Heterodimerization of Substance P and μ-Opioid Receptors Regulates Receptor Trafficking and Resensitization*

The μ-opioid receptor (MOR1) and the substance P receptor (NK1) coexist and functionally interact in nociceptive brain regions; however, a molecular basis for this interaction has not been established. Using coimmunoprecipitation and bioluminescence resonance energy transfer (BRET), we show that MOR1 and NK1 can form heterodimers in HEK 293 cells coexpressing the two receptors. Although NK1-MOR1 heterodimerization did not substantially change the ligand binding and signaling properties of these receptors, it dramatically altered their internalization and resensitization profile. Exposure of the NK1-MOR1 heterodimer to the MOR1-selective ligand [d-Ala2,Me-Phe4,Gly5-ol]enkephalin (DAMGO) promoted cross-phosphorylation and cointernalization of the NK1 receptor. Conversely, exposure of the NK1-MOR1 heterodimer to the NK1-selective ligand substance P (SP) promoted cross-phosphorylation and cointernalization of the MOR1 receptor. In cells expressing MOR1 alone, β-arrestin directs the receptors to clathrin-coated pits, but does not internalize with the receptor. In cells expressing NK1 alone, β-arrestin internalizes with the receptor into endosomes. Interestingly, in cells coexpressing MOR1 and NK1 both DAMGO and SP induced the recruitment of β-arrestin to the plasma membrane and cointernalization of NK1-MOR1 heterodimers with β-arrestin into the same endosomal compartment. Consequently, resensitization of MOR1-dependent receptor functions was severely delayed in coexpressing cells as compared with cells expressing MOR1 alone. Together, our findings indicate that MOR1 by virtue of its physical interaction with NK1 is sequestered via an endocytotic pathway with delayed recycling and resensitization kinetics.

The formation of homo-and heterodimeric receptors seems to be a general principle for many, if not all, G protein-coupled receptors (GPCRs), 1 and for some GPCRs, e.g. the GABA B and the T1 taste receptor. It appears to be an absolute requirement for functional activity (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11). We have previously shown that opioid and somatostatin receptors exist as constitutive homodimers when expressed alone and as constitutive heterodimers when coexpressed in human embryonic kidney (HEK) 293 cells (12)(13)(14). Heterodimerization of somatostatin receptor subtypes modulates their ligand binding, signaling, and trafficking properties, e.g. the sst 2A -sst 3 somatostatin receptor heterodimer behaved like the sst 2A homodimer but it did not reproduce the pharmacological characteristics of the sst 3 homodimer, suggesting that physical interaction of sst 3 with sst 2A induced a functional inactivation of the sst 3 subtype (13). Unlike that observed for the sst 2A -sst 3 heterodimer, sst 2A -opioid receptor (MOR1) heterodimerization did not significantly affect the ligand binding or coupling properties but promoted cross-modulation of phosphorylation, internalization, and desensitization of these receptors (14). Given the high level of sequence homology between opioid and somatostatin receptors existing in domains proposed to contribute to the dimerization interface (i.e. transmembrane helices) it was not unexpected that these closely related receptors can form heterodimers. However, so far little is known about dimerization of opioid receptors with distantly related GPCRs.
A major prerequisite for the physiological assembly of heterodimeric GPCRs in vivo is their coexpression in the same cell. MOR1 and NK1, the principal receptor for substance P (SP), coexist and functionally interact in pain-processing brain regions (15)(16)(17)(18)(19)(20)(21)(22). SP is released from nociceptive primary afferents in the spinal and trigeminal dorsal horn, where it activates spino-thalamic projection neurons. Recently, NK1 and MOR1 have been shown to coexist in trigeminal dorsal horn neurons, suggesting that MOR and NK1 functionally interact in these neurons during nociceptive neurotransmission. Furthermore, MOR1 and NK1 are highly expressed in brain regions implicated in depression, anxiety, and stress, but also in other regions such as the nucleus accumbens, which mediate the motivational properties of drugs of abuse including opioids. Interestingly, the rewarding effects of morphine are absent in mice lacking the NK1 receptor (20,21). Despite these observations, a molecular basis for the interactions between opioid and substance P receptors has not been established.
Here, we report that the MOR1 receptor forms stable heterodimers with the NK1 receptor. MOR1-NK1 heterodimerization did not substantially change ligand binding and signaling properties of the MOR1 receptor, but it dramatically altered its trafficking and resensitization profile.

