Roles of Phosphorylation-dependent and -independent Mechanisms in the Regulation of M1 Muscarinic Acetylcholine Receptors by G Protein-coupled Receptor Kinase 2 in Hippocampal Neurons*

When co-expressed with the inositol 1,4,5-trisphosphate biosensor eGFP-PHPLCδ, G protein-coupled receptor kinase 2 (GRK2) can suppress M1 muscarinic acetylcholine (mACh) receptor-mediated phospholipase C signaling in hippocampal neurons through a phosphorylation-independent mechanism, most likely involving the direct binding of the RGS homology domain of GRK2 to Gαq/11. To define the importance of this mechanism in comparison with classical, phosphorylation-dependent receptor regulation by GRKs, we have examined M1 mACh receptor signaling in hippocampal neurons following depletion of GRK2 and also in the presence of non-Gαq/11-binding GRK2 mutants. Depletion of neuronal GRK2 using an antisense strategy almost completely inhibited M1 mACh receptor desensitization without enhancing acute agonist-stimulated phospholipase C activity. By stimulating neurons with a submaximal agonist concentration before (R1) and after (R2) a period of exposure to a maximal agonist concentration, an index (R2/R1) of agonist-induced desensitization of signaling could be obtained. Co-transfection of neurons with either a non-Gαq/11-binding (D110A) GRK2 mutant or the catalytically inactive D110A,K220RGRK2 did not suppress acute M1 mACh receptor-stimulated inositol 1,4,5-trisphosphate production. However, using the desensitization (R2/R1) protocol, it could be shown that expression of D110AGRK2 enhanced, whereas D110A,K220RGRK2 inhibited, agonist-induced M1 mACh receptor desensitization. In Chinese hamster ovary cells, the loss of Gαq/11 binding did not affect the ability of the D110AGRK2 mutant to phosphorylate M1 mACh receptors, whereas expression of D110A,K220RGRK2 had no effect on receptor phosphorylation. These data indicate that in hippocampal neurons endogenous GRK2 is a key regulator of M1 mACh receptor signaling and that the regulatory process involves both phosphorylation-dependent and -independent mechanisms.

Despite many years of investigation we still have an incomplete understanding of how cholinergic inputs modulate neuronal function in the hippocampus. Nevertheless, it has been clearly shown that cholinergic innervation of the hippocampus is widespread (1,2) and that cholinergic deficits (caused by lesioning, pharmacological blockade, or gene knock-out) produce an array of disorders in learning and memory (3)(4)(5).
Transgenic approaches have helped to define the key roles of M 1 muscarinic acetylcholine (mACh) 1 receptors in cholinergic regulation of hippocampal function (5)(6)(7)(8), and gaining a better understanding of the physiological and pathophysiological regulation of this mACh receptor subtype in hippocampal neurons remains a key objective.
Desensitization of G protein-coupled receptor (GPCR) signaling following continuous or repeated agonist challenge is believed to be initiated by the phosphorylation of specific serine and/or threonine residues within the third intracellular loop and/or C-terminal tail of the receptor (9,10) and is a crucial mechanism for reducing (or "switching") signaling (11). Receptor desensitization is usually mediated by either second messenger-activated kinases (e.g. protein kinase C (PKC)) or GPCR kinases (GRKs). Receptor phosphorylation leads to the recruitment of arrestin proteins, which bind and physically prevent interaction between the GPCR and G protein (10,12). Receptor-arrestin complexes are also involved in the initiation of receptor internalization and may act as signaling scaffolds to assemble transduction pathways (12,13).
Recently it has been shown that at least two members of the GRK family, GRK2 and GRK3, are able to suppress G␣ q/11coupled receptor/phospholipase C (PLC) signaling even in the absence of kinase activity (14 -16). GRK2-mediated, phosphorylation-independent receptor regulation has now been demonstrated not only in recombinant systems but also for endogenous receptors in cell lines (17) and in primary, cultured hippocampal neurons (18). Phosphorylation-independent receptor regulation by GRK2 is mediated through a specific interaction of GTP-loaded G␣ q/11 with the RGS homology (RH) domain situated at the N terminus of GRK2 (14,15,19). Mutational and crystallographic analyses have identified specific amino acid residues within the RH domain that mediate G␣ q binding that map almost exclusively to the ␣5 helix of the RH domain of GRK2 (19 -21).
