Cholinergic Modulation of Sevoflurane Potency in Cortical and Spinal Networks In Vitro 

All procedures were approved by the animal care committee (Eberhard-Karls-University, Tuebingen, Germany) and were in accord with the German law on animal experimentation. Spinal cord slices were prepared from embryos of pregnant 129/SvJ mice (days 14 and 15) according to the method described by Braschler et al .21Neocortical slices were obtained from 2- to 5-day-old 129/SvJ mice as previously reported.22,23 

Excised slices were placed on a coverslip and embedded in a plasma clot (Sigma, Taufkirchen, Germany). The coverslips were transferred into plastic tubes containing 0.75 ml of nutrient fluid and incubated with 95% oxygen and 5% carbon dioxide at 36.0°C for 1–2 h. One hundred milliliters of nutrient fluid consisted of 25 ml horse serum (Invitrogen, Karlsruhe, Germany), 25 ml Hanks’ Balanced Salt Solution (Sigma), and 50 ml Basal Medium Eagle (Sigma). For spinal slices, nutrient fluid included 10 nm Neuronal Growth Factor (Sigma). The roller tube technique was used to culture the tissue.24After 1 day in culture, antimitotics (10 μm 5-fluoro-2-deoxyuridine, 10 μm cytosine-b-d-arabino-furanoside, 10 μm uridine [all from Sigma]) were added to reduce proliferation of glial cells. Slices were used after 21 days in vitro  for extracellular recordings.

Spontaneous action potential activity was recorded as reported previously.13,25,26In spinal slices, extracellular recordings were performed from visually identified interneurons located in the ventral horn area.27–29In brief, slices were perfused with artificial cerebrospinal fluid consisting of 120 mm NaCl, 3.3 mm KCl, 1.13 mm NaH²PO⁴, 26 mm NaHCO³, 1.8 mm CaCl², and 11 mm D-glucose. The artificial cerebrospinal fluid was bubbled with 95% oxygen and 5% carbon dioxide. Glass electrodes with a resistance of approximately 2–5 MΩ were filled with artificial cerebrospinal fluid and were introduced into the tissue until extracellular spikes exceeding 100 μV in amplitude were visible and single-unit or multiunit activity could be clearly identified. The noise (peak-to-peak) amplitude was usually approximately 50 μV. All experiments were performed at 34°–36°C.

Test solutions containing sevoflurane were obtained by dissolving the liquid form of the anesthetic in artificial cerebrospinal fluid, which was equilibrated with 95% oxygen and 5% carbon dioxide. A closed, air-free system was used to prevent evaporation. Anesthetic levels are given as multiples of minimum alveolar concentration (MAC) of an inhaled anesthetic required to suppress movement in response to noxious stimulation in 50% of subjects. We used the EC⁵⁰values for general anesthesia proposed by Franks and Lieb.30Therefore, we assumed that 1 MAC corresponds to an aqueous concentration of 0.35 mm sevoflurane.26 

Sevoflurane was administered via  bath perfusion using gastight syringe pumps (ZAK Medicine Technique, Marktheidenfeld, Germany), which were connected to the experimental chamber via  Teflon tubing (Lee, Frankfurt, Germany). The flow rate was approximately 1 ml/min. To ensure steady state conditions, recordings during anesthetic treatment were conducted 10–15 min after starting the perfusate change. The calibration of the recording system was performed as previously reported.31 

Data were bandpass filtered between 3 and 10 kHz as acquired on a personal computer using the Digidata 1200 analog-to-digital/digital-to-analog interface (Axon Instruments, Union City, CA). Records were in addition stored on a Sony data recorder PC 204A (Racal Elektronik, Bergisch Gladbach, Germany). Further analysis was performed using self-written software in OriginPro version 7 (OriginLab Corporation, Northampton, MA) and MATLAB 6.5 (The MathWorks Inc., Natick, MA).

Data analysis was conducted as described previously.26After close inspection of the data, a threshold was set manually to avoid artifacts produced by baseline noise. The mean firing rate was obtained from single-unit or multiunit recordings and defined as the number of action potentials above threshold divided by the recording time of 180 s. In addition to the effects of sevoflurane on mean firing rates, we normalized the data to eliminate acetylcholine-induced changes of the basal activity. The data were normalized by subtracting the firing rate monitored in the presence of sevoflurane from the control rate (absence of the anesthetic) for each experiment. The resulting difference was multiplied by 100 and divided by the control rate. Hence, a value of 0% indicates sevoflurane lacked any inhibitory drug action and a 100% depression indicates that not a single action potential occurred in the presence of the anesthetic.