H] SP and [ 3 H] DAMGO
were from PerkinElmer Life Sciences (Zavenstein, Belgium). Dithiobis-(succinimidylpropionate) (DSP) was purchased from Pierce. Mouse monoclonal anti-T7 antibody and mouse monoclonal anti-T7 antibody covalently coupled to Sepharose beads were obtained from Novagen (Madison, WI), rat monoclonal anti-HA antibody was purchased from Roche Diagnostics (Mannheim, Germany), mouse monoclonal anti-HA antibody covalently coupled to Sepharose beads were from Covance (Berkeley, CA), and polyclonal guinea pig anti-NK1 antibody was from Chemicon (Hofheim, Germany). In addition, polyclonal rabbit anti-T7, anti-HA, and anti-MOR1 (9998) antibodies were used which have been generated in our laboratory and characterized extensively (13,14,30,31). All polyclonal rabbit antisera were affinity-purified against their immunizing peptides using the Sulfo-Link coupling gel according to the instructions of the manufacturer (Pierce).
Cell Culture and Transfections-HEK 293 cells (HEK) obtained from ATCC were transfected with plasmids containing either the MOR1 receptor or the NK1 receptor using the calcium phosphate precipitation method. The wild-type HA epitope-tagged rat -opioid receptor MOR1 with puromycin resistance was generated as described previously (12). The full-length coding region of the rat NK1 cDNA was isolated by RT-PCR from rat brain total RNA using Pfu Turbo DNA polymerase (Stratagene, Gebouw, CA). The sense primer contained a HindIII and the antisense primer a XbaI restriction site. Purified PCR products were digested and ligated into a pcDNA3.1 expression vector encoding a neomycin resistance and the T7 epitope tag sequence MASMTG-GQQMG upstream to the HindIII insertion site. Double strand sequencing was employed to verify the resulting construct. To generate stable lines coexpressing two differentially epitope-tagged receptors, cells expressing HAMOR1 were subjected to a second round of transfection using LipofectAMINE (Invitrogen, Karlsruhe, Germany) and selected in the presence of 500 g/ml G418 and 1 g/ml puromycin (Sigma). 6 clones expressing HAMOR1 alone, 4 clones expressing T7NK1 alone, and 6 clones coexpressing T7NK1 and HAMOR1 were generated. Receptor expression was monitored using saturation ligand binding assays as described below. In addition, quantitative Western blot analysis was carried out to ensure that clones coexpressing a ϳ1:1 ratio of NK1 and MOR1 protein were selected. Furthermore, double immunofluorescent staining was performed in order to validate that NK1 and MOR1 were coexpressed within the same cells. The B max and K D values of cells that were used throughout this study are given in Table I.
BRET Assay-The NK1 and MOR1 coding sequences were amplified using sense and antisense primers harboring unique PstI and HindIII sites. Purified PCR products were digested and ligated into humanized pGFP-N3 and Rluc-N3 expression vectors (BioSignal Packard Biosciences, Montreal, Canada). For BRET measurements, HEK 293 cells were transiently transfected using LipofectAMINE. Forty-eight hours post-transfection, cells were detached with PBS/EDTA and resuspended in PBS containing 0.1% glucose (w/v) and 2 g/ml aprotinin. Cells were then transferred to 96-well microplates (white Optiplate from BioSignal Packard Biosciences) at a density of 100,000 cells/well. Deep Blue C (coelenterazine; BioSignal Packard Biosciences) was added at a final concentration of 5 M, and readings were collected using a Fusion microplate analyzer (BioSignal Packard Biosciences) that allows the sequential integration of the signals detected in the 330 -490 nm and 485-545 nm windows using filters with the appropriate band pass. The bioluminescence resonance energy transfer (BRET) signal is determined by calculating the ratio of the light emitted by the receptor-GFP (485-545 nm) over the light emitted by the receptor-Rluc (330 -490 nm). The values were corrected by subtracting background signal detected in non-transfected cells.
