Fluid secretion in interlobular ducts isolated from guinea-pig pancreas

Abstract

1. Pancreatic HCO₃⁻ and fluid secretion were studied by monitoring luminal pH (pH_L) and luminal volume simultaneously in interlobular duct segments isolated from guinea-pig pancreas. The secretory rate and HCO₃⁻ flux were estimated from fluorescence images obtained following microinjection of BCECF-dextran (70 kDa, 20 μM) into the duct lumen.

2. Ducts filled initially with a Cl⁻-rich solution swelled steadily (2.0 nl min⁻¹ mm⁻²) when HCO₃⁻²/CO was introduced, and the luminal pH increased to 8.08. When Cl⁻ was replaced by glucuronate, spontaneous fluid secretion was reduced by 75 %, and pH_L did not rise above 7.3.

3. Cl⁻-dependent spontaneous secretion was largely blocked by luminal H₂DIDS (500 μM). We conclude that, in unstimulated ducts, HCO₃⁻ transport across the luminal membrane is probably mediated by Cl⁻-HCO₃⁻ exchange.

4. Secretin (10 nM) and forskolin (1 μM) both stimulated HCO₃⁻ and fluid secretion. The final value of pH_L (8.4) and the increase in secretory rate (1.5 nl min⁻¹ mm⁻²) after secretin stimulation were unaffected by substitution of Cl⁻.

5. The Cl⁻-independent component of secretin-evoked secretion was not affected by luminal H₂DIDS. This suggests that a Cl⁻-independent mechanism provides the main pathway for luminal HCO₃⁻ transport in secretin-stimulated ducts.

6. Ducts filled initially with a HCO₃⁻-rich fluid (125 mM HCO₃⁻, 23 mM Cl⁻) secreted a Cl⁻-rich fluid while unstimulated. This became HCO₃⁻-rich when secretin was applied.

7. Addition of H₂DIDS and MIA (10 μM) to the bath reduced the secretory rate by 56 and 18 %, respectively. Applied together they completely blocked fluid secretion. We conclude that basolateral HCO₃⁻ transport is mediated mainly by Na⁺-HCO₃⁻ cotransport rather than by Na⁺-H⁺ exchange.

Studies on isolated pancreatic ducts have identified various ion channels and transporters that may be involved in transepithelial HCO₃⁻ secretion by this tissue (Case & Argent, 1993). Measurements of intracellular ion concentrations (Zhao et al. 1994; Villanger et al. 1995; Ishiguro et al. 1996a; De Ondarza & Hootman, 1997) and electrophysiological data (Gray et al. 1993; Novak & Greger, 1988a,b) indicate that the accumulation of intracellular HCO₃⁻ across the basolateral membrane of the duct cells may be achieved both by H⁺ extrusion via a Na⁺-H⁺ exchanger and/or H⁺-ATPase, and also by HCO₃⁻ uptake via a Na⁺-HCO₃⁻ cotransporter. HCO₃⁻ secretion across the luminal membrane is generally thought to involve the parallel operation of a Cl⁻-HCO₃⁻ exchanger and a cAMP-activated Cl⁻ channel that is believed to be the cystic fibrosis transmembrane conductance regulator (CFTR).

Uncertainty, however, still remains concerning the relative contributions of these various transport proteins to secretin-stimulated HCO₃⁻ secretion. A mathematical model of the pancreatic duct epithelium, based on data obtained from the rat (Sohma et al. 1996), predicts the secretion of a pancreatic juice with a HCO₃⁻ concentration of about 75 mM - as is observed in vivo in that species - but it cannot explain the much higher HCO₃⁻ concentrations achieved in many other species (e.g. guinea-pig, dog and human). There may therefore be significant species differences in the relative contributions of the different channels and transporters to secretin-evoked secretion.

In a previous paper (Ishiguro et al. 1996b), we described a new experimental protocol for detecting HCO₃⁻ secretion across the duct cell epithelium by monitoring the luminal pH of interlobular ducts isolated from the guinea-pig pancreas. In so doing, we demonstrated the presence of an apparently Cl⁻-independent, luminal efflux pathway for HCO₃⁻ which is stimulated by secretin. Furthermore we showed that, in secretin-stimulated ducts, transepithelial HCO₃⁻ transport requires extracellular (basolateral) Na⁺ but is not exclusively dependent on either a Na⁺-H⁺ exchanger or H⁺-ATPase activity. This was consistent with other data suggesting that Na⁺-HCO₃⁻ cotransport makes a substantial contribution to HCO₃⁻ uptake across the basolateral membrane (Ishiguro et al. 1996a).

