Binding of Calmodulin to the D2-Dopamine Receptor Reduces Receptor Signaling by Arresting the G Protein Activation Switch*

Signaling by D2-dopamine receptors in neurons likely proceeds in the presence of Ca2+ oscillations. We describe here the biochemical basis for a cross-talk between intracellular Ca2+ and the D2 receptor. By activation of calmodulin (CaM), Ca2+ directly inhibits the D2 receptor; this conclusion is based on the following observations: (i) The receptor contains a CaM-binding motif in the NH2-terminal end of the third loop, a domain involved in activating Gi/o. A peptide fragment encompassing this domain (D2N) bound dansylated CaM in a Ca2+-dependent manner (KD ∼ 0.1 μm). (ii) Activation of purified Gαi1 by D2N, and D2receptor-promoted GTPγS (guanosine 5′-(3-O-thio)triphosphate) binding in membranes was suppressed by Ca2+/CaM (IC50 ∼ 0.1 μm). (iii) If Ca2+ influx was elicited in D2 receptor-expressing HEK293 cells, agonist-dependent inhibition of cAMP formation decreased. This effect was not seen with other Gi-coupled receptors (A1-adenosine and Mel1A-melatonin receptor). (iv) The D2 receptor was retained by immobilized CaM and radiolabeled CaM was co-immunoprecipitated with the receptor. Specifically, inhibition by CaM does not result from uncoupling the D2 receptor from its cognate G protein(s); rather, CaM directly targets the D2 receptor to block the receptor-operated G protein activation switch.


Introduction
Dopamine acts as a neuromodulator (rather than a neurotransmitter) in the central nervous system because dopamine controls the propensity of a neuron to fire action potentials. The receptors for dopamine belong to the class of G protein coupled receptors. Five receptor subtypes representing two subfamilies have been identified by molecular cloning; D 1 /D 5 receptors stimulate adenylyl cyclase activity whereas D 2 , D 3 and D 4 receptors couple to G proteins of the G i/o class to inhibit adenylyl cyclase. G i/o -mediated signal transduction in excitable cells is also known to inhibit voltage activated N-type Ca 2+ -channels and to gate inwardly rectifying K + -channels (GIRK) via the release of βγ-subunit; the former effect was also demonstrated for D 2 -dopamine receptors (for a review, see 1). DARPP-32, an inhibitor of protein phosphatase 1. DARPP-32 is dephosphorylated on D 2dopamine receptor activation and thus becomes active; this effect is strongly enhanced by Ca 2+ /CaM through activation of calcineurin (9). These examples suggest that the signal transduced by Ca 2+ /CaM and signaling initiated by the intracellular D 2 -receptor overlap and may add to each other.
We have found in the primary peptide sequence of the human D 2 -dopamine receptor a CaMbinding motif which is located in the N-terminus of the third cytoplasmic loop of the receptor.
In the present work we report that CaM can convey a Ca 2+ -signal directly to the receptor through binding to this receptor domain. When Ca 2+ /CaM binds to the receptor, it antagonizes signaling by the receptor at the level of receptor-mediated G protein turnover. Based on these observations, we propose to extend the concept of a cross-talk between CaM activation and D 2 -receptor signaling; by binding to the receptor CaM exerts feed-back inhibition to downtone the signaling efficiency of the D 2 -receptor. In the presence of repetitive Ca 2+ oscillations, i.e. when the neuron is actively firing action potentials, CaM is suggested to suppress overt enhancement of dopamine receptor signal transduction in striatal nerve cells. [ 125 I]epideprid was obtained from the Austrian Research Centre (Seibersdorf, Austria).

Materials
Peptides derived from the amino acid sequence of the NH 2 -terminal and of the COOHterminal part of the third intracellular loop of the human D 2 -dopamine receptor were synthesized by solid-phase peptide synthesis as described (10).

Determination of cAMP formation
HEK293 cells were grown to confluence in 6-well plates. The adenine nucleotide pool was labeled by incubating the cells for 16 h with [ 3 H]adenine (2 µCi/well). After that the medium was replaced and the cells were pre-incubated for 1h with 100 µM of the phosphodiesterase inhibitor rolipram. The production of cAMP was stimulated by the addition of 25 µM forskolin; receptor-mediated inhibition of cAMP formation was assessed in the absence and presence of the Ca 2+ -ionophore A23187 (calcimycin) at a concentration of 3 µM (Ca 2+ concentration in the assay medium = 1.8 mM) and of the receptor agonists at the indicated concentrations. Accumulation of cAMP was allowed to proceed for 15 min at room tempera-Recombinant, myristoylated Gα i-1 was produced in Escherichia coli and purified from bacterial lysates as described in (14).

Radioligand binding experiments
Receptor-promoted G protein activation was determined by measuring the association rate of [ 35 S]GTPγS in Hek293 membranes expressing the D 2 -dopamine and the A 1 -adenosine receptor as described (11); the A 1 -adenosine receptor was used as a control instead of the Mel 1a -receptor. The latter only weakly stimulates GTPγS-binding because of its tight association with G proteins (15 proceeded at 30°C and was terminated by the addition of ice-cold stop buffer at the indicated was carried out as described (11). Before the binding assay, cell membranes were washed with 10 mM EGTA as described above. The binding reaction was carried out for 90 min at

