Section-specific H⁺ flux in renal tubules of fasted and fed goldfish

The maintenance of acid–base homeostasis is one of the most important physiological processes in all organisms as it ensures proper enzyme and protein function (Somero, 1986; Riggs, 1988). The kidney has the predominant function of generating and transporting HCO₃⁻ and ammonia (here referred to as the sum of NH₃ and NH₄⁺) in mammals, and thus ultimately regulates any metabolic acid–base disturbance by adjusting net acid (both titratable acid and NH₄⁺) and base (HCO₃⁻) excretion (Hamm et al., 2015). In teleost fishes, by contrast, the predominant organ involved in regulating acid–base disturbances is the gill (Perry and Gilmour, 2006), while the kidney seems to play only a small but significant role in acid–base balance. For instance, the kidney of the freshwater rainbow trout, Oncorhynchus mykiss, is known to compensate for environmental disturbances that induce acidosis, such as exposure to low environmental pH (McDonald and Wood, 1981), hyperoxia (Wheatly et al., 1984), hypercapnia (Perry and Fryer, 1997) and exhaustive exercise (Wood, 1988), as well as for a feeding-induced alkaline tide (Bucking et al., 2010). The mechanisms of renal ammonia and acid–base regulation in teleosts, however, remain poorly studied to date. This difficulty stems, in part, from the structural heterogeneity and architectural complexity of the teleost kidney, which contains multiple tissue types and a convoluted tubule (Sakai, 1985). Consequently, most physiological studies have only investigated acid–base regulation by the kidney as a whole (e.g. Wood et al., 1999; Lawrence et al., 2015; Wright et al., 2014). However, recently, we (Fehsenfeld and Wood, 2018) micro-dissected the kidney of the goldfish, Carassius auratus, enabling us to profile mRNA and protein expression of renal epithelial transporters involved in acid–base homeostasis along the renal tubule and its different segments.

In general, renal ionoregulation in freshwater teleosts involves epithelial transporters that are important for the re-absorption of ions to minimize solute loss into the dilute external medium (Wheatly et al., 1984; Curtis and Wood, 1992; Perry et al., 2003b). Transport of primary ions and electrolytes is coupled to the transport of acid–base equivalents, hence linking ionoregulation and acid–base balance in the teleost kidney (Perry et al., 2003b; Wheatly et al., 1984; Dantzler, 2016). Many of the same transporters, including V-type H⁺-ATPase (V-ATPase), Na⁺/H⁺ exchanger (NHE), Na⁺/HCO₃⁻ co-transporter and HCO₃⁻/Cl⁻-exchanger, can also be found in the teleost gill epithelium (Perry et al., 2003b). Similarly, these transporters are involved in acid–base regulation in the mammalian kidney (Weiner and Verlander, 2013).

In fasting C. auratus, the kidney produces alkaline urine, resulting in net renal base excretion. Exposure of animals to a low pH environment, however, results in net renal acid excretion (Lawrence et al., 2015). In contrast to the rainbow trout, the goldfish is agastric, like the freshwater killifish Fundulus heteroclitus (Wood et al., 2010), having an alkaline digestive tract, and therefore exhibits a systemic acidic tide after feeding. We (Fehsenfeld and Wood, 2018) were able to show that a systemic acid-load (metabolic acidosis) caused by feeding resulted in increased acid excretion by the goldfish kidney, as indicated by a decrease in urine pH and an increase in urine inorganic phosphate and ammonia concentrations. This process was in part accomplished by differentially adjusting expression levels of both mRNA and protein of a variety of transporters in the proximal, distal and connecting tubule of the goldfish nephron (Fehsenfeld and Wood, 2018).

