Nickel and extracellular acidification inhibit the water permeability of human aquaporin-3 in lung epithelial cells.

Nickel is a common cause of pneumoconiosis. Here, we show that nickel inactivates aquaporin (AQP)-3, the water channel expressed apically in epithelial cells of human terminal airways. Human AQP3 was transiently transfected into human lung cells, and water permeability was measured in transfected and neighboring untransfected cells. Incubation with NiCl2 rapidly, dose-dependently, and reversibly decreased water permeability in AQP3-expressing cells. Acidification of the extracellular medium also caused rapid, dose-dependent, and reversible inhibition of AQP3. Sensitivity of AQP3 to nickel was lower at alkaline pH than at neutral and acidic pH. Cells transfected with human AQP4 and AQP5, which are also expressed in airway epithelia, were insensitive to nickel and extracellular acidification. Zinc and cadmium, other common causes of pneumoconiosis, had no effect on the water permeability of AQP3. Three extracellular residues, Trp128, Ser152, and His241, were responsible for the blocking effect of nickel on human AQP3. Ser152 was identified as a common site for nickel and pH sensitivity. His53, Tyr124, and His154 were also involved in regulation of AQP3 by extracellular pH. In addition, the aromatic side chain of His154 was shown to be important for the water permeability of AQP3. Our results imply that nickel and extracellular pH may modulate lung water clearance and that defective water clearance may be an early component of nickel-induced lung disease.

Nonenzymatic regulation of ion channels by Ni 2ϩ and other divalent cations or by pH is a well established phenomenon with many important physiological and pathophysiological implications. Less is known about nonenzymatic regulation of water channels, aquaporins (AQPs) 1 (1). Mercury inhibits most mammalian water channels via binding to cysteine residues (2)(3)(4) and has been an important tool in studies of AQPs. Gold and silver were recently reported to inhibit a water channel from human erythrocytes, presumably AQP1, but a molecular basis for this inhibition has not been revealed (5). The question of whether Ni 2ϩ and other divalent ions known to regulate the activity of ion channels modulate the activity of AQPs has, to our knowledge, not yet been addressed.
Nickel is widely used in modern industry (reviewed in Ref. 6). Inhalation is the primary route of occupational exposure to nickel and other heavy metals, and inhalation of nickel compounds is a common cause of pneumoconiosis (6,7). AQP3, AQP4, and AQP5 are expressed in the airway epithelia (8 -10). AQP3 is located at the apical membrane of human lung epithelium (10). Here, we have examined the effects of Ni 2ϩ on the water permeability of human AQP3, AQP4, and AQP5 expressed in a human lung cell line. Since AQP3 has, when expressed in oocytes, been reported to be pH-sensitive (11), we also examined the effect of extracellular acidification. We show that Ni 2ϩ and pH regulate the water permeability of human AQP3, but not of human AQP4 and AQP5. We also address the question of whether Ni 2ϩ and pH may interact in the regulation of human AQP3.
Identification of the molecular sites responsible for the Ni 2ϩ and pH sensitivity of AQP3 is important for future development of therapeutic agents. Histidine, with a pK a of ϳ6.5, is the most likely molecular target for regulation by pH and is also a preferential site for Ni 2ϩ binding. By performing a series of mutations of extracellular histidines and amino acids considered to interact with histidine, we identified several molecular determinants of pH and Ni 2ϩ sensitivity and at least one common determinant of Ni 2ϩ and pH sensitivity.

MATERIALS AND METHODS
DNA Constructs-cDNA fragments encoding full-length AQP3 and the long form of AQP4 were obtained by amplification from the human lung QUICK-Clone cDNA library (Clontech). cDNA encoding human AQP5 was a generous gift from P. Agre (Department of Biological Chemistry, Johns Hopkins University School of Medicine, Baltimore, MD). The cDNA fragments were used for creation of two types of cDNA constructs: constructs that expressed a water channel fused with green fluorescent protein (GFP) and constructs that expressed a water channel and GFP as separate proteins present in the same cell. For the first type of construct, cDNA fragments were subcloned in-frame into the pEGFP-N2 vector for AQPs tagged with GFP at the COOH terminus and in-frame into the pEGFP-C2 vector for AQPs tagged with GFP at the NH 2 terminus. For the second type of construct, cDNA fragments were subcloned into the pIRES2-EGFP vector (Clontech).
