Regulation of vagally-evoked respiratory responses by the lateral parabrachial nucleus in the mouse

Vagal sensory inputs to the brainstem can alter breathing through the modulation of pontomedullary respiratory circuits. In this study, we set out to investigate the localised effects of modulating lateral parabrachial nucleus (LPB) activity on vagally-evoked changes in breathing pattern. In isoflurane-anaesthetised and instrumented mice, electrical stimulation of the vagus nerve (eVNS) produced stimulation frequency-dependent changes in diaphragm electromyograph (dEMG) activity with an evoked tachypnoea and apnoea at low and high stimulation frequencies, respectively. Muscimol microinjections into the LPB significantly attenuated eVNS-evoked respiratory rate responses. Notably, muscimol injections reaching the caudal LPB, previously unrecognised for respiratory modulation, potently modulated eVNS-evoked apnoea, whilst muscimol injections reaching the intermediate LPB selectively modulated the eVNS-evoked tachypnoea. The effects of muscimol on eVNS-evoked breathing rate changes occurred without altering basal eupneic breathing. These results highlight novel roles for the LPB in regulating vagally-evoked respiratory reflexes.


Introduction
Vagal sensory inputs to the pontomedullary brainstem contribute to the patterning of breathing via regulatory reflexes (e.g., the Hering-Breuer reflex; Dutschmann et al., 2014) and mediate defensive respiratory behaviours (e.g., the cough reflex; Baekey et al., 2001;Canning et al., 2004). Additionally, vagal sensory information may ascend through ponto-medullary synaptic relays to the forebrain where it contributes to the generation of respiratory-related homeostatic emotions such as the perception of breathing adequacy and urge to cough (Van den Bergh et al., 2012;Ando et al., 2014).
Understanding of the functional organisation of central circuits that integrate and regulate vagal sensory information is important, as alterations in function from physiological to pathological conditions may contribute to the symptoms of respiratory disease. For example, chronic cough, a respiratory disorder thought to affect 10% of the global population (Chung et al., 2022), is associated with aberrant peripheral vagal sensory inputs to the brain from the airways as well as altered central nervous system integration of this input Audrit et al., 2017;Song et al., 2021), contributing both to morbidity in disease and decrease in quality of life (Brignall et al., 2008;Song et al., 2015;Chamberlain et al., 2015). However, an incomplete understanding of the physiology and neuroanatomy of central respiratory-related sensory-motor networks still often precludes any effective medical treatment of neurologic respiratory disorders.
Vagal sensory neurons terminate in two medullary brainstem nuclei, the nucleus of the solitary tract (NTS) and the paratrigeminal nucleus (Pa5) (McGovern et al., 2015a;McGovern et al., 2015b;Driessen et al., 2018). Neurons in these nuclei in turn have projections to a wide range of medullary and pontine neuron pools, many of which are involved in autonomic and respiratory control. In particular, neuroanatomical tracing has revealed that the lateral parabrachial nucleus (LPB) receives and relays vagal sensory inputs derived from the NTS and Pa5 in rats (McGovern et al., 2015a;McGovern et al., 2015b;Driessen et al., 2018). An established respiratory-related parabrachial region is the pontine respiratory group (PRG). The PRG, located in the rostral pons, predominately includes the Kölliker-Fuse nuclei (KF) as well as specific lateral parabrachial (LPB) and medial parabrachial (MPB) nuclei (Dutschmann and Dick, 2012). The PRG receives a rich supply of vagal sensory information via medullary relay sites, and is well known for its role in respiratory pattern generation and respiratory reflex physiology (Lumsden, 1923;Cohen and Wang, 1959;Bonis et al., 2011Bonis et al., , 2013Dutschmann and Herbert, 2006;Dutschmann et al., 2009;Dutschmann et al., 2021).
In this study we provide evidence for a role for the caudal lateral parabrachial nucleus in regulating vagally derived respiratory-relevant sensory information in mice, seemingly independent of any role in ongoing respiratory pattern generation.

