Central Nervous System Sodium Channels Are Significantly Suppressed at Clinical Concentrations of Volatile Anesthetics

A stably transfected Chinese hamster ovary cell line (CNaIIA-1) was used. The vector used for transfection contained a gene conferring resistance to the amino-glycoside antibiotic G418.

The cells were grown in RPMI medium (Gibco, Grand Island, NY) containing 10% fetal bovine serum and 1% penicillin/streptomycin mixture (Gibco) as well as 200 micro gram/ml G418 to select for transfected cells. Cells were cultured in 25 cm²polystyrene culture flasks (Corning, Corning, NY) at 37 degrees C in room air containing 5% CO². For electrophysiologic recordings, cells were transferred to 60 mm Petri dishes (Becton Dickinson, Lincoln Park, NJ).

Cells were used 2-3 days after the transfer, before the cell layer became confluent. For electrophysiologic measurements, the culture medium was replaced by an extracellular solution containing 130 mM NaCl, 4 mM KCl, 1.5 mM CaCl², 1.5 mM MgCl², 5 mM glucose, and 5 mM HEPES, adjusted to pH 7.4 with CsOH. All solutions were filtered through 0.22 micro meter-filters (Millipore, Bedford, MA) before use.

Recordings were made at room temperature (24.4+/-1.2 degrees C). Because we expressed anesthetic potencies not as minimum alveolar concentration (MAC) values (strongly temperature-dependent) but as aqueous concentrations (relatively temperature-insensitive [14]), the concentrations can be extrapolated to body temperature.

Sodium currents were studied using the whole-cell configuration of the patch-clamp recording technique, [15]using a standard patch-clamp amplifier (Axopatch 200, Axon Instruments, Foster City, CA) controlled by commercially available software (pCLAMP, Axon Instruments) on a standard personal computer. Currents were filtered at 5 kHz, digitized, and recorded to hard disk. Capacitive transients and series resistance were measured and compensated using the amplifier's internal compensation circuitry; active series resistance compensation was used to compensate 60-85% of the series resistance. Average series resistance before compensation was 3.5+/-1.5 M omega; the average cell capacitance was 18.1+/-8.6 pF.

Patch-clamp pipettes were pulled from micropipette glass (Drummond, Broomall, PA) and filled with an intracellular solution containing 10 mM NaCl, 90 mM CsF, 60 mM CsCl, and 6 mM HEPES, adjusted to pH 7.4 with CsOH.

Cells with currents larger than 6 nA and smaller than 0.5 nA were excluded because of increasing series resistance error or possible contamination by possible small endogenous sodium currents in Chinese hamster ovary cells. [11]The average current of the 75 cells included in this study was 2.4+/-1.5 nA.

Volatile anesthetic solutions were prepared by injection of defined amounts of anesthetics into PTFE-capped glass vials filled with extracellular solution. The vials were vigorously vortexed for 2 min and sonicated in a bath sonicator for 1 h, yielding consistent anesthetic concentrations. The anesthetic solutions were applied via a glass-PTFE-perfusion system and a superfusion pipette (flow rate 0.5-0.8 ml/min) close to the cell. Anesthetic concentrations of halothane, isoflurane, enflurane, desflurane, and sevoflurane were measured in some experiments in the perfusion reservoir at the start of the perfusion and after 3 and 7 min using gas chromatography. The degree of evaporation from the open-perfusion reservoir is different for each anesthetic (largest for desflurane with a decrease in aqueous concentration of 25% after 7 min). No correction was made for this differential evaporation; all concentrations reported are those at the beginning of perfusion, and actual anesthetic concentrations at the cells will be lower.

Solutions of diethylether and ethanol were prepared in the same way, but concentrations were calculated from the amount of anesthetic injected into the glass vial.

Curve fits were computed using a least-squares algorithm (Marquardt-Levenberg) of commercially available software (Sigmaplot, Jandel Scientific, Corte Madera, CA). Data are+/-SEM, unless noted otherwise.

A standard way to calculate suppression of sodium currents by drugs is to determine peak inward current from current-voltage plots. [9,10]Sodium currents are elicited by voltage steps to various test potentials (here -60 to +100 mV) from a hyperpolarized holding potential (here -120 mV), at which channel inactivation is removed. Sodium currents elicited with this protocol are inhibited by volatile anesthetics, as exemplified by desflurane (Figure 1). The suppression occurs within seconds and can be reversed by washing out the anesthetic with anesthetic-free extracellular solution. Reversal was complete in most experiments, even after repeated drug application. Higher concentrations of anesthetic tended to lead to incomplete reversal and, therefore, were applied last in experiments in which multiple concentrations were applied. The voltage dependence of sodium current activation was unchanged by the anesthetics.

