|Home | About | Journals | Submit | Contact Us | Français|
Mu opioid receptors (MORs) are G protein-coupled receptors (GPCRs) that mediate the physiological effects of endogenous opioid neuropeptides and opiate drugs such as morphine. MORs are co-expressed with neurokinin 1 receptors (NK1Rs) in several regions of the central nervous system (CNS) that control opioid dependence and reward. NK1R activation affects opioid reward specifically, however, and the cellular basis for this specificity is unknown. We found that ligand-induced activation of NK1Rs produces a cell autonomous and non-reciprocal inhibition of MOR endocytosis induced by diverse opioids. Studies using epitope-tagged receptors expressed in cultured striatal neurons and a neuroblastoma cell model indicated that this heterologous regulation is mediated by NK1R-dependent sequestration of arrestins on endosome membranes. First, endocytic inhibition mediated by wild type NK1Rs was overcome in cells over-expressing β-arrestin2, a major arrestin isoform expressed in striatum. Second, NK1R activation promoted sequestration of β-arrestin2 on endosomes, whereas MOR activation did not. Third, heterologous inhibition of MOR endocytosis was prevented by mutational disruption of β-arrestin2 sequestration by NK1Rs. NK1R-mediated regulation of MOR trafficking was associated with reduced opioid-induced desensitization of adenylyl cyclase signaling in striatal neurons. Further, heterologous regulation of MOR trafficking was observed in both amygdala and locus coeruleus neurons that naturally co-express these receptors. These results identify a cell autonomous mechanism that may underlie the highly specific effects of NK1R on opioid signaling and suggest, more generally, that receptor-specific trafficking of arrestins may represent a fundamental mechanism for coordinating distinct GPCR-mediated signals at the level of individual CNS neurons.
G-protein coupled receptors (GPCRs) mediate a wide variety of physiological functions and are the targets of a vast array of both therapeutic and abused drugs. Opioid receptors comprise a subfamily of GPCRs that are activated both by endogenously produced opioid neuropeptides and exogenous alkaloid drugs such as morphine (Evans 2000). The mu opioid receptor (MOR) is the primary target mediating analgesic, euphoric, and reinforcing effects of morphine (Matthes et al., 1996). After exposure to opioid peptides, MOR is phosphorylated and associates with β-arrestin (also called non-visual arrestin), which modulates opioid signaling and functions as an adaptor protein to promote MOR endocytosis via clathrin-coated pits (Keith et al., 1996; Whistler and von Zastrow, 1998; Zhang et al., 1998; Oakley et al., 1999). This series of events is thought to contribute fundamentally to controlling MOR-mediated signaling under conditions of prolonged or repeated ligand exposure (Koch et al., 1998; Qiu et al., 2003). While morphine promotes relatively little regulated endocytosis of MOR compared to opioid peptides in several cell models, it does strongly produce endocytosis in striatal neurons, a brain region important for reward processing (Heimer et al., 1982; Keith et al 1998; Whistler et al., 1999; Bushell et al., 2002; Haberstock-Debic et al., 2003; Trafton and Basbaum, 2004; Haberstock-Debic et al., 2005). Therefore, regulated endocytosis of MOR may play an important role in the actions of endogenous MOR in response to both peptides and drugs such as morphine.
Whereas MORs are major direct targets receptors of opiate drug action, various other GPCRs modulate opioid function in vivo. Of these, the neurokinin 1 receptor (NK1R) has been found to modulate MOR-dependent responses with a remarkably high degree of specificity. Mice lacking NK1Rs are insensitive to the rewarding properties of morphine, while the rewarding effects of food and cocaine are preserved (Murtra et al., 2000; Ripley et al., 2002). The antinociceptive effects of morphine remain intact in NK1R knock-out animals, further supporting the specificity of NK1R function in modulating opiate reward (De Felipe et al, 1998). Moreover, MORs and NK1Rs are highly expressed in several brain regions relevant to reward processing, including the ventral striatum and amygdala (Pickel et al., 2000; Gadd et al., 2003; Nakaya et al., 2004; Jabourian et al., 2005; Poulin et al., 2006).
Given the known physiological interaction between the neurokinin and opioid neurochemical systems, and the co-expression of NK1R and MOR in relevant brain regions, we considered the possibility that these distinct receptor systems might interact functionally at the level of individual neurons. In the present study, we investigated this hypothesis in several cell populations, using both recombinant and natively expressed receptors. Our results identify a cell-autonomous mechanism by which NK1Rs specifically modulate MOR trafficking and signaling, and also suggest a more general principle of heterologous regulation of diverse GPCRs in the CNS.
The amino-terminally FLAG (DYKDDDD) epitope-tagged murine mu opioid receptor (F-MOR) was described previously (Keith et al., 1996). The rat NK1 neurokinin receptor (cDNA generously provided by Dr. Nigel Bunnett, University of California, San Francisco, San Francisco, CA) was tagged in its amino-terminal extracellular domain with an HA (YPYDVPDYA) epitope-tag using PCR (HA-NK1R). Truncation at residue 355 was accomplished by introducing a stop codon using PCR (HA-NK1-355x). Tagged receptor constructs were cloned into pCAGGS (Niwa et al., 1991) for expression in cultured neurons, and pcDNA3.1 (Invitrogen) for expression in neuroblastoma 2A cells (N2A). A bovine arrestin 3 (β-arrestin2) construct, tagged with EGFP as described previously (Mundell and Benovic, 2000), was generously provided by Dr. J. Benovic (Thomas Jefferson University). All constructs were confirmed by DNA sequencing (Elim Biopharmaceuticals, San Francisco, CA).
