Assembly of an SAP97-AKAP79-cAMP-dependent Protein Kinase Scaffold at the Type 1 PSD-95/DLG/ZO1 Motif of the Human β1-Adrenergic Receptor Generates a Receptosome Involved in Receptor Recycling and Networking*

Appropriate trafficking of the β1-adrenergic receptor (β1-AR) after agonist-promoted internalization is crucial for the resensitization of its signaling pathway. Efficient recycling of the β1-AR required the binding of the protein kinase A anchoring protein-79 (AKAP79) to the carboxyl terminus of the β1-AR (Gardner, L. A., Tavalin, S. A., Goehring, A., Scott, J. D., and Bahouth, S. W. (2006) J. Biol. Chem. 281, 33537-33553). In this study we show that AKAP79 forms a complex with the type 1 PDZ-binding sequence (ESKV) at the extreme carboxyl terminus of the β1-AR, which is mediated by the membrane-associated guanylate kinase (MAGUK) protein SAP97. Thus, the PDZ and its associated SAP97-AKAP79 complex are involved in targeting the cyclic AMP-dependent protein kinase (PKA) to the β1-AR. The PDZ and its scaffold were required for efficient recycling of the β1-AR and for PKA-mediated phosphorylation of the β1-AR at Ser312. Overexpression of the catalytic subunit of PKA or mutagenesis of Ser312 to the phosphoserine mimic aspartic acid both rescued the recycling of the trafficking-defective β1-ARΔ PDZ mutant. Thus, trafficking signals transmitted from the PDZ-associated scaffold in the carboxyl terminus of the β1-AR to Ser312 in the 3rd intracellular loop (3rd IC) were paramount in setting the trafficking itinerary of the β1-AR. The data presented here show that a novel β1-adrenergic receptosome is organized at the β1-AR PDZ to generate a scaffold essential for trafficking and networking of the β1-AR.

The sympathetic nervous system mediates its regulatory effects through G protein-coupled receptors (GPCR) 3 related to the family of ␣and ␤-adrenergic receptors. Among these receptors is the ␤ 1 -AR, which is coupled to the G s -cyclic AMP axis and plays a major role in transmitting sympathetic regulation to cardiac, renal, vascular, and other organs (2,3).
Persistent activation of the ␤ 1 -AR or other GPCR causes their desensitization and internalization via clathrin-coated pits or caveolae into early endosomes (4 -6). Internalized GPCR are either recycled back to the cell surface for another round of signaling or retained for degradation by lysosomal or proteasomal pathways (7)(8)(9). Characterization of the players involved in these distinct outcomes is the purpose of this study.
Recycling and resensitization of the ␤ 1 -AR are dependent upon two motifs; one is the ESKV sequence in the carboxylterminal tail, and the other is the region surrounding Ser 312 in the 3rd IC of the ␤ 1 -AR (10,11). The ESKV tetrapeptide conforms to a type I (PSD-95/DLG/ZO1) PDZ ligand (i.e. X(S/ T)X⌽, where X at positions Ϫ1 and Ϫ3 is any amino acid, and ⌽ at position 0 is a hydrophobic amino acid) (12,13). Mutagenesis of the type 1 PDZ or Ser 312 to alanine prevented the recycling and resensitization of the ␤ 1 -AR (10,11). Concerning Ser 312 , we determined that this residue is specifically phosphorylated by PKA and that the activity of PKA was required for recycling and resensitization of the human ␤ 1 -AR (11).
These results indicate that two distinct motifs are involved in recycling of the ␤ 1 -AR, but they do not explain how they crosstalk to one another to coordinate the sequence of events involved in recycling of this GPCR. A major breakthrough in identifying the mechanism of cross-talk between these two motifs was the identification of AKAP79 as the AKAP involved in recycling of the ␤ 1 -AR in HEK-293 and other cell lines (1). AKAP79 promoted the targeting of PKA to the ␤ 1 -AR by binding to the carboxyl-terminal 53 amino acids (between residues 425 and 477) of the ␤ 1 -AR (1). Here we report that the binding domain of AKAP79 to the ␤ 1 -AR overlaps with its type 1 PDZ motif. However, the binding between the PDZ and AKAP79 is indirect and involves the MAGUK protein SAP97 that simultaneously binds to AKAP79 and type 1 PDZ to target PKA to the bic acid for 2 h at 25°C. For saturation binding experiments, ICYP concentrations ranging between 5 and 300 pM were used to calculate the K D and the B max values for ICYP binding by parametric fitting of the data by using the Prism 4 software (GraphPad Corp.).
Acid Strip Confocal Recycling Microscopy Protocol-HEK-293 cells expressing the FLAG-or Myc-tagged WT ␤ 1 -AR or ␤ 1 -AR⌬PDZ were grown on poly-L-lysine-coated glass coverslips and serum-starved at 37°C for 1 h in DMEM supplemented with 25 mM HEPES, pH 7.4. The receptors were labeled with fluorescein isothiocyanate-conjugated anti-FLAG M2 IgG (10 g/ml) for 1 h at 37°C. Cells were treated with 10 M isoproterenol for 30 min at 37°C to promote agonist-mediated ␤ 1 -AR internalization. Then the cells were chilled to stop endocytosis and exposed to 0.5 M NaCl, 0.2 M acetic acid, pH 3.5, for 4 min on ice to remove antibody bound to extracellular ␤ 1 -AR (1,15,16). Cultures were then incubated with culture medium supplemented with 100 M of the ␤-antagonist alprenolol at 37°C for 10, 20, 30, or 45 min to establish the recycle time. After each time period, the coverslips were rinsed and fixed in 4% paraformaldehyde with 4% sucrose in PBS, pH 7.4, for 10 min at room temperature. Confocal fluorescence microscopy was performed on coded slides using a Zeiss Axiovert LSM 510 (100 ϫ 1.4 DIC oil immersion objective), and the immunocytochemical data were analyzed to determine the recycle time (1).
FRET Microscopy-Double stable cell lines expressing AKAP79-CFP and ␤ 1 -AR-YFP or SAP97-CFP and ␤ 1 -AR-YFP were established. In some cases, HEK-293 cells were trans-fected with the desired plasmids using the Lipofectamine reagent (Invitrogen) for 24 -36 h. After transfection, cells were plated on poly-L-lysine-covered coverslips for 24 h, fixed with 4% paraformaldehyde, pH 7.4, and mounted onto glass slides in Fluoromount G mounting media (Electron Microscopy Sciences, Hatfield, PA). Coverslips were sealed with clear nail polish and imaged using the sensitized emission or the acceptor photobleaching methods described in Gardner et al. (1). After image acquisition, the LSM 510 FRET macro tool was used to calculate FRETN values. FRETN is a measure of FRET that is normalized for the concentrations of donor and acceptor fluorophores and therefore represents a fully corrected measure of FRET (17)(18)(19). In this method the corrected FRET value for each pixel is calculated and then divided by concentration values for donor and acceptor (18 -20). FRETN was calculated on a pixel-by-pixel basis for the entire image.
In addition to FRETN microscopy, we performed acceptor photobleaching FRET microscopy. This method measures changes in the intensity of the donor channel that are observed upon complete photobleaching of the acceptor (YFP) by a 514 nm argon laser (1, 16 -18). From each photobleaching session, an image set consisting of time-lapse recordings of donor and acceptor channel intensities was obtained. FRET was recorded by examining the loss of quenching of CFP during YFP photobleaching, followed by an analysis of these images by the LSM FRET tool version 1.5 (AIM software release 3.2) to calculate the FRET efficiencies using selected area averages for donor CFP before and after bleaching. FRET efficiencies (%) are presented as the means Ϯ S.E. from 3 to 10 separate acquisition experiments on 5-10 images per experiment.
Co-immunoprecipitations and Pulldown Assays-Co-immunoprecipitations between FLAG-or Myc-tagged ␤ 1 -AR, SAP97, AKAP79, or the RII ␣-subunit of PKA were performed as follows. Cells stably expressing the indicated FLAG-tagged or Myc-tagged ␤ 1 -AR were lysed in radioimmune precipitation assay buffer (1), and the insoluble cellular debris was removed by centrifugation at 14,000 ϫ g av for 15 min at 4°C. After equalizing protein concentrations across all samples, lysates were added to M2 anti-FLAG-or anti-Myc-agarose beads at 4°C with gentle rotation for 4 h. Control experiments were performed by incubating lysates with preimmune IgG at the same concentration for 4 h at 4°C. The immune complexes were washed three times in radioimmune precipitation assay buffer and eluted from the beads with 40 l of 2ϫ Laemmli sample buffer containing 20 mM dithiothreitol. Resolved proteins and lysate inputs were separated by SDS-PAGE under denaturing conditions and electroblotted to nitrocellulose. Identical gels were run and transferred for separate detection of receptor, AKAP79, SAP97, or the RII␣ subunit of PKA by Western blotting.
