Internalization and Src Activity Regulate the Time Course of ERK Activation by Delta Opioid Receptor Ligands*

The present study showed that delta opioid receptor ( (cid:1) OR) ligands Tyr-Ticpsi [CH 2 -NH]Cha-Phe-OH (TICP) and ICI174864 behaved as inverse agonists in the cyclase pathway but induced agonist responses in the ERK cascade. Unlike ligands that behaved as agonists in both pathways, and whose stimulation of ERK was marked but transient (10 min), ERK activation by ICI174864 and TICP was moderate and sustained, lasting for more than 1 h in the case of TICP. Biochemical experiments showed that duration of ERK activation by agonists and “dual efficacy ligands” was inversely correlated with their ability to trigger receptor phosphorylation and degradation. Thus, although TICP stabilized (cid:1) ORs in a conformation that did not incorporate 32 P, was not a substrate for tyrosine kinase Src, and was not down-regulated following prolonged exposure to the drug, the conformation stabilized by D -Pen-2,5-enkephalin (DP- DPE) incorporated 32 P, was phosphorylated by Src, and suffered degradation within the firs t 2 h oftreatment. of endocytosis

Occupation of G protein-coupled receptors by agonist ligands has two distinct consequences, the generation of an intracellular signal and the concomitant activation of a series of regulatory mechanisms that modulate receptor responsiveness over time. The chain of regulatory events triggered by agonist occupation of the receptor has been extensively characterized and has led to an established model of desensitization in which phosphorylation of the receptor by G protein-coupled receptor kinases is the first step in the process (1,2). Phosphorylation then promotes the recruitment of ␤arrestin (3,4), which is responsible for uncoupling the receptor from the G protein (5) and for its targeting to clathrin-coated pits. From there receptors will be removed from the cell surface via dynamin-dependent endocytosis (6). Once inside the cell the receptor is either degraded or is quickly redirected to the cell membrane (7) for a new signaling cycle.
Despite the overwhelming evidence supporting this tightly knit model of activation and desensitization, there are also increasing observations indicating that activation and regulatory phenomena can be dissociated. For example, antagonist ligands for cholecystokinin (8) and endothelin receptors (9) selectively induce internalization without causing neither receptor activation nor phosphorylation. Agonists for parathyroid hormone type 1 receptor stabilize an active state that promotes signaling but does not recruit ␤arrestin or induce internalization (10). In contrast, certain ␤ 2 -adrenergic receptor (␤ 2AR ) 1 ligands that preclude G protein activation are still able to recruit ␤arrestin to the receptor (11).
Ligands that stabilize G protein-coupled receptors in a conformation that prevents activation of the G protein are classified as inverse agonists and are commonly thought to induce an inactive conformation of the receptor (12,13). More recently, some of these drugs have been described as "proteans" or "dual efficacy ligands," referring to their ability to display both agonist and inverse agonist behavior (11, 14 -16). For example, we have recently shown that ICI118551 and propranolol, two ligands of the ␤ 2AR , display dual efficacy, because they behave as inverse agonists in the cyclase pathway but produce agonist responses in the ERK cascade (11).
The observation that some "inverse agonists" may produce agonist responses indicates that the conformation they stabilize is not inactive, but rather a signaling state that is distinct from the one stabilized by classic agonists. If receptor states stabilized by agonists and dual efficacy ligands are distinct, then one would expect that the responses that they elicit would also be regulated in a distinct manner. The present study focused on this question, assessing whether agonistic responses generated by dual efficacy ligands for the ␦OR are regulated as agonist responses induced by its classic agonists. Results show 1 The abbreviations used are: ␤ 2AR , ␤ 2 -adrenergic receptor; DMEM, Dulbecco's modified Eagle's medium; ␦OR, delta opioid receptor; DPDPE, D-Pen-2,5-enkephalin; ERK, extracellular signal-regulated kinase; MAPK, mitogen-activated protein kinase; PP2, 4-amino-5-(4chlorophenyl)-7-(t-butyl)pyrazolo [3,4-d] that ERK activation by dual efficacy ligands like TICP and ICI174864 was considerably longer, although more modest than the response induced by agonists such as SNC-80 and DPDPE. Differences in time course were associated with the distinct ability of dual efficacy ligands to stabilize ␦ORs in an ERK-stimulating conformation that eluded regulatory steps typically triggered by highly efficacious agonists.
DNA Constructs-The human ␦OR cDNA was subcloned into the pcDNA3 expression vector (Invitrogen) as described previously (18) and was tagged at the C-terminal end using Clontech site-directed mutagenesis kit to remove the stop codon and introduce the sequence coding for the FLAG epitope (DYKDDDDK). The construction was confirmed by restriction enzyme mapping and DNA sequencing, and its signaling properties were shown to be identical to those of the wild type ␦OR (19,20). A truncated mutant of the murine ␦OR (␦OR344T) was kindly provided by Dr. M. von Zastrow (University of California at San Francisco). Wild type and inhibitory mutant forms of c-Src (K295R/Y527F) were a gift from Dr. Bouvier's laboratory (Université de Montréal).