We have previously used confocal imaging to detect the translocation of eGFP-PH PLC␦ as an IP 3 biosensor (18,22) to investigate M 1 mACh receptor regulation of PLC activity in hippocampal neurons. Using this approach, we showed that GRK2 inhibits M 1 mACh receptor signaling in hippocampal neurons via a phosphorylation-independent mechanism (18) 1 The abbreviations used are: mACh, muscarinic acetylcholine; PLC, phosphoinositide-specific phospholipase C; GPCR, G protein-coupled receptor; PKC, protein kinase C; CHO, Chinese hamster ovary; MCh, methacholine; IP 3 , inositol 1,4,5-trisphosphate; GRK, G protein-coupled receptor kinase; RH, RGS homology; eGFP, enhanced green fluorescent protein; HA, hemagglutinin; PDBu, phorbol 12,13-dibutyrate; ERK, extracellular signal-regulated kinase; MEK, mitogen-activated protein kinase/ERK kinase; JNK, c-Jun N-terminal kinase; AM, acetoxymethyl ester; BAPTA, 1,2-bis(O-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid; BAPTA-AM, 1,2-bis(O-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid tetra(acetoxymethyl) ester. most likely involving the direct binding of the RH domain of GRK2 to G␣ q/11 -GTP. Although these data highlight a novel mechanism by which GRK2 can regulate M 1 mACh receptor signaling in neurons, they do not establish whether GRK2mediated receptor phosphorylation is also an important mechanism in M 1 mACh receptor desensitization. Here we have used a number of approaches to delineate the mechanism by which M 1 mACh receptor signaling is regulated by GRK2 in hippocampal neurons. Our data indicate that endogenous GRK2 is a key regulator of M 1 mACh receptor signaling utilizing both phosphorylation-dependent and -independent mechanisms to regulate receptor activity.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-Hippocampal neurons from 1-dayold Lister-hooded rat pups were isolated as described previously (22,23). Briefly, isolated hippocampi were dissociated with Pronase E (0.5 mg ml Ϫ1 ) and thermolysin (0.5 mg ml Ϫ1 ) in a HEPES-buffered salt solution (130 mM NaCl, 10 mM HEPES, 5.4 mM KCl, 1.0 mM MgSO 4 , 25 mM glucose, and 1.8 mM CaCl 2 , pH 7.2) for 30 min. Tissue fragments were further dissociated by trituration in HEPES-buffered salt solution containing DNase I (40 g ml Ϫ1 ). Following centrifugation and further trituration, the cells were plated onto poly-D-lysine (50 g ml Ϫ1 )-treated 25-mm glass coverslips. For the first 72 h, the cells were cultured in Neurobasal medium (Invitrogen) supplemented with B27 and 10% fetal calf serum. Cytosine arabinoside (5 M) was added after 24 h, and after 72 h the cells were transferred to serum-free medium. The cultured neurons were transfected on day 5 with a 3:1 ratio of either vectorcontrol or GRK constructs to eGFP-PH PLC␦ , respectively, using the Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's instructions. Typically, neuronal transfection rates of Յ5% were achieved using this technique. HEK293 and CHO-K1 cells were grown in ␣-minimal essential medium, supplemented with 10% fetal calf serum, penicillin (100 units ml Ϫ1 ), streptomycin (100 g ml Ϫ1 ), and amphotericin B (2.5 g ml Ϫ1 ) (Invitrogen). All of the cells were maintained at 37°C, under 5% CO 2 in humidified conditions. Measurement of PLC Activity in Neurons and Assessment of M 1 mACh Receptor Desensitization-PLC activity was assessed using the agonist-stimulated translocation of eGFP-tagged pleckstrin homology domain of PLC␦1 (eGFP-PH PLC␦ ) and was visualized using an Olympus Optical FV500 scanning laser confocal IX70 inverted microscope as described previously (18,22). All of the experiments were undertaken in the presence of tetrodotoxin (500 nM) to block action potential-dependent synaptic activity. M 1 mACh receptor desensitization was assessed as described previously (18). Briefly, the neurons were challenged with an approximate EC 50 concentration (10 M, R1) of the mACh agonist methacholine (MCh) for 30 s, followed by a 5-min washout to allow recovery of PIP 2 , intracellular Ca 2ϩ stores, and eGFP-PH PLC␦ fluorescence to basal levels. Next a maximal concentration of MCh (100 M) was applied for 1 min to induce receptor desensitization. Finally, following another 5-min washout, neurons were rechallenged with a second pulse of MCh (10 M, R2) for a further 30 s. Receptor desensitization was determined as the reduction in peak IP 3 (16), and the receptors were immunoprecipitated using a specific rat monoclonal anti-HA antibody (3F10; Roche Applied Science) and electrophoretically resolved as described previously (16). The autoradiograms were documented and analyzed using the GeneGenius system and software (Syngene, Cambridge, UK). To examine the effects of GRK2 manipulation on M 1 mACh receptor phosphorylation, the cells were co-transfected with HA-tagged M 1 mACh receptor and either wild-type GRK2, D110A GRK2, K220R GRK2, or D110A,K220R GRK2. Receptor expression was equalized following [ 3 H]NMS binding prior to gel loading. For co-transfection experiments wild-type GRK2, D110A GRK2, K220R GRK2, or D110A,K220R GRK2 expression was determined by Western blotting using a rabbit polyclonal anti-GRK2 antibody (Santa Cruz, Santa Cruz, CA).
Assessment of Antisense Suppression of Endogenous GRK2 Expression-The ability of a specific antisense GRK2 construct (almost the full length of rat GRK2 cloned into pcDNA3 in an antisense direction) (23) to suppress endogenous GRK2 expression was initially determined after transfection of HEK293 cells with 1, 2, or 3 g of either pcDNA3 (control) or antisense GRK2 using Lipofectamine 2000, according to the manufacturer's instructions. After 72 h the cells were lysed, and GRK2 expression was determined by Western blotting using a rabbit polyclonal anti-GRK2 antibody (Santa Cruz). To examine the specificity of the GRK2 antisense construct, the expression levels of GRK3 and GRK6 were also examined in the same samples by Western blotting using rabbit polyclonal anti-GRK3 and anti-GRK6 antibodies (Santa Cruz).
