KCNK3 channel is important for the ventilatory response to hypoxia in rats

To clarify the contribution of KCNK3/TASK-1 channel chemoreflex in response to hypoxia and hypercapnia, we used a unique Kcnk3 - deficient rat. We assessed ventilatory variables using plethysmography in Kcnk3 - deficient and wild-type rats at rest in response to hypoxia (10% O 2 ) and hypercapnia (4% CO 2 ). Immunostaining for C-Fos , a marker of neuronal activity, was performed to identify the regions of the respiratory neuronal network involved in the observed response.Under basal conditions, we observed increased minute ventilation in Kcnk3 - deficient rats, which was associated with increased c-Fos positive cells in the ventrolateral region of the medulla oblon-gata. Kcnk3 - deficient rats show an increase in ventilatory response to hypoxia without changes in response to hypercapnia. In Kcnk3 - deficient rats, linked to an increased hypoxia response, we observed a greater increase in c-Fos-positive cells in the first central relay of peripheral chemoreceptors and Raphe Obscurus . This study reports that KCNK3/TASK-1 deficiency in rats induces an inadequate peripheral chemoreflex, alternating respiratory rhythmogenesis, and hypoxic chemoreflex.


Introduction
In mammals, the regulation of breathing is crucial for maintaining correct blood oxygenation and eliminating carbon dioxide (CO 2 ).This regulation is controlled by the central and peripheral regions.The regulation of the membrane potential by ion channels contributes to breathing regulation via central and peripheral respiratory chemoreception.Among the ion channels, background K + channels play an essential role in maintaining respiratory control (Bayliss et al., 2015;Buckler, 2015).Background K + channels are mainly characterised by the two-pore domain potassium (K2P) channel family.K2Ps function as background K + conductance channels that help stabilise resting membrane potentials of different cell types (Enyedi and Czirják, 2010).Moreover, non-canonical sensing mechanisms have been observed in K2P channels with rapid gating kinetics (similar to many voltage-gated K + channels).These K2P properties suggest they contribute to rapid neuronal action potentials (Schewe et al., 2016).
K2P channels comprise two subunits, each with two pore domains that form a single pore highly selective for K + (Enyedi and Czirják, 2010).Among K2P, the potassium channel subfamily K + member 3 (KCNK3), also called TWIK-related acid-sensitive K + (TASK-1), and KCNK9 or TASK-3 are both described in mice as contributing to the regulation of breathing, using double-or simple-deficient mice for task-1 or task-3 (Buehler et al., 2017;Jungbauer et al., 2017).TASK-1 and TASK-3 have several characteristics with the background K + current, including minor voltage sensitivity, extracellular pH sensitivity, resistance to classic K + channel blockers, and nonsensitivity to cytoplasmic calcium concentrations (Le Ribeuz et al., 2020).
The regulation of peripheral breathing is primarily mediated by the carotid body (Rausch et al., 1991).The predominant channels in rodent carotid body glomus cells are KCNK3/TASK-1 and KCNK9/TASK-3.Electrophysiological and pharmacological approaches have highlighted their presence in the rat carotid body type-1 cells (Buckler et al., 2000;Buckler, 2007).Buckler et al. showed that the TASK channels play a central role in the response of arterial chemoreceptors to hypoxia, acidosis, and hypercapnia (Buckler et al., 2000).Later, using kcnk3/task-1 or kcnk9/task-3 knockout mice, the same group demonstrated that both KCNK3/TASK-1 and KCNK9/TASK-3 are inhibited by hypoxia, cyanide, or mitochondrial uncouplers, and that they are coupled to both oxygen and metabolic signalling pathways in these cells (Turner and Buckler, 2013).Moreover, kcnk3/task-1 knockout mice showed an almost 50% reduction in hypoxia and CO2 (by 68%)-induced increases in carotid sinus nerve chemoafferent discharge recorded (Trapp et al., 2008).In the carotid body, a hypoxia-induced decrease in TASK-1/TASK-3 function initiates plasma membrane depolarisation, resulting in voltage-gated calcium channel activation and the consecutive activation of neurotransmitters that excite the glossopharyngeal nerve.The afferent glossopharyngeal nerve projects to the pontomedullary respiratory centre through the nucleus of the solitary tract, which allows for adjusting ventilation (López-Barneo et al., 2008).Furthermore, the KCNK3/TASK-1 channel is functionally expressed at multiple levels in brainstem respiratory-related neurones such as airway motoneurons or presumptive chemoreceptor neurones (Bayliss et al., 2001(Bayliss et al., , 2015)).However, controversial results have been reported regarding the role of KCNK3/TASK-1 in controlling breathing in response to hypoxia and hypercapnia in mice.task1 -/-mice have a relatively normal CO 2 -chemoreflex, suggesting that these channels are not essential for central respiratory chemosensitivity (Mulkey et al., 2007).In contrast, Trapp and co-workers showed in task1 -/-and task1/task3 double knockout mice a 50% and 68% reduction in ventilatory responses to hypoxia and CO 2 , respectively, compared to wild-type mice.The ventilatory responses to hypoxia and CO 2 were unchanged in task3 -/-mice, suggesting that the TASK-3 channel is not required for central respiratory responses while TASK-1 is crucial for the ventilatory regulation in responses to hypoxia and hypercapnia (Trapp et al., 2008).As rats and mice have different adaptation mechanisms in breathing (Arias-Reyes et al., 2021).Moreover, KCNK3/TASK-1 is not functional in mouse pulmonary vasculature contrary to humans and rats (Manoury et al., 2011;Antigny et al., 2016;Kitagawa et al., 2017;Lambert et al., 2019).task1 -/-mice developed hyperaldosteronism (Heitzmann et al., 2008), whereas aldosterone production is unchanged in Kcnk3-deficient rats (Lambert et al., 2019).These examples highlight crucial differences between mice and rats concerning KCNK3/TASK-1, which could exist for ventilatory response to hypoxia or hypercapnia in rats.Moreover, gain-of-function mutations in the KCNK3 gene were recently identified in patients with sleep apnoea (Sörmann et al., 2022).Patients with sleep apnoea have altered breathing during sleep and develop pulmonary hypertension in 17-50% of patients with sleep apnoea (Bady et al., 2000;Yan et al., 2021), and KCNK3/TASK-1 dysfunction leads to pulmonary hypertension (Antigny et al., 2016;Lambert et al., 2019).
To clarify the role of KCNK3/TASK-1 in the control of breathing and to determine the role of KCNK3 in the regulation of breathing in rats, we used Kcnk3-deficient rats.This study employed a recent and unique Kcnk3 Δ94ex1/Δ94ex1 rat line with a deficiency in the KCNK3 protein (Lambert et al., 2019(Lambert et al., , 2021) ) and characterised the consequence of Kcnk3 deficiency in the control of breathing in rats.

