Volatile Anesthetics Activate a Leak Sodium Conductance in Retrotrapezoid Nucleus Neurons to Maintain Breathing during Anesthesia in Mice

The protocol was reviewed and approved by the Committee of Animal Welfare of West China Hospital of Sichuan University (Chengdu, China), and relevant aspects of the Animal Research Reporting In Vivo Experiments (ARRIVE) guidelines were followed throughout. C57BL/6 mice were housed in standard conditions with a 12-h light/dark cycle and free access to food and water. To measure the breathing parameters, 12-week-old mice (injected with virus at 8 weeks, see below) were used. Neonatal mice at 7 to 12 days old were used for patch-clamp recordings.

The mice were anesthetized with ketamine/xylazine (60/10 mg/kg) and then transcardially perfused with 4% paraformaldehyde in 0.1 M phosphate-buffered saline, pH 7.4. The perfused brains were removed and put into 4% paraformaldehyde solution overnight, followed by 30% sucrose for 1 day. Transverse sections of brain stem (–5.5 to –7.5 mm to Bregma) were cut (12 μm) using a freezing microtome (CM1850; Leica, USA). Sections were double-labeled by incubating at 4°C overnight with primary antibodies, including sodium leak channel (1:800, mouse, SMC-417, StressMarq Biosciences, Canada), NeuN (1:400, rabbit, ab104225, Abcam, Cambridge, USA), NeuN (1:400, mouse, MAB377, Merck Millipore, Darmstadt, Germany), and Phox2b (1:500, rabbit, A92903N, Thermo Fisher, USA). Then the sections were incubated with secondary antibodies at room temperature for 2 h: Alexa Fluor 647 goat anti-mouse 1:200 (115-605-003) and Alexa Fluor 488 goat anti-rabbit 1:200 (111-545-003; Jackson ImmunoResearch, USA). All photographs were captured using a Zeiss AxioImager Z.2. and prepared using ImageJ (National Institutes of Health, USA).

Brain stem slices containing the retrotrapezoid nucleus area were prepared as previously described.²⁵  Briefly, neonatal mice were decapitated under ketamine/xylazine (60/10 mg/kg) anesthesia, and transverse brain stem slices (300 μm) were cut using a vibratome (VT1000 A; Leica) in ice-cold substituted Ringer solution containing 260 mM sucrose, 3 mM KCl, 5 mM MgCl₂, 1 mM CaCl₂, 1.25 mM NaH₂PO₄, 26 mM NaHCO₃, 10 mM glucose, and 1 mM kynurenic acid. The slices were incubated for 30 min at 37°C and subsequently at room temperature in normal Ringer solution 130 mM NaCl, 3 mM KCl, 2 mM MgCl₂, 2 mM CaCl₂, 1.25 mM NaH₂PO₄, 26 mM NaHCO₃, and 10 mM glucose. Both the substituted and normal Ringer solutions were equilibrated with 95% O₂, 5% CO₂, with pH 7.35. Two or three neurons were recorded from each animal. In total, 69 wild-type neonatal mice (postnatal 7 to 12 days, 32 males and 37 females) and 36 sodium leak channel short hairpin RNA–injected mice (postnatal 10 to 12 days, 17 males and 19 females) were included in the study.

Each individual brain slice was mounted in a recording chamber submerged in a continuously perfused external solution (~2 ml/min) and bubbled with 95% O₂, 5% CO₂, pH 7.35. Recordings were established using pipettes that had a resistance of 5 to 6 MΩ. Electrophysiologic recordings were conducted using an Axopatch 700B amplifier and Digidata1440 digitizer linked to a computer running pClamp 10.2 software (Molecular Devices, Sunnyvale, USA). The currents were sampled at 20 kHz and filtered at 10 kHz. Recordings were performed in either cell-attached or whole-cell configurations at room temperature (23 to 25°C). Holding currents and conductance were monitored over time by delivering –60 mV voltage steps every 170 ms. The current–voltage (I–V) relationship was determined using voltage steps between –60 and +80 mV (Δ10 mV). The pipette internal solution used for recording spontaneous firing rate (in cell-attached mode) contained the following: 120 mM KCH₃SO₃, 4 mM NaCl, 1 mM MgCl₂, 0.5 mM CaCl₂, 10 mM HEPES, 10 mM EGTA, 3 mM Mg-ATP, and 0.3 mM GTP-Tris, pH 7.2 (adjusted with potassium hydroxide). Sodium leak channel -mediated currents were measured in whole-cell voltage clamp (V_holding = –60 mV) using a cesium (Cs⁺)-based internal solution which contained the following: 104 mM Cs CH₃SO₃, 1 mM MgCl₂, 0.5 mM CaCl₂, 30 mM tetraethylammonium (TEA)-Cl, 10 mM EGTA, 3 mM Mg-ATP, 0.3 mM GTP-Tris, 10 mM HEPES, pH 7.2 (adjusted with CsOH). The external solution was the same as the normal Ringer solution but with the addition of TEA-Cl (20 mM), 4-aminopyridine (5 mM), tetrodotoxin (500 nM), bicuculline (10 μM), picrotoxin (100 μM), and cyanquixaline (10 μM) to block K⁺ channels, Na_v, and isolate retrotrapezoid nucleus neurons from synaptic/extra synaptic GABAergic and AMPA inputs. All recordings were maintained for several minutes to establish stable baseline conditions. Series resistance was typically less than 10 MΩ. The neuron was discarded and the data were not collected if the series resistance exceeded 15 MΩ. For wild-type neurons, evoked firing rates of less than 1.0 Hz under 10% CO₂ were excluded.