Animals and surgical procedures
Experiments were approved by the University of Melbourne Animal Ethics Committee and performed on adult male and female C57BL/6 mice (14-31 g) of at least 4 weeks of age (n = 37). Mice were group housed in standard environment in the Biomedical Science Animal Facility with ad libitum access to food and water.
For each experimental preparation a mouse was placed into an anaesthesia induction chamber with an inflow of 4% isoflurane in oxygen at 0.24 litre(L)/minute (min). Upon loss of righting reflex, the mouse was then placed in a supine position on a thermostatically controlled heat pad (37⁰C, Physitemp TCAT-2LV Controller with rectal probe) and fitted with a nose cone for maintenance anaesthesia (2% isoflurane at 0.24 L/min). Limbs were secured to the heat pad to facilitate access to the neck and chest. A gauze was placed between the animal and the heat pad to prevent direct skin contact with the heat source.
The left and right vagi exhibit asymmetric respiratory responses in rats (Stauss, 2017), and as such, all procedures were performed on the right side for consistency. Once a stable breathing rate was achieved and the paw withdrawal and palpebral reflexes confirmed to be absent, the neck was shaved clean. A midline incision in the skin was made from 3 mm above the sternum and terminating caudally to the larynx. The skin was reflected with incisions extending ventrally to near the forelimb and ear and the two submandibular salivary glands overlaying the trachea were separated from each other using blunt dissection. The right submandibular gland was then lateralised and eventually fully separated from all surrounding tissues using a combination of blunt dissection and cauterisation. The sternocephalicus muscle was then removed, allowing unobstructed access to the vagus nerve which was exposed and separated from the surrounding tissues via gentle blunt dissection. A thin silk thread was looped under the vagus to allow for manipulation.
The mouse was then flipped onto a prone position and mounted in the stereotaxic frame fitted with an anaesthetic nose cone for maintenance anaesthesia as above. The skull was manipulated in the stereotaxic frame until Bregma and Lambda were on a horizontal plane (+/-0.050 mm). A micromanipulator (Mini-151 micromanipulator, Narishige, Japan) was used to guide and position custom-made vagus nerve stimulating electrodes (stainless steel, bare tips: 76.2 µm diameter, coated wire: 139.7 µm diameter, A-M Systems) next to the vagus nerve and the nerve was placed onto the electrode using the silk thread. The electrodes were then retracted a sufficient distance to ensure that they were isolated from surrounding tissues and connected to an electrical stimulator (SD9 stimulator, Grass Medical Instruments, Rhode Island, USA).
For placement of diaphragm electromyograph (dEMG) electrodes, the right upper lateral abdomen and area surrounding the ribcage was shaved clean. A short subcostal incision was made along the right upper quadrant abdomen below the ribcage. The ribcage was then lifted and access to the diaphragm was visually verified. If required, the liver was gently pushed medio-posteriorly. Two tungsten tip copper wire electrodes (custom made) were hooked onto the diaphragm. A steel tipped copper electrode was inserted into the left hind leg, which functioned as a reference potential. The dEMG activity was received by a Pre-Amplifier (NL844 Pre-Amplifier, Digitimer, Hertfordshire, UK), amplified (Neurolog System model NL900D, Digitimer, Hertfordshire, UK), and then digitised (Micro1401 Mk2 A-D converter, CED, Cambridge, UK), before the activity was recorded and saved on a personal computer running Spike 2 Version 9 (CED, Cambridge, UK).
For microinjections into the parabrachial nucleus, the top of the head was shaved clean. A midline incision was made on the scalp from 1 mm anterior to Bregma to 3 mm posterior to Lambda. A Pneumatic Pico-Pump (PV820, World Precision Instruments) was connected to its complementary nozzle (PicoNozzle, WPI) with the nozzle fixed onto the stereotaxic frame arm. The arm was positioned at a 29.5-degree holder angle. A glass capillary pipette (Borosilicate glass capillary, 1B100-4, World Precision Instruments) pulled by a pipette puller (Model P-2000, Sutter instruments) was attached to nozzle mounted on the stereotaxic frame arm. A hand-held drill (Foredom 110 Volt micro motor 1070) with a Carbide-Tungsten tip (1.0 mm diameter, DU70.10, Circuit Imprimé Français) was used to drill into the skull a window of approximately 2×2mm in size to allow micropositioning of the glass capillary into the LPB.

Experimental design
Prior to commencement of the experiment, isoflurane was reduced (0.8-1% in oxygen at 0.24 L/min flowrate) and dEMG was observed for 5 min for breathing rate to fully adapt to the lower level of aesthesia. Next, the optimal eVNS voltage (1.2-2.4 V) was identified, defined here as the minimum volts that could elicit complete eVNS-induced apnoea with the following parameters: 1 millisecond (ms) pulse with 8 s (sec) train duration at 16 Hertz (Hz). This was performed to ensure the signal strength was strong enough to activate C-fibres (McAllen et al., 2018).
After a 5 min baseline period of recording, an injection of sterile saline (0.9% NaCl, 100-150 nl over 5-10 s) was made into LPB sites using the Pneumatic PicoPump at ejection pressure 10-20 psi at target coordinates (medial-lateral: − 1.200 to − 1.300 mm, anterior-posterior: − 3.070 to − 3.160 mm, dorsal-ventral: − 3.900 to − 4.000 mm relative to Bregma; 29.5-degree holder angle; Fig. 1). The pipette tip was left in place for 5 min to ensure full diffusion at the injection site and limit spread of the injectate back along the injection tract. After retracting the pipette, post-saline eVNS was performed at the optimal voltage using the following parameters: an 8 s train duration, 1 ms pulse width at different frequencies (1 Hz, 2 Hz, 4 Hz, 8 Hz, 12 Hz, 16 Hz, 20hz and 24 Hz) with a 4 min inter-stimulus interval (Fig. 1). Following this, the pipette was frontloaded with muscimol (1.744 mmol/L) mixed with red retrobeads (1:50 dilution, Lumafluor) and a 100-150 nl injection was made into the same LPB site coordinates using the Pneumatic PicoPump at ejection pressure 10-20 psi over the course of 5-10 s (Fig. 1). The pipette tip was again left in place and after retracting, the post-muscimol eVNS was performed with the same parameters as used for post-saline eVNS (Fig. 1). Muscimol, a GABAA agonist, was chosen as microinjection into the brain results in pharmacological inhibition of neurons located in the injection site, interrupting any relay circuits utilizing the local neurons, as we have previously shown (Mazzone et al., 2020).

