Dual allosteric modulation of opioid antinociceptive potency by a_2A-adrenoceptors

Abstract

Opioid and α₂-adrenoceptor (AR) agonists are analgesic when administered in the spinal cord and show a clinically beneficial synergistic interaction when co-administered. However, α₂-AR antagonists can also inhibit opioid antinociception, suggesting a complex interaction between the two systems. The α_2A-AR subtype is necessary for spinal adrenergic analgesia and synergy with opioids for most agonist combinations. Therefore, we investigated whether spinal opioid antinociception and opioid-adrenergic synergy were under allosteric control of the α_2A-AR. Drugs were administered intrathecally in wild type (WT) and α_2A-knock-out (KO) mice and antinociception was measured using hot water tail immersion or substance P behavioral assays. The α_2A-AR agonist clonidine was less effective in α_2A-KO mice in both assays. The absence of the α_2A-AR resulted in 10–70-fold increases in the antinociceptive potency of the opioid agonists morphine and DeltII. In contrast, neither morphine nor DeltII synergized with clonidine in α_2AKO mice, indicating that the α_2AAR has both positive and negative modulatory effects on opioid antinociception. Depletion of descending adrenergic terminals with 6-OHDA resulted in a significant decrease in morphine efficacy in WT but not in α_2A-KO mice, suggesting that endogenous norepinephrine acts through the α_2A-AR to facilitate morphine antinociception. Based on these findings, we propose a model whereby ligand-occupied versus ligand-free α_2A-AR produce distinct patterns of modulation of opioid receptor activation. In this model, agonist-occupied α_2A-ARs potentiate opioid analgesia, while non-occupied α_2A-ARs inhibit opioid analgesia. Exploiting such interactions between the two receptors could lead to the development of better pharmacological treatments for pain management.

1. Introduction

1.1 Adrenergic modulation of nociception

The adrenergic system is a critical modulator of nociception, including at the spinal level where nociceptive information is relayed from the periphery to the central nervous system. The activation of spinal adrenoceptors by norepinephrine (NE) released from descending noradrenergic fibers or by exogenously administered α₂-adrenoceptor (α₂-AR) agonists generates an antinociceptive effect (Eisenach et al., 1996). These receptors are located both pre- and post-synaptically on sensory neurons and in the spinal cord where they inhibit neurotransmitter release and hyperpolarize secondary dorsal horn neurons (Fairbanks et al., 2009). The α-adrenergic receptor family is divided into α₁- and α₂-adrenoceptors, and there are three α₂ receptor subtypes: α_2A, α_2B and α_2C. Poor subtype selectivity among α₂-AR ligands makes it difficult to distinguish the functions of each subtype; genetically modified mice therefore play a critical role in elucidating the relative roles of each subtype.

The α_2A-AR is the predominant mediator of the antinociceptive effects of α₂-adrenoceptor agonists at the spinal level, although the α_2C-AR is also involved depending on the ligand tested (Fairbanks et al., 2002; Stone et al., 1997). Elsewhere in the central nervous system, the α_2A-AR also mediates the anti-hypertensive and sedative effects of α₂-AR-selective agonists such as clonidine and dexmedetomidine, limiting their use as analgesic agents (Lakhlani et al., 1997; MacMillan et al., 1996). Thus, spinal administration of α₂-AR agonists is a useful strategy to minimize such unwanted side effects.

1.2 Opioid-adrenergic interactions in pain and antinociception

Activation of spinal µ- and δ-opioid receptors, MOPr and DOPr, respectively, have the potential to mediate analgesia. Interactions between opioid and adrenergic systems in the spinal cord are well documented. Of clinical importance is the interaction between opioid and α₂-AR agonists such as morphine and clonidine: while co-administration results in analgesic synergy, the interaction remains additive for sedative and cardiovascular side effects, thus widening the therapeutic window between desired and undesired effects (Stone et al., 2014). The identification of receptor subtypes mediating such synergistic interactions is critical for understanding these mechanisms and for development of novel synergy-based therapies. MOPr, DOPr, α_2A- and α_2C-AR have been shown to mediate opioid-adrenergic synergy depending on the opioid-α₂-AR agonist combination used (Chabot-Dore et al., 2014). Morphine/clonidine synergy - one of the most clinically-relevant and widely studied combinations – requires the activation of MOPr (Chabot-Doré et al., 2013; Guo et al., 2003), but the α₂-AR subtype necessary for this interaction has not been clearly identified. Morphine also interacts synergistically with spinally administered NE (Roerig et al., 1992), suggesting that the endogenous adrenergic system regulates opioid-mediated antinociception, but again the α₂-AR subtype(s) involved remain(s) unknown. Inhibition of morphine antinociception by α₂-AR antagonists has also been observed; the regulation of opioid antinociception by adrenoceptors is therefore bidirectional and complex (Browning et al., 1982; Iglesias et al., 1992; Morales et al., 2001; Stone et al., 1997).

1.3 Intra-receptor allostery potentially regulates opioid-adrenergic interactions

Opioid receptors and adrenoceptors belong to the G protein-coupled receptor (GPCR) superfamily of transmembrane receptors and both preferentially couple to Gα_i/o. GPCRs can exist as homo- and hetero-oligomers in cells and oligomerization can affect many aspects of GPCR function, including their maturation during biosynthesis, surface expression, ligand binding, signal transduction and internalization, all of which may have important pharmacological implications (Terrillon and Bouvier, 2004). These functional and pharmacological effects can result from allosteric interactions between receptors in an oligomer and, in some cases, only require the mere presence of the allosteric partner. For example, activation of the dopamine receptor D2 (DRD2) produces an anorexigenic effect that requires the presence, but not the activation of the ghrelin receptor GHSR1a (Kern et al., 2012).

In order for allosteric interactions to be relevant in vivo, receptors of interest must be co-expressed and physically interact in the tissue or cell of interest (Kenakin et al., 2010). An extensive body of functional and anatomical evidence points towards the co-localization of α_2A-AR and MOPr or DOPr receptors in the spinal cord (see Chabot-Doré et al. (2014) for a detailed review) and MOPr and α_2A-AR have been shown to heterodimerize in brainstem nuclei (Sun et al., 2015). DOPr and α_2A-AR co-localize in the substance P (SP)-expressing DRG neuron subpopulation and interact both physically and functionally when co-expressed in vitro (Riedl et al., 2009; Rios et al., 2004). Similarly, MOPr and α_2A-AR are expressed in DRG neurons (Riedl et al., 2009; Stone et al., 1998; Wang et al., 2010) and co-localize extensively in cultured DRG neurons (Tan et al., 2009). Furthermore, these receptors form dimers in heterologous expression systems as detected using bioluminescence resonance energy transfer (BRET; (Jordan et al., 2003), fluorescence resonance energy transfer (FRET; (Vilardaga et al., 2008), or co-immunoprecipitation (Jordan et al., 2003; Zhang and Limbird, 2004). Cross-desensitization between morphine and clonidine and cross-internalization of MOPr and α_2A-AR following agonist treatment in cultured DRG neurons further support the capacity of these receptors to interact (Tan et al., 2009). Moreover, it was shown in expression systems that the morphine-bound MOPr allosterically inhibits the α_2A-AR (Vilardaga et al., 2008). Therefore, α_2A-AR/MOPr and α_2A-AR/DOPr hetero-oligomers likely form in the spinal cord, making this neuronal population a good substrate to test whether the α_2A-AR allosterically regulate opioid antinociception.

1.4 Study aims

The objective of this study was to determine if the α_2A-adrenoceptor allosterically regulates spinal opioid antinociception and opioid-adrenergic synergy. The antinociceptive efficacy of spinal morphine, the opioid subtype-selective agonists [d-Ala²]-deltorphin II (DeltII) and [dAla², NMe-Phe⁴, Gly-ol⁵]-enkephalin (DAMGO), and the interaction between morphine/clonidine and DeltII/clonidine were therefore determined in the presence and absence of the α_2A-AR in mice in vivo. We further investigated the role of α_2A-AR in the endogenous regulation of morphine antinociception by NE. Our findings lead us to propose a mechanism for the distinct patterns of allosteric regulation of opioid receptor-mediated antinociception by ligand-occupied versus ligand-free α_2A-adrenoceptors.

2. Materials and Methods

2.1 Animals

All procedures were approved by the Animal Care Committee at McGill University and conformed to the ethical guidelines of the Canadian Council on Animal Care. All efforts were made to minimize animal suffering, to reduce the number of animals used, and to utilize alternatives to in vivo techniques, if available. We used C57BL/6 mice (Charles River, Quebec, Canada) as wild-type (WT) controls. Mice with a targeted gene deletion introducing a premature stop codon in the Adra2a gene were developed on a mixed C57BL/6 × FVB/129 background (Altman et al., 1999). Congenic α_2A-KO mice backcrossed to a standard C57BL/6 background were obtained from Jackson Laboratory (Bar Harbor, Maine, stock #004367). Mice were bred in house and genotyping was performed to assign Adra2a genotypes in WT and α_2A-KO mice as described (Peterhoff et al., 2003). Mice were housed in groups of 2–5 and maintained in ventilated cages on a regular 12 hour light/dark cycle and given ad libitum access to food and filtered water. Aged-matched 3–6 month old males were used in all studies and experimenters were blind to both genotype and treatment. Animals were re-randomized to control for the effect of previous drug exposure and re-tested up to 3 times with a minimum drug washout period of 4 days between experiments. The re-randomization across groups minimizes the potential that previous drug exposure confounds subsequent results. The potential effects of repeated handling were analyzed in the vehicle-treated mice and no effects of previous handling were observed in either the SP or the tail immersion assays.

