We developed a new apparatus, the virtual gland (VG), for measuring the rate of fluid secretion (Jv), its composition, and the transepithelial potential (TEP) in cultured epithelial cells under open circuit. The VG creates a 10-μl chamber above the apical surface of epithelial cells on a Costar filter with a small hole leading to an oil-filled reservoir. After the chamber is primed with a fluid of choice, secreted fluid is forced through the hole into the oil, where it forms a bubble that is monitored optically to determine Jv and collected for analysis. Calu-3 cells were mounted in the VG with a basolateral bath consisting of Krebs-Ringer bicarbonate buffer at 37°C. Basal Jv was 2.7 ± 0.1 μl·cm−2·h−1 (n = 42), and TEP was −9.2 ± 0.6 mV (n = 33); both measures were reduced to zero by ouabain (n = 6). Jv and TEP were stimulated 64 and 59%, respectively, by 5 μM forskolin (n = 10), 173 and 101% by 1 mM 1-ethyl-2-benzimidazolinone (n = 5), 213 and 122% by 333 nM thapsigargin (n = 5), and 520 and 240% by forskolin + thapsigargin (n = 6). Basal Jv and TEP were inhibited to 82 and 63%, respectively, with 10 μM bumetanide (n = 5), 71 and 82% with 100 μM acetazolamide (n = 5), and 47 and 56% with 600 μM glibenclamide (n = 4). Basal Jv and TEP were 52 and 89% of control values, respectively, after HCO3− replacement with HEPES (n = 16). The net HCO3− concentration of the secreted fluid was close to that of the bath (25 mM), except when stimulated with forskolin or VIP, when it increased (∼80 mM). These results validate the use of the VG apparatus and provide the first direct measures of Jv in Calu-3 cells.
- cystic fibrosis
- submucosal gland
- serous cell
patients with the genetic disease cystic fibrosis (CF) typically die from unremitting lung infections. Susceptibility to these infections is generally considered to arise from dysfunction of electrolyte and fluid transport across airway epithelia. Transport dysfunction is caused by loss of function of cystic fibrosis transmembrane conductance regulator (CFTR), an anion channel found primarily in the apical membranes of epithelial cells. Ciliated surface epithelia and submucosal gland serous cells express CFTR, and both of these cell types presumably contribute to the etiology of CF airway disease, but glands are the major secretory organs of healthy lungs. When provoked, they produce copious amounts of antibiotic-rich mucus that traps pathogens and inhibits their growth while they are swept from the airways (28). Mucus clearance normally keeps the airways sterile, and it has been proposed that altered gland function in CF airways impairs the mucus shield (1, 15–17, 21, 23, 26, 49, 52, 54).
Airway glands can be studied in isolation, and a great deal has been learned about their regulation (42). Gland mucus secretion has been quantified extensively (1, 6, 15–17, 23, 25, 26, 38, 47–50), but because glands are difficult to study with electrophysiological methods, electrophysiological study of cultured monolayers of gland cells and cell lines has mainly been used (31, 55–58). However, except in rare cases (22, 45), fluid transport by cell sheets has not been quantified. Therefore, we sought a simple method that would help bridge the gap between studies of native glands and cultured cell sheets.
We have developed a novel method for measuring fluid secretion across epithelial cell sheets and have used it to quantify the amount and selected features of fluid secretion by Calu-3 cells, a widely studied surrogate for airway submucosal gland serous cells (39). Our method differs in several respects from prior studies of the rate of fluid secretion (Jv) in airway epithelia (22, 45). In the best of these, Miller et al. (34) used an elegant modification of the capacitance probe technique (51) to measure fluid absorption across the retinal pigment epithelium with an accuracy of 0.5–1.0 nl/min (34); subsequently, their method was used to clarify important points of transport in several systems (5, 10, 37), including gland cells from normal subjects and CF patients (22). In the capacitance probe apparatus, the epithelial cell sheet separates two large volumes (∼12 ml) of fluid. The chambers are sealed except for a single thin column of fluid that moves relative to a probe that detects the capacitance between it and the surface of the fluid. The advantages of the capacitance probe are its sensitivity and ability to measure fluid flow in either direction. A disadvantage is that the large volumes of fluid used essentially clamp the fluid composition, so that, unlike the natural situation, the cells do not determine the composition of their apical fluid. The large volumes also mean that the composition of the secreted fluid is unknown.
The apparatus introduced here, which we term a “virtual gland” (VG), is advantageous for studies of secretory epithelia, because it allows the formation and collection of the secreted fluid. The VG has about the same sensitivity as the capacitance probe and is simpler to use, and because the bath is an open system, secretion rates are not altered by variations in the height, volume, or temperature-related volume changes of the bath. In addition to its ability to measure the composition of the secreted fluid, this system has the additional advantage that the cells can respond to that fluid, as we expect they do in real glands. The ionic concentration of the secreted fluid will help set the apical transmembrane potential, and secreted mediators, for example, ATP (8), can act back on the cells to influence the secretory process.