The present study aimed to further investigate the mechanism of renal acid–base regulation in response to feeding in C. auratus. Micro-dissection of intact kidney tubules in combination with the scanning ion-selective electrode technique (SIET) (Smith et al., 1994; Piñeros et al., 1998) allowed direct measurement of H⁺ flux at the basolateral surface of distinct segments of the renal tubule. Measurements from fasted and fed goldfish were used to demonstrate the effects of feeding on the acid–base transport physiology of the proximal, distal and connecting tubule. Furthermore, with the application of specific inhibitors, we were able to identify epithelial transporters involved in renal acid–base regulation by promoting and/or inhibiting H⁺ flux in the various renal tubule segments. We hypothesized that V-ATPase, NHE, HCO₃⁻ transporters (Na⁺/HCO₃⁻ co-transporter and/or HCO₃⁻/Cl⁻-exchanger) and carbonic anhydrase (the enzyme converting CO₂ to H⁺ and HCO₃⁻ and vice versa) contribute to the H⁺ flux through the renal epithelium. Furthermore, we hypothesized that (i) H⁺ flux would differ among the segments of the renal tubule and (ii) would be differentially affected by the inhibitors, as a result of the differential expression of transporters (at the mRNA and protein level) in each tubule segment, as shown in our previous study (Fehsenfeld and Wood, 2018).

Goldfish (∼5 cm, 2.5–3.0 g) were obtained from commercial suppliers and were held in 20 l aquaria at room temperature at a maximum density of 1 fish per 2 l for a minimum of 7 days before the experiments began. The recirculating and filtered tap water composition was (in mequiv l⁻¹): 0.6 Na⁺, 0.8 Cl⁻, 1.8 Ca²⁺, 0.3 Mg²⁺, 0.05 K⁺; titration alkalinity 2.1; hardness ∼140 mg l⁻¹ as CaCO₃ equivalents. Water parameters (pH 7.4, ammonia, nitrate and nitrite) were closely monitored and water changes were performed 2–3 times a week as necessary. Fish were fed to satiation 2–3 times a week with commercial flaked food (Nutrafin Max, Hagen, Montreal, QC, Canada). For experiments, fish were either fasted for 96 h (‘fasted’) or fasted for 93 h, then fed to satiation and used 3 h after feeding (‘fed’). Each animal was killed by cephalic concussion. All procedures were conducted under the approval of the University of British Columbia Animal Care Committee (licence no. A14-0251) and the guidelines of the Canadian Council on Animal Care.

Immediately following the experimental procedures described above, kidneys were removed and transferred into control goldfish Ringer solution containing (in mmol l⁻¹): 100.0 NaCl, 2.5 KCl, 1.5 CaCl₂·2H₂O, 1.0 MgCl₂·6H₂O, 5.0 NaHCO₃, 0.7 NaH₂PO₄·H₂O, 10 glucose, pH 7.3 (modified after Hoar and Hickman, 1975). First, larger clusters of tubules were pulled from the Wolffian duct as it runs through each half of the kidney. Subsequently, minuten pins glued to needles and screwed into syringes for better handling were used to carefully isolate single tubules including the glomerulus, proximal tubule, distal tubule and connecting tubule out of these clusters, as described in Fehsenfeld and Wood (2018). One intact tubule was transferred into the lid of a 5 cm Nunc cell culture dish and pinched down onto the dish at both ends using the needles. Lids, rather than the bottom part of the dishes, were used because tubules would stick to them due to apparent adhesive forces holding them in place that did not seem to be present in the bottom dish. The preparation was then transferred into the scanning ion-selective electrode technique (SIET) set-up and measurements were started immediately in control goldfish Ringer solution.

To determine control H⁺ flux in fasted and fed animals, renal tubules were scanned in control goldfish Ringer solution in 100 µm increments starting at the glomerulus and ending at the connecting tubule. To test for potential effects of time on tissue viability, the goldfish Ringer solution was renewed after the first scan, the tubules incubated for 10 min (to mimic addition of inhibitors as described below), and then rescanned at the same positions. For investigation of the effects of inhibitors on H₊ flux, the tubules of fed goldfish were incubated with the respective inhibitor for 10 min after the control scan, and then rescanned at the same locations as for the control Ringer solution. Inhibitors and their concentrations (and targets) were: 10 mmol l⁻¹ 2,4,6-triaminopyrimidine (TAP; paracellular pathway), 100 µmol l⁻¹ 5-(N-ethyl-N-isopropyl)amiloride (EIPA; NHE), 100 µmol l⁻¹ 4,4'-diisothiocyanatostilbene-2,2'-disulfonate (DIDS; HCO₃⁻/Cl⁻-exchanger and/or Na⁺/HCO₃⁻ co-transporter), 5 µmol l⁻¹ bafilomycin (V-ATPase) and 1 µmol l⁻¹ acetazolamide (carbonic anhydrase). There was no effect of the inhibitors at the indicated concentrations on the slope/responsiveness of the H⁺ microelectrodes.

