|Home | About | Journals | Submit | Contact Us | Français|
Opiates are analgesics of choice in the treatment of chronic pain, but their long-term use leads to the development of physiological tolerance. Thus, understanding the mechanisms modulating the response of their receptor, the μ opioid receptor (μOR), is of great clinical relevance. Here we show that heterodimerization of μOR with δ opioid receptors (δOR) leads to a constitutive recruitment of β-arrestin2 to the receptor complex resulting in changes in the spatio-temporal regulation of ERK1/2 signaling. The involvement of β-arrestin2 is further supported by studies using β-arrestin2 siRNA in cells endogenously expressing the heterodimers. The association of β-arrestin2 with the heterodimer can be altered by treatment with a combination of μOR agonist (DAMGO) and δOR antagonist (TippΨ), and this leads to a shift in the pattern of ERK1/2 phosphorylation to the pattern observed with μOR alone. These data indicate that, in the naive state, μOR-δOR heterodimers are in a conformation conducive to β-arrestin-mediated signaling. Destabilization of this conformation by cotreatment with μOR and δOR ligands leads to a switch to a non-β-arrestin-mediated signaling. Taken together, these results show for the first time that μOR-δOR heterodimers, by differentially recruiting β-arrestin, modulate the spatio-temporal dynamics of opioid receptor signaling.
The opioid receptors μOR (μ-opioid receptor), δOR (δ-opioid receptor), and κOR (κ-opioid receptor) (1) are members of the family A of G-protein-coupled receptors (GPCRs). They are involved in many biological responses, including analgesia, miosis, bradycardia, feeding, and hypothermia (2). Activation of these receptors leads to inhibition of adenylyl cyclase activity, phosphorylation of extracellular signal-regulated kinase 1/2 (ERK), activation of K+ currents, and inhibition of Ca2+ channels (3). Previous studies have shown that morphine functions primarily by activating μOR (3). Furthermore, studies have suggested that μOR interacts with δOR, and this leads to changes in receptor properties (4). For example, mice treated with δOR antagonists exhibit diminished morphine tolerance and dependence (5, 6). δOR knockout animals do not develop morphine tolerance (6), and reducing the surface insertion of δOR abolishes morphine tolerance (7). In addition, chronic morphine treatment up-regulates δOR (8), which leads to changes in μOR function (7).
Recent studies have directly examined the interaction between μOR and δOR using a variety of techniques such as coimmunoprecipitation and bioluminescence resonance energy transfer, and have shown that dimerization affects ligand binding and receptor signaling (9, 10, 11). These studies have suggested that the μOR-δOR heterodimer could represent a functional unit distinct from μOR or δOR. This is supported by the fact that μOR agonist-induced signaling can be enhanced by cotreatment with δOR-selective ligands, including δOR-selective antagonists (that do not elicit a signal when administered alone) (9). This raises the exciting possibility that the signaling pathways activated by μOR-δOR heterodimers are distinct from those activated by μOR or δOR alone. In this study, we show that μOR-δOR heterodimers recruit β-arrestin, which leads to changes in the spatio-temporal dynamics of opioid-mediated ERK activation. These findings have major clinical relevance since modulation of ERK phosphorylation has been shown to play an important role in regulating pain and analgesia (12).