Concentration-response curves were established from averaged suppression data from at least six cells for each anesthetic (Figure 2). Suppression was calculated as the reduction of the maximum inward current (which occurs, depending on the cell, in the test potential range of -10 to +10 mV), expressed as percent of control. Data for all anesthetics were fit with simple hyperbolic curves, assuming 100% maximum suppression. The logarithm of the IC⁵⁰values obtained (Table 1) correlates to the logarithm of the octanol/water partition coefficients [16]in the same linear manner as the Meyer-Overton correlation of MAC values (compare to Figure 6). Ethanol, which can be used as an (intravenous) anesthetic adjuvant, [17]was included in the experiments to extend the range of partition coefficients. For the anesthetics studied, linear regression fits to the data yielded slopes of -1.05 (r²= 0.956) for peak sodium current suppression and -0.99 (r²= 0.996) for concentrations yielding anesthesia (represented as MAC values, converted to aqueous concentrations [16]). On average for the anesthetics studied, 50% block of peak sodium current requires threefold higher concentrations than human MAC values.

Steady-state inactivation is a physiologically important property of sodium channels determining the availability of the channels at a certain membrane potential. The standard voltage protocol used to measure steady-state inactivation comprises prepulses to varying potentials (here, 500 ms to potentials between -150 and -10 mV) and a constant test pulse (here, -10 mV). Before application of an anesthetic, currents (shown for an exemplary cell in Figure 3) decrease at prepulse potentials positive to about -80 mV. This phenomenon can be explained by a voltage-dependent transition from closed to inactivated channels, the latter being unavailable for opening. The voltage-dependent distribution between the two states represents steady-state inactivation and is described by a Boltzmann function. Data were fitted to the function f(V) = I^max- {I sub max/exp [-z *F* (V - V^h)/RT]}, which is characterized by three parameters, with I^maxmaximum current at hyperpolarized potentials, z the slope parameter, and V^hthe midpoint potential at which the function assumes its half-maximal value (average control value for all experiments was -54.0 mV); F is the Faraday constant, R the gas constant, and T the absolute temperature. Currents are normalized to I^maxbefore addition of the anesthetic.

After application of an anesthetic (demonstrated for ether in Figure 3(B)), currents are reduced at all potentials as expected from the experiments described above, and additionally, steady-state inactivation is shifted in the hyperpolarizing direction (leftward).

These results reveal two effects of volatile anesthetics on sodium currents: (1) a voltage-independent current suppression at hyperpolarized potentials and (2) a hyperpolarizing (leftward) shift in steady-state inactivation, which leads to an additional, voltage-dependent current reduction in the voltage range of channel inactivation.

The concentration-dependence of the shift in V^his shown for the six anesthetics studied in Figure 4. For sevoflurane, the effect appears to saturate at higher concentrations (Figure 4(B)). In most experiments, however, currents at high anesthetic doses became too small for accurate determination of V^h, and it was not possible to determine whether the effect saturated for all anesthetics.

In most experiments, no complete reversal of the shift could be achieved during washout, probably because of the limited time of washout (5-10 min) or irreversible anesthetic effects. A hyperpolarizing shift of V^hin the absence of drugs has been reported for other whole-cell preparations. [18,19]However, in our cells, after the first 5 min of the experiment, no shift of steady-state inactivation with time (up to 30 min, data not shown) was found in control experiments without anesthetic application.

Potential-Dependence of Anesthetic Potency Leads to IC⁵⁰Values at Resting Membrane Potentials Comparable to Clinical Anesthetic Concentrations

The hyperpolarizing shift of channel inactivation induced by volatile anesthetics leads to a voltage-dependence of anesthetic potency. From channel inactivation plots normalized to I^maxbefore addition of the anesthetic (Figure 3(B)), concentration-response curves at each potential can be calculated. Plotting the IC⁵⁰values yielded by those calculations versus the prepulse potential (Figure 5) demonstrates an increase in anesthetic potency in the voltage region of channel inactivation. The data in Figure 5imply that the voltage dependence is similar for all anesthetics studied.

IC⁵⁰values at -120 mV, at which no channels are inactivated, and at -60 mV, near the resting potential and firing threshold of neuronal cells, are compared in Table 1with concentrations yielding anesthesia. At -120 mV, IC⁵⁰values for sodium channel suppression are much larger than clinical anesthetic potencies, whereas at -60 mV, the values are similar.

Both at -60 and at -120 mV, the IC⁵⁰values correlate well with octanol/water partition coefficients on logarithmic scales (Figure 6); slopes for the linear regression fits are -1.05 (r²= 0.918) at -60 mV and -1.00 (r²= 0.933) at -120 mV.

