Abstract
Chronic morphine treatment produces behavioral and cellular opioid tolerance that has been proposed to be caused by attenuated μ-opioid receptor (MOR) recovery from desensitization (resensitization). The process of MOR resensitization is thought to require βarrestin-2 (βarr-2)-dependent trafficking of desensitized receptors to endosomal compartments, followed by recycling of resensitized receptors back to the plasma membrane. However, there is little direct evidence for this, particularly in native neurons. This study used whole-cell patch-clamp recording in locus ceruleus (LC) neurons from wild-type (w.t.) and βarr-2 knock-out (k.o.) mice to examine whether βarr-2/dynamin-dependent trafficking is required for MOR resensitization in neurons from opioid-naive and morphine-treated mice. Surprisingly, recovery of MOR from acute desensitization in LC neurons does not require βarr-2- or dynamin-dependent trafficking. To the contrary, MOR resensitization was accelerated by disruption of either βarr-2 or dynamin function. Chronic morphine treatment caused cellular MOR tolerance and concurrently impaired MOR resensitization in neurons from w.t. mice, as expected from previous studies, but neither occurred in neurons from βarr-2 k.o. mice. Moreover, the impairment of MOR resensitization caused by chronic morphine was reversed in w.t. neurons when G-protein-coupled receptor kinase-2 (GRK2) or dynamin function was disrupted. Together, these results establish that βarr-2/dynamin-dependent receptor regulation is not required for MOR resensitization in LC neurons. Furthermore, chronic morphine treatment modifies GRK2–βarr-2–dynamin-dependent MOR trafficking to impair receptor resensitization, thereby contributing to opioid tolerance in LC neurons by reducing the number of functional receptors on the surface membrane.
Introduction
Opioids are potent and effective analgesics, but their utility is greatly limited by the development of tolerance (Buntin-Mushock et al., 2005). Analgesic opioids mediate their effects by activating μ-opioid receptor (MOR) (Kieffer and Gavériaux-Ruff, 2002). Sustained agonist activation of MOR initiates rapid regulatory events, including receptor desensitization and trafficking, that are thought contribute to the behavioral opioid tolerance that develops during prolonged opioid administration (von Zastrow, 2001). MOR regulation resembles that of the well characterized β2 adrenoceptor. Briefly, G-protein-coupled receptor kinase-2 (GRK-2) phosphorylation of the agonist-bound β2 adrenoceptor enhances its affinity for βarrestin-2 (βarr-2) binding, triggering receptor endocytosis via clathrin–dynamin-dependent mechanisms (Gainetdinov et al., 2004). After endocytosis, receptors are resensitized and recycled (von Zastrow et al., 2003). For MOR, βarr-2-dependent endocytosis and recycling are considered essential for receptor resensitization (Law et al., 2000; Koch et al., 2005), but there is no direct evidence for this. Indeed, blocking MOR endocytosis with concanavalin-A in neurons did not prevent MOR resensitization (Arttamangkul et al., 2006), suggesting that recycling may not be necessary for resensitization.
The failure of morphine to induce MOR endocytosis (Keith et al., 1996; Sternini et al., 1996) was proposed to cause morphine tolerance (Law et al., 2000; Koch et al., 2005) because desensitized MOR accumulates on the surface membrane in which it cannot resensitize. Such a mechanism is inconsistent with the in vivo finding that morphine tolerance is attenuated in βarr-2 knock-out (k.o.) mice in which MOR endocytosis is presumably impaired (Bohn et al., 1999, 2000, 2002). Alternatively, attenuated opioid tolerance in βarr-2 k.o. mice may be caused by the loss of βarr-2-mediated MOR desensitization in neurons. However, studies in single locus ceruleus (LC) (Dang et al., 2009) and dorsal root ganglion (DRG) (Walwyn et al., 2007) neurons have shown that MOR desensitization is unaffected by βarr-2 deletion. Therefore, it remains unclear how βarr-2-dependent MOR regulation contributes to cellular opioid tolerance in native neurons.
To determine the association between βarr-2-dependent MOR regulation and cellular tolerance, we used whole-cell patch-clamp recordings to study MOR function in brain slices containing LC neurons from chronic morphine-treated wild-type (w.t.) and βarr-2 k.o. mice. The LC comprises a uniform population of MOR-expressing noradrenergic neurons in which homologous desensitization of MOR coupling to GIRK channels (GIRK) and cellular tolerance are well established (Connor et al., 2004). The results show that MOR resensitization does not require βarr-2-dependent MOR trafficking or dynamin-dependent receptor endocytosis. MOR resensitization was actually more efficient when βarr-2-dependent receptor trafficking was disrupted. Chronic morphine treatment produced cellular tolerance in LC neurons from w.t. but not βarr-2 k.o. mice. Concurrently, MOR recovery from brief acute desensitization was impaired in neurons from morphine-treated w.t. but not βarr-2 k.o. mice. This impairment of MOR resensitization in w.t. neurons was reversed when GRK-2 or dynamin function were disrupted. Together, these results show that βarr-2-dependent trafficking is not required for MOR resensitization in LC neurons but does contribute to impaired receptor resensitization associated with cellular opioid tolerance.
