Differential Modulation of μ- and δ-Opioid Receptor Agonists by Endogenous RGS4 Protein in SH-SY5Y Cells*

Regulator of G-protein signaling (RGS) proteins are a family of molecules that control the duration of G protein signaling. A variety of RGS proteins have been reported to modulate opioid receptor signaling. Here we show that RGS4 is abundantly expressed in human neuroblastoma SH-SY5Y cells that endogenously express μ- and δ-opioid receptors and test the hypothesis that the activity of opioids in these cells is modulated by RGS4. Endogenous RGS4 protein was reduced by ∼90% in SH-SY5Y cells stably expressing short hairpin RNA specifically targeted to RGS4. In these cells, the potency and maximal effect of δ-opioid receptor agonist (SNC80)-mediated inhibition of forskolin-stimulated cAMP accumulation was increased compared with control cells. This effect was reversed by transient transfection of a stable RGS4 mutant (HA-RGS4C2S). Furthermore, MAPK activation by SNC80 was increased in cells with knockdown of RGS4. In contrast, there was no change in the μ-opioid (morphine) response at adenylyl cyclase or MAPK. FLAG-tagged opioid receptors and HA-RGS4C2S were transiently expressed in HEK293T cells, and co-immunoprecipitation experiments showed that the δ-opioid receptor but not the μ-opioid receptor could be precipitated together with the stable RGS4. Using chimeras of the δ- and μ-opioid receptors, the C-tail and third intracellular domain of the δ-opioid receptor were suggested to be the sites of interaction with RGS4. The findings demonstrate a role for endogenous RGS4 protein in modulating δ-opioid receptor signaling in SH-SY5Y cells and provide evidence for a receptor-specific effect of RGS4.

consensus" domain that bind to G␣ subunits and act as GTPase-accelerating proteins to increase the rate of GTP hydrolysis. This results in a decrease in the lifetime of the active G␣-GTP and free G␤␥ subunits and limits signaling to downstream effectors (5)(6)(7)(8). The mechanisms by which RGS proteins selectively modulate G protein-mediated receptor signal transduction pathways, especially opioid receptor signaling, are beginning to unfold (9 -12). The foundation for the function and selectivity of RGS proteins in regulating opioid signaling lies in their ability to interact with opioid receptors and their cognate G proteins. In general, the selectivity or the preference of an RGS protein for a particular receptor is determined by a variety of factors, including tissue-specific expression and precise interaction with the intracellular domains of receptor proteins, G protein subunits, and effectors as well as other pathway-specific components (13).
The effects of RGS proteins on opioid receptor signaling have been examined in several systems. The findings are not always consistent, probably due to the different methodologies used. It has been shown that members of the RZ, R4, and R7 subfamilies (7) of RGS proteins play crucial roles not only in terminating acute opioid agonist action but also in opioid receptor desensitization, internalization, recycling, and degradation (3,14), thereby affecting opioid tolerance and dependence (15)(16)(17)(18). Much work has been performed with RGS4, because it is a smaller RGS protein with a structure consisting of the RGS consensus (box) sequence and a small N terminus (19,20). It also has a wide distribution in the brain, especially in brain regions important for opioid actions, including the striatum, locus coeruleus, dorsal horn of the spinal cord, and cerebral cortex (21). In vitro RGS4 has been shown to reverse ␦-opioid receptor agonist-induced inhibition of cAMP synthesis in membranes prepared from NG108-15 cells (6). Overexpression of RGS4 in HEK293 cells also attenuated morphine-, [D-Ala 2 ,N-Me-Phe 4 ,Gly-ol 5 ]enkephalin (DAMGO)-, and [D-Pen 2 ,D-Pen 5 ]enkephalin (DPDPE)-induced inhibition of adenylyl cyclase (22,23). Co-expression of RGS4 with GIRK1/GIRK2 channels in Xenopus oocytes reduced the basal K ϩ current and accelerated the deactivation of GIRK channels activated by -opioid receptor agonist U69593 (24). Although these previous studies have provided evidence that RGS4 can negatively regulate opioid receptor signaling, they do not confirm a functional role for endogenous RGS4 in endogenous, nontransfected systems.
Human neuroblastoma SH-SY5Y cells endogenously express and ␦-opioid receptors and a variety of G␣ i/o proteins (25)(26)(27). Here we show that RGS4 is abundantly found at both the mRNA and protein levels in these cells. Consequently, we used SH-SY5Y cells to examine the hypothesis that RGS4 negatively modulates opioid receptor signaling under physiological conditions. The endogenously expressed RGS4 level in SH-SY5Y cells was reduced using lentiviral delivery of short hairpin RNA (shRNA) targeting the RGS4 gene. This resulted in changes in ␦but not -opioid receptor-mediated signaling to adenylyl cyclase and the MAPK pathway. These findings argue for a selective interaction of RGS4 with the ␦-opioid receptor. To test this, we expressed FLAG-taggedand ␦-opioid receptors together with a construct for a stable, proteosome-resistant RGS4 protein in HEK293T cells. Co-immunoprecipitation indicated that the ␦-opioid but not the -opioid receptor was closely associated with RGS4, providing further evidence for a selective interaction between RGS4 and ␦-opioid receptor signaling.
Cell Culture-SH-SY5Y, C6 rat glioma cells, and HEK293T cells were purchased from ATCC (Manassas, VA). They were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and penicillin (100 units/ml), streptomycin (100 g/ml) under 5% CO 2 at 37°C. SH-SY5Y cell lines expressing G␣ o constructs with a PTX-insensitive mutation (G␣ o PTXi) or both PTX-and RGS-insensitive mutations (G␣ o RGSi/PTXi) were generated by transfecting the appropriate plasmid DNAs (see below) using Lipofectamine 2000. Stable cell lines were selected and maintained in the presence of 0.4 mg/ml Geneticin.
