G Protein-coupled Receptor Kinase and β-Arrestin-mediated Desensitization of the Angiotensin II Type 1A Receptor Elucidated by Diacylglycerol Dynamics*

Receptor desensitization progressively limits responsiveness of cells to chronically applied stimuli. Desensitization in the continuous presence of agonist has been difficult to study with available assay methods. Here, we used a fluorescence resonance energy transfer-based live cell assay for the second messenger diacylglycerol to measure desensitization of a model seven-transmembrane receptor, the Gq-coupled angiotensin II type 1A receptor, expressed in human embryonic kidney 293 cells. In response to angiotensin II, we observed a transient diacylglycerol response reflecting activation and complete desensitization of the receptor within 2-5 min. By utilizing a variety of approaches including graded tetracycline-inducible receptor expression, mutated receptors, and overexpression or short interfering RNA-mediated silencing of putative components of the cellular desensitization machinery, we conclude that the rate and extent of receptor desensitization are critically determined by the following: receptor concentration in the plasma membrane; the presence of phosphorylation sites on the carboxyl terminus of the receptor; kinase activity of G protein-coupled receptor kinase 2, but not of G protein-coupled receptor kinases 3, 5, or 6; and stoichiometric expression of β-arrestin. The findings introduce the use of the biosensor diacylglycerol reporter as a powerful means for studying Gq-coupled receptor desensitization and document that, at the levels of receptor overexpression commonly used in such studies, the properties of the desensitization process are markedly perturbed and do not reflect normal cellular physiology.

The family of heptahelical receptors, also called seven-transmembrane receptors (7TMRs) 4 or G protein-coupled recep-tors, regulates myriad physiological and pathological signal transduction pathways. Despite a large diversity of ligands, including catecholamines, chemokines, lipids, peptides, odorants, and photons, these receptors share remarkable similarity in their intracellular regulatory mechanisms. Classically, receptor function is initiated by ligand-induced activation of heterotrimeric G proteins. This is followed by receptor inactivation, mediated by receptor phosphorylation by G protein-coupled receptor kinases (GRKs) and other kinases, ␤-arrestin recruitment, and receptor internalization (1,2). This inactivation results in desensitization, the loss of agonist efficacy following sustained stimulation.
Until recently our appreciation of the kinetics of 7TMR molecular pharmacology has been limited to extrapolation from either complex in vivo readouts such as changes in blood pressure or from biochemical techniques such as phosphoinositide hydrolysis assays (3), which cannot be performed in live, unperturbed cells. Thus, it has not been possible to observe receptor level desensitization in the continuous presence of agonist, the situation that most closely parallels many physiological circumstances. Consequently, despite a vast literature on 7TMR desensitization, uncertainty remains regarding how this process modulates receptor signaling kinetics.
The advent of fluorescent biosensor technology and the increasing number of intracellular signal indicators afford the opportunity to refine our understanding of heptahelical receptor signal kinetics. Indeed, these novel approaches have elucidated the spatiotemporal details of heptahelical receptor signaling at subcellular and sub-second scale (4 -8). These powerful new technologies include fluorescent small molecule indicators (9), green fluorescent protein-fused protein localization (10), and most recently resonance energy transfer-based measurements of protein modification, protein conformational change, and protein-protein interactions (11).
Such biosensors have the potential to provide a new appreciation of 7TMR desensitization kinetics. Thus, we set out to develop an assay system that affords sensitive and non-disruptive measurement of 7TMR signal regulation. This would allow us to test the specificity and kinetics of GRK and ␤-arrestin functions for receptor signal output. Using the angiotensin II type 1A receptor (AT1 A R) as a model, we designed a system that elucidates the kinetics of receptor regulation by real-time detection of the second messenger diacylglycerol (DAG). This has enabled us to determine the rate of receptor desensitization in the continuous presence of a constant ligand concentration and to test how changing concentrations of receptor, GRKs, and ␤-arrestins alter desensitization kinetics.