Methods
Animals and surgical procedures -The animal facility was licenced by the French Ministry of Agriculture (agreement N • C92-019-01).This study was approved by the Committee on Ethics of Animal Experiments (CEEA26; CAP Sud).Animal experiments were approved by the French Ministry of Higher Education, Research, and Innovation (N • 7757).Animal experiments were performed in accordance with the guidelines from Directive 2010/63/EU on 22 September 2010 of the European Parliament on the protection of animals used for scientific purposes, and complied with the French institution's animal care and handling guidelines.

Kcnk3-deficient rats
Kcnk3-deficient rats were generated using the CRISPR/Cas 9 technique with a specific sgRNA-rKCNK3 and Cas9 mRNA (Remy et al., 2017;Lambert et al., 2019Lambert et al., , 2021) ) targeting the first exon of the Kcnk3 gene, to induce a shift in the reading frame of exon 1 of the Kcnk3 gene.We used a strain with 94 bp deleted in the first exon of Kcnk3 (Δ94ex1), as previously described (Lambert et al., 2019(Lambert et al., , 2021)).In one new-born rat, a deletion of 94 bp (Δ94ex1) was found, which resulted in an out-of-frame shift in the open reading frame, leading to a premature stop codon and generation of a completely different aa (amino acid) sequence.However, premature stop codons can cause mRNA degradation (Chang et al., 2007).The deletion of 94 bp in the mRNA was not associated with the absence of mRNA, indicating the absence of mRNA degradation (Chang et al., 2007).We studied homozygous rats (Kcnk3 Δ94ex1/Δ94ex1 , also called Kcnk3-deficient rats) and wild-type (WT) littermates.Only male rats were analysed in this study at three months old.
As we described (Lambert et al., 2019(Lambert et al., , 2021)), a putative translation of the truncated mRNA could produce a truncated 90-aa protein instead of the 411 aa of the wild-type (WT) protein and share only the first 14 aa with the WT protein.Sequencing of the Kcnk3 mRNA from Kcnk3 +/+ and Kcnk3 Δ94ex1/Δ94ex1 rats confirmed the deletion of the 94 bp and an aberrant protein sequence with eight potential premature stop codons (Lambert et al., 2019).
The founder animal with the Δ94ex1 deletion was crossed with a WT partner, and the deletion was transmitted to the offspring, as shown by genotypic DNA analysis, demonstrating that the rats were either Kcnk3 +/+ , heterozygous Kcnk3 Δ94ex1/+ , or homozygous Kcnk3 Δ94ex1/ Δ94ex1 .In this study, we used only Kcnk3 +/+ and homozygous Kcnk3 Δ94ex1/Δ94ex1 .
In vivo measurement of ventilatory variables -Ventilatory variables were measured non-invasively in unanaesthetised and unrestrained animals by adapting the technique developed by Drorbaugh and Fenn (Drorbaugh and Fenn, 1955) and modified by Bartlett and Tenney (Bartlett and Tenney, 1970).Briefly, a whole-body flow barometric plethysmograph was used to obtain breathing variables using a differential pressure transducer (model DP 45-18, Validyne Engineering, Northridge, CA, USA) placed between a recording chamber and a reference chamber of the same volume.The pressure signal was sent to a demodulator (model CD15, Validyne Engineering, Northridge, CA), and the breathing variables were recorded using a Spike 2 data analysis system (CED, Cambridge, UK).Rats were placed in the recording chamber ventilated with air at room temperature (21-22 • C).Thus, whole-body plethysmography provided a measurement of respiratory frequency (f R in cycle per min, c/min), mean total time of one breath (Ttot, in seconds), tidal volume (V T in µl), and minute ventilation (Ve in ml/min).Tidal volume and minute ventilation were normalised by the weight (V T , µl/g and Ve, ml/g/min).The irregularity score (IS) reflects the variability in the duration of respiratory cycles (Voituron et al., 2009).The delta of Ve was obtained by calculating the difference between the values obtained under normoxia, hypoxia, and hypercapnia at different times for each animal.The Ti/Ttot and V T /Ti ratios were also C.-H. Yegen et al. determined to indicate breathing time and an index of inspiratory drive (Milic-Emili and Grunstein, 1976).
