Loss-of-function of chemoreceptor neurons in the retrotrapezoid nucleus: what have we learned from it?

Abstract

Central respiratory chemoreceptors are cells in the brain that regulate breathing in relation to arterial pH and PCO₂. Neurons located at the retrotrapezoid nucleus (RTN) have been hypothesized to be central chemoreceptors and/or to be part of the neural network that drives the central respiratory chemoreflex. The inhibition or ablation of RTN chemoreceptor neurons has offered important insights into the role of these cells on central respiratory chemoreception and the neural control of breathing over almost 60 years since the original identification of acid-sensitive properties of this ventral medullary site. Here, we discuss the current definition of chemoreceptor neurons in the RTN and describe how this definition has evolved over time. We then summarize the results of studies that use loss-of-function approaches to evaluate the effects of disrupting the function of RTN neurons on respiration. These studies offer evidence that RTN neurons are indispensable for the central respiratory chemoreflex in mammals and exert a tonic drive to breathe at rest. Moreover, RTN has an interdependent relationship with oxygen sensing mechanisms for the maintenance of the neural drive to breathe and blood gas homeostasis. Collectively, RTN neurons are a genetically-defined group of putative central respiratory chemoreceptors that generate CO₂-dependent drive that supports eupneic breathing and stimulates the hypercapnic ventilatory reflex.

1. Molecular identification of chemoreceptor neurons in the retrotrapezoid nucleus—an evolving definition.

Breathing stimulation elicited by acidifying the ventral lateral surface of the medulla oblongata immediately caudal to the trapezoid body is among the earliest evidence that this region contains respiratory chemoreceptors (Mitchell et al., 1963a, b). Several subsequent loss-of-function experiments using cooling (Budzinska et al., 1985; Forster et al., 1995; Ohtake et al., 1995), occlusion of blood supply artery (Kuwana et al., 1990), local anesthetic (procaine) application (Cherniack et al., 1979), coagulation (Schlaefke et al., 1979), chemical (Akilesh et al., 1997; Nattie et al, 1988) and electrolytic lesions (Nattie et al., 1991) of this region also highlighted its role in breathing regulation and central respiratory chemoreception. In 1989, Smith and colleagues identified in cats a group of very superficial neurons that projected to respiratory areas in the dorsal and ventral medulla, coining the term “retrotrapezoid nucleus” (RTN) (Smith et al., 1989). Further studies in rats using c-Fos showed neurons in the RTN were CO₂-stimulated (Berquin et al., 2000; Teppema et al., 1994; Teppema et al., 1997) and functional studies by Eugene Nattie’s group (Nattie and Li, 2002) targeting neurokinin type-1 receptor-expressing neurons in the RTN provided evidence that these neurons contribute to the stimulation of breathing by CO₂. In 2004, in vivo and in vitro electrophysiological recordings characterized a population of glutamatergic neurons in the RTN that increased their firing frequency in proportion to changes in pH/PCO₂ even after a reduction in synaptic transmission (Mulkey et al., 2004). At this time, the definition of an RTN neuron was largely based on a combination of functional properties and anatomical features.

Two years later, Stornetta et al., (2006) reported that the glutamatergic CO₂-activated neurons of the RTN were strongly PHOX2B-immunoreactive. This discovery was greatly aided by advances in the understanding of the genetic basis for central congenital hypoventilation syndrome (CCHS), which is primarily caused by mutations of the PHOX2B and causes central sleep apnea and severely blunts the respiratory response to CO₂ (Amiel et al., 2003; Weese-Mayer et al., 2010). Based on this work, RTN neurons could be increasingly better defined as acid-activated, glutamatergic (i.e., expressing the vesicular glutamate transporter type 2; VGlut2+), PHOX2B-expressing neurons that lacked the defining markers of catecholaminergic, GABAergic, glycinergic, serotonergic and cholinergic neurons (Guyenet and Bayliss, 2022; Guyenet et al., 2019). In addition to providing a possible explanation for the respiratory phenotype of CCHS, the expression of PHOX2B by putative chemoreceptor neurons in the RTN became a useful marker for future studies facilitating anatomical investigation and genetic-targeting (Abbott et al., 2009; Dubreuil et al., 2008; Marina et al., 2010). However, PHOX2B is also expressed by myriad neurons in the brainstem of adult rodents, including non-chemosensitive neurons in the RTN region such as the C1 catecholaminergic neurons (Kang et al., 2007). This is an experimental limitation when implementing genetically-targeted approaches to study respiratory function. Because of this limitation, it is often necessary to conduct many additional experiments to parse out the specific role of chemosensitive neurons from those controlling other physiological functions. This problem provided the motivation to search for a more specific marker to distinguish RTN central chemoreceptors from other types of PHOX2B+ neurons.