Primary striatal neurons and primary amygdala neurons were dissected from embryonic day 17–18 rat embryos taken from pregnant Sprague-Dawley rats. The striatum (caudate-putamen and nucleus accumbens) were dissected based on the criteria of Ventimiglia and Linsday (1998). The amygdala was identified as described by Altman and Bayer (1995). Dissected tissue was dissociated in 1x trypsin/EDTA solution (Invitrogen, Gaithersburg, MD) for 15 minutes before 1mL of trypsin inhibitor was added for 5 minutes at room temperature. Cells were washed and triturated in DMEM plus 10% fetal calf serum (FCS; Invitrogen). Mechanical trituration of tissue was performed using a glass pipette. Neuronal transfections were performed using electroporation (rat neuron nucleofector system; Amaxa Biosystems, Gaithersburg, MD) immediately after dissociation and plated on poly-L-lysine-coated (1mg/mL in 0.1 M sodium borate buffer, pH 8.5) glass coverslips washed previously in 70% nitric acid and rinsed over 2–3 days. Electroporation was conducted using 5 × 106 dissociated cells, 3 μg of plasmid DNA, and 100 μL of rat neuron nucleofector solution. After transfection, neurons were kept in pre-warmed RPMI media (Cell culture facility, University of California San Francisco, San Francisco, CA) for 10 minutes for recovery before being transferred onto 24-well plates. Media on the cells was replaced with Gibco Neurobasal media (Invitrogen) supplemented with B27 (Gibco) and L-glutamine 24 hours post transfection. Neurons were maintained 7–10 days in culture before assaying.
Neuroblastoma 2A cells were obtained from University of San Francisco cell culture facility and maintained in Gibco DMEM (Invitrogen) media supplemented with 10% FCS and penicillin/streptomycin. Cells were plated on poly-D-lysine coated (50μg/mL in water) glass coverslips onto 24-well plates for immunocytochemistry and 6 cm dishes for biotinylation assays. Transfections were performed using a cationic lipid transfection reagent (Effectene; Qiagen, Hilden, Germany). Cells were transfected when they were at 50% confluency and assayed 24 hours post transfection.
F-MOR and HA-NK1R distribution was visualized after dual immunocytochemical labeling with antibodies recognizing the distinct epitope tags. Rabbit anti-FLAG polyclonal antibody (0.2μg/mL; Sigma, St. Louis, MO) and mouse anti-HA (5μg/mL; Covance) antibodies were added to the culture medium of transfected cells, and surface labeling was carried out for 30 minutes before an additional 30 minute incubation in the presence or absence of the indicated ligand(s). After quick washes with either TBS (for neurons) or PBS (for N2A cells), cells were fixed with 4% paraformaldehyde dissolved in Ca2+-free PBS supplemented with 5% sucrose for 15 minutes. Specimens were permeabilized and blocked for 20 minutes in a solution containing either 0.1% saponin (for neurons) or 0.1% Triton X-100 (for N2A cells), and 2% BSA dissolved in the appropriate buffered saline. Secondary labeling was carried out using Alexa594 goat anti-rabbit IgG (for F-MOR) and Alexa488 donkey anti-mouse IgG (for HA-NK1R) antibodies (2μg/mL; Molecular Probes, Invitrogen), prepared in the same blocking solution and incubated for 30 minutes. Cells were then washed extensively with the appropriate buffered saline and mounted onto glass slides for fluorescence microscopy. Epifluorescence microscopy was carried out using a Nikon (Melville, NY) Diaphot microscope equipped with a 60x/numerical aperture (NA) 1.4 objective, mercury arc lamp illumination and standard dichroic filter sets (Omega Optical, Brattleboro, VT). Images were collected with a CCD camera (Princeton Instruments, Trenton, NJ) and analyzed using MetaMorph® software (Molecular Devices, Sunnyvale, CA). Confocal microscopy was carried out with a Zeiss (Oberkochen, Germany) LSM510 instrument, equipped with a 63x/NA 1.4 objective and using a pinhole diameter of 1 Airy disc.
To quantify ligand effects observed with epifluorescence microscopy, we used a previously described assay to assess ratiometric staining of surface and internalized receptors (Haberstock-Debic et al., 2005). In brief, cells were incubated with Alexa594 (Molecular Probes)-conjugated M1 anti-FLAG monoclonal antibody (5μg/mL; Sigma, St. Louis, MO) for 30 minutes (to selectively label F-MOR present in the plasma membrane) and then incubated in the absence or presence of the indicated ligand(s) for an additional 30 minutes. Cells were fixed and incubated with Alexa488-conjugated donkey anti-mouse IgG (Molecular Probes) for 30 minutes under nonpermeabilized conditions (2% BSA in TBS) to selectively detect only those labeled F-MORs remaining at the plasma membrane. Therefore, internalized receptors were labeled with only Alexa594, while surface F-MORs were labeled with both Alexa594 and Alexa488. The percentage of receptor internalization induced by the indicated agonist exposure was calculated from background-subtracted Alexa594 (red) and Alexa488 (green) fluorescence intensity values according to the following formula: % internalized = [1−(green/red ratio in agonist treated)/(green/red ratio in untreated)] × 100. For each experiment, 50 neuronal cell bodies or N2A cells were analyzed per condition. Results were compiled from five experiments for each cell type (and four for the N2A truncated HA-NK1R experiments). For experiments in primary cultures, these represented separate animals and culture preparations. Statistical analysis of differences between experimental groups was performed using unpaired Student’s t-test.
Colocalization of HA-NK1R and HA-NK1 355x with β-arrestin2-EGFP (in both localization and over-expression experiments) was obtained through a similar staining protocol as described above, except that rat anti-HA (5μg/mL; Roche) was used as primary antibody and Alexa594 conjugated anti-rat IgG (Molecular Probes) was used as secondary antibody. Triple localization of F-MOR, HA-NK1R, and β-arrestin2-EGFP was carried out using Alexa594 conjugated to M1 anti-FLAG and rat anti-HA primary antibodies, and HA-NK1R was visualized with Alexa647 goat anti-rat secondary antibody (Molecular Probes). Cells were imaged using the same confocal microscopy settings as described above. For colocalization experiments, control imaging of single-labeled cells was performed to confirm lack of detectable bleed through between channels.
