Extra-gastric expression of the proton pump H⁺/K⁺-ATPase in the gills and kidney of the teleost Oreochromis niloticus

In the aquatic environment, fishes are exposed to many challenges that require regulation of ions and acid–base balance (Evans et al., 2005; Marshall and Grosell, 2005). In fresh water (FW), fishes compensate for the diffusive loss of ions by active uptake of ions, driven by pumps (Evans et al., 2005). They also maintain water homeostasis by getting rid of osmotically gained water by excreting copious amounts of dilute urine (Marshall and Grosell, 2005). In FW teleost fishes, acid–base transport takes place across the gill epithelium via exchange mechanisms located in ionocytes [mitochondrion-rich ‘chloride’ cells (CCs)] and possibly in pavement or respiratory cells (PVCs) (Evans et al., 2005; Wilson, 2011). The two main pumps that have been identified in fish gill to drive direct and indirect ion fluxes are the basolateral Na⁺/K⁺-ATPase that provides a sodium motive force and an apical vacuolar-type proton-ATPase that provides a proton motive force (Evans et al., 2005; Hwang et al., 2011). However, Choe et al. (2004) have provided some evidence of a role of the H⁺/K⁺-ATPase (HKA) in ion regulation in elasmobranch fish gills.

The main site of HKA expression is the stomach in vertebrates (Shin et al., 2009; Wilson and Castro, 2010). HKA is composed of HKα1 (gene: ATP4A) and HKβ (ATP4B) subunits (Pedersen and Carafoli, 1987). The β-subunit is unique to the gastric HKA and is involved in stabilizing cellular targeting of the transporter, while the larger α-subunit catalyses ATP hydrolysis and ion translocation. The gastric HKA HKα1 subunit belongs to the large subgroup of the P-type ATPase IIc family that also includes the non-gastric/colonic HKA (HKα2) and Na⁺/K⁺-ATPases (NKα) (Pedersen and Carafoli, 1987).

Most studies of ion and/or acid–base regulation in FW and saltwater (SW) teleost studies have focused on Na⁺, Cl⁻ and Ca⁺ ions (Evans et al., 2005). However, K⁺ is also an ion of interest, as it is an essential and natural element that plays a critical role in nerve, muscle and many other vital cell functions, such as metabolism, growth and repair (Talling, 2010). In addition, K⁺ affects the kidney's ability to reabsorb bicarbonate as the main extracellular buffer to metabolic acids. During periods of positive potassium balance, potassium efflux pathways in SW tilapia (Furukawa et al., 2012) and FW zebrafish (Abbas et al., 2011) and medaka (Horng et al., 2017) have been studied showing that orthologues to the mammalian renal outer medullary potassium channel (ROMK) are present apically in branchial and skin ionocytes. In FW larval medaka, Horng et al. (2017) have proposed a paracellular uptake pathway between skin keratinocytes that requires a high transepithelial potential (TEP) to operate using a SIET (scanning ion-selective electrode technique) approach. Studies by Eddy (1985) and Gardaire and colleagues (Gardaire and Isaia, 1992; Gardaire et al., 1991) suggest the possibility of branchial uptake of K⁺ from the water in rainbow trout, although the mechanism remains unknown. According to Caplan (1998), H⁺/K⁺-ATPase is certainly also involved in extra-gastric roles such as renal acid–base regulation in tetrapods including mammals. Welling (2013) and Gumz et al. (2010) additionally summarized that kidney HKA has an established role in potassium and acid–base regulation, which controls potassium absorption, and could potentially contribute to efficient renal potassium excretion.

Proton pump inhibitors (PPIs) and acid pump antagonists (APAs) are drugs that are the mainstay treatments for all acid-related diseases by inhibiting acid secretion with varying efficiency (Codina and DuBose, 2006; Shin et al., 2009). Omeprazole was the first drug of the PPIs class to be introduced into clinical use (in 1989) that inhibits the gastric HKA by covalent binding; hence, the duration of its effect is longer than expected from its levels in the blood. Unlike PPIs such as omeprazole, APAs do not require acid activation.

Oreochromis niloticus is also a well-studied species for fish biology and is widely used in aquaculture globally (Maclean et al., 2002). Therefore, it was chosen as a model species of significance as its genome has been sequenced and annotated (ensembl.org). The O. niloticus genome only has the HKα1 gene atp4a (Castro et al., 2014) and lacks the non-gastric proton pump paralogue HKα2 gene atp12a, which simplifies analysis of the results. The apparent HKα2 gene (atp12a) teleost orthologue (ENSHHUG00000034153) in Hucho hucho has been incorrectly annotated in ensembl.org and although at least two orthologues do exist in the ancient Polypteriform ray-finned fish Erpetoichthys calabaricus (ENSECRG00000015423 and ENSECRG00000016690), none are found in teleost genomes (J.M.W., unpublished observation).

The potential significance of extra-gastric HKA expression in teleost fishes remains to be resolved. In this paper, we provide the first evidence for the expression of the gastric proton pump HKA (genes: atp4a and atp4b) in a teleost fish and demonstrate its potential function in potassium regulation. The central objective was to investigate the role of this pump in extra-gastric potassium regulation in the gill and kidney of the teleost fish O. niloticus. Rb⁺ was used as a surrogate flux marker for K⁺ and two experimental approaches were taken: in vivo and ex vivo Rb⁺ uptake assays to analyse HKA activity in the gill and kidney. To determine whether HKA was specifically involved in Rb⁺ uptake, we performed a pharmacological inhibition assay of Rb⁺ uptake with the gastric HKA inhibitors omeprazole and SCH28080. In addition, a tissue viability assay confirmed the stability of the ex vivo preparations. Specifically, the lactate dehydrogenase (LDH) assay was used as an indicator of cell damage by measuring the activity of this enzyme in the culture medium leaked from damaged cells. Finally, the expression of Atp4a and Atp4b in the gill and kidney was established using immunohistochemistry (IHC) with custom-made antibodies to identify the presence and subcellular location of these subunit proteins within these tissues, followed by validation by western blotting and confirmation of atp4a and atp4b expression by RT-PCR.