Subsequent experimental studies found that RTN neurons (then defined as CO₂-activated VGlut2+, PHOX2B+, TH- and ChAT- neurons) expressed the mRNA for the neuropeptide Galanin (Gal), being expressed in roughly 50% of RTN neurons (Stornetta et al., 2009). Thus, the utility of Gal for genetic targeting of RTN neurons for functional studies is potentially limited. However, Gal has been identified in PHOX2B+ neurons in primates and in excitatory neurons in the human homolog of the rodent RTN (Lavezzi et al., 2012; Rudzinski and Kapur, 2010; Levy et al., 2021; Levy et al., 2019). Therefore, Gal identifies a subtype of RTN neurons that is conserved across species.

In 2017, single-cell RNA sequencing and in situ hybridization were used to systematically characterize the transcriptome of individual PHOX2B+ RTN neurons (Shi et al., 2017). This study offered several key insights into the genes expressed by RTN neurons. First, it confirmed expression of GPR4 and TASK-2 in PHOX2B+ RTN neurons; these membrane proteins mediate the activation of these cells by H⁺ and CO₂ (Gestreau et al., 2010; Kumar et al., 2015). Second, it demonstrated that PHOX2B+ RTN neurons with chemoreceptor properties selectively express the mRNA for the neuropeptide Neuromedin B (Nmb) (Shi et al., 2017). This observation extended an earlier study by Li et al. (2016) that showed an enrichment of Nmb and the co-expression of Nmb with PHOX2B in the RTN. Shi et al. observed that Nmb-expressing neurons in the RTN (RTN^Nmb) are pH/CO₂ sensitive, glutamatergic and do not express markers genes for catecholaminergic other nearby neurons. Importantly, at least 80% of RTN^Nmb neurons express one or both acid-sensitive proteins TASK-2 and GPR4, and all neurons expressing these proton sensors in the RTN express Nmb (Shi et al., 2017). Clearly, there is an impressive alignment between the presence of Nmb and the expression of proteins that are proposed to confer intrinsic H⁺ sensitivity in the RTN neurons. However, it remains possible that a subset of Nmb neurons are not central chemoreceptors because around 20% of Nmb cells do not express detectable GPR4 or TASK-2, and a similar proportion do not exhibit detectable c-Fos expression following hypercapnia (Shi et al., 2017). Also, whether Nmb expression identifies central chemoreceptor neurons in the RTN in species other than rodents remains an open question, as Nmb expression was not detected in the vicinity of the RTN in humans based on 2 samples (Levy et al., 2021).

Presently, Nmb provides the best standalone criterion to identify putative chemoreceptor neurons in the RTN of mice and rats and provides a useful genetic target for these neurons (Souza et al., 2023) (Fig. 1A–C). Furthermore, the expression of Nmb appears to be stable throughout post-natal period permitting identification of RTN neurons throughout development (Shi et al., 2021) (Fig. 1A–C). Hence, the terminology “RTN neurons” defines perifacial PHOX2B+, excitatory neurons that are activated by CO₂ and contribute to the respiratory response to hypercapnia (see criteria in Gonye & Bayliss, 2023). Furthermore, all “RTN neurons” express Nmb and a frank majority express GPR4 and TASK-2. However, in the future, this definition might have to be refined to accommodate new information about the connectivity and potential functional specialization of subclasses of these neurons.

Figure 1.