Under physiologic conditions, stimulation of a cell rarely occurs as a single signal but rather as trains of frequent stimuli. Pharmacologic block of these frequent stimuli can be quantitatively different from stationary block if channel recovery from inactivation is slowed by a drug that interacts with the inactive state of the channel. As indicated in Figure 7for halothane, volatile anesthetics cause such a slowing of channel recovery (i.e., returning to the resting channel conformation) after a depolarization, consistent with fast association and dissociation rates of anesthetic binding to the inactive state of the channel. This effect was the same for all anesthetics studied at comparable concentrations (data not shown). Channel recovery without anesthetics and at low anesthetic concentrations could be fitted well with single exponential functions, whereas at higher concentrations, multiple exponentials were necessary to fit the data.

Prolonged channel recovery after a depolarization, such as shown in Figure 7, leads to use-dependent block if the intervals between depolarizing pulses become too short to allow for complete recovery to the resting state of the channel from the inactive state. We assessed use-dependent block with trains of 20 depolarizations from -85 to 0 mV applied at 5 Hz. All examined anesthetics caused use-dependent block; Figure 8(B) shows desflurane as an example: Increasing anesthetic concentrations caused more pronounced use-dependent current reduction and a decrease in the time constant of this use-dependent block.

To further investigate the mechanism underlying use-dependent block, we studied its dependence on the parameters of the pulse protocol (Figure 9). Higher pulse frequency (Figure 9(A)), longer pulse duration (Figure 9(B)), more depolarized holding potential, and more depolarized pulse potential all lead to more pronounced use-dependent block. These manipulations increase the fraction of time sodium channels spend in the inactivated state, and the more pronounced block is therefore consistent with a preferential interaction of volatile anesthetics with inactivated sodium channels.

The results presented here indicate that volatile anesthetics can significantly suppress sodium currents through voltage-gated sodium channels at clinically relevant concentrations. There are at least two distinct mechanisms by which these anesthetics suppress CNS sodium channel currents: (1) a potential-independent suppression of resting or open sodium channels and (2) a hyperpolarizing shift in the voltage-dependence of sodium channel inactivation. This latter modification of sodium channel function causes channels to inactivate at more hyperpolarized potentials, resulting in a voltage-dependent anesthetic suppression of sodium channel currents. As a consequence of these combined interactions, IC⁵⁰values for sodium channel suppression are much lower at potentials close to neuronal resting membrane potentials (-60 mV²⁰) than at the more hyperpolarized potentials used to determine peak current reduction. At the depolarized potentials, current suppression occurs at anesthetic concentrations similar to those yielding anesthesia (Figure 6and Table 1).

Several factors need to be considered when interpreting the anesthetic concentrations used in the present experiments. These experiments were conducted at room temperature, allowing a direct comparison with previous pharmacologic examinations of these channels, [12]as well as with other anesthetic studies of sodium channels that generally have been carried out at room or lower temperatures. [1,3-5,9,10]Although we reported aqueous anesthetic concentrations, which are relatively temperature-insensitive, [14]sodium channel properties may shift with temperature. For example, it has been shown in peripheral and/or nonmammalian sodium channel preparations that sodium channel kinetics change with higher temperatures and that steady-state inactivation is shifted toward more depolarized potentials. [21,22]However, these experiments were conducted with animals whose body temperature is lower than mammals, and thus, it is difficult to predict how such changes might affect the anesthetic interactions reported here.

The voltage-independent anesthetic suppression of sodium channels at hyperpolarized potentials is similar to that found in previous studies examining peak current reduction [9,10]and occurs at similar anesthetic concentrations. The volatile anesthetic modification of sodium channels found in the current experiments is consistent with a state-dependent interaction best described by the modulated receptor hypothesis of drug interaction. [23,24]In this model, different channel states (at least three for sodium channels: resting, open, and inactivated) are assumed to have different affinities for drug binding. The hyperpolarizing shift of steady-state inactivation (Figure 4), the use-dependent block of the channels (Figure 8and Figure 9), and the slowing of the rate of recovery from inactivation (Figure 7) can be explained qualitatively by assuming that anesthetics bind more strongly to the inactivated state of the channel rather than to its resting state. [25]The observed use-dependent block could be due to binding to the open state of the channel, but the increasing inhibition found when increasing the time the channel spends in the inactivated state (longer depolarizations, Figure 9(B)) indicates preferential binding to the inactivated state. The modulated receptor model has been proposed to explain the interactions of other drugs, such as local anesthetics, antiarrhythmics, and anticonvulsants, with the sodium channel, [26]and these agents may share common or overlapping binding sites.

A less specific type of interaction also could take place. For example, a change in the physicochemical properties of the cellular lipid membrane could result in an alteration of the free energy differences between the inactivated state of the channel and either its resting or its open state. In this case, the steady-state equilibrium between the channel conformations would be altered. With either explanation, anesthetic potency would depend on the available conformational states of the channel at different membrane potentials and thereby result in the potential-dependent anesthetic potencies reported here.

