Respirometry and cutaneous oxygen flux measurements reveal a negligible aerobic cost of ion regulation in larval zebrafish (Danio rerio)

For most fishes, maintaining and regulating internal ion balance is essential for survival. Many adaptations exist in freshwater (FW) and seawater (SW) fishes to counteract the ionic challenges of their native environments. Most FW teleost fishes are hyperionic and hyperosmotic relative to their external environment. These ionic and osmotic differences favour the passive loss of ions by diffusion and the continual gain of water through osmosis (Evans et al., 2005). To combat ion loss, FW teleost fishes absorb Na⁺, Cl⁻ and Ca²⁺ across cutaneous (larvae) and branchial (adults) epithelia using specialized, mitochondrion-rich (MR) cells termed ionocytes. Depending on subtype (see below), ionocytes express specific channels, exchangers and ATP-dependent transporters to facilitate ion uptake (Dymowska et al., 2012; Evans, 2011; Evans and Claiborne, 2009; Evans et al., 2005; Gilmour and Perry, 2009; Guh et al., 2015; Hwang, 2009, 2010; Hwang and Chou, 2013; Hwang and Lee, 2007; Hwang and Lin, 2013; Hwang and Perry, 2010; Hwang et al., 2011; Marshall, 2002; Marshall and Grosell, 2006; Perry et al., 2003; Perry and Gilmour, 2006). The metabolic demands of these cells are presumed to be high due to the abundance of mitochondria and ATP-consuming transporters (Dymowska et al., 2012; Zikos et al., 2014). However, currently there is no consensus on the metabolic costs of ion regulation in FW fishes.

Indeed, comprehensive reviews of the literature (Bœuf and Payan, 2001; Ern et al., 2014; Kirschner, 1995) suggest cost estimates ranging from less than 1% to 50% of metabolic rate for FW fishes. A common method for determining estimates has been to compare salinity-related differences in metabolism for a given species with the assumption that osmoregulatory costs are close to zero within an iso-osmotic environment. Using this method, costs for rainbow trout (Oncorhynchus mykiss) in FW were 20% of metabolic rate (Rao, 1968), 19% for the Nile tilapia (Oreochromis niloticus; Farmer and Beamish, 1969), 50% for the brown bullhead (Ameiurus nebulosus; Furspan et al., 1984), ‘negligible’ for the striped mullet (Mugil cephalus; Nordlie and Leffler, 1975) and potentially less than 12.5% of metabolic rate for the European perch (Perca fluviatilis; Christensen et al., 2017).

In contrast, studies that used theoretical models for determining osmoregulatory costs have estimated much lower costs. For example, by measuring the rate of total ion uptake, the electrical gradient between the water and blood of the fish and the basal metabolic rate, Eddy (1982) estimated that 1% of basal metabolic rate of FW rainbow trout (O. mykiss) and less than 4% of basal metabolic rate in goldfish (Carassius auratus) was attributed to osmoregulation. Moreover, these results are supported by another model presented by Kirschner (1995), which concluded that rainbow trout allocate less than 2% of metabolic rate for osmoregulation in FW. Finally, a study that examined the O₂ consumption of excised gills from FW-adapted cutthroat trout (Oncorhynchus clarkii) found that following exposure to bafilomycin A1 and ouabain (H⁺-ATPase and Na⁺/K⁺-ATPase inhibitors, respectively), gill tissue O₂ consumption dropped by 37%. This was translated to a 1.8% cost of whole animal O₂ uptake dedicated to NaCl uptake (Morgan and Iwama, 1999).

