Pharmacological Consequence of the A118G μ Opioid Receptor Polymorphism on Morphine- and Fentanyl-mediated Modulation of Ca²⁺Channels in Humanized Mouse Sensory Neurons

• The most common single nucleotide polymorphism of the human μ opioid receptor gene, A118G, is associated with differences in opioid analgesic effects in postoperative pain

• In mouse sensory neurons expressing the 118GG gene, morphine's potency and efficacy were reduced compared with the 118AA version, mirroring clinical observations and suggesting sensory neurons as a site for this effect

COMMON targets of opiate analgesics are often μ opioid receptors (MOR). Potentially functional genetic variation is present within the human MOR gene (OPRM1). For instance, several single-nucleotide polymorphisms are present within both coding and noncoding regions.1,–,3The most common functional polymorphism occurs at position 118 in exon 1.2–3The mutation results in an exchange from asparagine to aspartate at position 40 (N40D) on the extracellular N-terminal domain. The asparagine residue is one of five putative glycosylation sites on the MOR. Several clinical studies have reported that the A118G polymorphism may play a role in opiate effectiveness in the treatment of acute and chronic pain,4,–,13susceptibility to drug addiction,14,–,18stress,19and suicide.20Findings from in vitro  studies have suggested that this polymorphism affects receptor binding characteristics,21messenger RNA expression levels,22and ion channel activation.1However, under some experimental conditions this polymorphism showed no effect on function23or expression levels.24 

Recently we reported the generation of genetically modified, “humanized” mouse lines that are homozygous either for the wild-type (118AA) or variant (118GG) human OPRM1 allele and express the respective MOR.25It is known that mesolimbic dopamine neurons are under tonic γ aminobutyric acid-ergic inhibition that can be lifted through activation of MOR on γ aminobutyric acid-ergic terminals in the ventral tegmental area. It is further known that alcohol administration results in release of endogenous opioids in the ventral tegmental area, and through the cascade outlined ultimately in striatal dopamine release. Using this as a functional readout, and employing brain microdialysis, we found that 118GG mice exhibited a fourfold greater peak dopamine response following an alcohol challenge, demonstrating that A118G variation is functional in this animal model.25This study, however, left unresolved the molecular mechanism mediating the functional role of the A118G polymorphism, since binding of classic μ-agonists (β-endorphin, DAMGO) as well as signaling induced by these were unaffected by genotype.

The purpose of the present study was to extend this investigation in order to determine the pharmacological profile of two clinically relevant MOR ligands not examined in the original study, morphine and fentanyl, on the MOR-stimulated coupling to voltage-gated Ca²⁺channels in sensory neurons. These ion channels play a crucial role in nociceptive signaling.26Of particular relevance, studies of spontaneous nociception indicate the minor 118G allele to be a gain-of-function variant.27However, studies of morphine analgesia suggest functional consequences of the mutation to be just the opposite, with requirement of higher postoperative morphine doses.2 

Both humanized mouse lines were obtained as previously described.25Briefly, the line homozygous for the 118A allele (h/mOPRM-118AA, wild-type) was generated by replacing exon 1 of the mouse MOR gene with the corresponding human sequence. The line homozygous for the human 118G allele (h/mOPRM-118GG, mutant) was generated by employing site-directed 118A→G mutagenesis on the same construct, followed by replacement of exon 1. Except for the A-G exchange, therefore, both lines are genetically identical. The humanized mouse lines were generated on C57/BL6 background and the colony is maintained by h/m OPRM1-118AG breeding in order to generate littermates of all three genotypes: h/m-OPRM1-118AA, h/m-OPRM1-118AG, and h/m-OPRM1-118GG. The allele-frequencies in the offspring followed Mendelian distribution (24.1% h/m-OPRM1-118AA, 25.3% h/m-OPRM1-118GG, and 50.6% h/m-OPRM1-118AG) and were equally distributed with regards to sex of offspring (52.2% female, 47.8% male). Currently, the line is in the seventh generation of breeding, and no adverse phenotype has arisen during this time. Heterozygous mice were not studied so that both extremes could be compared and allow us to maintain a focused interpretation of the data.