Experiments with α_2C-KO mice and their controls were approved by the Institutional Animal Care and Use Committee of the University of Minnesota. α_2C-KO mice were developed at Stanford University, Palo Alto, California (Link et al., 1995) and purchased from Jackson Labs (Bar Harbor, Maine, stock # 002512) after 17 generations of backcrossing to C57BL/6 background. α_2C-KO and C57BL/6 mice WT controls were bred and housed in University of Minnesota, Minneapolis where experiments were performed.

Some of the experiments performed in WT mice have been previously published as part of another study (Chabot-Doré et al., 2013). Three stains of mice were tested in parallel – WT, DOPr-KO and α_2A-KO mice – so that they could share a WT control group, thus reducing animal use and maximizing resources. As a result, the WT data from Figures 2A–D (2^nd row), Figure 3D,E and Figure 4A,B and the α_2A-KO clonidine dose-response curve in the substance P assay in Figure 2A were previously published (Chabot-Doré et al., 2013). These data are included here to provide the necessary WT control data needed to put the current α_2A-KO experiments in context.

Figure 2. Dose-response analysis of spinal clonidine-, morphine-, DeltII- and DAMGO-mediated antinociception in WT and α_2A-KO mice.

Dose-response curves were constructed for each drug administered i.t. in the hot water tail immersion assay (upper row) and the SP behavioral assay (middle row). A) Clonidine was efficacious in the tail immersion assay and the SP behavioral assay in WT mice but clonidine potency and efficacy were reduced in α_2A-KO mice in both assays. B) Morphine was more potent in α_2A-KO mice than in WT mice in both assays. C) DeltII was more potent in α_2A-KO mice than in WT mice in both assays D) DAMGO was more potent in α_2A-KO mice in the tail immersion assay but not in the SP behavioral assay. E) Clonidine was equally efficacious and potent in WT and α_2C-KO mice. F) 100 nmol of SNC 80 was more effective in α_2A-KO mice in the tail immersion assay compared to WT or vehiclecontrols. A–D) Dose-response curves generated with the SP behavioral assay in WT mice and the clonidine dose response curve in α_2A-KO mice were published in Chabot-Doré et al. (2013) as part of a study conducted in parallel with this one. DeltII and DAMGO antinociception were measured with a cumulative dosing protocol in the tail immersion ED₅₀ values are reported in Table 2. *p ≤ 0.01.

Figure 3. Synergistic interaction between clonidine and morphine in WT but not α_2A-KO mice.

Dose-response curves for spinal morphine, clonidine and their combination in WT mice in the hot water tail immersion assay (A) and the SP behavioral assay (D). B, E) Isobolographic analysis of the morphine and clonidine interaction in WT mice: the morphine ED₅₀ value with lower CI is plotted on the y axis, and the clonidine ED₅₀ value with lower CI is on the x axis. The measured experimental ED₅₀ value (●) for the drug combination was lower than the theoretical additive ED₅₀ value (○), indicating that morphine and clonidine interacted in a synergistic manner (p < 0.05). In α_2A-KO mice, since clonidine failed to reach 50% efficacy in the hot water tail immersion assay (C) or in the SP behavioral assay (F), isobolographic analysis was not performed. However, when clonidine was added to morphine in a 1:1 ratio, the ED₅₀ values for morphine and clonidine were not significantly different than morphine alone (p < 0.05), showing that adding clonidine to morphine did not change its potency (F). The calculated experimental and theoretical ED₅₀ values for the morphine and clonidine combinations are reported in Table 3. D, E were published in Chabot-Doré et al. (2013) as part of a study conducted in parallel with this one.

Figure 4. Synergistic interaction between clonidine and DeltII in WT but not α_2A-KO mice.

(A) Dose-response curves of spinal DeltII, clonidine and their combination in WT mice in the SP behavioral assay. (B) Isobolographic analysis of the DeltII and clonidine interaction in WT mice: the morphine ED₅₀ value with lower CI is plotted along the y axis, and the clonidine ED₅₀ value with lower CI along the x axis. The measured experimental ED₅₀ value (●) for the drug combination was lower than the theoretical additive ED₅₀ value (○), indicating that DeltII and clonidine interact in a synergistic manner (p < 0.05). (C) In α_2A-KO mice, since clonidine reached less than 50% efficacy isobolographic analysis was not performed. However, the doseresponse curves of DeltII and of clonidine+DeltII overlapped and ED₅₀ values were not significantly different (p < 0.05), showing that adding clonidine to DeltII did not change its potency. The calculated experimental and theoretical ED₅₀ values for the DeltII+clonidine combinations are reported in Table 3. A, B were published in Chabot-Doré et al. (2013) as part of a study conducted in parallel with this one.

2.2 Drug preparation

Morphine sulphate was purchased from Medisca Pharmaceutica (St-Laurent, QC, Canada). Clonidine HCl and [d-Ala², NMe-Phe⁴, Gly-ol⁵]-enkephalin (DAMGO) were purchased from R&D Systems (MN, USA) and were dissolved in saline. [d-Ala²]-deltorphin II (DeltII; R&D Systems) was dissolved in acidified saline (0.9% NaCl, 0.05 M acetic acid). For cumulative dose-response curve experiments evaluating DAMGO and DeltII, artificial cerebrospinal fluid (ACSF, Harvard Apparatus, MA, USA) with 0.05 M acetic acid was used. (+)-4-[(αR)-α-((2S,5R)-4-Allyl-2,5-di-methyl-1-piperazinyl)-3-methoxybenzyl]-N,N-diethyl-benzamide (SNC 80; R&D Systems) stock and working solutions were dissolved in saline with 0.3% tartaric acid. Substance P (SP) was purchased from AnaSpec (CA, USA) and concentrated stocks were dissolved in acidified saline. Working SP solutions were dissolved in saline at a total dose of 15 ng/5µl alone or mixed with other drugs. 6-hydroxydopamine (6-OHDA, Sigma) was dissolved in saline with 0.2 mg/ml ascorbic acid and administered i.t. at 20 µg/5 µl. Vehicle solutions consisted of the diluents used for the respective drug tested. Drug doses administered intrathecally are expressed as total nmol or pmol in 5µl and drug combination doses are graphed as total drug dose (i.e. the sum of each drug).

2.3 Drug administration

Intrathecal (i.t.) drug administration was done by direct lumbar puncture in a volume of 5 µl adapted from the method of Hylden and Wilcox (1980) in conscious mice. Briefly, mice were immobilized in a cotton cloth and held by the pelvic girdle while a 30 gauge 1/2 inch needle mounted on a 50 µl Luer lip syringe (Hamilton, Reno, NV) was inserted in the intervertebral space and directed forward inside the vertebral column to deliver 5 µl of drug.

2.4 Behavioral assays

Animals were acclimatized to the testing room in their home cage for 45–60 minutes and to the testing chambers for 60 minutes before testing, if applicable.

2.4.1 Tail immersion assay

Mice were immobilized in a cotton cloth and the bottom 2/3 of the tail was immersed in a 49°C hot water bath. The latency for animals to withdraw their tails from the water was recorded with a hand-held stopwatch and cutoff time was set to 12 seconds to avoid tissue damage. To obtain reliable and stable measures, mice were given at least two training sessions on separate days during which baseline measurements were taken three times. Measurements were taken before (baseline latency) and after drug administration (experimental latency) and expressed as the percent of the maximal possible effect (MPE): % MPE = (Experimental Latency – Baseline Latency × 100) / (12 sec – Baseline Latency).

2.4.2 Radiant heat paw withdrawal assay

Thermal heat threshold was measured by focusing a radiant heat beam (IITC Life Science Inc, Woodland Hills, CA) onto the center of the plantar surface of the hind paw and measuring the latency to withdraw the hind paw (Hargreaves et al., 1988). Three measurements at 60 minute intervals were used to calculate an average. A 17 second cutoff was set to prevent tissue damage.

2.4.3 Mechanical sensitivity

Calibrated von Frey filaments (Stoelting Co., IL, USA) were applied on the plantar surface of the hind paw for 4 seconds or until paw withdrawal, and the 50% threshold to withdraw (grams) was calculated according to the up-down method (Chaplan et al., 1994) and as described by Millecamps et al. (2011).

2.4.4 Substance P behavioral assay

The antinociceptive action of single drugs and their combinations was tested in the SP behavioral assay developed by Hylden and Wilcox (1981). Briefly, 15 ng of SP was administered i.t. alone (control) or co-administered with a single drug or a drug combination (experimental) in a volume of 5µl. The number of caudally directed biting, licking and scratching behaviors were counted for 1 min and results are expressed as the percent inhibition of SP-induced behaviors: % Inhibition = (Control – Experimental) × 100/Control.

Control values represent the average number of SP-induced behaviors elicited by 15 ng of SP collected from 2–3 mice of each strain on the different experimental days (WT: n = 13; α_2A-KO: n =16).

