Direct Activation of Sleep-Promoting VLPO Neurons by Volatile Anesthetics Contributes to Anesthetic Hypnosis

Summary

Background

Despite seventeen decades of continuous clinical use, the neuronal mechanisms through which volatile anesthetics act to produce unconsciousness remain obscure. One emerging possibility is that anesthetics exert their hypnotic effects by hijacking endogenous arousal circuits. A key sleep-promoting component of this circuitry is the ventrolateral preoptic nucleus (VLPO), a hypothalamic region containing both state-independent neurons and neurons that preferentially fire during natural sleep.

Results

Using c-Fos immunohistochemistry as a biomarker for antecedent neuronal activity, we show that isoflurane and halothane increase the number of active neurons in the VLPO, but only when mice are sedated or unconscious. Destroying VLPO neurons produces an acute resistance to isoflurane-induced hypnosis. Electrophysiological studies prove that the neurons depolarized by isoflurane belong to the subpopulation of VLPO neurons responsible for promoting natural sleep, while neighboring non-sleep-active VLPO neurons are unaffected by isoflurane. Finally, we show that this anesthetic-induced depolarization is not solely due to a presynaptic inhibition of wake-active neurons as previously hypothesized, but rather is due to a direct postsynaptic effect on VLPO neurons themselves arising from the closing of a background potassium conductance.

Conclusions

Cumulatively, this work demonstrates that anesthetics are capable of directly activating endogenous sleep-promoting networks and that such actions contribute to their hypnotic properties.

Introduction

General anesthetics have been used to manipulate consciousness in patients for nearly 170 years, but it is still not known how these drugs impart hypnosis. At the molecular level, the number of possible effector sites is staggering: dozens of molecules are known to be sensitive to anesthetic agents, including many types of ion channels (reviewed in [1, 2]), gap junction channels [3], and G protein-coupled receptors [4]. Furthermore, it is clear that there is no single molecular site of action shared by all anesthetic agents [1]. Thus, the actions of general anesthetics must be understood in the context of neural anatomy and network connectivity.

Though anesthetic-induced hypnosis and natural sleep are distinct states, they share many similarities (reviewed in [5, 6]), leading to the increasingly popular theory that anesthetics may induce hypnosis by acting on endogenous arousal neural circuitry [7]. Much of the recent research has focused on anesthetics inhibiting wake-active nuclei such as the tuberomammillary nucleus [7, 8], but it remains unclear to what extent sleep-promoting nuclei, such as the ventrolateral preoptic nucleus (VLPO), are involved in generating the hypnotic state.

The VLPO is a predominately sleep-active nucleus containing GABAergic and galaninergic neurons that project to many arousal-promoting nuclei throughout the neuroaxis [9]. Several general anesthetics including chloral hydrate, propofol, various barbiturates, dexmedetomidine, and isoflurane have been shown to increase the number of active VLPO neurons [7, 10, 11]. Yet, ablation of VLPO neurons, which would be predicted to produce resistance to anesthesia, is known to cause an accrual of sleep debt [12] and has recently been reported to cause increased sensitivity to isoflurane anesthesia [13]. Thus it remains unclear whether VLPO activation contributes to anesthetic-induced hypnosis or if it is a secondary effect unrelated to behavioral state [14].

In the present study, we demonstrate that isoflurane dose-dependently increases the number of active VLPO neurons, but not at a sub-sedative dose or when the animals' behavioral state is reversed via pressure reversal. By using whole-cell recordings in hypothalamic slices, we identify isoflurane-activated neurons as belonging specifically to the putative sleep-promoting subpopulation of VLPO neurons. We demonstrate that isoflurane acts directly on these neurons to reduce a basal potassium conductance and thereby increase inward (depolarizing) current. Finally, targeted lesioning of the VLPO produces an acute resistance to induction by isoflurane. These results are consistent with anesthetic agents acting on the endogenous arousal neural circuitry to produce hypnosis, and suggest that the VLPO plays a critical role in anesthetic induction.

