Impaired central respiratory chemoreflex in an experimental genetic model of epilepsy

Abstract

Key points

• It is recognized that seizures commonly cause apnoea and oxygen desaturation, but there is still a lack in the literature about the respiratory impairments observed ictally and in the post‐ictal period.

• Respiratory disorders may involve changes in serotonergic transmission at the level of the retrotrapezoid nucleus (RTN).

• In this study, we evaluated breathing activity and the role of serotonergic transmission in the RTN with a rat model of tonic–clonic seizures, the Wistar audiogenic rat (WAR).

• We conclude that the respiratory impairment in the WAR could be correlated to an overall decrease in the number of neurons located in the respiratory column.

Abstract

Respiratory disorders may involve changes in serotonergic neurotransmission at the level of the chemosensitive neurons located in the retrotrapezoid nucleus (RTN). Here, we investigated the central respiratory chemoreflex and the role of serotonergic neurotransmission in the RTN with a rat model of tonic–clonic seizures, the Wistar audiogenic rat (WAR). We found that naive or kindled WARs have reduced resting ventilation and ventilatory response to hypercapnia (7% CO₂). The number of chemically coded (Phox2b⁺/TH⁻) RTN neurons, as well as the serotonergic innervation to the RTN, was reduced in WARs. We detected that the ventilatory response to serotonin (1 mm, 50 nl) within the RTN region was significantly reduced in WARs. Our results uniquely demonstrated a respiratory impairment in a genetic model of tonic–clonic seizures, the WAR strain. More importantly, we demonstrated an overall decrease in the number of neurons located in the ventral respiratory column (VRC), as well as a reduction in serotonergic neurons in the midline medulla. This is an important step forward to demonstrate marked changes in neuronal activity and breathing impairment in the WAR strain, a genetic model of epilepsy.

Key points

• It is recognized that seizures commonly cause apnoea and oxygen desaturation, but there is still a lack in the literature about the respiratory impairments observed ictally and in the post‐ictal period.

• Respiratory disorders may involve changes in serotonergic transmission at the level of the retrotrapezoid nucleus (RTN).

• In this study, we evaluated breathing activity and the role of serotonergic transmission in the RTN with a rat model of tonic–clonic seizures, the Wistar audiogenic rat (WAR).

• We conclude that the respiratory impairment in the WAR could be correlated to an overall decrease in the number of neurons located in the respiratory column.

Abbreviations

ANOVA

analysis of variance

AP

anterioposterior (brain co‐ordinate)

BötC

Bötzinger complex

CNS

central nervous system

cVRG

caudal ventral respiratory group

cSI

categorized severity index

f_R

respiratory rate

HCO₃

bicarbonate

5‐HT

5‐hydroxytryptamine (serotonin)

i.d.

internal diameter

i.p.

intraperitoneal

i.v.

intravenous

LC

locus coeruleus

NTS

nucleus of the solitary tract

NK1R

neurokinin‐1 receptor

o.d.

outer diameter

$P_{\mathrm{a},\mathrm{C}{\mathrm{O}}_2}$

partial pressure of arterial carbon dioxide

$P_{\mathrm{a},{\mathrm{O}}_2}$

partial pressure of arterial oxygen

PBS

phosphate‐buffered saline

Phox2B

paired‐like homeobox 2b

preBotC

pre‐Bötzinger complex

RTN

retrotrapezoid nucleus

rVRG

rostral ventral respiratory group

SEM

standard error of the mean

SUDEP

sudden unexpected death

TH

tyrosine hydroxylase

${\dot{V}}_{\mathrm{E}}$

ventilation

VRC

ventral respiratory column

V_T

tidal volume

WAR

Wistar audiogenic rat

Introduction

Sudden unexpected death (SUDEP) is the most common cause of mortality in patients with refractory epilepsy, and although the causes and potential mechanisms are still unknown (Massey et al. 2014), it is well accepted that SUDEP is frequently linked to cardiac arrest. Furthermore, while it is recognized that seizures commonly cause apnoea and oxygen desaturation (Bateman et al. 2008), there is still a lack of explanation about the respiratory problems observed ictally and in the post‐ictal period. The present study used the Wistar audiogenic rat (WAR) strain, which is a genetically selected reflex seizure model. In this model, acute high‐intensity acoustic stimulation induces tonic–clonic seizures and chronic acoustic stimulation induces limbic seizures (Garcia‐Cairasco et al. 1996; Dutra Moraes et al. 2000; see details below).

Neurons that produce serotonin (5‐HT) are restricted to a small number of nuclei in the brainstem (Jacobs et al. 2002). They project to virtually every region of the central nervous system (CNS), where they influence a broad spectrum of brain functions, including breathing (Ptak et al. 2009). The retrotrapezoid nucleus (RTN) is an important region of the brain that receives inputs from the 5‐HT neurons (Mulkey et al. 2007). The RTN is an integrative site that primarily regulates ventilation to maintain CO₂ homeostasis (henceforth called central respiratory chemoreflex; Mulkey et al. 2004; Takakura et al. 2006; Guyenet & Bayliss, 2015). Disruption of serotonergic input to respiratory centres such as the RTN is thought to contribute to respiratory impairments associated with multiple neurological disorders, including epilepsy (Buchanan & Richerson, 2010).

Although most of the documented respiratory deficits are seen as an immediate consequence of seizures, we hypothesize that ventilation and chemoreflexes are reduced in a genetic model of epilepsy (Doretto et al. 2003), and that these effects will be altered in chronically stimulated (kindled) rats. In the present study, our main goal was to describe and characterize anatomical and physiological brainstem chemosensory mechanisms in naive and kindled animals of the WAR strain.

Methods

Animals

Experiments were performed in 22 adult male Wistar rats (12 naive and 10 sound exposed chronically) and 25 male WARs (7 naive and 18 sound exposed chronically (kindled)) weighing 310–420 g. WARs are obtained from the breeding stock of the Department of Physiology of the Ribeirão Preto School of Medicine, University of São Paulo, where this strain has been developed and continuously maintained. Animals were used in accordance with the guidelines approved by the Animal Experimentation Ethics Committee of the School of Medicine of Ribeirão Preto (protocol number 080/2014) and the Institute of Biomedical Science at the University of São Paulo (protocol number: 108/2014).

WAR model

The kindling protocol (auditory chronic stimulation) was the same as that used by Dutra Moraes and colleagues (2000). In this protocol, Wistar (N = 10) and WARs (N = 18) were placed in an acrylic, acoustically isolated, chamber (height: 32 cm, diameter: 30 cm), located inside a sound proof chamber (45 × 45 × 40 cm) where, after 1 min of acclimation, they were submitted to a 120 dB broadband noise stimulus with a maximum of 1 min duration. After stimulation, the animals were kept in the cage for one more minute, and returned to their home cages. We repeated this procedure 20 times, twice a day (in the morning and in the late afternoon), for 10 days. During the sessions, the animal behaviour was recorded for posterior behavioural analysis. If the animals seized, the seizure were scored using a mesencephalic seizure severity index (Fig. 1), modified from Garcia‐Cairasco et al. (1996) as a categorized severity index (cSI) by Rossetti et al. (2006) or Racine´s seizure severity index (Racine et al. 1972), for limbic seizures (LI) occurring in kindled animals (Fig. 1). Sound stimulation was interrupted as soon as we observed generalized tonic seizures of score cSI = 4 (Fig. 1). Animals were observed for one additional minute. The animals were considered kindled (N = 13 out of 18) when at least three limbic seizures were observed after the sound stimulus. A control group (N = 10) consisted of Wistar rats that were manipulated equally as the rats submitted to the repeated acoustic stimuli protocol. Although a small fraction of Wistar rats responds with seizures to acoustic stimulation, only Wistar rats without any response were considered as controls.