To assess to what extent antisense treatment suppresses endogenous expression of GRK2 in neurons, hippocampal neurons were transfected with a 1:3 ratio of either eGFP and pcDNA3 (control) or eGFP and antisense GRK2. After 72 h, the neurons were fixed for 10 min using 4% paraformaldehyde prior to permeabilization with phosphate-buffered saline containing Triton X-100 (0.2%, for 5 min). Nonspecific antibody binding was blocked using phosphate-buffered saline containing Triton X-100-containing goat serum (10%, 30 min). The neurons were incubated overnight at 4°C with a polyclonal anti-rabbit GRK2 antibody (Santa Cruz). After washing, the neurons were incubated with an Alexa Fluor 546 anti-rabbit IgG (Molecular Probes; 1:400, for 40 min) prior to washing and mounting. GRK2 staining was visualized using an Olympus Optical FV500 scanning laser confocal IX70 inverted microscope, using excitation at 488 nm for eGFP and 543 nm for Alexa Fluor 546. GRK2 expression was determined as the mean fluorescence intensity of all eGFP-expressing neurons.
Creation of Non-G␣ q -binding Mutants-The aspartate residue at position 110 of GRK2 was mutated to an alanine using QuikChange (Stratagene) following the manufacturer's instructions. Briefly, the following primers were used to create D110A GRK2 and D110A, K220R GRK2 mutants (forward primer, GTC TGC AGC CGA GAG ATC TTC GCG ACC TAC ATC ATG AAG GAG; reverse primer, CTC CTT CAT GAT GTA GGT CGC GAA GAT CTC TCG GCT GCA GAC), creating a new NruI endonuclease site, which was used to identify positive mutants. The presence of the correct mutation was determined following DNA sequencing. Wild-type GRK2, D110A GRK2, K220R GRK2, and D110A,K220R GRK2, were Myc-tagged following cloning into pcDNA3.1/Myc-His (Invitrogen) between the HindIII and EcoRI sites.
Assessment of GRK2/G␣ q Binding-The ability of wild-type GRK2, D110A GRK2, K220R GRK2, and D110A,K220R GRK2 to bind to G␣ q was determined as follows. HEK293 cells were transfected with either 2 g of D110A GRK2, K220R GRK2, and D110A,K220R GRK2 or 1 g of the constitutively active G␣ q mutant Q209L G␣ q using GeneJuice according to the manufacturer's instructions. After 48 h the cells were lysed on ice for 10 min with the following buffer: 20 mM Tris, pH 7.4, 100 mM NaCl, 3 mM MgCl 2 , 0.5 mM EDTA, 0.05% (v/v) Igepal, 0.2 mg/ml benzamidine, 0.1 mg/ml leupeptin, 0.5 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol. Insoluble material was pelleted by centrifugation (14,000 ϫ g, 10 min, 4°C). Supernatants from cells transfected with Q209L G␣ q were mixed with supernatants from cells expressing wild-type GRK2, D110A GRK2, K220R GRK2, or D110A, K220R GRK2. Following rolling for 1 h at 4°C, the specific mouse monoclonal anti-Myc antibody 9E10 was added for 1 h at 4°C to immunoprecipitate Myc-tagged GRK2. Next 150 l of protein A-Sepharose was added, and the samples were placed on a roller for 30 min at 4°C. The samples were then washed three times with lysis buffer minus Igepal and EDTA, prior to resuspension in 2ϫ SDS-PAGE loading buffer. The samples were heated for 3 min at 85°C before separation by SDS-PAGE (10% acrylamide gel). Each sample (25 l) was loaded onto a gel for G␣ q detection, and a further 10 l was loaded onto another gel for detection of Myc expression. Separated protein was transferred to nitrocellulose, and G␣ q expression was detected using a rabbit polyclonal anti-G␣ q (1:2500 dilution); Myc expression was determined using a rabbit polyclonal anti-Myc antibody (1:1000 dilution; New England Biolabs). Protein expression was determined by the addition of ECL reagent (Amersham Biosciences) according to the manufacturer's instructions and exposure to Hyperfilm (Amersham Biosciences).
Measurement of Single Cell Ca 2ϩ Concentration-Following loading with Fluo-3 AM (4 M, 1 h), the neurons were excited at 488 nm, using an Olympus Optical FV500 scanning laser confocal IX70 inverted microscope. The cells were incubated at 37°C using a temperature controller and microincubator (PDMI-2 and TC202A; Burleigh, UK) and perfused at 5 ml/min with Krebs buffer (119 mM NaCl, 4.7 mM KCl, 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 4.2 mM NaHCO 3 , 10 mM HEPES, 11.7 mM glucose, and 1.3 mM CaCl 2 , pH 7.4). The images were captured using an oil immersion ϫ100 objective. Cytosolic Ca 2ϩ levels were measured as the relative change in fluorescence detected in an area of interest as described previously (24). Drugs were applied via the perfusion line for the times stated under "Results." Data Analysis-Concentration-response curves were fitted using nonlinear regression analysis using Prism, version 3.0 (GraphPad Software Inc., San Diego, CA). All of the data were analyzed using one or two-way analysis of variance, followed by Bonferroni's post-hoc test (Excel 5.0; Microsoft, Redmond, WA). Significance was accepted when p Ͻ 0.05.

