Breathing under Anesthesia: A Key Role for the Retrotrapezoid Nucleus Revealed by Conditional Phox2b Mutant Mice

Respiratory depression associated with sedation and anesthesia is a persistent and puzzling problem in anesthesiology.^1–4  In the last decade, particular attention has been paid to the respiratory depressant effects of opioids, which are particularly strong compared to those of other agents. Nonetheless, many if not all drugs used for anesthesia and analgesia can produce potentially severe respiratory depression.^5,6  Although we probably cannot fully prevent respiratory depression during anesthesia,⁷  active research is underway to identify respiratory stimulants capable of minimizing this effect without overly affecting anesthesia.⁴  Knowledge about the toxicity and efficacy of these drugs is still insufficient to allow clinical use.⁴  Identifying the cell types and neural pathways that sustain breathing during deep anesthesia may facilitate the development of effective respiratory stimulants.

Lesion studies suggest that breathing maintenance during anesthesia may involve the drive supplied by the retrotrapezoid nucleus to the respiratory central pattern generator.^5,8,9  The retrotrapezoid nucleus is a small group of glutamatergic neurons located in the rostral ventrolateral medulla, whose activation by arterial carbon dioxide stimulates breathing.¹⁰  In anesthetized cats and rats, neuronal lesions produced in the retrotrapezoid nucleus region by the neurotoxin kainic acid or by electrolysis caused apnea.^11–13  However, both these lesion-induction techniques lack selectivity, as recently pointed out by Souza et al.¹⁴  Studies of saporin-substance P injections showed that the retrotrapezoid nucleus neurons responsible for stimulating breathing during anesthesia were carbon dioxide–sensitive noncatecholaminergic neurons that expressed Phox2b,¹⁴  a master gene for the development of autonomic neurons.¹⁵  In anesthetized rats, 70% destruction of these retrotrapezoid-nucleus carbon dioxide–sensitive chemoreceptors substantially raised the carbon dioxide apneic threshold, thereby predisposing to apnea.¹⁶ 

To look for further evidence that the retrotrapezoid nucleus is pivotal in breathing maintenance during anesthesia, we examined breathing patterns in anesthetized mice with selective genetic depletion of the retrotrapezoid nucleus neurons. Loss of retrotrapezoid nucleus neurons has been reported in mouse models of congenital central hypoventilation syndrome, a life-threatening disorder characterized by hypoventilation during sleep and absence of the ventilatory response to carbon dioxide.^17,18  The disease-causing mutations in congenital central hypoventilation syndrome are generally polyalanine repeat expansion mutations of PHOX2B.¹⁹  Knock-in mice bearing the 7-alanine expanded allele (i.e., 27Ala) of PHOX2B (the most frequent mutation in patients¹⁹ ) exhibit massive loss of retrotrapezoid nucleus neurons, lack carbon dioxide chemosensitivity, and die within hours after birth.²⁰  In contrast, in the conditional mouse model with the +7Ala repeat mutation targeted to the retrotrapezoid nucleus (Phox2b^27Alaki/+ mice, hereafter designated “mutants”), despite massive selective loss of retrotrapezoid nucleus neurons and absence of carbon dioxide chemosensitivity at birth, survival is normal and carbon dioxide chemosensitivity recovers partially in adulthood.²¹  We hypothesized that mice with selective genetic depletion of retrotrapezoid nucleus neurons would be prone to severe respiratory failure during anesthesia.

All experimental protocols were approved by local and national ethics committees (Ministry of Higher Education and Scientific Research, Directorate-General for Research and Innovation, Paris, France), in accordance with the European Communities Council Directive 2010/63/EU for animal care. The experiments in the main study were conducted in mutant mice and their wild type littermates at postnatal day (P) 8, which approximately corresponds to term in humans,²²  and P14 to P16, corresponding to infancy in humans.²²  Complementary studies were conducted at P29 to P31, corresponding to adolescence,²³  and at P60 to P62, corresponding to early adulthood. The mice that survived anesthesia were killed at P8 by decapitation and beyond P8 by cervical dislocation, without anesthesia. The experiments were conducted in mice of both sexes. The proportions of males and females at birth were not significantly different in the wild types (125 males and 97 females in total; male-to-female ratio, 1.29) and mutants (86 males and 87 females in total; male-to-female ratio, 0.99, Fisher exact test; P = 0.192).

