Introduction

Ammoniotelic fish, such as rainbow trout (Oncorhynchus mykiss), excrete ammonia as their primary nitrogenous waste product, mainly through the gills (Wilkie 2002; Ip and Chew 2010). The gills, followed by the kidney, have received the primary focus in terms of mechanistic analyses for ammonia handling, whereas the gut has received relatively little attention. This is surprising considering that ammonia excretion increases greatly after feeding (e.g., Brett and Zala 1975; Bucking and Wood 2008; Zimmer et al. 2010), and intestinal digestive processes create an abundant ammonia source. Notably, chyme (the partially digested food matter within the intestine) has been shown to contain approximately 1–2 mmol l−1 ammonia (or higher) following a meal in various teleost species (Bucking and Wood 2012; Bucking et al. 2013a; Rubino et al. 2014; Pelster et al. 2015; Wood et al. 2019), which represents a significant ammonia challenge to the system, considering that this concentration is an order of magnitude greater than normal plasma levels (Bucking and Wood 2008, 2012). Indeed, after feeding in rainbow trout, plasma ammonia levels increase greatly in the hepatic portal vein which connects the intestinal drainage to the liver (Karlsson et al. 2006), suggesting that ammonia is absorbed from the chyme and/or synthesized within the intestinal tissue. Using an in vitro gut sac technique, Rubino et al. (2014) demonstrated that both processes occur, and their measurements suggested that close to half of the whole animal ammonia excretion after feeding could originate from the combination of these two processes. Similar observations were made using the plainfin midshipman (Porichthys notatus), a marine teleost (Bucking et al. 2013a).

Although it is becoming clearer that ammonia is absorbed across the intestine in teleosts, the possible mechanisms remain poorly understood. Rhesus (Rh) glycoproteins (ammonia channels) are expressed in mammalian intestine (Handlogten et al. 2005; Weiner 2006; Worrell et al. 2008). At least one Rh glycoprotein occurs in the trout intestine (Rhbg, a basolateral ammonia channel), and its mRNA expression increases in response to both feeding and high ammonia exposure (Nawata et al. 2007; Bucking and Wood 2012). In addition, Bucking et al. (2013b) demonstrated immunohistochemical localization of multiple Rh isoforms along the intestine of the plainfin midshipman. Rubino et al. (2015) observed mucosa to serosa transport of an Rh-permeable analogue, 14C-methylammonia, in competition with ammonia in trout gut sacs, further suggesting Rh glycoprotein involvement. In a preliminary pharmacological survey, Rubino et al. (2015) also reported that the Na+:K+:2Cl− (NKCC) co-transporter inhibitor bumetanide (applied mucosally), as well as the Na+:K+ ATPase (NKA) inhibitor ouabain (applied serosally) and low [Na+] treatments (applied to both surfaces simultaneously) inhibited the serosal flux of ammonia (Jsamm, i.e., appearance in the serosal solution). A variety of other Na+ transport blockers (applied mucosally for Na+:H+ exchange, Na+ channels, and H+-ATPase) were without effect on Jsamm. These results suggested the involvement of active transport, overall Na+ dependence, and the specific involvement of NKCC, which would contribute to the Na+ dependence. However, Rubino et al. (2015) studied only preparations from fed fish and did not measure either the mucosal flux of ammonia (Jmamm, i.e., disappearance from the mucosal solution) or the endogenous ammonia production rate (Jtamm = Jsamm − Jmamm) of the intestinal tissue, so alternate interpretations are possible. These might include undetected responses in Jmamm and/or Jtamm., and/or mechanisms that are only present when induced by feeding.

In the present study, we again used the in vitro trout gut sac system to investigate in greater detail the Na+ dependence, as well as possible K+ dependence, of intestinal ammonia transport, with parallel experiments on fed and fasted animals. The focus on K+ originated from the bumetanide results of Rubino et al. (2015) in which ammonia flux (but only Jsamm was measured) was reduced by this inhibitor, as well as the known ability of NH4+ to substitute for K+ in other systems. For example, in mammalian models, NH4+ can directly substitute for K+ on NKCC, which is postulated as a dominant mechanism of ammonia uptake (Worrell et al. 2008). The ammonium ion (NH4+) shares a similar hydrated ionic radius and charge as K+, and, therefore, may compete with K+ for binding to a variety of transport proteins, including NKCC, NKA, and K+ channels (reviewed by Abdoun et al. 2007; Ip and Chew 2010). Specific experiments focused on the effects of altered [Na+] and [K+] concentrations, ouabain, bumetanide, and the general K+-channel blocker barium on simultaneous measurements of Jmamm, Jtamm, and Jsamm. We also examined the mRNA expression and immunohistochemical localization of NKCC in fasted and fed fish, as well as the ability of NH4+ to support intestinal NKA activity, in replacement of K+.

Our specific hypotheses were that trans-intestinal ammonia fluxes would be perturbed by both [Na+] and [K+] manipulations, as well as by treatments with ouabain, barium, and bumetanide, and that responses would differ between preparations from fed and unfed trout. In particular, we hypothesized that evidence supporting a key role for NKCC in ammonia transport would be obtained, such as inhibition of Jmamm by bumetanide, by low mucosal [Na+], and stimulation by high mucosal [Na+] and low mucosal [K+]. We also predicted that that NKCC would be localized to the apical (mucosal) surface by immunohistochemistry in this freshwater species, which would complement findings from marine teleost fish (summarized by Grosell 2011), and that its mRNA expression would increase with feeding. We also predicted that NH4+ would be able to successfully replace K+ in activation of intestinal NKA activity. Overall, the present study advances our understanding of ammonia transport in the teleost intestine.

Materials and methods

Experimental animals

Rainbow trout (Oncorhynchus mykiss, 150–250 g from Humber Springs Trout Hatchery, Ontario, Canada, were kept in 500-L tanks (density of 30 fish per tank), with a flow-through (1.8 L min−1) of dechlorinated Hamilton tapwater (moderately hard). Composition (in mequiv L−1) was [Na+] = 0.6, [Cl−] = 0.8, [Ca2+] = 1.8, [Mg2+] = 0.3, [K+] = 0.05; titration alkalinity = 2.1; pH ~ 8.0; hardness ~ 140 mg L−1 as CaCO3 equivalents; temperature = 12.5–14 °C, background ammonia concentration ≤ 10 µmol L−1). Fish were on a maintenance feeding regime, administered as approximately 3% of their body mass fish three times per week (Martin Profishent Aquaculture Nutrition, Tavistock, ON, Canada; crude protein 45%, crude fat 9%, crude fiber 3.5%) and allowed to acclimate for 4 weeks prior to experiments. All procedures followed the regulations outlined by the Canada Council for Animal Care under McMaster University Animal Utilization Protocol 09-04-10.

