Bicarbonate and Chloride Secretion in Calu-3 Human Airway Epithelial Cells

The inherited disease cystic fibrosis (CF) ¹ is characterized by secretion of a thick viscous mucus that plugs the submucosal glands and small airways. This leads to chronic airway infections and is caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) (Boat et al., 1989). The predominant site of CFTR expression in the human lung is the serous cells of the submucosal glands (Jacquot et al., 1993; Engelhardt et al., 1994). Serous cells account for 60% of the cellular volume of the submucosal gland in human airways (Basbaum et al., 1990). Stimulation of an isotonic fluid secretion from the serous cells contributes to the hydration of the secretions from the mucous cells, thereby forming the low viscosity mucus that lines the conducting airways. Serous cells are also a major source of antimicrobial enzymes and peptides that help maintain an aseptic environment in the lungs (Basbaum et al., 1990). Salt concentration can influence the activity of these antimicrobial agents and it was recently suggested that altered salt concentration in the airway surface fluid may contribute to chronic airway infection in CF (Smith et al., 1996). Thus, the serous cells make a significant contribution to the volume, composition, and consistency of the submucosal gland secretions and represent a potentially important target in CF therapy. These considerations indicate the importance in understanding the mechanisms of fluid and electrolyte transport by serous cells.

Shen et al. (1994) screened 12 cell lines derived from lung adenocarcinomas in an attempt to identify a cell line that displayed electrophysiological properties consistent with human airway serous cells. They identified the Calu-3 cell line as being serous cell in nature, forming a monolayer with a transepithelial resistance of ∼100 Ω · cm², expressing high levels of CFTR and responding to both cAMP- and Ca²⁺-mediated agonists with changes in net transepithelial ion transport as measured by short circuit current (I_sc) (Finkbeiner et al., 1993; Shen et al., 1994). Several studies have produced variable results in the basal and stimulated transport properties of the Calu-3 cells and the ionic basis of the responses to secretory agonists remains unsettled (Shen et al., 1994; Illek et al., 1997; Moon et al., 1997; Singh et al., 1997; Lee et al., 1998). In this report, we present studies with Calu-3 cells that displayed a low basal I_sc (13 μA cm⁻²) and robust sustained responses to secretory agonists enabling the measurement of isotopic fluxes. The results demonstrate that Calu-3 cells, when stimulated by forskolin, secrete HCO⁻₃ by a Cl ⁻-independent, Na⁺-dependent, 4,4′-dinitrostilben-2,2′-disulfonic acid (DNDS)–sensitive, electrogenic mechanism. Secondly, when stimulated by 1-ethyl-2 benzimidazolinone (1-EBIO), an activator of the basolateral membrane Ca²⁺-activated K⁺ channels (K_Ca) (Devor et al., 1996), HCO⁻₃ secretion is reduced and the Calu-3 cells secrete predominately Cl ⁻ by a bumetanide-sensitive, electrogenic mechanism.

Calu-3 cells were grown in Dulbecco's modified Eagle's medium and Ham's F-12 (1:1) supplemented with 15% fetal bovine serum and 2 mM glutamine. The cells were incubated in a humidified atmosphere containing 5% CO₂ at 37°C. For measurements of short-circuit current (I_sc), Calu-3 cells were seeded onto Costar Transwell cell culture inserts (0.33 cm²) or Snapwell inserts (1.1 cm²). Both the Transwell and Snapwell inserts were collagen-coated overnight with 0.01% human placenta collagen type VI (Sigma Chemical Co.). On day one, the medium bathing the apical surface was removed to establish an air interface. Apical medium was removed and the cells fed every 48 h. After ∼7–14 d, the cells formed a confluent monolayer that held back fluid, thus maintaining an apical air interface. Short circuit current measurements were performed after an additional 14–28 d in culture. Patch-clamp experiments were performed on single cells plated onto glass cover slips 18–48 h before use.

For measurements of I_sc, the bath solution contained (mM): 120 NaCl, 25 NaHCO₃, 3.3 KH₂PO₄, 0.8 K₂HPO₄, 1.2 MgCl₂, 1.2 CaCl₂, and 10 glucose. Mannitol was substituted for glucose in the mucosal solution to eliminate the contribution of Na⁺ glucose cotransport to I_SC as previously reported by Singh et al. (1997). The pH of this solution was 7.4 when gassed with a mixture of 95% O₂–5% CO₂ at 37°C. For the Cl⁻-free solution, equimolar Na-gluconate replaced NaCl, 1 mM Mg-gluconate replaced MgCl₂, and 4 mM Ca-gluconate replaced CaCl₂. Calcium was increased to 4 mM to compensate for the Ca²⁺ buffering capacity of the gluconate. The HCO⁻₃-free buffer consisted of (mM): 145 NaCl, 3.3 KH ₂PO₄, 0.8 K₂HPO₄ 1.2 MgCl₂, 1.2 CaCl₂, 10 HEPES, pH adjusted with NaOH, 10 glucose or mannitol and was gassed with air. For the Na⁺-free Cl⁻-free solution, equimolar N-methyl-d-glucamine–gluconate replaced NaCl, choline-HCO₃ replaced NaHCO₃, 1 mM Mg-gluconate replaced MgCl₂, and 4 mM Ca-gluconate replaced CaCl₂. This solution contained 10 μM atropine to block the cholinergic effect of choline (Muallem et al., 1988).

The effects of forskolin and 1-EBIO on apical membrane Cl⁻ currents (I_Cl) were assessed after permeabilization of the serosal membrane with nystatin (360 μg/ml), and the establishment of a mucosa-to-serosa Cl⁻ concentration gradient. Serosal NaCl was replaced by equimolar Na-gluconate and Ca²⁺ was increased to 4 mM with Ca-gluconate. Nystatin was added to the serosal membrane 15–30 min before the addition of drugs. Successful permeabilization of the basolateral membrane was based upon the recording of a current consistent with the mucosal-to-serosal flow of negative charge. The effect of 1-EBIO on basolateral membrane K⁺ currents (I_K) was assessed after permeabilization of the apical membrane with nystatin (180 μg/ml) for 15–30 min, and establishment of a mucosa-to-serosa K⁺ concentration gradient. For measurements of I_K, mucosal NaCl was replaced by equimolar K-gluconate, while serosal NaCl was substituted with equimolar Na-gluconate. Calcium and Mg²⁺ salts were replaced as above.