Data are expressed as means±s.e.m. (N), where N represents the number of animals. All statistical analyses were performed with PAST3 (Hammer et al., 2001) or Minitab^® Statistical Software (State College, PA, USA). Data were checked for normal distribution with the Shapiro–Wilks test and homogeneity of variances with the F-test. If data did not meet these criteria for parametric testing, they were log transformed.

To investigate the effect of feeding (Fig. 1) and/or time (Fig. 2) on tubular H⁺ flux, the mean H⁺ flux for each segment (as indicated by horizontal lines in Figs 1 and 2 for proximal, distal and connecting tubule) was calculated for fasted and fed animals, respectively. Data were then analyzed by two-way ANOVA using a general linear model as implemented in Minitab 18 (Fig. 1: factor 1=fasted versus fed, factor 2=tubule segment; Fig. 2: repeated measures two-way ANOVA, factor 1=1st run versus 2nd run, factor 2=tubule segment). Subsequently, pairwise comparisons were conducted using Tukey's post hoc test.

For the inhibitor experiments (Figs 3–6), Student's paired t-tests were applied for each inhibitor in each section. Data were considered significantly different at P≤0.05. All graphs were generated with Inkscape 0.92.0 r15299 (https://www.inkscape.org).

In fasted goldfish, H⁺ efflux was observed from the tubule lumen and/or the epithelial cells over the basolateral epithelium to the external solution (Fig. 1B). Mean H⁺ flux rates did not differ significantly between the different segments of fasted animals (Fig. 1A). Two-way ANOVA revealed a significant effect of both tubule segment and feeding status on H⁺ flux, but not their interaction. Tukey's post hoc analysis identified H⁺ flux in the distal tubule of fed animals to be significantly lower (−57%) than that in the proximal and connecting tubule of fed animals. Mean H₊ flux increased significantly in fed animals compared with fasted animals in all three tubule segments (Tukey's post hoc analysis), and the difference was most pronounced in the proximal tubule (3.5-fold versus 2.4- and 2.2-fold increases in the distal and connecting tubule, respectively) (Fig. 1).

H⁺ flux during two consecutive runs with control saline in tubules of fed animals did not decrease significantly over time (Fig. 2; repeated measures two-way ANOVA). There was, however, a significant effect of tubule segment according to findings in fed tubules as mentioned above, with the distal tubule having significantly lower H⁺ flux compared with both the proximal and connecting tubule (Tukey's post hoc analysis). There was no significant interaction of time and tubule segment (P=0.7).

Consequently, tubules from fed animals were used in testing the effects of inhibitors. The application of 2,4,6-triaminopyrimidine (TAP; 10 mmol l⁻¹) (Fig. 3) to inhibit the paracellular pathway resulted in a dramatic ∼90% decrease of H⁺ flux in all three segments. Targeting epithelial transporters and enzymes with specific inhibitors differentially affected H⁺ flux along the tubule. Fig. 4 shows results for the proximal tubule, Fig. 5 shows results for the distal tubule and Fig. 6 shows those for the connecting tubule. Inhibiting NHE with EIPA decreased H⁺ flux by ∼55% in the proximal and connecting tubule, but did not affect H⁺ flux in the distal tubule. Applying DIDS to inhibit HCO₃⁻ transporters resulted in a decrease of H⁺ flux by 70% in the proximal tubule, and approximately 50% in the distal and connecting tubule. When V-ATPase was blocked by bafilomycin, H⁺ flux decreased by 67% in the proximal tubule and 45% in the distal tubule, while no effect was observed in the connecting tubule. Targeting carbonic anhydrase with acetazolamide decreased H⁺ flux by roughly 60% in all three segments.

