Contributions of carotid bodies, retrotrapezoid nucleus neurons and preBötzinger complex astrocytes to the CO2‐sensitive drive for breathing

Current models of respiratory CO2 chemosensitivity are centred around the function of a specific population of neurons residing in the medullary retrotrapezoid nucleus (RTN). However, there is significant evidence suggesting that chemosensitive neurons exist in other brainstem areas, including the rhythm‐generating region of the medulla oblongata – the preBötzinger complex (preBötC). There is also evidence that astrocytes, non‐neuronal brain cells, contribute to central CO2 chemosensitivity. In this study, we reevaluated the relative contributions of the RTN neurons, the preBötC astrocytes, and the carotid body chemoreceptors in mediating the respiratory responses to CO2 in experimental animals (adult laboratory rats). To block astroglial signalling via exocytotic release of transmitters, preBötC astrocytes were targeted to express the tetanus toxin light chain (TeLC). Bilateral expression of TeLC in preBötC astrocytes was associated with ∼20% and ∼30% reduction of the respiratory response to CO2 in conscious and anaesthetized animals, respectively. Carotid body denervation reduced the CO2 respiratory response by ∼25%. Bilateral inhibition of RTN neurons transduced to express Gi‐coupled designer receptors exclusively activated by designer drug (DREADDGi) by application of clozapine‐N‐oxide reduced the CO2 response by ∼20% and ∼40% in conscious and anaesthetized rats, respectively. Combined blockade of astroglial signalling in the preBötC, inhibition of RTN neurons and carotid body denervation reduced the CO2‐induced respiratory response by ∼70%. These data further support the hypothesis that the CO2‐sensitive drive to breathe requires inputs from the peripheral chemoreceptors and several central chemoreceptor sites. At the preBötC level, astrocytes modulate the activity of the respiratory network in response to CO2, either by relaying chemosensory information (i.e. they act as CO2 sensors) or by enhancing the preBötC network excitability to chemosensory inputs.


Introduction
Respiratory CO 2 chemosensitivity is fundamentally important for the control of breathing.Under resting conditions, the central respiratory network that generates the rhythm of breathing requires a certain level of CO 2 to be active (Nielsen & Smith, 1952).Indeed, under most physiological conditions, lung ventilation is directly proportional to the amount of CO 2 produced during metabolism (Phillipson et al., 1981).Specialized chemoreceptors located in the carotid bodies (and aortic bodies in some species) and in the central nervous system are responsible for respiratory CO 2 /pH chemosensitivity (Heymans & Bouckaert, 1930;Heymans & Neil, 1958;O'Regan & Majcherczyk, 1982).There is evidence that approximately one-third of the ventilatory response to CO 2 is mediated by the peripheral chemoreceptors, whereas the rest of the response is attributed to the actions of CO 2 /H + on central respiratory chemoreceptors residing in the brainstem (Bisgard et al., 1976;Dahan et al., 2008;Forster et al., 2008;Guyenet, 2014;Heeringa et al., 1979;Mitchell et al., 1963Mitchell et al., , 2006;;Pan et al., 1998;Richerson, 1995;Takakura et al., 2011).
Current models of central respiratory chemosensitivity (the mechanism that controls breathing in accord with brain parenchymal P CO2 /pH) propose that specialized neurons in the retrotrapezoid nucleus (RTN) and medullary raphé are primarily responsible for central CO 2 sensing, which is detected via proxy of extracellular pH changes (Guyenet et al., 2019;Kumar et al., 2015;Teran et al., 2014).An alternative hypothesis of 'distributed central chemosensitivity' suggests that central respiratory sensitivity to CO 2 is mediated by inputs from several brainstem sites (including the RTN and raphé), with each site providing a fraction of the tonic drive to breathe in eucapnia and a proportion of the overall response to hypercapnia (Forster et al., 1997;Nattie, 1999Nattie, , 2000)).The RTN may act not only as a CO 2 sensor but also as an upstream integration centre for CO 2 -sensitive inputs to the preBötzinger complex (pre-BötC), the key site of inspiratory rhythm generation (Del Negro et al., 2018;Smith et al., 1991).The preBötC is also capable of increasing lung ventilation in response to acidification (Feldman et al., 2003;Koizumi et al., 2010;Krause et al., 2009;Solomon, 2003a;Solomon et al., 2000).Experimental studies in rats involving localized acidification or lesions of the preBötC neurons suggested that the preBötC contributes ∼20%-25% of the ventilatory response to CO 2 (Grey et al., 2001;Nattie, 2000).Impairment of the signalling function of preBötC astrocytes by interfering with the exocytotic release of astroglial signalling molecules results in a similar 20%−25% reduction of the CO 2 -induced ventilatory response (Sheikhbahaei, Turovsky et al., 2018).
This study was designed to define the role of the pre-BötC mechanism(s), focusing specifically on preBötC astrocytes, in central respiratory CO 2 chemosensitivity.Using two experimental models -conscious and anaesthetized adult rats -we compared the respiratory responses to the increases in inspired CO 2 before and after a series of interventions (applied separately or in combination) that included (i) the blockade of exocytotic release mechanisms in preBötC astrocytes by virally driven expression of tetanus toxin light chain (TeLC) (Angelova et al., 2015;Rajani et al., 2018;Sheikhbahaei, Turovsky et al., 2018), (ii) inhibition of RTN neurons transduced to express G i -coupled designer receptors exclusively activated by designer drug (DREADD Gi ) by application of clozapine-N-oxide (Basting et al., 2015;Huckstepp et al., 2015;Korsak et al., 2018;Marina et al., 2010), and (iii) surgical denervation of the carotid bodies bilaterally (Sheikhbahaei et al., 2017;Smith et al., 2015).We show that in the anaesthetized state, disruption of preBötC astroglial signalling reduced the respiratory response to hypercapnia by ∼30%; this effect was similar to that produced by the removal of inputs from the peripheral chemoreceptors or inhibition of RTN neurons.In conscious animals under hyperoxic conditions (to minimize the activity of the carotid bodies), blockade of preBötC astroglial signalling reduced the CO 2 -induced increases in the respiratory frequency by ∼20%, whereas inhibition of RTN neurons reduced the increases in tidal volume by the same degree, ∼20%.These data support the hypothesis of distributed central CO 2 chemosensitivity.At the preBötC level, astrocytes contribute to the CO 2 -sensitive drive to breathe, either by relaying chemosensory information (i.e.acting as CO 2 sensors) or by enhancing preBötC network excitability to chemosensory inputs.