siRNA and Plasmid Transfections-GRK2, 3, 5, and 6 siRNA were all previously described (12). Transfection of GRK2 siRNA plus 100 ng of pcDNA3-DAGR into HEK-TetOn-AT1 A R cells was performed using Gene Silencer (Gene Therapy Systems) as previously described (12,14). To compensate for GRK2 siRNAmediated receptor up-regulation, 50 ng/ml doxycycline was added to control samples and no doxycycline was added to GRK2 siRNA samples, resulting in equivalent expression levels. For ␤-arrestin1/2 silencing via siRNA, control siRNA or siRNA targeting both ␤-arrestins1 and 2 (15) plus 100 ng of pcDNA3-DAGR were transfected into HEK-TetOn-AT1 A R cells using Gene Silencer. 24 h after transfection, cells were treated with 100 ng/ml doxycycline to induce a modest level (170 fmol/mg) of angiotensin receptor expression, which was not different between the two siRNA conditions. For ␤-arrestin and GRK2 overexpression experiments, the 1 g of the appropriate ␤-arrestin plasmid or 0.5 g of pRK5-bGRK2 was cotransfected with 100 ng of pcDNA3-DAGR into a 10-cm plate of AT1 A R-293 cells using FuGENE 6 (Roche Applied Science) according to the manufacturer's instructions.
Cell Culture-HEK-293 cells were maintained in minimum essential Eagle's medium (M2279; Sigma) plus 10% fetal bovine serum and penicillin/streptomycin. Where applicable, G418 was used at 500 g/ml for selection and 150 g/ml for maintenance and hygromycin was used at 250 g/ml for selection and 150 g/ml for maintenance. AT1 A R-stable HEK-293 cells (AT1 A R-293) have been described previously (14) and were cultured in the presence of 100 g/ml zeocin (Invitrogen).
Creation of Tet-inducible HEK-TetOn-AT1 A R Cells-HEK-293 cells were transfected with pTET-On and subjected to G418 (500 g/ml) selection for 3 weeks. Stable clones were selected based on their ability to inducibly express the AT1 A R from transient transfection of pTRE2-hyg-HA-AT1 A R and subsequent treatment with 1 g/ml doxycycline or vehicle for 48 h. A second round of transfection was performed with pTRE2-hyg-HA-AT1 A R, and cells were subjected to hygromycin selection (200 g/ml) for 3 weeks. Positive clones were selected based on low background and ability to induce AT1 A R expression in response to doxycycline. Receptor expression was determined in all experiments using previously described standard protocols for binding assays using fractional occupancy (16). Protein concentrations were determined by the Bradford assay (Pierce). Constitutive AT1 A R expression in these cell lines was very low (10 -150 fmol/mg total protein) but up-regulated in a dose-dependent fashion upon addition of the tetracycline analogue doxycycline to 500 -1500 fmol/mg total protein. Careful titration experiments revealed that in several clonal transfectant cell lines we could reliably select for specific receptor density across a range of ϳ10-fold. Receptor concentrations reported as fmol/mg protein convert to number of receptors/cell as follows: 100 fmol/mg protein equals ϳ90,000 receptor/cell, based on our measurement of 1.5 ng of protein/ HEK-293 cell.
Immunoblotting-To confirm either protein overexpression or silencing, a portion of the cells used for imaging were lysed and run on SDS-PAGE gels according to standard protocols. Western blots were performed using the following antibodies, endogenous ␤-arrestins with A1CT at 1:3000, GRK2 with GRK2-C15 at 1:1000 (S-562; Santa Cruz Biotechnology), and transiently expressed ␤-arrestins with FLAG M2 at 1:1000 (F1804; Sigma).
Chemiluminescent detection was performed with horseradish peroxidase-coupled secondary antibody (Amersham Biosciences) and SuperSignal West Pico reagent (Pierce). Chemiluminescence was quantified by a charge-coupled device camera (Syngene ChemiGenius2); representative images are shown as inverted grayscale.
Reagents-FURA2-AM was purchased from Molecular Probes (F1221) and was used with 15 min of preincubation at 1 g/ml. Valsartan was obtained from Novartis and was used at a concentration of 1 M. Angiotensin II was purchased from Sigma (A9525) and was used at 100 nM concentration in all experiments. Phorbol 12-myristate 13-acetate was purchased from Sigma (P1585) and used at a concentration of 1 M. Doxycycline hyclate was purchased from Sigma (D9891) and used at concentrations ranging from 50 ng/ml to 1 g/ml as indicated. Calphostin C was purchased from Calbiochem and used at 2 M.