Results are given as mean ± SEM. Concentration–response curves were fitted by Hill equations using OriginPro version 7. Estimated EC⁵⁰values were derived from these fits. Statistical significance of differences between concentration–response curves were assessed via  an F test, as previously reported.14For statistical analysis between two data groups, the Student t  test was used.

The effects of sevoflurane and acetylcholine on action potential activity monitored in neocortical slices are illustrated in figure 1. Both substances altered neuronal firing patterns as anticipated: action potential activity was decreased by sevoflurane but increased by acetylcholine (fig. 1A). The depression of neuronal activity caused by sevoflurane was displayed as a decrease in the frequency of bursts and a decrease in firing rates within these bursts (figs. 1B and C). Although the bursts were prolonged by the anesthetic, the overall time of neuronal silence was substantially lengthened. This pattern of sevoflurane effects on cortical network activity corresponds largely to the pattern observed with other ether anesthetics such as isoflurane and enflurane, as reported previously.13Acetylcholine prolonged the duration of active states as observed with sevoflurane, but in contrast to the anesthetic, it increased average firing rates by elevating action potential activity within bursts (figs. 1C and D).

Concentration-dependent effects of sevoflurane and acetylcholine in neocortical and spinal tissue slices are depicted in figure 2. Levels of neuronal activity have been normalized to facilitate comparison. Sevoflurane reduced spontaneous activity of cortical and spinal neurons concentration-dependently within a range of clinically relevant concentrations. The anesthetic completely depressed neuronal activity at high concentrations in both preparations. Half-maximal effects were observed at 0.29 ± 0.01 mm in cortical and 0.17 ± 0.02 mm in spinal cultures corresponding to 0.83 ± 0.03 and 0.49 ± 0.05 MAC, respectively. The EC⁵⁰value for depression of ongoing activity in spinal cultures obtained from mice turned out to be approximately 50% higher compared with the EC⁵⁰value reported previously for rats (0.32 ± 0.01 MAC).26In contrast to sevoflurane, acetylcholine increased firing rates in both cortical and spinal slices up to twofold to threefold. Cholinergic enhancement of action potential activity was more pronounced in cortical compared with spinal slices. Saturating effects were observed between 10 and 100 μm acetylcholine in both preparations.

The results presented so far do not provide an answer to the question whether acetylcholine is synthesized and released in cortical and spinal slice cultures. Therefore, we investigated the effects of the reversible acetylcholinesterase inhibitor neostigmine as well as atropine, a competitive antagonist at muscarinic receptors, on ongoing neuronal activity. In these experiments, acetylcholine was not added to the artificial cerebrospinal fluid. The results are summarized in figure 3. Neostigmine significantly increased action potential activity, demonstrating the existence of a cholinergic tone. The latter implication was corroborated by the finding that atropine reduced action potential firing of cortical and spinal neurons by approximately 50%. These experiments imply that the increase in ongoing neuronal activity evoked by bath applied acetylcholine (fig. 2) displays an effect generated on top of a basal tone caused by intrinsically released acetylcholine.

Concentration-dependent effects of sevoflurane in the presence and absence of acetylcholine are presented in figure 4. Acetylcholine augmented neuronal firing rates in cortical slices at all sevoflurane concentrations tested, causing a concentration-dependent rightward and upward shift of the sevoflurane concentration response curve (fig. 4A). In spinal slices, enhancement of action potential activity caused by 10 μm acetylcholine appeared less pronounced compared with cortical slices (fig. 4B).

