Sedative but Not Analgesic α²Agonist Tolerance Is Blocked by NMDA Receptor and Nitric Oxide Synthase Inhibitors

The experimental protocol was approved by the Animal Care and Use Committee of the Veterans Affairs Palo Alto Health Care System, Palo Alto, California. Male Sprague-Dawley rats (B&K, Fremont, CA) weighing 250–350 g were used. The rats were stratified to match the distribution of the weights in the groups as closely as possible. All tests were performed between 10 am and 6 pm. The number of animals for each experiment is listed in the figure legends.

Rats were made tolerant to the anesthetic action of an α²agonist, dexmedetomidine, as previously described. 29Briefly, dexmedetomidine was chronically administered to rats using Alzet^®osmotic minipumps (model 2002 or 1007D; Alza, Palo Alto CA), which discharge their contents at a mean pumping rate of 0.48 ± 0.02 μl/h. The pumps were inserted subcutaneously during isoflurane anesthesia in the dorsal thoracic region and loaded to administer 5 μg · kg⁻¹· h⁻¹for 7 days to induce hypnotic tolerance and 10 μg · kg⁻¹· h⁻¹for 14 days to induce analgesic tolerance. These dosing schedules were previously found to be optimal to generate a tolerant hypnotic or analgesic state. 28When MK-801 or Nω -nitro-l-arginine (NO²-arginine) was coadministered with dexmedetomidine, these compounds were included in the same pump. Previous studies 28have reported that the behavioral responses measured in sham surgery animals did not differ from rats implanted with pumps containing saline; therefore, the former were used. The appropriateness of substituting sham surgery for saline-filled pumps was also confirmed for analgesia studies. For the MK-801 and NO²-arginine hypnotic experiments, the pumps were removed 1 day before behavioral testing. This was to ensure that the effects observed were caused by the challenge dose of dexmedetomidine. Preliminary experiments found that removing the pumps did not qualitatively alter the results. In all other experiments, the pumps were not removed before testing.

Hypnotic response to dexmedetomidine, was defined by the loss of the rat’s righting reflex, and its duration was measured in minutes and referred to as sleep time. The duration of the loss of righting reflex was assessed as the time from the rat’s inability to right itself when placed on its back until the time when it spontaneously reverted completely to the prone position. The hypnotic response test was performed between 10 am and 6 pm as described previously. 29 

Nociceptive response was assessed by the tail-flick response to a noxious thermal stimulus 40 min after administration of the dexmedetomidine challenge dose. A high-intensity light beam was focused on the tail, and the time for the rat to move its tail out of the light was recorded as tail-flick latency. This method has been described in the previous report. 6The latencies from three sites on the tail were averaged. A cut-off time of 10 s was predetermined to prevent tissue damage. Baseline measurements consisted of a set of three tail-flick determinations at 2-min intervals. Baseline tail-flick latencies ranged between 3 and 4 s.

The NOS inhibitor NO²-arginine (Sigma, St. Louis, MO) and the NMDA antagonists MK-801 (RBI, Natick, MA) and ketamine (Sigma) were diluted in normal saline and were acutely administered intraperitoneally or chronically by means of Alzet^®osmotic minipumps (model 2002 or 1007D).

To minimize the number of pumps implanted into a rat, it was desirable to determine whether two drugs could be effectively administered in one pump. To test whether the concentrated solutions of MK-801 and NO²-arginine had an effect on the chemical stability of dexmedetomidine, MK-801 (9.6 mg/ml) or NO²-arginine (5 mg/ml) were mixed with dexmedetomidine (6 mg/ml) and incubated at 37°C for 14 days. These solutions were then diluted 40-fold to achieve a 150-μg/kg dose of dexmedetomidine, and these were injected into drug-naïve rats. The duration of loss of righting reflex was measured.

Loss of righting reflex and tail-flick data were analyzed using analysis of variance followed by post hoc  Bonferroni tests or the Dunnett multiple comparison test or t  test when appropriate.

