Calmodulin interacts with the third intracellular loop of the serotonin 5-hydroxytryptamine1A receptor at two distinct sites: putative role in receptor phosphorylation by protein kinase C.

The serotonin 5-HT(1A) receptor couples to heterotrimeric G proteins and intracellular second messengers, yet no studies have investigated the possible role of additional receptor-interacting proteins in 5-HT(1A) receptor signaling. We have found that the ubiquitous Ca(2+)-sensor calmodulin (CaM) co-immunoprecipitates with the 5-HT(1A) receptor in Chinese hamster ovary fibroblasts. The human 5-HT(1A) receptor contains two putative CaM binding motifs, located in the N- and C-terminal juxtamembrane regions of the third intracellular loop of the receptor. Peptides encompassing both the N-terminal (i3N) and C-terminal (i3C) CaM-binding domains were tested for CaM binding. Using in vitro binding assays in combination with gel shift analysis, we demonstrated Ca(2+)-dependent formation of complexes between CaM and both peptides. We determined kinetic data using a combination of BIAcore surface plasmon resonance (SPR) and dansyl-CaM fluorescence. SPR analysis gave an apparent K(D) of approximately 110 nm for the i3N peptide and approximately 700 nm for the i3C peptide. Both peptides also caused characteristic shifts in the fluorescence emission spectrum of dansyl-CaM, with apparent affinities of 87 +/- 23 nm and 1.70 +/- 0.16 microm. We used bioluminescence resonance energy transfer to show that CaM interacts with the 5-HT(1A) receptor in living cells, representing the first in vivo evidence of a G protein-coupled receptor interacting with CaM. Finally, we showed that CaM binding and phosphorylation of the 5-HT(1A) receptor i3 loop peptides by protein kinase C are antagonistic in vitro, suggesting a possible role for CaM in the regulation of 5-HT(1A) receptor phosphorylation and desensitization. These data suggest that the 5-HT(1A) receptor contains high and moderate affinity CaM binding regions that may play important roles in receptor signaling and function.

Although previous studies have thoroughly detailed the coupling of the 5-HT 1A receptor to heterotrimeric G proteins and intracellular second messengers, no studies have investigated the possible role of additional receptor-interacting proteins in 5-HT 1A receptor signaling. Growing evidence suggests that GPCRs can bind a variety of proteins, which can subsequently modify receptor signaling, internalization, and/or interaction with G protein subunits. For example, several GPCRs contain C-terminal PDZ motifs that permit them to interact with multiple intracellular proteins, including the Na ϩ /H ϩ exchanger regulatory factor and post-synaptic density protein PSD-95 (20). The isoform of 14-3-3 protein has been shown to bind and regulate the surface expression of different ␣ 2 -adrenergic receptor subtypes (21). Other reported GPCR-interacting proteins include endothelial nitric-oxide synthase (22), the small nonheterotrimeric G proteins ARF and RhoA (23), and the tyrosine phosphatases SHP-1 and SHP-2 (24).
Calmodulin (CaM) is a ubiquitous intracellular Ca 2ϩ -sensor that plays an important role in several downstream GPCR * This work was supported by grants from the Department of Veterans Affairs (Merit Award to J. R. R.), the National Institutes of Health Grants DK52448 and GM63909 (to J. R. R) and DK59950 and MH64795 (to A. K. G.), a pre-doctoral fellowship from the American Heart Association, Mid-Atlantic Affiliate (Grant 0215195U to J. H. T.), and laboratory endowments jointly supported by the MUSC Division of Nephrology and Dialysis Clinics, Inc. (to J. R. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. signaling pathways. CaM can bind to and modulate a diverse array of cellular proteins, including enzymes, ion channels, transcription factors, and cytoskeletal proteins. Recently, CaM has been shown to bind to the epidermal growth factor receptor (25), to platelet glycoprotein VI (26), and to some GPCRs (27). The first GPCR that was shown to interact with CaM was the metabotropic glutamate subtype 5 receptor (mGluR5), which contains a CaM-binding site in a region of the extended C terminus of the receptor that is also known to bind G protein ␤␥ subunits and to contain a PKC phosphorylation site (28). Wang et al. (27) showed that the -opioid receptor interacts with CaM at a C-terminal juxtamembrane region in the third intracellular loop. They concluded that CaM binding reduces G protein coupling, probably through a competitive mechanism. Similarly, Bofill-Cardona et al. (29) showed that CaM interacts with an N-terminal juxtamembrane region of the D 2 -dopamine receptor third intracellular loop, resulting in a blockade of the receptor-operated G protein activation switch. These examples indicate that CaM interactions may play important and diverse roles in GPCR signaling, although those roles remain largely undefined.