A. Sagittal view of the brainstem of a mouse (~1.4 mm lateral of midline) indicating the location of retrotrapezoid neurons labeled with a Cre-dependent expression of mScarlet (pseudo-colored green) in a Neuromedin-B (Nmb) Cre mouse. Scale bar: 300 μm B. Inset from (A) showing the distribution of Nmb+ neurons. Scale bar: 300 μm C. Coronal view of the ventrolateral brainstem (Bregma level: −6.47 mm) showing RTN neurons labeled with a Cre-dependent expression of mCherry in a Neuromedin-B (Nmb) Cre mice. Note that the labeling is very selective to neurons surrounding the facial motor nucleus (7N) as demonstrated in Souza et al., 2023. Scale bar: 300 μm. Abbreviations: py, pyramidal tract; sp5, spinal trigeminal tract. D. Breathing recordings in normoxia and under hypercapnia in rats with ablation of RTN neurons using the Substance-P conjugated with saporin (SSP-SAP) and their respective control (from Souza et al., 2018). E. Breathing recordings in normoxia and under hypercapnia in mice with genetically-targeted ablation of RTN neurons and their controls (from Souza et al., 2023). F and G. Minute-ventilation (V_E) during normoxia and different levels of hyperoxic-hypercapnia in rats (from Souza et al, 2018) and mice (from Souza et al., 2023) after lesion of RTN neurons and their respective controls. Despite different methods and species, lesion of RTN neurons results in impairment of the hypercapnic ventilatory response. H. Changes in V_E in response to hypercapnia (CO₂=9%) and the relationship with the number of surviving Nmb+ neurons in the RTN after injections of SSP-SAP in adult rats (from Souza et al., 2018). A majority of Nmb+ neurons need to be eliminated in order to reduce V_E in response to CO₂. I. Summary of compensatory mechanisms after lesion of RTN neurons in rodents. J and K. Effect of hyperoxia after lesion of RTN neurons in mice (from Souza et al., 2023). Note that when exposed to hyperoxia mice with lesion of RTN neurons present apneas. L. Acute inhibition of RTN neurons using optogenetics (RTN neurons are inhibited only during laser on period). Acute inhibition of RTN neurons causes instant hypoventilation in normoxia an effect that is enhanced by hyperoxia and hypercapnia. On the contrary, optogenetic inhibition of RTN neurons has a modest effect on breathing under hypoxic conditions, when respiratory alkalosis sets in (from Souza et al., 2023).

According to the distributed chemoreceptor theory (Nattie et al., 2012), the central component of the ventilatory to hypercapnia is derived from the effects of pH and CO₂ on multiple sites of chemoreception in the brain that collectively determine respiratory motor activity. Putative chemoreceptor sites include midbrain serotonergic neurons (Severson et al., 2003), the locus coeruleus (Biancardi, 2008), the nucleus of the solitary tract (Fu et al., 2017), and the RTN. Evidence for the distributed chemoreceptor theory is based on intrinsic H⁺ sensitivity of neurons in the aforementioned brain regions, and by the partial reduction in the hypercapnic ventilatory reflex (HCVR) following focal inhibition of each region. More work is required to understand how distinct sites of chemoreception interact to stimulate breathing. However, genetically-targeted cell ablation of the RTN neurons in mice reduces the sensitivity of the HCVR by 94% (Souza et al., 2023), suggesting these neurons are a lynch-pin for the HCVR and other putative respiratory chemoreceptor sites play a modulatory or condition-specific role in this response.