Cell Culture and Transfection-HEK293s cells were transfected using the calcium-phosphate precipitation method and clones stably expressing full-length or truncated receptors were selected using 400 g/ml G418. Cell lines stably expressing full-length ␦ORs and wild type c-Src were similarly selected, following Lipofectamine transfection (Invitrogen). The dominant inhibitory form of c-Src (K295R/Y527F) was transiently transfected (0.25-3 g of DNA) onto cell lines expressing the full-length ␦OR using polyethyleneimine as described previously (21). Cells were grown and maintained in complete DMEM containing 10% (v/v) fetal bovine serum, 1000 units/ml penicillin, 1 mg/ml streptomycin, and 1.5 g/ml fungizone in a humidified atmosphere of 5% CO 2 at 37°C.
Phosphorylation and Immunoprecipitation of FLAG-tagged Receptors-For 32 P incorporation assays, cells were incubated for 2 h in phosphate-free DMEM, after which [ 32 P]orthophosphoric acid was added at a final concentration of 1 mCi/ml, and incubation was allowed to proceed for an additional hour. At this time, DPDPE (1 M), TICP (1 M), or vehicle (0.01% Me 2 SO) were added to the incubation medium for 30 min. Cells were then recovered, and membranes were prepared as indicated below and finally suspended in solubilization buffer (0.5% n-dodecyl-maltoside (w/v), 25 mM Tris-HCl, pH 7.4, 2 mM EDTA, 140 mM NaCl, 5 g/ml leupeptin, 5 g/ml soybean trypsin inhibitor, 10 g/ml benzamidine, 2 g/ml aprotinin, 0.5 mM phenylmethylsulfonyl fluoride, and 2 mM 1,10-phenantroline). Following agitation at 4°C for 60 min, the solubilized fraction was centrifuged at 12,000 ϫ g for 60 min, and the receptor was immunoprecipitated from the supernatant fraction using anti-FLAG M2 antibody resin. 40 l of antibody-coupled resin equilibrated in solubilization buffer and supplemented with 0.1% bovine serum albumin (w/v) were used to purify the receptor overnight at 4°C under gentle agitation. The next morning the resin was pelleted, washed twice with 500 l of solubilization buffer and four times with 500 l of modified solubilization buffer (containing 0.1% instead of 0.5% n-dodecyl-maltoside (w/v)). The receptor was then eluted by incubating the resin for 10 min at 4°C with 100 l of modified solubilization buffer containing 175 g of FLAG peptide/ml. This elution was repeated three times, and the eluates were combined and concentrated by membrane filtration over Microcon-30 concentrators (Millipore). SDS sample buffer was then added, and samples were used for SDS-PAGE. A similar immunoprecipitation procedure was used to assess Tyr phosphorylation of ␦ORs.
SDS-PAGE and Western Blotting-SDS-PAGE was performed as described by Laemmli using a 4% stacking gel and 10% separating gel. Proteins resolved in SDS-PAGE were then transferred (50 mA, 16 h, Bio-Rad Mini-Trans Blot apparatus) from the gels onto nitrocellulose (Amersham Biosciences). In the case of 32 P incorporation studies, membranes were first exposed for autoradiography (BIOMAX films, Eastman Kodak Co.). When assessing Tyr phosphorylation of ␦ORs, mem-branes were probed overnight at 4°C with monoclonal antibodies raised against phosphorylated Tyr (1:500, PY99, Santa Cruz Biotechnology, Santa Cruz, CA). In both cases antisera directed against the FLAG M2 antibody (1:1000, Sigma) were used to detect the total amount of receptor protein present in each sample. Horseradish peroxidase-conjugated antimouse secondary antibodies (1:4000, Sigma) and chemiluminescence detection reagents (PerkinElmer Life Sciences) were used to reveal the blotted proteins, and relative intensities of the labeled bands were analyzed by densitometric scanning using MCID (Imaging Research Inc). Receptor phosphorylation was expressed as the ratio between phosphorylation and FLAG signals to normalize to the amount of receptor protein present in each sample.
For detection of ERK1/2 activation, cells were grown in 6-well plates and serum-starved overnight. The day of the experiment they were cultured for 2 h in serum-free medium and then exposed to different ligands. Following treatment, cells were washed with ice-cold phosphate-buffered saline, and whole cell extracts were prepared by lysis in SDS sample buffer. Samples were sonicated and then boiled for 5 min before loading for SDS-PAGE. Phospho-ERK1/2 detection was done by probing membranes with antiphospho-ERK1/2 antibody (1:1,000, Santa Cruz Biotechnology). Total ERK protein was determined after stripping by using 1:20,000 dilution anti-ERK1/2 antibody (Santa Cruz Biotechnology). Secondary antimouse (1:5,000, Sigma) and antirabbit (1: 40,000, Amersham Biosciences) horseradish-conjugated antibodies were used to visualize proteins by chemiluminescence. ERK1/2 phosphorylation was normalized according to protein contents by expressing results as the ratio between pERK1/2 and total ERK1/2.