Ethical approval
All the experiments were performed in adult Sprague-Dawley rats in accordance with the European Commission Directive 2010/63/EU (European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes), the UK Home Office (Scientific Procedures) Act (1986) and the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the respective Institutional Animal Care and Use Committees.The rats were housed in a temperature-controlled facility (22-25°C) with a 12 h light-dark cycle (lights switched on at 7:00 AM).Water and laboratory rodent food chow were provided ad libitum.This research was conducted in accordance with the animal ethics checklist outlined in the Journal's instructions for authors.

S. SheikhBahaei and others
J Physiol 602.1

Adeno-associated viral vector for the expression
of DREADD Gi in RTN neurons.RTN neurons were transduced to express DREADD Gi under the transcriptional control of human synapsin (hSyn) promoter (AAV2-hSyn-DREADD Gi -mCitrine; 5.6 × 10 12 unit ml −1 ; UNC Vector Core, Chapel Hill, NC, USA).Acute inhibition of RTN was then achieved by application of the DREADD ligand CNO (Armbruster et al., 2007;Gomez et al., 2017;Rogan & Roth, 2011).hSyn promoter was used; therefore, all neurons at the site of the injection were expected to express the transgene.The efficacy and specificity of this approach in inhibiting RTN neuronal activity have been demonstrated previously (Huckstepp et al., 2015;Korsak et al., 2018).Adeno-associated viral vector (AAV)-induced expression of ChR2-EGFP (4 × 10 12 unit ml −1 , UNC Vector Core) in RTN neurons was used in the control group of animals.

Viral gene transfer
Young male rats (60−80 g) were anaesthetized with an intramuscular injection of ketamine (60 mg kg −1 ) and medetomidine (250 μg kg −1 ).Supplemental anaesthesia was administered, and adequate anaesthesia was ensured throughout the surgery through the absence of a withdrawal response to a paw pinch.Animals were placed in a stereotaxic frame (tooth bar −18 mm below the interaural line), and the dorsal surface of the brainstem was exposed by separating the neck muscles at the mid-line followed by an incision of the atlanto-occipital membrane.The RTN was targeted bilaterally with two microinjections per side (100-120 nl each, ∼50 nl min −1 ) of DREADD Gi -mCitrine or ChR2-EGFP-expressing AAVs using the following coordinates from calamus scriptorius: 1.7 mm lateral from the mid-line and 3.7 mm ventral, with the micropipette holder arm positioned at 24°and 16°for each injection (Korsak et al., 2018;Marina et al., 2010).Anaesthesia was reversed with atipamezole (1 mg kg −1 , i.m.).No complications were observed after the surgery, and the animals gained weight normally.Carprofen (5 mg kg −1 per day subcutaneously) was given for postsurgical analgesia for 3 days, and the animals were allowed to recover for 7 days.After a recovery period of 3−4 weeks (the animals reaching the body weights of ∼200−250 g), AVV constructs designed to induce the expression of EGFP-TeLC or CatCh-EGFP in astrocytes were microinjected bilaterally into the preBötC following the same anaesthetic and surgical protocol.The microinjections were centred at the preBötC region and delivered using the following coordinates: 2 mm lateral, 0.9 mm rostral and 2.7 mm ventral from the calamus scriptorius (Rajani et al., 2018;Sheikhbahaei, Turovsky et al., 2018).Experiments were conducted 5−7 days after the injections (Fig. 1).