During inside-out patch-clamp recordings, the bath contained (mM): 145 K-gluconate, 5 KCl, 1 MgCl₂, 1 EGTA, 0.78 CaCl₂, (free Ca²⁺ = 400 nM), and 10 HEPES, pH adjusted to 7.2 with KOH. The pipette solution contained (mM): 140 K-gluconate, 5 KCl, 1 CaCl₂, 1 MgCl₂, and 10 HEPES, pH adjusted to 7.2 with KOH. For outside-out recordings, the bath contained 1 mM CaCl₂ in the absence of any added EGTA, while the pipette solution Ca²⁺ was buffered to 200 nM with EGTA (0.71 mM Ca²⁺, 1 mM EGTA).

Transwell inserts were mounted in an Ussing chamber (Jim's Instruments). Snapwell inserts were mounted in Ussing chambers (NaviCyte), and the monolayers were continuously short-circuited after fluid resistance compensation using automatic voltage clamps (558C-5; Iowa Bioengineering). Transepithelial resistance (R_T) was measured by open-circuiting the monolayer, or with a 2-mV bipolar pulse and the resistance calculated by Ohm's law. Forskolin, 1-EBIO, clotrimazole, 293B, and acetazolamide were added to both sides of the monolayers at the indicated concentrations. Bumetanide and charybdotoxin (CTX) were added only to the serosal bathing solution.

20 min after the Snapwell filters were mounted in Ussing chambers, isotopes (³⁶Cl, ²²Na, or ⁸⁶Rb) were added to the bath solution on one side of the monolayers. After an additional 20 min, by which time isotopic fluxes had reached a steady state, two 0.4-ml samples were taken from the unlabeled side and fresh unlabeled solution of equal volume was added. This time was considered time = 0 (T₀), and samples were taken thereafter at 15-min intervals for the next 75 min. When the effects of forskolin, 1-EBIO, or forskolin plus 1-EBIO were studied, the drugs were added to the serosal and mucosal sides at T₃₀ and fluxes before (T₀ − T₃₀) and 15 min after the drug additions (T₄₅ − T₇₅) were compared. Isotope activities were determined in a Packard liquid scintillation counter. All samples were weighed and these volumes were used to correct the chamber volume and to calculate the unidirectional ion fluxes using standard equations (Bridges et al., 1983). The net residual ion flux (J^R_net) was calculated from the difference in I _sc and the net fluxes of Cl⁻, JCl_net; Na⁺, J^Na_net; and Rb ⁻, J^Rb_net, where J ^R_net = I_sc − (J^Na_net + Ξ^Rb_net − Ξ^Cl_net ).

Single channel currents were recorded in the inside-out and outside-out patch-clamp recording configuration using a List EPC-7 amplifier (Medical Systems) and recorded on videotape for later analysis as described previously (Devor and Frizzell, 1993). Pipettes were fabricated from KG-12 glass (Willmad Glass Co.). All recordings were done at a holding voltage of −100 mV. The voltage is referenced to the extracellular compartment as the standard method for membrane potentials. Inward currents are defined as the movement of positive charge from the extracellular compartment to the intracellular compartment, and are presented as downward deflections from baseline in all recording configurations.

Single channel analysis was performed on records sampled after low-pass filtering at 400 Hz. Data records for all experimental conditions were at least 60-s long. The nP_o (the product of the number of channels, n, and the channel open probability, P_o) of the channels was determined using Biopatch software (3.11; Molecular Kinetics). nP_o was calculated from the mean total current (I) divided by the single channel current amplitude (i), such that nP_o = I/i. i was determined from the amplitude histogram of the current record.

Nystatin was a generous gift from Dr. S. Lucania (Bristol Meyers-Squibb). 293B (trans-6-cyano-4-(N-ethylsulfonyl-N-methylamino)- 3-hydroxy-2,2-dimethyl-chroman) was a generous gift from Dr. Rainer Greger (Albert-Ludwigs-Universtat, Freiberg, Germany). 1-EBIO was obtained from Aldrich Chemical Co. Acetazolamide, clotrimazole, and bumetanide were obtained from Sigma Chemical Co. Forskolin was obtained from Calbiochem. DNDS was from Pfaltz and Bauer. Charybdotoxin was obtained from Accurate Chemical and Scientific Corp. and made as a 10 μM stock solution in standard bath solution. 1-EBIO, 293B, and clotrimazole were made as >1,000-fold stock solutions in DMSO. Nystatin was made as a 180 mg/ml stock solution in DMSO and sonicated for 30 s just before use. Forskolin and bumetanide were made as 1,000-fold stock solutions in ethanol. Cell culture medium was obtained from GIBCO BRL.

All data are presented as means ± SEM, where n indicates the number of experiments.

In total, we evaluated 216 filters with standard bath solutions on the mucosal and serosal membrane surfaces. The basal I_sc and R_T under these conditions averaged 13 ± 0.8 μA · cm⁻² (range 2–21 μA · cm⁻²) and 353 ± 14 Ωcm² (range 187–667 Ωcm²), respectively. Forskolin (2–10 μM) induced, in all filters tested (n = 109), a damped oscillatory response that became stable and sustained after 5–10 min at a plateau value of 66 ± 4 μA · cm⁻² (range 50–103 μA · cm⁻²). A representative current trace is shown in Fig. 1 A. The increase in I_sc caused by forskolin was accompanied by a decrease in R_T to an average of 189 ± 7 Ωcm⁻² (range 111–333 Ωcm⁻²). Bumetanide (20 μM), an inhibitor of the NaK2Cl cotransporter, caused only a small inhibition of the forskolin stimulated I_sc (Δ −4.9 ± 1.3 μA · cm⁻², n = 11). The failure of bumetanide to inhibit the forskolin-stimulated increase in I_sc suggests that the NaK2Cl cotransporter does not contribute to the I_sc, and this raised the question whether the I_sc was due to Cl⁻ secretion. Additional experiments were performed to establish the ionic basis of the forskolin-stimulated I_sc.