Plasmid Constructs-The FLAG-tagged mouse -opioid receptor construct was a gift from Dr. Lakshmi Devi (Mount Sinai School of Medicine, New York). FLAG-tagged human ␦-opioid receptor construct (29) and the HA-RGS4C2S construct (30) were as described previously. A /␦-C chimeric receptor comprising amino acids 1-339 of the rat -opioid receptor and 322-372 of the human ␦-opioid receptor C-terminal domain and a ␦/-C chimeric receptor comprising amino acids 1-321 of human ␦-opioid receptor and 340 -398 of the rat -opioid receptor C-terminal domain have been described earlier (23). Both chimeras have the FLAG epitope tag at the N terminus. Human PTX-insensitive G␣ o (PTXi with a Cys to Gly mutation at amino acid 351) was from the Missouri S&T cDNA Resource Center. The RGS-insensitive construct (G␣ o RGSi/PTXi) was generated by a Gly to Ser mutation at amino acid 184 of PTXi G␣ o (31).
Reverse Transcription (RT)-PCR-Total RNA was prepared from SH-SY5Y or C6 cells using the VersaGENE TM RNA purification system (Gentra Systems, Minneapolis, MN) and then subjected to RT-PCR with SuperScript TM One-Step TM RT-PCR System according to the supplier's manual (Invitrogen). Primers for detection of RGS4 were designed from the RGS4 coding region as follows: sense primer, 5Ј-GAAGTCAA-GAAATGGGCTGAATC-3Ј; antisense primer, 5Ј-CGGT-TCTGAGATACGAGAC-3Ј. The primers were first checked by amplifying RGS4 plasmid DNA to make sure that the correct size of the PCR product was achieved with the expected size of 502 bp. Total RNA (200 ng) was used with primers (0.3 M each) and MgSO 4 (1.2 mM) in a 25-l volume. The reverse transcription was performed by heating RNA at 65°C for 10 min and then at 45°C for 30 min, followed by PCR with 30 cycles at 95°C for 30 s, 45°C for 45 s, and 72°C for 1 min. The RT-PCR products were separated by electrophoresis on a 1.8% agarose gel, stained with ethidium bromide, and photographed using a Kodak Image Station 440.
Design and Construction of Lentivirus Encoding shRNA against RGS4-The shRNA lentiviral delivery system developed by Dr. Didier Trono (32) was used. In brief, four targeting sites were designed based on the mouse RGS4 gene (33) as follows: site 1, 5Ј-AGCTTTCCTGAAGTCGGAA-3Ј; site 2, 5Ј-AAGAGGTGAACCTGGACTC-3Ј; site 3, 5Ј-CTTCCT-CAAGTCTCGATTC-3Ј; site 4, 5Ј-AATGCCAAGACTCTAT-GCT-3Ј. Sites 3 and 4 are identical between mice and humans; site 1 has two nucleotides different; and site 2 has one nucleotide different. The four shRNA oligonucleotides against RGS4 were constructed into the pLVTHM lentivector by direct cloning of annealed shRNA at Mlu1-Cla1 sites. The gene for green fluorescent protein (GFP) is encoded by the vector pLVTHM. Lentivirus was produced by co-transfecting each pLVTHM-shRNA construct individually with the other components of the virus, including pMDLg/pRRE, pRSV-Rev, pRRL, and pMD2G into low passage number HEK293T cells using Lipofectamine 2000 in OPTI-MEM medium. Lentiviruses were harvested from the supernatant, concentrated by centrifugation (35,000 rpm), resuspended in phosphate-buffered saline (pH 7.2), and flash-frozen in liquid N 2 . The concentrated lentiviruses were stored in aliquots at Ϫ80°C. A titration for each lentivirus was obtained by infecting HEK293T cells with serially diluted virus in 24-well plates, and the number of GFP-express-ing cells was counted to calculate transducing units. There were ϳ3 ϫ 10 7 -10 8 transducing units/ml for each lentivirus.
Western Blot for RGS4 Protein-Lysates were prepared from SH-SY5Y cells, HEK293T cells, or striatal tissue from wild-type or RGS4 knock-out mouse brain (34) by suspension in ice-cold radioimmune precipitation lysis buffer containing 1ϫ phosphate-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS with a fresh addition of 1ϫ protease inhibitor mixture mix (0.15-0.3 trypsin inhibitor units/ml aprotinin, 1 mM sodium orthovanadate, 1 mM benzamidine, 1 g/ml pepstatin, 10 g/ml phenylmethylsulfonyl fluoride, and 2 g/ml leupeptin). The suspension was passed 10 times through a 21-gauge needle, 10 g/ml phenylmethylsulfonyl fluoride was added, and the mixture was incubated for another 1 h on ice and finally centrifuged at 10,000 ϫ g for 10 min. The supernatant was saved as total cell lysate, aliquoted, and stored at Ϫ20°C. Protein (ϳ50 g) was subjected to SDS-PAGE on a 12% minigel and transferred to an Immobilon TM -P transfer membrane. The membrane was blocked with 1% bovine serum albumin in TBST (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.05% Tween 20) for 1 h and incubated with U1079 RGS4 antiserum (34) at a 1:10,000 dilution for 1 h at room temperature or overnight in the cold room. After three consecutive washes with TBST (for 10 min each), the membrane was incubated with a 1:10,000 dilution of secondary antibody (goat anti-rabbit IgG-horseradish peroxidase) for 1 h at room temperature. Prestained SDS-PAGE protein standards (Bio-Rad; Precision Plus protein standards, Kaleidoscope TM ) were used to determine the size of detected proteins. The membranes were stripped and reblotted with anti-␤-actin antibody at a 1:1000 dilution as an internal control for protein loading. Proteins were visualized by chemiluminescence with SuperSignal West Pico (Pierce) and exposed to x-ray film.