For immunocytochemical localization of endogenous receptors, cultured amygdala neurons (grown ~7 days in culture on poly-L-lysine coated glass coverslips) were incubated with 10μM of both DAMGO and substance P for 30 minutes followed by a 10 minute fixation in ice-cold methanol. Cells were blocked in PBS containing 5% goat serum, 0.3% Tween-20, and 0.1% Triton X-100 in PBS for 45 minutes.
Staining for endogenous MOR and NK1R was performed using an affinity-purified rabbit antibody recognizing the carboxyl terminus of MOR1 (Keith et al, 1998) and a guinea pig receptor antibody recognizing the distinct carboxyl-terminus of NK1 (1:5000; AB15810, Millipore), respectively. Permeabilized specimens were co-incubated with both antibodies at 25°C for 2.5 hrs and then extensively washed in PBS after primary incubation. Redistribution of receptors was visualized after secondary labeling with Alexa594-conjugated goat anti-rabbit IgG (for F-MOR) and Alexa488-conjugated donkey anti-mouse IgG (for HA-NK1R) antibodies (2μg/mL; Molecular Probes) prepared in the same blocking solution as above and incubated for 60 minutes.
N2A cells were grown to 50% confluency on 6 cm dishes and transfected with F-MOR and HA-NK1R, or F-MOR and the truncated HA-NK1R (HA-NK1 355x). Transfected cells were washed with cold Ca2+- and Mg2+-free PBS and then incubated in 0.3mg/mL Sulfo-NHS-SS-biotin (Pierce) in PBS at 4°C for 20 minutes. Cells were washed with TBS and placed in DMEM medium for 10 minutes at 37°C prior to incubation for 30 minutes under the indicated conditions. Cells were then washed on ice with TBS, to remove and quench residual biotinylation reagent, and remaining cell surface biotinylated receptors were stripped at 4°C two times for 15 minutes each using 100mM MESNA (Sigma-Aldrich) in 50mM Tris 8.8, 100mM NaCl, 1mM EDTA, and 0.2% BSA. Stripped cells were then quenched with iodoacetamide buffer (22 mg/mL iodoacetamide in TBS) for 10 minutes at 4°C. Cells were extracted with lysis buffer containing 1mg/mL iodoacetamide in IP buffer (0.2% Triton X-100, 150mM NaCl, 25 mM KCl, 10 mM Tris 7.4) with 0.1mM EDTA and a Complete mini EDTA-free protease inhibitor cocktail tablet (1 tablet/10mL; Roche). Cell debris was removed by centrifugation at 13,000 × g for 15 minutes at 4°C. Extracts containing equal amounts of cellular protein (determined using Coomassie Plus protein assay from Pierce) were incubated with immobilized streptavidin beads (Pierce) on a rotator overnight at 4°C. Protein-bound streptavidin beads were centrifuged at 3,000 × g for 1 minute at 4°C and washed twice with IP buffer. Proteins were eluted and denatured in SDS sample buffer with 10% βmercaptoethanol and separated by SDS/PAGE. Proteins were transferred to nitrocellulose and biotinylated proteins were detected by immunoblotting with anti-FLAG M1 mouse antibody (5μg/mL; Sigma) followed by anti-mouse IgG horseradish peroxidase conjugate (1:5000, Amersham) and chemiluminescence detection using ECL reagent (Amersham).
Dissociated striatal neurons were co-transfected with F-MOR and HA-NK1R and maintained for 7–10 days in culture before assaying. Prior to agonist treatment, neurons were incubated in media containing the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX, Sigma) at a concentration of 1 mM. To assess acute signaling response, neurons were exposed for 3 minutes to 3μM forksolin and 1 mM IBMX, in the presence 10μM morphine, 10μM substance P, or both. To assess MOR desensitization, neurons were preincubated in the presence morphine, substance P or both for 30 minutes. The indicated agonists were then washed out and the ability of neurons to subsequently respond to opiate was assessed by rechallenge for 10 minutes with 100nM morphine in the presence of 3μM forskolin and 1 mM IBMX. After agonist incubations, cells were immediately washed once with ice cold PBS before lysing in 0.1M HCl with 0.1% Triton X-100. Lysates were collected and particulates removed by centrifugation 1000 × g for 5 minutes at room temperature. The concentration of cAMP present in clarified lysates was determined according to the non-acetylated version of the Correlate EIA™ Direct Cyclic AMP Enzyme Immunoassay Kit protocol (Assay Designs, Ann Arbor, MI). Acute signaling response was calculated from ligand-induced inhibition of forskolin-stimulated cAMP production relative to that measured in cells challenged with forskolin alone. Desensitization was estimated by the decrease in morphine-induced inhibition of cAMP accumulation measured in the rechallenge compared to that measured in cells not previously exposed to opiate. For examination of substance P effects on desensitization, only experiments in which morphine-driven MOR desensitization was reliably detected in the absence of substance P were included in the analysis. In each individual experiment, all manipulations and determinations were carried out in triplicate and averaged. The number of independent experiments included in each comparison is indicated in the text. Statistical analyses of differences among experimental conditions were performed using unpaired t-test.