The dose–response relations for the hypnotic and analgesic effect of dexmedetomidine are shown in figure 1. The hypnotic effect of dexmedetomidine dose dependently increased in sham-treated animals but was almost completely absent in rats treated for 7 days, even when large doses were administered (fig. 1A). A biphasic dose–response curve for this action of dexmedetomidine, with a maximal efficacy of approximately 300 μg/kg, has been previously shown and has been shown to be caused by a stimulatory action of dexmedetomidine mediated by activation of α¹receptors. 30Exposure to dexmedetomidine for 14 days shifted the analgesic dose–response curve for dexmedetomidine approximately twofold and reduced the maximal effect (fig. 1B).

Premixing of dexmedetomidine with MK-801 or NO²-arginine followed by incubation at 37°C for 14 days to simulate the situation of these drugs being coadministered by one pump did not decrease the effectiveness of dexmedetomidine to cause sleep (fig. 2). This shows that dexmedetomidine remains chemically intact after prolonged exposure to these agents and documents the validity of using a single pump to deliver two drugs in further experiments.

Acute administration of MK-801 to drug-naïve animals did not affect sleep time (fig. 3A). When tolerance had developed, acute administration of MK-801 did not affect the expression of tolerance (fig. 3B). Chronic administration of MK-801 had no effect on the sleep time induced by acute administration of dexmedetomidine (fig. 3C), but coadministration of MK-801 with dexmedetomidine was able to prevent the development of tolerance (fig. 3D). In this experiment, the osmotic pumps were removed 1 day before behavioral testing.

Ketamine had a similar profile in that when acutely administered, it did not affect the sleep time caused by acute injection of dexmedetomidine (fig. 4A). Acute administration of 10 or 20 mg/kg ketamine could not reverse tolerance that was previously established (fig. 4B), but ketamine reversed tolerance to the sedative actions of dexmedetomidine when administered concurrently (fig. 4C). This same dose of ketamine had no effect by itself. In this experiment, the Alzet pumps were left in place for behavioral testing.

When administered to drug-naïve animals, NO²-arginine increased sleep time only at high doses (fig. 5A). When tolerance had developed by 7-day administration of dexmedetomidine, acute administration of low doses of NO²-arginine that did not affect sleep time in drug-naïve animals did not reverse its expression (fig. 5B). When NO²-arginine was coadministered with dexmedetomidine, the induction of tolerance to the hypnotic effects of dexmedetomidine was attenuated (fig. 5C). Treatment with 1.25 μg · kg⁻¹· h⁻¹NO²-arginine only (last column) did not affect dexmedetomidine sleep time.

Acute administration of MK-801 15 min before the challenge dose of dexmedetomidine suppressed its analgesic action at the two highest doses (fig. 6A). To determine whether MK-801 could affect the expression of tolerance, it was acutely administered to control and tolerant animals. A low dose of MK-801 that alone did not have any affect on dexmedetomidine-induced analgesia did not reverse the expression of tolerance (fig. 6B). A higher dose of MK-801 (400 μg/kg) that antagonized the analgesic action of dexmedetomidine in control animals was also unable to reverse tolerance. Coadministration of a dose of MK-801 that prevented the development of tolerance to the hypnotic effect of dexmedetomidine and was the maximum dose tolerated by the rat had no effect on the development of analgesic tolerance (fig. 6C). This dose of MK-801 had no effect on tail-flick latency when administered alone (data not shown).

Acute administration of NO²-arginine did not affect the analgesic action of dexmedetomidine (50 μg/kg intraperitoneally;fig. 7A). To determine whether NO²-arginine could affect the expression of tolerance, it was acutely administered to control and tolerant animals. NO²-arginine (1 and 20 mg/kg intraperitoneally) did not reverse the expression of tolerance (fig. 7B). Coadministration of a dose of 4 μg · kg⁻¹· h⁻¹NO²-arginine, a dose that prevented the development of tolerance to the hypnotic effect of dexmedetomidine, had no effect on dexmedetomidine-induced analgesia in control animals and had no effect on analgesic tolerance (fig. 7C). This dose of NO²-arginine had no effect on tail-flick latency when administered alone (data not shown). Increasing the dose of NO²-arginine further to 8 μg · kg⁻¹· h⁻¹was also ineffective in reversing tolerance.

This study shows that although NMDA receptors and NOS have a role in the development of tolerance to the hypnotic effect of dexmedetomidine, they do not seem to play a significant role in the development of analgesic tolerance. Neither NMDA receptor nor NOS inhibitors can affect the expression of hypnotic and analgesic tolerance. This indicates that there are likely multiple mechanisms by which behavioral tolerance to α²agonists is achieved. We have previously shown that acute administration of the L-type Ca²⁺channel antagonist nifedipine normalized the hypnotic response to dexmedetomidine in the α²-tolerant rats, but coadministration was ineffective in preventing the induction of this tolerance. 31 

Our previous work has elucidated key sites for the intracellular signaling of the hypnotic and analgesic action of α²agonists. The locus ceruleus (LC) seems to be a pivotal site for producing α²-agonist–induced hypnosis. 32However, the analgesic effects of α²agonists are mediated spinally, supraspinally, and peripherally. 6,33,34The analgesic action of dexmedetomidine injected directly into the LC results in activation of α²adrenoceptors in the spinal cord because this analgesia can be blocked by intrathecal injection of the α²antagonist atipamezole and also by intrathecal administration of pertussis toxin, 6which ribosylates and thereby inactivates defined species of G proteins. Intrathecal administration of α²agonists, such as clonidine 35,36or dexmedetomidine, 6,37also produce analgesia. These data suggest that spinal α²adrenoceptors are a final common pathway in antinociception.

A possible reason for the differences in the development of tolerance for the two behavioral responses may be the fact that the anatomic sites at which the responses are transduced are not the same. Biochemical changes associated with the hypnotic tolerant state have been found in the LC. Our previous studies of the LC 38showed that tolerance to the hypnotic response was associated with changes in α²adrenoceptor affinity, ribosylation of guanine nucleotide regulatory proteins (G proteins) ex vivo , and a decreased sensitivity of forskolin-stimulated adenylyl cyclase to inhibition by dexmedetomidine. Although we assume that tolerance is dependent on these changes, they have not been established definitively. The fact that compared with hypnotic tolerance, analgesic tolerance is achieved only with extended administration of high doses of dexmedetomidine and has different pharmacology may be because of the multiple sites of analgesic action. Therefore, although supraspinally mediated processes may be attenuated, the signaling may still be intact at analgesic sites in the spinal cord and periphery.

We have characterized some of the biochemical changes associated with tolerance in the LC, but we have not yet characterized what transpires in the spinal cord, the final common pathway for analgesic mechanism. Profound tolerance to the analgesic action of intrathecally administered dexmedetomidine has been shown, 39indicating that changes in spinal cord play a role in analgesic tolerance. A full understanding of the process underlying α²analgesic tolerance will depend on a more detailed study of the biochemical processes involved.

The α²adrenoceptor system is thought to have many similarities with the opioid receptor system based on their similarities in physiologic actions in the LC 40,41and spinal cord 42,43and in their actions on nociceptive processing in the spinal cord. 44,45Experimental findings from models of opioid tolerance were initially hypothesized to be directly relevant to the situation with the α²adrenoceptor. However, in contrast to α²analgesic tolerance, development of opioid analgesic tolerance is sensitive to NMDA receptor antagonists 20–23and NOS inhibitors. 24,25,46Opioid tolerance develops, at least in part, at the spinal level because spinal administration of morphine is reversed by coadministration of MK-801 47,48and with the presumably selective neuronal NOS inhibitor, 7-nitroindazole. 48The lack of effect of NMDA and NOS inhibitors in α²analgesic tolerance indicates that the mechanisms underlying α²and opioid tolerance are likely different. This could explain why in some studies, no analgesic cross-tolerance between opioid and α²agonists was found, 49,50although cross-tolerance for opioid and α²agonists has been observed for peripheral antinociception. 51Others have found evidence that even in a cell culture system in which both opioids and α²agonists inhibit cAMP production, opioid and α²tolerance do not share common pathways. 27Therefore, even two drug classes apparently sharing a common second messenger system may have different mechanisms of tolerance.

Data from these studies with the pharmacologic probes (MK-801, ketamine, and NO²-arginine) strongly suggest that tolerance has at least two distinct phases (induction and expression) as is seen in other forms of synaptic plasticity (e.g. , long-term potentiation), and that the components of the signaling pathway follow in an orderly temporal sequence. The fact that the efficacy of α²agonists may decrease over time and be differentially affected by coadministration of other classes of drugs may also have clinical ramifications for intensive care unit sedation.

The authors thank the Medical Research Council, London, United Kingdom, and the National Institutes of Health, Bethesda, Maryland.