Immunoprecipitation of the epitope-tagged D 2 -dopamine receptor
Membranes prepared from the cells stably expressing c-myc-D 2 R were washed with EGTA as described above; subsequently the membranes (2 mg) were solubilized with 0.6 % cholate (the ratio of detergent to membrane protein was 3:1) in HME buffer containing 750 mM NaCl and protease inhibitors (Pefabloc, Boehringer Mannheim). The insoluble material was collected by centrifugation at 35.000*g for 20 min. The supernatant was concentrated over a porous polycarbonate membrane (Amicon, cut off size = 30kDa) and subsequently, 0.1% digitonin was added, the concentration of cholate adjusted to 0.12% and the concentration of NaCl to and were applied to a SDS-polyacrylamide gel containing 6 M urea. The supernatant, the last out of three washes and the immunoprecipitate were analyzed. The c-myc-tagged D 2 R was detected by immunoblotting using a polyclonal antiserum directed against c-myc; in addition, the blot was probed with a G protein β-subunit-specific rabbit antiserum (17). Alternatively, the immunoprecipitate that had been prepared as described above was resuspended and gently reasonably close to 1 (Fig. 4C). The true affinity was approximated by extrapolating to infinitely low concentrations of dansyl-CaM, i.e. to the y-axis intercept. This calculation gave a K D of 80 nM, a value similar to the IC 50 for CaM observed in membranes (see Fig. 2C). In the absence of Ca 2+ , addition of D2N also induced an increment in the fluorescence of dansyl-CaM; however, the affinity was substantially lower than in the presence of Ca 2+ (Fig. 2C, ❍).

Calmodulin inhibits the activation of Gα i-1 by D2N
Neither Ca 2+ nor Ca 2+ /CaM had any appreciable effect on the rate of GTPγS-binding to purified (recombinant) Gα i-1 (Fig. 5A), a reaction limited by the release of prebound GDP. Ca 2+ /CaM (Fig. 5D). In contrast to the marked inhibition of G protein activation (see Fig. 2), CaM only very modestly reduced the number (but not the affinity) of high-affinity agonist binding sites. Thus, CaM did not interfere with the ability of the agonist-liganded receptor to form a complex with its cognate G protein(s) but selectively blocked the subsequent reaction step in signal transduction, i.e. the G protein turnover catalyzed by the active receptor.

Co-immunoprecipitation of the epitope tagged D 2 -dopamine receptor and CaM
In order to demonstrate a physical interaction of the D 2 -receptor with CaM we tagged the Nterminus with the c-myc epitope and expressed the epitope-tagged receptor in HEK293 cells.
EGTA-pretreated membranes were solubilized and the receptor was immunoprecipitated with a monoclonal antibody directed against the c-myc epitope. As a control we solubilized membranes from non-transfected cells and subjected the extract -in parallel -to the immuno-  (Fig. 6C). For illustrative purposes we depicted the last wash step; thereafter, the addition of EGTA released 10 and 25% of the radiolabeled D 2 -receptors added to CaMagarose and CaM-sepharose, respectively.
Thus, on CaM-sepharose -more than on CaM-agarose -a significant amount of radioactivity remained bound even after chelation of free Ca 2+ . This radioactivity was released by boiling in SDS and nominally amounted to ~15% of the total receptor-bound radioactivity (Fig. 6C).
In order to prove that this fraction was liganded to the receptor, we chose two approaches.
First, we determined the non-specific binding of [ 125 I]epideprid to CaM-sepharose using a matched amount of radioactivity; this non-specific binding was negligible. As a control, the same experiment was also performed with the A 1 -adenosine receptor expressed in HEK293 cells where similar amounts of the ligand [ 3 H]DPCPX were retained on the CaM-sepharose in the absence or presence of the receptors (1.0% vs.1.2% of total, not shown). This confirms that binding of CaM is a property specific to the D 2 -receptor which is not shared by the A 1adenosine receptor. Secondly, the CaM-sepharose was loaded with an epitope-tagged D 2receptor. After carrying out the wash steps, receptor-specific immunoreactivity was recovered by boiling the matrix in SDS (inset in Fig. 6C). We stress that the immune reactive bands were probed with an anti-HA antiserum and that the HA-tagged receptor had been transiently expressed in COS-7 cells. Thus, immunoprecipitation (Fig. 6A) as well as immobilization on CaM-sepharose (Fig. 6C, inset) yielded similar migration patterns of the D 2 -dopamine receptor and these were independent of the epitope-tag and of the cellular source.

Discussion
In the present work, we show that Ca In the tertiary structure of the membrane-bound receptor, the CaM interaction site is located adjacent to (the putative) transmembrane domain 5, removed by only two peptide bonds. We have also used as a control the peptides D2N' and D2N'', the sequences of which have been shifted toward the C-terminus of the 3 rd loop thereby curtailing the putative CaM-binding motif by 4 and 7 N-terminal residues, respectively. From these peptides it is evident that the motif has to be completely represented (i.e. including the flanking N-terminal residues) for CaM binding; the control peptides do not bind to Ca 2+ /CaM (Fig. 3C) and do not activate G i either (not shown). Thus, functionally important residues are found in the 3 rd cytoplasmic loop/α-helix-boundary and this is also true for many other receptors (27). The current view holds that receptor activation results in an enhanced tertiary interaction of these cue residues (28,29). This conformational change activates the cognate G protein and may similarly facilitate the docking of CaM which makes it a ligand-regulated process.
Our evidence suggests that indeed the activated receptor binds CaM even when it is engaged in the high-affinity ternary complex (agonist/receptor/G protein complex); based on the following data we conclude that CaM does not disturb G protein recognition but impedes the receptor-induced activation switch. (i) The binding of CaM and the G protein α-subunit to the receptor peptide is not mutually exclusive; the receptor-peptide combines simultaneously with       Fig.6