Available online at www.sciencedirect.com

008) 1209–1220www.elsevier.com/locate/cellsig

Cellular Signalling 20 (2

Different kinases desensitize the human δ-opioid receptor (hDOP-R) in theneuroblastoma cell line SK-N-BE upon peptidic and alkaloid agonists

Nicolas Marie 1, Benjamin Aguila 2, Ahmed Hasbi 3, Audrey Davis,Philippe Jauzac, Stéphane Allouche ⁎

Université de Caen, Laboratoire de biologie moléculaire et cellulaire de la signalisation, UPRES-EA 3919,IFR 146 ICORE, avenue côte de Nacre, 14032 Caen, France

Received 29 October 2007; received in revised form 13 February 2008; accepted 15 February 2008Available online 19 February 2008

Abstract

In a previous work, we described a differential desensitization of the human δ-opioid receptor (hDOP-R) by etorphine (a non-selective and alkaloidagonist) and δ-selective and peptidic agonists (DPDPE ([D-Pen2,5]enkephalin) and deltorphin I (Tyr-D-Ala-Phe-Asp-Val-Val-Gly-NH2)) in theneuroblastoma cell line SK-N-BE (Allouche et al., Eur. J. Pharmacol., 371, 235, 1999). In the present study, we explored the putative role of differentkinases in this differential regulation.

First, selective chemical inhibitors of PKA, PKC and tyrosine kinases were used and we showed a significant reduction of etorphine-inducedopioid receptor desensitization by the bisindolylmaleimide I (PKC inhibitor) while genistein (tyrosine kinase inhibitor) was potent to impairdesensitization induced by the different agonists. When the PKA was inhibited by H89 pretreatment, no modification of opioid receptordesensitization was observed whatever the agonist used.

Second, we further studied the role of G protein-coupled receptor kinases (GRKs) and by usingwestern-blot experiments we observed that only theGRK2isoformwas expressed in the SK-N-BE cells. Next, the neuroblastoma cellswere transfectedwith thewild typeGRK2 or its dominant negativemutant GRK2-K220R and the inhibition on cAMP level was determined in naïve and agonist-pretreated cells. We showed that over-expression of GRK2-K220R totallyabolished etorphine-induced receptor desensitization while no effect was observed with peptidic agonists and over-expression of GRK2 selectively impairedcAMP inhibition promoted by etorphine suggesting that this kinase was involved in the regulation of hDOP-R activated only by etorphine.

Third, correlation between functional experiments and phosphorylation of the hDOP-R after agonist activation was assessed by western-blot using thespecific anti-phospho-DOP-R Ser363 antibody.While all agonistswere potent to increase phosphorylation of opioid receptor, we showed no impairment ofreceptor phosphorylation level after PKC inhibitor pretreatment. Upon agonist activation, no enhancement of receptor phosphorylationwas observedwhenthe GRK2 was over-expressed while the GRK2-K220R partially reduced the hDOP-R Ser363 phosphorylation only after peptidic agonists pretreatment.

In conclusion, hDOP-R desensitization upon etorphine exposure relies on the GRK2, PKC and tyrosine kinases while DPDPE and deltorphin Imediate desensitization at least via tyrosine kinases. Although the Ser363 was described as the primary phosphorylation site of the mouse DOP-R,we observed no correlation between desensitization and phosphorylation of this amino acid.© 2008 Elsevier Inc. All rights reserved.

Keywords: Opioid receptor; Desensitization; Phosphorylation; GRK; Second-messenger kinase

⁎ Corresponding author. Laboratoire de Biochimie, Centre Hospitalier etUniversitaire, Avenue côte de nacre, 14033 Caen Cedex, France. Tel.: +33231064560; fax: +33 231064985.

E-mail address: allouche-s@chu-caen.fr (S. Allouche).1 Present address: CNRS, UMR7157, INSERM, U705, Université Paris

Descartes, Université Paris 7, Laboratoire de neuropsychopharmacologie desaddictions, Faculté de pharmacie, 4 avenue de l'Observatoire, 75006 Paris, France.2 Recipient of a fellowship from the Conseil Régional de Basse-Normandie.3 Present address: Departments of Pharmacology, University of Toronto and

the Center for Addiction and Mental Health, Toronto, Ontario, Canada.

0898-6568/$ - see front matter © 2008 Elsevier Inc. All rights reserved.doi:10.1016/j.cellsig.2008.02.010

1. Introduction

Opioid receptors belong to the G protein-coupled receptor(GPCR) superfamily and are divided in three subtypes: μ(MOP-R), δ (DOP-R) and κ (KOP-R). Their activation byendogenous or exogenous ligands mediate numerous effectsincluding analgesia at both spinal and supraspinal level andinvolves both MOP- and DOP-R [1]. The use of opiates in painmanagement is limited by the appearance of tolerance to

1210 N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

analgesic effects when these compounds are chronicallyadministrated. This leads to increase the dosage to obtain thesame analgesia but consequently favours the risk of adverseeffect such as fatal respiratory depression. One of the hypothesisto explain this tolerance is the desensitization of opioid receptor,defined as a loss of its responsiveness following chronicactivation [2]. Indeed, MOP- and DOP-R desensitization wasobserved in rat chronically treated with morphine [3]. In miceknocked-out for arrestin-3, a crucial protein involved in opioidreceptor desensitization, no sign of tolerance to analgesic effectsnor MOP-R desensitization were noted upon chronic morphinetreatment demonstrating the direct relationship between opioidreceptor desensitization and tolerance [4].

Mechanisms of GPCR desensitization were initiallydescribed for the β2-adrenergic receptor and subsequent studieson opioid receptors demonstrated that almost the samemechanisms occurred: the first step involves phosphorylationof the complex receptor-agonist by GRKs that triggers arrestinsrecruitment on the receptor and subsequently uncouplesreceptor from its cognate G proteins [5].

Pei et al. [6] reported the first demonstration for a link betweenopioid receptor phosphorylation and desensitization using [32P]orthophosphate metabolic labeling. Indeed, when transfected inHEK293, mouse DOP-R (m DOP-R) underwent a rapidphosphorylation (as soon as 3 min) upon DPDPE exposurecorrelated with its desensitization measured on the inhibition ofadenylate cyclase. Moreover, both DPDPE-induced receptordesensitization and phosphorylation are strongly reduced whenthe mutant of GRK2, GRK2-K220R, was over-expressedsuggesting a major role of this kinase in mDOP-R desensitizationand phosphorylation. Other groups also reported that the GRKsfamily is involved in the desensitization of the MOP-R in neuronsof the nucleus raphe magnus [7] and in the KOP-R expressed inCOS-7 cells (African green monkey kidney cell line) [8]. Forinstance, in HEK293 cells, GRK2 over-expression potentiates theability of morphine to phosphorylate MOP-R and reduced itsability to inhibit adenylate cyclase [9]. By using non-specificinhibitors, we also demonstrated the involvement of one or moreGRKs in desensitization and phosphorylation of the hDOP-Rendogenously expressed in the neuroblastoma cell line SK-N-BE[10].

