Isoflurane Differentially Modulates Medullary ON and OFF Neurons While Suppressing Hind-limb Motor Withdrawals

This study was approved by the University of California, Davis, animal care and use committee. Experiments were conducted in 19 isoflurane-anesthetized adult male Sprague-Dawley rats (weight, 375–550 g), housed individually on a 12-h light–dark cycle (lights on at 07:30) and given free access to food and water. Animals were not given neuromuscular blocking drugs at any time.

Rats were placed in an acrylic box, and anesthesia was induced with 4% isoflurane (Minrad Inc., Buffalo, NY). The rat was then placed on mask anesthesia (2.0%), and a tracheostomy was performed. A 14-gauge catheter was inserted into the trachea, and the animal was mechanically ventilated with isoflurane mixed in 100% O²for the remainder of the experiment. Jugular vein and carotid artery cannulations were performed for fluid administration and blood pressure monitoring, respectively. We continuously monitored rectal body temperature and blood pressure (model PB-240; Puritan-Bennett Corp., Hazelwood, MO). Endtidal carbon dioxide and anesthetic concentration were also monitored throughout the experiment with a calibrated Ohmeda Rascal II analyzer (Helsinki, Finland). Inspired and expired anesthetic concentrations were always within 0.1% of one another during the time of testing. Mean arterial pressure was always maintained above 80 mmHg with lactated Ringer’s solution (Abbott Laboratories, Chicago, IL) when necessary. Rectal temperature was maintained at 36.5 ± 1°C with a lamp connected to a variable power supply.

A midline craniotomy 2–3 mm in diameter was performed immediately caudal to the lambda to permit electrode insertion into the RVM. Two midline incisions were made to expose the C2 and S1–S2 spinous processes, to which vertebral clamps were later applied to stabilize the animal during electrophysiologic recording. After the animal’s iMAC determination (see MAC Measurement), the rat was then placed in the stereotaxic frame (Trent H. Wells, South Gate, CA) and affixed via  the vertebral clamps and ear bars.

Each animal’s iMAC was determined after all surgical procedures, before electrophysiologic recording. We allowed the animal to equilibrate to an anesthetic concentration that corresponded to approximate MAC values for isoflurane (1.1–1.3%). The iMAC value was determined by applying a supramaximal mechanical stimulus (30 cm hemostat that delivered 1.7 N/mm²) midway down the length of the tail. The clamp was applied and oscillated (rotated back and forth approximately ±30°) at approximately 2 Hz for 1 min or until gross purposeful movement was observed during the 1 min of clamping. 10Head turning toward the stimulus, multilimb movement, or both were interpreted as a positive response, whereas single limb withdrawals and tonic limb or neck extensions were interpreted as a negative response. 10Depending on the response, the anesthetic concentration was increased or decreased by 0.2%. After an equilibration time of 15–20 min, the clamp was reapplied. This process was continued until two anesthetic concentrations were found that just permitted and just prevented movement. The average of these values was iMAC. 10 

After the animal was fixed in the stereotaxic frame, a 10- to 13-MΩ tungsten microelectrode (FHC Inc., Bowdoinham, ME) was positioned on the midline of the cerebellar surface 10–12 mm caudal to bregma, then advanced manually 8 mm ventral, and then advanced in 5-μm increments with a microdrive (D. Kopf Instruments, Tujunga, CA) until a neuron of interest was isolated. All neurons were searched, isolated, and tested for their responsiveness to noxious pinch and heat at an anesthetic depth of 0.8–0.9 MAC. Only neurons that exhibited a discernable increase (approximately 200%) or decrease (approximately 50%) in firing rate to noxious pinch applied to the tail and hind paw and that in addition responded to noxious heat (50°C, 10 s) were selected for further testing. Noxious pinch was not applied to the hind paw used for formal testing of neuronal/withdrawal responses to heat.

