A role for dopamine in control of the hypoxic ventilatory response via D2 receptors in the zebrafish gill

Dopamine is a neurotransmitter involved in oxygen sensing and control of reflex hyperventilation. In aquatic vertebrates, oxygen sensing occurs in the gills via chemoreceptive neuroepithelial cells (NECs), but a mechanism for dopamine in autonomic control of ventilation has not been defined. We used immunohistochemistry and confocal microscopy to map the distribution of tyrosine hydroxylase (TH), an enzyme necessary for dopamine synthesis, in the gills of zebrafish. TH was found in nerve fibers of the gill filaments and respiratory lamellae. We further identified dopamine active transporter (dat) and vesicular monoamine transporter (vmat2) expression in neurons of the gill filaments using transgenic lines. Moreover, TH‐ and dat‐positive nerve fibers innervated NECs. In chemical screening assays, domperidone, a D2 receptor antagonist, increased ventilation frequency in zebrafish larvae in a dose‐dependent manner. When larvae were confronted with acute hypoxia, the D2 agonist, quinpirole, abolished the hyperventilatory response. Quantitative polymerase chain reaction confirmed expression of drd2a and drd2b (genes encoding D2 receptors) in the gills, and their relative abundance decreased following acclimation to hypoxia for 48 h. We localized D2 receptor immunoreactivity to NECs in the efferent gill filament epithelium, and a novel cell type in the afferent filament epithelium. We provide evidence for the synthesis and storage of dopamine by sensory nerve terminals that innervate NECs. We further suggest that D2 receptors on presynaptic NECs provide a feedback mechanism that attenuates the chemoreceptor response to hypoxia. Our studies suggest that a fundamental, modulatory role for dopamine in oxygen sensing arose early in vertebrate evolution.


INTRODUCTION
An animal's ability to detect low oxygen, or hypoxia, and respond appropriately is crucial for survival.The hypoxic ventilatory response (HVR) in vertebrates is an important physiological response to low partial pressure of oxygen (PO 2 ), which involves compensatory changes in ventilation to raise arterial PO 2 (Perry et al., 2009;Powell et al., 1998).In the mammalian carotid body, dopamine is an important neuromodulator involved in oxygen sensing and control of reflex hyperventilation.Hypoxic activation of carotid body oxygen chemosensory type 1 (glomus) cells involves the modulation of K + conductance, membrane depolarization, and resulting Ca 2+ -dependent neurotransmitter release to act on sensory terminals of the carotid sinus nerve (González et al., 1994;Kumar & Prabhakar, 2012;López-Barneo et al., 1988;Nurse, 2010).Dopamine, released by type 1 cells, has been shown to have autocrine-paracrine actions on carotid body activation in many species via G-protein-coupled dopamine D 2 receptors on both presynaptic type 1 cells and postsynaptic afferent terminals (Nurse, 2010).
At the presynaptic type 1 cell, the inhibitory actions of dopamine act to decrease further neurotransmitter release, thereby mediating carotid body hypoxia signaling (Benot & López-Barneo, 1990).A role for dopamine in mediating the ventilatory responses to hypoxia in aquatic vertebrates is less understood; however, studying systems in ancestral vertebrates allows us a unique look into the evolutionary origin of such an important physiological response and how it has evolved across different environments.
In fish, oxygen sensing occurs in the gills via chemoreceptive neuroepithelial cells (NECs; Figure 1).As water flows through the gill basket during breathing, NECs sense changes in oxygen and initiate HVR during hypoxia (Perry et al., 2009).NECs are found in the filament epithelium near the efferent filament artery and respond to a decrease in PO 2 by inhibition of background K + channels, membrane depolarization, and Ca 2+ -dependent vesicular recycling (Jonz et al., 2004;Zachar et al., 2017a).The latter is consistent with neurotransmitter release into the synaptic cleft, which may lead to activation of sensory nerve fibers.NECs are polymodal, such that they are sensitive to changes in O 2 , CO 2 , and H + (Abdallah, Jonz, et al., 2015;Jonz, 2018;Qin et al., 2010) and are characterized by immunoreactivity to serotonin (5-hydroxytryptamine [5-HT]) and synaptic vesicle protein, SV2 (Jonz & Nurse, 2003).Due to these characteristics, and the homology between the site of the carotid body and the first gill arch in fish, NECs are believed to be structurally and functionally similar to type 1 cells (Milsom & Burleson, 2007;Zachar & Jonz, 2012).Recent evidence, however, suggests that NECs may be homologues of oxygensensitive pulmonary neuroepithelial bodies (NEBs; Hockman et al., 2017).
In contrast to the well-described neurochemistry of the carotid body (Nurse, 2010), currently there is no direct evidence for which neurotransmitters are released within the gill during hypoxia, and the receptors they act on remain uncharacterized.5-HT, the most abundant neurotransmitter in gill NECs, and acetylcholine (ACh) have both been suggested as excitatory neurotransmitters in the gills (Jonz & Nurse, 2003;Porteus et al., 2012;Zachar et al., 2017b).Early evidence for a role of dopamine in the gills was shown in isolated gills of rainbow trout (Oncorhynchus mykiss), where dopamine caused a small, brief burst in chemoreceptor activity followed by a mild inhibition of receptor discharge (Burleson & Milsom, 1995).In live, whole-animal experiments using zebrafish larvae, application of dopamine decreased ventilation frequency, suggesting inhibitory effects of dopamine on ventilation (Shakarchi et al., 2013).Recent single-cell RNA-sequencing in zebrafish indicated expression of D 2 -like receptors in gill cells, including NECs (Pan et al., 2022); however, a mechanism for control of ventilation during hypoxia via an inhibitory dopaminergic pathway in the gill has not been defined.
The goal of the present study was to evaluate a potential role for dopamine in hypoxia signaling in zebrafish by establishing a site for the synthesis and storage of dopamine in the gill and characterizing the distribution of D 2 receptors.We identify the presence of tyrosine hydroxylase (TH) and the dopamine active transporter (DAT), involved in the synthesis or transport of dopamine, in sensory nerve fibers using immunohistochemistry and transgenic lines.In addition, we co-localize dopamine D 2 receptors with oxygen-sensitive NECs and provide evidence for presynaptic D 2 receptors in the modulation of the hyperventilatory response to hypoxia.

