Catecholamines modulate the hypoxic ventilatory response of larval zebrafish (Danio rerio)

Gill ventilation in fish is regulated to match metabolic demand with appropriate rates of respiratory gas transfer, and to enhance branchial gas exchange efficiency during environmental stressors. For example, during hypoxia (lowered levels of ambient O₂), hyperventilation (an increase in gill water flow) enhances oxygenation of the blood flowing through the gills (Perry et al., 2009; Perry and Gilmour, 2002). Hypoxic hyperventilation is achieved by increases in respiratory frequency (f_V) and/or amplitude depending on species (for reviews see Perry et al., 2009). Collectively, these changes in breathing patterns are termed the hypoxic ventilatory response (HVR). Although predominantly studied in adult fish, the HVR also occurs during early development prior to full maturation of the gill (Pan et al., 2019). In zebrafish larvae between 4 and 15 dpf, the increases in f_V are coupled with increased beating of the pectoral fins (Zimmer et al., 2020). The coupled increases in branchial and fin movements enable elevated water flow through the buccal cavity, rudimentary gills and across cutaneous surfaces. Although the role of the pectoral fin movements in hypoxic zebrafish larvae is unresolved (Zimmer et al., 2020), the branchial component of the HVR clearly increases hypoxia tolerance in larvae at 7 dpf, and presumably beyond, by increasing O₂ extraction from the buccal cavity (Pan et al., 2019).

The HVR in fish is likely to be initiated via activation of peripheral O₂ chemoreceptors (for reviews, see Jonz and Nurse, 2009; Jonz et al., 2015; Milsom, 2012; Perry et al., 2009). The cellular identity of piscine peripheral O₂ chemoreceptors and the underlying mechanisms whereby they sense and transduce O₂ signals have been most extensively studied in zebrafish (Jonz et al., 2004; Jonz and Nurse, 2003, 2005; Pan et al., 2021). However, regardless of species, the peripheral O₂ chemoreceptors in adult fish are thought to reside on or within the gills and orobranchial cavity, where they are well situated to detect changes in ambient and circulating levels of O₂. In adult rainbow trout (Oncorhynchus mykiss), serotonin-containing neuroepithelial cells (NECs) situated distally on the efferent edge of the gill filament were first identified as candidate O₂ chemoreceptors (Dunel-Erb et al., 1982). The O₂-sensing properties of NECs subsequently were confirmed in adult zebrafish (Jonz and Nurse, 2003). Although there is scant in vivo evidence, the NECs are generally considered the initial sites of O₂ sensing and subsequent downstream signaling culminating with the HVR (Jonz et al., 2015; Perry et al., 2009). Prior to innervation of the branchial NECs during early development (Jonz and Nurse, 2005), it was suggested that cutaneous NECs in zebrafish larvae may function as the peripheral O₂ chemoreceptors (Coccimiglio and Jonz, 2012). Regardless of the cellular identity and location of the peripheral O₂ chemoreceptors, afferent nerve fibres to respiratory control centres within the central nervous system (CNS) likely convey the hypoxic signal that elicits the HVR. The most likely candidates for the afferent nerves are cranial nerves IX and X because they appear to initiate most cardiorespiratory responses to hypoxia including the HVR (reviewed by Milsom, 2012) and rhythmic stimulation of the ganglion of cranial nerve X can entrain the ventilation response (De Graaf and Roberts, 1991).

The activation of the CNS respiratory control centres is the initial step of the HVR, which yields the rapidly occurring short-term phase of the hyperventilatory response. The short-term phase of the HVR, although ill-defined in fish (Porteus et al., 2011), is the period during which peak increases in ventilation are achieved. In adult zebrafish, the peak increases in f_V typically (although see Mandic et al., 2021 for an exception) occur within the first 60 min of continuous hypoxia exposure (Mandic et al., 2019; Vulesevic et al., 2006). The short-term phase of the HVR has been studied more thoroughly in zebrafish larvae (Mandic et al., 2019; Pan et al., 2021, 2019). In larvae younger than 15 dpf, the HVR is characterized by a rapid increase in f_V followed by a decline to a stable plateau phase in which f_V remains intermediate between baseline and peak levels until normoxia is restored (Mandic et al., 2019; Pan et al., 2021, 2019).