The point mutations in the extracellular loops of human AQP3 were generated by PCR-based mutagenesis using wild-type cDNA as a template. The presence of each point mutation and absence of other modifications were confirmed by sequence analysis of the whole insert.
The transmembrane structure of human AQP3 was predicted using TMHMM Version 2.0 (12). The protein sequences were aligned with ClustalW Version 1.81 (13).
Cell Culture-The human bronchial epithelial cell line BEAS-2b (subpassages 10 -36; European Collection of Cell Cultures, Center for Applied Microbiology and Research, Salisbury, Wiltshire, UK) was cul-tured on coverslips (Bioptechs, Butler, PA) coated with collagen type I and fibronectin (Sigma, Stockholm, Sweden) in Dulbecco's modified Eagle's medium/nutrient mixture F-12 (1:1; Invitrogen, Paisley, Scotland, UK) containing 0.5 units/ml penicillin and 50 g/ml streptomycin and supplemented with 10% heat-inactivated fetal bovine serum and 2 mM L-glutamine. On the second day of culture, the cells were transiently transfected with cDNA constructs (see above) using CLONfectin (Clontech) according to the manufacturer's protocol. Experiments were performed on the fourth day of culture.
Measurement of Water Permeability-The water permeability (P f ) was measured using a method that we recently described in detail (14). The method allows one to determine the P f in individual cells within cell monolayers and to compare the P f in cells that do and do not express GFP-labeled proteins. Briefly, the coverslips with the transfected cells were mounted in a closed perfusion chamber (Focht Live Cell Chamber System; Butler, PA) on the stage of a Zeiss 410 inverted laser scanning microscope and scanned every 2 s with excitation at 488 nm and emission at 515-525 nm. At the beginning of every P f measurement, an image showing the distribution of GFP-tagged proteins was recorded. Then, with the specimen remaining on the stage of the microscope, the cells were loaded with calcein (Molecular Probes Europe, Leiden, The Netherlands). To measure the P f , the cells were perfused with 300 mosM PBS (137 mM NaCl, 0.9 mM CaCl 2 , 0.49 mM MgCl 2 , 2.7 mM KCl, 1.5 mM KH 2 PO 4 , and 8.1 mM Na 2 HPO 4 , pH 7.4). The perfusate was then switched to 200 mosM PBS (osmolarity was decreased by reducing the concentration of NaCl to 87 mM). To determine the P f at a different pH, the pH of both the 300 and 200 mosM perfusion solutions was adjusted to the required value. In solutions at pH 8.0 and higher, CaCl 2 and MgCl 2 were omitted. In experiments with NiCl 2 , ZnCl 2 , and CdCl 2 , phosphates in all solutions were replaced with 10 mM HEPES, and 1.5 mM KCl was added to preserve potassium concentration.
The P f was calculated using the following equation (14): The time constant was calculated for every cell from the curve showing changes in fluorescence intensity inside the cell during osmotic swelling. 1 Ϫ b/V 0 represents the osmotically active portion of the cell volume. This portion was calculated using direct volume measurements performed using the Imaris Image Processing Toolkit (Bitplane, Zurich, Switzerland) on stacks of images recorded through the calcein-loaded cells at 300 mosM and through the same cells after complete swelling in 200 mosM PBS. For BEAS-2b cells, the osmotically active portion of the cell volume was found to be 0.15 Ϯ 0.02 (n ϭ 7). The constant ␥ was calculated as the slope of the relative fluorescence versus the relative osmolarity calibration curve. For BEAS-2b cells, it was found to be 0.63. (A/V) 0 is the initial cell surfaceto-volume ratio. It was calculated using Imaris on stacks of images recorded through the calcein-loaded cells incubated in 300 mosM PBS. For BEAS-2b cells, the ratio was found to be 5111 Ϯ 286 cm Ϫ1 (n ϭ 9). V w is the partial molar volume of water (18 cm 3 /mol), and ⌬ 0 is the initial osmotic gradient (outside-inside), which was 10 Ϫ4 mol/cm 3 in our experiments.
The basal P f in BEAS-2b cells and the P f in transfected cells could vary depending on the lot of CLONfectin used for the transfection and the cell culture subpassage. Therefore, we statistically compared only the P f in cells that were at a similar subpassage and transfected with the same CLONfectin preparation.