To assess Src activation, cells were grown in 100-mm Petri dishes and prepared for the experiment as described for ERK1/2. Following treatment with different ligands cells were washed, harvested, and solubilized in precipitation assay buffer (50 mM Tris-HCl, pH 7.4, 1% Triton X-100, 0.25% deoxycholate acid, 150 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 5 g/ml soybean trypsin inhibitor, 10 g/ml benzamidine, 1 g/ml aprotinin, 1 mM Na 3 VO 4 ) at 4°C for 90 min. After centrifugation of non-solubilized debris at 12,000 ϫ g for 20 min samples were concentrated, suspended in SDS sample buffer, and separated in SDS-PAGE. Anti-phospho-Src (Y416) monoclonal antibody (Upstate Biotechnology Inc.) at a dilution of 1:1000 was used to determine the presence of activated Src and total amount of protein loaded was detected by probing with antibody anti-Src (1:250, Upstate Biotechnology Inc.).
cAMP Accumulation Assays-Cells were labeled overnight (16 h) with 1 Ci/ml of [ 3 H]adenine in complete DMEM medium. The day of the experiment radioactive medium was replaced with fresh DMEM, cells were mechanically detached and thoroughly washed (three times) with phosphate-buffered saline (4°C), and viability was assessed using trypan blue (mortality was never higher than 5%). 5 ϫ 10 5 cells were then incubated for 20 min at 37°C in 300 l of assay mixture containing phosphate-buffered saline, 25 M forskolin, 2.5 M isobutylmethylxanthine, and different drugs at the indicated concentrations. At the end of the incubation period, the assay was terminated by adding 600 l of ice-cold solution containing 5% trichloroacetic acid, 5 mM ATP, and 5 mM cAMP. [  [ 35 S]GTP␥S binding assays were carried out on whole cell membrane preparations as described previously (20). Cells were suspended in lysis buffer (25 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 2 mM EDTA, 5 g/ml leupeptin, 5 g/ml soybean trypsin inhibitor, and 10 g/ml benzamidine) and homogenized with a Polytron homogenizer (Ultra-Turrax T-25, Janke and Kunkel) using three bursts of 5 s at maximum setting. Homogenates were centrifuged at 700 ϫ g for 5 min, and the supernatant was further centrifuged at 27,000 ϫ g for 20 min. Pellets were washed twice in lysis buffer and were immediately resuspended in [ 35 S]GTP␥S assay buffer (50 mM Hepes, 200 mM NaCl, 1 mM EDTA, 5 mM MgCl 2 , 1 mM dithiothreitol, 0.5% bovine serum albumin, and 3 M GDP, pH 7.4) to yield 10 g of protein/tube. [ 35 S]GTP␥S was used at 50 nM, and nonspecific binding was determined in the presence of 100 M GTP. The test compound SNC-80 was introduced at a final concentration of 100 nM and incubation was allowed to proceed for one hour at RT. The reaction was terminated by rapid filtration onto Whatman GF/C glass filters pre-soaked in water. Filters were washed twice with ice-cold wash buffer (pH 7) containing 50 mM Tris, 5 mM MgCl 2 , and 50 mM NaCl, and the radioactivity retained was determined by liquid scintillation.
Data Analysis-Statistical analysis and curve fitting were done using Prism 2.01 (GraphPad, San Diego, CA).

Comparison of the Effects of ␦OR Ligands in cAMP and ERK
Signaling Cascades-It has been previously shown that certain ligands for ␤ 2ARs display dual efficacy, inducing inverse agonist responses in the cAMP signaling pathway, but producing agonist effects in the ERK cascade (11). To determine whether this type of dual behavior was specific to G s -coupled receptors or could be extended to receptors coupled to G i/o proteins, different ligands for the ␦OR were compared in adenylyl cyclase and ERK signaling pathways. In the cAMP pathway ligands produced effects that spanned the complete spectrum of efficacy ranging from agonism to inverse agonism. At maximally effective concentrations (1 M) SNC-80 and DPDPE were highly efficacious agonists, morphine, TIPP, and naloxone were partial agonists, while ICI174864 and TICP displayed typical inverse agonist responses. Fig. 1A shows these different ligands ranked according to magnitude and vectorial aspects of their efficacies (SNC-80 Ն DPDPE Ͼ MOR Ն TIPP Ն Nx Ͼ ICI174864 Ͼ TICP). In contrast with the diversity of responses observed in cAMP accumulation assays all drugs tested in the ERK cascade behaved as agonists, except for naloxone that was neutral. Indeed, ERK phosphorylation was induced not only by drugs that behaved as agonists in the cyclase cascade but also by TICP and ICI174864, which had produced inverse agonist responses when tested in this pathway. Moreover, when ranked according to the magnitude of their effect on ERK phosphorylation, TICP, the most efficacious inverse agonist in the cyclase pathway was now more effective than partial agonists TIPP and morphine in activating Dual efficacy ligands for the ␤ 2AR produce ERK activation via ␤arrestin recruitment and independently of G protein activity (11). To determine whether this was also the case for G i/ocoupled receptors, cells were treated overnight with PTX, and ERK activity was assessed the following day. Without modifying basal activity of the kinase (pERK/ERK total ratio in controls: 0.6 Ϯ 0.1; following PTX: 0.6 Ϯ 0.1), PTX abolished ERK stimulation by classic agonist DPDPE and by dual efficacy ligand TICP (Fig. 1C). These results indicate not only that ERK stimulation by DPDPE and TICP requires G i/o protein activity but also that simple inactivation of spontaneous G i/o signaling cannot account for ERK stimulation. Neither TICP, ICI174864, nor classic agonists were able to evoke ERK activation in non-transfected cells (not shown), confirming that ligand-induced stimulation of ERK signaling was specifically mediated by the ␦OR.