Figure 1. Evolution of audiogenic kindling in WARs.

Time course of audiogenic kindling progression in WARs, using the mesencephalic seizure severity index (cSI) and limbic seizure severity index (LI). Animals were exposed to 20 high‐intensity acoustic stimuli (twice per day) for 10 days. Observe the significant decrease (one‐way ANOVA RM P = 0.0051, F = 2.096), of the mesencephalic cSI, contrasting with the concomitant significant increase (one‐way ANOVA RM, P < 0.0001; F = 5.943), of the LI.

Surgery and anaesthesia

Rats were anaesthetized with an intraperitoneal injection of ketamine (80 mg (kg body weight)⁻¹) combined with xylazine (7 mg (kg body weight)⁻¹) and placed in a stereotaxic frame (model 900; David Kopf Instruments, Tujunga, CA, USA). Unilateral stainless steel cannulas (0.6 mm o.d.) were implanted into the RTN using the coordinates: 2.5 mm caudal to lambda, 1.8 mm lateral to the midline and 7.5 mm below dura matter. The cannulas were fixed to the cranium using dental acrylic resin and jeweler screws. Rats received a prophylactic dose of penicillin (30,000 IU) given intramuscularly and a subcutaneous injection of the analgesic Ketoflex (ketoprofen 1%, 0.03 ml rat⁻¹) post‐surgically. After the surgery, the rats were maintained in individual boxes with free access to tap water and food pellets (Guabi rat chow; Paulínia, SP, Brazil).

Measurement of lung ventilation

Respiratory rate (f _R; breaths min⁻¹) and tidal volume (V _T; ml kg⁻¹) in conscious, freely moving rats were measured by whole‐body plethysmography as described in detail previously (Favero et al. 2011; Takakura et al. 2014). All experiments were performed at room temperature (24–26°C). Freely moving rats were kept in a Plexiglas recording chamber (5 l) that was flushed continuously with a mixture of 79% nitrogen (N₂) and 21% oxygen (O₂) (unless otherwise required by the protocol) at a rate of 1–1.3 l min⁻¹. Volume calibration was performed during each ventilation measurement throughout the course of the experiments by injecting a known volume of air (1 ml) inside the chamber. V _T was calculated using the formula described by Malan (1973) and used in previous studies (Biancardi et al. 2008; Favero et al. 2011; Takakura et al. 2014). Ventilation (${\dot{V}}_{\mathrm{E}}$) was calculated as the product of ${\dot{V}}_{\mathrm{T}}$ and f _R and is presented at ambient barometric pressure and at body temperature, which it is saturated with water vapour. Concentrations of O₂ and carbon dioxide (CO₂) in the chamber were monitored online using a fast‐response O₂/CO₂ monitor (ADInstruments). The pressure signal was amplified, filtered, recorded and analysed off‐line using Powerlab software (Powerlab 16/30, ML880/P; ADInstruments). Rectal temperature was measured before and at the end of the experiments and the values were averaged. Measurements of f _R and V _T were taken when breathing stabilized in the last 2 min before exposure to the stimulus and during the 2‐min period at the end of each stimulus. Changes in the f _R, V _T and ${\dot{V}}_{\mathrm{E}}$ (${\dot{V}}_{\mathrm{E}}$ = f _R × V _T; in ml min⁻¹ kg⁻¹) were averaged and expressed as means ± SEM.

Chemoreflex analysis

Unanaesthetized WARs (N = 7 naive and N = 13 kindling) and Wistar rats (N = 12 naive and N = 10 acoustic stimulation) were allowed at least 30 min to acclimate to the chamber environment at normoxia and normocapnia (21% O₂, 79% N₂ and < 0.5% CO₂) before ventilation was measured. For the physiological experiments, hypercapnia was induced by titrating CO₂ into the respiratory mixture up to 7% (50% O₂, balanced with N₂) for 10 min. On the other hand, Fos‐like immunoreactivity evoked by hypercapnia was studied in conscious, unrestrained adult rats. Before the experiments, rats were placed in a Plexiglass chamber (5 l) that was continuously flushed with a mixture of 79% N₂ and 21% O₂ at a rate of 1 l min⁻¹, to allow them to become acclimated to the environmental stimuli associated with the chamber and to minimize unspecific Fos expression. Wistar rats or WARs were first acclimated for 45 min in the chamber and after exposed to hypercapnia 7% CO₂–50% O₂, balanced with N₂ or normoxic control 21% O₂ for 3 h, as previously demonstrated (Berquin et al. 2000; King et al. 2013). At the end of the stimulus, the rats were immediately deeply anaesthetized with pentobarbital (60 mg (kg body weight)⁻¹) and transcardially perfused. All experiments were performed at room temperature (24–26°C).

Blood gases and pH analysis

Arterial pH, partial pressure of arterial carbon dioxide ($P_{\mathrm{a},\mathrm{C}{\mathrm{O}}_2}$), partial pressure of arterial oxygen ($P_{\mathrm{a},{\mathrm{O}}_2}$) and plasma bicarbonate (HCO₃ ⁻) were measured using a portable blood gas analyser (I‐Stat, Abbott Laboratory, NJ, USA). The analysis required 150 μl of blood samples, which was dripped from a polyethylene tubing (PE‐10 connected to PE‐50; Clay Adams, Parsippany, NJ, USA) previously implanted into the abdominal aorta through the femoral artery under i.p. ketamine combined with xylazine anaesthesia 20 h prior to experimentation (Barna et al. 2014).

Intraparenchymal injections

Serotonin creatinine sulfate (Sigma Chemicals Co., St Louis, MO, USA; 1 mm in sterile saline pH 7.4) was injected (50 nl) unilaterally using 5 μl Hamilton syringes connected by polyethylene tubing (PE‐10) to the injection needle 1.5 mm longer than the guide cannulas implanted into the brain. The concentration of serotonin used in these experiments was based on the EC₅₀ response elicited by 5‐HT injections into the hypoglossal nucleus or into the RTN of anaesthetized rats (Fenik & Veasey, 2003; Hawryluk et al. 2012). The needle tip was placed 200 μm below the caudal edge of the facial motor nucleus. Postmortem histological inspection of the location of dye verified the correct placement of the injections in all cases.

Histology

Brain tissue was obtained from 18 WARs (N = 7 naive and N = 11 kindled) and 12 Wistar rats (N = 5 naive and N = 7 sound exposed) that had been deeply anaesthetized with pentobarbital (60 mg (kg body weight)⁻¹) and perfused transcardially with 250 ml of PBS followed by 500 ml of 4% paraformaldehyde in 100 mm phosphate buffer. According to our previous publications (Barna et al. 2014; Takakura et al. 2014), the brains were removed from the skull, cryoprotected by overnight immersion in a 20% sucrose solution in phosphate‐buffered saline at 4°C, sectioned in the coronal plane at 40 μm on a sliding microtome and stored in cryoprotectant solution (20% glycerol plus 30% ethylene glycol in 50 mm phosphate buffer, pH 7.4) at −20°C for up to 2 weeks while awaiting histological processing. All histochemical procedures were done using free‐floating sections according to previously described protocols (Barna et al. 2012, 2014).