Chinese hamster ovary (CHO) K1, human embryonic kidney (HEK) −293, Neuro2A, and SKNSH cells from ATCC (Manassas, VA, USA) were maintained in F12 or DMEM + 10%FBS at 37°C in a humidified 5% CO2 incubator. CHO-δOR (CHO cell line stably expressing FLAG-δOR) is described elsewhere (13). Rabbit polyclonal anti-FLAG antibody, Tyr-D-Ala-Gly-N-Me-Phe-Gly-ol (DAMGO), Tyr-D-Ala-Phe-Glu-Val-Val-Gly (Deltorphin II), and G418 were from Sigma (St. Louis, MO, USA). D-Phe-Cys-Tyr-D-Trp-Orn-Thr-Pen-Thr-NH2 (CTOP) was from Peninsula Inc. (Belmont, CA, USA). Calphostin C and pertussis toxin were from Calbiochem (San Diego, CA, USA). Anti-ERK polyclonal, antiphospho-ERK monoclonal, antiphospho Stat3 (Ser-727, polyclonal, antiphospho-p90rsk polyclonal, antilamin A/C polyclonal, anti-HA monoclonal, and antiphospho-δOR (Ser-363, polyclonal antibodies were from Cell Signaling Technology Inc. (Danvers, MA, USA). Anti-HA polyclonal antibody was from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Rabbit polyclonal β-arrestin2 antibody (A2CT) was a gift from Drs. R. Lefkowitz (Duke University, Durham, NC, USA) and L. Bohn (Ohio State University College of Medicine, Columbus, OH, USA). The anti-β-arrestin polyclonal antibody was from Calbiochem. Monoclonal anti-GAPDH antibody was from Novus Biologicals (Littleton, CO, USA). Horseradish peroxidase-conjugated secondary antibodies were from Amersham Biosciences (Arlington Heights, IL, USA). Sulfo-NHS-biotin and avidin-coupled agarose were from Pierce (Rockford, IL, USA). Tyr-TicΨ (CH2NH)-Phe-Phe (TippΨ) was a gift from Dr. Peter Schiller (Institut de Recherches Cliniques de Montreal, Canada). The HA-μOR plasmid was a gift from Dr. Liu Chen (Temple University School of Medicine, Philadelphia, PA, USA).
Transfections were performed when cells were 80–90% confluent using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions, with 2.5 μg plasmid DNA (unless otherwise indicated).
Cells were lysed for 1 h in lysis buffer (1% Nonidet P-40, 10% glycerol, 300 mM NaCl, 1.5 mM MgCl2, 1 mM CaCl2, and 50 mM Tris-Cl, pH 7.4) containing protease inhibitor cocktail (Sigma). For immunoprecipitation, 100–200 μg of protein was incubated with 1 μg of polyclonal anti-FLAG or monoclonal anti-HA antibody with 10% v/v protein A-agarose (Sigma) overnight at 4°C. The beads were washed twice with lysis buffer and phosphate-buffered saline (PBS) containing 10 mM EDTA. Proteins were eluted in 50 μl of 2 × Laemmli buffer containing 1% 2-mercaptoethanol. Proteins were resolved by 10% SDS-PAGE and subjected to Western blot using monoclonal anti-HA or polyclonal anti-β-arrestin antibodies. Chemiluminescence detection was performed using the SuperSignal West Pico reagent (Pierce, Rockford, IL, USA). Multiple exposures of immunoblots were scanned and densitometric analysis was carried out within linear range by using Image J software. GraphPad PRISM software was used for data analysis.
CHO, CHO-δOR, or HEK-293 cells were transfected with HA-μOR plasmid. Cells were seeded on 24-well plates 24 h post-transfection. The next day, the cells were starved for at least 6 h in serum-free medium prior to stimulation with 100 nM DAMGO for the indicated times. In some cases cells were preincubated for 30 min with the indicated kinase inhibitor, followed by treatment with DAMGO in the presence of this inhibitor. Cells were solubilized by directly adding 1 × SDS buffer prewarmed to 65°C, followed by sonication with a microtip for 5 s. For each transfection, protein determination (Bradford) was carried out, then 30 μg protein was separated by 10% SDS-PAGE and transferred to nitrocellulose membranes for immunoblotting with mouse monoclonal antiphospho-p44/42 MAPK (anti-pERK, 1:1000) and rabbit polyclonal antip44/42 MAPK (anti-ERK, 1:1000).
For ERK localization, CHO and CHO-δOR cells growing on 6-well plates were transfected with HA-μOR. The next day cells were plated on 14 mm coverslips. On the following day, cells were starved for at least 6 h and stimulated with 100 nM DAMGO for the times indicated. Cells were stained with the monoclonal pERK antibody according to the manufacturer’s protocol. The cells were incubated with DAPI for 5 min, then washed four times with PBS.