2. Cell ablation of RTN neurons

The effect of RTN lesion on the control of breathing has been extensively studied in adult rodents and largely consistent across different methods used for neuronal ablation. One widely adopted approach to eliminate RTN neurons utilizes local microinjections of the ribosomal-toxin Saporin conjugated to the analogue of substance-P (SSP-SAP), which results in cell death of neurokinin receptor 1-expressing neurons in a concentration-dependent manner (Nattie and Li, 2002; Souza et al., 2018; Takakura et al., 2014; Takakura et al., 2008). Eliminating roughly 70% of RTN neurons using SSP-SAP reduces the apneic threshold in the anesthetized mechanically-ventilated adult rats, which indicates that the effect of CO₂ on breathing is mediated by the RTN (Takakura et al., 2008). In freely-behaving adult rats, injections of SSP-SAP in the RTN cause alveolar hypoventilation at rest and reduces the hypercapnic ventilatory reflex (HCVR; Fig. 1 D and F) with no substantial effect on the hypoxic ventilatory reflex (HVR) (Nattie and Li, 2002; Souza et al., 2018; Takakura et al., 2014). Importantly, the degree of HCVR impairment is inversely related to the number of Nmb+ neurons remaining after injections of SSP-SAP, with a complete loss of the reflex requiring >75% of RTN^Nmb neurons to be eliminated (Souza et al., 2018) (Fig. 1 H). Hence, SSP-SAP provides an effective and accessible approach to eliminate RTN neurons in rats. However, these microinjections also destroy neighboring neurokinin receptor 1-expressing neurons that could have contributed to the observed deficits in breathing (Souza et al., 2018; Souza et al., 2019; Takakura et al., 2014), and also destroys glial cells (Lin et al., 2013).

Genetically-targeted methods offer a more selective approach to ablate neurons with a known genetic identity. Taking advantage of the selective expression of Nmb by RTN neurons, we developed a transgenic mouse that expresses Cre-recombinase in Nmb+ cells (Souza et al., 2023). Using this transgenic mouse, we microinjected an adeno-associated virus in the RTN that promotes cell-apoptosis only in cells expressing Cre. This approach, which eliminated over 90% of RTN^Nmb neurons without affecting other cell types, caused hypoventilation at rest and a severely blunted HCVR (Souza et al., 2023) (Fig. 1 E and G). Thus, lesion of RTN neurons produces alveolar hypoventilation in both rats and mice irrespective of the cell ablation approach. In both instances, hypoventilation was characterized by a reduced tidal volume, elevated PaCO₂ and reduced PaO₂ (Souza et al., 2018; Souza et al., 2023) (Fig. 1 I). The hypoventilation produced by RTN lesion in rats and mice leads to a fully compensated respiratory acidosis (i.e., arterial pH is unchanged due to a renal retention of bicarbonate in both species) (Fig. 1 I). A notable difference between SSP-SAP lesions of the RTN in rats and the genetically-targeted approach in mice concerns the respiratory pattern. In rats, lesion of RTN neurons leads to a higher respiratory frequency and reduced tidal volume in normoxic conditions compared to controls (Souza et al., 2018). In mice, eliminating RTN neurons reduces tidal volume without changing respiratory frequency during eupneic breathing in normoxic conditions compared to controls (Souza et al., 2023). However, RTN^Nmb lesion in mice increased the variability of the respiratory cycles under normoxic conditions, whereas this phenotype was only evident under conditions of hyperoxia in rats after lesion of the RTN neurons using SSP-SAP (Souza et al., 2018). This difference could be due to the effects of SSP-SAP on other neurokinin type-1 receptor-expressing neurons that control breathing such as those in the Bötzinger complex (Souza et. al., 2019) or simply a species difference.

One clear similarity between the effects of RTN neuron ablation in rats and mice is the increased reliance of breathing on oxygen-sensitive mechanisms. Following the lesion of RTN neurons in both rats and mice, exposure to hyperoxia leads to severe hypoventilation compared to intact controls (Souza et al., 2018; Souza et al., 2023) (Fig. 1 I–K). The effect of hyperoxia in these experiments reveals that breathing is highly reliant on oxygen-sensitive drive in the absence of RTN neurons. This oxygen-sensitive drive might originate from an increase in the activity of peripheral chemoreceptor and/or central oxygen sensors such as astrocytes (Angelova et al., 2015). Supporting this theory, lesion of RTN neurons using SSP-SAP and simultaneous carotid body denervation produces a greater reduction in resting minute-ventilation compared to the lesion of RTN neurons alone (Takakura et al., 2014). Notably, despite being greatly reduced, breathing persists when RTN-lesion animals are exposed to hyperoxia or after carotid body ablation. This hyperoxia-resistant and weak residual drive to breathe may rely on O₂-independent mechanisms such as the stimulatory effect of CO₂ on the carotid bodies (Lopez-Barneo, 2022) or alternative central respiratory chemoreceptors (Nattie and Li, 2012).