Propofol was purchased from AstraZeneca (UK). Isoflurane and sevoflurane were obtained from Abbott Pharmaceutical Co., Ltd. (China). Saturated stock solutions of isoflurane (10 to 12 mM) and sevoflurane (4 to 6 mM) were prepared by adding liquid isoflurane and sevoflurane into exocellular solution and shaken overnight. Concentrations of saturated stock solutions were confirmed by gas chromatography.²⁶  Stock solutions of isoflurane, sevoflurane, or propofol (10 mg/ml) were diluted in extracellular solution to the desired final concentrations in gas-tight glass syringes by exocellular solution. Isoflurane at 0.30 mM and sevoflurane at 0.35 mM²⁷  were used as the predicted minimum alveolar concentration (MAC) for rodents. Propofol at 20 μM was used in the patch-clamp recordings because this concentration is relevant to the plasma EC₅₀ concentration of propofol in vitro.^28,29 

The mice were randomly assigned to an experimental group using a random number table generator. For the neonatal mice used in the patch-clamp recordings, the pups were anesthetized with 2% isoflurane and fixed in a stereotaxic frame (RWD, Life Science, China). A micropipette loaded with adenovirus pDC316–CMV–ZsGreen1–U6–short hairpin RNA (sodium leak channel) or pDC316–CMV–ZsGreen1–U6–short hairpin RNA (scrambled; 2 × 10¹⁰ transducing units/ml; Xuanzun Bioscience, China) was guided to the retrotrapezoid nucleus area (2.5 mm caudal to lambdoid, 1 mm lateral to the midline, and 4.0 mm ventral to the surface).The rate of injection was 500 nl/min, with a total volume of 100 nl.

For the behavioral experiments, the mice were anesthetized with 2% isoflurane and fixed in a stereotaxic frame. A small hole was drilled in the skull, and a pipette filled with pAAV2-H1-short hairpin RNA–(sodium leak channel)–CAG–enhanced green fluorescent protein or pAAV2–scrambled–CAG–enhanced green fluorescent protein virus (2 × 10¹³ transducing units/ml; Taitool Bioscience, China) was injected bilaterally into the retrotrapezoid nucleus area (6.5 mm caudal to bregma, 1.3-1.4 mm lateral to the midline, and 5.2-5.5 mm ventral to the surface). The rate of injection was 500 nl/min, with total volume of 200 nl. After injection, all the mice were allowed to recover from anesthesia under a heating pad and treated with ampicillin (100 mg/kg, subcutaneously), atipamezole (2 mg/kg, subcutaneously), and ketoprofen (4 mg/kg, subcutaneously). The sodium leak channel and scrambled short hairpin RNA were selected as previously described:²²  AAGATCGCACAGCCTCTTCAT (sodium leak channel) and GCTCAGTACGATCATACTCAC (scrambled). The adult mice were available for respiratory activity recording 4 weeks after injection, and the pups were used for patch-clamp recordings 3 days after injection.