Not only is there a wide range in the current estimates of the metabolic cost of ionic regulation in fishes, to our knowledge, there are no data concerning such costs in larval stages. The ionoregulatory mechanisms underlying the absorption of Na⁺, Cl⁻ and Ca²⁺ uptake in zebrafish larvae have been extensively investigated and are summarized in a number of comprehensive reviews (Evans, 2011; Guh et al., 2015; Hwang, 2009; Hwang and Lee, 2007; Hwang and Perry, 2010; Hwang et al., 2011; Kumai and Perry, 2012). In particular, the Na⁺ uptake pathways in larval zebrafish are well characterized and share the reliance on basolateral Na⁺/K⁺-ATPase for ultimate entry into the blood. Basolateral Na⁺/K⁺-ATPase activity is considered a central contributor to the ionoregulatory costs in FW fishes (Kirschner, 1995). Three apical pathways for Na⁺ transport in zebrafish larvae are proposed: (i) electroneutral Na⁺/H⁺ exchange via a Na⁺/H⁺ exchanger 3b (NHE3b; slc9a3.2) (Esaki et al., 2007), (ii) H⁺-ATPase (HA) activity linked with a putative epithelial Na⁺ channel (potentially an acid-sensing ion channel) (Dymowska et al., 2015; Zimmer et al., 2018) and (iii) Na⁺/Cl⁻ cotransport (NCC) facilitated by slc12a10.2 (Wang et al., 2009). The NHE3b- and HA-facilitated pathways are expressed in HA-rich (HR) cells and the NCC-mediated pathway resides in NCC ionocytes (Guh et al., 2015). The contribution of each pathway may differ depending on the prevailing environmental conditions (Hwang and Lee, 2007; Shih et al., 2012; Yan et al., 2007). Notably, Na⁺ uptake capacity in zebrafish larvae is increased markedly upon or after exposure to waters of low pH or low Na⁺ content (Kumai and Perry, 2011; Kumai et al., 2011; Kwong and Perry, 2016; Shih et al., 2012). Such increases in Na⁺ uptake are expected to increase metabolic cost given the active nature of Na⁺ uptake.

The aim of this study was to determine the aerobic costs associated with Na⁺ regulation in larval zebrafish. It was hypothesized that changes in the rate of Na⁺ uptake would significantly influence the O₂ consumption of the Na⁺-transporting ionocytes and thus affect whole-body oxygen consumption (Ṁ_O2) as well as cutaneous O₂ flux at the yolk sac due to its high density of ionocytes. In addition to exploiting the naturally occurring spatial distribution of cutaneous ionocytes, this hypothesis was tested using an integrated approach whereby rates of Na⁺ uptake were manipulated and/or ionocyte numbers altered by exposing fish to low Na⁺ or acidic water and after morpholino knockdown of Foxi3a [transcription factor responsible for ionocyte specification and differentiation (Hsiao et al., 2007)] or optical ablation of HR cells.

Wild-type (WT) zebrafish (Danio rerio) were housed at the University of Ottawa aquatic care facility. Fish were maintained in plastic aquaria that were constantly supplied with aerated, dechloraminated City of Ottawa tap water at 28°C [referred to as ‘system water’ (in mmol l⁻¹): 0.25 Ca²⁺, 0.78 Na⁺, 0.4 Cl⁻, 0.025 K⁺, 0.15 Mg²⁺; pH 7.6]. Fish were kept on a 14 h:10 h light:dark photoperiod and fed until satiation with a no. 1 crumble Zeigler diet (Aquatic Habitats, Apoka, FL, USA) once a day. Unless otherwise stated, for all experiments, zebrafish at 4 days post-fertilization (dpf) were used. Larvae were obtained by breeding adult zebrafish in plastic 2-liter breeding traps with a perforated insert. For breeding, 8–10 fish at a ratio of two females to one male were placed in breeding traps and left overnight. The following morning, embryos were collected using a fine mesh sieve and were placed in 50 ml Petri dishes at a density of 30 embryos per dish containing different media, based on experimental protocols (see below), and held in an incubator set to 28.5°C. Media in the Petri dishes were replaced daily until experimentation at 4 dpf. All experiments were conducted in compliance with the Canadian Council of Animal Care guidelines and after approval by the University of Ottawa Animal Care Committee (protocols BL-2118 and BL-1700).