Results

Hypnotic doses of isoflurane or halothane increase expression of c-Fos in a subset of VLPO neurons

To determine whether VLPO neurons were active during volatile anesthetic-induced hypnosis, we exposed mice to oxygen with or without volatile anesthetics for two hours, either during the period of maximal activity following lights-out (“dark phase”), or during the period of maximal sleep following lights-on (“light phase”). Following sacrifice, we analyzed immunohistochemical expression of c-Fos, a marker of antecedent neuronal activity. Consistent with previous reports [15, 16], and in contrast to most brain regions [5, 6, 17, 18], the VLPO of non-anesthetized mice sacrificed during the light phase had a two-and-a-half-fold increase in the number of c-Fos positive nuclei compared to mice sacrificed during the dark phase (p < 0.001; Figure 1). Similar increases were also observed for sedative and hypnotic levels of isoflurane: two-hour exposures to 0.6% isoflurane produced an increase of 215% ± 79% (p < 0.001) in c-Fos positive counts, and 1.2% isoflurane produced an increase of 179% ± 28% (p < 0.01). To determine the generalizability among inhaled volatile agents we tested an equipotent dose of halothane at 1% [19]. Halothane similarly increased the number of c-Fos reactive neurons in VLPO (150% ± 33%, p < 0.01). There were no significant differences in c-Fos counts between 0.6% isoflurane, 1.2% isoflurane, and 1.0% halothane (p > 0.05). Conversely, a sub-sedative dose of 0.3% isoflurane had no effect (76% ± 50%, p > 0.05). To determine whether natural sleep and anesthesia have an additive effect on the number of c-Fos cells in VLPO, we exposed mice in their light phase to 1.2% isoflurane and found no difference in c-Fos expression compared to a 1.2% isoflurane dark phase exposure (decrease of 12% ± 7%, p > 0.05).

Figure 1. VLPO c-Fos immunoreactivity was increased following exposure to volatile anesthetics.

Panels A-F depict sample coronal sections through the hypothalamus showing staining of c-Fos (brown nuclei) following a 2 hr exposure to (A) oxygen control, dark phase; (B) 0.3% isoflurane, dark phase; (C) 0.6% isoflurane, dark phase; (D) 1.2% isoflurane, dark phase; (E) 1.0% isoflurane at 70 atm, light phase; (F) 1.0% halothane, dark phase. Scale bar in A corresponds to 50 μm and applies to all panels A-F. (G) Bar graph summarizing c-Fos expression in VLPO. Counts are per unilateral VLPO in a 10 μm slice. Cell counts were analyzed using an ANOVA with post-hoc t-tests using the Bonferroni correction for multiple comparisons. Error bars represent SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001 as compared to the non-anesthetized dark phase control; †††, p < 0.001 as compared to 1.2% isoflurane during light phase.

To explore the behavioral relevance of VLPO activation during hypnosis, we used pressure reversal to separate anesthetic exposure from hypnosis. While there are no known chemical antagonists capable of reversing surgical planes of anesthesia [20], it has been well known for over half a century that exposing organisms to high external pressures can reverse the actions of many anesthetics [21-23]. Specifically, 70 atmospheres (atm) is known to decrease isoflurane potency by more than 50% in mice [22]. After being anesthetized with 1.0% isoflurane at 1 atm for 30 minutes, the mice were pressurized to 70 atm with helium gas and remained at pressure for 2 hours, during which time the mice regained their righting reflex, as observed through a view port. Pressure reversal resulted in an 85% ± 2% reduction in c-Fos positive neurons in VLPO compared to mice anesthetized with 1.2% isoflurane at 1 atm (p < 0.001). Consistent with the behavioral reversal of anesthesia, pressure reversal fully reverted levels of c-Fos expression back to that of wakefulness (Figure 1) with no differences among the isoflurane-exposed 70 atm pressure-reversed group, the 1 atm awake oxygen control group, and the 70 atm pressure-awakened oxygen control group (p > 0.05).