Beside GRKs, other kinases were demonstrated to beinvolved in opioid receptor desensitization. For instance, PKCactivation by phorbol esters promotes phosphorylation of DOP-[6] and MOP-R [11,12] but these kinases are involved neither inagonist-induced phosphorylation [6,11,12] nor receptor desen-sitization as demonstrated by PKC inhibitors or PKC-depletedcells. On the other hand, Mestek et al. [13] showed a potentiationof MOP-R desensitization when PKC was activated by phorbolester. More recently, Narita et al. [14] demonstrated the in-volvement of PKCγ in DAMGO-induced MOP-R desensitiza-tion in mice spinal cord.

As acute activation of opioid receptors promotes adenylatecyclase inhibition, it is unlikely that PKA would play a role indesensitization. However, after chronic morphine exposure, acompensatory increase in intracellular level of cAMP is observedwhich could activate PKA [15]. In vitro studies, demonstrated the

ability of morphine in promoting MOP-R phosphorylation in thepresence of catalytic subunits of PKA, but functional assays failedto demonstrate any role for this kinase in receptor desensitization[16,17].

In the human neuroblastoma SK-N-BE, we previouslyshowed a differential behaviour of the DOP-R in term ofdesensitization upon etorphine exposure compared to peptidicagonists (DPDPE and deltorphin I) [18]. So, we conducted thisstudy to investigate if this differential desensitization could relyon differential kinases recruitment by those agonists. Using twokinds of strategies (chemical kinase inhibitors and a dominantnegative mutant of GRK) to inhibit PKA, PKC, GRK2 andtyrosine kinases, we demonstrated that PKC, GRK2 andtyrosine kinases were involved in etorphine-induced desensiti-zation whereas DOP-R desensitization promoted by DPDPEand deltorphin I involves tyrosine kinases. Finally, we studiedthe differential involvement of kinases in hDOP-R phosphor-ylation state by using an anti-phospho-DOP-R Ser363 antibodyin western-blot experiments but we failed to observe anycorrelation between phosphorylation and desensitization.

2. Materials and methods

2.1. Cell culture

SK-N-BE cells, endogenously expressing hDOP-R, or SK-N-BE FLAG-tagged hDOP-R cells, over-expressing the hDOP-R fused to the FLAG motif atits carboxy-terminus tail [19], were maintained in Dulbecco's modified Eagle'smedium (DMEM), supplemented with 10% fetal calf serum (Invitrogen), 1%antibiotic-antimycotic mixture (Sigma), 2 mM L-glutamine and 1 mg/ml G418only for the SK-N-BE FLAG-tagged hDOP-R cells at 37 °C in a water-saturatedatmosphere containing 5% CO2.

2.2. Transfection of SK-N-BE cells

Transfections were realized using the calcium phosphate precipitationmethod on cells seeded in 24-well plates (functional assays) or in 25 cm2 flaskfor western-blot experiments. Cells were serum-starved for 6 h. Plasmid DNAs(pCMV5, pCMV5/GRK2, pCMV5/GRK2-K220R) were precipitated for 30 minat room temperature in an HEPES-buffered saline solution then the mixture wasadded to cells for 12–18 h (9 μg of DNA/350000 cells). Monolayers were thenwashed with phosphate-buffered saline (PBS) and incubated with freshcomplete medium for 24 h prior to functional or biochemical assays. Forwestern-blot experiments, transfected SK-N-BE cells were harvested in lysisbuffer (50 mM Tris–HCl, 0.1% (v/v) Nonidet P-40, pH=7.4) and sonicated.After determination of protein concentration using the Bradford assay (Bio-Rad), equal amounts of proteins were resolved on 10% SDS-PAGE. The over-expression of GRK2 and GRK2-K220R were revealed using anti-GRK2antibody (see GRKs immunoblotting section). Then, membranes were strippedand incubated with an anti-tubulin antibody to check the loading protein in eachlane. For functional experiments, transfected SK-N-BE cells were incubated inthe presence of [3H]adenine as described below.

In phosphorylation experiments, the SK-N-BE FLAG-tagged hDOP-R cellswere transfected either with pCMV5, pCMV5/GRK2 or pCMV5/GRK2-K220Rby nucleofection (Amaxa biosystems™) as described in details below.

2.3. Determination of cAMP accumulation

Inhibition of adenylate cyclase was determined by measuring [3H]cAMPaccumulation. Cells were seeded in 24-well plates at a density of 100000–150000 cells/well in the presence of 0.6 μCi/well [3H]adenine (PerkinElmerLife Sciences) for 12–15 h. Then, cells were pre-incubated in DMEM/20 mMHEPES with or without different kinase inhibitors [5 μMH89 (Calbiochem) for

1211N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

30 min; 1 μM Bisindolylmaleimide I (Bis I) (Calbiochem) for 30 min; 10 μMgenistein (Calbiochem) for 15 min] and incubated with or without differentagonists, in the presence of 1 mM isobutylmethylxanthine (IBMX) and 40 μMforskolin for 5 min at 37 °C. The reaction was stopped by removing the mediumand by addition of 5% (w/v) trichloroacetic acid. [3H]cAMP was separated bychromatography on acid alumina columns, mixed with 8 ml of scintillationmixture (PicoFluor-40, Packard), before assaying in a scintillation counter(Packard). Percentage of inhibition was calculated according to the followingformula: (1-(cpm agonist−cpm basal) / (cpm forskolin−cpm basal))×100,where cpm basal was determined in a medium containing only IBMX andcpm forskolin in the presence of forskolin+IBMX. All experiments were carriedout at least three times in triplicate.

Agonist concentrations used in functional assays produce almost the sameand maximum inhibitory effect on adenylate cyclase. For desensitization, cellswere pretreated with different conditions giving a maximal and similardesensitization [18] (see table below).

Agonist

Concentration Pretreatment time

etorphine

100 nM 60 min DPDPE 100 nM 30 min deltorphin I 10 nM 30 min

2.4. GRKs immunoblotting

GRKs expression was studied in SK-N-BE cells and human tissues wereused as control. Human brain and liver were obtained from cancer surgery andcorresponded to the adjacent tissue of the tumor. Homogenization of tissues wasperformed at 4 °C in 10 mM Tris–HCl, pH=7.4 using a Potter–Elvejhem tissuegrinder. Homogenates were centrifuged at 100 000 g for 35 min at 4 °C, then theresulting supernatant (cytosolic fraction) was frozen at −80 °C and the pelletswere suspended by sonication in the same buffer and transferred in 1.5 mlmicrofuge tubes. The crude membrane fraction was obtained after a secondcentrifugation at 20000 g for 20 min at 4 °C and frozen at −80 °C.