Experiments were designed to detect local cutaneous oxygen flux (J_O2) at the yolk sac epithelium of 4 dpf larval zebrafish in response to various experimental manipulations using a scanning micro-optrode technique (SMOT) (Hughes et al., 2019; Zimmer et al., 2020). The purpose of the J_O2 measures was to provide a direct assessment of the aerobic metabolism of the ionocytes with the highest resolution possible using current available techniques. The SMOT system (Applicable Electronics, Falmouth, MA, USA) consists of a detector system and a CMC-4 motion control unit connected to a fiber optic oxygen optrode (PreSens Precision Sensing, Regensburg, Germany). Adjustments of system settings are made using ASET-LV4 software (Applicable Electronics; http://www.applicableelectronics.com/overview). Oxygen optrodes were constructed from fiber optic cables that were flame-pulled to a 40 µm tip and coated with Pt(II) meso-tetra(pentafluorophenyl)porphyrin (Pt-TFPP; Frontier Scientific, Newark, DE, USA). The optrode was connected to the detector system that emits an excitation light through a blue LED (λ=400 nm), which excites the Pt-TFPP coating and detects fluorescence emission. The fluorescent emission from Pt-TFPP is quenched in the presence of oxygen.

To apply SMOT, a set-up in which fish could be secured in place was necessary. Larvae (2–4 dpf, depending on the experimental series) were anesthetized, individually, in a solution of 0.20 mg ml⁻¹ tricaine methane sulfonate (MS-222; Syndel Laboratories Ltd, Nanaimo, BC, Canada) for 5 min. Tricaine used in low doses as in this study does not affect O₂ consumption in chinook salmon (Oncorhynchus tshawytscha) larvae but does prevent pectoral and opercular movements (Rombough, 1988). Afterwards, fish were transferred to a modified Petri dish (Hughes et al., 2019) filled with the appropriate treatment solution, depending on the experimental series. The modified dish contained a thin bottom layer of Sylgard 184 silicon elastomer (Dow Corning, Midland, MI, USA). A thin strip of Sylgard was placed over the tail of the larva, with both ends of the strip secured to the bottom of the dish with Austerlitz 0.20 mm minutien insect pins (Entomoravia, Slakov u Brna, Czech Republic). The head of the larva was secured in place with a second, thicker strip of Sylgard that acted as a buttress (Hughes et al., 2019). This method of securing the larva in place was necessary because initial experiments demonstrated that the larva would otherwise drift as the probe moved during measurements. To ensure that measurements were made at ionocyte-expressing epithelia, larvae were vitally stained with 1 μM MitoTracker Red CMXRos (MitoROS; Thermo Fisher Scientific, Burlington, ON, Canada) for 10 min and 0.05 mg ml⁻¹ of Concanavalin A (ConA), a lectin that binds specifically to HR cells of zebrafish larvae (Lin et al., 2006), conjugated to Alexa 488 (Thermo Fisher Scientific) for 30 min. A control experiment designed to test the effect of the cell stains on cutaneous J_O2 was performed by measuring J_O2 across the yolk sac extension of stained and unstained larvae. No statistically significant effect was found (N=5; P=0.154; Fig. S1).

Before each experiment, the optrode was calibrated in zero solution (20 g l⁻¹ anhydrous Na₂SO₃) and air-saturated media that matched the corresponding experimental series. SMOT measurements were conducted by initially positioning the optrode directly above (5–10 µm) the epithelium of the yolk sac. The first measurement of O₂ partial pressure (P_O2) was taken at this location. The probe was then set to follow an excursion distance of 100 µm in the z-plane to take the second measurement. Measurements were recorded for 2 s followed by a waiting period of 5 s. This was repeated five times for each specific point on the epithelium; time between replicates was set to 10 s. For all experiments, P_O2 was measured at three equidistant points along the yolk sac extension (area where ionocytes are present) and at three adjacent points along the posterior trunk (area where ionocytes are absent) for a single larva (N=1). Observations were made using a Nikon SMZ 1500 (Nikon Instruments, Melville, NY, USA) fluorescence dissection microscope.

Whole-body O₂ consumption was measured in 2 or 4 dpf larvae to complement the localized measurements of J_O2 using SMOT. The system (Loligo Systems, Viborg, Denmark) consisted of a 24-well, glass microplate in which each chamber had a volume of 80 µl (well inner diameter, 4.5 mm). Each well contained a non-invasive O₂ sensor spot, which was scanned by a 24-channel optical fluorescence O₂ microplate reader. The microplate was placed in a temperature-controlled (29.9°C) water bath held on a shaker (continuous circular oscillations set to 30 rpm with a deviation of 25 mm) to prevent the formation of unstirred layers of water surrounding each larva. The sensors were calibrated using zero solution and air-saturated water (see above) after sealing the microplate using PCR tape, a silicone pad and a compression block. Care was taken to remove air bubbles from the wells before recording P_O2. Following calibration, wells were rinsed thoroughly and larvae from each respective treatment were assigned randomly to wells (one larva per well). P_O2 was recorded continuously for approximately 30 min.