To evaluate the hypoxic ventilatory response (HVR) and hypercapnic ventilatory response (HcVR), air was replaced with a hypoxic (10% O 2 ) or hypercapnic (4% CO 2 ) gas mixture for 10 min.Briefly, measurements were performed under normoxia (21% O 2 ) for 30 min, followed by hypoxia for 10 min (measures performed at 1'30, 2'30, 5, and 10 min), followed by a 30 min recovery period under normoxia, then hypercapnia for 10 min (measures performed at 1'30, 2'30, 5, and 10 min), followed again by a recovery period under normoxia (10 min) as described (Voituron et al., 2011).To analyse the response to hypoxia and hypercapnia, the ventilatory parameters were compared with the period of normoxia immediately preceding them.In this way, as the respiratory parameters are expressed as a function of the previous period of normoxia, the hypothetical influence of hypoxia was subtracted.
Analysis of c-Fos expression -The rib cage was opened after anaesthetising the rats with an intraperitoneal mixture of ketamine (100 mg/kg) and xylazine (10 mg/kg).Thus, rats were placed in a box ventilated with air (Normoxic condition) or hypoxic gas (O 2 10%balanced N 2 ) mixture for 2 hr.Transcardiac perfusion of 0.1 M phosphate-buffered saline (PBS) was performed for 10 min at a flow rate of 5 ml/min.The animals were then fixed by transcardial perfusion of 4% paraformaldehyde for 15 min.The whole brain of each animal was kept in 4% paraformaldehyde (PFA) for 48 h to optimise post-fixation.The 4% PFA was replaced by a 30% sucrose solution prepared in 0.1 M PBS at 4 • C. Sections of 40 µm thickness were made in the cryostat and then preserved in a cryoprotector based on sucrose and 0.2 M PBS to be stored at − 20 • C. Immunostaining was performed on one of every two sections, and the c-Fos protein was detected using standard procedures for Fos-Like Immunohistochemistry on free-floating sections (Perrin--Terrin et al., 2016).Endogenous peroxidases were neutralized with H 2 O 2 3%.Sections were then placed in 0.1 M PBS supplemented with 0, 3% Triton X-100% and 1% bovine serum albumin (BSA).Sections were then incubated for 48 h at 4 • C with a mouse monoclonal primary antibody directed against the c-Fos protein (sc-166940, Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) diluted 1/2000 in 0.1 M PBS supplemented with 0,3% Triton X-100 and 0,5% BSA.Sections were then incubated at room temperature for 2 h with a biotinylated horse anti-mouse IgG (BA 2000, Vector Laboratories) diluted 1/1000 in 0.1 M PBS supplemented with 0,3% Triton X-100 and 0,5% BSA and for 1 h with Avidin-Biotin-Peroxidase complex (VECTASTAIN, Elit PK-100 standard, ZE0622).Peroxidase activity was detected with 0.015% 3, 34-diaminobenzidine tetrahydrochloride, 0.04% nickel ammonium sulfate, and 0.006% H 2 O 2 in 0.05 M Tris buffer (pH 7,6).The reaction was stopped by rinsing with PBS once a brown colouration was visible.Sections were washed, mounted on silane-treated slides, air-dried, dehydrated with absolute alcohol, cleared with xylene, and covered with coverslips.
Control sections were treated in parallel without primary or secondary antibodies.No labelling was observed under these conditions.
We previously demonstrated that some neurones do not express c-Fos (Perrin-Terrin et al., 2016).However, this method helps determine the activation of neural pathways, such as respiratory rhythm adaptation, during chemical challenges (Dragunow and Faull, 1989;Perrin-Terrin et al., 2016).As c-Fos staining could also be present in glial cells, we called these cells C-Fos + cells in the present study.

Statistical analysis
All statistical tests were performed using the GraphPad Prism software (GraphPad, version 9.0 for Windows).All values are reported as mean ± standard error of the mean.All data were verified for normal distribution using the Shapiro-Wilk normality test.Differences between groups were analysed using an unpaired t-test for experiments with more than six samples.When the conditions of the parametric tests were not met, we used the Mann-Whitney test.Differences between two or more groups were assessed using a two-way ANOVA.Differences were considered statistically significant at p < 0.05.

Breathing at rest
We used whole-body plethysmography to record breathing patterns of freely moving WT and Kcnk3 Δ94ex1/Δ94ex1 rats under room air.At rest, minute ventilation was significantly increased in Kcnk3 Δ94ex1/Δ94ex1 rats due to increased V T (Table 1).Neither respiratory frequency nor variability of the cycle duration expressed by the irregularity score (IS), which reflects the variability in the duration of respiratory cycles, was significantly different between WT and Kcnk3 Δ94ex1/Δ94ex1 rats (Table 1).The Ti/Ttot ratio (indication of breathing time) was unchanged in Kcnk3 Δ94ex1/Δ94ex1 rats as compared to that in WT rats, while the V T /Ti ratio (index of inspiratory drive) was augmented (Table 1).

Consequences of Kcnk3 deficiency in rats on chemosensitivity to O 2 and CO 2
As described in the Materials and Methods section, in our experimental protocol, we successively exposed the same animal to normoxia and hypoxia, and after the recovery time, to hypercapnia and then normoxia.Before comparing the effects of hypercapnia exposure, we compared the plethysmography values 30 min after hypoxia exposure during the recovery period.As detailed in Table 2, we found similar differences between WT and Kcnk3 Δ94ex1/Δ94ex1 rats, as indicated by the increased VT, Ve, and Vt/Ti ratio (Table 2).
Concerning the response to hypercapnia, we found that, in WT and Kcnk3 Δ94ex1/Δ94ex1 rats, hypercapnia significantly increased V T and f R without any differences between both groups (Fig. 1).Therefore, the increase in Ve in response to CO 2 was similar between WT and Kcnk3 Δ94ex1/Δ94ex1 rats (Fig. 1).