We generated conditional, tissue-specific, knock-in Phox2b^27Alacki/+ mice as previously described.²¹  Briefly, upon cre recombinase-mediated recombination, mouse Phox2b exon 3 was replaced by the mutated human exon 3 bearing the seven-residue expansion. Because human and mouse Phox2b protein sequences are identical, the encoded protein is identical to mouse Phox2b except for the extension of the polyalanine stretch. Offspring with the recombined locus were produced by crossing Egr2^cre/+ males with Phox2b^{27Alacki/27Alacki} females. Genotyping was performed with tail DNA after completion of the protocol and data processing. To detect the presence of the Phox2b^27Alacki allele, the primers GCCCACAGTGCCTCTTAACT and CTCTTAAACGGGCGTCTCAC were used, yielding bands of 330 bp for the wild type gene, 474 bp for the mutated gene, and 380 bp for the recombined allele. To detect cre, the primers AAATTTGCCTGCATTACCG and ATGTTTAGCTGGCCCAAATG were used, yielding a band of 200 bp.

In the main study, ketamine anesthesia was administered to mice at P8 (8 wild types, 13 mutants) and P14 to P16 (59 wild types, 42 mutants) and in complementary studies at P29 to P31 (45 wild types, 36 mutants) and P60 to P62 (12 wild types, 19 mutants). We chose ketamine as an anesthetic/analgesic agent with minimal effects on central respiratory drive.²⁴  We also tested propofol and fentanyl at P8 (propofol: 17 wild types, 7 mutants; fentanyl: 13 wild types, 16 mutants) and P14 to P16 (propofol: 10 wild types, 10 mutants; fentanyl: 11 wild types, 7 mutants). Both propofol and fentanyl markedly depress respiration.²⁴  Ketamine, propofol, and fentanyl are widely used for pediatric and adult anesthesia/sedation and act mainly by different mechanisms.^5,25 

All agents were administered by intraperitoneal injection in the lower left quadrant of the abdomen, near the midline and umbilicus, taking care to avoid any visible milk spots. A 19-mm/30-gauge needle was inserted at an angle of 45° to the abdominal wall. The maximal volume administered was 20 µl/g. All injections were performed by the same experimenter.

The doses were as follows: ketamine, 100 mg/kg at P8, 150 mg/kg at P14 to P16, and 250 mg/kg at P29 to P31 and P60 to P62; propofol, 100 and 150 mg/kg at P8 and P14 to P16, respectively; and fentanyl, 1 mg/kg at P8 and P14 to P16. These doses were previously shown to rapidly produce deep anesthesia at all studied ages, as assessed by the loss of righting and tail-pinch responses,²⁶  with no or minimal mortality in wild type mice. They were at the upper end of the intraperitoneal ketamine,²⁷  propofol,²⁸  and fentanyl²⁹  dose ranges previously used in mice. The investigators were blinded to genotype, which was visually indiscernible. The genotypes were determined after completion of each protocol and therefore, cardiorespiratory data processing and mortality recordings were performed without previous knowledge of genotypes.

In a separate experiment, we analyzed the possible involvement of the serotonergic system (which is not affected by Phox2b mutation²⁰ ) in the response to anesthesia in mutants and wild type littermates. In rodents, serotonin produces an excitatory effect on the pre-Bötzinger complex, mediated by 5-hydroxytryptamine 2 receptors.³⁰  We tested serotonin system involvement by using the 5-hydroxytryptamine 2_A/C antagonist ketanserin at P29 to P31 (10 mg/kg intraperitoneal, combined with 150 mg/kg ketamine in 25 wild types, 16 mutants, vs. ketamine alone, in 22 wild types, 7 mutants). The litters were alternately assigned to the ketanserin and control groups.