Ethics statement
All wild-type and transgenic zebrafish were bred and maintained at the Laboratory for the Physiology and Genetics of Aquatic Organisms, University of Ottawa.Zebrafish were kept at 28 • C on a 14:10-h light:dark cycle (Westerfield, 2007).Larvae were obtained by breeding from 12month-old adults and maintained in 150-mm Petri dishes until 2 days postfertilization (d.p.f.).All procedures for animal use were carried out in accordance with institutional guidelines according to protocol BL-3666 and guidelines provided by the Canadian Council on Animal Care.
Adult zebrafish were euthanized by concussion and decapitated.Larvae were euthanized by hypothermic shock by immersion in an ice bath for 20 min.

Immunohistochemistry and antibody characterization
Techniques for tissue extraction and immunolabeling were carried out as previously described (Jonz & Nurse, 2003).Whole gill baskets were removed and immersed in phosphate-buffered solution (PBS) containing (mM): NaCl 137, Na 2 HPO 4 15.2, KCl 2.7, and KH 2 PO 4 1.5 at pH 7.8 (Bradford et al., 1994).Gill baskets were fixed by immersion in 4% PBS (pH 7.8).After three rinses in PBS, gill baskets were then separated into individual arches.Gill arches were incubated in primary antibodies for 24 h at 4 • C, rinsed with PBS three times at 3 min, and immersed in secondary antibodies for 1 h at room temperature in darkness.
Dissection and removal of gill arches are illustrated in Figure 1a-c.
no. 111-095-003; Cedarlane).zn-12 and SV2 antibodies were monoclonal and raised in mouse (RRID: AB_2315387 and RRID: AB_531908; Developmental Studies Hybridoma Bank, University of Iowa).zn-12 targets membrane fractions from adult zebrafish CNS and recognizes an HNK-1-like epitope (manufacturer specifications).Western blot analysis demonstrated that zn-12 and HNK-1 antibodies label similar bands ranging in molecular weight from 60 to 248 kDa (see Metcalfe et al., 1990).SV2 antibodies were raised against synaptic vesicles from the elasmobranch electric organ and bind to a transmembrane glycoprotein of ∼95 kDa on the cytoplasmic side of synaptic vesicles in endocrine and neurosecretory cells (Buckley & Kelly, 1985;manufacturer specifications).SV2 and zn-12 were used at 1:100 and targeted by goat anti-mouse secondary antibodies conjugated with Alexa 594 at 1:100 (cat.no.A11005; Invitrogen).Labeling by these antibodies in the zebrafish gill has been previously characterized (Jonz & Nurse, 2003).
Cells and nerve fibers of the filament epithelium expressing D 2 receptors were identified using polyclonal antibodies at 1:100 that targeted amino acids 24−34 near the ligand-binding domain of rat D 2 receptor (Tomé et al., 2004;cat.no.AB1558; Millipore; RRID: AB_90775).Prior to antibody incubation, antigen retrieval was performed to facilitate anti-D 2 binding.Gill baskets were immersed in a buffer containing 10 mM sodium citrate and 0.05% Tween 20 at pH 6.0, then incubated for 10 min at 98 • C. D 2 receptor immunoreactivity was observed using goat anti-rabbit secondary antibodies conjugated with FITC (1:50).To control for the specificity of D 2 in our preparation, we generated a custom peptide corresponding to the immunogen sequence (GSEGKADRPHY; Millipore).Preadsorption of the control peptide with the D 2 antibodies blocked all immunolabeling in gill tissue.
All antibodies were diluted using PBS-TX.Gill arches were rinsed with PBS, mounted onto glass microscope slides with Prolong Diamond antifade mountant (Thermo Fisher Scientific), and covered with a 1.2mm glass coverslip.In some experiments, D 2 -positive cells were further stained with the nuclear dye, 4′,6-diamidino-2-phenylindole (DAPI).

Transgenic lines
Two lines of transgenic zebrafish were used in the present study.The Tg(dat:tom20 MLS-mCherry) line was used to visualize dopaminergic neurons.In this line, the regulatory elements of the dopamine transporter gene (dat) were targeted to a reporter, mCherry, after fusion with the mitochondrial localizing signal (MLS) of Tom20 (Noble et al., 2015).Transgenic ET(vmat2:GFP) zebrafish were previously described by Wen et al. (2008)  slc18a2) and was used to visualize NECs (Pan et al., 2021).VMAT2 is an integral membrane protein that mediates storage of monoamines, such as dopamine and 5-HT, into synaptic vesicles.To determine the distribution of dopaminergic expression relative to serotonergic NECs in these lines, adult homozygous Tg(dat:tom20 MLS-mCherry) fish were crossed with homozygous adults from ET(vmat2:GFP) to generate double transgenic offspring containing both mCherry and GFP reporter genes.Double transgenic fish were raised to 3 months before harvesting gill tissue.Complete gill arches were mounted onto microscopic slides without additional staining.

