Isoflurane and Nociception: Spinal α2AAdrenoceptors Mediate Antinociception while Supraspinal α1Adrenoceptors Mediate Pronociception

These experiments were reviewed and approved by the Subcommittee on Animal Studies (Veterans Affairs Palo Alto Health Care System, Palo Alto, CA) and were in accordance with the provisions of the Animal Welfare Act, the Public Health Service (PHS) Guide for the Care and Use of Laboratory Animals, and Veterans Affairs Policy. All neuroablative and immunolesioning experiments were performed in adult male Sprague-Dawley rats (240–260 g) obtained from B&K Universal (Fremont, CA). Additional behavioral studies were performed in adult (20–30 g) male mice. Various genetically engineered mice strains were examined, including the following: (1) D79N mice with a nonfunctioning α^2Aadrenoceptor caused by a point mutation in its gene substituting aspartic acid by asparagine at amino acid residue #79; (2) α^2A−/− null mice with a knockout of the α^2A-adrenoceptor gene; (3) α^2C−/− null mice with a knockout of the α^2C-adrenoceptor gene; and (4) their wild-type controls. All of these strains were on a congenic C57BL/6J background. The α^2B−/− null mice had a knockout of the α^2B-adrenoceptor gene on a hybrid C57BL/6J and 129SvJ background; as their controls we used generationally matched wild-type mice on the same hybrid background (C57BL/6J × 129SvJ). Production of the α^2A−/−, α^2B−/−, α^2C−/−, and D79N mice have been described previously. 11–13 

All behavioral testing was performed in a blinded manner, and the experimental groups were tested simultaneously to ensure identical gas exposure conditions. Using a heating blanket, the tail temperatures were maintained within 0.5°C of 30°C.

Tail-flick latencies were determined from the mean of three (in rats) or two (in mice) consecutive latencies using a tail-flick apparatus (Columbus Instruments, Columbus, OH). A different patch of the middle (rat) or distal (mice) third of the tail was exposed to the light beam each time to minimize the risk of tissue damage. The same light stimulus intensity was used for all experiments in a given strain, having been preset at an intensity that elicited a mean latency of 3.5 s in room air. Latency measurements were taken only when the rat or mouse was calmly resting while being gently held under a towel, and a cutoff time of 10 s was used to prevent tissue injury.

The loss of righting reflex was assessed by placing the rat on its back and determining if the animal could right itself within 60 s.

Behavioral studies were performed in a Plexiglas chamber large enough to contain the tail-flick device and equipped with rubber flap iris diaphragm air seals. Antinociceptive testing was always performed after 30-min isoflurane exposure because we had demonstrated a maximal antinociceptive effect in mice and rats at this time interval (fig. 1). Fresh gas flow (rate varied between 3–10 l/min) was introduced into the chambers via  an inflow port; two fans were used to achieve adequate mixing within the chamber, and gases were purged by vacuum. Oxygen concentration in the chamber was maintained at 30% atmospheres (atm), and isoflurane at 1.2% atm (for intact rats), which is the approximate MAC concentration of isoflurane for both rats and humans. 14,15Because the 1.2% concentration of isoflurane did not have a significant antinociceptive effect on the tail-flick assay in the C57BL/6J mice, a 1.7% isoflurane concentration was used in the mice. Control exposure was with room air. An airway gas monitor (Model 254; Datex, Helsinki, Finland) was used to continuously monitor the concentrations of isoflurane, oxygen, and carbon dioxide in the chamber, and flow rates were adjusted to maintain the desired concentrations. Temperature in the chamber was controlled by a heating blanket, and the tail and rectal temperatures were monitored before each behavioral test. Rectal temperature was maintained within 0.5° of 36.5°C.

The rats were anesthetized with sodium pentobarbital (50 mg/kg, administered intraperitoneally), and a laminectomy was performed at T7–T8. The spinous processes and lamina were removed to expose a circular region of dura approximately 5 mm in diameter. The spinal column was stabilized and the spinal cord completely transected between T7 and T8. Postoperatively, the rats were injected subcutaneously with 5 ml saline and intramuscularly with 3 mg/kg enrofloxacin daily for the next 2 days. Sham-operated rats underwent the same surgical procedure without laminectomy or spinal transection.

Postoperative care included daily manual bladder emptying, soft bedding on the cage floor that was changed daily, with food and water supplied on the floor of the cage to ensure accessibility. Animals were also weighed and washed daily. The rats were monitored for postoperative complications, and there was no evidence of wound or bladder infection, skin breakdown, or excessive weight loss. Drug effects on tail-flick latencies were assessed at 6 and 7 days after surgery (crossover design), and then all rats were immediately killed.

