Influence of Nociception and Stress-induced Antinociception on Genetic Variation in Isoflurane Anesthetic Potency among Mouse Strains

All experiments were approved by the local animal care and use committee (Faculty of Science, McGill University, Montreal, Quebec, Canada). Naive, adult mice (6–12 weeks old) of both sexes were used. They included outbred CD-1:Crl mice and inbred (i.e. , genetically identical) mice of the following 10 strains: 129P3, A, AKR, BALB/c, C3H/He, C57BL/6, C57BL/10, CBA, DBA/2, and RIIIS (all J substrains). All mice were bred in our vivarium at McGill University for no more than three generations from breeding pairs obtained from Charles River Canada (Boucherville, Quebec, Canada) or The Jackson Laboratory (Bar Harbor, ME). Mice were housed in standard polycarbonate shoebox cages (2–4 mice/cage), in a temperature-controlled (21°± 2°C) environment and with ad libitum  access to food (Harlan Teklad 8604, Indianapolis, IN) and tap water. Sample sizes were n = 8–33/strain in each experiment for inbred strains and n = 8–65 in each experiment for CD-1 mice, with both sexes equally represented. Sample sizes in drug-treated (saline vs.  naloxone) mice were n = 4–11/strain/drug.

The tail-clip test of Takagi et al.  12was used, which is itself a modification of the tail-pinch test of Haffner.20In one experiment, naive mice of all strains were injected subcutaneously with either naloxone (10 mg/kg; Sigma, St. Louis, MO) or saline (10 ml/kg volume) and returned to their home cage for 20 min before testing. This high naloxone dose was chosen for its demonstrated ability to block opioid receptors of all classic types (μ, δ, and κ). Each mouse was lightly restrained in a cloth/cardboard holder, and an artery clip (exerting approximately 500 g of force) was applied to the tail 1–2 cm from the base. The mouse was immediately removed from the holder onto a table top (in room air) and the latency to lick, bite, or grab the clip (i.e. , the latency to attempt to remove the clip) or even bring the nose to within 1 cm of the clip was measured with a stopwatch to the nearest 0.1 s, after which the clip was immediately removed. A maximum cutoff latency of 120 s was imposed, at which time the clip was removed to prevent injury.

The results obtained with the tail clip described above might be specific to a weaker nociceptive stimulus. To evaluate this possibility, all experiments were repeated in naive mice using a stronger clip, exerting approximately 2,000 g of force. The cutoff latency in these experiments was 60 s.

Studies were performed in an airtight Plexiglas chamber (75 cm × 28 cm × 20 cm) equipped with two rubber flap iris diaphragm air seals to allow the experimenter to slide his or her arm inside. Mice were tested in groups of 10–20. Isoflurane in oxygen was introduced into the chamber (3 l/min) at one end. The concentration of isoflurane was measured with a gas analyzer (Datex Capnomac Ultima; Datex-Ohmeda, Helsinki, Finland) calibrated before use with gas samples provided by the manufacturer. Gas was sampled from two sites at the cage floor level close to the animals and away from the gas inflow site. Rectal temperature was measured intermittently and was maintained at 36.0°–38.0°C using an air heater/fan. The fan served to achieve adequate mixing of gas within the chamber. Mice from multiple strains and both sexes were tested simultaneously to counterbalance strains and sex across sessions.

Starting at 0.62% atm, isoflurane partial pressure was increased in the following sequence: 0.62, 0.75, 0.88, 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, and 1.8% atm. To more rapidly achieve the next desired concentration, we transiently increased the oxygen flow (up to 10 l/min) and vaporizer dial setting (up to 5%). The concentration at each level was kept constant for at least 20 min before behavioral testing to permit equilibration of anesthetic concentration in the mice. Pilot studies revealed that 30- and 40-min equilibration periods yielded equivalent data in CD-1 mice. Testing of individual mice was discontinued after two consecutive concentrations eliciting no response. Testing mice at all these concentrations took 6–8 h/day. However, pilot experiments (data not shown) revealed that using descending or up-down sequences of concentrations yielded highly similar strain means, suggesting that our data were not strongly influenced by either the direction of anesthetic presentation or timing factors.

MAC^LORRwas tested at isoflurane partial pressures ranging from 0.62 to 1.1%.

At each partial pressure, mice were placed in a supine position and observed for 30 s for their ability or inability to fully right themselves (with all four hind paws on the floor).

Minimum alveolar concentration was tested at isoflurane partial pressures ranging from 0.75 to 1.8%. The starting partial pressure for each mouse was determined by that mouse's MAC^LORRalready obtained, i.e. , mice were not tested for tail-clip responses at partial pressures below their individual MAC^LORR. At each partial pressure, the tail clip used in the baseline nociceptive assay (either 500 g or 2,000 g) was applied to the mouse's tail, again approximately 1–2 cm from the base, and mice were observed for 30 s for gross purposeful movement of the head, body, or extremities. We used only one tail-clip intensity (500 g or 2,000 g) per session.

Strain and sex differences in tail-clip response latencies in air were evaluated by analysis of variance, followed by t  tests, Tukey's post hoc  test, or both, as appropriate. MAC^LORRand MAC data were analyzed both by logistic regression and by the bracketing technique described by Sonner,21as the average of the last response/first no-response partial pressures. A criterion α level of 0.05 was used in all cases.

Inbred strains are virtually isogenic within strain, and thus narrow-sense heritability (h  ²) can be estimated from the between-strain genetic variation (V^a) and the within-strain (error) variation (V^e) using the formula h  ²= V^a/(V^a+ V^e),22which is an estimate of the population intraclass correlation coefficient. Because strains were chosen randomly with respect to the trait, these values are likely accurate estimates of the true trait heritabilities.23 

The extent to which one can infer the influence of a common set of genes on two traits can be estimated from the correlation of strain means, assuming that these strains are housed and tested equivalently,23as was the case here. Genetic correlations among (the normally distributed) strain means were assessed using the Pearson product-moment correlation coefficient (r ). Use of the Spearman rank correlation coefficient (not shown) yielded qualitatively similar conclusions.

There were no significant main effects of sex or significant strain × sex interactions for any dependent measure in the strain survey. Therefore, data from both sexes were collapsed for all further analyses. It should be noted, however, that the main effect of sex for both MAC^LORRand 500-g tail-clip MAC approached significance (F^1,197= 2.1, P = 0.14 and F^1,216= 3.0, P = 0.09, respectively), with male mice showing a tendency toward increased anesthetic sensitivity in both cases. For all dependent measures, strain sensitivities were unimodally distributed, with distributions not deviating statistically from normality (not shown).

Latency to attempted removal of the 500-g tail clip was robustly strain dependent (F^10,177= 11.4, P < 0.001), as shown in table 1and as we have previously reported.9The heritability of this trait was estimated as h  ²= 0.39. These responses are very similar to those reported previously in 10 of the same 11 strains using the same 500-g tail clip,9with a correlation between the strain means in the two experiments of r = 0.75 (P < 0.05). In that study, we also showed that there was no correlation whatsoever between tail diameter and response latency.9 

Latency to attempted removal of the 2,000-g tail clip was also strain-dependent (F^10,129= 4.4, P < 0.001), as shown in table 1. The heritability of this trait was estimated as h  ²= 0.25. As might be expected, removal latencies for this stronger clip were generally shorter (grand mean, 13.0 s) than removal latencies for the lighter, 500-g clip (grand mean, 22.6 s), except for CD-1 mice that showed no change and CBA and DBA/2 mice, which had inexplicably longer latencies. Because of these outliers, the genetic correlation between 500-g and 2,000-g clip latencies in all 11 strains was just significant (r = 0.60, P = 0.05).

The mouse strains displayed highly strain-dependent MAC^LORRsensitivity (F^10,208= 11.1, P < 0.001), as shown in figure 1Aand table 1. The heritability of this trait was estimated as h  ²= 0.35.

The mouse strains displayed highly strain-dependent 500-g tail-clip MAC means (F^10,227= 10.5, P < 0.001), as shown in figure 1Band table 1. The heritability of this trait was estimated as h  ²= 0.32.

Similarly, highly strain-dependent 2,000-g tail-clip MAC means were observed (F^10,150= 34.5, P < 0.001), as shown in figure 1Cand table 1. The heritability of this trait was estimated as h  ²= 0.70. As would be expected, MAC values for this stronger clip were higher, in every strain, than the corresponding values using the weaker clip. The genetic correlation between 500-g and 2,000-g clip MAC means in all 11 strains was highly significant (r = 0.72, P < 0.05).

Correlations between awake state (uninjected) tail-clip latencies, MAC, and MAC^LORRare shown in figure 2. Because what are being correlated are strain means, not individual values, these are genetic correlations (with some contamination by nonisogenic CD-1 mice, in which variability may be both genetic and nongenetic) in which high values imply similar genetic determination of correlated traits. A robust and significant (even after false discovery rate correction24) negative genetic correlation (r =−0.76, P < 0.01) was obtained between 500-g tail-clip latency in room air and isoflurane 500-g tail-clip MAC, such that mouse strains showing lower sensitivity (i.e. , longer latencies to attack the clip) in air showed higher sensitivity to isoflurane (i.e. , lower concentrations required to abolish responding). Similarly, a significant negative genetic correlation (r =−0.58, P < 0.05) was obtained between 2,000-g tail-clip latency in room air and isoflurane 2,000-g tail-clip MAC. By contrast, tail-clip latency and isoflurane MAC^LORRwere uncorrelated genetically (r =−0.10 for 500-g clip; r =−0.26 for 2,000-g clip). MAC^LORRand MAC to isoflurane exhibited a modest (r = 0.56 for 500-g clip; r = 0.52 for 2,000-g clip) but nonsignificant positive correlation with each other.

Strain surveys of room air tail-clip latencies were performed as before, but with saline or naloxone pretreatment. For both 500-g and 2,000-g clips, analysis of variance revealed significant main effects of strain, drug, and a strain × drug interaction (all P  values < 0.001). Saline injection itself produced no appreciable effects on response latencies in most strains; correlations with latencies of uninjected mice (table 1) were r = 0.69 and r = 0.86 for the 500-g and 2,000-g clips, respectively. In virtually all strains, mice injected with 10 mg/kg naloxone showed reduced tail-clip response latencies compared with saline-treated mice (figs. 3A and B). However, naloxone treatment did not appreciably alter the relative  sensitivities among strains, as shown by the high correlations between saline- and naloxone-treated strain means (r = 0.90 and r = 0.73 for 500-g and 2,000-g clips, respectively). The amount of tail clip-related SIA blocked by naloxone administration was estimated as the difference between the saline- and naloxone-treated latencies. Quantifying SIA as the percent change relative to saline yielded qualitatively similar results (data not shown). The magnitude of SIA itself correlated highly with uninjected, saline-injected, and naloxone-injected latencies (all r  values > 0.62). Because of this, the genetic correlations between SIA magnitude, MAC, and MAC^LORRwere very similar to those described above (SIA vs.  MAC: r =−0.67 and r =−0.71 for 500-g and 2,000-g clips, respectively; SIA vs.  MAC^LORR: r =−0.21 and r =−0.48 for 500-g and 2,000-g clips, respectively). The correlations between SIA and MAC (fig. 3C) were highly significant even after false discovery rate correction.24 

Significant between-strain differences were demonstrated in tail-clip nociception, tail clip-induced SIA, isoflurane MAC^LORR, and isoflurane tail-clip MAC, representing prima facie  evidence for genetic determination of variability in these traits. Strain means in every case were normally distributed, suggestive of multigenic inheritance. The data suggest that strain differences in isoflurane MAC derive to a considerable extent from strain differences in the underlying sensitivity to the noxious mechanical stimulus and/or are related to the magnitude of SIA produced by stimulus application. Strain differences in isoflurane MAC^LORRshow no such codetermination with nociception or SIA and are likely mediated by separate genetic factors. The possible genetic relation between MAC and MAC^LORRremains unclear, because we obtained suggestive but nonsignificant correlations.

Mouse strain differences in tail-clip MAC have been investigated previously.6,7These studies featured eight strains in common to those tested here (assuming that the CH3/J strain referred to, a designation unknown to us, is indeed the C3H/HeJ strain used here). Mean isoflurane MAC values in these eight strains ranged from 1.30 to 1.66 (median, 1.40). Although the force exerted by the clip used in those investigations was not stated, it might be inferred that their clip was also approximately 2,000 g, because we obtained MAC values in these eight strains using the 2,000-g clip ranging from 0.98 to 1.54 (median, 1.36). Importantly, the two data sets show an almost significant genetic correlation (r = 0.65, P = 0.08). Although our study did not investigate any other inhalant anesthetic besides isoflurane, Sonner et al.  6demonstrated that strain mean MAC values for desflurane correlated significantly with those for isoflurane. In turn, significant correlations between desflurane and nitrous oxide MAC among rat strains have also been demonstrated.5To our knowledge, there are no published reports of inbred strain sensitivity to isoflurane-induced MAC^LORRto compare to the current data.

The very strong correlation between our 500-g and 2,000-g tail-clip MAC values (r = 0.72) suggests that the genetic variation in MAC is not specific to stimulus strength. It is commonly assumed that variability in MAC should decrease as stimulus strength increases, with a “supramaximal” stimulus yielding MAC values that no longer depend on stimulus intensity.3,25Ethical constraints prevented us from testing this hypothesis directly by using a yet-stronger stimulus, because the 2,000-g clip already was producing obvious (but modest) tissue damage. However, although the absolute differences between strains in their room air latencies to remove the 2,000-g clip were smaller than those using the 500-g clip, the isoflurane MAC strain differences using the 2,000-g clip were actually larger than those using the 500-g clip, covering a range of 0.60 and 0.34% atm, respectively (table 1). More to the point, the latencies of uninjected, fully conscious mice to remove the 2,000-g tail clip were obviously far from immediate in all but a few strains.

The decidedly nonzero latencies to remove tail clips in the awake state are at odds with the idea that these are supramaximal stimuli. Naloxone was very effective at reducing these latencies, especially at 2,000 g, suggesting that opioid-mediated SIA produced by application of the clip itself is partly responsible for the noninstantaneous response to the clip. Although easily explainable, the ability of an acute pain stimulus to induce SIA to itself  has actually been rarely if ever directly reported in the pain literature. In the vast SIA literature,26nonpainful stressors (e.g. , forced swim, restraint) are usually employed, and when a painful stimulus such as foot shock is used as the stressor, the nociceptive measurement is generally made at the tail. It has long been known that “pain inhibits pain,” and an animal model of this phenomenon has been developed, called diffuse noxious inhibitory controls .27,28However, in existing behavioral diffuse noxious inhibitory control paradigms, a pain stimulus in one part of the body is demonstrated to inhibit pain in another  part.29,30We should note that most modern studies examining the effect of naloxone treatment on “basal” nociceptive sensitivity in humans and animals have yielded negative findings,31,32and we have found no effect of 10 mg/kg naloxone on 49°C tail-withdrawal test latencies (unpublished data, Shad B. Smith, B.S., and Jeffrey S. Mogil, Ph.D., McGill University, Montreal, Quebec, Canada, November 2001). Incidentally, our unpublished data rule out the possibility that restraint is producing the SIA in question, because mice in that study were restrained for far longer (approximately 1 min) than was the case here (< 10 s), with no evidence of SIA.

Although it is unmistakable that both tail clips produced strong SIA, it should be recognized that even in the presence of naloxone blockade of this SIA, significant (and at the 500-g clip, robust) strain differences in awake nociceptive responsiveness persisted, and those (naloxone-treated) latencies correlated equally highly with tail-clip MAC. Nonopioid forms of SIA are well known,33,34and it is possible that the “residual” nonzero latencies might be due to SIA unaffected by opioid receptor blockade with naloxone. However, because multiple neurochemical systems have been implicated in nonopioid SIA,35this is not a trivial matter to test.

Genetic variability in baseline nociception is highly correlated with genetic variability in SIA; we have noted this correlation before using thermal nociceptive stimuli and drug antinociception.13In turn, both nociception and SIA are genetically correlated with MAC. To the extent that one believes that awake state tail-clip latencies are truly noninstantaneous, our data suggest that either something is wrong with the concept of the “supramaximal” stimulus in basic science studies of anesthesiology, or the stimuli being used in such studies are not actually supramaximal. In the majority of existing studies, the force being exerted is not reported, not known, or both.36Conversely, if one assumes that in the absence of SIA (as would surely be the case during anesthesia) responses to the tail clips would be instantaneous, we have still demonstrated herein that the magnitude of this “unrealized” SIA can predict the variable potency of isoflurane. As such, the genetic factors underlying isoflurane potency have little to do with the anesthetic itself (e.g. , genetic alterations at the anesthetic's molecular target) but instead are related to the operation of brain mechanisms of pain modulation. That is, although it is doubtful that SIA mechanisms are active in unconscious mice, it seems that the analgesic components of both isoflurane anesthesia and the stress response are mediated by similar pathways. This is somewhat surprising, because previous studies of naloxone administration during inhalant anesthesia indicate no effect on MAC.37However, antinociception from electrical stimulation of the periaqueductal gray matter—a phenomenon thought to be mediated by the same mechanisms as SIA38—can decrease anesthetic requirements in humans.39Also, isoflurane at clinically relevant concentrations has been demonstrated to interact in a complex fashion with descending supraspinal mechanisms of pain modulation40in addition to its actions within the spinal cord.

The current findings may have implications for existing and future studies of anesthetic mechanisms in animals, because any number of genes/proteins implicated in anesthetic MAC by pharmacologic and/or transgenic studies4also have demonstrated effects on nociception and/or antinociception per se .41,42We recommend that investigators consider incorporating room air testing of noxious stimulus responsivity into studies of inhalant MAC. To the extent that isoflurane MAC really is a function of nociceptive sensitivity to the stimulus itself (or to SIA that is itself correlated with nociceptive sensitivity), many isoflurane sensitivity genes will also be nociceptive sensitivity genes. We have conducted a systematic study during the past few years of the genetics of nociception in the mouse.43–45These experiments have shown that the modality of the noxious stimulus is the inherited unit, i.e. , different sets of genes are responsible for variable sensitivity to thermal nociception, chemical nociception, mechanical allodynia (increased sensitivity to evoking mechanical stimuli after injury), and thermal hyperalgesia (increased sensitivity to evoking thermal stimuli after injury).8,9,46The status of acute mechanical nociception within this scheme has yet to be clarified, because tail-clip latencies showed very high genetic correlations with most thermal assays, but von Frey filament withdrawal thresholds did not.9We have identified a number of quantitative trait loci containing genes responsible for variability in nociceptive and analgesic sensitivity47–50and, in some cases, have provided evidence in support of particular candidate genes within the genetically linked (i.e. , coinherited) regions.47,50–52A quantitative trait locus mapping study of tail-clip mechanical nociception is ongoing, using an F²hybrid intercross between the A/J and C3H/HeJ mouse strains. As has been pointed out, currently identified genes underlying differences in the sensitivity to some inhalant anesthetic effects in Caenorhabditis elegans  may not be relevant in vertebrates.6 

The discovery of genes relevant to anesthetic potency may have considerable implications for basic scientific research into physiologic mechanisms of anesthesia and ultimately for clinical anesthesia.

The authors thank Sarah Harbottle, B.S. (Department of Psychology, McGill University, Montreal, Quebec, Canada), for preliminary data collection and Samson A. Sotocinal, Ph.D. (Department of Bioresource Engineering, McGill University), for construction of the induction chamber.