In this study, we examined the effects of anesthetics on rat brain IIa sodium channel alpha subunits expressed in stably transfected Chinese hamster ovary cells. The type IIa sodium channel is the most prominent sodium channel type in adult brain, although two other subtypes of sodium channel have been found. [12]The expression of this alpha-subunit in mammalian cell lines, without coexpression of beta subunits, is sufficient to obtain the normal physiologic [11]and pharmacologic [12]behavior of sodium channels found in situ. Further, this rat brain sodium channel has greater than 97% structural identity with the parallel human brain sodium channel, [13]indicating that the results obtained should be pertinent to the human CNS channel and making it a viable experimental model for examining anesthetic interactions with sodium channels. A recent report [27]indicated that the beta 1 subunit, when coexpressed with the alpha subunit in these cells, did not significantly shift channel inactivation.

The anesthetic effects on the expressed CNS sodium channels found in these experiments are quantitatively different from those previously reported. This parallels the finding that volatile anesthetics have larger effects on conduction in mammalian brain axons than peripheral nonmammalian preparations. [4]The concentrations needed to suppress peak inward sodium current when inactivation is removed by a hyperpolarized holding potential are somewhat lower than those reported for suppression of peak sodium currents in the squid giant axon. [9]However, in common with previous findings in squid axon, [1]suppression of rat brain sodium channels by volatile anesthetics follows the Meyer-Overton correlation with anesthetic solubility in lipophilic solvents.

The anesthetics studied also had a much larger effect on the voltage-dependence of inactivation of brain sodium channels compared to that reported for the squid giant axon [9]or the toad sciatic nerve. [28]In this latter preparation, diethylether exhibited no use-dependent block, [28]whereas halothane did so only at high stimulus frequencies. [29]These differences may reflect species variability in anesthetic modification of channel inactivation, as has been reported for n-alkanols. [19].

In determining the contributions of potential molecular targets to general anesthesia, it is generally assumed that "significant" targets must have a k¹/2 of action close the clinical concentrations of anesthetics. [14]The voltage dependence of anesthetic effects on sodium channels complicates the comparison of clinical anesthetic potency with sodium current suppression. At potentials close to the threshold potential of action potential firing (around -55 mV at the axon hillock [20]), about 50% of the sodium current is suppressed by clinical concentrations of volatile anesthetics, and therefore by the criterion given above, sodium channels should be reconsidered as a possible anesthetic target.

However, the relevance of this or any other molecular anesthetic action to anesthesia also needs to be related to changes occurring at the cellular level. [14]The assumption of a correlation between a 50% effect on a given target and its relevance to anesthesia may not be correct.

Further complicating the interpretation of molecular anesthetic actions on sodium channels, these proteins have several cellular physiologic functions that may be differentially affected by a common molecular anesthetic modification. Despite the significant anesthetic suppression of sodium channel current found in the present study, it has been established at the cellular level that anesthetics have only minor effects on axonal conduction at these same concentrations. [30,31]This may be due to the "safety factor," defined as the ratio of current supplied by incoming action potentials to that required to sustain propagation, which is normally quite high. [32]It has been estimated that more than 70% of sodium channels must be blocked to prevent action potential firing. [33]Further, a block of sodium channels in nerve and muscles would be fatal, as found in cases of puffer fish poisoning. [34]Nonetheless, the use-dependent block of sodium channels at higher stimulation frequencies leads to anesthetic impairment of axonal conduction. [27]At lower stimulus frequencies, use-dependent block would increase the threshold for action potential firing, with concomitant changes in neuronal firing patterns and frequencies.

Additionally, suppression of sodium currents should lead to an elevation in neuronal firing threshold, [35]and consistent with our results, volatile anesthetics have been shown to elevate the threshold potential of neurons at clinical concentrations. [2,36,37]Changes in threshold are probably most effective in disrupting neuronal function in the region of the axon hillock, although changes in the excitability of the axon have been implicated in general anesthesia. [3].

Finally, information in the CNS is encoded not as the amplitude or duration of action potentials but as their frequency, [20]and thus, increasing the threshold potentials needed to fire an action potential, added to use-dependent anesthetic block, could disrupt the flow of information in the CNS and thereby contribute to the clinical state of anesthesia.

In summary, we have presented evidence that clinical concentrations of volatile anesthetics significantly suppress currents through sodium channels at physiologic membrane potentials. The relevance of these molecular interactions to the clinical state of anesthesia needs to be further examined at the cellular level.

The authors thank Dr. O. S. Andersen, Dr. E. Bennett, Dr. J. P. Dilger, Dr. S. R. Levinson, and Dr. B. W. Urban, for discussions, and Dr. L. Palmer, for loan of equipment.