Effects of Sevoflurane on Excitatory Neurotransmission to Medullary Expiratory Neurons and on Phrenic Nerve Activity in a Decerebrate Dog Model

The research was approved by the Medical College of Wisconsin Animal Care Committee (Milwaukee, WI) and conformed with standards set forth in the National Institutes of Health Guide for Care and Use of Laboratory Animals.

Anesthesia in dogs was induced by mask ventilation with sevoflurane, and the trachea was intubated with a cuffed endotracheal tube; from that time, the lungs were mechanically ventilated with an oxygen–sevoflurane (1.3–1.8 minimum alveolar concentration [MAC]) mixture throughout the surgical procedures. We have previously described the surgical preparation in detail. 7,8In short, after bilateral vagotomy and bilateral pneumothorax the animals were decerebrated 9and paralyzed (0.1 mg/kg pancuronium followed by 0.1 mg · kg⁻¹· h⁻¹). The medulla oblongata and the phrenic nerve were prepared for recording. Esophageal temperature was maintained at 38.5 ± 1°C. Mean arterial pressure was kept above approximately 100 mmHg (if necessary, with 0.5–5 μg · kg⁻¹· min⁻¹phenylephrine). Protocols were performed only during steady-state conditions for blood pressure.

Data recording and collection, as well as the drugs used, are described in detail in our previous study. 7A multibarrel compound glass electrode was used to simultaneously record extracellular neuronal action potential activity before and during pressure ejection of respective neurotransmitters onto expiratory neurons of the caudal ventral respiratory group. 5,6,8Bipolar hook electrodes were used to record multifiber activity from the C5 rootlet, and the phrenic neurogram was obtained from the moving average of this activity.

The protocols were performed under hyperoxic (fraction of inspired oxygen [Fio²] > 0.8) and steady-state hypercapnic conditions (arterial carbon dioxide tension, 50–60 mmHg). A typical protocol consisted of two separate picoejection periods (part 1, 200 μm NMDA; part 2, 200 μm bicuculline) at each anesthetic dose level (i.e.,  0 MAC, 1 MAC sevoflurane, and 0 MAC end control).

The peak neuronal discharge frequency (Fⁿ) was measured for 10–20 respiratory cycles during a preejection control period (F^con). Then the glutamate receptor agonist NMDA was applied in increasing dose rates until an increase in peak Fⁿof at least 40 Hz was achieved. Typically, picoejection durations of 6–8 min with 2 or 3 dose rates were needed. Sufficient time was allowed for Fⁿto return to the control level.

To quantify the effect of NMDA on the postsynaptic receptor, we determined the dose rate that caused a 40-Hz increase in peak Fⁿ(i.e.,  a 40-Hz net increase at 0 MAC sevoflurane) and designated it as D^40Hz. To confirm the linearity of the dose–response curves in this range, we also determined the net increase at half this dose rate and designated it as 1/2 D^40Hz. Next, we determined the corresponding net increases from the dose–response curves at 1 MAC sevoflurane. Then all net increases were normalized to the 40-Hz net increase at 0 MAC, which was assigned a value of 100%. A two-way, repeated-measures analysis of variance with main factors of level of anesthesia (0 or 1 MAC sevoflurane) and neurotransmitter status (preejection control vs.  NMDA response) was performed (SuperANOVA; Abacus Concepts, Inc., Berkeley, CA).

After recovery from NMDA, the GABA^Areceptor antagonist bicuculline was picoejected until complete block of GABA^Aergic inhibition occurred (i.e.,  when an increase in picoejection dose rate did not result in any further increase in Fⁿ). Typically, picoejection durations of 5–10 min with several increasing dose rates were required. Upon discontinuation of bicuculline, complete postejection recovery was awaited, which typically required 30–45 min. Then sevoflurane was introduced to a depth of 1.0 MAC (2.4 vol%), 10and after an equilibration time of 15 min, both parts of the protocol were repeated in the same fashion. It was ensured that peak bicuculline dose rates during anesthesia always matched or exceeded those during the anesthetic-free run. After recovery from bicuculline, the anesthetic was discontinued, complete washout was awaited, and the picoejection protocol was repeated at 0 MAC to obtain end controls. A complete neuron protocol (0-MAC level, the anesthetic wash-in, the 1-MAC level, and the return to 0-MAC level end control) required approximately 4 h.