Breathing variables were measured noninvasively by using a battery of four custom-made, whole-body flow barometric plethysmographs (fig. 1, A–D), as previously described.²¹  Data were collected in all P8 mice (38 wild types, 36 mutants) and a large subset of P14 to P16 mice (69 wild types, 47 mutants). Each plethysmograph was composed of two 100-ml Plexiglas chambers (fig. 1B) immersed in a thermoregulated water bath to maintain the temperature at 32°C to prevent anesthesia-induced hypothermia. A 200 ml · min^-1 flow of dry air (Brooks airflow stabilizer, Urlo, The Netherlands) was injected in each chamber. The differential pressure between the chambers (GE Sensing transducer, France; range, ± 0.1 millibar) was converted into a digital signal at a sampling rate of 100 Hz and processed by using Labview software (National Instruments, USA). The apparatus was calibrated before each session by using a built-in pump incorporating a microsyringe (Ito Inc., Japan), which injected a sinusoidal airflow into the animal chamber with maximal amplitude 2 µl and frequency 6 Hz. We measured breath duration (s) and tidal volume (ml · g^-1) on a breath-by-breath basis to calculate minute ventilation (tidal volume · breath duration^-1 · 60, ml · min^-1 · g^-1). The limitations of the plethysmographic method in newborn mice have been discussed elsewhere.³¹  Briefly, plethysmography is effective for measuring breath duration (or frequency) and apneas but only provides semiquantitative measurements of tidal volume and minute ventilation because of gas-compression effects related to airway resistance.³¹  In particular, upper airway obstruction may affect tidal volume and ventilation values. By allowing simultaneous measurements in the same pup, and in four pups simultaneously, our experimental setup contributed to reduce experimental variability and sample size, thereby meeting the ethical requirements of reducing and refining animal use in research.

The plethysmograph chambers were equipped with electrocardiography recording platforms composed of four rectangular gold electrodes insulated from one another and embedded in the floor of the chamber (fig. 1, B–E). Conduction was enhanced by using electrode hydrogel (Sekisui Plastics, Japan). Signals were digitized at a sampling rate of 1,000 Hz (16 bits, PCI-6229, National Instruments, USA). An electrocardiograph signal was obtained when at least three paws were in contact with three different electrodes or, occasionally, when the pup was lying on the floor.

In a separate study, P14 to P16 mice anesthetized with 150 mg/kg ketamine (22 wild types, 7 mutants) were mechanically ventilated with air until the anesthesia wore off. The negative pressure ventilator comprised a set of chambers connected to a common negative-pressure generator allowing for simultaneous ventilation of up to 60 mice. Each mouse was placed for 2 h in a custom setting that allowed mechanical ventilation by intermittent extrathoracic negative pressure. The head of the animal was carefully placed in a thin latex ring, which was then affixed to a polycarbonate chamber so that the body was inside and the head outside. An intermittent negative-pressure generator alternately connected the chamber to the atmosphere or to a low-pressure vacuum source. Ventilation frequency was set at 3 Hz, duty cycle at 0.2, and peak negative pressure in the chamber at 14 cm H₂O. This pressure value was chosen to generate respiratory movements whose amplitude and frequency approximately corresponded to those during spontaneous breathing in mice of the same strain and age, according to preliminary pneumotachography measurements (not shown).

The study design is summarized in figure 1F. In Phase 1 (preanesthesia carbon dioxide test, 15 min), mice were randomly taken from each litter in successive groups of four, weighed, and placed together in the plethysmograph chambers. After 5 min of familiarization with the chamber, baseline breathing and heart rate were recorded over 5 min. Then, the airflow through the plethysmograph was switched to 8% CO₂ + 21% O₂ + 71% N₂ at the same flow rate (200 ml/min per chamber) for 5 min, after which the flow was switched back to normoxia for 5 min. In Phase 2 (air breathing under anesthesia, 30 min), the four mice were extracted from the plethysmograph chambers, given anesthetics, and replaced simultaneously in the plethysmograph chambers within 4 min after the injection, for a 30-min baseline recording. Phase 3 (posttreatment carbon dioxide test, 15 min) was identical to Phase 1. The primary outcome of the study was mortality caused by anesthesia in the mutants compared to the wild types. Mortality was assessed immediately after completion of Phase 3, based on cardiorespiratory signal monitoring while the mice were in the plethysmograph and on immobility, unresponsiveness to pinch, and discoloration. Survivors were reunited with their dams. Adolescent and adult mice (P29 to P31 and P60 to P62) did not undergo plethysmography; they were anesthetized with ketamine, and mortality was then assessed for 60 min. All the experiments were performed during the day, between 9:00 am and 6:30 pm.