Lentiviral Infection of SH-SY5Y Cells-Cells plated (at ϳ80% confluence) in 35-mm dishes were infected with a mixture of the four lentiviral stocks encoding shRNA against RGS4 with 6 g/ml Polybrene in Dulbecco's modified Eagle's medium. A control cell line was obtained using lentivirus stock encoding shRNA against GFP (from Dr. Didier Trono). The stable cell lines were generated by passaging cells into larger dishes, and GFP expression was used as an indicator for the presence of shRNA throughout the cell culture maintenance.
cAMP Accumulation-Cells were plated to 80 -90% confluence and treated with 10 M retinoic acid for 6 -8 days. On the day of the assay, cells were washed once with fresh serum-free medium, and the medium was replaced with 1 mM 3-isobutyl-1-methylxanthine (IBMX) in serum-free medium for 15 min at 37°C and then replaced with medium containing 1 mM IBMX, 30 M forskolin, and varying concentrations of SNC80 or morphine for 5 min at 37°C. Reactions were stopped by replacing the medium with ice-cold 3% perchloric acid, and samples were kept at 4°C for at least 30 min. An aliquot (0.4 ml) from each sample was removed, neutralized with 0.08 ml of 2.5 M KHCO 3 , vortexed, and centrifuged at 15,000 ϫ g for 1 min to pellet the precipitates. Accumulated cAMP was measured by radioimmunoassay in a 10 -15-l aliquot of the supernatant from each sample following the manufacturer's instructions and calculated as pmol/g protein accumulation of cAMP.
MAPK Analysis-Cells were plated in 24-well plates for 5-7 days and serum-starved for 24 h. Then they were washed once with fresh serum-free medium and stimulated with 100 nM morphine, 1 M SNC80, or distilled H 2 O, respectively, for 5 min at 37°C. The reaction was stopped by adding 0.1 ml of ice-cold SDS sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 50 mM dithiothreitol, and 0.01% bromphenol blue), and samples were removed from the wells, boiled for 5 min, and then subjected (10 -15 l each) to electrophoresis using a 12% SDS-PAGE minigel, followed by transfer to an Immobilon TM -P membrane for Western blotting. The blot was probed with a 1:6000 dilution of anti-phospho-p44/42 MAPK (Thr 202 /Tyr 204 ) antibody and visualized using horseradish peroxidase-conjugated anti-rabbit IgG, followed by enhanced chemiluminescence detection to measure the activated phospho-ERK and then exposed to Kodak x-ray film. To assure equal loading, the same membranes were stripped and reblotted with a 1:4000 dilution of anti-p44/42 MAPK antibody to measure total ERK levels. MAPK activity was calculated as normalized arbitrary units of phosphorylated MAPK (ERK1/2) over total ERK1/2 by densitometry analysis of films in the linear range of exposure using a Kodak Image Station 440.
Co-immunoprecipitation-FLAG-taggedor ␦-opioid receptor or chimeric /␦-opioid receptor cDNA was transiently transfected together with mutant HA-RGS4C2S cDNA at a 1:1 ratio of plasmid DNAs into HEK293T cells using Lipofectamine 2000. At 48 h post-transfection, cells were lysed in a buffer containing 50 mM Tris-HCl, pH 7.6, 5 mM MgCl 2 , 0.5% Nonidet P-40, 0.1% Lubrol, 1 mM EDTA, 1 mM sodium orthovanadate, and 1ϫ protease inhibitor mixture mix and then passed through a 21-gauge needle 10 times. Equal amounts of protein lysates (500 g each) were subjected to immunoprecipitation. The lysates were depleted of nuclei and cell debris by centrifugation for 10 min at 14,000 rpm in an Eppendorf centrifuge and precleared once with 20 l of protein A/G plus agarose for 30 min at 4°C on a rocking platform and then incubated with or without (protein A/G plus agarose only) anti-HA antibody-conjugated agarose beads (10 l each) for 4 h at 4°C with rocking. Subsequently, the immune complexes were washed four times in NET buffer (50 mM Tris-HCl, pH 7.6, 5 mM EDTA, 150 mM NaCl, 0.5% Nonidet P-40). Finally, complexes were dissociated by heating for 5 min in SDS sample buffer, resolved on a 10% SDS-polyacrylamide minigel, transferred to an Immobilon TM -P transfer membrane, and blotted with anti-FLAG M 1 antibody at a 1:2000 dilution. U11079 RGS4-specific antiserum at a 1:10,000 dilution was used to detect RGS4 protein. Secondary antibodies (anti-mouse or anti-rabbit) were used at a 1:10,000 dilution. Proteins were visualized by enhanced chemiluminescence and exposed to Kodak x-ray film.
Deglycosylation of Co-immunoprecipitated FLAG-tagged ␦-Opioid and Chimeric ␦/-Opioid Receptor-Immunoprecipitated FLAG-tagged ␦-opioid and chimeric ␦/-opioid receptors with RGS4 were eluted from the anti-HA antibodyagarose-conjugated beads with 1% (w/v) SDS, 1ϫ G7 reaction buffer, 1ϫ glycoprotein denaturing buffer in a total volume of 10 l, incubated for 15 min at room temperature (22°C), and then denatured by boiling for 10 min at 100°C. The elutants were diluted 5-fold with 1ϫ G7 reaction buffer and 1% Nonidet P-40 in a total volume of 50 l. Deglycosylation was performed with 3 l of N-glycosidase F (500,000 units/ml) overnight at 30°C before termination of the reaction by the addition of 4ϫ SDS sample buffer and 50 mM dithiothreitol. Proteins were separated by SDS-PAGE and analyzed by Western blot with anti-FLAG M 1 antibody at 1:2000 dilution. Anti-mouse secondary antibody was used at a 1:10,000 dilution. Proteins were visualized by enhanced chemiluminescence and exposed to Kodak x-ray film.