Immunocytochemistry-Cells were grown onto poly-L-lysine-coated coverslips overnight. For single immunofluorescence, cells were preincubated with affinity-purified rabbit anti-T7 or anti-HA antibody at a concentration of 1 g/ml at 4°C for 2 h and then either not exposed or exposed to various agonists. After fixation, bound primary antibodies were detected with cyanine 3.18 (Cy3)-conjugated anti-rabbit antibodies (1:200, Jackson ImmunoResearch, West Grove, PA). For double immunofluorescence, cells were preincubated with mouse anti-T7 and rabbit anti-HA antibodies at 4°C for 2 h. After fixation, bound primary antibodies were detected with a mixture of cyanine 3.18 (Cy3)-and cyanine 5.18 (Cy5)-conjugated secondary antibodies. Cells were then dehydrated, cleared in xylol and permanently mounted in DPX (Fluka, Neu-Ulm, Germany). To examine trafficking of ␤-arrestin, cells stably coexpressing T7NK1 and HAMOR1 were transiently transfected with an expression plasmid containing GFP-tagged ␤-arrestin2 (kindly provided by Dr. M. G. Caron; Duke University Medical Center, Durham, NC). After 48 h, cells were treated, fixed and then permanently mounted in Vectashield (Vector Laboratories, Burlingame, CA). Specimens were examined using a Leica TCS-NT laser scanner confocal microscope.
Internalization Assays-Cells were seeded at a density of 2 ϫ 10 5 cells per well onto poly-L-lysine-treated 24-well plates. On the next day, cells were preincubated with 1 g of affinity-purified rabbit anti-T7 or anti-HA antibody for 2 h in OPTIMEM 1 (Invitrogen) at 4°C. Cells were then treated with 1 M SP or 1 M DAMGO in OPTIMEM 1 for 2 h. Subsequently, cells were fixed and incubated with peroxidase-conjugated anti-rabbit antibody (1:1000, Amersham Biosciences) for 2 h at room temperature. After washing, plates were developed with 250 l of ABTS solution (Roche Applied Science). After 10 -30 min, 200 l of the substrate solution from each well was transferred to a 96-well plate and analyzed at 405 nm using a microplate reader (BioRad).
Radioligand Binding Assays-Saturation binding assays were performed on membrane preparations from stably transfected cells as described (12)(13)(14). The dissociation constant (K D ) and number of Determination of Receptor Desensitization and Resensitization by ERK Assays-Cells were seeded at a density of 1 ϫ 10 5 per well onto poly-L-lysine-treated 24-well dishes, grown in DMEM medium containing 0.5% fetal calf serum overnight and then pretreated with OPTIMEM 1 for 2 h. Cells were then incubated with 1 M SP or 1 M DAMGO for 0, 1, 2, 4, or 6 h in OPTIMEM 1. For resensitization studies, cells were exposed to 1 M DAMGO for 4 h followed by an additional incubation period of either 0, 10, 20, 30, 40, 50, or 60 min in the absence of agonist. Cells were then exposed to either SP, DAMGO or lysophosphatidic acid (LPA) for 5 min at 37°C. Incubation was terminated by removal of the culture medium, and subsequent addition of 150 l of boiling SDS-sample buffer. The samples were assayed as previously described (13,14).
Data Analysis-Data from ligand binding and ERK assays were analyzed by non-linear regression curve fitting using GraphPad Prism 3.0 software. Statistical analysis was carried out using the two-tailed paired t test. p values Ͻ 0.05 were considered as statistically significant.

Characterization of NK1-MOR1 Heterodimerization by Coimmunoprecipitation-To examine the dimerization of NK1
and MOR1 receptors, we produced HEK 293 cells stably coexpressing T7-tagged NK1 receptors and HA-tagged MOR1 receptors. Quantitative Western blot analyses revealed that 4 of 6 clones coexpressed similar levels of NK1 and MOR1 receptors (not shown). Saturation binding experiments revealed that NK1-MOR1 double stable clones used for further studies express nearly equivalent numbers of SP and DAMGO binding sites (Table I). While coexpressing cells had a similar high affinity for SP as cells expressing T7NK1 alone, these cells exhibited an ϳ3-fold higher affinity for DAMGO as cells expressing HAMOR1 alone (Table I). Competition binding assays also showed that SP did not compete with [ 3 H]DAMGO binding, and DAMGO did not compete with [ 3 H]SP binding in membranes prepared from T7NK1-HAMOR1 cells. When these cells were treated with the cross-linking agent DSP, solubilized, and subjected to immunoprecipitation using T7 affinity beads, the anti-T7 antibody detected a monomeric and a dimeric form of the NK1 receptor (Fig. 1A). When immunoprecipition was carried out with HA affinity beads, the anti-HA antibody detected a monomeric and a dimeric form of the MOR1 receptor (Fig. 1B). In addition, a band migrating at ϳ 50 kDa was seen which may correspond to the nonglycosylated form of the MOR1 receptor (12). When T7NK1 receptors were immunoprecipitated using T7 affinity beads, the rat HA antibody detected a monomeric and a dimeric form as well as a nonglycosylated form of the HAMOR1 receptor only in extracts prepared from cells coexpressing T7NK1 and HAMOR1, but not in extracts prepared from a mixture of cells expressing these receptors separately (Fig. 1C). The coprecipitation of T7NK1 and HAMOR1 suggests that these receptors may form heterodimeric or higher order heterooligomeric complexes in coexpressing cells.