Analysis of the Subcellular Distribution of GFP-tagged AQP3-To determine whether the effects of acidic pH or NiCl 2 were due to a changed amount of AQP3-GFP in the plasma membrane, the transfected cells were mounted in the same chamber as used for P f measurements in the control solution. A stack of images with a vertical displacement of 0.4 m was recorded. The solution was then changed to a pH 5.5 solution or a solution containing 1 mM NiCl 2 , and a new stack of images was recorded. The ratio of GFP signal in the plasma membrane to that in the adjacent cytosol was measured using ImageJ software (Research Services Branch, National Institute of Mental Health, Bethesda, MD). 2 The subcellular localization of AQP3 was compared in the same cells in two different solutions. Images at the same cell height were taken for the comparison. No swelling or shrinkage of the cells was observed in response to low pH or NiCl 2 .
Measurement of Intracellular pH-Cells were loaded with the pHsensitive fluorescent dye 2Ј,7Ј-bis(2-carboxyethyl)-5(6)-carboxyfluorescein (BCECF) by incubation with 10 M BCECF/acetoxymethyl ester (Molecular Probes Europe) for 30 min at 37°C. The cells were mounted in the same chamber used for P f measurements (see above) on a Zeiss Axiovert 135 microscope using a ϫ40/1.4 epifluorescence oil immersion objective. Fluorescence was alternately excited at wavelengths of 440/10 nm and 470/10 nm; emission fluorescence was collected with a 510/30-nm band pass filter. Data were recorded with a GenIISys image intensifier system connected to a CCD camera (MTI CCD72, Dage-MTI, Inc., Michigan City, IN) and analyzed using RatioTool software from ISee Imaging Systems (Raleigh, NC). At the end of each experiment, the intracellular pH was calibrated using buffer solutions containing 150 mM KCl, 5 mM MES or HEPES, and 10 M nigericin adjusted to pH 6.0, 6.5 (MES), 7.0, 7.5, and 8.0 (HEPES).
Data Presentation and Analysis-Data are presented as means Ϯ S.E. Statistical analysis was performed using Student's t test. A difference of p Ͻ 0.05 was considered statistically significant.

RESULTS
Distribution and Water Permeability of Human AQP3 Transfected into Lung Cells-Human AQP3 was transiently transfected into a bronchial epithelial cell line (BEAS-2b). GFP was used to identify cells expressing AQP3. In cells transfected with AQP3 tagged with GFP at the COOH terminus (AQP3-GFP), a distinct GFP signal was observed in the plasma membrane, whereas the signal was very weak or undetectable in the cytoplasm (Fig. 1a). The plasma membrane signal was evenly distributed along the apical and basal sides of the cells (Fig. 1b). In cells transfected with AQP3 and GFP as separate proteins (AQP3ϩGFP), the GFP signal was distributed throughout the cytoplasm (Fig. 1, c and d). At neutral and alkaline pH, cells expressing either AQP3-GFP or AQP3ϩGFP had a significantly (6 -12-fold) higher P f than neighboring cells lacking a GFP signal. From this, we concluded that the AQP3-GFP fusion protein was a fully functional water channel.
Sensitivity of Human AQP3 to pH and Ni 2ϩ -Acidification of the extracellular solution dose-dependently decreased the P f in cells expressing AQP3-GFP without affecting untransfected cells (Fig. 1e). At pH 5.5, the P f in transfected cells was not different from that in untransfected cells. The pH effect was similar in cells expressing AQP3-GFP and in cells expressing AQP3ϩGFP. Inhibition of AQP3-mediated water permeability by acidic pH was rapid and reversible ( Fig. 1, f and g). There was no detectable change in the distribution of human AQP3-GFP in cells exposed to low extracellular pH. The ratio of GFP signal in membranes to that in the cytosol was 2.52 Ϯ 0.27 before and 3.00 Ϯ 0.27 (n ϭ 8) after 1 min in pH 5.5 solution, when the AQP3-mediated water permeability was completely inhibited. Notably, the intracellular pH was only little affected when the extracellular pH was reduced from 7.4 to 5.5 (7.34 Ϯ 0.01 and 7.27 Ϯ 0.01, respectively; n ϭ 58).
Nickel also decreased the P f in cells expressing human AQP3-GFP in a dose-dependent manner (Fig. 2a). The effect was rapid and reversible (Fig. 2b). The P f in untransfected cells was not affected by Ni 2ϩ . There was no detectable change in the distribution of AQP3-GFP in cells exposed to Ni 2ϩ . The ratio of GFP signal in membranes to that in the cytosol was 3.89 Ϯ 0.34 before and 4.05 Ϯ 0.31 (n ϭ 8) after 1 min in solution with 1 mM NiCl 2 , when the AQP3-mediated water permeability was dramatically down-regulated.
The effect of Ni 2ϩ was pH-dependent ( Fig. 2c). At neutral and acidic pH, the AQP3-mediated water permeability was completely inhibited by 1 mM NiCl 2 . At pH 7.4 and 8.0, the P f in transfected cells was decreased by Ni 2ϩ , but remained significantly higher than that in untransfected cells.
Specificity of the pH and Ni 2ϩ Effect on Human AQP3-AQP4 and AQP5 are, like AQP3, expressed in the epithelial cells of the lower airways of the human lung (10). BEAS-2b cells were transfected with either human AQP4 or AQP5 using cDNA constructs encoding GFP and the water channels as separate or fusion proteins. The water permeability was increased 6 -8-fold in cells expressing AQP4 or AQP5 compared with untransfected cells. Lowering the extracellular pH to 5.5 or exposing these cells to Ni 2ϩ had no effect on the P f (Fig. 3, a  and b).
Like nickel, zinc and cadmium are well documented causes of pneumoconiosis (6, 7). Cells transfected with AQP3-GFP were incubated with 1 mM ZnCl 2 or CdCl 2 . The P f in cells expressing AQP3-GFP or in untransfected cells was not affected by either of these compounds (Fig. 3c).
Extracellular Determinants of the Ni 2ϩ and pH Sensitivity of Human AQP3-To examine the molecular determinants of Ni 2ϩ and pH sensitivity, we performed a series of point muta- FIG. 1. Distribution and water permeability of AQP3 expressed in BEAS-2b cells. a, AQP3 tagged with GFP at the COOH terminus was located in the plasma membranes of transfected cells. b, the XZ projection image shows that AQP3-GFP was present in both the apical and basal sides of the cells. c and d, in cells transfected with AQP3 and GFP as separate proteins, the GFP signal was distributed in the cytoplasm, labeling cells that express AQP3. e, the water permeability in cells expressing AQP3-GFP (E) was regulated by extracellular pH; the P f in untransfected cells (‚) was not affected. Data from 17 to 39 cells were used for each point of the curves. f and g, regulation of AQP3 by pH was rapid and reversible. f, measurement of the water permeability in individual cells. Fluorescence intensity was monitored in cells loaded with calcein. Vertical arrows indicate when the osmolarity was changed from 300 to 200 mosM (2) and back to 300 mosM (1). The slope of the curves after the change in osmolarity is proportional to the rate of cell swelling and is a measure of the water permeability in the cells. The obvious difference in the rate of swelling of transfected (trace A) and untransfected (trace B) cell at pH 7.4 disappeared at pH 5.5 and reemerged when the extracellular pH was returned to 7.4. g, shown is the mean P f in transfected (black bars; n ϭ 15) and untransfected (white bars; n ϭ 28) cells from the experiment described for f.  8 -17). b, the effect of Ni 2ϩ on cells expressing AQP3-GFP (black bars; n ϭ 25) was reversible. The time scale and setup of the experiment were similar to those shown in Fig. 1f. The P f measurement was first performed under control conditions and then in the presence of 1 mM NiCl 2 and after 1 min of washout with 300 mosM PBS. Nickel had no effect on untransfected cells (white bars; n ϭ 41). c, the effect of Ni 2ϩ on cells expressing AQP3-GFP was pH-dependent. The P f was first measured under control conditions and then in the same cells in the presence of 1 mM NiCl 2 . At pH 5.5, 6.0, and 7.0, the water permeability in cells expressing AQP3-GFP (n ϭ 7-16) was decreased by Ni 2ϩ to the level of neighboring untransfected cells. At pH 7.4 and 8.0, the P f in transfected cells was also decreased by NiCl 2 , but was significantly higher than that in neighboring untransfected cells. tions in the three extracellular loops of AQP3 (Fig. 4). The extracellular loops were chosen for the following reasons. First, the regulation of AQP3 by Ni 2ϩ and pH was rapid and reversible. Second, the changes in extracellular pH that inhibited the water permeability of AQP3 did not lead to any substantial changes in intracellular pH. Third, inhalation is the main route for Ni 2ϩ exposure. AQP3 is expressed apically in the human airways, and its extracellular loops will be the first target for nickel binding.
The following amino acids were chosen for mutation: histidine, which can interact with Ni 2ϩ (15) and which is the only amino acid with a pK a in the range for AQP3 pH sensitivity; serine, aspartate, tryptophan, and tyrosine, since they may interact with histidine (16 -18). The results from the mutation studies are summarized in Table I, and the most relevant results are presented in Figs. 5 and 6. All mutants except D219A and W231A (Fig. 5, a and b) were targeted to the plasma membranes of the cells. In two cases (S49A and H154A), mutant AQP3 was water-impermeable. In five cases (H53A, H53F, H53Y, Y124A, and H154Y), the maximal water permeability in cells expressing mutant AQP3 was significantly decreased compared with that in cells expressing wild-type AQP3.
Three residues, Trp 128 and Ser 152 in the second extracellular loop and His 241 in the third extracellular loop, were identified as determinants of AQP3 Ni 2ϩ sensitivity (Fig. 5, c and d). The P f in cells expressing AQP3(W128A), AQP3(S152A), or AQP3(H241A) was similar to that in cells expressing wild-type AQP3. Ni 2ϩ had no effect on the water permeability of any of these mutants.
None of AQP3 mutants completely abolished the pH dependence of the water channel. In cells expressing AQP3(H53A), AQP3(Y124A), or AQP3(H154F), the range of pH sensitivity was shifted to more alkaline values (Fig. 6, a-c). In cells expressing AQP3(S152A), the range of pH sensitivity was shifted to more acidic pH (Fig. 6d). The water permeability of wild-type AQP3 was completely inhibited at pH 5.5. At this pH, the P f in cells expressing AQP3(S152A) was 3.9-fold higher compared with that in the surrounding untransfected cells. DISCUSSION The biological importance of membrane transporters is, to a large extent, dependent on their capacity to be regulated. Here, we show that the water permeability of human AQP3 expressed in human lung cells is regulated by changes in extracellular pH and by Ni 2ϩ . Ser 152 in the second extracellular loop of AQP3 was identified as a common determinant of Ni 2ϩ and pH sensitivity. Our findings have several important potential implications for the understanding of lung physiology and pathophysiology, as well as for the physiology of other organs in which AQP3 is expressed, such as the kidney.
It is well documented that Ni 2ϩ can up-or down-regulate the activity of ion channels (19 -24), but this is, to the best of our knowledge, the first demonstration of Ni 2ϩ regulation of the activity of a water channel. The effect of Ni 2ϩ was immediate, reversible, and dependent on at least three extracellular residues. In studies of the mechanisms by which Ni 2ϩ modulates the activity of cyclic nucleotide-gated channels and the epithelial sodium channel, extracellular histidines were found to be the molecular determinants of Ni 2ϩ sensitivity (19,24). Nitrogen in the imidazole side chain of histidine is the preferred target for Ni 2ϩ binding (15). Human AQP3 has one histidine in the first extracellular loop, two in the second, and one in the FIG. 4. Extracellular determinants of the pH and Ni 2؉ sensitivity of human AQP3. The numbers indicate amino acid residues that were mutated. Black circles show the residues participating in the regulation of AQP3 by extracellular pH. Light gray circles show the residues participating in the regulation of AQP3 by Ni 2ϩ . Ser 152 (dark gray circle) is involved in both pH-and Ni 2ϩ -dependent regulation of AQP3. FIG. 3. Specificity of the effects of acidification and Ni 2؉ on AQP3. a, the water permeability of human AQP3, but not of AQP4 and AQP5, was decreased by acidic extracellular pH. b, the water permeability of AQP3, but not of AQP4 and AQP5, was decreased by 1 mM NiCl 2 . c, nickel, but not zinc or cadmium, decreased the water permeability of AQP3. NiCl 2 , ZnCl 2 , and CdCl 2 were used at a concentration of 1 mM. Data are differences between the P f in transfected and surrounding untransfected cells, normalized to the corresponding control (n ϭ 12-34 cells).
third. In this study, we found that one of these four extracellular histidine residues, His 241 , is crucial for the Ni 2ϩ sensitivity of human AQP3. The side chain of tryptophan has a structure that is similar to that of the imidazole ring of histidine. Therefore, we reasoned that tryptophan may also be a candidate for Ni 2ϩ binding. Indeed, we found that substitution of Trp 128 with alanine completely abolished the Ni 2ϩ sensitivity of AQP3. Negatively charged amino acids such as aspartate and glutamate can also bind Ni 2ϩ (25). None of the examined extracellular aspartates proved to be important for the regulation of AQP3 by Ni 2ϩ . Glutamate residues are not present in the extracellular loops of human AQP3. Ser 152 was the third residue in the extracellular loops of AQP3 that was also found to be crucial for Ni 2ϩ sensitivity. This residue, as well as Tyr 124 , was studied because the effect of Ni 2ϩ was pH-dependent. Mutation of His 241 , Trp 128 , or Ser 152 led to complete elimination of Ni 2ϩ sensitivity. This indicates that these three amino acids together are essential for Ni 2ϩ binding to human AQP3.
Mutation of His 53 to alanine attenuated and of His 154 to alanine abolished the water permeability of AQP3. We hypothesized that the imidazole rings of His 53 and/or His 154 may participate in stacking interactions, important for the proper conformation of the water path. To address this question, we replaced histidine with the aromatic amino acid phenylalanine. We found that AQP3(H154F), but not AQP3(H53F), had the same maximal water permeability as wild-type AQP3.
Human AQP3 was pH-sensitive in the range from pH 5.5 to 7.0. The imidazole group of histidine is the only amino acid side chain affected within this pH range. At pH 5.5, the group is protonated and positively charged, whereas at pH 7.0, it is electrically neutral. His 53 from the first extracellular loop and His 154 from the second extracellular loop were found to partic-ipate in the regulation of the water permeability by pH. Mutation of Ser 152 was also associated with a modification of AQP3 pH sensitivity. Serine is a common partner of histidine in a number of enzymes that have an active-site motif known as the catalytic triad (reviewed in Ref. 16). We suggest that, in AQP3, Ser 152 may play a role similar to that in the catalytic triad of enzymes. Tyrosine has been shown to closely interact with histidine in the photosystem of cyanobacteria (17). Cells expressing AQP3(Y124A) had significantly lower maximal water permeability than cells expressing wild-type AQP3. It is possible that the aromatic side chain of Tyr 124 may participate in stacking interactions with histidine or some other aromatic residue of AQP3.
The finding that AQP3 water permeability is pH-dependent is in line with previous observations by Zeuthen and Klaerke (11). Two other mammalian AQPs have been found to be pHsensitive: AQP0, which is expressed in the lens (26), and AQP6, which is expressed in intercalated cells of kidney collecting ducts (27). In contrast to AQP3, the maximal water permeability for AQP0 and AQP6 is low at pH 7.5 and significantly increases at acidic pH. A histidine residue in the first extracellular loop was identified as a determinant of the pH sensitivity of AQP0. Amino acid residues controlling the pH sensitivity of AQP6 were not identified. No attempt has been previously made to identify the molecular determinants of AQP3 pH sensitivity.
Alignment of the protein structures showed that all amino acid residues involved in the regulation of AQP3 by Ni 2ϩ or pH are absent in AQP4 and AQP5. Neither AQP4 nor AQP5 was Ni 2ϩ -or pH-sensitive. Nickel and acidic pH can therefore be used as tools to discriminate between the activities of AQP3 on one hand and AQP4 and AQP5 on the other.
This study was performed on a bronchial epithelial cell line derived from the human lung. Nickel is widely used in modern industry, and inhalation is the primary route of occupational nickel exposure. High doses of nickel have been suggested to predispose to asthma, lung fibrosis, and lung cancer (reviewed in Ref. 6). Could interaction of nickel with AQP3 contribute to one or more of these conditions? AQP3 is, like AQP4 and AQP5, abundantly expressed in the airway epithelium. AQP3-null mice have few signs of lung dysfunction except for a small but significant reduction of airway humidification (28,29). Mice lacking AQP4 and AQP5 also have few signs of lung dysfunction. This has raised the question of whether lung AQPs are important for airway fluid transport. However, there appears to be important species differences with regard to the lung expression of AQPs and particularly with regard to AQP3 expression. AQP3 is present in the apical membranes of bronchial epithelial cells in the human lung, but not in the rodent lung (8 -10). Hence, the extracellular loops of AQP3 will, in the human lung, face the airway lumen. Here, we found that Ni 2ϩ binds to the extracellular loops of human AQP3 and that AQP3 is sensitive to extracellular pH. Other investigators have shown that the epithelial sodium channel, which is essential for apical sodium entry into the airway epithelium, is also Ni 2ϩ -sensitive (24). Taken together, these observations are compatible with the concept that the initial stages in human lung disease caused by nickel are related to the effect of Ni 2ϩ on apical water and ion transporters. The interaction between nickel and pH becomes interesting in this context. The effect of Ni 2ϩ was more pronounced at acidic and neutral pH than at alkaline pH. Thus, it may be important to ensure that the airway liquid is alkaline in the situation of nickel exposure.
The pH dependence of human lung AQP3 may be of great clinical importance. The pH of airway surface liquid is ϳ6.8, and it responds rapidly to changes in systemic acid-base status a The P f was not different from the P f in untransfected cells. b The effect of Ni 2ϩ could not be measured since the P f was not different from the P f in untransfected cells.
c The effect of Ni 2ϩ was abolished. (30,31). In a recent study in which human airway surface epithelial cells were used, it was found that the addition of mast cells causes substantial increases in proton secretion (32).
In line with this, it was found that the pH of exhaled airway vapor condensate is significantly lower in patients with acute asthma than in healthy controls (33). In our study, AQP3 water permeability decreased sharply when the pH fell below 7.0. Interestingly, there is evidence that airway surface liquid pH may be dynamically regulated and that activation of the adenylate cyclase-cAMP pathway results in alkalinization, whereas treatment with histamine and ATP results in acidification of airway surface liquid (32,34,35). AQP3 is expressed in several other organs, including the kidney, salivary and lacrimal glands, and skin (9, 36 -38). The most prominent phenotype in AQP3 knockout mice is a decrease in urinary concentrating capacity, resulting in severe nephrogenic diabetes insipidus (39). Hence, it will be important to explore the relationship between renal medullopapillary pH and urinary concentrating capacity. Nickel is one of the most common causes of contact allergic dermatitis (6,40). Nickel inhibition of AQP3 may be a contributor to this condition. AQP3 and another member of the aquaporin family, AQP7, have recently been found in dendritic cells, which play an important role in the innate immune system (41,42). Mercury, which is well known blocker of AQPs (2-4) and a cause of contact dermatitis (6,43), inhibits the regulatory cell volume decrease after macropinocytosis in dendritic cells (42). Given the finding that nickel blocks the water permeability of AQP3, it is tempting to speculate that perturbation of volume regulation of dendritic cells is an important contributor to nickel dermatitis and to the inflammatory component of nickel pneumoconiosis. FIG. 5. Mutations affecting the Ni 2؉ sensitivity of human AQP3. Data are presented as differences between the P f in transfected cells and the corresponding untransfected cells from the same coverslips (n ϭ 13-28 cells). a and b, AQP3(W231A) was retained in the endoplasmic reticulum of the cells. AQP3(D219A) had a similar distribution. c and d, in contrast to cells expressing wild-type AQP3 (WT), the P f in cells expressing AQP3(W128A), AQP3(H241A), or AQP3(S152A) was not decreased by 1 mM NiCl 2 .
FIG. 6. Mutations affecting the pH sensitivity of human AQP3. Data are presented as differences between the P f in transfected cells and the corresponding untransfected cells from the same coverslips (n ϭ 12-74 cells). a, the P f in cells expressing AQP3(H53A) or AQP3(H53F) was significantly lower than that in cells expressing wild-type AQP3 (WT). The range of pH sensitivity was shifted to higher pH values in cells expressing AQP3 (H53A). b, the mutation H154A rendered human AQP3 water-impermeable. The maximal P f in cells expressing AQP3(H154F) was similar to that in cells expressing wild-type AQP3, but was achieved at higher pH values. c, the mutation Y124A decreased the maximal water permeability of AQP3 and shifted the pH sensitivity curve to higher pH values. d, the water permeability in cells expressing AQP3(S152A) was less responsive to acidic pH. At pH 5.5, the P f in transfected cells was 3.9-fold higher than that in untransfected cells, whereas in cells expressing wild-type AQP3, the water channel-mediated P f was completely inhibited.