Halothane-induced Hypnosis Is Not Accompanied by Inactivation of Orexinergic Output in Rodents

All studies were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania (Philadelphia, Pennsylvania) and were conducted in accordance with the National Institutes of Health guidelines.*Adult C57BL6/J male mice (Jackson Labs, Bar Harbor, ME) aged 8–12 weeks were placed on a 12-h light-dark cycle and given food and water ad libidum . On experimental days, the mice were placed in loss-of-righting-reflex chambers19and exposed to either 1.0% halothane dissolved in 100% oxygen (halothane group, n = 8) or in 100% oxygen (wake control group, n = 8) in the 2 h just after lights out (denoted by “ZT12–ZT14,” where ZT is Zeitgeber time—a 24-h clock defined by the onset of light, which occurs at ZT0) during their period of maximal activity.20Twenty-six orexin/ataxin-3 mice aged 70 ± 5 days and 13 sibling controls aged 65 ± 9 days of both genders were used in righting reflex behavioral assays. For the electrophysiology experiments, 37 adult male Sprague-Dawley rats (250–300 g, Harlan, Indianapolis, IN) were kept on a 12-h light-dark cycle (lights on at 0800) for at least 1 week after arriving in the lab before experiments, and were given food and water ad libidum .

All mice were immediately killed by cervical dislocation followed by rapid brain harvest, postfixation in 4% paraformaldehyde, and paraffin embedding. Sections were cut at 10 μm and stained for c-Fos (rabbit polyclonal antibody PC05 [Calbiochem, San Diego, CA]; corresponding to amino acids 4–17 of the human c-Fos, at 1:1000) and prepro-orexin (mouse monoclonal antibody MAB763 [Chemicon, Temecula, CA]; corresponding to amino acids 34–66 of the human orexin precursor, at 1:4000) according to standard protocols as previously described.15Fluorescent secondary antibodies were an Alexa 594-labeled goat antirabbit (Invitrogen, Carlsbad, CA, at 1:200) for detection of c-Fos and an Alexa 488 goat antimouse (Invitrogen, at 1:200) for detection of prepro-orexin. Three to six sections centered on the perifornical hypothalamus were counted per brain. Counting was performed by a blinded experimenter and confirmed by a second blinded experimenter.

Before electrophysiological recordings, the animals were anesthetized with halothane or isoflurane in air via  spontaneous respiration. Animals were intubated with a tracheal cannula, continuously anesthetized with either 0.9–1.2% halothane (for subsequent LC recordings under halothane) or 1.0–1.4% isoflurane (for subsequent recordings under isoflurane) using a Riken FI21 refractometer (AM; Bickford, Wales Center, NY) to confirm delivered anesthetic concentration, and placed in a stereotaxic frame with the incisor bar lowered to place the skull approximately 12 degrees from the horizontal plane. Body temperature was maintained at 34–36°C with a thermistor-controlled heating pad. Animals recorded during their dark period had their eyes covered with black tape to avoid any light input during setup, surgery, and recording.

During LC recordings and microinjections, the animals were maintained under stable levels of either 0.8–0.9% halothane or 1.0–1.1% isoflurane. These equipotent doses correspond to 1.3 times the ED⁵⁰dose required for loss of righting reflex in rats,21,22and were sufficient to permit in vivo  recordings in the stereotaxic frame without movement artifact. For recording of baseline firing frequencies during the active or rest periods, the animals were maintained exclusively under the influence of one or the other anesthetic. To examine the effects of the orexin-1 receptor antagonist SB-334867-A, the animals were initially anesthetized with either halothane or isoflurane and then switched to the alternative anesthetic. In animals initially anesthetized with halothane (n = 4), LC neurons were recorded until a neuron displaying inhibition to SB-334867-A was found. Subsequently, while recording from the same neuron, the inhaled anesthetic was switched to isoflurane. Rats initially anesthetized with isoflurane underwent a similar switch to halothane while continuously recording from an LC neuron (n = 2 animals). A minimum of 1.5 h with fresh gas flow exceeding minute ventilation passed to allow the rats to equilibrate with the new inhaled anesthetic.

