinophilin and arrestin 3 are capable of interacting with the same pool of ␣ 2A -AR in the context of the cell.
From extant data in the literature we know that arrestin 3 favors binding to GRK-phosphorylated receptors following receptor activation (25)(26)(27). We asked whether arrestin 3 could still prevent agonist-simulated spinophilin binding to the ␣ 2A -AR when the receptor cannot be phosphorylated by GRK. The ␣ 2A -AR is a substrate for GRK2 at the LEESSSS sequence in the N-terminal half of the 3i loop (28). We tested the ability of arrestin 3 to compete for spinophilin binding to a mutant ␣ 2A -AR with the LEESSSS sequence deleted (10), which as shown in Fig. 6B can not be effectively phosphorylated follow-   (11,12) and others (23) suggest a possible regulatory cycle for these interactions. The interactions of the ␣ 2A -AR with 14-3-3 can be disrupted by phosphorylated Raf peptides (11), a consequence of ␣ 2A -AR activation of the Ras cascade (22,29). Agonist occupancy of the ␣ 2 -AR appears to stabilize agonist interaction with spinophilin (12). This interaction may be transient since phosphorylation of Ser 296 -Ser 299 by GRK leads to arrestin 3 interaction with the receptor and presumed hyperphosphorylation of the receptor-arrestin complex. Furthermore, as shown in Fig. 6, arrestin 3 competes for spinophilin binding to the phosphorylated receptor, but not a mutant ␣ 2 -AR (⌬LEESSSS) that lacks the GRK sites for phosphorylation of the receptor. Both agonist dissociation from the receptor and phosphatasecatalyzed dephosphorylation may favor interactions of the ␣ 2A -AR with 14-3-3, although we have no direct evidence to affirm or refute this aspect of the regulatory cycle.
ing agonist stimulation. This ⌬LEESSSS ␣ 2A -AR has been demonstrated to have the same turn-over rate (21) and G protein coupling efficiency (10) as the wild type ␣ 2A -AR. Like that of the wild type receptor, agonist occupancy of the ⌬LEESSSS ␣ 2A -AR also enriches the amount of Myc-tagged spinophilin in the receptor-immunoisolated complex (Fig. 6C, lane 2 versus lane 1), but to an even stronger extent (4.43-fold in Fig. 6C versus 1.95-fold in Fig. 6A). However, exogenously expressed arrestin 3 does not decrease either the basal association (before agonist treatment, Fig. 6C, lane 3 versus lane 1) or enhanced association (following agonist occupancy, Fig.  6C, lane 4 versus lane 3) of spinophilin to the ⌬LEESSSS ␣ 2A -AR receptor. These data suggest that receptor phosphorylation increases the probability of arrestin 3 competition for spinophilin binding to the ␣ 2A -AR. This is likely a consequence of the known increase in affinity of arrestin for GRKphosphorylated receptors (25)(26)(27), perhaps due to the creation of another contact site for arrestin stabilization upon phosphate incorporation into the tetra-serine LEESSSS sequence.
In summary, the present data suggest that a surface presented by non-contiguous sequences within the predicted third intracellular loop of the ␣ 2A -AR interacts with spinophilin, arrestin 3, and 14-3-3. This surface interchangeably interacts with these three proteins in an agonist-regulated fashion, likely due to agonist-induced alterations in the structure of the 3i loop; however, GRK phosphorylation of the ␣ 2A -AR 3i loop preferentially drives the interaction with arrestin 3. Timeresolved fluorescence studies will likely be required to more definitively outline the regulatory cycle and its functional consequences.