2.4.5 Analysis of baseline sensory thresholds

Comparison of raw baseline values (means ± SEM) obtained in behavioral assays between WT and α_2A-KO were compared with a two-sided unpaired parametric T-test with a 95% confidence level (GraphPad Software, Inc.).

2.5 In vivo pharmacology

2.5.1 Dose-response curves of morphine, clonidine and their combinations

Mice were treated with single drug doses administered i.t. and measurements were taken with the SP behavioral assay or 10 minutes post injection with the tail immersion assay. Morphine or DeltII were combined with clonidine in 5 µl at an equi-effective 1:1 ratio and are expressed as total drug dose. Each dose represents a group of 4–12 animals.

2.5.2 Cumulative dose-response curves of DAMGO and DeltII

A total of four consecutive injections were performed with increasing doses of the drug tested. For DAMGO, we injected 0.003, 0.07, 0.3 and 1 nmol i.t. and for DeltII, we injected 1, 3, 10 and 20 nmol i.t. Mice were tested 3 min after i.t. injection and re-injected 2 minutes later, allowing for a total of 5 minutes between each injection. Vehicle-treated mice were included to confirm that the repeated injections did not change the tail flick latencies (data not shown). Initial experiments indicated that these peptide agonists rapidly lose their efficacy following i.t. injection with a return to baseline at 30 minutes. The same group of WT (n = 12) and α_2A-KO (n = 12) mice was used to test both DAMGO and DeltII by allowing a washout period of one week between experiments.

2.5.3 Morphine antinociception in 6-OHDA-lesioned mice

Pre-lesion baseline tail flick latencies were measured in groups of 7–10 WT or α_2A-KO mice before injecting them i.t. with 20 µg of 6-OHDA or vehicle solution. 3 days post-6-OHDA administration, baseline tail flick latencies were measured and full morphine dose-response curves were obtained using a cumulative drug administration protocol with a maximum of five consecutive injections of increasing morphine doses. Morphine doses were injected at 5 minutes intervals and tail flick latencies were measured at 4 minutes post injection. Because of the potency difference between WT and α_2A-KO mice, we used a different range of morphine doses to construct the cumulative dose-response curve in each strain: in WT mice, we used 1, 3, 6, 10 and 30 nmol i.t.; in α_2A-KO mice, we used 0.1, 0.3, 1, 3 and 6 nmol i.t.

2.5.4 Dose-response and isobolographic analysis

Dose-response curve graphs were generated with GraphPad Prism 6.0. A minimum of 4 animals were used per dose and each dose point is expressed as the mean % MPE or % inhibition ± S.E.M. ED₅₀ values and 95% confidence interval (CI) were calculated using a minimum of three doses in the linear portion of each dose-response curve following the method of Tallarida and Murray (1987) with the FlashCalc 4.5.3 pharmacological statistics software package generously supplied by Dr. Michael Ossipov. Statistical comparisons of potencies are based on the confidence limits of the ED₅₀ values.

Isobolographic analysis is the method of choice for evaluating drug interactions (Tallarida, 2006). As a pre-requisite for this analysis, each drug alone or in combination must be at least 50% efficacious. Drug combination ratios were chosen according to the relative potency of each drug by determining an approximately equally effective potency ratio between the agonists based on their respective ED₅₀ values. To test for interactions between agonists, the ED₅₀ values and S.E.M. of all dose-response curves were arithmetically arranged around the ED₅₀ value using equation (ln(10) × ED₅₀) × (S.E. of log ED₅₀) (Tallarida, 1992). This manipulation is required to perform an isobolographic analysis to evaluate if an interaction is synergistic, additive or sub-additive (Tallarida, 1992). When testing an interaction between two drugs, a theoretical additive ED₅₀ value is calculated for the combination based on the dose-response curves of each drug administered separately. This theoretical value is then compared by t test with the observed experimental ED₅₀ value of the combination. An interaction is considered synergistic if the experimental ED₅₀ is significantly less (P < 0.05) than the calculated theoretical additive ED₅₀.

Visualization of drug interactions can be facilitated by the use of an isobologram, i.e. a graphical representation of isobolographic analysis. The isobologram depicts the ED₅₀ value of each drug as the x-y-intercept. The line connecting these two points depicts the dose combination expected to yield 50% efficacy if the interaction is purely additive and is called the theoretical additive line. The theoretical additive ED₅₀ is determined mathematically and plotted on this line with its CI spanning perpendicularly from the line. The experimental ED₅₀ for the combination is plotted at the corresponding x,y co-ordinates along with its 95% confidence interval for comparison with the theoretical additive ED₅₀ value. All dose-response and isobolographic analyses were performed with FlashCalc 4.5.3.

2.5.4 SNC80 antinociception

WT and α_2A-KO mice were tested in the hot water tail flick assay (n=4–7). We measured the effect of 100 nmol of SNC80 or vehicle (i.t.) at 3 minutes post-injection since our pilot experiments determined that SNC80 effect is maximal at 3 minutes and returned to baseline values at 10 minutes (data not shown). Comparison between SNC80 and vehicle treatment in WT and α_2A-KO was done with a 2-way ANOVA followed by a Bonferroni multiple comparison T test with GraphPad Prism 6.0.

2.6 Tissue collection

Mice were deeply anesthetized with isofluorane and decapitated. Spinal cords were ejected through the rostral end of the vertebral canal by injecting PBS from the caudal end. DRGs were dissected out under a stereomicroscope. Tissues were frozen on dry ice and stored at −80°C.

2.7 Quantitative gene expression analysis of opioid receptors

Total RNA from spinal cords and DRGs was extracted and DNase digested (RNeasy Lipid Tissue Mini Kit, Quiagen). Conversion of mRNA to cDNA and subsequent quantification of DOPr and MOPr RNA by real time PCR using SYBR-Green was performed at the Genome Québec RNomic Center (Sherbrooke, QC, Canada). Primer sequences for opioid receptor genes (Oprm1, Oprd1) and housekeeping genes (Gapdh, Hptr1, Ubc) are shown in Table 1. Four mice of each strain were used and amplification reactions were performed in triplicate for each tissue sample. Relative expression and quality control evaluation was computed using the qBase method (Hellemans et al., 2007) which used three reference housekeeping genes (Gapdh, Hptr1, Ubc) to normalize expression values for each sample. Results were then normalized to the WT condition, giving it an arbitrary relative expression value of 1. Comparisons of relative gene expression between WT and α_2A-KO mice were done in GraphPad Prism 6.0 with a two-sided unpaired Mann-Whitney U test.

Table 1.

Sequence of forward (F) and reverse (R) primers used for quantitative PCR analysis of gene expression in WT and α_2A-KO mouse SC and DRG.

Gene	Primer	Sequences (5’ → 3’)
Oprm1	F R	TGGTCACAGCCATCACCATCATG CATCAGGTAGTTAACACTCTGAAAGGGCA
Oprd1	F R	GTCCCTCGCCCTAGCCATCG GCCAGATTGAAGATGTAGATGTTGGTGG
Gapdh	F R	TGACGTGCCGCCTGGAGAAA AGTGTAGCCCAAGATGCCCTTCAG
Hptr1	F R	GCTTGCTGGTGAAAAGGACCTCTCGAAG CCCTGAAGTACTCATTATAGTCAAGGGCAT
Ubc	F R	CGTCGAGCCCAGTGTTACCACCAAGAAGG CCCCCATCACACCCAAGAACAAGCACAAG

2.8 Western blot analysis

Spinal cords were collected from age-matched WT and α_2A-KO mice (n = 3 per strain) and homogenized with an automated tissue homogenizer (Precellys 24, Bertin Technologies) with 1.4 mm ceramic beads (Mo Bio Laboratories) in homogenizing buffer (25 mM Tris HCl pH 7.4, 1 mM EDTA, 2 mM MgCl₂) supplemented with proteinase inhibitor cocktail (Roche). The homogenate was centrifuged 4 min at 1000g and the supernatant containing the total protein fraction was denatured by boiling in sample buffer (TrisHCl 50 mM, 2% sodium dodecyl sulfate (SDS), 10% glycerol, 2.5% β-mercaptoethanol, bromophenol blue, pH 6.8). 40 µg of each sample was run in a 10% acrylamide SDS-PAGE and proteins were transferred onto a nitrocellulose membrane. Immunoblotting of the membranes was performed with 5% non-fat dry milk in TBST (0.1M Tris-HCl, 0.9% NaCl, 0.1% Tween 20, pH 7.4) as a blocking and antibody diluent solution. MOPr-immunoreactive (ir) bands were detected with a rabbit polyclonal antisera raised against amino acids 384–398 of the predicted MOR1 sequence (NHQLENLEAETAPLP; (Arvidsson et al., 1995)) diluted 1:2500 followed by an HRP-coupled mouse-anti-rabbit IgG (1:10 000; Jackson ImmunoResearch). Membranes were stripped and reprobed to detect glyceraldehyde-3-phosphate dehydrogenase (GAPDH)-ir band with a mouse monoclonal antisera (1:10 000; MAB374, Millipore) and an HRP-coupled goat-anti-mouse IgG (1:10 000; Jackson ImmunoResearch). Immunoreactivity was revealed by enhanced chemiluminescence and visualized by exposing membranes to a light sensitive film. Densitometry was performed using ImageJ64 analysis software (NIH) where the 67 kDa MOPr band from each mouse sample was normalized to GAPDH levels. Comparison of protein expression levels was done by comparing average normalized densitometry values between WT and α_2A-KO mice in GraphPad Prism 6.0 with a two-sided unpaired Mann-Whitney U test.