Materials and Methods
Electrophysiology.
Whole-cell recordings were made from 167 routinely genotyped, 5- to 16-week-old male βarr-2 knock-out mice from Drs. Lefkowitz and Caron (Duke University, Durham, NC)) or their wild-type littermates (w.t. C57BL/6 background). LC slices were prepared as described previously for rats (Osborne and Williams, 1995). Briefly, mice were anesthetized with isoflurane (4% in air) and decapitated, and horizontal vibratome (Leica VT1000) sections (200–220 μm) were prepared at 4°C and then incubated in extracellular solution (ACSF, 60 min, 37°C). ACSF contained the following (in mm): 126 NaCl, 2.5 KCl, 2.4 CaCl2, 1.2 MgCl2, 1.2 NaH2PO4, 21.4 NaHCO3, and 11.1 glucose (95% O2/5% CO2 at 37°C). Pipette (2–4 MΩ) solution contained the following (in mm): 115 K-MES, 20 KCl, 1.5 MgCl2, 10 BAPTA, 5 HEPES, 4 Mg-ATP, and 0.4 Na-GTP, pH 7.3–7.4. Voltage-clamp recordings of visualized LC neurons (infrared Nomarski optics; holding potential, −55 to −60 mV) were acquired using Axograph X (Axograph Scientific) and filtered at 10–100 Hz. Bestatin (10 μm) and thiorphan (1 μm) were included in all experiments using met5-enkephalin (ME). All experiments were approved by the Royal North Shore Hospital/University of Technology Sydney Ethics Committee, which complies with National Health and Medical Research Council of Australia guidelines.
For recording in LC neurons, inhibitory peptides or microcystin were backfilled into the recording electrodes. Soon after achieving whole-cell access (≤5 min), GIRK activation in LC neurons by a low concentration of naloxone (NA) (3 μm) or ME (300 nm) was assessed to determine the control response. Neurons were then dialyzed with inhibitory peptide or microcystin for up to 60 min, during which NA (3 μm) or ME (300 nm) activation of GIRK was assessed periodically. No changes in NA (3 μm) or ME (300 nm) activation of GIRK currents were detected by the loading of inhibitory peptides (100 μm) or microcystin (25 nm) (data not shown).
MOR internalization.
Mouse AtT20 neuroblastoma cells stably transfected with N-terminal Flag-tagged ì-opioid receptors (FLAG-MOR) as described previously (Borgland et al., 2003) were grown to 90% confluence in DMEM containing fetal bovine serum (10%), penicillin/streptomycin (50 U/5 μg/ml) and G418 (5 μg/ml), passaged, and seeded onto poly-l-lysine (100 μg/ml)-coated glass coverslips (∼50% density) in 35 mm dishes. The day after passage, coverslips were placed in extracellular HEPES-buffered saline (HBS) solution [in mm: 150 NaCl, 2.5 KCl, 1.8 CaCl2, 1 MgCl2, 10 HEPES, 10 glucose, pH 7.3 (330 ± 5 mOsm)] at room temperature and allowed to equilibrate for 30 min before the commencement of the experiment. Patch-clamp recording pipettes were filled with internal solution containing 5 mm NaCl, 130 mm KCl, 10 mm EGTA, 20 mm HEPES, 2 mm CaCl2, 5 mm MgATP, 0.2 mm NaGTP, 1.5 mg/ml biocytin alone (Sigma) or biocytin plus either a myristilated dynamin inhibitory peptide (DYNi) (100 μm; Tocris Bioscience) or a GRK2 inhibitory peptide (GRK2i) (100 μm, pH 7.3, 285 ± 5 mOsm). Electrodes were made with fire-polished borosilicate glass (A-M Systems) and had an average resistance of ∼2.5 MΩ. Individual cells were loaded with inhibitor/biocytin using a whole-cell configuration of the patch-clamp technique and by diffusion from the pipette shaft. Cells were voltage clamped at −60 mV and filled for 1, 5, or 10 min. Intracellular exposure to GRK2 or dynamin inhibitory peptides for 5 min or less did not produce reliable inhibition of endocytosis (data not shown). Patch electrodes were then carefully removed from the cell, and the cell was allowed to recover for 10 min. Cells were then transferred into HBS containing ME (30 μm), bestatin (10 μm), and thiorphan (1 μm) and incubated at 37°C for 30 min to induce internalization of FLAG-MOR. Incubations were stopped in ice-cold PBS (0.1 m, pH 7.2), fixed in 4% paraformaldehyde for 20 min, and then rinsed thoroughly with PBS at 4°C. Cells were permeabilized ∼24 h after fixation with ice-cold, absolute methanol for 5 min and then rinsed with PBS. They were then incubated with 10% normal goat serum for 1 h at room temperature and then for 2 h at room temperature with an anti-FLAG polyclonal antibody (2 μg/ml) raised in rabbit (Sigma). After three 10 min rinses with PBS, cells were incubated for 90 min with avidin-conjugated Alexa Fluor-647 (1:500; Invitrogen) and an anti-rabbit Alexa Fluor-488-conjugated secondary antibody (1:200; Invitrogen) diluted with 2% normal goat serum in PBS. Coverslips were washed three times with PBS and then mounted onto slides with Fluoromount (Sigma) mounting medium. Images were acquired with an Olympus FV300 confocal microscope using a 60× oil objective and a 1.5× digital zoom, and extent of internalization was scored by an observer blind to peptide exposure during patch-clamp recording.