Imaging-For all imaging experiments, 24 h prior to assay cells were split and plated to imaging dishes precoated with fibronectin. Cells were washed once and placed in imaging buffer (125 mM NaCl, 5 mM KCl, 1.5 mM MgCl 2 , 1.5 mM CaCl 2 , 10 mM glucose, 0.2% bovine serum albumin, 10 mM HEPES, pH 7.4) and imaged in the dark on a stage heated to 37°C. Images were acquired on a Zeiss Axiovert 200 M microscope (Carl Zeiss MicroImaging, Inc.) with a Roper Micromax cooled charge-coupled device camera (Photometrics) controlled by Slide-Book 4.0 (Intelligent Imaging Innovations). Cyan fluorescent protein (CFP) and fluorescence resonance energy transfer (FRET) images were obtained through a 436/20 excitation filter (20 nm bandpass centered at 436 nm), a 455DCLP (dichroic longpass mirror), and separate emission filters (480/30 for CFP and 535/30 for FRET). Yellow fluorescent protein (YFP) inten-sity was imaged through a 500/20 excitation filter, 515LP dichroic mirror, and 535/30 emission filter. All optical filters were obtained from Chroma Technologies. Excitation and emission filters were switched in filter wheels (Lambda 10 -2; Sutter Instruments). Integration times were varied between 100 and 300 ms to optimize signal and minimize photobleaching.

RESULTS
To assess AT1 A R activation and desensitization kinetics via real-time measurement of DAG, we used the fluorescent biosensor DAGR (diacylglycerol reporter) (8). DAGR consists of the DAG-binding C1 domain from protein kinase C ␤II (amino acids 37-152) fused between the cyan and yellow variants of green fluorescent protein, monomeric CFP and monomeric YFP (Fig. 1A) (8). DAGR exhibits intramolecular FRET, which depends upon the distance and orientation between the two fluorophores (8). In addition, DAGR molecules can undergo intermolecular FRET whenever DAGR concentration is high enough to allow close proximity (Ͻ10 nm). When DAG is generated by the activation of phospholipase C, DAGR translocates from the cytosol to bind DAG in the plasma membrane. This translocation increases the effective concentration of DAGR at the membrane, resulting in increased intermolecular FRET. This mechanism is distinguished from conformational change-induced intramolecular FRET changes (8) by its dependence on DAGR concentration. Whereas intramolecular, conformation-dependent FRET changes are concentration independent, intermolecular FRET is proportional to the amount of DAGR present. Increasing DAGR concentration, measured by the FRET-independent YFP intensity of single transiently transfected cells, causes increasing FRET changes between untreated cells and cells treated with a saturating dose of the DAG analogue phorbol 12-myristate 13-acetate (supplemental Fig.  S1A), confirming the mechanism of action of DAGR.
We then used DAGR to monitor the dynamics of Angiotensin II (AngII) stimulation of the AT1 A R. To avoid concerns of AT1 A R dysregulation caused by expression above physiological levels, we generated clonal HEK-293 cell lines stably transfected with tetracycline-inducible AT1 A R to control receptor expression. In HEK-TetON-AT1 A R cells expressing moderate amounts of AT1 A R (150fmol/mg protein) and transfected with DAGR, time lapse DAGR experiments reveal both the kinetics and subcellular location of DAG signals (Fig. 1B). Prior to agonist addition, DAGR is cytosolic (shown with the FRET-independent YFP image) and FRET is low (blue pseudocolor). Thirty seconds after AngII addition, DAGR translocates from the cytosol to the plasma membrane, where FRET increases (red pseudocolor). Quantification of the FRET ratio time lapse (FRET intensity/monomeric CFP intensity, normalized to base line) in single cells showed that AngII led to a rapid increase in DAG (peaking after ϳ15 s), followed by a slower return to base line (after ϳ5 min) (Fig. 1C). This DAG transient was completely blocked by the angiotensin receptor blocker valsartan.
Because DAGR functions by binding DAG and could potentially buffer DAG concentrations, we addressed concerns that DAGR could perturb normal AT1 A R signaling. We evaluated signal output downstream of DAG by measuring angiotensinstimulated protein kinase C activity with the fluorescent biosensor CKAR (8). 100 nM AngII stimulated transient protein kinase C (PKC) activity, which was entirely dependent on phospholipase C, DAG, and PKC as assessed by a panel of inhibitors (supplemental Fig. S1B). When CKAR is co-expressed with a red fluorescent analogue of DAGR (mRFP1-C1-mRFP1) that translocates after AngII stimulation to a similar extent as DAGR, PKC activity was indistinguishable from CKAR expressed with free mRFP1 (Fig. 1D). This indicates that the amount of DAG bound by DAGR did not significantly affect DAG-dependent signaling and thus validates DAGR as a nondestructive assay of AT1 A R signaling.