Our working hypothesis, therefore, is that pancreatic duct cells in the guinea-pig, and probably other species, are able to raise the luminal HCO₃⁻ concentration to a higher value than in the rat (a) because accumulation of HCO₃⁻ across the basolateral membrane is driven by electrogenic Na⁺-HCO₃⁻ cotransport in addition to Na⁺-H⁺ exchange, and (b)because they have an additional, Cl⁻-independent mechanism for HCO₃⁻ efflux across the luminal membrane.

To evaluate the relative contributions of these various transporters to secretion, it is necessary to quantify the net secretory flux of HCO₃⁻. Measurements of luminal pH (pH_L) only provide values for the luminal HCO₃⁻ concentration and, therefore, only a qualitative indication of HCO₃⁻ secretion up to the time where pH_L reaches its steady-state value. A quantitative estimate of the HCO₃⁻ flux has to take into account the changes in luminal volume that result from fluid flow to the lumen. In this paper we describe a new method for measuring luminal volume (and therefore fluid secretory rate) and luminal pH simultaneously in isolated interlobular duct segments. This has allowed us to quantify HCO₃⁻ and fluid secretion, and examine the effects of Cl⁻ substitution and selective inhibition of apical and basolateral transporters. From these results we conclude that secretin-stimulated fluid secretion by guinea-pig pancreatic ducts involves the participation of a Na⁺-HCO₃⁻ cotransporter at the basolateral membrane and a largely Cl⁻-independent HCO₃⁻-efflux pathway at the luminal membrane.

METHODS

Isolation and culture of interlobular ducts

Female Hartley guinea-pigs (350–450 g) were killed by cervical dislocation. The body and tail of the pancreas were removed and interlobular ducts (diameter 100–150 μm) were isolated and cultured overnight as described previously (Ishiguro et al. 1996a).

Solutions

Bath solution

The standard Hepes-buffered solution contained (mM): 140 NaCl, 5 KCl, 1 CaCl₂, 1 MgCl₂, 10 D-glucose and 10 Hepes, and was equilibrated with 100 % O₂. The standard HCO₃⁻-buffered solution contained (mM): 115 NaCl, 5 KCl, 1 CaCl₂, 1 MgCl₂, 10 D-glucose and 25 NaHCO₃, and was equilibrated with 95 % O₂- 5 % CO₂. Cl⁻-free solutions were made by replacing Cl⁻ with glucuronate. All solutions were adjusted to pH 7.4 at 37°C.

Luminal injection solution

The standard solution injected into the duct lumen contained (mM): 139 NaCl, 5 KCl, 1 CaCl₂, 1 MgCl₂, 10 D-glucose and 1 Hepes. Cl⁻-free solutions were made by replacing Cl⁻ with glucuronate. These luminal injection solutions were adjusted to pH 7.2 at 37°C. The high-HCO₃⁻ solution contained (mM): 14 NaCl, 5 KCl, 1 CaCl₂, 1 MgCl₂, 10 D-glucose, 125 NaHCO₃ and 1 Hepes, and was equilibrated with 95 % O₂- 5 % CO₂ (pH 8.2) prior to microinjection. All luminal injection solutions also contained a dextran conjugate of the pH-sensitive fluoroprobe 2′7’-bis(2-carboxyethyl)-5(6)-carboxyfluorescein (BCECF-dextran, 70 kDa) at a concentration of 20 μM.

Measurement of luminal pH (pH_L)

During overnight culture both ends of the interlobular duct segments sealed spontaneously thus isolating the luminal space from the bathing medium. The lumen of the cultured interlobular ducts was micropunctured as described previously (Ishiguro et al. 1996b). The duct was first immobilized with a suction pipette within a 700 μl perfusion chamber on an inverted microscope (Olympus IX) and superfused at 3 ml min⁻¹ at 37°C. Thereafter the lumen was punctured with a double-barrelled (theta-glass) micropipette (tip diameter 10–12 μm) and a small volume of oil was injected. The pipette tip was pushed forward through the oil droplet and the luminal fluid was withdrawn and replaced by the injection solution containing BCECF-dextran. The puncture pipette was removed once the duct had been filled sufficiently to regain its cylindrical shape (Fig. 1A). Any possible leakage of the luminal solution at the puncture site was prevented by the oil droplet.