Sections were incubated free‐floating in primary antibodies in 0.1 m Tris‐buffered saline, pH 7.4, with 10% horse serum and 0.1% Triton X‐100 for 24 h at 4°C, unless otherwise noted. Serotonin terminals were identified with an antibody directed against serotonin, which gave a more robust labelling of terminals than antibodies directed against tryptophan hydroxylase or the serotonin transporter. Serotonin was detected with a rat monoclonal antibody (1:100; 48–72 h incubation in 10% horse serum/0.3% Triton; Millipore, Billerica, MA, USA; MAB352, clone YC5/45) followed by Cy3‐tagged donkey anti‐rat IgG (1:200; Jackson Immuno Research). Serotonergic cell bodies were identified with a mouse monoclonal antibody against tryptophan hydroxylase (TrpH; 1:1000; Sigma; T0678, clone WH‐3) followed by a Cy3‐tagged goat‐anti‐mouse IgG (1:200; Invitrogen). Phox2b immunoreactivity was detected in combination with substance P receptor (neurokinin‐1 receptor (NK1R)) immunoreactivity using two rabbit antibodies in a sequential method as described previously (Rosin et al. 1993). First, Phox2b was identified in cell nuclei using a rabbit antibody (1:800; generously provided by J. F. Brunet, École Normale Supérieure, Paris, France (Pattyn et al. 1997); 24 h incubation in buffer with 0.5% TNB (a casein‐blocking reagent, Tyramide Signal Amplification kit, Perkin‐Elmer, Boston, MA, USA) and 0.3% Triton X‐100) followed by biotinylated donkey anti‐rabbit (1:500; Jackson ImmunoResearch) and avidin–biotin reagents (Vectastain Elite Kit, Vector Laboratories, Burlingame, CA, USA) and detection with nickel DAB (nickel ammonium sulfate/3,3′‐diaminobenzidine) to produce a blue‐black product. Second, NK1R was detected with a rabbit antibody (1:5000, Millipore, AB5060; incubation for 48 h in buffer with 0.5% TNB and 0.3% Triton X‐100) followed by biotinylated donkey anti‐rabbit (1:500; Jackson ImmunoResearch) and avidin–biotin reagents but reacting optimally with DAB to produce a light brown product.

Tyrosine hydroxylase (TH) was detected using a mouse antibody anti‐TH (MAB 318; Millipore; dilution 1:1000) and Fos was detected using rabbit antibody to Fos (Ab‐5; Calbiochem; dilution 1:10000). Sections were incubated for 24 h at room temperature and all diluted in PBS containing 10% normal horse serum (008‐000‐001; Jackson ImmunoResearch Laboratories) and 0.3% Triton X‐100. After several rinses, they were transferred to an appropriate affinity purified biotinylated secondary antibodies goat anti‐rabbit (BA‐1000, Vector, dilution 1:500) for Fos and donkey anti‐mouse (715‐065‐151, Jackson ImmunoResearch Laboratories, dilution 1:500) for TH, all diluted in PBS containing 1% normal horse serum and 0.3% Triton X‐100 incubated for 24 h at room temperature, rinsed again and exposed to Extravidin (E2886; Sigma‐Aldrich; dilution 1:2000) for 4 h at room temperature. Peroxidase reactions were visualized using the glucose oxidase procedure and DAB tetrahydrochloride as chromogen for TH and associated with 0.5% nickel sulfate for Fos immunostaining. Sections were rinsed again in PBS, mounted in sequential rostrocaudal order onto gelatin‐coated slides dehydrated through a series of ascending concentrations of ethanol, transferred into xylene, and coverslipped with DPX (06522; Sigma Aldrich) mountant for histology.

The antibodies against Fos, TH, serotonin, tryptophan hydroxylase, substance P (NK1), and Phox2b have been used previously and their specificity has been established in rat brain in previous studies in our laboratory (Barna et al. 2012, 2014; Takakura et al. 2014).

Cell mapping, cell counting and imaging

Cell mapping, counting and photography were done using the ImageJ software (public domain program available from the NIH; http://rsb.info.nih.gov/ij/) with a Zeiss Axioskop 2 microscope (Oberkochen, Germany) with 12‐bit colour CCD camera (CoolSnap, Roper Scientific, Tucson, AZ, USA); resolution 1392 × 1042 pixels as described previously (Barna et al. 2014; Takakura et al. 2014; Tuppy et al. 2015). Briefly, section alignment between brains was done relative to a reference section. To align sections around the RTN level, the most caudal section containing a cluster of facial motor neurons was identified in each brain and assigned the level bregma −11.6 mm, according to the atlas of Paxinos and Watson (Paxinos & Watson, 2005). Levels rostral or caudal to this reference section were determined by adding a distance corresponding to the interval between sections multiplied by the number of intervening sections. Six sections that included the RTN were analysed. The same method was also used to identify the bregma level of the locus coeruleus (LC), medullary raphe, nucleus of the solitary tract (NTS), Bötzinger complex (BötC), pre‐Bötzinger complex (preBotC), rostral ventral respiratory group (rVRG) and caudal ventral respiratory group (cVRG). However, for densitometric analysis of NK1R immunoreactivity, the region was defined by circle 350 μm diameter long, the top centred on the ventral edge of the nucleus ambiguus or nucleus retroambiguus and nine sections of the region of ventral respiratory column (VRC) were analysed. All the counts were made bilaterally in the brain regions according to a previous study (Tuppy et al. 2015).

Statistical analysis

Audiogenic kindling progression (cSI and LI values) was evaluated by means of one‐way repeated measures (RM) ANOVA. All other data sets were tested for normality using the Shapiro‐Wilk test, then differences within and between groups were determined using one‐ or two‐way RM ANOVA with Bonferroni multiple comparisons. Student's t test and a non‐parametric Kruskall‐Wallis test were also performed as required. All values are expressed as means ± SEM and significance differences indicated when P < 0.05.

Results

Characterization of the genetic model of epilepsy: the Wistar audiogenic rat (WAR) strain

Most of the WARs exposed to acute high‐intensity (120 dB) auditory stimulation develop tonic–clonic (mesencephalic) seizures and after repeated acoustic stimulation (audiogenic kindling) they develop limbic seizures, both varying in severity along the kindling progression (Garcia‐Cairasco et al. 1996; Dutra Moraes et al. 2000; Romcy‐Pereira & Garcia‐Cairasco, 2003). It is important to point out that none of our Wistar rats (controls) developed mesencephalic seizures (measured by the categorized seizure severity index (cSI); Rossetti et al. 2006) or limbic seizures (measured by the limbic seizure severity index (LI); Racine et al. 1972) during chronic auditory stimulation (2 stimulations per day for a total of 20 stimulations). In contrast, all WARs (N = 18) used in these experiments displayed mesencephalic seizures during high‐intensity auditory stimulation. The mean cSI at day 1 of kindling (first stimulation) was 5.88 ± 0.40, where 15 out of 18 (83.3%) displayed tonic–clonic seizures. The mean cSI at day 10 of kindling (twentieth stimulation) was 4.05 ± 0.74. Among these animals, 10 out of 18 (55.5%) exhibited limbic seizures, usually intermixed with wild running behaviour and tonic–clonic seizures. Those limbic seizures consisted usually of a combination of orofacial automatisms, head and forelimb myoclonus (Racine et al. 1972) and kindled WARs displayed a mean cSI at the twentieth stimulus of 1.44 ± 0.36, clearly a reduction when compared with day 1 (1st stimulation). The first limbic seizures were observed in two animals at the 6th stimulation. Two additional animals displayed the first limbic seizures at the 7th stimulation and the last two animals presented limbic seizures at the 9th stimulation. In all cases, during audiogenic kindling progression, limbic seizure scores increased slightly from the first appearance while the mesencephalic cSI decreased gradually (Fig. 1). Audiogenic kindling in WARs, therefore, showed a progressive tonic–clonic seizure attenuation (one‐way RM ANOVA; P = 0.0051, F = 2.096) with cSI decrease, concomitantly with the appearance of limbic behaviours and LI increase (one‐way RM ANOVA; P < 0.0001; F = 5.943) (Fig. 1). In summary, mesencephalic seizures co‐existed with limbic seizures. For example, some animals displayed wild running followed by tonic–clonic and limbic seizures afterwards. Additionally, some animals displayed, for example, post‐tonic limbic clonic seizures episodes. It will be important to note that we stopped the sound stimulus as soon as the first tonic posture appeared (opisthotonus) following Garcia‐Cairasco and colleages demonstration (Garcia‐Cairasco et al. 1996), or at a maximum of 1 min. The animals displaying opisthotonus usually went into clonic seizures at the removal of the sound stimulation; in contrast, animals displaying wild running or limbic seizures froze at the removal of the sound, at the maximum of 1 min.