The experiments designed to isolate single trigeminal ganglion (TG) neurons from humanized wild-type (118AA) and mutant (118GG) mice were approved by the Penn State College of Medicine Institutional Animal Care and Use Committee of Hershey, PA. The mice were first anesthetized with carbon dioxide and then decapitated with a laboratory guillotine. The TG were quickly removed and cleared of connective tissue in ice-cold Hank's balanced salt solution (Sigma-Aldrich, St. Louis, MO). Enzymatic dissociation of the ganglia was performed in modified Earle's balanced salt solution containing 0.6 mg/ml collagenase (Roche Applied Science, Indianapolis, IN), 0.4 mg/ml trypsin (Worthington Biochemical, Lakewood, NJ) and 0.1 mg/ml DNase (Sigma-Aldrich) for 40 min at 35°C in a shaking water bath. Afterward, the neurons were dispersed by vigorous shaking and centrifuged twice at 130 X g for 6 min. The isolated TG neurons were resuspended in minimal essential medium, supplemented with 10% fetal calf serum, 1% penicillin-streptomycin, and 1% glutamine (all from Invitrogen Corp., Carlsbad, CA), and plated onto 35 mm poly-L-lysine coated dishes. The dispersed neurons were then stored in a humidified atmosphere containing 5% CO²/95% air at 37°C.

Whole cell Ca²⁺channel currents were recorded employing the variant of the patch-clamp technique. The recording pipettes, fabricated from borosilicate glass (Garner Glass Co., Claremont, CA), were pulled on a P-97 micropipette puller (Sutter Instrument Co., Novato, CA). The Ca²⁺currents were acquired with an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA), equipped with an 18-bit analog-to-digital converter board (HEKA Instruments, Bellmore, NY). The currents were analog filtered at a frequency of 2 kHz (–3dB, 4-pole low-pass Bessel filter), and digitized with custom-designed S5 software developed by Stephen R. Ikeda, M.D., Ph.D. (Chief, Laboratory of Molecular Physiology, National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD).

The concentration-response relationships were determined by the sequential application of the MOR agonists at different concentrations in ascending order, and full recovery of the Ca²⁺current amplitude was required before exposure to another concentration. Furthermore, to minimize desensitization, no more than 3 concentrations of agonists were applied to each neuron. Percent Ca²⁺current inhibition was determined as [1-(peak current in the presence of drug/peak current before drug)]* 100%. The concentration-response curves were generated by pooling the results for each concentration and fitting the points to the Hill equation: I = I^MAX/{1 + (IC⁵⁰/[ligand])nH}, where I is the percentage inhibition, I^MAXis the maximum inhibition of the Ca²⁺current, IC⁵⁰is the half-inhibition concentration, [ligand] is the agonist concentration, and nH is the Hill coefficient. It should be noted that not all TG neurons tested exhibited Ca²⁺current inhibition in the presence of agonists. Overall, approximately 50–60% of neurons tested in both groups exhibited coupling between MOR and Ca²⁺channels.

The external recording solution consisted of (in mM): tetraethylammonium hydroxide 145, methanesulphonic acid 140, HEPES 10, glucose 15, CaCl²2.5, tetraethylammonium chloride 15, and TTX 0.0003. The pH was adjusted to 7.40 with tetraethylammonium hydroxide. The pipette solution contained (in mM): N-methyl-D-glucamine 110, tetraethylammonium hydroxide 25, HCl 20, EGTA 11, HEPES 10, CsCl²20, CaCl²1, MgATP 4, Na²GTP 0.3, and Na-Creatine phosphate 7. The pH of this solution was adjusted to 7.20 with tetraethylammonium hydroxide. Both stock solutions of morphine and fentanyl (both from Sigma-Aldrich) were prepared in water and diluted in the external recording solution to their final concentration. The agonists were applied to the neuron under study with a custom-designed gravity-fed microperfusion system.