Single-unit activity was amplified and displayed on an oscilloscope, as well as digitized using two personal computer systems. On one computer, single-unit activity was discriminated and saved for use with off-line template-matching software. 11On another computer, raw single-unit activity was digitized and displayed simultaneously with hind-limb force and thermode temperature, using Powerlab and Chart software (AD Instruments, Grand Junction, CO). Firing rate histograms were constructed on the Powerlab system using the Histogram spike discrimination software plug-in for Chart (AD Instruments, Grand Junction, CO). At the end of the experiment, recording sites were marked with electrolytic lesions by passing anodal DC current through the electrode. The brain was then excised, placed in 10% formalin, cut transversely in 40-μm-thick sections, counterstained with cresyl violet, mounted, and coverslipped.

The method for isometric hind-limb withdrawal force measurement was identical to our previous study. 12After a neuron was selected for analysis, the hind paw was strapped with tape to a feedback-regulated Peltier ther-mode (Thermal Devices Inc., Golden Valley, MN) with a heated surface area of 0.42 cm². The midportion of the plantar hind-paw surface was in contact with the ther-mode, which was left in place throughout the testing period. The Peltier device was attached to one end of a horizontally oriented, counterbalanced lever, the midpoint of which had a freely moving swivel. The other end of the lever was attached to a force transducer (model FT03; Grass Instruments, West Warwick, RI) via  a piece of thread. The thread and the center of the hind paw were attached to opposite ends of the lever at equal distances from the pivoting point of the lever (the swivel), and the suture was in vertical alignment with the action of the force transducer. Thus, when the animal pulled its hind limb up with the attached thermode (withdrew), the other end of the lever pulled down on the force transducer. Before testing began, 50–60 g passive tension was imposed on the hind limb by adjusting the height of the lever. Force transducer output (calibrated in grams) was digitized and recorded on a personal computer, along with thermode temperature and single-unit activity (AD Instruments). After collection of neuronal and withdrawal responses to noxious heat, some animals had one or more limbs each attached to a force transducer to record neuronal and withdrawal responses to pinch, clamping, or both of the hind paw or tail. However, responses to mechanical stimuli were used for diagnostic purposes only and were not routinely recorded and formerly tested.

At least 15 min was allowed to lapse between changes in anesthetic concentration before retesting. The order in which different anesthetic concentrations were tested was randomized. We waited at least 10 min after any type of noxious stimulus before recording spontaneous activity for 7–15 min at 0.65, 0.85, and 1.15 iMAC. The isoflurane concentrations used for testing were chosen because they span a concentration range that begins where motor withdrawals to graded noxious thermal stimuli are reliably elicited on nearly every trial (0.65 MAC) and ends at an isoflurane concentration where withdrawals to noxious heat are abolished on nearly every trial (1.15 MAC). At each of these isoflurane concentrations, graded noxious thermal stimuli (48°, 51°, and 54°C; 10 s) were then delivered to the hind paw in ascending order at 5-min intervals, from a baseline temperature of 35°C. For each stimulus, simultaneous neuronal and withdrawal responses to heat were measured. Spontaneous activity was recorded 1 min before each heat stimulus. Neuronal responses were quantified by summing the number of impulses during the 60 s after the onset of the heat stimulus. Hind-limb withdrawals were detected if there was an increase of 0.5 g or more in tension. Peak force and force integrated over time (area under the curve) were calculated for each withdrawal. Spontaneous activity was analyzed using a two-factor analysis of variance (MAC × unit). Neuronal or withdrawal responses at different stimulus temperatures and different MAC concentrations were analyzed by a three-factor analysis of variance (MAC × temperature × unit, or animal) with post hoc  Tukey multicomparisons. Statistical comparisons were made using commercial software (SPSS, Inc., Chicago, IL). In all cases, a P  value of less than 0.05 was considered statistically significant.

Results were obtained from a total of 21 units (10 ON cells and 11 OFF cells) in animals that had a mean isoflurane MAC value of 1.2 ± 0.1% (SD) (n = 19). Histologic verification of 8 ON-cell and 9 OFF-cell recording sites showed that ON and OFF cells were located in the nucleus raphe magnus and surrounding nucleus reticularis magnocellularis and were depicted in section templates from a rat brain atlas 13(fig. 1).