2.9 [³H]-DeltII saturation and competition binding assay

WT and α_2A-KO mouse spinal cords were homogenized with a handheld rotor stator homogenizer (TissueRuptor, Quiagen) in cold homogenization buffer (50 mM TrisHCl, 2.5 mM EDTA, pH 7.0) supplemented with protease inhibitors (Roche). The homogenate was centrifuged 10 minutes at 1000 g at 4°C and the supernatant was centrifuged at 48 000 g for 20 minutes at 4°C. The pellet was resuspended in binding buffer (50 mM TrisHCl, 3 mM MgCl₂, pH 7.4 with protease inhibitor) and incubated at 37°C for 15 min to dissociate receptor-bound endogenous opioid peptides and degrade monoamines. The membranes were re-centrifuged at 48 000 g and resuspended in binding buffer.

Binding experiments were performed with the DOP-selective radioligand deltorphinII (d-Ala²), [Tyrosyl-3,5-³H] ([³H]-DeltII, PerkinElmer, MA, USA). In saturation binding experiments, 200 µg of spinal cord membrane protein from WT or α_2A-KO mice was combined with five concentrations of [³H]-DeltII (0.3–10 nM) in a final sample volume of 300 µl. Nonspecific binding was determined by the addition of the nonselective opioid receptor antagonist naloxone (10 µM, R&D systems). In competition binding experiments, 200 µg of spinal cord membrane protein was combined with 1 nM of ³H-DeltII and the DOR-selective antagonist naltrindole (R&D Systems) at concentrations ranging from 1 µM to 100 fM. Samples were prepared in triplicate and incubated for 60 min in a 37°C water bath. Spinal cord membranes were collected by filtration on a grade GF/C glass microfiber filter (Whatman) that was previously soaked overnight in binding buffer with 0.5% polyethyleneimine and washed three times with ice-cold washing buffer (25 mM TrisHCl, pH 7.4). Filters were transferred into 5 ml polypropylene tube and filled with 3 ml of Ecolume liquid scintillation cocktail (MP Biomedicals, OH, USA). Tubes were quantified with a Wallace Winspectral 1414 liquid scintillation counter for saturation binding experiments and a Beckman Coulter LS 6500 for competition binding experiments. All ligand binding analyses were done in GraphPad Prism 6.0. For saturation binding, total [³H]-DeltII binding and non-specific binding data were fitted using a global nonlinear regression model and non-specific binding was subtracted from total binding to obtain a specific binding curve. B_max and K_d with 95% CI values were obtained by nonlinear regression of specific binding. Competition binding data were fitted to a one-site non-linear regression model to determine naltrindole K_i and IC₅₀ values with 95% CI. Strain differences were evaluated with an extra sum-of-squares F test to compare the best fit values of B_max, K_d and K_i.

2.10 HPLC analysis of catecholamines

Spinal cords from WT and α_2A-KO mice treated with 6-OHDA or vehicle solution were collected and processed for HPLC analysis (n = 3 per experimental group). Spinal cord samples were weighed and then sonicated in 150 µl of solution containing six parts of 0.2 M perchloric acid and one part of antioxidant solution (Kankaanpaa et al., 2001). The homogenates were centrifuged at 17,000 g for 35 min at 4 °C, and the supernatant was filtered using Millex-HV syringe filters (PVDF-membrane, 0.45 µm pores; Millipore). The concentrations of norepinephrine (NE), dopamine (DA) and epinephrine were determined with a UHPLC (Dionex UltiMate3000 RS, Thermo Scientific) combined with a coulometric array detector (Coularray 5600A, Thermo Scientific). Mobile phase consisted of 10 % acetonitrile in a 0.1 M phosphate solution (pH 3.0), supplemented with 1 mM EDTA and 1.5 mM 1-octanesulfonic acid, and the analytes were separated in a column (Acclaim™ 120, C18, 3 um, 2.1 × 100 mm, Thermo Scientific) that was kept at 30 °C. Eight coulometric cells of the detector were set to 50–450 mV. The data was analyzed with the CoulArray software using peak area as a measure for the amount of analyte. Each value was normalized against the weight of the sample to get final values as ng/mg tissue. Comparison between vehicle and 6-OHDA treatment in WT and α_2A-KO spinal cords was done with a 2-way ANOVA followed by a Tukey multiple comparison T test with GraphPad Prism 6.0.

3. Results

3.1 α_2A-KO mice are hypersensitive to noxious heat

The baseline sensitivity of α_2A-KO mice compared to WT mice to assess the role of the α_2A-AR in normal thermal and mechanical nociception. Compared to WT mice, α_2A-KO mice were more sensitive to heat stimuli in the tail immersion assay (Figure 1A; WT: 2.9 ± 0.2 sec. (n=22); α_2AKO: 1.4 ± 0.1 sec. (n=16); p<0.0001). Similarly, α_2A-KO mice were more sensitive than WT mice to heat from a radiant light beam focused on the hind paw (Figure 1B; WT: 6.8 ± 0.4 sec. (n=15); α_2A-KO: 5.6 ± 0.2 sec. (n=23); p = 0.007). There were no differences between WT and α_2A-KO mice in 50% mechanical withdrawal thresholds (Figure 1C; WT: 1.8 ± 0.2 g (n=17); α_2A-KO: 1.7 ± 0.1 g (n=16)) or in the number of nocifensive behaviors induced by intrathecally administered SP (Figure 1D; WT: 37 ± 4 counts (n=13); α_2A-KO: 30 ± 3 counts (n=16)). Together, these data indicate that α_2A-KO mice are more sensitive to heat stimuli than WT mice, but that mechanical sensitivity and responsively to intrathecal SP were not affected.

Figure 1. Comparison of thermal, mechanical and SP-induced nociceptive thresholds between WT and α _2A-KO mice.

Tail flick latencies measured in the hot water (49°C) tail immersion assay (A) and paw withdrawal latencies measured with the radiant heat assay (B) were more rapid in α_2A-KO mice. C) 50% mechanical threshold measured with von Frey filaments were similar on the hind paws of WT and α_2A-KO mice. D) The number of nocifensive behaviors induced by administration of 15 ng of SP i.t. were not significantly different between WT and α_2A-KO mice. Two-tailed unpaired t-test, ** p < 0.01, *** p < 0.0001.

3.2 Spinal adrenergic and opioid antinociception in α_2A-KO and WT mice

The role of α_2A-AR in spinal antinociception mediated by clonidine, morphine, DeltII and DAMGO were evaluated in the hot water tail immersion assay (Figure 2, top row) and in the SP behavioral assay (Figure 2, second row) by comparing the ED₅₀ values for each drug in WT and α_2A-KO mice (Table 2). All drugs dose-dependently increased tail flick latencies and reduced SP-induced behaviors in WT and α_2A-KO mice (Figure 2) with the exception of clonidine, which had reduced potency and efficacy in the α_2A-KO mice (Figure 2A). In contrast, clonidine potency and efficacy in the SP assay was not different in α_2C-KO mice compared to WT controls (Figure 2E). These data implicate the α_2A-AR in clonidine antinociception, although we cannot exclude a role for the α_2B-AR subtype.

Table 2.

Comparison of ED₅₀ values (nmol (95% CI)) for spinal morphine, DeltII and DAMGO in WT and α_2A-KO mice in the tail immersion and SP behavioral assays.

Assay	Strain	Clonidine	Morphine	DeltII	DAMGO
Tail immersion assay	WT	4.2 (2.0–4.7)	3.1 (2.0–4.7)	21.2 (11.4–39.2)	0.17 (0.09–0.32)
	α2A-KO	N/A§	0.26 (0.18–0.38)	4.3 (3.4–5.5)	0.06 (0.03–0.10)
	Potency ratio	N/A	12:1*	5:1*	3:1*
SP behavioral assay	WT#	0.4 (0.2–0.7)	0.93 (0.52–1.66)	0.15 (0.05–0.44)	0.0034 (0.0017–0.007)
	α2A-KO	7.0§ (2.9–16.6)	0.12 (0.03–0.46)	0.0021 (0.0006–0.008)	0.0041 (0.0026–0.0065)
	Potency ratio	1:18*	8:1*	70:1*	1:1
	α2C-WT	21.4 (9–51)
	α2C-WT	12.5 (4.5–35)

^*ED₅₀ significantly different, WT vs. α_2A-KO, t-test (p < 0.05)

^§Efficacy < 50%.

^#Values reported in Chabot-Doré et al. (2013)

Morphine had similar efficacy, but was more potent in α_2A-KO mice than in WT mice in both assays (Figure 2B). The potency of the DOPr-selective peptide agonist DeltII was also increased in α_2A-KO mice compared to WT mice (Figure 2C). The limited solubility of DeltII prevented us from testing it at higher doses to assess maximal efficacy. We also tested the antinociceptive response to the non-peptide DOPr-selective agonist SNC80 (Figure 2F). A high dose of SNC80 administered i.t. (100 nmol) elicited antinociception in both strains compared to vehicle but with greater efficacy in α_2A-KO compared to WT mice (2 way ANOVA; strain: F_{(1, 37)} = 4.48, p = 0.04, dose: F_{(1, 37)} = 13.6, p = 0.0007, interaction: F_{(1, 37)} = 0.78, p = 0.38). The antinociceptive response of the MOPr-selective peptide agonist DAMGO was more potent in α_2A-KO mice than in WT mice in the tail flick assay, but no difference in potency was observed in the SP behavioral assay (Figure 2D). Overall, these experiments indicate that in the absence of the α_2AAR, opioid spinal antinociception is potentiated.