Nile tilapia, Oreochromis niloticus (Linnaeus 1758), were supplied by Sand Plains Aquaculture facility (London, ON, Canada), with mass ranges of 5–25 g for in vivo and 50–200 g for ex vivo experiments. Upon arrival, the fish were placed in 180 l tanks supplied with dechlorinated city water and reverse osmosis water at a room temperature of 25±1°C. Fish were fed twice daily with Ewos pellets ad libitum. All experiments were approved by Wilfrid Laurier University's Animal Care Committee (R14002 and R18003) and followed the guidelines from the Canadian Council on Animal Care.

Rubidium offers several advantages that makes it a suitable replacement for potassium for flux studies. Rubidium has a similar physicochemical behaviour and characteristics to those of potassium. For instance, both are alkali metals and their hydration radii are similar. Also, Rb⁺ at high concentrations of exposure has limited toxicity compared with potassium. In the environment and in the body, Rb⁺ is present at very low concentrations; therefore, accumulation within the organism during flux studies is indicative of uptake from the water during exposure (Dietz and Byrne, 1990; Sanders and Kirschner, 1983; Tipsmark and Madsen, 2001; Mähler and Persson, 2011; Wilcox and Dietz, 1995).

Rubdium uptake rate was determined by measuring rubidium accumulation in muscle over time using terminal sampling. Oreochromis niloticus (n=16, 6–15 g) were captured randomly from the 180 l holding tanks using a hand net, then moved to one of four 3 l tanks (n=4 per tank) on flow through with aeration. The water was maintained at ∼pH 7.8. Fish were acclimated overnight, and the Rb⁺ flux was initiated by stopping flow to the tanks and adding a specific final concentration of Rb⁺ (0.1, 0.5, 1 or 2.5 mmol l⁻¹ RbCl) to each tank. An initial water sample of 10 ml was taken to check the water [Rb⁺]. At 0, 3, 6 and 12 h, O. niloticus were killed with an overdose of MS222 (0.2 g l⁻¹; Syndel Laboratories, Nanaimo, BC, Canada) and epaxial muscle samples (usually 0.2–1 g) were excised for measurement of [Rb⁺]. Sample processing is described in ‘Analytical techniques: Rb⁺ kinetics’, below.

Omeprazole was used as a proton pump inhibitor to evaluate the activity of HKA for this experiment by comparing sham-injected controls and omeprazole­-injected fish. Treatment fish were injected with 5 mg kg⁻¹ omeprazole dissolved in DMSO and 0.9% sodium chloride, 12 h before and at the start of the Rb uptake periods. The control fish were injected with an equivalent volume of 0.9% NaCl with 2% DMSO. For injection, fish were anaesthetized with 0.2 g l⁻¹ of MS222, weighed and the injection volume calculated. Oreochromis niloticus were moved to individual 500 ml tanks with aeration and flow to the tanks was stopped after the second injection. Rb⁺ was added to each tank at a final concentration of  0, 0.1, 0.5, 1 or 2.5 mmol l⁻¹ at time zero. After 3 or 6 h, fish (n=4) were killed with an overdose of MS222, their length and mass were measured, and muscle samples were collected as described above in ‘Rb⁺ kinetics’.

Muscle samples were weighed, dried at 60°C for 72 h, dissolved in 2.5 ml concentrated HNO₃, and diluted 50× with Milli-Q water and spiked with CsCl at a final concentration of 0.2%. Rb⁺ concentration was determined using a PinAAcle 900T Atomic Absorption Spectrophotometer (Perkin Elmer, Waltham, MA, USA) in flame mode. Rubidium standards were prepared in 0.2% CsCl and 2% HNO₃ in Milli-Q water in a range from 0 to 9.6 μmol l⁻¹.

Rb⁺ kinetics were characterized from Rb⁺ uptake rate at the tested concentrations of 0.1, 0.5 and 1 mmol l⁻¹ using Michaelis–Menten kinetics with GraphPad Prism (version 6.01 for Windows, GraphPad Software, La Jolla, CA, USA). V_max (maximum velocity) and K_m (substrate concentration at which the reaction rate is half V_max) were calculated by linear regression analysis of the Lineweaver–Burk plot equation plotted as 1/substrate (Rb concentration) on the x-axis and 1/velocity (uptake rate) on the y-axis. The x-intercept represents −1/K_m and the y-intercept is 1/V_max.

