Specimens of A. testudineus (20–50 g) were procured from a fish farm in Singapore. Fish specimens were acclimated and maintained as described in Chen et al. (2017). Feeding was stopped 2 days before the experiment (Tay et al., 2006). Approval to carry out the experimental procedures has been received from the Institutional Animal Care and Use Committee, National University of Singapore (IACUC 021/10 and 098/10).

Control fish (N = 10; N = 4 for molecular work, N = 3 each for Western blot and immunofluorescence microscopy) were submerged in 25 volumes (v/w) of fresh water. For seawater exposure, fish were progressively acclimated to a daily increment in salinity from fresh water to seawater (salinity 30) through a 6-day period (day 1: fresh water, day 2: salinity 10, day 3: salinity 15, day 4: salinity 20, day 5: salinity 25, day 6: salinity 30), and then maintained in salinity 30 for 1, 3 or 6 days (N = 18; N = 12 for molecular work, N = 3 each for Western blot and immunofluorescence microscopy). Waters of different salinities were prepared by mixing seawater with fresh water in different proportions. Fish were subjected to 100 mmol l⁻¹ NH₄Cl in fresh water (N = 18; N = 12 for molecular work, N = 3 each for Western blot and immunofluorescence microscopy) following the procedures of Chen et al. (2017). Fish were anesthetized and killed before tissue sampling according to the methods described in Chen et al. (2017). Gills were excised immediately and snap-frozen in liquid nitrogen before storing at −80°C or processed for immunofluorescence microscopy as described below.

Total RNA was extracted from gill samples and purified as described by Chen et al. (2017). RNA concentration was subsequently determined with a BioSpec-nano (Shimadzu, Tokyo, Japan) and the quality of the RNA samples determined electrophoretically before using for cDNA synthesis.

The partial nhe3 sequence was obtained using gene-specific primers (forward: 5′-GTCCACGTCAACGAGGTC-3′ and reverse: 5′-ATTCCACTTGTCCCTCATGTAG-3′) designed from the conserved regions of various nhe3 mRNA sequences from fish species obtained from National Centre Biotechnology Information (NCBI). The PCR reaction was prepared using the methods and thermocycling conditions stated in Chen et al. (2017). The PCR product of the estimated size was extracted and purified using the FavorPrep™ Gel Purification Mini Kit (Favorgen Biotech Corporation, Ping-Tung, Taiwan) and sequenced.

RACE-PCR was carried out with the Advantage® 2 PCR kit (Clontech Laboratories), using gene-specific primers (5′ RACE: 5′-AGCCCGTGTCGAGGATGATTTGAG-3′ and 3′ RACE: 5′-GAACACAGGCTTCATCCTCCTCACAC-3′), which were designed based on the partial sequence of nhe3. The thermocycling conditions used were identical to those of Chen et al. (2017). RACE-PCR products were separated electrophoretically and multiple rounds of sequencing were performed in both directions. Analysis of the assembled sequence was performed using Bioedit v7.1.3 and the complete cDNA coding sequence of nhe3 has been deposited to Genbank with accession number KU555941.

The deduced Nhe3 amino acid sequence was aligned and compared with Nhe3/NHE3 amino acid sequences of Oryzias latipes, Oreochromis mossambicus, Xenopus tropicalis, Homo sapiens, and the most thoroughly characterized isoform, NHE1, from H. sapiens, using BioEdit. The transmembrane (TM) domains were determined using the PSIPRED secondary structure prediction server (http://bioinf.cs.ucl.ac.uk/psipred/). After aligning the sequences with ClustalX2, phylogenetic analysis was performed, using maximum likelihood analysis and 100 bootstrap replicates, with Phylip v3.6. Nhe3/NHE3 amino acid sequences of selected animal species were obtained from Genbank or UniProtKB/TrEMBL with the following accession numbers: Oncorhynchus mykiss Nhe3 (ABO32815.2), Danio rerio Nhe3a (ABU68834.1), D. rerio Nhe3b (ABU68830.1), Tribolodon hakonensis Nhe3 (BAB83083.1), Cynoglossus semilaevis Nhe3 (XP_008329339.1), O. mossambicus Nhe3 (BAF80347.1), O. latipes Nhe3 (XP_011479629.1), X. tropicalis Nhe3 (XP_004915375.1), Hynobius nigrescens Nhe3 (BAF76797.1), Oryctolagus cuniculus NHE3 (AAA31420.1), Fukomys damarensis NHE3 (KFO38498.1), Rattus norvegicus NHE3 (AAA41702.1), Mus musculus NHE3 (XP_006517084.1), Macaca fascicularis NHE3 (XP_005556582.1), H. sapiens NHE3 (NP_004165.2), and Patiria pectinifera Nhe3 (ABQ10588.1) as the outgroup.