Respiratory activity was measured using a whole-body plethysmograph system (Data Scientific International, USA), utilizing an animal chamber maintained at room temperature and ventilated with air (0.5 l/min) at 10:00 am to 4:00 pm. For the original behavioral experiments, male adult mice (20 to 25 g, ~8 weeks old, n = 17, with 7 in the sodium leak channel -short hairpin RNA group and 7 in the scrambled-short hairpin RNA group, 3 used for location of fluorescence) were used. For the additional experiments in response to peer review, both male and female adult mice (20 to 25 g, ~8 weeks old; 3 females and 4 males in the sodium leak channel –short hairpin RNA group and 3 females and 4 males in the scrambled–short hairpin RNA group) were used. All the mice were individually placed into the chamber and allowed 2 h to acclimate before the starting the measurements. Respiratory activity was recorded using Buxco (USA) FinePointe software 2.0 for a period of 10 min in room air followed by exposure (10 min/condition) to graded increases in carbon dioxide (0, 3, and 5% CO₂ with balance O₂) under control conditions and in the presence of general anesthetics. MAC values (for loss of righting reflex) of isoflurane and sevoflurane were used as 0.8 and 1.5% in mice, respectively.³⁰  The intraperitoneal dose of propofol was 70 mg/kg (ED₅₀ for loss of righting reflex in mice).³¹  Respiratory frequency (breaths/min), tidal volume (ml/g, normalized to body weight), and minute ventilation (ml/min/g, normalized to body weight) were measured during a 30-s period of relative quiescence during the last minute of each condition. One technician who was blinded to the animal groups measured the respiratory outputs in vivo, and another researcher analyzed the data.

Sodium leak channel mRNA was measured by quantitative real-time polymerase chain reaction. The retrotrapezoid nucleus area was dissected under a microscope. Total RNAs was extracted using the Eastep Super RNA extraction kit (Promega, China). cDNA was synthesized with a GoScript reverse-transcription kit (Promega). Then the cDNA was used as a template to quantify the sodium leak channel expression level using GoTaq qPCR Master Mix (Promega) and specific primers (Sangon Biotech, China) according to the manufacturer’s protocol. The forward and reverse primers of sodium leak channel were 5´-GTCCTGACGAATCTCTGTCAGA-3´ and 5´-CTGAGATGACGCTGATGATGG-3´, respectively. The glyceraldehyde-3-phosphate dehydrogenase gene was used as an internal control. The polymerase chain reaction conditions were as follows: 3 min at 95°C and 40 cycles of 15 s at 95°C; 30 s at 55°C; and 30 s at 72°C.

The sample sizes of the electrophysiologic recordings were based on similar experimental designs, and no power calculation was conducted. Power analysis was used to determine the number of animals used in each experiment in vivo. Preliminary results (n = 4) showed that under room air conditions, the respiratory frequency of sodium leak channel genetically silenced mice and control mice was 249.3 ± 11.2 and 183.2 ± 6.3 breaths/min, respectively. Based on this to detect a 20% change in our primary outcome measures with 80% power (α = 0.05; β = 0.10), we needed seven animals per group for the plethysmography experiments. Note that the preliminary data used for power analysis were not included in the final data; therefore, the power analysis based on the preliminary test was not an interim analysis for final analysis, and the P values were not adjusted. Electrophysiologic data were analyzed using Clampfit 10.2 software (Molecular Devices, USA) and GraphPad Prism 8 (GraphPad Software, USA). All data are presented as means ± SD, and outliers, if any, were also included in the analyses. Statistical analysis was performed using SPSS version 23.0 (SPSS Inc., USA) and GraphPad Prism 8. The normality of the data was tested by the Shapiro–Wilk test. Repeated measures data were analyzed by two-way ANOVA with a Bonferroni post hoc test. Two-tailed independent-samples or paired Student’s t tests were used for comparison between control and anesthetized animals. No animal data were missing or excluded from the behavioral analysis. Approximately 10 to 15% cells from wild-type mice were discarded due to a change of firing rate of less than 1 Hz in response to 10% CO₂ in the patch-clamp recordings. The specific test used for each comparison is reported in the figure legend. Statistical significance was set at P < 0.05.

To determine whether sodium leak channel contributes to isoflurane modulation of retrotrapezoid nucleus neurons, we first confirmed the expression and function of sodium leak channel in Phox2b-positive retrotrapezoid nucleus neurons. Consistent with previous work,^22,23  we found that sodium leak channel was expressed by Phox2b-immunoreactive neurons in the retrotrapezoid nucleus (fig. 1A). We also confirmed that sodium leak channel was expressed by neurons functionally identified as retrotrapezoid nucleus (fig. 1B) chemoreceptors based on their firing response to CO₂/H⁺. In cell-attached voltage-clamp mode, chemosensitive retrotrapezoid nucleus neurons were spontaneously active (1.5 ± 1.0 Hz) under control conditions (5% CO₂) and respond to 10% CO₂ with at least a 1.0 Hz increase in firing rate (3.2 ± 1.0 Hz, P < 0.001, n = 10; fig. 1, C and D).