The antidopamine β-hydroxylase-saporin (DβH-saporin) immunotoxin (Advanced Targeting Systems, San Diego, CA) is injected intracerebroventricularly and permanently destroys noradrenergic neurons in the locus coeruleus and the A5 and A7 brainstem nuclei over 14 days. 16,17After DβH-saporin treatment, the tyrosine hydroxylase–positive noradrenergic neurons in the locus coeruleus completely disappear; residual staining is observed in 29% of the A5 and 26% of the A7 noradrenergic neurons. 1Rats were anesthetized with intraperitoneal injection of pentobarbital (50 mg/kg). While the skull was fixed in a stereotaxic apparatus, the animal was injected with DβH-saporin (3 μg/3 μl) or saline (3 μl) into the lateral ventricle as previously described. 1 

After nociceptive testing, all rats were anesthetized with pentobarbital, transcardially perfused, decapitated and the brain removed, fixed, cryoprotected, sliced into 40-μm-thick sections, and every third section of the brainstem (from caudal periaqueductal gray to rostral medulla) was retained for immunohistochemical analysis. Sections were stained using antibodies for tyrosine hydroxylase as previously described, 1and tyrosine hydroxylase–positive neurons in the A5, locus coeruleus, and A7 nuclei were counted, and the aggregate for all sections for each nucleus in each rat was derived. The investigator performing the counting was blinded to the treatment.

Intrathecal injections were performed using a modified single percutaneous injection technique. 18Rats were lightly anesthetized with isoflurane, and then a 27-gauge needle was inserted at the L5–L6 intervertebral space. Intrathecal placement was confirmed by a slight tail twitch. All rats were injected with a 10-μl volume of either drug or vehicle and, after allowing 20 s for the injectant to disperse, the needle was slowly withdrawn. While developing this technique we injected 20 μl of 5% lidocaine intrathecally in six rats, which consistently caused a transient hind-paw paralysis.

To perform intracerebroventricular injections, the rats were cannulated stereotaxically (Plastics One Inc., Roanoke, VA) in the left lateral ventricle as previously described. 19Rats were injected with a 10-μl volume of either drug or vehicle over a 2-min period using a Harvard 22 infusion pump (Harvard Apparatus Inc., South Natick, MA). At the completion of the behavioral experiment, the correct placement of the cannula was confirmed histologically in each rat with Evans blue dye.

To demonstrate that isoflurane-induced prolongation of tail-flick latency was not caused by loss of consciousness, the time course for the development of isoflurane-induced hypnosis and antinociception were established. Loss of righting reflex and tail-flick latencies were measured in rats during exposure to air and then after 5, 10, 15, 30, and 90 min of isoflurane (1.2% atm) exposure.

To determine whether isoflurane-induced antinociception was mediated at a spinal or supraspinal level, rats underwent either spinal transection or sham surgery. Six days later, baseline tail-flick latencies were determined, and after 30 min exposure to isoflurane, the tail-flick latencies were repeated. Three different concentrations of isoflurane were tested on consecutive days: 0.8, 1.0, and 1.2% atm.

The contribution of noradrenergic brainstem neurons to isoflurane-induced antinociception was determined by immunolesioning the noradrenergic neurons. Rats were injected with the immunotoxin DβH-saporin (3 μg/3 μl) or with saline (administered intracerebroventricularly). Fifteen days after the intracerebroventricular injection, the baseline tail-flick latencies were determined, and after a 30-min exposure to either isoflurane or air, the tail-flick latencies were repeated. After the tail-flick measurement, the rat was anesthetized, and the brain was harvested and immunostained for tyrosine hydroxylase to evaluate the efficacy of noradrenergic lesioning.