Goldfish, C. auratus, produce an alkaline urine (pH 7.8–8.5) under fasted conditions (Lawrence et al., 2015; Fehsenfeld and Wood, 2018). This indicates renal reabsorption of H⁺ from the tubule lumen into the extracellular fluid/plasma and, indeed, this is what we measured in the present study by applying SIET to the proximal, distal and connecting tubule (Fig. 1). In response to the acidic tide caused by feeding, however, urine pH was observed to drop by up to 0.6 units (Fehsenfeld and Wood, 2018); under these conditions the kidney reversed this flux to net H⁺ secretion. Surprisingly, the results here appear to show the opposite effect, suggesting a pronounced increase in H⁺ reabsorption in response to feeding that was highest in the proximal tubule (3.5-fold versus ∼2.3-fold in distal and connecting tubule with reference to fasted conditions; Fig. 1). When the paracellular pathway was blocked with TAP, however, the observed H⁺ reabsorption was reduced to 10% of the control values for fed animals in all segments (Fig. 3). The teleost kidney has been reported to express a multitude of tight junction proteins: cytosolic scaffolding proteins cingulin and ZO-1 (Kolosov et al., 2014), as well as transmembrane proteins like occludin (Chasiotis and Kelly, 2008; Kolosov et al., 2017) and over 35 claudins (Kolosov et al., 2013), and the tricellular proteins ILDR2 and tricellulin (Kolosov and Kelly, 2013, 2018; Kolosov et al., 2017). Despite the fact that many tight junction proteins have been detected in the kidney, their individual role and distribution in the different regions of the fish nephron are unknown. Chasiotis and Kelly, however, showed that occludin was expressed in the distal and not the proximal tubule of goldfish (Chasiotis and Kelly, 2008) and the African clawed frog Xenopus laevis (Chasiotis and Kelly, 2009). Accordingly, rather than indicating a direct reabsorption of protons, the observed H⁺ flux probably represents H⁺ back-flux via the paracellular pathway down the concentration gradient established by active H⁺ secretion into the small closed tubular lumen.

We are confident that the tissue was intact and that this H⁺ back-flux was not due to tissue deterioration, because we took particular care to avoid touching the tubules themselves during the microdissection (i.e. they were not pulled on or squeezed). Rather, needles were mainly used to carefully tease apart the connective tissue. Additionally, we carefully inspected every tubule visually during and after the microdissection for any potential rips. Measurements were highly reproducible, and tubules of fasted animals generally had very low H⁺ flux.

As it was not possible to directly measure apical H⁺ flux, the observed H⁺ back-flux through the tight junction can be interpreted as a proxy for apical H⁺ secretion. Consequently, a proton gradient directed from the lumen to the plasma, resulting in the observed H⁺ back-flux, can be explained by the following: (1) active apical ion transport resulting in higher [H⁺] in the tubular lumen compared with extracellular [H⁺] and/or (2) ‘trapped’ urine of fed animals having increased [H⁺] in the lumen (pH 7.2) compared with the extracellular fluid (pH 7.85 of fasted animals; Fehsenfeld and Wood, 2018). Differences in the magnitude of H⁺ back-flux in the different segments of the tubule isolated from fed animals (Fig. 1A) can be attributed to a combination of regional alterations in (i) paracellular permeability, i.e. the tight junctions, and (ii) active ion transporter (epithelial channels and pumps) expression. In the intact kidney of fed freshwater fishes, such back-flux of H⁺ through the paracellular cleft would be minimized by continuous production and flow of copious amounts of dilute urine that would probably increase in the face of a systemic acidosis (Lawrence et al., 2015).

While the recent study by Fehsenfeld and Wood (2018) established a general molecular basis for renal acid–base regulation in different tubule segments of the goldfish kidney, the SIET measurements of the present study allowed us to establish real-time H⁺ flux data. Furthermore, by application of specific inhibitors, our results allow us to compare the observed H⁺ flux to the presence of epithelial transport proteins and further characterize their role in renal acid–base balance in the different tubule segments. Note that we elected to test the potential inhibitors only on preparations from fed animals, because H⁺ flux rates were low in fasted animals, so we cannot eliminate the possibility that different response patterns would have been seen in the latter. Here, we suggest models for general mechanisms of transport of acid–base equivalents (i.e. H⁺, HCO₃⁻ and NH₄⁺) in the proximal (Fig. 7), distal (Fig. 8) and connecting tubule (Fig. 9). No inhibitors are currently available for Rhesus proteins, the major transporters for NH₃/NH₄⁺. They are, however, likely to play a role in acid–base regulation, so we have incorporated them into our proposed models where appropriate. The models take into account inhibitor results from the present study, protein and mRNA expression data from Fehsenfeld and Wood (2018), as well as aspects from the mammalian models for renal acid–base (Hamm et al., 2013) and ammonia transport (Weiner and Verlander, 2013, 2017). While depicted here as one cell, it is possible that certain processes are instead associated with multiple, rather specific cell types as in the mammalian system.