Table 1
Breathing variables of WT and Kcnk3 Δ94ex1/Δ94ex1 rats in baseline conditions.
WT n = 10 Regarding the ventilatory response to hypoxia, the behaviour of the WT and Kcnk3 Δ94ex1/Δ94ex1 rats differed.Indeed, in WT rats, we observed a characteristic biphasic response with an increase in Ve (peaking 1 min after the onset of hypoxic exposure), followed by a relative decline (rolloff, Fig. 2).This response was due to the biphasic evolution of V T without a change in f R .However, in Kcnk3 Δ94ex1/Δ94ex1 rats, the V T remained elevated throughout the exposure to hypoxia (Fig. 2), leading to a sustained response to hypoxia with an absent or delayed roll-off.

Anatomical substrate responsible for the observed ventilatory response to hypoxia
We sought to identify the ponto-bulbar structures of the respiratory network involved in the observed differences between the mutants and WT rats.Therefore, we performed immunohistochemical detection of the c-Fos protein (black arrow) (Table 3, Fig. 3), which could be considered a marker of cellular activity.
Under normoxic conditions, basal c-Fos expression was observed in the studied structures (Table 3, Fig. 3).Although there was no difference in most structures, we observed an increase of the c-Fos positive cells in baseline conditions between WT and Kcnk3 Δ94ex1/Δ94ex1 rats in the levels of rVLM, cVLM, and RPa (Table 3, Fig. 3).
In WT rats, hypoxic exposure led to an increase in c-Fos expression in cNTS, rVLM, cVLM, RPa, and ROb, while in Kcnk3 Δ94ex1/Δ94ex1 rats, hypoxia led to an increase in the number of c-Fos positive cells in c and mNTS, as well as in cVLM and ROb (Table 3, Fig. 3).

Discussion
In this study, we report several essential findings related to the effects of Kcnk3 deficiency on breathing control.I) Under basal conditions, Ve increases in Kcnk3 Δ94ex1/Δ94ex1 rats.II) Kcnk3 Δ94ex1/Δ94ex1 rats developed a potentiated ventilatory response to hypoxia, while the ventilatory response to hypercapnia was unaltered.III) The increased response to hypoxia in Kcnk3 Δ94ex1/Δ94ex1 rats is associated with a higher number of c-Fos positive cells in cNTS, cVLM, and ROb.

Role of KCNK3/TASK-1 in breathing control
In a mouse model, the ventilation at rest was similar between WT and task1 -/-mice (Trapp et al., 2008).In task1 − /− mice, only modified respiration was observed in male mice (Jungbauer et al., 2017).In task3 − /− mice, however, discrete changes in respiration were observed in both sexes.Double task1/task3 knockout (KO) mice exhibit increased basal ventilation caused by an enlarged tidal volume and peak expiratory and inspiratory flow (Buehler et al., 2017).This phenotype was similar to that of task1 -/-mice (Jungbauer et al., 2017).
In contrast, we found that Kcnk3 Δ94ex1/Δ94ex1 rats had a higher Ve than WT rats.We only observe a difference in V T but not in terms of f R , which would indicate that this increase in Ve was due to a change in the gain of the motor output.This is supported by the increased V T /Ti ratio, which suggests a strengthening of ventilatory control.In parallel, we observed a difference in c-Fos expression in the cVLM.The VLM is a neuronal column ventral to the ambiguus nucleus that extends from the pyramidal decussation to the caudal edge of the facial nucleus (Voituron et al., 2006;Huckstepp et al., 2015;Baum et al., 2018).The ventrolateral medulla contains the ventral respiratory group (VRG) (Bianchi 1971).An increase in phrenic activity amplitude was correlated with an increase in caudal VRG inspiratory neuron activity (Morris et al., 1996), suggesting that strengthening of the ventilatory drive could be linked to the increase in activity observed in the cVLM.
In contrast, we observed an increase in c-Fos activity in the rVLM in the control condition.Catecholaminergic neurons (C1) in the rVLM modulate sympathetic outflow.Indeed, rVLM contains sympathoexcitatory neurones (Guyenet, 2006), which could be responsible for the increased heart rate and right ventricular systolic pressure recently observed in Kcnk3 Δ94ex1/Δ94ex1 rats (Lambert et al., 2019).