Total RNA (4 μg) from gill samples was reverse transcribed, and qPCR reaction prepared and performed as described by Chen et al. (2017), using 0.2 μmol l⁻¹ of gene-specific qPCR primers (forward: 5′-ACAGTAGAAACACAACAAGGCA-3′ and reverse: 5′-CACCACTCTCACCTTCATCAG-3′). Threshold cycle, as C_t value, was collected at each elongation step. Melt curve analysis was performed, and the PCR products were separated in a 2% agarose gel to verify the presence of a single product.

Absolute quantification was adopted in this study, as it was crucial to assess the transcript levels of nhe3 across different experimental conditions in the gills of A. testudineus, which was not possible to achieve through relative quantification. In order to produce a pure amplicon (standard) of a defined region of nhe3 cDNA from the gills of A. testudineus, PCR was performed with the gene-specific qPCR primers and cDNA as a template according to the method described by Gerwick et al. (2007). After gel separation and purification as described above, the nhe3 fragment was cloned using pGEM®-T Easy vector (Promega Corporation, Madison, WI, USA) and the presence of the insert was confirmed by sequencing. The recombinant clone was quantified using the BioSpec-nano (Shimadzu), and serially diluted (from 10⁶ to 10² copies/2 μl) to obtain the standard curve using the default settings of the StepOnePlus™ Software v2.3 (Thermo Fisher Scientific, Waltham, MA, USA). The PCR efficiency for nhe3 was 90.8%. The quantity of transcript in each sample was determined from the linear regression line derived from the standard curve.

A rabbit polyclonal anti-Nhe3 antibody was designed against residues 736–749 (GKSPDRSRSYHSGD) of the deduced Nhe3 sequence from A. testudineus (GenScript, Piscataway, NJ, USA). Anti-Nhe3 (10 μg) was pre-incubated with the immunizing peptide of Nhe3 (150 μg) in a final volume of 10 ml at 25°C for 1 h to validate the specificity of the antibody.

Gill samples were prepared for Western blotting following the methods described in Chen et al. (2017). Briefly after homogenization, the homogenate was centrifuged and protein concentration of the supernatant determined (Bradford, 1976) before diluting to 5 μg μl⁻¹ with Laemmli buffer (Laemmli, 1970). Proteins (50 μg) were separated by SDS-PAGE [8% acrylamide (for anti-Nhe3) or 12% acrylamide (for anti-actin) as the resolving gel and 4% acrylamide as the stacking gel] and transferred onto a PVDF membrane. The detection of Nhe3 and actin (Developmental Studies Hybridoma Bank, Iowa City, IA, USA) were performed using Pierce™ Fast Western kit, SuperSignal™ West Pico Substrate (Thermo Fisher Scientific), following the manufacturer's instruction. After transfer, primary antibody incubation (anti-Nhe3 antibody, 1:500 dilution or anti-actin antibody, 1:10,000 dilution) was performed as described by Chen et al. (2017) and subsequently incubated with an optimized concentration of anti-rabbit or anti-mouse horseradish peroxidase-conjugated secondary antibody, respectively, for 15 min at 25°C. Bands for both Nhe3 and actin were visualized by chemiluminescence and processed following the methods described by Chen et al. (2017). Band intensities were quantified using ImageJ (version 1.41, NIH) and the results (arbitrary densitometric units) were presented as the relative protein abundance of Nhe3 normalized with the protein abundance of actin.