Consistent with previous studies,^22,23  we found that pharmacologic blockade of sodium leak channel by bath application of gadolinium (50 μM) decreased baseline activity of chemosensitive retrotrapezoid nucleus neurons from 1.2 ± 0.2 to 0.2 ± 0.1 Hz (P = 0.001, n = 5; fig. 1, E, F, and G), whereas activation of sodium leak channel by application of substance P (10 μM) increased firing rate from 0.7 ± 0.2 to 3.2 ± 0.9 Hz (P = 0.005 by paired t test, n = 5; fig. 1, E, F, and G).

Once a chemosensitive cell was identified, we went on to characterize the neuronal firing response to isoflurane. We found that exposure to isoflurane at both subanesthetic (0.20 to 0.25 mM, ~0.8 MAC) and anesthetic (0.42 to 0.50 mM, ~1.5 MAC) concentrations increased the firing rate of Phox2b-expressing retrotrapezoid nucleus neurons from 1.5 ± 0.8 to 2.9 ± 0.8 Hz (P = 0.020, n = 5, fig. 2, B and C) and 1.0 ± 0.4 to 2.7 ± 0.4 Hz (P < 0.001, n = 5, fig. 2, A, D, and E), respectively. The effects of isoflurane on retrotrapezoid nucleus neuronal activity was dose-dependent (P = 0.041, fig. 2F). Furthermore, bath application of gadolinium (50 μM) decreased isoflurane-stimulated neuronal activity of Phox2b neurons in the retrotrapezoid nucleus from 2.7 ± 0.5 to 0.2 ± 0.1 Hz (P = 0.002, n = 5; fig. 2, G and J). In whole-cell current-clamp mode and in the presence of tetrodotoxin (500 nM) to block neuronal action potentials, isoflurane at an anesthetic concentration of 0.42 to 0.50 mM (~1.5 MAC) depolarized resting membrane potential from −59.8 ± 2.5 to −55.0 ± 3.3 mV (P = 0.028, n = 5; fig. 2K). In addition, sodium leak channel was genetically silenced by sodium leak channel -short hairpin RNA in brain slices, and neurons with ZsGreen fluorescence were selected for recording (Supplemental Digital Content, fig. 1, A and B, https://links.lww.com/ALN/C456). Isoflurane at 0.42-0.50 mM (~1.5 MAC) did not increase firing rate of neurons in the retrotrapezoid nucleus in brain slices from sodium leak channel genetically silenced mice (Supplemental Digital Content, fig. 1C, https://links.lww.com/ALN/C456). Of note, ~33% of the sodium leak channel -silenced neurons were not spontaneously active and failed to respond to carbon dioxide. These results show that sodium leak channel regulates basal activity and transmitter modulation of retrotrapezoid nucleus neurons and is a candidate substrate responsible for isoflurane activation of neurons of the retrotrapezoid nucleus.

The effects of isoflurane on sodium leak channel -mediated currents in retrotrapezoid nucleus Phox2b neurons was characterized in whole-cell voltage-clamp (V_holding = −60 mV) using a Cs⁺-based internal solution. Consistent with our cell-attached voltage-clamp data (fig. 2F), we found that isoflurane activation of sodium leak channel conductance was concentration-dependent (P = 0.008; fig. 3F). Isoflurane at 0.20 to 0.25 mM (~0.8 MAC) increased holding current from −68.9 ± 6.7 to −109.8 ± 11.6 pA (P = 0.005, n = 5; fig. 3B) and increased conductance from 2.5 ± 0.8 to 3.7 ± 1.2 nS (P = 0.014, n = 5; fig. 3B). Higher concentrations of isoflurane (0.42 to 0.50 mM, ~1.5 MAC) increased holding current from −75.0 ± 12.9 to −130.1 ± 34.9 pA (P = 0.002, n = 6; fig. 3C) and increased conductance from 1.8 ± 0.5 to 3.6 ± 1.0nS (P = 0.001, n = 6; fig. 3D). Bath application of substance P (10 μM) increased holding current by 166.7 ± 56.4% and increased conductance by 189.1 ± 70.1% (fig. 3E). Replacing extracellular sodium with N-methyl d-glucamine apparently decreased holding current and conductance by 56.3 ± 22.7% (fig. 3H) and 28.6 ± 18.7% (fig. 3I), respectively. The effects of isoflurane on holding current and conductance were eliminated when extracellular Na⁺ was replaced with N-methyl d-glucamine (fig. 3H). Furthermore, isoflurane increased the slope of the I–V relationship, which appears voltage-independent and reverses potential near 0 mV (fig. 3, J and K). The effects of isoflurane on holding current and conductance was diminished in retrotrapezoid nucleus neurons in brain slices from sodium leak channel genetically silenced mice (fig. 3, L and M).