We defined the peak Fⁿfor the preejection control period as F^con, and for the maximal Fⁿunder bicuculline block as F^e. F^eis a measure of the uninhibited overall glutamatergic excitatory drive to the neuron, whereas F^conrepresents this drive reduced by the prevailing basal GABA^Aergic inhibition. To calculate the change in overall excitatory drive, the data were normalized to F^eat the 0-MAC level, which was assigned a value of 100%. A two-way, repeated measures analysis of variance with main factors of level of anesthesia (0 or 1 MAC sevoflurane) and blocking status (preejection control vs.  maximal bicuculline block) was used.

The prevailing GABA^Aergic inhibition was described by the inhibitory constant α which is defined as follows:α=[F^e− F^con]/F^e. The values for F^eand F^conwere obtained for the 0-MAC level (F^con0, F^e0) and the 1-MAC level (F^con1, F^e1) from the experimental runs. Then they were used in the calculation of the anesthetic-induced effects on overall excitation (ΔF^e=[F^e1− F^e0]/F^e0) and overall inhibition (Δα=[α1−α0]/α0). All results are given as mean ± SD, and P < 0.05 was used to indicate significant differences unless stated otherwise.

Peak phrenic nerve activity (PPA), which is a neural index of the magnitude of the tidal volume, 11was measured in arbitrary units as the peak height of the phrenic neurogram in the anesthetic-free control state and during steady-state 1 MAC sevoflurane anesthesia. PPA data obtained from our previously published halothane study were similarly analyzed for comparison purposes. All PPA data were normalized to the values obtained during the anesthetic-free control state, which was assigned a value of 100%. These normalized data were used in the subsequent statistical analyses to compare the two agents.

Twenty animals were studied, and 16 complete neuron protocols consisting of 0-, 1-, and 0-MAC end control levels were obtained.

Figure 1shows a representative example of an expiratory neuronal response to increasing picoejection dose rates of the glutamate receptor agonist NMDA before (trace A) and during (trace C) 1 MAC sevoflurane. A detailed analysis of these runs revealed that at 0 MAC a 40-Hz net increase was reached at a dose rate of 0.72 pmol/min. The net increase at the same dose rate at 1 MAC sevoflurane was 43.8 Hz even though F^conwas profoundly depressed from 115 to 35 Hz (i.e.,  by 70%).

The pooled normalized net increases in peak Fⁿ, which were produced by NMDA, for 16 complete neuron protocols are summarized in figure 2. One MAC sevoflurane reduced the NMDA-induced net increase from 100 to 82.3 ± 53.7% at D^40Hz, but this was not statistically significant (P = 0.17). The 0-MAC end control value for D^40Hz(filled square in fig. 2) was 88.3 ± 53.6%, which was also not significantly different from the 0-MAC control (filled circle in fig. 2). The corresponding normalized 1/2 D^40Hzdata were not statistically different. A slope analysis for the 0–1/2 D^40Hzand the 1/2 D^40Hz–D^40Hzvalues confirmed linearity of the NMDA responses over the full dose range.

Figure 1(traces B and D) shows the effects of bicuculline ejection on the same neuron as described earlier in this study for the NMDA responses. At 0 MAC, bicuculline increased peak Fⁿfrom 115 to 247 Hz, yielding a GABA^Aergic inhibitory constant α⁰= 0.53. This means that the prevailing endogenous GABA^Aergic inhibition attenuated neuronal output by 53% during the control state. At 1 MAC, bicuculline increased peak Fⁿfrom 35 to 162 Hz, yielding an α¹value of 0.79. Thus, for this neuron, 1 MAC sevoflurane enhanced the GABAergic inhibition by 49%. The overall excitatory drive, which was measured in the absence of GABA^Aergic inhibition (i.e.,  during maximal block with bicuculline) was decreased from 247 to 162 Hz or by 34%.

The pooled data for 16 complete neuron protocols are shown in figure 3. Analysis of the pooled data shows that 1 MAC sevoflurane increased the mean α value from 0.47 ± 0.07 to 0.56 ± 0.11 (i.e.,  1 MAC sevo-flurane enhanced overall GABA^Aergic inhibition by 18.7 ± 20.6% (Δα, fig. 3,,bottom , center bar). At the same time, 1 MAC sevoflurane depressed overall excitation by 19.5 ± 16.2% (P < 0.01, ΔF^e, fig. 3, bottom , left). In addition, peak F^conwas significantly reduced from 53.0 ± 6.4 to 36.6 ± 11.9% (fig. 3, top , F^con) (i.e.,  by 26.7 ± 22.4%[ΔF^con, fig. 3, bottom , right).