Histology and imaging
At the conclusion of experiments, mice were killed with intraperitoneal injection of sodium pentobarbital (150-200 mg/kg) and transcardially perfused with 40 ml phosphate-buffered saline (PBS, pH 7.4) followed by 40 ml of 4% paraformaldehyde (PFA). The brain was removed from the skull, post-fixed in 4% PFA at 4⁰C overnight and stored in 0.01% PFA at 4⁰C until sectioning.
Tissues were cryoprotected in 30% sucrose at 4⁰C for two nights, then submersed in optimal cutting temperature compound (OCTC, Tissue-Tek, California, USA) and snap frozen with dry ice. Cryostat-cut sections (50 µm thickness; Leica CM1860 UV) were collected in serial order in PBS (room temperature) and then mounted onto gelatin-coated slides. The slides were coverslipped with Fluroshield (Sigma-Aldrich, Fluroshield F6182) and then stored in the dark at 4⁰C until imaging on an upright fluorescent microscope (Leica DM6 B microscope with Leica DFC7000T camera) connected to a computer running LAS X (Leica Application Suite X). Imaging parameters such as exposure duration and gain were appropriately adjusted to minimise background noise and increase signal strength.
Photomicrographs of each brain section containing the parabrachial nucleus were viewed in FIJI (version 1.53k) and cross-referenced with a mouse brain atlas (Paxinos and Franklin, 2001) to identify the red retrobead-marked muscimol injection sites. Distributions of muscimol injections were mapped onto the mouse brain atlas and defined on a region-by-region basis. Injections were not arbitrarily compartmentalised to a single point, but instead any region which was labelled with red retrobeads following injection was considered as having been a target of the injection. As such, individual injections often incorporated several regions. All regions hit by an injection had the accompanying physiology considered as part of the average response for that region. Injections were grouped firstly by presence of retrobeads in the LPB, or at a close off-target region, and secondly by rostro-caudal level of the LPB (caudal:

Data analysis
The dEMG datasets were first selected through a set of criteria to ensure the validity of subsequent inter-subject comparison. Specifically, datasets were only included in final analysis if they had a successful surgical setup (n = 18 excluded), stable baseline breathing of greater than 60 breaths per minute (BPM) that persisted across the duration of the experiment (n = 1 excluded), and if muscimol microinjection entered the 4th ventricle resulting in change of baseline respiration (n = 2 excluded, change in baseline respiration interpreted as spread of muscimol through the 4th ventricle to respiratory pattern generating areas). Using these criteria, n = 16 experiments were included, n = 12 of which displayed injectate present in the LPB (injection sites also located in the medial parabrachial n = 2, the cuneiform nucleus (CnF) n = 1, or the fastigial nucleus (FN) n = 1).
Analysis of dEMG data was performed offline using Spike 2 Version 9 (CED, Cambridge, UK). Raw dEMG data was first rectified and smoothed. Amplitude thresholds were set whereby any dEMG signals that surpassed the threshold were automatically determined as a breath. For each eVNS the numbers of breaths during pre-stimulation 10 s period ('pre-stimulation' bin), during the 8 s stimulation ('stimulation' bin) and during post-stimulation 10 s period ('post-stimulation' bin) were counted. The counts for each bin were exported and converted to breaths per minute (BPM). Additionally, the dEMG amplitude of each breath was measured (millivolts) and the average amplitude for each bin was calculated. In cases where there was no breath, no amplitude was calculated. Finally, the percentage change in BPM (ΔBPM) and amplitude (ΔAMP) in response to eVNS were calculated to facilitate comparisons across experiments.
All statistical analyses were performed using GraphPad PRISM 9.2.0 (GraphPad Software, California, USA). Because eVNS evoked biphasic effects on breathing frequency, dependent on stimulation frequency, tachypnoea (increased ΔBPM; ΔBPM of greater than 0%) and apnoea (decreased ΔBPM; ΔBPM 0% to − 100%) were analysed separately. Thus, the maximal effect (E max ) was defined as the maximum evoked decreased ΔBPM during eVNS ( Fig. 2A), whereas the minimal effect (E min ) was defined as the maximum evoked increased ΔBPM ( Fig. 2A). Post-saline and post-muscimol E max and E min were compared using paired t-tests.
Whether the saline and muscimol datasets are significantly different to each other was also assessed utilising non-linear regressions. This was accomplished by assessing the comparative improvement in the sum of squares relative to degrees of freedom following separation of the merged data into saline and muscimol datasets. Specifically, all values were merged onto a single dataset, a non-linear regression was fit onto the merged data, and a sum of squares was obtained to assess the goodness of fit. This 'goodness of fit' was assessing the null hypothesis that the two datasets could be best explained by a single non-linear regression. The merged data was then split into its saline and muscimol dataset components, and a non-linear regression was fit onto both of them separately. The two regressions had their sum of squares calculated and added together. This 'goodness of fit' would assess whether the increased degrees of freedom created by separating the data into saline The maximally evoked tachypnoea and apnoea were set as Emin and Emax, respectively for the mean % change in respiratory frequency at different stimulus frequency. Paired t-tests compared post-saline Emin/Emax to post-muscimol Emin/Emax. A non-linear regression (solid red line) fit to the post-saline ΔBPM curve (constrained to not go above 0) was compared to its postmuscimol ΔBPM counterpart. (B) Change in pre-stimulation respiratory rate and dEMG amplitude over time. Pre-stimulation respiratory rate (BPM) and dEMG amplitude (mV) did not significantly change over the course of repeated eVNS trials or in response to muscimol injection. (C) Extra sum-of-squares F tests revealed a significant difference between post-saline and post-muscimol non-linear regressions suggesting muscimol injection decreased the efficacy of eVNS at evoking apnoea. (D) Extra sum-of-squares F tests revealed a significant difference between post-saline and post-muscimol non-linear regressions suggesting muscimol injection had a significant effect on ΔAMP (E) Emax muscimol injection data showed a significant abolishment of eVNS-induced apnoea. Note there were overlapping five datasets with a fully induced apnoea and no response to muscimol. (F) Emin muscimol injection data did not show a significant change in eVNS-induced tachypnoea. (G) Summary of individual injection rostro-caudal distributions. Error presented as S.E.M., n = 12. * p < 0.05, **** p < 0.0001. and muscimol datasets led to a proportionally significant decrease in the sum of squares. This was specifically assessed by comparing the sum of squares of the merged and separated datasets via extra sum-of-squares F tests. Statistical significance was defined as p < 0.05. Non-linear regressions fit onto the BPM data were constrained to a maximum value of 0 ( Fig. 2A), as eVNS-induced tachypnoea did not have a sufficient range of different eVNS frequency data points to accurately plot a regression curve. Amplitude data was analysed with the same approach with no constraints.
The average length of post-stimulation apnoea was calculated, and group means were compared, using repeated measures 2-way ANOVA. Pre-stimulation respiratory rate and dEMG amplitude were compared with repeated measures 1-way ANOVA instead, as eVNS was not a variable pre-stimulation. All data sets present represent the mean ± S.E. M. unless explicitly marked otherwise.

Baseline respiratory measures in isoflurane anaesthetised and instrumented mice
The baseline respiratory parameters of all animals used in this study were recorded prior to stimulation and these measures remained stable over the course of the experiment (Fig. 2B), indicating that neither the effects of repeated eVNS nor microinjections of saline or muscimol impacted baseline respiratory activity. Pre-stimulation respiratory rate or dEMG amplitude did not significantly change over the course of the experiment (repeated measures 1-way ANOVA, no main effect of repeated stimulation over time for BPM, F (15, 11) = 0.9397, p = 0.4491, or AMP, F (15, 11) = 1.496, p = 0.2449; Fig. 2B). No significant difference in respiratory parameters was observed when comparing pre-stimulation bins of the last post-saline stimulation and the first post-muscimol stimulation (paired t-test, 24 Hz saline prestimulation vs 1 Hz muscimol pre-stimulation, no effect of muscimol microinjection for BPM, t(11) = 1.19, p = 0.26, or AMP, t(11) = 0.77, p = 0.46; Fig. 2B).