Figure 1. Bright-field and fluorescence images of an interlobular duct segment microinjected with BCECF-dextran.

A, bright-field image of a sealed duct segment after micropuncture and replacement of the luminal fluid with a solution containing BCECF-dextran. The oil droplet prevents leakage of the luminal solution after withdrawal of the pipette. The bar indicates 200 μm. B, the same duct after 10 min of stimulation with 10 nM secretin showing the increase in luminal volume resulting from fluid secretion. C and D, fluorescence images of the luminal space corresponding to A and B, respectively. E and F, binary images obtained by transformation of C and D, respectively.

The pH of the duct lumen (pH_L) was estimated by microfluorometry using a modification of the method described previously (Ishiguro et al. 1996b). Briefly, the ducts were illuminated alternately at excitation wavelengths of 430 and 480 nm. Luminal pH was calculated from the fluorescence ratio (F₄₈₀/F₄₃₀) measured at 530 nm. Calibration was performed in situ by varying the pH of the superfusate after first permeabilizing the ducts with digitonin (2.5 mM for 2 min).

Measurement of luminal volume and fluid secretory rate

Along with the measurement of pH_L, images of BCECF-dextran fluorescence were obtained at 1 min intervals using a CCD camera. The fluorescence images were subsequently transformed to binary images using ARGUS-50 software (Hamamatsu Photonics, Hamamatsu, Japan). Figure 1 shows bright-field images taken at the beginning of a representative experiment (Fig. 1A) and after stimulation with secretin (Fig. 1B). Panels C and D are the corresponding images of BCECF-dextran fluorescence from the lumen, and panels E and F are the binary images obtained from panels C and D. The threshold intensity for the transformation to the binary image was set automatically by the software.

In order to determine the secretory rate, the initial values for the length (L₀), diameter (2R₀) and projected area (A₀) of the duct lumen were measured in the first image of the series. The initial volume of the duct lumen (V₀) was calculated, assuming cylindrical geometry, as πR₀²L₀. The luminal surface area of the epithelium (E) was taken to be approximately 2πR₀L₀ (neglecting the ends of the cylinder since L₀ >> R₀). In subsequent images of the series, the luminal image area A was expressed as relative area (A/A₀). A representative experiment is illustrated in Fig. 2A. Relative volume (V/V₀) was calculated from relative area (A/A₀) using V/V₀= (A/A₀)^3/2. This assumes that R/R₀=L/L₀, i.e. that the duct swells radially and longitudinally by the same fraction. This assumption was supported by more detailed measurements of R and L in several duct image series. Measurements from the experiment in Fig. 2A are shown in Fig. 2B. The rate of fluid secretion was then calculated at 1 min intervals from the increment in duct volume and expressed as the secretory rate per unit luminal area of epithelium (nl min⁻¹ mm⁻²) (Fig. 2C).

Figure 2. Measurement and calibration of fluid secretory rate in interlobular ducts.

A, changes in the relative area (A/A₀) of the luminal space in a duct exposed first to HCO₃⁻/CO₂ and then stimulated with 10 nM secretin in Cl⁻-free conditions. B, changes in relative length (L/L₀, •) and relative radius (R/R₀, ○) corresponding to the changes in A/A₀ shown in A. C, changes in secretory rate calculated from the changes in A/A₀ shown in A. The values for R₀, L₀, E, V₀ and the final volume of this duct were 70 μm, 692 μm, 0.30 mm², 10.6 nl and 17.5 nl respectively. D, calibration of fluid secretory rate showing the relationship between the cumulative injected volume and the calculated increase in volume in three individual ducts. Initial volumes of the ducts were 5.2 nl (○), 8.6 nl (•) and 11.2 nl (▵). The straight line is the line of identity.

In order to validate the calculation of absolute duct volumes from fluorescence images, direct calibration of luminal volume was performed using a nanolitre-syringe driver (Sutter Instruments, San Francisco, CA, USA). In these experiments, the tip of the micropuncture pipette was retained in the duct lumen. The duct was superfused with a Hepes-buffered, HCO₃⁻-free solution (under which conditions no change of luminal volume was observed), and then 0.5 nl volumes of the BCECF-dextran solution were injected repeatedly. Figure 2D shows representative calibration data from three individual ducts in which the volume increase calculated from the change in the fluorescence image area is plotted against the cumulative injected volume. The data points lie close to the line of identity thus confirming the calibration of the system.