Suppression of GRK2 Expression Using an Antisense GRK2
Construct-The GRK2 antisense construct used in this study has previously been shown to cause an approximate 75% suppression of endogenous GRK2 expression when stably expressed in NG108 -15 cells (25). To examine the effectiveness of the GRK2 antisense in a transient system, HEK293 cells were transfected with 1, 2, or 3 g of GRK2 antisense (or the pcDNA3 vector) for 72 h. Western blot analysis showed that 1 g of the GRK2 antisense was sufficient to cause an 80% suppression of endogenous GRK2 expression in HEK293 cells 72 h post-transfection (Fig. 1, A and B). The GRK2 antisense construct produced specific depletion of endogenous GRK2, because GRK3 and GRK6 expression levels were unaffected (Fig.  1A). Note that the GRK3 antibody used detects both GRK2 (upper band) and GRK3 (lower band) and that antisense GRK2 selectively depletes GRK2 without affecting GRK3 immunoreactivity (Fig. 1A). To determine whether antisense treatment was able to deplete endogenous GRK2 expression in neurons, the cultures were co-transfected with eGFP and pcDNA3 (control) or eGFP and antisense GRK2 for 72 h. Analysis of all neurons co-expressing eGFP indicated that the mean intensity of GRK2 immunoreactivity was reduced by Ն60% following antisense GRK2 treatment ( Fig. 1, C and D), whereas GRK3 immunoreactivity was not affected (data not shown).

Effects of GRK2 Suppression on M 1 mACh Receptor Signaling in Hippocampal
Neurons-To examine the effects of GRK2 on M 1 mACh receptor signaling endogenous GRK2 was specifically depleted following transfection with antisense GRK2 for 72 h. Expression of antisense GRK2 did not affect IP 3 production stimulated by an acute agonist addition when compared with pcDNA3-transfected, control neurons ( Fig. 2A). In agreement with our previous findings using the agonist pretreatment protocol, which compares responses to submaximal agonist concentrations before (R1) and after (R2) a maximal agonist application (100 M, for 60 s), R2 was ϳ40% less than R1 following desensitization with 100 M MCh in the presence of pcDNA3 (Fig. 2, B and D). However, in neurons expressing the antisense GRK2, R2 was virtually identical to R1, indicating that down-regulation of the endogenous GRK2 protein and GRK expression determined by immunoblotting with rabbit polyclonal antibodies recognizing GRK2, GRK3, or GRK6. Note that anti-GRK3 also detects GRK2, and it is the latter protein that is down-regulated in lane 2. The blots shown are representative of experiments performed on at least three separate occasions. B, densitometric analysis of GRK2 expression in HEK293 cells transfected with either pcDNA3 or antisense GRK2 constructs. The data are shown as the means Ϯ S.E. for 5 separate experiments. The antisense GRK2 construct significantly suppressed endogenous GRK2 expression (**, p Ͻ 0.01) relative to pcDNA3-transfected HEK293 cells. Hippocampal neuronal cultures were transfected with a 1:3 ratio of eGFP and pcDNA3 or eGFP and antisense GRK2 for 72 h, prior to fixation and immunocytochemical analysis of GRK2 expression. C, representative images showing the effects of either pcDNA3 or antisense GRK2 on endogenous GRK2 expression. In both cases, the image on the left shows eGFP-transfected neurons, and that on the right shows GRK2 immunostaining. D, cumulative data showing mean intensity of GRK2 immunostaining for all neurons expressing eGFP and co-expressing either pcDNA3 or antisense GRK2. The data are represented as the mean intensities Ϯ S.E. for 24 and 21 neurons for pcDNA3 and antisense GRK2 treatments, respectively. Antisense GRK2 treatment significantly suppressed endogenous GRK2 expression (**, p Ͻ 0.01) compared with pcDNA3-transfected neurons.
largely prevents M 1 mACh receptor desensitization (Fig. 2, C and D). These data strongly suggest that GRK2 is a key endogenous mediator of M 1 mACh receptor desensitization in hippocampal neurons.
A C-terminal Fragment of GRK2 (GRK2-ct) Does Not Alter M 1 mACh Receptor Signaling-Agonist activation of receptors leads to a translocation of GRK2 to the plasma membrane, a process that requires binding of free G␤␥ subunits to the Cterminal domain (Gly-495 to Leu-689) of bovine GRK2 (26,27). Because G␤␥ subunits are able to directly stimulate PLC signaling (27,28), overexpression of GRK2 may lead to inhibition of PLC signaling through sequestration of free G␤␥ subunits. To assess whether inhibition of PLC signaling seen with the overexpression of GRK2 or K220R GRK2 was due to sequestration of free G␤␥ subunits, the C-terminal 194 amino acids of GRK2 (GRK2-ct) were co-expressed with eGFP-PH PLC␦ . Overexpression of GRK2-ct, confirmed by Western blotting (data not shown), did not affect the acute IP 3 response to single concentrations of MCh (3, 10 or 100 M, for 30 s; data not shown), and had no effect on the R2/R1 ratio assessed using the M 1 mACh receptor desensitization protocol (data not shown).