RTN lesions also increase sigh frequency at rest in rats and mice; this phenomenon is also a likely consequence of hypoxemia because hyperoxia restores sigh frequency to control levels in mice (Souza et al., 2023). These results suggest that RTN neurons might not be involved in hypoxia-induced sighs, although other evidence indicates that a portion of the RTN^Nmb neurons regulate emotional sighing (Li et al., 2020). Altogether, these data suggest that after lesion of RTN neurons, hypoxemia activates O₂-sensitive mechanisms that drive breathing and compensate, at least in part, for low alveolar ventilation (Fig1. I–K).

The effects of RTN neuron ablation were also explored in other behaviors related to CO₂-sensing. For example, chronic lesions of RTN lesions also impair arousal from non-REM sleep in response to CO₂ (Souza et al., 2019). However, CO₂-induced arousal persists after RTN ablation, indicating that the carotid bodies or CO₂-responsive CNS neurons other than RTN contribute to this reflex (Abbott and Souza, 2021; Kaur et al., 2020; Kaur and Saper, 2019). Interestingly, avoidance and freezing responses to hypercapnia are unaffected by RTN ablation (Souza et al., 2023). In summary, RTN neurons are an obligatory link of the central hypercapnic ventilatory reflex but their contribution to other behaviors elicited by hypercapnia ranges in importance from modest (arousal from sleep) to negligible (avoidance, freezing).

3. Genetic mutations affecting the development of RTN neurons

RTN neurons differentiate from cells within the rhombomere 5 of the embryonic hindbrain relying on the sequential expression of four transcription factors Egr-2, Phox2b, Lbx1, Atoh1 for correct differentiation and migration (Goridis and Brunet, 2010; Guyenet and Bayliss, 2022; Isik and Hernandez-Miranda, 2022). Conditional mutations have been used to study the effects of impaired RTN development on breathing (Goridis and Brunet, 2010). For example, Cre-dependent excision of Phox2b (Phox2b^lox/lox) in cells expressing Egr2 and Lbx1 leads to a loss of RTN neurons and an impairment of the fictive respiratory response to pH in reduced preparations (Dubreuil et al., 2009). Similarly, the conditional disruption of Phox2b in Atoh1-expressing neurons or the disruption of Atoh1 in PHOX2B-expressing neurons leads to a defective development of RTN neurons (Huang et al., 2012; Ruffault et al., 2015). In these mutant mice, the mortality rate is higher at birth and the ventilatory response to CO₂ is impaired, especially during the neonatal period, with partial recovery in adulthood (Ruffault et al., 2015).

RTN neurons have also been disrupted during development by introducing mutations in the Phox2b gene that are associated with CCHS such as polyalanine repeated expansion mutations (PARM) and the less common, non-polyalanine repeat expansion mutations (NPARM) form (Patwari et al., 2010; Weese-Mayer et al., 2010; Zhou et al., 2021). Heterozygous mice carrying a Phox2b PARM mutation (i.e., Phox2b^27Ala/+) lack most RTN neurons, display severe hypoventilation, and experience high mortality at birth (Dubreuil et al., 2008). When the same mutation is restricted to Egr-2 derived progenitors (via the Cre-expression conditioned to Egr-2; Phox2b^27alacki), the mutant protein PHOX2B^27ala is confined to developing RTN neurons and a small number of neurons in the peri-trigeminal region. Phox2b^27alacki mice also lack most RTN neurons at birth, but these mice are viable despite exhibiting a severe deficit of the HCVR during the post-natal period. The HCVR partially recovers within two weeks of birth (40% of that of the controls) via mechanisms that are yet to be determined (Ramanantsoa et al., 2011). Like rats and mice subject to RTN lesions in adulthood (Souza et al., 2018; Souza et al., 2023), Phox2b^27alacki mice present severe hypoventilation when exposed to hyperoxia, indicating that eupneic breathing is driven mostly by oxygen-sensitive mechanisms (Ramanantsoa et al., 2011). A recent study also explored how RTN neurons and breathing are affected by NPARM of the Phox2b gene (Phox2b^Δ8) confined to neurons expressing Atoh1 (Ferreira et al., 2022). Transgenic mice carrying Phox2b^Δ8 presented fewer RTN neurons than controls and an attenuated HCVR only in the newborn, which could be explained by the incomplete disruption of RTN function (Ferreira et al., 2022). Another mutation found in CCHS patients is the Lbx1 frameshift (Lbx1^FS) which, when conditional expressed in Egr-2 neurons results in a complete loss of RTN neurons, modest hypoventilation at rest and an impaired HCVR in newborn mice (Hernandez-Miranda et al., 2018).