Biphasic effects of eVNS on respiratory pattern
The eVNS following saline microinjections produced a biphasic effect on respiratory rate, dependent on the frequency of stimulation, accompanied by modest effects on respiratory amplitude ( Fig. 2A). At low stimulation frequencies (1-4 Hz), a rapid-onset tachypnoea was observed, which peaked at 38.2 ± 9.9% increase in mean respiratory rate at 4 Hz (n = 12; Fig. 2A). Increasing eVNS frequency progressively transitioned the evoked tachypnoea into a reduction in respiratory rate and eventual partial or complete apnoea at higher stimulation frequencies. Thus, at 16 Hz and 24 Hz eVNS, respiratory rate was reduced by − 76.3 ± 6.6% and − 80.0 ± 4.2% respectively, (n = 12; Fig. 2A), an effect that continued into the post-stimulation period for an average of 1.928 ± 0.19 s in all preparations.

Effect of lateral parabrachial nuclei muscimol microinjections on eVNS-evoked respiratory responses
First, we assessed pooled data from all animals (n = 12) independent of injection location in different LPB sub-nuclei. As discussed above, post-saline eVNS induced biphasic changes in respiratory rate, with tachypnoea followed by apnoea as eVNS frequency increased (Fig. 2C). In contrast, muscimol injection into LPB partially reduced the apnoeic effect of eVNS, whilst not significantly altering the tachypnoea at low stimulation frequencies (Fig. 2C). This resulted in a significant difference between the eVNS post-saline and post-muscimol datasets that could not be explained by a single non-linear regression (comparison of a non-linear regression fit onto merged data vs two non-linear regressions on the separated saline/muscimol datasets through extra sum-of-squares F tests, F(3, 186) = 6.292, p = 0.0004; Fig. 2C), along with a significant increase in E max (E max paired t-test, t(11) = 2.242, p = 0.0465; Fig. 2E) but not E min (paired t-test, t(11) = 0.1119, p = 0.9129; Fig. 2F). There was a significant effect on dEMG amplitude following muscimol microinjection into the LPB over the course eVNS stimulation frequency (comparison of a non-linear regression fit onto merged data vs two non-linear regressions on the separated saline/ muscimol datasets through extra sum-of-squares F tests, F(4167)= 3.053, p = 0.0184; Fig. 2D). Furthermore, whilst the post stimulation apnoea did significantly increase in duration with increasing eVNS frequency, muscimol microinjection did not significantly affect the apnoea duration (repeated measures 2-way ANOVA, main effect of increasing eVNS frequency, F (1, 7) = 22.83, p = 0.0001, no main effect of muscimol injection, F (1, 7) = 2.454, p = 0.1455, data not shown).
Despite the significant groupwise effects noted with this first level of analysis, it was also evident that there existed inter-subject variability in the magnitude of post-muscimol Emax or Emin effect sizes (Fig. 2E). This prompted a second level analysis to interrogate the relationship between the anatomical location of microinjection sites within the LPB and muscimol effect on Emax or Emin (Fig. 2G, Supplementary Figures 1  and 2). Histological analysis of injection sites showed that individual microinjections of muscimol often spread across more than one subregion within the LPB, albeit the pattern was different on subject-bysubject basis due to varied injection site co-ordinates (Fig. 2G). To account for this variability of injection site the LPB was divided into rostral, intermediate, and caudal subregions for the purpose of grouping preparations based on the histological presence of injectate ( Fig. 2G; also see Fig. 1).
In contrast, muscimol microinjections that reached the right intermediate LPB (n = 6) or right rostral LPB (n = 6) did not alter eVNSinduced apnoeic respiratory responses ( Fig. 4A; Fig. 5A). Non-linear regression analysis did not show significant difference between the merged and separated frequency-response curves (intermediate LPB; comparison of a non-linear regression fit onto merged data vs two nonlinear regressions on the separated saline/muscimol datasets through extra sum-of-squares F tests, F(3,90)= 1.938, p = 0.1291; Fig. 4B; rostral LPB; comparison of a non-linear regression fit onto merged data vs two non-linear regressions on the separated saline/muscimol datasets through extra sum-of-squares F tests, F(3,90)= 1.369, p = 0.2573; Fig. 5B).
Similarly, muscimol microinjection into the intermediate LPB or rostral LPB did not alter dEMG amplitude (intermediate LPB, comparison of a non-linear regression fit onto merged data vs two non-linear regressions on the separated saline/muscimol datasets through extra sum-of-squares F tests, F(4,81)= 0.7261, p = 0.5767; Fig. 4C; rostral LPB, comparison of a non-linear regression fit onto merged data vs two non-linear regressions on the separated saline/muscimol datasets through extra sum-of-squares F tests, F(4,83)= 0.2235, p = 0.9246; Fig. 5C) or the E max (intermediate LPB, t(5) = 0.7638, p = 0.4795; Fig. 4D; rostral LPB E max , t(5) = 1.037, p = 0.3472; Fig. 5D). Notably, there was a significant attenuation in the E min following muscimol   In two preparations, microinjections of muscimol did not reach any part of the LPB or pontine respiratory group. One of the missed injections reached the FN, with rostral spread to the CnF, while the second missed injection encompassed the caudal CnF, as well as along the edge of the external cortex of the inferior colliculus (ECIC) (Supplementary Figure 3). In these two preparations there was no significant effect of muscimol on eVNS-evoked respiratory responses (#165 data; non-linear regression for saline R 2 = 0.5266, fit onto muscimol dataset R 2 = 0.2576, comparison of the fit through extra sum-of-squares F tests F (3,10)= 0.1116, p = 0.9513; #111 data; non-linear regression for saline R 2 = 0.9298, fit onto muscimol dataset R 2 = 0.9760, comparison of the fit through extra sum-of-squares F tests F(3,10)= 0.9144, p = 0.4685; Supplementary Figure 3). Due to the low number of off-target microinjections E min and E max analysis could not be performed, however assessment of the data suggests no apparent difference for E max (#165 Saline E max = − 91.0 vs Muscimol E max = − 93.0; #111 Saline E max = − 100.0 vs Muscimol E max = − 100.0).