Sub-confluent SK-N-BE were harvested in 10 mM Tris–HCl, 1 mM EDTA,pH=7.4 and sonicated. After a centrifugation at 20000 g for 20 min at 4 °C, theresulting supernatant was frozen at −80 °C (cytosolic fraction) and the pelletwas suspended in the same buffer and centrifuged in the same conditions. Theresulting pellet (membrane fraction) was kept at −80 °C until use.

Equal amounts of protein (cytosolic and membrane fractions) were resolvedon 10% SDS-PAGE. Proteins were next electrophoretically transferred ontonitrocellulose membranes. Identification of various GRK isoforms was realizedusing polyclonal anti-GRK2, 3, 4, 5 and 6 antibodies (1/1000, Santa CruzBiotechnology). Immuno-reactive bands were detected on X-Omat films(Kodak) using chemiluminescence kit (PerkinElmer Life Sciences).

2.5. hDOP-R phosphorylation on Ser363

Phosphorylation experiments were conducted on SK-N-BE FLAG-taggedhDOP-R cells [19] since a too weak immuno-labelling was observed inpreliminary studies using SK-N-BE cells endogenously expressing opioidreceptors. Cells were grown until ∼80% of confluence and exposed or not(naïve) in serum-free medium to various ligands for 15 min at 37 °C (100 nMetorphine, 100 nM DPDPE, 10 nM deltorphin I, 1 μM natrindole alone or incombination with the different agonists). When using the PKC inhibitor, cellswere pretreated or not in the presence of 1 μMBis I for 30 min and then exposedor not (naïve) to the different agonists for 15 min at 37 °C. In GRK experiments,cells were transfected by nucleofection (Amaxa biosystems™) either with thepCMV5 alone (mock), the wild type GRK2 or the dominant negative mutantGRK2-K220R using the A20 program in the L buffer as recommended by themanufacturer. We opted for this technology to improve the transfection yield inpreviously transfected cells by comparison with the calcium phosphateprecipitation method. These conditions ensure a sufficient expression of theGRK2 and the GRK2-K220R to observe a putative role of these proteins onreceptor phosphorylation.

Then, cells were harvested in lysis buffer, sonicated, and proteinconcentrations were determined using the Bradford assay. Equal amounts of

proteins were loaded on 10% SDS-PAGE and transferred onto nitrocellulosemembranes. To make an accurate comparison of the phosphorylation levelbetween the different conditions of treatment, we also checked the proteinloading by staining the nitrocellulose using Ponceau S. Phosphorylation of thehDOP-R on Ser363 was assessed by using specific rabbit antibody for phospho-DOP-R Ser363 (1/1 000, Cell Signaling). After incubation with a horseradishperoxidase-conjugated goat anti-rabbit secondary antibody, immunoreactivebands were revealed using the enhanced chemiluminescence system (Super-Signal West Pico Chemiluminescent Substrate, Pierce).

Quantification of western-blot was realized by densitometric analysis usingBio1D software (Vilbert Lourmat, France).

3. Results

We previously showed that one or more GRKs wereinvolved in etorphine-induced desensitization and phosphor-ylation of hDOP-R in SK-N-BE cells [10]. This work wasconducted to precisely identify the kinases responsible fordesensitization and phosphorylation of hDOP-R induced bypeptidic agonists, DPDPE and deltorphin I, and the alkaloidagonist etorphine.

3.1. PKA is not involved in hDOP-R desensitization

We first determined if PKA could be involved in hDOP-Rdesensitization, using H89, a potent and selective PKA inhibitor[20]. In absence of H89 pretreatment, the three opioid agonistswere able to inhibit adenylate cyclase from ∼30% to ∼50%(Fig. 1). These inhibitory effects were significantly diminishedby H89 pretreatment (5 μM, 30 min) only for etorphine (49.8±1.8% vs 37.3±2,1%, t-test, Pb0.01). When cells were treatedfor 60 min with etorphine or 30 min with DPDPE or deltorphinI, the inhibition of adenylate cyclase was strongly reduced oreven a slight stimulation of forskolin-activated adenylatecyclase was evidenced. Those data demonstrated hDOP-Rdesensitization (Fig. 1). This desensitization was not influencedby H89 treatment even in the case of etorphine ruling out therole of PKA in hDOP-R desensitization in the SK-N-BE cells.

3.2. PKC reduced etorphine-induced desensitization

As PKC was shown to be involved in opioid receptor-induced desensitization in different cellular models as well as invivo (see introduction), we evaluated the inhibition of thiskinase family by a broad spectrum PKC inhibitor, the Bis I[21,22]. When cells were pretreated or not with Bis I, thedifferent opioid agonists promoted an adenylate cyclaseinhibition from ∼40% to ∼60% indicating no PKC involve-ment in their inhibitory action on the cAMP pathway in naïvecells. As expected, chronic stimulation by the various agoniststriggers hDOP-R desensitization (Fig. 2) as observed in theFig. 1. When cells were pretreated with the PKC inhibitor(1 μM, 30 min), we were able to observe a significant reductionof etorphine-induced hDOP-R desensitization by more than30% (83.5±6.7% vs 50.8±3.1%, t-test, Pb0.05; Fig. 2). Incontrast, in Bis I-treated cells, DPDPE and deltorphin I wereable to promote hDOP-R desensitization to a similar extent as inBis I-untreated cells (Fig. 2).

Fig. 1. PKA is not involved in hDOP-R desensitization. SK-N-BE cells were preincubated (+) or not (−) for 30 min with 5 μMH89 and pretreated (hatched bars) or not(solid bars) with etorphine (Eto) (100 nM, 60 min), DPDPE (DP) (100 nM, 30 min) or deltorphin I (Delto) (10 nM, 30 min). At the end of the pretreatment, wedetermined the ability of each agonist to inhibit cAMP accumulation. Data represent means±S.E.M of 3 to 4 different experiments performed in triplicate. Statisticalsignificance was determined by Student's t-test (⁎⁎Pb0.01, H89 non-treated vs H89-pretreated cells).

1212 N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

Fig. 2. Role of PKC in etorphine-induced hDOP-R desensitization. SK-N-BE cells were preincubated (+) or not (−) for 30 min with 1 μMBis I and pretreated (hatchedbars) or not (solid bars) with etorphine (Eto) (100 nM, 60 min), DPDPE (DP) (100 nM, 30 min) or deltorphin I (Delto) (10 nM, 30 min). At the end of the pretreatment,we determined the ability of each agonist to inhibit cAMP accumulation. Data represent means±S.E.M of 3 to 4 different experiments performed in triplicate.Statistical significance was determined by Student's t-test (⁎⁎Pb0.01, Bis I non-treated vs Bis I-pre-treated cells).