Based upon their behavioural responses to chronic auditory stimulation our animals were classified in the following groups: (i) Wistar naive and WAR naive: not exposed to high‐intensity auditory stimulation; (ii) Wistar resistant: exposed to high‐intensity auditory stimulation, but did not develop seizures (as noted before, a subpopulation of Wistar rats, near 10% from the vivarium, can eventually display mild audiogenic seizures, but they were excluded from our analysis) and (iii) WAR limbic: 13 out of 18 (72.2%) displayed at least three limbic seizures during the audiogenic kindling protocol.

Respiratory effects elicited by hypercapnia in WARs and Wistar rats

The first series of experiments were performed to characterize baseline breathing patterns in the WAR strain. In adult WARs, we observed a lower ventilation (${\dot{V}}_{\mathrm{E}}$) compared to Wistar rats (332 ± 105 vs. Wistar 505 ± 36 ml kg⁻¹ min⁻¹, P < 0.001) due to a lower tidal volume (V _T) (3.5 ± 0.8 vs. Wistar 5 ± 0.78 ml kg⁻¹, P < 0.001) (Fig. 2 A and C). There was no difference in baseline respiratory frequency (f _R) when we compared WARs and Wistar rats (P = 0.137) (Fig. 2 B). A lower ${\dot{V}}_{{\mathrm{O}}_2}$ (19 ± 0.8 vs. Wistar 24 ± 1 ml kg⁻¹ min⁻¹, P < 0.01) and the ${\dot{V}}_{\mathrm{E}}$/${\dot{V}}_{{\mathrm{O}}_2}$ ratio (16.9 ± 3 vs. Wistar 24 ± 4 ml kg⁻¹ min⁻¹, P < 0.01) were also observed in the WAR strain, without changes in the resting body temperature (36.3 ± 0.2 vs. Wistar 36.8 ± 0.3°C, P = 0.254) (Fig. 2 G–I).

Figure 2. Respiratory deficits in a genetic model of audiogenic seizure, the WAR strain.

Whole‐body plethysmography was used to measure tidal volume (V _T, ml kg⁻¹), respiratory rate (f _R, breaths min⁻¹) and minute ventilation (${\dot{V}}_{\mathrm{E}}$, ml kg⁻¹ min⁻¹) in unanaesthetized Wistar or WAR animals during exposure to normoxia or hypercapnia (7% CO₂). Summary data plotted as V _T (A), f _R (B) and minute ventilation (${\dot{V}}_{\mathrm{E}}$; C) show that WAR animals have a decreased hypercapnia‐induced increases in V _T and ${\dot{V}}_{\mathrm{E}}$. Changes in tidal volume (V _T; D), breathing frequency (f _R; E), minute ventilation (${\dot{V}}_{\mathrm{E}}$; F), oxygen consumption (${\dot{V}}_{{\mathrm{O}}_2}$; G), ${\dot{V}}_{\mathrm{E}}$/${\dot{V}}_{{\mathrm{O}}_2}$ relationship (H) and body temperature (I) in Wistar (naive or kindling) or WAR (naive or kindling). ^*Different from Wistar strain; N = 7–13/group of rats). [Colour figure can be viewed at wileyonlinelibrary.com]

We then challenged both WAR and Wistar animals with an increase in ambient (chamber) CO₂. This was done under hyperoxic (50% O₂) conditions, in order to attenuate the contribution of peripheral chemoreceptors to the hypercapnic chemoreflex ventilatory response (HCVR) (Lahiri & DeLaney, 1975 a,b; Takakura et al. 2014). It is important to note that hyperoxia caused a small decrease in baseline ${\dot{V}}_{\mathrm{E}}$ in both WARs and Wistar rats, which would be expected with suppression of peripheral chemoreceptor input. Ventilation (${\dot{V}}_{\mathrm{E}}$) in WARs was significantly lower (813 ± 341 vs. Wistar 1661 ± 177 ml kg⁻¹ min⁻¹, P < 0.001) when compared with Wistar rats during hyperoxic hypercapnia (Fig. 2 C and F). The reduction in the HCVR in WARs was solely attributable to a lower V _T (Fig. 2 A and D), and both WARs and Wistar rats increased f _R to equal degrees at 7% CO₂ (P = 0.137) (Fig. 2 B and E).

We did not observe significant changes in resting ${\dot{V}}_{\mathrm{E}}$ or the increase in ${\dot{V}}_{\mathrm{E}}$ produced by hypercapnia in both naive and kindling WAR animals (Fig. 2).

Arterial blood gases and pH were not different between genotypes, i.e both WARs and Wistar rats, indicating that there is no compensation by chronic metabolic acid–base changes (Table 1).

Table 1.

Changes in blood gas parameters in Wistar and WAR strains in the rest condition

Groups	pH	$P_{\mathrm{a},\mathrm{C}{\mathrm{O}}_2}$ (mmHg)	$P_{\mathrm{a},{\mathrm{O}}_2}$ (mmHg)	HCO3− (mm)
Wistar naive (N = 12)	7.35 ± 0.01	40 ± 3	83± 6	22 ± 3
Wistar kindling (N = 10)	7.37 ± 0.04	39 ± 3	85± 5	23 ± 4
WAR naive (N = 7)	7.38 ± 0.03	43 ± 4	82± 7	24 ± 2
WAR kindling (N = 13)	7.35 ± 0.02	44 ± 5	81± 4	25 ± 6

Values are means ± SEM. $P_{\mathrm{a},\mathrm{C}{\mathrm{O}}_2}$, arterial partial pressure of CO₂; $P_{\mathrm{a},{\mathrm{O}}_2}$, arterial partial pressure of O₂; HCO₃ ⁻, bicarbonate.

Effect of hypercapnia on Fos expression by a population of chemoreceptor neurons

Next, we designed protocols to confirm and extend our interpretation of the previously described experiments by using the proto‐oncogene product Fos, used as a reporter of cell activation (Barna et al. 2012; 2014). The CO₂‐sensitive neurons of the RTN region belong to a cell group with a well‐defined phenotype characterized by the presence of VGLUT2 mRNA, a Phox2b‐ir nucleus and the absence of both tyrosine hydroxylase (TH) and choline acetyl transferase (ChAT) (Stornetta et al. 2006). In practice, the chemoreceptors RTN neurons can be identified histologically as Phox2b‐positive and TH‐negative cells (Phox2b⁺/TH⁻) within the region of interest because every Phox2b‐ir neuron within this region contains VGLUT2 mRNA (Stornetta et al. 2006) and the Phox2b⁺/TH⁻ are separated from surrounding ChAT‐positive neurons, e.g. the facial motor neurons (Takakura et al. 2014). In addition, hypercapnia is known to induce Fos expression in the RTN region of animals (Sato et al. 1992; Teppema et al. 1994; Fortuna et al. 2009).