4–6 week old mice were anesthetized with isoflurane and the brain was removed and sliced horizontally (200 μm thickness) using a vibratome (Leica, Nussloch, Germany) in ice-cold artificial cerebro-spinal fluid (ACSF) containing the following (in mM): 126 NaCl, 2.5 KCl, 1.2 MgCl2, 1.2 NaH2PO4, 2.4 CaCl2, 21.4 NaHCO3 and 11 glucose as previously described (Arttamangkul et al., 2008). Slices including the locus coeruleus were allowed to warm up to 34°C in oxygenated ACSF containing (+) MK-801 (10 μM, Sigma-Aldrich, St. Louis, MO) for 15 minutes and then incubated in a solution containing Alexa 594-conjugated M1 antibody. The tissue was visualized with an upright microscope (Olympus, Center Valley, PA.) equipped with a custom-built two-photon apparatus, as described previously (Arttamangkul et al., 2008). Data were acquired and collected using Scan Image Software (Pologruto et al., 2003). A z-series was collected at 1 μm intervals for 15 μm. Drugs were applied by perfusion. All experiments were done at 35°C. Analyses were done off-line using Image J (NIH) software as described in detail previously (Arttamangkul et al., 2008). For the control condition, integrated intensity data were obtained from labeled slices before drug application. This fluorescence intensity was defined as total surface-accessible receptor immunoreactivity (C). Integrated fluorescence intensity measured after drug perfusion, followed by calcium free ACSF containing ethylene glycol-bis(β-aminoethylether) N,N,N′,N′-tetraacetic acid (0.5 mM EGTA, 10 min), was defined as internalized receptor immunoreactivity (I). Percentage of internalization was calculated by (I/C) × 100.
To begin to investigate possible regulatory interactions between mu opioid receptors and neurokinin 1 receptors, we examined the subcellular distribution of receptors when co-expressed in dissociated striatal neurons. Cultured striatal neurons were co-transfected with FLAG-tagged mu opioid receptors (F-MOR) and HA-tagged neurokinin 1 receptors (HA-NK1R), and the localization of each receptor was examined by dual-label epifluorescence microscopy. In the absence of ligand (untreated condition), antibody-labeled F-MOR was observed in a distribution indicative of plasma membrane localization (Fig 1A, top row, left panel). Similar surface localization was also observed with HA-NK1R (middle panel). The overlay image confirmed an overlap in distribution between F-MOR and HA-NK1R on the plasma membrane (right panel). Examination of the cell body at higher magnification emphasized the overlap in F-MOR and HA-NK1R localization (insets). Exposure of cells to 10μM morphine (MS) for 30 minutes produced a substantial redistribution of labeled F-MOR into a punctate intracellular pattern (Fig 1A, middle row, left panel), consistent morphine-induced endocytosis of MOR shown previously in these neurons (Haberstock-Debic et al., 2005). In contrast, morphine did not detectably affect the plasma membrane localization of labeled HA-NK1R (middle panel and overlay). Conversely, activating HA-NK1Rs with substance P (SP) resulted in selective internalization of HA-NK1Rs, while F-MORs remained predominantly in the plasma membrane (Fig 1A, bottom row).
We next examined the effects of activating both receptors by exposing neurons to morphine and substance P simultaneously. Labeled HA-NK1R redistributed rapidly into a punctate pattern in the combined presence of both agonists, indistinguishable from the redistribution observed in neurons exposed to substance P alone (Fig 1B top row, compared to Fig 1A bottom row). Surprisingly, F-MORs co-expressed on these same cells remained localized predominantly in the plasma membrane, in contrast to the pronounced redistribution observed in the presence of morphine alone (Fig 1B top row, compared to Fig 1A, middle row). This selective redistribution of HA-NK1R was highlighted in higher magnification of a representative region of the cell body (inset). To investigate whether this non-reciprocal inhibition of F-MOR redistribution was dependent on the presence of HA-NK1Rs, we identified cells on the same coverslip expressing only F-MOR and not HA-NK1Rs. In these neurons, incubation with both morphine and substance P produced robust redistribution of F-MOR into a punctate pattern, similar to what was observed with morphine alone (Fig 1B, bottom row). Together, these results show that morphine-induced redistribution of F-MOR in neurons is specifically inhibited by simultaneous activation of co-expressed HA-NK1Rs by substance P.
To determine whether the agonist-mediated receptor redistribution observed by immunofluorescence is the result of receptor endocytosis, we used a previously described dual-labeling method to distinguish between surface-localized and internalized MORs (Haberstock-Debic et al., 2005). F-MORs present in the plasma membrane were initially surface-labeled using Alexa594-conjugated anti-FLAG monoclonal antibody. After 30 minutes of ligand incubation, neurons were fixed and receptors remaining in the plasma membrane were selectively labeled with Alexa488-conjugated secondary antibody (Fig 2A, middle column). Internalized receptors appeared red, whereas plasma membrane-localized receptors labeled with both fluorochromes appeared yellow in the overlay image (Fig 2A). Representative images of striatal neurons revealed that a significant fraction of F-MOR present initially in the plasma membrane internalized following exposure of neurons to morphine for 30 minutes (Fig 2A, top row). In contrast, incubation of neurons with both morphine and substance P resulted in little internalization of F-MOR, demonstrated by an overlap in staining of total receptor and remaining surface receptor (Fig 2A, bottom row). Ratiometric quantification in multiple neurons (as described in Materials and Methods) confirmed the significant inhibitory effect of HA-NK1R activation on regulated endocytosis of F-MOR (Fig 2B, black bars).
The enkephalin analog D-Ala2-N-Me-Phe4-Glycol5-enkephalin (DAMGO) is known to drive MOR endocytosis more readily and to a greater degree compared to morphine in many cell types (Keith et al 1998; Whistler et al., 1999; Bushell et al., 2002; Trafton and Basbaum, 2004). Thus we tested whether activation of NK1Rs receptors can also inhibit endocytosis of MOR induced by DAMGO. Incubation of striatal neurons with 10μM DAMGO for 30 minutes produced substantial internalization of F-MOR, which was somewhat larger in absolute magnitude than that induced by morphine (Fig 2B, white bars). Co-application of substance P resulted in a less pronounced, but still significant, inhibition of F-MOR endocytosis. Thus, NK1R-mediated inhibition of MOR trafficking is not restricted opiate drugs such as morphine, and can also affect regulated endocytosis induced by opioid peptide.
