Spinal Carbonic Anhydrase Contributes to Nociceptive Reflex Enhancement by Midazolam, Pentobarbital, and Propofol

The study protocols, approved by the Faculty of Medicine Animal Care Committee (University of Calgary, Calgary, Alberta, Canada), were designed to comply with the guidelines of the International Association for the Study of Pain and the Canadian Council for Animal Care.

Male Sprague-Dawley rats (weight, 300–500 g) were instrumented with chronic indwelling intrathecal catheters, with the distal catheter ports located in the lumbar region as previously described. 16,17Briefly, under general anesthesia with halothane (2–3% in oxygen), rats were positioned prone in a stereotactic frame. A length of polyethylene-10 catheter (Becton Dickinson Co., Parsippany, NJ) was inserted through an incision in the atlantooccipital membrane to a position of 8 cm caudal to the cisterna magna at the level of the lumbar subarachnoid space and secured with dental cement. Postoperatively, animals free from motor deficits were allowed to recover from surgery for a week before testing. Animals were housed in groups of two or three, with free access to food and water and a 12-h light–dark cycle. Each animal was used for up to three experiments with a rest of at least 1 week between each experiment. Correct location of catheters was confirmed by injection of methylene blue dye at postmortem examination.

The studies were designed to test three hypotheses. First, we sought evidence that intrathecal pretreatment with CAI would block the nociceptive reflex enhancement associated with systemic anesthetic administration. Second, by administering the CAI into the lumbar cerebrospinal fluid (CSF), we examined whether the block of nociceptive reflex enhancement would occur in the hind limbs but not the forelimbs, suggesting a drug action localized to the lumbar cord or nerves. Finally, we evaluated whether pretreatment with intrathecal CAI produced a similar degree of blockade of the hind limb hyperreflexia in animals sedated with pentobarbital, midazolam, and propofol. In testing the first hypothesis, the role of CA was evaluated as suggested by Maren 18using two CAIs, acetazolamide and ethoxyzolamide. Dose–response relations were evaluated in concentration ranges that are thought to be specific for CAI with few if any other effects. 18Acetazolamide was used in concentrations of 0.2, 2, 20, and 200 μm, and ethoxyzolamide was used in concentrations of 5 and 50 μm.

The observer, blinded to the nature of the intrathecal drug injected, measured paw withdrawal latency (PWL) in the forelimbs and hind limbs using a custom-made Hargreaves box 19as previously described. 7The testing chamber is a box (19 × 28 × 29 cm) with acrylic walls and a glass floor. The floor is heated by a projector lamp bulb (Radius tungsten halogen lamp, model EJY, 19 V, 80 W; General Electric, Glen Allen, VA) that projects through an aperture (5 × 10 mm) in the cover of a housing installed below the glass floor. The circuit of the device consists of a photocell aimed at the aperture from within the housing, the projector bulb, a timer, and a switch. To make a measurement, the observer positions the aperture under the paw and closes the circuit with the switch. When the glass has heated sufficiently to evoke a withdrawal response, the interruption of the light falling on the photocell opens the circuit, turning off the timer and the bulb. With our device, paw withdrawal latencies can be determined within 0.1 s. Paw withdrawal latencies in untreated animals range from 9–12 s, with a cutoff time of 14 s. Measurements were made sequentially in the four limbs at each measurement time described in the next section.

Each rat was allowed to acclimatize in the test chamber until it was resting quietly, which usually took 10–15 min. After control measurements of PWL were recorded, the intrathecal catheters were injected with 10 μl artificial cerebrospinal fluid (aCSF), either plain or containing acetazolamide or ethoxyzolamide. Fifteen minutes after intrathecal drug injection, animals received an intraperitoneal injection of one of the three sedative drugs: pentobarbital (30 mg/kg), propofol (50 mg/kg), or midazolam (1.9 mg/kg). The effects of sedative drug injection in each animal were represented by the mean value of the paw withdrawal latency and the mean sedation score calculated from measurements made 5, 15, 25, 35, 45, and 55 min after the intraperitoneal injection. For each animal, one mean value for the PWL following intraperitoneal drug injection was calculated by averaging the measurements obtained during the 55 min after intraperitoneal drug injection. For comparisons between drugs, the individual PWL measurements were expressed as a percentage of the predrug injection value in each animal and averaged over the 55-min postinjection time.

After each measurement of PWL, the observer determined the presence or absence of the righting reflex and assigned a sedation score. The righting reflex was considered present if the animal made any effort to reassume the prone position within 1 min after having been placed supine. The sedation score was determined by the response to light touch with a fingertip to the whiskers (0 = fully responsive, 1 = slow response, 2 = unresponsive to whisker stimulation).