Electrophysiological classification of VLPO neurons

Forty-nine VLPO neurons were recorded from acutely prepared preoptic hypothalamic slices obtained from 40 anesthetic-naïve mice. Neurons were classified according to their response to a bath application of 100 μM norepinephrine (NA): 15 neurons were strongly depolarized by NA (5.41 mV ± 0.97 mV, p < 0.01) and were classified as NA(+), while 34 were hyperpolarized (-3.61 ± 0.80 mV, p < 0.01) and were classified as NA(-) (Figure 2). In keeping with previous literature [24], the majority (79%) of NA(-) neurons had clearly-discernible low threshold spikes (LTS) following hyperpolarization (Figure 2B, insert), whereas few NA(+) cells (13%) exhibited an LTS. The NA(-), LTS VLPO neurons are believed to be the putative sleep-promoting subset, while their neighboring NA(+) neurons that lack LTS are state-indifferent. Morphological analysis of biocytin-filled neurons revealed that all NA(+) cells were bipolar, while NA(-) cells tended to be multipolar, as has previously been reported [24-26]. Basic membrane properties such as membrane resistance and action potential characteristics did not differ between these two neuronal subtypes (Table 1).

Figure 2. Putative sleep-active VLPO neurons (NA(-)) were depolarized by ex vivo exposure of isoflurane.

Panels A and B show sample traces from a NA(+) (A) and a NA(-) (B) VLPO neuron during exposure to norepinephrine (NA) and isoflurane, with insets depicting membrane responses to hyperpolarizing current injections (-100, -80, and -60 pA). The majority of NA(-) neurons (79%) showed clear evidence of a rebound low-threshold spike (B, arrowhead in inset), whereas few of the NA(+) neurons (13%) exhibited this phenomenon. (C) Bar graph depicting change in firing rate relative to baseline. (D) Bar graph depicting change in membrane potential relative to baseline. Recordings were performed on 15 NA(+) and 34 NA(-) neurons and analyzed using two-way ANOVAs with post-hoc t-tests using the Bonferroni correction for multiple comparisons. Error bars represent SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001; all comparisons were made against baseline.

Table 1. Basic electrophysiological membrane properties.

Basic membrane properties did not differ between NA(+) and NA(-) VLPO neurons.

	RMP (mV)	IR (MΩ)	Tau (ms)	AP threshold (mV)	AP amplitude (mV)	AP duration (ms)
NA(+) (n=15)	-57.7 ± 2.6	595.8 ± 51.6	23.4 ± 2.1	-33.9 ± 1.7	52.9 ± 2.1	1.6 ± 0.1
NA(-) (n=34)	-56.8 ± 1.0	602.8 ± 34.0	23.8 ± 1.4	-31.2 ± 0.7	49.6 ± 2.0	1.7 ± 0.1

Isoflurane directly depolarizes VLPO NA(-) neurons by decreasing potassium conductance

Two concentrations of isoflurane dissolved in artificial cerebrospinal fluid (aCSF) were used: 240 μM (roughly equivalent to 1.2% in oxygen) and 480 μM (roughly equivalent to 2.4% in oxygen). Neither dose had a significant effect on NA(+) neurons, though there was a consistent trend towards a reduction in firing rate (-0.46 Hz ± 0.28 and -0.76 Hz ± 0.54, respectively) and membrane potential (-1.96 mV ± 1.88 and -2.83 mV ± 3.59, respectively). However, as shown in Figure 2, exposing NA(-) neurons to 240 μM and 480 μM isoflurane resulted in a depolarization (+6.36 mV ± 1.20, p < 0.05; +11.56 mV ± 2.24, p < 0.001) and an increase in firing rate (+1.29 Hz ± 0.29, p < 0.001; +2.53 Hz ± 0.52, p < 0.001), nearly tripling or quadrupling their basal firing rates.