To help elucidate the ionic basis of the forskolin- induced increase in I_sc, we performed unidirectional ion flux measurements with ³⁶Cl, ²²Na, or ⁸⁶Rb; the latter was used as a measure of K⁺ movements. The Cl⁻ flux studies are shown in Fig. 1 B and are summarized together with Na⁺ and Rb⁺ fluxes in Table I. As in the previous experiments, there was a small basal I_sc under control conditions of ∼8 μA · cm⁻² (i.e., 0.3 μEq · cm⁻² · h⁻¹) that was stimulated 6–10-fold by forskolin in the subset of 36 filters used for the flux studies. Under control conditions, there was no net movement of Cl⁻ or Rb⁺ and a small net absorption of Na⁺. Forskolin increased both unidirectional fluxes of Cl⁻ four- to fivefold (Fig. 1 B). Both Rb⁺ fluxes were increased 1.5-fold, but forskolin had no effect on the fluxes of Na⁺ (Table I). Because both unidirectional fluxes of Cl⁻ and Rb⁺ were increased to a similar extent, there was no net flux of Cl⁻ or Rb⁺ caused by forskolin. The difference between I_sc and the net flux of each ion was calculated and is given in Table I as J^R_net. Because there was no net flux of Cl ⁻ or Rb⁺ under control or forskolin conditions, neither of these ions account for the basal or forskolin-stimulated I_sc. However, the net absorption of Na⁺ fully accounts for the control, basal I_sc, and a small portion (15%) of the I_sc in the forskolin-stimulated cells. When the flux studies for Cl⁻, Na⁺, and Rb⁺ were combined to calculate theJ^R_net using the mean I_sc (control 0.31 ± 0.053 μEq · cm⁻² · h⁻¹; forskolin 2.60 ± 0.144 μEq · cm⁻² · h⁻¹, n = 36) for the studies in Table I, the control J^R_net was −0.12 ± 0.11 μEq · cm⁻² · h⁻¹ and the forskolin J^R_net was 2.37 ± 0.189 μEq · cm⁻² · h⁻¹. These results demonstrate that the forskolin-induced increase in I_sc cannot be accounted for by the net transepithelial secretion of Cl⁻ or the absorption of Na⁺ or K⁺. Rather, the increase in I_sc caused by forskolin must be attributed to the net movement of an unmeasured ion, often referred to as the net residual ion flux, J^R_net. Because HCO⁻₃ is the only remaining ion of significant concentration, J ^R_net is likely to be due to the net secretion of HCO⁻₃ and additional experiments were performed to test this hypothesis.

Ion substitution experiments were performed to help further establish the ionic basis of the forskolin-stimulated I _sc. Consistent with the failure of bumetanide to inhibit the forskolin-stimulated I_sc and the J^Cl_net of only 0.09 ± 0.257 μEq · cm⁻² · h⁻¹, substitution of Cl⁻ with gluconate caused only a partial reduction of the response to forskolin (Fig. 2 A). Similar to the control response, the I_sc response to forskolin in Cl⁻-free solution was rapid in onset with a transient peak and a sustained plateau of 46 ± 1.6 μA · cm⁻² (n = 24) (Fig. 2 A). The subsequent addition of Cl⁻ (30–60 mM) to the mucosal or serosal solution did not cause a further increase in I_sc (data not shown). As in the Cl⁻ containing solution, bumetanide (20 μM serosal) had no effect on the forskolin-stimulated I_sc (Δ 0.15 ± 0.76 μA · cm⁻², n = 6) (Fig. 3). In contrast, removal of HCO⁻₃ from the mucosal and serosal bathing solutions resulted in a greatly diminished response to forskolin (Fig. 2 B). After a transient response, I _sc was increased by only 4 ± 1 μA · cm⁻² (n = 10) in HCO⁻₃-free solutions. Substitution of Na ⁺ with N-methyl-d-glucamine, Cl⁻ with gluconate, and NaHCO₃ with choline HCO₃ also resulted in a greatly reduced response to forskolin. Forskolin caused a transient increase in I_sc without a sustained plateau in the Na⁺-free, Cl⁻-free, HCO⁻₃- containing solution (Fig. 2 C), which resembles the response in HCO⁻₃-free media. However, the subsequent addition of Na ⁺ (30 mM) to the serosal but not the mucosal solution caused a sustained increase in I_sc of 24 ± 1.0 μA · cm⁻² (n = 12) in forskolin-stimulated cells (Fig. 4). Addition of Na⁺ (30 mM) to the serosal solution be-fore forskolin caused a small decrease in I_sc Δ −7.6 ± 0.2 μA · cm⁻² (n = 12) as expected for the serosal-to-mucosal diffusion of a cation. This decrease in I_sc was reversed and I_sc rose to a sustained level of 23 ± 0.8 μA · cm⁻² (n = 12) with the subsequent addition of forskolin. Thus, the forskolin-stimulated increase in the I_sc was Cl⁻ independent but Na⁺ and HCO⁻₃dependent.

The above results are consistent with forskolin-stimulated net secretion of HCO⁻₃. To further test this hypothesis, the pharmacological sensitivity to various inhibitors of HCO⁻₃ transport were evaluated. The carbonic anhydrase inhibitor, acetazolamide (1 mM mucosal and serosal), caused a 27% decrease (a reduction of 13 ± 1 μA cm², n = 6) in the forskolin-stimulated I_sc in Cl⁻-free solutions (Fig. 3). DNDS (3 mM), an inhibitor of Cl⁻/HCO⁻₃ exchangers and Na ⁺:HCO⁻₃ cotransporters, was without effect when added to the mucosal solution ( Δ = 0.2 μA · cm⁻², n = 6), but caused an inhibition of 56% (Δ −26 ± 1 μA · cm⁻², n = 6) when added to the serosal side in Cl⁻-free solutions. Similar results were obtained in Cl⁻-containing solutions (Δ −2.5 ± 1.3 μA · cm⁻², n = 6 mucosal; Δ −27 ± 2 μA · cm⁻², n = 6 serosal). The half maximal inhibitory concentration (K_i) for serosal DNDS was 300 μM. The inhibitory effects of serosal DNDS and acetazolamide were additive, together causing a 75% decrease in I_sc. The Na⁺-K⁺-ATPase inhibitor, ouabain (100 μM), caused an immediate and complete inhibition of the forskolin-stimulated I_sc. Neither CTX (50 nM), a blocker of Ca²⁺ activated K⁺ channels (Garcia et al., 1995), nor 293B (100 μM), a blocker of the cAMP/ PKA activated K⁺ channel (KvLQT1; Lohrmann et al., 1995; Loussouarn et al., 1997) inhibited the forskolin-stimulated I_sc. The nonselective K⁺ channel blocker, Ba²⁺ (5 mM serosal side), inhibited the forskolin-stimulated I_sc by only 10 ± 2 μA · cm⁻² (n = 6).

The requirement for serosal Na⁺, the inhibition by ouabain, and the partial inhibition by serosal DNDS suggests some of the secreted HCO⁻₃ is mediated by the uptake of HCO⁻₃ across the basolateral membrane on a Na ⁺:HCO⁻₃ cotransporter. ² The partial inhibition of I_sc by acetazolamide suggests some of the secreted HCO⁻₃ originates from a metabolic source. The Cl ⁻ independence and the failure of mucosal DNDS to inhibit I_sc suggests the exit of HCO⁻₃ across the apical membrane is not mediated by a Cl ⁻/HCO⁻₃ exchanger.