Oreochromis niloticus (n=3) were randomly sampled from the 180 l holding tanks using a hand net, and killed with an overdose of MS222 as described above for sample collection. The gill arches and kidney were excised and rinsed in Ringer's solution [in mmol l⁻¹:140 NaCl, 15 NaHCO₃, 1.5 CaCl₂, 1.0 NaH₂PO₄, 0.8 MgSO₄, 5.0 d-glucose and 5.0 N-(2-hydroxyethyl)-piperazinepropanesulfonic acid (Hepps); 310 mosmol kg⁻¹, pH 7.8] equilibrated with 99% O₂/1% CO₂. Each gill arch was then cut transversely into blocks of 3–5 pairs of filaments (5–10 mg), and kidney tissue was cut into blocks of 4–8 mg. Pieces were placed in 24-well culture plates and incubated in 1 mmol l⁻¹ Rb⁺ Ringer’s solution for up to 30 min in various RbCl concentrations (0, 0.1, 0.5 and 1 mmol l⁻¹, and a higher concentration of 2.5 mmol l⁻¹). Following incubation, tissues were washed free of Rb⁺, using Tris-sucrose buffer (in mmol l⁻¹: 2.5 KCl, 1.5 CaCl₂, 1.0 KH₂PO₄, 0.8 MgSO₄, 10 Tris, 260 sucrose, pH 7.8), equilibrated with 99% O₂/1% CO₂. The samples were then blotted gently on filter paper, weighed to the nearest 0.1 mg and placed in 1.5 ml tubes. Sample ion extractions were performed overnight in 0.5 ml of 5% trichloroacetic acid (TCA) at 24°C. The next day, samples were homogenized with a bead homogenizer (Precellys 24, Bertin Technologies SAS, Montigny-le-Bretonneux, France) and centrifuged (14,000 g for 15 min at 4°C; Thermo Scientific Sorvall Legend Micro 21R Microcentrifuge, Thermo Scientific, Waltham, MA, USA). Finally, the supernatants were diluted 50× with Milli-Q water and spiked with CsCl (0.2% final concentration). Rubidium standards prepared in 2% TCA, 0.2% CsCl ranged from 1.2 to 12.25 μmol l⁻¹. Rb⁺ concentration was measured by PinAAcle 900T Atomic Absorption Spectrophotometer (Perkin Elmer) in flame mode at 780.8 nm with a slit width of 2.0 nm. The rate of Rb⁺ uptake is expressed as μmol g⁻¹ wet mass h⁻¹. Additional gill, kidney, stomach, anterior intestine, rectum, liver, brain, testis, urinary bladder and spleen samples were collected at the time of killing for western blotting and PCR analyses.

Gill and kidney samples were transferred to 24-well microplates and incubated with 1 mmol l⁻¹ Rb⁺ Ringer solution equilibrated with 99% O₂/1% CO₂ for 30 min and omeprazole was added to the solution at a final concentration of 0.5 mmol l⁻¹ from a stock solution of 19.2 mmol l⁻¹ in methanol. An initial time course experiment over 0, 10, 20 and 30 min to determine an optimal inhibition time showed a linear uptake of Rb⁺ over time in the presence or absence of omeprazole. Omeprazole inhibition of Rb⁺ uptake with different concentrations of Rb⁺ (0.5, 1, 1.5 and 3.5 mmol l⁻¹ RbCl) was then carried out; the control was an equivalent volume of methanol.

Gill tissue Rb⁺ uptake was also assessed in the presence of SCH28080 (Santa Cruz Biotech) added to a final concentration of 0.2 mmol l⁻¹ from a stock solution of 36.1 mmol l⁻¹ dissolved in DMSO. An initial experiment was carried out for the time course of Rb⁺ uptake at 0, 10, 20 and 30 min. We then evaluated Rb flux and inhibition rates with different Rb⁺ concentrations (0.5, 1, 1.5 and 3.5 mmol l⁻¹ RbCl); the control was an equivalent volume of DMSO.

LDH assays were performed to evaluate LDH activity in the tissue (intracellular) and the leakage of enzyme into the medium (extracellular), which can be used as a marker of dead or damaged cells following treatment. Total LDH was calculated as the sum of extracellular LDH plus the intracellular (tissue) LDH. To measure extracellular LDH, 1 ml of culture medium (Rb⁺ Ringer's solution) was transferred from the 24-well microplates into 1.5 ml tubes and snap frozen on dry ice. To measure intracellular LDH, tissues were rinsed with phosphate-buffered saline (PBS; in mmol l⁻¹: 137 NaCl, 2.7 KCl, 8.5 Na₂HPO₄, 1.5 KH₂PO₄, pH 7.4), blotted dry and homogenized in 500 μl PBS to release all intracellular LDH and then centrifuged. The LDH in the supernatant was assayed on the same day or frozen at −80°C for later measurement.

In a 96-well plate, 10 μl of sample (tissue or medium) was added to each well containing 200 µl of 0.32 mmol l⁻¹ NADH dissolved in phosphate buffer (PB; 50 mmol l⁻¹ KH₂PO₄, pH 7.4). Then, 10 μl of 7.36 mmol l⁻¹ of pyruvate in PB was added and a linear (negative) slope obtained using a microplate reader (SpectroMax M2 plate reader; Molecular Devices, San Jose, CA, USA) at 340 nm for 10 min, using SOFTmax Pro 4.0 software. LDH leakage was expressed as a percentage of LDH activity in the extracellular medium versus the total activity.

Gill, kidney and stomach tissues collected from the in vivo experiments were placed in embedding cassettes in 3% paraformaldehyde in PBS (pH 7.4) fixative for 24 h. The tissues were then transferred to decalcification solution (30% formic acid, 13% sodium citrate) for 48 h, and then stored in 70% alcohol (histological grade) for at least 24 h. Tissues were then embedded in paraffin using a Shandon Citadel 1000 processor (Thermo Scientific, Pittsburgh, PA, USA). The paraffin blocks were sectioned (5 μm) using a Leica RM2125 RTS microtome (Leica Biosystems, Wetzlar, Germany), collected on to aminopropylsilane-coated slides (aminopropylsilane from Sigma-Aldrich, St Louis, MO, USA), air dried and stored for future use or prepared for IHC.