Phrenic data **were used when peak phrenic amplitude (PPA) for the 0-MAC end control period recovered to the preanesthesia control level. To maximize the number of data sets, we used phrenic data from several related studies with the same experimental setup (decerebration, mechanical ventilation, hyperoxic hypercapnia) and the same anesthetic sequence (0 MAC, 1 MAC, 0 MAC) but divergent picoejection protocols. We analyzed 20 protocols with sevoflurane, which were derived from 28 animals from the current study and an ongoing study. Twenty-four data sets with halothane were derived from 36 dogs, also from two studies (a study of halothane and expiratory neurons 7and an ongoing study).

Both anesthetics significantly reduced PPA (fig. 4,,right ). Sevoflurane reduced PPA significantly more than halothane (60.1 ± 17.6%vs.  42.5 ± 18.4%, P < 0.01, unpaired two-sided t  test). Both anesthetics depressed expiratory neuronal activity to a similar extent (ΔF^con[sevoflurane], 26.7 ± 22.4%vs. ΔF^con[halothane], 27.9 ± 10.6%7(fig. 4, left , nonsignificant). Both anesthetics depressed PPA significantly more than F^con(sevoflurane, P < 0.001; halothane, P < 0.01 [unpaired two-sided t  test).

The current study shows that sevoflurane reduced the activity of expiratory premotor neurons in the brainstem of decerebrate dogs. This reduction resulted from a depression of glutamatergic excitation without affecting the postsynaptic glutamate receptor excitability. Enhancement of GABA^Aergic inhibitory neurotransmission contributed approximately equally to the depressant effect of sevoflurane. The current study cannot address whether the GABA^Aergic effects are presynaptic or postsynaptic in nature.

The magnitude of the reduction of overall glutamatergic excitation by sevoflurane was not different than that found previously for halothane (P = 0.08, unpaired two-sided t  test) 7. Thus, it is likely that both anesthetics affect similar mechanisms that result in the inhibition of presynaptic glutamate release. These mechanisms may include a reduction in excitatory drive to the presynaptic neurons, as it has been described by Perouansky et al.  12for the excitation of inhibitory interneurons, and inhibitory effects on one or more steps in the cascade that controls exocytosis of glutamate. 13 

Our in vivo  study showed an enhancement of overall GABA^Aergic inhibition by 1 MAC sevoflurane, although we could not apportion the amount of presynaptic and postsynaptic contributions. Additional studies would be necessary to delineate the effects of sevoflurane and halothane on the postsynaptic GABA^Areceptor function in our respiratory neuron model.

An enhancement of the response to GABA and a small direct activation by sevoflurane have been described for the GABA^Areceptor. 1,14In acutely dissociated rat hippocampal neurons during whole cell voltage clamp, Kira et al.  1showed that sevoflurane alone could increase the Cl⁻current (I^Cl). However, this effect is negligible at 1 MAC sevoflurane (i.e.,  0.35-mm concentration 15) in aqueous solution and only becomes noticeable at bath concentrations of 3 MAC and greater. At 1 MAC, sevoflurane not only enhanced the GABA-induced I^Cl, 14but also accelerated the activation phase of I^Cl, suggesting an increase in the apparent affinity of the GABA^Areceptor to GABA. 1Similarly, whole cell patch-clamp studies by Jenkins et al.  16on mouse fibroblasts transfected with GABA^Areceptor subunits suggested a direct binding of sevoflurane to the receptor channel protein. In addition, the authors reported that there was no difference in the degree of GABA^Aergic potentiation between sevoflurane and halothane at 37°C. In accordance, we also found that there was no difference between the enhancement of overall inhibition by 1 MAC sevoflurane and 1 MAC halothane 7in expiratory premotor neurons (P = 0.18, unpaired two-sided t  test).

The phrenic neurogram represents the collective neuronal output of the phrenic motoneuron pool, and PPA is considered to be a good neural index for the magnitude of tidal volume. 11Our data show that PPA is more depressed than expiratory premotor neurons (fig. 4). This suggests either a differential depression of inspiratory and expiratory premotor neurons by the anesthetics or additional depressant effects downstream at the phrenic motoneuronal level. However, we could show in an intact preparation that a 1-MAC halothane increase (from 1 to 2 MAC) depressed inspiratory and expiratory premotor neuronal activity to the same extent and significantly less than PPA. 17Similarly, in the decerebrate preparation halothane depressed PPA more than expiratory premotor neuron activity. We assume that this differential depression is in part caused by an additional anesthetic-induced depression of synaptic transmission from the bulbospinal premotor neurons to the phrenic motoneurons, as well as direct depression of the phrenic motoneurons.