Discussion
Due to the well-established roles of the LPB nuclei as important sensory integration and relay sites, the present study aimed to specifically assess the role of the LPB in regulating vagal sensory-evoked changes in breathing pattern in mice. Through focal inhibition of neurotransmission with muscimol microinjection across different rostrocaudal levels of the LPB, we identified a caudal LPB site as having a novel role in determining eVNS-induced apnoeic responses. In addition, we observed an inhibition of eVNS-evoked tachypnoea following muscimol injection of a region in the intermediate LPB. These findings suggest that the LPB plays important roles in regulating respiratory responses evoked by vagal sensory neuron activation and support the notion that the LPB may be intimately involved in the regulation of reflex and higher order vagal sensory-evoked responses.

Functional and neuroanatomical evidence supporting a role for the LBP in modulating vagally-mediated respiratory reflexes
The present study identified the caudal/intermediate LPB as having a significant role in vagal sensory relay. eVNS-induced apnoea was abolished post-muscimol microinjection into the caudal LPB, while inhibition of the intermediate LPB area significantly attenuated tachypnoeic activity. This study's finding that interruption of local LPB activity altered respiratory response to eVNS is well supported by the literature. Indeed, topographically organised projection pattern is a robust and well accepted finding for many sensory inputs, including respiration (McGovern et al., 2015a;McGovern et al., 2015b). That said, the finding of a notable role of the caudal LPB was a surprising discovery, as the vast Fig. 6. Summary of anatomical distribution of significant attenuation of apnoeic or tachypnoeic response to eVNS following muscimol microinjection. LPB in dark brown, scp in grey. Anatomical mapping of attenuation of eVNS-induced apnoea (E max ) results are indicated in solid line and the mapping of the attenuation of eVNSinduced tachypnoea (E min ) in dashed line. Asterix symbolises the degree of significance, * p < 0.05** p < 0.01. majority of studies on breathing have not assessed this portion of the LPB. Gustatory and vagal inputs have previously been found to converge in the caudal LPB of rats, with neurons responsive to eVNS typically located in what was described as 'caudal lateral' portions of the LPB (Hermann and Rogers, 1985). However, a role for these vagal inputs in regulation of respiration via the caudal LPB was not assessed at that point, with other studies investigated the caudal LPB in terms of gustatory and taste sensory relay in rats and hamsters (Norgren and Pfaffmann, 1975;Travers and Smith, 1984;Halsell and Frank, 1991;Baird et al., 2001).
Consistent with the topographically organised functional data obtained in the present study, prior neuroanatomical tracing has identified that projections from 'general visceral' and 'respiratory' NTS areas differentially innervate the intermediate and rostral LPB (Herbert et al., 1990). In addition, there is evidence that vagal afferent relay neurons in the Pa5 and the NTS, regions well known to be associated with respiratory-relevant vagal sensory relay function, synapse with neuroanatomically separate rostral and intermediate LPB populations, further supporting divergent medullary relay projection pattern in the LPB (McGovern et al., 2015b;Driessen et al., 2018). However, in all tracing studies performed to-date, none have directly or indirectly assessed the caudal LPB to the authors' knowledge.
Physiological data obtained through local inhibition of medullary relay sites in previous studies mirror our results, suggesting a functional, in addition to neuroanatomical, link between the medulla and LPB. The medullary regions of the NTS and Pa5 are the primary sensory afferent termination sites of vagal sensory neurons. Interruption of their relay of sensory afferent information prevents respiratory pattern response to sensory stimuli, without altering eupnoeic respiratory pattern (Vardhan et al., 1993;Driessen et al., 2015). This was demonstrated via localised injection of muscimol into the NTS, which prevented apnoea normally induced by stimulation of cardio-pulmonary vagal afferent C-fibres in rats (Vardhan et al., 1993). The Pa5 was directly assessed in guinea pigs; muscimol inhibition of the Pa5 prevented laryngeal vagal sensory stimulation from inducing apnoea . Future study is needed to directly assess the voracity of a functionally relevant NTS/Pa5-caudal/intermediate LPB circuit, however current literature strongly suggests that such a circuit may indeed underpin the results observed in this study.