Both data and statistical analyses were carried out with Igor Pro (WaveMetrics, Inc., Lake Oswego, OR) and Prism (GraphPad Software, Inc., La Jolla, CA) software packages, respectively. The acquired curved fits (using the Hill equation) for each data set were compared employing an F test with a P  value <0.05 considered statistically significant. Further, the plotted data shown in figure 1was compared employing a two-tailed, unpaired t  test with a P  value <0.05 considered statistically significant.

Mice were housed (3–5 per cage) in climate-controlled (22°± 2°C) rooms on a normal 12 h light cycle (lights on 7 AM) with food and water available ad libitum  until the day of testing. The mice were separated into individual cages the morning before testing. These procedures were approved by the Gallo Center Institutional Animal Care and Use Committee (Emeryville, CA) and were in accordance with National Institutes of Health guidelines for the Humane Care and Use of Laboratory Animals.

Antinociception was assessed using the hotplate latency test. Individual mice were placed on the hotplate, set at 52°± 0.02°C, and observed for the first behavioral signs of nociception (time to lick/flick their front and hind paws and jump from the hotplate). A cut-off time of 60 s was used to minimize tissue damage. Predosing reaction times were the average of the three readings taken approximately 15 min apart. After subcutaneous administration of morphine (3 mg/kg) or saline, antinociceptive testing was performed at the following times: 15, 30, 45, 60, 75, 90, and 180 min. Hotplate latency times were then converted to % maximal possible effect (MPE) using the following formula28:

All data are presented as the mean (± SEM). The %MPE was calculated using Microsoft Excel (Microsoft Corp., Redmond, WA) and Prism was used to calculate the area under the curve. Comparisons between experimental groups were made for the area under the %MPE versus  time curve (extent and duration of morphine antinociception; %MPE area under the curve) and performed using the Student t  test as implemented in the SigmaStat (Aspire Software International, Ashburn, VA) statistical analysis package.

In the first set of experiments, the biophysical parameters of TG neurons isolated from 118AA and 118GG mice were compared. Figure 1A shows the neuron capacitance, a measure of cell size, between both 118AA and 118GG groups of neurons. The results show that the two groups were nearly identical (two-tailed, unpaired t  test, P = 0.92). Further, figure 1B shows the Ca²⁺current density of both neuron sets. The statistical comparison of this parameter showed that there were no significant differences between both groups (two-tailed, unpaired t  test, P = 0.32).

Figure 2shows the current voltage (I-V) relationships of acutely dissociated TG neurons from both 118AA and 118GG mice. The Ca²⁺currents were recorded employing the whole cell variant of the patch-clamp technique and elicited with a 70 ms depolarizing step to various test pulse potentials from a holding potential of −80 mV. The current amplitude was measured isochronally 10 ms from the beginning of the test pulse. The normalized I-V curves shown in figure 2, A and C were recorded from 118AA (n = 36 to 58) and 118GG (n = 42 to 56) neurons, respectively. Both curves indicate that the inward current began to activate near −40 mV and reached a peak amplitude at −15 mV. Figure 2, B and D depict superimposed Ca²⁺current traces from a representative 118AA and 118GG neuron, respectively. As a result, the morphine and fentanyl concentration-response relationships described below were measured at the test potential to −15 mV from a holding potential of −80 mV. Overall, these results indicate that acutely isolated TG sensory neurons exhibited similar physical and Ca²⁺channel characteristics.

In this set of experiments, the morphine concentration-response relationship was determined. Figure 3shows the time courses of peak Ca²⁺current amplitude obtained before and during application of morphine (10 μM) in 118AA and 118GG sensory neurons, respectively. The Ca²⁺currents were evoked every 10 s with a 50 ms test pulse to −15 mV from a holding potential of −80 mV (shown in fig. 3B). The peak current amplitude was measured isochronally 10 ms after the initiation of the test pulse. Figure 3B shows the numbered Ca²⁺current traces evoked with this paradigm and correspond to those plotted in figure 3A. Exposure of the neuron to morphine (trace 2) resulted in approximately 45% inhibition of the Ca²⁺current. Removal of morphine resulted in a rapid recovery of the current amplitude. The time course of a 118GG neuron in figure 3C shows the effect of morphine on the peak Ca²⁺current and the corresponding numbered traces are shown in figure 3D. The application of morphine to the neuron led to a 27% inhibition of the current, and the time of recovery after agonist removal was fast.