1213N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

Fig. 3. GRK2 is the only GRK expressed in the SK-N-BE cell line. The crudemembrane fraction of human brain (HB) and liver (HL), cytosolic (BEc) ormembrane (BEm) fractions prepared from SK-N-BE cells were prepared asdescribed in materials and methods. Proteins were resolved on 10% SDS-PAGEand the presence of GRKs was revealed using specific antibodies against eachisoform.

1214 N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

3.3. GRK2 is involved in etorphine-induced desensitization

3.3.1. Expression pattern of GRKs in SK-N-BE cellsUsing non-specific GRKs inhibitors (heparin, Zn2+), we

previously demonstrated the involvement of this kinase familyin etorphine-induced desensitization of hDOP-R in SK-N-BEcells [10]. So, in a first set of experiments, we sought todetermine which GRK isoforms were expressed in both cy-tosolic (BEc) and membrane (BEm) fractions of this cell line.A western-blot approach was used to screen SK-N-BE cells(BE) for the presence of GRK2, 3, 4, 5 and GRK6 using eitherthe humain brain (HB) or liver (HL) as positive controls [23]. Itshould be noted that the presence of GRK1 and 7 were notinvestigated as these kinases are specifically found in retinaand pineal gland [24]. Immunoblots shown in Fig. 3 dem-

Fig. 4. GRK2 is involved in etorphine-induced hDOP-R desensitization. A.Whole cell extracts of SK-N-BE cells transfected with pCMV5, pCMV5/GRK2or pCVM5/GRK2-K220R were prepared as described in materials and methodsand probed for GRK2 and tubulin. B. Cells transfected with pCMV5 (Mock),pCMV5/GRK2 (GRK2) or pCMV5/GRK2-K220R (K220R) were pretreated(hatched bars) or not (solids bars) with etorphine (Eto) (100 nM, 60 min),DPDPE (DP) (100 nM, 30 min) or deltorphin I (Delto) (10 nM, 30 min). At theend of the pretreatment, we determined the ability of each agonist to inhibitcAMP accumulation. Data represent means±S.E.M of 3 to 4 differentexperiments performed in triplicate. Statistical significance was determined byANOVA followed by Bonferroni's test (⁎⁎⁎ Pb0.001, Mock- vs GRK2-transfected cells; ### Pb0.001 Mock- vs K220R-transfected cells).

onstrated that all the GRKs assayed were expressed either inthe brain (i.e. GRK2, 4 and 6) or in liver (GRK3 and 5).Regarding the SK-N-BE cells, we studied the expression ofGRKs both in cytosolic and membrane fractions and obtained

Fig. 5. Role of tyrosine kinases in hDOP-R desensitization. SK-N-BE cells were preincubated (+) or not (-) for 15 min with 10 μM genistein and pretreated (hatchedbars) or not (solid bars) with etorphine (Eto) (100 nM, 60 min), DPDPE (DP) (100 nM, 30 min) or deltorphin I (Delto) (10 nM, 30 min). At the end of the pre-treatment,we determined the ability of each agonist to inhibit cAMP accumulation. Data represent means±S.E.M of 3 to 4 different experiments performed in triplicate.Statistical significance was determined by Student's t-test (⁎⁎Pb0.01, ⁎Pb0.05, genistein non-treated vs genistein-pretreated cells).

1215N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

Fig. 6. hDOP-R phosphorylation on Ser363 induced by the different agonists. TheFLAG-tagged hDOP-R SK-N-BE cells were pretreated or not with 10 μMnaltrindole for 5 min (panel A) or 1 μMBis I for 30 min (panel B) then exposed ornot for 15 min in the presence of 100 nM etorphine or DPDPE or 10 nMdeltorphin I. In the panels C and D, the FLAG-tagged hDOP-R SK-N-BE cellswere transfected by nucleofection either with the pCMV5 alone (Mock), the wildtype GRK2 or the dominant negative mutant GRK2-K220R as described inmaterials and methods. 48 h later, cells were exposed or not for 15 min in thepresence of 100 nM etorphine or DPDPE or 10 nM deltorphin I. At the end ofagonist exposure, whole cells extracts were prepared as described in materials andmethods, resolved on a 10% SDS-PAGE and hDOP-R phosphorylation wasdetected using anti-phospho-DOP-R Ser363 antibody. Panel E, western-blotobtained in mock, GRK2 or GRK2-K220R transfected cells were quantified bydensitometry using Bio1D software. Results are expressed as the % of control andrepresent means±S.E.M of 3 to 6 different experiments. Statistical significancewas determined by ANOVA followed by Bonferroni-Dunn test (⁎, Pb0.05).

1216 N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

similar results. As depicted in the Fig. 3, the GRK2 is the singleGRK member detected in this cell line.

3.3.2. Functional assays in cells over-expressing GRK2 or itsdominant negative mutant GRK2-K220R

In order to characterize the role of GRK2 in hDOP-Rdesensitization, transient transfections were conducted with thewild type GRK2 or its dominant negative mutant (GRK2-K220R)in SK-N-BE cells. Cells were transiently transfected using thecalcium phosphate precipitation method with pCMV5, pCMV5/GRK2 or pCMV5/GRK2-K220R and first, over-expression ofthese proteins was checked by western-blot. Fig. 4A showed thatcells transfected with pCMV5/GRK2 and pCMV5/GRK2-K220Rdisplayed a strong increase in the immuno-reactivity againstGRK2 of about∼2–4 folds compared to pCMV5-transfected cellsand when referred to the tubulin immunoreactivity.

Regarding the functional assays, in cells transfected with theempty plasmid (Mock), the three opioid agonists were able toinhibit the adenylate cyclase by ∼50% and promoted adesensitization by ∼70 to 95% after 30 min exposure forDPDPE and deltorphin I or 60 min for etorphine (Fig. 4B) asobserved in non-transfected SK-N-BE cells (Figs. 1 and 2). Incells over-expressing the wild type GRK2, the inhibitory effect ofetorphine on adenylate cyclase was dramatically reducedcompared to mock-transfected cells since almost no inhibitionon cAMP was measured. For the two peptidic agonists, GRK2over-expression did not modify their ability to inhibit theadenylate cyclase in naïve cells as well as in agonist-pretreatedcells (Fig. 4B). The over-expression of the dominant negativemutant of GRK2, GRK2-K220R, did not modify the inhibition ofadenylate cyclase produced by the different agonists in naïvecells but totally impairs hDOP-R desensitization promoted byetorphine pretreatment (Fig. 4B). In contrast, in cells over-expressing the GRK2-K220R mutant, hDOP-R desensitizationinduced by DPDPE and deltorphin I exposure reached the samelevel as in SK-N-BE cells transfected with the plasmid alone(Fig. 4B). Taken altogether, these results demonstrated theselective involvement of GRK2 in hDOP-R desensitizationpromoted by etorphine but neither with DPDPE nor deltorphin I.