In summary, the development of RTN neurons is exquisitely sensitive to Phox2b mutations found in CCHS. Mutations in Phox2b affecting predominantly RTN neuron development cause hypoventilation and loss of HCVR at birth. We hypothesize that these mutations are survivable because the breathing motor pattern generator is intact, and breathing is sustained by respiratory chemoreceptors other than RTN neurons. Viable mutants with loss of RTN neurons gradually recover roughly 40% of the HCVR of adult controls. This recovery is unexplained. Hypotheses include the following: a small fraction of RTN neurons survives in these mutants and develops new connection; CO₂ sensing by the carotid bodies is enhanced; or other central chemoreceptors are recruited. One limitation of targeting RTN neurons based on genetic lineage is that mutations linked to specific genes also affect the development of other classes of neurons besides RTN, which could contribute to the respiratory deficits and impaired CO₂ chemosensitivity (Goridis and Brunet, 2010).

4. Acute inhibition of RTN neurons

In freely-breathing conditions, the lesion of RTN neurons leads to compensatory responses that minimize the disruption in alveolar ventilation and blood gas homeostasis. These compensatory mechanisms obscure the contribution of RTN neurons to the drive to breathe at rest. For this reason, approaches that allow for acute, reversible, and repeatable silencing of RTN neurons are important to test their contribution to eupneic breathing under various physiological circumstances while minimizing potential compensatory mechanisms. To date, two approaches have been used to try to reach this goal: optogenetics and chemogenetics. These approaches utilize genetically-targeted expression of actuators that are activated by light (optogenetics) or receptor agonists (chemogenetics) to reduce cell excitability in a temporally-controlled fashion. An important difference between these approaches is the precision of temporal control and the mechanism of action. A majority of optogenetic methods provide instantaneous (<10 ms) control of cell activity owing to light-mediated ion channel activity. By contrast, chemogenetics relies on G protein-coupled receptors signaling pathways and the systemic administration of a small molecule agonist resulting in a relatively slow-onset (tens of seconds) and long-lasting inhibitory effect (min to hours) on cell activity.

Inhibitory optogenetics was first applied to RTN neurons in rats using viral vectors encoding the inhibitory opsin Archaerhodopsin (Arch), an outward proton pump, under the control of a Phox2a/b promoter (PRSx8)(Basting et al., 2015; Burke et al., 2015). More recently, this inhibitory opsin has been targeted to RTN neurons using a Cre-dependent vector encoding Arch associated with the transgenic Nmb-Cre mice (Souza et al., 2023). In both species, optogenetic inhibition of RTN neurons produces acute and reversible hypoventilation owing to a reduction in tidal volume and respiratory frequency (Fig 1 L) (Basting et al., 2015; Burke et al., 2015; Souza et al., 2023). Importantly, the magnitude of the instantaneous ventilatory inhibition is inversely proportional to blood pH; i.e., the effects of RTN inhibition increase as arterial pH decreases either pharmacologically by using acetazolamide or by increasing PaCO₂ and is reduced under conditions of respiratory alkalosis such as hypoxia (Fig 1 L) (Basting et al., 2015; Souza et al., 2023). Additionally, hypoventilation during acute inhibition of RTN neurons is accentuated by hyperoxia and carotid body denervation (Basting et al., 2016; Souza et al., 2023). Together, these data indicate that RTN neurons drive respiratory activity at rest in direct proportion to the arterial PCO₂ within a physiological range (Basting et al., 2015). Also, when oxygen-sensing mechanisms are silenced or ablated, RTN inhibition has a more pronounced effect on breathing, further emphasizing the interdependent relationship between RTN neurons and the peripheral chemoreceptor function (Guyenet et al., 2018).