The morphine concentration-response curves for 118AA and 118GG sensory neurons are plotted in figure 3E. The data points in both groups were fit to the Hill equation. The IC⁵⁰(nM), maximum current inhibition (%, ± SEM) and Hill coefficient (± SEM) for 118AA neurons were: 47 [19 to 115], 53 ± 2.4 [47 to 59], and 0.70 ± 0.16 [0.29 to 1.10] (95% confidence intervals are in brackets); and 260 [54 to 1250], 39 ± 2.6 [26 to 52], and 0.53 ± 0.22 [-0.004 to 1.06] for 118GG neurons, respectively. The plots show that in 118AA sensory neurons, morphine exhibited slightly more than a fivefold greater potency and a higher efficacy than 118GG neurons. The statistical comparison of both fits showed that they were significantly different (P < 0.0001).

In another set of experiments, a morphine concentration-response curve was obtained in TG neurons isolated from the parent C57/BL6 line (i.e. , not modified genetically). The data showed that the curves for 118AA and parent mice neurons were similar (data not shown). The IC⁵⁰(nM), maximum current inhibition (%, ± SEM), and Hill coefficient (±SEM) values were 99 [22 to 434], 50 ± 5.0 [37 to 62], and 1.23 ± 0.26 [-0.81 to 3.28], respectively (n = 4 to 19 neurons; 95% confidence intervals are in brackets). Statistical comparison of 118AA and parent wild-type neuron curve fits did not reach significance (P = 0.69). Thus, the TG neurons isolated from homozygous AA mice and the parent colony did not show detectable differences in the morphine-mediated Ca²⁺current inhibition.

In the next series of experiments, we examined whether the fast-acting opioid agonist fentanyl would exhibit a similar pharmacological profile with regard to Ca²⁺channel inhibition. Figure 4shows the time course of peak Ca²⁺current inhibition in 118AA (fig. 4A) and 118GG (fig. 4C) neurons before and during fentanyl (10 μM) application. Ca²⁺channel currents were evoked with a 50 ms test pulse to −15 mV from a holding potential of −80 mV applied every 10 s. The traces shown in figure 4, B and D correspond to the numbers for 118AA and 118GG neurons, respectively. As described for morphine in the “Time Course of Morphine Block and Concentration-response Relationships” section (see results in fig. 3) in 118AA neurons, exposure to fentanyl led to a 52% (fig. 4A) and 61% (fig. 4C) inhibition of Ca²⁺currents. It can be seen that following fentanyl removal the period of recovery was longer when compared with that observed with morphine (fig. 3, A and C). Figure 4E is a plot that compares the fentanyl concentration-response relationships for 118AA and118GG neurons. The data points were also fitted to the Hill equation as described above (see Materials and Methods). It can be seen that the curves for both group of neurons were nearly identical. The IC⁵⁰(nM), maximum current inhibition (%, ± SEM), and Hill coefficient (± SEM) for 118AA neurons were: 62, 52 ± 15.1 [4 to 100] and 0.55 ± 0.66 [-1.56 to 2.65] for 118AA neurons, respectively; whereas in 118GG neurons these values were: 102, 79 ± 89 [-204 to 363] and 0.31 ± 0.77 [−2.15 to 2.77]. The 95% confidence intervals are shown in brackets. Comparison of both curve fits did not show a significant difference (P = 0.63).

Because of the significant pharmacological shift of the morphine concentration responses in 118GG relative to 118AA neurons, we next measured the antinociceptive response to morphine in both humanized mouse lines. Morphine (3 mg/kg) was administered subcutaneously and the change in response latency was measured and converted to %MPE (fig. 5A) as described above (see Materials and Methods). In addition, the %MPE area under the curve was calculated to determine the extent and duration of morphine antinociception for both genotypes. The data depicted in figure 5B show that 118AA mice had a significantly (P < 0.05) greater %MPE area under the curve when compared with 118GG. These results suggest that the 118GG-carrying mice exhibited an impaired analgesic response with morphine.