3.4. Role of tyrosine kinases in hDOP-R desensitization

As PKA, PKC and GRK2were not found to be responsible forhDOP-R desensitization promoted by peptidic agonists, weinvestigated the role tyrosine kinases in this process. Moreover,several studies reported a role for these kinases in DOP-Rphosphorylation [25], KOP-R signalling [26] as well as in MOP-R down-regulation [27]. In order to investigate the role of tyrosinekinases in hDOP-R desensitization, SK-N-BE cells werepretreated or not using genistein (10 μM; 15 min), known as abroad spectrum tyrosine kinases inhibitor [28]. As previouslyobserved in Figs. 1, 2 and 4B and in absence of genistein pre-treatment, opioid agonists induced a cAMP accumulationinhibition by ∼50%. After genistein pretreatment, we noticed asignificant reduction of the ability of deltorphin I to inhibit cAMPaccumulation (45.3±3% vs 28.1±2.5%, t-test, Pb0.01) withoutany effect on the other agonists (Fig. 5). As expected, agonist

1217N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

pretreatment leads to hDOP-R desensitization as revealed by areduction of their inhibitory effects on adenylate cyclase (Fig. 5).After sustained agonists exposure, genistein pretreatment wasshown to significantly increase the ability of the three opioidligands to inhibit cAMP accumulation (etorphine : 24.4±3% vs41.2±2.4%, t-test, Pb0.01; DPDPE : 5.5±3.8% vs 28.2±1.4%,t-test, Pb0.01; deltorphin I : 2.6±2.4% vs 10.1±1.3, t-test,Pb0.05) reflecting a decrease in hDOP-R desensitization (Fig. 5).

3.5. hDOP-R phosphorylation

Data obtained in functional experiments suggest that etorphine-induced hDOP-R desensitization would require GRK2, PKC and

Fig. 7. Model for the differential hDOP-R desensitization and phosphorylationupon etorphine and peptidic agonists activation. Upon etorphine activation,hDOP-R phosphorylation on Ser363 does not involve GRK2 or PKC but a non-identified kinase. However, GRK2 has a major role in desensitization on thecAMP pathway with minor roles of PKC and tyorine kinases. These differentkinases would phosphorylate hDOP-R on different residues or other signalingmolecules. When hDOP-R is activated by DPDPE or deltorphin I (peptidicopioid agonists), the receptor is partially phosphorylated on Ser363 by theGRK2 but not PKC. However, neither GRK2 nor PKC mediate opioid receptordesensitization but rather a tyrosine kinase and other non-identified kinases.

tyrosine kinases. In contrast, DPDPE and deltorphin I promotehDOP-R desensitization via tyrosine kinases. The next step was toanswer to the following question: is the differential hDOP-Rdesensitization due to a difference in phosphorylation of the opioidreceptor. The hDOP-R phosphorylation state was determined in theSK-N-BE cells over-expressing the FLAG-tagged hDOP-R byusing the anti-phospho-DOP-R Ser363 antibody since initial studiesperformed in SK-N-BE cells showed a too weak immuno-labelingfor an accurate phosphorylation level determination. In the absenceof receptor activation, almost no significant labeling was visualized(Fig. 6A, naïve). When hDOP-R were activated by the differentagonists during 15 min, we were able to observe a strong immuno-reactivity at about 40 kDa by 2.5 to 3.1 fold over basal conditions(Fig. 6A and E, Eto 15, DP 15 and Del 15). This agonist-mediatedphosphorylation of the S363 residue was completely blocked by co-administration with the DOP-R antagonist naltrindole at 10 μM(Fig. 6A,+10 μM naltrindole). When PKC were inactivated by aprior treatment during 30 min with 1 μMBis I (same conditions asdescribed in functional experiments), no significant modification ofthe ability of the different agonists to stimulate hDOP-Rphosphorylation was observed (Fig. 6B). In a last set ofexperiments, to investigate the role of GRK2 in the agonist-induced receptor phosphorylation, the FLAG-tagged hDOP-R SK-N-BE cells were transiently transfected eitherwith the empty vector(mock), the wild type kinase or its dominant negative mutantGRK2-K220R. As depicted in the Fig. 6C and E, the strong over-expression of the bovine GRK2 did not modify the etorphine andthe DPDPE-induced hDOP-R phosphorylation. Surprisingly, inthese conditions but in the presence of deltorphin I, we ratherobserved a significant decrease by 45% of the hDOP-Rphosphorylation (Fig. 6C and E). After ensuring that the dominantnegative mutant GRK2-K220R was expressed to a greater extentover the endogenous GRK2, we examined the phosphorylationstate of the hDOP-R. We showed that this mutant efficientlyimpaired the Ser363 phosphorylation induced by DPDPE anddeltorphin I but not by etorphine (Fig. 6D and E).

4. Discussion

Opioid receptors, and especially MOP-R, desensitization wasshown to be crucial in the development of tolerance. Indeed, byusing knock-out mouse lacking arrestin-3, Bohn and collaborators[4] impairedMOP-R desensitization and consequently tolerance tochronic morphine treatment. As described for most of the GPCRs,opioid receptor phosphorylation is closely associated withdesensitization. A lot of work has been dedicated to identify thekinases responsible for opioid receptors phosphorylation. Pre-viously, we described a difference in hDOP-R desensitizationbetween etorphine, on one hand, and DPDPE and deltorphin I, onthe other hand [18] that could be related to ability of these agoniststo promote a different hDOP-R trafficking [19] but also a differentDOP-R phosphorylation. In order to gain insight in the molecularmechanisms of hDOP-R desensitization and to know if a putativedifference in kinase recruitment could be at the origin of thisdifferential desensitization, we determined which kinase(s) couldbe involved in hDOP-R desensitization and phosphorylationinduced by etorphine, DPDPE and deltorphin I.