In vitro gut sac experiments

Gut sac experiments were performed to quantify ammonia fluxes using mucosal and serosal solutions of varying composition, with methods similar to those of Rubino et al. (2014). A common protocol was employed across all treatments. In unfed treatments, trout were fasted for 1 week prior to experimentation. Fed trout were of same group as the unfed fish; however, they were given a single satiating meal (~ 3–5% body mass) and allowed 24 h to digest. In each group, randomly selected fish were euthanized with a terminal overdose of 0.07 g L−1 pH-neutralized MS222. The body cavity was revealed through a ventral midline incision from the pectoral fins to the anus. The entire gut was then excised, freed from connective and adipose tissue, and placed in an ice-cold bath of Cortland saline (Wolf 1963; in mmol L−1: NaCl 124, KCl 5.1, CaCl2 1.6, MgSO4 0.9, NaHCO3 11.9, NaH2PO4 3, glucose 5.5, pH = 7.4). The bile duct was tied off using 2-0 silk to prevent bile spillover into the intestine. The entire gut was flushed internally with Cortland saline to remove chyme. The whole intestine from a single fish was used for a given treatment and the anterior, mid, and posterior intestines were separated from each other, and tied off at one end. A flared polyethylene tube (Intramedic Clay-Adams PE-60; Becton-Dickenson and Company, Sparks, MD) was inserted into the open end and tied in place with 2-0 silk. The desired mucosal saline, depending upon the treatment (see below), was then infused into each individual gut section via syringe. Repeated withdrawal and reinfusion ensured thorough mixing. A small volume of this mixed solution was taken, marked as the initial mucosal solution, and stored at − 20 °C for later analysis. Sections were filled with saline until taut, at a similar tension to that of a fed intestine filled with chyme, and the tube was sealed. The control mucosal solution was Cortland saline modified to contain 1 mmol L−1 ammonium chloride (NH4Cl), which was previously determined to be a physiologically relevant concentration of ammonia in chyme 24-h post-feeding (Rubino et al. 2014). The sacs were rinsed with Cortland saline, blotted dry, weighed to 0.0001 g accuracy for initial weight (Wi), and then immersed into 15-mL (mid and posterior intestines) or 50-mL (anterior intestine) plastic centrifuge tubes containing saline, which varied in composition depending upon the treatment (see below). This saline was termed the serosal solution. Control serosal saline was an unmodified Cortland saline.

A 2-h flux period began upon submersion of the sacs into their respective serosal solutions. Serosal solutions were bubbled with a 99.7% O2:0.3% CO2 gas mixture for the entire flux period, mimicking physiological PCO2 (2.25 Torr) as well as maximizing O2 availability. At the end of the 2-h flux, gut sacs were immediately removed, and a 5-ml sample of the serosal solution stored at − 20 °C for future analysis. Sacs were then blotted dry and weighed again (Wf), and the mucosal solution was collected and stored at − 20 °C for future analysis. The empty preparation was thoroughly blotted again and weighed again for empty weight (We). Surface area of the tissue was then measured, using graph paper as outlined by Grosell and Jensen (1999), and as done in our previous studies (Rubino et al. 2014, 2015) and many others (e.g., Nadella et al. 2014; Pelster et al. 2015). These surface area measurements did not incorporate contributions of luminal villi or intestinal folds. Control treatments were conducted in parallel with all the series listed below.

Series 1: Na+ series

Treatments of ouabain, low mucosal and serosal [Na+] (LSMN), and high mucosal [Na+] (HMN) were employed.

For the ouabain treatment, 10−4 mol L−1 ouabain was added to both the serosal and mucosal solutions. Ouabain was dissolved in a small volume of Cortland saline through heating, and then added to the stock solution used in the experiments. The concentration of ouabain used in this study is within the range used in other in vitro studies (e.g., rat astrocyte cultures, Hosoi et al. 2002; trout gill NKA activity assays McCormick 1993).

In the LSMN series, NaCl was removed from the Cortland saline in both the mucosal and serosal solutions and substituted with the inert osmolyte mannitol. This reduced [Na+] to about 15 mmol L−1. The osmolality was carefully balanced by adding mannitol until the exact target value was achieved, as measured using an osmometer (Wescor 5100C; Logan, UT, USA), so that it was the same in mucosal and serosal solutions, and like that of an unmodified Cortland saline. NaCl was reduced on both sides, because earlier studies demonstrated rapid backflux of Na+ from the serosal to mucosal solutions if the serosal concentration was not reduced simultaneously (e.g. Nadella et al. 2014).

HMN treatment involved doubling the NaCl concentration to 248 mmol L−1, yielding a [Na+] of about 263 mequiv L−1 in the mucosal solution. Osmolality of the serosal solution was again balanced using mannitol and confirmed using an osmometer, so that it was the same as in the mucosal solution. This was done to ensure that any differences were observed because of increased [Na+] only, and not through differences in osmotic pressure.

Series 2: bumetanide

Bumetanide (10−4 mol L−1) was added only to the mucosal solution. The concentration of bumetanide used in this study is similar to that used previously in fish epithelial transport studies (e.g. Marshall 1986). The drug was first dissolved in a carrier solvent, DMSO, and then added to Cortland saline, so that the final mucosal solution contained 0.1% DMSO. The serosal solution was Cortland saline modified to contain only 0.1% DMSO. Control treatments in this series used Cortland saline modified to contain 0.1% DMSO in both the mucosal and serosal solutions to assess for any differences in fluxes emerging solely from the use of DMSO. Differences in osmolality were negligible.

Series 3: K+ series

Low mucosal [K+] (LMK), low serosal [K+] (LSK), high mucosal barium (HMB), and high serosal barium (HSB) treatments were employed. In the LMK and LSK treatments, KCl was not added to the mucosal or serosal saline, respectively, which is represented by the treatment names. Osmolality was balanced using a small amount of mannitol. In the high barium treatments, 10−2 mol L−1 BaCl2 was added to either the mucosal or serosal saline, and osmolality was again balanced using mannitol. It is known that this high level of Ba2+ is needed to block many K + channels when applied externally (Yellen 1987; Taglialatela et al. 1993).