In the present studies, we introduce the VG method and use it to provide direct evidence that Calu-3 cells secrete fluid in open-circuit conditions. In Ussing chamber experiments, they produce a basal short-circuit current (Isc) of 20–40 μA/cm2 that is mainly the result of HCO3− secretion. Stimulation with forskolin causes a further increase in HCO3− secretion, whereas stimulation with agents that hyperpolarize the cell, such as thapsigargin and 1-ethyl-2-benzimidazolinone (1-EBIO), recruit mainly Cl−-mediated increases in Isc (4, 31, 36, 39).
We determined the magnitude and properties of Calu-3 fluid secretion in response to various mediators and inhibitors and compared the results with predictions made from Isc studies (4, 11–14, 18, 19, 27, 30–32, 35, 36, 39, 40, 43, 44, 53) and with secretion by native airway submucosal glands (21, 23–26, 54).
The Calu-3 cell line was obtained frozen from the American Type Culture Collection (Rockville, MD). After they were thawed, the cells were grown at 37°C in T25 tissue culture flasks (Costar, Pleasanton, CA) containing 1:1 Dulbecco's modified Eagle's medium-Ham's F-12 plus 15% fetal bovine serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM glutamine in an atmosphere of 5% CO2-95% O2. Cells were passaged with 1:8 dilution and plated at 5 × 105 cells/cm2 onto Snapwell filters (12 mm diameter, 0.4 μm pore size, 1 cm2 growing area; Costar, Cambridge, MA) that had been coated with human placental collagen (Sigma). Cells were grown at the air-liquid interface for ≥14 days before use. In the present study, the mean age of the cells was 21 ± 1 days (n = 70).
The prototype VG is shown in Fig. 1. A Costar Snapwell filter containing confluent cells is assembled as shown so that a 1.7-mm-thick plastic barrier creates a small-volume apical chamber above the apical surface of the cells. The barrier separates a thin film (∼90 μm deep, ∼10 μl) of apical fluid from a water-saturated mineral oil layer (500 μl). A 0.6-mm-diameter hole in the barrier serves as a virtual duct to convey the apical fluid into the oil-filled collection chamber, where it forms a spherical bubble with a diameter that can be measured optically at regular intervals and converted to a volume; the change in volume over time provides a direct measure of Jv (see below). A second smaller hole serves as a voltage port (see below).
Before assembly, the apical surface of the cells was rinsed twice with Krebs-Ringer bicarbonate buffer (KRB) without glucose and then twice with the same solution containing 10 μM blue dextran. The filter cup and the collection chamber were then filled with the blue dextran solution, and the collection chamber was pressed gently but firmly into the Snapwell filter cup while the filter was supported against a flat, KRB-covered surface to prevent it from being distended by the pressure of assembly. Excess apical fluid was expressed through the hole into the collection chamber; the purpose of having fluid of the same composition in that chamber at this stage is to prevent the formation of bubbles in the apical chamber, which can otherwise occur during assembly. Because of variations in dimensions of the filters, cell layers, and assembly procedures, assembled VGs have apical chambers of variable volumes, which must be measured in each experiment.
To estimate the volume of the apical chamber, the bubble of apical fluid was collected at intervals during secretion and the blue dextran concentration ([BD]) was determined colorimetrically. The apical chamber volume is then given by where VA is volume of the apical chamber, VSF is volume of secreted fluid, [BD]SF is [BD] in the collected fluid bubble, and [BD]initial is starting [BD] (initially 10 μM, and then reset after each measurement). In the present study, the estimated mean volume of the apical chamber was 10.0 ± 1.0 μl (n = 60).
The assembled Snapwell filter and collection chamber were placed in a fabricated plastic bath holder so that the cell layer was positioned just at the surface of the bath KRB solution (volume ∼5 ml) to eliminate any significant positive or negative hydrostatic pressure. The bath KRB was continuously gassed with 5% CO2-95% O2 and maintained at 37°C with a temperature controller (model TS-4, Sensortek) that was coupled to the bath chamber by an aluminum jacket machined to fit the chamber. The KRB composition was (in mM) 115 mM NaCl, 2.4 mM K2HPO4, 0.4 mM KH2PO4, 25 mM NaHCO3, 1.2 mM CaCl2, and 2 mM glucose (pH 7.4), with osmolarity adjusted to 290–295 mosM. For HCO3−-free experiments, all HCO3− in the KRB was replaced with 25 mM HEPES or 1 mM HEPES + 24 mM NaCl that had been pregassed with humidified 100% O2. The HEPES solutions were adjusted to pH 7.4 after they were gassed with O2, and all HEPES experiments were performed with continuous O2 gassing.