Larvae were observed under a Nikon SMZ 1500 microscope equipped with the 49008-ET-mcherry, Texas Red filter cube set (Chroma Technology, Bellow Falls, VT, USA) and images of the fluorescently stained, apical surface of the yolk sac were captured using a 1.3-megapixel CMOS sensor connected to µEYE cockpit software (IDS, Obersulm, Germany; https://en.ids-imaging.com/ids-software-suite.html). Cells were counted manually, and surface area of the yolk sac was determined by ImageJ 1.51 open source software (National Institutes of Health, Bethesda, MD, USA; https://imagej.nih.gov/ij/download.html). Epithelial J_O2 was measured across the yolk sac of each larva. A correlation analysis was performed to determine relatedness by plotting MR cell density (x-axis) against J_O2 (y-axis).

In our initial studies, it was found that J_O2 was lowest near the yolk sac extension. Thus, all subsequent SMOT measurements were performed across the yolk sac extension because the lower ‘background’ J_O2 would allow for greater sensitivity to detect changes in J_O2 resulting from metabolic changes in ionocytes.

Zebrafish were reared in reconstituted medium resembling system water, composed of the following (in mM): 0.15 MgSO₄·7H₂O, 0.02 K₂HPO₄, 0.05 KH₂PO₄, 0.25 CaCl₂, 0.4 Na₂SO₄; pH 7.6. Unless otherwise mentioned, larvae were reared under N-Na conditions (800 µmol l⁻¹) and maintained at pH 7.6. Rearing medium was titrated daily to pH 7.6 using dilute KOH or H₂SO₄ and was replaced daily. In one experiment, larvae were reared in N-Na adjusted to pH 4 (low pH; L-pH) using H₂SO₄ and were acutely transferred to pH 7.6 at 4 dpf, just prior to experimentation. In a second experiment, larvae reared in N-Na were acutely transferred to L-Na conditions (5 µmol l⁻¹ Na⁺; 2.5 µmol l⁻¹ Na₂SO₄) just prior to experimentation and a separate group of larvae were reared in L-Na and acutely transferred to N-Na conditions prior to experimentation. Na⁺ uptake, whole-body respirometry and SMOT measurements were performed on larvae from all treatments.

Knockdown of the transcription factor foxi3a (NCBI reference sequence: NM_198917.2) was used to prevent ionocyte differentiation in larvae (Hsiao et al., 2007). However, the effect of foxi3a knockdown lasted only until 2 dpf (J.J.P., unpublished observations) and therefore experiments on foxi3a morphants (individuals experiencing morpholino knockdown) were performed at 2 dpf. Knockdown was achieved by the microinjection of an anti-sense morpholino oligonucleotide. Embryos were injected at the 1-cell stage with 4 ng of either a sham morpholino (5′-CCTCTTACCTCAGTTACAATTTATA-3′; Gene Tools, Philomath, OR, USA) that has no biological target in zebrafish, or a morpholino targeting the translation start site of foxi3a [5′-CCTTCAACAAAGAGAAACGGGGAAGA-3′ (Hsiao et al., 2007); Gene Tools] suspended in 1 nL of Danieau buffer [in mmol l⁻¹: 58 NaCl, 0.7 KCl, 0.4 MgSO₄, 0.6 Ca(NO₃)₂, 5.0 Hepes; pH 7.6] containing 0.05% Phenol Red for visualization. At 2 dpf, zebrafish were dechorionated with fine forceps prior to Na⁺ uptake, microrespirometry and SMOT measurements. After 2 dpf, ionocyte density in foxi3a morphants gradually increased across the larval yolk sac. The sparsely spaced ionocytes accommodated the tip of the SMOT probe and made it possible to make flux measurements over areas where ionocytes were present or absent on the yolk sac extension.