Paraffin sections of gill samples from A. testudineus were prepared following the methods described by Chen et al. (2017). Antigen retrieval was carried out with 0.05% citraconic anhydride in a boiling water bath (Namimatsu et al., 2005). After blocking with 5% BSA/TPBS, sections were incubated with: (1) the rabbit anti-NKAαRb1 polyclonal antibody (a pan-specific antibody originally designed by Ura et al., 1996 for labeling Nka α-subunits) co-labeled with the mouse anti-Nkaα1b monoclonal antibody (Abmart, Shanghai, China; Ching et al., 2013) or the mouse anti-Nkaα1c polyclonal antibody (Genscript; Chen et al., 2017) to examine whether Nkaα1b and Nkaα1c were present in different types of ionocyte, (2) the custom-made rabbit anti-Nhe3 antibody (Genscript) co-labeled with the mouse anti-Nkaα1a monoclonal antibody (Abmart), or the mouse anti-Nkaα1b monoclonal antibody (Abmart), or the mouse anti-Nkaα1c polyclonal antibody (Genscript) to determine the type of Nka α-isoform associated with Nhe3 in the gills of A. testudineus, and (3) the mouse anti-Nkaα1b antibody (Abmart) or the mouse anti-Nkaα1c antibody (Genscript) co-labeled with the rabbit anti-Rhcg1 antibody (kindly provided by S. Hirose) or the rabbit anti-Rhcg2 antibody (Genscript; Chen et al., 2017) to examine if they were localized to the same type of ionocyte.

The rabbit polyclonal anti-Rhcg1 antibody was raised against residues 425–488 of D. rerio Rhcg1 (Nakada et al., 2007). The anti-Nkaα1a and anti-Nkaα1b antibodies have been validated to be isoform-specific with no cross-reactivity previously by Ching et al. (2013) while the specificities of the anti-Rhcg1 and anti-Rhcg2 antibodies have been validated formerly through a peptide competition test by Chen et al. (2017). To validate the specificity of anti-Nkaα1c binding in this study, anti-Nkaα1c (6.25 μg) was pre-incubated with the immunizing peptide of Nkaα1c (62.5 μg) provided by Genscript, in a total volume of 10 ml at 25°C for 1 h before performing Western blotting and the results are shown in Supplementary Figure S1.

The anti-Nkaα1a and anti-Nkaα1b antibodies were diluted 1:1,000 in 5% BSA/TPBS (blocking buffer). The anti-Rhcg1 and anti-Rhcg2 antibodies were diluted 1:500 in 5% BSA/TPBS. The anti-Nkaα1c antibody was diluted 1:200 in 5% BSA/TPBS and the anti-Nhe3 antibody was diluted 1:100 in 5% BSA/TPBS. Secondary antibody incubations were performed as described by Chen et al. (2017). Both primary and secondary antibody incubations were performed at 37°C for 1 h. After rinsing thrice with TPBS, sections were mounted and cured in ProLong® Gold Antifade mounting medium (Thermo Fisher Scientific) before viewing and images were captured following the methods described in Chen et al. (2017). Adjustments, if necessary, were done to the entire image without altering the integrity of the data.

All analyses were performed using the statistics software, SPSS version 18 (IBM Corporation, Armonk, NY, USA). Levene's test was used to verify the homogeneity of variances. One-way analysis of variance was performed with multiple comparisons of means by Tukey's test (for equal variances) or Dunnett's T3 (for unequal variances) for means acquired from absolute quantification. Differences in relative band intensities from Western blot were analyzed with Independent Samples T-test. All statistical tests were two-tailed and the significance level was set at P < 0.05.

The complete cDNA coding sequence of nhe3 obtained from the gills of A. testudineus comprised 2,628 bp (Genbank accession number KU555941), coding for 876 amino acids with an estimated molecular mass of 97 kDa (Figure 1). The deduced Nhe3 sequence had 12 TM regions (Figure 1). A phylogenetic analysis revealed that Nhe3 of A. testudineus was more closely related to the Nhe3 sequences of teleosts than those of amphibians and mammals (Figure 2). An alignment of Nhe3 from A. testudineus with various animal species indicated highly conserved residues (E252, D257, and E383; corresponding to E262, D267, and E391 of human NHE1) involved in coordinating Na⁺ binding (Figure 1). Coordination of H⁺ binding and regulation of intracellular pH are known to be performed by the following residues in NHE: E131, R440, G455, G456, H480, and H500, and five of these residues (except for H480) were conserved in Nhe3 of A. testudineus. Nhe is a target for inhibition by amiloride (Benos, 1982); of the six residues identified as crucial for determining amiloride sensitivity in NHE1, three residues were conserved in Nhe3 of A. testudineus (F151, G164, and E338; corresponding to F161, G174, and E346 of human NHE1) (Figure 1). Notably, there were three amino acid substitutions (L153F, H341S, and G344S; corresponding to L163, H349, and G352 of human NHE1) within the amiloride-binding site in the Nhe3 sequence of A. testudineus.