We compared Fos expression in the RTN neurons in four groups of Wistar and WAR animals. The control group (Wistar normocapnia, W‐N: N = 3) was maintained under hypocapnic (50% O₂, balance N₂) conditions. The second group (Wistar hypercapnia, W‐H: N = 4) was subjected to hypercapnia (7% CO₂, 50% O₂, balance N₂). The third rat group (WAR normocapnia, WAR‐N: N = 4) was maintained under hypocapnic (50% O₂, balance N₂) conditions and the fourth group (WAR hypercapnia, WAR‐H: N = 4) was maintained under hypercapnic conditions (7% CO₂, 50% O₂, balance N₂). The number of Fos⁺/TH⁻ neurons were identified and counted in a one‐in‐six series of transverse sections (1 section every 240 μm). Counts were made on both sides of the brain and throughout the portion of the ventrolateral medulla (RTN region) (Fig. 3 A). Figure 3 B shows the RTN region in a control rat (Wistar) and Fig. 3 C in WARs that were exposed to high levels of CO₂ in a period of 3 h. Due to the antibody species competition with our Phox2b antibody, we had to run our experiments to highlight that the Fos immunoreactivity (Fos⁺) was absent in the TH⁺ cells, suggesting that the neurons activated by hypercapnia were probably Phox2b immunoreactive (Stornetta et al. 2006).

Figure 3. Hypercapnia triggers Fos expression by the RTN in Wistar and WAR strains.

A, transverse sections of the medulla oblongata illustrating the region in which Fos‐ir neurons were identified and counted. The numbers at the left of the hemisections refer to the location of the sections caudal to the bregma level (in mm) according to the atlas of Paxinos & Watson (Paxinos & Watson, 2005). The case represented in A is from a Wistar rat (left side of the hemisection) and a WAR (right side of the hemisection) that had been exposed to hypercapnia for 3 h. Cell were counted bilaterally. The area located ventral to and immediately caudal to the facial motor nucleus (VII) delineates the parafacial region that the RTN neurons (Fos⁺TH⁻) were identified and counted. B and C, Fos expression in the RTN region after hypercapnia in Wistar (B) and WAR (C). White arrows represent Fos⁺/TH⁻ neurons, while black arrow indicates TH⁺. D, total number of Fos⁺ RTN neurons counted per rat in the same 4 groups of rats (Wistar normoxia (W‐N); Wistar hypercapnia (W‐H); WAR normoxia (WAR‐N) and WAR hypercapnia (WAR‐H). ^*Different from normoxia groups (P < 0.05). ⁺Different from Wistar rats that were exposed to hypercapnia (P < 0.05). Scale bars: 1 mm in A; 40 μm C (applies to B and C); 20 μm in C´. Abbreviations: NA, nucleus ambiguus; py, pyramidal tract; Sp5, spinal trigeminal tract; IO, inferior olive; VMS, ventral medullary surface; VII, facial motor nucleus. [Colour figure can be viewed at wileyonlinelibrary.com]

The ventrolateral medulla contains two clusters of Fos⁺ neurons centred predominantly within the rostral aspect. These include the catecholaminergic neurons (C1 neurons) and the non‐catecholaminergic, presumptive RTN chemoreceptors (Takakura et al. 2008, 2014) (Fig. 3 B and C).

The Fos‐ir was expressed in both catecholaminergic (identified by TH immunoreactivity) and non‐catecholaminergic neurons (TH⁻) (Fig. 3 A–D). The non‐catecholaminergic cells were generally located lateral to the TH‐ir neurons and under the facial motor nucleus (Fig. 3 A). More specifically, immunoreactive nuclei were present in the area located at the rostral end of the ventral respiratory column called the RTN region. Based on our previous experience, we believe that the neurons activated by hypercapnia belong to a cell group with a well‐defined phenotype characterized by the presence of VGLUT2 mRNA and Phox2b‐immunoreactivity and the absence of both TH and choline acetyl transferase (ChAT) activity (Stornetta et al. 2006; Takakura et al. 2014).

Based on the above description, hypercapnia caused a large increase in the number of Fos⁺ cells in presumably RTN neurons (Fos⁺/TH⁻) (Fig. 3). Under these circumstances, the number of Fos⁺/TH⁻ immunorreactive cells were lower in the WAR‐H versus the W‐H group (P < 0.01) (Fig. 3 D). Summing the cells identified at all levels of the region of interest represented in Fig. 3 A and multiplying this number by 6, produced an uncorrected total number of Fos⁺/TH⁻ RTN neurons counted per rat. After applying a 0.81 Abercrombie correction factor previously determined on identically prepared histological material (Takakura et al. 2008), more accurate estimates of the actual number of Fos⁺ RTN neurons in the four rat groups were: W‐N: 54 ± 12; W‐H: 545 ± 37; WAR‐N: 34 ± 8 and WAR‐H: 189 ± 36. The difference between the experimental groups and the control groups was highly significant by one‐way ANOVA (Fig. 3 D).

We also counted the number of Fos⁺ neurons located in the raphe region (raphe pallidus/parapyramidal region (RPa/ppy); raphe obscurus (ROb)), within the nucleus of the solitary tract (NTS) and in the locus coeruleus (LC) (Fig. 4 A–I). The number of Fos⁺ neurons in the RPa/ppy (442 ± 37 vs. W‐H 976 ± 21, P < 0.05) and in the ROb (562 ± 58 vs. W‐H 1876 ± 157, P < 0.01) were lower in WAR‐H group compared to W‐H group (Fig. 4 A, D and G). The number of Fos⁺ neurons in the NTS was not different from the WAR strain (263 ± 25 vs W‐H. 279 ± 43, P = 0.146) (Fig. 4 B, E and H). In contrast to the lower number of neuronal cells activated in the RTN and raphe, the number of Fos⁺ neurons in the LC was higher compared with the W‐H group (212 ± 26 vs. W‐H 46 ± 11, P < 0.01) (Fig. 4 C, F and I). We found that 88% of the total Fos⁺‐activated neurons by hypercapnia (186 ± 13 vs. total Fos⁺ 212 ± 26, P = 0.001) belongs to the noradrenergic cells located in the LC region.

Figure 4. Different Fos expression elicited by hypercapnia in the raphe pallidus, nucleus of the solitary tract and locus coeruleus.

A–F, Fos expression in the raphe pallidus (A and D), nucleus of the solitary tract (B and E) and locus coeruleus region (C and F) after hypercapnia in Wistar rats (A–C) and WAR (D–F). White arrows represent Fos⁺ neurons, while black arrows indicate TH⁺ neurons. G–I, total number of Fos⁺ neurons counted per rat in the same 4 groups of rats (Wistar normoxia (W‐N); Wistar hypercapnia (W‐H); WAR normoxia (WAR‐N) and WAR hypercapnia (WAR‐H). ^*Different from normoxia groups (P < 0.05). ⁺Different from Wistar rats that were exposed to hypercapnia (P < 0.05). Scale bars: 40 μm in D applies to A, D, B and E; 20 μm in E´ applies to B´ and E´; 60 μm in F applies to C and F. Abbreviations: AP, area postrema; cc. central canal; LC, locus coeruleus; NTS, nucleus of the solitary tract; ROb, raphe obscurus; RPa/ppy, raphe pallidus/parapyramidal region; 4V, fourth brain ventricle. [Colour figure can be viewed at wileyonlinelibrary.com]

Lower number of neurons within the ventral respiratory column in WARs

The following experiments were designed to further assess the effect within the several types of neurons involved in the respiratory control on the WAR model.