Anesthesia-induced mortality was compared between mutant and wild type mice, at each age and with each anesthetic agent, by applying the two-tailed Fisher exact test. The experimental unit was the animal (n refers to the number of animals in each group). Weights and baseline cardiorespiratory variables were compared by two-tailed independent t tests. The time-course of the ventilatory response to hypercapnia was analyzed by averaging minute ventilation over consecutive 30-s periods throughout the 15-min plethysmographic recording (5 min air, 5 min hypercapnia, 5 min air) at P8 and P14 to P16. These data were compared between mutant and wild type mice using two-way repeated measures ANOVA with genotype (wild types vs. mutants) as a between-subject factor and time period (1 to 30) as a repeated factor. Bonferroni post hoc tests were performed to compare time-matched minute ventilation data between genotypes. For all parametric tests, normality of data was assessed by the Shapiro–Wilk test. Data are mean ± SD in the text, tables, and figures. No statistical power calculation was conducted before the study. The sample size was based on our previous experience with respiratory variability in this genetic model. All analyses were done using R Studio v1.1.243 and JMP Pro 13 (SAS Institute, Inc., USA). Differences were considered statistically significant if P < 0.05. Sex had no statistically significant effect in any analyses and will not be mentioned further.

At P8 and P14 to P16, the mutants (n = 36 and n = 47, respectively) weighed slightly less than their wild type littermates (n = 38 and n = 69, respectively; table 1) but were normal in general appearance and behavior. First, we verified that the Phox2b^27Alacki/+ mutation effectively disrupted the carbon dioxide–sensitive drive, as previously shown,²¹  by performing whole-body plethysmography at P8 and P14 to P16. Mutant and wild type mice at P8 did not differ significantly regarding baseline breathing variables (i.e., before anesthesia, Phase 1, table 1). At P14 to P16, mutant mice showed significantly longer breath durations (i.e., lower breathing frequencies) and marginally smaller ventilation values than wild type mice. As expected, at P8, mutants lacked any ventilatory response to carbon dioxide, in contrast to wild type mice (genotype-by-time interaction, P < 0.001; fig. 2A). A slight response was present at P14 to P16 but was significantly smaller than in the wild type mice (genotype-by-time interaction, P < 0.001; fig. 2B). This disruption of carbon dioxide sensitivity in mutant mice confirms previous results.²¹  Wild type mice sustained a regular breathing pattern and a vigorous response to hypercapnia (fig. 2, A and B).

Ketamine anesthesia produced high mortality rates in mutant mice at all studied ages, including adulthood, whereas all or most wild type mice survived (table 2). The sedative effect extended beyond completion of Phase 3 in all surviving mice, including mutants.

The cause of death was first determined by analyzing the cardiorespiratory recordings. Figure 3 shows a typical sequence of respiratory events in a P14 to P16 mutant mouse given ketamine. A period of eupneic breathing (fig. 3A) was followed by a sequence of short clusters of gasps, in pairs (“double gasps”) or triplets (“triple gasps”), separated by apneas (fig. 3B). This sequence was followed by a series of gasps of decreasing amplitude ending with terminal apnea (fig. 3C), usually during Phase 2 (within 13 min after ketamine administration). However, the time to terminal apnea varied widely across individuals, with some mice dying before being placed in the plethysmograph (i.e., within 4 min after ketamine administration). The electrocardiograph signal consistently outlasted the terminal gasps, although the heart rate was reduced (fig. 4). These findings support respiratory arrest as the cause of death, as opposed to cardiac arrest.

That death was due to respiratory arrest in ketamine-exposed mutants was further confirmed by exposing a separate group of P14 to P16 mutants (n = 7) and wild type littermates (n = 22) to mechanical ventilation immediately after a 150-mg/kg ketamine injection. The mice were ventilated until the effects of the anesthesia wore off, as reflected by gross body movements. The end of anesthesia varied widely among mice, from 60 min to 90 min after ketamine injection. In contrast to the high mortality rate caused by 150-mg/kg ketamine in nonventilated mutants (32 of 42, 76%; table 2), all ventilated mutants survived anesthesia (exact Fisher test, ventilated vs. nonventilated mutants, P < 0.001), as did all wild type mice, confirming respiratory arrest as the cause of ketamine-induced death in nonventilated mutant mice. Monitoring for 3 days after anesthesia confirmed that all ventilated mice survived and appeared normal, ruling out any major delayed effect of ketamine anesthesia.