To test the contribution of sodium leak channel in retrotrapezoid nucleus chemoreceptors to isoflurane modulation on breathing, sodium leak channel -specific short hairpin RNA in an adeno-associated viral delivery system was bilaterally injected into the medial portion of the retrotrapezoid nucleus region. Approximately 4 weeks after injection, we measured baseline breathing and the ventilatory response to carbon dioxide in sodium leak channel genetically silenced mice and scrambled-short hairpin RNA (control) mice (Supplemental Digital Content, fig. 2, A and B, https://links.lww.com/ALN/C456). Consistent with a previous study,²²  mice in which sodium leak channel was genetically silenced in the retrotrapezoid nucleus exhibited normal respiratory activity under control conditions but showed a diminished minute ventilatory response to 5% CO₂ (1.9 ± 0.6 vs. 2.9 ± 0.8 ml/min/g, P = 0.026, n = 7, Supplemental Digital Content, fig. 2B, https://links.lww.com/ALN/C456) compared with mice injected with scrambled–short hairpin RNA. This respiratory phenotype was primarily mediated by a decrease in respiratory frequency (197.5 ± 48.1 vs. 264.4 ± 56.0 breaths/min, P = 0.033, n = 7; Supplemental Digital Content, fig. 2B, https://links.lww.com/ALN/C456). We then applied quantitative real-time polymerase chain reaction (Supplemental Digital Content, fig. 2C, https://links.lww.com/ALN/C456) and immunofluorescence staining (Supplemental Digital Content, fig. 3, https://links.lww.com/ALN/C456) to confirm expression of sodium leak channel. The mRNA levels of sodium leak channel in the retrotrapezoid nucleus were decreased by 42.0 ± 11.8% (P = 0.007, n = 6 per group; Supplemental Digital Content, fig. 2C, https://links.lww.com/ALN/C456) in the mice that received sodium leak channel –short hairpin RNA compared with the mice that received scrambled short hairpin RNA.

Baseline breathing and the ventilatory response to carbon dioxide under isoflurane were compared between control (scrambled–short hairpin RNA) and sodium leak channel genetically silenced mice (n = 7/group). To test the concentration-dependent effects of isoflurane on respiratory output, isoflurane at subanesthetic (0.8%, ~MAC for loss of righting reflex) and anesthetic (1.5%, ~MAC of immobility)³⁰  concentrations were tested. The subanesthetic concentration of isoflurane (0.8%) increased the respiratory output in control mice by increasing both the respiratory frequency (from 166.6 ± 20.7 to 202.0 ± 17.7 breaths/min, P = 0.002; fig. 4D) and minute ventilation (from 1.9 ± 0.4 to 2.5 ± 0.6 ml/min/g, P = 0.002; fig. 4F). Conversely, this same level of isoflurane decreased both the respiratory frequency (P = 0.008; fig. 4D) and minute ventilation (P = 0.049; fig. 4F) in sodium leak channel genetically silenced mice. These results are consistent with our cellular data and strongly suggest that sodium leak channel in retrotrapezoid nucleus neurons contributes to the excitatory effect of isoflurane on respiration at subanesthetic concentrations. Compared with 0.8% isoflurane, 1.5% isoflurane obviously decreased the respiratory output in control mice, including decreasing the respiratory frequency, tidal volume, and minute ventilation (fig. 4, D, E, and F).

To test the ventilatory responses to carbon dioxide under isoflurane anesthesia, control and sodium leak channel genetically silenced mice were successively exposed to 0.8 or 1.5% isoflurane followed by graded increases in hyperoxic hypercapnia of 0, 3, and 5% CO₂, which corresponded to arterial carbon dioxide levels of 45.4 ± 5.1, 69.5 ± 2.0, and 74.9 ± 1.2 mmHg, respectively (fig. 4G). Compared with control mice, sodium leak channel genetically silenced mice showed a diminished ventilatory response to carbon dioxide under both subanesthetic and anesthetic concentrations of isoflurane. This phenotype was primarily due to suppression of respiratory frequency. For example, under 0.8% isoflurane and in 5% CO₂, sodium leak channel genetically silenced mice breathed at a rate of 227.7 ± 37.6 breaths/min, whereas control mice showed a frequency of 278.2 ± 42.7 breaths/min (P = 0.037; fig. 4J), and this corresponded with a diminished minute ventilatory output (3.0 ± 0.5 vs. 3.9 ± 0.7 ml/min/g, P = 0.026 by two-tailed independent sample t test; fig. 4L). Similarly, under 1.5% isoflurane in 5% CO₂, sodium leak channel genetically silenced mice also showed a diminished breathing frequency (146.2 ± 29.3 vs. 175.2 ± 11.2 breaths/min; P = 0.031 by two-tailed independent samples t test vs. control mice; fig. 4J) and lower minute ventilation (1.4 ± 0.3 vs. 1.9 ± 0.3 ml/min/g; P = 0.007 by two-tailed independent samples t test vs. control mice; fig. 4L).