Further interactions between acetylcholine and sevoflurane were characterized by normalizing the changes in neuronal activity caused by the anesthetic in the absence and presence of acetylcholine (see Materials and Methods). The plots in figures 4C and Dshow the concentration-dependent effects of sevoflurane ranging to 100% inhibition, corresponding to a total depression of firing rates in both preparations. In cortical slices, acetylcholine attenuated sevoflurane potency in a concentration-dependent manner. The EC⁵⁰was increased from 0.83 ± 0.03 MAC (0 μm acetylcholine) to 1.59 ± 0.32 MAC (1 μm acetylcholine) and 2.68 ± 0.57 MAC (10 μm acetylcholine) (fig. 4Cand table 1). Both concentration–response curves of sevoflurane measured in the presence of acetylcholine differed significantly from the concentration–response curve in the absence of acetylcholine (P < 0.001, F test). In contrast to neocortical networks, acetylcholine did not reduce the potency of sevoflurane in depressing spinal neurons (0.49 ± 0.05 MAC in the absence compared with 0.47 ± 0.10 MAC in the presence of acetylcholine; fig. 4Dand table 1). The concentration–response curves did not differ significantly (P > 0.1, F test).

How can we explain the finding that acetylcholine reduces sevoflurane potency in cortical but not in spinal neurons? We hypothesized that this difference is related to distinct molecular targets of sevoflurane in cortical and spinal networks as well as to acetylcholine-induced changes in the efficacy of GABA^Areceptor–mediated inhibition. In particular, we hypothesized that sevoflurane depresses neuronal activity in cortical networks predominantly via  enhancing GABA^Areceptor function, as demonstrated previously for isoflurane and enflurane.13In contrast to the neocortex, GABA^Areceptors are only a minor target of sevoflurane in the spinal cord.26,32,33 

This hypothesis was tested as follows: Quantitative effects of acetylcholine were mimicked in spinal and neocortical slices by applying bicuculline, a specific GABA^Areceptor antagonist. Thereby, two effects exerted by acetylcholine were simulated, namely an increase in neuronal activity and a decrease in the efficacy of GABA^Areceptor–mediated inhibition. According to the hypothesis proposed in the last paragraph, bicuculline is expected to decrease sevoflurane potency in cortical but not in spinal slices. A concentration of 1 μm bicuculline was used in the experiments because it approximately doubled ongoing neuronal activity in cultured cortical and spinal slices (fig. 5A) and can consequently be regarded as equipotent with regard to the effects on network activity induced by acetylcholine. In neocortical slices, bicuculline induced an increase in EC⁵⁰of sevoflurane from 0.83 ± 0.03 to 1.44 ± 0.26 MAC (fig. 5B). Both concentration–response curves differed significantly (P < 0.01, F test). In contrast to the neocortex, bicuculline did not significantly reduce the potency of sevoflurane in depressing spinal neurons (0.49 ± 0.05 MAC in the absence compared with 0.42 ± 0.05 MAC in the presence of bicuculline; fig. 5C). In accord with the EC⁵⁰values, the concentration–response curves were not different (P > 0.1, F test).

In the following step, we investigated whether the depressant effects of sevoflurane in neocortical slices can be completely antagonized by higher bicuculline concentrations (100 μm) as reported previously for other ether derivates.34A concentration of 100 μm bicuculline abolished the depression of neuronal network activity by 0.26 mm (corresponding to 0.75 MAC) sevoflurane almost completely (from 49.61 ± 3.41% in the absence to 2.87 ± 5.13% in the presence of bicuculline; P < 0.001, n = 10). This result points to enhanced GABAergic synaptic inhibition being the predominant mechanism mediating the depressant effects of 0.75 MAC sevoflurane in neocortical slices (fig. 6).

In the last set of experiments, the interactions between sevoflurane and bicuculline were studied to elucidate how a decrease in GABA^Areceptor–mediated inhibition alters the potency of sevoflurane in cortical and spinal networks. The results displayed in figure 5support the proposed mechanism that acetylcholine attenuates the impact of GABAergic inhibition. To corroborate this hypothesis, the effects of etomidate, an almost selective modulator at GABA^Areceptors, were investigated. The selectivity of etomidate at GABA^Areceptors has been demonstrated previously for both cortical and spinal networks in vitro  and in vivo .35,36Therefore, the hypothesis to be tested was that acetylcholine reduces the relative inhibition of neuronal activity exerted by etomidate not only in cortical but also in spinal slices. Effects of etomidate on cortical and spinal neurons were investigated at a concentration of 1.5 μm. This concentration decreased neuronal activity by approximately 60% in both cortical and spinal networks in the absence of acetylcholine. In cortical slices, the relative inhibition in the presence of acetylcholine was reduced from 63.4 ± 6.8 to 29.7 ± 13.7% (n = 10, P < 0.05), whereas acetylcholine decreased the relative inhibition of 1.5 μm etomidate to depress spinal neurons from 65.5 ± 3.9% to 47.0 ± 4.1% (n = 8, P < 0.01). The results clearly demonstrate that in contrast to sevoflurane, the relative inhibition of etomidate was reduced not only in cortical but also in spinal networks, supporting the hypothesis that acetylcholine attenuates the impact of GABAergic inhibition in both preparations.