Analytical procedures

Ammonia

Collected mucosal samples were thawed, and immediately deproteinized using ice-cold 20% perchloric acid. Samples were then pH neutralized using 1 mol L−1 KOH. Quantification of ammonia in the serosal and deproteinized mucosal solutions were performed using a commercial assay (Raichem Cliniqa ammonia assay, glutamate dehydrogenase method) modified for use in a microplate (read at 340 nm) by scaling down the reagents used in the assay. Standard curves were generated for every group of measured samples. Samples and standards were treated similarly.

NKA activity

After trout were fasted for a week, they were fed a single satiating meal, and euthanized 24 h later via terminal overdose of neutralized MS-222. This same protocol was used for the fed treatments for NKCC mRNA expression and immunohistochemistry (see below). Samples of the gill, anterior, mid, and posterior intestine were taken, rinsed with Cortland saline as above, flash frozen in liquid N2, and stored at − 80 °C for the NKA assay. Frozen tissue was homogenized in an ice-cold EGTA-deoxycholate buffer and NKA activity was determined based on a protocol established by McCormick (1993). Activity was normalized to protein content determined via a commercial assay (Bio-Rad, Hercules, CA, USA), using a standard curve based on bovine serum albumin (Sigma, St. Louis, MO, USA).

NKCC mRNA expression

Total RNA was extracted from homogenized anterior, mid, and posterior intestine samples of unfed and fed fish using TriZol (Invitrogen, Burlington, ON). Total RNA was quantified, and purity was measured spectrophotometrically (Nanodrop ND-1000, Nanodrop Technologies, Wilmington, DE, USA). RNA integrity was then assessed by running samples on a 1% agarose gel stained with RedSafe (FroggaBio, Toronto, ON). The cDNA was synthesized using an oligo(dT17) primer, and Superscript II reverse transcriptase (Invitrogen, Burlington, ON). The mRNA expression levels of NKCC1a (GenBank: DQ864492) were then determined from this synthesized cDNA using quantitative polymerase chain reaction (qPCR) using the following primer pair:


Forward: 5′-AAC TTT GTG GAT CCG AGT GG-3′

Reverse: 5′-TAT CAG CTT GTC CCC CAG AG-3′.


Reactions were 20 µl consisting of 4 µL of 5 × diluted cDNA, 4 pmol of both forward and reverse primers, 0.8 µl of 10× diluted ROX dye, and 10 µL of SYBR Green (Bio-Rad) and were performed using an MX3000P qPCR System (Stratagene, Cedar Creek, TX). To verify the formation of a single PCR product, melt curve analysis was performed on the samples, alongside no template controls, and non-reverse transcribed controls. Relative gene expression was generated from a standard curve, through serial dilution of a randomly selected sample, and was normalized to β-actin, which was expressed equally in unfed and fed fish.

Immunohistochemistry

Anterior and posterior intestine segments were excised from unfed and fed fish. Cross-sectional rings (< 0.5 cm) were cut and immersion-fixed in 10% neutral buffered formalin solution, pH 7.3 overnight at 4 °C and then stored in 70% ethanol. Tissue was later processed for paraffin embedding, and sectioned (5 μm). Sections were collected on APS (aminopropylsilane) coated slides and immunolabelled as described in Wilson et al. (2007).

Briefly, sections were dewaxed, and antigen retrieval performed (0.05% citraconic anhydride pH 7.3 for 30 min at 100 °C) and blocked with 5% normal goat serum in 0.05% Tween-20/phosphate-buffered saline, pH 7.4 (TPBS). Sections were probed with the NKCC/NCC mouse monoclonal antibody T4 (1:100; Xu et al. 1994; Hiroi et al. 2005) alone or in combination with the pan-specific NKA-α subunit rabbit polyclonal antibody αR1 (1:500; Wilson et al. 2007) overnight at 4 °C. Following a series of rinsing steps with TPBS, sections were incubated with goat anti-mouse Dylight 594 and goat anti-rabbit Alexa Fluor 488 (1:500, both from Jackson ImmunoResearch Laboratories, West Grove, PA, USA) for 1 h at 37 °C. Following rinsing with TPBS, nuclei were counter-stained with DAPI (4′,6-diamidino-2-phenylindole) and coverslips mounted with 1:1 glycerol:PBS. Sections were viewed and images were collected with a Leica DM6000B photomicroscope (Wetzlar, Germany).

Calculations

Serosal and mucosal ammonia flux rates

To calculate the serosal ammonia flux rate (Jsamm, in µmol cm−2 h−1, i.e., appearance in the serosal saline), the following formula was used:

$${\text{Js}}_{\text{amm}} = \left[ {\left( {{\text{Ts}}_{\text{ammf}} {-}{\text{Ts}}_{\text{ammi}} } \right)} \right] \times {\text{Vs}}]/{\text{SA}} \times t,$$
(1)

where Tsammf and Tsammi are the final, and initial ammonia concentrations (µmol L−1) in the serosal solution, Vs is the volume of the serosal solution (L), SA is the surface area of the intestinal sac (cm2), and t is time (h). All fluxes were positive, indicating flux of ammonia into the serosal saline.

The mucosal ammonia flux rate (Jmamm, in µmol cm−2 h−1, i.e., disappearance from the mucosal saline) was calculated using the following formula:

$${\text{Jm}}_{\text{amm}} = \left[ {\left( {{\text{Tm}}_{\text{ammi}} \times {\text{Vm}}_{\text{i}} } \right){-}\left( {{\text{Tm}}_{\text{ammf}} \times {\text{Vm}}_{\text{f}} } \right)} \right]/{\text{SA}} \times t ,$$
(2)

where Tmammf and Tmammi are the final and initial ammonia concentrations (µmol L−1) in the mucosal solution, and Vmf and Vmi are the final and initial volumes of the mucosal solution (L). Vmf was calculated as Wf − We, and Vmi as Wi − We (In vitro gut sac experiments, above). Positive Jmamm fluxes indicate ammonia flux out of the mucosal solution, whereas negative fluxes represent ammonia flux into the mucosal solution.

Total tissue ammonia production rates

Total tissue ammonia production rates (Jtamm; µmol cm−2 h−1), representing the net rate of endogenous ammonia production by the gut tissue, was calculated using the formula:

$${\text{Jt}}_{\text{amm}} = {\text{Js}}_{\text{amm}} {-}{\text{Jm}}_{\text{amm}} .$$
(3)

The calculation assumes that there is no net change in tissue ammonia content. Rubino et al. (2014) suggested that this assumption is valid, as tissue ammonia content remained unchanged over the duration of their gut sac experiments.