There were significant increases in the mRNA expression levels of nhe3 in the gills of A. testudineus after 1 day (by 130-fold), 3 days (by 172-fold), or 6 days (by 105-fold) of exposure to seawater, as compared to the freshwater control (Figure 3A). By contrast, a significant decrease (by 52%) in the transcript level of nhe3 was observed in the gills of A. testudineus exposed to 100 mmol l⁻¹ NH₄Cl for 3 days, although it returned to control levels on the sixth day (Figure 3B).

Immunoreactive band of Nhe3 was detected near the molecular mass of ~85 kDa which was slightly less than the expected molecular mass of 97 kDa. The protein abundance of Nhe3 increased significantly (by 2.6-fold) in the gills of A. testudineus exposed to seawater for 6 days (Figure 4A), but remained unchanged in the gills of fish exposed to 100 mmol l⁻¹ NH₄Cl for 6 days (Figure 4B). The validity of antibody binding was confirmed by a peptide competition test (Figure 4A).

The expression of Nhe3 was undetectable in the gills of A. testudineus kept in fresh water or exposed to 100 mmol l⁻¹ NH₄Cl in fresh water (Figures 5A, 6A–C). When acclimated to seawater, the gills of A. testudineus expressed Nkaα1b and Nkaα1c in the basolateral membranes of two distinct types of seawater-inducible ionocyte (Supplementary Figure S2). Results obtained from the gills of the seawater-exposed fish revealed for the first time that Nhe3 had a different subcellular localization in two distinct types of ionocyte which were immunoreactive to either the anti-Nkaα1b antibody (Figure 5B) or the anti-Nkaα1c antibody (Figure 5C). Nhe3 was detected in the apical membranes of a large proportion of Nkaα1b-labeled ionocytes (Figure 5B), but it was expressed in the basolateral membranes of Nkaα1c-labeled ionocytes (Figure 5C). In either case, there were ionocytes that were labeled with only Nhe3, and these ionocytes probably expressed the other seawater-specific Nkaα isoform (either Nkaα1b or Nkaα1c) in the basolateral membranes (Figures 5B,C). Additionally, the majority of the Nkaα1b-labeled ionocytes in the gills of the seawater-acclimated fish expressed Rhcg1 (Figure 7A) and Rhcg2 (Figure 7B) on the apical and basolateral membranes, respectively. By contrast, the Nkaα1c-labeled ionocytes in the gills of the seawater-acclimated fish expressed only basolateral Rhcg2 (Figure 8B) without apical Rhcg1 (Figure 8A).

Similar to members of the NHE family, Nhe3 from A. testudineus comprised an N-terminal domain, with 12 TM regions, which is mainly responsible for the catalysis of amiloride-sensitive electroneutral Na⁺/H⁺ exchange. It also had a C-terminal domain that serves a regulatory function (Wakabayashi et al., 1992). Residues involved in the coordination of cation binding and intracellular pH regulation, which are characteristic of the well-studied human NHE1 (Murtazina et al., 2001; Wakabayashi et al., 2003; Slepkov et al., 2007), were highly conserved in the Nhe3 of A. testudineus.

Site-specific mutagenesis demonstrates that the acidic side chains of E262, D267, and E391 are essential for NHE activity (Murtazina et al., 2001). Mutation of E262 and D267 effectively eliminates transport activity while mutation at E391 reduces NHE activity partially, thereby showing the importance of the oxygen atom in the side chains of these negatively-charged amino acids in Na⁺ binding (Boyer, 1988; Murtazina et al., 2001). An arginine residue (A440) and two glycine residues (G455 and G456) have been identified to be critical for intracellular pH sensing as they reside in the putative “pH_i sensor” site in NHE1, and regulate the pH_i sensitivity of NHE1 in opposite directions (Wakabayashi et al., 2003). E131 is another residue crucial in regulating Na⁺ transport and the pH sensitivity of NHE1. Charge reversal mutations of E131 result in a shift in the intracellular pH dependence of NHE1 activity and eliminate NHE1 response to a hyperosmotic medium (Hisamitsu et al., 2007). As the imidazole side chains of histidine residues can be protonated, mutations at H480 and H500 in NHE3 alters the binding of protons at these positions (Cha et al., 2002). This significantly affects the pH sensitivity of the mutant NHE3 as its activity threshold is shifted to a more acidic region. Of the nine residues stated above as important for NHE activity, eight of them (with the exception of H480) were present in Nhe3 of A. testudineus. The replacement of H480 in Nhe3 of A. testudineus is probably a common phenomenon as substitution at this position was observed in all the Nhe3 sequences of teleosts examined here. These highly conserved residues further confirm the importance of the Na⁺ and H⁺ binding sites and the “pH_i sensor” site in Nhe3 of A. testudineus and suggests that the ion exchange activity of Nhe3 in A. testudineus probably functions via a similar mechanism of action.