The dose of pentobarbital (30 mg/kg) used in the current study was the same as that used previously. 7Pilot evaluations suggested that the duration of sedation after doses of midazolam of 1.9 mg/kg and propofol of 50 mg/kg would be similar to that induced by 30 mg/kg pentobarbital. Intrathecal CAI concentrations in animals treated with midazolam and propofol (determined in the initial pentobarbital studies) were 20 μm acetazolamide and 50 μm ethoxyzolamide.

All data are presented as mean ± SD. The effect of time on PWL during the 55-min observation period was evaluated with the Pearson product–moment correlation coefficient. Effects of intrathecal CAI and limb selection were evaluated by paired t  test. Using pentobarbital as the sedative drug, the dose–effect relation between intrathecal CAI concentration and PWL were examined with two-factor analysis of variance (ANOVA) using drug concentration as the first factor and upper versus  lower limbs as the second factor. The effects of intrathecal acetazolamide on nociceptive hyperreflexia in pentobarbital- and propofol-treated animals were compared directly in the same animals using a randomized block design to assign each animal to receive either pentobarbital or propofol. Results from the latter study were analyzed with two-factor ANOVA using drug as the first factor and limb as the second factor influencing the PWL. To compare the effects of intrathecal CAI administration among the three sedative drugs, results from animals that received either 20 μm acetazolamide or 50 μm ethoxyzolamide intrathecally were pooled for each sedative drug and then compared with two-factor ANOVA using the sedative drug as the first factor and limb as the second factor.

Forty-three rats were enrolled in the study. Nine control animals without catheters were given midazolam or propofol without any CAI. Intrathecal catheters were successfully inserted and maintained in 27 animals; 76 studies were performed on these animals. Results are presented as mean ± SD.

Paw withdrawal latency before any drug administration was approximately 11 s and was similar in the forelimbs and the hind limbs (table 1). Intrathecal injection of CAI had only small effects on PWL. The highest concentration of acetazolamide tested (20 μm) had no significant effect on PWL. Intrathecal injection of 50 μm ethoxyzolamide was associated with a 4% reduction in PWL in both forelimbs and hind limbs (table 1;P = 0.01), without a significant difference between the forelimbs and hind limbs (P = 0.377). In two animals, intrathecal injection of 10 μl aCSF containing 200 μm acetazolamide caused agitation; this dose was therefore removed from the protocol. None of the remaining doses of acetazolamide or ethoxyzolamide produced any obvious behavioral effects.

Midazolam, pentobarbital, and propofol had sedative effects during the 55-min observation period after intraperitoneal injection (figs. 1–3). Animals that received pentobarbital were more deeply sedated that those treated with midazolam or propofol (P < 0.001;table 2).

In control animals (no intrathecal CAI treatment), PWL in both forelimbs and hind limbs fell rapidly to 60–70% of control values (fig. 1and table 3) and remained stable during the 55-min observation period. In both the control animals and animals treated with intrathecal CAI, there was no significant correlation between time and PWL during the first 55 min after injection of any of the three sedative drugs (P > 0.3).

In animals pretreated with intrathecal injection of 20 μm acetazolamide or 50 μm ethoxyzolamide, forelimb PWL values were similar to values observed in control animals (table 2). In contrast, in the hind limbs, intrathecal acetazolamide and ethoxyzolamide injection reduced the nociceptive hyperreflexia induced by midazolam and pentobarbital (figs. 2 and 3and table 3). In propofol-treated animals, intrathecal acetazolamide but not ethoxyzolamide decreased hind limb hyperreflexia.

Pretreatment with intrathecal acetazolamide produced a concentration-dependent reduction (P = 0.002) in hind limb hyperreflexia (fig. 4). Both 5 and 50 μm ethoxyzolamide reduced hind limb hyperreflexia (P < 0.01) to a similar degree.

When pentobarbital and propofol were compared directly in the same animals, 20 μm intrathecal acetazolamide blocked hind limb hyperreflexia to a greater degree during sedation with pentobarbital than with propofol (two-factor repeated-measures ANOVA, interaction term, P = 0.003;table 3). Since the temporal profiles of results for 20 μm acetazolamide and 50 μm ethoxyzolamide were similar (figs. 2 and 3), the data were pooled for comparison of the results between midazolam, pentobarbital, and propofol. The effect was significantly less (P = 002) with propofol (65 ± 10% in hind limbs vs . 71 ± 10% in forelimbs) than with either midazolam (67 vs . 87 ± 10%) or pentobarbital (60 ± 9 vs . 83 ± 9%).