Since fluid secretion was negligible in the absence of HCO₃⁻, it was also possible to determine the coefficient of variation in the duct volumes obtained by this method from series of images of the same duct. For three ducts, each examined over a 5 min period, the values were 2.2, 0.3 and 0.9 %.

Materials

BCECF-dextran and dihydro-4,4′-diisothiocyanatostilbene-2,2′-disulphonic acid (H₂DIDS) were obtained from Molecular Probes; ouabain, forskolin, NaCN and sodium iodoacetate from Sigma; secretin (porcine) from Peptide Institute (Osaka, Japan); and N-methyl-N-isobutylamiloride (MIA) from Research Biochemicals International (Natick, MA, USA).

Statistics

Averaged data are presented as the means ±s.e.m. unless otherwise indicated. Tests for statistically significant differences were made with Student's t test for paired or unpaired data as appropriate.

RESULTS

Luminal pH and secretory rate in the presence of Cl⁻

The luminal fluid that had accumulated during overnight culture of the ducts was removed by micropuncture and replaced by a weakly buffered, HCO₃⁻-free, Cl⁻-rich solution containing the pH-sensitive fluoroprobe BCECF-dextran. The collapsed ducts were filled until they had just regained a cylindrical shape after which the puncture pipette was removed in order to isolate the duct lumen.

Ducts were initially superfused with a Hepes-buffered HCO₃⁻-free solution. During this equilibration period, pH_L did not change significantly and no fluid secretion was observed (0.01 ± 0.17 nl min⁻¹ mm⁻², mean ±s.e.m., n = 5; Fig. 3A and B). Switching the bath solution from the HCO₃⁻-free solution to one containing 25 mM HCO₃⁻ (equilibrated with 5 % CO₂) caused pH_L first to decline rapidly as a result of CO₂ diffusing from the bath to the luminal fluid. Thereafter pH_L increased and after 4 min it had reached 8.08 ± 0.06 and was still increasing. During this period, a constant secretory rate (2.02 ± 0.52 nl min⁻¹ mm⁻²) was observed. These results indicate that, with 148 mM Cl⁻ initially present in the lumen, there was a significant spontaneous secretion of HCO₃⁻-rich fluid. The net HCO₃⁻ flux, J_HCO3, which was calculated from the secretory rate J_v and the rate of change of pH_L (see Appendix), was 1.38 ± 0.15 nmol min⁻¹ mm⁻² during the final minute of this period. From the ratio of J_HCO3 and J_v, the effective concentration of HCO₃⁻ in the secreted fluid was estimated to be 790 ± 74 mM. Since salt-water coupling in the pancreatic duct is believed to generate a near-isosmotic fluid (Case et al. 1968) the HCO₃⁻ concentration in the secreted fluid would not be expected to exceed about 150 mM. Because the calculated value was so much greater than this, we conclude that more than 80 % of the HCO₃⁻ flux during the early part of the experiment was due to Cl⁻-HCO₃⁻ exchange and did not therefore generate any flow of fluid to the lumen.

Figure 3. Luminal pH and secretory rate in the presence of Cl⁻.

Ducts filled with a weakly buffered, HCO₃⁻-free, Cl⁻-rich (148 mM) solution containing 20 μM BCECF-dextran were initially superfused with a Hepes-buffered HCO₃⁻-free solution. After a 3 min period, the bath solution was switched to a solution containing 25 mM HCO₃⁻ equilibrated with 5 % CO₂. After a further 4 min, 10 nM secretin was applied. A, changes in luminal pH; one of five experiments. B, changes in secretory rate; means ±s.e.m. of all five experiments. One-way analysis of variance showed significance at the P < 0.005 level.

When 10 nM secretin was then added to the bath, pH_L increased further and reached 8.37 ± 0.06 after 8 min. The fluid secretory rate increased by a further 1.50 ± 0.19 nl min⁻¹ mm⁻² to a value of 3.52 ± 0.41 nl min⁻¹ mm⁻² and then remained constant during the 8 min period of secretin stimulation (Fig. 3A and B). The HCO₃⁻ flux calculated for the final minute of this period was 0.96 ± 19 nmol min⁻¹ mm⁻². By this stage, the effective concentration of HCO₃⁻ in the secreted fluid was 277 ± 26 mM, indicating that at least 40 % of the HCO₃⁻ flux was probably still the result of Cl⁻-HCO₃⁻ exchange rather than net secretion.