Confocal microscopy
Whole-mount preparations were examined using an upright micro-

Chemical screening
Chemical screening assays were carried out as previously described (Coe et al., 2017;Rhabar et al., 2016).Larvae ranged from 14 to 16 d.p.f.Normoxic f v was obtained after larvae had settled for 10 min.Larvae were exposed to hypoxic solution, the addition of 50 μL of a test drug, or a combination of hypoxia and drug.All solutions contained 0.04 mg/mL tricaine.For hypoxia, the plate was placed in a hypoxic incubator (Forma 3110; Thermo Fisher Scientific) for 7 min, into which 100% N 2 was injected to reduce O 2 levels to 1% (8 mmHg).
PO 2 was measured with a built-in thermal conductivity O 2 sensor.In other experiments, the D 2 antagonist, domperidone (cat.no.D122; Sigma-Aldrich), or the D 2 agonist, quinpirole (cat.no.Q102; Sigma-Aldrich), was tested.Domperidone was applied at concentrations from 50 to 300 μM for 2 min, whereas quinpirole was co-administered with hypoxia at concentrations from 10 to 200 μM.Both drugs target D 2 receptors in the mammalian carotid body (Carroll et al., 2005;Tomares et al., 1994).Domperidone and quinpirole are poorly soluble in water; therefore, they were first dissolved in dimethyl sulfoxide (DMSO) and diluted to produce a final concentration of <0.5% DMSO.For controls, we included 0.5% DMSO in solution when hypoxia was tested alone (see Figure 5c).We previously showed that DMSO at or below this concentration had no effect upon f v (Coe et al., 2017).For recovery, each larva was transferred to a new well in normal solution and f v was recorded after 5 min.
Analysis of variance (ANOVA) with repeated measures was used to compare f v between control, treatment, and recovery at each concentration.A Bonferroni post hoc test was then used to determine whether mean f v was significantly different between control and treatment groups (p < .05).Curves for dose versus response were prepared to display concentration-dependent trends and to estimate the effective concentration of each drug that gave a half-maximal f v response (EC 50 ).Percent maximum f v for domperidone (at 300 μM) and percent maximum inhibition for quinpirole (at 150 μM) were calculated to normalize the data following procedures described in Rahbar et al. (2016) and Coe et al. (2017).For estimation of EC 50 , a line constrained from the origin at zero to the maximal response was fit to the data using a nonlinear ligand binding model with least squares following the equation: y = b max ⋅x h /(k d h + x h ) (Prism v5.0, RRID: SCR_002798; GraphPad Software Inc.).

Acclimation to chronic hypoxia
A total of 64 adult zebrafish were used in acclimation experiments.
Groups of eight zebrafish were placed in 2.5-L tanks and acclimated to hypoxia (35 mmHg) for 24 h, 48 h, 72 h, or 7 days.Groups of eight control zebrafish were simultaneously maintained in identical tanks at normoxia (∼160 mmHg) for the same periods of time.For acclimated fish, water PO 2 was gradually lowered over 8 h at a rate of ∼16 mmHg/h by introducing a mixture of compressed air and 100% N 2 from a gas mixer (Pegas 4000 MF; Columbus Instruments) and delivered through a porous air stone.Water temperature was kept at 28 • C by placing tanks in a temperature-controlled water bath, and 50% water changes were performed in each tank every other day.

Aquatic surface respiration assays
Experiments assessing the effects of hypoxic acclimation on aquatic surface respiration (ASR) behavior were performed following Abdallah, Thomas, et al. (2015).After 24-h, 48-h, 72-h, or 7-day acclimation to hypoxia (see Section 2.6), fish were allowed to recover in normoxic water for 1 h.A 1-L test chamber was marked with gradations on the outside measuring the distance (in mm) from the bottom to the air-water interface, or surface.100% N 2 was delivered to the test chamber via polyethylene tubing and a porous stone until the desired water PO 2 was achieved.Water PO 2 ranged from ∼160 to 10 mmHg and was verified before and after each trial with an O 2 meter (Model 550A; YSI).Individual zebrafish were introduced into the test chamber and were allowed to recover from handling stress for 5 min.A 5-min trial was then performed, after which zebrafish were transferred to a normoxic tank for an additional 5 min for recovery.Zebrafish were reintroduced to the test chamber after recovery and an additional trial was performed at a lower PO 2 .In this manner, zebrafish were exposed to successive bouts of progressively more severe acute hypoxia (100, 50, 30, and 15 mmHg).Fish taking on a posture in which the dorsal aspect of the head lied just at or below the surface with the body at a slight angle (as described by Kramer & McClure, 1982;Perry et al., 2009;Timmerman & Chapman, 2004) were considered to be performing ASR.Videos were analyzed blindly by observation post hoc, and the cumulative time (s) spent in ASR at each level of hypoxia was counted manually.