To determine whether isoflurane was acting pre- or post-synaptically on VLPO NA(-) neurons, we used two independent methods to block synaptic transmission: (1) a combination of the non-selective AMPA and kainate receptor antagonist 6,7-dinitroquinoxaline-2,3-dione (DNQX, 20 μM), the NMDA receptor antagonist (2R)-amino-5-phosphonopentanoate (AP5, 100 μM), and the GABA_A receptor antagonist bicuculline (Bic, 20 μM); and (2) aCSF enriched in Mg²⁺ (9.68 mM) and depleted in Ca²⁺ (160 μM) to block synaptic vesicular release. Both methods were effective at completely abolishing all postsynaptic events (Figure 3A), but the isoflurane-induced depolarization and activation of NA(-) VLPO neurons persisted (Figure 3B&C). In the presence of DNQX/AP5/Bic, 240 μM isoflurane (n=7) produced an increase in firing rate (+1.30 Hz ± 0.85, p < 0.05) and depolarized the resting membrane potential (+6.23 mV ± 1.63, p < 0.05). In the presence of high Mg²⁺ and low Ca²⁺, 240 μM isoflurane (n=3) produced an increase in firing rate (+1.19 Hz ± 0.30, p < 0.05), but the depolarization of the membrane potential was not significant (+6.47 mV ± 2.83, p > 0.05); 480 μM isoflurane (n=3) produced an increase in firing rate (+2.36 Hz ± 0.54, p < 0.01) and significantly depolarized the membrane potential (+12.64 mV ± 6.15, p < 0.01).

Figure 3. Activation of NA(-) VLPO neurons by isoflurane persisted in the presence of synaptic blockade.

(A) Sample voltage-clamp traces from VLPO neurons inhibited by norepinephrine (NA(-)) show that both excitatory (downward deflections) and inhibitory (upward deflections) postsynaptic currents were eliminated by the administration of 20 μM DNQX, 100 μM AP5, and 20 μM bicuculline (n=7), or by the replacement of normal artificial cerebrospinal fluid (aCSF) with a mixture where Ca²⁺ content was reduced to 160 μM and Mg²⁺ content increased to 9.68 mM (n=3). (B & C) Isoflurane, at both 240 μM and 480 μM, was effective at increasing the firing rate (B) and depolarizing the resting membrane potential (C) in NA(-) neurons, both in normal aCSF and in preparations where all postsynaptic activity has been abolished. Data were analyzed using two-way ANOVAs with post-hoc t-tests using the Bonferroni correction for multiple comparisons. Error bars represent SEM. n.d., not determined; *, p < 0.05; **, p < 0.01; ***, p < 0.001; all comparisons were made against baseline.

Isoflurane (240 μM) had no effect on the frequency or amplitude of excitatory or inhibitory postsynaptic currents (EPSCs, IPSCs), and only altered the kinetics of IPSCs, with a prolonged half-width (p < 0.05; Figure 4B and Table S1). In contrast, 240 μM isoflurane produced a pronounced inward current of 12.57 pA ± 2.01 when NA(-) neurons were held at -60 mV (p < 0.001). This shift was preserved when 240 μM isoflurane was introduced in the presence of synaptic inhibitors 20 μM DNQX, 100 μM AP5, and 20 μM bicuculline, resulting in an inward current of 13.68 pA ± 3.17 (p < 0.001) compared to 4.62 ± 3.66 (p > 0.05, n.s.) in the presence of the inhibitors alone (Figure 4A&C).

Figure 4. Pre- and postsynaptic effects of isoflurane upon VLPO NA(-) neurons.

(A) Sample voltage-clamp traces from a NA(-) neuron clamped at -60 mV showing an inward current during isoflurane exposure, which persisted during synaptic blockade. (B) Average excitatory post synaptic current (EPSC; top) and inhibitory post-synaptic current (IPSC; bottom) traces recorded from a NA(-) neuron (100 traces averaged). Isoflurane (240 μM) increased IPSC half-width, but had no effect on EPSC frequency or amplitude (see Table S1). (C) Administration of 240 μM isoflurane increased inward current in NA(-) cells held at a fixed voltage (-60 mV), an effect that persisted during synaptic blockade with 20 μM DNQX, 100 μM AP5, and 20 μM bicuculline (n=7). Data were analyzed using two-way ANOVAs with post-hoc t-tests using the Bonferroni correction for multiple comparisons. Error bars represent SEM. ***, p < 0.001; all comparisons were made against baseline.

The isoflurane-induced depolarization observed in NA(-) neurons was accompanied by an increase in membrane resistance (+377.5 MΩ ± 106.4, p < 0.05) in four out of four neurons examined (Figure 5A&B). To determine whether the increased resistance and depolarizing inward current were due to a decreased conductance of chloride or potassium, the chloride reversal potential was shifted from -75 mV to -15 mV, but this had no effect on the magnitude of the inward current (data not shown). However, when NA(-) neurons were held at the reversal potential for potassium (E_K = -100 mV), the inward current was drastically reduced (0.65 pA ± 0.52, p < 0.01) and did not differ from 0 pA (p > 0.05; Figure 5C).