LC recordings and microinjections were performed using standard methods as previously described.23,24In brief, a 5-mm-diameter hole was drilled in the skull above the LC and the dura was reflected. A glass recording micropipette (tip diameter, 2–4 μm; 10–20 MΩ impedance) was filled with 2% pontamine sky blue dye in 0.5 m Na-acetate and aimed at the LC (4 mm caudal to lambda, and 1.2 mm lateral to midline). Signals were amplified and continuously displayed on an oscilloscope as unfiltered and filtered (0.3–10 kHz bandpass). LC neurons (5.8–6.5 mm ventral to the skull surface) were tentatively identified during recording using well-established criteria: An entirely positive, notched waveform 2 milliseconds or more in duration in unfiltered records; a slow, tonic, spontaneous discharge (0.5–6.0 Hz); and a strong phasic activation followed immediately by long-duration inhibition (0.5–1.0 s) in response to noxious stimuli (electrical foot shock stimulation).25After identification of putative LC neurons, spike amplitude was continuously monitored to ensure stable recordings. Footpad stimulation was delivered through paired 26-gauge needles inserted in the medial aspect of the contralateral hind paw and was given as monophasic 0.5-millisecond pulses of 90 V. Stimuli were given once every 2 s for a total of 50 repetitions. Spikes of single neurons were discriminated, and digital pulses were led to a computer for online data collection using a laboratory interface and software (CED 1401, SPIKE2; Cambridge Electronic Design, Cambridge, England). Response magnitudes to foot shock stimulation were calculated as the total number of spikes recorded from 20 to 100 milliseconds after stimulation minus the expected number of spikes (average number of spikes/s recorded 1 min before stimulation).

Injection pipettes (tip diameter, 12–20 μm) were fashioned from microdispensing capillary tubes as described previously.26,27The thin lumen of these pipettes allowed accurate measurement of microinjection volume (accuracy within 15 nl). The shank of the pipette was bent and attached to the recording pipette using ultraviolet light-curing dental resin (Filtek Supreme, Scotchbrand Multipurpose Adhesive; 3M, St. Paul, MN), with its tip recessed approximately 100 μm.

Microinjections were made by pneumatic pressure through the infusion barrel of the compound recording/infusion microelectrode (Picospritzer III, Parker Instruments, Cleveland, OH). After isolating a single LC neuron, baseline firing frequencies were recorded for at least 150 s. After this baseline period, solutions were administered by applying pressure pulses of 30–100-millisecond duration at 40 psi every 15 s for a total of 150–225 s. With this procedure, each pulse delivered approximately 12 nl and the total volume applied was 120–180 nl. Care was taken to ensure that microinjection did not interfere with unit recordings by constantly monitoring spike amplitude during delivery. Typically, neurons could be held long enough to observe firing frequencies return to baseline (washout) within 30–300 s.

SB-334867-A (100 μm; Tocris, Ellisville MO) was dissolved at its final concentration in artificial cerebrospinal fluid, consisting of (in mm) 112 NaCl, 3.1 KCl, 1.2 MgSO⁴, 0.4 NaH²PO⁴, and 25 NaHCO³, and was either used fresh or stored at −20°C for no longer than 1 week before use. This dose of 100 μm was chosen based on previously published in vivo  electrophysiological studies.28,29At the end of the recordings, micropipette penetrations were marked by iontophoretic ejection of dye from the recording pipette (−20 μA, 50% duty cycle for 15 min). Brains were snap-frozen in a solution of isopentane at −80°C, and coronal sections through the LC (40 μm thick) were cut on a cryostat, mounted on gelatinized glass slides, and stained with neutral red.

Induction and emergence from halothane was defined behaviorally using the respective loss and return of the righting reflex. The mice were placed in cylindrical gas-tight controlled environment chambers arrayed in parallel. After 90 min of habituation with 100% oxygen each day on two successive days, anesthesia was induced with a Dräger model 19.1 halothane vaporizer (Draeger Medical, Inc., Telford, PA) using 12 incremental increases in halothane dissolved in 100% oxygen. Halothane gas concentration was determined in triplicate during the last 2 min of each step. Initial concentration was 0.41%. After 15 min at each concentration to allow for equilibration of the mouse with the anesthetic vapors, the concentration of halothane was increased by 8 ± 3% of the preceding value. Peak halothane concentration was 1.06%. At the end of each 15-min interval, the cylindrical chambers were rotated 180 degrees. A mouse was considered to have lost the righting reflex if it did not turn itself prone onto all four limbs within 2 min. After the last mouse lost its righting reflex, the halothane concentration was increased one more time before measurements of emergence time, which was defined as the duration that elapsed until each mouse regained its righting reflex by turning prone onto all four paws. Mouse temperature was maintained between 36.6 ± 0.6°C by submerging the controlled environment chambers in a 37°C water bath. To minimize both the number of mice used and the number of anesthetic exposures, induction and emergence from halothane in orexin/ataxin-3 mice and sibling controls were performed during the same experiment. Sensitivity to anesthetic induction and emergence from isoflurane and sevoflurane has been previously reported.15 