Chronic morphine treatment.
Mice were lightly anesthetized with isoflurane as described previously (Bagley et al., 2005a). Three subcutaneous injections of a sustained release preparation of morphine [free base 300 mg/kg suspended in 0.1 ml of Arlacel A (mannide monooleate), 0.4 ml of light liquid paraffin, and 0.5 ml of 0.9% w/v NaCl] on alternate days over a 5 d period. Vehicle mice were injected with suspension lacking morphine.
Statistics and curve fitting.
Data were analyzed using PRISM (GraphPad Software). Values are given as mean ± SEM. Statistical comparisons were made with t tests (for two groups) or two-way ANOVA with Bonferroni's post hoc tests and considered significant if p < 0.05.
Reagents.
Reagents were from the following: ME, thiorphan, and bestatin were from Sigma and naloxone, morphine base, and GSK were from RBI. The GRK2i sequence W643-S670 (Ac-WKKELRDAVREAQQLVQRVPKMKNKPRS-NH2, >95%) was custom synthesized by AUSPEP and dynamin inhibitory peptide by Tocris Bioscience.
Results
βArrestin-2 deletion accelerates the rate of MOR resensitization
Desensitization was defined here as the loss of MOR function that develops rapidly (several minutes) during sustained application of high concentrations of agonists (Connor et al., 2004). Activation of MOR with a supramaximal concentration of the endogenous opioid ME (30 μm, ≥10 min) has been widely reported to produce maximal desensitization and robust MOR endocytosis in LC neurons (Osborne and Williams, 1995; Bailey et al., 2004; Dang and Williams, 2005; Arttamangkul et al., 2006; Dang et al., 2009). Desensitization of MOR-activated GIRK currents in LC neurons provides a rapid, reliable assay of G-protein activation that does not involve loss of channel function because agonists at other GPCRs are still able to fully activate GIRK currents when MOR is desensitized in these cells (Connor et al., 2004). We previously reported that the rate and extent of agonist-induced acute desensitization of GIRK currents are nearly identical in LC neurons from w.t and βarr-2 k.o. mice and that GIRK activation can still be achieved by α2 adrenoceptor agonist, confirming that desensitization of MOR-activated GIRK is mostly homologous (Dang et al., 2009). Figure 1 confirms that acute desensitization of MOR-activated GIRK currents was unaffected by βarr-2 deletion (Dang et al., 2009). As reported previously in detail (Dang et al., 2009), the peak GIRK current activated by ME (30 μm) was slightly smaller in βarr-2 k.o. than w.t. mice (77 ± 9 pA, n = 13 vs 94 ± 7 pA, n = 15, respectively; p > 0.157), as was the response to a submaximal concentration of ME (300 nm; 21 ± 3 pA, n = 24 vs 26 ± 3 pA, n = 16, respectively; p > 0.22) (Fig. 1A–C). The potency of ME was also similar in βarr-2 k.o. and w.t. LC neurons [pEC50 from nonlinear logistic fitting: 5.74 vs 5.88 (−log M), respectively]. This suggests little or no change in efficacy of MOR coupling to GIRK in LC neurons from βarr-2 k.o. as reported previously (Dang et al., 2009). Desensitization of the peak response after 10 min exposure to ME (30 μm) did not differ between βarr-2 k.o. and w.t. LC neurons (51 ± 4% of the initial peak, n = 13 vs 54 ± 3%, n = 15, respectively) (Fig. 1A). MOR desensitization determined using a submaximal probe concentration of ME (300 nm) applied immediately after washout of the supramaximal concentration is identical [29 ± 6% (n = 4) vs 28 ± 6% (n = 7) of the prepulse response, respectively] (Fig. 1A). These findings confirm that a submaximal probe concentration of ME (300 nm) provides a sensitive measure of MOR function that is comparable across genotypes (Connor et al., 2004).