Measuring Jv and collecting the secreted fluid.
Dynamic measurements of Jv were obtained by optical monitoring of the growth of the fluid bubble in the oil (Fig. 1, E and F, and Fig. 2). We used a Wild microscope fitted with a digital video camera (Logitech, Fremont, CA) and captured images automatically at fixed intervals (typically 5 min) using software supplied with the camera. An optical grid (0.5 mm; Fig. 1E) was present throughout the procedure for calibration. The diameter or areas of digital images of bubbles were measured with a commercial version of NIH Image software (Scion, Frederick, MD), and the formula for a sphere was used to convert the measures to volume. Jv was determined for each 5-min period as follows: [(volume of bubble n + 1) − (volume of bubble n)]/time. For presentation of “continuous” data (see Fig. 5), two 5-min periods were averaged. For summary data in response to agonists or inhibitors, a 20-min period of secretion starting with the peak response was used. In rare cases when the response continued to increase (agonists) or decrease (inhibitors) until the end of the 30- to 40-min epoch, the last 20 min of the epoch were used.
All measured bubbles were subsequently collected between oil blocks in constant-bore capillary tubing (usually 10 μl volume; Drummond Scientific, Broomall, PA), and the linear dimension of the collected fluid column was again measured at ×60 using a Wild binocular microscope. The volume is then given by the formula for a cylinder or merely as a fraction of the preset total tube volume. The two measurement procedures showed good agreement for bubble volumes under 3.0 μl; the mean collected volume was 96.4 ± 0.4% of the digitally calculated volume (195 bubbles, 60 experiments). As bubbles increase in size, the spherical shape becomes oblate, and the assumption of spherical shape leads the volume to be overestimated. The shapes of the bubbles were directly observed during calibration procedures by imaging them from the top and from the side via a prism (Melles Griot, Rochester, NY). The volumes used for all calculations of Jv were taken from bubbles under 3 μl and were checked and, if necessary, corrected from measurements in constant-bore capillaries. Collected fluid was stored frozen for subsequent physical and biochemical analyses.
Measurement of TEP difference.
TEP was measured via a voltage port consisting of a 0.36-mm-diameter hole filled with a wick of conducting material (we used a buffer-saturated fragment of absorbent paper) that established electrical continuity with the apical fluid and allowed the TEP to be measured with a microelectrode while preventing bulk flow of fluid (Fig. 1E). The microelectrode was pulled with a vertical puller, and the tip was broken to give low impedance when filled with bath solution. It was coupled via silver-silver chloride wire to a high-impedance microelectrode amplifier (Getting Instruments). During the course of these experiments, we discovered that Calu-3 cells are very sensitive to silver chloride, showing a rapid drop in TEP and Jv in response to a ground wire in the small-volume bath (for similar findings with smooth muscle see Ref. 20Fig. 1G). A comparison of TEP measured at the voltage port and in the actual fluid bubble showed that they were virtually identical.
Measurement of pH.
Collected fluid was placed in micro chambers designed to accommodate the tip of a mini-combination pH electrode (WPI, Sarasota, FL), and pH was measured under conditions of constant temperature and nominally 100% humidity (no evaporation was detected during the course of the measurements). Because the apical chamber must be primed with an added fluid, early collections are mixtures of the secreted and the priming fluid, and appropriate corrections must be applied. We start by knowing the volume of the apical chamber and its starting pH; we then measure the volume of secreted fluid and the pH of the collected fluid. The pH values are converted to HCO3− concentrations ([HCO3−]) and then to nanomoles of HCO3− in the apical and collection chamber (i.e., the fluid bubble). The increase in nanomoles of HCO3− divided by the secreted volume (collection chamber volume only) is the average [HCO3−] secreted over the sample interval.
Forskolin, thapsigargin, acetazolamide, bumetanide, and ouabain were obtained from Sigma. 1-EBIO was purchased from Aldrich Chemical (Milwaukee, WI). Ouabain was dissolved in deionized water at a stock concentration of 10 mM. To obtain a 10 mM stock solution, bumetanide was dissolved in 200 mM sodium hydroxide; in spite of this high concentration of NaOH in the stock, the bath pH was unchanged by addition of the diluted aliquot. All other drugs were dissolved in DMSO at the following concentrations: forskolin, 10 mM or 50 mM; thapsigargin, 1 mM; acetazolamide, 200 mM; 1-EBIO, 2 M; and glibenclamide, 300 mM. In control experiments, we observed that DMSO levels >0.15% depressed basal Jv, so we kept the final concentration of DMSO at ≤0.15% where possible.
Values are means ± SE; n indicates the number of experiments. Statistical difference was determined by Student's t-test. P < 0.05 was considered statistically significant.