Recently, we described the generation of two humanized mouse lines in which the native MOR's exon 1 was replaced with human exon 1 that encompasses the polymorphism.25Our initial study presented evidence for the 118G allele to be a gain-of-function variant with regard to dopamine response to an alcohol challenge, but did not address whether it moderates morphine responses relevant to analgesia. Thus, the present study examined the pharmacological profile of MOR in sensory neurons isolated from homozygous mice expressing either the 118AA or 118GG allele. The main finding was a differential response of 118GG sensory neurons to morphine when compared with 118AA neurons. Neurons expressing the variant allele exhibited a five-fold decrease of the morphine-mediated Ca²⁺current inhibition potency (IC⁵⁰) as well as a 26% decrease in efficacy. Furthermore, when both groups of mice were tested for morphine requirement on a hotplate assay, the 118GG homozygous mice displayed a diminished response for up to 3 h when compared with 118AA mice. These behavioral assays strongly support our electrophysiological observations, which suggest that the mutant MOR exhibits an altered morphine pharmacological profile. These results are in agreement with several clinical studies that have shown that carriers of the mutant allele require an increase of the dosing requirements to achieve analgesia or adequate pain control.2–3,29 

It is known that, unlike most opiates (i.e. , fentanyl, etorphine, DAMGO, and β-endorphin), morphine exhibits a unique pharmacological profile with a propensity to promote slower and incomplete receptor internalization, with a typical exposure time greater than 5 min.30In our electrophysiological experiments, exposure of the cells to morphine was generally less than 40 s (fig. 3, A and C). Hence, the observed difference is not likely a result of significant altered receptor internalization rates. Nevertheless, our data suggest that the physiologic response of the opioid receptor variants is ligand-dependent and that morphine's “recognition” of the mutant receptor diminishes its analgesic effectiveness.

Given the heterogeneous nature of sensory neurons, the electrophysiological experiments were performed in cells with comparable biophysical characteristics, including cell size, Ca²⁺channel current density, and peak current potential. Thus, it is unlikely that the effects observed with morphine were a result of employing different cell subpopulations. In addition, the electrophysiological data obtained from sensory neurons isolated from the parent mice were not significantly different from the 118AA group. Differences in MOR density between 118AA and 118GG neurons are also unlikely to account for our findings. Binding assays, which measured receptor density in several brain regions, showed no significant differences in our mouse model.25Moreover a recent study, also performed in postmortem human brain areas involved in nociception, found that receptor density of MOR was not different between 118AA and 118GG carriers.24However, another report found lower MOR messenger RNA in postmortem human brain tissue from 118G individuals compared with both heterozygous or 118AA subjects.22In this study, however, protein levels were not assessed.

The A118G single nucleotide polymorphism has been reported to be associated with elevating27or decreasing11,31the pain threshold of the 118G allele carriers. Healthy volunteers carrying the variant allele, for example, have been shown to have higher pain (i.e. , pressure) threshold when compared with homozygous 118A subjects.27On the other hand, it has been reported that the morphine requirement for postcesarean analgesia and pain scores in patients homozygous for the 118A allele were significantly lower than carriers of the 118G allele (either 118AG or 118GG).11This conflicting effect of the polymorphism on pain threshold may be a result of an endogenous opioid that exerts opposite effects on the variant receptors (i.e. , higher receptor affinity) than that observed with morphine. For instance, a recent study in healthy volunteers found that carriers of the 118G allele displayed an enhanced cortisol response to naloxone but a dampened response to psychosocial stress compared with 118A carriers.32The authors suggested that with a higher MOR affinity, 118G allele carriers have a greater inhibitory opioidergic tone on corticotrophin-releasing hormone neurons in the arcuate nucleus. As a result, exposure to naloxone led to the enhanced cortisol response.32Thus, the blunted response to psychologic stress may result from a higher affinity of the mutant receptor, for this native opioid ligand brings about a differential response from the hypothalamic-pituitary-adrenal axis. It is, therefore, tempting to speculate that under normal conditions, the 118G allele imparts a “protective” effect on carriers with a higher pain threshold. However, once that threshold is reached, 118G carriers will require higher amounts of opiates to achieve adequate pain control33that may, unfortunately, not be obtained. As a result, the likelihood of addiction is increased and detrimental to this patient population.