Statistical analyses

Data have been expressed as mean ± SEM (N number of fish). In Table 1, comparisons made between control unfed and control fed fish (significant differences represented by asterisks), and between controls, and DMSO controls (significant differences represented by daggers) were conducted using a Student’s unpaired t test with Bonferroni correction. Controls across all treatments were pooled for these analyses. In Figs. 1, 2, 3, 4, 5, 6, 7, 8, and 9, comparisons made between controls (of either unfed or fed fish) and experimental treatments within Series 1, 2, or 3 (significant differences represented by asterisks) were conducted by a one-way ANOVA, followed by Dunnett’s post hoc test. In Table 2, for the NKA activity analysis, comparisons in activities made between K+ and NH4+ at the same concentration (significant differences represented by asterisks) were conducted via Student’s unpaired t test, while comparisons of NKA activity among different intestinal sections (significant differences represented by letters) were conducted by a repeated-measures ANOVA. For the NKCC gene expression, comparisons made between unfed and fed fish (significant differences indicated by asterisks) were conducted by a Student’s unpaired t test. In all cases, significance was accepted at the P < 0.05 level.

Table 1 Control and DMSO control Jsamm, Jmamm, and Jtamm values for the anterior, mid, and posterior intestine of unfed and fed fish
Full size table
Fig. 1
figure 1

Serosal (Jsamm) and mucosal (Jmamm) ammonia flux rates (µmol/cm2/h) of the anterior (aN = 5, 4, 5, 4, dN = 4, 4, 4, 4), mid (bN = 6, 5, 5, 5, eN = 4, 4, 4, 4), and posterior intestine (cN = 5, 5, 5, 5, fN = 5, 4, 4, 4) of unfed fish for the Na+ series. Exposures included unfed controls (black bars) and unfed ouabain, low mucosal and serosal sodium (LMSN), and high mucosal sodium (HMN) treatments (white bars). “Asterisk” significant (P < 0.05) comparing the unfed treatments to the unfed controls

Full size image
Fig. 2
figure 2

Serosal (Jsamm) and mucosal (Jmamm) ammonia flux rates (µmol/cm2/h) of the anterior (aN = 5, 4, 4, 4, dN = 3, 4, 4, 4), mid (bN = 4, 5, 4, 4, eN = 3, 4, 4, 4), and posterior intestine (cN = 5, 5, 4, 4, fN = 5, 3, 4, 4) of fed fish for the Na+ series. Exposures included fed controls (black bars) and fed ouabain, low mucosal and serosal sodium (LMSN), and high mucosal sodium (HMN) treatments (white bars). “Asterisk” Significant (P < 0.05) comparing the fed treatments to the fed controls

Full size image
Fig. 3
figure 3

Total tissue ammonia production rates (Jtamm: µmol/cm2/h) of the anterior (aN = 4, 4, 4, 4, dN = 3, 4, 4, 4), mid (bN = 4, 4, 4, 4, eN = 3, 4, 4, 4), and posterior intestine (cN = 5, 4, 4, 4, fN = 5, 3, 4, 4) of unfed and fed fish for the Na+ series. Exposures included unfed and fed controls (black bars) and unfed and fed ouabain, low mucosal and serosal sodium (LMSN), and high mucosal sodium (HMN) treatments (white bars). “Asterisk” significant (P < 0.05) comparing the unfed and fed treatments to the respective unfed and fed controls

Full size image
Fig. 4
figure 4

Serosal (Jsamm) and mucosal (Jmamm) ammonia flux rates (µmol/cm2/h) of the anterior (aN = 5, 4, dN = 5, 4), mid (bN = 5, 5, eN = 5, 4), and posterior (cN = 5, 5, fN = 4, 4) intestine of unfed fish for the DMSO and bumetanide treatments. Exposures included unfed DMSO controls (black bars) and unfed bumetanide treatment (white bars). “Asterisk” significant (P < 0.05) comparing the unfed bumetanide treatment to the unfed DMSO control treatment

Full size image
Fig. 5
figure 5

Serosal (Jsamm) and mucosal (Jmamm) ammonia flux rates (µmol/cm2/h) of the anterior (aN = 5, 5, dN = 4, 4), mid (bN = 5, 5, eN = 4, 4), and posterior (cN = 5, 5, fN = 5, 4) intestine of fed fish for the DMSO and bumetanide treatments. Exposures included fed DMSO controls (black bars) and fed bumetanide treatment (white bars). “Asterisk” significant (P < 0.05) comparing the fed bumetanide treatment to the fed DMSO control treatment

Full size image
Fig. 6
figure 6

Total tissue ammonia production rates (Jtamm: µmol/cm2/h) of the anterior (aN = 5, 4, dN = 4, 4), mid (bN = 5, 4, eN = 4, 4), and posterior intestine (cN = 4, 4, fN = 4, 4) of unfed and fed fish for the DMSO and bumetanide treatments. Exposures included unfed and fed DMSO controls (black bars) and unfed and fed bumetanide treatment (white bars). “Asterisk” significant (P < 0.05) comparing the unfed and fed bumetanide treatments to the respective unfed and fed DMSO controls

Full size image
Fig. 7
figure 7

Serosal (Jsamm) and mucosal (Jmamm) ammonia flux rates (µmol/cm2/h) of the anterior (aN = 4, 4, 4, 5, 5, dN = 4, 5, 5, 5, 5), mid (bN = 5, 5, 5, 4, 5, eN = 3, 5, 5, 5, 5), and posterior intestine (cN = 5, 5, 5, 5, 5, fN = 5, 5, 5, 5, 5) of unfed fish for the K+ series. Exposures included unfed controls (black bars) and unfed low mucosal potassium (LMK), low serosal potassium (LSK), high mucosal barium (HMB), and high serosal barium (HSB) treatments (white bars). “Asterisk” significant (P < 0.05) comparing the unfed treatments to the unfed controls

Full size image
Fig. 8
figure 8

Serosal (Jsamm) and mucosal(Jmamm) ammonia flux rates (µmol/cm2/h) of the anterior (aN = 4, 5, 5, 5, 5, dN = 3, 4, 4, 5, 5), mid (bN = 4, 4, 5, 5, 5, eN = 4, 5, 5, 5, 5), and posterior intestine (cN = 5, 5, 5, 5, 5, fN = 5, 5, 5, 5, 5) of fed fish for the K+ Series. Exposures included fed controls (black bars) and fed low mucosal potassium (LMK), low serosal potassium (LSK), high mucosal barium (HMB), and high serosal barium (HSB) treatments (white bars). “Asterisk” significant (P < 0.05) comparing the fed treatments to the fed controls