Experimental groups
(1) The control group of rats was transduced to express CatCh-EGFP in preBötC astrocytes and ChR2-EGFP in RTN neurons (control, n = 5).
(2) The first experimental group of rats was transduced to express CatCh-EGFP in preBötC astrocytes and DREADD Gi in RTN neurons (RTN-DREADD GI , n = 5).CNO was then administered to inhibit RTN neurons transduced to express DREADD Gi to determine RTN contribution to the hypercapnic ventilatory response.CO 2 challenges were applied before and after denervation of the carotid bodies.The hypercapnic ventilatory response remaining after the inhibition of RTN and denervation of the carotid body was interpreted to represent the response mediated by other central CO 2 -chemosensitive sites, including the preBötC.
(3) The second experimental group of rats transduced to express TeLC in preBötC astrocytes and DREADD Gi in RTN neurons (preBötC-TeLC/RTN-DREADD Gi , n = 5).A contribution of preBötC astrocytes to the hypercapnic ventilatory response was suggested by the data reported previously (Beltrán-Castillo et al., 2017;Sheikhbahaei, Turovsky et al., 2018).In this group of animals CO 2 challenges were applied before and after denervation of the carotid bodies and subsequently after RTN inhibition following CNO application.The hypercapnic ventilatory response remaining in these animals after RTN inhibition and carotid body denervation was interpreted to represent the response mediated by central CO 2 -chemosensitive sites, other than RTN and the preBötC; that is, the difference in the ventilatory response to CO 2 between the experimental animals in groups 2 and 3 reflected the contribution of the preBötC astrocytes.

Experimental design
All experiments and data analyses were conducted by investigators who were blinded to the identity of the experimental groups.
Experiment 1.The aim was to determine the relative contribution of RTN neurons and preBötC astrocytes to the hypercapnic ventilatory response in conscious rats.
Respiratory activity in conscious rats was assessed using whole-body plethysmography (Enhorning et al., 1998;Mortola & Frappell, 1998;Sheikhbahaei et al., 2017).Animals were placed in a Plexiglass recording chamber (∼1 l, temperature: 22−24°C) that was flushed continuously (1.2 l min -1 ) with humidified gas mixture containing >60% O 2 (balanced with N 2 ) to reduce the drive from the peripheral chemoreceptors (Chavez-Valdez et al., 2012;Gonzalez et al., 1994).To reduce the potentially confounding influence of circadian variation in physiological responses, all experiments were conducted at the same time of the day (between 11:00 AM and 3:00 PM).Concentrations of O 2 and CO 2 in the chamber were monitored online using a fast-response O 2 /CO 2 analyser (ML206, AD Instruments, Colorado Springs, CO, USA).To determine the control response to CO 2 , all animals received an injection of saline (0.3-0.4 ml, i.p.).The animals were allowed to acclimatize to the chamber environment for at least 1 h before baseline respiratory activity was measured.Hypercapnia was then induced by stepwise increases in CO 2 concentration in the respiratory gas mixture from <0.3% to 3% and 6% (since 3% CO 2 did not trigger a significant effect on ventilation in these experiments, we only report the responses evoked by 6% CO 2 ).Each CO 2 concentration was maintained for 5 min.Animals were left to recover for 2 days, after  which the CO 2 challenge was repeated after the systemic administration of CNO (2 mg kg −1 , i.p.) to inhibit RTN neurons transduced to express DREADD Gi .Treatment of preBötC-TeLC/RTN-DREADD Gi rats with CNO allowed us to assess the combined contribution of RTN neurons and preBötC astrocytes to the hypercapnic ventilatory response.The contribution of preBötC astrocytes was determined by comparing the responses recorded in preBötC-TeLC/RTN-DREADD Gi and RTN-DREADD Gi rats prior to CNO administration (Fig. 1).
Experiment 2. The aim was to determine the relative contribution of the carotid body chemoreceptors, RTN neurons, and preBötC astrocytes to the hypercapnic ventilatory response in anaesthetized rats.Animals were anaesthetized with urethane (1.3 g kg −1 , i.v.) after cannulation of the femoral artery and vein under 3% isoflurane.Adequate anaesthesia was ensured throughout the experiment through continuous monitoring of the arterial blood pressure stability and the absence of a withdrawal response to a paw pinch.Supplemental anaesthesia was administered as needed.Blood gases were monitored (Table 1), the trachea was cannulated, and the animal was mechanically ventilated with a gas mixture containing ∼30% O 2 and ∼70% N 2 .End-tidal levels of O 2 and CO 2 were monitored using a fast-response O 2 /CO 2 analyser (ML206, AD Instruments).Phrenic nerve activity (PNA) and EMG ABD were recorded as measures of central inspiratory and expiratory activities, respectively (Huckstepp et al., 2015;Marina et al., 2010).The carotid sinus nerves (CSN) were dissected for subsequent bilateral sectioning.Core body temperature was kept at ∼37°C using a servo-controlled heating blanket.P O2 , P CO2 , and pH of the arterial blood were measured prior to each CO 2 challenge.After surgical preparation, each animal was allowed to stabilize for ∼30 min.To determine the baseline inspiratory and expiratory responses to CO 2 , phrenic nerve and EMG ABD activities were recorded for 5 min followed by the application of the first hypercapnic challenge (10%-12% CO 2 in the inspired gas mixture for 5 min; Fig. 2).The contribution of preBötC astrocytes to the hypercapnic ventilatory response was determined by comparing the peaks of CO 2 -evoked PNA and EMG ABD responses between the preBötC-TeLC/RTN-DREADD Gi and RTN-DREADD Gi experimental animals prior to the administration of CNO.
After ∼20 min recovery from the first CO 2 challenge (to allow blood gases to return to normal levels), the inputs from the carotid bodies were removed by bilateral sectioning of the CSNs.CO 2 -evoked PNA and EMG ABD responses were determined 20 min after carotid body denervation (CBD).The differences in CO 2 -evoked PNA and EMG ABD responses before and after CBD revealed the contribution of the peripheral chemoreceptors to the hypercapnic respiratory response in this preparation.Shortly after the blood gases returned to baseline levels (∼20 min), the rats were given CNO (Tocris Bioscience, Minneapolis, USA, 2 mg kg −1 , i.v.) followed by a third hypercapnic challenge 20 min later.The contribution of RTN neurons to the hypercapnic ventilatory response was determined by assessing the differences in CO 2 -evoked PNA and EMG ABD responses between the control and RTN-DREADD Gi experimental animals after the administration of CNO.