The above results are consistent with the conclusion that forskolin stimulation causes the electrogenic secretion of HCO⁻₃. To further test this hypothesis, we performed experiments to determine whether forskolin caused an alkalinization of the apical solution. Calu-3 cells were studied under open circuit conditions with a small volume of fluid (100 μl) on the apical surface (1.1 cm²) and 5 ml of continuously gassed (95% O₂/ 5% CO₂) NaCl, NaHCO₃ buffer, pH 7.4, on the serosal side. Cells were incubated without or with forskolin (2 μM) and the apical solution collected after 90 min. The apical sample was thoroughly gassed before measuring its pH with a miniature pH electrode. Studied in this manner, we found forskolin caused an alkalinization of the apical solution to a pH of 7.8 ± 0.06 (n = 6), whereas control untreated filters showed a small acidification of the apical solution, pH 7.3 ± 0.05 (n = 6). The forskolin-stimulated alkalinization of Δ0.5 pH over a 90-min period corresponds to the net movement of HCO⁻₃ of 1.7 μeq · cm⁻² · h⁻¹ or 46 μA · cm⁻², a value in good agreement with the forskolin-stimulated increase in I_sc of 53 μA · cm⁻² under short circuit conditions.³ Based on these pH measurements, the ion flux measurements, the ion substitution studies, and the pharmacology studies, we conclude that the forskolin-induced I_sc response in Calu-3 cells is due to the net secretion of HCO⁻₃ by a Cl ⁻-independent Na⁺-dependent, and DNDS-sensitive electrogenic mechanism.

We previously demonstrated that the novel benzimidazolinone, 1-EBIO, induced a sustained transepithelial Cl⁻ secretory response in rat colonic mucosa, human colonic T84 cells, and murine airway epithelia (Devor et al., 1996). CTX and clotrimazole inhibited the 1-EBIO– stimulated Cl⁻ secretion consistent with the activation of basolateral membrane K⁺ channels that was confirmed in permeabilized monolayers (Devor et al., 1996, 1997). Moreover, patch clamp studies demonstrated 1-EBIO activates an inwardly rectifying, calcium activated, CTX, and clotrimazole-sensitive K⁺ channel (Devor et al., 1996, 1997). Permeabilized monolayers revealed 1-EBIO also activates an apical membrane Cl⁻ conductance (Devor et al., 1996). The studies reported here were performed to determine if 1-EBIO would have similar effects on Calu-3 cells.

In 46 experiments, 1-EBIO (1 mM) increased I_sc from a basal value of 8 ± 0.8 to 62 ± 4 μA · cm⁻² with only a modest decrease in R_T (control 397 ± 21 Ωcm² vs. 1-EBIO 336 ± 20 Ωcm²). A current trace of a typical I_sc response to 1-EBIO is shown in Fig. 5 A. The response was rapid in onset and sustained over a long period. Dose–response studies revealed the half maximal effective concentration of 1-EBIO was ∼500 μM. Consistent with the activation of the K_Ca channels, CTX (50 nM) inhibited 47% of the 1-EBIO–stimulated I_sc. The half maximal effective concentration of CTX was 3.2 nM (n = 4). Clotrimazole (10 μM), a nonpeptide inhibitor of K_Ca, also inhibited 87.6 ± 1.9% (n = 5) of the response to 1-EBIO with a K _i of 1.2 μM (n = 5). Bumetanide (20 μM) inhibited ∼50% of the 1-EBIO–stimulated I_sc (Table II). DNDS and acetazolamide caused only small (<10%) decreases in the 1-EBIO–stimulated I_sc.

Unidirectional fluxes of ³⁶Cl revealed that 1-EBIO caused the net secretion of Cl⁻ (Fig. 5 B and Table II). As in previous experiments (Fig. 1 B), there was no net secretion of Cl⁻ in control monolayers. 1-EBIO caused a sixfold increase in the serosal-to-mucosal flux of Cl⁻ without altering the mucosal-to-serosal flux leading to net Cl⁻ secretion. Moreover, the net secretion of Cl⁻ fully accounted for the increase in I_sc caused by 1-EBIO, leaving a small J^R_net of only 0.25 ± 0.263 μEq · cm⁻² · h⁻¹. Bumetanide inhibited the serosal-to-mucosal flux of Cl⁻ and thereby caused a 70% inhibition in J ^Cl−_net in 1-EBIO–stimulated monolayers.

The above results demonstrate Calu-3 cells secrete HCO⁻₃ when stimulated by forskolin and Cl ⁻ when stimulated by 1-EBIO. In the next series of experiments, we evaluated the effects of 1-EBIO on forskolin stimulated monolayers. As in the previous experiments, forskolin increased I_sc from a control value of 6.8 ± 0.7 to 67 ± 4.3 μA · cm⁻² (n = 12) without causing the net secretion of Cl⁻ and leaving a J^R_net nearly equal to the change in I _sc (Fig. 6 and Table III). 1-EBIO further increased I_sc to 114 ± 5 μA · cm⁻² (Fig. 6 and Table III). Similar results were obtained if the order of the addition of forskolin and 1-EBIO were reversed. CTX inhibited 79 ± 2% (n = 8) and bumetanide inhibited 80 ± 1% (n = 5) of the forskolin plus 1-EBIO–stimulated I_sc. When added to the forskolin-stimulated cells, 1-EBIO caused a twofold increase in the serosal-to-mucosal flux of Cl⁻ and a J ^Cl−_net that was nearly equal to the I _sc (Fig. 6 and Table III). Thus, 1-EBIO caused a 70% decrease in the forskolin-stimulated J^R_net. These results suggest 1-EBIO can switch the forskolin-stimulated Calu-3 cells from HCO⁻₃- to Cl ⁻-secreting cells.

One hypothesis to explain the effects of 1-EBIO on Calu-3 cells is the activation of basolateral membrane K⁺ channels that would tend to hyperpolarize the membrane potential. The inhibition of the 1-EBIO response by CTX and clotrimazole support this hypothesis. Hyperpolarization of the membrane potential would increase the driving force for anion exit of both HCO⁻₃ and Cl ⁻ across the apical membrane. However, hyperpolarization of the basolateral membrane potential would also tend to decrease the driving force for basolateral membrane HCO⁻₃ entry on the Na ⁺:HCO⁻₃ cotransporter, whose Na ⁺ to HCO⁻₃ stoichiometry is reported to be 1:2 or 1:3 in various cell types (Boron and Boulpaep, 1989). A second hypothesis, and one that is not mutually exclusive with the former hypothesis, is that 1-EBIO activates apical membrane anion channels that were not activated by forskolin and that the 1-EBIO-activated channels allow for the preferential exit of Cl⁻ over HCO⁻₃. To test these hypotheses, we performed studies on permeabilized monolayers.