Antigenic peptides specific for Atp4a and Atp4b were determined from predicted O. niloticus amino acid sequences from the Nile tilapia genome (Ensembl ENSONIG00000005974 and ENSONIG00000003180, respectively). To avoid cross-reactivity with NKAα and β subunits, the predicted peptide sequences were screened against O. niloticus Atp1a1 (uniprot.org I3K3G3) and Atp1b3a (ENSONIG00000027418.1), respectively. Peptide prediction, synthesis and conjugation to the carrier protein KLH were performed by Genetel (Madison, WI, USA). The O. niloticus Atp4a carboxyl-terminal peptide DEIRKLGVRRHPGSWWDQELYY (amino acids 999–1020) and Atp4b amino terminal peptide MATLKEKRTCGQRCEDFG (amino acids 1–18) were selected. Two chickens per antigenic peptide were used and IgYs were isolated from egg yolks as Atp4b-B1349 and Atp4b-B1350 antibodies and Atp4a-B1351 and Atp4a-B1352 antibodies. Pre-immune isolated IgYs served as negative controls. Screening of staining in tilapia stomach gastric glands was used as a positive control of antibody specificity (Table S1).

Tissues were dewaxed at 60°C for 20 min, passed through three xylene baths (5 min) and an alcohol series of 2×100% (5 min) and 70% (3 min), with or without antigen retrieval. This included 0.05% citraconic anhydride for 30 min at pH 7.4 at 100°C (Namimatsu et al., 2005) followed by 1% SDS/PBS (pH 7.4) treatment (Brown et al., 1996). The sections were then washed three times with deionized water (5 min) and TPBS (5 min; 0.05% Tween-20 in PBS). Sections were blocked in a humidity chamber (20 min) with BLØK (Life Technologies) or 1% fish skin gelatin (Sigma-Aldrich) in TPBS, then probed with the primary antibodies Atp4a-B1352 and Atp4b-B1349 at 1:5000 and 1:500 for the gill and kidney, respectively, and mouse monoclonal antibody for Na⁺/K⁺-ATPase (NKA) α5 (1:100; clone a5, Developmental Studies Hybridoma Bank, Iowa, USA); chicken pre-immune sera diluted 1:5000 or 1:500 for the gill and kidney, respectively, were the negative controls (Burry, 2011). Slides were incubated for 1–2 h at 37°C in a humidity chamber, then washed three times with TPBS (5, 10, 15 min) with intermittent agitation. The secondary antibodies were goat anti-mouse IgG Alexa 555 (1:500) and goat anti-chicken IgY Alexa 488 (1:500) applied for 1–2 h at 37°C, followed by a final series of washes with TPBS, adding 5 μl DAPI (4′,6-diamidino-2-phenylindole) to the second wash buffer. Double sequential chicken labelling experiments were also done using the secondary antibodies goat anti-chicken IgY Fab Alexa 488 (1:500), and goat anti-chicken IgY CF640 (1:500) antibodies and a Fab blocking step (Negoescu et al., 1994). Coverslips were mounted with 1:1 glycerol PBS, pH 7.4, containing 0.1% NaN₃. Sections were viewed on a Leica DM5500 B microscope with a Hamamatsu C11440 Orca-Flash 4.0 digital camera.

Frozen gill, kidney, stomach, anterior intestine, liver, testis, urinary bladder and spleen collected from fish used in the ex vivo experiments were thawed and homogenized in ice-cold imidazole buffer (IB; 50 mmol l⁻¹ imidazole, pH 7.5) with a Precellys24+ bead homogenizer (6500 rpm, 2×10 s; Bertin) and centrifuged at 5000 g for 10 min at 4°C. The supernatant (S1) was decanted and centrifuged at 21,100 g for 20 min at 4°C, and the second supernatant (S2) was removed, and the pellet resuspended in IB by sonication. An equal volume of the homogenized pellet was mixed with 2× Laemmli's buffer, heated at 70°C for 10 min and stored at 4°C. Total protein in the resuspended pellet was measured with the BCA assay, and samples in Laemmli's buffer were then adjusted to 1 μg μl⁻¹. Western blotting was performed as described in Wilson et al. (2007) using a Bio-Rad Tetra mini electrophoresis system (Hercules, CA, USA) with a 10% resolving gel and 4% stacking gel. Transfer to PVDF membranes (Immun-Blot^®, Bio-Rad) was performed using a Hoeffer TE22 wet cell (1 h at 100 V). Membranes were dried, marked, stained with 0.5% Ponceau S, and blocked with 10% BLOTTO in TTBS (Tris-buffered saline with 0.05% Tween-20) before probing with Atp4a and Atp4b antibodies. Preimmune IgY was used as a negative control. Goat anti-chicken HRP-conjugated secondary antibody (1:50,000) was used to detect immunoreactive bands by enhanced chemiluminescence (Clarity, Bio-Rad) using an imager (c300, Azure Biosystems, Dublin, CA, USA).

Total RNA was extracted from a panel of tilapia organs (gill, kidney, stomach, anterior intestine, rectum, liver and brain) using the Bio-Rad Aurum Total RNA extraction kit with on-column DNaseI digestion according to the manufacturer's instructions. The concentration and purity of the RNA were assessed by spectrophotometry using a Biotek Take-3 plate and only samples with 260 nm/280 nm ratios between 1.8 and 2.2 were processed further. For each sample, 1 μg of RNA was converted to cDNA using the Applied Biosystems™ High-Capacity cDNA Reverse Transcription Kit according to the manufacturer's instructions.

PCR was run using BESTaq master mix (Applied Biological Materials Inc.) with either atp4a- or atp4b-specific primers at 250 nmol l⁻¹ each in 10 μl reaction volumes with 0.5 μl of cDNA (Table S2). The PCR consisted of an initial denaturation at 94°C for 3 min, followed by cycling steps of denaturation at 94°C for 10 s, annealing at 60°C for 10 s and extension at 72°C for 5 s for a total of 30 cycles. Subsequently, 5 μl of each sample were run on 2% agarose TBE (89 mmol l⁻¹ Tris-borate and 2 mmol l⁻¹ EDTA, pH 8.3) gels containing GelRed (Biotium Inc., Fremont, CA, USA) at 50 V for 80 min using a Multipet-one electrophoresis apparatus. Gels were imaged using an Azure c300 imager.