The cholinergic system is an important modulatory neurotransmitter system in the brain.37Activation of cholinergic innervation of the cortex has been implicated in sensory processing, learning, and memory. The application of the reversible acetylcholinesterase inhibitor tacrine has been demonstrated to promote plasticity and learning in the motor cortex of healthy volunteers.38At the cellular level, high concentrations of the cholinergic agonist carbachol have been shown to have a strong desynchronizing action on neuronal activity measured in cultured cortical neurons.39In layer V of the rat visual cortex, acetylcholine both increases excitability and depresses synaptic transmission by affecting different GABAergic interneurons via  distinct cholinergic receptors.40Furthermore, there is ample evidence that cholinergic stimulation decreases GABA release in cortical networks by activating muscarinic and nicotinic receptors located on presynaptic terminals.9,10 

Moreover, interactions of reversible and irreversible blockers of acetylcholinesterase with muscarinic receptors affect GABAergic transmission in hippocampal neurons in vitro .41But how will a decrease in GABA release alter anesthetic mechanisms? We predict that the potency of GABAergic anesthetics in depressing neuronal network activity is quantitatively linked to the extracellular GABA concentration. This hypothesis is based on the fact that at clinically relevant concentrations, anesthetics frequently act as positive modulators at GABA^Areceptors. Therefore, they require the presence of GABA at the agonist binding site for affecting GABA^Areceptor function.42–44The experimental finding that acetylcholine impairs the potency of sevoflurane in decreasing neuronal activity in cortical slices is therefore well explained by a modulatory action of sevoflurane at GABA^Areceptors.

Unlike in neocortex, sevoflurane potency remained unaffected by acetylcholine in spinal cord slices. What is the reason for this brain region–specific difference? In the current study, we have provided evidence that the GABA^Areceptor is a major target for sevoflurane in neocortical microcircuits. However, we recently demonstrated that in cultured spinal slices sevoflurane depresses neuronal activity via  multiple molecular targets and that modulation of GABA^Areceptors is not the major mechanism by which sevoflurane decreases the excitability of spinal neurons.26The different actions of sevoflurane on neocortical and spinal networks are clearly shown in figure 5. The specific GABA^Areceptor antagonist bicuculline increases neuronal basal activity at a concentration of 1 μm in neocortical and spinal slices by approximately twofold, indicating that in both networks action potential firing is under substantial GABAergic control. However, bicuculline reduces sevoflurane potency only in neocortical slices, supporting the hypothesis that acetylcholine affects sevoflurane potency just minimally in spinal slices because GABA^Areceptors are not a predominant target of the anesthetic.

However, the finding that acetylcholine did not affect the potency of sevoflurane in depressing action potential firing of spinal slices can alternatively be explained by assuming that acetylcholine does not affect GABA^Areceptor–mediated inhibition in the spinal cord. To rule out this possibility, interactions between acetylcholine and the almost selective GABA^Areceptors modulator etomidate were investigated, hypothesizing that acetylcholine did not affect etomidate potency in spinal slices. The hypothesis had to be abandoned because acetylcholine decreased the potency of etomidate in both spinal and cortical networks. The finding that acetylcholine significantly affects etomidate potency in spinal slices clearly argues against the possibility that acetylcholine does not modulate GABA^Areceptor–mediated inhibition in the spinal cord.