Full size image
Fig. 9
figure 9

Total tissue ammonia production rates (Jtamm: µmol/cm2/h) of the anterior (aN = 4, 4, 4, 5, 5, dN = 3, 4, 4, 5, 5), mid (bN = 3, 5, 5, 4, 5, E: N = 4, 4, 5, 5, 5), and posterior intestine (cN = 5, 5, 5, 5, 5, fN = 5, 5, 5, 5, 5) of unfed and fed fish for the K+ Series. Exposures included unfed and fed controls (black bars) and unfed and fed low mucosal potassium (LMK), low serosal potassium (LSK), high mucosal barium (HMB), and high serosal barium (HSB) treatments (white bars). “Asterisk” significant (P < 0.05) comparing the unfed and fed treatments to the respective unfed and fed controls

Full size image
Table 2 Enzyme activity values for NKA in the gill, and the anterior, mid, and posterior intestine of fed fish
Full size table

Results

The effect of feeding and DMSO on Jsamm, Jmamm, and Jtamm

Under control conditions, feeding elicited a significant increase in Jsamm only in the mid intestine (Table 1). Similarly, Jmamm remained unchanged in all sections in response to feeding; however, Jtamm was significantly decreased in the fed posterior intestine. When 0.1% DMSO was present, Jsamm was significantly reduced in all sections in unfed fish, with a similar trend for Jmamm, but the latter was significant only in the posterior intestine. The inhibitory effect of DMSO on ammonia transport was still present in fed fish (significant for Jsamm in mid intestine, and for Jmamm in mid and posterior intestine). Feeding significantly increased Jsamm in all three sections in the presence of DMSO, with a similar trend in Jmamm, but the latter was significant only in the anterior intestine. Notably, in the presence of DMSO, only Jtamm of the anterior intestine was significantly increased in fed fish.

Na+ series

Ouabain (10−4 mol L−1) significantly decreased Jsamm in all sections in unfed fish (Fig. 1), with decreases ranging from 75% in the posterior intestine (Fig. 1c), to 27% in the mid intestine (Fig. 1b). Ouabain also essentially eliminated or even reversed Jmamm in all sections, though the effect was significant only in the posterior intestine (Fig. 1f). This inhibitory effect of ouabain on ammonia transport was less pronounced in fed fish, where significant decreases in both Jsamm and Jmamm were observed only in the mid intestine (Fig. 2b, e). Jtamm was not significantly affected by ouabain in any section, under either unfed or fed conditions (Fig. 3).

The effects of the LMSN treatment were generally very similar to those of ouabain. LMSN caused significant decreases in Jsamm of all intestinal sections in unfed fish (Fig. 1a–c). These decreases, compared to controls, ranged from 33% in the anterior intestine (Fig. 1a), to 63% in the posterior intestine (Fig. 1c). LMSN similarly eliminated or even reversed Jmamm in all sections of unfed fish, though the inhibition was significant only in the anterior and posterior intestines (Fig. 1d, f). Fed fish also experienced decreases in Jsamm, with significant effects in the mid and posterior intestine (Fig. 2b, c). LMSN exposure had more profound inhibitory effects on Jmamm compared to those of ouabain in fed fish, with elimination or reversal of flux in all three sections (Fig. 2d–f). As with ouabain, Jtamm was not significantly affected by LMSN in any section, under either unfed or fed conditions (Fig. 3).

HMN was generally inhibitory to ammonia transport, but did not have as large of an impact as observed with ouabain and LMSN treatments. In unfed fish, significant decreases in both Jsamm and Jmamm were observed only in the posterior intestine (Fig. 1c, f). In fed fish, the mid intestine was the only section to experience decreases in both Jsamm and Jmamm with HMN treatment (Fig. 2b, e). HMN had no significant effect on Jtamm in any section, under either fed or unfed conditions, similar to the preceding two treatments (Fig. 3).

Bumetanide series

DMSO (0.1%) was required as a vehicle for bumetanide (10−4 mol L−1). As DMSO in itself caused significant effects on ammonia transport (see above, Table 1), all bumetanide responses were evaluated relative to the DMSO control treatment. Bumetanide had strong inhibitory effects on Jsamm in all three sections in unfed fish (Fig. 4a–c), without significant effects on Jmamm except in the anterior intestine where Jmamm was actually elevated (Fig. 4d). These actions of bumetanide correlated with marked reductions in Jtamm in all sections which were significant only in the anterior and mid intestine (Fig. 6a, b). In general, qualitatively similar effects of bumetanide (reductions in Jsamm and Jtamm, unchanged Jmamm) were seen in fed fish (Figs. 5, 6), though feeding abolished the significant reductions in Jsamm (Fig. 4a) and Jtamm (Fig. 6a) seen in the anterior intestine of unfed fish and reversed the stimulation of Jmamm (Fig. 4d) to a slight but significant inhibition (Fig. 5d).

K+ series

Compared to the Na+ series (cf. Figs. 1, 2), the treatments employed in this series had less profound impacts on Jsamm throughout the intestine, whereas virtually all of them markedly inhibited Jmamm, except in the anterior segment of unfed fish (Figs. 7, 8).

In unfed fish, LMK significantly inhibited Jsamm only in the posterior intestine, with a 61% decrease from the control value (Fig. 7c). However, LMK essentially eliminated Jmamm in the mid and posterior intestine (Fig. 7e, f) but was without effect in the anterior intestine (Fig. 7d). In fed fish, LMK had no significant effects on Jsamm in any section (Fig. 8a–c), whereas Jmamm was again inhibited, with virtual total blockade in the anterior and posterior segments (Fig. 8d, f). LMK had no significant effects on Jtamm in any section (Fig. 9).

LSK showed generally similar responses LMK. In unfed fish, there was a comparable significant decrease in Jsamm only in the posterior intestine (Fig. 7c), and a virtual elimination of Jmamm only in the mid and posterior intestine (Fig. 7e, f). In fed fish, Jsamm was again unaffected by LSK in all sections (Fig. 8a–c), whereas Jmamm was reduced or reversed in all sections (Fig. 8d–f). Again, Jtamm was generally unaltered throughout, except for an increase in the fed posterior intestine (Fig. 9).