Experiments were then conducted to determine the effect of βarr-2 deletion on MOR recovery from acute desensitization. Receptor resensitization was assessed by comparing MOR activation of GIRK by ME (300 nm) before and at various times after desensitization. Consistent with previous reports (Osborne and Williams, 1995; Dang and Williams, 2004), MOR activation of GIRK exhibited slow but progressive recovery after the washout of ME (30 μm; 10 min). In neurons from w.t. mice, MOR resensitization approached full recovery after 70 min (Fig. 1B,D). In contrast, MOR resensitized more rapidly in neurons from βarr-2 k.o. mice (Fig. 1C,D) (p < 0.0001, two-factor ANOVA). MOR activation of GIRK fully recovered after 20–30 min in neurons from βarr-2 k.o., whereas at the same time points, neurons from w.t. mice exhibited only 45 ± 3% (n = 16) and 51 ± 3% (n = 18) recovery, respectively (Fig. 1D).
Disruption of βarrestin-2-dependent receptor trafficking accelerates MOR resensitization
The finding that MOR resensitization is more efficient in LC neurons from βarr-2 k.o. mice suggests that MOR resensitization does not require βarr-2-dependent receptor trafficking and, presumably, receptor endocytosis. To corroborate this, manipulations were applied to prevent βarr-2-dependent MOR endocytosis in LC neurons from w.t. mice. MOR endocytosis was disrupted either upstream (GRK2 inhibition) or downstream (dynamin inhibition) of βarr-2 association with the MOR (Gainetdinov et al., 2004). First, neurons from w.t. mice were loaded intracellularly with a GRK2i peptide via the recording electrode (100 μm) to disrupt GRK2 activation by Gβγ subunits (Koch et al., 1993; Li and Wang, 2001; Dang et al., 2009).
Confirmation that GRK2i peptide disrupted βarr-2-dependent endocytosis was performed in cultured cells because quantification of endocytosis of native MOR in LC is not readily achieved using confocal microscopy (Scavone and Van Bockstaele, 2009), particularly after preparation of brain slices and whole-cell recording procedures. Immunohistochemical staining of native MOR after recording LC neurons was not sufficiently sensitive to determine disruption of internalization by GRK2i (data not shown). However, under the same recording conditions, intracellular application of GRK2i peptide prevented endocytosis of Flag-tagged MOR (Borgland et al., 2003) in cultured AtT20 cells (see Fig. 3). Nine cells in total were filled with GRK2i/biocytin. Of these, GRK2i completely blocked ME-induced internalization of FLAG-MOR in six cells, one cell showed marginal internalization of FLAG-MOR, and the remaining two cells showed moderate but reduced internalization, suggesting that the peptide disrupts MOR internalization as expected. No internalization was observed in untreated, biocytin-filled cells not exposed to ME (see Fig. 3A) (n = 4) but was observed in all biocytin-filled cells exposed to ME (see Fig. 3B) (n = 6).
As reported previously (Dang et al., 2009), disruption of GRK2 function alone did not affect MOR desensitization in neurons from w.t. or βarr-2 k.o. mice (Fig. 2B). As predicted from results in βarr-2 k.o. mice, disruption of GRK2 function in neurons from w.t. mice significantly increased the rate of MOR resensitization (Fig. 2B) (two-factor ANOVA, p < 0.0001, all points to 30 min GRK2i vs w.t. control). Whereas MOR activation of GIRK approaches full recovery after 70 min under control conditions, complete recovery was observed after 20–30 min in neurons filled intracellularly with GRK2i.