Filters mounted in the VG showed high and variable Jv immediately after they were mounted but declined to relatively stable values within 30–60 min. Therefore, we ignored the first 60 min of data and followed basal secretion for up to 6 additional hours. Summary data for four long-term basal secretion experiments are shown in Fig. 3, and mean basal secretion rates for a much larger sample taken 60–90 min after the filters were mounted are shown in Fig. 5. For all data, the average basal secretion rate was 2.7 ± 0.1 μl·cm−2·h−1 (n = 42) and the average basal TEP was −9.2 ± 0.6 mV (n = 33).
Several control experiments were carried out to verify that the observed fluid accumulation represents active secretion. When filters without cells or filters with killed cells were placed in the VG apparatus, no fluid accumulation was observed (data not shown). Ouabain blocks Na+-K+-ATPase and eliminates all Isc in Ussing chamber experiments with Calu-3 cells (44). When applied to Calu-3 cells in the VG, 10 μM ouabain caused a slow decline in TEP (time constant = 28 ± 1 min, n = 6) and Jv (time constant = 39 ± 3 min, n = 6). After 90 min, both values reached zero.
Responses to agonists.
An abundant literature describes Isc responses of Calu-3 cells to forskolin, thapsigargin, and 1-EBIO. Forskolin elevates intracellular cAMP concentration and produces variable increases in Isc that are inversely proportional to the level of basal Isc. Thapsigargin elevates intracellular Ca2+ concentration by inhibiting the sarcoplasmic reticulum Ca2+-ATPase (46), and 1-EBIO activates CFTR and basolateral K+ channels (2, 3). All these agonists stimulated increases in fluid secretion from Calu-3 cells in the VG.
Forskolin (5 μM) increased Jv from 2.4 ± 0.3 to 4.0 ± 0.4 μl·cm−2·h−1 (64 ± 16%, n = 10) and increased TEP from −7.6 ± 0.8 to −12.1 ± 1.2 mV (59 ± 16%, n = 10; Fig. 4). VIP (1 μM) increased Jv from 3.1 ± 0.4 to 5.4 ± 0.8 μl·cm−2·h−1 (71 ± 25%, n = 4). Thapsigargin (333 nM) increased Jv from 2.0 ± 0.2 to 6.4 ± 0.9 μl·cm−2·h−1 (213 ± 44%, n = 5) and increased the average TEP from −8.2 ± 1.2 to −18.2 ± 3.3 mV (122 ± 40%, n = 5; Figs. 4 and 5). In some experiments, we noted oscillations in the TEP as has been reported for the Isc after stimulation with thapsigargin (Fig. 1G), but the sampling limits for determining Jv did not allow us to determine whether it showed similar oscillations. When forskolin and thapsigargin were used in combination, the response was synergistic, increasing Jv from 2.0 ± 0.5 to 12.2 ± 1.0 μl·cm−2·h−1 (520 ± 50% of basal Jv) and TEP from −8.2 ± 1 .1 to −28.0 ± 2.2 mV (240 ± 30%, n = 6; Figs. 4 and 5). 1-EBIO stimulates large Isc increases from Calu-3 cells in Ussing chamber experiments. For Calu-3 cells mounted in the VG, 1 mM 1-EBIO increased fluid secretion from 3.2 ± 0.4 to 8.6 ± 1.0 μl·cm−2·h−1 (173 ± 33%, n = 5) and TEP from −10.1 ± 1.1 to −20.3 ± 3.5 mV (101 ± 34%, n = 5; Fig. 4).
Responses to inhibitors.
Effects of five inhibitory conditions on basal Jv and TEP are summarized in Fig. 6.
Bumetanide inhibits the Na+-K+-2Cl− cotransporter, which is responsible for transporting Cl− into the cell across the basolateral membrane. To estimate the contribution of Na+-K+-2Cl− cotransporter-dependent Cl− secretion to Jv and TEP, we tested the effects of 10 μM bumetanide on basal and stimulated secretion. Bumetanide reduced basal Jv from 2.5 ± 0.3 to 2.1 ± 0.3 μl·cm−2·h−1 (18 ± 11%) and TEP from −9.2 ± 0.6 to −5.8 ± 1.1 mV (37 ± 11%, n = 5; Fig. 6). The response to forskolin was not affected by bumetanide, but the response to thapsigargin was eliminated: addition of thapsigargin after bumetanide failed to increase Jv (from 2.1 ± 0.3 to 2.2 ± 0.5 μl·cm−2·h−1, 3 ± 22%) or TEP (from −5.8 ± 1.1 to −6.9 ± 1.7 mV, 20 ± 30%, n = 5). When bumetanide was applied after thapsigargin stimulation, it inhibited ∼100% of the stimulated Jv and TEP (n = 5; Fig. 5). When bumetanide followed 1-EBIO stimulation, it had differential effects on Jv and TEP. Bumetanide inhibited most, but not all, of the ΔJv, leaving Jv ∼35% higher than basal values. In contrast, TEP was inhibited to a level 65% below the basal value (n = 5; data not shown). We have no explanation for this unusual effect. Because of the large response to forskolin + thapsigargin, we were interested to see the effect of bumetanide on these agents used in combination. Bumetanide reduced Jv from 11.6 ± 1.1 to 6.8 ± 0.7 μl·cm−2·h−1 (−42 ± 6%, n = 5) and TEP from 21.3 ± 4.1 to 9.7 ± 1.7 mV (−55 ± 8%, n = 5). We noted that the Jv remaining after bumetanide was significantly (P < 0.001) greater than the basal value of 2.0 ± 0.5 μl·cm−2·h−1, suggesting that, in the presence of forskolin, thapsigargin can slightly increase the secretion of a bumetanide-insensitive component, which is most likely HCO3−.
Acetazolamide, which inhibits carbonic anhydrase, was previously shown to reduce the basal Isc of Calu-3 cells by about one-third (44). In the VG, 100 μM acetazolamide reduced basal Jv from 2.3 ± 0.2 to 1.6 ± 0.3 μl·cm−2·h−1 (29 ± 11%) and TEP from −8.2 ± 1.1 to −6.7 ± 1.1 mV (18 ± 14%, n = 5; Fig. 6). Stimulation by 5 μM forskolin was blocked after acetazolamide; neither Jv nor TEP showed a significant increase: ΔJv = 10 ± 20% and ΔTEP = 10 ± 10% (n = 5, not significant).
HEPES (HCO3−-free condition).
Filters were mounted in HCO3−-free medium (bath and apical fluid, with 1 or 25 mM HEPES to maintain pH) and gassed with O2. In this condition, we measured basal secretion 30–60 min after the filters were mounted. The average basal Jv in HCO3−-free medium (bath and apical fluid) was 1.4 ± 0.2 μl·cm−2·h−1 (n = 16) vs. 2.7 ± 0.1 μl·cm−2·h−1 in Krebs solution (n = 42), a reduction of 50% (P < 0.005), but TEP was reduced by only 12% [−8.1 ± 0.8 (n = 16) vs. −9.2 ± 0.6 mV (n = 33), not significant; Fig. 6]. In the absence of HCO3−, forskolin was without effect and did not alter the gradual decline in Jv or TEP (n = 8); this finding is in agreement with observations of Isc in Ussing chambers (4). The subsequent addition of thapsigargin caused large increases in Jv (from 0.7 ± 0.2 to 5.1 ± 1.7 μl·cm−2·h−1, 598 ± 228%, n = 6) and TEP (from −5.4 ± 1.5 to −21.9 ± 5.3 mV, 303 ± 97%, n = 6), but these increases were transient and were followed by rapid declines to baseline values (Fig. 7).
Bumetanide plus HCO3−-free conditions.
Addition of bumetanide to Calu-3 cells mounted in zero HCO3− reduced the basal secretion rate from 1.2 ± 0.3 to 0 μl·cm−2·h−1 within 30 min (n = 4), but TEP was only reduced from −7.7 ± 0.9 to −4.8 ± 1.1 mV (37 ± 14%, n = 5; Figs. 6 and 8). Addition of ouabain reduced TEP to 0 mV within 60 additional min.
Glibenclamide is a nonspecific inhibitor of CFTR (41). Addition of 600 μM glibenclamide basolaterally reduced basal Jv and TEP by 49 and 46%, to 1.8 ± 0.3 μl·cm−2·h−1 and −6.8 ± 1.9 mV (n = 4), respectively. Addition of forskolin after glibenclamide produced no increase in Jv or TEP (n = 3), but subsequent addition of thapsigargin caused a 98% increase in Jv to 2.7 ± 0.5 μl·cm−2·h−1 (n = 4). However, this occurred in the presence of a decrease in TEP (mean change of −21% to −4.5 ± 0.5 mV). Jv changes in response to thapsigargin + forskolin were highly variable in the presence of glibenclamide (0–284%), and the maximum Jv was only 3.6 μl·cm−2·h−1 vs. a mean Jv of ∼12 μl·cm−2·h−1 in the absence of glibenclamide. In three experiments, glibenclamide was added after forskolin, and the mean result was a reduction of Jv to basal levels. For maximum effectiveness, glibenclamide is usually added bilaterally; in these experiments, it was only added basolaterally.
Relation between mean Jv and mean TEP in the VG.
The TEP difference ranged from 0 to −38 mV and showed an approximately monotonic relation to Jv, with each 2.8 mV of TEP equating to a Jv of ∼1 μl·cm−2·h−1 (Fig. 9). If secretion is purely isotonic (150 mM), 1 μl·cm−2·h−1 should correspond to ∼4 μA, which would give a mean resistance for the filters of 700 Ω·cm2.
The peak Jv in response to every agonist was always delayed relative to the peak TEP, and decreases of TEP for every inhibitor also always preceded a decrease in Jv, indicating some physical capacitance in the system. Inspection of many records indicated a mean delay of ∼10 min between TEP and Jv; so the above correlations were always made with that time correction. If all measures are taken together, it appeared that fluid secretion by Calu-3 cells always required an electrogenic component, however little it might be. On the other hand, as shown in Fig. 8, it was possible to maintain a TEP in the absence of active secretion. We never saw fluid absorption with Calu-3 cells, although absorption was readily apparent with other cell types and could be induced in Calu-3 cells by making the bath hypertonic.
Comparison of Calu-3 responses in the VG with Calu-3 Isc responses and native gland secretion.
A comparison of responses to agonists for Calu-3 cells in the VG vs. the Ussing chamber is shown in Table 1, which also includes the response for these same agonists in native glands. Agonists and antagonists can only be added basolaterally in the VG and for real glands. Jv and Isc responses were qualitatively similar; the main difference compared with glands is that thapsigargin had little effect on native glands, whereas carbachol had little effect on Calu-3 cells. If Calu-3 cell responses to thapsigargin are compared with gland responses to carbachol, responses to carbachol and forskolin are completely occlusive in glands, whereas they are additive and, possibly, synergistic in Calu-3 cells. We hypothesize that activation of muscarinic receptors in glands cannot be mimicked merely by elevating intracellular Ca2+ concentration.
[HCO3−] in secreted fluid.
An advantage of the VG system over prior methods is that the secreted fluid can be collected and assayed for any desired property. However, because the apical chamber must be primed with an added fluid, early collections are mixtures of the secreted and the priming fluid, and appropriate corrections must be applied (see methods). We measured the pH of collected fluid with a miniature pH electrode and used the Henderson-Hasselbalch equation to convert the values to [HCO3−] (Fig. 10). With a bath pH of 7.4 (25 mM HCO3−), the mean [HCO3−] of basal secretions was 36 ± 9 mM (n = 12). Forskolin or VIP stimulation increased [HCO3−] [to 74 ± 21 mM (n = 5) and 92 ± 39 mM (n = 4), respectively], whereas thapsigargin stimulation decreased [HCO3−] to 17 ± 4 mM (n = 6), and thapsigargin + forskolin produced an [HCO3−] of 23 ± 4 mM (n = 4). According to these data, the secreted fluid was similar to bath pH, except when stimulated with forskolin or VIP, when it became markedly more basic.
The levels of basal secretion observed in the VG varied considerably. For the experiments reported in Fig. 4, mean basal Jv was slightly >2 μl·cm−2·h−1, and direct, continuous measures of pH taken with pH-sensitive microelectrodes during secretion indicated that the basal secretions were more acidic than the bath (data not shown). In contrast, pH of the collected basal secretions from earlier experiments indicated a more basic secretion. When the two sets of data were compared, it became apparent that basal Jv was much higher in the earlier experiments than in recent experiments. Therefore, we compared the calculated [HCO3−] of basally secreted fluid as a function of Jv. As shown in Fig. 11, the function was positive.
We have introduced a novel method to study fluid secretion by epithelial cell sheets in open-circuit conditions and applied it to fluid secretion by the Calu-3 cell line.
Overview of the VG method.
The VG 1) provides a direct measure of fluid secretion, 2) operates in open-circuit conditions, 3) allows the cells to form their own apical fluid, 4) allows the cells to respond to the composition of that fluid, 5) allows the fluid to be collected for analysis, and 6) allows the TEP difference to be measured.
Open-circuit studies are important adjuncts to Isc studies, because ion flows can differ markedly in the two circumstances (10). In addition, the VG method can, in principle, identify electrically silent mechanisms of fluid transport that are missed in Isc studies. The apical fluid compartment seen by the cells has a relatively small volume. Therefore, the cells can create a distinctive apical fluid similar to that they would experience in vivo, which cannot happen in the Ussing chamber or capacitance probe methods, where the fluid composition is clamped by the large volumes. Because the apical fluid in the VG can approximate the composition of native fluid, the cellular ion transport mechanisms may be influenced in a similar manner by the altered ion gradients that develop, as well as by secreted mediators such as ATP (8, 9) and the myriad proteins released by Calu-3 cells (33, 59). For all these reasons, the VG method can provide an essential link between Ussing chamber experiments and studies of fluid transport by intact glands.
Limitations of the VG method.
Limitations of the VG method are as follows: 1) In its present form, the VG does not measure epithelial resistance, which weakens inferences about possible contributions of electrically silent transport mechanisms to fluid secretion. 2) The VG would also be improved by reducing the volume of the apical chamber from 10 to 2 μl, which would correspond to a chamber height of ∼18 μm. This was achieved in some experiments but at increased risk of cell damage. 3) The apical chamber is relatively inaccessible, so in this study all reagents were added only to the basolateral side. 4) The secretion of fluid as a bubble into an oil layer eliminates evaporation and makes possible a sensitive optical assay of Jv, but lipophilic components of the fluid may be lost. 5) The present system requires the development of skills for assembly of the VG and for collection of the secreted fluid. All these limitations can be addressed with future development of the system. With these advantages and disadvantages in mind, what have we learned about Calu-3 cells from this initial VG study?
Basal secretion by Calu-3 cells in the VG.
Calu-3 cells are widely used as models for airway gland serous cells and are interesting, because anion transport in the Ussing chamber is rich in HCO3− rich or Cl−, depending on the mode of stimulation (4). In Ussing chamber experiments, Calu-3 cells show various levels of basal Isc, which is characterized by relative insensitivity to bumetanide (36). Calu-3 cells in the VG always displayed a resting Jv that averaged 2.7 μl·cm−2·h−1. The Jv varied greatly among different filters (Fig. 11), but for any given filter, Jv was stable for many hours after an initial brief period of higher secretion that probably was induced by aspects of the mounting procedure. Bumetanide reduced basal Jv by only 18%, whereas replacement of HCO3− with HEPES buffer reduced basal Jv by 52%, and the subsequent addition of bumetanide reduced Jv to zero. We interpret these results to mean that basal Jv is mediated by Cl− and HCO3− secretion, and when transport of one anion is inhibited, the other can partially compensate. The [HCO3−] of basally secreted fluid was positively related to Jv and, at lower Jv, could actually be less than the bath concentration. This seems inconsistent with other evidence for high levels of HCO3− secretion, and we have evidence from pH-stat experiments that acid secretion is also occurring (29).
Forskolin was used as surrogate for the natural transmitter VIP (23), which was only used in a few experiments. Forskolin increased Jv to 4.0 μl·cm−2·h−1, and this increase was completely eliminated by prior treatment with acetazolamide or replacement of HCO3− with HEPES, indicating that the stimulation of Jv by forskolin was entirely mediated by increased HCO3− transport (4) and that, unlike basal Jv, Cl− secretion could not substitute for the inhibited HCO3− transport. The estimated [HCO3−] of forskolin- or VIP-stimulated fluid was ∼80 mM.
Thapsigargin was used as a surrogate for the natural transmitter ACh, because ACh or carbachol causes only transient increases in Isc in Calu-3 cells (39). Thapsigargin increased Jv to 6.4 μl·cm−2·h−1, and this increase was completely eliminated by prior treatment with bumetanide, consistent with the increased Jv being mediated by Cl− secretion. However, the pH of fluid stimulated by thapsigargin was 7.23, which suggests that HCO3− secretion was not eliminated but simply became a smaller fraction of anion secretion.
Combined Jv response to forskolin + thapsigargin.
The largest increases in Jv were produced by combining forskolin and thapsigargin, which gave a mean Jv of 12.2 μl·cm−2·h−1. This combined response was inhibited 42% by bumetanide and 64% by HCO3− replacement with HEPES, and the combination entirely eliminated the response. The pH of fluid secreted in response to these two agonists was 7.4 (23 mM HCO3−). The interpretation of these results is not direct because of the possibility that inhibition of one anion transport pathway leads to compensatory increases in the other, but given the pH measure, it appears that this condition also represents a combination of HCO3−- and Cl−-mediated fluid secretion.
What is the status of CFTR in these conditions?
Ussing chamber (36) and patch-clamp (7) data suggest that CFTR is the only apical anion channel in Calu-3 cells. The presence of basal secretion and the ability of thapsigargin to stimulate secretion are consistent with CFTR being active in the basal state. However, because forskolin alone stimulated secretion and greatly augmented thapsigargin-stimulated secretion, one of two things must be true: 1) CFTR single open channel probability is rate limiting for secretion and is not fully active at rest, or 2) forskolin must stimulate other transport pathways that contribute to Jv.
pH ([HCO3−]) of secreted fluid.
To summarize the results across all conditions (Fig. 10), the estimated [HCO3−] in the secreted fluid was higher than that in the bath for forskolin or VIP stimulation but was close to bath values during basal secretion and secretion stimulated by thapsigargin alone or in combination with forskolin. These results vary from expectations that forskolin- or VIP-mediated secretion will be mediated by nearly pure HCO3− secretion and thapsigargin-stimulated secretion by nearly pure Cl− secretion (4). We hypothesize that the lower-than-predicted levels of HCO3− detected in basal and forskolin-stimulated conditions occur because of parallel secretion of acid that neutralizes some of the HCO3− (29). We further hypothesize that the higher-than-predicted levels of HCO3− observed during thapsigargin-mediated secretion occur because only a portion of intracellular [HCO3−] is derived from the voltage-sensitive Na+-HCO3− cotransporter (thapsigargin-induced hyperpolarization makes the transport of anions into the cell less favorable), with the remainder being generated from CO2 via acetazolamide-sensitive processes that are not voltage sensitive.
Comparison with glands.
The thickness of Calu-3 cell monolayers varies from 17 to 34 μm, with an average of ∼20 μm (39), so the volume of Calu-3 cells on a 1.1-cm2 filter is ∼2 μl. The mean secretion rates observed in this study ranged from 2.7 μl·cm−2·h−1 for basal secretion to ∼11 μl·cm−2·h−1 for secretion in response to forskolin + thapsigargin. These rates are equivalent to 120% and 550% of the estimated Calu-3 cell volume each hour. Direct correlations of human gland submucosal volumes and secretion rates for a selected sample of glands gave average values of 145 nl for gland volume and 150 nl/h for secretion rate (sustained response to carbachol stimulation), which is equivalent to ∼100% of the gland volume each hour (N. S. Joo, unpublished observations). Although there are many uncertainties in these measures (the gland volumes include lumens and nonsecreting ducts), the comparison suggests that Calu-3 cells are capable of exceeding native gland secretion rates.
Apical environment in glands and in the VG.
Human submucosal glands are multiply branched structures, with serous cells at the distal ends of the tubules. The VG is an attempt to reproduce an apical fluid environment that is more similar to that found in the lumens of serous cell tubules in native glands. How close are we?
Direct observations before stimulation, as well as images of sections of fixed tissues, indicate that gland tubule lumens may be closed (empty) or distended to varying degrees. In a typical tubule, the outer diameter is ∼60 μm, lumen diameter is ∼20 μm, and cell height is ∼20 μm (Fig. 12, A and B). One way to pose the geometric issues in glands and the VG is to consider apical fluid volume relative to cell volume or surface area. Use of the observed gland tubule dimensions to calculate volumes (considering tubules as cylinders) shows that the epithelial cell volume for a given length of tubule is nearly eight times the resting lumen volume (this ratio will increase at the ends of the tubules and, of course, approaches infinity as the lumen volume approaches zero). In our present experiments, the mean estimated apical volume in the VG experiments is 10 μl, whereas the cell volume is 2 μl, or 0.2 times the apical fluid volume (Fig. 12C). Thus, in glands, the ratio of cell volume to lumen volume is ≥40-fold greater than in the VG. With regard to cell surface-to-apical (or luminal) volume ratio, the cell surface area in the VG is ∼1 cm2 (108 μm2) and the apical volume is 10 μl (104 nl), giving a surface-to-volume ratio of 10,000 μm2/nl. In a gland with a 20-μm-diameter 100-μm-long lumen, the apical surface area is 6,283 μm2 and the lumen volume is 31,416 μm3 (0.031 nl), giving a surface-to-volume ratio of ∼200,000 μm2/nl. Thus the ratio of cell surface to lumen volume is ∼20 times greater in gland tubules than in the VG.
The conclusion from these considerations is that our apparatus, while achieving a 300-fold improvement over a 3-ml Ussing chamber, nevertheless causes a considerable dilution of the serous cell secretions compared with real glands.
The VG apparatus provides a new method for direct measurement of fluid secretion by secretory epithelial cell sheets under conditions that approximate natural conditions found in gland lumens. Initial experiments with the Calu-3 cell line confirm expectations that most fluid secretion is mediated by the electrogenic secretion of HCO3− and Cl−, with forskolin/VIP stimulating an HCO3−-rich fluid and thapsigargin, alone or in combination with forskolin, stimulating a Cl−-rich fluid (4). However, secretions never reached the extreme pH values expected from purely HCO3−- or purely Cl−-mediated secretion, and additional work is required to account for our observations. If Calu-3 cells accurately reflect the properties of gland serous cells, then the acidic pH of gland secretions under all conditions (21, 25) must arise from secondary modifications of the secreted fluid downstream from the serous acini.
This work was supported by National Institutes of Health Grants DK-51817 and HL-60288 and by the Cystic Fibrosis Foundation.
We thank Kimberly Winges and Dennis Lee for help with cell culture and Ramsey Asmar, Teresa Au, Anthony Ko, and Esteban Gomez for contributions to developmental work on the VG apparatus.
Present address of T. Irokawa: Dept. of Respiratory & Infectious Disease, Postgraduate Div., Tohoku University School of Medicine, 1-1 Seiryo-machi, Aoba-ku, Sendai 980-8574, Japan (E-mail: [email protected]).
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 © 2004 the American Physiological Society