Data analysis
Analysis of whole-body plethysmography recordings (experiment 1).The duration (T TOT ) and amplitude (tidal volume, V T ) of the respiratory cycles were determined after the animals had habituated to the plethysmography chamber environment for at least 1 h.The average baseline T TOT was calculated for the respiratory cycles recorded during periods of calm wakefulness and/or quiet sleep recorded over a 30-min period and was used to determine the respiratory frequency (number of breaths per minute, ƒ R ) as described (Sheikhbahaei et al., 2017;Sheikhbahaei, Turovsky et al., 2018).V T (normalized to the body weight) was determined by measuring the amplitude of Figure 2. Hypercapnic respiratory responses in anaesthetized and artificially ventilated rats Time-condensed records illustrating changes in the amplitude of the integrated phrenic nerve activity (∫PNA) and abdominal EMG (∫EMG ABD ) in response to sequential hypercapnic challenges in rats transduced to express DREADD Gi in RTN (retrotrapezoid nucleus) neurons (RTN-DREADD Gi ).The initial response to the hypercapnic challenge (5 min, 10%−12% CO 2 ) was induced in rats with intact carotid bodies.The carotid body nerves were then bilaterally sectioned [CBD (carotid body denervation)], and the rat was exposed to the second hypercapnic challenge.Shortly after blood gas levels returned to normal physiological ranges, CNO was applied systemically (2 mg kg −1 , I.V.), followed by the third hypercapnic challenge.Animals were left to recover for at least 20 min between each experimental manipulation.Changes in PNA recorded during hypercapnia before (initial) and after each treatment (CBD and application of CNO) were normalized with respect to resting activity (100%) and expressed as the percentage increase in PNA (CO 2 -evoked PNA).Expanded recordings of PNA and EMG ABD are shown, illustrating late-expiratory abdominal activity evoked by CO 2 that was blocked by CNO application.
[Colour figure can be viewed at wileyonlinelibrary.com]J Physiol 602.1 pressure changes in the chamber.Minute ventilation (V E ) was calculated using the following formula: Analysis of data obtained in anaesthetized preparations (experiment 2).Electrophysiological data were digitized (3 kHz sampling rate) and analysed offline (Spike2, Cambridge Electronic Design, Cambridge, UK).Peak PNA responses to CO 2 (Fig. 2) before and after CBD, and then after CNO administration in conditions of CBD, were compared and expressed as percentage changes from the control CO 2 response recorded at baseline (Fig. 2).Peak increases in PNA were also analysed and expressed as percentage increases in the inspiratory burst amplitude relative to the resting activity.Phrenic nerve discharge frequency was not analysed as it was entrained by the ventilator frequency.

Statistical analysis
The data were reported as mean (SD) and compared by paired t test, unpaired t test or one-way ANOVA using GraphPad Prism 9 (RRID:SCR_002798).Each data point on figures represents data obtained from one animal.

Targeting the RTN neurons to express DREADD Gi
The extent of bilateral DREADD Gi or control transgene expression in the RTN was examined in all the experimental animals.To aid the identification of transduced cells, anti-GFP immunostaining was performed (Huckstepp et al., 2015;Korsak et al., 2018).Transduced neurons were identified within the targeted RTN region, extending in the rostrocaudal direction between −10.8 and −12.1 mm from bregma (Fig. 3A).Fig 3B shows a representative image taken from the centre of the RTN.Transduced cells were counted in a section with maximum expression (Korsak et al., 2018), which revealed 61 (28) transduced neurons per section at this level.No staining was observed in the neighbouring Bötzinger complex (BötC), and no co-localization with TH immunoreactivity was observed (Fig. 3C), suggesting that the brainstem catecholaminergic neurons were not transduced or were expressing the transgenes at a very low level.

Targeting preBötC astrocytes to express TeLC
The nucleus ambiguus (NA, immunostained for ChAT) was used as an anatomical landmark to aid in the identification of the preBötC region, which is located ventral to the semi-compact division of the nucleus (NAsc; Fig. 3D).The extent of the bilateral TeLC expression in preBötC astrocytes of all the experimental animals was histologically reconstructed from serial coronal sections after immunohistochemical detection of EGFP (Fig. 3D).Transduced astrocytes were found rostrocaudally from −12.1 to −13.4 mm relative to bregma, with the peak density of expression observed at the preBötC level (Fig. 3D).A representative high-magnification confocal image illustrating TeLC expression in preBötC astrocytes (−12.8 mm relative to bregma) is shown in Fig. 3E.TeLC expression driven by this vector was observed exclusively in astrocytes, as reported previously (Angelova et al., 2015;Dutta et al., 2018;Rajani et al., 2018;Sheikhbahaei, Turovsky et al., 2018)
The ventilatory response to hypercapnia was reduced in rats expressing TeLC in preBötC astrocytes.The peak ƒ R response to 6% inspired CO 2 was reduced by 18% [168 (18) min −1 vs. 204 (20) min −1 in control animals; p = 0.021, unpaired t test; Fig. 4A], and peak V E response was reduced by 21% (p = 0.018, unpaired t test; Fig. 4A).TeLC expression in the preBötC had no effect on V T under any conditions (Fig. 4A).
Inhibition of RTN neurons after CNO administration (2 mg kg −1 , i.p.) had no effect on resting ventilation (in conditions of 60% inspired oxygen).Peak increases in ventilation in response to 6% CO 2 were significantly reduced when RTN neurons were inhibited (Fig. 4B).After CNO administration, CO 2 -induced increases in V T and V E were smaller in the RTN-DREADD Gi group; V T was reduced by 18% (p = 0.006, unpaired t test), and V E was similarly reduced by 18% (p = 0.010, unpaired t test).Inhibition of RTN neurons had no effect on the respiratory frequency under any conditions (Fig. 4B).
Combined blockade of exocytotic release mechanisms in preBötC astrocytes by virally driven expression of TeLC, and inhibition of RTN neurons expressing DREADD Gi following systemic administration of CNO (in the preBötC-TeLC/RTN-DREADD Gi group), led to a 65% reduction in the ventilatory response to CO 2 (p < 0.001, unpaired t test; Fig. 4B).These data suggest that in conscious animals in conditions when the drive from the peripheral chemoreceptors is reduced by hyperoxia, RTN neurons mediate the CO 2 -induced increases in tidal volume, whereas preBötC mechanisms contribute to the increases in respiratory frequency.