In addition to GRK2 and βarr-2, LC neurons also express GRK3, GRK5, and GRK6 (Erdtmann-Vourliotis et al., 2001) and βarr-1 (Fan et al., 2003). It is possible that MOR endocytosis can occur via βarr-1-dependent mechanism after phosphorylation by other GRKs. To test this, experiments were conducted with DYNi peptide to prevent dynamin-dependent endocytosis, which encompasses all βarrestins, and clathrin-dependent endocytosis and perhaps caveolar-mediated endocytosis (Doherty and McMahon, 2009). When DYNi peptide was loaded into w.t. neurons via the recording electrode, MOR resensitization was dramatically accelerated (Fig. 2C,D), with nearly complete recovery after 10 min (90 ± 8%, n = 6). The increased rate of recovery with DYNi in w.t. LC neurons was slightly faster than in βarr-2 k.o. neurons without DYNi (p < 0.001, two-factor ANOVA). Figure 2E shows that disruption of dynamin function with DYNi peptide also slightly facilitated MOR resensitization in neurons from βarr-2 k.o. mice (two-factor ANOVA, p < 0.01, but post hoc tests for individual time points did not differ). In contrast, recovery in the presence of DYNi did not differ between w.t. and βarr-2 k.o. neurons (two-factor ANOVA, p > 0.66). This suggests that there may be some dynamin-dependent MOR endocytosis in LC neurons from βarr-2 k.o. mice, perhaps mediated by βarr-1, that partially contributes to the attenuation of MOR resensitization.
To confirm that DYNi peptide can block dynamin function, DYNi was loaded into cultured AtT20 cells using the same loading conditions as that used for electrophysiological recordings. Confocal imaging of MOR endocytosis in AtT20 cells shows that DYNi peptide robustly blocked dynamin-dependent endocytosis (Fig. 3). DYNi completely inhibited internalization in six of eight cells filled with DYNi/biocytin. In the remaining two cells, slight internalization could be visualized, suggesting that the peptide disrupts MOR internalization as expected.
To further establish whether MOR endocytosis is required for MOR resensitization, LC neurons from w.t. mice were treated with hypertonic sucrose (50 mm) before and during ME desensitization. We have reported previously that treatment of slices with hypertonic sucrose (50 mm) impairs vesicle recycling and presumably endocytosis (Bagley et al., 2005b). Similar to blocking dynamin-dependent endocytosis with DYNi peptide, hypertonic sucrose did not affect MOR activation of GIRK currents or acute desensitization but significantly enhanced the rate of MOR resensitization (Fig. 2E) (two-factor ANOVA, p < 0.0001). Together, these results strongly suggest that MOR resensitized more efficiently when MOR endocytosis was disrupted.
In the absence of MOR endocytosis, MOR resensitization is the same for LC neurons from w.t. and βarr-2 k.o. mice
If βarr-2-mediated MOR trafficking and endocytosis causes the difference in the rate of recovery in neurons from w.t. and βarr-2 k.o. mice, then the rate and magnitude of MOR resensitization should be similar for neurons from w.t. and βarr-2 k.o. mice under conditions for which MOR endocytosis is minimal. To test this hypothesis, MOR resensitization of LC neurons from w.t. and βarr-2 k.o. mice was determined after MOR activation of GIRK was briefly desensitized for ≤2 min using 30 μm ME to achieve a more rapid peak response than could be achieved with 10 μm during very brief agonist exposure. Similar results were obtained in several cells exposed to 10 μm ME (data not shown). This treatment has been shown to produce significant MOR desensitization but cause little or no detectable MOR endocytosis (Borgland et al., 2003; Arttamangkul et al., 2006). Indeed, desensitization treatment that causes little or no MOR endocytosis mediates indistinguishable MOR desensitization and resensitization in neurons from w.t. and βarr-2 k.o. mice (Fig. 4). Neurons from both w.t. and βarr-2 k.o. mice exhibited nearly complete recovery after 10 min (93 ± 6 and 81 ± 6%, respectively). Together, the results show that βarr-2-dependent receptor endocytosis attenuates MOR resensitization and is not an obligatory component of the resensitization process as thought previously.
Inhibition of protein phosphatase function prevents MOR resensitization
Dephosphorylation of desensitized GPCRs is considered essential for receptor resensitization (Osborne and Williams, 1995; Zhang et al., 1997a; Oakley et al., 1999; Qiu et al., 2003). Inhibition of dephosphorylation by intracellular application of the relatively selective protein phosphatase 1 and 2A inhibitor microcystin (Gehringer, 2004) was reported to slow MOR resensitization in rat LC neurons (Osborne and Williams, 1995), but this could have disrupted either intracellular recycling mechanisms or cell surface resensitization. To test whether receptor dephosphorylation is required for MOR resensitization in LC neurons when arrestin-dependent mechanisms are blocked, microcystin (25 nm) was included in the recording electrode, as described previously (Osborne and Williams, 1995). Figure 5 shows that accelerated MOR resensitization (GRK2i in βarr-2 k.o. neurons) was significantly suppressed in the presence of microcystin (two-factor ANOVA, p < 0.0001). After a partial recovery in the first 5 min after desensitization (ME at 30 μm, 10 min), no additional recovery developed for the remainder of the experiment. These results suggest that dephosphorylation of desensitized MOR is essential for complete recovery of receptor function, and this occurs efficiently, independent of GRK2–βarr-2-dependent MOR trafficking.