Pulldown Assays-The human ␤ 1 -AR cDNA was digested with SmaI and XhoI to isolate the carboxyl-terminal fragment encoding the amino acids between 425 and 477. This fragment and the corresponding carboxyl-terminal tail fragment of the ␤ 2 -AR were cloned into the pGEX-4T-2 GST vector (GE Healthcare) and amplified in BL-21 Escherichia coli cells. HEK-293 cells that were transfected either with empty or with myc-SAP97 expressing vector were lysed with in 0.2% Triton X-100 in PBS supplemented with protease inhibitors. After 16,000 ϫ g av centrifugation of cell lysates, GST or ␤ 1 -AR c-tail or ␤ 2 -AR c-tail -GST fusion proteins were added to aliquots of the supernatants. Twenty l of glutathione-agarose beads (50% slurry in H 2 O) were added after mixing for 30 min at 4°C. The mixture was mixed for another 2 h at 4°C. After washing three times with the same lysis buffer, the proteins were eluted from beads with sample buffer (containing 2.5% ␤-mercaptoethanol). Eluates were separated on a 4 -15% gel and analyzed for SAP97 by immunoblotting. Far Western blots were performed to detect the interaction between FLAG-␤ 1 -AR (WT and ⌬PDZ) and SAP97 according to Hall (21) with a few modifications. In brief, the purified receptor was slot-blotted onto a dry nitrocellulose membrane in a volume of 100 l (20 l at a time was added) under vacuum (manifold II). The membrane was wetted in 5 ml of Tris-buffered saline containing 0.2% Tween 20 (TBST) and then blocked with TBST containing 10 mg/ml bovine serum albumin (TBST-BSA) for 1 h at 22°C by gentle shaking. Affinity purified Myc-SAP97 was hybridized with the membranes in TBST-BSA for 16 h at 4°C with gentle rocking. The membrane was washed five times (5 min each with TBST) and probed with 1:1000 dilution of anti-Myc IgG (9E-10 monoclonal antibody) for 1 h at 22°C in TBST-BSA. After washing the blot five times with TBST, it was probed with horseradish peroxidase-conjugated anti-mouse IgG (Pierce) in TBST-BSA for 20 min at room temperature. The blot was washed five times in TBST (5 min each) and then developed using enhanced chemiluminescence.
Cyclic AMP Accumulation and Adenylyl Cyclase Assays-HEK-293 cells stably expressing the various ␤ 1 -AR constructs in 6-well plates were switched to DMEM ϩ 25 mM HEPES for 2 h. Appropriate drugs in DMEM/HEPES, supplemented with 300 M of the phosphodiesterase inhibitor isobutylmethylxanthine, were added to the cells for 10 min at 37°C. The reaction was stopped, and 1 ml of 0.1 N cold HCl was added followed by freezing of the entire plate in liquid nitrogen. Frozen plates were quickly thawed at 65°C to break the cells, and the cell extract was lyophilized. The dry pellet was resuspended in assay buffer, and cyclic AMP was quantified by radioimmunoassay (RIANEN Assay System; PerkinElmer Life Sciences). For the determination of adenylyl cyclase activity, membranes were prepared from cells without phenylmethylsulfonyl fluoride, and the activity of adenylyl cyclase in response to increasing concentrations of isoproterenol was determined (14,16). The concentration-response curves to isoproterenol were fitted by nonlinear regression using Prism 4.1 software (GraphPad Corp.) in order to determine the concentration of isoproterenol that generated 50% of the maximal response (EC 50 ) for each ␤ 1 -AR construct.
Adenylyl Cyclase Assays for ␤-AR Desensitization and Resensitization-HEK-293 cells stably expressing the various ␤ 1 -AR constructs were divided into four sets. The first and second sets were used as control for desensitization and the third and fourth sets for resensitization assays. Cells for desensitization were exposed to 1 mM ascorbic acid (control) or 10 M isoproterenol for 10 min at 37°C and then processed for the preparation of membranes. The third set was used as the control for resensitization and the fourth set for resensitization assays. Cells for resensitization were exposed either to 1 mM ascorbic acid (control) or to 10 M isoproterenol for 3 h at 37°C and then incubated with 100 M alprenolol for 1.5 h at 37°C, followed by the preparation of membranes. Adenylyl cyclase activities in these membranes were determined (14,16), and the K act Ϯ S.E. for each ␤ 1 -AR was calculated using the Prism 4 program, and statistical comparisons were analyzed using Prism 4 and Instat programs (GraphPad Corp.).
Phosphorylation and Phosphopeptide Mapping of the ␤ 1 -AR-To determine the effect of disrupting the ␤ 1 -AR PDZ or downregulation of SAP97 on isoproterenol-mediated phosphorylation of the ␤ 1 -AR, HEK-293 cells expressing the WT ␤ 1 -AR were transfected with the SAP97 siRNA vector or its scrambled control. On the day of the experiment, the ATP pools were labeled with 200 Ci of 32 PO 4 /ml for 1.5 h, and the cells were stimulated with either 1 mM ascorbic acid or 10 M isoproterenol in 1 mM ascorbic acid for 10 min at 37°C. After cell lysis, equivalent amounts of proteins in each supernatant were incubated with M2 anti-FLAG-agarose beads at 4°C for 5 h. The resins were washed in radioimmune precipitation assay buffer, and the eluted proteins were resolved by SDS-PAGE. The gels were transferred to nitrocellulose, and amounts of 32 P incorporated into the ␤ 1 -AR were determined by electronic counting with Packard Instantimager TM . The bands corresponding to phosphorylated ␤ 1 -AR protein on the filter were cut out and submerged in 70% (v/v) formic acid containing 100 mg per ml of cyanogen bromide (Science Lab Chemicals, Kingswood, TX) for 1.5 h at room temperature (1). At the end of the digestion, the samples were lyophilized and dissolved in Tricine sample buffer. Then 5 l from each sample was spotted onto a GF/C filter pre-moistened with 10% trichloroacetic acid. The filters were mounted on a filtration manifold and washed three times with 5 ml of 10% trichloroacetic acid to remove the free 32 P. After drying, the counts/min of 32 P/filter were determined by liquid scintillation spectrometry. Equal counts/min (1,200 Ϯ 50) of 32 P were loaded per lane and subjected to electrophoresis on 16% acrylamide gels in Tricine cathode buffer. At the end of the run the gel was electroblotted to nitrocellulose, and the filters were counted by the Instantimager TM and then exposed to an x-ray film overnight.
Biotinylation Assay of ␤ 1 -AR Recycling with Cleavable Biotin-Cells expressing the WT ␤ 1 -AR with siRNA to SAP97 or its control were surface-biotinylated with 1.5 mg/ml sulfo-NHS-SS-biotin (Pierce) in Hanks' balanced salt solution with Ca 2ϩ and Mg 2ϩ at 4°C (1). Biotinylated cells were exposed to isoproterenol for 30 min and then cooled to 4°C to stop membrane trafficking, and the remaining surface biotin was quantitatively cleaved with glutathione. After cleavage, warm DMEM was added, and cells were incubated at 37°C for 15, 30, and 60 min to allow internalized receptor to recycle before the cells were cooled to 4°C and incubated with glutathione cleavage buffer for a second time to ensure complete cleavage of any newly appearing surface biotin. At the end of each time point, the cells were scraped into detergent-free lysis buffer, sonicated, and then centrifuged at 100,000 ϫ g av for 20 min at 2°C. The membrane pellet was dissolved in lysis buffer supplemented with detergents and recentrifuged at 100,000 ϫ g av for 20 min at 2°C. The supernatant was collected, and equal amounts of protein from all samples were mixed with 50 l of BSA-blocked ultralink-neutravidin beads (Pierce) to isolate the biotinylated pro-teins. The resin was extracted, and the extracts were subjected to immunoblotting with anti-FLAG antibody to determine the density of ␤ 1 -AR.