At sub-MAC isoflurane concentrations, ON cells were characterized by an excitatory response to hind paw and tail pinch and to noxious thermal hind paw stimulation, whereas OFF cells were inhibited by these stimuli. Individual neuronal and withdrawal responses to noxious mechanical and thermal stimuli are shown for an ON cell at 0.65 MAC isoflurane in figures 2A–Cand for an OFF cell at 0.85 MAC isoflurane in figures 2D and E.

Spontaneous activity was recorded in 10 ON cells and 9 OFF cells for at least 7 min. From 0.65 to 1.15 MAC, isoflurane significantly inhibited spontaneous activity of ON cells by 70% (from 15.0 ± 10.0 [SD] impulses/s to 4.5 ± 6.8 impulses/s; P < 0.001). In contrast, from 0.65 to 1.15 MAC, isoflurane significantly enhanced spontaneous activity of OFF cells by 138% (from 18.2 ± 14.7 to 43.4 ± 21.7 impulses/s; P < 0.001). The most substantial changes in both ON- and OFF-cell spontaneous activity occurred between 0.85 and 1.15 MAC, where immobility occurred. Effects of isoflurane on spontaneous activity are shown in figure 3Afor ON cells (n = 10) and in figure 3Bfor OFF cells (n = 9). From 0.65 to 0.85 MAC, mean ON-cell spontaneous activity was depressed by 36% (from 15.0 ± 10.0 impulses/s to 10.2 ± 9.0 impulses/s), and OFF-cell spontaneous activity was increased by 31% (from 18.2 ± 14.7 to 23.9 ± 19.1 impulses/s. Between 0.65 and 0.85 MAC, changes in ON- and OFF-cell spontaneous activity were insignificant when each class of neuron was analyzed separately. However, when ON and OFF cells were pooled together and the absolute value of changes in spontaneous activity between 0.65 and 0.85 MAC were compared, a significant change was found (P < 0.003, paired t  test). We felt justified in pooling these data because the amount of reflex facilitation or depression is determined by both ON- and OFF-cell activity combined. 14,15 

Noxious heat–evoked responses of ON cells were significantly depressed by isoflurane at 1.15 MAC compared with 0.65 and 0.85 MAC (P < 0.001), whereas OFF-cell responses were significantly enhanced (P < 0.001). When the change in activity evoked by heat was calculated by subtracting spontaneous activity recorded during 60 s before the stimulus from the response (number of total impulses fired over 60 s after the onset of heat), isoflurane depressed the heat-evoked change in activity for ON cells but not for OFF cells over all stimulus temperatures. Thus, heat caused a decrease in OFF cell activity at 1.15 MAC that was not significantly different from 0.65 and 0.85 MAC, except for responses to the lowest stimulus intensity of 48° (P < 0.01). Individual examples of isoflurane effects are shown for an ON cell in figure 4Aand an OFF cell in figure 4B, with raw traces of withdrawal force shown above the responses where withdrawals occurred. Mean heat-evoked changes in activity (spontaneous activity subtracted) are shown in figure 5Afor ON cells (n = 10) and in figure 5Bfor OFF cells (n = 11).

Withdrawal force (both peak force and area under the curve) was significantly depressed by isoflurane from 0.65 to 0.85 MAC (P < 0.001). At 0.85 MAC, the stimulus–response function showed a rightward shift compared with withdrawals at 0.65 MAC, which together with a decrease in withdrawal incidence indicated an increase in withdrawal threshold by isoflurane (fig. 6). From 0.65 to 0.85 MAC, the smaller changes in ON- and OFF-cell activity were accompanied by a 55% decrease in withdrawal force across all stimulus temperatures. At 1.15 MAC, withdrawals occurred in only 2 of 21 trials (in 19 animals), in which the two withdrawals were small and only occurred at the highest stimulus intensity of 54°C. Thus, at 1.15 MAC, the large changes in ON- and OFF-cell spontaneous activity (and heat-evoked activity of ON cells) were accompanied by a nearly complete absence of motor withdrawals to noxious thermal stimuli.

ON- and OFF-cell responses were associated better with the motor withdrawal than the heat stimulus per se . Figure 7shows an individual example in which the animal was subjected to a 54°C stimulus in three consecutive trials. The two trials that elicited a withdrawal also evoked an ON-cell response. In the trial in which no withdrawal occurred, the ON cell did not respond. Raw traces of single-unit activity and withdrawal force for the first and third trials are shown on expanded time scales. Note that the pattern of ON-cell discharge reflects the pattern of withdrawal force and that the magnitude of the response corresponds to the magnitude of the withdrawal from trial to trial. For the first 30 s after a 48°C heat stimulus, the mean excitatory response of ON cells was significantly greater in animals that withdrew from the stimulus (315 ± 377 [SD] impulses/30 s) than for animals that had no detectable withdrawal (38 ± 84 [SD] impulses/30 s; P < 0.05, unpaired t  test). The mean inhibitory response of OFF cells was also greater in animals that showed a withdrawal response (−425 ± 382 [SD] impulses/30 s) than for animals that did not (−107 ± 203 [SD] impulses/30 s; P < 0.05, unpaired t  test). Mean responses of ON cells are shown for animals that had a withdrawal response (fig. 8A) and animals that did not (fig. 8B). Mean inhibitory responses of OFF cells are shown for animals that had a withdrawal response (fig. 8C) and for animals that did not (fig. 8D).

The current results provide additional evidence for the role of RVM ON and OFF cells in nociceptive reflex modulation and further suggest that the effects of isoflurane on these classes of RVM neurons might contribute to isoflurane-induced motor depression. Previous studies show that episodes of increased OFF-cell activity, decreased ON-cell activity, or both were coupled with decreases in the efficacy of a noxious stimulus to elicit a nociceptive reflex, and vice versa . 16This is consistent with the current results, in which isoflurane suppressed ON cells (figs. 3A, 4A, and 5A) and facilitated OFF cells (figs. 3B and 4B) while depressing withdrawal responses to noxious heat (fig. 6). Previous studies suggest OFF-cell responses are reflex related because they correspond better to tail-flick latency than to the stimulus per se , 17as confirmed currently. When animals were subjected to the same heat stimulus at sub-MAC anesthetic concentrations, excitatory ON-cell responses and inhibitory OFF-cell responses were significantly larger for those animals that exhibited a withdrawal (fig. 7). Furthermore, increases in both ON-cell responses and withdrawal force occurred with increases in the thermal stimulus intensity.

The relative amounts of ON- and OFF-cell spontaneous activity have been shown to be important in determining nociceptive responsiveness, such that ON-cell firing facilitates and OFF-cell firing inhibits a nociceptive reflex. 14Moreover, hyperalgesic states are marked by increased ON-cell activity, 9,18whereas increased OFF-cell activity occurs during analgesia. 7A recent study used a double-pulse laser stimulation paradigm, 15which showed that the amount of reflex facilitation corresponded to the combined amount of both ON- and OFF-cell activity at the time of the second pulse. Because ON and OFF cells seem to function in concert to determine the efficacy of a nociceptive reflex, the small changes in ON- and OFF-cell spontaneous activity we observed between 0.65 and 0.85 MAC, when combined (67%), could have accounted for much of the 55% decrease in force production seen between these two anesthetic concentrations. Larger changes in spontaneous activity that occurred in both ON and OFF cells from 0.85 to 1.15 MAC (fig. 3) were accompanied by a nearly complete absence of withdrawals to noxious heat (fig. 6). Although we found that the absolute amount of heat-evoked decrease in mean OFF-cell activity remained more or less unchanged by isoflurane (fig. 5B), it is possible that at 1.15 MAC, isoflurane increased OFF-cell spontaneous activity to a point at which heat-evoked decreases in tonic firing are insufficient to permit disinhibition of the reflex (fig. 4B).