␦OR Ligands Differ in Their Kinetics of ERK Activation-To determine whether the time course of ERK activation by classic agonists differed from that of dual efficacy ligands, cells were exposed to a maximally effective concentration (1 M) of each drug, and ERK phosphorylation was measured following increasing periods of time. Two main types of kinetic profiles could be recognized. One was characteristic of highly efficacious ligands like SNC-80 and DPDPE, which produced quick and pronounced ERK activation that peaked within 5 min ( Fig.  2A), decaying right away with a calculated half-life (t1 ⁄2 ) of ϳ2 min (Fig. 2, B and C). The other type of response, induced by partial agonists and dual efficacy ligands was less pronounced but more sustained, decaying with a t1 ⁄2 that ranged between 11 and 14 min (Fig. 2, B and C). Among ligands inducing sustained responses, the effect of TICP could be distinguished from the rest of the drugs in the same category, because its effect was more pronounced and particularly more sustained (p Ͻ 0.001 two-way ANOVA; Fig. 2, A and B).
The Time Course of ERK Activation by Highly Efficacious Agonists and Dual Efficacy Ligands Is Correlated with Desensitization Parameters-One of the primary checkpoints that Procedures." Statistical significance of drug effects on cAMP production was established by comparing the amount of cAMP counts obtained in the presence of each ligand to cAMP counts produced in the control situation (all drugs differed from control as determined using one-way ANOVA and Dunnett's post hoc test; p Ͻ 0.05; not shown). Drug effects as they appear in the figure are expressed as percent change with respect to total amount of cAMP produced in the absence of ligand (percent change in cAMP accumulation ϭ {[cAMP ligand Ϫ cAMP no ligand ]/cAMP no ligand } ϫ 100) and correspond to mean Ϯ S.E. of at least nine experiments carried out in triplicates. Differences among drug effects were established by comparing percent changes induced by different ligands using one-way ANOVA and Tukey's post-hoc test. B, HEK293 cells stably expressing full-length ␦ORs were serumstarved (16 h) prior to exposure to saturating concentrations (1 M) of the indicated ligands for 5 min following which ERK signaling was assessed by immunoblot. Band immunoreactivity was quantified using MCID to measure optical density, and ERK1/2 phosphorylation was normalized to the amount of protein loaded per lane, by expressing the data as a ratio of phosphoERK1/2 over total ERK1/2 optical density. The statistical significance of drug effects was established by comparing the ratio obtained in the presence of each ligand to the ratio obtained in the basal condition (all drugs except naloxone differed from basal using one-way ANOVA and post hoc Dunnett's test; p Ͻ 0.05; not shown). Drug effects as they appear in the figure were expressed as the percentage of the basal ratio (percentage of basal ϭ ([pERK/totalERK ligand ]/[pERK/totalERK no ligand ]) ϫ 100), and represent mean Ϯ S.E. of at least seven experiments. Statistical differences that appear in the figure were established using one-way ANOVA followed by Tukey's post-hoc test. Immunoblots above the histogram bars correspond to representative examples of results obtained for each of the indicated drugs. C, effect of PTX (100 ng/ml for 16 h) on DPDPE or TICP-induced ERK phosphorylation. Cells were serum-starved and exposed or not to PTX prior to treatment with either DPDPE or TICP (1 M for 5 min). Results, expressed as in B, correspond to mean Ϯ S.E. of four independent experiments. Statistical difference between drug effects obtained in the presence and absence of PTX was determined using Student's t test and appear in the figure.
controls drug effects over time is the receptor itself. In particular, ␦OR signaling efficacy is regulated by phosphorylation of C-terminal Ser/Thr residues (22,23). To determine whether differences in the time course of ERK activation could be related to the distinct ability of different ligands to trigger phosphorylation of ␦ORs cells were exposed for 30 min to DPDPE or TICP (1 M) in the presence of [ 32 P]orthophosphoric acid. Receptors were immunopurified, resolved on SDS-PAGE, and transferred onto nitrocellulose membranes that were first exposed for autoradiography and then used for Western blot analysis using an anti-FLAG M2 antibody. Immunoblots revealed two broad bands at Х55 and Х46 kDa, corresponding to mature and immature monomeric forms of the receptor, respectively (24). Autoradiograms showed that 30-min incubation with DPDPE increased 32 P incorporation by the Х55-kDa species, but this effect was absent for TICP. Thus, at a time when the ERK response for the agonist was no longer present, ␦ORs were heavily phosphorylated. In contrast, ERK activation by the dual efficacy ligand was still at its maximum, and no phosphorylation of the receptor could be detected.
Phosphorylation is an initial step in the process of desensitization, but if exposure to an agonist is allowed to proceed long enough, ␦ORs will start to be targeted for degradation (7). Hence, to confirm whether the different time course of ERK activation by DPDPE and TICP also correlated with later events in the process of desensitization, cells were treated for 2 h either with the agonist or the dual efficacy drug. Following treatment the total amount of ␦OR protein present in membrane preparations was assessed by immunoblot (Fig. 3B). Although incubation with TICP caused no detectable change in the mature receptor species (Х55 kDa), there was a decrease of the corresponding immunoreactive band following treatment with DPDPE. These results confirm that differences in the time course of ERK activation by DPDPE and TICP is inversely correlated with the ability of each ligand to trigger different events within the process of desensitization.