Normal central chemoreflex
The KCNK3/TASK-1 channel and other K2P channels participate in central CO 2 /pH chemoreception because of their localisation in various chemosensitive central regions, including the ventrolateral medulla, medullary dorsal, caudal raphe, pontine locus coeruleus, and ventral respiratory group, and because they are sensitive to the external pH and oxygen (Bayliss et al., 2001(Bayliss et al., , 2015;;Washburn et al., 2003).In a rat model using the non-selective KCNK3/TASK-1 blocker anandamide, Li Q and colleagues suggested that the KCNK3/TASK-1 channel may act as a chemosensor for central respiration, contributing to pH-sensitive respiratory effects (Li et al., 2020).
However, the role of KCNK3/TASK-1 in the respiratory response to hypercapnia remains controversial.
The KCNK3/TASK-1 channel is not involved in central CO 2 chemosensing.Moreover, task1 -/-, task3 -/ , -or double task1/task3 knockout mice have a relatively normal CO 2 -chemoreflex, suggesting that these channels are not essential for central respiratory chemosensitivity (Mulkey et al., 2007).Mulkey et al. showed that the pH sensitivity of 5HT raphe neurones was abolished in TASK channel-null mice, indicating that 5HT raphe neurones are not involved in CO2 the chemoreflex (Mulkey et al., 2007).In contrast, Trapp et al. measured altered ventilation in task1 -/-and task1/task3 double knockout mice (Trapp et al., 2008).Similarly, to the work of Mulkey et al., we found that exposure to hypercapnia led to a profound increase in ventilation in all animal groups.Furthermore, ventilatory response to hypercapnia was preserved in Kcnk3 Δ94ex1/Δ94ex1 rats.Similar to that in mice, we demonstrated that KCNK3/TASK-1 was no important in the response to hypercapnia in rats.
However, our results in rats differed from those obtained in task1 -/- mice, which displayed a blunted respiratory response to hypoxia (Trapp et al., 2008).These differences could be explained by the fact that rats and mice have different adaptation mechanisms (Arias-Reyes et al., 2021), without any doubt regarding our experimental methodology as we and others have already used similar experimental procedures (Voituron et al., 2011;Lucking et al., 2018).In contrast to humans and rats, KCNK3/TASK-1 channels are not functional in mouse pulmonary vasculature (Manoury et al., 2011;Antigny et al., 2016;Kitagawa et al., 2017;Lambert et al., 2019).Moreover, task1 -/-mice are characterised by hyperaldosteronism (Heitzmann et al., 2008), whereas aldosterone production remains unchanged in Kcnk3-deficient rats (Lambert et al., 2019), confirming the crucial differences related to KCNK3/TASK-1 in mice and rats.Moreover, sensitivity to CO 2 differs among rat strains, highlighting the fact that physiological mechanisms, including CO 2 sensitivity, are genetically regulated (Hodges et al., 2002).Values are means ± SEM of the respiratory frequency (fR, c/min), tidal volume (VT; µl/g), minute ventilation (Ve, ml/g/min), mean total time of one breath (Ttot, sec), breathing time (Ti/Ttot), index of the inspiratory drive (Vt/Ti) and the, Irregularity score (IS).Data were analysed using an unpaired t-test.* indicates p < 0.05; ** indicates p < 0.01.