Chemogenetic inhibition of RTN neurons using designer receptors activated only by designer drugs (DREADDs) or allatostatin is reported to have negligible effects on resting breathing in freely behaving rats (Marina et al., 2010; SheikhBahaei et al., 2023) and mice (Herent et al., 2023; Li et al., 2020), contrasting with the effects of optogenetic inhibition. However, chemogenetic inhibition of RTN neurons blunts the recruitment of abdominal muscles by CO₂ (Marina et al., 2010; SheikhBahaei et al., 2023) consistent with the potent excitatory effects of RTN neurons on active expiration (Souza et al., 2020). Furthermore, chemogenetic inhibition of RTN neurons appears to be capable of attenuating the hyperpnea associated with exercise (Herent et al., 2023), suggesting that RTN neurons are involved in the ventilatory response to increased metabolic demand.

In sum, acute and reversible inhibition of RTN neurons using optogenetics results in a marked reduction in ventilation when inhibition is performed both at rest and during hypercapnia, consistent with the effects of the lesion of RTN neurons. Chemogenetic inhibition of RTN neurons in freely behaving animals, in contrast, has modest effects on breathing relative to lesion and optogenetic inhibition of RTN neurons. There are some considerations that may account for this discrepancy, including selectivity of RTN neuron targeting and the nature of the method used to inhibit RTN neurons. First, RTN neurons intermingle with a variety of excitatory and inhibitory neurons making the results of studies using pan-neuronal promoters (for example, synapsin) to targeted RTN neurons difficult to interpret. Second, inhibition of RTN neurons with the commonly used DREADD, HM4Di, should theoretically preserve the intrinsic response of these neurons to pH, as well as paracrine and synaptic inputs. Third, because of the long-lasting effects of the chemogenetic approach, peripheral chemoreceptor activation likely counteracts the respiratory effects of chemogenetic inhibition of RTN neurons.

5. Disruption of neurotransmission in RTN neurons

Genetic models and RNA interference methods (e.g., short hairpin RNA; shRNA) have been recently introduced to characterize the action of some of the multiple neurotransmitters utilized by RTN neurons. The importance of glutamatergic signaling for RTN neurons function has been demonstrated using a Cre-dependent conditional knock-out of the gene Slc17a6, that encodes the vesicular transporter for glutamate type 2 (VGluT2^(flox/flox) mice)(Holloway et al., 2015). Local microinjection of a Cre-expressing vector driven by the Phox2b promoter, and a Cre-dependent excitatory opsin was used to optogenetically stimulate the transduced neurons (Holloway et al., 2015). In mice with intact VGluT2 (controls), optogenetic stimulation of RTN produced a robust increase in ventilation that was absent when RTN neurons lacking VGluT2 (VGluT2^(flox/flox)) were stimulated (Holloway et al., 2015). The release of glutamate from RTN neurons is also important for the HCVR as eliminating VGluT2 in Atoh1/Phox2b expressing neurons (VGlut2^lox/lox; Atoh1^FRTCre; P2b::FLPo) reduces the HCVR of neonatal mice (Ruffault et al., 2015). Resting breathing was only mildly reduced at rest (9%) by the genetic elimination of VGluT2 in RTN neurons in this model (Ruffault et al., 2015), again suggesting that powerful compensatory mechanisms mitigate the effects of any form of insult to RTN neurons. In sum, these studies show that glutamate release is critical for the stimulatory effect of RTN neurons on breathing and, in particular, the HCVR.