Receptor Binding Assay-Membranes from SH-SY5Y cells expressing shRNA against RGS4 or GFP were prepared essentially as described previously (25) Data Analysis-Data from at least three separate experiments, each carried out in duplicate (except where stated), are presented as means Ϯ S.E. For analysis, each data point was treated as a separate value. Data were compared by two-way analysis of variance with Bonferroni posttest analysis of variance and differences considered significant if p was Ͻ0.05. Concentration-response data were fitted and compared using the simultaneous nonlinear regression curve fitting package in GraphPad Prism, version 5.0 (GraphPad Software, La Jolla, CA).

RGS4 Expression in SH-SY5Y
Cells-RGS4 expression in SH-SY5Y cells was confirmed by both RT-PCR (Fig. 1A) and Western blot (Fig. 1B). Using RGS4-specific primers for the coding region, a strong 502 bp band was detected in SH-SY5Y cells that was not present in C6 rat glioma cells, previously reported not to express RGS4 (12). Using the validated U1079 RGS4-specific antibody (34), a strong band was detected at 28 kDa in HEK293T cells overexpressing RGS4, striatal tissue from wild-type mouse brain, and SH-SY5Y cells. This band was absent in striatal tissue from RGS4 knock-out mice, C6 cells, and untransfected HEK293T cells. In SH-SY5Y cells, there were two smaller bands detected which may indicate the presence of short forms of RGS4 or metabolism products. In order to confirm specificity, the RGS4 antibody was preabsorbed with purified full-length RGS4 protein (Fig. 1D). This completely blocked the 28 kDa band and two smaller bands in SH-SY5Y cells, leaving some low intensity nonspecific bands.
Development of a SH-SY5Y Cell Line with Reduced Expression of RGS4-To study the role of RGS4, we developed a SH-SY5Y cell line stably expressing shRNA against RGS4 to continuously block RGS4 protein expression. Four lentiviral stocks encoding shRNA targeted to four different sites on the RGS4 gene with GFP marker were used to infect SH-SY5Y cells. Over 90% of the SH-SY5Y cells were infected with lentivirus, as indicated by the GFP marker ( Fig. 2A). The stable SH-SY5Y cell line expressing shRNA against RGS4 showed much reduced RGS4 expression as determined by SDS-PAGE followed by Western blot analysis using the U1079 RGS4 antiserum, confirming RGS4 knockdown and providing an RGS4-deficient cell. In contrast, RGS4 protein was easily detectable in SH-SY5Y control cells stably expressing shRNA against GFP (Fig. 2B).
To determine if opioid receptor levels were changed in the stable SH-SY5Y cell lines expressing shRNA against RGS4 or GFP, receptor binding assays were performed with the selective -ligand [ 3 H]DAMGO (2.5 nM) and selective ␦-ligand [ 3 H]DP-DPE (4 nM) (25). As shown in Fig. 2C, the levels of expression were the same in both cell lines withand ␦-opioid receptors expressed in a 3.5:1 ratio, similar to previous reports (25,35). This was also confirmed by Western blot analysis of separated proteins for the expressedand ␦-opioid receptors before and FIGURE 1. RGS4 expression in SH-SY5Y cells. A, RGS4 mRNA expression by RT-PCR. Total RNA prepared from SH-SY5Y or C6 glioma cells was subjected to RT-PCR using RGS4-specific primers. The PCR product was separated on a 1.8% agarose gel, stained, and photographed. The expected 502-bp PCR product was detected in SH-SY5Y (SY5Y) cells but not in C6 glioma cells (C6). B, RGS4 protein expression by Western blot. Lysates were prepared from HEK293T cells overexpressing RGS4 (OE), striatal brain tissue from wild type (WT) or RGS4 knock-out mice (KO), SH-SY5Y cells (SY5Y), C6 glioma cells (C6), or untransfected HEK293T cells (HEK) and resolved (50 g each) on a 12% SDS-PAGE minigel for Western blot analysis with U1079 anti-RGS4 antibody, as described under "Experimental Procedures." C, the same blot was stripped and blotted with anti ␤-actin antibody with similar intensity in each lane except in HEK293T cells overexpressing RGS4, which used 100-fold less protein. D, purified full-length RGS4 protein completely neutralized the anti-RGS4 antibody, causing loss of the 28 kDa band and two smaller bands in the immunoblot of lysates from SH-SY5Y cells.
after retinoic acid differentiation (data not shown), similar to a previous report (36).
Differential Effects of RGS4 on Inhibition of Forskolin-stimulated cAMP Accumulation by SNC80 and Morphine-RGS4deficient and control SH-SY5Y cells were first differentiated with 10 M retinoic acid treatment for 6 -8 days, and then adenylyl cyclase activity was measured by the accumulation of cAMP stimulated by forskolin (30 M) in the presence of the phosphodiesterase inhibitor IBMX (1 mM). Cells differentiated with 10 M retinoic acid showed increased inhibition of cAMP accumulation by opioid agonists (ϳ20%) (data not shown) and so allowed for an improved agonist/noise ratio in this assay (37). The levels of forskolin-stimulated cAMP accumulation were similar in control and RGS4-deficient cells (5.3 Ϯ 1.1 versus 5.9 Ϯ 0.9 pmol/g protein, respectively). The degree of inhibition of forskolin-stimulated cAMP accumulation by the -opioid agonist morphine at 100 nM was the same in both cell lines (35 Ϯ 6% in RGS4-deficient cells and 33 Ϯ 7% in control cells), but the effect of the selective ␦-opioid agonist SNC80 at 1 M was significantly increased to 47 Ϯ 4% in the RGS4-deficient cell line compared with 30 Ϯ 4% in control cells expressing shRNA against GFP (p ϭ 0.005) (Fig. 3A).
The above concentrations of morphine and SNC80 are submaximal. The difference observed with SNC80 could be explained by a change in potency or a change in maximal effect of the agonist. Concentration-effect curves for both morphine and SNC80 to inhibit forskolin-stimulated cAMP accumulation were determined as shown in Fig. 3, B and C, respectively. The forskolin-stimulated values in control cells or RGS4-deficient cells in the absence of either SNC80 or morphine were not different (2.42 Ϯ 0.05 pmol/g protein in control cells versus 2.34 Ϯ 0.03 pmol/g protein in RGS4-deficient cells). Morphine was consistently more potent and gave higher maximal effects (85 and 87% inhibition, respectively) than SNC80 in control cells expressing endogenous levels of RGS4 and in RGS4deficient cells. There was no difference in the concentrationresponse curves for morphine between the two cell lines (Fig.  3B). In contrast, the maximum effect and EC 50 values for SNC80 were significantly different (F(2,80) ϭ 35.48, p ϭ 0.0001) when compared by simultaneous curve fitting (Graph-Pad prism, version 5.0) after constraining the forskolin-only values (Fig. 3C). In the RGS4-deficient cells, there was a significant increase in the maximal inhibition of forskolin-stimulated cAMP accumulation to 62 Ϯ 3.5% compared with 41 Ϯ 5% in control cells with a 4-fold leftward shift of the EC 50 value (from 348 nM in control cells to 84.7 nM in RGS4-deficient cells).

Reversal of the SNC80 Gain of Function by Transient Transfection of Mutant HA-RGS4C2S in SH-SY5Y Cells with Reduced
Expression of RGS4-In order to determine that the increased inhibition of forskolin-stimulated cAMP accumulation in the RGS4-deficient SH-SY5Y cells by SNC80 was indeed due to the loss of RGS4 protein, we replenished RGS4 in these cells. It is known that RGS4 is unstable due to high proteasomal degradation (37). Consequently, we used a stable RGS4 construct with a mutation (cysteine to serine) at the second amino acid of the N terminus and with a C-terminal HA tag (HA-RGS4C2S) (30). Following transient transfection of the mutant HA-RGS4C2S into the RGS4-deficient SH-SY5Y cells, the expressed HA-RGS4C2S was easily detectable using anti-HA antibody (Fig. 4A), with no bands detected in control cells transfected with the pcDNA3.1 vector alone. The inhibitory effect of SNC80 in the RGS4-deficient SH-SY5Y cells on forskolin-stimulated cAMP accumulation was reversed significantly from 68 Ϯ 7 to 57 Ϯ 7.9% (p ϭ 0.03) following transfection of the stable RGS4 (Fig. 4B). The effect of HA-RGS4C2S transfection was not greater due to the low transfection efficiency of SH-SY5Y cells using Lipofectamine 2000 (ϳ15%), as monitored by transfection of pEGFP plasmid DNA under the same conditions. In contrast, transient transfection of HA-RGS4C2S had no effect on the degree of inhibition of forskolin-stimulated cAMP by morphine.   JULY 3, 2009 • VOLUME 284 • NUMBER 27

JOURNAL OF BIOLOGICAL CHEMISTRY 18361
-opioid receptor coupling to adenylyl cyclase was generally sensitive to other RGS proteins expressed in these cells. Accordingly, SH-SY5Y cells stably expressing G␣ o with a PTX-insensitive mutation (G␣ o PTXi) or with both PTXand RGS-insensitive mutations (G␣ o RGSi/PTXi) were used to study the inhibition of forskolin-stimulated cAMP accumulation by a submaximal concentration of the -opioid receptor agonists morphine and DAMGO (Fig. 5). After overnight treatment with PTX (100 ng/ml) to uncouple endogenous G␣ proteins, both morphine and DAMGO showed increased inhibition of forskolin-stimulated cAMP accumulation in cells expressing RGS-insensitive G␣ o compared with cells expressing RGS-sensitive G␣ o . The effect was only significant with morphine (p Ͻ 0.05), possibly because of the higher efficacy of DAMGO (31). These results indicated that -opioid receptor agonists are sensitive to global inhibition of RGS protein activity in SH-SY5Y cells.
Differential Effects of RGS4 on Stimulation of MAPK by SNC80 and Morphine-To determine if the differential effect of RGS4 onand ␦-agonist signaling is specific to inhibition of adenylyl cyclase, we examined whether MAPK regulation by opioid receptor agonists was also modulated by endogenous RGS4 protein. Stimulation of ERK phosphorylation by morphine (100 nM) and SNC80 (1 M) was measured in both RGS4deficient and control SH-SY5Y cells (Fig. 6). The basal level of ERK phosphorylation, measured as normalized arbitrary units

. Increased inhibition of forskolin-stimulated cAMP accumulation by SNC80 but not by morphine in RGS4-deficient SH-SY5Y cells.
A, the ␦-agonist (SNC80) but not the -agonist (morphine) showed enhanced inhibition of forskolin-stimulated cAMP in RGS4-deficient SH-SY5Y cells. Cells were treated with 10 M retinoic acid 6 -8 days before the assay and pretreated with 1 mM IBMX in serum-free medium for 15 min. Cells were then incubated with 30 M forskolin, 1 mM IBMX, and 100 nM morphine or 1 M SNC80 for 5 min at 37°C, and accumulation of cAMP (pmol/g protein) was measured by radioimmunoassay, as described under "Experimental Procedures." *, p ϭ 0.005 for SNC80 in RGS4-deficient cells compared with control cells. Shown is concentration response for morphine (B) or SNC80 (C) to inhibit forskolin-stimulated cAMP accumulation in control (solid line and open symbols) and RGS4-deficient cells (dashed line and solid symbols), expressed as pmol/g protein accumulation of cAMP from 4 -6 experiments each performed in duplicate (where error bars are not shown, they are smaller than the symbols). The procedure was carried out as described in A and under "Experimental Procedures" with varying concentrations of morphine (10 Ϫ9

. Transfection of a stable mutant of RGS4 reverses increased inhibition of adenylyl cyclase by SNC80 in RGS4-deficient SH-SY5Y cells.
SH-SY5Y cells stably expressing shRNA against RGS4 were treated with 10 M retinoic acid for 6 days and then transfected with either pcDNA3.1 (vector) or an HA-tagged stable RGS4 construct (HA-RGS4C2S). A, at 2 days post-transfection, protein lysates were prepared and subjected to SDS-PAGE, followed by Western blot analysis using anti-HA antibody. B, forskolin-stimulated cAMP accumulation (pmol/g protein, n ϭ 4 in duplicate, mean Ϯ S.E.) in the absence or presence of morphine (100 nM) or SNC80 (1 M) was carried out at 2 days post-transfection, as described in Fig. 3A. *, p Ͻ 0.05 compared with SNC80 effect in cells expressing vector alone.
of phosphorylated ERK over total ERK, was similar in both cell lines as was the level of total ERK. Morphine stimulated a similar, ϳ50%, increase in ERK phosphorylation in both control and RGS4-deficient cells. SNC80 was less effective in stimulating ERK phosphorylation than morphine, giving only a 12 Ϯ 9% increase in control cells. However, the level of ERK phosphorylation by SNC80 in the RGS4-deficient cells was markedly increased to 89 Ϯ 20% (Fig. 6) (p Ͻ 0.05).
Co-immunoprecipitation of ␦but Not -Opioid Receptors with RGS4 in HEK293T Cells-We could not identify opioid receptor antibodies suitable for immunoprecipitation of endogenous receptors in SH-SY5Y cells. Therefore, to study the possible interaction ofor ␦-opioid receptors with RGS4, HEK293T cells were co-transfected with FLAG-taggedor ␦-opioid receptor together with the mutant HA-tagged RGS4C2S construct. Subsequently, HA-RGS4C2S protein was immunoprecipitated with anti-HA antibody-conjugated agarose beads, and the immune complexes were resolved and blotted with an anti-FLAG M 1 antibody to detect any FLAG-tagged or ␦-opioid receptors.
The expression of the FLAG-taggedand ␦-opioid receptors in HEK293T cells was easily detectable on Western blots by the presence of multiple bands following SDS-PAGE of whole cell lysates using anti-FLAG M 1 antibody (Fig. 7A) compared with lysates from cells transfected with vector alone (pcDNA3.1). The two main bands detected in cells expressing the ␦-opioid receptor were at ϳ75-85 and 37-42 kDa, with a weak band at 50 -55 kDa observed only after longer exposure. In the -opioid receptor-expressing cells, there were three main bands at ϳ42-45, 55, and 90 -95 kDa. The lower molecular mass bands probably represent nonglycosylated receptors. In preparations from bothand ␦-expressing cells, there were higher molecular mass bands identified, possibly representing receptors in protein aggregates. Similar complex Western blot patterns have been pre-viously observed for FLAG-tagged opioid receptors expressed in HEK293 cells (38).
Following immunoprecipitation of lysates from HEK293T cells expressing the ␦-opioid receptor with anti-HA antibodyconjugated beads, the ␦-opioid receptor band at ϳ75-85 kDa was identified by SDS-PAGE using anti-FLAG M 1 antibody. This band was not observed following immunoprecipitation with agarose beads alone or when the immunoprecipitation experiment was performed in mock-transfected cells (Fig. 7A). In contrast, there was no evidence for co-immunoprecipitation of any -opioid receptor protein by anti-HA antibody from HEK293T cells co-expressing HA-RGS4C2S and the -opioid receptor. This was true even with a much longer exposure time of the blot and with double the amount of -opioid receptor plasmid DNA in the transfection. Agonist stimulation of the ␦-opioid receptor (SNC80; 1.0 M) or -opioid receptor (morphine; 1.0 M) for 15 min did not enhance the degree of the co-immunoprecipitation (data not shown).
We were unable to perform the reverse experiment to study RGS4 in immune complexes precipitated with FLAG M 1 antibody against the receptors due to the presence of an IgG light chain band at the same molecular mass as RGS4 (ϳ27 kDa). However, the presence of expressed HA-RGS4C2S in the immune complex following immunoprecipitation with anti-HA antibody was confirmed by reblotting with the RGS4specific antibody (U1079), after stripping the transfer membrane (Fig. 7B).  or GFP (control) were grown for 5-7 days and serum-starved for 24 h before the assay. Cells were treated with 100 nM morphine, 1 M SNC80, or distilled H 2 O as a control for 5 min at 37°C. The samples were harvested in 1ϫ SDS sample buffer and subjected to a 12% SDS-polyacrylamide minigel for MAPK analysis by Western blot using anti-phospho-p44/42 MAPK (ERK1/2) antibody for the detection of phosphorylated ERK. Membranes were then stripped and reblotted with anti-p44/42 MAPK (ERK1/2) antibody to measure total ERK levels. MAPK activity was expressed as normalized arbitrary units (a.u.) of phosphorylated ERK over total ERK by densitometry analysis. A, mean Ϯ S.E. of five individual, independent experiments; B, a representative Western blot. *, SNC80 produced a statistically significant (p Ͻ 0.05) enhancement of MAPK activity in RGS4-deficient cells compared with control cells.

Co-immunoprecipitation of RGS4 and Chimeric ␦/-Opioid
Receptor-A previous report has suggested that the C-tail ofand ␦-opioid receptors is an interacting site for RGS4 protein (39). To further confirm the C-tail as a site for interaction of the ␦-opioid receptor with RGS4, we used chimeric /␦Cand ␦/C-opioid receptors both with an N-terminal FLAG epitope (23). The chimeric opioid receptor constructs were co-expressed with the stable RGS4 mutant HA-RGS4C2S in HEK293T cells. SDS-PAGE of lysates from cells expressing the FLAG-tagged receptor chimeras showed very similar patterns (Fig. 8A) as FLAG-tagged ␦and -opioid receptors (Fig. 7A). In lysates from cells transfected with the /␦C opioid receptor, two main bands at ϳ90 -95 and 42-45 kDa were observed with some higher molecular weight bands. Treatment of cell lysates with anti-HA antibody-conjugated agarose beads co-immunoprecipitated a protein of ϳ90 -95 kDa in the transfected chimeric /␦C opioid receptor cells that was not present in the control HEK293T cells transfected with vector alone (pcDNA3.1) (Fig. 8A). RGS4C2S was present in the immune complex (Fig. 8B). These findings indicate that the C-tail of the ␦-opioid receptor contributes to the site of interaction with RGS4. Western blotting following electrophoretic separation of lysates from HEK293T cells expressing the ␦/C chimeric receptor construct gave two main bands at ϳ75-85 and 37-42 kDa. The protein at 75-85 kDa was co-immunoprecipitated by anti-HA antibody (Fig. 8A) together with RGS4C2S (Fig. 8B). Since we found no biochemical evidence for interaction of the -opioid receptor and RGS4, this would suggest a role for other regions of the ␦-opioid receptor, possibly the third intracellular loop (39), as well as the C terminus as interacting sites for RGS4 protein.
Deglycosylation of FLAG-tagged ␦-Opioid and Chimeric ␦/-Opioid Receptor Immunoprecipitated with RGS4-Experiments depicted in Figs. 7A and 8A showed that we were able to immunoprecipitate the larger species (ϳ75-85 kDa) but not smaller species (ϳ37-42 kDa) of the ␦-opioid receptor with RGS4C2S, although the smaller species was abundantly detected in whole cell lysates. This could be because the higher apparent molecular mass species is the mature glycosylated plasma membrane form. N-Glycosylation is necessary to maintain expression of fully functional and stable ␦-opioid receptor molecules at the cell surface, whereas lower molecular mass forms of the receptor are probably intracellular precursors (40). To support this idea, precipitated immune complexes were subjected to N-glycosidase F digestion (Fig. 9). This resulted in a shift in the apparent molecular mass of both immunoprecipitated FLAG-␦ and chimeric FLAG-␦/ receptors compared with the immune complexes without N-glycosidase F digestion.

DISCUSSION
In this study, we used lentiviral delivery of shRNA to specifically suppress the level of endogenously expressed RGS4 protein in human neuroblastoma SH-SY5Y cells. This suppression of RGS4 message and protein enhanced the inhibition of cAMP accumulation and especially the stimulation of MAPK activity by the ␦-opioid receptor agonist SNC80 but not by the -opioid receptor agonists morphine or DAMGO. The effect on SNC80 was reversed by overexpression of a stable RGS4 mutant (HA-RGS4C2S). An interaction between RGS4 and an N-glycosylated mature form of the ␦-opioid receptor was confirmed by co-immunoprecipitation of HA-tagged RGS4 with FLAGtagged ␦-opioid but not FLAG-tagged -opioid receptor when the proteins were transiently co-expressed in HEK293T cells. These results provide novel evidence for opioid receptor type specificity of the action of RGS4 in modulating G protein signaling in SH-SY5Y cells and demonstrate for the first time that endogenous RGS4 differentially modulatesand ␦-opioid receptor signaling in an endogenous system.
The effect of RGS4 knockdown on ␦-opioid signaling to adenylyl cyclase was only modest. A similar small effect on opioid inhibition of adenylyl cyclase has been reported in cells overexpressing RGS4 (22,23), and we have confirmed in HEK293T cells that overexpressed HA-RGS4C2S caused only a 4-fold rightward shift in the concentration-effect curve for SNC80 (data not shown). This could indicate that RGS proteins other than RGS4 are involved in modulating ␦-opioid receptor signaling. On the other hand, RGS proteins act to decrease the lifetime of active G␣-GTP but do not stop the G protein cycle. Indeed, even in SH-SY5Y cells after elimination of all RGS activity by the use of RGS-insensitive G␣ subunits, there was only a limited effect on -opioid signaling, in agreement with previous data in other cell systems (31). In contrast to adenylyl cyclase, the effect on ␦-opioid receptor signaling to MAPK was much more robust, possibly suggesting a preferential action of RGS4 for this signaling pathway. We have previously shown a differential ability of RGS protein activity to modulate cell signaling pathways (31).
The finding of a negative modulatory role for RGS4 on ␦-opioid receptor signaling in SH-SY5Y cells agrees with the results of Hepler et al. (6), who reported that exogenously added RGS4 successfully reversed inhibition of cAMP accumulation in NG108 -15 cells by the ␦-opioid agonist [Leu 5 ]enkephalin, and with findings that GST fusion proteins of the C-tail and third intracellular loop peptides of the ␦-opioid receptor can interact with RGS4 (39). Conversely, these results are in contrast to the observation that transiently expressed RGS4 in HEK293T cells did not modulate the ability of the ␦-opioid agonist DPDPE to inhibit adenylyl cyclase via a transiently expressed ␦-opioid receptor (23). Indeed, that study found a lack of activity of RGS proteins in modulating ␦-opioid signaling to adenylyl cyclase with the exception of RGS9. However, we could not identify the expression of RGS9 protein in SH-SY5Y cells by Western blot using a highly selective anti-RGS9 antibody (41) (data not shown).
The finding that -opioid receptor signaling to adenylyl cyclase or to the MAPK pathway is not modulated by RGS4 in SH-SY5Y cells is consistent with our finding that RGS4 and the -opioid receptor expressed in HEK293T cells did not co-immunoprecipitate. However, our findings are inconsistent with previous reports using heterologous expression systems. For example, overexpression of RGS4 in HEK293 cells caused a significant attenuation of -opioid receptor-mediated inhibition of adenylyl cyclase (22,23). In addition, RGS4 accelerates the deactivation of G protein-coupled inwardly rectifying K ϩ channel after termination of -opioid receptor signaling in Xenopus oocytes (42). There is also evidence for an interaction of RGS4 with GST-fusion C-terminal tail peptides of the -opioid receptor (39). Then again, a major difference between the current study and previous work is that we have knocked down endogenously expressed RGS4 in the neuroblastoma SH-SY5Y cell line that endogenously expresses bothand ␦-opioid receptors. This provides a model much closer to the endogenous situation compared with cells heterologously overexpressed with RGS4 and/or receptors. Our finding that -opioid signaling is altered in SH-SY5Y cells stably expressing RGSinsensitive G␣ o compared with cells expressing wild-type G␣ o demonstrates that -opioid receptor signaling is sensitive to the GTPase-accelerating protein activity of RGS proteins expressed in SH-SY5Y cells. Consequently, it is possible that in SH-SY5Y cells -opioid receptor signaling is insensitive to or in some way protected from RGS4.
An alternative scenario for our observed selectivity of RGS4 and one that explains previous results in heterologous expression systems is that RGS4 does modulate both ␦and -opioid signaling in SH-SY5Y cells but that other RGS proteins expressed in these cells can substitute for RGS4 at the -opioid but not at the ␦-opioid receptor (i.e. there is a functional redundancy at the -opioid but not at the ␦-opioid receptor). This could be due to the lack of expression of alternative RGS partners for the ␦-opioid receptor in SH-SY5Y cells (e.g. RGS9 is not expressed, as discussed above). Alternatively, there may be a higher stringency of interaction between RGS4 and the ␦-opioid receptor. Certainly, it is known that the ␦-receptor is cou- FIGURE 9. Deglycosylation of the FLAG-tagged ␦and chimeric ␦/-opioid receptors immunoprecipitated by HA-RGS4C2S. Transfected FLAG ␦or chimeric FLAG ␦/-opioid receptor in HEK293T cells was immunoprecipitated with HA-RGS4C2S using anti-HA antibody-conjugated agarose beads as in Figs. 7A and 8A. The immune complexes were eluted from the anti-HA antibody-agarose-conjugated beads, denatured, and subjected to the deglycosylation with (ϩ) or without (Ϫ) N-glycosidase F (PNGase F), as described under "Experimental Procedures." The reaction was terminated, proteins were resolved with SDS-PAGE, and Western blot analysis was performed using anti-FLAG M 1 antibody. Only the larger molecular mass species (ϳ75-85 kDa, as indicated by arrows) of both the FLAG ␦ and the chimeric FLAG ␦/ receptors were precipitated compared with the whole cell lysates (WL), and this migrated faster following N-glycosidase F digestion (ϩ) compared with the samples without N-glycosidase F digestion (Ϫ). The vector alone control cells (V) transfected with pcDNA3.1 and treated similarly on the right showed smaller background bands, indicating possible IgG heavy chain (IgG HC) and light chain (IgG LC) due to nonspecific immunoprecipitation. pled tightly to G␣ proteins (43) and also has a high level of agonist-independent activity, indicative of tight coupling (44,45). RGS4 could thus be an integral part of a ␦-opioid receptor-G␣-G␤␥-RGS4 signaling complex (39).
Indeed, our immunoprecipitation results do indicate that the ␦-opioid receptor has a more robust interaction with RGS4 than the -opioid receptor in the heterologously expressed HEK293T cells. This is probably due to interaction with the third intracellular loop and/or C-tail of the ␦-opioid receptor in agreement with a report that RGS4 interacts directly, probably through its N-terminal region (46), with GST fusion proteins of the C-tail and third intracellular loop of the ␦-opioid receptor but only interacts with GST-fusion C-terminal tail peptides of the -opioid receptor (39). The presence of two interacting sites between RGS4 and the ␦-opioid receptor, the C terminus and third intracellular loop, could explain why we observe coimmunoprecipitation of the ␦-opioid but not the -opioid receptor, although RGS4 has been reported to act as a GTPaseaccelerating protein for -opioid signaling in HEK293 cells (22,23). Certainly, the presence of multiple contact sites between the ␦-opioid receptor and RGS4 could be responsible for the observed interaction and/or the high stringency in the SH-SY5Y cell system.
We have previously shown thatand ␦-opioid receptors share the same G proteins in SH-SY5Y cells (25), but the present results suggest that these G proteins are differentially regulated by RGS4. This conundrum can be explained by postulating that RGS4 is closely associated in a complex with the ␦-opioid receptor as discussed above rather than the G␣ protein, in accordance with the immunoprecipitation experiments and previous studies with purified proteins (39). Indeed, several studies have identified receptor-specific effects of RGS protein action. For example, only RGS3 expression suppressed gonadotropin-releasing hormone-induced IP 3 responses (10), and in pancreatic acinar cells, microinjected RGS4 selectively inhibited calcium signals induced by muscarinic but not cholecystokinin receptors (11). In addition, Wang et al. (9) showed, using ribozyme-mediated gene knockdown, that RGS3 is a negative modulator of ERK activation by muscarinic m3 receptors but that endogenous RGS5 is a negative modulator of AT1a receptor-mediated activation of ERK in both A-10 cells and primary culture of aorta smooth muscle cells. However, since we were unable to show a direct interaction between endogenous ␦-opioid receptors and endogenous RGS4 in SH-SY5Y cells, we cannot rule out the possibility that the RGS4-␦-opioid receptor interaction is indirect (48), perhaps mediated by an intermediate scaffold, such as spinophilin. Spinophilin is known to bind to several G protein-coupled receptors at the third intracellular loop and to RGS4 (49).
In conclusion, the current findings demonstrate that endogenous RGS4 in SH-SY5Y cells differentially modulates the ability ofand ␦-opioid receptor agonists to inhibit adenylyl cyclase and stimulate the MAPK pathway, possibly due to a high stringency of interaction between RGS4 and the ␦-opioid receptor. This is in contrast to studies in heterologous overexpression systems and shows the value of lentiviral delivery of shRNA targeting RGS genes to permit the analysis of RGS function and its receptor interactions in endogenous systems. Given that RGS4 is widely expressed in many brain regions (21), including the amygdala, nucleus accumbens, and caudate-putamen, where ␦-opioid receptors are also highly expressed (47), the selective RGS4 modulation of ␦over -opioid receptor signaling may play a significant role in modulation of ␦-opioid receptor-mediated behaviors.