Different NHE isoforms display varying affinity for its inhibitors, amiloride and its analogs, in the following order of sensitivity: NHE1 ≥ NHE2 > NHE5 > NHE3 > NHE4 (Masereel et al., 2003; Orlowski and Grinstein, 2004). Six amino acids (F161, L163, G174, E346, G352, and H349) are implicated as residues important for inhibitor binding and hence, NHE1 function. L167F (corresponding to L163 in human NHE1) mutation in hamster increases the resistance of NHE1 to amiloride but has no effect on Na⁺ transport (Counillon et al., 1993). In fact, the leucine at this position of all the Nhe3/NHE3 sequences examined in this study was replaced with either phenylalanine or tyrosine. Hence, this difference could be characteristic of Nhe3/NHE3 and contribute to the difference in amiloride affinity between NHE1 and NHE3 (Masereel et al., 2003; Orlowski and Grinstein, 2004). Likewise, the substitution of H349 with serine was common among all the Nhe3/NHE3 examined in this study, and H349S in NHE1 is known to cause a change in its amiloride sensitivity (Wang et al., 1995).

Two working models have been proposed for apical Na⁺ uptake in the gills of freshwater teleosts, depending on the fish species and the environment of its habitat: (1) an apical H⁺-ATPase electrically coupled to Na⁺ uptake via ENaC and (2) an electroneutral exchange of Na⁺ and H⁺ by apical Nhe (Hwang and Lee, 2007). For instance, stenohaline zebrafish depends largely on H⁺-ATPase for Na⁺ uptake, but the euryhaline medaka relies on Nhe3 instead (Guh et al., 2015). A relatively high expression of Nhe3 in relation to Na⁺ uptake in a freshwater environment have been confirmed in many teleosts (Wu et al., 2010; Kumai and Perry, 2011; Shih et al., 2012). By contrast, the expression of nhe3/Nhe3 was weak in the gills of freshwater A. testudineus. Hence, it is uncertain whether Nhe3 plays a significant role in branchial Na⁺ absorption in A. testudineus in a hypoosmotic environment. This could perhaps be elucidated in the future by examining the effects of fresh water with low [Na⁺] on the expression levels of nhe3/Nhe3 in its gills.

By contrast, the expression levels of branchial nhe3/Nhe3 increased substantially in the seawater-acclimated A. testudineus. Although the observed molecular mass (85 kDa) for Nhe3 in A. testudineus is slightly smaller than the estimated size (97 kDa), it is not an uncommon phenomenon. “Gel-shifting” is usually observed for membrane proteins and can be attributed to the effects of differential detergent binding and folding when the protein is electrophoresed by SDS-PAGE (Rath et al., 2009). The increase in the expression of nhe3/Nhe3 in the gills of A. testudineus during seawater exposure is unsurprising as branchial Nhe3 immunoreactivity has been detected in many fish species acclimated to seawater (Evans et al., 2005; Watanabe et al., 2008; Hwang et al., 2011; Christensen et al., 2012; Seo et al., 2013), thereby implying its involvement in H⁺ secretion during acid-base regulation in a hyperosmotic environment (Claiborne et al., 2002; Choe et al., 2005). Increases in respiratory and metabolic rates in a hyperosmotic environment can result in acidosis and hence, the prompt elimination of any excess H⁺ from the system via Nhe3 is essential for the maintenance of acid-base balance (Seo et al., 2013). The uptake of Na⁺ in exchange for the excretion of H⁺ would undoubtedly aggravate the ionic load that a fish in seawater will encounter. However, the resulting influx of Na⁺ may account for only a small portion of the total intracellular concentration of Na⁺ and thus, may be “worth” the additional energetic costs to maintain acid-base balance (Evans, 1984). Furthermore, the high Na⁺ concentrations in the ambient seawater would favor Na⁺/H⁺ exchange (Potts, 1994), down a favorable Na⁺ concentration gradient generated by the basolateral Nka (Choe et al., 2002).