The pore forming antibiotic nystatin was used to permeabilize the apical membrane and a transepithelial mucosal-to-serosal K ⁺ gradient was established. After permeabilization, 1-EBIO increased I_K, and this was inhibited by both CTX (Fig. 7 A) and clotrimazole (B). In 17 experiments, 1-EBIO (1 mM) increased I_K an average of 91 ± 9 μA · cm⁻² and this was inhibited 66 ± 2% by CTX (50 nM, n = 10) and 95 ± 2% by clotrimazole (10 μM, n = 7). Thus, 1-EBIO does activate basolateral membrane K⁺ channels. In contrast, forskolin (2 μM) failed to cause an increase in I_K. After the establishment of a mucosal-to-serosal Cl⁻ gradient, the addition of nystatin to the serosal membrane elicited an absorptive I_Cl of 58 ± 9 μA · cm⁻² (n = 24, Fig. 8). Thus, in contrast to the measurements of I_K, treatment of the monolayers with nystatin appears to uncover or activate a substantial basal I_Cl. Similar results were observed in T84 cells studied under the same experimental conditions (Devor et al., 1996). Therefore, this effect of nystatin is not unique to Calu-3 cells. The mechanisms involved in this nystatin induced increase in I_Cl are unknown. The subsequent addition of forskolin (10 μM) to the nystatin-treated monolayers increased I_Cl by an additional 186 ± 15 μA · cm⁻² (n = 7) (Fig. 8 A). 1-EBIO failed to cause any further increase in I_Cl in the forskolin treated monolayers. However, 1-EBIO alone when added to the nystatin-treated monolayers increased I_Cl by an additional 74 ± 11 μA · cm⁻² (n = 6) and forskolin further increased I_Cl by an additional 110 ± 12 μA · cm⁻² (n = 6; Fig. 8 B).

Thus, both forskolin and 1-EBIO when added alone can activate an apical membrane Cl⁻ conductance in nystatin-treated Calu-3 monolayers. Forskolin caused a 2.5-fold greater increase in I_Cl compared with the 1-EBIO response. The lack of specific Cl⁻ channel blockers (Schultz et al., 1999) prevents us from determining whether the same channel or different Cl⁻ channels are activated by forskolin and 1-EBIO. However, when forskolin and then 1-EBIO was added, the effects on I_Cl were not additive, suggesting that forskolin alone can maximally activate the apical Cl⁻ conductance. Therefore, the effect of 1-EBIO in causing the switch from HCO⁻₃ secretion to Cl ⁻ secretion appears to result from the activation of basolateral membrane K⁺ channels and decreased driving force for HCO⁻₃ entry across the basolateral membrane. This hypothesis will be considered further in the discussion.

The above results indicate that Calu-3 cells express K ⁺ channels with similar pharmacological characteristics to the K⁺ channels we described previously in T84 cells (Devor and Frizzell, 1993; Devor et al., 1996, 1997) and that this conductance may be important in altering the driving force for HCO⁻₃ entry across the basolateral membrane that elicits Cl ⁻ secretion in Calu-3 cells. Thus, we wished to characterize this K⁺ channel at the single channel level. Inward and outward single-channel currents observed on excision of membrane patches into a symmetric K⁺ bath containing 400 nM free Ca²⁺ are shown in Fig. 9 A. Channel activity showed no obvious voltage dependence and required Ca²⁺ in the bath (data not shown). The average current–voltage for four such patches is shown in Fig. 9 B (•). Single channel currents were inwardly rectified with average chord conductance values of 31 ± 2 pS at −100 mV and 9 ± 0.2 pS at +100 mV. The K⁺-to-Na⁺ selectivity of this channel was assessed by replacing 100 mEq pipette K⁺ with Na⁺; P_K/P_Na was calculated from the Goldman-Hodgkin-Katz relation. Replacing pipette K⁺ with Na⁺ shifted the reversal potential by −20 mV (n = 4; Fig. 9 B, ○). A shift of −27 mV is predicted for a perfectly K⁺ selective electrode. From these data, the calculated K⁺-to-Na⁺ selectivity ratio is 5.5:1. This conductance and K⁺:Na⁺ selectivity values are similar to what has been previously reported for a Ca²⁺-activated K⁺ channel in T84 cells (Devor and Frizzell, 1993; Tabcharani et al., 1994; Roch et al., 1995) as well as primary cultures of canine tracheal epithelial cells (Welsh and McCann, 1985; McCann et al., 1990).

We previously demonstrated that 1-EBIO directly activated the K_Ca of T84 cells in excised patch-clamp recordings (Devor et al., 1996). Thus, we determined whether 1-EBIO would similarly activate K_Ca in excised, inside-out single channel patch-clamp recordings from Calu-3 cells. The effect of 1-EBIO (200 μM) on one patch is shown in Fig. 10. Under control conditions (400 nM free Ca²⁺ in the bath), minimal K_Ca channel activity was observed. 1-EBIO produced a large increase in channel activity that was readily reversible after washout of the 1-EBIO. In 14 inside-out recordings, 1-EBIO increased nP_o from 0.08 ± 0.02 to 1.68 ± 0.39. These results indicate that this channel, as in T84 cells, is responsible for the increase in the basolateral membrane K⁺ conductance and I_sc during an 1-EBIO–mediated secretory response.

We demonstrate above that the 1-EBIO–induced basolateral membrane K⁺ conductance is sensitive to block by CTX and clotrimazole (Fig. 7). We therefore determined whether these inhibitors would block the channel in excised outside-out and inside-out patches. The effect of CTX (50 nM) on K_Ca in an outside-out patch is shown in Fig. 11 A. When holding the patch at −100 mV, addition of CTX to the outside of the channel resulted in a complete inhibition of channel activity. This block was voltage dependent and was partially relieved by voltage clamping the patch to +100 mV. The inhibition by CTX was completely reversible. Similar results were obtained in three additional outside-out patches. Clotrimazole (10 μM) also completely inhibited K_Ca activity, reducing nP_o from 1.59 ± 0.24 to 0.05 ± 0.02 (n = 6; Fig. 11B). Thus, results from these K⁺ channel blocker experiments further indicate that 1-EBIO is activating this inwardly rectifying Ca²⁺-activated K⁺ conductance in Calu-3 monolayers resulting in the stimulation of Cl⁻ secretion and the inhibition of HCO⁻₃ secretion.

The results of our studies with Calu-3 cells demonstrate that forskolin stimulates the net secretion of HCO⁻₃. Forskolin consistently caused an increase in I_sc to a new sustained plateau. Ion flux studies revealed that this increase in I_sc could not be explained by the net transport of Na⁺, Rb⁺, or Cl⁻, leaving HCO⁻₃ secretion as the likely basis for the increase in I _sc. Ion substitution experiments demonstrated HCO⁻₃, but not Cl ⁻, was required to elicit a sustained increase in I_sc with forskolin. In addition, Na⁺ was required in the serosal bath to elicit a forskolin response. Inhibitor studies revealed that the forskolin response was sensitive to ouabain, indicating a role for the Na⁺/K⁺-ATPase. The forskolin response was also sensitive to DNDS on the serosal side but not the mucosal side, indicating a role for a basolateral membrane Na⁺:HCO⁻₃ cotransporter or Cl⁻: HCO⁻₃ exchanger. However, because Cl⁻ was not required and serosal Na⁺ was, the effects of DNDS are likely to result from the inhibition of a basolateral membrane Na⁺:HCO⁻₃ cotransporter. Acetazolamide caused a partial inhibition of the forskolin response, consistent with some of the secreted HCO⁻₃ arising from metabolic sources. The ion flux studies failed to show evidence of net secretion of Cl ⁻ in response to forskolin, and bumetanide did not inhibit the I_sc response. Thus, forskolin did not cause the net secretion of Cl⁻ across Calu-3 cells under short circuit conditions. Rather, we conclude forskolin causes the net secretion of HCO⁻₃ by a Cl ⁻-independent, Na⁺-dependent, and DNDS-sensitive electrogenic mechanism in Calu-3 cells. The forskolin-stimulated alkalinization of the mucosal bathing solution of Calu-3 cells, studied under open circuit conditions, lends further support to this conclusion.

Although forskolin did not stimulate the net secretion of Cl⁻, it did cause a fivefold increase in both unidirectional fluxes of Cl⁻ (Fig. 1 B and Table I) and it is of interest to understand the mechanisms that underly these changes. Our first interpretation was that forskolin increased the transcellular passage of Cl⁻ in both directions. Thus, the opening of CFTR would allow for both the exit and entry of Cl⁻ across the apical membrane. The NaK2Cl cotransporter in the basolateral membrane would allow the entry of Cl⁻ leaving one to explain how Cl⁻ exits the cell in the serosal-to-mucosal direction. However, bumetanide did not alter the unidirectional fluxes, consistent with the lack of change in the forskolin-stimulated I_sc. Thus, the NaK2Cl cotransporter does not appear to mediate the entry of Cl⁻ across the basolateral membrane in the forskolin-stimulated monolayers. We next entertained the possibility that Cl⁻ may move across the basolateral membrane on a Cl⁻:HCO⁻₃ exchanger. However, the increases in both unidirectional fluxes in response to forskolin were still observed in HCO⁻₃-free buffer. Thus, the increased fluxes do not depend on extracellular HCO⁻₃. Because this experiment does not exclude the possibility that a basolateral membrane anion exchanger is operating in a Cl ⁻:Cl⁻ exchange mode, we examined the effects of serosal DNDS (1 mM) on the Cl⁻ fluxes. DNDS cause a 70% decrease in both unidirectional fluxes in the forskolin-stimulated monolayers. Therefore, the increase in Cl⁻ fluxes caused by forskolin can largely be accounted for by a Cl⁻:Cl⁻ exchange across the basolateral membrane and the exit and entry of Cl⁻ via CFTR across the apical membrane.

The studies with 1-EBIO demonstrated the Calu-3 cells are not limited to the secretion of HCO⁻₃ , but rather they can also be stimulated to secrete Cl⁻. 1-EBIO, like forskolin, consistently caused a sustained increase in I_sc. ³⁶Cl flux studies showed the 1-EBIO–stimulated increase in I_sc could be fully accounted for by the net secretion of Cl⁻. In addition, both the increase in I_sc and the net secretion of Cl⁻ were inhibited by bumetanide. Studies on permeabilized Calu-3 monolayers revealed 1-EBIO activates both a basolateral membrane K⁺ conductance and an apical membrane Cl⁻ conductance as previously shown in studies on T84 cells (Devor et al., 1996). CTX and clotrimozole both inhibited the 1-EBIO I_sc response as well as the 1-EBIO–activated K⁺ current in permeabilized monolayers. Patch-clamp studies demonstrated the presence of an intermediate conductance, inwardly rectified, Ca⁺-activated K⁺ channel in Calu-3 cells that was activated by 1-EBIO and blocked by CTX and clotrimozole. We and others have also identified a Ca⁺-activated K⁺ channel with identical biophysical properties and pharmacological profile in T84 cells (Devor and Frizzell, 1993; Tabcharani et al., 1994; Roch et al., 1995; Devor et al., 1996). Moreover, Welsh and McCann (1985) and McCann et al. (1990) have already shown that this channel is expressed in native airway epithelial cells and is therefore not just in epithelial cell lines. Recently, three different groups have cloned the same K⁺ channel, variously referred to as hIK-1, hSK4, and hIK (Ishii et al., 1997; Joiner et al., 1997; Jensen et al., 1998). These channels have identical biophysical properties and pharmacological profile to the channel observed in canine tracheocytes, T84 cells, and Calu-3 cells. Northern blot analysis has confirmed the presence of the mRNA for hIK-1 in T84 and Calu-3 cells (Devor, D.C., unpublished results). Thus, we conclude that one site of action of 1-EBIO is the activation of hIK-1 in the basolateral membrane of Calu-3 cells. Permeabilization of monolayers demonstrated 1-EBIO also activates an apical membrane Cl⁻ channel; however, the identity of the apical membrane Cl⁻ channel that is activated by 1-EBIO is less certain. Haws et al. (1994) have reported the predominant Cl⁻ channel observed in Calu-3 cells is a low conductance channel with properties consistent with those of CFTR. 1-EBIO is a benzimidazolinone and other benzimidazolinones (e.g., NS004 and NS1619) have been reported to activate CFTR (Gribkoff et al., 1994; Champigny et al., 1995). Thus, it is possible that the Cl⁻ channel activated by 1-EBIO in Calu-3 cells is CFTR. However, further studies will be necessary to confirm this hypothesis.