NHE is generally believed to promote apical H⁺ secretion (and Na⁺ absorption) in the vertebrate kidney and specifically in the proximal tubule as the major site for renal acid excretion (Gilmour and Perry, 2009; Hamm et al., 2013; Ivanis et al. 2008; Teranishi et al. 2013). Furthermore, protein abundance of NHE (isoform 3) in the cortical brush-border apical membrane of the proximal tubule was reported to increase following chronic metabolic acidosis in rats (Ambühl et al., 1996). Likewise, in goldfish renal tubules, NHE3 protein was located in the basolateral membrane of all tubule segments of fasted goldfish, but increased in abundance and relocated to the apical membrane in the proximal tubule when animals experienced a systemic acid load as a result of feeding (Fehsenfeld and Wood, 2018). This indicates a similar mechanism in the goldfish kidney to that in the mammalian kidney.

Blockade of NHE(s) with the EIPA in the present study resulted in a decrease of H⁺ flux in the goldfish renal tubule, verifying its role in renal acid–base balance. Surprisingly, however, this was only observed in the proximal (Fig. 4) and connecting tubule (Fig. 6) of fed goldfish, and not in the distal segment (Fig. 5). It has to be kept in mind that the inhibitor may have acted not only on basolateral but also on apical NHE3 because of its permeability through the tight junctions (Van Itallie et al., 2008; Hou et al., 2013). Consequently, the decrease of H⁺ flux in the proximal tubule when inhibited by EIPA is in line with NHE-mediated apical proton secretion. Alternatively or additionally, EIPA may have blocked a different isoform of NHE, for example NHE1 that appears to be expressed in the basolateral membrane of the proximal tubules in mammals (Hamm et al., 2013; Bobulescu et al., 2005). NHE may be present in the distal segment of goldfish renal tubule, but in accordance with our measurements is probably not involved in acid–base regulation in this region, as has also been suggested for the mammalian distal tubule (Hamm et al., 2013).

An electrogenic Na⁺/HCO₃⁻ co-transporter (NBCe1A/NBC1) in basolateral membranes of the proximal tubule is crucial for the reabsorption of base in mammals (Hamm et al., 2013; Weiner and Verlander, 2013; Handlogten et al., 2015). NBC1 has also been identified in the trout kidney and was proposed to mediate increased basolateral reabsorption of HCO₃⁻ in response to a respiratory acidosis caused by hypercarbia (Perry et al., 2003a). Two anion exchangers have been identified in mammalian renal cells: AE1 (SLC4A1) is present in the basolateral membrane of A-type intercalated cells in the collecting duct, whereas pendrin (SLC26A4) is found in the apical membrane of the proximal tubule and B-type intercalated cells of the collecting duct (Hamm et al., 2013; Soleimani et al., 2001). Members of the SLC26 family of membrane transporters including pendrin have also been identified in zebrafish gills (Bayaa et al., 2009) and in V-ATPase-rich cells in the elasmobranch branchial epithelium (Piermarini and Evans, 2001; Tresguerres et al., 2005; Reilly et al., 2011). In response to a metabolic alkalosis (feeding-induced alkaline tide), elasmobranch gills translocated pendrin into the apical membrane with the help of cytoplasmic carbonic anhydrase (Tresguerres et al., 2007; Roa et al., 2014). NBC1 and AE1, in contrast, were expressed basolaterally not only in the proximal but also in the distal and connecting tubule in goldfish (Fehsenfeld and Wood, 2018).

Unfortunately, no specific inhibitor can distinguish between NBC(s), AE(s) and pendrin (Soleimani et al., 2001; Bayaa et al., 2009). The application of DIDS, as chosen here to block HCO₃⁻ transporters in general, resulted in a decrease of proton flux in all three segments but was most pronounced in the proximal segment (70% inhibition of H⁺ flu, versus 50% in distal and connecting tubules; Figs 4, 5 and 6). With basolateral AE1 protein abundance being decreased in all three segments upon feeding (and with the most pronounced decrease in the proximal tubule; Fehsenfeld and Wood, 2018), the effect observed with the application of the inhibitor DIDS basolaterally might mostly be due to targeting NBC1. Blocking NBC1-promoted reabsorption of HCO₃⁻ into the extracellular fluid would result in an increase of extracellular H⁺ and therefore diminish the H⁺ gradient and the H⁺ back-flux as observed here. This finding emphasizes the importance of renal reabsorption of HCO₃⁻ for acid–base balance during metabolic acidosis, especially in the proximal tubule.

Apical V-ATPase is generally believed to promote H⁺ secretion in branchial acid-secreting cells such as peanut lectin agglutinin negative (PNA⁻) mitochondria-rich cells of the trout gill epithelium and in proton pump-rich cells of zebrafish gills (Gilmour and Perry, 2009). In contrast, branchial base-secreting cells seem to possess a basolateral V-ATPase as proposed for trout PNA positive (PNA⁺) mitochondria-rich cells (Gilmour and Perry, 2009) and the branchial epithelium of stingrays (Piermarini and Evans, 2001). A similar distinction between cells can be made in the vertebrate kidney, with apical V-ATPase in A-type intercalated cells (PNA⁻; co-expressed with basolateral anion exchanger AE1) of the distal tubule, connecting segment and the collecting duct of the kidney cortex functioning as acid-secreting cells, whereas V-ATPase is more ubiquitously expressed in B-type intercalated cells (PNA⁺; co-expressed with apical pendrin) and is often localized in the basolateral membrane (Brown and Breton, 2000; Brown et al., 2009). In goldfish tubules, V-ATPase appeared to localize exclusively to the basolateral membrane of all three investigated segments of this study (Fehsenfeld and Wood, 2018). A basolateral V-ATPase could indeed be important for the alkalinization of the urine under control conditions (Lawrence et al., 2015).

Bafilomycin decreased H⁺ flux in the proximal (Fig. 4) and distal tubule (Fig. 5), confirming a role for V-ATPase in acid–base regulation of these tubule regions. With a basolateral V-ATPase presumably pumping H⁺ into the plasma, inhibiting this pathway would accumulate intracellular protons and result in a shift of the equilibrium for the intracellular carbonate system. Consequently, the generation of intracellular CO₂ would increase while diminishing both intracellular H⁺ and HCO₃⁻. The observed decrease in net H⁺ flux upon blocking V-ATPase could then be explained by the general decrease of intracellular protons available for basolateral, and possibly apical, transport, while also less HCO₃⁻ could be reabsorbed basolaterally via Na⁺/HCO₃⁻ co-transporter.

Carbonic anhydrase is crucial for acid–base balance in virtually all systems (Gilmour and Perry, 2009). In addition to the predominant cytosolic isoform of the enzyme (CAII-like/CAc) (Esbaugh et al., 2005; Gilmour and Perry, 2009), the kidney of freshwater teleosts expresses at least one membrane-bound carbonic anhydrase (CAIV) (Georgalis et al., 2006). CAIV is located apically and available to the tubular lumen in both mammals (Hamm et al., 2013; Schwartz et al., 2000) and fishes (Perry et al., 2003b; Georgalis et al., 2006), where it is proposed to help titrate filtered HCO₃⁻ with secreted H⁺, thereby effecting HCO₃⁻ reabsorption from the urine. Intracellularly, carbonic anhydrase converts intracellular CO₂ into HCO₃⁻ and H⁺, which is then returned to the extracellular fluid or secreted into the urine, respectively.

Blocking carbonic anhydrase in the present study decreased H⁺ flux by 60% in all three investigated segments, indicating that the conversion of CO₂ contributes considerably to the secretion and/or reabsorption of acid–base equivalents, respectively. The inhibitor acetazolamide, however, is membrane permeable and does not distinguish between different isoforms of the enzyme. Consequently, the present results confirm that carbonic anhydrase is critical for acid secretion and/or base reabsorption in all three parts of the goldfish tubule but do not reveal the site of action.

We thank Ryan Belowitz for help with fish maintenance. The authors would like to thank Dr Scott P. Kelly (York University, Toronto, ON, Canada) for donation of 2,4,6-triaminopyrimidine.