Does M 1 mACh Receptor Phosphorylation by GRK2 Mediate Receptor Desensitization?-Suppression of endogenous GRK2 expression reverses M 1 mACh receptor desensitization in hippocampal neurons; however, it is unclear whether this is due to inhibition of PLC signaling via binding of G␣ q/11 and/or GRK2mediated receptor phosphorylation. To determine whether GRK2-mediated receptor phosphorylation is required for M 1 mACh receptor desensitization, we introduced a single point mutation D110A to create both GRK2 and K220R GRK2 mutants, which are incapable of binding G␣ q/11 (19). When transiently transfected, expression levels of the GRK2, D110A GRK2, K220R GRK2, and D110A,K220R GRK2 constructs were similar in HEK293 cells (data not shown). To assess GRK2/G␣ q/11 binding, Myc-tagged GRK2, D110A GRK2, K220R GRK2, and D110A,K220R GRK2 proteins were mixed with constitutively active Q209L G␣ q . Immunoprecipitation of Myc-tagged the GRK2 constructs followed by immunoblotting for G␣ q indicated a strong binding of G␣ q to wild-type GRK2 and K220R GRK2 (Fig. 3A). However, introduction of the D110A mutation prevented Q209L G␣ q binding to D110A GRK2 or D110A,K220R GRK2 (Fig. 3A). Initial experiments indicated that the acute agonist-stimulated IP 3 response was unaffected when D110A GRK2 or D110A,K220R GRK2 proteins were expressed, but in common with our previous findings (18), introduction of K220R GRK2 caused a marked suppression of this response in hippocampal neurons (Fig. 3B). These data suggest that without an intact RH domain, GRK2 cannot inhibit M 1 mACh receptor signaling through binding to GTP-bound activated G␣ q/11 . As the acute agonist-stimulated IP 3 response was normal in D110A GRK2-or D110A,K220R GRK2-expressing neurons, we could now assess whether GRK2-mediated phosphorylation is necessary for M 1 mACh receptor desensitization. Assessing the R2/R1 ratio using the M 1 mACh receptor desensitization protocol, it could be shown that expression of D110A GRK2 significantly enhances M 1 mACh receptor desensitization, whereas expression of the catalytically inactive D110A,K220R GRK2 protein significantly diminishes M 1 mACh receptor desensitization (Fig. 3C). These data indicate that GRK2-mediated phosphorylation is required for the agonistdependent attenuation of the IP 3 signal mediated by the M 1 mACh receptor in hippocampal neurons.
Agonist-stimulated Phosphorylation of the M 1 mACh Receptor-Definitive assessment of whether inhibition of the G␣ q/11 -GRK2 binding interaction alters agonist-stimulated M 1 mACh receptor phosphorylation in hippocampal neurons should ideally be made in this neuronal preparation. However, because of the lack of a suitable antibody capable of immunoprecipitating the endogenous rat M 1 mACh receptor, we have expressed HA-tagged M 1 mACh receptors in CHO cells to provide an insight into this key question. Preliminary studies indicated that agonist-stimulated M 1 mACh receptor phosphorylation peaks at approximately 10 min, and therefore this time was used for subsequent experiments (Fig. 4A). The rate of phosphorylation of the rat M 1 mACh receptor was comparable with that observed previously for the human M 1 mACh receptor stably expressed in CHO cells (29). MCh-stimulated M 1 mACh receptor phosphorylation was increased approximately 2-fold in the presence of wild-type GRK2 or D110A GRK2 (Fig. 4, B and E) compared with receptor phosphorylation observed in vector (pcDNA3)-transfected cells. Expression of kinase-dead K220R GRK2, or D110A,K220R GRK2 mutants had no effect upon MCh-stimulated M 1 mACh receptor phosphorylation (Fig. 4, C  and E). These data confirm the lack of catalytic activity in the K220R mutants but also suggest that, at least in CHO cells, endogenous GRK2 may not be responsible for agonist-stimulated M 1 mACh receptor phosphorylation.