NK1 and MOR1 Receptors Form Heterodimers in Living Cells-Detergent solubilization of cells during coimmunoprecipitation studies could promote artifactual aggregation of hydrophobic proteins such as receptors. We therefore analyzed NK1-MOR1 heterodimerization in living cells using BRET. This technique is a proximity assay based on nonradiative transfer of energy between a bioluminescent donor (Rluc) and a fluorescent acceptor (GFP) that allows real time monitoring of protein-protein interaction in living cells (33)(34)(35)(36)(37). To assess NK1 and MOR1 homo-and heterodimerization, fusion constructs linking the receptor carboxyl-terminal tail to either Rluc or GFP were cotransfected in HEK 293 cells, and the transfer of energy between the two partners was assessed following the addition of Deep Blue C. Upon oxidation of the luciferase substrate, the enzyme emits light with a peak at 400 nm that can excite GFP, which, in turn, re-emits fluorescence with a peak at 510 nm but only if the two partners are within the BRET-permissive distance (Ͻ100 Å). The BRET signal is determined by calculating the ratio of the light emitted by the Receptor-GFP over the light emitted by the Receptor-Rluc. As shown in Fig. 2, a strong BRET signal was detected in cells expressing MOR1-Rluc/MOR1-GFP and in cells expressing MOR1-Rluc/NK1-GFP indicating that heterodimers between MOR1 and NK1 also form in living cells. No significant BRET was detected when MOR1-Rluc was coexpressed with soluble GFP, confirming the selectivity of the detected signals. No significant changes were detected in the measured MOR1-Rluc/ NK1-GFP BRET levels when the cells were stimulated with either DAMGO, SP, or a mixture of DAMGO and SP (not shown).
Endocytotic Trafficking of the NK1-MOR1 Heterodimer-We next examined the consequences of NK1-MOR1 heterodimerization on agonist-induced receptor endocytosis using a quantitative enzyme-linked immunosorbent assay and immunocytochemistry. As depicted in Fig. 3A, quantitative analysis of receptor internalization after treatment with either 1 M SP or 1 M DAMGO revealed that NK1 and MOR1 were selectively internalized only in response to their cognate ligands in cells  expressing these receptors alone. In cells coexpressing NK1 and MOR1, NK1 was internalized in response to both SP and DAMGO. Conversely, MOR1 internalized after DAMGO as well as after SP (Fig. 3C). As shown in Fig. 3, B and D, both NK1 and MOR1 receptors were predominantly confined to the plasma membrane in untreated cells (Control). After 30 min exposure to SP, a loss of NK1 receptors from the plasma membrane and a robust internalization was observed in cells expressing the NK1 receptor alone (Fig. 3B, SP). In contrast, the distribution of NK1 receptors did not change when these cells were treated with DAMGO. Incubation with DAMGO but not SP induced internalization of MOR1 receptors in cells expressing MOR1 alone (Fig. 3B, DAMGO). The confocal images clearly showed that upon DAMGO treatment the MOR1 internalized and clustered in smaller vesicle-like structures then the NK1 after SP incubation indicating that these receptor have a different endocytotic fate. As depicted in Fig. 3D, SP induced a robust redistribution of both NK1 and MOR1 when these receptors were expressed in the same cells. Similar, after treatment with DAMGO, MOR1 as well as NK1 were internalized (Fig. 3D, DAMGO). The confocal images of coexpressing cells revealed that SP and DAMGO induced cointernalization of NK1 and MOR1 into the same endosomal compartment.