To further analyze the respiratory effects of ketamine in mutants, we first compared breathing variables in mutants and their wild type littermates throughout the recording session (see Supplemental Digital Content 1, https://links.lww.com/ALN/B908, showing breathing variables in wild types and mutants from Phase 1 to Phase 3, and distinguishing surviving and nonsurviving mutants in Phase 1). Before ketamine treatment, the mutants had no ventilatory response to carbon dioxide. In both genotypes, ketamine depressed minute ventilation, mainly by decreasing tidal volume and increasing breath duration. However, the wild types tended to recover pre-ketamine levels for all variables, whereas ventilation in the mutant survivors remained depressed throughout Phases 2 and 3. The inspiratory-to-expiratory ratio was also markedly affected by ketamine in mutant survivors, suggesting impaired respiratory phase timing. Then, we analyzed individual plots of nonsurviving mutants (Supplemental Digital Content 2, https://links.lww.com/ALN/B909, which is a zoom of Phase 2, including wild types, mutant survivors, and mutant nonsurvivors with available respiratory data until the onset of gasping). No clear differences in respiratory variables were observed between surviving and nonsurviving mutants.

We then examined whether survival of a subgroup of ketamine-exposed mutants was due to the serotonergic drive to the respiratory central pattern generator. Serotonergic terminals are found throughout brainstem respiratory regions, and both serotonin and the peptides released by serotonergic cells modify the activity of many types of respiratory neurons.⁸  In a separate experiment, we addressed the contribution of the serotonergic drive (which is not affected by Phox2b gene mutation²⁰ ) by combining 10 mg/kg ketanserin, a serotonin 5-hydroxytryptamine 2_A/C antagonist, with 150 mg/kg ketamine, administered to P29 to P31 mice (25 wild types, 16 mutants), and comparing the findings to those in a group given only ketamine (22 wild types, 7 mutants).

The 150 mg/kg ketamine dose was nonlethal in P29 to P31 mutants (table 3). However, when combined with ketanserin, ketamine caused death in 11 of 16 (69%) mutants (exact Fisher test, ketanserin-treated vs. untreated mutants, P = 0.005). All but one ketanserin-treated wild type mouse survived (table 3). These results support a role for the serotonin system in protecting mutants against ketamine-induced respiratory arrest. However, our attempt to decrease mortality in ketamine-exposed mutants by using the 5-hydroxytryptamine 2 agonist 1-(2,5-dimethoxy-4-iodophenyl)-2-aminopropane in doses of 0.75 to 1.5 mg/kg did not yield statistically significant results (0.75 mg/kg: P > 0.999; 1 mg/kg: P = 0.121; 1.5 mg/kg: P = 0.184).

We then examined whether the lethal effects of ketamine anesthesia in mutants extended to other anesthetic/analgesic agents with different mechanisms of action. We exposed mutants and their wild type littermates to propofol (P8: 17 wild types, 7 mutants, 100 mg/kg; and P14 to P16: 10 wild types, 10 mutants, 150 mg/kg) or fentanyl (P8: 13 wild types, 16 mutants, 1 mg/kg; and P14 to P16, 11 wild types, 7 mutants, 1 mg/kg; table 3). Both agents produced very high mortality rates in mutants but no mortality in wild type mice (table 4), further supporting the mutants’ vulnerability to anesthesia previously suggested by ketamine-induced mortality. The respiratory traces in propofol-exposed mice displayed similar patterns of gasps and apneas as did those seen with ketamine (see Supplemental Digital Content 3, https://links.lww.com/ALN/B910, showing respiratory arrest in a P15 mutant mouse after propofol administration). Fentanyl-exposed mice exhibited severely disrupted ventilation from the onset of the recording, also followed by a sequence of gasps that gradually diminished in size until terminal apnea (see Supplemental Digital Content 4, https://links.lww.com/ALN/B911, showing respiratory arrest in a P15 mutant mouse after fentanyl administration).

In this study, ketamine, propofol, and fentanyl caused respiratory arrest in most mice with selective genetic loss of retrotrapezoid nucleus neurons, in doses that were safe in their wild type littermates. These results support a pivotal role for the retrotrapezoid nucleus in maintaining spontaneous breathing during deep anesthesia. The serotonin system probably contributed to sustain breathing during anesthesia in retrotrapezoid nucleus neuron-depleted mice.