Increases in ON-cell activity and decreases in OFF-cell activity evoked by noxious stimuli usually occur shortly before the onset of a nociceptive reflex, and thus the RVM was initially proposed to have a permissive, or “gating,” role in reflex initiation. However, with rapid thermal laser stimulation, ON-cell bursts and OFF-cell pauses usually begin shortly after the onset of tail-flick electromyographic activity. 15This suggests that ON and OFF cells are not likely to play a role in reflex initiation, but rather they may play a role in modulating the reflex after its initiation or in response to subsequent noxious stimuli. In the current study, the more slowly rising contact heat stimuli (12°C/s) elicited ON- and OFF-cell responses that usually (although not always) preceded withdrawals (figs. 2 and 7). Given that noxious stimuli encountered by an animal in its natural environment are of variable intensity and speed of onset, it is possible that ON cells, OFF cells, or both might have a permissive (reflex-initiating) role as well as the capability to modulate nocifensive behavior after its initiation.

It is possible that ON-cell and OFF-cell activity reflects, rather than causes, modulation of nociceptive motor output. However, microinjection of opioids into the RVM causes both ON-cell inhibition and OFF-cell facilitation while inhibiting the tail-flick reflex, 19arguing in favor of a direct role for the RVM in modulating nociceptive reflexes. Furthermore, when RVM neurons were recorded through the electrode used for stimulation, OFF cells were most frequently recorded close to sites at which electrical stimulation elicited the greatest degree of tail-flick reflex inhibition. 20Moreover, local microinjection ofα¹-noradrenergic agonists excites ON cells 21while producing hyperalgesia. 22However, a role for ON cells in modulation of nociceptive reflexes is still somewhat less clear than for OFF cells. Kynurenate microinjection into the RVM (to block excitatory amino acid transmission) selectively blocks ON-cell activity without changing tail-flick latency, while OFF-cell activity and its inhibition by noxious heat remains unaffected. 23Overall, the majority of evidence from previous studies suggests that the current changes in ON- and OFF-cell activity by isoflurane contributed to the depression of limb withdrawal force.

Nearly all previous studies of RVM ON and OFF neurons measured reflex latencies alone (but see Foo and Mason 15). However, the precise latency of the neuronal response is inherently difficult to determine because of the cyclic spontaneous firing patterns in ON and OFF cells. We currently opted for a paradigm that addressed the role of ON- and OFF-cell activity in relation to the magnitude and pattern of motor withdrawals. Using this model, we found that ON- and OFF-cell responses correspond to the presence and magnitude of the withdrawal (figs. 5, 7, and 8).

The specific neuronal populations involved in depression of sensorimotor processing by anesthetics are largely unknown. Using MAC determination, several anesthetics have been shown to block movement via  a spinal action, albeit with varying degrees of supraspinal effects. 1–3We currently tested graded isoflurane effects on descending modulation by quantifying the motor output in terms of withdrawal force, in lieu of using an “all-or-none” MAC determination. We have recently validated the thermal hind-limb withdrawal force method as relevant to MAC while permitting quantification of graded anesthetic effects on motor and neuronal responses simultaneously. 12The current results confirm the model’s clinical relevance in that withdrawals were greatly reduced or abolished just above each animal’s MAC value.

The current data support our earlier hypothesis 24that under conditions of high anesthetic concentration in the brain, ON cells are relatively more depressed than OFF cells, leading to a net balance favoring descending inhibition in the spinal cord. Additional limited evidence in support of this hypothesis derives from a previous study, 25which showed that isoflurane at presumed peri-MAC concentrations depressed ON cells to a greater extent than OFF cells, which were variably affected. The current study, which showed a dose-dependent increase in OFF-cell spontaneous activity, had the added advantage of determining each animal’s iMAC value before testing to control for individual differences in sensitivity to isoflurane.

Barbiturate anesthetics have been shown to drastically change response characteristics of RVM neurons compared to the awake state, in which only low-threshold excitatory ON neurons are found, whereas high-threshold neurons and OFF cells are not found until the animal is anesthetized. 26In unstimulated awake animals, others have found OFF cells, which were characterized by facilitatory effects of morphine. 27These previous studies suggest that OFF cells may become profoundly inhibited by noxious stimuli in the awake state and are more resistant to inhibitory effects of noxious stimuli in the anesthetized or analgesic state.