The gills of the seawater-acclimated A. testudineus express two distinct types of seawater-inducible Nkaα-immunoreactive ionocyte labeled with either Nkaα1b or Nkaα1c (Supplementary Figure S2), which probably serve disparate physiological functions in seawater. While the Nkaα1c-immunoreactive ionocytes also express basolateral Nkcc1a and apical Cftr indicating that they are responsible for active salt excretion, the Nkaα1b-immunoreactive ionocytes lack Nkcc1a and Cftr, suggesting that they are involved in some other physiological processes which have yet to be defined (unpublished findings from X.L. Chen and Y.K. Ip). Of note, it has been established that A. testudineus increases the production and excretion of ammonia when exposed to seawater (Chang et al., 2007). The energy required for increased syntheses of certain branchial proteins and transport of ions during seawater acclimation can be supported by increases in protein degradation and amino acid catabolism. It is possible that the Nkaα1b-immunoreactive ionocytes take part in ammonia excretion if they express apical Nhe3 together with some sort of ammonia transporters.

Ammonia excretion in fishes is achieved largely through the transport of NH₃ down a favorable P_NH3 gradient across the branchial epithelium, with the excreted NH₃ being “trapped” by H⁺ excreted through the apical H⁺-ATPase and/or Nhe (Weihrauch et al., 2009; Wright and Wood, 2009; Hwang et al., 2011). Using specific morpholino oligonucleotides to knock down Rhcg1 translation in zebrafish embryos, Shih et al. (2008) showed that NH4+ efflux occurred favorably in H⁺-ATPase-rich cells and there was a drop in NH4+ efflux upon knockdown of Rhcg1 or H⁺-ATPase, or inhibition of H⁺-ATPase with bafilomycin or Nhe with EIPA. However, in some teleosts such as the mummichog (Edwards et al., 2005) and medaka (Wu et al., 2010), H⁺-ATPase is not the key player in H⁺ secretion. In fact, it has been demonstrated in embryos of medaka that Nhe3 and Rhcg1 were co-localized to a type of ionocyte, and ammonia excretion is inhibited by EIPA, but not bafilomycin, implying that Nhe3 plays a critical role in ammonia excretion (Wu et al., 2010). Using a scanning ion-selective electrode to measure H⁺ gradients, Liu et al. (2013) detected an acidic boundary layer at the yolk-sac surface of seawater-acclimated medaka larvae. They demonstrated that nhe3, rhbg, rhcg1, and rhcg2 expression levels were up-regulated in the medaka larvae exposed to ammonia in seawater, suggesting that these transporters were involved in H⁺-facilitated ammonia excretion. Furthermore, Rhcg1 and Nhe3 have been co-localized in Nkaα-immunoreactive ionocytes in the gill, kidney, and skin of the mangrove killifish, whereby the apical Nhe3 apparently facilitates ammonia excretion (Cooper et al., 2013). More importantly, there are evidence to support a positive correlation between NH4+ excretion and Na⁺ uptake in fish gill and skin (Shih et al., 2008; Wu et al., 2010; Cooper et al., 2013; Liu et al., 2013).

As expected, Nhe3 was localized to the apical membrane of the Nkaα1b-immunoreactive ionocytes, which co-expressed apical Rhcg1 and basolateral Rhcg2, in the gills of A. testudineus exposed to seawater. Therefore, it is logical to infer that H⁺-facilitated ammonia excretion occurs in seawater-exposed A. testudineus to augment increased ammonia production, which does not occur in the freshwater control. In seawater, NH₃ in the blood probably enters the ionocyte through the basolateral Rhcg2 and exits via the apical Rhcg1 to the external medium. The excreted NH₃ can be “trapped” as NH4+ by the H⁺ excreted through the apical Nhe3. This generates an outwardly-directed ΔP_NH3 to facilitate branchial ammonia excretion in A. testudineus during seawater exposure. Nkaα1b is essential to transport the intracellular Na⁺ to the blood, in order to maintain intracellular Na⁺ homeostasis and provide the driving force necessary for Nhe3 to function. The excess Na⁺ would finally exit the branchial epithelium paracellularly. In this way, A. testudineus is able to effectively excrete ammonia and regulate the intracellular pH of those ionocytes participating in the process.