Larvae at 4 dpf reared under N-Na and pH 7.6 were stained with MitoRos and ConA and then anesthetized in a solution of 0.20 mg ml⁻¹ MS-222. Larvae were embedded at room temperature with 1.8% low-melting point agar (Bioshop Canada, Burlington, ON, Canada) in the bottom of a 10 mm Petri dish. After the agar solidified, the Petri dish was filled with MS-222-containing N-Na water and the larvae were examined using a single-photon, scanning confocal laser microscope (A1R⁺, Nikon Instruments). A 404.6 nm laser set to 100% power was used to ablate patches of ionocytes along the yolk sac extension; ablation was confirmed visually by the absence of fluorescence in the ablated areas. Exposure time was set to 32.2 s and scan speed was set to 1/32. These settings were shown previously to successfully ablate melanocytes on larval zebrafish (Yang et al., 2004). Following cell ablation, larvae were removed from the agar bed using fine forceps and left to recover for 30 min in fresh reconstituted water prior to SMOT measurements.

All statistical analyses were performed using SigmaPlot (version 11.0; Systat Software, Chicago, IL, USA; https://systatsoftware.com/products/sigmaplot/). Data are reported as means±standard error of the mean (s.e.m.). Statistical significance of treatment effects was evaluated through two-way and one-way ANOVA followed by a Holm–Šidák post-hoc test or a Student's t-test. Statistical significance was accepted at P≤0.05. Specific details of statistical analyses are included in corresponding figure captions.

Epithelial J_O2 was measured within three regions of the yolk sac (Fig. 1A) near small clusters of MR cells (Fig. 1B). The regions were arbitrarily designated as follows: anterior (region anterior to the apex of the yolk sac), middle (posterior to the apex of the yolk sac, but anterior to the yolk sac extension), and posterior (yolk sac extension). MR cell density (including HR cell density) was determined for each larva. MR cell density was significantly lower in the posterior portion of the yolk sac but was not significantly different for HR cells across the yolk sac (Fig. 2A). Furthermore, the posterior region also displayed the lowest J_O2, revealing a declining anterior-to-posterior trend in J_O2 along the yolk sac (Fig. 2B); however, there was no relationship (R²=0.006) between J_O2 and ionocyte density across the yolk sac (Fig. 2C).

Larvae were reared in normal (pH 7.6; mass 0.221±0.012 mg, N=9) or low (pH 4; mass 0.249±0.025 mg, N=6) pH conditions but were transferred to pH 7.6 immediately prior to Na⁺ uptake, Ṁ_O2 or J_O2 measurements. Na⁺ uptake rate was significantly higher in larvae that were reared at pH 4; however, Ṁ_O2 was significantly lower (Fig. 3A,B). J_O2 at the yolk sac extension or trunk was not significantly affected by L-pH acclimation (Fig. 3C).

In response to acute transfer from N-Na to L-Na conditions, Na⁺ uptake rate was significantly reduced, but was stimulated in larvae reared in L-Na conditions (mass 0.2409±0.246 mg, N=6) and transferred to N-Na conditions (Fig. 4A). Despite these changes in Na⁺ uptake rates, Ṁ_O2 (Fig. 4B) and J_O2 (Fig. 4C) measured under the same conditions were not significantly different across treatment groups. J_O2 was not significantly different between the yolk sac extension and trunk under any of the conditions (Fig. 4C).

At 2 dpf, ionocytes were present across the yolk sac extension of the sham larvae (Fig. 5A), but not the foxi3a morphants (Fig. 5B). This led to a large reduction in Na⁺ uptake compared with the sham larvae (Fig. 6A). However, whole-body Ṁ_O2 (Fig. 6B) and J_O2 (Fig. 6C) were not affected by foxi3a knockdown. Interestingly, a significant difference between J_O2 measured at the yolk sac extension and trunk was detected in 2 dpf larvae, but this difference was unaffected by foxi3a knockdown (Fig. 6C).

Noticeably, at 59 h post-fertilization (hpf), ionocyte density at the yolk sac epithelium continued to increase for the sham larvae (Fig. 7A) and the morphants (Fig. 7B). At this time, regional J_O2 at sites with and without ionocytes was not significantly different (Fig. 7C).

Using single-photon confocal microscopy, cell ablation was carried out on the epithelium of 4 dpf larval zebrafish. Using MitoTracker staining as a guide (Fig. 8A), sites along the yolk sac extension were ablated (Fig. 8B) with the goal of destroying ionocytes, while other sites were left unaffected. There were no significant differences in J_O2 between ablated and unablated sites across the yolk sac extension (Fig. 8C).

The purpose of this study was to assess the aerobic costs of Na⁺ uptake in larval zebrafish. A major component of the experimental design consisted of manipulating water chemistry to increase or decrease the rate of Na⁺ uptake. This protocol provided an opportunity to examine the relationship between Na⁺ uptake rate and Ṁ_O2/cutaneous J_O2 as well as cutaneous J_O2 at regions of low and high ionocyte abundance. Furthermore, through foxi3a knockdown, ionocyte differentiation was delayed, causing a near elimination of Na⁺ uptake. The results clearly demonstrated that Ṁ_O2 and J_O2 were not correlated with rates of Na⁺ uptake or ionocyte density, thereby demonstrating that Na⁺ uptake does not incur measurable aerobic cost in zebrafish larvae. Ultimately, the findings of this study support the notion that ionic regulation incurs a relatively low aerobic cost in FW fishes (Eddy, 1982; Kirschner, 1995; Morgan and Iwama, 1999; Nordlie and Leffler, 1975).

It is important to acknowledge that the results of this study, strictly speaking, are valid only for zebrafish larvae (at 4 dpf) and may not necessarily apply to other species or other developmental stages, including adults. In rapidly developing larvae, growth presumably commands a significantly larger portion of the total energy budget in comparison with adults. Although it is conceivable that the high rates of anabolic processes in larvae masked the energetic costs of ionic regulation (Na⁺ uptake), we consider this unlikely because the high rates of Ṁ_O2 in larvae are roughly matched by equally high rates of Na⁺ uptake (compare Fig. 3A of this study with fig. 7B in Zimmer and Perry, 2020). Thus, the ratio of Na⁺ uptake to Ṁ_O2 is more-or-less constant (3.4–4.4%) in larvae and adults (derived using Ṁ_O2 data from Fig. 3B of this study and table1 in Mandic et al., 2020). An additional caveat is that measuring Ṁ_O2 using whole-body respirometry may not necessarily reveal the true aerobic cost of ionic regulation if changes in O₂ consumption by ionocytes occurring during periods of experimentally altered Na⁺ uptake are masked by re-allocation of the energy budget such that overall Ṁ_O2 remains unchanged. For this reason, the conclusions of the present study were heavily reliant on the results of the SMOT experiments in which measurements of J_O2 were localized to small clusters of ionocytes at the surface of the yolk sac epithelium and thus were unaffected by the constraints of energy re-allocation.

Despite the finding of a decreasing ionocyte density (Fig. 2A) and a reduction in J_O2 (Fig. 2B) across the yolk sac in the anterior-to-posterior direction, the two factors were not correlated (Fig. 2C). Although boundary layer P_O2 can be an indicator for the regional O₂ demand of organisms such as rainbow trout larvae (Ciuhandu et al., 2007), our results demonstrate that epithelial J_O2 measured at the yolk sac surface was not influenced by ionocyte density. It is possible that the anterior-to-posterior trend in declining J_O2 was related to metabolic demand of underlying tissues or organs. For example, Hughes et al. (2019) reported that epithelial J_O2 in zebrafish larvae was highest around the heart/gill area and decreased towards the trunk. Thus, the higher J_O2 observed in the anterior yolk sac in the present study may have resulted from the underlying energetically demanding tissues, including the heart and developing gills. Additionally, given the reliance of J_O2 on blood flow (Hughes et al., 2019), the increased rates of J_O2 in the anterior region may reflect differences in perfusion.