From the current results, it is not possible to conclude whether isoflurane exerted its effects directly in the RVM. However, changes in RVM activity by isoflurane cannot be accounted for entirely by a spinal effect. That isoflurane has little to no effect on evoked responses of dorsal horn neurons in the peri-MAC range 12,24,28,29argues against a spinally mediated action of isoflurane on blocking ascending input to the RVM from the spinal cord. It is possible that reductions in dorsal horn spontaneous activity by isoflurane could account for reductions in withdrawal force. However, in our previous study, 12dorsal horn neurons exhibited higher spontaneous firing rates under supra-MAC concentrations of halothane, where little to no movement occurs, than under sub-MAC concentrations of isoflurane, where movement does occur. Also, one would expect that any spinal action would depress ON- and OFF-cell responses to noxious heat equally, which was not the case. However, because OFF-cell spontaneous activity was lower under sub-MAC compared with a supra-MAC isoflurane concentration, the effects of isoflurane on OFF-cell responses to heat may have been underestimated because of a possible saturation of the OFF-cell inhibitory response at lower isoflurane concentrations.

Isoflurane may act at other supraspinal sites to mediate its effect on RVM neurons. One such site is the midbrain periaqueductal gray, which projects to the RVM as part of a descending antinociceptive circuit to mediate opiate-induced analgesia. 30Interestingly, there are also ON and OFF cells in the periaqueductal gray with properties similar to those in the RVM. 31The similar actions of isoflurane and morphine on ON and OFF cells raise the possibility that isoflurane could act on endogenous opioidergic systems in the periaqueductal gray, the RVM, or both to decrease the force of movement elicited by a noxious stimulus. MAC per se  is not affected by naloxone for several volatile anesthetics, 32,33but sevoflurane MAC increases by approximately 20% in μ-opioid receptor knockout mice compared with wild types. 34However, another study showed that μ-opioid receptor knockout mice exhibit significant increases in N -methyl-d-aspartate glutamate receptor messenger RNA. 35It is therefore unclear whether the MAC increase in knockout mice was due to absence of opiate receptors or to an N -methyl-d-aspartate receptor–mediated increase in neuronal excitability. The effect of isoflurane on noradrenergic systems is also a possibility. Norepinephrine excites ON cells, although it is without effect on OFF cells. 21It was reported that isoflurane (up to approximately 1 MAC) caused a supraspinally mediated facilitation of tail-flick latencies, which was partially reversed by selective ablation of noradrenergic neurons. 36 

Currently, when spontaneous firing was subtracted from heat-evoked responses, there was a marked, concentration-dependent reduction in the ON-cell burst, whereas there was a much lesser reduction of the OFF-cell pause that occurred in relation to reflex movement (fig. 5). Because ON-cell activity is dependent on glutamatergic transmission 23and OFF-cell inhibition is μ-aminobutyric acid type A receptor mediated, 16peri-MAC concentrations of isoflurane might therefore have a more pronounced glutamate-depressant action than a γ-aminobutyric acid–enhancing action. This is consistent with a recent study, which shows relatively greater effects of halothane on glutamatergic transmission than on γ-aminobutyric acid–mediated transmission. 37 

In general, results from the current study as well as our previous study 29suggest that descending modulation of nociceptive processing remains functionally intact at sub-MAC isoflurane concentrations but becomes depressed or abolished along with noxious stimulus–evoked movement within the immediate peri-MAC concentration range (i.e. , approximately 0.8–1.2 MAC). Further studies are necessary to fully understand brainstem–spinal cord interactions involved in nociceptive sensorimotor processing and the specific sites and mechanisms by which anesthetics act in this circuitry to ablate movement.

The authors thank Cindy Clayton, D.V.M. (Technician, Department of Anesthesiology and Pain Medicine; University of California, Davis, Davis, California), and Mirela Iodi-Carstens, B.A. (Technician, Section of Neurobiology, Physiology, and Behavior, University of California, Davis, Davis, California), for their excellent technical assistance.