Our group has previously reported a role for CaM in 5-HT 1A receptor signaling to extracellular signal-regulated protein kinase (ERK) MAPKs (18,30). 5-HT 1A receptor-mediated ERK activation is inhibited by chelation of intracellular Ca 2ϩ , CaM inhibitors, and expression of the CaM-sequestering protein calspermin. CaM appears to play a role in ERK activation by modulating receptor endocytosis, a step required for the activation of the mitogen-activated protein kinases kinase, MEK, which phosphorylates and activates ERK. The fact that CaM plays a vital role at such a proximal step in the 5-HT 1A receptor signaling pathway suggests that perhaps CaM exerts its effects by interacting with the receptor itself.
In this work, we report that CaM co-immunoprecipitates with the 5-HT 1A receptor in CHO fibroblasts. A search of the primary sequence revealed that the human 5-HT 1A receptor contains two potential CaM binding motifs, located in the Nand C-terminal juxtamembrane regions of the third intracellular loop of the receptor. Both motifs contain consensus PKC phosphorylation sites and are important for G i/o protein coupling, indicating that these regions of the receptor likely play an important regulatory role in receptor function. The purpose of this work was to determine whether the two putative CaM binding sites could bind to CaM, and if so, whether this would alter some parameter of receptor function.
Plasmids and Transfections-The yellow fluorescent protein expression vector eYFP-N1 was obtained from Clontech (San Jose, CA). The Renilla Luciferase protein expression vectors RLuc-N1 and RLuc-C1 were kindly provided by J. Yordy (MUSC, Charleston, SC). The calmodulin coding sequence without its stop codon was amplified from HeLa cell cDNA using sense and antisense primers containing unique XmaI and SacII restriction sites. The fragment was then subcloned in-frame into the RLuc-N1 and RLuc-C1 vectors. The 5-HT 1A receptor coding sequence was amplified from pcDNA3.1 containing the entire 5-HT 1A receptor coding sequence and 3Ј untranslated region using sense and antisense primers containing unique BaMH1 and XhoI restriction sites (19). The fragment was then subcloned in-frame into the eYFP-N1 vector. All vector constructs were verified by DNA sequencing.
Cell Culture and Transfection-CHO-K1 (CHO) cells expressing the 5-HT 1A receptor were maintained in F-12/Ham's medium, supplemented with 10% fetal calf serum, streptomycin (100 g/ml), penicillin (100 units/ml), and gentamicin (400 g/ml) at 37°C in a 5% CO 2enriched, humidified atmosphere. 24 -48 h before each experiment, cells were switched to serum-free medium containing 0.5% fatty acidfree bovine serum albumin (Sigma). For transient transfections, CHO cells were plated in 6-well plates at 1 ϫ 10 5 cells per well, and cultured for 24 h in F-12/Ham's medium supplemented with 10% fetal calf serum. Cells were then transferred to antibiotic-and serum-free F-12/ Ham's medium and transfected using LipofectAMINE 2000 reagent according to the manufacturer's instructions (Invitrogen). The final amount of transfected DNA for a single well was ϳ2 g. After transfection, cells were cultured in F-12/Ham's medium supplemented with 0.5% bovine serum albumin or 10% fetal calf serum for 48 h to allow for protein expression. For stable transfections, CHO cells were plated in a 12-well plate and cultured in F-12/Ham's medium supplemented with 10% fetal calf serum until they reached ϳ90 -95% confluence. Cells were transfected (1 g of DNA/well) as previously described and cultured for 48 h. Cells were then sub-cultured into 100-mm dishes, and media was replaced with fresh F-12/Ham's medium supplemented with 10% fetal calf serum and 400 g/ml Geneticin. Cells were cultured for 1-2 weeks to allow for the selection of stable clones.
Gel Shift Assays-Gel shift analysis of CaM-peptide complexes was performed using urea-polyacrylamide gel electrophoresis, as described by Erickson-Viitanen and Delgrado (33). Reactions (30-l total volume) containing 300 pmol of CaM (ϳ5 g) and increasing amounts of peptide (0 -3000 pmol) were incubated in 100 mM Tris-HCl, pH 7.5, 4 M urea, and either 0.1 mM CaCl 2 or 1 mM EGTA at 22°C for 30 min. 15 l of a 50% glycerol/0.1% bromphenol blue loading buffer was added to each reaction, and the samples were resolved on 14% polyacrylamide gels containing 4 M urea and either 0.1 mM CaCl 2 or 1 mM EGTA in the running buffer. Protein was visualized by staining with Gel-code blue (Pierce) staining reagent.