All data were tested for normality. Statistical testing was performed using Prism 4.0c software (Graph Pad, La Jolla, CA). Fos immunoreactivity in orexin-positive neurons spanning the dorsomedial through the perifornical hypothalamus was analyzed with a Mann–Whitney U test and is reported as a median plus interquartile range. Electrophysiological spike frequencies were averaged into 10-s bins using recording software (SPIKE 2; Cambridge Electronic Design) and exported to an Excel (Microsoft Corporation, Redmond, WA) spreadsheet. After stable recordings were established and before drug application, a 150-s window of time was chosen to average baseline impulse activity and compare with average frequencies during drug application. Normality of these data sets was confirmed using Kolmogorov-Smirnov, D’Agostino and Pearson omnibus, and Shapiro-Wilk normality testing. Significance was tested by paired two-tail t  tests before and during drug application. To obtain EC⁵⁰and Hill slope constants, the log of volatile anesthetic gas concentration versus  the fraction of the population having lost the righting reflex plots were generated and fit with a nonlinear dose–response curve with a variable slope by using Prism 4.0c (Graph Pad). EC⁵⁰and Hill slope constants are reported as means of two independent trials with 95% confidence limits. Emergence time data generated immediately after each anesthetic exposure are reported as a mean ± SEM.

To determine whether anesthetic-induced hypnosis must be accompanied by inhibition of wake-active orexinergic neurons, we examined the effect of an obtunding dose of halothane on nuclear c-Fos expression as a surrogate of neuronal activity. Adult C57BL/6J mice were exposed to either an oxygen control or to anesthetizing doses of halothane 1.0% in oxygen for 2 h beginning at lights out, the period of maximal wakefulness, and killed for immunohistochemical analysis. During halothane 37.7% (28.5–45.7%) of dorsomedial and perifornical orexinergic neurons exhibited c-Fos–positive nuclear staining. This was not significantly different from nonanesthetized circadian wakeful controls 42.4% (34.5–47.1%) (n = 4 animals/group with 4-5 slices/animal; U = 7.000, P = 0.89; by Mann–Whitney U test, fig. 1).

We tested whether LC firing rates were similar during the period when nocturnal rodents are most awake and active (ZT13–ZT19, 1–7 h after lights out). During general anesthesia produced by isoflurane, representative in vivo  recordings demonstrate that LC neurons fired significantly slower than under a comparable dose of halothane (2.95 ± 0.27 Hz halothane, n = 29 neurons, 12 animals vs.  1.88 ± 0.17 Hz isoflurane, n = 42 neurons, 7 animals, unpaired two-tail t  test: P = 0.001, t = 3.989, df = 69, fig. 2), consistent with preserved LC orexinergic input under halothane but not under isoflurane anesthesia. Rest period (ZT5–11) firing frequencies were not significantly different in LC neurons recorded from halothane and isoflurane anesthetized animals (1.7 ± 0.2 Hz halothane, n = 16 neurons, 8 animals vs.  1.38 ± 0.12 Hz isoflurane, n = 31 neurons, 5 animals, unpaired two-tail t  test: P = 0.17, t = 1.262, df = 30, fig. 2).