Impaired peripheral chemoreflex in Kcnk3-deficient rats
In our WT rats, hypoxia induced the classic biphasic ventilatory response, characterised by an increase in ventilation up to a hyperventilation peak, followed by a relative "roll-off" ventilatory depression, where ventilation decreases from the hyperventilation peak but remains above pre-hypoxic values (Maxová and Vízek, 2001).In Kcnk3 Δ94ex1/Δ94ex1 rats, the first hyperventilation was similar to WT rats.However, roll-off was not observed.Similarly, Trapp et al. demonstrated that after 3 min under hypoxia, the hypoxia-induced increase in ventilation in mice was reduced in task1 -/-mice (Trapp et al., 2008).We found that WT rats had increased cNTS while Kcnk3-deficiency increase cNTS and mNTS.The cNTS and mNTS are significant projection sites from peripheral chemoreceptors (Torrealba and Claps, 1988;Finley and Katz, 1992), suggesting that in addition to carotid bodies, the cNTS and mNTS contribute to breathing regulation in Kcnk3-deficient rats.Indeed, carotid body function is compromised in task1 -/ -mice (Trapp et al., 2008).In addition, Buckler et al. demonstrated that KCNK3/TASK1 channels may be involved in regulating peripheral chemoreceptor breathing via hypoxic inhibition of KCNK3/TASK-1 channels in carotid body glomus cells (Buckler, 2007(Buckler, , 2015)).
As previously mentioned, KCNK3/TASK-1 is expressed in some brainstem respiratory neurones (Bayliss et al., 2001(Bayliss et al., , 2015)).In Kcnk3 Δ94ex1/Δ94ex1 rats (Sprague Dawley strain), we found a more marked increase of c-Fos expression in the level of the NTS.At the same time, in this rat strain, orexin neurones sense extracellular pH changes via TASK channels and participate in ventilatory regulation via projections on the NTS (Wang et al., 2021).Further experiments in Kcnk3 Δ94ex1/Δ94ex1 rats are needed to understand the consequence of Kcnk3 deficiency entirely.
In WT rats, we observed an increase in the number of c-Fos positive cells in RPa and ROb, as previously observed in response to tissue hypoxia (Bodineau and Larnicol, 2001).In Kcnk3 Δ94ex1/Δ94ex1 rats, ROb was more activated under hypoxic conditions than in WT rats.The RPa and ROb contain neurones that project to phrenic motoneurons (Loewy and McKellar, 1981;Holtman et al., 1990;Sasek et al., 1990) and structures that generate and modulate the CCR, such as preBötC (Ptak et al., 2009).Therefore, we cannot exclude the possibility that this activation of ROb could be implicated in the observed ventilatory response in Kcnk3 Δ94ex1/Δ94ex1 rats.

Possible consequences of inadequate hypoxic chemoreflex
The fact that the KCNK3/TASK-1 channel contributes to breathing regulation could also have consequences in the development of sleep apnoea, a disorder that affects more than 1 billion people (Benjafield et al., 2019).Interestingly and linked to our present study, Sörmann and colleagues recently identified nine de novo gain-of-function mutations in KCNK3 that lead to global developmental delay, hypotonia, a range of structural malformations, and sleep apnea (Sörmann et al., 2022).Until this discovery, and due to its expression in the hypoglossal motor nucleus, the link between KCNK3/TASK-1 and sleep apnoea had already been hypothesised (Gurges et al., 2021) but not demonstrated.Furthermore, patients with sleep apnoea experience abnormal, interrupted breathing during sleep, and 17-50% of patients with sleep apnoea develop pulmonary hypertension (Bady et al., 2000;Yan et al., 2021).Although the link between pulmonary hypertension and sleep apnoea remains controversial, nocturnal oxygen drops damage nitric oxide synthesis and cause vascular remodelling, possibly leading to pulmonary hypertension (Dempsey et al., 2010).Moreover, heterozygous loss-of-function mutations in KCNK3 have been identified in patients with pulmonary arterial hypertension (Ma et al., 2013;Higasa et al., 2017;Olschewski et al., 2017;Le Ribeuz et al., 2020), and we found that Kcnk3-deficient rats were predisposed to the development of pulmonary hypertension.Indeed Kcnk3 Δ94ex1/Δ94ex1 rats developed more severe pulmonary hypertension than WT rats after chronic hypoxia exposure (3 weeks at 10% of O 2 ) (Lambert et al., 2019).Here we found that in Kcnk3 Δ94ex1/Δ94ex1 rats, unlike in WT-rats, an absence of the roll-off respiratory response to hypoxia exposure indicates that Kcnk3 deficiency impaired response to hypoxia which could contribute to increased susceptibility to the development of lung diseases, including pulmonary hypertension (Lambert et al., 2019(Lambert et al., , 2021) ) and possibly sleep apnoea.The increased response to hypoxia in Kcnk3-deficient rats could predispose animals to the activation of hypoxia-inducible factor 1 (HIF-1α).HIF-1α is a crucial factor of pulmonary vascular remodelling occurring in PAH (Ryan and Archer, 2015;Lei et al., 2016), and we found an overactivation in HIF-1α in the lung from Kcnk3-deficient rats compared to that in WT, which predispose Kcnk3-deficient rats to severe PH under chronic-hypoxia exposure (Lambert et al., 2019).
In patients with PAH, a more pronounced increase in minute ventilation than in healthy subjects has been reported during acute hypoxia.This phenomenon is also observed during exercise-induced hypoxia in patients with PAH, which could ameliorate with ventilation-perfusion ratio abnormalities induced by pulmonary vascular disease (Groth et al., 2018).However, further experiments are required to clarify this point.