Blot Overlay Assays-Peptides (1-100 nmol) were immobilized to PVDF membranes by slot-blot and washed twice with 100 mM Tris-HCl, pH 7.5. The membranes were blocked with 5% bovine serum albumin in 100 mM Tris-HCl, pH 7.5, containing 0.1% Tween 20 for 1 h at room temperature, and then were incubated with 0.5 g/ml biotinylated CaM in the presence of either 0.1 mM CaCl 2 or 1 mM EGTA overnight at 4°C. The PVDF membranes were then washed 3ϫ in the same buffer without CaM, followed by incubation with alkaline phosphatase-conjugated avidin for 1 h at room temperature. Detection was with a chemiluminescent reagent.
Surface Plasmon Resonance-Real-time binding and kinetic analyses were performed at the MUSC Biomolecular Resource Facility on a BIAcore 3000 biosensor system (Pharmacia Biosensor AB) using surface plasmon resonance (SPR) measurements. Carboxymethylated sensor chips (type CM5) were activated with a 1:1 mixture of 0.2 M Nethyl-NЈ-(3-dimethylaminopropyl)carbodiimide and 0.05 M N-hydroxysuccinimide in water. Synthetic peptides (100 g/ml, 31.5 nM in 10 mM sodium acetate at pH 4.8) were then immobilized on the sensor chips using an amine coupling kit (BIAcore) as described by the supplier. Unreacted sites were blocked with 1 M ethanolamine (pH 8.5). The SPR signals from the immobilized peptides generated BIAcore response units ranging from 400 to 720. Control flow cells were activated and blocked in the absence of protein. Binding was evaluated over a range of CaM concentrations (15.6 nM to 1 M) in 150 mM NaCl, 100 mM HEPES (pH 7.4), and 400 M CaCl 2 under a continuous flow of 5 l/min at 25°C. 10 l of CaM-containing solutions was pulsed over the surface of the chip for 2 min using the kinject command. Binding of CaM to peptide-immobilized flow cells was corrected for binding to control flow cells. Flow cells were regenerated by passing over running buffer without CaCl 2 . Binding data were fitted to a 1:1 Langmuir binding model under steady-state conditions using BIAevaluation version 3.1 software (BIAcore).
Fluorometric Measurements with Dansyl-CaM-Dansyl-CaM was synthesized according to the method of Bertrand et al. (34). Briefly, 10 g of CaM was incubated with ϳ1 g dansyl-chloride for 1 h at 4°C. Dansyl-CaM was purified from unincorporated dye using a Centricon TM concentrator with a molecular mass cutoff of 10,000 Da. Measurement of absorbance at 340 nm (molar extinction coefficient: 3,400 M Ϫ1 cm Ϫ1 ) gave an incorporation of ϳ1.3 dansyl units per CaM molecule. Fluorescence emission spectra of dansyl-CaM were measured from 400 to 600 nm using a SLM 8000 TM C spectrofluorometer (Aminco-Bowman) with an excitation wavelength of 340 nm. Test peptides (0 -8 M) were incubated with dansyl-CaM in 100 mM Tris-HCl, pH 7.5, supplemented with 0.1 mM CaCl 2 for 2 h at room temperature. The concentration of dansyl-CaM (0.1-0.5 M) was varied, and concentration-response curves were generated for fluorescence enhancement at each dansyl-CaM concentration. The apparent K D values for each concentration of dansyl-CaM were fit to the Hill equation by linear regression to calculate true affinities.
Bioluminescence Resonance Energy Transfer-To determine whether CaM and the 5-HT 1A receptor come into close proximity in intact cells, we used bioluminescence resonance energy transfer (BRET), a technique that detects close proximity of proteins using energy transfer between luminescent and fluorescent tags. A bioluminescent donor source (Renilla reniformis luciferase, RLuc) can transfer energy to an acceptor fluorophore (yellow variant of Aequorea green fluorescent protein, YFP) within a radius of ϳ50 Å, and this transfer is virtually undetectable at distances greater than 100 Å (35). CHO-5-HT 1A R cells were transfected as described above and cultured in F-12/Ham's media for 48 h to allow for protein expression. Cells were detached with PBS/1 mM EDTA and distributed into a 96-well plate at 1 ϫ 10 5 cells/well. Fluorescence measurements were acquired using a Victor 2 multilabel plate reader (PerkinElmer Life Sciences). In some cases, cells were incubated in the presence of inhibitors for 20 min, followed by the addition of 1 M 8-OH-DPAT for 5 min. Coelenterazine h was then added to a final concentration of 5 M, and sequential measurements were made with filters at 460 Ϯ 25 nm and 525 Ϯ 25 nm. The BRET ratio was calculated as the ratio of light emitted at 525 nm (YFP) over the light emitted at 460 nm (Luciferase).