␤-Arrestin2 Trafficking in NK1-and MOR1-coexpressing Cells-MOR1 and NK1 have been reported to profoundly differ in their patterns of ␤-arrestin trafficking. For MOR1, ␤-arrestin directs the receptors to clathrin-coated pits. However, MOR1-␤-arrestin complexes are relatively unstable and dissociate at or near the plasma membrane. Consequently, ␤-arrestin is excluded from MOR1-containing vesicles (25,38). In contrast, NK1 and ␤-arrestin form stable complexes, remain associated, and internalize together into the same endocytotic vesicles (25)(26)(27)(28). We therefore examined ␤-arrestin trafficking in cells coexpressing MOR1 and NK1 after transient transfection of GFP-labeled ␤-arrestin2. Cells were exposed to either 1 M SP or 1 M DAMGO for 30 min and the subcellular distribution of the receptor proteins and ␤-arrestin2-GFP was analyzed by confocal microscopy. As shown in Fig. 4, NK1 (left  panel, Control, red) and MOR1 (right panel, Control, red) were clearly confined to the plasma membrane whereas ␤-arrestin2 (Control, green) was distributed throughout the cytoplasm but excluded from the nucleus in untreated cells. Like that seen in cells expressing NK1 alone, the NK1 receptor was internalized together with ␤-arrestin2-GFP into large clusters of early endosomes after 30 min exposure to SP in cells coexpressing NK1 and MOR1 (Fig. 4, left panel, SP). Interestingly, under these conditions the MOR1 receptor was cointernalized together with NK1 into same endosomes, which cocontained ␤-arrestin2-GFP (Fig. 4, right panel, SP). Moreover, unlike that seen in cells expressing MOR1 alone, the MOR1 receptor was also internalized together with ␤-arrestin2-GFP into large clusters of early endosomes after 30 min exposure to DAMGO in cells coexpressing NK1 and MOR1 (Fig. 4, left panel, DAMGO). Under identical conditions the NK1 receptor was cointernalized together with MOR1 into the same endosomes, which cocontained ␤-ar-restin2-GFP (Fig. 4, right panel, DAMGO). These results suggest that activation of the NK1-MOR1 heterodimer by both SP and DAMGO induced the formation of stable ␤-arrestin complexes and cotrafficking of the receptor heterodimer and ␤-arrestin into the same endosomal compartment thereby promoting a functional switch of MOR1 from a class A to a class B receptor.
Phosphorylation of the NK1-MOR1 Heterodimer-Formation of stable complexes between ␤-arrestin and GPCRs strongly depends on the presence of clusters of phosphate acceptor sites (defined as serine/threonine residues occupying three consecutive positions or three out of four positions) within the carboxylterminal tail of the receptor. These clusters are remarkably conserved in their position within the carboxyl-terminal domain and serve as primary sites of agonist-dependent receptor phosphorylation (26,27). While such clusters of phosphate acceptor sites are present in the NK1 receptor, the carboxylterminal tail of MOR1 does not contain such a motif. To delineate a mechanistic basis for the observed switch of MOR1 from a class A to a class B receptor, we assessed whole cell receptor phosphorylation in response to either SP or DAMGO in cells coexpressing NK1 and MOR1. As shown in Fig. 5, SP induced a rapid and robust phosphorylation of the NK1 receptor monomer (ϳ4-fold over basal). Interestingly, SP exposure of the NK1-MOR1 heterodimer also significantly increased phosphorylation of the MOR1 receptor monomer (ϳ3-fold over basal). As expected DAMGO produced a rapid and robust phosphorylation of the MOR1 receptor monomer (ϳ4.5-fold over basal). DAMGO also significantly increased phosphorylation of the NK1 receptor monomer (ϳ2-fold over basal) indicating that activation of the MOR1 subunit of the NK1-MOR1 heterodimer resulted in cross-phosphorylation of the NK1 subunit and visa versa. This cross-phosphorylation was not simply due to crossreactivity of the agonists, because it was not observed in cells expressing either T7NK1 or HAMOR1 alone (not shown). Thus, NK1-MOR1 cross-phosphorylation may provide a plausible explanation for the altered ␤-arrestin trafficking in response to activation of this heterodimeric receptor.