Figure 5. Isoflurane exposure increased the membrane resistance of VLPO NA(-) neurons due to a reduction in potassium conductance.

(A) Sample trace from a NA(-) neuron showing membrane voltage response to bath administration of 240 μM isoflurane; membrane resistance was monitored by measuring the magnitude of the downward voltage deflections in response to -30 pA injected current pulses. Exposure to 240 μM isoflurane depolarized the NA(-) neuron and increased membrane resistance. This increase in resistance was maintained when negative current was applied to return the cell to its baseline resting potential in the presence of isoflurane (not shown). (B) Isoflurane significantly increased membrane resistance in all four NA(-) neurons (p < 0.05). (C) Using voltage clamp techniques, the inward current observed during isoflurane exposure was diminished when the neurons were held at E_K (-100 mV). The current produced by isoflurane at -100 mV was not significantly different from 0 pA. Error bars represent SEM. *, p < 0.05; **, p < 0.01.

Lesions of VLPO neurons produce initial resistance to isoflurane-induced hypnosis

A targeted galanin-saporin was used to lesion VLPO bilaterally (n=15), with an unconjugated saporin used for sham lesions (n=7) (Figure 6A-D). The lesioned group had an average neuronal loss of 91% ± 2% in VLPO including the expected specific loss of galanin neurons (Figure S1). At six and twenty-four days following surgery, all mice were assessed in parallel for their sensitivity to isoflurane using a loss of righting reflex (LORR) paradigm. As depicted in Figure 6E, at six days lesioned mice required higher doses of isoflurane to lose their righting reflex than those of their sham cohorts: the effect of the lesion was a right shift in the LORR dose-response curve, with a shift of the ED₅₀ from 0.85% (95% confidence interval [CI]: 0.84%–0.86%) to 0.95% (CI: 0.93%–0.96%) (Table S2), a dose by which all sham animals had lost their righting reflex. At 24 days this effect was reversed in VLPO-lesioned mice, which then demonstrated a left-shift of the LORR dose-response curve and a significantly decreased ED₅₀ of 0.78% (CI: 0.75%–0.80%). VLPO lesions produced no effect on body weight or core body temperature (data not shown; F(1,37)=1.292, p > 0.05 and F(1,37)=0.1810, p > 0.05, respectively).

Figure 6. Mice with targeted VLPO lesions became acutely resistant to isoflurane-induced hypnosis.

4× DAPI (A,C) and 20× NeuN (B,D) images of a sample sham (A,B) and lesioned (C,D) animal (see also Figure S1). Lesion efficacy was assessed via cell counts of NeuN-positive cells; the white boxes in (B) and (D) are 250 μm tall by 400 μm wide and correspond to the location used for counting. Two animals out of 17 in the lesion cohort that had less than 80% cell loss in the VLPO compared to shams were excluded from analysis (n=7 in the sham group). Average neuronal loss in VLPO for lesioned animals was 91% ± 2%. (E) Dose-response curves showing the fraction of unconscious animals as judged by a loss of righting reflex in response to stepwise increases in isoflurane concentrations at 6 and 24 days post-surgery; the horizontal axis is displayed on a logarithmic scale. VLPO lesions produced a significant rightward shift of the induction dose-response curve at day 6, and a significant leftward shift at day 24 (Table S2). (E, Insert) Bracketed ED₅₀ values as calculated by averaging the last concentration at which an animal had an intact righting reflex with the first concentration at which an animal lost its righting reflex. Error bars represent SEM. *, p < 0.05; ***, p < 0.001.