RESULTS
Characterization of AKAP79 Binding to the ␤ 1 -AR-Myctagged ␤ 1 -AR constructs with progressive deletions within their carboxyl termini were co-expressed with FLAG-AKAP79 in order to localize by co-immunoprecipitations the sequence in the ␤ 1 -AR that bound AKAP79 (Fig. 1A). Deletion of the amino acids between 425 and 441 in the carboxyl-terminal tail of the ␤ 1 -AR had little effect on the immunoprecipitation of AKAP79 by this ␤ 1 -AR mutant. Deletion of the sequence between 425 and 463 significantly reduced the immunoprecipitation of AKAP79, indicating that the binding site between AKA79 and the ␤ 1 -AR partially overlapped with this sequence. Mutagenic inactivation (⌬) of each residue in the type 1 PDZ sequence (ESKV) between amino acids 474 and 477 to alanine (␤ 1 -AR⌬PDZ) completely inhibited the interaction between the ␤ 1 -AR and AKAP79, confirming that the AKAP79 interacting site overlapped with the type 1 PDZ sequence. The interactions between AKAP79 and the ␤ 1 -AR are involved in recycling of the agonist-internalized ␤ 1 -AR back into the cell membrane in a process termed "resensitization" (1,11). This process is involved in trafficking of the agonist-internalized ␤ 1 -AR back to the cell membrane (Fig. 1B, images a-e). Inhibition of the binding of AKAP79 to the ␤ 1 -AR by inactivating the PDZ pre- Lysates represent 5% of the total extract, whereas immunoprecipitations (IP) represent 30% of the total volume. B, HEK-293 cells stably expressing FLAG-WT ␤ 1 -AR (images a-e) or FLAG-␤ 1 -AR⌬PDZ, in which the last four amino acids in the carboxyl terminus of the ␤ 1 -AR were mutated to alanine (images f-j), were cultured on glass bottom slides. Recycling of the WT ␤ 1 -AR and the ␤ 1 -AR⌬PDZ in response to 10 M isoproterenol (n ϭ 3) was conducted as described under "Experimental Procedures." Each scale bar represents 5 m.
vented the recycling of the agonist-internalized ␤ 1 -AR⌬PDZ ( Fig. 1B, images f-j). In other experiments we determined that the binding parameters of [ 125 I]ICYP to the WT ␤ 1 -AR or to the ␤ 1 -AR⌬PDZ were comparable ( Table 1). The effect of mutagenesis of the PDZ on receptor coupling efficacy to G s was assessed by measuring basal and isoproterenol-stimulated increases in cyclic AMP accumulation in whole cells and in adenylyl cyclase activities in membranes prepared from cells expressing comparable densities of each ␤ 1 -AR construct (Fig.  2). Basal levels of cyclic AMP in cells stably expressing the empty pcDNA3.1 vector were 8 Ϯ 3 and 14 Ϯ 2 pmol/mg of protein in cells expressing the WT ␤ 1 -AR or the ␤ 1 -AR⌬PDZ, respectively ( Fig. 2A). The EC 50 value for the accumulation of cyclic AMP in response to isoproterenol was 7 Ϯ 1.5 nM for the WT ␤ 1 -AR and 9 Ϯ 2 nM for the WT ␤ 1 -AR⌬PDZ ( Fig. 2A; p Ͼ 0.05). Basal levels of adenylyl cyclase activities in membranes prepared from cells expressing the WT ␤ 1 -AR were 17 Ϯ 3 pmol/min/mg, whereas in cells expressing the ␤ 1 -AR⌬PDZ, these levels were 14 Ϯ 3 pmol/min/mg (p Ͼ 0.05). The EC 50 value for the activation of adenylyl cyclase in membranes expressing either the WT ␤ 1 -AR or the ␤ 1 -AR⌬PDZ were comparable at 0.1 Ϯ 0.02 M, and maximal activation of adenylyl cyclase was 82 pmol/min/mg protein by either of these receptors (Fig. 2B). Thus, the ⌬PDZ mutation had no effect on the coupling efficacy of the ␤ 1 -AR to G s .
To characterize the mechanism by which AKAP79 binds to the ␤ 1 -AR, full-length WT ␤ 1 -AR or ␤ 1 -AR⌬PDZ was hybridized to immobilized AKAP79 (Fig. 3, A and B). In these far Western assays, no direct binding of AKAP79 to either ␤ 1 -AR construct was observed, indicating that these proteins did not interact directly. An alternative mechanism that can account for indirect binding of AKAP79 to the ␤ 1 -AR is through their mutual association with MAGUK proteins (22). MAGUK proteins related to the PSD/SAP family (PSD-95/SAP90, SAP97/ hdlg, Chasyn-110/PSD-93, and SAP102) share a common domain organization consisting of three PDZ domains in their amino-terminal half that bind to type 1 PDZs and Src homology 3 and guanylate kinase-like domains at their carboxyl terminus that bind to AKAP79 and other proteins (22)(23)(24). MAGUK proteins related to PSD-95 and MAGI-II have been shown to bind to the ␤ 1 -AR PDZ (25,26). Therefore, HEK-293 cells were probed for the expression of these proteins, but none of them was detected in this cell line. Consequently, we explored whether other MAGUK-related proteins were expressed in HEK-293 cells by probing cell extracts with a pan-PDZ antibody (Upstate Biotechnology, Inc.), which identified a prominent immunoreactive species with an apparent molecular mass of 110 -116 kDa (data not shown). This protein was confirmed as human SAP97 by Western blotting with a monoclonal antibody to human SAP97 (Fig. 3C) and with a polyclonal anti-  Table 1, were stimulated with 0, 1, 10, 50, and 100 nM of isoproterenol for 5 min at 37°C, followed by determining the levels of cyclic AMP as described under "Experimental Procedures." B and C, isoproterenol-mediated activation of adenylyl cyclase in membranes prepared from cells expressing the WT ␤ 1 -AR or its point mutants, and chimera were determined. The EC 50 values for isoproterenol in activating adenylyl cyclase for each ␤ 1 -AR construct are reported in Table 1.

TABLE 1 Ligand binding properties of ͓ 125 I͔iodocyanopindolol to wild-type and mutated ␤ 1 -AR stably expressed in HEK-293 cells
Binding of ICYP to 0.5 g of membranes derived from HEK-293 cells expressing the various ␤ 1 -AR constructs was measured as described under ЉExperimental Procedures.Љ For saturation binding experiments, ICYP concentrations ranging between 5 and 300 pM were used. Saturation isotherms were analyzed by one-or two-site models to determine the K D and B max values of ICYP. Dose-response curves of isoproterenol-mediated stimulation of adenylyl cyclase were determined in membranes prepared from these cells, as described under ЉExperimental Procedures.Љ These curves were analyzed by nonlinear regression (Prism 4.0) to calculate the EC 50 Ϯ S.E. for each construct (n ϭ 3).

Construct
B max K D for ICYP EC 50 value for isoproterenol-mediated activation of adenylyl cyclase FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7

Trafficking of the ␤ 1 -AR by a PDZ-anchored Receptosome
JOURNAL OF BIOLOGICAL CHEMISTRY 5089 SAP97 antibody (data not shown). Far Western assays between the ␤ 1 -AR and SAP97 showed direct binding between SAP97 and the full-length ␤ 1 -AR, which was abrogated when the PDZ domain in the ␤ 1 -AR was mutated (Fig. 3D). Pulldown assays between SAP97 and glutathione S-transferase (GST) fusions of full-length carboxyl termini of the ␤ 1 -AR and the ␤ 2 -AR indicated that SAP97 preferentially associated with the carboxyl terminus of the human ␤ 1 -AR (Fig. 3E).
To further confirm that the ␤ 1 -AR could bind to SAP97, we determined whether they were co-localized. Myc-SAP97 was transiently transfected into a HEK-293 cell line stably expressing 1.1 Ϯ 0.2 pmol/mg protein of YFP tagged to the carboxyl terminus of the WT ␤ 1 -AR, and their distribution was determined by dual-labeling confocal microscopy (Fig. 3F). In control cells, Cy-3-labeled SAP97 (Fig. 3F, red) was co-localized with YFP-labeled ␤ 1 -AR (yellow) at the cell surface of HEK-293 cells (Fig. 3F, image a). In addition, SAP97 was distributed into other intracellular compartments (24). Exposing these cells to isoproterenol caused the internalization of the ␤ 1 -AR without altering the cellular distribution of SAP97 (Fig. 3F, image b). After the removal of isoproterenol, the internalized ␤ 1 -AR trafficked back into the cell membrane and was co-localized with SAP79 (Fig. 3F, images c-f).