We next asked if the pronounced inhibitory effect of NK1R activation on MOR trafficking could be observed in a cultured cell model that is advantageous for mechanistic and biochemical studies. Typically, morphine drives endocytosis of MORs poorly in transformed cell lines (Keith et al., 1996; Alvarez et al., 2002). Nevertheless, by surveying a number of neurosecretory cell types, we found rapid morphine-mediated endocytosis of MORs in mouse-derived Neuro2A (N2A) cells. We first assessed the subcellular translocation of F-MOR and HA-NK1Rs with antibody labeling of cells transiently co-expressing these two receptors under various agonist conditions. Confocal microscopy demonstrated a predominantly plasma membrane distribution of both receptor types in the absence of ligand (Fig 3A, top row). Exposure to 10μM morphine for 30 minutes selectively promoted redistribution of F-MORs to intracellular vesicles, whereas HA-NK1Rs remained predominantly on the cell surface (Fig 3A, second row). Conversely, incubation with substance P resulted in selective internalization of HA-NK1Rs but not F-MORs (Fig 3A, third row). Further replicating the results obtained in striatal neurons, simultaneous incubation of N2A cells with both morphine and substance P resulted in selective internalization of HA-NK1Rs and little visible redistribution of F-MOR from the plasma membrane (Fig 3A, bottom row). Quantification by ratiometric labeling confirmed the inhibitory effect of HA-NK1R activation on morphine-induced endocytosis of F-MORs (Fig 3B, black bars). DAMGO-induced internalization of F-MOR was also significantly inhibited by HA-NK1R activation in N2A cells, and this was even more pronounced than observed in striatal neurons (Fig 3B, white bars).
To further confirm the HA-NK1R-dependent inhibition of F-MOR endocytosis, we used a biochemical assay to specifically measure the fate of surface-biotinylated receptors in a population of cells co-expressing both F-MOR and HA-NK1R. Cells were surface-biotinylated and then subjected to agonist incubation for 30 minutes. Biotin attached to receptors remaining in the plasma membrane was selectively cleaved using a membrane-impermeant reducing agent and the internalized (still biotinylated) pool of F-MORs was selectively detected following isolation on streptavidin beads. Very little biotinylated signal was detected in cells incubated in the absence of opioid agonist, confirming the efficiency of the cleavage reaction and, as expected, both morphine and DAMGO produced significant internalization of F-MORs (a representative immunoblot is shown in Fig. 3C, compare lane 1 to lanes 3 and 5). F-MOR internalization induced by both opioid agonists was dramatically reduced, essentially to the level of untreated control cells, when cells were incubated in the presence of substance P together with either morphine or DAMGO (Fig 3C, compare lanes 3 and 5 with lanes 4 and 6).
What is the mechanism underlying the HA-NK1R-dependent inhibition of MOR endocytosis? Given that NK1R is a Gq-coupled receptor, one possibility is endocytic inhibition could occur as a consequence of Gq-mediated signaling via phospholipase C (PLC). This hypothesis is potentially consistent with a previous report that protein kinase C (an effector of the Gq-PLC pathway) can inhibit MOR internalization (Ueda et al., 2001), but is contradicted by the recent finding that activation of a distinct Gq-coupled receptor (the 5HT2A serotonin receptor) enhances rather than inhibits MOR internalization (Lopez-Gimenez et al., 2008). Further arguing against this hypothesis, we found that blocking PLC activity in N2A cells using the specific inhibitor U73122 did not prevent NK1R-dependent inhibition of MOR endocytosis (Supplemental Fig. 1). Another hypothesis is that the observed endocytic inhibition is caused by depletion of functional arrestin activity from the cytoplasm by sequestration on the endosome membrane. Whereas many GPCRs (including MOR) rapidly dissociate from arrestins upon endocytosis, a subset of GPCRs (including the NK1R) are capable of driving pronounced sequestration of arrestins on endosomes (Oakley et al., 2001). Further, endosomal sequestration by such GPCRs was shown previously to mediate heterologous inhibition of endocytosis of non-sequestering GPCRs in non-CNS cells (Klein et al. 2001; Schmidlin et al. 2002). To examine whether depletion of arrestin might mediate HA-NK1R-dependent inhibition of MOR endocytosis in neuronal cells, we first asked if increasing cytoplasmic arrestin by over-expression is sufficient to rescue morphine-induced endocytosis of MOR. To do so we focused on arrestin 3 (β-arrestin2), which is the major non-visual arrestin expressed in the striatum (Gurevich et al., 2002). EGFP-tagged β-arrestin2 was transfected into N2A cells co-expressing F-MOR and HA-NK1Rs, and triple color confocal microscopy was used to detect and localize each of the proteins specifically. In contrast to cells expressing arrestins at endogenous levels, co-activation of HA-NK1Rs in cells over-expressing β-arrestin2 failed to inhibit morphine-induced endocytosis of F-MOR (Fig 4). This was indicated by the ability of both F-MOR and HA-NK1Rs to endocytose rapidly to an overlapping population of endocytic vesicles in cells exposed to both morphine and substance P, and these endocytic vesicles also visibly sequestered EGFP-tagged β-arrestin2 from the cytoplasm.
To determine if arrestin sequestration occurs in N2A cells, we co-expressed HA-NK1R and EGFP-tagged β-arrestin2 and examined arrestin redistribution after substance P treatment. Immunofluorescence analysis confirmed that activation of HA-NK1R for 30 minutes was sufficient to promote rapid and pronounced redistribution of EGFP-tagged β-arrestin2 to receptor-containing endosomes (Fig 5A, left column). This effect was specific to HA-NK1Rs because activation of F-MORs with either morphine or DAMGO alone did not produce detectable sequestration of arrestin on endosomes (not shown). By truncating a distal portion of the HA-NK1R tail (HA-NK1 355x), we were able to disrupt substance P-induced arrestin sequestration on endosomes (Figure 5A, right column). This portion of the NK1R tail was previously shown to be required for endosomal sequestration of arrestins in non-neural cells (Oakley et al., 2001). Importantly, substance P-induced receptor endocytosis was not blocked by this mutation (Fig 5A, bottom right).