The transient nature of the AngII-stimulated DAG signal reflects rapid rates of AT1 A R desensitization as well as DAG clearance by diacylglycerol kinases. Receptor desensitization is mediated by either GRK receptor phosphorylation and ␤-arrestin recruitment (1) or by second messenger-regulated kinase receptor phosphorylation (identified for the AT1 A R as PKC) (17)(18)(19). To test the effect of receptor phosphorylation on DAG, we generated a mutant angiotensin II receptor with all 13 serines and threonines in the carboxyl-terminal tail (subsequent to the seventh transmembrane helix) mutated to alanine, named "13A". These residues are required for agonist-induced AT1 A R phosphorylation, ␤-arrestin recruitment, and receptor internalization (20 -24). We compared the signal kinetics of WT and 13A AT1 A R and found the two display strikingly different patterns of DAG generation ( Fig. 2A). Compared with the transient DAG signal from WT AT1 A R, 13A AT1 A R exhibits a larger and prolonged DAG signal. This difference is consistent with a rapid (Ͻ5 min) phosphorylation-mediated desensitization of the AT1 A R. This result is not unique to HEK-293 cells, as preliminary comparison of WT and 13A in TetOn-U2osteosarcoma cells gave the same result (data not shown). Similar results were found in HEK-293 cells with CYPHR, a FRET sensor for phosphatidylinositol bisphosphate/inositol triphosphate, that like DAGR reflects phospholipase C activity (8) (supplemental Fig. S1C). Consistent with these results, AT1 A R internalization, a phosphorylation-dependent process (22), is rapid and nearly complete for cells expressing wild-type (WT) receptor, whereas mutant 13A AT1 A R completely fails to internalize (Fig. 2B). We also evaluated calcium signals as a more distal measure of receptor function. In contrast to DAG, calcium dynamics do not correlate well with receptor desensitization; WT and 13A receptors gave similar calcium responses (Fig. 2C), consistent with post-receptor mechanisms of calcium regulation (reviewed in Refs. 25,26). We thus conclude that DAGR can measure G q -coupled receptor regulation with high temporal fidelity, which allows us to explore the regulation of receptor desensitization.
Using tetracycline induction of receptor expression, we then determined the effect of receptor density on DAG kinetics and receptor desensitization. As already noted in Fig. 1C, we found that at low receptor expression (Ͻ100 fmol/mg protein, equivalent to 90,000 receptors/cell) AngII-stimulated DAG is transient, returning to base line after 2-3 min (Fig. 3A). This represents a surprisingly rapid and profound desensitization of the AT1 A R. In contrast, at increasingly higher, but still moderate, receptor expression (up to 375 fmol/mg protein), the rapid phase of desensitization is progressively lost, leading to sustained DAG. An expanded time scale (Fig. 3B) shows that the initial rate of DAG accumulation correlates with receptor expression and illustrates how rapidly desensitization occurs (within 30 s) at low receptor density.
We next tested the effect of altering 13A AT1 A R expression on DAG. As expected, this mutant fails to desensitize even at very low density (30 fmol/mg protein), and the primary effect of increasing receptor expression is increased rate of DAG generation and higher maximum DAG signal (Fig. 3C). An expanded time scale (Fig. 3D) shows the effect of receptor expression on rate of accumulation and the contrast with the kinetics of FIGURE 2. DAGR reports AT1 A R desensitization. A, AngII-induced diacylglycerol from the wild-type AT1 A R (WT) rapidly returns to base line, whereas the signal from a nonphosphorylatable, non-desensitizing AT1 A R (13A) remains elevated. Receptor surface expression was comparable for WT and 13A (150 fmol/mg protein). B, 100 nM AngII induces rapid internalization of wild-type AT1 A R (WT) but fails to internalize a nonphosphorylatable AT1 A R mutant (13A). C, calcium dynamics display a more complicated relationship to desensitization, with only an elevated plateau of intracellular calcium to distinguish WT from 13A. All data are means with standard errors from three to five experiments.
