Aquaporin (AQP)8-facilitated transport of NH3 has been suggested recently by increased NH3 permeability in Xenopus oocytes and yeast expressing human or rat AQP8. We tested the proposed roles of AQP8-facilitated NH3 transport in mammalian physiology by comparative phenotype studies in wild-type vs. AQP8-null mice. AQP8-facilitated NH3 transport was confirmed in mammalian cell cultures expressing rat or mouse AQP8, in which the fluorescence of a pH-sensing yellow fluorescent protein was measured in response to ammonia (NH3/NH4+) gradients. Relative AQP8 single-channel NH3-to-water permeability was ∼0.03. AQP8-facilitated NH3 and water permeability in a native tissue was confirmed in membrane vesicles isolated from testes of wild-type vs. AQP8-null mice, in which BCECF was used as an intravesicular pH indicator. A series of in vivo studies were done in mice, including 1) serum ammonia measurements before and after ammonia infusion, 2) renal ammonia clearance, 3) colonic ammonia absorption, and 4) liver ammonia accumulation and renal ammonia excretion after acute and chronic ammonia loading. Except for a small reduction in hepatic ammonia accumulation and increase in ammonia excretion in AQP8-null mice loaded with large amounts of ammonia, there were no significant differences in wild-type vs. AQP8-null mice. Our results support the conclusion that AQP8 can facilitate NH3 transport but provide evidence against physiologically significant AQP8-facilitated NH3 transport in mice.
- water transport
- transgenic mouse
aquaporin (AQP)8 has been proposed to be a potentially important water transporter in multiple tissues, particularly in the gastrointestinal tract. Northern blot and RT-PCR analyses showed strong AQP8 transcript expression in salivary gland, liver, pancreas, small intestine, colon, and testis, with lower expression in kidney and heart (7, 25, 38). However, available anti-AQP8 antibodies have been poor because AQP8 has few polar residues at its NH2 and COOH termini. Using AQP8-knockout mice as negative control, immunolocalization studies in mice with a rabbit polyclonal antibody against the mouse AQP8 NH2 terminus sequence confirmed AQP8 protein expression in liver, colon, and testis (38). However, phenotype studies of AQP8-null mice were negative, with only minor differences between wild-type and AQP8-null mice in testicular size, even after an attempt to expose subtle phenotype differences by physiological stresses to gastrointestinal organs and kidney as well as codeletion of other AQPs (38).
Several laboratories have recently reported evidence for facilitated transport of NH3 by AQP8 (14, 16, 22), suggesting a new possible role for AQP8 in mammalian physiology. Jahn et al. (16) found that human AQP8 complemented ammonia transport deficiency in a yeast mutant (Δmep1–3) deficient in ammonia transport. Using pH-sensitive microelectrodes, Holm et al. (14) reported a 30–40% increased rate of cytoplasmic alkalinization after NH3 exposure in AQP8-expressing compared with control Xenopus oocytes. Liu et al. (22) suggested that human AQP8 was also permeable to methylammonium based on NH3 inhibition of methylammonium transport in oocytes expressing human AQP8. Facilitated transport of NH3 by a membrane protein was somewhat of a surprise because of the generally high intrinsic NH3 permeability of most membranes and consequent unstirred layer effects. Nevertheless, the potential consequences of AQP8-facilitated ammonia permeability in mammalian physiology include AQP8-dependent hepatic, renal, and colonic NH3 accumulation and/or excretion.
In this study we confirmed NH3 transport by mouse AQP8 in mammalian cells and in a native mouse tissue and then systematically tested the physiological significance of AQP8-facilitated NH3 transport by comparative phenotype studies of ammonia (NH3/NH4+)-related hepatic, renal, and colonic functions in wild-type vs. AQP8-knockout mice. AQP8-knockout mice do not express detectable AQP8 transcript or protein yet are phenotypically indistinguishable from wild-type mice except for testicular enlargement in the male knockout mice (38). Although hepatic ammonia accumulation and renal excretion of nonmetabolized ammonia were altered in AQP8-knockout mice after mice were loaded with large amounts of ammonia, the differences were small and other studies were negative, providing evidence against physiologically important AQP8-facilitated NH3 transport in mice.
Cell culture and transfection.
The cDNAs encoding full-length rat AQP8 and mouse AQP8 were fused downstream from c-myc in pcDNA6 (Invitrogen). Plasmid inserts were fully sequenced. Fischer rat thyroid (FRT) epithelial cells expressing the yellow fluorescent protein YFP-H148Q (9) were grown at 37°C in a 5% CO2-95% air atmosphere with F-12 medium containing 500 μg/ml G418. Cells were transfected with the rat or mouse AQP8 plasmid with Lipofectamine 2000 (Invitrogen) with standard protocols. Cells stably expressing rat or mouse AQP8 were selected with 10 μg/ml blasticidin S (Invitrogen) and confirmed by immunostaining with anti-c-myc antibody. For measurements of water and NH3 permeability, cells were suspended by agitation in PBS (Ca2+, Mg2+ free) containing 0.04% EDTA and washed twice (100 g, 10 min) in PBS.
Knockout mice deficient in AQP8 in a CD1 genetic background were generated by targeted gene disruption as described previously (38). All animal procedures were approved by the University of California, San Francisco, Committee on Animal Research.
Isolation of testis plasma membrane vesicles.
Testes were homogenized as described previously (38), and a postnuclear supernatant was mixed with an equal volume of 2.3 M sucrose to obtain a 1.4 M sucrose fraction. A sucrose density gradient was prepared consisting of 0.5 ml of 2.0 M sucrose, 1 ml of 1.6 M sucrose, 2 ml of 1.4 M sucrose fraction containing homogenate, 2 ml of 1.2 M sucrose, and 0.5 ml of 0.8 M sucrose. The gradient was centrifuged for 2.5 h at 37,500 rpm in an SW 50.1 rotor, and 1-ml fractions were collected. The top fraction containing mainly plasma membranes was collected and centrifuged at 37,500 rpm for 1 h. Membrane pellets were resuspended and used for water and NH3 permeability measurements.
Water permeability measurements.
Stopped-flow measurements of cell and vesicle osmotic water permeability were carried out on a Hi-Tech Sf-51 instrument. The kinetics of decreasing cell/vesicle volume was measured from the time course of 90° scattered light intensity at 530-nm wavelength. Cells or vesicles were suspended in PBS and subjected to a 250 mM inwardly directed gradient of sucrose. In some experiments, cells were incubated with 0.3 mM HgCl2 for 5 min before measurements. Osmotic water permeability coefficients (Pf) were computed from the time course of scattered light intensity by standard procedures (30).
Ammonia permeability measurements.
FRT cells (5 × 105 cells/ml) in suspension buffer (mM: 110 NaCl, 5 KCl, 25 HEPES, pH 7.4) were mixed in the stopped-flow apparatus with an equal volume of ammonia-containing buffer (mM: 85 NaCl, 5 KCl, 25 NH4Cl, 25 HEPES, pH 7.4). YFP fluorescence was excited with a 500-nm interference filter and detected with a 535-nm filter. The apparent ammonia permeability coefficient (Pammonia, cm/s) was computed as described previously (37) from an exponential time constant (τ) fitted to the fluorescence time course: Pammonia = 1/τ [(S/V)/10(pHf−pKa)]−1, where S/V is vesicle surface-to-volume ratio pKa is 9.25, and pHf is final solution pH after mixing determined from the increase in fluorescence. For measurements on fractionated testis plasma membrane vesicles, membrane pellets were resuspended at ∼10 mg protein/ml in suspension buffer containing 20 μM BCECF-acid (Molecular Probes) and passed 20 times though a 30-gauge needle. External dye was removed by overnight dialysis in suspension buffer at 4°C. BCECF-containing membrane vesicles were diluted to ∼2 mg protein/ml in suspension buffer and mixed in the stopped-flow apparatus with an equal volume of ammonia buffer titrated to give pH 8 after mixing with an equal volume of the vesicle suspension. BCECF fluorescence was excited with a 485-nm interference filter and detected with a >515-nm cut-on filter.
Ammonia concentration in blood, urine, and tissue homogenates.
Blood was obtained by tail bleeding, and plasma was separated from blood cells by centrifugation. Urine was obtained by gentle abdominal massage to induce voiding. Brain, liver, kidney, and testis homogenates were obtained by homogenizing tissues in a fivefold excess of distilled water with a Tissue Tearor (Biospec Products) and obtaining a supernatant after centrifugation at 13,200 g. Ammonia was assayed by UV absorbance with a commercial ammonia assay kit (Boehringer Mannheim/R-Biopharm) according to the manufacturer's instructions. Ammonia concentration in tissues was shown as mmol/kg tissue wt (calculations were done by assuming tissue density as 1).
Blood CO2 and pH measurements.
Mice were anesthetized with 80 mg/kg ketamine and 10 mg/kg xylazine. Blood samples were obtained by ventricular puncture. Blood Pco2 and pH were measured immediately on a pH/blood gas analyzer (Chiron Diagnostics).
Acute ammonia loading experiments.
Mice were anesthetized with 80 mg/kg ketamine and 10 mg/kg xylazine. Body temperature was kept at 36–37°C with a heating lamp and a heating pad. PE-10 tubing was inserted into a jugular vein for infusion of NH4Cl (150 mM; 0.1 ml/10 g body wt). Blood and/or tissue samples were obtained at specified time points.
Chronic ammonia loading experiments.
Experiments were performed as described by Chambrey et al. (3). Mice were adapted to metabolic cages for 3 days and provided with deionized water ad libitum and standard laboratory mouse chow. Two 24-h urine samples were collected (under light mineral oil) to determine daily urinary solute excretion. The mice were then given 0.28 M NH4Cl in place of water. Urine output, ammonia concentration, pH, and titratable acid were measured daily. Mice were killed after 5 days, at which time blood was taken for assay of ammonia concentration, pH and Pco2.
Colonic ammonia transport.
After solid food was withheld for 24 h (giving 5% glucose in water instead), mice were anesthetized as described above. The colon was exposed by a midline abdominal incision. Proximal and distal colon (12- to 20-mm length) was tied with 5-0 nylon suture. For measurement of ammonia permeability, 0.2 ml of warmed modified PBS containing 0.1 M NH4Cl and 0.5% blue dextran (as a volume marker) was injected into the lumen of proximal and distal colon with a 27-gauge needle. Twenty-microliter fluid samples were withdrawn at specified times for assay of ammonia concentration.
Cells were fixed with 4% paraformaldehyde in PBS for 10 min, treated with 0.2% Triton X-100 in PBS for 10 min, and incubated with monoclonal anti-c-myc antibody (1:500) for 1 h and Cy3-conjugated goat anti-mouse IgG (1:200; Sigma) for 30 min.
Stably transfected FRT cell lines expressing rat or mouse AQP8 were established to measure AQP8-facilitated NH3 permeability. Immunofluorescence indicated that the transfected cells expressed rat and mouse AQP8 in a plasma membrane pattern (Fig. 1A). No staining was seen in nontransfected (control) FRT cells. AQP8-facilitated osmotic water permeability was measured from the time course of scattered light intensity in response to a 250 mM inwardly directed osmotic gradient of sucrose. The light scattering signal amplitude is inversely related to cell volume. Representative original light scattering data are shown in Fig. 1B for FRT cells without and with AQP8 expression, under control conditions (Fig. 1B, left) and after brief incubation with HgCl2, an inhibitor of aquaporin-facilitated water transport (Fig. 1B, center). Experiments were done initially at a low temperature (10°C), where the strongly temperature-dependent lipid-mediated water permeability is minimized. The data are presented on two contiguous timescales to show the initial rate of decreasing cell volume as well as the maximal signal change after full osmotic equilibration. Osmotic equilibration (half time) was slow (∼2 s) in control FRT cells (Fig. 1B, top). Water permeability was rapid (∼0.2 s) in cells expressing rat AQP8 (Fig. 1B, middle) and mouse AQP8 (Fig. 1B, bottom). HgCl2 reduced water permeability strongly in AQP8-expressing cells but had no effect on control cells (Fig. 1B, middle). As expected, osmotic equilibration was faster at 25°C than at 10°C (Fig. 1B, right), particularly in control FRT cells because of the greater Arrhenius activation energy for lipid vs. AQP-mediated water transport. Water permeability coefficients are summarized in Fig. 1C. These data indicate strong functional plasma membrane expression of mouse and rat AQP8 in transfected FRT cell lines.
NH3 permeability was measured in control and AQP8-expressing FRT cell lines, each of which expresses cytoplasmic YFP-H148Q, from the kinetics of luminal pH in response to rapid mixing with NH4Cl-containing solutions. The pH-dependent increase in fluorescence caused by rapid NH4+/NH3/pH equilibration was measured by stopped-flow fluorimetry. Rapid NH3 influx increases cytoplasmic pH, producing an increase in YFP fluorescence. NH3 permeability was significantly greater in AQP8-expressing than control cells (Fig. 2A, left). NH3 permeability coefficients (Pammonia) are summarized in Fig. 2B. NH3 permeability in all cells was strongly temperature dependent (Fig. 2A, right). These experiments demonstrate AQP8-facilitated NH3 transport in transfected mammalian cells.
AQP8 is expressed at the plasma membrane in spermatogenic cells in testis (2, 17, 38). Figure 3A shows that osmotic water equilibration was significantly faster in plasma membrane vesicles of testis from wild-type vs. AQP8-null mice. Because membranes from multiple cell types are present in the preparation, osmotic equilibration is increased in only a fraction of membranes, precluding computation of absolute permeability coefficients. NH3 transport across BCECF-loaded plasma membrane vesicles from testis was measured from the kinetics of luminal pH in response to rapid mixing with NH4Cl solutions. As in the cell studies, rapid NH4+/NH3/pH equilibration produces a pH-dependent increase in fluorescence. Figure 3B shows a component of relatively rapid NH3 permeability in vesicles from wild-type mice compared with AQP8-null mice, confirming functional AQP8-facilitated NH3 permeability in a native tissue.
To study the consequences of AQP8-facilitated NH3 permeability in mice in vivo, we first measured baseline blood and tissue ammonia concentrations. Ammonia (NH3 + NH4+) concentration was greater in skeletal muscle (∼4 mmol/kg) than in blood (∼0.2 mM). There was no significant difference in ammonia concentrations in blood, liver, brain, kidney, and muscle in wild-type vs. AQP8-null mice (Fig. 4A). Also, neither urine ammonia concentration nor urine pH was different (Fig. 4B). Renal ammonia clearance was similar in AQP8-null mice (187 ± 24 ml/day) and wild-type mice (161 ± 19 ml/day). Finally, blood gas analysis showed no significant differences in Pco2 or pH in wild-type vs. AQP8-null mice (Fig. 4C).
To stress ammonia handling in vivo, plasma ammonia concentration was measured after an acute intravenous NH4Cl load. Different quantities of 150 mM NH4Cl were infused and serial blood samples collected at specified times over 1 h. After ammonia loading, plasma ammonia concentration increased abruptly and returned to baseline over ∼30 min in wild-type and AQP8-null mice (Fig. 5A). Figure 5B shows ammonia concentrations in liver at different times after low- and high-dose ammonia infusion. At 15 min after the high-dose ammonia load, the ammonia concentration in AQP8-null mice was somewhat lower than that in wild-type mice, although the difference was abolished by 30 min (Fig. 5B).
Chronic high-dose ammonia loading experiments were done as a maximal stress to potentially expose defects in hepatic uptake/metabolism of ammonia in AQP8 deficiency. Mice were given 280 mM NH4Cl in water as their only fluid source, as described previously (3). The wild-type and AQP8-null mice did not develop gross signs of hepatic encephalopathy. Blood gas analysis showed no significant differences in Pco2 and pH in wild-type vs. AQP8-null mice after chronic NH4Cl loading (Fig. 6A, left and center). Interestingly, blood ammonia concentration was greater in AQP8 null-mice after chronic NH4Cl loading (Fig. 6A, right). As expected, ammonia content was increased substantially in multiple tissues after NH4Cl loading (Fig. 6B). Ammonia content was significantly lower in liver of AQP8-null vs. wild-type mice and greater in kidney. Figure 7A shows remarkably increased urinary ammonia excretion during chronic ammonia loading, with a greater increase in the AQP8-null mice. Urinary excretion of titratable acid was increased during NH4Cl but not different in wild-type vs. AQP8-null mice (Fig. 7B). Urinary pH was mildly reduced (Fig. 7C). Together, these results suggest that in high-dose, chronic NH4Cl loading, impaired hepatic ammonia uptake in AQP8 deficiency results in reduced hepatic ammonia metabolism and consequent increased serum and renal ammonia.
AQP8 is expressed in apical membrane in crypt epithelial cells in proximal colon but not in transverse or distal colon (38). NH3 permeability was measured in closed colonic loops from the disappearance kinetics of NH4Cl. Figure 8 shows rapid disappearance of ammonia from loops in proximal colon, but without significant difference in wild-type vs. AQP8-null mice.
The purpose of this study was to determine the potential role of AQP8-facilitated NH3 transport in mammalian physiology. These experiments were motivated by reports that rat and human AQP8, as well as plant AQP8 homologs, could transport NH3 in heterologous expression and reconstituted systems (14, 16, 22). Our study utilized stably transfected epithelial cells and AQP8-knockout mice, which compared with wild-type mice have normal development, growth, and appearance (38). Our strategy was to confirm and characterize mouse AQP8-facilitated NH3 transport and then to test a series of phenotypes in wild-type vs. AQP8-knockout mice in which AQP8-facilitated NH3 transport could be important.
Ammonia exists in aqueous solutions in two molecular forms, NH3 and NH4+, which are likely transported by different molecular mechanisms. Protein-mediated transport of NH4+ has long been recognized (5, 19). Many proteins, including Na+-K+-ATPase (32), H+-K+-ATPase (4), Na+-K+-2Cl− cotransporter (8), and K+-Cl− cotransporter (1), have been shown to transport NH4+. For each of these proteins, NH4+ transport is believed to occur through substitution for K+ or, in the case of Na+/H+ exchangers, for H+ (12). Less attention has been given to NH3 transport, which was assumed to occur passively across the lipid bilayer by a solubility-diffusion mechanism. However, recently the Rh glycoprotein family has been implicated in ammonia transport (23, 24, 26). Members of the Rh ammonia transporter family, RhAG, RhBG, and RhCG, appear to mediate ammonia transport in most if not all living organisms (18, 34). Crystallographic analysis and reconstitution studies suggest that AmtB/Rh transports NH3 but not NH4+. This protein contains a narrow central hydrophobic pore element ∼20 μm long, which allows passage of NH3 but not NH4+ (18, 20).
Jahn et al. (16) recognized that human AQP8 shares the “aqua-ammoniaporin” signature found in plant AQPs that transport ammonia. With a functional complementation assay in a yeast mutant deficient in ammonia transport (Δmep1–3), plant AQP TIP2 and mammalian AQP8 were found to be ammonia permeable. A follow-up study by the same group using Xenopus oocytes confirmed the transport of NH3 by AQP8 and TIP2 (14). Evidence was also reported for apparent NH4+ transport by these proteins, although it was not possible to resolve whether NH4+ transport occurred through the AQP itself vs. secondary effects related to rapid NH3 transport.
In this study, AQP8-facilitated NH3 permeability was confirmed in stably transfected mammalian cells expressing rat or mouse AQP8. These measurements used a pH-sensing YFP to follow cytoplasmic pH changes resulting from NH3 transport and consequent instantaneous equilibration among NH3, NH4+, and pH. Our study did not address possible NH4+ transport by AQP8. NH3 and osmotic water permeability were significantly greater in AQP8-expressing than control cells. Quantitative analysis of permeability data showed a relative NH3-to-water single-channel permeability of ∼0.03. AQP8-facilitated NH3 and water permeability was also demonstrated in plasma membrane vesicles from mouse testis, confirming AQP8-facilitated NH3 in a native tissue.
In mammals, ammonia metabolism and transport are critical for whole body acid-base balance. Under basal conditions, blood, urine, and tissue ammonia concentrations in AQP8-null mice and wild-type mice did not differ significantly; nor did blood pH or Pco2. Additional studies were done under stress, measuring responses to acute intravenous ammonia infusions done with different ammonia loads. After an acute single NH4Cl infusion, blood ammonia concentration increased promptly and then returned to near baseline over 30 min. There was little effect of AQP8 deletion on the peak ammonia concentration or its kinetics, which is not unexpected based on data from the older literature indicating that skeletal muscle, which does not express AQP8, is a primary determinant of early ammonia redistribution in response to bolus administration (28).
The liver is an important organ for ammonia metabolism. Ammonia uptake and metabolism to urea or glutamine are essential for prevention of hyperammonemia and hepatic encephalopathy. RhBG is expressed strongly in hepatocytes and RhCG in bile duct epithelia (35, 36). We previously reported (38) AQP8 expression in the plasma membrane of mouse hepatocytes. Here we found similar baseline liver ammonia content in wild-type and AQP8-null mice, as well as similar blood and urinary ammonia, suggesting minimal if any role of hepatic AQP8-facilitated NH3 transport under normal physiological conditions. After single bolus infusion of a large amount of NH4Cl, there was a small but significant reduction in liver ammonia content in AQP8-null mice at 15 min, consistent with a small component of AQP8-facilitated NH3 transport by hepatocytes. In response to chronic, high-dose NH4Cl loading we found higher blood and urine ammonia in APQ8-null mice and reduced hepatic ammonia accumulation. These results are consistent with a small amount of AQP8-facilitated hepatic ammonia uptake, at least under nonphysiological conditions of high-dose chronic NH4Cl loading. It might be interesting to study ammonia handling in double-knockout mice lacking hepatic Rh protein(s) and AQP8.
Ammonia metabolism in kidney is critical for acid-base homeostasis (34). Two-thirds of net acid secretion in the urine involves nontitratable NH4+ excretion, with ammonia concentrations sometimes exceeding 50 mM. Renal ammonia transport involves multiple nephron segments, including the proximal tubule, thick ascending limb of the loop of Henle, and collecting duct (6, 11, 13, 31). AQP8 is thought to be expressed in kidney (7, 38), although its localization is unclear. Our previous data (38) indicated unimpaired urinary concentrating ability in AQP8-null mice. The data here showing normal baseline acid-base parameters, and appropriate renal response to acute and chronic NH4Cl loading, provide evidence against significant AQP8-facilitated ammonia handling in kidney.
Ammonia is produced in the gastrointestinal tract and absorbed into the portal circulation. In the colon, luminal ammonia is produced predominantly from metabolism of urea produced by enterocyte-mediated amino acid metabolism (12). In colon crypt cells, the basolateral Na+-K+-2Cl− cotransporter and the Na+/H+ exchanger are thought to be important for NH4+ transport (27). AQP8 is expressed at the luminal membranes of crypts in proximal colon (38). We found here that AQP8 deletion did not affect ammonia absorption in closed loops of colon after NH4Cl infusion. Ammonia transport was quite rapid, which may be related to the expression of RhBG and RhCG in the colon (12).
We note that the ammonia concentration measured here in mouse blood (without ammonia loading; 195 ± 21 μM) is greater than that in human blood (generally <50 μM). We confirmed the much lower ammonia concentrations in freshly drawn human blood samples assayed identically to the mouse blood. Studies of blood ammonia concentrations in mice have varied considerably with reported values of (μM) 46 ± 11 (21), 54 ± 7 (3), 80 ± 5 (10), 232 ± 67 (29), 316–571 (33), and 554 ± 124 (15). This considerable variation is probably due to a combination of factors, including different mouse strains, diets, and technical factors in the assay. We conclude that blood ammonia concentration is higher at baseline in mice vs. humans.
In summary, the data here indicate that mouse AQP8 facilitates water and NH3 transport, although the absolute contribution of AQP8 to NH3 transport is smaller than that to water transport. Although NH3 transport by aqua-ammoniaporins is of potential importance in plants and lower organisms, our data in mice argue against a physiologically important role of AQP8-facilitated ammonia transport.
This work was supported by National Institutes of Health Grants DK-35124, HL-59198, EY-13574, EB-00415, DK-72517, and HL-73856 (to A. S. Verkman), and DK-66194 (to B. Yang), a Research Development Program grant (R613) from the Cystic Fibrosis Foundation (to A. S. Verkman), and Grant 0365027Y from the American Heart Association (to B. Yang).
We thank Liman Qian for mouse breeding and genotype analysis.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society