LC neurons from βarr-2 knock-out mice elude cellular tolerance after chronic morphine treatment
Chronic morphine treatment induces cellular tolerance, defined here as the rightward shift in the concentration–response curve for MOR effector signaling, which is associated with reduced efficacy of both MOR activation of GIRK (Christie et al., 1987) and inhibition of voltage-gated calcium channels in LC (Connor et al., 1999) and various other neurons (Bagley et al., 2005a). However, the mechanisms responsible are still unknown. Previous studies have reported that morphine analgesic tolerance is greatly attenuated in βarr-2 k.o. mice, as is tolerance to MOR-activated GTPγS binding in brainstem membranes (Bohn et al., 2000, 2002). To determine whether this occurs for MOR-activated GIRK in LC neurons, w.t. and βarr-2 k.o. mice were treated with a sustained release preparation of morphine or vehicle for 6 d as described previously (Connor et al., 1999). LC slices were prepared and maintained in the absence of morphine for 1 h before experimental recordings. Figure 6 shows that the concentration–response curve for ME activation of GIRK currents in neurons from morphine-treated w.t. mice is depressed and shifted to the right, representing MOR tolerance in single neurons (two-factor ANOVA, drug treatment effect, p < 0.0001). This is consistent with previous reports for LC neurons from morphine-treated rats (Christie et al., 1987; Dang and Williams, 2004). When the same experiments were conducted in neurons from morphine-treated βarr-2 k.o. mice, the concentration–response curves were superimposed (two-factor ANOVA, p > 0.65), indicating that cellular tolerance was abolished.
Impaired MOR resensitization in LC neurons from chronic morphine-treated mice is rescued by disruption of βarr-2-dependent MOR regulation and dynamin-dependent endocytosis
Previous studies have shown that, concurrently with cellular morphine tolerance in LC neurons (Christie et al., 1987), chronic morphine treatment modified MOR regulation such that receptor resensitization was impaired, even after a very brief desensitization treatment with ME (Dang and Williams, 2004). Attenuated MOR resensitization could contribute to opioid tolerance by reducing the number of functional receptors at the cell surface. The mechanism of impaired MOR resensitization after morphine treatment is not known. Having shown that βarr-2-dependent MOR trafficking and dynamin-dependent endocytosis can attenuate MOR resensitization in LC neurons from opioid-naive mice, experiments were conducted to determine whether these processes contribute to impaired MOR resensitization after chronic morphine treatment. To accomplish this, MOR resensitization from brief desensitization treatment (as shown in Fig. 4) was determined in neurons from chronic morphine-treated (6 d) w.t. and βarr-2 k.o. mice. In opioid-naive mice, neurons from both w.t. and βarr-2 k.o. mice exhibited the same extent of desensitization (Dang et al., 2009) and indistinguishable rates of resensitization (Fig. 4). Consistent with what had been reported for LC neurons from morphine-treated rats (Dang and Williams, 2004), recovery of MOR activation of GIRK from a very brief (≤2 min) desensitization treatment was greatly attenuated in neurons from morphine-treated w.t. mice compared with w.t. control (vehicle-treated) mice (Fig. 7). However, chronic morphine treatment did not impair MOR resensitization in neurons from βarr-2 k.o mice, suggesting that βarr-2-dependent processes cause the impaired MOR resensitization after chronic morphine treatment. To validate this, neurons from chronic morphine-treated w.t. mice were loaded intracellularly with GRK2i or DYNi. Rapid MOR resensitization was rescued in neurons from chronic morphine-treated w.t. mice when either GRK2 or dynamin function was disrupted (Fig. 7). Together, these results show that chronic morphine treatment impairs MOR resensitization in LC neurons in a GRK2–βarr-2–dynamin-dependent manner. This finding may provide an explanation for the lack of cellular MOR tolerance in LC neurons from βarr-2 k.o. mice.
Discussion
This study establishes that MOR resensitization from acute desensitization does not require βarr-2 or dynamin-dependent MOR trafficking. To the contrary, βarr-2-dependent trafficking and presumably endocytosis attenuate MOR resensitization under basal conditions. Although it was not possible to quantify whether the manipulations used during patch-clamp recording (GRK2i and DYNi) disrupt endocytosis of native MOR in LC neurons, the same manipulations and recording conditions did block endocytosis of Flag-tagged MOR in cultured cells, suggesting that the inhibitory peptides function as expected and reported (Dang et al., 2009). Interestingly, βarr-2-dependent mechanisms are involved in the impairment of MOR resensitization in LC neurons caused by chronic morphine treatment. When βarr-2 is deleted, cellular opioid tolerance in LC neurons is abolished and recovery of MOR function from a brief desensitization treatment resembles that seen in neurons from opioid-naive mice.