Statistical Analysis-Results shown represent the means Ϯ S.E. of the number of experiments indicated in each case. Statistical analysis was performed by student's t test.

Interaction of CaM with the Serotonin 5-HT 1A Receptor-We
have previously shown that CaM plays roles in numerous 5-HT 1A receptor signaling pathways, including stimulation of MAPKs (18,30), activation of Na ϩ /H ϩ exchange (19), and receptor internalization (30). We wondered if, in addition to its distinct roles in downstream signaling pathways, CaM might play a more proximal role in 5-HT 1A receptor signaling by interacting directly with the receptor through one of its intracellular domains. To that end, we immunoprecipitated the 5-HT 1A receptor from CHO fibroblasts that were transfected to stably express this receptor at physiological levels (36). As shown in Fig. 1, a CaM-specific antibody detected a distinct band at ϳ19 kDa, which corresponds to the Ca 2ϩ -bound mobility of CaM under denaturing conditions. This band was not present when immunoprecipitation was performed with nonspecific-IgG. The density of the CaM band was similar to or slightly decreased in samples treated with the 5-HT 1A receptorselective agonist 8-OH-DPAT. These results suggest that in CHO cells CaM is constitutively complexed with the 5-HT 1A receptor.
Identification of Putative CaM Binding Regions in the Third Intracellular Loop of the 5-HT 1A Receptor-Using a computer search program that identifies putative CaM-binding sites based on evaluation criteria such as hydropathy, ␣-helical propensity, residue charge, helical class, residue weight, and hydrophobic residue content (37), we identified two putative CaM binding regions in the protein sequence of the 5-HT 1A receptor. Both putative CaM-binding domains were localized to the third intracellular loop of the receptor, at the N-and C-terminal juxtamembrane regions ( Fig. 2A). As a general rule, CaM binding regions are characterized by the presence of several hydrophobic residues interspersed with several positively charged residues, often forming amphipathic ␣-helices (38,39). CaM binding regions described to date have been divided into several motifs based on the distance between key hydrophobic residues. The N-terminal CaM binding region was identified as a 1-12 motif, with key hydrophobic residues at positions 1 and 12, whereas the C-terminal CaM binding region in the 5-HT 1A receptor was classified as a 1-8- whereas no binding was observed for the MLCK control peptide (Fig. 3). Binding was inhibited by removing Ca 2ϩ from the incubation mixture and replacing it with 1 mM EGTA, although moderate binding to the i3N peptide in the presence of EGTA was still observed.
To determine if the peptides can form high affinity complexes with CaM, we performed polyacrylamide gel electrophoresis in the presence of 4 M urea. The presence of 4 M urea dissociates lower affinity and nonspecific protein-protein interactions (K D Ͼ 100 nM). A constant amount of CaM (300 pmol) was incubated with increasing amounts (75-3000 pmol) of the peptides and subsequently analyzed by nondenaturing gel electrophoresis (Fig. 4). In the presence of Ca 2ϩ , the addition of the 5-HT 1A i3N (Fig. 4B), but not the i3C peptide (Fig. 4C), pro-duced an upward shift in the migration of CaM, as did a peptide corresponding to the CaM binding region of MLCK (Fig. 4A). In contrast, no peptide-CaM complexes were formed when Ca 2ϩ was chelated with EGTA. These data suggest that the i3N peptide binds CaM with high affinity (better than 100 nM), whereas the i3C peptide likely binds with significantly lower affinity. The mobility shift induced by the MLCK peptide was complete at a peptide:CaM molar ratio of ϳ5:1. The 5-HT 1A i3N peptide induced gel shifts that were incomplete up to a peptide: CaM molar ratio of 10:1. The inability of the 5HT 1A peptides, and even the MLCK peptide, which is known to bind CaM with very high affinity (ϳ6 pM), to cause complete shifts at a molar ratio of 1:1 is in line with previous reports in which peptides from the CaM binding regions of the mGluR5 (40)  dopamine receptor (29) were also unable to cause complete shifts under the same conditions.