The contribution of carotid body chemoreceptors, RTN neurons and preBötC astrocytes to hypercapnic respiratory responses in anaesthetized rats
Blockade of exocytotic release mechanisms in preBötC astrocytes or the expression of DREADD Gi in the RTN did not affect the homeostatic control of blood gases (Table 1).The magnitude of the initial (control) response to CO 2 (5 min), expressed as a percentage increase from the baseline PNA amplitude (% PNA), was not significantly different between the groups of experimental animals [controls: increase 114 (54)%; RTN-DREADD Gi : To confirm that we could effectively inhibit the RTN neuronal population, we evaluated the effects of CNO administration on CO 2 -induced increases in the EMG ABD activity.CNO administration after CBD effectively abolished recruitment of late-expiratory EMG ABD activity evoked by CO 2 in both RTN-DREADD Gi and preBötC-TeLC/RTN-DREADD Gi experimental animals (Figs 2 and 5), consistent with the previously reported data (Marina et al., 2010).
To determine the contribution of RTN neurons to the inspiratory component of the hypercapnic response, we analysed the evoked responses using two complementary methods.First, we compared the magnitude of CO 2 -induced PNA increases in RTN-DREADD Gi and preBötC-TeLC/RTN-DREADD Gi to that recorded in the control group.A between-subjects analysis showed that CO 2 -evoked PNA responses were reduced after CNO administration by ∼47% in RTN-DREADD Gi rats (p < 0.001, unpaired t test; Fig. 7A) and by ∼70% in preBötC-TeLC/RTN-DREADD Gi animals (p < 0.001, unpaired t test; Fig. 7A).
Second, we performed a within-subject analysis of the CO 2 -induced PNA responses obtained before and after CBD and CNO administration where the responses were expressed as percentages of the first (control) and second hypercapnic responses (after CBD).In two groups of experimental animals transduced to express DREADD Gi in the RTN (RTN-DREADD Gi and preBötC-TeLC/RTN-DREADD Gi ), CNO administration led to a significant reduction in the hypercapnic respiratory response (reduction by ∼54% when compared to the initial response; p = 0.001, paired t test; Fig. 7B; and by ∼37% when compared to the CO 2 response recorded after CBD; p = 0.002, paired t test; Fig. 7B).These data suggest that in anaesthetized rats in the absence of the carotid body input, the RTN mechanisms contribute approximately one-third to the overall respiratory response to CO 2 .
To evaluate the contribution of preBötC astrocytes to the hypercapnic respiratory response, we next compared the magnitude of the CO 2 -induced increases in PNA between the RTN-DREADD Gi and preBötC-TeLC/ RTN-DREADD Gi experimental animals.The first response to CO 2 in the preBötC-TeLC/RTN-DREADD Gi group was not significantly different from that recorded in the RTN-DREADD Gi group [increase by 71 (21)% vs. 103 (45)%, n = 5 per group; p = 0.18, unpaired t test; Fig. 8; see also Fig. 5A].Blood gas analysis showed no differences between preBötC-TeLC/RTN-DREADD Gi and the other two experimental groups (Table 1).After CBD, the PNA increases induced by CO 2 were reduced by ∼31% in the preBötC-TeLC/RTN-DREADD Gi group compared to that in the RTN-DREADD Gi group (p = 0.024; unpaired t test).These data suggest that in anaesthetized rats, in the absence of the carotid body input, preBötC astrocytes contribute approximately one-third to the overall respiratory response to CO 2 .Subsequent RTN inhibition by application of CNO reduced the amplitude of CO 2 -induced increase in PNA by a further 42% (p = 0.012, unpaired t test; Fig. 8).