Quantitative polymerase chain reaction
Whole gill baskets were removed from zebrafish acclimated to hypoxia or normoxic controls after 48 h, as described in Section 2.6, and immediately frozen on dry ice.Total RNA from gill baskets was extracted using Trizol reagent (Life Technologies) and quantified using a Nan-oDrop 2000c UV-Vis Spectrophotometer (Thermo Fisher Scientific).
cDNA was generated using 1000 ng of RNA from whole gill baskets using QuantiTect Reverse Transcription Kit (Qiagen) following manufacturers protocol.To check for genomic DNA contamination, a no-template negative control and a no reverse transcriptase negative control were included.
mRNA gene expression of dopamine receptor-2 (drd2a and drd2b) in the gill was assessed by real-time reverse transcription quantitative polymerase chain reaction (RT-qPCR) using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad).Expression of reference gene, elongation factor 1a (ef1a), was stable between gill baskets of normoxia and hypoxia groups and was therefore used to normalize drd2a and drd2b mRNA expression.Standard curves were generated using a serial dilution of pooled cDNA to optimize primer reaction conditions.Quantitative polymerase chain reaction (qPCR) included 1 μL cDNA, 1 μL specific forward primer, 1 μL specific reverse primer, 7 μL nuclease-free water, and 10 μL Universal SYBR Green Supermix for a total reaction volume of 20 μL.Each reaction included an initial step at 98 • C to activate the enzymes in the mix, followed by 40 repeats of a denaturation step at 95 • C and an annealing/extension step at the optimized temperature for each primer pair (Table 1).A melt step from 66 to 95 • C in 0.5 • C increments was included at the end of the reaction to check the specificity of the amplicon produced.Each biological sample was run in triplicate, and the relative abundance of mRNA was calculated using the ΔΔC t method (Livak & Schmittgen, 2001).Amplification efficiencies and R 2 values are included in Table 1.Statistical analysis was carried out using the Mann-Whitney U test with Prism software.

Dopaminergic innervation in the zebrafish gills
Gills were removed from zebrafish and prepared as whole mounts to observe NECs and associated innervation (Figure 1d,e).Immunohisto-TA B L E 1 Dopamine receptor primer pair conditions used for mRNA quantification by real-time reverse transcription polymerase chain reaction.Note: Primer sequences for dopamine receptors drd2a and drd2b were designed using Primer3 software.ef1a was used as a reference gene.Abbreviations: F, forward primer; R, reverse primer.

Gene
chemical labeling with anti-TH indicated extensive localization of the enzyme throughout the nerve supply in the gills (Figure 2).TH-positive nerve fibers emanated from the gill arches and coursed through the gill filaments via superficial and deep nerve bundles, located above and below the efferent filament artery (eFA), respectively (arrowheads, Figure 2a,b,d,e).These nerves formed a plexus in the filaments that encircled the eFA and extended to the interlamellar regions and the respiratory lamellae (Figure 2d,e).In addition, TH-positive nerve fibers were closely associated with 5-HT-positive NECs of the filaments and lamellae (arrows, Figure 2a,c,d,f).To confirm whether the identified TH-positive nerve fibers were similar to those previously identified in the zebrafish gills, which are predominantly of extrabranchial origin (Jonz & Nurse, 2003), we used the zebrafish-specific neuronal antibody, zn-12.TH-positive nerve fibers of the filaments and lamellae were co-labeled with both markers (arrowheads, Figure 3).Chain neurons (ChNs; see Jonz & Nurse, 2003), as part of the intrabranchial deep nerve bundle, were also visible and were positive for both TH and zn-12.
In order to further examine dopaminergic activity in the gills, we produced a line of transgenic zebrafish by crossing the Tg(dat:tom20 MLS-mCherry) and ET(vmat2:GFP) lines (Figure 4).This allowed visualization of cells or nerve fibers expressing dat (Noble et al., 2015) and vmat2 (Pan et al., 2021).Nerve fibers expressing dat were observed in the gill filaments, including the interlamellar regions (arrowheads, Figure 4a,b,d,e) and surrounding the eFA (Figure 4g,h).
The pattern of dat expression in nerve fibers was punctate.This may have been due to restriction of the reporter gene, mCherry, to mitochondria in cells that express dat (Noble et al., 2015).Mitochondria may be highly concentrated in sensory nerve terminals, including those that innervate oxygen chemoreceptors (Bollé et al., 2000;González et al., 1994).dat expression was not detected in nerve fibers of the lamellae.NECs of the gill filaments and lamellae expressed vmat2, but only NECs of the filaments came in close contact with dat nerve fibers (arrows, Figure 4a,c,d,f).Intrabranchial ChNs were also found to have vmat2 expression (Figure 4d,f,g,i).
Notably, we did not observe expression of dat in intrabranchial neurons.

Targeting dopamine D 2 receptors affected the ventilatory response
Chemical screening assays were used to assess D 2 receptors as putative targets of endogenous dopamine release.Exposure of zebrafish to domperidone to block D 2 receptors had no effect at low concentrations, but significantly increased f v at higher concentrations, with a ∼1.6-fold change above control values at 200 and 300 μM (ANOVA, Bonferroni, p < .05; Figure 5a).The effects of domperidone were dose dependent with an estimated EC 50 of 30 μM (Figure 5b).Since D 2 blockade caused an increase in f v , as does hypoxia (Rahbar et al., 2016), we hypothesized that activation of D 2 receptors would have the opposite effect.We therefore administered quinpirole while animals were exposed to acute hypoxia.Zebrafish displayed a robust ventilatory response to hypoxia alone (with 0.5% DMSO included for control), where f v was significantly elevated from 36.8 ± 6.4 to 92.4 ± 12.5 breaths per minute following acute exposure (ANOVA, Bonferroni, p < .05; Figure 5c).At all concentrations tested, quinpirole abolished the hyperventilatory response to acute hypoxia, and even suppressed f v below control values at 150 and 200 μM (Figure 5c).The EC 50 for quinpirole was estimated to be <10 μM (Figure 5d).