Previously, we have reported 7that systemic administration of acetazolamide blunts the nociceptive hyperreflexia induced by pentobarbital without altering the prepentobarbital reflex latency. In the current study, the results show (table 1) that intrathecal administration of 20 μm acetazolamide did not alter forelimb or hind limb reflex latency. Following treatment with pentobarbital, forelimb withdrawal latencies in animals treated with 20 μm acetazolamide (60 ± 8%) were similar to those of control animals pretreated intrathecally with aCSF (59 ± 11%;table 3). In contrast, hind limb reflex latencies in the acetazolamide-treated animals remained at 92 ± 18% of control values (table 3). The observation that lumbar intrathecal injection of acetazolamide blocked pentobarbital-induced hyperreflexia in the hind limbs but not the forelimbs is consistent with a spinal site of action. The results suggest that spinal CA makes a major contribution to the hyperreflexia induced by systemic administration of pentobarbital, midazolam, and propofol. These findings also support a specific physiologic role for spinal CA and bicarbonate ion since the effective concentrations of acetazolamide were within the values conventionally considered to inhibit CA without producing ancillary side effects. 18 

The results following intrathecal injection of ethoxyzolamide show some important differences with those with acetazolamide. Intrathecal injection of 50 μm ethoxyzolamide produced a 10% reduction in reflex latency in both the forelimbs and hind limbs. Since this was not seen with either systemic or intrathecal treatment with acetazolamide, these findings suggest that ethoxyzolamide may have additional effects on reflex latency. The latter may be due to either systemic effects following reabsorption from CSF or to redistribution to CSF that bathes the forelimb neural structures.

The current findings provide support for a role of CA-dependent, GABA^Areceptor–mediated neuromodulation in nociceptive signal transfer in the spinal cord. Although we are not aware of previous reports of this phenomenon in the spinal cord, many laboratories 10,12,13,15,20,21–23have investigated a similar process referred to as the anion-shift hypothesis 10,13in mammalian hippocampal circuits. Briefly, the anion-shift hypothesis states that under conditions that prolong the open time of the GABA^Areceptor channel, there is a shift of ion flux from Cl⁻to HCO³⁻through the channel. This shift results in an HCO³⁻-mediated depolarizing excitatory current that converts the GABA-mediated postsynaptic membrane response from inhibitory to excitatory. The consequence is facilitation of synaptic transmission and synaptic plasticity. 13Extracellular alkaline pH transients lasting approximately 30 s have been observed following GABA^Areceptor stimulation with GABA or muscimol 21in comparison to the excitatory effect that persists for at least an hour in the current study. The GABA^Areceptor–mediated ion shift can be initiated by an activity-related increase in GABA release, by prolonging the open time of the channel pharmacologically or by activation of CA. 9,10,13Voipio and Kaila 14have referred to these modulatory processes as activity-dependent, bicarbonate-mediated excitatory “volume” transmission. Carbonic anhydrase activators enhance, while CAIs impair spatial learning and memory in conscious animals. 24,25Sun and Alkon 10have reviewed recent studies that indicate that in the hippocampus, carbonic anhydrase plays a key role in signal transmission in neuromodulation of signal transmission in hippocampal neural circuits and “functions as an effective attentional gate.” We are unaware of any previous reported evidence supporting CA gating of nociceptive transmission in the spinal cord.

It is of interest that intrathecal CAI had a smaller effect on hind limb nociceptive reflex latency after propofol injection. Although propofol induced a similar magnitude of hyperreflexia as with pentobarbital and midazolam, intrathecal administration of CAI did not reduce the hyperreflexia induced in the lower limbs as much as it did with pentobarbital and midazolam. Since midazolam and propofol produced similar levels of sedation (table 2), the difference in CAI effect on hind limb hyperreflexia in the propofol group cannot be simply explained by the differences in levels of sedation. Midazolam, pentobarbital, and propofol all alter the decay kinetics of GABA^Areceptor–mediated currents by slowing deactivation and/or desensitization of GABA^Areceptors. 26–28Perhaps a detailed understanding of the differences in functional and pharmacological properties among these drugs can help to explain the difference in CAI sensitivity that we have found in the current study.

The findings of the current study may have clinical relevance to the neural mechanism(s) of some aspects of the excitement phase of anesthesia. Excitement during emergence from anesthesia presents a number of management problems, such as arterial hypertension, agitation, delirium, and increased responsiveness to pain. Our findings suggest the latter component may be due to increased transmission of nociceptive impulses achieved by interaction with a CA-mediated gating system similar to that seen in hippocampal pathways. 9We have previously presented evidence 29to support a role for bicarbonate-mediated, CA-dependent mechanisms in the enhancement of synaptic transmission by pentobarbital in rat hippocampus in vitro . The high degree of correlation between the pharmacodynamics of nociceptive reflex enhancement and activation of the hippocampal electroencephalogram during barbiturate sedation is consistent with a common pharmacological mechanism for these two “excitatory” phenomena. 30 

In summary, spinal CA contributes to nociceptive hyperreflexia induced by the systemic administration of pentobarbital, midazolam, and propofol. The smaller effect of spinal CAI in propofol-treated animals may indicate that there are additional important factors mediating hyperreflexia induced by propofol. The current findings are consistent with the concept that pentobarbital, midazolam, and, to a lesser extent, propofol interact with mechanisms of carbonic anhydrase gating of nociceptive impulse transmission.