3.3 Morphine-clonidine spinal antinociceptive synergy requires α_2A-AR

We tested if the synergistic interaction between the MOPr-preferring morphine and clonidine requires the α_2A-AR by evaluating this combination in WT and α_2A-KO mice in the hot water tail immersion and SP behavioral assays. In WT mice, the calculated ED₅₀ values for spinal morphine and clonidine were within one order of magnitude in both assays; we therefore combined them at a 1:1 ratio to test their interaction (Figure 3A, D). The potency of the combination was increased compared to each drug administered alone in the tail immersion assay (Figure 3A) and the SP behavioral assay (Figure 3D). Isobolographic analysis demonstrated that morphine combined with clonidine interacted synergistically in both assays (Figure 3B, E, Table 3). The interaction between morphine and clonidine was also assessed α_2AKO mice. Since a minimum of 50% efficacy is required for isobolographic analysis and clonidine did not reach this threshold in either assay, isobolographic analysis was not performed. As an alternative, we combined morphine and clonidine in the α_2A-KO mice in the same ratio (1:1) used in WT mice to test whether the addition of clonidine would alter the morphine doseresponse. The ED₅₀ values for the morphine + clonidine combinations were not significantly different from morphine alone in either assay (Figures 3C, F, Table 3), suggesting that the α_2AAR is required for the synergistic interaction between morphine and clonidine.

Table 3.

Experimental and Theoretical ED₅₀ values (nmol (± 95% SEM)) for drug combinations in WT and α_2A-KO mice.

Drug	Assay	WT # Experimental	Theoretical	Interaction	α2A-KO Experimental	Theoretical	Interaction
Morphine + Clonidine	TI	0.79 (±0.35)	3.52 (±1.67)	Synergistic*	0.29 (±0.13)	N/A	No interaction§
	SP	0.058 (±0.029)	0.56 (±0.27)	Synergistic*	0.071 (±0.044)	N/A	No interaction§
DeltII + Clonidine	SP	0.0010 (±0.0005)	0.22 (±0.19)	Synergistic*	0.009 (±0.015)	N/A	No interaction§

^*Indicates that the experimental ED₅₀ value < theoretical ED₅₀ value (t-test, p < 0.05)

^§Since clonidine did not reach 50% efficacy in α_2A-KO mice, isobolographic analysis was not performed (N/A). We therefore compared the ED₅₀ value for the drug combinations to the ED₅₀ values of DeltII or morphine alone (t test, p > 0.05).

^#SP behavioral assay values reported in Chabot-Doré et al. (2013).

3.4 Deltorphin II-clonidine spinal antinociceptive synergy requires α_2A-AR

To evaluate the role of the α_2A-AR in interactions with DOPr agonists, the DOPr agonist DeltII was tested in combination with clonidine in WT and α_2A-KO mice. In WT mice, DeltII and clonidine inhibited SP-induced behaviors with similar potency. We therefore co-administered these agonists at a 1:1 mixture reflecting their equi-effective ratio. The potency of the combination was increased compared to either drug administered alone (Figure 4A). Isobolographic analysis revealed that the experimental ED₅₀ value of the drug combination was significantly lower than the theoretical additive ED₅₀ value; the drug interaction is therefore synergistic (Figure 4B, Table 3). In α_2A-KO mice, as clonidine did not inhibit more than 50% of SP-induced behaviors, isobolographic analysis was not performed. However, using the same 1:1 ratio to combine DeltII and clonidine as in WT mice, we tested if the addition of clonidine to DeltII modulated its potency in α_2A-KO mice. Clonidine did not shift the DeltII dose-response curve, suggesting that there is no interaction between DeltII and clonidine in α_2A-KO mice (Figure 4C, Table 3).

3.5 MOPr expression level is unchanged in α_2A-KO mice

MOPr is the primary receptor mediating morphine antinociceptive effects and an increase in its expression could render morphine more potent. MOPr mRNA and protein expression in WT and α_2A-KO were therefore examined. Quantitative PCR analysis of the cDNA corresponding to MOPr mRNA was compared in DRG and spinal cord extracts from WT and α_2A-KO mice. No significant differences in MOPr expression between WT and α_2A-KO mice were observed in either DRG (WT = 1.0 ± 0.1, α_2A-KO = 1.2 ± 0.1, p = 0.15; Figure 5A) or spinal cord (1.0 ± 0.05, α_2A-KO = 1.1 ± 0.04, p = 0.47; Figure 5B). We further analyzed the expression of MOPr by western blot analysis of purified membrane extracts from WT and α_2A-KO spinal cords (Figure 5C). The relative MOPr density, normalized to GAPDH, was not different between WT and α_2A-KO mice (nonparametric t test, p = 0.7; Figure 5D). We conclude that neither MOPr mRNA nor protein expression were altered in α_2A-KO. Thus changes in MOPr expression do not explain the increased potency of morphine in α_2A-KO mice compared to WT mice.

Figure 5. Analysis of MOPr mRNA and protein expression in dorsal root ganglia (DRG) and spinal cords (SC) from WT and α_2A-KO mice.

Quantitative PCR analysis of the Oprm1 gene transcript from WT and α_2A-KO DRG (A) and SC (B) showed no significant strain differences in MOPr mRNA levels. Relative expression was obtained by normalizing expression values to three internal housekeeping reference genes (Gapdh, Hptr1, Ubc) and then normalized with the WT value of that tissue as a reference. C) Representative western blot showing the 67 kDa MOPr-immunoreactive (ir) bands that were used to quantify MOPr levels in the spinal cord of WT and α_2A-KO mice. D) Densitometry analysis of western blot MOPr-immunoreactive bands relative to GAPDH showed no strain difference between WT and α_2A-KO.

3.6 DOPr expression level is unchanged and DeltII binding affinity is unaffected in α_2A-KO mice

As the antinociceptive responses to DeltII and SNC80 were enhanced in α_2A-KO mice, we tested the hypothesis that DOPr expression levels were upregulated in these mice. We analyzed DOPr mRNA expression in spinal cord and DRG levels in WT and α_2A-KO mice by quantative PCR. No significant differences were found in the relative expression of DOPr receptors in DRGs between WT (1.00 ± 0.04) and α_2A-KO mice (0.79 ± 0.09; p = 0.07; Figure 6A). Similarly, relative DOPr expression in spinal cord tissues from WT mice (1.00 ± 0.08) was not different from that of α_2A-KO mice (1.10 ± 0.05; P = 0.33; Figure 6B).

Figure 6. Analysis of DOPr mRNA and [³H]-DeltII binding in WT and α_2A-KO mice.

Quantitative PCR analysis of the Oprd1 gene transcript from WT and α_2A-KO DRG (A) and spinal cord (B) mRNA extracts showed no strain differences in DOPr mRNA expression. Relative expression was obtained by normalizing expression values to three internal house keeping reference genes (Gapdh, Hptr1, Ubc) and then normalized with the WT value of that tissue as a reference. C) Saturation ligand binding performed on spinal cord membrane protein extracts from WT and α_2A-KO mice showing specific binding obtained by subtracting nonspecific from total [³H]-DeltII binding. The binding was concentration-dependent and no significant strain difference was detected. D) Competition of 1 nM [³H]-DeltII binding to spinal cord membranes by naltrindole was similar in WT and α_2A-KO mice. CPM = counts per minute. DPM = disintregrations per minute.

Saturation binding assays were conducted to measure the number of binding sites and the affinity of [³H]-DeltII in SC membrane preparations from WT and α_2A-KO mice. This method was selected as an alternative to western blot analysis since the selectivity of DOPr antibodies is controversial (Scherrer et al., 2009; Wang et al., 2010). [³H]-DeltII bound to membrane preparations in a concentration-dependent manner (Figure 6C) and the total number of specific binding sites was similar in both strains (WT: B_max = 373 (319–426) CPM, α_2A-KO: B_max = 309 (201–417) CPM; nonlinear regression, p = 0.14) and apparent dissociation constant (WT: K_D = 0.9 (0.4–1.4) nM, α_2A-KO: K_D = 1.1 (−0.4–2.6) nM; nonlinear regression, p = 0.68) were not significantly different. We further evaluated [³H]-DeltII affinity using competition binding assays. The affinity of naltrindole, a DOP-selective antagonist, determined by competition with [³H]-DeltII in WT and α_2A-KO was similar in both strains (p = 0.29; Figure 6D). Naltrindole competitively inhibited [³H]-DeltII binding with a K_I value of 0.17 (0.12–0.25) nM in WT membrane preparations and a K_I value of 0.14 (0.11–0.18) nM in α_2A-KO membrane preparations. Thus, the absence of the α_2A-AR did not alter DeltII affinity or DOPr expression levels in a manner that would explain the increased DOPr-mediated antinociceptive potency or efficacy observed in α_2A-AR mice.