Implications of eVNS recruitment of different vagal fibre populations for LPB modulation of vagal respiratory reflexes
In the present study, eVNS at low frequencies induced a tachypneic effect which was consistently present at a 2-4 Hz stimulation frequency. Stimulation frequencies > 4 Hz slowed breathing and evoked apnoeic periods which persisted for the duration of high frequency eVNS and often extended into a post-stimulation apnoea. This stimulus-frequency dependent biphasic effect of eVNS likely reflects a differential recruitment of vagal sensory C-and A-fibres reflexes.
Tachypnoea in response to eVNS is commonly associated with activation of reflexes mediated by C-fibre vagal sensory nerve populations  which are normally silent or exhibit slow, irregular firing pattern (Coleridge and Coleridge, 1984) but can be recruited during the processing of pulmonary nociception (Mazzone and Undem, 2016). Our findings suggest that C-fibre reflexes may be more efficiently activated at lower stimulus frequencies. However, the central circuitry involved in vagal c-fibre-evoked tachypneic responses is poorly investigated and is further complicated by the fact that nodose C-fibres and jugular C-fibres have primary relay in the NTS and Pa5, respectively (Nassenstein et al., 2010). Our observation that inhibition of the intermediate LPB reduces vagally mediated tachypneic responses provides new insight into the central organisation of this reflex. By contrast, the apnoea evoked by higher frequency stimulation resembles the cessation of breathing characteristic of the Hering-Breuer reflex (Kubin et al., 1985), which has well-described circuitry involving large, myelinated A-fibre (pulmonary stretch receptor) inputs onto NTS pump cells that relay to the PRG where 'inspiratory off' responses are mediated through modulation of the pontomedullary respiratory network activity (Dutschmann and Dick, 2012). Interestingly, our data would also argue that a role exists for the caudal LPB in modulating respiratory responses mediated by this reflex pathway. Whether the caudal LPB and PRG interact to regulate apnoeic reflexes, or represent parallel distinct pathways, is presently unknown.
Overall, the biphasic response patterns observed in the present study are largely in accordance with previous findings in rats (Zaaimi et al., 2008) and mice . However, some minor differences were observed. eVNS was more effective in reducing the breathing rate (84.6% reduction) compared to optogenetic activation of mouse A-fibre populations (72% reduction) . Contrastingly, the tachypnoeic effect of low frequency eVNS (31.3% increase) was not as large as the 68% increase in respiratory rate with optogenetic activation of mouse C-fibre populations . This effect likely originates from the stimulation parameters employed, with eVNS effects at higher stimulus frequencies reported to range from complete apnoea (Stettner et al., 2007;Song et al., 2011), to a 23.7% decrease in breathing rate (Zaaimi et al., 2008). Furthermore, whereas optogenetic stimuli can be selectively targeted to single sensory neuron populations, the net functional responses mediated by eVNS presumably reflect the dominant net effect of activating reflexes mediated by multiple axon types simultaneously (Schertel et al., 1986;Bonham and Joad, 1991;Vardhan et al., 1993;Ravi and Singh, 1996;Paton, 1998;Chang et al., 2015). Indeed, stimulation frequency was the only parameter to be modified in the present study, delivered at an optimum voltage to activate all vagal sensory fibre types, limiting the direct interpretation as to whether eVNS-evoked effects arise from C-and A-fibre reflexes that are differentially activated in a frequency-dependent manner.