Surface Plasmon Resonance-Interactions between CaM and the 5-HT 1A receptor i3N and i3C peptides were further examined by SPR using BIAcore 2000 sensor technology. The peptides were immobilized to a carboxymethylated CM5 sensor chip using amine-coupling chemistry at typical densities of 400 -720 resonance units. No significant binding was observed between CaM and the chip surface or between CaM and the negative control MLCK mutant peptide (data not shown). In contrast, CaM rapidly and reversibly interacted with both the i3N and i3C peptides in the presence of Ca 2ϩ , but not in the presence of EGTA, showing typical saturation curves (Fig. 5). Kinetic data yielded complex results with neither i3N or i3C peptides fitting accurately to a 1:1 Langmuir model. This was in part due to very high on/off rates, which were beyond the limitations of the software. Interactions were instead evaluated using steady-state analysis in which equilibrium responses (R eq ) were plotted against CaM concentration. Affinity calculations yielded K D values of ϳ110 nM and 700 nM for the i3N-CaM and i3C-CaM interactions, respectively. These results are consistent with our gel shift studies shown in Fig. 4.
Interaction of 5-HT 1A i3 Loop Peptides with Dansyl-CaM-To evaluate the binding affinities of i3N and i3C peptides with CaM in solution, we measured changes in the fluorescence emission spectrum of dansyl-CaM. Ligand binding to dansyl-CaM is thought to shield the fluorophore from the aqueous environment, and this can be detected as an enhancement in fluorescence emission and a blue shift of the emission peak to a lower wavelength. This is clearly illustrated in Fig. 6A. In the absence of Ca 2ϩ , dansyl-CaM displays weak fluorescence, with an emission peak of ϳ520 nm. In the presence of Ca 2ϩ , the fluorescence emission is enhanced and shifted to a lower wavelength. Addition of i3N and i3C peptides further enhanced dansyl-CaM fluorescence and shifted the emission peak to below 500 nm, whereas the negative control MLCK mutant peptide was completely ineffective (data not shown). We then examined the concentration-dependent binding of i3N and i3C peptides to dansyl-CaM. Fig. 6 (B and C) shows the fluorescence emission of various concentrations of dansyl-CaM (0.1, 0.2, 0.3, 0.4, and 0.5 M) incubated with increasing concentrations of i3N or i3C peptides in the presence of Ca 2ϩ . Data points were fit by nonlinear least squares analysis to the Boltzmann one-site binding equation. Notably, apparent affinity measurements were dependent on the concentration of dansyl-CaM, with higher concentrations of dansyl-CaM requiring increased concentrations of peptide to induce enhancements in fluorescence. This can be attributed to depletion of free peptide, a phenomenon that is most apparent at low concentrations of dansyl-CaM. This can be corrected by calculating affinities over a range of dansyl-CaM concentrations and extrapolating to an infinitely low dansyl-CaM concentration, at which point depletion is essentially nonexistent. Accordingly, affinity measurements derived from peptide binding curves were plotted against the concentration of dansyl-CaM (Fig. 6, D and E). These values fell onto straight lines, which when extrapolated to the y-axis, gave apparent affinities of 87 Ϯ 23 nM and 1.70 Ϯ 0.16 M for the i3N and i3C peptides, respectively. These values are in line with estimated affinities that were determined using SPR.
Association of 5-HT 1A R-YFP and CaM-RLuc in CHO Cells Assessed by BRET-Having shown that CaM can interact with synthetic peptides derived from the 5-HT 1A receptor i3 loop, we next set out to show that this interaction occurs in vivo. This type of analysis has been lacking in previously published reports of CaM-GPCR interactions, which have relied entirely on the use of solubilized cellular extracts and in vitro binding experiments. To demonstrate close physical interaction be- tween the 5-HT 1A receptor and CaM within living cells, we used bioluminescence resonance energy transfer (BRET), a technique that detects close proximity of proteins using energy transfer between luminescent and fluorescent tags. We tagged the C terminus of the full-length human 5-HT 1A receptor with YFP and stably expressed it in CHO cells. These cells were then transfected with CaM fused to either the N or C terminus of RLuc, and the BRET ratio was determined as the ratio of light emitted by YFP (525 Ϯ 25 nm) over that emitted by RLuc (460 Ϯ 25 nm) following the addition of the membrane-permeable RLuc substrate, coelenterazine. Before each experiment, total YFP fluorescence was measured (excitation ϭ 485 Ϯ 15 nm) to verify equal protein expression between individual constructs and between experiments. As shown in Fig. 7, cells transfected with RLuc alone produced an insignificant BRET signal (BRET ratio ϭ 0.03), which can primarily be attributed to nonspecific energy transfer and bleed-through of the luminescent signal into the fluorescent filter-set. In contrast, a significant BRET signal was observed in cells transfected with CaM fused at either the N terminus (BRET ratio ϭ 0.18) or C terminus (BRET ratio ϭ 0.15) of RLuc. The fact that both fusions were capable of producing a significant BRET signal is not surprising, in that CaM is a relatively symmetrical protein with N-and C-terminal EF hand motifs. These data strongly support the idea that CaM and the 5-HT 1A receptor interact with each other in a constitutive manner, as was suggested by the results of the co-immunoprecipitation experiments shown in Fig. 1.