Phox2b immunoreactivity (Phox2b‐ir) was examined within the RTN regions in both WAR and Wistar animals. The number of Phox2b (Phox2b⁺) nuclei counted in one‐in‐six series of 40‐μm brain sections per rat into the RTN was lower (274 ± 18 vs. Wistar 968 ± 28, P < 0.05) in WAR animals compared to Wistar rats (N = 5; Fig. 5 A and B). The remaining Phox2b⁺ nuclei within the RTN region in WAR animals belonged to the dorsal cap area of RTN (Fig. 5 B and C). The Phox2b⁺ neurons located in the thin band of myelin‐poor brain parenchyma, composing the marginal layer, were almost absent (Fig. 5 B and C). The number of Phox2b⁺ neurons within the NTS was not different between both strains (i.e WAR and Wistar) (WAR 1224 ± 153 vs. Wistar 1120 ± 137, P > 0.05) (Fig. 5 D–F).

Figure 5. Phox2b expression by the RTN in Wistar and WAR strains.

Phox2b expression in the RTN region in Wistar (A) and WAR (B). White arrows represent Phox2b⁺/TH⁻ neurons, while black arrows indicate Phox2b⁺/TH⁺. C, total number of Phox2b⁺/TH⁻ RTN neurons counted per rat in the same 2 groups of rats (Wistar normoxia (W‐N) and WAR normoxia (WAR‐N)). Phox2b expression in the NTS region in Wistar (D) and WAR (E). White arrows represent Phox2b⁺/TH⁻ neurons, while black arrows indicate Phox2b⁺/TH⁺. F, total number of Phox2b⁺/TH⁻ NTS neurons counted per rat in the same 2 groups of rats (Wistar normoxia (W‐N) and WAR normoxia (WAR‐N)). ^*Different from Wistar group (P < 0.05). Scale bar in E = 40 μm and applies to all panels. Abbreviations: NTS, nucleus of the solitary tract; RTN, retrotrapezoid nucleus; VMS, ventral medullary surface. [Colour figure can be viewed at wileyonlinelibrary.com]

In our WAR model, we also analysed the substance P receptor immunoreactivity (neurokinin 1 receptor: NK1R‐ir) present within the ventral respiratory column (VRC) (Fig. 6). WAR had a lower NK1R immunoreactivity within the Bötzinger (BötC) (35 ± 3% of control; P = 0.033), preBötzinger complex (preBotC) (31 ± 2% of control; P = 0.037) and in the premotor neurons located in the ventral respiratory group (rVRG (21 ± 0.9% of control; P = 0.041) and cVRG (32 ± 1.3% of control; P = 0.034)) compared to Wistar rats (Fig. 6 A–I). NK1R staining in dorsal brainstem regions such as the solitary tract (98 ± 2% of control; P = 0.125) and motor nucleus of the vagus appeared normal (95 ± 3.4% of control; P = 0.243).

Figure 6. NK1R‐expressing neurons in the ventral respiratory column in Wistar and WAR strains.

Representative photomicrographs of Wistar rats (A, C, E and G) or WAR (B, D, F and H). A–H, NK1R immunoreactive within the ventral respiratory group, more specifically in the Bötzinger complex (BotC; A and B); preBötzinger complex (PreBotC; C and D); rostral ventral respiratory group (rVRG; E and F) and caudal ventral respiratory group (cVRG; G and H). I, analysis of the extent of NK1R immunoreactivity in the regions outlined by the dashed lines in A–H. NK1R immunoreactivity is expressed as percentage of the immunoreactivity on the control group (Wistar group). ^*Different from Wistar group (P < 0.05). Abbreviation: Amb, nucleus ambiguus. Scale bar in H = 200 μm and applies to all panels. [Colour figure can be viewed at wileyonlinelibrary.com]

Reduction of the serotonin terminals within the RTN in the WAR strain

As described above, RTN neurons can be identified histologically by the presence of the transcription factor Phox2b (Stornetta et al. 2006; Takakura et al. 2008, 2014). According to Stornetta and colleagues (2006), these cells are located ventrally on the facial motor nucleus with a major concentration at the caudal end of this nucleus. A fraction of the chemosensitive neurons reside in the marginal layer (50 μm thick layer of neuropil located below the spinocerebellar tract), and the rest are found mostly within and dorsal to the spinocerebellar tract (dorsal cap of RTN) (Stornetta et al. 2006). The chemosensitive neurons, regardless of the location of their somata, have extensive dendrites both within the marginal layer and in the dorsal cap of RTN (Mulkey et al. 2004; Weston et al. 2004). Both of these regions contained a dense network of serotonin‐immunoreactive (ir) terminals (Fig. 7), and are rich in substance P‐ir terminals (Takakura et al. 2008, 2014). Within the marginal layer, a large proportion of the substance P‐containing terminals were also positive for serotonin, whereas only a few contained the adrenaline‐synthesizing enzyme PNMT (Rosin et al. 2006).

Figure 7. Serotonergic innervation of the ventral medullary surface in Wistar and WAR strains.

The ventral medullary surface (VMS) of the retrotrapezoid nucleus (RTN) receives input from terminals labelled by markers for a wide variety of neurotransmitters. A and B, serotonin (5HT)‐ir varicosities in the VMS of brainstem section (∼11.6 mm caudal to bregma) in Wistar and WAR strains. C and D, examples of serotonergic neurons (TrpOH revealed with with Cy3) in the raphe obscurus (ROb) in Wistar and WAR. E, number of 5HT neurons that were identified in the raphe regions (raphe obscurus (ROb) and raphe pallidus/parapyramidal region (RPa/ppy)). ^*Different from Wistar group (P < 0.05). Scale bar in D = 100 μm and applies to all panels. [Colour figure can be viewed at wileyonlinelibrary.com]

The major source of the serotonergic input to RTN is located in brain areas containing dense concentrations of serotonergic cells (the raphe nuclei). According to a previous study, all subdivisions of the raphe from the caudal medulla oblongata to the caudal midbrain (raphe obscurus, raphe pallidus, parapyramidal area, raphe magnus, raphe interpositus, and pontine, median, and dorsal raphe) send projections to the RTN region (Rosin et al. 2006; Mulkey et al. 2007). In the present study, we focused on the raphe obscurus and into the raphe pallidus/parapyramidal region, midline structures that extend throughout the medulla oblongata. They are suspected to play a particularly prominent role in facilitating respiratory, autonomic, and somatic motor outflows (Jacobs et al. 2002; Richerson, 2004; DePuy et al. 2013). We found that the number of 5‐HT varicosities within the marginal layer of the RTN region were lower in WAR animals compared to Wistar rats (Fig. 7 A and B). The number of tryptophan hydroxylase immunoreactive neurons in the ROb (WAR 79 ± 14 vs. Wistar 568 ± 22, P < 0.001) and in the RPa/ppy (WAR 199 ± 6 vs. Wistar 428 ± 13, P < 0.01) were also lower in WAR compared to Wistar rats. This observation may explain the reduction in the serotonergic innervation in the RTN region in WAR animals (Fig. 7 A–E).

Exogenous injection of serotonin into the RTN in WAR and Wistar rats

In Wistar rats, unilateral injection of serotonin (1 mm, 50 nl) into the RTN of conscious animals increased ${\dot{V}}_{\mathrm{E}}$ (819 ± 88 vs. saline 514 ± 45 ml kg⁻¹ min⁻¹, P < 0.01) by increasing both V _T (7.2 ± 0.87 vs. saline 5.2 ± 0.54 ml kg⁻¹, P < 0.01) and f _R (127 ± 4 vs. saline 99 ± 5 breaths min⁻¹, P < 0.01) (Fig. 8 A–C). In the WAR model, unilateral injection of serotonin into the RTN of conscious animals also increased ${\dot{V}}_{\mathrm{E}}$ (498 ± 102 vs. saline 323 ± 34 ml kg⁻¹ min⁻¹, P < 0.05) by increasing both V _T (4.7 ± 0.67 vs. saline: 3.4 ± 0.5 ml kg⁻¹, P < 0.05) and f _R (112 ± 3 vs. saline 95 ± 2 breaths min⁻¹, P < 0.05) (Fig. 8 A–C). However, it is important to note that the stimulatory effect of serotonin on breathing was blunted in WAR animals (T ₈ = 7.145; P = 0.0126) (Fig. 8 A–C). Injections of serotonin were placed unilaterally in the RTN in these rats (Fig. 8 D). Briefly, the injection centre was 250 μm below the facial motor nucleus and 200 μm rostral to the caudal end of this nucleus, targeting the region that contains the highest density of CO₂‐sensitive RTN neurons (Mulkey et al. 2004; Takakura & Moreira, 2011, 2013; Takakura et al. 2011, 2013, 2014).