In fish gills, Nhe3 is usually localized to the apical membranes of ionocytes, and basolateral Nhe3 localization has not been reported before. This is the first report on Nhe3 being expressed in different membranes (apical or basolateral) of two different types of ionocyte in the gills of a fish. Of note, NHE1 is ubiquitously expressed in the basolateral membrane of epithelial cells (Donowitz et al., 2013) including branchial ionocytes of fishes (Claiborne et al., 1999). However, the Nhe isoform localized to the basolateral membrane of the Nkaα1c-immunoreactive ionocytes in A. testudineus could not be Nhe1 as the epitope for Nhe3 was uniquely designed against the deduced amino acid sequence of Nhe3, and did not match the Nhe1 sequence obtained from the gills of A. testudineus (X.L. Chen and Y.K. Ip, unpublished results). Furthermore, based on the NCBI non-redundant protein database, the epitope sequence selected to raise the custom-made anti-Nhe3 antibody displayed high similarity with equivalent sequences of Nhe3/NHE3 from various animals, as revealed by the top 10 hits of the BLAST results (Table 1). Although the Nkaα1c-labeled ionocytes displayed basolateral Rhcg2-labeling, the lack of apical Rhcg1-labeling indicates that they do not have the same physiological function as the Nkaα1b-immunoreactive ionocytes. At present, the exact function of basolateral Nhe3 in the Nkaα1c-immunoreactive ionocytes of the seawater-exposed A. testudineus is unknown, but we speculate that it could serve to regulate blood pH and maintain pH homeostasis in the Nkaα1c-immunoreactive ionocytes of A. testudineus.

During emersion or ammonia exposure, active ammonia excretion is one of the most effective ways to ameliorate ammonia toxicity in fishes, as it can maintain low internal ammonia concentrations and prevent the brain from ammonia intoxication (see Chew et al., 2006; Ip and Chew, 2010; Chew and Ip, 2014 for reviews). Some air-breathing fishes can actively excrete ammonia through their gills; these include the climbing perch (Tay et al., 2006), the giant mudskipper (Periophthalmodon schlossei; Randall et al., 1999; Chew et al., 2003, 2007; Ip et al., 2004a) and the African sharptooth catfish (Ip et al., 2004b). Periophthalmodon schlossei is capable of lowering the pH of the external medium (Chew et al., 2003, 2007), and branchial H⁺ excretion can prevent a back flux of NH₃ during active NH4+ excretion (Ip et al., 2004a). However, there are differences in the mechanisms of active ammonia excretion between the gills of A. testudineus (Loong et al., 2011; Ip et al., 2012a,b) and the gills of P. schlosseri (Chew et al., 2014b,a; Chew and Ip, 2017), as the former is incapable of lowering the pH of the external medium in response to environmental ammonia (Ip et al., 2012b). The current model of active branchial ammonia excretion in A. testudineus denotes that ammonia may permeate through the basolateral Nkcc1 (Loong et al., 2011) or the basolateral Rhcg2 (Chen et al., 2017) and exit the Nkaα1c-immunoreactive ionocytes as NH4+ through the apical Rhag (Chen et al., 2017), driven by a trans-apical membrane electrical potential generated by the efflux of HCO3- or Cl⁻ via the apical Cftr (Ip et al., 2012b).

Indeed, our results indicate that Nhe3 is unlikely to play a significant role in active NH4+ transport in the gills of A. testudineus. The mRNA expression levels of nhe3 in the gills of A. testudineus remained low during 6 days of exposure to 100 mmol l⁻¹ NH₄Cl in fresh water, with a significant decrease on day 3. The low branchial expression of nhe3 (<30 copies per ng total RNA) implies that Nhe3 is unlikely to be of physiological significance. This is further supported by the lack of significant changes on the protein abundance of branchial Nhe3, and the undetectable level of apical Nhe3 in the branchial epithelium, in A. testudineus exposed to ammonia. Taken together, our results corroborate the report on the inability of A. testudineus to lower the pH of the ambient water during ammonia exposure (Ip et al., 2012b).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.