In addition to being glutamatergic, RTN neurons express mRNA of neuropeptides such as Neuromedin-B (NMB), Pituitary adenylate cyclase-activating peptide (PACAP), Galanin (GAL) and Enkephalin (ENK) (Shi et al., 2017). Yet the role of peptidergic signaling by RTN neurons is largely unknown. Based on mRNA content, the most abundantly expressed RTN peptide is PACAP (Shi et al., 2017) and recent studies suggest that the release of this peptide from RTN neurons seems especially important for breathing homeostasis during development (Shi et al., 2021). To test the role of this peptide in breathing regulation, Shi et al., (2021) used an shRNA-producing vector driven by PRSx8 to reduce PACAP expression in RTN neurons (Shi et al., 2021). This procedure reduced the HCVR, seemingly by attenuating the excitatory effect of RTN neurons on the pre-Bötzinger complex (Shi et al., 2021). Importantly, PACAP deletion in RTN neurons has no effect on VGlut2 expression by these neurons or on their intrinsic chemosensitivity (Shi et al., 2021). The expression of PACAP mRNA peaks in RTN neurons during the first 3 days after birth, which has been suggested to reflect an important role for this peptide in driving breathing during the early post-natal period (Shi et al., 2021). The function of the other RTN neuropeptides have not been thoroughly investigated including that of NMB. NMB release from RTN neurons has been proposed to contribute to emotional sighing (Li et al., 2016; Li et al., 2020), but direct evidence using the downregulation of NMB in RTN neurons is still lacking.

6. Glial cell disruption in the RTN

Evidence is growing that astrocytes in the RTN have a role in central respiratory chemoreflex by sensing changes in arterial pH and stimulating nearby neurons (Gourine et al., 2010). Undoubtedly, astrocytes from the RTN region are stimulated by changes in pH/PCO₂ (Gourine et al., 2010). However, to our knowledge, key experimental evidence showing the loss-of-function of astrocytes in the RTN is lacking. For example, experiments using selective glia ablation or impairing their function while preserving RTN neuronal integrity would provide valuable evidence for the role of astrocytes in the central respiratory chemoreflex. In fact, blocking purinergic neurotransmission was successful in reducing the respiratory output derived from astrocyte stimulation (Gourine et al., 2010; Wenker et al., 2012). The effects of purinergic blockade could be directly by inhibiting neurons (Gourine et al., 2010), by the changes in local vascular reactivity (Hawkins et al., 2017) or a combination of both. Indirect evidence using the knockout of the sodium-bicarbonate cotransporters (NBCe1s), one of the glial bicarbonate transporters, produced no changes in HCVR (Li et al., 2023) but a reduction in the ventilatory response to metabolic acidosis (Brady et al., 2023). Finally, when interpreting data from RTN lesions by SSP-SAP, it’s important to note this toxin also destroys glial cells (Lin et al., 2013) although the neurons are the major cell type affected. In a more refined approach, genetically-targeted approaches that selectively kill RTN neurons and preserve glial cells still showing a great impairment of the HCVR (Souza et al., 2023).

7. Conclusions

The experiments described in this review use complementary loss-of-function approaches to investigate the role of RTN neurons in the generation of PaCO₂-dependent respiratory drive in rodents. The most invariable outcome of disrupting RTN neurons, regardless of method, is an impaired ventilatory response to CO₂, including the recruitment of active expiration. This evidence indicates that RTN neurons are either a critical central respiratory chemoreceptor or that they integrate various inputs that are responsible for the HCVR. Selective knock-down of GPR4 and TASK-2 in RTN neurons would be useful to distinguish these possibilities. Another common finding is that deficits in breathing caused by disrupting RTN neurons are often buffered by oxygen-sensitive mechanisms that probably involve peripheral chemoreceptors but are not fully understood. Despite the commonalities, there is a debate on the relative contribution of RTN neurons to the many aspects of breathing regulation based on the distinct outcomes across different methods. We suggest these discrepancies can be attributed to the degree of which RTN neurons are inhibited or ablated, owing to either ineffective targeting or silencing. For this reason, it is important to evaluate the total proportion of RTN neurons that are disrupted or targeted and provide evidence that RTN neurons are effectively silenced, ideally in in vivo conditions. Moreover, in all the loss-of-function approaches of RTN neurons, it’s imperative to take into consideration the compensatory effects that maintain blood gas homeostasis, in particular in freely behaving animals. Overall, the functional studies using loss-of-function methods have provided strong support for the hypothesis that RTN neurons are an indispensable cluster of cells for the stimulatory effects of CO₂ on breathing.

Highlights.

• Neurons in the retrotrapezoid nucleus (RTN) are the most well characterized group of cells with putative central respiratory chemoreceptor function.

• Loss-of-function of RTN has been studied using different methods, each with their own particular limitations.

• Central respiratory chemoreception requires RTN neurons.