This study originally only investigated the effects of isoflurane. Sevoflurane and propofol were also analyzed in response to peer review. Propofol at 20 μM did not affect the holding currents (from −55.0 ± 16.9 to −53.3 ± 19.5 pA, P = 0.254) or conductance (from 2.7 ± 0.8 to 2.7 ± 0.9 nS, n = 6, P = 0.377; fig. 5, A and B). Sevoflurane at concentrations of 0.25 to 0.32 mM (~0.8 MAC, loss of righting reflex) enhanced sodium leak channel -mediated holding currents (from −44.97 ± 12.5 to −87.11 ± 30.5 pA, n = 5, P = 0.002) and conductance (from 1.4 ± 0.6 to 2.9 ± 0.6 nS, n = 5, P = 0.001; fig. 5, C and D). Accordingly, the firing rate of retrotrapezoid nucleus Phox2b neurons was suppressed by 20 μM propofol from 1.7 ± 0.7 to 0.6 ± 0.4 Hz (P = 0.013, n = 5; fig. 5, E and F). In contrast, sevoflurane at 0.25 to 0.32 mM increased the neuronal firing rate from 1.4 ± 0.5 to 2.6 ± 0.6 Hz, P = 0.006, n = 7; fig. 5, G and H). Bath application of gadolinium (50 μM) during exposure to sevoflurane suppressed neural activity (fig. 5, G and H).These results indicate that sevoflurane but not propofol can enhance sodium leak channel conductance in retrotrapezoid nucleus Phox2b neurons.

Subanesthetic concentrations (~1 MAC for loss of righting reflex) of isoflurane (~0.8%; fig. 6A) and sevoflurane (~1.5%; fig. 6C) increased the respiratory output in control mice. Compared with control mice, sodium leak channel genetically silenced mice exhibited a lower breathing frequency under isoflurane (209.3 ± 36.6 vs. 150.8 ± 31.55 breaths/min, P < 0.001, n = 7; fig. 6A) or under sevoflurane (214.9 ± 27.4 vs. 176.5 ± 32.7 breaths/min, P = 0.022, n = 7; fig. 6C). Sodium leak channel genetically silenced mice exhibited lower minute ventilation under isoflurane (2.7 ± 0.6 vs. 2.1 ± 0.5 ml/min/g, P = 0.047, n = 7; fig. 6A) or under sevoflurane (2.1 ± 0.3 vs. 1.4 ± 0.6 ml/min/g, P = 0.014, n = 7; fig. 6C). However, propofol at 70 mg/kg (i.p. ED₅₀ of loss of righting reflex) suppressed breathing (fig. 6E).

Ventilatory responses to increased carbon dioxide under 0.8% isoflurane (fig. 6B) or under 1.5% sevoflurane (fig. 6D) were diminished in sodium leak channel genetically silenced mice, as revealed by higher breathing frequency and minute ventilation in control mice than in sodium leak channel genetically silenced mice. However, genetic silencing expression of sodium leak channel in the retrotrapezoid nucleus produced no effect on breathing frequency, tidal volume, or minute ventilation in response to increased carbon dioxide under propofol anesthesia (fig. 6F).

For comparisons between sexes, there was no difference in respiratory frequency or minute ventilation responses to carbon dioxide between male (n = 4) and female (n = 3) mice. The following P values were obtained: for control mice, P = 0.542 for respiratory frequency, and P = 0.262 for minute ventilation (by two-way repeated measures ANOVA); for sodium leak channel genetically silenced mice, P = 0.241 for respiratory frequency, and P = 0.411 for minute ventilation (by two-way repeated measures ANOVA). Under anesthesia by isoflurane, sevoflurane or propofol, there was no difference in respiratory frequency between male and female mice: control mice, P = 0.175 for isoflurane, P = 0.292 for sevoflurane, and P = 0.671 for propofol (by two-tailed independent samples t test); sodium leak channel genetically silenced mice, P = 0.649 for isoflurane, P = 0.825 for sevoflurane, and P = 0.555 for propofol (by two-tailed independent samples t test).