Unlike the effects described with morphine above, the fentanyl pharmacological profile of 118AA and 118GG sensory neurons was similar. These results differ from two clinical studies that have reported that 118G allele carriers required more12or less5fentanyl for anesthesia than 118A carriers. It is unclear why fentanyl-mediated Ca²⁺channel inhibition would be comparable in both types of sensory neurons in our mouse model and exert a different effect in the clinical setting.5,27Our model system consists of acutely dissociated neurons, and thus, the influence of other neurons (either pre- or postsynaptically) or glial cells is removed. Further, under our recording conditions, the temporal resolution is in the ms to s range and the exposure of the neurons to the receptor agonists is limited. The conditions observed in vivo  are significantly more complex. In the clinical setting, for instance, there are varied routes of administration; morphine metabolism that is influenced by gender, age, disease state; and different rates of elimination.

In addition, we previously found that the DAMGO and β-endorphin pharmacology were also similar in sensory neurons between 118AA and 118GG allele carriers.25These findings differed from a study that reported a threefold increase in potency by β-endorphin with regard to activating G protein-gated inwardly rectifying K⁺channels in frog oocytes expressing 118G MOR when compared with 118A receptors.1The difference between morphine and these three other opioid receptor agonists may be because of differences in the physical binding to the receptors (see Discussion); in the signaling components (i.e. , heterotrimeric G proteins) that couple each MOR subtype and effectors (i.e. , ion channels and enzymes); or variations between each model system employed (i.e. , native vs.  heterologous expression levels).

Our humanized mouse line model differs from that of Blendy et al. , who first reported a murine model of the A118G polymorphism.34In that model, the equivalent nucleotide substitution was performed at position 112 (A112G) of the MOR gene. The result was a similar amino acid substitution (N38D) at the N-terminus and the loss of the putative glycosylation site. However, whereas the receptor polymorphism in humans results in the loss of 1 of 5 putative glycosylation sites (our mouse model), the polymorphism in Blendy et al. 's model removes 1 of 4 glycosylation sites. Nevertheless, our electrophysiological and behavioral studies with morphine are in agreement with their observations of the morphine-mediated antinociception with the hot-plate assay. In a cumulative dosing paradigm, they found that 112GG carriers exhibited a significantly lower maximal possible effect of morphine.34These results suggest that morphine's effects in mutant receptor carriers will be manifested with either three or four glycosylation sites. The loss of the glycosylation site may be significant in decreasing the opioid receptor's ability to dimerize with other opioid receptors.35For example, the loss of the glycosylation site in β1-adrenergic receptors has been reported to markedly decrease both the ability of the receptor to dimerize and reach the cell surface.36 

Previous in vitro  studies that examined how the A118G polymorphism affected signaling mechanisms and receptor function relied on the heterologous expression of the receptors. Some of the reported observations indicated that the results were affected by cell type and the employed transfection method.17,19,37The advantage of the humanized mouse model used in our study is that the underlying mechanisms of the opiate intervariability observed clinically can be investigated directly in cells from tissue that is involved in pain transmission.

In summary, this is the first electrophysiological study to employ an animal model and examine the functional consequence of the A118G polymorphism in sensory neurons with regard to opioid-mediated Ca²⁺current inhibition. The experiments were performed in neurons that had comparable biophysical parameters. The data for the morphine pharmacological profile of the 118GG carriers illustrated a diminished potency and efficacy resembling the clinical, behavioral, and in vitro  observations. On the other hand, no significant pharmacological differences between both groups of neurons were observed when fentanyl was employed. From the data we conclude that the use of the humanized mouse line will be a valuable tool to further explore the signal transduction pathways that mediate the changes in morphine pharmacology in 118GG patients.