Figure 8. Respiratory responses to exogenous injection of serotonin (5HT) into the RTN in Wistar and WAR strains.

A–C, summary data (N = 5 rats) show effect of unilateral RTN injections of saline (50 nl) and serotonin (5HT; 1 mm, 50 nl) on tidal volume (V _T; A), breathing frequency (f _R; B) and minute ventilation (${\dot{V}}_{\mathrm{E}}$; C) in awake unrestrained rats. D, histological section showing the location of drug injections within the RTN. Abbreviations: py, pyramidal tract; Sp5, spinal trigeminal tract; VII, facial motor nucleus. Scale bar = 100 μm. ^*Different from saline; ⁺Different from 5HT in Wistar strain. [Colour figure can be viewed at wileyonlinelibrary.com]

Discussion

The data presented in this manuscript demonstrate a severe respiratory dysfunction as well as an impairment in the ventilatory CO₂ sensitivity in an experimental genetic model of tonic–clonic seizures, the Wistar audiogenic rat (WAR). The changes in baseline breathing and the ventilatory response to CO₂ were strongly correlated to the reduced number of neurons in the ventral respiratory column (VRC) and serotonergic neurons within the raphe region. Our findings suggest an epilepsy model with depressed neuronal activity in specific territories of the respiratory network (central chemoreception) strongly correlated with reduced breathing output.

Genetic model of audiogenic epilepsy: WAR model

During seizures, specific their brain areas become hyperactive with the synchronization of the neuronal activity. Audiogenic seizures, induced by high‐intensity auditory stimulation, are initiated within specific auditory structures such as the inferior colliculus (Ross & Coleman, 2000; Faingold, 2012). In animals susceptible to audiogenic seizures, it is known that repetition of the acoustic stimulus favours the recruitment of limbic structures, leading to a change in behavioural and EEG pattern of seizures, henceforth called audiogenic kindling (Marescaux et al. 1987; Garcia‐Cairasco et al. 1996). The WAR is a strain selected from Wistar audiogenic susceptible progenitors based on the phenotypic seizure response to high‐intensity auditory stimulation (Doretto et al. 2003). Audiogenic seizures from WARs appear as a result of the hyperactivation mainly of the inferior colliculus (Garcia‐Cairasco, 2002). The repetition of stimulus evokes changes in seizure behavioural repertoire together with the appearance of epileptic EEG activity in limbic structures, such as the amygdala, hippocampus and cortex, similar to that observed in other sound‐susceptible rat strains (Marescaux et al. 1987; Naritoku et al. 1992; Garcia‐Cairasco et al. 1996; Dutra Moraes et al. 2000; Romcy‐Pereira & Garcia‐Cairasco, 2003; Galvis‐Alonso et al. 2004). The mechanisms by which audiogenic kindling occurs are largely unknown. Current literature indicates that the inferior colliculus activates (besides known primary thalamic and cortical auditory pathways) the amygdala and hippocampus (Kraus & Canlon, 2012; Tupal & Faingold, 2012). These circuits are thought to have a role in the temporal processing of sound and sound‐related memories (Squire et al. 2001; Liberman et al. 2009).

Respiratory deficits in the WAR strain

There is evidence that in seizure events, in which the airflow was measured, central apnoeas with O₂ desaturation and increased levels of CO₂ were observed in approximately 50% of the cases (Bateman et al. 2008). We were able to record baseline breathing in a freely moving awake genetically selected model of epilepsy, the WAR strain. We found that WAR animals have a dramatic reduction in baseline V _T and ${\dot{V}}_{\mathrm{E}}$. The animals are apparently not sick because they maintain their body weight, the daily water intake and body temperature. Our data are also in agreement with previous data from Granjeiro and colleagues (2016) showing reduced ventilatory responses after exposing WARs to either hypoxia and hypercapnia.

Looking at oxygen consumption, we noticed a significant reduction in the ${\dot{V}}_{{\mathrm{O}}_2}$ and in the relation ${\dot{V}}_{\mathrm{E}}$/${\dot{V}}_{{\mathrm{O}}_2}$ when comparing WARs to the control Wistar animals. It is possible that WARs are not able to compensate their metabolism, resulting in a reduction in oxygen consumption. Additionally, we noticed a reduction in the number of 5‐HT neurons located in the raphe region that could also explain a reduction in the baseline breathing and possibly a reduction in oxygen consumption. Another potential interpretation of the reduction in baseline breathing is the fact that in the WAR strain we found a dramatic reduction in the number of NK1‐ir neurons in VRC which includes the chemoreceptors neurons in the RTN and the excitatory preBötzinger (preBotC) neurons. Both types of neurons are critical components of breathing automaticity (Guyenet et al. 2010 a,b; Feldman et al. 2013; Guyenet & Bayliss, 2015).

It is well established in the literature that RTN neurons drive breathing in resting conscious mammals. Marina and colleagues (2010) found that RTN inhibition did not change resting ventilation. However, amongst other study groups, we have shown pharmacologically or by genetic deletion of the RTN region a breathing reduction, at rest, during quiet waking (Li et al. 2006; Ramanantsoa et al. 2011; Takakura et al. 2013, 2014; Basting et al. 2015). In the present study, a reduction of 71% of the Phox2b⁺/TH⁻‐expressing neurons of the RTN could account for the breathing reduction at rest. It is also important to notice that it is unlikely that the reduced 5‐HT neurons would contribute to the reduced resting ventilation. Since others have shown in the 5‐HT neuron deletion models (Lmx1b and pet‐1 KO mice), there is no effect on resting ventilation despite losing all or some 5‐HT neurons (Hodges et al. 2008, 2011).

Central respiratory chemoreception in the WAR strain

Central chemoreception refers to the detection of changes in CO₂/H⁺ within the brain and the associated effects on breathing. Here, we found that WAR animals have a reduction of nearly 50% of the ventilatory response to CO₂. The reduction in the chemosensory control of breathing is related to the selective lack of labelling (or cell reduction) of more than 70% of Phox2b⁺/TH⁻ (i.e. chemoreceptors) neurons within the RTN region or to the decrease in 5‐HT neurons in the medullary raphe. We have two interpretations for the fact that we have almost 70% ablation of the Phox2b⁺/TH⁻ neurons in the RTN region and 50% reduction of the breathing response to CO₂.