3.7 Endogenous norepinephrine is not necessary for the potentiation of morphine antinociception in α_2A-KO mice

Disruption in α_2A-AR function can result in elevated NE levels in the central nervous system (Lakhlani et al., 1997). This is due to the loss of negative feedback by presynaptic α_2A-AR autoreceptors, which normally attenuate NE release. Since co-administration of morphine with NE at the spinal cord results in analgesic synergy (Roerig et al., 1992), we hypothesized that elevated NE levels, due to the loss of feedback inhibition in α_2A-KO, could contribute to the increased spinal morphine efficacy observed in these mice. To test this hypothesis, the toxin 6-OHDA was injected intrathecally to eliminate spinal catecholaminergic fibers that release NE in the spinal cord (Fasmer et al., 1986). In WT mice, morphine efficacy was reduced in mice treated with 6-OHDA compared to mice treated with vehicle, confirming the role of endogenous NE in the antinociceptive effect of morphine (Figure 7A). In α_2A-KO mice, increased morphine potency was observed as before (WT vs. α_2A-KO; vehicle-treated) and 6-OHDA treatment had no effect on morphine potency or efficacy in these animals (Figure 7A, Table 5).

Figure 7. Effect of selective elimination of spinal catecholaminergic nerve terminals on morphine antinociception and heat nociception.

WT and α_2A-KO mice were injected i.t. with either 6-OHDA or vehicle to eliminate catecholaminergic nerve fibers containing NE. A) Morphine dose-response curves were constructed using a cumulative dosage protocol. 6-OHDA reduced morphine response in WT mice, but not in α_2A-KO mice. ED₅₀ values are reported in Table 5. B) HPLC analysis of spinal content in norepinephrine (NE) and dopamine (DA) in ng/mg of spinal cord tissue. Vehicle-treated α_2A-KO mice have lower NE and DA levels than WT mice. 6-OHDA treatment completely eliminated NE in both strains. C) Tail flick latency values were lower in 6-OHDAtreated WT mice compared to vehicle-treated mice. The 6-OHDA treatment had no effect on baseline latencies of α_2A-KO mice. *** p < 0.001 ** p < 0.01.

Table 5.

ED₅₀ values for morphine in WT and α_2A-KO mice treated with 6-OHDA (i.t.).

Strain	Treatment	ED50 (95% CI)
WT	Vehicle	6.2 nmol (3.9 – 9.9)
	6-OHDA	No efficacy
α2A-KO	Vehicle	0.48 nmol (0.36 – 0.64)
	6-OHDA	0.55 nmol (0.33 – 0.91)

The elimination of catecholaminergic fibers may also affect levels of neurotransmitters other than NE. We therefore performed HPLC analysis to evaluate the effect of 6-OHDA treatment on levels of NE, epinephrine and dopamine (DA) in spinal cords of WT and α_2A-KO mice (Figure 7B). Levels of NE recovered from spinal cord supernatant were lower in vehicle-treated α_2A-KO mice compared to vehicle-treated WT mice. However, in both strains, NE was undetectable in 6-OHDA-treated mice (2 way ANOVA; strain: F_{(1, 8)} = 18.0, p = 0.003, treatment: F_{(1, 8)} = 1083, p = 0.0001, interaction: F_{(1, 8)} = 18.0, p = 0.003). The amount of DA in the spinal cord was much lower compared to NE in both strains and α_2A-KO mice and 6-OHDA treatment did not significantly alter DA levels in either mouse strain (2 way ANOVA; strain: F_{(1, 8)} = 10, p = 0.01, treatment: F_{(1, 8)} = 5, p = 0.06, interaction: F_{(1, 8)} = 1.2, p = 0.3). Epinephrine was undetectable in the spinal cord of all experimental groups. Thus, α_2A-KO mice have lower catecholamine levels in their spinal cord, but 6-OHDA treatment selectively eliminated NE, which was in both cases, the most abundant catecholamine neurotransmitter.

Together, these results show that increased morphine potency was maintained in α_2A-KO mice despite the selective lesion of noradrenergic fibers, suggesting that the loss of negative feedback inhibition of NE in the α_2A-KO mice is not responsible for the effects on morphine antinociception.

3.8 Descending noradrenergic inhibition of noxious heat sensitivity is mediated through the α_2A-AR

Descending noradrenergic fibers modulate endogenous inhibition of nociceptive stimuli at the spinal cord level by activating α₂-ARs (Gilsbach et al., 2009). We hypothesized that the increased sensitivity to heat observed in α_2A-KO mice (Figure 1) was due to a loss of sensitivity to descending endogenous NE inhibition in these mice. To test this, tail flick latencies from WT and α_2A-KO mice in which descending NE fibers were eliminated were compared (Figure 7B). 6-OHDA-treated WT mice were more sensitivity to heat compared to vehicle-treated mice (WT vehicle = 1.9 ± 0.2 sec., WT 6-OHDA = 1.1 ± 0.1 sec., p = 0.01), confirming the role of descending noradrenergic fibers in heat nociception. In α_2A-KO mice, no such difference was observed in tail flick latencies between 6-OHDA-and vehicle-treated mice (α_2A-KO vehicle = 1.5 ± 0.2, α_2A-KO 6-OHDA = 1.2 ± 0.1, p = 0.14), suggesting that the α_2A-AR subtype mediates endogenous NE modulation of heat nociception.

4. Discussion

4.1 The α_2A-adrenoceptor modulates heat nociception

The hypersensitivity to heat observed in α_2A-KO mice compared to WT mice in the tail flick and radiant heat assays suggests that heat nociception is endogenously modulated by the α_2A-AR under normal conditions. These data also demonstrate that the hypersensitivity is not site-specific, since α_2A-KO mice were more sensitive to heat applied on either the tail or the plantar surface of the hind paw. Changes in thermal thresholds were not observed in previous studies using α_2A-D79N mice where a point mutation rendered the α_2A-AR dysfunctional and decreased its expression levels (Lakhlani et al., 1997; MacMillan et al., 1996; Malmberg et al., 2001; Surprenant et al., 1992). Thus endogenous inhibition of sensitivity to noxious heat could require the physical presence but not necessarily the normal signaling properties of the α_2A-AR. Differences in heat nociception were also not previously reported using α_2A-KO mice in the radiant heat assay (Lähdesmäki et al., 2003). It is possible that heat hyperalgesia in α_2A-KO mice depends on the experimental paradigm which could differentially activate C and Aδ fibers (Yeomans and Proudfit, 1996). Consistent with previous reports, a strain difference in tactile threshold was not observed (Lähdesmäki et al., 2003; Malmberg et al., 2001). These results suggest a modality-specific role for the α_2A-AR in the endogenous modulation of nociception. Descending noradrenergic pathways contribute to endogenous inhibition of pain through activation of α₂-ARs. α₂-AR antagonists decrease tail flick and hot plate latencies in rats (Sagen and Proudfit, 1984), suggesting that adrenergic receptors help set the threshold for thermal nociception. A decreased sensitivity to heat but not to mechanical stimuli was reported in rats and mice following the lesioning of catecholaminergic fibers with 6-OHDA (Fasmer et al., 1986; Kuraishi et al., 1983). Furthermore, Jasmin et al. (2002) observed a similar thermal, but not mechanical, hypersensitivity in dopamine β-hydroxylase-KO mice, which lack the enzyme that converts dopamine to NE. Similarly, increased sensitivity to heat was noted in mice with elevated extracellular NE levels resulting from a genetic deletion of the NE transporter (Bohn et al., 2000). These studies all suggest that endogenous NE modulates heat nociception. Here, we confirm the increase in heat sensitivity in WT mice following an i.t. 6-OHDA treatment and show that similar changes are not matched in 6-OHDA-treated vs. vehicle-treated α_2A-KO mice. Thus, the α_2A-AR is involved in the inhibition of noxious heat via descending noradrenergic fibers in the spinal cord in naïve animals.

4.2 Unifying model: The α_2A-adrenoceptor may allosterically modulate spinal opioid antinociception

In this study, a set of seemingly contradictory observations regarding regulation of opioid antinociception by the α_2A-AR are presented. These observations are summarized in Figure 8A using morphine as an example, although similar observations were made with DeltII. First, morphine and DeltII potency were enhanced in the absence of the α_2A-AR. Second, we report that morphine and DeltII synergy with clonidine is mediated by the α_2A-AR, illustrating that opioid antinociception is enhanced when the α_2A-AR is activated by clonidine. Third, morphine antinociception was diminished following the depletion of spinal NE in WT, but not in α_2A-KO mice.

Figure 8. Proposed model of allosteric regulation of opioid receptors by the α _2A-adrenoceptor.