Effects of Expression of D110A -GRK2 and D110A,K220R GRK2 on M 3 mACh Receptor Signaling in SH-SY5Y Cells-We have previously shown that GRK2 can inhibit M 3 mACh receptor signaling in SH-SY5Y cells in a phosphorylation-independent manner (17). K220R GRK2 does not inhibit agonist-stimulated M 3 mACh receptor phosphorylation, indicating that endogenous GRK2 does not phosphorylate the endogenous M 3 mACh receptor in SH-SY5Y cells (17). Transient transfection of SH- to bind to a constitutively active Q209L G␣ q protein was assessed in HEK293 cells. Myc-tagged GRK2 constructs were mixed in vitro with Q209L G␣ q and incubated for 1 h at 4°C (see "Experimental Procedures"). Myc-GRK2 constructs were immunoprecipitated using a monoclonal anti-Myc antibody. Following SDS-PAGE separation/Western transfer, G␣ q protein was visualized a specific rabbit polyclonal anti-G␣ q antibody. Immunoprecipitation efficiency of the different of GRK2 constructs was assessed using a specific rabbit anti-Myc antibody. B, MCh-stimulated IP 3 generation was assessed in hippocampal neurons transfected with a 1:3 ratio of eGFP-PH PLC␦ to pcDNA3, D110A GRK2, D110A,K220R GRK2, or K220R GRK2 for 48 h. Expression of pcDNA3, D110A GRK2, or D110A,K220R GRK2 did not affect the IP 3 response to 3, 10, or 100 M MCh, whereas the K220R GRK2 construct essentially abolished the response to MCh. The data are shown as the peak increases in cytosolic fluorescence and are presented as the means Ϯ S.E. for 12-24 neurons in each case. C, cumulative data showing the effects of expressing the D110A GRK2 or D110A,K220R GRK2 constructs on M 1 mACh receptor desensitization in hippocampal neurons. The desensitization (R1/ MCh max /R2) protocol was performed as described under "Experimental Procedures." The data are presented as the means Ϯ S.E., for 8 -11 neurons taken from at least three separate hippocampal preparations. Expression of the D110A GRK2 construct significantly increased desensitization (*, p Ͻ 0.05), whereas the D110A, K220R GRK2 construct markedly decreased (**, p Ͻ 0.01) the degree of receptor desensitization. IP, immunoprecipitation; WB, Western blot. SY5Y cells with K220R GRK2 inhibited the acute IP 3 response to a single concentration of MCh (100 M, for 60 s); however, no inhibition of agonist-stimulated IP 3 production was observed in the presence of the D110A GRK2 or D110A, K220R GRK2 proteins (Fig. 5A). To examine whether expression of the non-G␣ q/11binding GRK2 mutants affects M 3 mACh receptor desensitization, we applied the pretreatment protocol where 10 M MCh was added for 1 min during R1 and R2, and the desensitizing concentration of MCh (100 M) was applied for 3 min, with 5-min wash periods between MCh additions (17). In agreement with our previous data, the R2 response was attenuated by approximately 40% compared with R1 in vector (pcDNA3)-transfected cells. Expression of either D110A GRK2 or D110A, K220R GRK2 proteins in SH-SY5Y cells had no effect on R2/R1 ratios (Fig. 5B), indicating that GRK2 is unlikely to play a significant role in M 3 mACh receptor desensitization in this human neuroblastoma cell line.
Role of ␤-Arrestins in M 1 mACh Receptor Signaling in Hippocampal Neurons-Initially, ␤-arrestin expression was determined after transfection with eGFP-tagged bovine ␤-arrestins 1 and 2. There was no detectable difference in the levels of expression of ␤-arrestins 1 and 2 (Fig. 6A), and expression was evident in hippocampal neurons for at least 15 days in vitro (data not shown). Non-eGFP-tagged bovine ␤-arrestin 1 and 2 constructs were co-transfected with eGFP-PH PLC␦ and effects on M 1 mACh receptor desensitization assessed using the R1/R2 protocol. Expression of ␤-arrestin 2 significantly increased the extent of M 1 mACh receptor desensitization, whereas expression of the ␤-arrestin 1 protein had no effect (Fig. 6B). These data suggest that recruitment of ␤-arrestin 2 may constitute an important step in M 1 mACh receptor desensitization.
Effects of Altered PKC Activity on M 1 mACh Receptor Signaling in Hippocampal Neurons-To determine whether PKC is involved in the regulation of M 1 mACh receptor signaling, we manipulated PKC activity using the protein kinase inhibitor staurosporine and PKC activator phorbol 12,13-dibutyrate (PDBu). To assess the acute effect of PDBu, the neurons were stimulated with two pulses of MCh (100 M, 30 s) separated by a 10-min washout period during which PDBu (1 M) was added for 3 min prior to the second pulse. This treatment resulted in a Ն50% suppression of the IP 3 response compared with vehicle control (Fig. 7, A and B). The effect of PDBu was prevented by pretreatment with staurosporine (1 M, added 15 min before the first MCh challenge; Fig. 7B). Staurosporine alone did not affect either the first or second response to MCh in hippocampal neurons (data not shown). Further, down-regulation of conventional and novel PKC isoenzymes through chronic PDBu (1 M, 24 h) treatment did not affect M 1 mACh receptor desensitization assessed using the R1/MCh max /R2 protocol. Thus, PKC down-regulation had no effect upon the acute MCh-stimulated IP 3 response (data not shown) and also had no effect upon the R2/R1 ratio (Fig. 7C). In addition, preincubation of neurons with staurosporine (1 M) also had no effect on the R2/R1 ratio (Fig. 7D). These data indicate that although acute PKC activation can inhibit PLC activity, PKC appears to play no role in the agonist-stimulated desensitization of M 1 mACh receptor signaling in hippocampal neurons.