Although probably not involved in the onset of the HVR, neuroendocrine agents may modulate the short-term response once hyperventilation is initiated (for a review, see Pan and Perry, 2020) to fine-tune branchial water flow as the hypoxia exposure continues. Among the various neuroendocrine factors, the catecholamines (adrenaline and noradrenaline) have received considerable attention (Perry et al., 1992; Randall and Perry, 1992; Randall and Taylor, 1991). There is abundant evidence that intravascular injections of catecholamines, which may be elevated naturally during hypoxia (Boutilier et al., 1988; Butler et al., 1978; Kinkead and Perry, 1991; Perry and Gilmour, 1996; Perry et al., 1991; Perry and Reid, 1992; Perry et al., 2004; Reid and Perry, 2003; reviewed by Randall and Perry, 1992), can increase ventilation in fish (see table 1 in Pan and Perry, 2020). These effects of catecholamines possibly arise from interactions with centrally located β-adrenoreceptors (β-ARs) (Randall and Taylor, 1991). Less clear, however, is whether catecholamines, even when increased, play any role in modulating the short-term HVR. Catecholamines are unlikely to act directly on peripheral O₂ chemoreceptors because although gill O₂ chemoreceptors in rainbow trout increase their discharge during hypoxia, they do not respond to changes in catecholamine levels (Burleson and Milsom, 1995). The β-AR antagonist, propranolol, is known to impair the HVR in rainbow trout (Aota et al., 1990) and also block catecholamine-induced activation of central respiratory neurons in dogfish (Squalus acanthias) (Randall and Taylor, 1991). Thus, it was suggested that β-adrenergic stimulation of CNS respiratory neurons via centrally released catecholamines might contribute to the HVR (Perry et al., 1992). Alternatively, the β-blockade of the HVR could arise from nonspecific effects because propranolol can inhibit O₂ chemoreceptor discharge in an in vitro first gill arch of rainbow trout (Burleson and Milsom, 1990) despite the inability of catecholamines to stimulate these peripheral chemoreceptors (Burleson and Milsom, 1995).

With the recent advances in gene editing techniques such as CRISPR/Cas9 knockout (Zimmer et al., 2019b) and Tol2 transgenesis coupled with genetically encoded indicators (Kawakami, 2004; Lin and Schnitzer, 2016), the zebrafish has emerged as an important model to assess physiological mechanisms in fish. Through measurements of f_V combined with knockout of the β1-AR using CRISPR/Cas9 and live cell imaging using a genetically encoded Ca²⁺ indicator, we tested the hypothesis that adrenaline, by interacting with the β1 sub-type of peripheral and/or central β-ARs, contributes to the HVR in zebrafish larvae once the peak breathing response is attained. Larvae were selected specifically for these experiments because they facilitate in vivo cellular Ca²⁺ imaging. Additionally, the previously characterised multi-phasic HVR of zebrafish larvae suggested the secondary involvement of neuroendocrine agents.

Adult zebrafish were housed in 10 liter polycarbonate tanks in a recirculating aquatic system (Aquatic Habitats, Apopka, FL, USA). Fish were maintained at 28°C under a 14 h:10 h light:dark cycle in dechloraminated city of Ottawa tap water (hereafter referred to as system water) and fed to satiation twice daily. Wild-type zebrafish were obtained from in-house stock at the University of Ottawa aquatic care facility. The β-AR knockout adrb1^−/− line used in the present study was developed and validated by Joyce et al. (2022). The Tg(p2rx3b:GCaMP6s) transgenic line was generated in house as described below. All lines are available upon request. Embryos were obtained using standard protocols (Westerfield, 2000) and reared in an incubator in 50 ml Petri dishes containing system water at 28.5°C; water was changed daily. All procedures for animal use and experimentation were carried out in compliance with the University of Ottawa Animal Care and Veterinary Service guidelines under protocol BL-226, which adhered to the recommendations for animal use provided by the Canadian Council for Animal Care.