Does the inhibition of LC firing frequency caused by the orexin-1 receptor antagonist SB-334867-A during the active period, previously observed under halothane-induced hypnosis29also occur when the animal is under isoflurane-induced hypnosis? To answer this, we used combined microinjection-electrophysiological recording electrodes in a total of six animals (fig. 3). To ensure that any negative data would not be the result of interanimal differences in either the active-period neuronal firing rate or responses to SB-334867-A, two animals were recorded under isoflurane anesthesia first and subsequently switched to halothane, while four animals were recorded under halothane anesthesia first, followed by isoflurane. Consistent with our previously published results,29100 μm SB-334867-A reduced LC impulse activity during the active period but not the rest period while the animals were under halothane anesthesia (2.66 ± 0.45 Hz preinjection, 1.58 ± 0.24 Hz SB-334867-A; paired two-tail t  test: n = 8, P = 0.008, t = 3.59, df = 7). Specifically, neurons with firing frequencies faster than the median frequency (2.76 Hz) were inhibited by SB-334867-A (3.59 ± 0.41 Hz preinjection vs.  1.94 ± 0.26 Hz SB-334867-A; paired two-tail t  test: n = 4, P = 0.001, t = 10.24, df = 3), whereas neurons firing slower than the median frequency were not (1.73 ± 0.45 Hz preinjection vs.  1.21 ± 0.32 Hz SB-334867-A; paired two-tail t  test: n = 4, P = 0.32, t = 1.20, df = 3). In contrast, SB-334867-A did not attenuate active-period LC impulse activity during isoflurane-induced hypnosis (1.88 ± 0.17 Hz preinjection, 2.0 ± 0.19 Hz SB-334867-A; paired two-tail t  test: n = 19, P = 0.3, t = 1.07, df = 18). Neurons firing faster than the median frequency during isoflurane anesthesia (1.89 Hz) were not significantly inhibited in the presence of SB-334867-A (2.49 ± 0.11 Hz preinjection, 2.50 ± 0.25 Hz SB-334867-A; n = 10, paired two-tail t  test: P = 0.96, t = 0.10, df = 9), nor were those firing more slowly (1.19 ± 0.12 Hz preinjection, 1.44 ± 0.17 Hz SB-334867-A; n = 9, paired two-tail t  test: P = 0.01, t = 3.5, df = 8). SB-334867-A did not attenuate rest-period LC impulse activity during isoflurane-induced hypnosis (1.38 ± 0.17 Hz preinjection, 1.5 ± 0.18 Hz SB-334867-A; paired two-tail t  test: n = 16, P = 0.63, t = 2.3, df = 15). Altogether, the electrophysiological results suggest that while orexinergic neurons can maintain the potential to excite downstream targets under halothane, orexinergic activity is blunted under isoflurane.

Having demonstrated that halothane-induced hypnosis arises despite preservation of activity in two wake-active groups, we hypothesized that orexin/ataxin-3 mice would show delayed emergence from halothane general anesthesia with no change in sensitivity to induction of anesthesia, as was the case for isoflurane and sevoflurane.15As expected (fig. 4A), induction sensitivity was indistinguishable (F^2,70= 0.21) for orexin/ataxin-3 mice (halothane EC⁵⁰= 0.87%, 95% CI = 0.84–0.90%, Hill slope = 13.2, 95% CI = 6.5–19.8) and wild-type sibling controls (halothane EC⁵⁰= 0.86%, 95% CI = 0.83–0.89%, Hill slope = 15.3, 95% CI = 7.6–23.1). However, after exposure to halothane, emergence times for halothane exposed orexin/ataxin-3 mice (20.0 ± 1.6 min) and wild-type sibling controls (18.8 ± 1.0 min) were also indistinguishable (P = 0.52, by unpaired t  test) (fig. 4B).

Our understanding of the mechanisms responsible for anesthetic-induced unconsciousness is at best rudimentary. The search for structures controlling the arousal state whose function is reversibly modulated by anesthetic drugs remains essential for unraveling anesthetic action. Conscious perception requires an awake brain, although wakefulness by itself is not sufficient for consciousness.30Recently, extensive insights have been made in the neurobiology of sleep-wake control.31,32The promotion, maintenance, and stabilization of wakefulness are functions of the ascending reticular activating system—a distributed neuroanatomic network with distinct structures in the brainstem, diencephalon, and forebrain.