Co-immunoprecipitations between SAP97 and a variety of ␤ 1 -AR constructs were used to determine whether SAP97 binds to the ␤ 1 -AR in a PDZ-dependent manner. FLAG-WT ␤ 1 -AR or its ⌬PDZ mutant was co-transfected into HEK-293 cells with Myc-SAP97 (Fig. 4, panel A, a-c). Precipitates of the WT ␤ 1 -AR, co-immunoprecipitated SAP97, and reciprocal immunoprecipitations of SAP97 precipitated the WT ␤ 1 -AR (Fig. 4, a  and b), but precipitates prepared from cells expressing the ␤ 1 -AR⌬PDZ failed to co-immunoprecipitate SAP97 (Fig. 4, panel A, c). Moreover, in precipitates of the WT ␤ 1 -AR, we detected co-immunoprecipitations of AKAP79 and the RII-␣ subunit of PKA (Fig. 4, panel A, d and e). In cells co-expressing FLAG-AKAP79 and Myc-SAP97 (Fig. 4, panel A, f and g), precipitates of AKAP79 co-immunoprecipitated SAP97 and vice versa, indicating that SAP97 and AKAP79 interacted with each other under these conditions (17,22). These data show that a quaternary complex composed of SAP97, AKAP79, and PKA was assembled at the type 1 PDZ in the carboxyl-terminal tail of the ␤ 1 -AR.
The organization of this scaffold was further studied using siRNAs to SAP97 (Fig. 4, panel B). We predicted that if SAP97 was a bridging molecule between the ␤ 1 -AR and AKAP79/PKA, consequently knockdown of SAP97 should destabilize the binding between the ␤ 1 -AR and AKAP79. Toward this, cells expressing 1.4 pmol/mg Myc-␤ 1 -AR and FLAG-AKAP79 were transfected with SAP97 siRNA or its scrambled control (Fig. 4,  panel B, h-m). In cells expressing the scrambled siRNA, ␤ 1 -AR precipitates co-immunoprecipitated the ␤ 1 -AR, AKAP79, and SAP97 (Fig. 4, h-j). In cells expressing SAP97 siRNA, precipitates of the ␤ 1 -AR failed to co-immunoprecipitate AKAP79 or SAP97 (Fig. 4, k-m). In addition, the expression of SAP97 was not detected either in the lysates or in the immunoprecipitates prepared from these cells, indicating that the siRNA effectively FIGURE 3. Identification of SAP97 as an interacting partner at the carboxyl-terminal type 1 PDZ of the ␤ 1 -AR. A and B, ␤ 1 -AR does not interact with AKAP79 in far Western blotting assays. FLAG-WT ␤ 1 -AR and FLAG-␤ 1 -AR ⌬PDZ (ϳ50 ng) were immobilized onto nitrocellulose filters. The filters were hybridized with 100 ng/ml FLAG-AKAP79, washed, and then probed with the anti-AKAP79 monoclonal antibody. C, HEK-293 cell lysates express SAP97. D, immobilized Myc-SAP97 (ϳ50 ng) interacts with FLAG-␤ 1 -AR but not with FLAG-␤ 1 -AR⌬PDZ in far Western assays. E, purified GST-␤ 1 -AR-carboxyl-terminal tail (between amino acids 425-477, left panel) binds preferentially to SAP97 in pulldown assays that were conducted as described under "Experimental Procedures." F, Myc-SAP97 was transiently transfected into a HEK-293 cell line stably expressing the WT ␤ 1 -AR-YFP. After 2 days, the cells were exposed to ascorbic acid (No Iso) or to 10 M isoproterenol for 30 min, acid-washed, and treated with 100 M alprenolol for 0, 10, 20, 30, and 45 min. At the end of each time period, the cells were fixed with 4% paraformaldehyde, permeabilized with 1% Triton X-100 in PBS, stained with Cy-3 conjugated to anti-Myc 9E-10 monoclonal antibody, and visualized by dual confocal microscopy (YFP, ex ϭ 514 nm, em ϭ 530LP; Cy3, ex ϭ 543 nm, em ϭ 560 BP) using LSM-510 multitracking configuration. IB, immunoblot; CBB, Coomassie Brilliant Blue; Iso, isoproterenol.
knocked down SAP97 levels (Fig. 4m). Therefore, because the SAP97 siRNA abolished the ability of the ␤ 1 -AR to co-immunoprecipitate AKAP79 and SAP97, it indicates that SAP97 is likely to serve as a bridging molecule between the ␤ 1 -AR and the AKAP79-PKA complex.
Characterization of the Association between the ␤ 1 -AR, SAP97, and AKAP79 by FRET Microscopy-Fluorescence confocal and FRET microscopy provide data relevant to the cellular distribution and proximity of the proteins under study (1,(17)(18)(19). To determine whether SAP97 and the ␤ 1 -AR interacted with one another, WT ␤ 1 -AR-CFP or -YFP and SAP97-YFP or -CFP were generated. The binding and G s -coupling parameters of ␤ 1 -AR-CFP and ␤ 1 -AR-YFP were determined by radioligand binding and adenylyl cyclase assays (Table 1 and Fig. 2C). These chimera bound ICYP with affinities (K D ) comparable with that of the WT ␤ 1 -AR (Table 1). ␤ 1 -AR-CFP and ␤ 1 -AR-YFP chimera displayed basal activities of adenylyl cyclase that were comparable with that of the WT ␤ 1 -AR (Fig.  2C). Moreover, isoproterenol generated a graded escalation in the activity of adenylyl cyclase in chimeric receptors that culminated in an ϳ6-fold increase in its activity, with EC 50 values that were comparable with those of the WT ␤ 1 -AR (Table 1).
Next, WT ␤ 1 -AR-CFP and SAP97-YFP were cotransfected into HEK-293 cells, and their interaction was assessed by acceptor photobleaching FRET microscopy (Fig. 5A). The imaging data indicated that the interactions between the WT ␤ 1 -AR and SAP97 were strong and displayed FRET efficiencies of 24 Ϯ 5.0% (Table 2). To assess the influence of the PDZ site on the distribution of the ␤ 1 -AR and its interaction with SAP97, FRET interaction efficiencies between the ␤ 1 -AR⌬PDZ-CFP and SAP97-YFP were analyzed by FRET microscopy (Fig. 5D). Data from several experiments did not reveal FRET interactions between the ␤ 1 -AR⌬PDZ and SAP97, confirming that the interaction between SAP97 and the ␤ 1 -AR was mediated through the PDZ. Similarly, AKAP79 interacted with the full-length WT ␤ 1 -AR with a moderate FRET efficiency of 12 Ϯ 1.1% (Fig. 5B), and this interaction was abolished when the PDZ site was mutated (Fig. 5E). Finally, SAP97 and AKAP79 interacted with a FRET efficiency of 19 -21%, indicative of high affinity interactions ( Fig. 5C and Table 2). Previously we have shown by FRET microscopy strong interactions between the RII␣ subunit of PKA and AKAP79 and between RII␣ and the WT ␤ 1 -AR (1). These data complement the immunoprecipitation results in Fig. 4 that have shown binding between the SAP97-AKAP79-PKA complex and the WT ␤ 1 -AR.
Additionally, we used FRETN microscopy to determine the efficiency of molecular interactions between the WT ␤ 1 -AR and the SAP97-AKAP79 complex in live cells (Fig. 6). This method generates high resolution, real time images of sensitized emission FRET using image subtraction, which allows comparison of FRET from multiple cells for a given acceptordonor pair (18,19). FRETN showed strong association between the WT ␤ 1 -AR and SAP97 (FRETN efficiency of 23 Ϯ 5%). Furthermore, strong interactions between SAP97 and AKAP79 (FRETN efficiency ϭ 21 Ϯ 2%) and between the WT ␤ 1 -AR and AKAP79 (FRETN efficiency ϭ 12 Ϯ 1.1%) were observed (Fig. 6,  B and C). Finally, in agreement with Nakagawa et al. (20), we confirmed the existence of SAP97 dimers because SAP97-CFP interacted with SAP97-YFP with a high FRET efficiency of 19.7 Ϯ 3% (Fig. 6D).