MOR resensitization does not require βarr-2-dependent endocytosis
This study suggests that, unlike the mechanisms of β2 adrenoceptor regulation (Zhang et al., 1997b; Oakley et al., 1999; Gainetdinov et al., 2004), MOR resensitization in LC neurons does not require βarr-2-dependent mechanisms. When MOR endocytosis was presumably impaired by deletion of βarr-2, disruption of GRK2 function, or disruption of dynamin function, MOR resensitization was similarly accelerated rather than inhibited. Furthermore, intracellular application of a phosphatase inhibitor in the absence of βarrestin-dependent receptor trafficking (GRK2 inhibition in βarr-2 k.o.) prevented full recovery of MOR function, suggesting that MOR dephosphorylation is required for MOR resensitization and that this can occur independently of βarr-2-dependent MOR trafficking and endocytosis. The ability of MOR to dephosphorylate and resensitize rapidly when GRK2 and βarr-2 processes are disrupted is consistent with the observation that blocking endocytosis with concanavalin A in cultured LC neurons did not prevent MOR resensitization (Arttamangkul et al., 2006). Together with the latter, the present findings challenge the notion that βarr-2-dependent internalization is required for MOR dephosphorylation and resensitization.
MOR activates GIRK via a membrane delimited interaction with Gβγ subunits released from heterotrimeric G-protein activation, providing a direct and rapid measurement of receptor function in intact cells (Connor et al., 2004). Comparison of changes in GIRK currents evoked by a submaximal concentration of ME (300 nm) applied before and after treatment with supramaximal ME (30 μm) provides a sensitive measure of receptor desensitization and resensitization (Connor et al., 2004). In this and other studies in LC and DRG neurons (Walwyn et al., 2007; Dang et al., 2009), the basal efficacy of coupling between MOR and Gβγ subunit modulated ion channels was slightly reduced in βarr-2 k.o. mice. This contrasts with original studies of GTPγS binding assays in membranes from several regions (Bohn et al., 1999, 2000, 2002) that found enhanced efficacy of MOR GTPγS binding in βarr-2 k.o. mice. The basis for the difference is unknown. It could involve different neuronal populations as suggested by behavioral studies (Raehal et al., 2005), signaling in an intact, native neuronal environment versus membranes, or perhaps different forms of regulation that might develop during protracted GTPγS assays [2 h (Connor et al., 2004)].
When MOR resensitization was assessed in neurons from opioid-naive w.t. mice, the rate of resensitization was similar to that reported previously for LC neurons from rat (Osborne and Williams, 1995) and similar to the rate of MOR recycling reported in cultured cells (Tanowitz and von Zastrow, 2003). This suggests that the slow resensitization observed in neurons from naive w.t. mice reflects the rate of recycling of endocytosed MOR back to the plasma membrane. However, when βarr-2-dependent MOR endocytosis is disrupted, MOR resensitization was greatly facilitated. Thus, under conditions in which endocytosis is impaired by inhibition of GRK2, deletion of βarr-2, or inhibition of dynamin, the increased rate of MOR resensitization presumably reflects the absence of intracellular trafficking of endocytosed receptors as determined for the β2 adrenoceptor (Oakley et al., 2000).
Some studies have implied that MOR recycling is a necessary step for MOR resensitization (Koch et al., 1998, 2001; Law et al., 2000; Qiu et al., 2003). Treatment of cells with monensin to inhibit endosomal recycling (Law et al., 2000; Qiu et al., 2003), truncated MOR mutants (Qiu et al., 2003), or MOR splice variants (Koch et al., 1998, 2001; Tanowitz et al., 2008) all reduced the recycling and resensitization of endocytosed MOR in parallel. Although the present findings show that MOR resensitization occurs efficiently when arrestin/dynamin-dependent mechanisms are blocked, the results remain compatible with the fact that, once receptors are endocytosed, relatively slow receptor recycling (Tanowitz et al., 2008) is necessary for recovery of MOR localization and signaling at the surface membrane.
Biochemical studies implicating a necessity for MOR recycling in resensitization (Koch et al., 1998, 2001; Law et al., 2000) were generally performed over timescales greatly exceeding acute desensitization of G-protein βγ subunit mobilization as measured by GIRK activation (see above), βarr-2 binding (Oakley et al., 2000), and endocytosis (Tanowitz and von Zastrow, 2003; Arttamangkul et al., 2006; Yu et al., 2009). Therefore, such methods cannot readily distinguish recovery of functional MOR at the cell surface from increased MOR surface density (and therefore function) resulting from recycling (Connor et al., 2004). In contrast, the findings that monensin also reduces the rate of MOR resensitization in LC neurons (Dang and Williams, 2004), together with concanavalin A, which blocks MOR endocytosis but not resensitization (Arttamangkul et al., 2006), are consistent with the current finding that βarr-2, dynamin-dependent mechanisms slow MOR resensitization.