The ventilatory responses to pharmacological treatments or hypoxia of 4 days post-fertilization (dpf) zebrafish larvae were assessed by monitoring ventilation frequency (f_V) as described previously (Pan et al., 2019). Briefly, after light anesthesia using 0.05 mg ml⁻¹ Tris-buffered MS-222 (Millipore Sigma, Cat#: E10521), an individual larva was placed in a microcapillary tube with an inner diameter of 1 mm. System water containing MS-222 (as above) at 28.5°C was delivered to the microcapillary tube by gravity at a rate of 1.6–1.8 ml min⁻¹. A previous study demonstrated that this level of anesthesia did not significantly reduce the ventilatory response of zebrafish larvae to hypoxia (Jonz and Nurse, 2005). Each larva was allowed to adjust to the chamber for 10 min prior to the start of the trial. During each trial, f_V was recorded (iPhone SE mounted onto a Zeiss Discovery V8 dissecting microscope) for 5 min under baseline conditions, followed by 15 min of treatment. For hypoxia trials, an additional 10 min of recovery phase was recorded after re-establishing normoxia. To determine f_V, either buccal or opercular movements (depending on the orientation of the fish in the chamber and the visibility of the mouth and/or operculum) were counted for each minute. The videos were coded and viewed without knowledge of the experimental conditions. The f_V values obtained for each larva were then normalized by subtracting the average baseline f_V (f_V during the final minute of the 5 min baseline period) of the treatment group/genotype to which the larva belonged. For pharmacological studies, the baseline f_V (f_V during the final minute of the 5 min baseline period), stable f_V (mean f_V during the final 10 min of the treatment phase), and max f_V (maximum f_V during the treatment phase) were calculated for statistical analyses. For hypoxia trials, the baseline f_V (f_V during the final minute of the 5 min baseline period), the max f_V during the onset phase (first 5 min of the hypoxia treatment during which f_V peaked), the mean f_V during the stable phase (final 10 min of hypoxia treatment during which f_V was stable), and the mean f_V during the final 5 min of the recovery phase were calculated for statistical analysis.

We first examined the effects of β-AR agonists and antagonists on f_V in 4 dpf wild-type larvae. Each larva was exposed at the beginning of the treatment phase of the ventilation trial to either 1, 10 or 100 µmol l⁻¹ (−)-adrenaline (+)-bitartrate salt (Millipore Sigma, Cat#: E4375) or 100 µmol l⁻¹ of the β-AR agonist (−)-isoproterenol hydrochloride (Millipore Sigma, Cat#: I6504). For co-treatment with adrenaline and the β-AR antagonist propranolol, each larva was pretreated with 10 µmol l⁻¹ (±)-propranolol hydrochloride (Millipore Sigma, Cat#: P0884) for 30 min prior to undergoing the ventilation trial. During the baseline phase of the ventilation trial, the fish chamber was irrigated with 10 µmol l⁻¹ (±)-propranolol hydrochloride, whereas 100 µmol l⁻¹ (−)-adrenaline (+)-bitartrate salt along with 10 µmol l⁻¹ (±)-propranolol hydrochloride were used during the treatment phase of the trial. The agonists and antagonists used in the ventilation trials were dissolved in system water and stabilised with the antioxidant ascorbic acid (0.1 g l⁻¹; Millipore Sigma, Cat#: A92902) and adjusted to pH 7.0 with NaOH.

This set of experiments examined the effects of β-AR agonists on f_V in 4 dpf adrb1^−/− mutants. To control for genetic background (Zimmer et al., 2019a), the wild types (adrb1^+/+) used in these trials were ‘cousins’ of the adrb1^−/− mutants. To achieve this, an initial in-cross of adrb1^+/− heterozygous zebrafish was performed, and offspring screened to obtain breeding stocks of sibling adrb1^+/+ or adrb1^−/− zebrafish. Embryos used for experiments were then bred from the adrb1^+/+ or adrb1^−/− breeding stock. Each larva was exposed during the treatment phase of the ventilation trial to 100 μM (−)-isoproterenol hydrochloride. The preparation of reagents and the protocol for f_V measurements were identical to those described above.

To examine the role of β-ARs in the HVR, 4 dpf adrb1^+/+and adrb1^−/− larvae were exposed to hypoxia (30 mmHg) with or without 50 µmol l⁻¹ (±)-propranolol hydrochloride. A pilot trial conducted with wild-types exposed to hypoxia (30 mmHg) and 10 µmol l⁻¹ (±)-propranolol hydrochloride showed that the f_V response of larvae to 10 µmol l⁻¹ propranolol was highly variable, and thus propranolol concentrations were increased to 50 µmol l⁻¹ in the final experiments. In trials containing propranolol, each larva was pre-treated with 50 µmol l⁻¹ (±)-propranolol hydrochloride for 30 min; this concentration of propranolol was maintained throughout all phases of the ventilation trial. The preparation of reagents and the protocol for f_V measurements were identical to those described above.

The p2rx3b promoter region was PCR-amplified using Q5 high-fidelity DNA polymerase (NEB, Cat#: M0491S) from wild-type zebrafish larval DNA extracted using PureLink™ genomic DNA mini kit (Invitrogen, Cat#: K182002) following the manual's procedures. The GCaMP6s reporter gene sequence was PCR-amplified using Q5 high-fidelity DNA polymerase from Addgene plasmid #59531 courtesy of Misha Ahrens (Vladimirov et al., 2014). Primers for amplifying the promoter and reporter gene sequence are provided in Table S1. The p2rx3b promoter region and GCaMP6s reporter gene sequence were then cloned into a Tol2 expression vector containing the myl7:egfp heart-specific transgenesis marker in the reverse orientation using restriction cloning. All restriction enzymes (Table S1) used were purchased from New England Biolabs. Transposase mRNA was synthesized from linearized transposase plasmid using mMESSAGE mMACHINE SP6 kit (Invitrogen, Cat#: AM1340). Both the Tol2 expression vector and the transposase plasmid were courtesy of Marc Ekker (University of Ottawa). To generate founder transgenic fish and establish the transgenic line in a transparent Casper background to facilitate live imaging, one-cell-stage wild type×Casper embryos were injected with ∼1 nl of 100 ng μl⁻¹ synthesized plasmid and 200 ng μl⁻¹ transposase mRNA, and screened for cardiac green fluorescent protein (GFP) expression at 2 dpf. To establish the line, founders with cardiac GFP expression were raised to adults and crossed with Casper adults to screen for embryos expressing the corresponding reporter gene in a Casper background.

All experiments were performed on 4 dpf Tg(p2rx3b:GCaMP6s) larvae. Imaging of larvae was performed with an A1R MP+ confocal microscope (Nikon Instruments Inc. USA) with a 20× water-immersion objective (0.75 NA, time-lapse imaging). The microscope was further equipped with a custom-made water perfusion system enabling switching of perfusion solutions into the imaging chamber. The imaging chamber (∼4 ml) was made of glass with an O₂ sensor spot attached (Loligo Systems, Denmark) to allow for O₂ measurement within the chamber using the Witrox 4 O₂ meter and AutoResp software (Loligo Systems, Denmark). An individual larva was mounted on a coverslip with its side facing upwards in 1% low melting point agarose (BioShop, Cat#: AGA101.25). The coverslip was placed on top of the imaging chamber with the larva facing downwards into the chamber. This setup allowed for the chamber to be sealed and for the larva to be imaged through the coverslip once water flow was started. A diagram of the imaging setup is presented in Fig. S1. Each trial consisted of 5 min of baseline and 15 min of treatment. Hypoxia trials consisted of an additional 10 min recovery in normoxic water. During the treatment phase, the chamber was supplied with either water (control), water containing 100 µmol l⁻¹ (−)-adrenaline (+)-bitartrate salt, hypoxic water (30 mmHg), or hypoxic water containing 50 µmol l⁻¹ (±)-propranolol hydrochloride. In trials containing propranolol, each larva was pretreated with 50 µmol l⁻¹ (±)-propranolol hydrochloride for 30 min before undergoing the Ca²⁺ imaging trial. Propranolol was maintained in the water during all phases of the trial. All solutions used in the trials were prepared as above. To quantify Ca²⁺ activity, images of the ganglia of nerves IX and X were acquired at a rate of 30 frames s⁻¹ with a field of view of 512×512 or 512×256 pixels with a spatial resolution of 322×322 or 322×161 μm (x×y).

Image data were motion-corrected before exporting raw traces for regions of interest using CaImAn under the Mesmerize platform (Giovannucci et al., 2019; Kolar et al., 2021). Baseline corrections for raw traces were performed in R using the baseline package (Liland et al., 2010) to account for the decrease in fluorescence intensity of the sensor during the trial. Ca²⁺ events, defined as a peak in [Ca²⁺]_i with a minimum of continuous 5 s decreases in [Ca²⁺]_i on either side of the peak, were detected using the ‘find peaks’ function within the GCalcium package in R (https://CRAN.R-project.org/package=GCalcium) and Ca²⁺ event frequency and mean [Ca²⁺]_i were obtained using a custom-written R script. See supplementary Materials and Methods for further details on calcium imaging data analysis.

Statistical analyses were conducted using Sigmaplot (Systat Software) on normalized data to account for differences in baseline values between different batches of fish used over the span of three years (data for Figs 1, 3 were collected in fall 2019, data for Figs 2, 4–6 were collected in summer 2022). Two-factor comparisons among multiple means were performed using a two-way ANOVA. The Holm–Šidák post hoc test was used if a significant interaction was detected. Statistical significance was set at P<0.05. To analyse the effects of hypoxia on ƒ_V in β1-AR-knockout larvae, a three-way ANOVA was conducted with genotype, treatment and time as the three factors. Because the three-way ANOVA was significant, a Holm–Šidák post hoc test was used to perform pairwise comparisons. All summary statistics are presented in Table S2.

At all adrenaline concentrations used (1­­–100 µmol l⁻¹), wild-type fish responded rapidly with an increase in f_V, with maximal (peak) increases occurring within 5 min of application (Fig. 1A). This first response phase was followed by a period of elevated, but stable f_V (average Δf_V over the final 10 min of treatment). The ventilatory responses to adrenaline were concentration dependent, with higher levels causing greater increases in maximal or stable Δf_V (Fig. 1B).

Microscopic examination of the Tg(p2rx3b:GCaMP6s) transgenic fish demonstrated that at 4 dpf, cranial nerves IX and X extend projections dorsally into the hindbrain region of the CNS and ventrally to terminate within the pharyngeal arch region (Fig. 2A). Under control conditions during which time the larva was imaged while submerged in system water only, mean [Ca²⁺]_i levels in the ganglia of nerves IX and X were stable throughout the 20 min of imaging (Fig. 2B–C,E). The frequency of spontaneous Ca²⁺ spikes (f_Ca²⁺) did not change during the 20 min of imaging (Fig. 2B,D,F). The addition of adrenaline to the water did not alter the Ca²⁺ traces, with mean [Ca²⁺]_i and Δf_Ca²⁺ in ganglia of nerves IX and X (Fig. 2B–F) unchanged from control conditions. To ensure that the Ca²⁺ measurement system was operational, mean [Ca²⁺]_i was recorded during exposure of fish to hypoxia. The representative recording in Fig. 2B clearly indicated an increase in mean [Ca²⁺]_i during hypoxia.

The increased f_V associated with adrenaline treatment was mimicked by exposure to the β-AR agonist isoproterenol (Fig. 3). Maximal f_V and stable f_V were elevated by 133±9 and 85±9 breaths min⁻¹, respectively, during isoproterenol treatment, changes which were similar to those elicited by adrenaline (106±17 and 72±18 breaths min⁻¹; Fig. 3). Conversely, the increases in f_V during adrenaline treatment were abolished by treatment with the β-AR antagonist propranolol (Fig. 3). Note that in this experimental series, the sudden switch to water containing adrenaline or adrenaline with propranolol at 5 min was accompanied by a transient decrease in f_V.

Isoproterenol caused similar increases in f_V in adrb1^+/+ and adrb1^−/− larvae. Maximal f_V increased by 76±8 and 68±11 breaths min⁻¹, respectively whereas stable f_V was increased to 60±8 and 42±7 min⁻¹. Thus, there were no significant differences in f_V when comparing adrb1^+/+ and adrb1^−/− larvae (Fig. 4). Interestingly, adrb1^−/− mutants did exhibit a significantly higher f_V during baseline conditions when compared to adrb1^+/+ larvae (52±4 in adrb1^−/− mutants compared to 28±4 breaths min⁻¹ in wild-types; Fig. 4; Table 1).

During hypoxia, adrb1^+/+ larvae rapidly increased f_V to peak levels within 2–3 min of exposure (112±14 breaths min⁻¹), which then decreased to a stable level above baseline f_V (46±11 breaths min⁻¹). Propranolol treatment before and during hypoxia resulted in an attenuation of maximal f_V during the hypoxia onset phase (68±17 breaths min⁻¹) and prevented the increase in f_V during the stable phase (−5±8 breaths min⁻¹, Fig. 5). The response of knockout larvae to hypoxia was similar to the response of wild-type larvae: as in the wild type, peak f_V was attenuated by propranolol (123±19 breaths min⁻¹ without propranolol compared to 88±29 breaths min⁻¹ with propranolol) and the sustained increase in f_V was prevented (42±17 without propranolol compared to −1±8 breaths min⁻¹ with propranolol; Fig. 5). Interestingly, the elevated baseline f_V exhibited in β1-AR mutants was absent following exposure to propranolol (52±4 without propranolol compared to 24±6 breaths min⁻¹ with propranolol, Table 1).

When exposed to hypoxia, there was an increase in mean [Ca²⁺]_i within the ganglia of cranial nerves IX and X (Fig. 6B,D). There was no change in f_Ca²⁺ during the onset phase of hypoxia exposure, although f_Ca²⁺ declined to levels below baseline during the stable phase of hypoxia and during recovery. Co-exposure to hypoxia and propranolol did not alter Ca²⁺ dynamics when compared with that of hypoxia exposure alone (Fig. 6).

The larval zebrafish exhibits a marked HVR (Jonz and Nurse, 2005), which enhances hypoxia tolerance (Pan et al., 2019). In the current study, we assessed the adrenergic control of breathing in zebrafish larvae with a specific focus on establishing (i) the location (central versus peripheral) of the β-ARs mediating the ventilatory response to adrenaline, (ii) the role of the β1-AR sub-type, and (iii) whether catecholamines are required to initiate and/or modulate the short-term ventilatory response to hypoxia.

Using a combination of transgenic fish with cell-specific genetically encoded Ca²⁺ sensors to monitor Ca²⁺ signalling in ganglia of cranial nerves IX and X, mutants lacking the β1-AR (adrb1^−/−) and conventional pharmacological techniques, the results demonstrated that adrenaline elicits a dose- dependent β-AR-mediated increase in f_V that does not require expression of β1-ARs. In addition, β-ARs, although not required for the initiation of the HVR, appear to be involved in setting the maximal increase in f_V and perhaps more importantly, in maintaining hyperventilation during the later (stable) phase of the short-term HVR. We propose that the modulatory effects of β-adrenergic stimulation on the HVR are mediated within the CNS and are not reliant on activation of β1-ARs. These findings are the first to implicate catecholamines in the control of breathing during early development in fish.

Exposure of larval zebrafish to adrenaline resulted in a dose-dependent increase in f_V, which was probably mediated by β-ARs because specific stimulation of β-ARs using isoproterenol elicited a similar increase in f_V, whereas blocking β-ARs with propranolol abolished the adrenaline-mediated increase in ventilation. Because the experimental design involved exposing the entire larva to adrenergic receptor agonists (e.g. adrenaline), it was not possible to distinguish between their peripheral versus central effects given that both peripheral and central ARs presumably would be activated. Indeed, pharmacological agents such as metronidazole with a similar a molecular mass to adrenaline, were shown to penetrate the CNS (Mathias et al., 2014). The experimental approach used in the present study is distinctly different from direct injection of adrenaline into the circulatory system as was done routinely in previous studies (see table 1 in Pan and Perry, 2020). The latter approach targets only peripheral ARs because adrenaline is unable to pass through the blood brain barrier, at least in goldfish (Carassius auratus) and rainbow trout (Busacker and Chavin, 1977; Nekvasil and Olson, 1986). Thus, in an attempt to better clarify the roles of peripheral versus central β-ARs, we examined the calcium dynamics in ganglia of cranial nerves IX and X during exposure of larvae to adrenaline. Expression of genetically encoded calcium indicators in these ganglia was driven by the p2rx3b promoter that has been shown to label peripheral sensory neurons (Kucenas et al., 2006), which extend into the pharyngeal arch region where they wrap around 5-HT positive NECs (Rosales et al., 2019 preprint), the presumptive O₂ chemoreceptors in fish. In addition, cranial nerves IX and X are thought to initiate most cardiorespiratory responses to hypoxia including the HVR (reviewed by Milsom, 2012). Thus, any ventilatory signals that are initiated in the periphery are likely transmitted to the CNS via cranial nerves IX and X. Because calcium participates in the transmission of depolarizing signals within and between neurons (Brini et al., 2014), the expectation is an increase in calcium activity within cranial nerves IX and X if ventilatory signals are transmitted from the periphery to the CNS. Indeed, such a response was observed when fish were exposed to hypoxia. However, our data showed that there was no difference in calcium dynamics within the ganglia of cranial nerves IX and X of larvae exposed to adrenaline compared with the controls, indicating indirectly that central β-ARs are likely to be responsible for the increase in ventilation caused by catecholamines at 4 dpf.

Previous experiments that examined the adrenergic control of breathing were not designed to evaluate the role of specific β-AR sub-types in mediating the ventilatory adjustments. To address this question, the ventilatory responses of adrb1^−/− mutant fish to β-AR activation (using isoproterenol) was assessed. The β1-AR was targeted for knockout owing to its high expression (relative to β2-ARs) in larval zebrafish as well as its high expression within tissues sensitive to hypoxia (e.g. gill, brain and heart) in adult zebrafish (Steele et al., 2011). Results obtained using the adrb1^−/− mutants indicate that the increased f_V accompanying the activation of β-ARs does not require the presence of the β1-AR sub-type.

Although the hyperventilatory response to β-AR stimulation was unaltered in the β1-AR-deficient fish, we cannot exclude the involvement of the β1-AR in wild-type fish in which all β-AR sub-types are present. The continued increase in f_V in the adrb1^−/− larvae merely indicates that the remaining β-AR sub-types are sufficient to promote the hyperventilatory response in the absence of β1-AR (i.e. a functional β1-AR is not required to elicit the increase in f_V). If the β1-AR is the predominant sub-type underlying the hyperventilatory response in the wild type, it is possible that β-AR redundancy allows the role of the β1-AR to be replaced by other β-ARs in the adrb1^−/− mutants. Similar arguments can be applied to the interpretation of the role of the β1-AR in the HVR (see below). This type of genetic compensation is not uncommon in loss-of-function mutants (Rossi et al., 2020); a previous study showed that at least within the heart, adrb1^−/− mutants exhibit elevated expression of β2- and β3-AR mRNA (Joyce et al., 2022). Thus, the relative importance of the β1-AR relative to the other sub-types in promoting the β-AR-mediated increases in f_V in wild-type fish remains uncertain and warrants further investigation.

As shown previously in larval zebrafish (Mandic et al., 2019; Pan et al., 2021, 2019), the HVR is multiphasic, with f_V reaching a maximum during the onset of hypoxia and then gradually declining to a stable level above normoxic conditions. Previous studies examining the role of catecholamines in modulating the HVR yielded mixed results, with propranolol attenuating the HVR in rainbow trout (Aota et al., 1990) whereas neither α nor β-AR antagonists altered the HVR in the Atlantic cod (Gadus morhua) (Kinkead et al., 1991). Our data from larval zebrafish showed that when co-exposed to hypoxia and propranolol, the peak f_V during the hypoxia onset phase was attenuated whereas the increase in f_V during the stable phase was abolished, suggesting a role for β-ARs in maintaining, but not initiating the HVR. However, similarly to hyperventilation evoked by catecholamine exposure, the β1-AR is not required for the HVR; both adrb1^+/+ and adrb1^−/− larvae exhibited identical ventilatory responses to hypoxia. As discussed previously for the catecholamine exposure experiments, we cannot dismiss the involvement of the β1-AR in wild-type fish in which all β-AR sub-types are present.

Previous studies probably examined only the peripheral effects of ARs through intravascular delivery of drugs, as the brain specifically excludes catecholamines, particularly adrenaline (Busacker and Chavin, 1977; Nekvasil and Olson, 1986), whereas our experimental approach examined ARs located both in the periphery and the CNS. However, the possibility of AR antagonists passing through the blood-brain barrier and affecting central ARs in previous studies cannot be ruled out.

Data obtained from calcium imaging of the ganglia of cranial nerves IX and X during hypoxia and propranolol treatment support our contention that β-ARs within the CNS modulate the HVR. Under hypoxia exposure, mean [Ca²⁺]_i increased gradually following the onset of hypoxia and remained elevated during the entire duration of exposure; f_Ca²⁺ was not elevated by hypoxia. Thus, it is possible that the mechanism for transmitting the HVR signal from the peripheral O₂ chemoreceptors to the CNS is via an elevation of mean [Ca²⁺]_i within cranial nerves IX and X. However, additional research will be required to establish a direct cause and effect relationship between the increase in mean [Ca²⁺]_i and the HVR. Regardless, the calcium dynamics under hypoxic conditions was not altered by propranolol, suggesting that the effects of propranolol on the HVR are mediated within the CNS. Because f_V was merely attenuated by propranolol during the onset phase of the HVR, it would appear that: (i) the initiation signal for the HVR from peripheral chemoreceptors does not require β-ARs and (ii) β-ARs within the CNS modulate this initiation signal by increasing f_V. This modulatory effect is also apparent in the stable phase of hypoxia exposure, when other mechanisms presumably are activated to provide an inhibitory signal to lower f_V to a stable level while remaining above normoxic levels. We suggest that by eliminating the excitatory signal from β-ARs, the inhibitory signals dominate the HVR resulting in the decrease of f_V during the stable phase of the HVR. Because propranolol prevented the increased f_V during the stable phase of the HVR in both adrb1^+/+ and adrb1^−/− larvae, it is apparent that activation of β2- and/or β3-ARs in the CNS provides an excitatory signal on ventilation during the stable phase of hypoxia exposure. This may also explain the elevated baseline ventilation in adrb1^−/− mutants that is absent after propranolol treatment because these mutants might have elevated expression of excitatory β2- and β3-ARs, as discussed previously.

Taken together, the results of the current study demonstrate that the HVR in zebrafish larvae is regulated by catecholamines interacting with β-ARs. Because adrenaline was ineffective at altering Ca²⁺ activity in the sensory neurons that are stimulated during hypoxia, we suggest that β-AR modulation of the HVR is mediated via catecholamines released and acting within the CNS. Conceivably, β-ARs could be located within the central pattern generator or the respiratory motor neurons because direct injection of catecholamines in the vicinity of these structures caused a marked change in the pattern of central respiratory drive measured as bursting activity in branchial nerves (Randall and Taylor, 1991). Additional studies are needed to clarify the role of β-ARs in modulating ventilation and to provide direct evidence for the central modulation of the HVR by catecholamines.

We thank the University of Ottawa aquatics facility staff for zebrafish care and the microscopy core for the help with in vivo calcium imaging.