This study identified sodium leak channel in chemosensitive retrotrapezoid nucleus neurons as a target of volatile anesthetics to activate respiratory activity. Specifically, we showed at the cellular level that isoflurane and sevoflurane increased excitability of retrotrapezoid nucleus phox2b neurons partly by enhancing sodium leak channel conductance. Further, at the whole-animal level, genetic silencing of sodium leak channel in the retrotrapezoid nucleus decreased the ventilatory responses to carbon dioxide and potentiated volatile anesthetic-induced respiratory depression. These results indicate that sodium leak channel in chemosensitive retrotrapezoid nucleus neurons are important determinants of respiratory activity during anesthesia of volatile anesthetics. In contrast, sodium leak channel-silenced mice showed no difference relative to control mice under propofol anesthesia. Understanding the cellular and molecular mechanisms contributing to maintenance of spontaneous breathing during general anesthesia is clinically relevant and may facilitate development of novel respiratory stimulants.

The retrotrapezoid nucleus is an important respiratory control center,^12,32  and a previous study identified chemosensitive Phox2b neurons in this region as required for maintaining breathing during general anesthesia.¹⁵  Neurons in this region that express Phox2b provide the main stimulus for breathing.^{12,13,33–35}  Isoflurane has been shown to stimulate the activity of retrotrapezoid nucleus Phox2b neurons by mechanisms involving inhibition of a THIK-1-like conductance and activation of an as-yet unidentified leak Na⁺ current.⁷  Our previous study indicates that pyramidal neurons expressing the sodium leak channel may be activated by isoflurane.²⁴  These results suggest the sodium leak channel may contribute to isoflurane activation of retrotrapezoid nucleus neurons and breathing. Consistent with this, here we found that isoflurane and sevoflurane at clinically relevant concentrations increased the excitability of retrotrapezoid nucleus Phox2b neurons in part by enhancing sodium leak channel conductance. Of note, our results cannot exclude involvement of potassium channels, such as THIK-1, that are also known to contribute to the excitatory action of isoflurane on retrotrapezoid nucleus Phox2b neurons.⁷ 

Compared with intravenous general anesthetics like propofol, volatile anesthetics moderately depress respiratory function.^11,36,37  This feature of volatile anesthetics makes them ideally suited for certain medical procedures needing spontaneous breathing such as intubation of a difficult airway and induction of pediatric anesthesia.^9–11  Therefore, understanding the mechanisms contributing to the maintenance of breathing during anesthesia of volatile anesthetics is critical for the safety of patients undergoing these types of procedures.

Sodium leak channel has been identified as an important modulator of respiratory function.^21–23,38  Global loss of sodium leak channel is lethal because of respiratory failure.^20,21  Loss of sodium leak channel from key elements of the respiratory circuit, including rhythmogenic neurons in the pre-Bötzinger complex, also results in lethal apnea.²³  Diminished expression of sodium leak channel in retrotrapezoid nucleus neurons severely compromises the ventilatory response to carbon dioxide.^23,39  Here we confirm the importance of the sodium leak channel to retrotrapezoid nucleus function by showing that neuronal activity of the retrotrapezoid nucleus is diminished by the nonselective sodium leak channel blocker gadolinium or genetic silencing of the sodium leak channel by short hairpin RNA and enhanced by substance P, further suggesting that the sodium leak channel is a critical modulator of excitability of retrotrapezoid nucleus Phox2b neurons. Although gadolinium as a pharmacologic probe for sodium leak channel has been widely used by previous studies,^21,40  its use as a nonspecific blocker of sodium leak channel in patch-clamp recordings is a limitation of the present study. To overcome this limitation, we delivered sodium leak channel –short hairpin RNA with adenovirus vector to silence sodium leak channel in brain slices, which helps strengthen our conclusions, although the genetic silencing of sodium leak channel is not specific to Phox2b neurons.

We observed two behavioral phenotypes caused by loss of sodium leak channel from retrotrapezoid nucleus neurons. First, consistent with previous work,²²  the loss of sodium leak channel from retrotrapezoid nucleus neurons minimally affected baseline breathing in awake mice but diminished the ventilatory response to carbon dioxide. Second, isoflurane and sevoflurane strongly suppressed respiratory activity in sodium leak channel genetically silenced mice compared with control mice. Meanwhile, propofol produced no effect on sodium leak channel -mediated channel conductance, and genetic silencing of sodium leak channel did not affect the respiratory output under propofol anesthesia. These results indicate that sodium leak channel regulates the excitability of retrotrapezoid nucleus Phox2b neurons and helps maintain breathing during anesthesia by volatile anesthetics but not by propofol.