Discussion

The putative sleep-active VLPO neurons project extensively throughout the central nervous system and play a critical role in arousal regulation. While most neurons within the brain are inactivated upon induction of general anesthesia, herein we demonstrate that the volatile anesthetics can directly activate putative sleep-promoting VLPO neurons. VLPO activation occurs in a dose-dependent fashion at concentrations that precede the onset of anesthetic hypnosis, but does not occur at a low, sub-sedative dose. Using ex vivo electrophysiology we further show that isoflurane only activates VLPO neurons that are hyperpolarized by NA and exhibit LTS, both of which are hallmarks of cells whose enhanced firing is believed to cause the onset and maintenance of natural sleep. Consistent with this, our c-Fos studies show no difference in VLPO c-Fos expression between mice sacrificed during a period of maximal sleep and anesthetized mice, and no additivity when anesthesia is delivered during the rest phase. To strengthen the correlation between anesthetic-induced activation of VLPO and the onset of behavioral hypnosis, we further establish that lesioning of the VLPO is associated with an acute resistance to anesthetic hypnosis. Conversely, when the anesthetic state is antagonized despite ongoing exposure to an anesthetic, c-Fos immunoreactivity in VLPO reverts to awake-like levels, suggesting that the VLPO may be capable of integrating arousal state inputs and needn't be influenced exclusively by the anesthetic.

VLPO activation is a common point of neuronal convergence for many anesthetics as well as sleep

Studies using c-Fos protein immunoreactivity as a marker of antecedent neuronal activity have demonstrated that propofol [7], barbiturates [7], dexmedetomidine [10], and chloral hydrate [11], among others, increase the number of c-Fos positive neurons in the VLPO, but whether this translates into increased neuronal activation has not been shown. In the current study we demonstrate that the volatile anesthetic isoflurane not only increased the expression of c-Fos in VLPO neurons, but but also enhanced neuronal excitation. Lu et al. [11] previously demonstrated that isoflurane exposure increases c-Fos expression in the VLPO, though the anesthetic dose was not measured. Here, we show that clinically-relevant hypnotic and sedative doses of isoflurane and halothane were effective at increasing c-Fos in the VLPO, while a sub-sedative dose was not. However, a major drawback to using c-Fos as an indicator of neuronal activity in VLPO is that the VLPO is not a homogenous nucleus. The VLPO consists of two very distinct subpopulations of neurons: two-thirds of the neurons are characterized as being primarily multipolar, having a low-threshold spike (LTS) produced by T-type calcium channels, being inhibited by the wake-promoting neurotransmitter NA, and these are the cells that are believed to be sleep-promoting; the remaining one-third are typically bipolar, do not show LTS, are excited by NA, and are state-indifferent [15, 24, 25]. Aside from electrophysiological differences, there are currently no known markers available to distinguish between these two subpopulations. Therefore, we used ex vivo whole-cell electrophysiology to establish that the NA(-), but not NA(+), VLPO neurons show increased excitability during exposure to isoflurane.

Isoflurane had no observable effect on NA(+) neurons: there was no significant effect on either membrane potential or firing rate at either isoflurane concentration, demonstrating divergent electrophysiological responses for adjacent NA(+) and NA(-) neurons. Though there was a trend towards a reduction in firing rate in NA(+) neurons, a floor effect may have prevented us from observing the full effects of the drug: both cell types had an average basal firing rate of approximately 1 Hz. Conversely, the putative sleep-active NA(-) neurons were profoundly affected by isoflurane: both doses were effective at depolarizing the neurons and increasing the firing rate. One possible mechanism by which isoflurane could be exciting VLPO NA(-) neurons is by acting presynaptically to increase glutamate release onto VLPO neurons. In single-cell dissociated neuronal preparations, propofol is reported to increase glutamate release from presynaptic terminals by acting on presynaptic GABA_A receptors, which results in a depolarization due to a shifted chloride gradient [25]. We found that isoflurane, however, had no effect on glutamatergic EPSC frequency (Table S1), ruling out this potential mechanism. Isoflurane did increase the halfwidth of GABA_A-mediated IPSCs, as has been observed in other neuronal populations [27]. However, isoflurane's potentiation of inhibitory GABAergic signaling should oppose rather than explain isoflurane's depolarizing effect on NA(-) neurons.