To evaluate α¹-adrenoceptor modulation of isoflurane-induced antinociception and to determine the site of this action, intact and spinal-transected rats were tested for baseline tail-flick latencies and then injected with either the α¹-adrenoceptor antagonist prazosin (2 mg/kg administered intraperitoneally; Sigma Chemical, St. Louis, MO) or vehicle (20% 2 hydroxypropyl-b-cyclodextrin in sterile water; RBI, Natick, MA). After 55 min in air, the tail-flick latencies were repeated. The next day the same procedure was repeated, except that 25 min after the injection of prazosin or saline, the rats were exposed to isoflurane (1.2% atm for intact rats, 0.8% atm for spinal-transected rats). After 30 min of isoflurane exposure, the tail-flick latencies were repeated. Another cohort of rats was tested for baseline tail-flick latencies and then intrathecally injected with either prazosin (30 μg/10 μl) or vehicle. Ten minutes after injection, the rats were exposed to either isoflurane (1.2% atm) or air for 30 min, and the tail-flick testing was repeated. This protocol was also used in another cohort of rats to examine the effects of intracerebroventricular injections of prazosin (30 μg/ 10 μl) or vehicle on tail-flick latency response to isoflurane. Doses of prazosin in the range of 8–30 μg/μl have been previously reported to completely block the effects of endogenous and exogenous α¹-adrenoceptor agonists when administered via  intrathecal and intracerebroventricular routes. 6,20,21 

To determine the dependence of isoflurane-induced antinociception on α²adrenoceptors, the α²-adrenoceptor antagonist yohimbine (2 mg/kg for intraperitoneal administration, 30 μg/10 μl for intrathecal and intracerebroventricular administration; Sigma Chemical) or vehicle (0.9% saline) was administered to intact and spinal-transected rats as described above for prazosin. Doses of yohimbine in the range of 8–30 μg/μl have been previously reported to completely block the effects of endogenous and exogenous α²-adrenoceptor agonists when administered via  intrathecal and intracerebroventricular routes. 6,21–24 

To identify the α²-adrenoceptor subtype mediating isoflurane antinociception, tail-flick testing was performed in α^2A−/−, α^2B−/−, α^2C−/−, D79N mice, and in their respective wild-type controls. Baseline latencies were determined in the gas chamber during room-air conditions; thereafter the mice were removed from the chamber, which was then equilibrated with the test gas mixture (1.7% atm of isoflurane). After 30 min, the mice were placed back in the gas chamber and exposed to the test gas mixture for 30 min. The nociceptive testing was then repeated.

All data are presented as mean ± SEM, and differences are considered significant at P < 0.05. Tail-flick latencies were compared using paired (air vs.  isoflurane) and unpaired (saline vs.  DβH-saporin) t  tests. A repeated-measures analysis of variance was used to test for the development of isoflurane antinociception over time, and a paired t  test was used to test for latency differences from baseline. A one-way analysis of variance was performed on the tail-flick latencies when comparing groups of mice, and an unpaired t  test was used to test for contrasts. Isoflurane antinociceptive effects for the tail-flick assay are shown as the percentage of the maximum possible effect:

After a 5-min exposure to isoflurane (1.2% atm), all rats had loss of righting reflex, but there was no significant effect on tail-flick antinociception (fig. 1). Isoflurane-induced antinociception for tail flick gradually developed over 30 min, achieving significance after 10 min.

After spinal cord transection, the antinociceptive effect of isoflurane was dramatically enhanced compared with sham-operated control rats (fig. 2A). When the rats were exposed to a 0.8% atm concentration of isoflurane for 30 min, there was a 770% increase in antinociceptive effect after spinal cord transection. A smaller increase in antinociceptive effect after spinal transection was observed with a 1.0% concentration (580%) and a 1.2% concentration (250%) of isoflurane.

Figure 2Billustrates that the antinociceptive effect of isoflurane (1.2% atm, 30 min) was significantly enhanced in the DβH-saporin–treated rats compared with saline-treated rats. There was no difference in baseline tail-flick latencies between control and immunolesioned rats.

Systemic administration of the α¹-adrenoceptor antagonist prazosin (2 mg/kg, administered intraperitoneally) had no effect on nociceptive thresholds in the intact or spinal-transected rats when tested in air (fig. 3A). Prazosin enhanced isoflurane-induced antinociception by 132% in the intact rats but had no effect in the spinal-transected rats (fig. 3B). When prazosin (30 μg/10 μl) was administered intrathecally or intracerebroventricularly in intact rats, it had no antinociceptive effect in air (fig. 3C). Intrathecal prazosin had no effect on isoflurane-induced antinociception, but intracerebroventricular prazosin enhanced isoflurane antinociception by 82% (fig. 3D).