If indeed differences in time course of ERK activation by agonists and dual efficacy ligands were due to their distinct ability to trigger regulatory mechanisms of receptor responsiveness, interfering with these mechanisms should transform ERK activation by the agonist, into the more prolonged type of response observed for the dual efficacy ligand. To test this assumption the time course of ERK activation by DPDPE was assessed in presence of sucrose, which is an inhibitor of clathrin-mediated endocytosis. Although sucrose did not turn ERK activation into a stable response, it prolonged the effect of DPDPE by increasing the decay t1 ⁄2 of activation from 2 to 12 min (p Ͻ 0.001 for interaction; two-way ANOVA; Fig. 4), a value that falls within the 11-to 15-min range observed for dual efficacy ligands. Another means to modify mechanisms regulating ␦OR responsiveness is to mutate amino acids that are implicated in the process. For ␦ORs, Ser/Thr residues located in the C-terminal domain of the receptor are the principal target for G protein-coupled receptor kinases, and their phosphorylation is an essential step in the desensitization of fulllength ␦ORs (22,23,25). To explore the contribution of these residues to the kinetics of ERK activation by DPDPE, experiments were repeated using a receptor truncated at its C terminus (␦OR344T). This approach also yielded results in which the DPDPE response decayed more slowly than in the fulllength receptor (t1 ⁄2 of 6 min; two-way ANOVA; p Ͻ 0.01 for interaction; Fig. 4). However, the effect of truncation was far less noticeable than that observed with sucrose on the fulllength ␦OR. Moreover, the addition of sucrose further prolonged the decay t1 ⁄2 for DPDPE in truncated receptors (t1 ⁄2 of 38 min; two-way ANOVA; p Ͻ 0.02 for interaction; Fig. 4). Agonists but Not Dual Efficacy Drugs Induce Tyr Phosphorylation of ␦ORs-The fact that the time course of ERK activation by DPDPE was only modestly prolonged by removal of the C terminus suggests that there could be a complementary mechanism capable of regulating ERK activation by agonists in the absence of C-terminal Ser/Thr residues. In this sense mutation of a Tyr residue located proximal to Ser/Thr amino acids of the C terminus has been shown to attenuate agonist-induced internalization and down-regulation of the receptor (26). Thus, it was deemed of interest to determine whether ␦ORs could be differentially phosphorylated at their tyrosine residues following exposure to agonists and dual efficacy ligands. To do so, cells expressing full-length receptors were incubated for 30 min either with DPDPE or TICP, and receptors were immunopurified and separated by SDS-PAGE. Immunoblots with antibodies that recognize phosphorylated Tyr residues showed that TICP and DPDPE distinctively modified phospho-Tyr contents of ␦ORs (Fig. 5A). Although 30-min incubation with TICP (1 M) produced no significant change, DPDPE induced an increase in immunoreactivity for phospho-Tyr in the band corre- FIG. 4. ERK activation by DPDPE may be modified to yield a prolonged decay t1 ⁄2 typical of TICP. HEK293 cells stably expressing wild type or truncated ␦ORs (0.5-1 pmol/mg of protein) were serum-starved (16 h) and on the day of the experiment were incubated or not with 0.4 M sucrose for 4 h, followed by exposure to saturating concentrations (1 M) of DP-DPE for the indicated times. Phos-phoERK1/2 immunoreactivity was measured as described in previous figures. Results are expressed as percentage of maximal response and correspond to mean Ϯ S.E. of at least seven experiments. Curves were fit to one-phase exponential decay using GraphPad Prism 2.01 and calculated t1 ⁄2 values appear in the figure to the right of representative immunoblots obtained for each condition. Statistical comparison among curves was done using two-way ANOVA, and p values are discussed in the text.
sponding to the mature receptor (Fig. 5A). Furthermore, the introduction of Src inhibitor PP2 (20 M) prior to exposure to DPDPE prevented the increase in phospho-Tyr content (Fig. 5B), indicating that a non-receptor Tyr-kinase of the Src family was involved in agonist-induced Tyr phosphorylation of ␦ORs. To determine whether failure of TICP to induce Tyr phosphorylation of the receptor was due to its incapacity to stimulate Src, cells overexpressing the kinase were exposed either to DPDPE or TICP and changes in Src activity assessed by immunoblot. Fig.  5C shows that both the agonist and the dual efficacy ligand increased immunoreactivity for the active form of Src, confirming that the observed difference in tyrosine phosphorylation of ␦ORs was not related to diminished capacity of TICP to stimulate Src.