As shown previously, co-immunoprecipitation of CaM with the 5-HT 1A receptor in CHO cells was not dependent on agonist treatment. However, co-immunoprecipitation often does not readily detect differences in protein association, particularly when the associations are weak or transient. We used BRET to determine whether receptor activation plays a role in the CaM-5-HT 1A receptor interaction. Exposure of cells to 1 M 8-OH-DPAT for 5 min had no effect on the constitutive BRET signal (Fig. 7B). To further characterize the BRET signal, cells were incubated for 20 min in the presence of several chemical inhibitors of CaM. A significant decrease in the BRET ratio was observed in the presence of three structurally distinct CaM antagonists, W-7, ophiobolin A, and calmidazolium chloride. In addition, the cell-permeable Ca 2ϩ -chelator, BAPTA-AM also decreased the BRET ratio. These data suggest that CaM needs ( 227 RKTVK 231 ) and i3C ( 341 RKTVK 345 ) CaM binding regions. The i3N and i3C regions correspond to typical CaM-binding domains, forming amphipathic ␣-helices composed of positively charged and hydrophobic residues, with few or no negatively charged residues. Phosphorylation at the target threonine residues in the i3N and i3C residues (Fig. 8A) adds a negative charge and would likely influence CaM binding. We tested this hypothesis using synthetic peptides (i3N-P and i3C-P), which are identical to the previously used peptides but are phosphorylated at threonines 229 and 343, respectively. The fluorescence emission spectrum of dansyl-CaM was evaluated in the presence of each phosphorylated peptide. As shown in Fig. 8B, the i3N-P peptide produced a significantly reduced shift in the emission peak of dansyl-CaM as compared with the i3N peptide. Likewise, the i3C-P peptide also produced a reduced shift in the emission peak of dansyl-CaM (Fig. 8C). These results suggest that the phosphorylated peptides have significantly reduced affinities for CaM compared with their unphosphoryl-ated counterparts and that the association of CaM with the i3N and i3C regions of the 5-HT 1A receptor may be regulated by phosphorylation reactions.
Interaction of CaM with the i3N and i3C Peptides Antagonizes Phosphorylation by Protein Kinase C-Activation of PKC by phorbol esters induces a rapid phosphorylation of the 5-HT 1A receptor (36). This phosphorylation is associated with desensitization of multiple signaling pathways, including the inhibition of adenylyl cyclase (36), K ϩ channel regulation (42), and hydrolysis of phosphoinositides (43). Lembo et al. (43) showed that this desensitization could be reversed by mutation of three putative PKC sites in the i3 loop of the 5-HT 1A receptor, two of which correspond to the sites in the i3N and i3C peptides. We tested the i3N and i3C peptides as substrates for PKC phosphorylation using in vitro phosphorylation assays. Purified rat brain PKC (comprised primarily of PKC-␣ and PKC-␤ isozymes) readily phosphorylated the i3N peptide (K m ϭ 4.21 Ϯ 1.11 M; V max ϭ 35.35 Ϯ 4.89 nmol/min/mg) and i3C peptide (K m ϭ 7.10 Ϯ 1.32 M; V max ϭ 10.10 Ϯ 1.00 nmol/min/ mg) as shown in Fig. 9. These values are similar to values described for other PKC substrates (44). To determine whether CaM can compete with PKC for access to the i3N and i3C phosphorylation sites, we repeated the in vitro phosphorylation assays in the presence of increasing concentrations of CaM. Phosphorylation of both peptides was dose-dependently decreased by CaM. Not surprisingly, the peptides were differentially sensitive to CaM binding, with phosphorylation of the i3N peptide being almost completely inhibited in the presence of 20 M CaM, whereas phosphorylation of the i3C peptide was inhibited by ϳ75% in the presence of 120 M CaM. These data suggest that the association of CaM with the i3N and i3C peptides can attenuate phosphorylation by PKC. DISCUSSION The purpose of this study was to characterize the interactions of CaM with two putative CaM-binding domains in the 5-HT 1A receptor. We have used a variety of methods to characterize the interactions in cells and in vitro, resulting in several novel observations. We have reported in this work that the serotonin 5-HT 1A receptor contains two putative CaM-binding sites in the juxtamembrane regions of the third intracellular loop, spanning amino acids 215-237 (i3N) and 328 -350 (i3C). We used blot overlays, gel shifts, and SPR to document interactions between both peptides and CaM and calculated affinities of ϳ100 nM and 1 M, respectively. To determine whether interactions between the holo-5-HT 1A receptor and CaM might occur in CHO cells, we used co-immunoprecipitation and BRET experiments. We showed that CaM immunoprecipitates with the 5-HT 1A receptor and resides in overlapping domains of CHO cells with the 5-HT 1A receptor, as determined by BRET. The presence of a strong BRET signal between CaM and the 5-HT 1A receptor suggests that the two proteins reside within 50 Å of each other, and are likely to physically interact. To our knowledge, this represents the first example of a CaM-GPCR interaction shown to occur in living cells. The interactions between the 5-HT 1A receptor and CaM depend on Ca 2ϩ binding and activation of CaM in that the Ca 2ϩ -chelator, BAPTA-AM, and three CaM inhibitors can abrogate the BRET signal. Moreover, the interaction between CaM and the 5-HT 1A receptor appears to be constitutive, in that agonist treatment does not alter significantly either the amount of CaM in 5-HT 1A receptor immunoprecipitates, or the BRET signal between 5-HT 1A receptor-eYFP and CaM-luciferase.
A wide variety of CaM-binding domains have been described in various proteins. The two putative CaM binding sites parallel typical CaM-binding domains, forming predicted amphipathic ␣-helices composed of basic and hydrophobic amino acid residues. In addition, both sites could be aligned with well established CaM binding motifs based on the location of key hydrophobic residues. The CaM binding region in the C-terminal portion of the the 5-HT 1A receptor i3 loop was classified as a 1-8-14 motif, in that it aligns well with other well defined 1-8-14 CaM binding motifs, such as those for neuronal and endothelial nitric-oxide synthase. The putative CaM-binding domain in the N terminus of the 5-HT 1A receptor i3 loop is similar in location to a CaM-binding domain identified in the Results shown represent the mean Ϯ S.E. for at least three separate experiments with each peptide. D 2 dopamine receptor, but the two sites show little sequence homology. The N-terminal 5-HT 1A receptor i3 loop binding region was classified into a relatively newly defined class of CaM binding motifs characterized by core hydrophobic residues at positions 1 and 12, and this peptide was well aligned with other 1-12 motif-containing proteins such as caldesmon, the N-methyl-D-aspartate receptor, and the phosphodiesterases. Although originally classified as a type 1B 1-8-14 motif (sequence read in the C-to N-terminal direction), the D 2 dopamine CaM binding region has been re-classified as a basic motif, characterized by a preponderance of positively charged lysine and arginine residues within the CaM-binding sequence.
The role of the 5-HT 1A receptor CaM-binding sites is unknown, however, the juxtamembrane N-and C-terminal regions of the 5-HT 1A receptor third intracellular loop have been implicated in numerous receptor functions. Both sites reside in regions of the i3 loop important for contacting heterotrimeric G proteins, with work by Albert et al. suggesting a model in which amphipathic ␣-helical regions of the i2 and i3 loop align to form hydrophobic G protein interaction sites (45). Structural predictions based on this model would likely suggest an interaction with CaM as well, based on similar hydrophobic-binding pockets. This is supported in part by studies showing that certain amphipathic peptides, such as the venoms mellitin and mastoparan, can interact with both CaM and G␣ subunits (46,47). Furthermore, interaction of CaM with the -opioid (OP 3 ) and D 2 dopamine receptors has previously been shown to regulate G protein activation (29). In the case of the OP 3 receptor, CaM was shown to compete with G protein binding, whereas in the case of the D 2 receptor, CaM was shown to block the receptoroperated G protein activation switch without uncoupling the receptor from G protein. It is likely that CaM plays similar roles in 5-HT 1A receptor signaling to G protein subunits.
