Opioid-induced hyperalgesia (OIH) is a syndrome of increased sensitivity to noxious stimuli, seen after both the acute and chronic administration of opioids, that has been observed in humans and rodent models. This syndrome may reduce the clinical utility of opioids in treating acute and chronic pain.
In these studies, the authors measured the propensity of 15 strains of inbred mice to develop mechanical manifestations of OIH. These data were subjected to in silico genetic analysis, which resulted in the association of haplotypic blocks within or near several known genes. Both pharmacologic agents and transgenic mice were used to confirm the functional association of the most strongly linked gene with OIH.
Both baseline mechanical nociceptive thresholds and the percentage changes in these thresholds after 4 days of morphine treatment were found to be highly strain dependent. The haplotypic blocks most strongly associated with the mechanical OIH data were located within the beta2 adrenergic receptor gene (beta2-AR). Using the selective beta2-AR antagonist butoxamine, the authors observed a dose-dependent reversal of OIH. Furthermore, deletion of the beta2-AR gene sharply reduced the mechanical allodynia present after morphine treatment in the wild-type mouse strain. Analysis of the associated beta2-AR haplotypic block identified single nucleotide polymorphisms potentially explaining in part the strain specific differences in OIH.
Genetic variants of the beta2-AR gene seem to explain some part of the differences between various strains of mice to develop OIH. The association of this gene with OIH suggests specific pharmacologic strategies for reducing the impact of OIH on patients consuming opioids.
ACCUMULATING evidence suggests that the administration of opioid analgesics leads not only to analgesia, but to a paradoxical sensitization to noxious stimuli that is particularly evident after the abrupt cessation of opioid administration or at the time of serum opioid nadir between regular doses. This phenomenon is referred to as opioid-induced hyperalgesia (OIH). Among the more important human studies documenting this effect are those demonstrating hyperalgesia in patients formerly addicted to opioids who are maintained on methadone when compared with matched controls not receiving methadone or other opioids.1–3This hyperalgesia was most pronounced immediately before daily methadone administration but was measurable even at time points closer to peak methadone serum concentrations.4A recent prospective trial in which sustained-acting morphine was given to patients with chronic low back pain demonstrated measurable hyperalgesia within 1 month of beginning therapy.5Other human data suggest that the short-term infusion of opioids such as the μ-opioid receptor agonist remifentanil followed by abrupt cessation exacerbates preexisting hyperalgesia.6–8Some evidence suggests this phenomena is due to the activation of N -methyl-d-aspartate receptors.7
More recently, rodent models have been used to study OIH. Many laboratories have reported mechanical allodynia, thermal hyperalgesia, or both after the acute administration of opioids such as heroin and fentanyl,9,10the chronic (days) administration of intrathecal morphine,11,12the local peripheral administration of morphine,13or the chronic administration of systemic opioids of several types.14–16Many mechanisms have been proposed to explain this type of sensitization, with some of the more commonly discussed possibilities including activation of N -methyl-d-aspartate receptors,14,17,18activation of facilitative descending pathways from the rostral ventromedial medulla (RVM),16the decreased reuptake of neurotransmitters from primary afferent fibers,19and the enhanced responsiveness of spinal neurons to nociceptive neurotransmitters such as substance P and glutamate.20,21Although not previously linked to OIH, the enhanced expression of β2-adrenergic receptors (β2-ARs) have been identified as adaptive changes occurring during chronic exposure to opioids.22,23Likewise, the functional enhancement of β2-AR signaling has been demonstrated after chronic morphine exposure in various nervous system tissues.24,25These observations will become relevant to the current studies. Although traditional pharmacologic, electrophysiologic, biochemical, and molecular techniques have been useful in the exploration of OIH, we are now in position to use murine genetics to identify genomic loci linked to this phenomenon.
A substantial and growing body of literature supports the conclusion that genetics influence pain sensitivity and analgesic responses. With respect to the consequences of chronic morphine administration, several reports explore the genetic basis of tolerance and dependence specifically.26–28The genetics of OIH are not well explored in part because of the barriers posed by genetic studies. Traditional murine mapping experiments have advanced our understanding of the genetic basis of pain, but the techniques generally used are quite time-consuming and accessible only to laboratories with substantial expertise in genetics and molecular biology. More recently, in silico techniques have been introduced, which allow mapping to be performed in much expedited fashion. These techniques rely on the availability of high-resolution single nucleotide polymorphism (SNP) databases. The computational algorithms then compare phenotypic trait data for a series of inbred mouse strains with the SNP alleles for those strains as organized into either genomic segments of arbitrary size29or, more recently, as organized into haplotypic blocks.30,31These now well-described techniques have proven useful in identifying chromosomal regions and even specific genes involved in many traits, including bone metabolism, alcohol withdrawal, immune system function, susceptibility to pulmonary injury, the expression of specific genes, and several other traits.29,30,32In these studies, we used in silico mapping to identify haplotypic blocks associated with OIH and confirmed a functional association for one haplotypic block corresponding to the β2-AR using independent techniques.
Materials and Methods
All animal experiments were done after approval of protocols by our Institutional Animal Care and Use Committee (Palo Alto, CA) and complied with the Guide for the Care and Use of Laboratory Animals available through the National Academy of Sciences (Washington, D.C.).33
Inbred Mouse Strains.
Inbred mouse strains were obtained from Jackson Labs (Bar Harbor, ME) at 7–8 weeks of age. Mice were kept a further 7–10 days from the date of arrival in our animal care facility before use to allow for acclimation. Mice were kept under pathogen-free conditions and were provided food and water ad libitum with a 12:12-h light:dark cycle. The strains used were 129/SvlmJ, A/HeJ, A/J, AKR/J, B10.D2-H2/oSNJ, BALB/cByJ, BALB/cJ, C3H/HeJ, C57BL/6J, DBA/2J, LP/J, LG/J, MRL/MpJ, NZB/BinJ, and NZW/LaCJ (15 strains).
FVB and FVB β2-AR congenic null mutants were obtained from a local breeding colony. The generation of these mice is described by Chruscinski et al. 34These mice were individually genotyped and used in our experiments at 7–8 weeks of age. Animal husbandry was otherwise identical to that used for the inbred strains.
After baseline nociceptive testing, morphine (Sigma Chemical, St. Louis, MO) was administered to mice subcutaneously 20 mg/kg twice per day on days 1–3. On day 4, the dose was increased to 40 mg/kg twice per day in 50- to 100-μl volumes of 0.9% NaCl similar to our previous protocols for generating opioid-induced hyperalgesia.14,35For OIH determinations, mice were assessed 16 h after the final dose of morphine.
The selective β2-AR antagonist butoxamine was obtained from Sigma Chemical and diluted in 0.9% NaCl before use. After the measurement of baseline mechanical thresholds in C57BL/6J mice, butoxamine was injected subcutaneously. Behavioral testing was repeated in 30 min, and the next higher dose of butoxamine in the series was then injected into the same mice to obtain cumulative dose–response information. Control mice received saline injections at the same time points. Pilot data confirmed 30 min to be a time of maximal drug effect.
Local hind paw injections were performed by lightly restraining the mice and injecting 5 μl of drug containing 0.9% NaCl subcutaneously into the central plantar area of the hind paw. For these injections, a 30-gauge needle and a microsyringe were used. Mice recovered for 10 min in their testing enclosures, which was observed to be a time of maximal drug effect.
Mechanical nociceptive thresholds were assayed using nylon von Frey filaments according to the “up–down” algorithm described by Chaplan et al. 36as we have used previously to detect allodynia after chronic opioid administration.14,35In these experiments, mice were placed on wire mesh platforms in clear cylindrical plastic enclosures of 10 cm in diameter and 30 cm in height. After 20 min of acclimation, fibers of sequentially increasing stiffness (0.2–2 g, seven fibers) were applied to the center of the plantar surface of a hind paw just distal to the first set of foot pads and left in place 5 s with enough force to bend the fiber slightly. Withdrawal of the hind paw from the fiber was scored as a response. When no response was obtained, the next stiffest fiber in the series was applied to the same paw; if a response was obtained, a less stiff fiber was next applied. Testing proceeded in this manner until four fibers had been applied after the first one causing a withdrawal response allowing the estimation of the mechanical withdrawal threshold using curve fitting of the response data.37Our index of mechanical OIH was calculated as the percentage decrease in baseline mechanical nociceptive threshold resulting from chronic morphine administration.
Thermal Withdrawal Latency.
Response latencies to noxious thermal stimulation were measured using the method of Hargreaves et al. 38as we have modified for use with mice.14In this assay, mice were placed on a temperature-controlled glass platform (29°C) in a plastic enclosure as described above. After 20 min of acclimation, a beam of focused light was directed toward the same area of the hind paw as described for the von Frey assay. The time to purposeful withdrawal of the foot from the beam of light was measured to 0.1 s. A 20-s cutoff was used to prevent tissue damage. In these experiments, the light beam intensity was adjusted to provide an approximate 10-s baseline latency for the C57BL/6 index strain before morphine treatment, and the same light intensity was used for all subsequent experiments. Two measurements were made per animal per test session.
In Silico Mapping
Association Studies Using Haplotypic Mapping.
Using HapMapper software developed by Roche Bioscience (Palo Alto, CA) and a 158,000 SNP database organized into haplotypic blocks for all strains tested,††30,31we attempted to determine associations between our OIH trait data and individual blocks. Briefly, this approach identifies haplotypic blocks whose pattern of genetic variation correlates with the distribution of trait values among the inbred strains. The actual correlation between trait values (mechanical OIH in this case) and the strain groupings for each haplotype block is determined using analysis of variance–based modeling. The resulting P values are used to rank the strengths of correlation for each block in the database. This technique has been used recently to identify genes associated with a number of different murine phenotypic traits.30At the time of analysis, this haplotypic map contained blocks corresponding to 2,171 genes.
All data are displayed as mean ± SEM unless otherwise noted. Dose–response data were fitted using a sigmoidal function with variable slope (Prism 4; GraphPad Software, San Diego, CA). Where repeated measures were used, analysis of variance was applied with post hoc t testing.
Strain Differences for OIH
Figure 1displays the mechanical baseline nociceptive responses and those observed after 4 days of morphine treatment. The differences in baseline nociceptive responses for the various strains were large, with the most mechanically sensitive strain having a baseline nociceptive threshold 16% of the least sensitive strain. As can also be seen in figure 1, the degree of mechanical allodynia acquired during morphine administration varied for the different strains. Table 1lists the mechanical allodynia caused by chronic morphine administration as a percentage change in baseline thresholds. Although some strains displayed only small mechanical changes, six strains had a greater than 80% reduction in mechanical nociceptive threshold. There was no correlation between baseline mechanical nociceptive thresholds and the degree of OIH developed by the individual strains.
In Silico Mapping Based on Haplotypic Analysis
HapMapper software and an expanded SNP database organized into haplotypic blocks were then used to analyze the data. In this analysis, the distribution of the propensity of the strains to develop mechanical hyperalgesia after morphine administration was compared with the partitioning which would be predicted for the strains given the allelic possibilities for each available haplotypic block. Higher correlation resulted from the strain-specific trait data more closely matching the pattern predicted by the haplotypic alleles for any given block. Figure 2Adisplays the results of the haplotypic mapping as a cartesian plot with the inverse of the P value plotted against the chromosomal location of the blocks. Figure 2Bdisplays the tabular results of the HapMapper program. Note that each of the color coded blocks under “Haplotype” represents the haplotype for one strain of mice, with the leftmost strain being the one with the lowest degree of mechanical OIH (129/SvJ) and the rightmost strain being the one displaying the greatest degree of mechanical OIH (MRL/MpJ). Our most highly correlated block overall, which corresponded to the β2-AR gene (Adrb2), possessed three possible haplotypic alleles represented as either a red, green, or blue block in figure 2B. One of the alleles was possessed by only one strain (NZW/LaCJ). The seven strains having the least tendency to develop mechanical sensitization were of one haplotype, whereas the seven more extensively sensitized strains had the other haplotypic allele. This particular haplotypic block was defined by a group of eight SNPs.
Effects of Selective β2-AR Antagonists on OIH
We next hypothesized that if mechanical allodynia after chronic morphine administration was supported by β2-AR, the administration of a β2-AR antagonist should reduce that allodynia. Control mice and mice rendered allodynic by the administration of morphine for 4 days were injected with various doses of the selective β2-AR antagonist butoxamine. The C57BL/6J strain was chosen for these experiments as this strain is commonly used in nociceptive testing, and mechanical OIH is prominent in this strain. The data presented in figure 3Ashow the dose-dependent reversal of the opioid-induced allodynia without effects of the drug on mechanical withdrawal thresholds in the control mice. The ED50for butoxamine reversal of the allodynia was 0.25 mg/kg (95% confidence interval, 0.08–0.75). Figure 3Bdemonstrates that butoxamine was capable of reversing the thermal manifestations of OIH as well. Experiments using the selective β2-AR antagonist ICI 118,551 provided similar results (data not shown).
OIH in β2-AR Null Mutant Mice
We next sought a pharmacologically independent approach to confirming or refuting a role for β2-AR in supporting the mechanical and thermal manifestations of OIH. In these experiments, β2-AR null mutants and littermate FVB wild-type mice were used. Figure 4presents data demonstrating that the wild-type and null mutant strains had similar baseline mechanical and thermal withdrawal thresholds but that after morphine treatment, the null mutant mice developed no discernible mechanical allodynia or thermal hyperalgesia. The wild-type mice, on the other hand, displayed statistically significant reductions in both the mechanical and thermal indices of OIH. Not all morphine-induced changes were different in the β2-AR null mutants, however. The weight loss which characterizes the chronic treatment of mice with morphine was similar for null mutants and wild-type mice (7.5% vs. 7.8%; no statistical difference).
Effects of Local Injections of β2-AR Ligands
Because reports in the literature suggest that peripherally expressed β2-ARs support states of enhanced mechanical nociceptive sensitivity,39,40we undertook studies directed at determining whether the peripheral administration of butoxamine could reduce mechanical OIH and, conversely, whether a selective β2-AR agonist could cause mechanical allodynia if injected locally. Figure 5Ademonstrates that the injection of a butoxamine into the plantar surface of the mouse hind paw fully reversed the mechanical sensitization in that paw. The 1 μg butoxamine dose chosen had minimal impact when administered systemically (fig. 3). As shown in figure 5, neither saline nor butoxamine had effects on baseline nociceptive thresholds. Figure 5Bshows that the selective β2-AR agonist terbutaline dose-dependently caused mechanical allodynia after local injection, supporting the hypothesis that this receptor is an important regulator of nociceptive sensitivity.
Structure of the β2-AR Haplotype Block Associated with Morphine-induced Allodynia
Figure 6provides a diagram of the haplotype block associated with morphine induced mechanical allodynia. This block begins on the 5′ side of the β2-AR gene on chromosome 18 and extends through the coding region to the 3′ side of the gene. The block contains eight SNPs. The single SNP located within the coding region of the gene is associated with a silent mutation and thus does not change the receptor's amino acid sequence.
The goals of this project were (1) to use in silico techniques to derive predictions as to what specific genes are associated with the propensity of mice to develop OIH, and (2) to pursue one of these predictions to confirm or refute a functional role in supporting OIH. The data presented in figure 1as well as table 1demonstrate strong interstrain differences in the propensity to develop mechanical sensitization. Subsequent in silico genetic mapping identified the β2-AR gene as a candidate gene involved in modulating mechanical OIH in mice (fig. 2). Pharmacologic and β2-AR null mutant mouse experiments provided results consistent with the genetic association (figs. 3–5). Ours were not the first efforts to use genetics to investigate pain-related traits. Although specific genes were not identified, previous studies demonstrated influences of genetics on nociceptive measurements of many types41–43and other consequences of chronic morphine administration, such as tolerance and dependence.26–28,44
The in silico mapping approach used a set of SNPs (158,000 at the time of analysis) organized into haplotypic blocks using an algorithm described by Wang et al. 31The significance of naturally occurring haplotypic blocks with respect to genomics and genomic mapping strategies has been reviewed in recent publications.45,46This technique took advantage of the naturally occurring linkage disequilibrium that exists for the SNPs corresponding to the 2,171 genes within the approximately 75 Mb of the murine genome for which SNP discovery had been undertaken. Because these blocks tend to be small in genomic terms, 39 Kb on average, linkage of phenotypic traits to these haplotypic blocks often results in the identification of specific genes associated with the traits of interest. Figure 2displays our results from these studies. From the group of haplotypic blocks present in the HapMapper database, the one corresponding to the β2-AR was most strongly associated. Although three haplotypes were represented in the 15 strains of mice, 14 strains had one of the two predominant haplotypes. One of the reasons for the high degree of association of this block with mechanical OIH was the partitioning of the strains into nonoverlapping groups by the two predominant haplotypes.
With a spectrum of data in hand implicating β2-AR in OIH, we must ask whether these observations fit with existing literature. One set of observations suggesting β2-AR might be involved in opioid-induced hypersensitivity states involves measurements of increased β2-AR density in the central nervous system after chronic exposure of rats to morphine22–24as well as the up-regulation of guanosine triphosphate binding proteins, the molecules used by β-AR to activate ion channels and second-messenger systems.47In cell culture systems, chronic exposure to morphine enhances stimulated cyclic adenosine monophosphate production by a mechanism involving an increased expression of β2-AR.25Although the mechanism was not well described, it is interesting to note that serum levels of cyclic adenosine monophosphate were significantly increased in rats treated chronically with morphine as well.48
Independently reported animal behavioral data supports possible roles for peripherally expressed β2-AR in states of enhanced pain sensitivity. These data lead us to test the hypothesis that the hind paw injection of β-AR antagonists could reduce mechanical OIH. For example, Khasar et al. 39demonstrated that both epinephrine and the more selective β-AR agonist isoproterenol caused a mechanical hyperalgesia when injected into the hind paws of rats. This sensitization was blocked by β-AR antagonists. In this study, the authors went on to provide evidence that the mechanical sensitization might be related to the sensitization of small diameter dorsal root ganglion neurons. In a later study, Aley et al. 40reproduced the previous data and provided further evidence that β2-AR was the likely receptor subtype responsible for the sensitization. In additional studies, cultured dorsal root ganglion neurons were found to respond to β2-AR stimulation with phosphorylation of extracellular signal–related kinase.40This type of phosphorylation has been linked to enhanced nociception by many laboratories. In addition, β2-ARs have been shown to enhance inflammation in models of arthritis49,50possibly involving enhanced production of tumor necrosis factor α.51
Our own data using the local hind paw injection of butoxamine and terbutaline to decrease and increase mechanical nociceptive sensitivity, respectively, are consistent with the peripheral β2-AR effects outlined above. An additional factor perhaps amplifying the role of peripheral β2-AR in supporting OIH is the increased levels of circulating catecholamines present after the acute administration of morphine and the increases during opioid abstinence in rodents.52–54Although we have focused on the periphery, it is possible that in the setting of OIH, β2-ARs expressed in other locations participate in modulating the nociceptive sensitization as well. Ultimately, our findings will need to be integrated in a model for OIH that includes the other brain and spinal cord level mechanisms which have been described.12,16,19,21
Although the results we obtained are notable for the association made with β2-AR, they are equally notable for the lack of association with genes coding for proteins well demonstrated to modulate OIH. The protein perhaps best associated with the modulation of OIH at this point is the N -methyl-d-aspartate receptor.7,9,10,12,14,17Haplotype blocks pertaining to the genes coding for the principal subunits of this receptor are represented in the database used by the haplotypic in silico mapping program we employed. The lack of a high-strength association with any of these subunits should not be interpreted as inconsistent with the existing pharmacologic data, however. It may be that the N -methyl-d-aspartate receptor is critical in the modulation of the OIH trait, but that the genetic variants of receptor subunits that exist do not lead to important functional differences in the resulting proteins and thus are not likely to be identified in this type of mapping study.
It is also notable that this data set did not lead to the identification of factors influencing the distribution or elimination of morphine, although haplotype blocks pertaining to various drug transporters and metabolic enzymes using morphine as a substrate were included. Simple morphine brain level measurements after acute administration did not predict mechanical OIH (data not shown). It is possible that circulating or brain morphine levels would influence other consequences of chronic morphine administration, however. Similarly, we have completed mapping studies using only morphine. Although other opioids can cause OIH, it is not clear that β2-AR would always play as prominent a role.
A number of factors limit the strength and meaning of these mapping results. The first is that although the sets of SNPs and haplotypic blocks used here were relatively large, they were not comprehensive. SNPs are the most common type of genetic polymorphisms. Although one of the advantages of haplotypic analysis is that not all SNPs need to be known to define the common haplotypes, the map we used was far from complete. Our haplotypic mapping likely analyzed only approximately one tenth of all murine genes at optimal resolution. Therefore, it is possible if not likely that some genes influencing OIH were not identified in these studies. Also, while the number of strains used was relatively large,15many more strains and corresponding SNP data would be needed to have the power to identify all haplotypic blocks associated with complex traits such as OIH, opioid tolerance, or pain sensitivity. An analysis of the strain dependence of power in haplotypic analysis has been published recently.31This analysis suggests that the use of 30–40 strains would very substantially enhance the power of this type of study and will be required for whole genome analyses. The use of 15 strains of mice would result in a power of approximately 0.8 in detecting causal genetic loci having genetic effects in the range of 0.5, i.e. , explaining 50% of the genetic variance. Therefore, continuing to accumulate strain specific data may allow us to identify more genes involved in this complex trait.
A list of the most useful tools and techniques available for neuroscience research would likely include pharmacologic approaches, electrophysiology, molecular biology, the generation of transgenic models, immunohistochemistry, and others. We are poised to add in silico murine genetic studies to this list. In this set of studies, a relatively simple paradigm for measuring OIH in various strains of readily available inbred mice was used to obtain a data set that was directly subjected to in silico genetic analysis. Having demonstrated the ability of β2-AR blockade to reduce or eliminate OIH in mice, we are now in position to translate these findings to human studies. It may be possible to determine the ability of β2-AR blockade to reduce OIH in human models.6–8Conceivably, the addition of β2-AR to a chronic opioid regimen might improve the long-term efficacy of this form of treatment.
The authors thank Dr. Drew Patterson (Assistant Professor, Department of Anesthesiology, Stanford University, Palo Alto, California) and Brian Kobilka (Professor, Department of Molecular and Cellular Physiology, Stanford University) for the kind gift of the β2adrenergic null mutant mice and littermates.