WT-AT1 A R-stimulated DAG transients. The initial rate of DAG accumulation (10 -20 s after agonist addition) is highly correlated with receptor expression (r 2 ϭ 0.98, supplemental Fig. S1D) and is similar for both WT and 13A AT1 A R. This indicates that at these low and moderate receptor concentrations, receptor expression is the limiting factor in a DAG response. Furthermore, the similarity of initial DAG accumulation for WT and 13A AT1 A R indicates that the 13A mutant elicits normal G protein coupling to phospholipase C.
Because AT1 A R desensitization is dramatically impaired even at moderate expression levels, we investigated the mechanisms of desensitization for this receptor. Because ␤-arrestins desensitize signaling by binding receptors and sterically interdicting G protein coupling, the desensitization capacity of ␤-arrestins could become stoichiometrically limited by AT1 A R expression levels above the concentration of endogenous ␤-ar-restins. To test this hypothesis, we used a stably transfected HEK-293 cell line with dramatic overexpression of the AT1 A R (1200 fmol/mg protein). Cells transfected with a vector plasmid display sustained DAG after AngII stimulation, typical of lost desensitization (Fig. 4A). However, we found that overexpression of either ␤-arrestin1 or 2 could fully restore normal desensitization kinetics to this cell line. This led us to conclude that 1) ␤-arrestins are the limiting component to AT1 A R desensitization at high receptor expression, and 2) both ␤-arrestin1 and ␤-arrestin2 are capable of desensitizing the AT1 A R. Because other work has shown that overexpressed, angiotensin-stimulated AT1 A R co-immunoprecipitates both endogenous ␤-arrestins (12), we have further concluded that the two ␤-arrestins are redundant for AT1 A R desensitization. To confirm the importance of receptor phosphorylation for ␤-arrestin-mediated AT1 A R desensitization, we repeated these experiments with 13A AT1 A R. As expected, exogenous ␤-arrestin expression did not restore desensitization to the nonphosphorylatable 13A AT1 A R (Fig.  4B). This result solidifies the role of phosphorylation as a necessary intermediate between receptors and ␤-arrestins in the desensitization process.
In contrast to ␤-arrestins, overexpression of GRK2, the kinase responsible for most AT1 A R phosphorylation (12), suppressed detectable DAG accumulation altogether (Fig. 4A). However, similar results for GRK2 were obtained using the 13A mutant receptor (Fig. 4B) and with a kinase-inactive K220M GRK2 mutant (27) for the WT AT1 A R (data not shown). These results suggest this effect is not related to GRK2 kinase activity and most likely corresponds to G protein sequestration ascribed to an RGS (regulator of G protein signaling)-like domain of GRK2 (28,29). Interestingly, overexpression of GRK3, the closest homologue of GRK2, did not suppress the DAG signal (data not shown). We confirmed G␣ q as the locus of the profound signal inhibition caused by GRK2 overexpression: G␣ q overexpression, presumably saturating the RGS activity of GRK, partially rescued the signal suppression (supplemental Fig. S2). Despite this, the finding that 13A AT1 A R completely fails to desensitize in the presence of endogenous levels of GRK2 ( Fig. 2A) suggests that GRK2 kinase activity, and not the RGS function, is the primary regulator of acute AT1 A R desensitization.
Given these results and the concern that results derived from overexpressed GRKs and ␤-arrestins may not reflect physiolog-ically relevant functions, we evaluated the effects of siRNAmediated silencing of the endogenous ␤-arrestin/GRK system. Although individual silencing of neither ␤-arrestin1 nor ␤-ar-restin2 significantly altered WT AT1 A R desensitization (data not shown), simultaneous silencing of both ␤-arrestins significantly abrogated desensitization in comparison to a non-functional control siRNA (Fig. 5A). Average ␤-arrestin silencing was 80%; we found it impossible to completely eliminate ␤-arrestin expression (Fig. 5B) but believe that our results can be qualitatively extrapolated to indicate that a large portion of AT1 A R desensitization in our HEK-293 cells is ␤-arrestin dependent. Importantly, AT1 A R expression was unchanged by ␤-arrestin silencing. In contrast, GRK2 silencing up-regulated surface AT1 A R expression 2-fold. To compensate for this, we treated control siRNA-transfected cells with a dose of doxycycline (0.05 g/ml) empirically determined to result in AT1 A R expression equivalent to that found after GRK2 silencing. In experiments with equivalent AT1 A R expression, desensitization is dramatically compromised by the loss of GRK2 expression (Fig. 5C). Relative to ␤-arrestin silencing, GRK2 silencing was very effective (Fig. 5D), averaging 90 -95%. In contrast to the dramatic effect of GRK2 silencing on desensitization, similarly effective silencing of GRK3, GRK5, or GRK6 had no significant effect on AT1 A R desensitization (supplemental Fig. S3). Thus, we conclude that GRK2 is the primary regulator of AT1 A R desensitization in our HEK-293 cells and that GRK2 functions predominantly through phosphorylation of the carboxyl-terminal tail of AT1 A R, inducing ␤-arrestin translocation and inhibition of receptor-G protein coupling.

DISCUSSION
This study analyzed real-time AT1 A R signal kinetics as a model for heptahelical receptor desensitization, using the fluorescent biosensor DAGR. DAGR is a non-disruptive reporter and does not alter receptor function as measured by the effect of a DAGR analogue on downstream DAG-dependent PKC activity (Fig. 1D). This contrasts with phosphatidylinositol hydrolysis, the standard second messenger assay for G q -coupled receptors, which requires lithium to block analyte degradation (3,30), altering downstream and feedback signaling. Other assays for G q -coupled signals include high pressure liquid chromatography, used to measure DAG directly (31), and a DAG assay using recombinant DAG kinase for the generation of 32 P-labeled phosphatidic acid from crude membranes (32). However, none of these assays are amenable to real-time analysis.
In contrast, sufficient dynamic sensitivity is readily obtained with calcium binding fluorophores such as FURA-2-AM that are commonly used to detect receptor activation in living cells. However, as shown in Fig. 2C, these indicators do not effectively discern differences in receptor desensitization, Additionally, receptor desensitization has been inferred from ␤-arrestin recruitment assays (10), but these require overexpressed receptor and ␤-arrestin and thus alter receptor function.
Future studies using DAGR to measure receptor signal kinetics might take advantage of G␣ 15 subunits or chimeric G␣ subunits that are commonly overexpressed to couple any 7TMR to phospholipase C and DAG/Ca 2ϩ generation (33). This may allow a more comprehensive evaluation of the kinetics and GRK specificity for 7TMR family regulation.
Despite the utility of DAGR as a non-disruptive sensor, the DAGR signal is a complex readout, dependent on the dynamic balance of DAG generation and clearance. We have isolated the acute phase of AT1 A R desensitization mediated by GRK2 phosphorylation of the receptor tail and ␤-arrestin recruitment, but it remains possible that dynamic regulation of DAG clearance by diacylglycerol kinase activity may play a role in shaping DAG signals. This should prove to be a fruitful area of further study.
Perhaps the most provocative findings revealed by this study relate to the relationship between the amount of receptor expression and desensitization. Typically, studies of the molecular regulation of receptors like the AT1 A R have been performed in transfected cellular systems with receptor densities of pmol/mg protein.
Here we have shown that these levels saturate the ability of cells to desensitize second messenger signaling. This reflects stoichiometric ␤-arrestin function via a mechanism of steric blockade; in contrast, GRKs function enzymatically and can compensate for receptor overexpression.
In contrast to the very high receptor densities frequently used in molecular studies, tissues and cell lines typically express Ͻ100 fmol/mg protein (equal to ϳ90,000 receptors/cell in this study) of endogenous AT1 A R (34 -39). As shown by DAGR transients in HEK-TetOn-AT1 A R cells (Fig. 3A), these expression levels are well within the range of the capacity of cells to desensitize receptor signaling. Consequently, we believe that the tetracycline-inducible cell system affords a biologically relevant level of receptor expression.
Regardless of receptor density, overexpression of GRK2 completely inhibited any DAG accumulation in response to AngII. This is unrelated to AT1 A R desensitization or GRK2 kinase function because an identical result was obtained with GRK2 K220R, a kinase-inactive mutant, and also by transfecting the wild-type GRK2 with the 13a AT1 A R. Instead, in the context of GRK2 overexpression, DAG signals are blocked through G␣ q sequestration by the RGS homology domain of GRK2 (28,29,40). Because AT1 A R desensitization requires both intact AT1 A R phosphoacceptor sites and GRK2 expression and the RGS homology function is independent of these phosphoacceptor sites, we conclude that the primary mechanism of AT1 A R desensitization is GRK2 phosphorylation of the AT1 A R carboxyl-terminal tail. However, because receptor, ␤-arrestin, and GRK isoform concentrations vary across cell type, we expect that other tissues and cell lines can use different GRKs to mediate AT1 A R desensitization, as recently shown for the ␤ 2 -adrenergic receptor (41). Additionally, in cells with sufficiently high concentrations of GRK2, the RGS homology domain function may be an important component of AT1 A R desensitization.