Initial experiments (Fig. 2) suggested that ionocyte O₂ consumption was negligible under N-Na and pH 7.6. However, we predicted that under conditions that challenge Na⁺ homeostasis, including exposure to L-pH or L-Na water, the aerobic demand of ionocytes might increase and thus be quantifiable using SMOT. Several changes to the ionoregulatory system occur under L-pH, which would be expected to increase the metabolic demand of Na⁺-transporting ionocytes. For example, HR cell density was shown to increase in 4 dpf zebrafish larvae acclimated to pH 4 water (Horng et al., 2009) and it follows that the increased number of mitochondrion-rich cells would lead to an increased O₂ demand. In addition, under chronic L-pH, HA mRNA and protein expression are increased (Chang et al., 2009; Horng et al., 2009; Lin et al., 2015; Yan et al., 2007), thus creating an increased demand for ATP to fuel the ATP-dependent transporter. Furthermore, L-pH rearing in zebrafish is accompanied by an increase in Na⁺ uptake capacity (Kumai et al., 2011) that may be related to endocrine responses, leading to increased plasma levels of prolactin and cortisol that promote Na⁺ uptake and increase HA activity (Kwong et al., 2014); these changes are assumed to come at an energetic cost to the larvae.

Thus, it was predicted that during periods of increased Na⁺ uptake rate after exposure to L-pH, O₂ consumption and epithelial O₂ flux would increase. Despite stimulating Na⁺ uptake capacity through the manipulation of environmental conditions, the results of microrespirometry and SMOT measurements did not support our hypothesis. Notably, under the L-pH treatment, Ṁ_O2 was reduced (Fig. 3B) and epithelial J_O2 was unaffected (Fig. 3C) in larvae that had been reared in pH 4 and acutely transferred to pH 7.6, even though Na⁺ uptake was significantly higher under these same conditions (Fig. 3A).

The discrepancy observed between whole-body Ṁ_O2 and regional J_O2 may be linked to a slightly hindered development. Although considered an acid-tolerant species (Kwong et al., 2014), capable of surviving in pH 4.0 water, the ideal rearing pH for zebrafish is 6.8–7.5 (Avdesh et al., 2012). In light of the negative effects of acidic water exposure on the physiology of fishes (McDonald, 1983; Wood, 1989), it is possible that the L-pH environment negatively affected growth. However, zebrafish at 4 dpf reared at pH 4.0 are only marginally shorter (e.g. 6% shorter in the study of JavadiEsfahani and Kwong, 2019) or largely unaffected (Horng et al., 2007). Although length was not measured in the current study, larvae that were raised at pH 4.0 were slightly heavier than those raised at pH 7.6. The small (or absent) effects of low pH on body length despite a marked reduction in whole body Ṁ_O2 may reflect energy re-allocation. Thus, as discussed above, the results of the regional measurement of J_O2 using SMOT in the specific vicinity of ionocyte clusters are crucial to our overall conclusions on the aerobic costs of ionic regulation.

Adult zebrafish acclimated to L-Na (0.04 mmol l⁻¹) exhibited an increase in gill Na⁺/H⁺ exchanger 3b (NHE3b: slc9a3b) mRNA expression and a decrease in HA (atp6v1aa) mRNA expression (Yan et al., 2007). These results suggest that under L-Na, Na⁺ uptake is managed principally by NHE3b in HR cells. Furthermore, using the scanning ion-selective electrode technique on 4 dpf zebrafish larvae, it was found that L-Na acclimation led to increased Na⁺ uptake presumably by HR cells (Shih et al., 2012). These results suggest that the increase in Na⁺ uptake observed in the current study in the L-Na-acclimated larvae after acute transfer to N-Na, was a result of increased NHE3b activity. The acute reduction of ambient Na⁺ accompanying the transfer of N-Na-acclimated larvae to a L-Na environment caused a decrease in Na⁺ uptake due to less substrate being available to the Na⁺/H⁺ exchanger (Fig. 4A). Overall, despite these bi-directional alterations to Na⁺ uptake, there were no accompanying differences in Ṁ_O2 (Fig. 4B) or J_O2 (Fig. 4C) among the treatments. Thus, any changes in O₂ uptake of the ionocytes are negligible compared with the bulk O₂ uptake of the rest of the body. Moreover, the fact that across all water chemistry conditions there were no differences in J_O2 between the trunk (no ionocytes) and the yolk sac extension (abundant ionocytes) argues for an insignificant contribution of ionocytes to bulk O₂ flux (Fig. 3C; Fig. 4C). Again, however, it must be considered that the growth-dominant metabolism of the larvae may lead to an energetic re-allocation when a reduction in ionoregulatory demands occurs.