A) Schematic showing dose-response curves obtained for spinal morphine under different experimental conditions. In WT mice, morphine produces a dose-dependent antinociceptive effect (blue solid line), which is reduced in mice treated with 6-OHDA (blue dashed line) and potentiated by the addition of clonidine (purple solid line). In α_2A-KO mice, morphine is more potent than in WT mice (red solid line) and neither the 6-OHDA treatment (red dashed line) nor the addition of clonidine (purple dashed line) shifted the morphine dose-response curve. B) In the proposed model, in WT mice, opioid receptors (OPr, green) are inhibited by the α_2A-AR (red) in its ligand-free inactive state, which results in decreased opioid receptor antinociceptive output. When an agonist like clonidine (Clon) or norepinephrine (NE) activates α_2A-AR, this inhibitory action on opioid receptors is removed, resulting in the synergism between the two agonistoccupied receptors. Under normal conditions, tonic NE release from descending noradrenergic fibers maintains an equilibrium state between ligand-free and agonist-occupied α_2A-AR. When NE is depleted, the equilibrium is shifted towards the inactive and inhibitory α_2A-AR, and when exogenous clonidine is administered, the equilibrium is shifted towards an active α_2A-AR state, which disinhibits the opioid system. In α_2A-KO mice, opioid receptors are not subjected to inhibition by the α_2A-AR regardless of the presence or absence of α_2A-AR ligands and are thus always fully activated.

The above findings lead us to propose allosteric regulation as the basis by which the α_2A-AR modulates spinal opioid receptor function in an activation state-dependent manner (Figure 8B). This model predicts that ligand-free α_2A-ARs inhibit opioid receptor function, as was observed after depleting spinal NE. The model also predicts that this inhibitory action is removed when the α_2A-AR is occupied by agonists, thereby potentiating opioid analgesia, such as when clonidine is given in combination with morphine or DeltII. Finally, it predicts that loss of the α_2A-AR in the KO animals removes this constitutive inhibitory action on opioid receptors, leaving the opioid receptors fully functional and suggesting at least two types of modulation – positive and negative - by the α_2A-AR. According to this model, the tonic release of NE in the spinal cord creates an equilibrium between the ligand-occupied and ligand-free α_2A-ARs that give rise to an intermediate response to opioid agonists in WT mice under normal conditions. Removing NE will favor the constitutive inhibition while the addition of α_2A-AR agonists, such as clonidine, shifts the equilibrium towards the occupied receptor, which potentiates opioid responses.

4.3 Functional consequences of allosteric interactions within GPCR oligomers

G protein-coupled receptors are in direct physical contact within hetero-oligomers and the binding of a ligand on one receptor can cause functional changes at the other receptor through allosteric interactions (Kenakin et al., 2010). State-dependent α_2A-AR regulation of opioid effects such as reported here are consistent with previously reported allosteric interactions. Vilardaga et al. (2008) showed that MOPr allosterically modulates α_2A-AR upon morphine binding by reducing conformational changes in the α_2A-AR and G_i activation by NE. The reduction in G_i signaling was accompanied by inhibition of the ERK1/2 signaling pathway. The ability of α_2A-AR to allosterically modulate MOPr was not investigated in that study; however, if the allosteric interaction is symmetric, our model would predict that the α_2A-AR has similar allsoteric effects on the MOPr. Jordan et al. (2003) observed a decrease in GTPγS binding and MAPK phosphorylation when MOPr/α_2A-AR-expressing cells were treated with morphine/clonidine, which could result from decreased G_i signaling output (Jordan et al., 2003). When activated individually, MOPr and α_2A-AR normally activate MAPK signaling pathways. In the presence of agonists for both receptors, the MOPr/α_2A-AR may switch to different signaling pathways as indicated by the suppression of G_i-protein coupling and downstream signaling. Protein kinase C (PKC) is a likely downstream target of the new signaling pathway engaged by the co-activation of MOPr and α_2A-AR since inhibiting PKC blocks the analgesic synergy between morphine and clonidine, but not the antinociceptive effect of each drug alone (Wei and Roerig, 1998). In our model, we postulate that the un-liganded α_2A-AR constitutively inhibits the opioid receptor and that activating the α_2A-AR would relieve this constitutive inhibition and engage alternative signaling pathways.

Allosteric interactions between GPCR heterodimers may change ligand binding properties, leading to negative or positive cooperativity that can be detected in binding experiments. For example, changes in the shape and slope of saturation and competition binding curves could reflect alterations in receptor density or ligand affinity (Albizu et al., 2006; Durroux, 2005). Our behavioral results and allosteric interaction model predict that, in the absence of the α_2A-AR, the number of binding sites or binding affinity would increase. Using [³H]-DeltII binding to spinal cord membrane preparations from WT and α_2A-KO mice, we saw a small decrease in maximal binding sites (B_max) in α_2A-KO mice, but ligand affinity (K_D and K_i) was not different from WT mice. These results suggest that the presence of an un-liganded α_2A-AR does not affect [³H]DeltII binding properties. However, allosteric interactions usually occur when the orthosteric binding site is occupied by an agonist or an antagonist. For example, Browning et al. (1982) reported a displacement of [³H]DADLE, a DOR agonist, by α₂AR ligands in rat brain membranes (Browning et al., 1982). Competition binding experiments with [³H]DeltII in the presence of α_2A-AR ligands would provide further insights into the hypothesized allosteric interactions.

Although we propose a model based on allosteric interactions, it is possible that functional interactions could be mediated in the absence of direct physical interactions between the receptors (e.g. via intracellular crosstalk). We cannot exclude other possibilities such as molecular crosstalk between the two receptors which might be mediated by constitutive activity of the α_2A-AR. That said, there is significant precedent in the literature for allosteric interactions between antagonist binding (which does not activate signaling) on one receptor partner modulating signaling via the other (Goupil et al., 2015; Wrzal et al., 2012). Such experiments could be conducted on the putative α_2A-AR/OPr dimers in future studies. Results from pharmacological studies using α₂-AR antagonists to investigate opioid-adrenergic interactions are not consistent with our observations in α_2A-KO mice, indicating that blocking the α_2A-AR is different from removing it. Combining morphine with the α₂-AR antagonists yohimbine or idazoxan has been reported to decrease morphine efficacy (Browning et al., 1982; Iglesias et al., 1992; Morales et al., 2001; Stone et al., 1997). Interestingly, this interaction was paralleled in binding studies on CNS membranes whereby α₂-AR ligands could displace radiolabeled µ or δ opioid ligands (Browning et al., 1982). These two classes of ligands do not bind the same receptor sites, again suggesting that an allosteric interaction rather than a direct competition affected opioid binding properties. Further, since yohimbine and idazoxan are antagonists for the α₂-AR rather than agonists, they would not be expected to activate cellular signaling and thus their effects would not simply be due to molecular crosstalk between the two receptor pathways. In our model, α₂-AR antagonists would counteract opioid effects by displacing NE bound to the α_2A-AR, thus stabilizing the receptor complex in an inactive conformation. This model, however, does not explain how ultra low doses of α_2A-AR antagonists have the opposite effect on morphine antinociception, i.e. an increase in morphine potency and prolonged antinociceptive effect (Milne et al., 2014). Finally, when measuring the effect of drug interactions at the behavioral level, one cannot exclude the possibility that the interaction involves distinct cells in the system under study.

4.4 Morphine is more potent in absence of the α_2A-AR

Other studies have examined the role of α_2A-AR in the antinociceptive action of morphine using genetically modified mice. In α_2A-D79N mice, spinal morphine potency in the warm water tail immersion assay was not different from WT mice, but a decreased potency in the SP assay was reported (Stone et al., 1997). In another study, no difference in systemic morphine antinociceptive effect was observed between WT and α_2A-D79N mice in the hot plate assay (Lakhlani et al., 1997). Our data report the opposite: the potency of morphine is augmented in α_2A-KO mice with both the tail immersion and the SP behavioral assay. This potentiation is not due to an upregulation of MOPr mRNA or protein. The divergent results between α_2A-D79N and α_2A-KO mice raise two possibilities: 1) morphine potentiation requires the complete absence of α_2A-AR and/or 2) the dysfunctional α_2A-D79N is capable of preventing morphine potentiation. A similar observation was reported in which the D2 dopamine receptor (DRD2) is allosterically regulated by a co-localized and un-liganded ghrelin receptor GHSR1a (Kern et al., 2012). The mere presence of GHSR1a, in the absence of its activation (since this brain region never sees ghrelin), was sufficient to modulate DRD2 signaling output and to produce an anorexigenic response to dopamine. In absence of GHSR1a, dopamine did not affect food intake in GHSR1a-KO mice. This supports our assertion that the opioid response observed in α_2A-KO mice is due to the loss of the negative allosteric influence of the unliganded α_2A-AR.

Using α_2A-KO mice, Ozdogan et al. (2006) observed an enhanced response to morphine (i.p.) and other partial opioid agonists with the tail flick assay, but not for the full agonist fentanyl or with the hot plate assay. This is consistent with our observation of greater potentiation of spinal morphine efficacy compared to the small potentiation of the full MOP agonist, DAMGO (i.t.) in α_2A-KO mice. In contrast, using a single morphine test dose (i.p.), the same group previously reported no strain difference in morphine effect (Ozdogan et al., 2004). The use of different doses, administration routes or behavioral assays could be a source of discrepancy between studies.