That the high mortality in the ketamine-exposed mutants was due to respiratory arrest was first established by examining the order of respiratory and cardiac arrests and was subsequently confirmed by showing that the mutants survived if given mechanical ventilation. In mutant mice, ketamine anesthesia produced apneas interspersed with clusters of gasps, a common marker of hypoxia. Gasps (and sighs) are generated by the pre-Bötzinger complex,³²  which does not express Phox2b and therefore is spared by Phox2b gene mutations.³³  Clusters of double or triple gasps have never been reported in ketamine-anesthetized animals; there is a single report in piglets sedated with 20 mg/kg intraarticular pentobarbital sodium administration.³⁴  However, this pattern is often reported in dying preterm infants and in infants who subsequently experienced sudden infant death syndrome.³⁵  Thus, in both these human cases and our mutant mice, the lack of successful autoresuscitation was not related to an inability to gasp.^35,36  Rather, the anesthetized mutants probably died of hypoxic hypoventilation caused by disruption of the excitatory drive sent by the retrotrapezoid nucleus to the central pattern generator.

The recurrent apnea in anesthetized mutants suggested that absence of the retrotrapezoid nucleus may have caused an increase in the carbon dioxide apneic threshold. In an earlier study, the carbon dioxide apneic threshold was increased by 70% destruction of carbon dioxide–sensitive chemoreceptors in the retrotrapezoid nucleus of anesthetized rats.¹⁶  Our mutants exhibited similar or higher loss rates.²¹  Furthermore, ketamine, propofol, and mu-opioid agonists also increase the carbon dioxide apneic threshold, in a dose-dependent manner.^25,37  Breathing was not impaired by anesthesia alone (as indicated by the breathing pattern in anesthetized wild types) or neuronal loss alone (as indicated by the breathing pattern in non-anesthetized mutants). However, when combined in anesthetized mutants, these two factors caused respiratory arrest, probably by increasing the carbon dioxide apneic threshold.

Few mutant mice survived ketamine, propofol, or fentanyl anesthesia, and mortality was highest in the youngest animals. Individual differences in the vulnerability to anesthesia possibly reflected variations in retrotrapezoid nucleus cell loss caused by the Phox2b^27Alacki/+ mutation.²¹  In previous studies, the number of residual cells in the retrotrapezoid nucleus of mutant mice was estimated at 15 to 20% of the mean value in wild types.²¹  This range is close to the threshold at which pharmacologic destruction of the retrotrapezoid nucleus disrupted phrenic nerve discharge in rats.¹⁶  In anesthetized and ventilated rats, bilateral 70% destruction of retrotrapezoid nucleus carbon dioxide–sensitive chemoreceptors (i.e., 30% spared neurons) substantially raised the carbon dioxide apneic threshold.¹⁶  Also, in rats with unilateral 70% retrotrapezoid nucleus neuron destruction, acute inhibition of the contralateral intact retrotrapezoid nucleus with muscimol instantly abolished phrenic nerve discharge.¹⁶  Cell counting was not done in the current study, but the mutant mice that survived anesthesia may have been those with more numerous spared cells in the retrotrapezoid nucleus and, therefore, with a stronger residual excitatory drive to the central pattern generator.¹⁴ 

The key role of the retrotrapezoid nucleus in sustaining breathing during anesthesia does not preclude contributions of other neuronal systems. In particular, activation of the serotonergic raphe neurons stimulates ventilation, and the full effects of carbon dioxide on breathing require activity of these neurons.¹⁰  Therefore, we examined whether the serotonin system contributed to sustain breathing during anesthesia in mutants. We found that the 5-hydroxytryptamine 2 receptor antagonist ketanserin caused death in mutants despite being used in doses that were safe in wild types when administered alone. This finding supports a role in breathing maintenance during anesthesia of the excitatory drive from serotoninergic neurons to the respiratory network (especially the pre-Bötzinger complex), via 5-hydroxytryptamine 2 receptors.