We used neonatal mice aged of 7 to 12 days for patch-clamp recordings because it is exceedingly difficult to perform visualized patching on retrotrapezoid nucleus neurons of the animals older than two weeks. We used adult mice for the behavioral tests because of the more stable and better signal to noise ratio obtained in older mice. This age disparity may raise concerns of developmental differences in the respiratory circuits between neonatal and older mice. However, we consider this a minor limitation because retrotrapezoid nucleus neurons in both neonatal and adult mice are intrinsically and strongly activated by CO₂/H^+33,41  and show similar responses to wake-on transmitters, including substance P,⁴²  which indicates similar respiratory circuits between ages by a mechanism involving sodium leak channel.^22,23  Furthermore, consistent with a previous study,²⁴  we also found that sodium leak channel is widely expressed in both neonatal and adult mouse brain. Therefore, it is unlikely that fundamental properties of retrotrapezoid nucleus neurons, including sodium leak channel activation by volatile anesthetics, is dramatically different between neonatal and adult mice.

This study originally measured breathing parameters in male adult mice. In response to peer review, we measured respiratory activity in both male and female adult mice in additional experiments. Both sexes of rodents are widely used in whole-body plethysmography, and no differences have been found between males and females in this regard.^15,22,43  Of note, we repeatedly measured 0.8% isoflurane in both sexes during additional experiments and found similar effects, also indicating that sex may not influence the respiratory measurements.

Although we investigated the contribution of retrotrapezoid nucleus Phox2b neurons here, modulation of respiratory function by volatile anesthetics likely involves multiple levels of the respiratory circuitry, including other chemoreceptor regions or the downstream respiratory rhythm generator.^7,8,44  Isoflurane can decrease respiratory output by depression of chemoreceptor regions, including the medullary raphe or the respiratory rhythm generator.^8,44,45  Therefore, carbon dioxide–stimulated ventilation is also inhibited by anesthetic concentrations of isoflurane, although isoflurane can still activate retrotrapezoid nucleus central chemoreceptors at this concentration. This inhibition can result from the net actions of isoflurane on the central respiratory rhythmical generator and chemoreceptors outside the retrotrapezoid nucleus, whereas retrotrapezoid nucleus Phox2b neurons are activated by isoflurane.

The sodium leak channel also contributes to other nonrespiratory pharmacologic effects of volatile anesthetics.^38,46–48  For example, subanesthetic concentrations of isoflurane can activate sodium leak channel in hippocampal pyramidal neurons, and our previous study showed that behavioral hyperactivity during isoflurane induction is diminished by forebrain genetic silencing of sodium leak channel.²⁴ Drosophila and nematode unc-79 (component of sodium leak channel complex) mutants are hypersensitive to the immobilizing effects of volatile anesthetics.⁴⁶  A clinical case reported that a 3-year-old child with a pathologic mutation of sodium leak channel showed hypersensitivity to volatile anesthetics, which caused aggravated respiratory depression and cardiac arrest.³⁸ 

A limitation of this study is that the body temperatures of the mice were not perfectly controlled during plethysmography. General anesthetics slightly decreased body temperatures of the mice, which may negatively modulate respiratory function (Supplemental Digital Content, table 1, https://links.lww.com/ALN/C456). It is impossible to put a heating pad under the chamber to control body temperatures of the animals because the temperature of the sealed chamber may change the pressure inside and greatly affects accuracy of the results. Instead, room temperature was kept throughout. However, we consider this a minor issue because a robust stable measurement of respiration was found. In addition, anesthetics-induced changes of body temperature were similar between control mice and sodium leak channel genetically silenced mice (Supplemental Digital Content, table 1, https://links.lww.com/ALN/C456), indicating that the contribution of sodium leak channel did not result from a decrease of body temperature. Of note, we may have underestimated the stimulatory effect of volatile anesthetics on breathing because anesthetics including isoflurane can cause a modest reduction in body temperature and consequently decrease metabolic and respiratory activity.

In summary, volatile anesthetics at clinically relevant concentrations can stimulate activity of Phox2b neurons in the retrotrapezoid nucleus by enhancing sodium leak channel conductance and help maintain respiratory output during exposure to volatile anesthetics. Likewise, genetic silencing of sodium leak channel in the retrotrapezoid nucleus can aggravate volatile anesthetic-induced respiratory depression.

Supported by grant No. 2018YFC2001800 (to Dr. Zhu and Dr. Ou) from the National Key R&D Program of China, Beijing, China, and grant Nos. 81771486 and 81974164 (to Dr. Zhou) from National Natural Science Foundation of China, Beijing, China. This work was also supported in part by grant nos. HL104101 (to Dr. Mulkey), HL137094 (to Dr. Mulkey), and NS099887 (to Dr. Mulkey) from the National Institutes of Health, Bethesda, Maryland.

The authors declare no competing interests.