Recent in vitro  studies have suggested the presence of voltage-independent, hyperpolarizing leak K⁺channels that are activated by inhalational anesthetics 18and can be found in cerebral nuclei and with an especially high density in the membrane of motoneurons in the brainstem and spinal cord. 19Talley et al.  19found that in rat hypoglossal motoneurons tandem-pore domain acid-sensitive K⁺(TASK)-1 channels contributed to a prominent pH-sensitive background K⁺current. Both halothane and sevoflurane increased the conductance of these K⁺channels, which produced dose-dependent hyperpolarizations that could be antagonized by decreasing the pH from 7.3 to 6.5. 20It is possible that such anesthesia-induced K⁺channels contribute to the pronounced depression of phrenic motoneurons compared with the expiratory premotor neurons in our study.

Interestingly, Sirois et al. 18found in hypoglossal motoneurons in a neonatal rat brainstem slice preparation that the outward K⁺current induced by 0.75% halothane (51.1 ± 6.5 pA) was 40% less than the current induced by 2.0% sevoflurane (71.3 ± 8.8 pA), which are considered equivalent anesthetic concentrations. These values suggest that sevoflurane may have a greater effect on TASK-1 channels than halothane. This could explain the finding that, in our preparation, at the 1-MAC level sevoflurane depressed phrenic motoneurons to a greater extent than halothane.

A similar, differential effect for the two anesthetics was reported by Doi et al.,  2who recorded respiratory-related neuronal activity in the feline nucleus ambiguus and found a significantly greater depression of spontaneous neuronal activity with 1 MAC sevoflurane than with 1 MAC halothane. The nucleus ambiguus contains a high number of respiratory-related laryngeal and pharyngeal motoneurons, 21and labeling reveals a high density of TASK-1 channels in this area. 19We consider it likely that the neuron sample of Doi et al.,  2contained mainly motoneurons and speculate that the greater depression of neuronal activity by sevoflurane was caused by a differential effect on TASK-1 channels.

TASK-1 transcripts are expressed at high levels in cranial and spinal motoneurons but not in all neurons. The role of these channels may be minimal in our premotor neurons, although direct data are lacking. Because activation of these channels was capable of producing up to a 50% increase in whole cell membrane conductance in hypoglossal motoneurons, 20this effect would reduce neuronal excitability. However, the NMDA-induced excitation of our expiratory premotor neurons was not altered by sevoflurane (fig. 2) or halothane. 7In addition, TASK-1 channels have been shown to be depressed by low pH, serotonin, and norepinephrine. 19Ejection of serotonin, norepinephrine, and acidified artificial cerebrospinal fluid on expiratory premotor neurons in the thiopental-anesthetized intact dog does not alter neuronal frequency (unpublished data) as would be expected if TASK-1 channels were present and functional. Thus, it seems unlikely that in these canine expiratory premotor neurons an anesthetic-induced activation of inhibitory K⁺channels contributed to the changes in excitatory drive (i.e.,  the depression of overall excitation).

Many of the limitations of our methodology have been discussed in our previous publications 7,8and elsewhere. 6,7Decerebration allowed us to compare the effect of 1 MAC sevoflurane with an anesthesia-free control state. This may explain differences between our results and other studies. Mutoh et al.  22studied the effect of sevoflurane on respiratory parameters in intact dogs under α-chloralose–urethane background anesthesia and also found that sevoflurane induced an increase in breathing frequency. In their model, however, this effect of sevoflurane could be abolished by vagotomy. In our decerebrate, vagotomized, and peripherally deafferented dogs sevoflurane still caused an increase in breathing frequency. Thus, we suggest that sevoflurane, like halothane, also has directly stimulating effects at the brainstem level, presumably in the rate-generating regions. 23 

In summary, sevoflurane depressed expiratory premotor neuron activity through a presynaptic reduction of glutamatergic excitation and an enhancement of GABA^Aergic inhibition. Sevoflurane depressed expiratory premotor neurons significantly less than phrenic motoneurons.

The authors thank Jack Tomlinson, Biology Laboratory Technician, Department of Anesthesiology, VA Medical Center, Milwaukee, Wisconsin, for excellent technical assistance.