The goal of the next series of experiments was to assess whether Src could differentially regulate ERK responses generated by DPDPE or TICP. To accomplish this, cells were pretreated with increasing concentrations of Src inhibitor PP2, and ERK stimulation was assessed following exposure to each of the two ␦OR ligands. It was found that concentrations of 20 -40 M PP2 had opposite effects on ERK activation by the dual efficacy ligand and the agonist. Although the effect of TICP was blocked (Fig. 6A), that of DPDPE was enhanced, and it was only at a concentration of 80 M that PP2 interfered with ERK activation by DPDPE (Fig. 6B). The inhibitory effect of low, specific concentrations of PP2 on the response to TICP is compatible with the notion that this drug induces ERK activation in a Src-dependent manner. On the other hand, the higher nonspecific concentrations needed to block the effect of DPDPE do not allow us to conclusively implicate Src as an intermediate in agonist-induced ERK stimulation. So, to overcome this problem of specificity, ERK activation by DPDPE was re-assessed by transfecting cells with increasing concentrations of a kinaseimpaired Src mutant (K296R/Y528F). This procedure generated a similar biphasic pattern as described for PP2, with low levels of the mutant (0.25-0.5 g of DNA) enhancing DPDPE responses and higher levels (3 g of DNA) inhibiting ERK activation by the agonist. The observed inhibition of DPDPE responses by high levels of inactive Src confirms the idea that activity of this kinase is necessary for ERK activation by agonists and is consistent with previous reports showing that pharmacological inhibition of Src interfered with agonist-induced ERK activation (27).
On the other hand, the fact that low levels of mutant Src or modest concentrations of PP2 enhanced ERK activation by FIG. 5. Tyrosine kinase Src phosphorylates ␦ORs occupied by DPDPE but not by TICP. A, HEK293s cells stably expressing FLAG-tagged ␦ORs were exposed for 30 min to either of the drugs, and the receptor purified as described before. Tyrosine phosphorylation was assessed by immunoblot by calculating the ratio between phospho-Tyr immunoreactivity of the mature species (shown in the right panel) and FLAG immunoreactivity of the same band (not shown). Results represent mean Ϯ S.E. of at least five experiments and are expressed as percentage of the ratio between phosphoTyr immunoreactivity and FLAG immunoreactivity obtained in basals. Statistical significance of differences between the two drugs was established using Student's t test, and the result of the analysis appears in the figure. B, 20 M PP2 or vehicle (Me 2 SO 0.01%) were introduced to cell cultures 1 h before the experiment followed by stimulation by DPDPE (1 M for 30 min) and assessment of pTyr immunoreactivity in purified receptors. Results are expressed as the percentage of the of phospho-Tyr/FLAG ratio obtained in basals of the corresponding control or PP2 condition and represent mean Ϯ S.E. of at least five experiments. Student's t test was used to compare phosphorylation induced in the presence and absence of the Src blocker. The result of the analysis appears in the figure. C, effects of DPDPE and TICP on Src activity. Cells were serumstarved overnight (16 h) and then exposed to the indicated drug (1 M) for 5 min. Src activation was assessed as described under "Experimental Procedures" by measuring phospho-Tyr 416 immunoreactivity. Src phosphorylation was normalized according to the amount of protein present in each sample by expressing the data as a ratio of phospho-Tyr 416 over total Src immunoreactivity. Results represent mean Ϯ S.E. of seven experiments and are expressed as pSrc/totalSrc ratio. Statistical significance of drug effects was determined using one-way ANOVA followed by Dunnett's post hoc test to compare drugs to basal. **, p Ͻ 0.01. DPDPE suggest that Src could also be involved in the negative regulation of agonist signaling. This possibility was assessed by pretreating cells with DPDPE (1 M for 30 min) in the presence or absence of PP2 and by then assessing the ability of SNC-80 to induce GTP␥ 35 S binding. By itself, PP2 (20 M for 1 h) enhanced the ability of SNC-80 (100 nM) to promote GTP␥ 35 S binding (controls: 169 Ϯ 17 fmol/mg; PP2: 209 Ϯ 17 fmol/mg; p Ͻ 0.05; Fig. 7A), but this effect was accompanied by a marked increase in basal nucleotide binding (controls: 86 Ϯ 9 fmol/mg; PP2: 128 Ϯ 17 fmol/mg; p Ͻ 0.05, Fig. 7A). Therefore, to avoid any possible confounding, subsequent comparisons of the ability of SNC-80 to promote GTP␥ 35 S binding were expressed as percentage changes with respect to the corresponding nonstimulated condition under study. As shown in Fig. 7B, the efficacy of SNC-80 to induce nucleotide binding was greatly reduced following pre-treatment with DPDPE. However, if PP2 was introduced into the incubation medium before DPDPE, the ability of SNC-80 to induce GTP␥ 35 S binding was not significantly modified, confirming that inhibition of Src activity had a protective effect against agonist-induced desensitization.
Finally, to specifically examine whether Src may have contributed to the distinct kinetic profile of ERK activation by DPDPE and TICP, the time course for DPDPE was assessed in presence of 20 M PP2. In the absence of Src inhibition, ERK phosphorylation by DPDPE had completely disappeared within the first 30 min of incubation. In contrast, in the presence of PP2, ERK activity at 30 min was still 18 Ϯ 5% of the maximal response (which corresponds approximately to a 95% increase above ERK activity in non-stimulated cells; p Ͻ 0.01; two-way ANOVA; Fig.  7C). Moreover, PP2 had a stabilizing effect on ERK activation by DPDPE, because following 1-h incubation with the agonist, phosphorylation of the kinase was not significantly changed from the value observed 30 min before (16 Ϯ 5% of maximal).

DISCUSSION
The present study provides new insight into the regulation of agonistic responses induced by double efficacy ligands. In particular, results presented indicate that when occupied by this type of dual ligands ␦ORs elude phosphorylation and induce prolonged activation of the ERK cascade. These properties contrast with those of highly efficacious agonists, whose activation of the ERK cascade is transient and correlated with heavy phosphorylation and degradation of the receptor.