Data were analysed as means±s.e.m. Two-way ANOVAs were used to test for differences between treatment groups and time or Rb⁺ concentrations for uptake/inhibition rate data. If a significant difference was found by ANOVA, a post hoc Holm–Šidák test was performed. In the case of non-normality and/or unequal variance, an equivalent, non-parametric test was used (Mann–Whitney rank sum test). A paired t-test was used to compare differences in the ex vivo gill SCH28080 experiment. Significance was accepted when P<0.05. All statistical analyses were carried out using SigmaPlot (v.11, Systat, San Jose, CA, USA).

Rubidium uptake was successfully measured over time at a water Rb⁺ concentration of 2.5 mmol l⁻¹. Rb⁺ uptake, measured as accumulation in white muscle, showed an increase over time to a maximum at 12 h (Fig. 1A). Using a range of Rb⁺ concentrations in the millimolar range, the data showed a Michaelis–Menten relationship and linear regression of the Lineweaver–Burk plot where the Rb⁺ V_max was 20.4 nmol g⁻¹ h⁻¹ and the K_m value was 0.4 mmol l⁻¹ (Fig. 1B). Rb⁺ uptake was not inhibited by omeprazole in vivo as there was no significant difference in Rb⁺ uptake rate between sham-injected (control) and omeprazole-injected O. niloticus (n=4, P=0.842) (Fig. 1C).

As the in vivo methods performed to evaluate the activity of gastric HKA in Rb⁺ uptake showed no effect, another methodology using ex vivo gill filaments was examined. While ex vivo experiments facilitate manipulation of the experimental conditions, tissue viability can be an issue. Therefore, tissue viability was monitored over the course of the experiments.

The appearance of the intracellular enzyme LDH in the culture medium was used as an indicator of tissue viability. Higher levels of LDH activity were found in gill compared with kidney tissue (Table 1). Gill and kidney showed stable viability throughout the examination phases, whether with control or treated O. niloticus tissues, even at the longer times examined (30 min). In gill preparations, the highest leakage measured was only 0.12% in control and 0.11% in omeprazole treatments and this was in the initial 0 min group. All other groups showed leakage of less than 0.01% and there was no significant relationship with time or treatment group. In the kidney, leakage rate was also very low. The highest leakage for control and omeprazole treatment groups was 0.06% and 0.04%, respectively. Consequently, no differences were found (P=0.693).

Rubidium uptake by the gill was measured over 30 min in media containing 1 mmol l⁻¹ RbCl (Fig. 2A). Omeprazole in ex vivo treated gills significantly inhibited Rb⁺ uptake compared with the control and no significant difference in tissue Rb⁺ concentration was observed between 10 and 30 min within this treatment group. In ex vivo kidney preparations, a similar trend was observed, with significant increases in tissue Rb⁺ concentration over 20 min in control tissue and a significant inhibition after 10 min in omeprazole-treated kidney tissue for the duration of the experiment (Fig. 2B; n=3, P≤0.001).

Flux measured over 15 min with Rb⁺ concentrations of 0.5, 1, 1.5 and 3.5 mmol l⁻¹ indicated a significant inhibition by omeprazole (n=3) in the gill (Fig. 2C) and kidney (Fig. 2D) (P≤0.001). In the gill, a significant inhibition was observed at 3.5 mmol l⁻¹ Rb⁺ while in the kidney, a significant inhibition was observed at 1.5 and 3.5 mmol l⁻¹ Rb⁺.

SCH28080-treated O. niloticus gill filaments showed a significant inhibition of Rb⁺ uptake at 10 min (n=3, 2.88±0.11 μmol g⁻¹ h⁻¹) compared with the control (5.75±1.22 μmol g⁻¹ h⁻¹; paired t-test, P≤0.001). Rb⁺ uptake in the kidney following treatment with SCH28080 was not measured.

The stomach was used as a positive control for IHC as all antibodies cross-reacted strongly with the gastric glands (Fig. S1). However, the Atp4a and Atp4b antibodies B1350 and B1351, respectively, showed additional non-specific staining in the stomach lumenal epithelium and were therefore not used for gill and kidney IHC staining experiments. The reported Atp4a and Atp4b localizations were performed with the B1352 and B1349 antibodies, respectively.

Optimized IHC staining conditions for the gill and kidney were found: (1) at dilutions of 1:5000 and 1:500, respectively; (2) using fish skin gelatin in the blocking medium instead of BLØK for gills tissue – no difference in staining was found in the kidney; (3) with 1–2 h incubation at 37°C rather than overnight at 4°C; and (4) without the need for staining with antigen retrieval steps [0.05% citraconic anhydride (Namimatsu et al., 2005) or 1% SDS/PBS (Brown et al., 1996)].

Ionocytes, identified by NKA staining, were present mainly in the filament epithelium with a round shape (Figs 3B and 4B). Immunolocalization of both Atp4a and Atp4b was found in the apical region of these ionocytes (Figs 3A,C and 4A,C). The appearance of staining varied from diffuse apical to restricted along the apical membrane.

The different nephron regions of the Nile tilapia kidney can be distinguished based on tubule cell morphology and NKA immunolocalization. In the proximal tubule, NKA staining was weak and restricted to the basal membrane; the distal tubule had the strongest NKA staining with an extensive basolateral location, while the collecting tubule/duct also had strong NKA staining in a basolateral location. HKA staining with both Atp4a and Atp4b antibodies was found apically in the collecting tubule and was absent from proximal and distal tubule regions (Fig. 5A,D,E,H).