Blocking all synaptic input onto VLPO neurons had little impact on isoflurane's effect on VLPO NA(-) neurons' membrane potential and firing rate. In the presence of antagonists blocking all presynaptic activity, or in the presence of high Mg²⁺ and low Ca²⁺ to prevent synaptic vesicular release, isoflurane still depolarized the NA(-) neurons and produced an increase in firing rate. Consistent with this, in the absence of any synaptic blockers the only significant effect of isoflurane in voltage-clamp mode (in addition to the aforementioned increase in GABA IPSC halfwidth) was an increase in inward current. These findings demonstrate that isoflurane must have a direct extrasynaptic effect on VLPO NA(-) neurons. While surprising, there is existing evidence for such an effect: isoflurane, like most volatile anesthetics, is a promiscuous drug known to affect a wide range of targets [28]. Of isoflurane's known receptor targets, the majority are modulated in a way that would be predicted to lead to an inhibitory effect, but isoflurane can depolarize cells by potentiating cation channels [29, 30] or by closing anion channels such as the tandem pore potassium channels K_2P12.1 and K_2P13.1 [31] (reviewed in [2, 17, 32]). The latter is a congruent mechanism to explain our isoflurane-induced increase in membrane resistance, which is accompanied by an inward current that is abolished at the reversal potential for potassium. Additionally it should be noted that we cannot exclude a voltage-gated potassium channel from underlying our effect as the cumulative effects of low open probability could be overcome by abundant channel expression. Hence, although the exact molecular mechanisms through which isoflurane depolarizes NA(-) VLPO neurons remains presently unknown, isoflurane's mechanism is clearly distinct from the disinhibition of VLPO that is thought to underlie dexmedetomidine's hypnotic actions [10].

VLPO activation modulates, but is not required for, anesthetic hypnosis

Both natural sleep and anesthetic-induced hypnosis produce similarly increased numbers of active neurons in VLPO, but there is no additive effect when naturally-sleeping mice are anesthetized with isoflurane. Furthermore, by using pressure reversal to decouple anesthetic exposure from its behavioral endpoint of hypnosis, our results demonstrate that VLPO activity correlates with the state of behavioral arousal, rather than with the presence or absence of the anesthetic. There are no specific antagonists of general anesthesia; however other acute interventions including microinjections targeting the centromedial thalamus, as well as systemic administration of physostigmine, doxapram, and methylphenidate all have partial efficacy [33-35]. Pressure reversal has been shown to be effective at a wide range of concentrations for many anesthetics [23]. Though isoflurane is capable of directly activating sleep-promoting VLPO neurons, when animals exposed to hypnotic levels of isoflurane were behaviorally aroused via pressure reversal c-Fos counts in VLPO were no different from awake controls, suggesting that the VLPO is integrating multiple arousal-related inputs.

The VLPO plays a critical role in promoting sleep [9], and the data presented here argue for a similar role in anesthetic-induced hypnosis. Yet, it is clear that activation of sleep-promoting VLPO neurons is not a requirement for sleep–i.e., eszopiclone-induced sleep does not increase VLPO c-Fos [36]–nor for the onset of anesthetic-induced hypnosis, as has been demonstrated for ketamine [11]. However, the sleep-active neurons of the VLPO form one site of convergence entwining the neurobiology of endogenous arousal state regulation with anesthetic-induced hypnosis. Consequently, the VLPO is a natural candidate for translating the physiological effects of pre-existing sleep deprivation upon resulting anesthetic hypersensitivity [37, 38].