ERK was activated not only by drugs like SNC-80, DPDPE, TIPP, and morphine, which also produced agonistic effects in the cyclase pathway, but also by ligands like TICP and ICI174864 that stimulated ERK activity despite displaying inverse agonist behavior in cAMP accumulation assays (Fig. 1,  A and B). Inverse agonists are commonly thought to produce their actions via an inactive receptor conformation that precludes G protein signaling (12,13). However, it is improbable that activation of the ERK cascade by TICP or ICI174864 could be due to inhibition of spontaneous G i/o signaling, because the inactivation of G i/o proteins by PTX failed to produce any effect on ERK activity (Fig. 1C). Moreover, the fact that PTX abolished ERK activation by TICP indicates that, similar to agonists, ERK activation by dual efficacy ␦OR ligands requires the activation of a G i/o protein. This property is in marked contrast with dual efficacy ligands described previously for the ␤ 2AR whose activation of the ERK cascade was not dependent on G proteins (11).
The fact that TICP could stimulate G i/o proteins to activate ERK signaling and simultaneously block G i/o activity regulating the cyclase pathway may be explained by the fact that ␦ORs are promiscuous receptors, capable of interacting with more than one G i/o protein subtype (28,29). Hence, it is possible for ligands like TICP or ICI174864 to simultaneously display opposite agonist and inverse agonist behaviors in ERK and cAMP cascades by, respectively, stabilizing a conformation of the receptor that activates one G i/o subtype but inactivates another. In contrast, classic agonists would stabilize a conforma-FIG. 6. Src inhibitor PP2 produced opposite effects on ERK activation by DPDPE and TICP. HEK293s cells stably expressing full-length ␦ORs (0.5-1 pm/mg of protein) were serum-starved overnight (16 h) and the day of the experiment exposed to indicated concentration of PP2 or vehicle (Me 2 SO, 0.01%) 1 h before addition of TICP (A) or DPDPE (B) (1 M for 5 min). PhosphoERK1/2 immunoreactivity was measured as described before and expressed as the percent change of phosphoERK/ERK total ratio obtained in the absence of ␦OR ligand. Results represent mean Ϯ S.E. of at least four experiments. Statistical significance of the effect of PP2 on DPDPE and TICP responses was established using one-way ANOVA followed by Dunnett's post hoc test to compare ERK activation in the absence of PP2 to activation induced in presence of increasing concentrations of the Src inhibitor. *, p Ͻ 0.05. C, HEK293s cells stably expressing full-length ␦ORs were transiently transfected with the indicated concentrations of DNA encoding for kinase-impaired Src mutant (K296R/Y528F). Results and statistical analysis are as in B. On top of the representative immunoblots for ERK activity are the blots corresponding to the total amount of Src immunoreactivity present in cell lysates following transfection with the indicated quantities of Src K296R/Y528F. tion that activates G i/o proteins responsible for the regulation of both pathways.
Consistent with the idea that classic agonists and dual efficacy ligands may stabilize ␦ORs in different conformations is the observation that the rank order of efficacy with which TICP, TIPP, and morphine modified cyclase signaling was reversed with respect to their efficacies to promote ERK activation (Fig. 1, A and B). In fact, reversal in rank order of efficacy (or potency) for ligands that regulate more than one signaling cascade via the same receptor is considered as one of the most compelling proofs in favor of a model of ligand-specific active receptor states (30,31). In particular, the observed swap in position between TICP on the one hand and TIPP and morphine on the other makes it possible to conclude that the conformation via which the dual efficacy ligand stimulated ERK signaling is different from the one(s) involved in ERK stimulation by the two partial agonists.
The observations that receptors stabilized by DPDPE but not those occupied by TICP incorporated [ 32 P] (Fig. 3A) and were a target for the tyrosine kinase Src (Fig. 5, A and B), further support the idea that ␦ORs may exist in ligand-specific conformations and point to the fact that these different receptor states have distinct desensitization properties. The response observed for DPDPE is in keeping with previous reports showing that highly efficacious agonists promote ␦OR phosphorylation by G protein-coupled receptor kinases and tyrosine kinases (23,25,27,32,33). On the other hand, TICP resembles morphine in its ability to induce a receptor conformation that is poorly phosphorylated by receptor kinases (32). Although the failure of morphine to trigger phosphorylation of different opioid receptors has been frequently attributed to its stabilization of a receptor conformation that differs from the one stabilized by more efficacious agonists (34), there is some controversy as to whether lack of receptor phosphorylation is not just the reflection of the low efficacy of the drug (35,36). Given the fact that DPDPE and TICP induced similar levels of Src activity (Fig. 5C) differential tyrosine phosphorylation of ␦ORs by the two ligands (Fig. 5A) cannot be attributed to drug efficacy, further pointing to the existence of distinct conformations that are differentially recognized as Src substrates. At the same time, similar magnitude of Src activation by TICP and DPDPE poses the question: why do they differ in their ability to stimulate the MAPK (Fig. 1B)? A possible explanation to this observation would be that DPDPE activates ERK via more than one pathway, whereas TICP would only depend on Src.