Available evidence suggests that one way in which general anesthetics transiently and reversibly impair consciousness is via  inhibition of wake-promoting and sustaining arousal centers.3,4,33–36In this article we demonstrate that inhibition of all such arousal centers is not necessary for anesthetic-induced unconsciousness. For instance, noradrenergic neurons in the LC and orexinergic neurons in the hypothalamus form part of the distributed neuroanatomic wake-promoting network.37During active wakefulness, orexinergic neurons in the dorsomedial and perifornical hypothalamus are known to fire rapidly; whereas during both nonrapid eye movement and rapid eye movement sleep, orexinergic neurons become nearly quiescent.38,39For orexinergic neurons, c-Fos protein expression reliably tracks antecedent electrophysiological activity.20In awake animals, c-Fos is expressed in the nuclei of dorsomedial and perifornical hypothalamic orexinergic neurons, whereas in anesthetic-induced hypnosis produced by isoflurane or sevoflurane, c-Fos expression is significantly reduced by 30 or 50%, respectively, to levels found during nonrapid eye movement sleep.15,20Here we show that for the prototype volatile alkane, halothane, anesthetic-induced hypnosis occurs despite ongoing wake-active levels of c-Fos expression and presumably of electrophysiological activity in the orexinergic population. Wake-active LC neurons also fire more rapidly during the active period (ZT12–24) than the rest period (ZT0–12) in halothane-anesthetized rats.40This is mediated at least in part by orexinergic inputs to the LC.29Ongoing depolarization of orexinergic neurons with subsequent release of orexin-A at its efferent projections during halothane anesthesia is sufficient to explain preserved activity in the LC, which is heavily innervated by orexinergic neurons,16depolarized by orexin-A,18and slowed during halothane anesthesia by local microinjection of the orexin antagonist SB-334867-A. This pattern of activity in orexinergic and LC neurons during halothane contrasts with that seen during comparable doses of isoflurane. We demonstrate that local microinjection of SB-334867-A during an isoflurane anesthetic had no effect on the LC, confirming the absence of orexinergic tone. In stark contrast to halothane’s effects, equipotent doses of isoflurane do not appear to preserve orexinergic activity as determined immunohistochemically and do not permit the expression of a diurnal rhythm of LC impulse activity.

Several methodologic issues must be considered when examining our results. First, despite its two decades of use as a neuronal activity tracer, immunohistochemical detection of c-Fos protein may not accurately reflect electrophysiological activity in all cell types.41For example, in populations that do not express c-Fos or in populations such as the LC, where basal c-Fos expression is extremely low during active wakefulness42,43at a period in time when LC neurons fire maximally, it may not be possible to detect an anesthetic-induced decrease in c-Fos. However, in the orexinergic neurons, c-Fos expression in the 2 h before killing has been validated as a surrogate of antecedent neuronal activity. Second, for all experiments that compare the effects among volatile anesthetics, equipotent doses must be defined. For our in vivo  extracellular recordings of LC neurons of rats acutely anesthetized with isoflurane or halothane, we used a dose of 1.3 times the ED⁵⁰for loss of righting in rats, based on published studies of uninstrumented animals.21,22While the actual ratio of isoflurane’s ED⁵⁰to halothane’s ED⁵⁰is numerically different than that of work by Imas et al .,44inspection of the 95% confidence limits in the original work by Kissin et al.  21,22is not inconsistent with more recent work. Nonetheless, we acknowledge that our chosen ratio relying on the increased potency of halothane relative to isoflurane may not be perfect. Third, compounding the question of anesthetic potency is the development of mild hypothermia in our rats, which developed despite several attempts to maintain euthermia, including warming blankets above and below all rats. As core temperatures did not differ between rats exposed to isoflurane or halothane, the major impact would be an increase in volatile anesthetic potency relative to our measured concentrations. Finally, with respect to studies in orexin/ataxin-3 transgenic mice, we acknowledge the potential for compensatory changes as a result of postnatal destruction of all orexinergic neurons. These animals and their sibling controls were studied at 10 weeks of age shortly after onset of narcolepsy and cataplexy.45Therefore, any potential compensatory adaptations should be minimized in animals that undergo embryogenesis and development with an intact orexin signaling system, relative to studies in mice with a constitutive knockout of the prepro-orexin gene or its receptors.

Studies with barbiturates, propofol, chloral hydrate, and ketamine have previously demonstrated that inactivation of the LC is not required for induction of general anesthesia, even though it is considered fundamental for dexmedetomidine.3,4,34,42Both indirect and direct assessments indicated that halogenated ether anesthetics such as isoflurane, as well as propofol and etomidate, do inhibit orexinergic neurons of the hypothalamus.15,42,46,47Meanwhile, halothane, similar to dexmedetomidine,34fails to depress activity of orexinergic neurons.