Effects of Manipulating [Ca 2ϩ ] i on M 1 mACh Receptor Desensitization in Hippocampal
Neurons-Assessment of receptor desensitization using measurements of PLC activity must take into account the fact that this enzyme is modulated by other factors, including intracellular Ca 2ϩ (28). To examine the possible role that Ca 2ϩ plays in the desensitization of the M 1 mACh receptor, we assessed the effect of the loading neurons with the Ca 2ϩ chelator BAPTA-AM. Pretreatment with BAPTA-AM (30 M, 20 min) caused a reduction in the peak IP 3 response following MCh stimulation, with a greater effect being observed at 10 M than at 30 M MCh (Fig. 8A). However, M 1 mACh receptor desensitization, assessed using the R1/ MCh max /R2 protocol, was unaffected in BAPTA-loaded cells (Fig. 8B). To assess the effectiveness of BAPTA-AM pretreat- ment, intracellular Ca 2ϩ levels were determined in neurons using the Ca 2ϩ indicator Fluo-3. Agonist stimulation (MCh, 100 M, 30 s) produced only a small rise in intracellular Ca 2ϩ when compared with that elicited by depolarization with KCl (40 mM) for 1 min (Fig. 8C). Following depolarization, the Ca 2ϩ response to a second addition of MCh (100 M, 30 s) was often, although not invariably, increased (Fig. 8C). Pretreatment of neurons with BAPTA-AM resulted in complete attenuation of the MCh-stimulated [Ca 2ϩ ] i response but only a modest attenuation of the depolarization-induced response (Fig. 8C). DISCUSSION We have previously shown using the IP 3 biosensor eGFP-PH PLC␦ that expression of GRK2 in hippocampal neurons leads to an almost complete suppression of M 1 mACh receptor-mediated IP 3 signaling and that this suppression is independent of kinase activity, because the catalytically inactive K220R GRK2 mutant also completely inhibited signaling (18). In common with a number of other groups (14,15,19,21), we proposed that the suppression of signaling is a consequence of the direct interaction of the N-terminal RH domain of GRK2 with GTPbound G␣ q/11 . In the present study we have examined whether the catalytic activity of GRK2 acts as an alternative or additional mechanism for GRK2-mediated suppression of M 1 mACh receptor signaling.
Because both GRK2 and K220R GRK2 are equally effective in suppressing M 1 mACh receptor signaling (18), another approach was required to determine whether endogenous GRK2 is involved in M 1 mACh receptor regulation. We have previously used a construct consisting of virtually the whole coding sequence for rat GRK2 cloned into pcDNA3 in an antisense direction. This construct was able to bring about a 75% depletion of endogenous GRK2 when stably expressed in NG108 -15 cells (24). Here, we have confirmed this finding by transfecting the GRK2 antisense construct into HEK293 cells and demon-strating that 72 h post-transfection, an 80% suppression of GRK2 could be observed without affecting the expression of other GRKs. In hippocampal neurons expression of the GRK2 antisense construct produced Ն60% depletion of endogenous GRK2 expression, which almost completely prevented the prestimulated desensitization of M 1 mACh receptor-mediated IP 3 signaling, implicating endogenous GRK2 as a key protein kinase responsible for attenuating the M 1 mACh receptor response to agonist.
If GRK2 acts to suppress M 1 mACh receptor signaling purely through a direct RH-G␣ q interaction, it would follow that depletion of endogenously expressed GRK2 in hippocampal neurons might lead to enhanced agonist sensitivity. However, no change in acute agonist-stimulated IP 3 responses was observed in neurons transfected with the GRK2 antisense construct. It is possible that GRK3, the expression of which is unaffected by the presence of GRK2 antisense, may be able to compensate for the loss of GRK2 and thus prevent a change in the sensitivity of M 1 mACh receptors to agonist stimulation from occurring. However, this would contrast with the inability of GRK3 to compensate in GRK2-depleted neurons in experiments evaluating agonist-mediated M 1 mACh receptor desensitization. Although the use of the GRK2 antisense construct highlights the requirement for endogenous GRK2 in the regulation of M 1 mACh receptor signaling in hippocampal neurons, this approach sheds little new light on the mechanism(s) involved.
Initially, GRK2 was thought to block PLC signaling because of its ability to bind free G␤␥ subunits, thus inhibiting G␤␥ stimulation of PLC␤ isoenzymes (27,30). However, as has previously been demonstrated in COS7 cells (27), M 1 mACh receptor-stimulated IP 3 formation is unaffected following overexpression of the C-terminal domain (Gly-495 to Leu-689) of bovine GRK2 in hippocampal neurons. These data suggest that G␤␥ subunits play little or no role in M 1 mACh receptor- mediated PLC activation in this cell type. More recent studies have shown that the N-terminal RH domain of GRK2 is responsible for inhibiting G␣ q/11 -coupled signaling to PLC (14,15). Indeed, the key amino acids involved in this interaction have been mapped to the ␣5 helix of the N-terminal domain of GRK2 (19). Mutation of several individual amino acid residues within this region prevents G␣ q binding to GRK2 (19) and restores the ability of the G␣ q/11 -coupled mGlu1a receptor to signal through PLC (21). In agreement with these findings, we show here not only that mutation of the RH domain prevented G␣ q binding, but also that the expression of the non-G␣ q -binding D110A GRK2 mutant in hippocampal neurons did not inhibit acute MChstimulated M 1 mACh receptor signaling. Crucially, abolition of the GRK2 RH domain-G␣ q interaction restored the agonist-stimulated IP 3 response and meant that we were able to show that expression of D110A GRK2 enhanced, whereas D110A,K220R GRK2 markedly decreased M 1 mACh receptor desensitization in hippocampal neurons. Very recently Iwata et al. (31) have used similar single and double GRK2 point mutations to delineate the requirement for both the RH domain and receptor kinase activity in the rapid termination of signaling by recombinantly expressed H 1 histamine receptors in HEK cells. Although it is difficult to make definitive conclusions regarding the effects of D110A mutants without directly determining their effects on M 1 mACh receptor phosphorylation in hippocampal neurons, we believe our data are the first to point to GRK2-mediated phosphorylation being required to mediate the desensitization of an endogenously expressed receptor in a native neuronal background.
Crystallographic analysis of the structure of GRK2 in a complex with G␤␥ (20) has revealed that this GRK is able to bind simultaneously to receptor and G␣ q , in addition to G␤␥, suggesting that both G␤␥, and G␣ q binding might be required for the correct targeting of GRK2 to the receptor. We have shown here that in the absence of G␣ q binding, GRK2 is still able to phosphorylate the M 1 mACh receptor, apparently as efficiently as wild-type GRK2, in CHO cells. These data suggest that the availability of G␤␥ subunits and receptor are sufficient to allow the efficient interaction with GRK2 binding and receptor phosphorylation. Indeed, RH domain mutation of GRK2, to eliminate the interaction with G␣ q/11 , does not reduce the ability of GRK2 to phosphorylate rhodopsin (19), and inhibition of GRK2/ G␣ q/11 interactions has also been shown not to alter the ability of GRK2 to bind avidly to the mGlu1a receptor (21).
Thus, we propose that in hippocampal neurons, M 1 mACh receptors are subject to two distinct modes of regulation by GRK2, as well as being likely substrates for GRK6-mediated phosphorylation (18). We have also shown here that PKC-and Ca 2ϩ -dependent mechanisms do not appear to have a role in agonist-evoked M 1 mACh receptor desensitization. GRK-mediated GPCR phosphorylation often acts as a cue for arrestin recruitment to the receptor (10,30). At present, evidence implicating arrestin involvement in M 1 mACh receptor signaling is contradictory, even with respect to data obtained in similar cell expression systems. Thus, Lee et al. (32) have reported that co-expression of M 1 mACh receptors with ␤-arrestin 1 or ␤-arrestin 2 did not enhance receptor sequestration in HEK293-tsA201 cells. However, expression of the dominant-negative ␤-arrestin 1 V53D mutant in HEK293 cells has been reported to suppress sequestration of M 1 mACh receptors (33). In another report, antisense suppression of arrestin expression in HEK293 cells attenuated desensitization of the endogenously expressed mACh receptor but not the P2Y 1 or P2Y 2 receptors (34). In the present study we have provided evidence that M 1 mACh receptor phosphorylation, mediated by GRK(s) (GRK2 and/or GRK6), promotes the selective binding of ␤-arrestin 2, because receptor desensitization was enhanced in the presence of ␤-arrestin 2 but not ␤-arrestin 1. In agreement with our findings Santini et al. (35) found that in RBL-2H3 cells expressing physiological levels of M 1 mACh receptors, agonist stimulation leads to the selective recruitment of ␤-arrestin 2, but not ␤-arrestin 1.
An increasing number of studies have now provided a substantial body of evidence for ␤-arrestins not only facilitating receptor internalization but also scaffolding critical components of the ERK/JNK cascades, including Raf-1, MEK1, and ERK1/2 (36 -38), and JNK3 (12,39). Thus, M 1 mACh receptor phosphorylation by GRK2 (and possibly other GRKs/protein kinases) and the recruitment of ␤-arrestin 2 may not only attenuate phosphoinositide signaling but also act as a template for switching the signaling outputs of the receptor to alternative pathways. In the central nervous system this may provide a mechanism through which M 1 mACh receptors can cause the sustained signaling necessary to bring about long term changes in neuronal activity (40,41).
Despite the presence of a RH domain in the N terminus of all GRKs, at present only GRK2 and GRK3 (and recently GRK4)  (42) have been shown to suppress GPCR signaling in a phosphorylation-independent manner (14,15,19,43). In the case of GRK4, whether this action results from the RH domain binding to G␣ i/o or the ␥-aminobutyric acid, type B receptor remains to be established. With the exception of mGlu1a receptor, where evidence has been provided to support a model whereby GRK2 recruitment displaces G␣ q/11 binding to the receptor on agonist addition (21), little is known about the physiological significance of phosphorylation-independent regulation of G␣ q/11 -coupled receptors. Indeed, as discussed above, structural analysis has revealed that GRK2 is able to bind G␣ q/11 , G␤␥, and receptor simultaneously, inferring that GRK2 might act to inhibit GPCR signaling by blocking the interaction of G␣ q/11 (and/or G␤␥) with PLC␤ prior to receptor phosphorylation and arrestin binding (20). Our data show that the RH domain of GRK2 can mediate a rapid attenuation of M 1 mACh receptor-stimulated signaling events; however, in the absence of GRK2 RH domainmediated signaling inhibition, the importance of receptor phosphorylation for the prolonged suppression (or signal switching) of M 1 mACh receptor signaling could be unmasked. In conclusion, GRK2 has been shown to inhibit M 1 mACh receptor signaling through phosphorylation-dependent and -independent mechanisms. It appears likely that these mechanisms are interactive and perhaps sequential with the phosphorylationindependent mechanism initially suppressing G␣ q/11 /PLC interactions prior to receptor phosphorylation and recruitment of ␤-arrestin 2, which can cause receptor internalization, perhaps to initiate alternative signaling cascades as a prelude to longer term plasticity of neuronal functions.