Responses to HMB and HSB treatments were virtually identical. In unfed as well as fed trout, there was strong inhibition of Jsamm in the posterior intestine by both (Figs. 7c,8c), and in the mid intestine of unfed animals by HMB only (Fig. 7b). Jmamm was eliminated by both HMB and HSB in the mid and posterior sections of unfed animals (Fig. 7e, f), and strongly inhibited in all three sections of fed animals (Fig. 8d–f). HMB and HSB had no effects on Jtamm in any section (Fig. 9).

The effects of NH4 + on the activation of Na+:K+-ATPase

NKA activity across the whole intestine was more than an order of magnitude higher than that observed in the gills (Table 2). The ability of NH4+ to support NKA activity was evaluated by replacing K+ with NH4+. For gill NKA, activity with both 1 mmol L−1 and 3 mmol L−1 NH4+ was greatly reduced relative to assays run with 1 mmol L−1 and 3 mmol L−1 K+. However, for intestinal NKA from all three sections, activity with NH4+ was comparable to that with K+ at both concentrations (Table 2).

The effects of feeding on the mRNA expression and immunohistochemical analysis of NKCC

Relative normalized gene expression for NKCC was altered by feeding status (Fig. 9). In both the anterior and posterior intestine, there were significant increases in gene expression (eightfold and fourfold, respectively) in response to feeding. Expression in the mid intestine remained unchanged.

Further exploration of the anterior and posterior intestines was performed by NKCC/NCC immunohistochemical staining (Fig. 10). The protein was apically located on the mucosal surface and feeding caused an increase in apical immunoreactivity in both sections. A stronger presence was observed in the anterior intestine compared to that of the posterior intestine. Interestingly, there was almost no appreciable apical staining in the posterior intestine of unfed fish, in contrast to the anterior intestine of unfed fish. Furthermore, no detectable immunoreaction was observed along the basal layer.

Fig. 10
figure 10

Relative normalized gene expression for the NKCC rainbow trout isoform for unfed and fed fish. “Asterisk” significant (P < 0.05) comparing fed sections to the respective unfed sections (N = 5 each section)

Full size image

Discussion

Overview

To our knowledge, this study, together with the preliminary findings of Rubino et al. (2015), presents the first observations demonstrating a potential linkage between intestinal ion transport and ammonia handling in a fed teleost fish. As reported by Rubino et al. (2014), ammonia transport tended to increase after feeding, though in the present study, this was prominent only in the preparations where DMSO was used as a vehicle. Furthermore, feeding versus fasting resulted in rather minor differences in the responses of ammonia transport to various experimental treatments, contrary to our original hypotheses. The most prominent of these was an attenuation of the inhibitory effects of ouabain. However, manipulations of both [Na+] and [K+] interfered with ammonia transport, in accord with our original hypotheses, as well as the observations of Rubino et al. (2015), confirming the Na+ dependence of ammonia transport. Furthermore, as we had predicted, increased expression of NKCC with feeding was confirmed at the mRNA level, with supporting immunohistochemical evidence. However, our specific hypothesis that NH4+ substitution for K+ on apical NKCC was a key step in the mucosal uptake of ammonia was not well supported. Most importantly, mucosally applied bumetanide had no detectable inhibitory effect on Jmamm except in the anterior intestine of fed trout and the responses to elevated mucosal [Na+] (HMN) and low mucosal [K+] (LMK) were opposite to those expected if mucosal NKCC was involved. Together, the addition of Jmamm and Jtamm measurements to the present study, that were unavailable in Rubino et al. (2015), allowed for these observations which resulted in a change to our model. Specifically, these results highlighted less dependence on NKCC for intestinal ammonia uptake (Fig. 12). Nevertheless, an interaction with K+ was observed. The inhibitory actions of barium indicated that this interaction probably reflected NH4+ passage through K+ channels. The inhibitory effects of ouabain confirmed an important contribution of NKA to the overall transport process, and at least in part this could explain why treatments that reduced serosal [K+] (LSK) and mucosal [Na+] (LMSN) were also inhibitory. Finally, our hypothesis that NH4+ would be able to successfully replace K+ on intestinal NKA was confirmed. This suggests that NKA also plays a role in scavenging NH4+ from the intestinal circulation for metabolism in the enterocytes, thereby preventing excessive increases in plasma ammonia levels after feeding. High mucosal [Na+] (HMN) and low serosal [K+] (LSK) would be expected to increase this basolateral “back-transport” of NH4+ by NKA, thereby contributing at least in part to the inhibitory effects on ammonia transport observed with these treatments.

Is NKCC involved in ammonia transport in the intestine?

NKCC mRNA was expressed in all three intestinal sections (Fig. 10) and may now be added to the list of gut transporters that are upregulated after feeding in freshwater fish (Kamunde and Wood 2003; Bucking and Wood 2012; Turner and Bucking 2017). Nadella et al. (2014) presented pharmacological evidence (inhibition of Na+ transport by furosemide) for NKCC function in gut sacs of recently fed freshwater trout, but only in the anterior intestine; there was no effect in the mid and posterior intestine. Conversely, pharmacological evidence (inhibition of Na+ transport by hydrochlorothiazide) for Na+:Cl− co-transporter (NCC) function was seen only in the mid and posterior sacs (Nadella et al. 2014). We are aware of no other previous information on this topic in freshwater teleosts (indeed NKCC is often omitted in freshwater models—e.g. Movileanu et al. 1998), but the results fit with many studies on the gut of seawater teleosts where both NKCC and NCC are significant contributors to apical Na+ uptake processes (reviewed by Grosell 2011). To observe if this mRNA expression resulted in detectable levels of NKCC protein expression, immunohistochemistry was performed and indicated the presence and apical location of a signal, particularly in the anterior intestine of freshwater trout. Furthermore, this signal showed increased expression after feeding, similar to the observations of increased mRNA expression following feeding (Fig. 11). One limitation in the performance of the immunohistochemistry is that the monoclonal antibody T4 which was utilized recognizes both NKCC and NCC (Xu et al. 1994). However, if, indeed, this protein is NKCC, then its apical location suggests that it would be of the absorptive type rather than the secretory type, as discussed by Hiroi et al. (2005), and in agreement with the literature on seawater teleosts (Loretz 1995; Grosell 2011). In the future, quantitatively assessing NKCC protein expression, such as through a western blot, could aid in determining the extent of the increase following feeding.

Fig. 11
figure 11

Double immunofluorescence localization of NKCC/NCC (red) and NKA (α subunit, green) with DAPI nuclear staining of the anterior (a, b) and posterior (c, d) intestine of fed (a, c) and unfed (b, d) rainbow trout. Arrows indicate brush border staining. Scale bar = 100 µm

Full size image

Bumetanide has been demonstrated to reduce ammonia uptake through inhibition of NKCC in multiple organ systems and cell types, including rabbit kidney (Kinne et al. 1986), mouse hybridoma cells (Westlund and Haggstrom 1998), and mouse neurons (Thrane et al. 2013). Our earlier suggestion that NKCC played a key role in intestinal ammonia transport (Rubino et al. 2015) was based on the inhibitory actions of bumetanide and low Na+ treatments on Jsamm. The more detailed measurements of the present study confirmed an inhibition of Jsamm by both bumetanide (Figs. 4, 5) and LMSN (Figs. 1, 2) in most sections, and also showed that LMSN reduced Jmamm to a comparable extent (Figs. 1, 2), but revealed that the inhibitory action of bumetanide on Jsamm can largely be explained by an inhibition of endogenous ammonia production (Jtamm) by the intestinal tissue itself (Fig. 6). There was no inhibition of Jmamm except for a slight reduction in the anterior intestine of fed fish (Figs. 4, 5), which greatly weakens the case for involvement of NKCC in the apical uptake step. Further evidence against NKCC involvement was a lack of stimulation of either Jmamm or Jsamm by elevated mucosal [Na+] (HMN) (Figs. 1, 2), and a lack of stimulation by low mucosal [K+] (LMK) (Figs. 7, 8). Indeed, in some sections, inhibitions were observed, whereas if NKCC was involved, increased ammonia transport would be expected to result from both treatments. The inhibitory actions of LMSN could be interpreted as support for NKCC, but equally well could have resulted from a reduction of intracellular [Na+], tending to slow down NKA, which provides the overall driving force for ammonia transport, as discussed subsequently.

One caveat with respect to the lack of an inhibitory effect of bumetanide on Jmamm is that this result may have been confounded by the required presence of the 0.1% DMSO vehicle. DMSO by itself generally inhibited both Jmamm and Jsamm, without effects on Jtamm (Table 1). As Jmamm values were generally close to zero in the presence of DMSO, a true inhibitory effect of bumetanide + 0.1% DMSO (relative to 0.1% DMSO alone) on Jmamm would have manifested as a switch to negative values, but this was generally not seen (Figs. 4, 5). However, since bumetanide strongly suppressed Jtamm, this might have limited the gradient for unidirectional ammonia backflux from tissue to lumen, preventing the response. The reason why 0.1% DMSO inhibited Jmamm and Jsamm is unclear, as this vehicle usually increases membrane permeability (Notman et al. 2006). Perhaps, there are normally significant backfluxes of ammonia at both surfaces, such that the observed Jmamm and Jsamm (which are net fluxes) are reduced. Similarly, the reason why bumetanide inhibited endogenous ammonia production (Jtamm) is unclear. Intestinal N-metabolism is complex (reviewed by Bergen and Wu 2009; Taylor et al. 2011). One possibility is reduced glutamine transport into mitochondria. At the concentration used here, bumetanide has been reported to act as a competitive inhibitor of glutamate transport into synaptic vesicles (Roseth et al. 1995). Glutamate dehydrogenase and glutaminase are thought to be the major routes of endogenous ammonia production in trout intestine (Rubino et al. 2014), so reduction of mitochondrial glutamate supply by bumetanide may have led to the reduction of Jtamm and, therefore, Jsamm.

Overall, the present data provide little support for ammonia uptake across the mucosal surface of the trout intestine by NKCC, except perhaps in the anterior intestine after feeding where NKCC is markedly upregulated. In comparable mammalian models such as a colonic crypt cell line, the role of apical NKCC appears to be solely secretory so as to promote ammonia backflux into the lumen, thereby preventing ammonia levels in the portal bloodstream from rising too high (Worrell et al. 2008). Similarly, in the rumen of sheep, there appears to be no role for NKCC in ammonia absorption, and the interaction of K+ with ammonia uptake can be explained by NH4+ entry through apical K+ channels (reviewed by Abdoun et al. 2007).

Are K+ channels involved in ammonia transport?

The findings of Abdoun et al. (2007) on ammonia transport across the rumen prompted our interest in the potential involvement of K+ channels, and led to the experiments with barium, a broad-spectrum inhibitor of K+ channels (Yellen 1987; Taglialatela et al. 1993; Choe et al. 2000). HMB has been reported to strongly inhibit both ammonia and K+ absorption in the perfused rat intestine in vivo (Hall et al. 1992). Mucosal quinidine (another broad-spectrum K+ channel blocker) reduced ammonia absorption across the sheep rumen in vitro (Bödeker and Kemkowski 1996). In most trout preparations, both HMB and HSB strongly inhibited Jmamm and Jsamm (Figs. 7, 8) without effects on Jtamm (Fig. 9). The anterior intestine was again an exception (Figs. 7a, d, 8a).

K+ channels are often highly permeable to NH4+ (Yellen 1987; Choe et al. 2000). Ion flow through K+ channels is determined by the prevailing electrochemical gradients (MacKinnon 2003). Considering the high intracellular [K+] and low mucosal [K+], the electrochemical gradient would likely be in the secretory direction for K+ through apical channels in the intestine (Loretz 1995; Movileanu et al. 1998). Indeed, in a marine teleost, an apical barium-sensitive K+ channel-mediating K+ secretion into the lumen has been identified (Musch et al. 1982; Loretz 1995). However, measured tissue ammonia concentrations are approximately twofold higher than the mucosal fluid concentrations in trout gut sac preparations (Rubino et al. 2014), and the potential difference across the apical membrane of enterocytes should be highly negative inside (see Bödeker and Kemkowski 1996; Abdoun et al. 2007). Therefore, there should be a strong electrochemical gradient for NH4+ uptake through K+ channels in the apical membrane, which would explain why HMB blocks Jmamm and Jsamm. The presence of basolateral K+ channels also seems likely, though there is also evidence for basolateral K+ exit by K+:Cl− co-transport in the intestine of marine teleosts (Smith et al. 1980; Loretz 1995; Movileanu et al. 1998; Grosell 2011). HSB would block basolateral K+ channels, but it seems unlikely that the prevailing electrochemical gradient would favor NH4+ efflux though such channels. More probably, blockade of these channels by HSB would increase intracellular [K+], thereby reducing NKA activity, and, therefore, the net driving force on NH4+ transport, explaining the inhibition of Jmamm and Jsamm. Overall, the present results suggest that ammonia uptake across the mucosal surface of the trout intestine occurs in part via NH4+ entry through apical K+ channels.

What is the role of NKA in ammonia transport?

Ouabain is a potent and irreversible inhibitor of NKA (Albers et al. 1968), so its actions in reducing both Jmamm (Fig. 1) and Jsamm (Fig. 2) without effects on Jtamm (Fig. 3) in most sections were in accord with the findings of Rubino et al. (2015), where only Jsamm was monitored. The inhibitory actions of LMSN (Figs. 1, 2) can also be explained by a slowing of NKA. These results strongly suggest that NKA plays an important role in energizing ammonia absorption. This finding also agrees with mammalian studies where anaerobiosis, 2,4-dinitrophenol, and sodium cyanide inhibited ammonia absorption in the ileum (Mossberg and Ross 1967), while ouabain inhibited ammonia secretion in colonic crypt cells (Worrell et al. 2008). It is interesting that the inhibitory effects of ouabain were reduced by feeding, which is known to cause an upregulation of intestinal NKA activity in fish (Kamunde and Wood 2003; Turner and Bucking 2017). The primary role of NKA is likely to establish the overall electrochemical gradient for transport. To the extent that apical NH4+ uptake by NKCC is involved, the low intracellular [Na+] and negative intracellular potential will energize this process. Similarly, the negative potential will help to drive NH4+ entry through apical K+ channels. While we suggest that transport across the membrane is facilitated by NKA, paracellular transport of ammonia due to movement of water across the epithelium due to osmotic gradients (Grosell 2011) and transcellular transport of water through aquaporins (e.g., AQP1, Wood and Grosell 2012) cannot be discounted. Furthermore, Rubino et al. (2015) identified reduced fluid transport rates across the intestine in gut sacs when using ouabain. Therefore, this reduction in fluid transport may contribute to the reduction in ammonia flux observed when this inhibitor is applied.

At the same time, basolateral NKA may play a very different role at the blood side, that of NH4+ uptake (‘scavenging’) from the portal blood so as to prevent circulating levels of ammonia from rising too high after feeding, and to ensure that there is a ready supply of substrate for glutamine synthesis in the enterocytes (Bergen and Wu 2009; Taylor et al. 2011; Wilson et al. 2013). This would occur by NH4+ competition at the extracellular K+ binding site on the enzyme. In support of this idea, we found that NH4+ was just as effective as K+ in supporting intestinal NKA activity, a phenomenon that was not seen with gill NKA (Table 2). The inability of NH4+ to support gill NKA activity confirms previous observations on trout (Salama et al. 1999; Wood and Nawata 2011), though in several other teleost species, gill NKA activity can be strongly increased by NH4+ (Randall et al. 1999; Wood et al. 2013, 2014). We are aware of no previous studies on NH4+ effects on intestinal NKA in fish. By this model, high NH4+ levels in the intestinal blood after feeding might help to improve the electrochemical gradient for NH4+ absorption, which may direct more ammonia into amino acid synthesis by the intestinal tissue through scavenging from the portal blood. This starts with the trapping of ammonia into glutamine by glutamine synthetase, which increases in activity in response to feeding in the intestine (Bucking and Wood 2012; Bucking et al. 2013b). Glutamine can then shuttle amino groups to other amino acids and can provide energy to the intestinal cells for oxidation or gluconeogenesis.

Concluding remarks and future directions

Rubino et al. (2015) presented a preliminary, rather speculative model describing ammonia transport in the intestinal tract of rainbow trout (see their Fig. 8). The present study clarifies and expands that model (Fig. 12). While the proposal of an apical NKCC is confirmed, the present more detailed analysis suggests that this transporter is not as important in ammonia absorption as earlier hypothesized, except perhaps in the anterior intestine. Rather, the interaction with K+ seems to occur because NH4+is taken up through barium-sensitive K+ channels on the apical surface. Mucosal NH4+ uptake via both NKCC and K+ channels is energized by basolateral NKA, which plays an additional role in scavenging NH4+ on the serosal surface. This occurs by transport on the extracellular K+ binding site of NKA, thereby ensuring intracellular substrate for amino acid synthesis in the intestinal tissue. NH4+is equally effective as K+ in supporting the activity of the enzyme; this may become important when ammonia levels in the intestinal circulation are elevated after feeding.

Fig. 12
figure 12

Updated schematic diagram as a conceptual model for transcellular and paracellular ammonia transport for all sections of the intestine, revised from earlier version of Rubino et al. (2015). Routes for ammonia uptake are energized by basolateral NKA providing the electrochemical gradients for secondary active transport. Ammonia is believed to occupy K+ transport sites on various transporters at both the apical and basolateral surface, though with less involvement of NKCC than previously suggested. It is possible for ammonia to also enter intestinal cells through either an apical or basolateral Rh glycoprotein. Alternative routes of uptake through solvent drag via fluid transport occurring through osmosis, or water uptake through aquaporins (Wood and Grosell 2012) may also occur. Once inside the cells, ammonia detoxification enzymes, such as GSase, may reduce intracellular ammonia concentrations (Bucking and Wood 2012; Rubino et al. 2014)

Full size image

However, many uncertainties remain, some of which can best be solved by electrophysiological studies using Ussing chambers and patch clamping, approaches which have proven very useful in understanding how Na+, Cl−, and HCO3− are transported in the gastrointestinal tracts of marine teleosts (Loretz 1995; Grosell 2011). However, some important questions can be profitably attacked using the simple gut sac approach. Principal among these is the potential role of non-ionic diffusion of NH3 along PNH3 gradients, perhaps facilitated by Rh proteins. The role of non-ionic diffusion issue has been extensively investigated in the gills of fish, yet remains controversial (Wilkie 2002; Wright and Wood 2009; Ip and Chew 2010).This mechanism has not yet been examined in the fish intestine, despite substantial evidence that it plays a role in ammonia transport in the mammalian gastrointestinal tract (Swales et al. 1970; Castell and Moore 1971; Brown et al. 1975; Abdoun et al. 2007). Non-ionic diffusion can be examined by establishing fixed PNH3 gradients across gut sac preparations, far more easily than across the gills. Furthermore, Rubino et al. (2015) argued that ammonia movement by solvent drag may also be important in the fish intestine; again, this can be addressed quite easily by establishing fixed osmotic gradients across gut sac preparations (e.g., Wood and Grosell 2012). Our understanding of ammonia transport and metabolism in the fish intestine remains in its infancy. This is a rich area for future investigation.