We used a targeted saporin to lesion galaninergic neurons within VLPO and an unconjugated saporin as a sham control. Previous studies lesioned VLPO with ibotenic acid or orexin-saproin [12, 13] and may have destroyed a different subset of neurons. However, with a similar loss of greater than 90% of all VLPO neurons, galanin-saporin treatment produced a short-term resistance to induction by isoflurane six days following the lesion, resulting in a right-shift of the LORR dose-response curve, shifting the ED₅₀ from 0.85% to 0.95% (Table S2). This is consistent with a published report showing that VLPO lesions produce a foreshortened duration of action of dexmedetomidine [10], but it is in contrast to a report showing increased burst suppression ratios–an EEG-dependent measure of anesthetic depth–during exposure to hypnotic levels of isoflurane due to an accumulation of lesion-produced sleep loss [13]. Because sleep loss is a known confound of anesthetic sensitivity [37, 38], our goal was to examine the effects of the lesion before any significant amount of sleep loss could accrue. Excitotoxic lesions of the VLPO produce pronounced sleep/wake architecture changes by seven days post-lesion [12], but targeted-saporin lesions have a lag time of four days before significant cell loss is observed [39]. Therefore, we chose six days following saporin lesion to maximize cell loss while minimizing sleep debt accrual. When we reexamined the mice 18 days later, the lesion group was more sensitive to induction by isoflurane with an ED₅₀ of 0.78% compared to 0.85% for the sham group (Table S2). Although we did not use polysomnography to measure sleep and wakefulness in our lesioned mice, we view our findings as consistent with those of Eikermann et al. [13]: VLPO lesions produce an acute resistance to isoflurane anesthesia that over time is reversed to a sensitization due to sleep-debt accumulation [12]. Importantly, despite our best attempts it is possible that an accruing sleep debt 6-days following targeted nanoinjection of galanin-saporin may have minimized a potentially larger magnitude shift in anesthetic resistance that could have resulted from a true acute pharmacologic or genetic inactivation of VLPO. Furthermore, the non-sleep-active NA(+) neurons are also galaninergic [40] and were thus also susceptible to the galanin-saprorin lesion; we would predict that lesioning solely the NA(-) VLPO neurons would produce a larger effect. Based upon preserved temperature and body weight regulation in our lesioned mice, our small volume nanoinjections did not spread to the nearby adjacent median preoptic area, which participates in thermoregulation [41] nor the ventromedial hypothalamus, which contributes to maintenance of body weight [42].

The relatively small magnitude of the effects of a VLPO lesion on the ED₅₀ of the LORR curve (12% of sham at 6 days), and the fact that this shift can be reversed if the animal's sleep architecture is allowed to be disrupted, suggests that though it may play an important role, the VLPO is not the sole nor master mediator of anesthetic-induced hypnosis. But this is not unique to anesthesia: VLPO lesioned rats still sleep [12]. Other sleep-promoting neurons have been discovered in the median preoptic nucleus [43], basal forebrain [44], and cortex [45]. Hence, it is likely that despite a VLPO lesion, other sleep-promoting groups remain intact as would any hypnotic component arising through the modulation of activity in these or other state-dependent neurons.

Conclusion

While anesthetics are known to increase transcription of the immediate early gene c-fos in VLPO neurons, it has not previously been established whether the activation is a consequence of the inhibition of wake-active sites or is independently due to actions of the anesthetic agents on VLPO neurons. In the current study, we show that for isoflurane this activation is a direct postsynaptic effect resulting from the closure of a background potassium conductance, and is specifically limited to the presumed sleep-active NA(-) subpopulation of VLPO neurons. Furthermore, the number of active neurons in the VLPO, as measured by c-Fos, corresponds to the behavioral state of the animal: mice that are naturally sleeping or exposed to hypnotic or sedative levels of a volatile anesthetic show relatively high numbers of c-Fos-positive neurons in the VLPO, whereas mice that are naturally awake or behaviorally reversed in the presence of an anesthetic show relatively fewer active neurons in VLPO. While there is still much debate as to how general anesthetics mediate hypnosis, the data presented here are consistent with volatile anesthetics acting via the endogenous sleep circuitry. The specificity to which isoflurane targets solely the sleep-active subpopulation of neurons in VLPO and the tight coupling of VLPO activity with behavioral state is suggestive of a causative, functionally significant relationship between the activation of VLPO by volatile anesthetics and the onset of hypnosis.

Experimental Procedures

Previously-described protocols were used for righting-reflex assessment of anesthetic sensitivity [18, 19, 46] and slice electrophysiology [47]. Full descriptions of these and other materials and methods are available in the online Supplemental Experimental Procedures.

Hypnotic doses of isoflurane directly depolarize a subset of hypothalamic neurons

This subset in VLPO is electrophysiologically identical to those that regulate sleep

Lesioning VLPO produces acute resistance to induction with isoflurane

Systems for promoting natural sleep and anesthesia are mechanistically entwined