This analysis, however, has several limitations. First, the 5-hydroxytryptamine 2 receptor agonist 2,5-dimethoxy-4-iodoamphetamine failed to significantly reduce ketamine-induced mortality in P14 to P16 mutants. Second, the role of the serotonin system revealed by ketanserin may reflect a compensatory adaptation to retrotrapezoid nucleus loss specific to mutant mice. Third, a previous report that knock-out mice with disrupted 5-hydroxytryptamine 2_A receptor signaling survived anesthesia with ketamine 150 mg/kg³⁸  would seem at variance with our findings. We focused on the 5-hydroxytryptamine 2 receptor because of its excitatory effect on respiratory rhythm generation, but the effects of 5-hydroxytryptamine 1,4 receptor agonists, which counteract opioid-induced ventilatory depression,^1,4,39  also deserve attention. Finally, we cannot discount that ketanserin increased ketamine-induced respiratory depression and mortality via its alpha1-adrenergic receptor blocking properties in residual retrotrapezoid nucleus cells⁴⁰  or in the pre-Bötzinger complex.⁴¹  This possible effect might explain why the 5-hydroxytryptamine 2 agonist 1-(2,5-dimethoxy-4-iodophenyl)-2-aminopropane did not decrease ketamine-induced mortality. The role for serotonin pathways in sustaining breathing during anesthesia clearly requires further investigation.

The critical role of the retrotrapezoid nucleus during deep anesthesia probably reflects its strategic location and function in the central pattern generator. One important characteristic of retrotrapezoid nucleus neurons is their monosynaptic input to the respiratory central pattern generator,⁴²  which contrasts with the chiefly polysynaptic inputs from other sources (e.g., the peripheral chemosensory drive, arousal-related suprapontine activation, and exercise-related central command). Polysynaptic pathways are strongly affected by anesthetics due to cumulation of effects along the signaling chain.⁴³  Furthermore, the pontine parabrachial-Koölliker-Fuse complex, which controls expiratory duration and the inspiratory on-switch,^44,45  has strong reciprocal connections with the retrotrapezoid nucleus.⁴⁶  The loss of retrotrapezoid input in mutants may have contributed to the disruption of respiratory phase timing in this group. Therefore, the cumulative effect of anesthetic agents on synaptic transmission, combined with retrotrapezoid neuron depletion, may account for the vulnerability of mutants to all anesthetics studied, despite their different cellular targets and properties.

The most parsimonious explanation of anesthesia-induced respiratory arrest in mutant mice is that massive loss of retrotrapezoid nucleus drive combined with silencing of polysynaptic respiratory pathways fully abolished the respiratory drive to the central pattern generator. However, our data do not rule out the possibility that respiratory arrest in the mutants involved mechanisms related to the specific cellular targets and respiratory central pattern generator effects of each anesthetic.^5,25  Ketamine is an N-methyl-d-aspartate receptor antagonist that also depresses ventilation via γ-aminobutyric acid type A and opioid receptors.⁴⁷  Propofol depresses ventilation via the γ-aminobutyric acid type A receptor. The glutamatergic pre-Boötzinger complex and the pontine parabrachial-Koölliker-Fuse complex, which, as noted above, control expiratory duration and the inspiratory on-switch,^44,45  express opioid receptors. Possibly, activation of one of these pathways may be sufficient to disrupt the function of a system made inherently unstable by retrotrapezoid neuronal loss.

Finally, in vitro studies recently showed that isoflurane exposure aggravated the toxic effects of the mutated PHOX2B gene (i.e., PHOX2B protein misfolding, aggregation, and loss of nuclear localization) in cultured cells with the +7Ala expansion.⁴⁸  Further research is needed to determine how these effects may affect the function of residual retrotrapezoid nucleus neurons in mutant mice.

The current results may explain why patients with congenital central hypoventilation syndrome, most of whom carry PHOX2B polyalanine expansions, are prone to major anesthesia-related complications that require profound perioperative precautions.⁴⁹  They also suggest that the often used combination of opioids and propofol, which act additively on anesthesia induction,⁵⁰  may be particularly detrimental to patients with congenital central hypoventilation syndrome.

Mice with selective loss of retrotrapezoid nucleus neurons were highly vulnerable to deep anesthesia induced by ketamine, propofol, or fentanyl. Most of them died of respiratory arrest when anesthetized, in contrast to their wild type littermates. These results confirm the pivotal role of the retrotrapezoid nucleus in breathing maintenance during deep anesthesia. They suggest that drugs capable of selectively activating retrotrapezoid nucleus neurons may hold promise for preventing anesthesia-induced respiratory depression.

Support was provided solely from institutional sources, Association Française du Syndrome d’Ondine (French Association of Ondine Syndrome; Paris, France) and Legs Poix (Poix Legacy Fund, Chancellery of Paris Universities; Paris, France).

The authors declare no competing interests.