1218 N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

In the present paper, we clearly demonstrated that etorphinetriggersDOP-R desensitization by recruiting different kinases, theGRK2, PKC and tyrosine kinases. In host cells, Pei et al. [6] werethe first to report that GRK2 mediated both phosphorylation anddesensitization of the mouse DOP-R upon DPDPE exposurewhile GRK5 over-expression only promoted phosphorylation ofthe receptor after agonist exposure suggesting a primary role ofonly GRK2 in desensitization. In contrast, the mDOP-R,endogenously expressed in the NG108-15 neuroblastoma×ratglioma hybrid cells, was shown to undergo desensitization uponDPDPE exposure after over-expression of GRK6 but not GRK2[29]. Such discrepancies could be due to different expression levelof either opioid receptors or GRKs between these cellular models.In the SK-N-BE cells, when we over-expressed the GRK2, nomore inhibitory effect of etorphine was observed in naïve cells.Such an artificial situation would increase the affinity between thereceptor and this GRK promoting a rapid phosphorylation anddesensitization. A more “physiological” approach to delineate therole of GRK2 in etorphine-induced hDOP-R desensitization wasconducted using the dominant negative mutant GRK2-K220R.When over-expressed, this mutant protein competed with theendogenous GRK2 and totally impaired the desensitizationinduced by etorphine. While we observed partial inhibition ofdesensitization in the presence of PKC and tyrosine kinaseinhibitors, this latter result clearly indicates that GRK2 is themajor kinase involved in hDOP-R desensitization promoted byetorphine. The data obtained in the present study are in goodagreement with previous results suggesting the role of a GRKwhen using non-specific GRK inhibitors [10]. Opioid receptorphosphorylation by GRKs is not sufficient to promote desensi-tization but requires additional partners such as arrestins.Recently, we demonstrated that over-expression of arrestin-2 inthe SK-N-BE cells strongly potentiated desensitization uponetorphine exposure [30] a result that comforts the GRK2activation by the alkaloid agonist. Surprisingly, the peptidicagonists, DPDPE and deltorphin I, which induced a rapid andprofound hDOP-R desensitization were unable to mobilize theGRK2 as shown by the lack of effect on adenylate cyclaseinhibition when this kinase or its dominant negative mutant wereover-expressed. These results would suggest that the binding ofalkaloid and peptidic agonists would rely on different regions ofthe DOP-R which consequently leads to a different conforma-tional changes exposing or not some of the putative phosphoryla-tion sites. In the case of etorphine, these amino acidswould be firstphosphorylated by the GRK2 while the hDOP-R activated byDPDPE or deltorphin I would be phosphorylated by tyrosinekinases. The mDOP-R was previously shown to be tyrosine-phosphorylated by different agonists including DPDPE andetorphine but in this study, only the role of this kinase on receptorinternalization and ERK1/2 activation were addressed [25]. In thepresent study, we can only speculate about the tyrosinephosphorylation of the hDOP-R but we cannot rule out thatother signaling proteins would be tyrosine-phosphorylated. Toour knowledge, our study is the first paper demonstrating that thedifferential desensitization of the hDOP-R previously observed[18] is linked to different molecular mechanisms. Concerning theMOP-R and by using a large panel of opioid ligands, Yu and

collaborators [31] reported a parallel between the phosphorylationlevel and desensitization a result that could have suggested theinvolvement of different kinases. This is only recently thatJohnson et al. [32] demonstrated that morphine and DAMGO,alkaloid and peptidic opioid agonists respectively, both inducedMOP-R desensitization on the G protein-coupled inwardlyrectifying potassium channel (GIRK) currents but DAMGO-induced desensitization is GRK2-dependent, whereas morphine-induced desensitization requires a PKC.

As activation of opioid receptor generally inhibits adenylatecyclase and thus decreases cAMP level, PKA is unlikely to beinvolved in desensitization. However, chronic exposure tomorphine results in a compensatory up-regulation of adenylatecyclase activity increasing cAMP level and therefore stimulatesPKA activity [15]. In vitro experiments showed thatMOP-R couldbe phosphorylated by the catalytic subunit of PKA in the presenceof morphine or levorphanol but not DAMGO or DADLE;however, this kinase was not involved in MOP-R desensitization[16]. In agreement with these data, our results showed no effect ofthe PKA inhibitor, H89, on desensitization induced by the threeagonists. In contrast, we observed that inhibition of PKC reducedselectively the etorphine-induced hDOP-R desensitization. Pre-viously, we showed that hDOP-R activation by etorphine in theSK-N-BE cells produced an increase of intracellular calcium [33]that could activate PKC [34] suggesting a role for this kinase inhDOP-R signaling. This is in contradictionwithmany authors whofound that even if PKC could phosphorylate opioid receptors[6,11,12], it is neither involved in agonist-induced phosphorylation[6,11,12] nor desensitization [12,35]. On the other hand, Narita etal. [36] showed that PKC could play a role in tolerance todeltorphin II-induced antinociceptive effect suggesting a role forthis kinase in DOP-R homologous desensitization. The role ofPKC in etorphine-induced hDOP-R desensitization is also in goodagreement with a previous paper of our laboratory in which wedescribed heterologous desensitization between DOP-R, α2-adrenergic and D2-dopaminergic receptors [37]. Indeed, whenSK-N-BE cells were pretreated with etorphine during 12-15 h, wewere able to observe a decrease of the inhibitory action of arterenoland 2-bromo-α-ergocryptine on the adenylate cyclase suggesting arole of second-messenger-dependent kinases.

Beside hDOP-R desensitization, we noted that PKA andtyrosine kinase inhibitors were also able to modulate theinhibitory actions of etorphine and deltorphin I on the adenylatecyclase in naïve cells. Indeed, pretreatment of SK-N-BE cellswith these kinase inhibitors was shown to reduce the cAMPinhibition produced by etorphine and deltorphin I without anyeffect for DPDPE. This suggests that these kinases would beone of the intra-cellular regulators of the basal signaling activityof hDOP-R. Phosphorylation of opioid receptor, signaltransduction proteins as G-proteins or regulators of signalingsuch as GRKs might account for these effects. The MOP- [38]and the DOP-R [10] were shown to be phosphorylated in thebasal state in absence of agonist exposure. While the tyrosinephosphorylation of DOP- [25] and KOP-R [26] appears to be apositive regulators of their activity, the dephosphorylation oftyrosine of MOP-R was shown to potentiate DAMGO-inducedpotassium channel activity [39]. So, we can suppose that when

1219N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

SK-N-BE cells are pretreated with genistein, the DOP-R is nottyrosine-phosphorylated which subsequently attenuates its sig-naling. However, the decrease of cAMP inhibition by genisteinwas only observed for deltorphin I but neither with etorphine norDPDPE supporting the hypothesis that each agonist mayrecognize specific spatial conformation of the DOP-R. Such aneffect of genistein on DOP-R has already been reported but withanother opioid agonist, [D-Thr2-Leu5-Thr6]enkephalin (DTLET)and on the activation of ERK1/2 pathway [25] so it is difficult tocompare such data with our study. A decrease of etorphine-induced cAMP inhibition was also noted when SK-N-BE cellswere pretreated with the PKA inhibitor H89 without any effect onthe inhibitory action of DPDPE or deltorphin I. As discussedabove, the reduced ability of etorphine to inhibit adenylatecyclase would result from modification of phosphorylation ofeither the receptor or proteins involved in its regulation. H89 waspreviously shown to impair p38 activation by DPDPE butindependently of modification of DOP-R phosphorylation level[40]. As PKAwas reported to phosphorylate GRK2 allowing itsinteraction via Gβγ and GPCR phosphorylation [41], we wouldhave expected a potentiation of etorphine-induced adenylatecyclase inhibition rather than a decrease as observed. PKA wasalso shown to negatively regulate interactions between GRK2and arrestins [42]. So, we can speculate that the basal activity ofadenylate cyclase would allow the GRK2 phosphorylation andthat H89 pretreatment would activate this kinase leading topartial opioid receptor desensitization. This hypothesis issupported by the lack of effect of H89 on DPDPE and deltorphinI adenylate cyclase inhibition since we demonstrated in thepresent study that peptidic agonist-induced hDOP-R desensitiza-tion was not GRK2-dependent.

The last part of this work was to correlate our functionaldata with the level of hDOP-R phosphorylation between thedifferent opioid agonists. We decided to study the DOP-Rphosphorylation at Ser363 since previous papers report thisamino acid as the primary phosphorylation site upon DPDPEactivation [43,44]. These experiments were conducted usingthe specific antibody directed against the phospho-Ser363 ofthe DOP-R that was previously validated in the CHO cellsover-expressing the hDOP-R [45]. We observed that thedifferent opioid agonists promoted almost a similar phosphor-ylation level and that this phosphorylation was specific ofhDOP-R as shown by the blockade with naltrindole, a DOP-Rselective antagonist. As expected, the PKC inhibitor Bis I wasunable to impair hDOP-R phosphorylation upon opioidagonists exposure since Xiang et al. [46] clearly demonstratedthat PKC activation by phorbol ester but not by DPDPE led tothe phosphorylation of the Ser344 and not the Ser363 of themDOP-R. When over-expressing the dominant negativemutant of GRK2, we noted a partial decrease in hDOP-Rphosphorylation for peptidic agonists but not etorphine. Thisreduction did not reach a statistical significance for etorphinewhile we clearly over-expressed the GRK2-K220R indicatinga potent competition between the endogenous GRK2 and themutant. These data suggest either that the mutant cannotcompletely impair the GRK2 activity or that the Ser363 is, atleast, partially phosphorylated by the GRK2 but also by

another kinase upon DPDPE and deltorphin I exposure. This isin contrast with data obtained by the groups of Pei and Lawwho established that both Thr358 and Ser363 were the GRK2phosphorylation sites [43,44]. However, those studies wereconducted with the mDOP-R over-expressed in the HEK cellswhile the present study was realized in the neuroblastoma cellsSK-N-BE with the hDOP-R. Comparison of the Ser/Thrresidues present in the carboxy-terminal tail of both murineand human DOP-R reveals substantial differences which areprobably critical. Indeed, we observed that the human receptorcontains the Ser343 and 346 which are absent in the mousereceptor and that the mouse receptor contains the Thr353 whichis lacking in the human receptor. The Ser cluster between 343and 346 in the hDOP-R but not in the mouse receptor wouldprobably be the target of kinases such as GRKs. However, toour knowledge there is no site-directed mutagenesis dataavailable for the hDOP-R supporting our hypothesis. Differ-ences in the Ser/Thr composition at the carboxy-terminal tail ofthe human and the mouse DOP-R would be crucial as evidencedfor the KOP-R. The ability of the κ agonist U50,488 tophosphorylate the human but not the rat KOP-R was demon-strated to rely on the presence of the Ser358 in the human receptor,which is absent the rat [47]. This demonstrates that even smalldifferences in amino acid composition would compromiseextrapolation of data between different species. So, we speculatethat while the Ser363 is phosphorylated in the human DOP-Rupon agonist activation, additional residues would be GRK2-phospho acceptor sites such as Ser343 or 346. Moreover, when theGRK2 was over-expressed in the SK-N-BE cells, instead of anincrease in hDOP-R phosphorylation, we observed either nomodification upon etorphine and DPDPE or a significantdecrease in the presence of deltorphin I. We hypothesize thatthe Ser343 or 346 would be the major phosphorylation sites ofGRK2 and that the phosphorylation of these latter amino acidswould reduce the accessibility of the GRK2 on Ser363 andconsequently, leading to a decrease of phosphorylation level atthis site. This supposition is supported by the lack of correlationbetween functional and phosphorylation data. Whereas the over-expression of both GRK2 and its dominant negative mutantimpair the etorphine-induced cAMP inhibition and the hDOP-Rdesensitization respectively, no modification on the phospho-Ser363 level was observed.

In conclusions, we showed that the differential desensitiza-tion of hDOP-R by etorphine, on one hand, and by DPDPE anddeltorphin I, on the other hand, previously observed on thecAMP pathway, would rely on the involvement of differentkinases: the GRK2, PKC and tyrosine kinases for etorphine andat least tyrosine kinases for DPDPE and deltorphin I (Fig. 7).While the Ser363 of the DOP-R is phosphorylated to a similarextent for the different opioid agonists, it is probably not themajor GRK2 phosphorylation site of the hDOP-R.

Acknowledgment

We thank Prof. S. Cottechia (Toxicology and PharmacologyInstitute, University of Lausanne, Switzerland) for generouslyproviding pCMV5/GRK2 and K220R.

1220 N. Marie et al. / Cellular Signalling 20 (2008) 1209–1220

References

[1] H. Fields, Nat. Rev. Neurosci. 5 (7) (2004) 565.[2] H.H. Loh, P.L. Tao, A.P. Smith, Synapse 2 (4) (1988) 457.[3] F. Noble, B.M. Cox, Br. J. Pharmacol. 117 (1) (1996) 161.[4] L.M. Bohn, R.R. Gainetdinov, F.T. Lin, R.J. Lefkowitz, M.G. Caron,

Nature 408 (6813) (2000) 720.[5] R.R. Gainetdinov, R.T. Premont, L.M. Bohn, R.J. Lefkowitz, M.G. Caron,

Annu. Rev. Neurosci. 27 (2004) 107.[6] G. Pei, B.L. Kieffer, R.J. Lefkowitz, N.J. Freedman, Mol. Pharmacol. 48

(2) (1995) 173.[7] A.H. Li, H.L. Wang, J. Neurochem. 77 (2) (2001) 435.[8] K. Raynor, H. Kong, J. Hines, G. Kong, J. Benovic, K. Yasuda, G.I. Bell,

T. Reisine, J. Pharmacol. Exp. Ther. 270 (3) (1994) 1381.[9] J. Zhang, S.S. Ferguson, L.S. Barak, S.R. Bodduluri, S.A. Laporte, P.Y.

Law, M.G. Caron, Proc. Natl. Acad. Sci. U. S. A. 95 (12) (1998) 7157.[10] A. Hasbi, J. Polastron, S. Allouche, L. Stanasila, D. Massotte, P. Jauzac,

J. Neurochem. 70 (5) (1998) 2129.[11] R. El Kouhen, O.M. Kouhen, P.Y. Law, H.H. Loh, J. Biol. Chem. 274 (14)

(1999) 9207.[12] L. Zhang, Y. Yu, S. Mackin, F.F. Weight, G.R. Uhl, J.B. Wang, J. Biol.

Chem. 271 (19) (1996) 11449.[13] A. Mestek, J.H. Hurley, L.S. Bye, A.D. Campbell, Y. Chen, M. Tian, J.

Liu, H. Schulman, L. Yu, J. Neurosci. 15 (3 Pt 2) (1995) 2396.[14] M. Narita, H. Mizoguchi, H. Nagase, T. Suzuki, L.F. Tseng, J. Neurosci. 21

(11) (2001) 3715.[15] V.C. Yu, W. Sadee, J. Pharmacol. Exp. Ther. 245 (1) (1988) 350.[16] S. Chakrabarti, P.Y. Law, H.H. Loh, J. Neurochem. 71 (1) (1998) 231.[17] Z. Wang, W. Sadee, Eur. J. Pharmacol. 389 (2–3) (2000) 165.[18] S. Allouche, M. Roussel, N. Marie, P. Jauzac, Eur. J. Pharmacol. 371 (2–3)

(1999) 235.[19] N. Marie, I. Lecoq, P. Jauzac, S. Allouche, J. Biol. Chem. 278 (25) (2003)

22795.[20] T. Chijiwa, A. Mishima, M. Hagiwara, M. Sano, K. Hayashi, T. Inoue, K.

Naito, T. Toshioka, H. Hidaka, J. Biol. Chem. 265 (9) (1990) 5267.[21] M.T. Romero-Avila, C.F. Flores-Jasso, J.A. Garcia-Sainz, Biochem. J. 368

(Pt 2) (2002) 581.[22] D. Toullec, P. Pianetti, H. Coste, P. Bellevergue, T. Grand-Perret, M.

Ajakane, V. Baudet, P. Boissin, E. Boursier, F. Loriolle, et al., J. Biol.Chem. 266 (24) (1991) 15771.

[23] K. Palczewski, Eur. J. Biochem. 248 (2) (1997) 261.[24] T. Metaye, H. Gibelin, R. Perdrisot, J.L. Kraimps, Cell. Signal. 17 (8)

(2005) 917.

[25] H.K. Kramer, M.L. Andria, D.H. Esposito, E.J. Simon, Biochem.Pharmacol. 60 (6) (2000) 781.

[26] S.M. Appleyard, J.P. McLaughlin, C. Chavkin, J. Biol. Chem. 275 (49)(2000) 38281.

[27] Y. Pak, B.F. O'Dowd, J.B. Wang, S.R. George, J. Biol. Chem. 274 (39)(1999) 27610.

[28] T. Akiyama, J. Ishida, S. Nakagawa, H. Ogawara, S. Watanabe, N. Itoh, M.Shibuya, Y. Fukami, J. Biol. Chem. 262 (12) (1987) 5592.

[29] J. Willets, E. Kelly, Eur. J. Pharmacol. 431 (2) (2001) 133.[30] B. Aguila, M. Roussel, P. Jauzac, S. Allouche, Brain Res. 1114 (1) (2006)

11.[31] Y. Yu, L. Zhang, X. Yin, H. Sun, G.R. Uhl, J.B. Wang, J. Biol. Chem. 272

(46) (1997) 28869.[32] E.A. Johnson, S. Oldfield, E. Braksator, A. Gonzalez-Cuello, D. Couch, K.

J. Hall, S.J. Mundell, C.P. Bailey, E. Kelly, G. Henderson, Mol. Pharmacol.70 (2) (2006) 676.

[33] S. Allouche, J. Polastron, P. Jauzac, J. Neurochem. 67 (6) (1996) 2461.[34] P.J. Parker, J. Murray-Rust, J. Cell Sci. 117 (2) (2004) 131.[35] H. Morikawa, K. Fukuda, H. Mima, T. Shoda, S. Kato, K. Mori, Br. J.

Pharmacol. 123 (6) (1998) 1111.[36] M. Narita, H. Mizoguchi, J.P. Kampine, L.F. Tseng, Br. J. Pharmacol. 118

(7) (1996) 1829.[37] N. Namir, J. Polastron, S. Allouche, A. Hasbi, P. Jauzac, J. Neurochem. 68

(4) (1997) 1764.[38] J.R. Arden, V. Segredo, Z. Wang, J. Lameh, W. Sadee, J. Neurochem. 65

(4) (1995) 1636.[39] J.P. McLaughlin, C. Chavkin, Mol. Pharmacol. 59 (6) (2001) 1360.[40] Z. Zhang, S.M. Xin, G.X. Wu, W.B. Zhang, L. Ma, G. Pei, J. Neurochem.

73 (4) (1999) 1502.[41] M. Cong, S.J. Perry, F.T. Lin, I.D. Fraser, L.A. Hu, W. Chen, J.A. Pitcher,

J.D. Scott, R.J. Lefkowitz, J. Biol. Chem. 276 (18) (2001) 15192.[42] S.J. Mundell, G. Pula, J.C. More, D.E. Jane, P.J. Roberts, E. Kelly, Mol.

Pharmacol. 65 (6) (2004) 1507.[43] J. Guo, Y. Wu, W. Zhang, J. Zhao, L.A. Devi, G. Pei, L. Ma, Mol.

Pharmacol. 58 (5) (2000) 1050.[44] O.M. Kouhen, G. Wang, J. Solberg, L.J. Erickson, P.Y. Law, H.H. Loh,

J. Biol. Chem. 275 (47) (2000) 36659.[45] E. Navratilova, M.C. Eaton, D. Stropova, E.V. Varga, T.W. Vanderah, W.R.

Roeske, H.I. Yamamura, Eur. J. Pharmacol. 519 (3) (2005) 212.[46] B. Xiang, G.H. Yu, J. Guo, L. Chen, W. Hu, G. Pei, L. Ma, J. Biol. Chem.

276 (7) (2001) 4709.[47] J. Li, J.G. Li, C. Chen, F. Zhang, L.Y. Liu-Chen, Mol. Pharmacol. 61 (1)

(2002) 73.