We next asked if disrupting arrestin sequestration prevents endocytic inhibition of MORs in cells expressing arrestins at endogenous levels. N2A cells expressing F-MOR and HA-NK1 355x were exposed to their respective agonists individually or in combination. As expected, incubation with morphine or substance P alone selectively promoted endocytosis of F-MORs or HA-NK1Rs, respectively (not shown). However, rapid endocytosis of both F-MORs and HA-NK1 355x receptors was also observed in cells exposed to both morphine and substance P, in contrast to the endocytic inhibition of MOR produced by co-activation of the wild type HA-NK1R (Fig 5B, top panels). Similarly, incubation with DAMGO also resulted in MOR endocytosis, even in the presence of substance P (Fig 5B, bottom panels).
Ratiometric quantification of F-MOR internalization in N2A cells co-expressing HA-NK1 355x confirmed that the degree of internalization observed in cells exposed to both morphine and substance P was not significantly different from that observed when exposed to morphine alone (Fig 5C, black bars). Similarly, the endocytic inhibitory effect on DAMGO-induced internalization was also greatly reduced (Fig 5C, white bars). Biochemical analysis using surface biotinylation of receptors further confirmed this loss of endocytic inhibition. Addition of substance P to the culture medium did not detectably reduce the internalized pool of F-MORs produced by either morphine or DAMGO (Fig 5D, compare lanes 3 and 5 with lanes 4 and 6). Together, these results strongly suggest that endosomal sequestration of arrestin by activated HA-NK1Rs is both necessary and sufficient to produce the observed inhibition of F-MOR endocytosis in N2A cells.
To determine whether a similar mechanism mediates the NK1R-dependent endocytic inhibitory effect in striatal neurons, we carried out the equivalent experiment in primary neuronal cultures co-expressing F-MOR and HA-NK1 355x. In contrast to the significant inhibition of morphine-induced endocytosis of F-MOR observed in cells produced by activating wild type HA-NK1Rs, substance P-mediated activation of HA-NK1 355x did not prevent morphine induced endocytosis of F-MOR (Fig 6 top panels, compare to Fig 1B). DAMGO-induced endocytosis of F-MOR was also not detectably impaired by activation of co-expressed HA-NK1 355x receptors (Fig 6, bottom panels).
To begin to assess the functional significance of this arrestin-dependent MOR-NK1R regulatory effect, we examined MOR signaling via heterotrimeric G proteins. Because MORs couple primarily to Gi, we used forskolin to stimulate adenylyl cyclase activity and examined MOR signaling via inhibition of forskolin-induced cAMP production in cultured striatal neurons co-expressing F-MORs and HA-NK1Rs. As expected, exposure of neurons to morphine for 3 minutes significantly inhibited cAMP production (Fig 7A). This acute signaling response was not detectably affected by co-activation of HA-NK1Rs with substance P. Further, consistent with HA-NK1R coupling primarily to Gq, substance P-mediated activation of HA-NK1Rs in the absence of opiate did not change cAMP levels detectably from the control (i.e., forskolin only) condition.
To examine whether NK1R-mediated inhibition of MOR endocytosis is associated with an effect on MOR desensitization, we assayed the ability of morphine to inhibit cAMP accumulation in neurons pre-exposed to agonist. Pre-incubation of neurons with morphine for 30 minutes strongly attenuated the ability of a second challenge of morphine to inhibit forskolin-stimulated cAMP accumulation in neurons, compared to the morphine-induced inhibition observed in opiate-naive neurons (Fig 7B, compare bars 2 and 3 from the left). Interestingly, this morphine-induced attenuation of subsequent opiate response was significantly reduced by inclusion of substance P in the pre-incubation (Fig. 7B, compare bars 3 and 4 from the left). We verified in control experiments that forskolin-stimulated cAMP accumulation measured in the absence of opioid did not differ significantly between the pre-treatment conditions when compared to drug-naïve neurons (Supplemental Fig. 2). Together, these results indicate that NK1R activation does not detectably affect acute opiate signaling, but significantly reduces functional desensitization of MOR signaling produced by opiate pre-exposure.
To determine whether a similar inhibition of MOR endocytosis can occur with receptors expressed at endogenous levels, we explored this phenomenon in several populations of neurons where both MOR and NK1Rs are reportedly co-expressed. In our cultures, immunocytochemical staining of dissociated striatal cultures revealed that MOR and NK1Rs were expressed largely in distinct neurons, making this preparation unfavorable for examining heterologous regulation of endogenous receptor trafficking (data not shown).
Another brain region that is known to co-express these receptors and modulate behavioral reward to opioids is the amygdala (Gadd et al., 2003; Nakaya et al., 2004; Poulin et al., 2006). We cultured primary rat amygdala neurons and found that a subset of these neurons were positive for both endogenous MOR and NK1R immunoreactivity, while others expressed only one but not the other receptor type. In neurons expressing MORs but not NK1Rs, incubation of cultures with both 10 μM DAMGO and substance P for 30 minutes resulted in a pronounced redistribution of MOR immunoreactivity, indicative of rapid internalization of MORs (Fig. 8a, top left). In contrast, redistribution of MORs was substantially reduced in neurons co-expressing both MORs and NK1Rs under the same agonist conditions, while NK1R endocytosis in response to substance P was unaffected (Fig. 8A, bottom row). We verified in control experiments that DAMGO alone stimulated a pronounced redistribution of MOR in both NK1R–positive and –negative neurons (not shown). These results suggest that inhibition of MOR internalization by NK1R activation also occurs in neurons that endogenously express these receptors.