De-and Resensitization of the NK1-MOR1 Heterodimer-NK1 and MOR1 differ in their intracellular signaling in that NK1 stimulates inositol trisphosphate formation via G q and MOR1 inhibits cAMP formation via G i . However, both NK1 and MOR1 stimulate a rapid and transient ERK1/2 phosphorylation. Analysis of mitogenic signaling of the NK1-MOR1 heterodimer revealed that SP and DAMGO produced similar doseand time-dependent responses in cells coexpressing T7NK1 and HAMOR1 as in cells expressing these receptors alone. We then examined the desensitization of ERK signaling of the NK1-MOR1 heterodimer. Cells coexpressing T7NK1 and HAMOR1 were preincubated in the presence or absence of either SP or DAMGO for 4 h. The medium was removed and the ability of SP, DAMGO, or LPA to stimulate ERK1/2 activity were either not exposed (Control) or exposed to 1 M SP or 1 M DAMGO for 30 min. Cells were subsequently fixed, fluorescently labeled with either anti-T7 or anti-HA antibodies, and the subcellular distribution of receptor proteins was examined by confocal microscopy. D, HEK 293 cells coexpressing T7NK1 and HAMOR1 were either not exposed (Control) or exposed to 1 M SP or 1 M DAMGO for 30 min. Cells were subsequently fixed, subjected to dual immunofluorescence labeling with a mixture of anti-T7 and anti-HA antibodies and the subcellular distribution of receptor proteins was examined by confocal microscopy. The upper panel shows NK1-Li, and the lower panel shows MOR1-Li. Shown are representative results from one of three independent experiments performed in duplicate. Note, that in untreated cells both NK1 and MOR1 were almost exclusively confined to the plasma membrane revealing extensive colocalization. Exposure of the NK1-MOR1 heterodimer to SP promoted cointernalization of MOR1 together with NK1. Conversely, exposure of the NK1-MOR1 heterodimer to DAMGO induced cointernalization of NK1 together with MOR1. The asterisks indicate a significant difference (p Ͻ 0.05) between T7NK1-HAMOR1-expressing cells and cells expressing these receptors alone (two-tailed Student's paired t test). Scale bars, 20 m.

FIG. 4. Agonist-induced ␤-arrestin trafficking in cells coexpressing NK1 and MOR1.
Left panel, HEK 293 cells coexpressing T7NK1 and HAMOR1 were transiently transfected with ␤-arrestin2-GFP and either not exposed (Control) or exposed to 1 M SP or 1 M DAMGO for 30 min. Cells were fixed, and T7NK1 receptors were labeled using anti-T7 antibodies. Right panel, HEK 293 cells coexpressing T7NK1 and HAMOR1 were transiently transfected with ␤-arrestin2-GFP and either not exposed (Control) or exposed to 1 M SP or 1 M DAMGO for 30 min. Cells were fixed, and HAMOR1 receptors were labeled using anti-HA antibodies. Note, in untreated cells both NK1 and MOR1 (shown in red) were almost exclusively confined to the plasma membrane whereas ␤-arrestin2 (shown in green) was distributed throughout the cytoplasm. Exposure of the NK1-MOR1 heterodimer to either SP or DAMGO promoted internalization of both NK1 and MOR1 together with ␤-arrestin2-GFP (yellow in overlay). Representative images from two independent experiments performed in duplicate are shown. Scale bar, 20 m. was determined. As depicted in Fig. 6, preincubation with either SP or DAMGO for 4 h significantly attenuated both NK1-and MOR1-dependent responses. In contrast, mitogenic signaling of the LPA receptor, a third receptor that is endogenously expressed in this system, was unchanged suggesting that the NK1-MOR1 heterodimer underwent homologous cross-desensitization under these conditions. Class A and class B receptors profoundly differ in their resensitization kinetics. Given our observation that NK1-MOR1 heterodimerization promoted a functional switch of MOR1 from a class A to a class B receptor, we compared resensitization of -opioid receptordependent responses in MOR1-expressing cells and in cells coexpressing NK1 and MOR1. Cells were preincubated in the presence or absence of DAMGO for 4 h. After a DAMGO-free incubation period the medium was removed and the ability of DAMGO to stimulate ERK1/2 activity was determined. The results depicted in Fig. 7 show that resensitization of MOR1mediated mitogenic signaling was severely delayed in cells coexpressing NK1 and MOR1 as compared with cells expressing MOR1 alone. DISCUSSION The neuropeptide SP and endogenous opioids, as well as their corresponding G protein-coupled receptors, are intimately involved in the regulation of nociceptive transmission (15)(16)(17)(18)(19)(20)(21)(22). At the level of the spinal cord, many SP-containing terminals target neurons that encode noxious stimuli. These neurons have recently been shown to contain both NK1 and MOR1 receptors (15,16,22). Increased NK1 receptor stimulation in mice lacking noradrenaline leads to reduced opioid efficacy (19). Electrical stimulation of afferent C-fibers is associated with an NK1-dependent increase in the excitability of spinal cord neurons, a phenomenon termed "wind-up." Wind-up is effectively abolished by opioids (18). Substance P-opioid chimeric peptides produce analgesia without formation of tolerance (17). In addition, MOR1 and NK1 are highly coexpressed in brain regions implicated in depression, anxiety and stress, but also in other regions such as the nucleus accumbens, which mediate the motivational properties of drugs of abuse including opioids. Interestingly, the rewarding effects of morphine are absent in mice lacking the NK1 receptor (20,21). Despite these observations, a molecular basis for the interactions between opioid and substance P receptors has not been established yet.
Here, we provide biophysical and biochemical evidence for heterooligomerization of the substance P and the -opioid receptor. The coimmunoprecipitation experiments carried out in the present study clearly demonstrate that NK1 and MOR1 receptors exist as constitutive heterooligomeric complexes at the plasma membrane when coexpressed in HEK 293 cells. The immunoprecipitation of T7-tagged NK1 receptors resulted in coprecipitation of HA-tagged MOR1 receptors only from coexpressing cells but not from a mixture of cells expressing these receptors separately, suggesting that NK1-MOR1 heterooligomers preexisted in these cells prior to cell lysis and were not artifactually formed during sample preparation. When cells FIG. 5. Agonist-induced cross-phosphorylation of the NK1-MOR1 heterodimer. HEK 293 cells coexpressing T7NK1 and HAMOR1 were exposed to 1 M SP or 1 M DAMGO for 20 min, and whole cell receptor phosphorylation was determined. T7NK1 was immunoprecipitated with rabbit anti-T7 antibodies, and HAMOR1 was immunoprecipitated with rabbit anti-HA antibodies. A, autoradiographs from representative experiments are shown. B, means Ϯ S.E. of three independent experiments quantified by phosphorimager analysis.
The asterisks indicate significant agonist-induced phosphorylation compared with basal levels in the absence of agonist (p Ͻ 0.05; two-tailed Student's paired t test). Note, phosphorylation of MOR1 was significantly increased above basal levels in the presence of the NK1-selective agonist SP. Conversely, phosphorylation of NK1 was significantly increased above basal levels in the presence of the MOR1-selective agonist DAMGO. The data were normalized to basal phosphorylation in the absence of agonist for each receptor monomer. The positions of molecular mass markers are indicated on the left (in kDa). were treated with the cross-linker DSP, precipitation of T7tagged NK1 receptors resulted in coprecipitation of HA-tagged MOR1 receptors with migration properties corresponding to that of the nonglycosylated MOR1 receptor (12). The fact that this lower molecular weight band was not seen in whole cell phosphorylation assays suggests that NK1-MOR1 heterooligomers may have already formed in the cytoplasmic reticulum. By utilizing BRET, we demonstrated that NK1 and MOR1 form constitutive heterooligomeric complexes in living cells. Recent evidence from quantitative BRET analysis of ␤ 1 -and ␤ 2 -adrenergic receptor homo-and heterodimerization suggests that most of the receptors expressed in cells exist as constitutive dimers and that, at least in undifferentiated fibroblasts, the proportion of homo-and heterooligomers is determined by the relative level of receptor expression (39). Given the frequent coexpression of MOR1 and NK1 in spinal cord neurons, a direct protein-protein interaction between these receptors is possible. Such a physical interaction could provide a plausible explanation for cross-talk between substance P and -opioid receptors during nociceptive transmission. Our analysis of the functional consequences of NK1-MOR1 heterodimerization revealed no substantial changes of the ligand binding and signaling properties of these receptors. In contrast, NK1-MOR1 heterodimerization markedly altered the internalization and desensitization profile of both receptors. Exposure of the NK1-MOR1 heterodimer to the MOR1-selective ligand DAMGO promoted cointernalization and cross-desensitization of the NK1 receptor. Conversely, exposure of the NK1-MOR1 heterodimer to the NK1-selective ligand SP promoted cointernalization and crossdesensitization of the MOR1 receptor.
Based on trafficking patterns and affinity for ␤-arrestin, GPCRs are categorized into two classes (23)(24)(25)(26)(27)(28)(29). Class A receptors include the -opioid, ␤ 2 and ␣ 1B adrenergic, endothelin A, and dopamine D1A receptors. These receptors bind to ␤-arres-tin2 with higher affinity than to ␤-arrestin1. In addition, their interaction with ␤-arrestin is transient. ␤-arrestin is recruited to the receptor at the plasma membrane translocates with it to clathrin-coated pits; however, the receptor-␤-arrestin complex dissociates upon internalization of the receptor, such that, as the receptor proceeds into the endosomal pool, the ␤-arrestin recycles back to the plasma membrane (26,38). Class B receptors, represented by the substance P, angiotensin AT 1a , neurotensin 1, and vasopressin 2 receptors bind to ␤-arrestin1 and ␤-arrestin2 with equal affinity. These receptors form stable complexes with ␤-arrestin, such that the receptor-␤-arrestin complex internalizes as a unit that is targeted to endosomes (23, 26 -28). While class A receptors recycle and resensitize rapidly, class B receptors recycle and resensitize slowly (25). Interestingly, in cells coexpressing NK1 and MOR1, we observed that both DAMGO and SP induced recruitment of ␤-arrestin to the plasma membrane and cointernalization of NK1-MOR1 heterodimers with ␤-arrestin into endosomes. Consequently, resensitization of MOR1-dependent receptor functions was severely delayed in coexpressing cells as compared with cells expressing MOR1 alone indicating that MOR1 by virtue of its physical interaction with NK1 is sequestered via a class B receptor-like pathway with delayed recycling and resensitization kinetics.
The structural features that dictate the stability of the receptor-␤-arrestin complex reside within specific clusters of putative phosphate acceptor sites (defined as serine/threonine residues occupying three consecutive positions or three out of four positions) in the carboxyl-terminal tail of the receptor (27). Such clusters are remarkably conserved in their position within the carboxyl-terminal tail with respect to the NPXXF motif (marking the end of the seventh transmembrane domain) and the end of the receptor (27). While such clusters of phosphate acceptor sites are present in the NK1 receptor, the carboxyl-terminal tail of MOR1 does not contain such a motif. In whole cell receptor phosphorylation assays, we demonstrate that SP exposure of the NK1-MOR1 heterodimer significantly increased phosphorylation of the MOR1 receptor monomer. Conversely, DAMGO exposure of the NK1-MOR1 heterodimer significantly increased phosphorylation of the NK1 receptor monomer. Thus, homologous cross-phosphorylation of the NK1-MOR1 heterodimer may provide a mechanistic basis for cotrafficking of NK1-MOR1 heterodimers with ␤-arrestin into endosomes.
The stability of the receptor-␤-arrestin interaction determines not only the recycling and resensitization profile of the receptor but also the mechanisms and functional consequences of downstream signaling (24). Tohgo et al. (29) have recently shown that activated ERK remains associated with stable class B receptor-␤-arrestin complexes, which, in turn, limits nuclear translocation of ERK and attenuates Elk1-driven transcription. In contrast, class A receptors promote nuclear translocation of activated ERK and stimulate Elk1-driven transcription (29). Thus, altered ␤-arrestin trafficking induced by NK1-MOR1 heterodimerization may lead to cytosolic retention of activated ERK and distinct patterns of ERK-dependent transcription. FIG. 7. Differential resensitization of -opioid receptor responses in MOR1-and NK1-MOR1-expressing cells. HEK 293 cells expressing HAMOR1 or coexpressing T7NK1 and HAMOR1 were serum-starved overnight and incubated in the presence or absence of 1 M DAMGO for 4 h. Cells were washed followed by a DAMGO-free interval of either 0, 10, 20, 30, 40, 50, or 60 min. Cells were then exposed to DAMGO for 5 min and subsequently lysed. Equal amounts of protein were resolved by SDS-PAGE, and levels of phosphorylated ERK1/2 were determined by immunoblotting. A, results were quantified by densitometric analysis. Data were normalized to total ERK2 and expressed as the fold ERK1/2 phosphorylation over the basal value in untreated cells. The maximum agonist-induced activation of ERK1/2 without agonist preincubation was defined as 100%. Values represent means Ϯ S.E. of three independent experiments performed in duplicate.
The asterisks indicate a significant difference (p Ͻ 0.05) between MOR1-expressing cells and cells coexpressing NK1 and MOR1 (twotailed Student's paired t test). B, representative immunoblots for MOR1-expressing cells (upper panel) and cells coexpressing NK1 and MOR1 (lower panel), respectively. The position of phospho-ERK1/2 (pERK1/2) is indicated on the right. The positions of molecular mass markers are indicated on the left (in kDa).
In conclusion, we provide biophysical, biochemical and functional evidence for heterodimerization of two distantly related receptors namely the substance P and the -opioid receptor. We show that heteromeric assembly of NK1 and MOR1 crossmodulates internalization and desensitization of both receptors. In addition, we demonstrate that MOR1 by virtue of its physical interaction with NK1 is sequestered via a class B receptor-like pathway with delayed recycling and resensitization kinetics. Altered ␤-arrestin trafficking in cells coexpressing MOR1 and NK1 could, thus, not only impact on opioid receptor resensitization but also on the long term cellular effects of opioids.