The role of GRK2 in AT1 A R phosphorylation and desensitization is consistent with data showing that GRK2, of all the ubiquitously expressed GRKs, is the single largest contributor to ␤-arrestin recruitment to the AT1 A R in our HEK-293 cells. 5 FIGURE 5. ␤-arrestin or GRK silencing by siRNA impairs desensitization measured by DAGR kinetics. A, GRK2 silencing by siRNA impairs desensitization, leading to increased and sustained DAG. Receptor levels were equalized (160 fmol/mg protein) with doxycycline (48 h 0.05 g/ml for control siRNA (CTL), none for GRK2 siRNA) to compensate for receptor up-regulation upon GRK2 silencing. B, a representative immunoblot of equal protein amounts from whole cell lysate shows that GRK2 silencing is Ͼ90%. ␤-actin immunoblots are shown as protein loading controls. C, ␤-arrestin1/2 silencing by siRNA impairs desensitization, leading to increased and sustained DAG. ␤-arrestin silencing had no effect on receptor expression (170 fmol/mg protein). D, a representative immunoblot of equal protein from whole cell lysate shows that ␤-arrestin1/2 silencing is Ͼ80%. All data are means with standard errors for three experiments.
However, this contrasts with findings in these same cells that GRK6 is the largest contributor to ␤-arrestin recruitment to the ␤ 2 AR receptor (41). This demonstrates a receptor specificity of GRK function. It is likely that this specificity varies among cell and tissue types, illustrated by the fact that ␤ 2 AR regulation by GRK6 is absent in U2-osteosarcoma cells, which instead use GRK2 and 3 to regulate ␤ 2 AR function (41). This complexity can be regulated further by "biased agonists" such as the AngII analogue 1 Sar, 4 Ile, 4 Ile-Ang-II, which recruits ␤-arrestins in the absence of G protein stimulation and is regulated by GRK6 instead of GRK2. 5 This may provide opportunities for novel, more subtle pharmacological manipulation of receptor systems beyond the classic spectrum of agonists and antagonists of G protein-mediated signaling. Indeed, as shown here, the effect of altering GRK activity on receptor signaling can be quite profound.
Of course, there are other means of altering receptor signaling. The receptor, GRK, and ␤-arrestin system comprise a mechanism for homologous desensitization that contrasts with heterologous desensitization mediated by receptor phosphorylation by second messenger kinases. For G␣ q -coupled receptors like the AT1 A R, PKC has been shown to mediate receptor phosphorylation and heterologous desensitization as measured by increased signaling after suppression of PKC activity (17)(18)(19). However, we did not evaluate heterologous, PKC-mediated desensitization with the DAGR reporter because of the complexities of the PKC feedback network. In addition to phosphorylating receptor, PKC has been found to phosphorylate and inhibit phospholipase C-␤ (42). Hence, PKC inhibition simultaneously increases positive signaling (by blocking phospholipase C-␤ phosphorylation) and decreases negative feedback (by blocking heterologous AT1 A R desensitization). Indeed, treatment with Ro-31-8425 delays the return of AngII-stimulated DAG to base line, but we have not determined the relative contributions of PKC to phospholipase C regulation and AT1 A R heterologous desensitization. The methods described here should provide fruitful exploration of these issues. However, this work suggests that independent of other signaling feedback mechanisms, GRK2 is required for AT1 A R desensitization in HEK-293 cells.
More generally, this work provides the technical and conceptual framework to evaluate receptor desensitization in a new light. Although the classical definition of desensitization involves measuring responses to repeated stimulation, we believe that kinetic measurements of receptor signaling in the presence of constant agonist provide a more accurate and physiologically relevant description of receptor regulation. Our data highlight the critical importance of receptor expression, GRK phosphorylation, and ␤-arrestin binding in determining the rate of receptor desensitization. The extrapolation of these methods to other cell lines, tissues, and receptors will provide a more comprehensive understanding of heptahelical receptor regulation and may provide more precise methods of evaluating drug candidates that influence the amplitude and kinetics of receptor signals.