To test if cross-inhibition occurs in another relevant population of CNS neurons, and to obtain quantitative data, we examined the effect of substance P on MOR internalization using locus coeruleus neurons from F-MOR Tg+/−, MOR−/− mice (Arttamangkul et al., 2008). This brain region consists primarily of noradrenergic neurons, which possess a high density of both endogenous MORs and NK1Rs (Cheeseman et al., 1983; Tempel and Zukin, 1987; Nakaya et al., 1994). In slices imaged in the absence of agonist, immunoreactive MOR was localized primarily in the plasma membrane, as indicated by the peripheral staining pattern in 2-photon optical sections (Fig. 8B, top left panel). Because morphine does not cause significant internalization in LC neurons, [Met]5enkephalin (ME) was used in these experiments (Arttamangkul et al., 2008). Bath application of 10μM ME for 15 minutes induced a redistribution of F-MOR receptors to a punctate pattern consistent with regulated endocytosis of MOR (Fig. 8B, top middle panel). This was confirmed by subsequent application of calcium-free EGTA solution, which dissociates antibody from surface-accessible receptors, leaving the internalized receptor pool specifically labeled (Fig. 8B, top right panel). Application of 10μM substance P to slices did not detectably change the plasma membrane localization pattern of labeled MOR (Fig. 8B, bottom left panel), indicating that activating the NK1Rs alone did not promote significant internalization of MOR. Subsequent application of ME, in the continued presence of substance P, produced some redistribution of MOR but the magnitude of this effect was visibly reduced compared to that observed in slices exposed to ME alone (Fig. 8B, middle panels). Consistent with this, the internalized fraction of labeled MOR, which is resistant to the EGTA strip condition, was also visibly reduced in slices exposed to ME in the presence of substance P (Fig. 8B, right panels). Quantification of these results over multiple experiments confirmed that substance P significantly inhibited ME-induced internalization of MOR in locus coeruleus neurons (Fig. 8C).
The present results demonstrate that NK1R activation mediates a pronounced and cell autonomous inhibition of MOR endocytosis induced by diverse opioid agonists. Whereas either morphine or DAMGO promote rapid endocytosis of MOR in striatal neurons, simultaneous activation of NK1Rs with substance P strongly inhibited this fundamental regulatory process. We identified a transformed neurosecretory cell line in which pronounced morphine-induced endocytosis of MOR also occurs, and observed a similar endocytic inhibition mediated by NK1R activation in these cells. Our results indicate that, in both cell types, inhibition of MOR endocytosis is mediated by NK1R-dependent sequestration of arrestins on endosomes. This cell-autonomous regulatory mechanism was also found to produce a functionally significant attenuation of morphine-induced desensitization of MOR signaling via adenylyl cyclase.
Furthermore, heterologous regulation of MOR trafficking was established in two additional populations of CNS neurons that mediate behavioral effects of opiate drugs and was observed in neurons expressing receptors at endogenous levels. Together, these observations strongly suggest that NK1Rs can significantly modulate both trafficking and downstream signaling of MORs in multiple populations of physiologically relevant CNS neurons.
Our findings thus identify a novel mode of functional interaction between MORs and NK1Rs. A previous study of these two GPCRs expressed in a non-neural cell model (HEK293 cells) reported that ligand-induced activation of either MOR or NK1R promoted co-endocytosis of both GPCRs by formation of MOR/NK1R heterodimers (Pfeiffer et al., 2003). Such heterodimer-driven endocytic “dragging” was not clearly evident in the present study, as opioid and neurokinin agonists selectively promoted endocytosis of MORs and NK1Rs, respectively. Further, several observations argue that the regulatory interaction revealed in the present study cannot be explained by heterodimer formation. First, activation of NK1Rs inhibited, rather than promoted, MOR endocytosis. Second, we did not observe rapid endocytosis of either MOR or NK1R in the absence of the appropriate receptor agonist. Third, the observed cross-regulatory effect was non-reciprocal, as NK1R activation inhibited MOR endocytosis but not vice versa. The present results do not exclude heterodimer-based regulation of opioid receptors by neurokinins, and we are aware that the degree to which various GPCRs form heteromeric complexes varies among cell types (Bouvier, 2001). Our findings argue strongly, however, that the distinct mechanism established in the present study plays a dominant role in mediating functional interaction between MOR and NK1Rs in CNS neurons.
Cell autonomous, non-reciprocal endocytic inhibition has been observed previously among various GPCRs in non-neural cells as well as myenteric neurons (Klein et al., 2001; Schmidlin et al., 2002), and is consistent with the existence of significant differences among individual GPCRs in their ability to mediate arrestin trafficking from the cytoplasm to the endosome membrane (Oakley et al., 1999). NK1Rs, in particular, are able to mediate arrestin sequestration on endosomes in HEK293 cells (Oakley, 2001; Schmidlin et al., 2002). To our knowledge the present results are the first to demonstrate that arrestin sequestration can mediate heterologous regulation of GPCR trafficking in CNS neurons. Second, we show that this links NK1R activation to regulation of MORs in multiple populations of CNS neurons relevant to the behavioral effects of opiate drugs, and that this regulation can be observed among endogenously expressed receptors. Thus, we believe that arrestin trafficking to endosome membranes likely represents a fundamental mechanism for integrating receptor-mediated signaling effects across distinct GPCRs in diverse neural cell types. Further, this function of arrestin trafficking in signal integration is clearly distinct from the previously proposed role of endosome-associated arrestins in scaffolding signaling complexes on the endosome membrane in response to activation of particular GPCRs (Lefkowitz et al., 2006). Although the actual stoichiometry of non-visual arrestins (or β-arrestins) relative to GPCRs is not clearly established in native cell types, our finding that endogenously expressed receptors mediate significant heterologous inhibition in both amygdala and locus coeruleus neurons suggests that the effective functional activity of β-arrestins is a limiting factor determining endocytic regulation of GPCRs in relevant CNS neurons. This conclusion, in addition to its specific relevance to opioid regulation, suggests a general principle that may underlie regulation between diverse GPCR-mediated signaling processes occurring simultaneously in the same neurons.