Effect of the ␤ 1 -AR PDZ and SAP97 on Agonist-mediated Phosphorylation of the ␤ 1 -AR-To study the role of the ␤ 1 -AR PDZ and its associated scaffold in signaling by the ␤ 1 -AR, we determined whether inactivation of the ␤ 1 -AR PDZ or knockdown of SAP97 produced comparable effects on ␤-agonist-mediated phosphorylation of the ␤ 1 -AR. Cell lines stably expressing comparable levels of FLAG-tagged WT ␤ 1 -AR (1.05 Ϯ 0.12 pmol/mg) and of FLAG-tagged ␤ 1 -AR⌬PDZ (1.2 Ϯ 0.1 pmol/ Lysates from these cells were immunoprecipitated (IP) with anti-Myc or anti-FLAG epitope antibodies. Precipitates were immunoblotted (IB) for ␤ 1 -AR and SAP97 as indicated. SAP97 co-immunoprecipitated with WT ␤ 1 -AR, but not with ␤ 1 -AR⌬PDZ. d and e, co-immunoprecipitation of ␤ 1 -AR and AKAP79-PKA complexes from HEK-293 cells. Lysates from HEK-293 cells transfected with Myc-SAP97 and FLAG WT ␤ 1 -AR were immunoprecipitated with FLAG antibodies. AKAP79 and the RII␣-subunit of PKA co-immunoprecipitated with the WT ␤ 1 -AR. f and g, co-immunoprecipitation of AKAP79 and SAP97 from HEK-293 cells. Lysates from HEK-293 cells transfected with Myc-SAP97 and FLAG AKAP79 were immunoprecipitated with FLAG or Myc antibodies. AKAP79 co-immunoprecipitated with SAP97. h-m, knockdown of SAP97 expression using siRNA eliminates the binding of FLAG-AKAP79 to Myc-␤ 1 -AR in HEK cells (k-m), but the interaction between AKAP79 and the ␤ 1 -AR is not eliminated in cells expressing the scrambled siRNA (h-j). In all experiments, lysates represented 5% of the total extract, whereas immunoprecipitations represent 30% of the total volume. FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 5091 mg) were used. Isoproterenol-mediated phosphorylation of the WT ␤ 1 -AR or ␤ 1 -AR⌬PDZ increased total phosphorylation of the WT ␤ 1 -AR by 6-fold and that of the ␤ 1 -AR⌬PDZ by 2-fold (Fig. 7A). These experiments were also performed in cells expressing the FLAG-tagged WT ␤ 1 -AR with scrambled or SAP97 siRNAs. Isoproterenol increased total phosphorylation of the WT ␤ 1 -AR by 5.7-fold in control or in cells expressing the scrambled siRNA (Fig. 7A, compare lanes 1 to 2 and 5  to 6). However, in cells expressing the ␤ 1 -AR⌬PDZ or in those co-expressing the WT ␤ 1 -AR and SAP97 siRNA, isoproterenol increased the phosphorylation of the ␤ 1 -AR by ϳ2.5-fold (Fig. 7A, compare lanes 3 to 4 and 7 to 8). Thus, inactivation of the ␤ 1 -AR PDZ or knockdown of SAP97 was roughly equivalent in inhibiting the phosphorylation of the ␤ 1 -AR by isoproterenol.

Trafficking of the ␤ 1 -AR by a PDZ-anchored Receptosome
The two major kinases that are involved in agonist-mediated phosphorylation of the ␤ 1 -AR are the GRK and PKA (27). Thus, the underlying cause for reduced phosphorylation of the ␤ 1 -AR⌬PDZ might be due either to inhibition of GRK, PKA, or both. The preferred substrates for phosphorylation by PKA are serine/threonine residues that are preceded by RX-or RRX-(where X is any amino acid and R is arginine). This organization is found solely around Ser 312 in the 3rd IC, which corresponds to RRPS 312 (11). The preferred substrates for phosphorylation by GRK are serine/threonine residues that are preceded by an acidic amino acid (28,29). This organization is found around four serine residues that reside exclusively in the carboxyl terminus of the ␤ 1 -AR (27). To identify the kinase affected by knockdown of SAP97 or by the inactivation of the PDZ, the phosphorylation of the carboxyl terminus versus that of the 3rd IC should be independently determined. Therefore, the phosphorylated ␤ 1 -AR was cleaved with cyanogen bromide that cleaves the full-length ␤ 1 -AR into a 10-kDa 32 P-labeled peptide that encompasses the 3rd IC and into a 15-kDa 32 Plabeled peptide that encompasses the carboxyl terminus of the ␤ 1 -AR (1). Previously we have shown that isoproterenol increased the phosphorylation of the 10-and 15-kDa peptides by ϳ6-fold (1). However, the counts/min derived from cyano- FIGURE 5. Acceptor photobleaching analysis of FRET interactions between SAP97, AKAP79, and the ␤ 1 -AR. A and B, cells co-expressing the WT ␤ 1 -AR-CFP with either SAP97-YFP or AKAP79-YFP were photobleached, and FRET was recorded by examining the loss of quenching of CFP during acceptor (YFP) photobleaching. FRET was observed between the WT ␤ 1 -AR and SAP97 and between the WT ␤ 1 -AR and AKAP79, with a FRET efficiency of 24 Ϯ 5 and 12 Ϯ 1.1%, respectively. C, FRET signals between AKAP79 and SAP97 were also recorded with FRET efficiency between 19 and 21%. These data verify that interactions between WT ␤ 1 -AR, AKAP79, and SAP97 have occurred with relatively high efficiency. D and E, however, in cells co-expressing the ␤ 1 -AR⌬PDZ-CFP with either SAP97-YFP or AKAP79-YFP no FRET interactions were observed, indicating that mutagenesis of the PDZ domain of the ␤ 1 -AR abrogated the interactions between the ␤ 1 -AR and either SAP97 or AKAP79. The color ruler shows the relationship between the pseudo-FRET color and the corresponding FRET efficiency reported in Table 2.

TABLE 2 FRET efficiencies in (%) as recorded by the acceptor photobleaching method in fixed cells
FRET efficiency was calculated using area averages for donor (D) before and after bleaching. FRET ϭ (D after ϪD before)/D after. Donor (D) and acceptor (A) threshold values were subtracted from all pixels before FRET calculation. The mean values of FRET efficiencies calculated for at least 10 specific regions of interest in each FRET pair from seven experiments were collected and analyzed with FRET tool software for LSM510 version 3.2.

Trafficking of the ␤ 1 -AR by a PDZ-anchored Receptosome
gen bromide cleavage of an equivalent amount of receptor protein from cells pre-exposed to ascorbic acid were insufficient to accurately estimate the incorporation of 32 P into these peptides under basal conditions (1). Therefore, the 32 P-labeled ␤ 1 -ARs on nitrocellulose filters were digested with cyanogen bromide, and the amounts of 32 P incorporated into the resulting peptides were determined. Then an equal number 32 P counts/min derived from isoproterenol-and ascorbic acid-treated cells were subjected to electrophoresis on 16% acrylamide/Tricine gels (Fig. 7B). In the samples derived from cells pre-exposed to ascorbic acid, we observed that the 10-and 15-kDa peptides were phosphorylated at a ratio of ϳ1:3 (Fig. 7B, lane 1). Isoproterenol increased the total phosphorylation of each peptide by ϳ6-fold (1) but did not alter their phosphorylation ratios, indicating that both the 10-and 15-kDa peptides were substrates for isoproterenol-mediated phosphorylation (Fig. 7B, lane 2). Cleavage of 32 P-␤ 1 -AR⌬PDZ from control or isoproterenoltreated cells, followed by loading the same number of 32 P counts/min as in Fig. 7B, lanes 1 and 2, generated the 15-kDa phosphopeptide only (Fig. 7B, lanes 3 and 4). Thus, mutagenesis of the PDZ abrogated basal and isoproterenol mediated phosphorylation of the 10-kDa peptide. The next series of experiments was conducted in cells expressing the WT ␤ 1 -AR with scrambled or SAP97 siRNA (Fig. 7, lanes 5-8). We have shown that siRNA-mediated knockdown of AKAP79 inhibited the phosphorylation of the 10-kDa peptide in response to isoproterenol (1). However, we were not able to estimate the effect of the AKAP79 siRNA on the peptides derived from ascorbic acid-treated cells. Thus, the 32 P-labeled WT ␤ 1 -AR from control or isoproterenol-treated samples was cleaved, and equal amounts of 32 P counts/min as in Fig. 7B, lanes 1-4, were subjected to electrophoresis. The 32 P-WT ␤ 1 -AR derived from control or isoproterenol-treated cells that expressed the scrambled siRNA generated the 10-and 15-kDa phosphopeptides at a ratio of ϳ1:3 (Fig. 7B, lanes 5 and  6). However, the 32 P-labeled WT ␤ 1 -AR derived from control or isoproterenol-treated cells that expressed the SAP97 siRNA generated the 15-kDa phosphopeptide only (Fig. 7B, lanes 7 and  8). Therefore, inactivation of the ␤ 1 -AR PDZ or knockdown of SAP97 both inhibited the phosphorylation of the 10-kDa peptide derived from the 3rd IC that contains the putative PKA-Ser 312 substrate. These results buttress our claim that inactivation of the PDZ inhibited the targeting of the AKAP79-PKA a] represent the total concentrations (interacting and noninteracting) of the donor-and acceptor-labeled species, respectively. FRET1 is proportional to the FRET signal from the specimen. Dfd is the donor signal that would take place if no FRET occurred and is therefore proportional to the total concentration of the donor. Afa is the acceptor signal that would take place if no FRET occurred and is therefore proportional to the total concentration of the acceptor. FRETN values for the various constructs are presented in Table 2.  1 and 2), whereas the ␤ 1 -AR⌬PDZ showed reduced phosphorylation in response to isoproterenol (lanes 3 and 4). siRNA-mediated knockdown of SAP97 also reduced the phosphorylation of the WT  1 and 2) or the ␤ 1 -AR⌬PDZ (lanes 3 and 4), were subjected to electrophoresis on 16% acrylamide gels in Tricine cathode buffer. These experiments were repeated in cells expressing the WT ␤ 1 -AR with scrambled (Scr, lanes 5 and 6) or SAP97 siRNAs (lanes 7 and 8). Electronic counting of 32 P incorporated in lane 2 indicated that the % of the counts/min in the 10-versus the 15-kDa band was 34% (190 -571 cpm, respectively). In lanes 3 and 4, ϳ653 cpm were counted in each lane that was exclusively located in the 15-kDa band. In lanes 5 and 6, the % counts/min in the 10versus the 15-kDa band was 30% (202-623 cpm, respectively). In lanes 7 and 8, ϳ693 cpm were counted in each lane that was exclusively located in the 15-kDa band. FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 complex to the ␤ 1 -AR and inhibited PKA-mediated phosphorylation of Ser 312 in the 3rd IC.