Acclimation to hypoxia reduced D 2 gene expression
Previous studies from single-cell RNA-sequencing in zebrafish indicated expression of D 2 -like receptors in gill cells, including NECs (Pan et al., 2022).In the present study, we sought to determine whether acclimation to hypoxia would change expression of paralogous genes encoding D 2 receptors (drd2a and drd2b), thereby implicating these receptors in the hypoxic response.To first determine a relevant timepoint of hypoxia acclimation to evaluate such changes in gene expression, we exposed hypoxia-acclimated and control zebrafish to progressive bouts of acute hypoxia and evaluated the amount of time spent performing ASR behavior (Figure 6).Acclimation to hypoxia for 48 h significantly reduced mean ± SEM cumulative time spent performing ASR during acute hypoxic exposure, compared with The EC 50 for quinpirole could not be reliably determined but was estimated at <10 μM.
unacclimated zebrafish, at PO 2 values below 30 mmHg (Mann-Whitney U test; p < .01; Figure 6).This reduction was not significant in fish acclimated to hypoxia for less than 48 h acclimation, thereby highlighting 48 h as a physiologically relevant timepoint of hypoxia exposure to evaluate changes in gene expression.Zebrafish were then exposed to hypoxia (35 mmHg) for 48 h, and qPCR was used to evaluate relative changes in drd2a and drd2b expression (Figure 7).We found a significant reduction by 3.08-fold in drd2a compared to control fish (Mann-Whitney U test, p < .01; Figure 7), as well as a significant reduction by 3.04-fold in drd2b (Mann-Whitney U test, p < .01; Figure 7).
Expression of the reference gene, ef1a, was stable between gill baskets of normoxia and hypoxia groups (Figure 7).

Immunohistochemical localization of D 2 receptors in the gills
Immunohistochemical labeling of anti-D 2 -R revealed localization of the receptor primarily to NECs of the efferent gill epithelium, and to nerve fibers of the lamellae (Figure 8).D 2 -R labeling was combined with anti-SV2 and zn-12 in order to simultaneously observe co-localization of D 2 -R with NECs and nerve fibers, respectively.Although NECs and nerves were both labeled by antibodies raised in mouse (and appear in magenta), their markedly different morphology allowed them to be easily distinguished from each other.D 2 -R co-localized with SV2 in NECs along the efferent filament epithelium (arrows, Figure 8a-c).Transverse optical sectioning and rotation of confocal images showed D 2 -R-positive NECs were located superficially to, and in close apposition with, the eFA (Figure 8d-i).Co-labeling with these antibodies also revealed sparse D 2 -R/zn-12-positive nerve fibers of the filament and prominent D 2 -R/zn-12-positive nerve fibers of the lamellae (arrowheads, Figure 8a-c).
Co-application of anti-D 2 -R with the nuclear marker, DAPI, identified D 2 -R labeling of cells of the afferent gill epithelium (Figure 9a,b).D 2 -R-expressing cells surrounded the afferent filament artery (aFA) in transverse optical section (Figure 9c,d).No labeling of SV2 was observed in the afferent filament epithelium (not shown).

F I G U R E 6
Acclimation to chronic hypoxia reduced the aquatic surface respiration (ASR) response to acute hypoxia.Adult zebrafish previously acclimated to hypoxia for various timepoints (24 h, 48 h, 72 h, or 7 days) were exposed to successive bouts of progressively more severe acute hypoxia (100, 50, 30, and 15 mmHg).Data points for all groups were connected with a continuous line for clarity.Acclimation to hypoxia for 48 h, 72 h, and 7 days significantly reduced mean ± SEM cumulative time in ASR during acute hypoxic exposure (squares) at PO 2 values below 30 mmHg, compared with unacclimated zebrafish (circles; Mann-Whitney U test; p < .01;N = 8 in control and acclimated groups for each PO 2 and each period of acclimation).

DISCUSSION
The present study demonstrates a role for dopamine in the gills of zebrafish as a modulator of the hyperventilatory response to hypoxia.
We used immunohistochemistry, confocal microscopy, and transgenic markers to localize elements of dopamine synthesis and transport to sensory nerve fibers that innervate NECs in the gill filaments.We further integrated pharmacological and molecular approaches to reveal presynaptic expression of D 2 receptors in chemoreceptive NECs and associated innervation.

Synthesis and storage of dopamine in nerve fibers of the gills
TH is a rate-limiting enzyme required for the synthesis of catecholamines, including dopamine.Localization of TH can therefore be used to elucidate potential sites of dopamine production in the gill.The presence of TH in gills appears to be heterogeneous across fish species.
In catfish (Ictalurus punctatus), isolated 5-HT-positive NECs were colabeled with antibodies against TH, suggesting that NECs may release 5-HT and dopamine in that species (Burleson et al., 2006).By contrast, a lack of TH labeling in the gills of goldfish (Carassius auratus), trout (O.mykiss), and Indian catfish (Heteropneustes fossilis) leaves a role for dopamine in these species less clear (Porteus et al., 2013;Zaccone et al., 2003).Expression of th2, one of the genes encoding TH, was identified in the gill region of zebrafish embryos using in situ hybridization, but the cells expressing the enzyme were not identified (Chen et al., 2009).
In the present study, we detected anti-TH labeling and dopamine active transporter (dat) expression in nerve fibers throughout the gill filaments and respiratory lamellae.Interestingly, many TH-and dat-positive nerve fibers were closely associated with NECs, and a previous study described these as afferent fibers of ganglionic neurons that carry chemosensory information to the central nervous system during hypoxic stimulation of NECs (Jonz & Nurse, 2003).By comparison with the rat carotid body, Finley et al. (1992) described afferent nerve fibers of the petrosal ganglion that innervate type I cells and contained both TH and DOPA decarboxylase (DDC), a second enzyme involved in dopamine synthesis.Although the primary source of dopamine in the carotid body is usually regarded as the type I cell (González et al., 1994;Itturiaga et al., 2009;Nurse, 2010) also present in NECs (Pan et al., 2021(Pan et al., , 2022)).Both are involved in the synthesis and storage of dopamine and serotonin.

Dopamine D 2 receptors mediate the response to hypoxia
Early evidence for a role of dopamine receptors in control of the HVR in fish was provided from isolated gills of rainbow trout, where dopamine caused a small and brief burst in chemoreceptor activity followed by a mild inhibition of receptor discharge (Burleson & Milsom, 1995).More recently, in zebrafish, exogenous application of dopamine decreased ventilation frequency as early as 7 d.p.f.(Shakarchi et al., 2013), and genes encoding members of the D 2 -like receptor family were expressed in gill NECs and neurons (Pan et al., 2022).In the carotid body, D 2 is the dominant type of dopamine receptor involved in modulation of the response to hypoxia (Carroll et al., 2005;Gauda, 2002;Itturiaga et al., 2009;Mir et al., 1984;Nurse, 2010).We therefore used a chemical screening assay (Rahbar et al., 2016) to test whether D 2 in the gills might affect the hypoxic response.We report an increase in ventilation frequency in zebrafish larvae in response to the D 2 receptor antagonist, domperidone.Further, when the D 2 receptor agonist, quinpirole, was administered, larvae failed to mount a hyperventilatory response to hypoxia.These data support the hypothesis that D 2 receptors inhibit the hyperventilatory response to hypoxia, as it does in the carotid body.Due to potential off-target effects inherent to whole-animal experiments, we used immunohistochemical and molecular approaches for a more specific characterization of D 2 in the gills.
qPCR confirmed that both gene transcripts encoding D 2 receptors were expressed in isolated gill tissue.Further, we found co-localization of D 2 receptors and chemosensory NECs in the efferent gill filament epithelium.The location of these receptors is consistent with the carotid body, where inhibitory D 2 receptors are found on chemosensory type 1 cells, which may suggest a similar mechanism of inhibition or modulation in the zebrafish gill.However, further work is needed to explore a possible mechanism of inhibition of the hypoxic signal via D 2 receptors.We also found the relative gene expression of both drd2a and drd2b in isolated gills decreased following hypoxia acclimation.
Similarly, Huey and Powell (2000) showed D 2 receptor mRNA expression in the carotid body decreased following hypoxia acclimation and suggested this may be one mechanism by which exposure to chronic hypoxia reduces inhibition of hypoxia signaling to enhance ventilatory acclimatization to hypoxia.Our findings are consistent with the mammalian carotid body and suggest gill D 2 receptors may be involved in acclimatization to hypoxia.
In Figure 10, we summarize a proposed model, in which the postsynaptic release of dopamine inhibits the NEC response to hypoxia.
During hypoxia, NECs depolarize, increase intracellular Ca 2+ concentration, and undergo synaptic vesicle recycling, as occurs during neurotransmitter secretion (Jonz et al., 2004;Zachar et al., 2017a).The excitatory neurotransmitter responsible for relaying the initial hypoxic signal from NEC to postsynaptic nerve terminal remains unconfirmed, but may be serotonin.Evidence demonstrating that NECs in zebrafish contain 5-HT, synaptic vesicles, the vesicular monoamine transporter (VMAT2), and selectively express genes encoding tryptophan hydroxylase and the serotonin transporter (SERT) for possible 5-HT synthesis and uptake, supports this idea (Jonz & Nurse, 2003;Pan et al., 2021Pan et al., , 2022)).We propose that the released neurotransmitter activates post- in petrosal neurons of rat carotid body, dopamine acts through D 2 to inhibit adenylyl cyclase and cyclic AMP, thus reducing membrane excitability (Zhang et al., 2018).Future physiological experiments on NECs may elucidate the intracellular pathways involved in reducing NEC activity via D 2 .Interestingly, we found D 2 receptor labeling was not limited to chemosensory NECs.Immunohistochemical labeling against D 2 receptors also revealed some D 2 -positive nerve fibers of the gill filaments that projected into the respiratory lamellae, and an unknown cell type in the afferent filament epithelium.A role for D 2 receptors at these sites in the gill is less clear.Though chemosensory NECs are primarily limited to the efferent filament epithelium (Jonz & Nurse, 2003), there are a number of cell types in the afferent filament epithelium that may be of interest in oxygen sensing and signaling.For example, ACh is a neurotransmitter thought to be involved in the facilitation of the HVR in zebrafish.In the zebrafish gill, the vesicular acetylcholine transporter (VAChT) has been localized to cells in the efferent filament epithelium as well as an abundance of cells in the afferent filament epithelium, suggesting the importance of ACh on both sides of the gill filament (Zachar et al., 2017b).Whether VAChT-positive cells in the afferent epithelium are also D 2 positive is currently not clear; however, given that carotid body type 1 cells contain both ACh and dopamine, it is possible that a similar organization of neurotransmitters exists in the gill.

CONCLUSION
The present study sought to evaluate a role for dopamine in hypoxia signaling within the zebrafish gill.We found evidence for postsynaptic synthesis and reuptake of dopamine in the gills and implicate As illustrated, zebrafish NECs contain 5-HT, synaptic vesicles, and the vesicular monoamine transporter, VMAT2 (Jonz & Nurse, 2003;Pan et al., 2021).NECs also express genes encoding tryptophan hydroxylase and the serotonin transporter (SERT) for possible 5-HT synthesis and uptake (Pan et al., 2022).Created with BioRender.com.
presynaptic D 2 receptors in control of the HVR in zebrafish.Given the location of D 2 receptors, it is possible that this control takes place at the level of the chemosensory NECs within the zebrafish gill; however, further work is needed to explore such a mechanism of inhibition.
The results of our studies are consistent with carotid body mediation of hypoxia signaling via D 2 receptors and thus suggest that an inhibitory or modulatory role for dopamine in oxygen sensing arose early in vertebrate evolution.
paraformaldehyde in PBS overnight at 4 • C. Tissues were removed and rinsed in PBS three times at 3 min before permeabilization for 24 h at 4 • C. Permeabilizing solution (PBS-TX) contained 2% Triton X-100 in Illustration of the whole-mount gill preparation in zebrafish and organization of the structures involved in oxygen sensing.(a) Gill baskets were removed ventrally from zebrafish, as shown.Rostral is at the top of the image and the gill filaments of gill arches 1−4 are labeled.E, eye; PF, pectoral fin.The scale bar represents 0.5 mm and applies to panels (b) and (c).An isolated gill basket is shown in panel (b), and a single gill shown in panel (c).GA, gill arch.The dashed box indicates a typical region of a gill filament (F) that was studied using confocal microscopy and is represented as a schematic in panel (d).(d) In this simplified representation, oxygen-sensitive neuroepithelial cells (NECs, green) are located along the filament (F) and receive sensory innervation (red).Nerve fibers form a plexus (not shown) around NECs and extend laterally to the secondary lamellae (L, small arrows), and large nerve bundles (large arrow) extend toward the gill arch.(e) Extrabrancial (left) and intrabranchial (right) innervation of filament NECs arise from the branchial nerve (BN) or superficial and deep proximal neurons (SPN and DPN), respectively.E is modified from Jonz and Nurse (2008), with permission from the authors.

F I G U R E 2
Confocal imaging of immunohistochemical localization of tyrosine hydroxylase (TH)-positive nerve fibers associated with serotonergic neuroepithelial cells (NECs).(a) NECs (arrows) labeled with anti-serotonin (5-HT, green) were closely associated with a plexus of nerve fibers (arrowheads) labeled by anti-TH (magenta) in the filaments (F) and lamellae (L).(b, c) TH and 5-HT labeling shown separately.(d-f) Images of gill filaments and lamellae are shown at higher magnification (upper panels) and tilted back 90 • (lower panels).In the latter, the transverse optical section revealed the close association between NECs and nerve fibers, and that the nerve fibers surrounded the efferent filament artery (eFA).Scale bar in panel (a) = 20 μm and applies to panels (b) and (c).Scale bar in panel (d) = 20 μm and applies to panels (e) and (f).
and were obtained from the Becker Laboratory at the University of Edinburgh.ET(vmat2:GFP) zebrafish contained a reporter gene for green fluorescent protein (GFP) under the expression of the vesicular monoamine transporter (vmat2 also known as F I G U R E 3 Confocal imaging of immunohistochemical co-localization of tyrosine hydroxylase (TH) with a zebrafish-specific neuronal marker.(a) Labeling with anti-TH (green) co-localized with nerve fibers labeled with zn-12 (magenta).(b, c) TH and zn-12 labeling shown separately.Both nerve bundles of the central filament (F) and all zn-12-positive nerve fibers (arrowheads) of the filaments and lamellae (L) were also TH positive.Some labeling by anti-TH did not co-localize with zn-12.An intrabranchial chain neuron (ChN) is indicated that was labeled with both markers.Scale bar in panel (a) = 10 μm and applies to all panels.
scope platform (FN1 [Nikon] or H101A ProScan [Olympus]) with motorized XYZ control and a confocal scanning system (A1RsiMP [Nikon] or FV100 BX61 LSM [Olympus]) equipped with continuous laser lines at 405, 488, and 561 nm.Images were viewed and captured with confocal imaging software, NIS Elements (Nikon).Each image is presented as a composite projection of multiple optical sections, each separated by 0.5−1.0μm, and combined to produce a total tissue thickness of 20−60 μm.Processing of images was made through the open-source software, Fiji (RRID: SCR_002285; Schindelin et al., 2012) and Corel Draw 10 (RRID: SCR_014235; Corel Corp.).Signals from red fluorescence markers were converted to magenta. figures.

F
Confocal imaging of nerve fibers expressing the dopamine active transporter (dat) and associated neuroepithelial cells (NECs) expressing the vesicular monoamine transporter (vmat2) in gills of the Tg(dat:mCherry)/ET(vmat2:GFP) line.(a) dat expression (mCherry, magenta) was observed in nerve fibers (arrowheads) that formed a plexus along the center of the gill filaments (F).Nerves were closely opposed to vmat2-positive NECs (green), labeled by GFP and indicated by arrows.L = lamellae.(b, c) dat and vamt2 expression shown separately.(d-f) At higher magnification, the dat-positive nerve plexus extended across the filament epithelium and reached the interlamellar regions.A chain neuron (ChN) and associated axons were labeled and are indicated.(g-i) Images from panels (d-f) tilted back 45 • (upper panels) and 90 • (lower panels).Rotation demonstrates the relationship between the dat-expressing nerve fibers and NECs in a transverse optical section.The plexus of dat nerve fibers wrapped around the efferent filament artery (eFA) and was closely associated with NECs.The ChN was located beneath the eFA, as part of the deeper nerve bundle that is comprised by intrabranchial neurons.Scale bar in panel (a) = 25 μm and applies to panels (b) and (c).Scale bar in panel (d) = 10 μm and applies to panels (e-i).F I G U R E 5 In chemical screening assays, domperidone (a dopamine D 2 receptor antagonist) increased breathing frequency (f v ), and quinpirole (a dopamine D 2 receptor agonist) abolished the hyperventilatory response to hypoxia.f v (in breaths per minute) was measured in zebrafish larvae at 14−16 days postfertilization (d.p.f.).(a) Mean f v ± SEM is shown for controls (C), after application of 50, 100, 200, or 300 μM domperidone and after recovery (R).Asterisks denote a significant increase in f v compared to controls (p < .05;ANOVA with repeated measures and Bonferroni posttest; N = 11, 10, 11, and 10 for each concentration, respectively).(b) A plot of concentration versus percent maximum f v response was fit with a nonlinear ligand binding model with least squares following the equation: y = b max ⋅x h /(k d h + x h ).The effective half-maximal concentration (EC 50 ) for domperidone was 30 μM.(c) Mean f v ± SEM for larvae exposed to hypoxia alone (including 0.5% DMSO for control; left panel) or hypoxia combined with 50, 150, or 200 μM quinpirole.N = 15 for hypoxia, and 12, 11, and 12 for each concentration, respectively.(d) A plot of the concentration versus percent maximum inhibition of the f v response was fit with the same equation as in panel (b).
, Finley et al. demonstrated that 41% of all carotid body afferents were dopaminergic and suggested that postsynaptic nerve terminals may additionally synthesize and release dopamine to modulate the type I cell response to hypoxia.In addition, oxygen-sensitive NEBs of the lung epithelium in rat were shown to receive innervation by TH-positive nerve fibers (Van Genechten et al., 2004), although a similar role for dopamine in NEBs has never been shown.Our characterization indicates that, in zebrafish, TH-and dat-positive nerve fibers that associate with gill NECs are capable of dopamine synthesis and reuptake and provide a potential mechanism for production and vesicular storage of dopamine in postsynaptic nerve fibers (Figure 10).We did not detect TH in NECs in the present study.Dopamine synthesis in gill NECs of zebrafish, however, remains a possibility since expression of ddc (the gene encoding DOPA decarboxylase) was reported in NECs and neurons in single-cell RNA-sequencing (Pan et al., 2022), and the vesicular monoamine transporter (VMAT2) is F I G U R E 7 Relative gene expression of drd2a and drd2b (encoding D 2 receptors) decreased following 48 h of chronic hypoxia.Data were normalized to the mRNA abundance of the reference gene, ef1a.Data were analyzed using a Mann-Whitney U test and means significantly different from control are indicated by asterisks (p < .01;N = 8).Relative abundance for reference gene ef1a in the gill of control (normoxia) and treatment (hypoxia).Data points presented are relative to the control group.Hypoxia had no effect on the mRNA abundance of ef1a (Mann-Whitney U test; p > .05;N = 7).
synaptic receptors, leading to subsequent release of dopamine from sensory nerve terminals into the synaptic cleft, where it acts upon presynaptic D 2 receptors.D 2 receptor activation by dopamine may then provide an intracellular pathway through which NEC excitation or secretion by hypoxia is reduced.In type I cells of rabbit carotid body, dopamine was shown to inhibit voltage-gated Ca 2+ channels, thereby reducing neurosecretion (Benot & López-Barneo, 1990).Moreover, F I G U R E 8 Characterization of D 2 -positive cells of the efferent gill epithelium.Confocal imaging of immunohistochemical labeling of dopamine receptor 2 (D 2 -R) with a zebrafish-specific neuronal marker (zn-12) and neuroepithelial cells (NECs) containing synaptic vesicle protein-2 (SV2).(a) Labeling with anti-D 2 -R (green) co-localized with NECs (arrows) labeled with SV2 (magenta) and nerve fibers labeled with zn-12 (magenta) in the filaments (F).(b, c) D 2 -R and SV2/zn-12 labeling shown separately.Some zn-12-positive nerve fibers (arrowheads) of the filament and lamellae (L) were also D 2 -R positive.(d-f) Co-localization of NECs and D 2 -R from panels (a-c) shown at higher magnification and tilted back 45 • .(g-i) Images from panels (a-c) tilted back 90 • .Rotation demonstrates D 2 -R-positive NECs in a transverse optical section oriented superficial to the efferent filament artery (eFA).Scale bar in panel (a) = 20 μm and applies to panels (b) and (c).Scale bar in panel (d) = 20 μm and applies to panels (e-i).

F
Characterization of D 2 -positive cells of the afferent gill epithelium.(a) Confocal imaging of immunohistochemical localization of dopamine receptor 2 (D 2 -R) on the afferent filament epithelium.(b) Co-application with DAPI-labeled nuclei of D 2 -R-positve cells.(c, d) Image from panel (a) was tilted back 45 • (c) and 90 • (d).Rotation demonstrates D 2 -R-expressing cells surround the afferent filament artery (aFA).Scale bar in panel (a) = 20 μm and applies to panels (b-d).F I G U R E 1 0 Model of the proposed role for dopamine in modulating the hypoxic response in gill neuroepithelial cells (NECs) of the efferent epithelium.As shown in the present study, postsynaptic nerve terminals in the gill filaments express tyrosine hydroxylase (TH) and the dopamine active transporter (DAT), involved in synthesis and uptake of dopamine (DA), respectively.When the NEC is stimulated during hypoxia: (1) Release of an unknown neurotransmitter from the NEC activates a postsynaptic receptor on the sensory nerve terminal.(2) Dopamine is then released from the nerve terminal and (3) activates presynaptic D 2 receptors of the NEC, or is taken up postsynaptically by DAT.(4) The former leads to inhibition of the NEC response to hypoxia.The excitatory neurotransmitter released by NECs has not been confirmed but may be serotonin (5-HT).