It is likely that the ability of NK1R activation to alter morphine’s endocytic activity can persist even in the absence of continuous substance P release. Recycling of NK1Rs is reported to be slow in various cell types (Grady et al., 1995; Grady et al., 1996; Wang et al., 2002). Dissociation of the β-arrestin-receptor complex is dependent on dephosphorylation of receptors in endosomes and important for resensitization of internalized NK1Rs (Garland et al., 1996). Prolonged retention of β-arrestins on endosomes, together with slow recycling of NK1Rs, could thus result in low cytoplasmic levels of this regulatory protein after an initial round of receptor activation. In addition, the lack of MOR desensitization as a result of inhibiting endocytosis alters the normal signaling response to drugs such as morphine. Thus, it is possible that the NK1R-mediated regulation of MOR endocytosis, in addition to its acute trafficking effects, contributes to longer-term plasticity in the regulatory profile of opioid drugs.
We were surprised to find a lack of overlap in MOR and NK1R expression on our dissociated striatal culture preparation. It is possible that this reflects developmental differences between our cultures (which are derived from embryonic rats) and adult animals used in previous studies (Pickel et al., 2000; Jabourian et al., 2005). However, we did observe that MORs and NK1Rs are co-expressed in a significant fraction of cultured amygdala neurons. This brain region is known to be important for mediating conditioned reward processes, expresses both MOR and NK1R, and ablation of NK1Rs in these neurons has previously been reported to reduce morphine reward behavior (Everitt et al., 1991; Gadd et al., 2003). We also observed substance P-induced heterologous inhibition of MOR internalization in an acute slice preparation of locus coeruleus, using a recently developed method that allowed quantitative assessment of MOR endocytosis (Arttamangkul et al., 2008). In addition to the amygdala, NK1R-expressing neurons in the locus coeruleus may also play a role in morphine-driven behaviors (Nestler et al., 1994) and in opiate withdrawal (Redmond and Krystal, 1984; Rasmussen 1990). Therefore, it is conceivable that the cell autonomous regulation established in the present study contributes to NK1R-dependent modulation of opiate responses evident from study of NK1R-knockout animals and animals in which NK1R function is chemically ablated (Gadd et al, 2003). There is also considerable co-expression of MOR and NK1R in the trigeminal dorsal horn (Aicher et al., 2000). Although we did not examine this region in the present study, and NK1Rs are not essential for morphine antinociception in vivo (De Felipe et al, 1998), it is possible that receptor co-expression could confer NK1R-dependent regulation on opioid signaling in these neurons as well.
The observed inhibitory effect of NK1R activation on MOR endocytosis and desensitization in CNS neurons suggests a novel mechanism by which neurokinin signaling may affect opioid responses in multiple brain regions. Important questions for future study include further examination of the effects of cell autonomous regulation on opioid-mediated neurophysiology and investigation into how interaction between these two neurochemical systems affects functional connectivity between other brain regions. For example, the central nucleus of the amygdala receives reciprocal dopaminergic afferents from the ventral tegmental area and is connected to the core of the nucleus accumbens in the striatum (Ungerstedt, 1971; Wallace et al., 1992). It is conceivable, therefore, that the ability of other GPCRs to alter the trafficking and signaling properties of MORs in response to opiates could have profound effects on neural circuits driving reward behavior. By establishing cell autonomous regulation of MOR by NK1R activation in several neuronal populations, the present study identifies a new principle by which diverse GPCR-linked signaling systems may be coordinately regulated in the CNS.
N2A cells co-expressing F-MOR and HA-NK1R were pre-incubated for 15 minutes with 10μM of either the PLC inhibitor (U73122; Sigma) or its inactive analog (U73433; Sigma) before adding the indicated agonist(s) and incubating cells for an additional 30 minutes in the continued presence of U73122 or U73433. (A) Representative epifluorescence micrographs showing localization of F-MOR (top row) and HA-NK1R (bottom row) in cells exposed to the active PLC inhibitor U73122 and incubated for 30 minutes in the presence of 10 μM DAMGO (+DG), 10 μM substance P (+SP) or both (+DG +SP). The PLC inhibitor did not prevent endocytosis of either receptor induced by its respective agonist when applied alone (compare top and bottom panels of the left and middle columns), nor did it prevent the SP-dependent inhibition of MOR endocytosis (compare top and bottom panels of the right column). (B) The corresponding micrographs from cells incubated in the presence of the inactive analog U73433, confirming that SP-dependent inhibition of MOR endocytosis was indistinguishable from that observed in the presence of the active PLC inhibitor (compare with the corresponding panels in part A of the figure). Scale bar, 20μM.
Striatal neurons co-expressing F-MOR and HA-NK1R were incubated for 30 minutes in the absence or presence of the indicated agonists, washed free of ligand as in Figure 7B, and re-challenged with forskolin alone for 10 minutes to stimulate cAMP accumulation. Pre-incubation with 10μM SP did not detectably change subsequent forskolin-stimulated cAMP accumulation compared to untreated neurons (n=4, p=0.4). No significant effect was observed after pre-incubation with 10μM morphine alone, compared to no pretreatment (n=5, p=0.1), or pre-incubation with both 10μM morphine and 10μM substance P (n=5, p=0.9).
The authors thank Nigel Bunnett and Marc Caron for valuable discussion and reagents, and James Hislop for critical comments on the manuscript. Y.J.Y. received support from a Gallo Predoctoral Fellowship in Alcohol and Addiction Studies and an NIH predoctoral fellowship (F31DA024530-01). These studies were supported by research grants from the NIH.