Trafficking of the ␤ 1 -AR by a PDZ-anchored Receptosome
Cross-talk between the ␤ 1 -AR PDZ Scaffold and Ser 312 -Two distinct domains in the ␤ 1 -AR, namely Ser 312 in the 3rd IC and the PDZ in the carboxyl-terminal tail, participate in imparting a recycling signal to the ␤ 1 -AR. If, as suggested earlier, one of the functions of the PDZ domain is to target PKA to the ␤ 1 -AR, then we hypothesized that overexpression of PKA might overcome the effect of inactivating the PDZ on recycling of the ␤ 1 -AR. To address this question, the recycling experiment for ␤ 1 -AR⌬PDZ was conducted under conditions of increased PKA activation (Fig. 8A). HEK-293 cells expressing 1.2 Ϯ 0.1 pmol/mg of FLAG-␤ 1 -AR⌬PDZ were transfected with the empty vector or with the vector expressing cPKA. In cells expressing the empty vector, the agonist-internalized ␤ 1 -AR⌬PDZ did not recycle (Fig. 8A, images d-g). In cells expressing cPKA, the ␤ 1 -AR⌬PDZ recycled back to the cell surface within 45 min from the removal of isoproterenol (Fig.  8A, images l-o). A boundary was drawn around the inner circumference of the cells in Fig. 8A in order to determine the distribution of pixels between membranous and intracellular compartments. The density of the pixels residing inside the boundary versus those residing outside the boundary was used as an index for internalized and membranous ␤ 1 -AR, respectively. The pixel data were plotted as a function of time after the removal of isoproterenol in order to calculate the recycling kinetics of the ␤ 1 -AR (Fig. 8B). The data indicate that in cells expressing the ␤ 1 -AR⌬PDZ with the empty pcDNA vector, the internalized ␤ 1 -AR⌬PDZ did not recycle (Fig. 8A, images d-g). However, in cells expressing the ␤ 1 -AR⌬PDZ and cPKA, the agonist-internalized ␤ 1 AR⌬PDZ recycled with a t 0.5 ϭ 20 Ϯ 5 min, indicating that super induction of cPKA overcame the FIGURE 8. Roles of the ESKV sequence in the carboxyl-terminal tail of the ␤ 1 -AR and Ser 312 in the 3 rd IC in regulating the recycling and resensitization of the ␤ 1 -AR in response to isoproterenol. A, recycling of the ␤ 1 -AR⌬PDZ in response to 10 M isoproterenol (Iso). HEK-293 cells stably expressing the ␤ 1 -AR⌬PDZ were mock-transfected (pcDNA, images a-h) or transfected with an expression vector for the catalytic subunit of PKA (images i-p). In addition, HEK-293 cells were transiently transfected with the point mutant of ␤ 1 -AR⌬PDZ in which the serine residue at position 312 was mutated to aspartic acid (S312D ␤ 1 -AR⌬PDZ, images q-y). Recycling of the ␤ 1 -AR in response to 10 M isoproterenol (n ϭ 3) for 30 min followed by an acid wash (A/W) to remove the antibody bound to extracellular ␤ 1 -AR were conducted as described under "Experimental Procedures." Each scale bar represents 5 m. B, summary of the results of the experiment in A are presented as line graphs of n ϭ 3. The isoproterenol-internalized ␤ 1 -AR⌬PDZ did not recycle (images d-h), whereas the ␤ 1 -AR⌬PDZ with cPKA and the S312D ␤ 1 -AR⌬PDZ recycling was rapid with a t 0.5 of 25 Ϯ 5 and 18 Ϯ 4 min, respectively. C, comparison of adenylyl cyclase activities in response to short term isoproterenol (short term desensitization) and in response to isoproterenol followed by antagonist (desensitization followed by resensitization) treatments in HEK-293 cells expressing WT, ⌬PDZ, S312D, and S312D⌬PDZ constructs of the ␤ 1 -AR. These experiments were replicated (n ϭ 3) each in triplicate. effect of the PDZ mutation on recycling of the ␤ 1 -AR. If the recycled ␤ 1 -AR⌬PDZ was inserted properly into the cell membrane, then Cy3-conjugated anti-FLAG IgG bound to the amino-terminal FLAG epitope should be oriented extracellularly. In this case, a second acid wash would strip Cy3-IgG from the recycled receptor population. In agreement with these assumptions, we observed reduced cell fluorescence in the recycled ␤ 1 -AR⌬PDZ in cPKA expressing cells, but not in pcDNA expressing cells (Fig. 8A, compare image p with h). It should be emphasized however, that a 30-min pretreatment with forskolin, which activates all the isoforms of adenylyl cyclase and markedly activates PKA, did not restore the recycling phenotype to ␤ 1 -AR⌬PDZ (data not shown), suggesting that chronic activation of PKA might be necessary to restore the recycling of the ␤ 1 -AR⌬PDZ.
If one of the functions of the ␤ 1 -AR PDZ is to facilitate the phosphorylation of Ser 312 by targeting PKA to the ␤ 1 -AR, then it is logical to hypothesize that replacement of Ser 312 with the phosphoserine mimic aspartic acid would restore the recycling of the ␤ 1 -AR⌬PDZ. This hypothesis was tested directly by generating a ␤ 1 -AR⌬PDZ construct in which the serine at position 312 was mutated to aspartic acid (S312D⌬PDZ). Indeed the agonist-internalized S312D⌬PDZ recycled efficiently (Fig. 8A, images t-w), with kinetics comparable with those of the WT-␤ 1 -AR (t 0.5 15 Ϯ 4 min) (Fig. 8B). Because recycling of the agonist-desensitized and internalized GPCR is a priori for its resensitization, we determined whether ␤ 1 -AR-mediated activation of adenylyl cyclase was functionally resensitized in those ␤ 1 -AR⌬PDZ constructs that were capable of recycling (Fig. 8C). Rapid desensitization of adenylyl cyclase in membranes expressing all the four ␤ 1 -AR constructs described in Fig. 8C was observed after 10 min of exposing the cells to isoproterenol, indicating that mutagenesis of either the PDZ or Ser 312 alone or in combination did not affect short term desensitization of the receptor. The resensitization assay involves the desensitization of the ␤ 1 -AR by exposing cells to isoproterenol for 3 h, followed by incubating the cells with 100 M of the ␤-antagonist alprenolol to induce the recycling of the ␤ 1 -AR and subsequent resensitization of its adenylyl cyclase activity (1,11). In this assay, we observed the resensitization of adenylyl cyclase activity in the WT ␤ 1 -AR, but the activity of adenylyl cyclase of the ␤ 1 -AR⌬PDZ was significantly desensitized (Fig. 8C). Functional resensitization of adenylyl cyclase activity of the ␤ 1 -AR⌬PDZ was restored in the context of S312D⌬PDZ construct, indicating that the modification of Ser 312 to its phosphoserine mimic "aspartic acid" resuscitated the recycling and resensitization of ␤ 1 -AR⌬PDZ.