Systemic administration of the α²-adrenoceptor antagonist yohimbine (2 mg/kg, administered intraperitoneally) reduced tail-flick latencies in the intact rats when tested in air (−14% maximum possible effect) but had no effect in the spinal-transected animals (fig. 4A). Systemic yohimbine completely blocked isoflurane-induced antinociception in the intact rats but had no effect in spinal-transected rats (fig. 4B). When yohimbine (30 μg/10 μl) was administered intrathecally or intracerebroventricularly in intact rats, it had no effect on nociceptive latencies in air (fig. 4C). Figure 4Dillustrates that intrathecal yohimbine reduced isoflurane-induced antinociception by 72% (P < 0.05), but intracerebroventricular yohimbine did not significantly reduce isoflurane antinociception (P = 0.16).

The antinociceptive effects of isoflurane with the tail-flick assay in the α^2A−/−, α^2B−/−, α^2C−/−, D79N mice, and in their respective wild-type controls are shown in figure 5. The isoflurane antinociceptive response was reduced by 54% in the α^2A−/− mice and 62% in the D79N mice (vs.  wild-type controls;P < 0.05), indicating that the α^2A-adrenoceptor subtype partially mediates isoflurane antinociception. Although the isoflurane-induced antinociceptive effect in the α^2B−/− mice was less than in the corresponding wild-type cohort, this was not significantly different (P = 0.10).

In attempting to elucidate the mechanisms for the analgesic effect of anesthetic agents, it is important to obviate any effect that loss of consciousness can exert on the antinociceptive assay. We previously reported dissociation between N²O-evoked antinociception and N²O-induced hypnosis since these occurred at different concentrations; this enabled us to distinguish the mechanism underlying these effects in experimental paradigms. However, for isoflurane, the hypnotic effect develops at a lower concentration than the antinociceptive effect. 25To separate these two effects, we used the radiant heat tail-flick assay, which measures the latency of a spinal withdrawal reflex to noxious heat, which is independent of hypnotic-induced decrement in purposeful movement.

Figure 1illustrates that continuous isoflurane inhalation (1.2% atm) gradually increased the tail-flick latency over a 30-min period. Numerous studies have shown that isoflurane, at concentrations up to 2.2% atm, has no effect on spinal or peripheral nerve evoked compound motor nerve action potential amplitudes or compound muscle action potential amplitudes, 26–28thereby negating motor paralysis as a mechanism for change in latency. The inhibitory effect of isoflurane on tail-flick latencies was not a result of the hypnotic–sedative effects of isoflurane, because the tail-flick latency was unchanged after 5-min exposure to isoflurane, at which time all the rats were unconscious and unable to right themselves when laid on their backs. Furthermore, after complete spinal cord transection, the tail-flick withdrawal response was intact, and the isoflurane analgesic effect was greatly enhanced compared with sham-operated controls (fig. 2A). Although the tail-flick response is clearly a spinal reflex, it is modulated by brainstem neurons with descending spinal projections that facilitate (decrease the latency) and inhibit (increase the latency) nociceptive processing in the spinal cord. The enhanced isoflurane antinociceptive effect we observed after spinal transection suggests that isoflurane evoked a descending pronociceptive effect on the tail-flick assay. Spinal transection blocked the descending facilitory effect of isoflurane, thus unmasking the intrinsic spinal antinociceptive effect of isoflurane on the tail-flick reflex.

Using DβH-saporin to selectively destroy the brainstem noradrenergic neurons, we observed an enhanced isoflurane antinociceptive effect, indicating that these neurons contribute to the pronociceptive effect of isoflurane (fig. 2B). Furthermore, the systemic administration of the α¹-adrenoceptor antagonist prazosin also enhanced the antinociceptive effect of isoflurane, and this effect was lost after spinal transection (fig. 3B), suggesting that activation of α¹adrenoceptors evoke the pronociceptive effects of isoflurane via  descending spinal pathways. Pronociceptive actions have been attributed to both supraspinal 6,7and spinal 7,23,29α¹adrenoceptors when activated by endogenous or exogenous adrenergic ligands. We demonstrated that supraspinally administered prazosin enhanced isoflurane-induced antinociception, but no effect was observed with the spinal administration of prazosin (fig. 3D), indicating that supraspinal α¹adrenoceptors mediate the isoflurane pronociceptive effect.