The process of MOR endocytosis involves sequential GRK phosphorylation of agonist bound receptor, binding of arrestins (particularly βarr-2), and dynamin-dependent receptor endocytosis (Kovoor et al., 1998; Gainetdinov et al., 2004). Given that LC neurons express βarr-1 (Fan et al., 2003), it is possible that neurons from βarr-2 k.o. mice can mediate MOR endocytosis via a βarr-1-dependent mechanism (Oakley et al., 2000). When DYNi was loaded into LC neurons from βarr-2 k.o. mice, MOR resensitization was further enhanced, suggesting that desensitization may have caused some MOR endocytosis in βarr-2 k.o. mice, possibly mediated by βarr-1. It is also possible that caveolar-mediated MOR endocytosis facilitates MOR resensitization in the absence of βarr-2-dependent process. This is unlikely because dynamin inhibition, which accelerated MOR resensitization, blocks clathrin-dependent and perhaps caveolar-mediated endocytosis (Doherty and McMahon, 2009).
βarr-2 trafficking is required for opioid tolerance in LC neurons
The finding that morphine treatment failed to produce cellular tolerance in LC neurons from βarr-2 k.o. mice is consistent with the findings that tolerance to DAMGO-stimulated GTPγS binding in brainstem and spinal cord membranes is attenuated in these animals (Bohn et al., 2000, 2002). This was attributed to ablation of βarr-2-dependent desensitization of MOR (Bohn et al., 1999, 2000), which was not found here in LC (Dang et al., 2009) or DRG (Walwyn et al., 2007) neurons, but the term “desensitization” did not refer to the same MOR regulatory process because rapid desensitization during sustained presence of agonist was not determined in the GTPγS assays (Bohn et al., 1999, 2000, 2002).
The results suggest that persistence of rapid recovery from desensitization after chronic morphine could contribute to the attenuation of behavioral opioid tolerance in βarr-2 k.o. mice if the mechanism found in LC generalizes to analgesia-related neurons. We propose that, after chronic morphine, βarr-2-dependent regulation of MOR is enhanced, slowing MOR resensitization, thereby shifting the equilibrium between receptor desensitization and resensitization to an accumulation of desensitized MOR that accounts for MOR tolerance. Therefore, ablation of βarr-2 in the k.o. mice facilitates resensitization and prevents cellular opioid tolerance in LC neurons. The precise mechanism of βarr-2-dependent impairment of resensitization during chronic morphine treatment in vivo is not known, but the present study shows that impaired MOR resensitization after a brief ME desensitization treatment can be rescued in LC neurons from morphine-treated w.t. mice, either by disrupting GRK2 function or inhibition of dynamin function. In light of recent reports effectively showing that morphine mediates βarr-2 recruitment to the MOR, albeit less efficiently than agonists such as DAMGO (McPherson et al., 2010), as well as inducing MOR endocytosis in neurons (Yu et al., 2009), these findings link the impairment of MOR resensitization and cellular opioid tolerance in LC to adaptations within the process of GRK2–βarr-2–dynamin-dependent MOR regulation rather than accumulation of inactive MOR at the cell surface as was suggested from studies in cultured HEK293 cells (Koch et al., 2005).
In conclusion, this study has established that βarrestin–dynamin-dependent endocytosis is not necessary for MOR resensitization as thought previously. Resensitization proceeds efficiently when βarrestin-dependent endocytosis is disrupted. βarr-2-dependent receptor endocytosis actually slows the rate of MOR resensitization, which is presumably attributable to the slow rate of recycling of endocytosed receptors. After chronic morphine treatment, βarr-2-dependent receptor regulation can contribute to cellular tolerance by impairing MOR resensitization. Although opioid actions in LC are not clearly linked to antinociception, if the mechanisms identified here can be generalized to neurons in pain pathways, it may lead to more rational therapies to limit analgesic tolerance in vivo.
Footnotes
This work was supported by National Health and Medical Research Council of Australia Program Grant 351446 and Fellowship 511914 (M.J.C.). We thank Drs. Lefkowitz and Caron for supply of the βarrestin-2 knock-out mice.
- Correspondence should be addressed to MacDonald J. Christie, Brain and Mind Research Institute, M02G, The University of Sydney, Sydney, NSW 2006, Australia. mac.christie{at}sydney.edu.au