CHO and CHO-δOR cells growing on 6-well plates were transfected with HA-μOR and β-arrestin2-EGFP plasmids. Cells were plated on 14 mm coverslips 24 h post-transfection. The next day, cells were fixed in −20°C methanol and immunostained with primary antibodies against HA (1:2000) and Flag (1:150), and against secondary Alexa 594-coupled donkey anti-mouse and Cy5-coupled donkey anti-rabbit antibodies. The same procedure was carried out with Neuro2A and HEK cells transfected with Flag-δOR, HA-μOR, and β-arrestin2-EGFP plasmids. Coverslips were mounted with Mowiol and visualized with a Leica TCS SP1 confocal microscope equipped with four external lasers (350, 488, 568, and 633 nm, Leica Microsystem). Images were acquired with an ×63/1.32 PL APO objective lens and analyzed in sequential scanning mode.
CHO or CHO-δOR cells expressing HA-μOR were grown in 6-well plates, serum starved for 6 h, and stimulated with 100 nM DAMGO for the indicated times. Nuclear and cytoplasmic fractions were prepared as described elsewhere (14). Immunoblotting for pERK and ERK was performed as described above. Immunoblotting of the nuclear protein lamin A/C with the anti-Lamin antibody (1:2000) and of the cytoplasmic protein GAPDH (1:2000) was performed to assess the purity of the fractions.
SKNSH cells at 70% confluency in 6-well plates were stimulated with 100 nM DAMGO, 20 nM TippΨ, or a mixture of both ligands for 5 min. After stimulation, cells were washed twice with ice-cold PBS/Ca2+/Mg2+ and incubated with sulfo-NHS-Biotin (Pierce) for 1 h at 4°C. Biotin was quenched by incubating the cells with 0.1 M glycine for 30 min at 4°C, then the cells were solubilized in immunoprecipitation lysis buffer for 30 min. The cell lysate was spun down to eliminate insoluble material and the supernatant was incubated with avidin-coupled agarose (Pierce) overnight at 4°C. The beads were washed four times in lysis buffer and eluted in 50 μl of 2 × SDS sample buffer. Proteins were resolved by 10% SDS-PAGE for immunoblotting using anti-β-arrestin2 antibody.
Chemically synthesized, double-stranded siRNAs targeting human β-arrestin2 and human μOR were purchased from Dharmacon (Lafayette, CO, USA). A nonsilencing RNA duplex was used as a control. SKNSH cells on 6-well plates were transfected with 100 nM of siRNA and HA-μOR plasmid. The next day cells were split into 24-well plates for pERK assays. HEK cells were similarly transfected with the siRNAs.
CHO or CHO-δOR cells were transfected with HA-μOR and Ly6E Stat3-response luciferase reporter construct (15). After 24 h, the cells were split into 6-well plates. The next day cells were starved for 6 h in serum-free media, stimulated for 30 min with 100 nM DAMGO, and lysed in lysis buffer (Promega, Madison, WI, USA). Firefly luciferase activity was measured using a fluorometer as described (15). Renilla luciferase activity was measured as a control.