Calu-3 cells secrete HCO⁻₃ in response to forskolin and Cl ⁻ in response to 1-EBIO. However, when the two agonists are added together, anion secretion is dominated by Cl⁻ secretion and there is a decrease in the net secretion of HCO⁻₃. Studies with primary cultures of human bronchial epithelial cells lead Smith and Welsh (1992) to suggest that airway epithelia may also switch between HCO⁻₃ and Cl ⁻ secretion. Ashton et al. (1991) have also suggested that pancreatic ductal epithelial cells can be differentially stimulated to secrete HCO⁻₃ or Cl⁻. The mechanisms that underlie the switch between HCO⁻₃ and Cl ⁻ secretion are largely unknown. Our results with Calu-3 cells offer some insight and suggest a model (Fig. 12) to explain how the same cell can secrete HCO⁻₃ when stimulated by forskolin and Cl ⁻ when stimulated by 1-EBIO or 1-EBIO plus forskolin.

The first tenet of the model is the presence of an anion channel in the apical membrane that can conduct both HCO⁻₃ and Cl ⁻. Whether there are two separate channel types, one favoring HCO⁻₃ and activated by forskolin and one favoring Cl⁻ and activated by 1-EBIO, or a single channel type that conducts both HCO⁻₃ and Cl ⁻ is not clear at this time. Nonselective anion channels have been reported but to our knowledge an epithelial anion channel that favors HCO⁻₃ over Cl ⁻ has not yet been described in the literature. Because HCO⁻₃ secretion is stimulated by forskolin, the anion channel mediating the secretion of HCO⁻₃ is likely to be activated by cAMP and PKA, as is CFTR. CFTR is highly expressed in Calu-3 cells (Finkbeiner et al., 1993; Shen et al., 1994) and activated by forskolin when measured by anion efflux methods and patch clamp analysis (Haws et al., 1994). Preliminary studies using impedance analysis have shown forskolin does activate an apical membrane anion conductance in Calu-3 cells (Bridges, R.J., unpublished observations). Patch-clamp anion selectivity studies have shown CFTR can conduct HCO⁻₃, although at a fraction (0.15–0.25) of the Cl ⁻ conductance (Gray et al., 1990; Poulsen et al., 1994; Linsdell et al., 1997). Heterologous expression of wt-CFTR but not ΔF508-CFTR in NIH3T3 fibroblasts and C127 mammary cells was shown to confer the cells with a Na⁺-independent, HCO⁻₃-dependent, forskolin-regulated intracellular pH recovery mechanism (Poulsen et al., 1994). Illek et al. (1997) have shown, in α-toxin– permeabilized monolayers of Calu-3 cells, the activation of a HCO⁻₃ current by cAMP with a similar HCO⁻₃ to Cl ⁻ selectivity as observed in the patch-clamp studies. In addition, Smith and Welsh (1992) demonstrated cAMP-stimulated HCO⁻₃ secretion across normal but not CF airway epithelia and they suggested HCO⁻₃ exit across the apical membrane is through the Cl ⁻ channel that is defectively regulated in CF. Thus, we propose that CFTR mediates the exit of HCO⁻₃ across the apical membrane of Calu-3 cells.

The involvement of an anion channel in HCO⁻₃ secretion is not a new concept. However, previous models have proposed the anion channel acts as a shunt pathway mediating the exit of Cl⁻ from the cell (Stetson et al., 1985). Luminal Cl⁻ is then thought to be used by an apical membrane Cl⁻:HCO⁻₃ exchanger that mediates the exit of HCO⁻₃ from the cell. Thus, this model for HCO⁻₃ secretion necessitates the presence of luminal Cl ⁻ for the apical membrane exit of HCO⁻₃. The studies with Calu-3 cells demonstrate Cl ⁻ is not required for the secretion of HCO⁻₃. Ishiguro et al. (1996) have recently reported results on HCO⁻₃ secretion in interlobular ducts from guinea pig pancreas that demonstrate agonist-stimulated HCO⁻₃ efflux at low (7 mM) luminal Cl ⁻ concentrations. These authors suggest their results are not easily reconciled with HCO⁻₃ transport across the luminal membrane being mediated by a Cl ⁻:HCO⁻₃ exchanger in parallel with a Cl ⁻ conductance. Rather, they too argue for a conductive, channel mediated, exit of HCO⁻₃ across the apical membrane (Ishiguro et al., 1996). Our findings are consistent with this hypothesis, and they suggest the Calu-3 cells will be a useful cell line to help further test this hypothesis as well as to determine the role of CFTR in apical HCO⁻₃ exit.

The second tenet of the model (Fig. 12) is the presence of an electrogenic Na ⁺:HCO⁻₃ cotransporter (NBC) in the basolateral membrane that mediates the entry of HCO⁻₃ into the cell. Boron and Boulpaep (1983) were the first to describe an electrogenic NBC with Na ⁺:HCO⁻₃ stoichiometry of 1:3 that mediates the exit of HCO⁻₃ across the basolateral membrane in the proximal tubule of the tiger salamander Ambystoma tigrinum. Romero et al. (1997) using mRNA from the tiger salamander kidney have recently expression cloned this NBC. The cloning of a human homologue of the renal NBC has also recently been reported (Burnham et al., 1997), as has a unique human pancreatic isoform (Abuladze et al., 1998). The stoichiometries of the cloned NBCs have not yet been established but Xenopus oocyte expression studies have shown the renal NBC is electrogenic, Na⁺- and HCO⁻₃-dependent, Cl ⁻-independent, and disulfonic stilbene–sensitive (Romero et al., 1997). These characteristics are shared by NBCs studied in kidney, glial, liver, pancreas, and colon (Boron and Boulpaep, 1989). Our studies with Calu-3 cells demonstrate that forskolin-stimulated HCO⁻₃ secretion also shares these characteristics, consistent with the presence of a NBC in the basolateral membrane. Preliminary reverse transcription–PCR and sequencing studies have shown Calu-3 cells express a NBC (Gangopadhyay and Bridges, unpublished observations) lending further support to this notion. Studies in progress are focused on ascertaining which of the NBC isoforms is expressed in Calu-3 cells as well as the membrane localization, apical versus basolateral, of the cotransporter. According to Fig. 12, we predict a basolateral membrane NBC with a Na ⁺:HCO⁻₃ stoichiometry that favors the entry of HCO⁻₃ when Calu-3 cells are stimulated by forskolin. Both the pancreatic and renal isoforms of the NBCs have consensus phosphorylation sites for protein kinase A and therefore may be regulated by cAMP-mediated agonists (Romero et al., 1997; Abuladze et al., 1998). Thus, in addition to the activation of an apical membrane anion channel (CFTR?), forskolin may also activate HCO⁻₃ entry on the NBC.