Although the predominant effect of isoflurane activation of the brainstem noradrenergic neurons is clearly pronociceptive, there is also an antinociceptive effect. Systemic administration of the α²-adrenoceptor antagonist yohimbine inhibited the antinociceptive effect of isoflurane, and this effect was lost after spinal transection (fig. 4B), suggesting that norepinephrine released by brainstem neurons acted on α²adrenoceptors to evoke an antinociceptive effect via  descending spinal pathways. Furthermore, spinally administered yohimbine reduced isoflurane-induced antinociception, but no effect was observed with the supraspinal administration of yohimbine or in the presence of spinal transection (fig. 4D), indicating that spinal α²adrenoceptors mediate an isoflurane-evoked descending spinal antinociceptive effect, similar to N²O-evoked analgesia. 2 

Pharmacologic evidence in rats 30,31and studies using D79N mice with dysfunctional α^2Aadrenoceptors 32–34have established that the α^2Asubtype mediates the antinociceptive response to α²-adrenoceptor agonists. Now we observe that mice deficient in the α^2A-adrenoceptor subtype or D79N mice with dysfunctional α^2Aadrenoceptors showed reduced isoflurane-induced antinociception, indicating that the spinal α^2A-adrenoceptor subtype mediates an antinociceptive response to isoflurane.

We described a noradrenergic-supraspinal α¹-adrenoceptor pathway that mediates the pronociceptive effect of isoflurane. In addition, we observed a noradrenergic-spinal α²-adrenoceptor pathway mediating an antinociceptive isoflurane effect. These opposing isoflurane effects are consistent with behavioral and electrophysiologic evidence that stimulating the pontine noradrenergic neurons simultaneously activates α¹-adrenoceptor pronociceptive and α²-adrenoceptor antinociceptive pathways. 6,23,29,35It has been reported that low concentrations of isoflurane have a pronociceptive effect on hind-paw withdrawal latencies in rats. 5When the isoflurane concentration is selectively lowered in brain, the systemic isoflurane concentration required for MAC is reduced by 43% in goats. 4These investigators concluded that a 1-MAC isoflurane concentration in the brain actually increased the systemic MAC requirement, which could indicate a supraspinal isoflurane pronociceptive effect.

The enhanced antinociceptive action of isoflurane after spinal cord transection is a novel finding, but several other lines of investigation support the hypothesis that a component of isoflurane-induced antinociception is generated intrinsically within the spinal cord neurons by a mechanism that is independent of cortical inhibition or descending spinal pathways. Isoflurane tail clamp MAC was unchanged in rats undergoing forebrain aspiration, indicating that isoflurane anesthetic action is primarily mediated in the midbrain or lower. 36Furthermore, when only the brain of the goat is perfused with isoflurane, the concentration required for MAC is increased 140%, another indication that isoflurane-induced MAC has an intrinsic spinal mechanism. 37In another study, 1 h after a spinal cord freeze lesion, the isoflurane MAC (measured by hind-paw withdrawal after tail clamp) modestly changed from 1.26 to 1.03% atm. 38Isoflurane also concentration-dependently (0.2–1.3%) inhibited the dorsal root evoked slow ventral root response in isolated neonatal rat spinal cord, electrophysiologic evidence of an isoflurane-evoked intrinsic spinal mechanism of nociceptive inhibition. 39 

In summary, the data from this investigation supports the hypothesis that, at clinically relevant concentrations, there are at least three concurrent components whereby nociception is modulated by isoflurane (fig. 6): (1) a noradrenergic neuron-supraspinal α¹-adrenoceptor–mediated descending pronociceptive mechanism; (2) a noradrenergic neuron-spinal α²-adrenoceptor–mediated descending antinociceptive mechanism; and (3) an intrinsic nonadrenergic spinal mechanism of analgesia.

Our interpretation of these data are that the α¹-adrenoceptor–mediated descending pronociceptive effect of isoflurane is much greater than its α²-adrenoceptor–mediated descending antinociceptive effect; thus, the combined descending spinal effect is predominantly pronociceptive. Normally, the aggregate descending spinal pronociceptive effect of isoflurane is concealed by the more potent intrinsic spinal cord antinociceptive action of isoflurane, which has no adrenergic basis. Noradrenergic lesioning or spinal cord transection eliminates the aggregate descending spinal pronociceptive effect, thus appearing to enhance the intrinsic spinal isoflurane antinociceptive action. The complexity of these interactions suggests that the mechanisms and sites of anesthetic action will not be easily defined.

The authors thank Rekha R. Rapaka, B.S. (Undergraduate Research Student, Biological Sciences, Stanford University, Stanford, CA), and Rebecca K. Berquist, B.A. (Undergraduate Research Student, Human Biology, Stanford University, Stanford, CA), for their invaluable technical assistance; and Lee E. Limbird, Ph.D. (Professor, Department of Pharmacology, Vanderbilt University, Nashville, TN), for providing us with the D79N transgenic mice.