The brain acetylcholine concentrations occurring during a severe organophosphate intoxication are hard to measure in vivo . However, in an outstanding study, Tonduli et al .3exposed freely moving rats to soman and acquired three sets of neurophysiologic data before and during the intoxication. They determined cortical acetylcholinesterase activity and acetylcholine concentrations by microdialysis and associated the parameters with electroencephalographic recordings as well as with power spectrum analysis of the γ band. Acetylcholine levels measured in 60-μl microdialysis fractions obtained from intoxicated rats ranged from 40 ± 8 pmol in subjects showing no seizure activity to 102 ± 21 pmol in rats displaying severe seizure activity,3corresponding to concentrations of 0.7 and 1.7 μm in the perfusate of the fraction, respectively. The concentration in extracellular space is always higher compared with the perfusate concentration, depending on the type of probe and the perfusion speed. In the study performed by Tonduli et al ., the microdialysis probes were perfused with 2 μl/min, and a fraction was collected every 30 min. With this perfusion speed, a relative recovery of 20% can be assumed, leading to concentrations between 3.5 and 8.5 μm in the extracellular space depending on the degree of soman intoxication. Therefore, the concentrations of 1 and 10 μm used in our experiments cover well the range of extracellular acetylcholine concentrations observed in intoxicated rats.

In an elegant study, Hudetz et al .45recently showed that neostigmine, a reversal blocker of acetylcholinesterase, inverts isoflurane-induced hypnosis in rats. In addition to these results, there is multiple evidence that cholinesterase blockers reverse anesthesia by increasing brain acetylcholine levels.46,47In the current study, we identified two putative mechanisms: First, we observed that acetylcholine increases neuronal basal activity in spinal and cortical slices by twofold to threefold (fig. 2). Because general anesthesia is based on a severe depression of central nervous functions, the excitatory actions exerted by acetylcholine must be counterbalanced by increasing anesthetic concentrations. Second, we showed that acetylcholine compromises anesthetic potency, a finding that is probably explained by an impairment of GABA^Areceptor–mediated inhibition. Both mechanisms, namely the elevation of neuronal basal activity (fig. 4) and the decrease in anesthetic potency, may significantly increase drug requirement for providing general anesthesia in patients with organophosphate intoxication. The pharmacologic management of organophosphate poisoning has extensively been discussed by Lallement et al .48,49Remarkably, benzodiazepines such as diazepam, which are potent modulators of GABA^Areceptors, are recommended for acute treatment of epileptiform seizure activity in intoxicated patients.7,49In rats, diazepam interrupts soman-induced seizures only when applied within the first 5–10 min after seizure onset.50It seems likely that a severe decrease in the potency of diazepam as caused by massive cholinergic overstimulation may attribute to this temporally limited effect.

The conclusion that cholinergic stimulation increases drug requirement for providing anesthesia is also corroborated by human studies using a reversible inhibitor of acetylcholinesterase. Meuret et al .47investigated the effects of physostigmine in human subjects anesthetized with propofol by assessing central nervous system function by use of the Auditory Steady State Response and Bispectral Index. Physostigmine restored consciousness with concomitant increases in the Auditory Steady State Response and Bispectral Index. Using a similar experimental approach, Plourde et al .51report that sevoflurane anesthesia is also antagonized by physostigmine.

In summary, the studies on the effects of anticholinesterases on experimental animals and human subjects clearly indicate that cholinergic overstimulation may considerably increase drug requirement for providing general anesthesia. However, raising anesthetic concentrations into a high-dose range in patients with organophosphorus intoxication may not always be an option, because these patients frequently display cardiac abnormalities and hemodynamic instability.1,7An alternative treatment may be the coapplication of anesthetics and antagonists of muscarinic acetylcholine receptors such as atropine. The rational for this option is that acetylcholine-induced reversal of hypnosis is largely mediated via  muscarinic receptors. Hudetz et al .45demonstrated that muscarinic agonists mimic the effects of anticholinesterases on the righting reflex and the electroencephalogram. This finding is corroborated by the observation of Douglas et al .52,53that cholinergic electroencephalographic arousal is largely mediated by M1 and M2 receptors. Without blocking muscarinic receptors, anesthesiologists may be unable to determine anesthetic requirement because acetylcholine concentrations in the central nervous system are unknown and may undergo gross changes in patients with organophosphorus intoxication. However, more direct studies to test these ideas in whole animals are required before recommendations for approaches to clinical care can be made.

The authors thank Claudia Holt (Technical Assistant, Eberhard-Karls-University, Tuebingen, Germany) for excellent technical assistance.