A relative preservation of orexinergic signaling during halothane exposure might be anticipated to bias other cortically projecting arousal systems in addition to the LC, such as the basal forebrain,46intralaminar, and midline thalamic nuclei,48and may begin to mechanistically explain how hypnotic doses of halothane produce less cortical electroencephalographic depression than comparable doses of isoflurane.7–9Such a finding is also consistent with the demonstration that intracerebroventricular addition of exogenous orexin-A during 1.2% isoflurane, when endogenous levels should be very low, changed the cortical electroencephalographic signature from a state of burst suppression, indicative of deep anesthesia, to a more aroused pattern.47 

Recently, suprahypnotic doses of sevoflurane have been shown to directly depolarize LC neurons recorded from a brainstem slice via  gap junction channels. Significantly reduced inward depolarizing currents were also elicited from LC neurons in the slice by hypnotic and subhypnotic levels of sevoflurane, as well as sub- to suprahypnotic doses of isoflurane, but not by halothane or propofol.49Our results with isoflurane and halothane suggest that the direct depolarizing contributions of volatile anesthetics (sevoflurane >> isoflurane >> halothane) at gap junctions measured in the chemically deafferented LC could be outweighed by orexinergic and other presently undefined afferent inputs to the LC in the intact animal. However, it should be noted that we did not measure the firing rates of LC neurons in intact rats exposed to equipotent hypnotic doses of sevoflurane.

We also demonstrate that mice lacking orexinergic neurons emerge from halothane anesthesia in a manner indistinguishable from their sibling controls. In contrast, such mice have delayed emergence from isoflurane and sevoflurane.15The neuronal substrates governing anesthetic emergence remain largely unknown. However, distinct patterns of emergence in narcoleptic-cataplexic mice after exposure to two different halogenated ethers, as compared with the halogenated alkane halothane, highlight an opportunity to exploit agent-specific diversity, leading ultimately to novel hypothesis testing. Although our results show that hypnotic doses of halothane do not depress firing of orexinergic neurons nor of LC neurons, this cannot account for the difference in recovery from anesthesia as compared with isoflurane/sevoflurane, because orexinergic neurons have been completely destroyed in adult orexin/ataxin-3 mice.45 

Redundancy in the neural control of arousal state may have evolved as a survival advantage to protect wakefulness. Both lesion and genetic knockout studies suggest the functional integration of arousal promoting networks is sufficient to overcome permanent destruction at one or more ascending reticular activating nodes.50,51However, transient anesthetic-induced inactivation at wake-promoting reticular activating loci  occurring either in parallel or at critical sites appears sufficient to induce anesthetic hypnosis.33,52,53Presumably, reactivation of wake-promoting reticular activating loci  is essential for anesthetic emergence.15,35,54Hence, to explain the normal emergence of orexin/ataxin-3 mice from halothane, we predict the existence of an afferent input to orexinergic neurons with the following features: it should show circadian changes in neuronal activity, should modulate activity of orexinergic neurons while directly or indirectly modulating LC activity, and should be inhibited by isoflurane/sevoflurane while remaining unaffected by equipotent hypnotic doses of halothane. This hypothetical wake active locus may also reinforce activity of LC neurons in addition to the indirect effects such a nucleus would promote by activating the orexinergic neurons.18Of the many potential candidates for such a group, basal forebrain cholinergic neurons are known to display state-dependent activity, send efferent projections to orexinergic neurons, and depolarize orexinergic neurons via  release of acetylcholine.55,56Manipulation of central cholinergic signaling modulates the behavioral expression of sleep, wake, and anesthetized states.14,57,58Pharmacologic enhancement of cholinergic neurotransmission using physostigmine, an acetyl cholinesterase inhibitor that crosses the blood-brain barrier, has been demonstrated to speed emergence from both intravenous anesthetics59–62and volatile anesthetics including isoflurane, sevoflurane, and halothane.63–65Although no study describes a direct comparison between the efficacy of physostigmine as a partial antagonist of halothane- versus  isoflurane- or sevoflurane-induced hypnosis, equipotent doses of halothane and isoflurane have been shown to significantly but differentially alter presynaptic cholinergic signaling in a rat synaptosome preparation.66Halothane antagonizes the inhibition of pharmacologically induced release of the inhibitory neurotransmitter γ-aminobutyric acid mediated by muscarinic receptors in striatum, whereas isoflurane does not. Whether such an effect persists elsewhere in the brain and whether it occurs with physiologic depolarization remains unknown, but this finding highlights the potential mechanistic differences between equipotent doses of halothane and isoflurane on efficacy at cholinergic sites.66Together, these facts make cholinergic basal forebrain neurons an attractive subject for current46,67,68and future investigations.

The authors wish to thank Dr. Andrew Ochroch, M.D., M.S.C.E. (Associate Professor, Department of Anesthesiology and Critical Care, University of Pennsylvania) for his expert statistical input.