Phosphorylation of Ser/Thr residues in the C-tail of ␦ORs is a major regulatory event that triggers the internalization (22,33) and desensitization (22,23) of the full-length receptor. Hence, it was reasoned that, if the differential phosphorylation of these residues by TICP and DPDPE contributed to their distinct kinetics of ERK activation, removal of the C terminus or interference with the process of internalization of the fulllength receptor should convey ERK activation by DPDPE some of the characteristics of the TICP response. Both of these interventions resulted in prolonged ERK stimulation by DPDPE, but only inhibition of internalization by sucrose prolonged the decay t1 ⁄2 of ERK activity to values within the range observed for dual efficacy ligands. Furthermore, the discrete increase in decay t1 ⁄2 associated with the effect of DPDPE in the truncated ␦OR could be further prolonged by sucrose, indicating that the time course of ERK responses generated by the truncated mutant was still dependent on internalization. The latter obser-  (16 h), and the day of the experiment exposed to PP2 (20 M) or vehicle (Me 2 SO, 0.01%) 1 h before addition of DPDPE (1 M) for the indicated times. ERK phosphorylation was measured as described previously, and results are expressed as the percentage of maximal response. They correspond to the mean Ϯ S.E. of at least four experiments. Curves were fit to one-phase exponential decay using GraphPad Prism 2.01. Calculated t1 ⁄2 values appear in the figure to the right of representative immunoblots obtained for each condition. Statistical comparison among curves was done using two-way ANOVA and p values are discussed in the text. vation, which is consistent with previous studies showing that in HEK293 cells this truncated mutant internalizes as the wild-type (37), stresses the determinant role played by receptor sequestration in the kinetics of ERK stimulation by agonists. On the other hand, the prolonged decay t1 ⁄2 associated with the stimulation of ERK by dual efficacy ligands (Fig. 2B) is consistent with previous results showing that 30-min exposure to ICI174864 did not change the total amount of ␦ORs present at the cell membrane (20).
Although inhibition of internalization slowed down the decay of ERK activity induced by the agonist, the response never attained the characteristic sustained profile observed for ERK activation by TICP ( Fig. 2A). This incapacity to recreate the complete "phenotype" of TICP stimulation, together with the fact that DPDPE produced a very transient activation of ERK in the truncated mutant, suggested that mechanisms different from phosphorylation of the C terminus could also contribute to the distinct kinetics of ERK stimulation by DPDPE and TICP.
Because only DPDPE stabilized ␦ORs in a conformation that was recognized by Src and given that tyrosine residues have been implicated in the regulation of ␦OR signaling (26,27), one possibility that was assessed was whether Src could distinctively regulate ERK activation by agonists and dual efficacy ligands. Low concentrations of Src inhibitor PP2 (20 -40 M) blocked ERK activation by TICP but enhanced the response to DPDPE (Fig. 6, A and B) revealing that, indeed, Src had a distinct effect on ERK responses elicited by the two types of drugs. The ability of low concentrations of PP2 (Fig. 6B) or very discrete amounts of inactive Src (Fig.  6C) to increase the magnitude of ERK activation by DPDPE may be interpreted as an indication that Src negatively controls ␦OR responsiveness to agonists, an assumption that was confirmed in GTP␥ 35 S binding assays, where PP2 was found to protect against agonist-induced desensitization. On the other hand, the fact that ERK activation by TICP was blocked at all levels of Src inhibition (Fig. 6A) not only indicates that this non-receptor tyrosine kinase is an intermediate in ERK stimulation by the dual efficacy ligand but also argues against a regulatory role of Src in the effects of these drugs. The observation that DPDPE-dependent activation of ERK was blocked by high concentrations of PP2 (Fig. 6B) or by the transfection of high quantities of inactive Src indicate that agonists do rely on Src for the stimulation of the MAPK. However, this effect is not apparent at the same level of Src inhibition at which the effect of TICP is blocked, due to the additional regulatory effect of Src on agonist responses.
The interpretation that Src may have contributed to determine the transient kinetics of ERK activation by DPDPE is supported by the fact that PP2 had a stabilizing effect on ERK stimulation by the agonist. Indeed, although PP2 had no significant effect on the decay t1 ⁄2 , it was shown to prolong DPDPE responses by preventing complete fading of ERK activity after the first 10 min of stimulation (Fig. 7C). The mechanism whereby Src inhibition had this stabilizing effect on DPDPE responses is not clear. Although Src has been implicated in the regulation of receptor endocytosis, including ␦ORs (27,38,39), the fact that the effect of PP2 had no resemblance to that of sucrose indicates that inhibition of sequestration is probably not the main mechanism involved. On the other hand, results from GTP␥ 35 S binding assays showing that PP2 enhanced agonist-induced nucleotide binding as well as basal GTP␥ 35 S binding activity (Fig. 7A) suggest that Src may also influence ␦OR signaling by reducing ␦OR-G protein coupling. It is possible then, that the stabilizing effect of PP2 on DPDPE-dependent activity could be linked to a better coupling between agonist stabilized receptors and the corresponding G protein.
In conclusion, this study showed that the agonistic responses of dual efficacy ligands for ␦ORs were more sustained and decayed much slower than those of classic agonists. These kinetics were associated with the distinct ability of dual efficacy ligands to stabilize ␦ORs in an active conformation that does not trigger the same regulatory mechanisms as classic, highly efficacious agonists.