Through the chicken–chicken double labelling technique, the Atp4a and Atp4b antibodies showed clear apical overlap in staining ionocytes (positive NKA immunoreactivity) in gills (Fig. 6) and tubular cells in the kidney collecting tubule segment (Fig. S2). Negative control pre-immune IgYs showed no staining comparable to either Atp4a or Atp4b antibodies and this was not affected by changes in the IHC conditions (Fig. S3). Also, pre-incubation of antibodies with excess peptides eliminated specific staining (data not shown).

The tilapia Atp4a and Atp4b peptides have a predicted molecular mass of 113 and 33 kDa, respectively. In Atp4a western blots, a diffuse band around 110 kDa was detected in the gill and a more compact band in stomach although not in the kidney (Fig. 7). Band immunoreactivity increased with protein loading and was not present in blots probed with preimmune IgY, as demonstrated in the gill (Fig. S4a). Atp4a expression at the predicted molecular mass was not found in other organs [anterior intestine, liver and spleen, although weak expression of a slightly smaller protein was detected in urinary bladder and testis, while in the latter a larger diffuse band (∼150 kDa) was also detected (Fig. S4b)]. We were unable to produce consistent western blotting results with our Atp4b antibodies and thus no results are shown.

RT-PCR confirmed that both atp4a and atp4b were expressed in the gill and kidney although at apparently much lower levels than in the stomach (Fig. 8). In brain, atp4a was also detectable, while weak atp4b expression was detected in anterior intestine. No expression was detected in rectum or liver samples for either HKA gene.

We show here, for the first time, the expression and potential role of HKA in the gill and kidney of a teleost, O. niloticus, functioning as an extra-gastric proton pump based on in vivo and ex vivo experimental approaches and immunohistochemistry. The in vivo and ex vivo experiments used rubidium as a surrogate flux marker for potassium uptake: carrier-mediated uptake was demonstrated in vivo and pharmacological inhibition was demonstrated ex vivo by omeprazole and SCH28080 in the kidney and gill. It is well recognized in tetrapods that the stomach is the primary organ for acid secretion mediated by the proton pump HKA. Consequently, in this study, the stomach was used as a positive control to validate custom-made HKA α and β subunit antibodies to confirm the molecular identity of the extra-gastric proton pump in the gill and kidney. Specific gastric gland immunoreactivity was observed in O. niloticus stomach, confirming antibody specificity, whereas staining was found apically in gill ionocytes and kidney collecting duct, consistent with the role of these cells in ion and acid–base regulation. The kidney immunoreactivity in the collecting duct segment was consistent with the contribution to K⁺ regulation through reabsorbing K⁺ and acidifying the tubular fluid, as seen in mammals (Koeppen, 2009; Wingo and Cain, 1993). Taken together with the findings for atp4a and atp4b RT-PCR, these results strongly support the presence (IHC) and function (Rb⁺ uptake) of HKA in gill and kidney potassium regulation.

We demonstrated that the uptake of potassium by the tilapia gill was carrier mediated. The V_max and the apparent K⁺ affinity K_m in O. niloticus indicated a direct, substrate-independent mechanism and/or carrier-mediated potassium uptake. In rainbow trout, K⁺ uptake has also been measured, although the mechanism has not yet been elucidated (Eddy, 1985; Gardaire and Isaia, 1992; Gardaire et al., 1991). Eddy (1985) reported that in trout, K⁺ uptake rate was only 3–5% of Na⁺ and Cl⁻ uptake (Eddy, 1975); however, in studies with lungfish and lamprey, Rb⁺ flux rate was 20% and 10% of Na⁺ and Cl⁻ flux, respectively (Doherty, 2016; Wilkie et al., 1998, 2007). In tilapia, Rb⁺ uptake rate was approximately 11 μmol kg⁻¹ h⁻¹, which was approximately 3% of Na⁺ uptake rate (Oreochromis mossambicus 360 μmol kg⁻¹ h⁻¹; Flik et al., 1989; Potts et al., 1967) and in agreement with the observation made by Eddy (1985).

Several classes of drugs have been investigated to evaluate the activation of the pump by inhibiting the acid secretion by HKA using, for example, omeprazole, ouabain, SCH28080 and vanadate (Hersey et al., 1988). However, in this study, only omeprazole and SCH28080 were tested for pharmacological inhibition as these PPIs and APAs have the highest binding affinity and covalently bind with HKA (Singh et al., 2013). They are widely used for treating all acid secretion-related disorders in the stomach, and omeprazole was the first clinically useful acid-activated drug. However, it cannot be activated at neutral pH, and its activation is related to the pH level in terms of other factors including dosage and duration of treatment (Shin and Sachs, 2010). According to Shin et al. (2009), SCH28080 was a compound developed to control acid secretion by gastric glands; however, unlike omeprazole, it does not require acid activation to effectively reduce the acid secretion of HKA. The role of the pumps in potassium uptake based on pharmacology was demonstrated by pharmacological inhibition experiments using a proton pump inhibitor at a concentration predicted to be sufficient for inhibition of HKA. However, only the ex vivo methods showed a definitive result, whereas the in vivo results showed no clear pharmacological support for the hypothesis that the gastric HKA plays a role in acid–base or K⁺ regulation by the gills of O. niloticus.