Discussion
In this study, we further investigated the role of the preBötC in the mechanisms of central respiratory CO 2 sensitivity.The data obtained show that under general anaesthesia, CO 2 sensitivity of the preBötC is mediated by astrocytes that contribute 31% to the overall respiratory response to hypercapnia.This contribution is similar to that of the peripheral chemoreceptors of the carotid body (28%) and the RTN mechanisms (37%).Experiments conducted in conscious animals showed that preBötC mechanisms mediate CO 2 -induced increases in the respiratory frequency, whereas RTN neurons are responsible for CO 2 -induced increases in tidal volume.
Removal of the carotid body input decreases the respiratory frequency at rest and increases the variability of breathing, and has a major impact on the ventilatory responses to hypoxia and hypercapnia (Andronikou et al., 1988;Angelova et al., 2015;Dahan et al., 2007;Forster et al., 2008;Nakayama et al., 2003;Sheikhbahaei et al., 2017;Smith et al., 2003;Timmers et al., 2003).The data obtained in the present study are in agreement with the results of previous studies, demonstrating that the chemosensory drive from the carotid bodies is responsible for ∼30% of the respiratory sensitivity to CO 2 (Andronikou et al., 1988;Dahan et al., 2007;Gautier & Bonora, 1979;Izumizaki et al., 2004;Olson et al., 1988).
Viral targeting of the RTN neurons to express inhibitory receptors was established and validated in our earlier studies (Huckstepp et al., 2015;Korsak et al., 2018;Marina et al., 2010).In this study, histological analysis of every sixth 50 μm section identified ∼60 transduced neurons per side, which translates to ∼3600 neurons expressing DREADD Gi .Although we used immunohistochemical detection of mCitrine, it is possible that we may have underestimated the number of transduced neurons.However, the application of CNO completely abolished active expirations, a result similar to that obtained in studies that used selective chemogenetic inhibition of Phox2b-expressing RTN neurons (Marina et al., 2010).This suggests that a critical number of RTN neurons were effectively inhibited in our experiments, consistent with the proposed key role of the RTN neurons in controlling the active expirations (Marina et al., 2010).The percentage reduction in CO 2 -evoked inspiratory (phrenic nerve activity) response after inhibition of DREADD Gi -expressing RTN neurons observed in this study was similar to that reported previously (Bhandare et al., 2020;Korsak et al., 2018;Marina et al., 2010).In conscious rats when the inputs from the peripheral chemoreceptors were reduced by hyperoxia, the inhibition of the RTN neurons had no effect on resting respiratory activity as previously observed in normoxic conditions (Marina et al., 2010).This observation differs from the data showing reduction in breathing frequency and tidal volume observed with optogenetic inhibition of RTN neurons under similar conditions (Basting et al., 2015;Burke et al., 2015), suggesting that RTN inhibition may have been more effective in these experiments.However, the CO 2 -evoked increases in tidal volume and minute ventilation were reduced by ∼20% in our experiments.In anaesthetized and artificially ventilated rats in conditions of bilateral CBD, RTN inhibition decreased the respiratory response to CO 2 by 37%, similar to that reported previously (Huckstepp et al., 2015;Marina et al., 2010).These results are consistent with the recently reported data, suggesting that in awake rodents RTN neurons may be less responsive to CO 2 than under general anaesthesia (Bhandare et al., 2020).
The preBötC has chemosensory properties (Solomon, 2003a(Solomon, , 2003b;;Solomon et al., 1999Solomon et al., , 2000)), and some pre-BötC neurons respond to acidification in vitro (Koizumi et al., 2010).PreBötC astrocytes can modulate the respiratory response to hypercapnia via the release of d-serine and/or ATP (Beltrán-Castillo et al., 2017;Sheikhbahaei, Turovsky et al., 2018;Turk et al., 2022).The data obtained in this study further support the earlier observations in conscious animals that astrocytes within preBötC rhythm-generating circuits contribute to CO 2 -induced respiratory responses (Sheikhbahaei, Turovsky et al., 2018).In conscious animals, interfering with exocytotic release mechanisms in preBötC astrocytes decreased baseline ƒ R and concomitantly V E but had no effect on V T .
These data support the conclusions reached by Grey et al. (2001) and Nattie (2000), who suggested that the contribution of preBötC mechanism(s) to the overall respiratory response to CO 2 is ∼20%-25%.Indeed, in conscious animals, lung ventilation during resting breathing (eucapnia) and at the peak of the CO 2 response was reduced by ∼20% when vesicular release mechanisms in preBötC astrocytes were blocked.Thus, astroglial signalling in the preBötC plays an important role in orchestrating the ventilatory response to CO 2 .Whether this reflects direct sensitivity of preBötC astrocytes to CO 2 or astroglial modulation of preBötC network excitability to chemosensory inputs remains to be established.We have not demonstrated that preBötC astrocytes are sensitive to changes in pH/CO 2 , but such sensitivity has been shown for other brainstem astrocytes, including astrocytes that reside in the RTN (Gourine & Dale 2022;Kasymov et al., 2013).
Simultaneous inhibition of RTN neuronal population and blockade of exocytotic release mechanisms in the pre-BötC astrocytes decreased minute ventilation by ∼35% in conscious rats, suggesting that the RTN and pre-BötC mechanisms may control distinct components of the ventilatory response (V T and ƒ R , components, respectively).In experiments conducted in anaesthetized rats, the effect of inhibiting RTN neurons and pre-BötC astrocytes on phrenic nerve response was found to be additive, resulting in ∼70% reduction of the CO 2 response.Although we cannot rule out the decreased excitability of the respiratory network due to the removal of the carotid body input, our results are consistent with previously reported data, suggesting that urethane anaesthesia may reduce the drive from some chemosensitive regions (Massey & Richerson, 2017).In our experimental conditions, carotid body denervation led to a 28% reduction in the response, whereas combined blockade of astroglial signalling in the preBötC, bilateral RTN inhibition and carotid body denervation reduced the CO 2 -induced response by about 70%.These data support the hypothesis of distributed CO 2 chemosensitivity, which proposes that the CO 2 -sensitive drive to breathe is mediated by interacting peripheral and central mechanisms, including the interactions between chemosensitive brainstem neurons and glial cells.

Experimental limitations
In this study, to determine the relative contribution of the RTN neuronal population to the hypercapnic respiratory S. SheikhBahaei and others J Physiol 602.1 response, we used a viral vector with a pan-neuronal promoter.We did not target the specific subpopulation of RTN neurons that express the specific markers of the chemosensitive neurons, and inhibition of other neurons within the region could potentially have changed the respiratory response to CO 2 (Cleary et al., 2021).The physiological data showing complete blockade of active expirations and significant (by ∼40%) reduction in the inspiratory response to CO 2 were found to be very similar to that reported in the preceding studies (Marina et al. 2010;Nattie & Li, 2002, 2006;Takakura et al., 2014), suggesting that a critical number of RTN neurons were successfully inhibited in these experiments.In addition, interpretation of these results is complicated because the role of the carotid body input in modulating the CO 2 sensitivity of both sites remains unknown.