The 5-HT 1A receptor i3N and i3C regions also contain phosphorylation sites previously implicated in receptor desensitization (36,41,49). Lembo et al. showed that phorbol ester-induced receptor desensitization could be reversed by mutation of three putative PKC sites in the i3 loop of the 5-HT 1A receptor (43). Two of these sites, 227 RKTVK 231 and 341 RKTVK 345 , reside in the i3N and i3C regions, respectively. New data in this report shows that both 5-HT 1A receptor i3N and i3C peptides can serve as substrates for rat brain PKC-induced phosphorylation. Moreover, CaM decreases the ability of PKC to phosphorylate either peptide in a concentration-dependent manner. This finding suggests that constitutive binding of CaM to either juxtamembrane region of the 5-HT 1A receptor i3 loop could prevent or greatly attenuate phosphorylation of the receptor by PKC, and if so, then CaM should also serve to attenuate PKCinduced desensitization of the receptor.
Significant evidence suggests that phosphorylation of CaMbinding domains and the binding of CaM to a target protein can be mutually exclusive and in some cases antagonistic. Phosphorylation of target proteins by PKC has been shown to reduce binding of CaM to myosin 1 (50), the plasma membrane Ca 2ϩ pump (51,52), and calcium transport-like (CaT1) protein (53). Likewise, binding of CaM to neural tissue-enriched acidic protein (NAP-22) inhibits phosphorylation by PKC (54). Minakami et al. showed that phosphorylation and CaM binding to peptides derived from the mGluR5 are antagonistic, indicating a possible regulatory role for CaM in GPCR signaling and responses (40). We have shown that phosphorylated peptides derived from the N-and C-terminal CaM binding regions of the 5-HT 1A receptor i3 loop have weakened affinities for CaM. In addition, association of CaM with the i3N and i3C peptides inhibited phosphorylation by PKC. These data suggest that PKC-induced phosphorylation and CaM binding to the jux-tamembrane regions of the 5-HT 1A receptor are antagonistic. Interestingly, the affinities of the i3N and i3C peptides for CaM (ϳ100 nM and 1 M) are both significantly stronger than their corresponding affinities for PKC (ϳ5 M and 7 M). This indicates that CaM could reasonably be expected to blunt phosphorylation of both i3N and i3C peptides by PKC within the 5-HT 1A holoreceptor in vivo.
Taken in aggregate, our findings implicate constitutive direct binding of CaM to juxtamembrane regions of the i3 loop of the 5-HT 1A receptor in blunting the phosphorylation of the receptor by PKC. The findings further suggest that CaM plays a role in preventing or attenuating PKC-induced (heterologous) desensitization of the 5-HT 1A receptor in CHO cells. Indeed, the 5-HT 1A receptor in CHO cells is relatively resistant to desensitization in that pretreatment with phorbol esters shifts the concentration-response curve for the inhibition of adenylyl cyclase less than one log to the right, without altering the maximum response (36).
Our findings do not rule out other roles for CaM in regulating the propagation of signals from the 5-HT 1A receptor or in regulating other processes such as receptor internalization. In that regard, we have previously shown that CaM is critical for the 5-HT 1A receptor to activate ERK, in a process involving agonist-induced receptor internalization (30). Similarly, Melien et al. (55) showed that ERK activation induced by norepinephrine and prostaglandin F2␣ was sensitive to pharmacological inhibitors of CaM in hepatocytes. Likewise, CaM has been shown to mediate activation of ERK by the -opioid receptor through a pathway involving the transactivation of the epidermal growth factor (EGF) receptor (56). A role for CaM has also been proposed in the general process of receptor endocytosis and trafficking in numerous models, including yeast (57) and human epithelial cells (32). Our group reported a role for CaM in 5-HT 1A receptor internalization, a required step for MEK and subsequent ERK activation (30). Conversely, pharmacological inhibitors of CaM have been shown to inhibit recycling and degradation of the EGF receptor without affecting its internalization, resulting in the accumulation of receptors in enlarged endosomal structures (48). Although roles for direct binding of CaM to a GPCR in signal initiation or in internalization and/or trafficking of the receptor have yet to be reported, we speculate that these effects are possible, if not likely.
In conclusion, we have identified the presence of two distinct CaM-binding sites in the serotonin 5-HT 1A receptor. These sites reside in juxtamembrane regions at the N and C termini of the large third intracellular loop of the receptor. In addition, we have shown that the 5-HT 1A receptor interacts with CaM in intact, living cells. Finally, we have shown that binding of CaM and phosphorylation by PKC of the 5-HT 1A receptor third intracellular loop juxtamembrane peptides are antagonistic in vitro. These results suggest that CaM may serve as a critical regulator of 5-HT 1A receptor function.