To lower the energy expended on Na⁺ uptake, ionocyte density was reduced experimentally (using foxi3a knockdown) to only a few scattered ionocytes across the whole body (Fig. 5B). Due to the elimination of O₂ uptake by the presumed metabolically active Na⁺-transporting cells, it was predicted that the foxi3a knockdown would reveal reductions in whole-body and cutaneous O₂ uptake rates and that the reductions in cutaneous J_O2 would be restricted to the yolk sac. Despite the predicted fall in Na⁺ uptake rate in the morphants compared with sham larvae (Fig. 6A), there was no difference in Ṁ_O2 (Fig. 6B) or J_O2 (Fig. 6C), providing further evidence that the amount of energy utilized by the ionocytes is negligible. Notably, there was a significantly higher J_O2 at the trunk versus the yolk sac of 2 dpf larvae in both treatments (Fig. 6C), which may reflect a higher rate of O₂ consumption in the rapidly developing trunk relative to the yolk sac. However, this difference was not observed at other developmental stages, nor was it observed in two previous studies that measured J_O2 in larvae at 4 dpf (Hughes et al., 2019; Zimmer et al., 2020). It is possible that differences in J_O2 between the trunk and yolk sac are greatest in the younger larvae (i.e. 2 dpf) when growth is presumably faster.

Subsequently, as ionocytes gradually developed at around 3 dpf, there was still no significant difference in J_O2 when measuring over sites with and without ionocytes (Fig. 7C). Considering that the larval epithelium consists of keratinocytes, mucous cells, club cells, ionocytes and undifferentiated cells (Chang and Hwang, 2011), it was reasonable to predict that sites with more mitochondrion-rich ionocytes would yield a higher J_O2 because the remaining mitochondrion-poor cell types are assumed to have a lower energy requirement.

In addition to morpholino knockdown of foxi3a, cell ablation was also used to reduce ionocyte density. Because ablation was not entirely specific, some surrounding surface epithelial cells and subepithelial cells were also removed. Despite this loss of cells, J_O2 was unaffected (Fig. 8C). These data are consistent with the results obtained through foxi3a knockdown. In both cases, cells that were presumed to be metabolically active were removed from the cutaneous epithelium, yet J_O2 was unaffected. Thus, the elimination of the basal O₂ uptake into ionocytes was undetectable through cutaneous J_O2 measurements using SMOT.

In this study, the combined approaches of radioactive Na⁺ tracing, whole-body microrespirometry and SMOT were used to assess the aerobic costs of Na⁺ uptake in zebrafish larvae. Together, these techniques were complementary and allowed a thorough testing of the hypothesis. Through Na⁺ tracing, it could be ensured that the larvae exhibited the predicted response in Na⁺ uptake with specific treatments. Thus, the results from SMOT and whole-body microrespirometry could be interpreted with greater confidence. Notably, SMOT is a recently developed and technologically advanced technique (Ferreira et al., 2020) in this area of research and thus it was important to also implement a more tested technique such as whole-body microrespirometry. Despite using arguably the most sensitive technique available for measuring epithelial J_O2 in larval fishes, in addition to whole-body respirometry, the results of this study do not support the hypothesis that a substantial portion of aerobic metabolism is dedicated to ion regulation. It must be noted that this study was performed on larvae, which may have different metabolic demands compared with adults. However, even after a drastic reduction in ionocyte density leading to a near elimination of Na⁺ uptake, there was no change in metabolic rate or oxygen flux. To date, the many studies that have assessed the metabolic costs associated with ionoregulation and osmoregulation in fishes have provided drastically different results, ranging from a negligible cost, to nearly 50% of metabolic rate. Our study supports the notion that the aerobic cost of Na⁺ uptake and maintenance of ionocytes, at least in larvae, is negligible.

We are grateful for the excellent imaging and microscopy technical support of Andrew Ochalski.