First, as described above, the present experiments showed a large decrease in the chemosensory control of breathing elicited by CO₂ in unrestrained awake WARs with peripheral chemoreceptors. It is possible that peripheral chemoreceptors can partially account for the breathing response elicited by CO₂. The role of carotid body in compensating for the lack of central CO₂ sensitivity is ambiguously reported in literature (Forster et al. 2000; Blain et al. 2009; Takakura et al. 2014). The entire respiratory pattern generator is strongly activated by peripheral chemoreceptors and therefore may be an obligatory intermediate for relaying the CO₂ response of the peripheral chemoreceptors in a situation when the RTN chemoreceptor neurons are lost (Finley & Katz, 1992). Second, compensation by some of the multiple sites postulated to function as CO₂ sensors in the brain (Nattie & Li, 2009; Nattie, 2011) can be inactive or without strong enough connections to the respiratory pattern generator compared with control Wistar animals. Candidate chemoreceptor sites are the serotonergic neurons in the raphe or the noradrenergic neurons in the locus coeruleus (LC) that have been reported to be pH sensitive in vivo and in vitro (Richerson, 2004; Biancardi et al. 2008; Kanbar et al. 2010; Iceman et al. 2013; Buchanan et al. 2015). We found a dramatic reduction in the number of serotonergic neurons in the raphe regions; however, our results demonstrate an increase in the number of noradrenergic neurons in the LC activated by hypercapnia in the WAR strain. The LC is considered to have a central respiratory chemoreceptor function because these cells are activated by hypercapnia in vivo and by acidification in vitro (Pineda & Aghajanian, 1997; Ritucci et al. 2005; Biancardi et al. 2008; Nichols et al. 2008; Nattie & Li, 2009). The LC is also activated by peripheral chemoreceptors (Elam et al. 1981; Teppema et al. 1997), but lesions of this structure have no effect on the peripheral respiratory chemoreflex (Biancardi et al. 2010). According to others, the chemoreceptor function of LC is at least 10‐fold smaller than that of the RTN, at least in Wistar animals (Kanbar et al. 2010). In WARs, it seems that the LC noradrenergic neurons are activated in an attempt to compensate for the decrease of the RTN and raphe chemoreceptor neurons in the control of breathing. The interpretation of these data requires further study.

There is also evidence that the glial cells, especially in the ventrolateral medulla, have a chemoreceptor function (Gourine et al. 2010; Wenker et al. 2010). It has been known since the early work of Fukuda and Honda (Fukuda & Honda, 1975; Fukuda et al. 1978) that a subset of glial cells located near the ventral medullary surface in the vicinity of the RTN is depolarized by bath acidification. To fully understand the contribution of glial cells in the control of breathing in our model, it will be important for future work to identify and selectively manipulate subsets of astrocytes within the ventral medullary surface of WARs.

Role of central 5‐HT neurons in respiratory control

Previous reports investigating the role of serotonergic neurons in control of baseline breathing and the ventilatory response to CO₂ have relied on either pharmacological or genetic approaches (Olson et al. 1979; Mitchell et al. 1983; Hodges et al. 2008; Depuy et al. 2011; Iceman et al. 2013; Takakura & Moreira, 2013; Buchanan et al. 2015). We demonstrated that the WAR strain has a dramatic reduction in the number of 5‐HT neurons in the raphe region. In addition, we also noticed a significant reduction in the 5‐HT innervation of the ventral medullary surface, as illustrated by a strong reduction of serotonergic varicosities.

The literature has shown evidence that 5‐HT neurons are chemoreceptors, and there is a clear increased firing rate of 5‐HT neurons and serotonin release during hypercapnia (Veasey et al. 1995; Richerson, 2004; Kanamaru & Homma, 2007; Massey et al. 2015). However, the relative contribution of 5‐HT neurons to the central chemoreflex elicited by CO₂ compared with other putative chemoreceptors (Mulkey et al. 2004; Takakura et al. 2006; Biancardi et al. 2008) remains unclear. The current data do not directly address whether 5‐HT neurons are respiratory CO₂ chemoreceptors, but they do address two key issues. First, our data show that a ventilatory response to increased CO₂ can occur in the condition of reduced numbers of 5‐HT neurons. Second, we demonstrated that 5‐HT neurons play an important role in the ventilatory response to increased CO₂ in our genetic model of epilepsy, the WAR strain.

The deficits reported here in the WAR strain, with a significant decrease of 5‐HT neurons, shed light on the normal role of the 5‐HT system in central control of breathing. The contributions and relative importance of the 5‐HT system in respiratory control may be underestimated in this model, presumably by other compensatory mechanisms. Our results indicate that the WAR strain has a reduction in baseline breathing, which makes it unable to adequately respond to some environmental challenges, such as a rise in CO₂. Thus, a defect in the 5‐HT system that diminishes its ability to respond to this exogenous stressor would naturally compromise a vulnerable individual.

It will be interesting to demonstrate in further studies if WAR defects in central breathing, associated with a failing 5HT system, are relevant to SUDEP, which has previously been linked to abnormalities in 5‐HT neurotransmission (Buchanan et al. 2014; Zeng et al. 2015), and has been associated with defects in control of breathing (Hodges et al. 2008; Buchanan et al. 2015; Feng & Faingold, 2015).

The 5‐HT system exerts a profound effect on respiratory activity (Mulkey et al. 2007; Ptak et al. 2009; Ray et al. 2011). Serotonergic neurons originating from raphe nuclei project to all levels of the respiratory network including the RTN, the pre‐Bötzinger complex, and brainstem and spinal respiratory premotor and motor neurons. Evidence also indicates that 5‐HT modulation of RTN neurons is determined by inhibition of KCNQ channels, most likely by a Gq‐coupled receptor mechanism (Hawryluk et al. 2012). Considering that loss of KCNQ channels can cause certain types of epilepsy and reduction in breathing automaticity (Hawryluk et al. 2012), future studies will be necessary to correlate, in the WAR model, if the breathing impairment is linked with reduction in the KCNQ activity.

Finally, our expectation was that audiogenic kindling would amplify the reduced ventilatory responses in naive WARs, because of the impact of the seizure experience (2 seizures per day, 10 days; 20 seizures total). In fact, our results did not show any differences between naive and kindled rats in their reduced ventilatory performance when exposed to hypercapnia. However, previous data from our laboratory have shown that systemic injection of a subconvulsant dose of pilocarpine (240 mg kg⁻¹) induces status epilepticus (SE) with earlier latency and with more deaths than SE in Wistar animals (380 mg kg⁻¹) (Garcia‐Cairasco et al. 2004). Comparing SE with audiogenic kindling, the latter has an interval for recovery between seizures, which is not observed in the continuous seizures of SE. Therefore, further experiments are planned with WARs exposed to other seizure induction protocols such as SE induced by pilocarpine, in order to evaluate their respiratory performance when exposed to hypercapnia.

Conclusion

We demonstrated for the first time a respiratory impairment in a genetic model of tonic–clonic seizures, the WAR strain. More importantly, we demonstrated an overall decrease in the number of neurons located in the VRC, as well as a reduction in serotonergic neurons in the midline medulla. This is an important step in directly demonstrating marked changes in neuronal activity and breathing impairment in specific territories of the respiratory network of the WAR model. It is still unknown if these respiratory alterations validate WARs as a suitable model to study increased risk factors and mechanisms associated with SUDEP.

Additional information

Competing interests

None declared.

Author contributions

L.T.T., A.C.T., N.G.C. and T.S.M. designed the experiments; L.T.T., J.A.C.O. and T.S.M. performed experiments; L.T.T., J.A.C.O., N.G.C. and T.S.M. analysed data; A.C.T., N.G.C. and T.S.M. wrote the paper. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by public funding from São Paulo Research Foundation (FAPESP: grants: 2014/22406‐1 to A.C.T.; 2007/50261‐4 to N.G.C.; FAPESP‐CINAPCE (2005/56447‐7 to N.G.C.), Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq: grants 471744/2011‐5 and 471263/2013‐3 to A.C.T.; 303121/2010‐6 to N.G.C. and 471283/2012‐6 to T.S.M.), PROEX‐CAPES, FAEPA and CNPq Research Fellowship (305533/2012‐6 to T.S.M. and 301651/2013‐2 to A.C.T.).