We and others have previously shown that μOR and δOR associate to form heterodimers that display new pharmacological properties (9–11, 16). In an effort to explore the signaling mechanisms of μOR-δOR heterodimers, we examined the ability of μOR-δOR to recruit β-arrestin. Since little is known about β-arrestin recruitment by μOR-δOR heterodimers, first we examined the subcellular localization of β-arrestin in HEK cells expressing μOR-δOR and compared it to cells individually expressing μOR or δOR. Confocal immunofluorescence studies showed that in cells expressing μOR or δOR, β-arrestin2 is located exclusively in the cytoplasm and does not colocalize with the receptors, which are found mostly at the plasma membrane. In contrast, in cells coexpressing μOR and δOR, β-arrestin2 is primarily located at the plasma membrane, where it colocalizes with the receptors (Fig. 1A). This suggests a constitutive recruitment of β-arrestin2 by the heterodimer. Quantification indicates that in > 94% of cells that coexpress μOR and δOR, β-arrestin2 is colocalized with the receptors (R. Rozenfeld and L. A. Devi, unpublished results). To ensure that coexpression of μOR and δOR leads to surface localization of β-arrestin2 independent of the cell system used, we carried out studies in two other cell lines. Examination of the neuroblastoma cell line Neuro2A also revealed a plasma membrane localization of β-arrestin2 only in cells coexpressing μOR and δOR, and not in cells expressing either one of these receptors alone (Fig. 1B). Similar results were obtained in CHO cells (Fig. 1C). To study the physical association of β-arrestin2 with μOR-δOR heterodimers, we carried out immunoprecipitation experiments. We find that β-arrestin co-immunoprecipitates with the receptors only in cells expressing μOR-δOR heterodimers but not in cells expressing only μOR or δOR (Fig. 1D). The fact that β-arrestin2 associates with μOR-δOR in the absence of ligand treatment suggests that μOR-δOR heterodimers constitutively recruit β-arrestin2.
β-Arrestins have gained recognition in recent years as modulators of signal transduction of many GPCRs predominantly by activating signaling pathways such as the ERK pathway (17). β-Arrestin2-dependent ERK activation has been shown to be slower in onset and more sustained than the classic (β-arrestin-independent) pathway of ERK activation (18–20). We thus examined heterodimer-mediated ERK phosphorylation and compared it to μOR-mediated ERK phosphorylation. For this, we examined the time course of DAMGO-induced ERK phosphorylation in HEK cells transiently expressing both μOR and δOR, and compared it with the time course in cells expressing only μOR. In cells expressing μOR, DAMGO treatment leads to a rapid and transient increase of phosphoERK (pERK), with a peak signal at ~3–5 min. The level of pERK rapidly declines to < 30% of the maximal response after 10 min (Fig. 2, left panel). In μOR-δOR cells, in addition to the first pERK peak, DAMGO treatment leads to a slower increase in pERK, with a sustained second phase at ~15–20 min (Fig. 2, right panel). This suggests that the sustained phase of pERK is mediated by the activation of μOR-δOR heterodimers.
To test this further, we examined the effect of varying the levels of μOR-δOR on the kinetics of pERK. For this, we used CHO cells stably expressing δOR (CHO-δOR) and transfected them with different amounts of HA-μOR cDNA. As anticipated, transfection with 5, 3.75, and 2.5 μg of HA-μOR cDNA yielded decreasing expression levels of μOR (lanes 1, 3, and 5 in Fig. 3A) without affecting δOR expression levels (Fig. 3A, IB: anti-FLAG). HA-μOR could be coimmunoprecipitated with δOR under the three transfection conditions (lanes 2, 4, and 6); however, the ratio of μOR associated with δOR to total μOR varied with the different expression levels of μOR, suggesting that μOR-δOR heterodimerization is modulated by the expression level of the receptors. Under conditions of low μOR expression (2.5 μg of cDNA), heterodimerization with μOR is favored compared with conditions where μOR is highly expressed (5 μg of cDNA); under these conditions, μOR appears to be mostly in homodimeric form (Fig. 3A). As the ratio of heterodimer to total μOR increased, we observed a shift in the peak of pERK. For example, at highest heterodimer levels pERK peaks at 15 min, and at lowest heterodimer levels it peaks at 3–5 min (Fig. 3B, C), a time course reminiscent of that seen in μOR alone (Fig. 3B, upper panel). To ensure that these changes in pERK kinetics were not mediated by δOR activation, we examined ERK phosphorylation in δOR-expressing cells on DAMGO treatment. Under these conditions, 100 nM DAMGO treatment did not lead to detectable ERK phosphorylation (data not shown). We also checked that DAMGO treatment did not lead to dissociation of the heterodimer by coimmunoprecititation experiments. We found that upon stimulation, μOR is still present in the δOR immunoprecipitate (Supplemental Fig. 2). Taken together, these results are consistent with the notion that the extent of heterodimerization regulates the dynamics of ERK signaling.
Several studies have reported that μOR-mediated ERK phosphorylation involves a PKC-dependent mechanism (21–23). We examined the involvement of PKC activity in μOR-δOR-mediated ERK phosphorylation using the PKC inhibitor, calphostin C. For this purpose, we examined DAMGO-induced ERK phosphorylation in CHO-δOR transfected with 2.5 μg of HA-μOR cDNA (to favor heterodimer formation) and compared it to cells expressing μOR alone. In vehicle-treated μOR cells (Fig. 4A, top panels), DAMGO treatment leads to ERK phosphorylation, with a peak at 3–5 min, whereas in μOR-δOR cells this treatment leads to a sustained phosphorylation at 10–20 min. In μOR-expressing cells, treatment with calphostin C inhibits ERK phosphorylation at early time points. While a component of the response at late time points is calphostin C-resistant, it represents only a minor portion compared with the total pERK response (Fig. 4A, lower panel). In contrast, in μOR-δOR-expressing cells, this treatment does not inhibit DAMGO-induced ERK phosphorylation at any time point examined (Fig. 4B, lower panel). On the contrary, pertussis toxin treatment inhibits ERK phosphorylation in both μOR and μOR-δOR cells (Supplemental Fig. 3). We examined the possibility that DAMGO-induced ERK phosphorylation by the heterodimer involves transactivation of δOR. For this, we used phosphorylation of δOR C-terminal tail as a readout of its activation. We find that δOR is not phosphorylated under any condition examined, except on direct activation by the specific agonist deltorphin II (Supplemental Fig. 4). This suggests that the changes in pERK kinetics seen in μOR-δOR cells are not due to transactivation of δOR.
Together with results showing that the early phase of ERK phosphorylation is PKC dependent but the pERK response at later times is not, these findings support the notion that the signal transduction pathways activated by the μOR-δOR heterodimer are distinct from the pathways activated by μOR.
To critically evaluate the role of β-arrestin2 in μOR-δOR heterodimer-mediated ERK phosphorylation, we used β-arrestin2 siRNA (to deplete endogenous β-arrestin2) in SKNSH cells that endogenously express μOR-δOR heterodimers (9, 11). In these cells, DAMGO treatment leads to a rapid peak of pERK at 5 min, followed by a second, less intense sustained phase of pERK at 10–15 min (Fig. 5). We find that treatment with β-arrestin2 siRNA leads to a substantial decrease in endogenous protein (Fig. 5, inset). Under these conditions, there is a significant decrease in the late sustained phase of ERK phosphorylation (Fig. 5). This treatment also leads to an increase in the early phase of ERK phosphorylation (Fig. 5). As a control, β-arrestin2 siRNA treatment did not affect the kinetics of ERK phosphorylation in HEK cells expressing μOR alone (not shown). These results showing that depletion of β-arrestin2 affects the μOR-δOR heterodimer-mediated signaling suggest that the heterodimer signals through a β-arrestin2-dependent pathway.
A provocative set of studies with other GPCRs has shown that a subset of ligands is able to specifically modulate the extent of arrestin association with their receptors (19, 24). We applied this to opioid receptor heterodimers, and examined the effect of a mixture of μOR and δOR specific ligands on the association of β-arrestin with the heterodimer. We chose the μOR agonist DAMGO and the δOR antagonist TippΨ, since in previous studies this combination was found to enhance G-protein coupling and signaling by the heterodimer (9, 11).
First, we examined whether treatment with these two ligands would lead to a dissociation of β-arrestin2 from the heterodimer. Immunoprecipitation experiments showed that in untreated cells or in cells treated with individual ligands, β-arrestin2 was associated with the heterodimer (Fig. 6A). In contrast, in cells treated with a combination of ligands, β-arrestin2 was not associated with the heterodimer, as evidenced by the lack of β-arrestin2 in the δOR immunoprecipitate (Fig. 6A). Next, using cell surface biotinylation, we examined whether this combination of ligands would lead to the dissociation of β-arrestin2 from endogenous receptors in SKNSH cells. In untreated cells or in cells treated with DAMGO or TippΨ, β-arrestin2 was present in the biotinylated fraction (Fig. 6B), suggesting that it associates with membrane proteins (i.e., opioid receptors). Confocal microscopy experiments also show that in SKNSH cells, a large fraction of transfected β-arrestin2-EGFP is localized at the plasma membrane (Supplemental Fig. 1). In contrast, in cells treated with a combination of ligands, β-arrestin2 was not detected in the biotinylated fraction. These results are consistent with the notion that, in the unstimulated state, heterodimerization stabilizes μOR-δOR in a conformation conducive to association with β-arrestin2. Dual occupancy of the heterodimer (by simultaneous treatment with the δOR antagonist and DAMGO) leads to destabilization of this conformation, resulting in a dissociation of β-arrestin2.
Next we examined the effect of β-arrestin2 dissociation (induced by the combination of ligands) on ERK phosphorylation. We find that cotreatment with DAMGO and TippΨ leads to a significant decrease in the late phase of ERK phosphorylation and to an increase in the early phase of ERK phosphorylation (Fig. 6C). This pattern is similar to that observed with β-arrestin2 siRNA treatment (Fig. 5). Taken together, these results suggest that dual occupancy of μOR-δOR (by cotreatment with μOR and δOR ligands) leads to changes in the conformation of the heterodimer that results in the dissociation of β-arrestin2, enabling rapid ERK phosphorylation.
In a parallel series of studies, we examined the effects of heterodimerization on δOR signaling. In CHO cells expressing δOR alone, treatment with deltorphin II leads to a rapid and transient increase in ERK phosphorylation (Fig. 7A). In contrast, in μOR-δOR-expressing SKNSH cells, this treatment leads to a slow and sustained late pERK response in addition to the rapid first peak (Fig. 7B, top panel). We examined whether heterodimerization with μOR could account for this pattern of ERK phosphorylation. For that we used siRNA to μOR to down-regulate the μOR expression level. SiRNA to μOR decreased μOR expression by 70–80% (Supplemental Fig. 5A) and abolished μOR-mediated ERK signaling (Supplemental Fig. 5B). Under these conditions, there is a significant decrease in the late sustained phase of deltorphin II-induced ERK phosphorylation (Fig. 7B). These results indicate that heterodimerization with μOR is necessary for the second phase of δOR-mediated signaling. We tested whether this regulation of δOR signaling by heterodimerization involved a β-arrestin2-dependent mechanism. SiRNA-mediated β-arrestin2 down-regulation leads to a similar shift in deltorphin II-induced pERK kinetics. Then we examined whether occupying μOR and δOR binding sites affects the time course of ERK phosphorylation. Cotreatment with deltorphin II and a μOR antagonist (10 nM CTOP) also leads to a shift in pERK kinetics to a pattern similar to that observed in cells expressing δOR alone.
Taken together, these results indicate that hetrodimerization affects the signaling properties of δOR, and this regulation involves a β-arrestin2-mediated pathway, suggesting a common β-arrestin2-mediated modulation of μOR and δOR signaling by heterodimerization.
We examined whether, in addition to changes in the kinetics of ERK phosphorylation, heterodimerization leads to alterations in the subcellular localization of pERK. In μOR-expressing cells, DAMGO-induced pERK is largely localized to a nuclear compartment, as revealed by confocal microscopy (Fig. 8A, left panel). In contrast, in μOR-δOR-expressing cells, the majority of pERK is confined to the cytoplasm (Fig. 8A, right panel). Nuclear localization was not seen even on prolonged (20 min) DAMGO exposure (not shown). To confirm the differential localization of pERK, we carried out subcellular fractionation studies. The purity of the cytoplasmic and nuclear fractions was verified by probing fractions with antilamin A/C antibody (nuclear marker) and anti-GAPDH antibody (cytoplasmic marker) (Fig. 7B). In μOR-expressing cells, μOR agonist treatment results in localization of the majority (>70%) of pERK to the nuclear fraction within 5 min. In contrast, in μOR-δOR cells the same treatment results in localization of < 10% of pERK to the nuclear fraction even after 15 min stimulation; the majority of pERK (>80%) is restricted to the cytoplasm (Fig. 8B). Thus, while opioid treatment in μOR-expressing cells results in mobilization of pERK to the nucleus, the same treatment in μOR-δOR heterodimer-expressing cells results in retention of pERK in the cytoplasm.
We next examined the functional consequences of differential localization of pERK in μOR vs. μOR-δOR cells by examining the phosphorylation of cytoplasmic (p90rsk) and nuclear (serine 727 of Stat3) pERK substrates (pERK-mediated Stat3 activation by μOR has been reported; ref. 25). We find that DAMGO treatment leads to a rapid and transient phosphorylation of p90rsk in μOR cells and a slow and sustained phosphorylation of p90rsk in μOR-δOR cells (Fig. 9A). This pattern is similar to that of ERK phosphorylation and agrees with the cytoplasmic localization of pERK in μOR-δOR cells. In contrast, Stat3 phosphorylation at its nuclear phosphorylation site, Ser-727 (26), is seen only in μOR-expressing cells and not in μOR-δOR-expressing cells (Fig. 9A), even after 30 min stimulation (not shown). This agrees with the nuclear localization of pERK in μOR cells. Next we examined whether the differential phosphorylation of Stat3 affects its transcriptional activity. Stat3-mediated transcriptional activity was measured using a Stat3 luciferase reporter gene. We find that DAMGO treatment leads to a significant increase in Stat3 activity only in μOR- but not in μOR-δOR-expressing cells (Fig. 9B). These results support the idea that μOR-δOR heterodimerization restricts pERK to the cytoplasm, thereby preventing its ability to activate nuclear substrates such as the transcription factor Stat3. Thus, the subcellular localization of pERK, which is critical to the outcome of downstream signaling pathways, is modulated by μOR-δOR heterodimerization.
In the present study, we show that heterodimerization with δOR alters μOR-mediated signaling, resulting in spatio-temporal changes in the dynamics of ERK phosphorylation. We find that the slow and sustained ERK phosphorylation (seen as a second phase) is mediated by the μOR-δOR heterodimer. In cells that express high levels of heterodimers (CHO cells, Fig. 3C), there is a robust second phase, and as the levels of the heterodimer are reduced, the relative robustness of the second phase decreases (Fig. 3A–C). Under these conditions there is a reciprocal increase in the levels of the μOR homodimer, and this corresponds to an increase in the first peak of ERK phosphorylation. Thus, the level of the first peak of ERK phosphorylation correlates with the level of homodimers, and the level of the second phase correlates with the level of heterodimers. This is further supported by our data with various cell systems expressing different levels of the heterodimer (Figs. 2, ,3,3, and and6).6). The μOR-δOR heterodimer-induced second phase of ERK response is mediated by β-arrestin2, and thus the heterodimer constitutes a unique signaling unit. Depletion of β-arrestin2 from cells expressing μOR-δOR leads to a pattern of ERK phosphorylation similar to that observed in cells expressing μOR alone. This is consistent with other studies that have shown altered G-protein coupling and channel activation by μOR-δOR heterodimers (16, 27).