Figure 1 .
Figure 1.Experimental paradigm After stereotaxic injections of AAV2-hSyn-DREADD Gi -mCitrine (or control vector in RTN) and AVV-sGFAP-EGFP-TeLC (or control vector in preBötC), hypercapnic ventilatory responses in each rat were recorded in conscious and anaesthetized states.[Colour figure can be viewed at wileyonlinelibrary.com]

Figure 3 .
Figure 3. Targeting RTN (retrotrapezoid nucleus) neurons to express DREADD Gi and preBötC astrocytes to express tetanus toxin light chain (TeLC) A, Schematic illustration of the extent of DREADD Gi expression after GFP immunostaining (within coloured regions shown schematically) in the ventrolateral regions of the medulla oblongata in representative reconstructed serial coronal brainstem sections of adult rats.Adeno-associated viral vectors designed to drive the expression of DREADD Gi -mCitrine were used to target RTN neurons bilaterally.The peak density of transduced neurons was within the RTN (extending rostrocaudally from bregma −12.1 and −10.8 mm), with limited expression in the immediately adjacent rostral and caudal areas of the ventrolateral medullary reticular formation, including the lateral parafacial area.B, Representative confocal images illustrating GFP immunostaining corresponding to the brainstem section −11.6 mm (from bregma).The expression of DREADD Gi -mCitrine in RTN neurons was amplified with anti-GFP immunostaining.C, Higher-magnification images showing that neurons transduced to express DREADD Gi (green, left panel) do not express TH (tyrosine hydroxylase, white arrowheads; red, middle and merged images).D, Schematic illustration from a representative adult rat brainstem showing the extent of TeLC expression (revealed by GFP immunostaining, green-coloured regions) in the ventrolateral medulla oblongata in astrocytes of the preBötC.Adenoviral vector designed to drive the expression of EGFP-TeLC was used to target preBötC regions bilaterally.E, A representative, high-magnification confocal image illustrating GFP immunolabeling (and therefore TeLC expression) at the preBötC region of maximum expression (−12.8 mm relative to bregma).VS, ventral surface.Abbreviations: py, pyramids; NAsc, semi-compact division of the nucleus ambiguous; IO, inferior olive; SP5, spinal trigeminal nucleus; VII, facial nucleus; VS, ventral surface.[Colour figure can be viewed at wileyonlinelibrary.com]

Figure 4 .
Figure 4. Contributions of RTN (retrotrapezoid nucleus) neurons and preBötC astrocytes to the development of the ventilatory response to CO 2 A, Group data illustrating the effect of DREADD Gi expression in RTN neurons (without CNO) and TeLC (tetanus toxin light chain) expression in preBötC astrocytes on CO 2 -induced increases in ƒ R , V T and V E in conscious rats under conditions of peripheral chemoreceptor inhibition by hyperoxia.B, Group data illustrating the effect of inhibition of DREADD Gi -expressing RTN neurons (administration of CNO, 2 mg kg −1 ; I.P.) and TeLC expression in preBötC astrocytes on CO 2 -induced increases in ƒ R , V T and V E in conscious rats.Each data point represents the measurement obtained in one animal.p-Values -unpaired t test.[Colour figure can be viewed at wileyonlinelibrary.com]

Figure 6 .
Figure 6.Contribution of the carotid body chemoreceptors to the hypercapnic respiratory response in anaesthetized rats A, Initial CO 2 -evoked (10%-12% CO 2 , 5 min) changes in the amplitude of phrenic nerve activity relative to baseline activity.B and C, Effects of carotid body denervation (CBD) on CO 2 -evoked increases in the amplitude of phrenic nerve activity in control and RTN-DREADD Gi animals (B) and combined in all animals from three experimental groups (C).Each data point represents the measurement obtained in one animal.p -unpaired t test (A) and paired t test (B and C).[Colour figure can be viewed at wileyonlinelibrary.com]

Figure 7 .
Figure 7. Contribution of the RTN (retrotrapezoid nucleus) neurons to the hypercapnic respiratory response in anaesthetized rats A, Summary data illustrating the effect of CNO (2 mg kg −1 , I.V.) on CO 2 -induced (10%-12% CO 2 , 5 min) increases in phrenic nerve activity in control, RTN-DREADD Gi and preBötC-TeLC/RTN-DREADD Gi experimental groups.B, Effects of CNO on CO 2 -evoked increases in the amplitude of phrenic nerve discharge compared to the initial responses (top) and responses evoked in conditions of CBD (carotid body denervation, bottom).Each data point represents the measurement obtained in one animal.p -unpaired t test (A) and paired t test (B).[Colour figure can be viewed at wileyonlinelibrary.com]

Figure 8 .
Figure 8. Contribution of preBötC (preBötzinger complex) astrocytes to the hypercapnic respiratory response in anaesthetized rats Group data illustrating the contribution of preBötC astrocytes to CO 2 -induced (10%-12% CO 2 , 5 min) increases in phrenic nerve activity in control conditions, after CBD (carotid body denervation) and then after the administration of CNO (2 mg kg −1 , I.V.) in RTN-DREADD Gi and preBötC-TeLC/RTN-DREADD Gi experimental groups.Each data point represents the measurement obtained in one animal.p -unpaired t test.[Colour figure can be viewed at wileyonlinelibrary.com]