4.5 DOP agonists are more potent and effective in α_2A-KO mice

In both the tail flick and the SP behavioral assays, DeltII antinociception was enhanced in α_2A-KO mice. This is in contrast with previous work showing that DeltII-mediated antinociception was unchanged in α_2A-D79N mice in the SP behavioral assay (Fairbanks et al., 2002; Stone et al., 2007; Stone et al., 1997). Since we have demonstrated that DOPr mediates DeltII efficacy in the SP behavioral assay (Chabot-Doré et al., 2013), we are confident that DOPr also mediates DeltII antinociception in α_2A-KO mice. However, because the antinociceptive effect of DeltII in the tail immersion assay may not be DOPr-mediated (Scherrer et al., 2009; van Rijn et al., 2012), we cannot conclude with certainty that the enhanced DeltII efficacy in α_2A-KO mice is mediated by DOPr in this assay. Nevertheless, the enhancement was observed with another DOPr-selective agonist, SNC80, demonstrating that the effect extends to another agonist. Upregulation of functional DOPr has been reported following inflammation, nerve injury, morphine treatment and prolonged exposure to ethanol (Cahill et al., 2003; Cahill et al., 2001; Gendron et al., 2006; Lucido et al., 2005; van Rijn et al., 2012). It is therefore possible that the absence of the α_2A-AR triggers a similar upregulation. Our results show that neither Oprd1 transcript levels nor the number of [³H]-DeltII binding sites were augmented in α_2A-KO mice compared to WT mice. Other mechanisms leading to DOPr upregulation – increased surface expression or change in signaling output – have not been investigated and therefore are not excluded.

4.6 The α_2A-AR is necessary for clonidine-mediated synergistic interactions with DeltII and morphine

The spinal analgesic effect of clonidine is attributed to the α_2A-AR (Fairbanks and Wilcox, 1999), an observation we confirmed with α_2A-KO mice in both the SP behavioral assay (Chabot- Doré et al., 2013) and the tail immersion assay. Other α₂-AR agonists such as dexmetomidine, UK 14,304 and ST-91 also mediate their antinociceptive effects at least in part through the α_2AAR (Lakhlani et al., 1997; Stone et al., 2007; Stone et al., 1997). While the synergistic interaction between DeltII and UK 14,304 was lost in α_2A-D79N mice (Stone et al., 1997), synergy is maintained when DeltII is combined with either ST91 (Stone et al., 2007) or moxonidine (Fairbanks et al., 2002). Therefore, the role of the α_2A-AR in synergy between α₂-AR agonists and DeltII depends on the adrenergic ligand tested, with the α_2C-AR contributing to this interaction in some cases (Fairbanks et al., 2002). In the current study, clonidine synergy with DeltII was absent in α_2A-KO mice, suggesting that the α_2A-AR is necessary for clonidine’s synergistic interaction with DeltII. In a nerve injury model, α₂-AR analgesic efficacy was bi-directionally modulated by activating or inhibiting DOPr, suggesting that this receptor pair is a valid therapeutic target for neuropathic pain (Aira et al., 2015).

Consistent with the DeltII results, the synergistic interaction between morphine and clonidine was also absent in the α_2A-KO mice. To our knowledge, this is the first report demonstrating that the α_2A-AR is necessary in this widely studied and clinically relevant drug interaction.

4.7 Endogenous norepinephrine modulates opioid antinociception

A subset of α_2A-ARs in the spinal cord are expressed on descending noradrenergic nerve fibers, where they act as autoreceptors regulating the release of NE via negative feedback inhibition (Gilsbach et al., 2009). The absence of α_2A-AR could disrupt this negative feedback inhibition, resulting in increased NE release which could potentiate the effects of spinally administered morphine. If elevated spinal NE levels mediate the increased opioid response in α_2A-KO, the 6-OHDA treatment would have decreased morphine potency and efficacy to the same level in both strains. Instead, the chemical lesion did not affect the potentiated morphine response in α_2A-KO mice. Therefore, the enhanced potency of morphine in α_2A-KO mice is not due to an excess of NE in α_2A-KO mice and the mechanism must be independent of endogenous NE levels. Since we demonstrated that morphine efficacy decreased in 6-OHDA-treated WT mice, but not in α_2A-KO mice, we also conclude that, under normal conditions, spinal morphine efficacy requires the action of endogenous NE on α_2A-AR.

The link between the endogenous noradrenergic system and opioid responses has been studied by manipulating spinal NE levels, but the adrenergic receptor requirement has not been adressed. Depleting spinal levels of NE by different methods has been shown to reduce the antinociceptive effect of morphine in rodents (Berge and Ogren, 1984; Jasmin et al., 2003; Zhong et al., 1985). Others have found that these treatments had no effect on morphine antinociception (Milne et al., 2008; Pappas et al., 1982; Sawynok et al., 1991). These discrepancies may be attributed to differences in experimental design such as the choice of behavioral assay and time point. Morphine efficacy is also decreased in mice lacking dopamine β-hydroxylase (Dbh), the enzyme that converts dopamine to NE (Jasmin et al., 2002). Interestingly, the reverse is also true: when NE levels are upregulated such as in NE transporter (NET)-KO mice, morphine efficacy is enhanced (Bohn et al., 2000). These studies point at a role for the noradrenergic system in modulating morphine antinociception. Our study further identifies the α_2A-AR as the receptor mediating this effect by showing that α_2A-KO mice are not responsive to a decrease in spinal NE levels.

4.8 Use of genetically modified animals

The interpretation of the results presented here has certain limitations. Our experimental approach relies primarily on the comparison of WT mice with a mouse strain carrying a genetic deletion in the Adra2A gene. This approach provides the opportunity to assess α_2A-AR function. For example, while clonidine is a non-selective α₂-AR agonist with affinity for all three α₂-AR subtypes (α_2A-AR, α_2B-AR, α_2C-AR), our results show that α_2A-AR is the principal mediator of clonidine analgesia and synergistic interaction with opioids in the spinal cord. The use of transgenic KO mice presents certain caveats that must be considered. For example, global KO suppresses the expression of α_2A-AR not only in DRGs and the spinal cord, but also in all tissues. Deletion of these receptors could therefore lead to compensatory mechanisms or dysfunctions in the expression of other genes or in the regulation of other physiological systems, introducing confounding factors in the interpretation of our results. The pattern of autoradiographic labeling of a non-selective α₂-AR ligand in the brain of α_2A-KO mice was similar to the distribution of α_2C-AR in WT mice, suggesting that there are no profound changes in other α₂-AR binding sites due to the deletion (Lahdesmaki et al., 2002). We also confirmed that the expression of MOPr and DOPr were unchanged in α_2A-KO mice, but the possibility that other genes are dysregulated cannot be ruled out. Given that no changes were observed between α_2A-KO and WT mice in a) mechanical nociceptive thresholds, b) sensitivity to intrathecal SP and c) the antinociceptive responses to DAMGO, some components of nociception and antinociception are normal in these mice.

4.9 Clinical implications

Uncovering the role of the α_2A-AR in the modulation of pain and opioid sensitivity by NE raises the possibility that an imbalance in adrenergic signaling in the spinal cord may be related to some chronic pain conditions. For example, the etiology of fibromyalgia syndrome (FMS) is still largely unknown. Cerebrospinal fluid (CSF) levels of SP are higher in FMS patients than in normal individuals (Russell, 1998), but the CSF levels of methoxyhydroxyphenylglycol, a metabolite of NE, are lower (Russell et al., 1992). This suggests that FMS patients have reduced noradrenergic tone, which reduces the inhibition of SP release from nociceptors and sensitizes the nociceptive circuit in the spinal cord. Our results suggest that low levels of NE in the spinal cord would also decrease the effect of endogenous or exogenous opioids in FMS patients and restoring NE levels could attenuate pain symptoms associated with FMS. Tramadol is a noradrenergic reuptake inhibitor and a weak MOP agonist that is efficacious for the treatment of FMS (Russell et al., 2000). According to our proposed model, the success of this treatment could be attributed to its dual action: 1) the reuptake inhibitor restores spinal NE levels, which favors the agonist-occupied form of the α_2AAR and releases the inhibition on MOPr so that 2) the action of tramadol on MOPr can be fully effective.

4.10 Conclusion

Here we propose a model of activation state-dependent allosteric modulation of opioid receptor function by α_2A-ARs. According to the model, activation of α_2A-ARs facilitates and the inactive form inhibits opioid antinociception. This model is supported by previous studies showing coexpression in vivo and physically interactions in vitro, and accounts for many of the seemingly contradictory opioid-adrenergic interactions observed in pre-clinical and clinical settings. Finally, this model has implications for understanding functional interactions between other coexpressed G protein-coupled receptor pairs in vivo.

Highlights.

• The role of α_2A-adrenoceptors (ARs) in spinal opioid antinociception was investigated.

• The α_2A-AR is necessary for analgesic synergy between clonidine and opioid agonists.

• Opioid antinociception is enhanced in α_2A-AR-KO mice.

• A model of activation state-dependent allosteric modulation is proposed

• The model has implications for allosteric modulation of other G protein-coupled receptor pairs.

Abbreviations

α₂-AR

α₂-adrenoceptor

α_2A-AR

α_2A-adrenoceptor

DeltII

[d-Ala²]-deltorphin II

DAMGO

[d-Ala², NMe-Phe⁴, Gly-ol⁵]-enkephalin

DOPr

delta opioid receptor

MOPr

mu opioid receptor

i.t.

intrathecal

Oprd1

mouse delta opioid receptor gene

Oprm1

mouse mu opioid receptor gene

DA

dopamine

DRG

dorsal root ganglia

SC

spinal cord

MPE

maximal possible effect

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

NE

norepinephrine

SP

substance P

6-OHDA

6-hydroxydopamine