Omeprazole is available in both oral and injectable forms (Worden and Hanna, 2017) and has few side effects (Schubert, 2017). Although less common, the injectable form is an effective means of gastric ulcer treatment and gastric acidification inhibition (Worden and Hanna, 2017). In the present study, 40 mg kg⁻¹ omeprazole was injected intraperitoneally overnight 12 h before the experiment and then at the beginning of the experiment. However, as no inhibition of Rb⁺ uptake was observed, we assumed that either there was not sufficient time for the drug to reach the gills, considering the duration of acid inhibition (48 h) allowing strong binding to HKA (Shin et al., 2009) or the drug was not activated, given that binding and inhibition of omeprazole are dependent on acidification, or maybe it was cleared and excreted by the fish. The second option is supported by a study of Shin and Sachs (2010), which indicates that the activation rate of omeprazole correlates with the pH in the cells; therefore, it is possible that under in vivo conditions, acidification was not sufficient to inhibit the gill HKA as the pH was not adjusted during the experiments. In future studies, in vitro measurements of HKA activity from the organs of treated fish would address this question of effective pump inhibition. It is also possible that the omeprazole was cleared as in mammalian studies, omeprazole is rapidly cleared from the plasma with a half-life of less than 1 h because of rapid hepatic metabolization (Larsson et al., 1985). However, it is also possible that alternative mechanisms for K⁺ uptake might compensate for the inhibition of branchial HKA such as the paracellular uptake mechanism described by Horng et al. (2017) in medaka.

As mentioned above, the kinetics of the accumulation of Rb⁺ over time suggested the presence of carrier-mediated uptake of Rb⁺. This promising observation led us to look for further pharmacological evidence of HKA using a different methodology, an ex vivo preparation. With this preparation, we had more control over the tissue and the experimental surroundings, including the pH, which is known to be an essential factor for the activation and binding of the proton pump inhibitors. The ex vivo findings along with the Rb⁺ uptake results from the in vivo experiment support the hypothesis that the gastric-type HKA plays a role in K⁺ and possibly acid–base regulation in O. niloticus. This is the first evidence of the presence of HKA in teleost gills and kidney. We then used omeprazole and SCH28080 as specific inhibitors for the gastric HKA, aimed at reducing Rb⁺ uptake and demonstrating the presence of the gastric HKA in the gills and kidney. Gumz et al. (2010), in their review, reported that the gastric HKA in tetrapod kidney is inhibited by SCH28080 and it is sensitive to omeprazole, which is supported by our results showing an inhibition of Rb⁺ uptake with these inhibitors in both the kidney and gill.

This type of ex vivo approach was developed and validated by Tipsmark and Madsen (2001) in the gill and kidney of salmonid fish with the inhibitor ouabain to study NKA. As we were studying a different pump (HKA) that is sensitive to different inhibitors (omeprazole and SCH28080), some modifications were made to the approach to fit the experiments. As expected, gill and kidney inhibition rates were significantly noticeable within 10 min of the start of the experiment, demonstrating that the inhibitors were blocking the uptake of Rb⁺ and reducing the pump activity as time increased. Although inhibition with SCH28080 was only tested on the gill at 10 min, it showed a high degree of inhibition (50%).

A potential challenge with all ex vivo preparations is tissue degradation. Although the tissues were capable of taking up Rb⁺ over 30 min, further compelling evidence for tissue integrity was provided by the LDH activity measurements. The leakage rate of this cytosolic enzyme into the medium was used as an indicator of cell damage and death. The resultant average leakage rate out of the tissue was extremely low (0.03%) during ex vivo experiments. However, the only point when the gills showed a leakage rate higher than the average was at time 0 h (0.1%), when residual activity from sample preparation would be expected. These activity levels were still very low, indicating the maintenance of tissue integrity during the course of the experiment and significantly no correlation with inhibitor treatments was observed. In comparable studies that measured LDH activity levels, the same range of values was reported in Mugil auratus (81 U g⁻¹ wet tissue mass; Krajnovic-Ozretic and Ozretic, 1987) and Fundulus grandis (51.8 U g⁻¹ wet tissue mass; Diaz, 2014).

The IHC results are the first to demonstrate that a gastric HKA in a FW teleost is expressed in extra-gastric sites where it may function in K⁺ absorption. Choe et al. (2004) were the first to detect expression of HKα1 (Atp4a) in the gills of a FW elasmobranch (Dasyatis sabina), and Hentschel et al. (1993) in the kidney of a marine elasmobranch (Scyliorhinus caniculus). These results suggested that the HKα1 isoform of elasmobranchs is similar to that in rat gastric parietal cells and renal intercalated cells, with 81.4% similarity to the rat HKα1 protein sequence. Smolka et al. (1994) had already confirmed the cross-reactivity of the antibodies used in these studies with proteins found in the gastric glands. While all the type IIc P-ATPases share homology, alignment of D. sabina with O. niloticus Atp4a isoforms from GenBank shows that the O. niloticus protein shares similarities in many regions, including the invariant phosphorylation site and the predicted transmembrane regions of type HKα1. Furthermore, the O. niloticus sequence is 86% identical to that of D. sabina.

Nile tilapia (O. niloticus) stomach was shown to be immunoreactive to antibodies generated against Atp4a and Atp4b, indicating that they recognize the O. niloticus gastric proton pump, which was confirmed by Atp4a western blotting and transcript expression of both subunits by RT-PCR. The results in gills indicate that gastric HKA immunoreactivity occurred apically in a subpopulation of mitochondrion-rich chloride cells or ionocytes. This cell type with its different subtypes in teleost fish gills is the site of ion uptake (Wilson and Laurent, 2002; Dymowska et al., 2012), although the subtypes that are responsible for K⁺ uptake had not been identified until this work. In the diadromous Atlantic stingray, Choe et al. (2004) identified a similar apical staining pattern in gill ionocytes (intercalated cells) for Atp4a using a heterologous antibody. However, the HKA uptake mechanism of ionocytes may not be the exclusive means of K⁺ uptake given that, in medaka, paracellular K⁺ uptake has been measured in-between keratinocytes in skin (Horng et al., 2017) with ionocytes instead having been identified as sites of K⁺ excretion (Abbas et al., 2011; Horng et al., 2017). The HKA-expressing ionocytes may constitute a new ionocyte subtype in FW tilapia; however, additional work is required to determine its relationship with Type I (NKA only), II (apical NCC and basolateral NBC1 and NKA) and III (apical NHE3 and basolateral NKCC and NKA) described in Mozambique tilapia (O. mossambicus) (Hwang et al., 2011; Dymowska et al., 2012).