Whether a NBC mediates entry or exit of HCO⁻₃ depends on the stoichiometry of the transporter, the membrane potential, and the concentrations of Na ⁺ and HCO⁻₃ inside and outside the cell. Sodium: HCO⁻₃ stoichiometries of 1:2 and 1:3 have been reported (Boron and Boulpaep, 1989), indicating that turnover of the NBC may result in the transfer of one or two negative charges across the membrane at usual membrane voltages. The 1:2 stoichiometry is associated with NBC-mediated HCO⁻₃ entry, whereas a 1:3 stoichiometry is consistent with HCO⁻₃ exit. If one assumes typical ion concentrations of 145 mM Na ⁺, 25 mM HCO⁻₃ outside, and 15 mM Na ⁺ and 15 mM HCO⁻₃ inside, then HCO⁻₃ will enter a cell on the NBC at membrane potentials less hyperpolarized than −85 mV when the Na⁺:HCO⁻₃ stoichiometry is 1:2 and −49 mV when it is 1:3. Membrane potentials more hyperpolarized than these valves will lead to HCO⁻₃ exit from the cells. Thus, the activation of basolateral membrane K ⁺ channels by 1-EBIO is expected to hyperpolarize the membrane potential, and this will inhibit the entry of HCO⁻₃ on the NBC. If the hyperpolarization is of sufficient magnitude, this change in driving force may drive HCO⁻₃ out of the cell across the basolateral membrane. Hyperpolarization will also tend to drive anions (HCO⁻₃ and Cl⁻) out of the cell across the apical membrane. However, because basolateral membrane entry of HCO⁻₃ becomes inhibited, this apical membrane hyperpolarization will favor Cl⁻ secretion. Therefore, we propose that the switch between HCO⁻₃ secretion and Cl ⁻ secretion is determined by the basolateral membrane potential. Differential regulation of the basolateral membrane potential by secretory agonists would provide a means of stimulating HCO⁻₃ or Cl ⁻ secretion. As shown in Fig. 12, CFTR could serve as both a HCO⁻₃ and a Cl ⁻ channel mediating the apical membrane exit of either anion depending on the nature of the anion provided by the basolateral membrane cotransporter mechanisms.

Why does forskolin fail to stimulate Cl⁻ secretion in Calu-3 monolayers? Cyclic AMP–stimulated Cl⁻ secretion is known to require the activation of both an apical membrane Cl⁻ conductance and a basolateral membrane K⁺ conductance; the former depolarizes and the latter repolarizes the membrane voltage to maintain a driving force for Cl⁻ exit (Halm and Frizzell, 1990). Permeabilization studies demonstrated forskolin does activate an apical membrane Cl⁻ conductance (Fig. 8), but that it fails to activate a basolateral membrane K⁺ conductance (Fig. 7). Thus, unless the basal K⁺ conductance can maintain the apical voltage above the Cl⁻ equilibrium potential (E_Cl < −35 mV, assuming intracellular Cl⁻ = 30 mM), Cl⁻ can not be secreted. Indeed, the expected high Cl⁻ conductance of the apical membrane of forskolin-stimulated Calu-3 cells would set the apical membrane voltage at E_Cl and this would provide the driving force for HCO⁻₃ exit since E _HCO3 is –13 mV (assuming intracellular HCO⁻₃ = 15 mM and extracellular = 25 mM).⁴ This electrical coupling may explain the apparent Cl⁻ dependence of HCO⁻₃ secretion in some epithelia and further emphasizes the importance of CFTR in Cl ⁻ and HCO⁻₃ secretion.

If the results we have obtained with Calu-3 cells accurately reflect the transport properties of native submucosal gland serous cells, then HCO⁻₃ secretion in the human airways warrants greater attention. Calu-3 cell HCO⁻₃ secretion in response to cAMP-mediated agonists is quite similar to that observed in pancreatic duct cells where mutations in CFTR have profound pathological effects. Pancreatic function in CF patients is characterized by impaired fluid, HCO⁻₃, and Cl ⁻ secretion by the ductal epithelial cells, the site of CFTR expression (Durie and Forstner, 1989; Marino et al., 1991). Impaired secretion ultimately leads to destruction of the pancreas by digestive enzymes in the obstructed ducts. The principle secreted ion by the ductal cells is HCO⁻₃, which drives Na ⁺ and water into the lumen by electrical and osmotic coupling. The secreted alkaline fluid serves to regulate the activities of the digestive enzymes and to flush them into the duodenal lumen. Secreted HCO⁻₃ is also thought to have an osmotic advantage (Hogan et al., 1994). With the aid of carbonic anhydrase, HCO⁻₃ can quickly combine with protons to make CO ₂ and H₂O, and thereby tend to make the fluid hypoosmotic. If the airway submucosal glands and surface epithelium function in an analogous manner, potential roles for HCO⁻₃ in the airways may include the processing, regulation, and clearance of submucosal gland–derived enzymes, mucus, and antimicrobial agents. Early studies have suggested mucus undergoes a transition from gel to sol at alkaline pH (Forstner et al., 1977) and HCO⁻₃ secretion could therefore aid in the clearance of mucus from the submucosal glands, a process that is impaired in CF. Airway serous cells also express abundant amounts of carbonic anhydrase (Basbaum et al., 1990), some of which may be of the type IV membrane-associated isoform that could convert the secreted HCO⁻₃ to CO₂ and H₂O in the lumen of the gland or in the airway surface fluid. The rapid loss of CO₂ during ventilation of the airways would favor a shift in the enyzmatic equilibrium toward the conversion of HCO⁻₃ to H₂O. The volatility of the HCO⁻₃ /CO₂ buffer system, especially at an air–liquid interface, while having potential physiological significance, will also make the investigation of HCO⁻₃ secretion in the airways a formidable challenge to the experimentalist. Studies with Calu-3 cells will provide a means to further investigate the mechanisms involved in serous cell HCO⁻₃ secretion, and perhaps with this knowledge how to better study HCO⁻₃ secretion in the intact airways.

We gratefully acknowledge the excellent technical assistance of Mr. Matthew Green, M. Cheng Zhang Shi, and Ms. Maitrayee Sahu. We thank Ms. Michele Dobransky for the superior secretarial assistance in preparing the manuscript.

This work was supported by Cystic Fibrosis Foundation (CFF) grant Devor96PO (to D.C. Devor) and National Institute of Diabetes and Digestive and Kidney Disease grants DK45970 (to R.J. Bridges) and DK46588 (to R.A. Frizzell). A.C. DeLuca is a CFF fellow (DeLuca98DO) and R.J. Bridges is a CFF Research Scholar (E841).