β-Arrestin-mediated ERK phosphorylation has been described for several GPCRs, albeit the majority of these studies have been in cells expressing only homodimers (for a review, see ref. 28). The best-studied example is that of the AT1 receptor, for which two distinct pathways of ERK activation have been uncovered: a PKC-dependent pathway that leads to transient ERK phosphorylation and targets pERK to the nucleus, and a β-arrestin-dependent pathway that leads to sustained ERK phosphorylation and targets pERK to the cytosol and endosomes (18). This spatial and temporal segregation of ERK activated by PKC and β-arrestin pathways has been shown to lead to the activation of distinct downstream signaling cascades (29, 30). This raises the possibility that conditions that selectively stimulate or inhibit one of these pathways could have significant physiological relevance. Indeed, the activation of distinct signaling pathways by specific ligands has been described recently for β2-adrenergic (24) and parathyroid hormone (20) receptors. In both studies, specific ligands were shown to stabilize their cognate receptor in conformations conducive to either G-protein-dependent or β-arrestin-dependent ERK phosphorylation. Here we show that heterodimerization can also stabilize receptors in a conformation conducive to activating a specific signaling pathway. This allows the same ligand (DAMGO) to activate distinct pathways of ERK phosphorylation (i.e., a PKC-dependent or a β-arrestin-dependent pathway), based on the absence or presence of the dimerizing partner (in this case, δOR). Note that β-arrestin-mediated ERK activation is sensitive to pertussis toxin, suggesting that this also requires G-protein activation. It is conceivable that the homo-and heterodimeric receptors are organized as higher order complexes that include G-proteins, scaffolding molecules (such as β-arrestin), and other signaling molecules. Within the complex, receptor activation could initiate a sequence of events involving G-protein activation, a component of which could induce changes in the conformation of β-arrestin needed to enhance ERK phosphorylation. Further studies are needed to validate such a model. Taken together, these results show for the first time that μOR-δOR heterodimers, by recruiting β-arrestin, modulate the dynamics of receptor signaling and its downstream effects. These findings have potential clinical relevance for modulating morphine effects.
Heterodimerization of μOR and δOR has been implicated in the regulation of morphine-induced analgesia (11). We previously showed that modulating μOR-δOR heterodimer activity by occupying the binding site of one protomer (δOR) changes the agonist-induced response of the partner (μOR) (9, 11). For example, coadministration of a δOR ligand with morphine results in an enhancement of morphine-induced analgesia. Animals lacking β-arrestin2 also show enhanced morphine-induced analgesia (31). Here we show that β-arrestin knockdown or dual occupancy of the heterodimer (that leads to dissociation of the heterodimer-β-arrestin complex) leads to a pattern of ERK phosphorylation similar to that of μOR. In preliminary studies we find that cotreatment with a δOR antagonist leads to changes in morphine-mediated signaling similar to those observed with DAMGO (R. Rozenfeld and L. A. Devi, unpublished results). This presents an opportunity to selectively target heterodimers with a combination of morphine and a δOR antagonist. This would lead to a switch in signaling from a μOR-δOR (β-arrestin-dependent) to a μOR (G-protein-dependent)-mediated pathway resulting in an increase in the analgesic response of μOR, and thus enhanced morphine effects.
Interactions between μOR and δOR have also been implicated in the development of morphine tolerance (32). Previous studies have reported the modulation of morphine tolerance by δOR using animals lacking δOR (6) or with impaired δOR expression (7), or by cotreatment with δOR ligands (5). This, taken with the finding that animals lacking β-arrestin2 fail to develop morphine tolerance (33), supports a role for μOR-δOR-mediated recruitment of β-arrestin2 in regulating opiate tolerance.
Taken together, these findings showing that heterodimerization of μOR and δOR modulates μOR agonist-induced ERK activation and its downstream signaling pathways suggest a provocative role for heterodimerization in opiate analgesia and tolerance.
We thank I. Gomes, F. Decaillot, and J. Morón for helpful discussion, and N. Abul-Husn for advice and help in writing this manuscript. Supported by National Institutes of Health grants (DA08863 and DA019521 to L.A.D. and R24 CA095823 to MSSM-Microscopy Shared Resource Facility).