Both HKA isoforms are localized to the colleting duct of mammalian kidney, and similarly, the gastric HKA immunoreactivity was also found in the collecting duct of O. niloticus, thus emphasizing its role in K⁺ reabsorption and possibly acid–base balance. This segment has the ability to acidify the tubular fluid by secreting H⁺ and reabsorbing K⁺, thus playing a role in acid–base and potassium regulation in tetrapods (Koeppen, 2009; Gumz et al., 2010). In contrast, in the kidney of a marine elasmobranch (Scyliorhinus caniculus), gastric HKA immunoreactivity was found in the late distal tubule as well as the proximal tubule (Hentschel et al., 1993).

The IHC results in O. niloticus gill are supported by both Atp4a western blotting and atp4a and atp4b RT-PCR. The absence of supporting western blotting data for Atp4b is not ideal but is also not without precedence when using antibodies (e.g. NHE2, Ivanis et al., 2008; NHE3b, Ito et al., 2014). Antibodies may not necessarily be cross-reactive with peptides in different formats (histological sections versus SDS-PAGE immunoblots) where protein conformation differs (native versus denatured, respectively) and therefore antibody binding sites or epitopes may change (Harlow and Lane, 1988). In the case of the kidney, while IHC and RT-PCR presented positive findings for both HKA subunits, western blotting for Atp4a was negative in contrast to the gill and stomach. However, the rationale for the lack of Atp4b antibody cross-reactivity cannot be applied to the case of Atp4a in the kidney. There is the possibility of insufficient sensitivity in the technique to detect the kidney Atp4a. Future work is needed to address this possibility.

The main finding of this study is that tilapia can take up potassium and that HKA is involved. Potassium regulation is important to animals. Animals, including humans, generally obtain potassium through food at levels that do not pose health risks; however, minor disturbances (increase–decrease) in potassium concentration might cause serious health problems. In mammals, for example, unfavourable effects might occur when they are subjected to higher than normal K⁺ plasma concentrations (>5.0 mmol l⁻¹: hyperkalaemia), or when K⁺ plasma concentrations are lower than the normal range (3.5–5.0 mmol l⁻¹: hypokalaemia). Both hyperkalaemia and hypokalaemia result from disruptions in transcellular homeostasis or in the renal regulation of K⁺ excretion (Talling, 2010). However, as a normal kidney can excrete hundreds of milliequivalents of K⁺ daily, excessive K⁺ intake is an uncommon cause of hyperkalaemia without other contributing factors. Therefore, if renal K⁺ excretion is impaired, whether through drugs, renal insufficiency or other causes, then excess K⁺ intake can produce hyperkalaemia. In contrast, K⁺ depletion (hypokalaemia) by the kidney is commonly caused by polyuria (higher rates of urination) and polydipsia (excessive thirst), which can lead to excessive drinking and dilution of K⁺ in the blood. Sustained K⁺ depletion can severely affect systems and organs, leading to cardiovascular and neurological disturbances and impaired muscle and kidney function. In aquatic animals, maintenance of potassium homeostasis is also vital. Although little work has been done in fishes, it would be predicted that strenuous exercise and a diet and environment (e.g. freshwater) poor in K⁺ would result in hypokalaemia whereas diets and environments (e.g. seawater) rich in K⁺ would be predicted to result in hyperkalaemia if regulatory mechanisms were impaired.

Based on our findings of an apical localization of HKA in gill ionocytes and kidney collecting duct, we predict the following model to explain HKA's role in K⁺ uptake (Fig. 9). HKA uses the energy from ATP hydrolysis to power the electroneutral exchange of intracellular H⁺ for extracellular K⁺ (Caplan, 1998). The uptake of K⁺ would be against its electrochemical gradient. Intracellular K⁺ would exit the cell down its electrochemical gradient across the basolateral membrane via an inwardly rectifying K⁺ channel (Evans et al., 2005). The IHC results show that these cells are also rich in basolateral NKA; however, its role in K⁺ uptake is likely to be indirect and/or related to driving other transport processes. It should be noted that K⁺ is but one of many ions that may be transported by gill ionocytes for ion and acid–base regulation (Evans et al., 2005).

In summary, this research is the first to demonstrate the expression and role of the gastric proton pump HKA as an extra-gastric pump (in the gill and kidney) in a teleost species. Evidence for this includes omeprazole- and SCH28080-induced inhibition of the gastric proton pump in O. niloticus gills and kidney, suggesting a role in K⁺ reabsorption and possibly acid–base regulation. Immunohistochemistry demonstrated that the pump was localized apically in the gill ionocyte and the kidney collecting duct segment. Future ex vivo and in vivo experiments in different teleost species are vital to better understand the role of HKA in teleost physiology and to determine how widespread it is. Of particular interest will be determining what role the pump plays in acid–base regulation.

We would like to thank Mr R. Busby of Sand Plains Aquaculture for providing tilapia used in this study and Drs J. C. McGeer, M. P. Wilkie and P. Craig for comments on the thesis on which this paper is based (Barnawi, 2017). The α5 monoclonal antibody developed by Douglas M. Fambrough at Johns Hopkins University was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology.