INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

TWO-THIRDS OF NET ACID secretion in the urine is via ammonium (NH) where urinary levels of total ammonia can easily reach values in excess of 50 mmol/l. Along the nephron, NH is synthesized and secreted in the lumen of the proximal tubule (22), reabsorbed in the thick ascending limb of Henle's loop (15, 22), and is secreted again into the lumen of the collecting duct (14, 19). The classic model of NH₃/NH transport across cell membranes assumes that NH transport occurs via channels (e.g., K⁺ channels) or transporters (e.g., Na-K-2Cl cotransport or Na/H exchange), but NH₃ permeation across cell membranes occurs by nonionic diffusion through the lipid phase of the membrane. This concept has been supported by experiments in many different cell types. However, several studies indicate that certain nephron segments have restricted diffusion to NH₃ (19, 22). This is particularly evident in the apical membrane of the medullary thick ascending limb of Henle where NH₃/NH transport plays an important role in acid-base homeostasis. More recently, several other cell membranes have been observed to have very low permeability to NH₃ (21, 22, 29, 33, 36). Although the reasons for these differences in NH₃ permeability are not clear, variations in lipid composition are certainly a possibility. Another strong possibility is that certain membrane proteins, such as aquaporins (AQPs), may facilitate NH₃ transport and account for some of the differences among membranes from different cell types.

Water, like ammonia, was traditionally thought to cross membranes of most cells by solubility diffusion through the lipid bilayer. The existence of membranes with restricted H₂O permeability as with many tight, barrier epithelia (21, 34, 41) and the discovery of H₂O-selective channels (27, 28) challenged the universality of this concept. AQPs belong to a family of intrinsic membrane proteins functioning primarily as H₂O channels that facilitate significant transmembrane transport of H₂O in response to small osmotic gradients. The AQP(s) family of mammalian H₂O channels is growing and includes 10 homologs so far. All members have a major and common function as selective pores through which H₂O crosses plasma membranes. They all contain structural motifs similar to AQP1, the most studied AQP (3), with amino acid identities ranging from 20 to 52% (35). They are widely distributed in tissues but with little overlap. Their functional and structural properties are being actively investigated and have been reviewed lately (2, 5, 20, 35). For the most part, AQPs appear to function in a manner that provides a transcellular route of H₂O transport. However, the selectivity of AQPs to transport of other solutes and/or gases has not been well studied. AQPs are now classified into the following two groups: those that primarily transport H₂O (orthodox AQPs) and others that can transport small molecules such as glycerol and urea (1, 2). In a recent study, CO₂ permeability in oocytes expressing AQP1 was found to be ~40% higher than in control oocytes, suggesting that CO₂ can pass through AQP1 (24). These findings were later confirmed in studies showing HgCl₂ inhibition of CO₂ transport through AQP1 in oocytes (10) and in studies in reconstituted proteoliposomes (26). These were the first studies to demonstrate transport of a gas through a channel.

The aim of this study was to determine whether expressing AQP1 could facilitate transport of NH₃. The experiments were conducted on oocytes because of the ease of expressing AQP and the very low permeability of oocytes to NH₃. Because NH₃ (a gas) has a molar volume (24.9 cm³/mol) that is similar to that of H₂O (18 cm³/mol), permeation of NH₃ through AQP1 was hypothesized.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

General Methods

Solutions. The standard bathing solution used was ND-96 medium containing (in mM) 96 NaCl, 2 KCl, and 1.8 CaCl₂ buffered with 5 HEPES to pH 7.5. The NH₃/NH solution contained 20 mM NH₄Cl (replacing NaCl) at the desired pH. Osmolarity of all solutions was ~200 mosmol/l. OR3 medium (GIBCO-BRL Leibovitz media) contained glutamate and 500 units each of penicillin/streptomycin, with pH adjusted to 7.5 and osmolarity adjusted to ~200 mosmol/l.

Isolation of oocytes. We harvested oocytes in stage 5/6 from female Xenopus laevis. Briefly, this was done by anesthetizing the frog by mild hypothermia in H₂O containing 0.2% tricaine (3-aminobenzoic acid ethyl ester; Sigma, St. Louis, MO). A 1-cm incision was made in the abdominal wall, one lobe of the ovary was externalized, and the distal portion was cut. The wound was closed by a few stitches in the muscular plane of the peritoneum using 5-0 catgut followed by two to three stitches in the abdominal skin using 6-0 silk. The excised piece of ovary containing oocytes was rinsed several times with Ca-free ND-96 solution until the solution was clear. The tissue was then agitated in ~15 ml sterile filtered Ca-free solution containing collagenase type 1A (Sigma) for 30-40 min. Free oocytes were rinsed several times with sterile OR3 medium, sorted, and then stored at 18°C.

Preparation of cRNA. Plasmid containing the appropriate template DNA was purified by the Wizard Plus Minipreps DNA Purification System (Promega, Madison, WI). The plasmid was then digested with an appropriate restriction enzyme that has a cleavage site downstream of the insert to produce a linear template, followed by proteinase K (1 mg/ml) digestion. DNA was then phenol-chloroform extracted two times followed by chloroform extraction and ethanol precipitation. cDNA was transcribed in vitro with T7 RNA polymerase. The in vitro synthesis of capped RNA (cRNA) transcripts was then accomplished using the mCAPTM RNA Capping Kit (Stratagene, La Jolla, CA). The concentration of cRNA was determined by ultraviolet absorbance, and its quality was assessed by formaldehyde-MOPS-1% agarose gel electrophoresis (31).

Injection of oocytes. Oocytes in OR3 medium were visualized with a dissecting microscope and were injected with 50 nl of cRNA for AQP1 (0.02 µg/µl, for a total of 1 ng of RNA). Control oocytes were injected with 50 nl of sterile H₂O. The sterile pipettes had tip diameters of 20-30 µm. They were backfilled with paraffin oil and connected to a Drummond nanoject displacement pipette (Drummond Scientific). Injected oocytes were used 3-5 days after injection with RNA.

Electrophysiological measurements in frog oocytes. The pH microelectrodes were of the liquid ion exchanger type, and the resin (hydrogen ionophore I, cocktail B) was obtained from Fluka Chemical (Ronkonkoma, NY). Single-barreled microelectrodes were manufactured as described earlier (30). Briefly, alumina-silicate glass tubings (1.5 mm OD × 0.86 mm ID; Frederick Haer, Brunswick, MD) were pulled to a tip <0.2 µm and dried in an oven at 200°C for 2 h. The electrodes were vapor silanized with bis(dimethylamino)dimethyl silane in a closed vessel (300 ml). The exchanger was then introduced into the tip of the electrodes by means of a very fine glass capillary. pH electrodes were backfilled with a buffer solution (4). The electrodes were fitted with a holder with an Ag-AgCl pellet attached to a high-impedance probe of a WPI FD-223 electrometer. The pH electrodes were calibrated in standard solutions of pH 6 and 8. The average slope of 50 electrodes used in our studies was 59.0 ± 1.0.

Two-electrode voltage clamp. Whole cell currents were recorded using two-electrode voltage clamp (OC-725; Warner Instruments, Hamden, CT). For those experiments, electrodes were pulled from borosilicate glass capillaries (OD 1.5 mm; Fredrick Haer) using a vertical puller (model 700C; David Kopf Instruments). Electrodes were filled with 3 M KCl solution and had resistances of 1-4 M. Bath electrodes were also filled with 3 M KCl and were directly immersed in the chamber. For current measurements, oocytes were clamped at 60 mV, and long-term readings of current were sampled at a rate of one per second. Inward flow of cations is defined by convention as inward current (negative current). For measurement of whole cell conductance, oocytes were periodically pulsed (6 times/min) with a constant current (100 nA), and voltage deflections were recorded. Whole cell conductance was calculated from the current-to-voltage ratio.

Curve fitting, statistics, and data analysis. Initial rates of change in intracellular pH (pH_i; dpH_i/dt) were determined from the slope of the line obtained by fitting pH_i vs. time to a linear regression line. In all the experiments, values were reported as means ± SE. Statistical significance was judged primarily from two-tailed Student's t-tests. Whenever feasible, measurements were determined under control and test conditions in the same cell, and each cell served as its own control (paired data); "n" is the number of observations and is shown in parentheses. Results are considered statistically significant at P  0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

NH₃/NH-Induced pH_i and V_m Changes in H₂O-Injected Oocytes

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Effect of NH3/NH on H2O-injected oocytes. In control oocytes (H2O injected), exposure to NH3/NH (20 mM) decreased intracellular pH (pHi) by 0.29 ± 0.05 (segment ab) and depolarized the cell by 48 ± 2.3 mV. The pHi decrease and depolarization of the cell are consistent with a significant NH influx. The absence of any initial pHi increase indicates low permeability to NH3. Removal of NH3/NH reversed these changes (segment bc).

Results from experiments similar to those of Fig. 1 indicated that steady-state V_m and pH_i of H₂O-injected oocytes bathed in HEPES-Ringer averaged 55 ± 1.1 mV (n = 30) and 7.31 ± 0.03 (n = 28), respectively. In the presence of NH₃/NH, the oocyte depolarized to 7 ± 1.8 mV (n = 22), and pH_i decreased to 7.03 ± 0.05 (n = 20). The rate of pH_i decrease caused by NH₃/NH was 9.2 ± 1.1 × 10⁴ pH/s (n = 18), whereas the rate of pH_i recovery upon removal of NH₃/NH was 2.4 ± 0.3 × 10⁴ pH/s (n = 10).

Effect of NH₃/NH on H₂O-Injected Oocytes in the Absence of External Na⁺

As shown in Fig. 2, exposing oocytes to NH₃/NH caused a pH_i decrease (segment abc) and depolarization, as observed previously in Fig. 1. In this experiment (as with several other experiments), there was a small pH_i increase (segment ab), consistent with an apparent small NH₃ permeability. These changes were fully reversible upon removal of NH₃/NH (segment cd). At point d, Na⁺ was removed from the external solution (replaced with N-methyl-D-glucamine), which caused a sustained hyperpolarization, but pH_i did not change (segment de). In the continued absence of Na⁺, addition of NH₃/NH still caused a pH_i decrease (segment ef) and great depolarization of the oocyte. pH_i and V_m recovered when NH₃/NH was removed (segment fg). In seven experiments, the NH₃- and/or NH-induced pH_i decrease in the absence of Na⁺ was 0.18 ± 0.06, and the rate of acidification was 6.4 ± 1.7 × 10⁴ pH/s. Although both values were less than the NH₃- and/or NH-induced pH_i decrease in the presence of Na⁺ (0.28 ± 0.05 and 9.2 ± 1.1 × 10⁴ pH/s, n = 20, respectively), the differences were not statistically significant (P > 0.05).

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Effect of NH3/NH on H2O-injected oocytes in the absence of external Na+. NH3- and/or NH-induced pHi decrease and depolarization are not inhibited in the absence of external Na+. In paired experiments, exposing control (H2O-injected) oocytes to 20 mM NH3/NH in the presence of Na+ decreased pHi by 0.15 ± 0.02 and substantially depolarized the cell by 45 ± 5.9 mV (segment abc). In this experiment, there was a small initial increase in pHi (segment ab) consistent with a small NH3 permeability. These effects were reversed upon removal of NH3/NH from the bathing solution (segment cd). Removal of external Na+ did not affect pHi (segment de) but moderately hyperpolarized the cell (change in Vm = 15 ± 2.6 mV). Exposing oocytes to NH3/NH in the absence of Na+ still decreased pHi by 0.18 ± 0.06 (segment ef) and depolarized the cell by 74 ± 3.3 mV. Removal of NH3/NH reversed these changes (segment fg).

In several membranes with low NH₃ permeability and fast NH entry, such as the thick ascending limb of Henle, one possible route of NH transport is thought to be through K⁺ channels (13, 21, 37). To check whether inhibiting K⁺ channels could block NH influx, we performed the experiments depicted in Fig. 3. As shown in this experiment, exposure of H₂O-injected oocytes to NH₃/NH in the absence of Na⁺ caused a pH_i acidification (segment ab), and V_m became more positive, as shown in Fig. 2. Removal of NH₃/NH reversed these changes with a prompt V_m recovery and a slow pH_i increase toward the initial steady-state value (segment bc). Readdition of external Na⁺ (at point c) did not affect pH_i, which continued to recover (segment cd), but the cell depolarized slightly. When Na⁺ was removed again, there was no change in pH_i but a small hyperpolarization of ~10 mV (segment de). Addition of Ba²⁺ (1 mM) depolarized the cell by ~33 mV but did not affect pH_i (segment ef). Addition of NH₃/NH in the presence of Ba²⁺ and the absence of Na⁺ still caused a substantial depolarization and acidification of the cell (segment fg). All changes were fully reversible (segment ghi). In three similar experiments, Ba²⁺ slowed down the rate of pH_i decrease but did not prevent the NH-induced depolarization or illicit an NH₃-induced intracellular alkalinization.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Effects of NH3/NH on H2O-injected oocytes in the presence of Ba2+. The first part of the experiment shows the usual and reversible effects of NH3/NH. In the absence of bath Na+, NH3/NH caused a decrease in pHi (segment ab) and a cellular depolarization, as shown earlier. pHi and Vm fully recovered upon removal of NH3/NH (segment bc) and addition of Na+ (segment cd). In the second part, Ba2+, in the absence of Na+, depolarized the oocyte by 30 mV but did not affect pHi (segment ef). In the continued presence of Ba2+, NH3/NH still caused a decrease in pHi (segment fg), and Vm became more positive. The changes in pHi and Vm were reversed upon removal of NH3/NH (segment gh) and Ba2+ (segment hi).

Effect of NH₃/NH on Oocytes Expressing AQP1 in the Absence of External Na⁺

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Effect of NH3/NH on oocytes expressing aquaporin (AQP) 1 in the absence of external Na+. In oocytes expressing AQP1, removal of Na+ from the bath hyperpolarized the oocyte but did not cause any significant change in pHi (segment ab). Exposure to NH3/NH in the absence of Na+ caused a small pHi increase (rather than a decrease) and significant depolarization (segment bc). The depolarization indicates that NH transport is probably not affected. The changes in Vm and pHi were reversed upon removal of NH3/NH.

Effect of NH₃/NH at High Bath pH on H₂O-Injected Oocytes

To maximize the NH₃-induced signal (in the face of a substantial NH-induced acidification), we conducted the experiments shown in Fig. 5. In H₂O-injected oocytes, NH₃/NH (20 mM) at external pH [bath pH (pH_B)] of 7.5 decreased pH_i (segment abc) and depolarized the cell, as previously described. In this experiment, as occasionally seen, a transient small alkalinization (segment ab), presumably resulting from NH₃ entry, was observed. At point c, switching the external solution to NH₃/NH (20 mM) at pH_B of 8.0 (which raises outside NH₃ ~3-fold) caused an increase in pH_i (segment cd) with no significant change in V_m, suggesting that the pH_i change is primarily the result of NH₃. When pH_B was subsequently raised to 8.5 (thus raising external NH₃ further at the expense of NH), pH_i increased again (segment de) with no significant change in V_m. The pH_i increase was reversed when external NH₃/NH was returned to pH 7.5 (segment ef).

View larger version (14K): [in this window] [in a new window]   Fig. 5.   Effect of NH3/NH at higher bath pH (8.0 and 8.5) on H2O-injected oocytes. Exposing control oocytes to NH3/NH (at bath pH 7.5) caused a small transient alkalinization (segment ab) followed by a substantial decrease in pHi (segment bc), and the cell depolarized significantly as shown previously. After a new steady state was reached, raising bath pH from 7.5 to 8.0 (and therefore increasing external NH3 concentration) caused a slow increase in pHi (segment cd) at a rate of 3.1 × 104 pH/min. Elevating bath pH further from 8.0 to 8.5 caused another increase in pHi (segment de) at a rate of 7.6 × 104 pH/min. Vm did not change significantly throughout, indicating that the pHi increase is the result of increased NH3 entry rather than a change in NH transport. Switching back to NH3/NH at pH 7.5 reversed the changes in pHi (segment ef).

Effect of NH₃/NH at High Bath pH on Oocytes Expressing AQP1

4

4

View larger version (13K): [in this window] [in a new window]   Fig. 6.   Effect of NH3/NH at higher bath pH (8.0) on oocytes expressing AQP1. Exposing AQP1 oocytes to NH3/NH (at bath pH 7.5) caused a small and slow increase in pHi (segment ab). Increasing bath pH from 7.5 to 8.0, thus increasing external NH3 concentration, resulted in a significant increase in pHi (segment bc) at a rate of 12 × 104 pH/min. Switching bath solution to NH3/NH at pH 7.5 reversed the change in pHi, causing a decrease in pHi (segment cd) with no effect on Vm. Both pHi and Vm fully recovered when bath solution was switched to control (segment de).

The change in pH_i (0.32 ± 0.06) at pH_B of 8.0 was significantly more than that in control oocytes (0.05 ± 0.02), and the rate of pH_i increase was more than two times faster (9.1 ± 1.5 vs. 4.3 ± 1.2 × 10⁴ pH/s). In fact, in AQP1 oocytes, the rate of the NH₃-induced increase at pH_B of 7.5 was faster than that at pH_B of 8.0 in control oocytes. During these maneuvers, V_m was stable and did not change significantly toward a more negative value; hence, the flux of NH was likely unchanged. Therefore, the difference in the change in pH_i (and the rate) in response to increased NH₃ concentration in the bath (as when pH_B is raised) can be used to estimate the change in NH₃ permeability secondary to AQP1 expression. These experiments demonstrate that expressing AQP1 significantly enhanced NH₃ permeability in oocytes.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

In investigating NH₃ transport and the role of AQP1, we measured changes in pH_i and V_m induced by exposing oocytes to solutions equilibrated with NH₃/NH (20 mM NH₄Cl). In such an approach, as NH₃ enters the cell, it combines with intracellular H⁺ to form NH, which will lead to an increase in pH_i. In contrast, NH entry into the cell would lead to a decrease in pH_i when NH dissociates intracellularly, releasing NH₃ and H⁺. Therefore, an increase in NH₃ permeability would cause an alkaline shift in pH_i. Whereas a decrease in NH permeability could cause an increase in pH_i, it would also lead to a change in V_m (with the cell expected to hyperpolarize). Relying on such measurements, the results of the present study indicate that 1) oocytes have low but finite permeability to NH₃, 2) oocytes have high apparent permeability to NH, and 3) expression of AQP1 increases permeability of NH₃ without substantially affecting NH transport in the oocyte.

We used Xenopus oocytes in this study because of the distinct advantages they provide. First, the oocyte is a very convenient system for expressing exogenous proteins and transporters. Second, it is easy to obtain long-time and stable measurements by microelectrodes. This is highly advantageous because pH_i measurements by microelectrodes are very reliable and more accurate than other methods. Third, pH_i changes that occur in the oocytes are relatively slow; therefore, accurate measurements can be obtained. This contrasts with measurements in other preparations in which NH₃-induced pH_i changes are extremely fast, as is the case in most mammalian cells (25), or in which NH₃ permeability varies dramatically (and the background permeability is very high) with different lipid composition as is the case with studies in lipid bilayers (40).

In the kidney, NH₃ transport is very important, yet not well studied. In the proximal tubule, significant NH₃ transport occurs in addition to NH transport (18). NH₃ transport is the predominant mode of entry of total ammonia in the collecting duct (21). The predominant mechanism of NH₃ transport in most tubular segments is presumed to be lipid phase diffusion through the membrane. However, NH₃ permeability varies considerably along the nephron. For example, whereas significant NH₃ diffusion occurs in the proximal tubule (18), permeability to NH₃ is substantially less and diffusion of NH₃ is more restricted in the cortical and medullary collecting ducts (16, 19, 22). Other studies (21) reported very little permeability of the apical membrane to NH₃ in the thick ascending limb of Henle's loop. Various studies proposed several mechanisms to account for this variability in NH₃ transport ranging from lipid solubility differences to specific channels or carriers to nonspecific pathways that may mediate NH₃ transport. Although speculative, the correlation between low H₂O permeability and low NH₃ permeability in the thick ascending limb and the collecting duct has been noted (21) and is suggestive that NH₃ may be transported through specific carriers, perhaps even H₂O channels.

In this study, the data indicate that, in control (H₂O-injected) oocytes and in AQP1 oocytes, permeability of NH is particularly high. The evidence for this is based on three observations. First, exposure of oocytes to NH₃/NH caused a substantial decrease in pH_i consistent with a big influx of NH and its subsequent release of intracellular H⁺, leading to intracellular acidification. Second, exposure to NH₃/NH caused a huge depolarization of the oocyte to near 0 mV, indicating an electrogenic pathway also consistent with NH influx. The amount of cell depolarization caused by 20 mM NH₄Cl (48 ± 1.9 mV) is substantially bigger than the depolarization caused by 20 mM K⁺ (18 ± 1.1 mV, n = 3) and is not inhibited by Ba²⁺ (see Fig. 3). Third, NH₃/NH exposure of oocytes induced an inward current in voltage-clamp experiments (23). These observations confirm previous studies which suggested that NH influx in oocytes occurs through a nonselective cationic channel (7, 8, 12).

It is also likely that at least a fraction of NH influx could occur through an electroneutral pathway such as Na-K-2Cl cotransport. Several studies have addressed this issue and concluded that the conductive pathway of NH is probably the major contributor to NH influx. Sasaki et al. (32) could only slightly inhibit NH-induced acidification in the presence of furosemide, and Burckhardt and Fromter (8) showed no significant effect of bumetanide on NH-induced V_m changes or cellular acidification. Our own experiments, with bumetanide, agree with the above studies.

This high permeability of NH and its effects on V_m and pH_i complicates examination of NH₃ permeability in the oocyte. The path of NH transport has been poorly defined and difficult to inhibit but is presumably a cationic conductive pathway. In preliminary experiments, cinnamate, diphenylamine-2-carboxylic acid, and bumetanide did not inhibit the NH-induced depolarization. Other inhibitors, including Cs⁺, tetramethylammonium, and quinidine were tried by other investigators (8, 12) and also did not inhibit the NH-induced depolarization of the cell. In the absence of a specific inhibitor of this NH pathway, an NH-induced pH_i decrease could conceivably mask an NH₃-induced pH_i increase. Our data, however, indicate that oocytes have a low but finite permeability to NH₃.

The low permeability of oocytes to NH₃ compared with that of NH is evident because there is usually no apparent NH₃-induced alkalinization. In some experiments, however, there was a small transient alkalinization that preceded the sustained pH_i decrease in response to NH₃/NH. However, our data indicate that, unlike other virtually NH₃-impermeable membranes such as the distal colon or gastric glands or even the medullary thick ascending loop, the oocyte membrane is permeable to NH₃. The evidences for this are the following observations. First, the significant decrease in pH_i in response to NH₃/NH cannot be achieved unless NH₃, generated intracellularly from the dissociation of NH, can exit the cell. In control oocytes, for example, NH₃/NH exposure causes a pH_i decrease of 0.29 ± 0.04 (n = 20). With an estimated buffering power of 12.4 mM/pH, calculated from CO₂ pulses (24), and assuming that this pH_i decrease is caused solely by the influx of NH, with no significant permeability to NH₃, intracellular NH concentration has to reach an impossibly high level of ~571 mM (see APPENDIX). In other words, for the influx of NH to cause this amount of decrease in pH_i, the newly generated NH₃ has to leave the oocyte, hence, a permeable membrane, thus driving the dissociation reaction of NH  NH₃ + H⁺ to the right. The second evidence is more direct and is derived from the protocol employed in the experiments of Figs. 5 and 6. In these experiments, increasing the ratio of NH₃ to NH by raising pH_B resulted in a direct increase in pH_i. Thus an increase in the NH₃ gradient across the membrane is reflected in an increase in the NH₃-induced pH_i change as expected if NH₃ is permeable through the membrane. This increase in pH_i is not likely caused by a decrease in NH influx because V_m was not affected at all. To further verify this latter point, we ruled out a major change in NH permeability in oocytes expressing AQP1 by two ways. First, in voltage-clamped experiments, we measured the current induced by NH in the presence and absence of expressed AQP1. Exposure to NH caused an inward current of 169 ± 20.4 nA (n = 3) in AQP1 oocytes and 145.3 ± 9.2 nA (n = 3) in H₂O-injected oocytes. These values were not significantly different (P > 0.05), and this small change in NH-induced current is opposite to what would be expected if NH conductance decreased. Second, we assessed total membrane conductance in the presence and absence of AQP1 from voltage deflections in response to a constant current pulse. These results show that total membrane conductance in oocytes expressing AQP1 (1.43 ± 0.09 µS, n = 3) was not statistically different (P > 0.05) from that of H₂O-injected oocytes (1.09 ± 0.15 µS, n = 3).

Our experiments on oocytes expressing AQP1 indicate that expressing the H₂O channel enhanced NH₃ transport with little or no effect on NH transport. The possibility that AQP1 may be permeable to NH₃ has its parallel in previous studies, which demonstrated that another gas, CO₂, can also be transported through AQP1 (10, 24, 26). NH₃ has a small molar volume (24.9 cm³/mol) that is close to that of H₂O (18 cm³/mol); therefore, it is conceivable that H₂O channels may also be permeable to NH₃ as well.

The permeability of AQP(s) in general to solutes other than H₂O remains controversial. Various AQP(s), although highly permeable to H₂O, have been reported to transport other solutes as well. Abrami and coworkers (1) reported that AQP1 has a low permeability to glycerol, ethelene glycol, and 1,3-propanediol. On the basis of their ability to transport glycerol and various other solutes, a set of AQP(s) is referred to as "aquaglyceroporins" and include several mammalian AQP(s), such as human AQP3, rat AQP7, and rat AQP9 (for a review see Refs. 2, 20, 35). Although CO₂ permeability through AQP1 (24) has been confirmed by other studies (10, 26), studies on red blood cells from AQP1-deficient mice failed to elucidate an effect on CO₂ transport (38). This last study also failed to show a change in NH₃ transport across red blood cells from AQP1-deficient mice. Another study (40) on proteoliposomes with reconstituted AQP1 did not show a significant effect on NH₃ permeability either. Both studies, however, were conducted on membranes with very high baseline permeability to NH₃. Nodulin 26, a plant AQP, was recently reported to transport NH₃ and NH (11). A more recent study on AQP6 indicates that it may function as an anion conductive pathway under certain conditions (39). Non-H₂O transport function of some AQP(s) is increasingly recognized as a property that may have important implications (2, 20, 35).

In our study, two lines of evidences indicate increased NH₃ permeability in oocytes expressing AQP1. First, NH₃- and/or NH-induced acidification (consistently observed in control oocytes) was usually absent when AQP1 was expressed (see Fig. 4). Because NH transport was presumably unchanged (voltage change not altered with AQP1), an alkalinization secondary to NH₃ influx likely masked the acidification from NH entry. Second, when oocytes were preequilibrated with NH₃/NH, raising pH_B (and therefore increasing the ratio of NH₃ to NH) induced a pH_i increase at a rate that was more than twofold faster in AQP1 oocytes compared with control oocytes. In fact, the rate of pH_i increase at pH 8.0 in AQP1 oocytes was faster than the rate of pH_i increase induced by raising pH_B to 8.5 in control oocytes. Both sets of experiments are consistent with an increased NH₃ transport when AQP1 is expressed.

Several observations indicate that the increased NH₃ transport with the expression of AQP1 may potentially be physiologically significant. For example, in the mammalian proximal tubule, where AQP1 H₂O channels are highly expressed, significant transport of NH₃ also occurs (17, 18). H₂O channels could, in principle, be a major component of the pathway of NH₃ transport in the proximal tubule. On the other hand, in the thick ascending limb, NH₃ transport across the apical membrane is limited because of a low relative permeability to NH₃. This limited NH₃ transport does correlate with an apparent lack of H₂O channels in the apical membrane of this segment, but other factors, such as the lipid composition of the membrane, may also play a role. Further along the nephron, in the collecting duct, entry of total ammonia is thought to be mediated by both NH₃ and NH across the basolateral membrane and NH₃ diffusion across the apical membrane. The basolateral membrane has AQP3 and AQP4, and the apical membrane has AQP2 water channels inserted in the presence of antidiuretic hormone vasopressin. The permeation of NH₃ through AQP2, -3, and -4 has not yet been documented but certainly deserves further investigation, particularly since this would represent a route for regulation of NH₃ transport via H₂O channels. In summary, there is correspondence between the presence of H₂O channels in various membranes along the nephron and NH₃ transport. The present findings indicate that H₂O channels may be important in mediating NH₃ transport, but this awaits further investigation. To date, AQP knockout animals have not demonstrated acid-base abnormalities, but whether defects in ammonia transport could be uncovered with more direct testing awaits further studies.