Conclusions
Using unique Kcnk3 Δ94ex1/Δ94ex1 rats, we reported that the KCNK3/ TASK-1 deficiency alters the effectiveness of structures involved in respiratory rhythmogenesis and the hypoxic chemoreflex arc, probably independently of carotid body structures.These alterations could have profound consequence for patients lacking functional KCNK3/TASK-1, including sleep apnoea and pulmonary hypertension.Nevertheless, further studies are needed to fully elucidate the contribution of KCNK3/ TASK-1 to breathing control and sleep apnoea.

Limitations
All commercially available anti-KCNK3 antibodies tested were unusable because when tested in kcnk3-knockout mice or in our Kcnk3deficient rats, they were either nonreactive or yielded the same signal as that in WT mice or WT rats (Schmidt et al., 2015;Murtaza et al., 2017;Lambert et al., 2019).This prevented any measurement of KCNK3 protein expression in Kcnk3-deficient rats (Lambert et al., 2019).Thus, we could not confirm the localisation of the KCNK3 protein in Kcnk3-deficient rats.
Moreover, we found that the ventilatory responses to WT and Kcnk3deficient rats 30 min after hypoxia protocol differ between basal normoxia conditions.To avoid misinterpretation regarding the role of KCNK3/TASK-1 channel in the ventilatory response to hypercapnia, further experiments should be done in different animals to ensure that KCNK3/TASK-1 is not involved in the hypercapnia response.

Fig. 1 .
Fig. 1.KCNK3/TASK-1 dysfunction did not modify the central chemoreflex.(A) Plethysmographic traces from the same WT and Kcnk3 Δ94ex1/Δ94ex1 rats when breathing air (normocapnia) or hypercapnic (Hc) gas mixture at 10 min (B) Evolution of the delta of Ve as a function of the time of exposure to hypercapnia in WT and Kcnk3 Δ94ex1/Δ94ex1 rats.(C-E) Histograms showing irregularity score (C), V T (D), and f R (E) change in WT (grey columns) and Kcnk3 Δ94ex1/Δ94ex1 (blue columns) rats.ns: not significant.n = 10 for WT and n = 9 for Kcnk3 Δ94ex1/Δ94ex1 .The black line and blue line correspond to basal values for WT and Kcnk3 Δ94ex1/Δ94ex1 , respectively.Experiments were analysed with 2-way ANOVA completed by Tukey post hoc test for multiple comparisons.ns indicates nonsignificant, *p < 0.05, **p < 0.01.

Fig. 2 .
Fig. 2. KCNK3/TASK-1 dysfunction alters the peripheral chemoreflex.(A) Plethysmographic traces from the same WT and Kcnk3 Δ94ex1/Δ94ex1 rats when breathing air (normoxia) or hypoxic (Hx) gas mixture at 5 min (B) Evolution of the delta of Ve as a function of the time of exposure to hypoxia in WT and Kcnk3 Δ94ex1/Δ94ex1 rats.(C-E) Histograms showing irregularity score (C), V T (D), and f R (E) change in WT (grey columns) and Kcnk3 Δ94ex1/Δ94ex1 (bleu columns) rats.n = 10 for WT and n = 9 for Kcnk3 Δ94ex1/Δ94ex1 .The black line and blue line correspond to basal values for WT and Kcnk3 Δ94ex1/Δ94ex1 , respectively.Experiments were analysed with 2-way ANOVA completed by Tukey post hoc test for multiple comparisons.ns indicates nonsignificant, *p < 0.05, **p < 0.01.

FundingF.
Antigny receives funding from the Fondation du Souffle et Fonds de Dotation Recherche en Santé Respiratoire from the Fondation Lefoulon-Delalande and the Fondation Legs Poix.The authors received funding from the French National Agency for Research (ANR) (grant no.ANR-18-CE14-0023).NV received funding from the French Foundation legs Poix, BQR, and IFRB programs of the university Sorbonne Paris Nord.Kcnk3-deficient rats were generated with financial support from Fondation Maladies Rares in the framework of small-animal models, and the rare disease program generated Kcnk3-deficient rats.M. Lambert was supported by the Therapeutic Innovation Doctoral School (ED569).

Table 2
Breathing variables of WT and Kcnk3 Δ94ex1/Δ94ex1 rats in normoxia condition measured 30 min after hypoxia exposure.