Characterization of the Role of SAP97 in Recycling of the Human ␤ 1 -AR-Thus far, we have shown that cross-talk between the ␤ 1 -AR PDZ domain and Ser 312 was involved in regulating the recycling and resensitization of the ␤ 1 -AR in HEK-293 cells. To determine the role of SAP97 in this phenomenon, the effects of SAP97 knockdown and overexpression on recycling of the WT ␤ 1 -AR were assessed (Figs. 9 and 10). In cells stably expressing 1.1 pmol/mg protein of WT ␤ 1 -AR-YFP, knockdown of SAP97 inhibited the recycling of the WT ␤ 1 -AR as determined by the confocal recycling assay (Fig. 9A, images  i-l). However, knockdown of SAP97 had variable effects on agonist-induced internalization of the ␤ 1 -AR as well, whereas in some cells SAP97 had no effect on internalization, and in others it reduced the internalization by ϳ35% (compare internal pixels in Fig. 9A, images h versus b). Therefore the effect of knockdown of SAP97 on trafficking of the ␤ 1 -AR was determined by surface biotinylation because in this assay the internalization and recycling data are derived from the entire cell population rather than from few imaged cells (1). HEK-293 cells stably expressing FLAG-tagged WT ␤ 1 -AR with scrambled or SAP97 siRNA were surface-biotinylated with cleavable biotin followed by quenching of excess biotin with glycine. The amount of biotin incorporated into the ␤ 1 -AR under this condition indexed total cellular ␤ 1 -AR biotinylation (Fig. 9B, lanes  1 and 6). The cells were then exposed to isoproterenol for 30 min, followed by cleavage of the remaining cell surface biotin (Fig. 9B, lanes 2 and 7). The amount of biotin recovered in this step indexed the amount of biotinylated ␤ 1 -AR that was internalized in response to isoproterenol, whereas the ratio of internal to total biotin indexed the percentile of total receptors that were internalized. Isoproterenol induced the internalization of ϳ60% of total ␤ 1 -AR in control cells, whereas the internalization of the ␤ 1 -AR in SAP97 siRNA-treated cells was reduced by ϳ20 (n ϭ 4). To initiate recycling, isoproterenol was replaced with the ␤-antagonist alprenolol, and the cells were warmed to 37°C for an additional 15, 30, or 60 min (Fig. 9B, lanes 3-5 and  8 -10). After each time period, the cells were cooled to 4°C, and biotin was cleaved for the second time to ensure cleavage of any newly appearing "recycled" surface biotin. Thus, the loss of biotin from the second cleavage step indexed the recycling of the ␤ 1 -AR. The data indicate that by 60 min, more than 90% of the biotin was lost from the ␤ 1 -AR in cells expressing the scrambled siRNA, reflecting membrane recycling and subsequent biotin cleavage (Fig. 9B, lanes 3-5). In contrast, the internalized (biotinylated) ␤ 1 -AR in cells expressing SAP97 siRNA was not changed even after 1 h from the removal of isoproterenol, reflecting their internal distribution (Fig. 9b, compare lanes 9 and 10 with lanes 4 and 5). Next, we quantified the amount of biotin remaining as a function of time after the removal of isoproterenol and determined that the ␤ 1 -AR recycled with a t 0.5 of 25 Ϯ 5 min (Fig. 9C).
In follow-up experiments, the effect of overexpression of Myc-SAP97 or Myc-PSD-95 in cells stably expressing 1.1 Ϯ 0.2 pmol/mg protein of WT ␤ 1 -AR-CFP on isoproterenol-mediated ␤ 1 -AR internalization and recycling was determined by confocal microscopy (Fig. 10A). The rationale for analyzing the effect of PSD-95 is that PSD-95 binds to the ␤ 1 -AR PDZ with high affinity and interferes with the internalization of the receptor, but its effect on recycling is unknown (26). However, because HEK-293 cells do not express PSD-95 (Fig. 10B), PSD-95 was overexpressed along with SAP97. Overexpression of SAP97 did not affect the rate or magnitude of isoproterenolmediated internalization of the ␤ 1 -AR (t 0.5 ϭ 5 min Ϯ 1 min). On the other hand, overexpression of PSD-95 markedly inhibited the magnitude (Ϫ50%) and rate of isoproterenol-mediated internalization of the WT ␤ 1 -AR (Fig. 10, A, images u-z, and C). Overexpression of SAP97 did not affect the rate or magnitude of ␤ 1 -AR recycling, but PSD-95 reduced both the rate and the magnitude of ␤ 1 -AR recycling (Fig. 10C). Thus, these MAGUK proteins exerted different effects on internalization, recycling, and resensitization of the ␤ 1 -AR. Finally, knockdown of AKAP79 or SAP97 in conjunction with FRET microscopy was conducted to determine the organization of the scaffolding complex that binds to the ␤ 1 -AR PDZ (Fig. 11). Knockdown of AKAP79 did not prevent the association between the WT ␤ 1 -AR-YFP and SAP97-CFP as assessed by acceptor photobleaching FRET microscopy (Fig. 11A). Knockdown of SAP97, however, abolished the interaction between WT ␤ 1 -AR and AKAP79, indicating that SAP97 served as bridging molecule between the ␤ 1 -AR and the AKAP79-PKA complex (Fig. 11, B and C).

DISCUSSION
Type 1 PDZ domains in the tails of GPCR such as the ␤ 1 -or the ␤ 2 -AR are necessary for efficient recycling of these receptors (10, 30 -32). The type 1 PDZ in the ␤ 2 -AR (DSLL) interacts with the PDZ-binding domain in the Na ϩ /H ϩ exchanger regulatory factor ((NHERF) also known as ezrin-radixin-moesin (ERM)-binding phosphoprotein-50 (EBP50)), and this interaction is required for recycling of the ␤ 2 -AR (31). Similarly, inactivation of the ESKV sequence of the ␤ 1 -AR by a variety of point mutations generated ␤ 1 -AR mutants that, although efficiently internalized in response to isoproterenol, were not recycled (10,32). The type 1 PDZ in the human ␤ 1 -AR has attracted wide attention because it interacted with numerous proteins related to the MAGUK family, and these interactions exerted different effects on its functions. The association of the ␤ 1 -AR with the MAGUK protein PSD-95, for example, inhibited the internalization of the ␤ 1 -AR but facilitated its interaction with N-methyl-D-aspartate (NMDA) receptors (26). Another family of PDZinteracting proteins was the MAGI-related proteins, which increased the magnitude of agonist-induced internalization of the ␤ 1 -AR (25,33). Two additional PDZ-binding proteins were shown to interact with the type 1-PDZ in the carboxyl-terminal tail of the ␤ 1 -AR. The first was a protein called the ␤ 1 -ARbinding partner, which is involved in regulating ␤ 1 -AR-mediated activation of extracellular signal-regulated kinases 1/2 (34). The other was the cystic fibrosis transmembrane conductance regulator-associated ligand, which is involved in surface expression of the ␤ 1 -AR (35).
These MAGUK proteins, however, are static entities because they lack ATP-binding and catalytic core motifs and are thought to function as adaptor proteins (23). Therefore, in addition to identifying SAP97 as a novel MAGUK protein that interacts with the ␤ 1 -AR PDZ, we have confirmed that an AKAP79-PKA complex binds to SAP97. This novel organiza-  6 -10). C, rate of biotinylated ␤ 1 -AR recycling in cells in which the expression of SAP97 was knocked down versus those with control SAP97 levels. Internalized WT ␤ 1 -AR did not recycle in HEK-293 cells in which SAP97 was knocked down, whereas it recycled with t 0.5 of 25 Ϯ 5 min in scrambled siRNA controls.
tion generates a dynamic MAGUK-␤ 1 -AR complex with ATP binding and catalytic core motifs to create a novel ␤ 1 -AR signalosome that broadens the range of functions attributed to this receptor.
In addition to its assigned role in recycling of the agonistinternalized ␤ 1 -AR, a central function of this ␤ 1 -adrenergic receptosome is to ensure potency, fidelity, and reversibility of ␤ 1 -AR signaling. Potency and fidelity of signaling are ensured by high affinity connections between the PDZ and the various members of the scaffold. These interconnections ensure that cyclic AMP, generated by agonist-mediated activation of the receptor-G s -adenylyl cyclase axis, activates the PKA bound to the PDZ domain of the ␤ 1 -AR (Fig. 11C). Activated PKA phosphorylates Ser 312 in the 3rd IC of the ␤ 1 -AR as well as any acceptor protein bound to the MAGUK or AKAP members of the scaffold (Fig. 11C). The cyclic AMP signal is quickly terminated by phosphodiesterases bound to AKAP79 or somewhere else in this scaffold (36). In addition to activating the G s -adenylyl cyclase axis, agonist binding to the ␤ 1 -AR causes conformational changes in the receptor that facilitate GRK-mediated phosphorylation of acidotrophic residues such as Ser 475 at position Ϫ2 in the middle of the ␤ 1 -AR PDZ (28,29). These modi-fications promote physical separation between the scaffold and the ␤ 1 -AR PDZ and facilitate the internalization of the receptor (1) (Fig. 3F). Thus, functional synergism between homologous desensitization through GRK and homologous resensitization through PKA are capable of generating several cycles of receptor activation, desensitization, and recovery, which potentially can maintain the signaling output from cells with low density of ␤ 1 -AR.
In the model described above, we proposed that one of the functions of the ␤ 1 -AR PDZ was to target PKA to the ␤ 1 -AR to facilitate the phosphorylation of Ser 312 . This is a novel idea because it hypothesizes that a major function of the PDZ-associated scaffold is the transmission of signaling information. This suggestion is novel because the hypothesis of Gage et al. (10), which states "that PDZ domain-mediated protein interactions are sufficient to promote rapid recycling of GPCR," does not take into account the involvement of other downstream elements in GPCR recycling. Thus, we proceeded to test our hypothesis that "PDZ domain-mediated protein interactions transmit recycling signal(s) that promote rapid recycling of the GPCR." At first we confirmed the involvement of the ␤ 1 -AR PDZ, SAP97, in addition to AKAP79 (1) in phosphorylating The effect of isoproterenol on the distribution of YFP pixels outside versus those inside a 300-nm boundary in images a-f, k-p, and u-z, was determined. Because the number of pixels inside the boundary progressively increased after the addition of isoproterenol, this was reflected as a reduction in the percentile of membranous receptors in the ϩ isoproterenol images. After the initiation of recycling (Ϫisoproterenol), there was a progressive decline in the pixels inside the 300-nm boundary and a corresponding increase in the pixels outside this boundary in images g-j, q-t, and aЈ-dЈ. The data are derived from n ϭ 2 experiments each utilizing 5-7 cell images per time point.
Ser 312 , thereby providing a vivid example of cross-talk between a domain in the extreme carboxyl terminus and another in the 3rd IC of the GPCR (Figs. 7 and 11C). If the phosphorylation of Ser 312 lies downstream from the PDZ in setting the trafficking itinerary for the ␤ 1 -AR, then the PDZ-generated recycling signal, which is phospho-Ser 312 , could be replicated either by super-induction of PKA or by mutagenesis of Ser 312 in the 3rd IC of the ␤ 1 -AR to the phosphoserine mimic aspartic acid (S312D). Our data in Fig. 8 showed that the ␤ 1 -AR⌬PDZ would recycle under conditions where the catalytic activity of PKA was chronically elevated or when its putative target Ser 312 in the ␤ 1 -AR⌬PDZ mutant was replaced by its phosphoserine mimic aspartic acid. These findings imply that Ser 312 is downstream from the PDZ and apparently occupies a more dominant position than the type 1 PDZ in setting the trafficking itinerary of the ␤ 1 -AR. We arrived at this conclusion because our model extended beyond the binary MAGUK-PDZ model into a quaternary model that incorporates an AKAP, which is involved in targeting PKA to the ␤ 1 -AR microdomain (Fig. 11C).
The role of PKA targeting in setting the trafficking itinerary of GPCR is a nascent field with relatively few reports. The role of PKA in translocating aquaporin-2 from intracellular compartments into cell membranes in response to vasopressin and in recycling of agonist-internalized NMDA receptors is well documented (15,37). Nevertheless, beyond these few examples, the significance of PKA and its targeting to the PDZ motif in trafficking of the agonist-internalized GPCR remains to be substantiated. Recently, the EBP-50-binding protein ezrin was identified as an AKAP-like protein, which binds to the RI subunit of PKA, instead of the more common RII subunit (38). These data suggest that EBP-50, which binds to the ␤ 2 -AR PDZ and is required for the recycling of the ␤ 2 -AR, targets an AKAPlike protein to the ␤ 2 -AR PDZ. Similarly, aquaporin-2 has a type 1 "G(S/T)KA" PDZ sequence at its extreme carboxyl terminus, and its recycling requires EBP-50 and involves AKAP/PKAmediated phosphorylation of Ser 256 (37). These observations suggest that PDZ-mediated targeting of PKA to the GPCR might be more common than expected and could be involved in the recycling of these and other proteins.
The identification of SAP97 as an organizer of a scaffold composed of AKAP79 and PKA converts the PDZ-binding domain from a static multiprotein-binding complex into a dynamic network with increased number, range, and intensity of signals that are transmitted via the ␤ 1 -AR. The PDZ domains of the PSD/SAP family of MAGUK proteins interact with the carboxyl-terminal type 1 PDZ motif found in a variety of membrane and intracellular proteins, including Shaker K ϩ channels, NMDA receptors, and the ␤ 1 -AR, which binds to PDZ-3 domain of SAP97 (33,39,40). The Src homology 3 and guanylate kinase domains are involved in the clustering activity of SAP97 because they bind to the GKAP/PAPAP/DAP family of postsynaptic density proteins as well as to many other signaling/scaffolding proteins, including AKAP79 and other PKAbinding proteins (22,41). SAP97 contains an additional L27 domain in its amino terminus that is involved in dimerization of SAP97 and in binding to other MAGUK proteins such as mLIN-2/CASK and DLG3 (20,23). These protein-protein interactions can therefore diversify the signaling of the ␤ 1 -AR and may explain some of the neuronal and cardiovascular functions attributed to the ␤ 1 -AR. For example, myocardial ␤ 1 -ARs selectively phosphorylate via PKA many substrates that are involved in regulating myocardial contractility through AKAPdependent mechanisms (42). In neurons, SAP97 and AKAP79 both interact with the GluR1 ␣-amino-3-hydroxy-5-methyl-4isoxazole propionic acid receptor subunit, and the interaction between SAP97 and AKAP79 might recruit PKA to ionotropic glutamate receptors (17,22). Thus it is conceivable that multiplexing between a ␤ 1 -AR-MAGUK-AKAP79 complex might scaffold ␣-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptors, L-type Ca 2ϩ channels, and other signaling molecules to the ␤ 1 -AR microdomain to facilitate their phosphorylation by the ␤ 1 -AR signaling pathway as has been observed in hippocampal neurons (43). These scaffold-mediated connections between the ␤ 1 -AR and voltage-gated channels, for exam- FIGURE 11. Characterization of the WT-␤ 1 -AR, SAP97, and AKAP79 scaffold by acceptor photobleaching FRET microscopy. A, FRET was obtained in HEK-293 cells expressing WT ␤ 1 -AR-YFP and SAP97-CFP when AKAP79 was knocked down with siRNA. B, FRET was not obtained in HEK-293 cells expressing WT ␤ 1 -AR-YFP and AKAP79-CFP when SAP97 was knocked down with siRNA. C, organization of the ␤ 1 -AR receptosome. Based upon our data, we hypothesize that the ␤ 1 -AR is in a macromolecular complex with SAP97, AKAP79, and the PKA tetramer. Agonist-induced activation of the ␤ 1 -AR generates cyclic AMP, which binds to PKA and releases the cPKA that phosphorylates Ser 312 in the 3rd IC of the ␤ 1 -AR. In addition, cPKA might phosphorylate other targets that are brought into the vicinity of the ␤ 1 -AR microdomain through their association with AKAP79 or SAP97. ple, might underlie the observed effects of the ␤ 1 -AR on synaptic plasticity (44,45) and in the formation of emotionally charged memories that could result in post-traumatic stress disorders (46 -48). The mechanism of ␤ 1 -selective ␤-blockers in blocking the reconsolidation of traumatic memories, especially those following acts of terror, might be explained by the mélange of proteins that are scaffolded directly or indirectly with the ␤ 1 -AR through this novel receptosome (49,50).