Possible organisation of LPB output circuitry modulating vagally mediated respiratory reflexes
The circuits through which LPB output neurons regulate breathing during vagus nerve stimulation are yet to be investigated. It is possible that muscimol microinjection into either the intermediate or caudal LPB may interrupt LPB outputs that directly project to the pontomedullary elements well known for modulating respiratory pattern. Such a conclusion would imply a function more in line with the role of the PRG, a known recipient of respiratory sensory inputs and effector on respiratory pattern generation. However, unlike inhibition of the PRG, muscimol injections into the caudal or intermediate LPB did not influence eupneic breathing rate. Thus, the LPB regions modulating vagal respiratory reflexes identified in the present study clearly do not form a core component of the respiratory central pattern generator. Instead, neurons in the intermediate and caudal LPB may be recruited by vagal sensory inputs to in turn modify respiratory pattern generation.
It is also possible that the pathway by which the LPB modulates vagally mediated changes in respiratory pattern is indirect, mediated through the ascent of sensory information to higher order brain regions, which in turn provide descending modulation of respiratory pattern generation. Indeed, a number of forebrain areas have been observed having roles in altering respiratory pattern, including in anaesthetised animals (Hassan et al., 2013;Bondarenko et al., 2014;Ajayi et al., 2018;Mazzone et al., 2020). Notably, the LPB is considered to be an important relay to the thalamus, and ultimately the insular and other cortical regions that lead to interoceptive perception of visceral sensation (Norgren and Leonard, 1973;Norgren, 1976;Davenport and Vovk, 2009;Han et al., 2018).

Technical considerations, limitations, future directions
One caveat in the present study is that muscimol injections were rarely confined to a single rostro-caudal level of the LPB. Injections must pass through LPB areas to reach more caudal targets, and microinjectate can spread from the initial injection site. Separating region-dependent effects relies on the assumption that if an injection is altering activity, other injections will specify which part of the injection is the effector region. This lack of independence in the regional data collected also prevented direct statistical comparison between responses evoked at differing rostro-caudal levels. While imperfect, this assumption holds for caudal LPB areas. This is as intermediate injections do not need to pass through caudal LPB areas, but caudal LPB injections pass through the intermediate LPB. As such the strongly localised effect to the caudal LPB area in terms of inhibition of eVNS-induction of apnoea appears robust. However, other approaches, for example using focal optical stimuli or electrophysiological recordings, would be useful to help refine LPB regions involved in the observed responses.
Further investigation would help interrogate the role of the caudal and intermediate LPB sites in modulating eVNS-induced respiratory responses. Indeed, careful consideration is needed in light of the fact that lesion of the neighbouring fastigial nucleus, can influence respiratory rate (Xu and Frazier, 1997). Notably, lesion of the FN was reported to have no effect on A-fibre induction of the Hering-Breuer reflex (Xu and Frazier, 1997), a finding that was corroborated by one experiment in the present study where muscimol that reached fastigial nucleus failed to alter eVNS-induced apnoea. Nevertheless, more studies are needed to further validate the caudal LPB result to conclusively separate it from the surrounding neuroanatomical landscape.
It is arguable that eVNS is a non-physiological stimulus as it will directly stimulate all vagal axons at regular rhythm and strength. This is notably distinct from physiological scenarios, where stimuli will be targeted to specific organs and optimal for specific sensory fibre types. It will be an important topic of future investigation to examine the role of LPB manipulations in respiratory responses to physiological stimuli, for example inhalation of airway irritants.
Finally, adaptation of evoked responses following repeated eVNS has been reported in rats (Siniaia et al., 2000;Dutschmann et al., 2009Dutschmann et al., , 2014. Moreover, prolonged exposure to anaesthesia has been found to reduce breathing rate over time in mice (Low et al., 2016). However, similarity between the intermediate and rostral LPB post-saline and post-muscimol frequency-response curves, in addition to the lack of any decrease over time in pre-stimulation breathing rate, suggests that these considerations were not major influences on the present study's results. Future study should attempt to further specify what ascending vagal pathway is being interrupted by muscimol microinjection into the caudal LPB. Previous investigation by our lab has identified two distinct ascending pathways for respiratory sensory information, one via the NTS and the other the Pa5. This would help identifying if only one or both pathways are interrupted, thus ascribing specific LPB physiological function to one or both medullary relay sites.

Conclusion
Modulatory function on respiratory activity in the LPB has traditionally been associated with the PRG and specifically the KF. Here we provide the first evidence that the caudal and intermediate LPB nuclei may also have roles, albeit distinct to those of the PRG, in the mediation of respiratory related vagal afferent inputs. This finding adds to a growing list of essential roles played by the LPB in regulating responses during hypercapnia and respiratory chemosensation, fluid and food intake, nociception, and thermoregulation. Experiments designed to understand how vagally mediated respiratory reflex inputs are integrated with the plethora of sensory modalities arriving in the LPB are needed to further contextualise the observed physiological findings.

CRediT authorship contribution statement
Conceptualisation: SBM, AAKM, MD. Experimental Procedures: RB, MT. Data Analysis: RB, Writingoriginal draft: RB, Writingreview and editing: All authors, Funding acquisition: SBM. All authors approved the final draft of the manuscript.

Declaration of Competing Interest
The authors declare they have no competing interest of relevance to this study.

Data Availability
Data will be made available on request.