In four similar experiments, spontaneous fluid secretion was followed for 20 min before secretin was applied (Fig. 4B). As in Fig. 3B, the secretory rate increased rapidly when the bath solution was switched from the Hepes solution to the HCO₃⁻/CO₂ solution. However, while the luminal pH remained at a high value (Fig. 4A), the rate of spontaneous fluid secretion declined slowly over the 20 min period from 1.93 ± 0.26 nl min⁻¹ mm⁻² during the first 5 min to 1.28 ± 0.12 nl min⁻¹ mm⁻² in the final 5 min (P < 0.05). Despite the accumulation of fluid in the lumen during this period, stimulation with 10 nM secretin was still able to evoke secretory rates comparable with the highest values obtained in the shorter experiments.

Figure 4. Time course of luminal pH and secretory rate during spontaneous secretion in the presence of Cl⁻.

Under the same experimental conditions as in Fig. 3, changes in luminal pH and fluid secretory rate were monitored for 20 min in the HCO₃⁻/ CO₂ bath solution before secretin (10 nM) was applied. A, changes in luminal pH; one of four experiments. B, changes in secretory rate; means ±s.e.m. of all four experiments.

Figure 7A (•) shows the concentration-response curve for the effects of secretin on the fluid secretory rate. The concentration of secretin required for a half-maximal increase in fluid secretion was approximately 0.9 nM (log[secretin]=−9.07 ± 0.94). Stimulation with forskolin (1 μM) also increased the fluid secretory rate, to 2.83 ± 0.17 nl min⁻¹ mm⁻² (n = 5; Fig. 7B), supporting our previous observation that an elevation of the intracellular cAMP concentration is sufficient to stimulate a full secretory response.

Figure 7. Effects of secretin and forskolin on fluid secretion.

A, concentration-response curve for the effects of secretin on the secretory rate in isolated interlobular ducts in the presence (•) and absence (○) of Cl⁻. Data are shown as the means ±s.e.m. of at least five experiments. One-way analysis of variance showed significant variations with concentration at the P < 0.0001 level in both sets of data. B, secretory rate in the absence of HCO₃⁻ (HCO₃⁻-free), under unstimulated conditions (spontaneous), during stimulation with secretin (10 nM) and in the presence of forskolin (1 μM). ▪, secretory rate in the presence of Cl⁻; □, secretory rate in the absence of Cl⁻. Means ±s.e.m. of at least five experiments.

Luminal pH and secretory rate in the nominal absence of Cl⁻

Our previous study demonstrated that ducts stimulated with secretin could alkalinize the luminal fluid, presumably as a result of HCO₃⁻ secretion, in the virtual absence of luminal Cl⁻ (Ishiguro et al. 1996b). The next experiment was designed to investigate whether fluid secretion could also be observed under these conditions. Ducts were superfused first with a HCO₃⁻-free, Cl⁻-free solution (in which all of the Cl⁻ was replaced by glucuronate) for at least 15 min. The lumen was then punctured and filled with a weakly buffered, Cl⁻-free solution containing BCECF-dextran. Switching the bath solution to a HCO₃⁻-buffered, Cl⁻-free solution acidified the lumen as before. But under these conditions (with Cl⁻ absent from both basolateral and luminal solutions) pH_L recovered only partially (Fig. 5A), and after 4 min the value was still less than 7.4 (7.30 ± 0.04, n = 5). The rate of fluid secretion (0.47 ± 0.17 nl min⁻¹ mm⁻²; Fig. 5B) was also significantly smaller than in the presence of Cl⁻. The HCO₃⁻ flux was greatly diminished under these conditions (0.22 ± 0.02 nmol min⁻¹ mm⁻²). These results therefore suggest that the spontaneous secretion of HCO₃⁻ and fluid into the lumen of the unstimulated ducts is largely dependent on the presence of Cl⁻.

Figure 5. Luminal pH and secretory rate in the absence of Cl⁻.

Ducts filled with a weakly buffered, HCO₃⁻-free, Cl⁻-free solution (replaced with glucuronate) containing 20 μM BCECF-dextran were initially superfused with a Hepes-buffered, HCO₃⁻-free, Cl⁻-free solution. After a 3 min equilibration period, the bath solution was switched to a solution containing 25 mM HCO₃⁻ equilibrated with 5 % CO₂. After a further 4 min, 10 nM secretin was applied. A, changes in luminal pH; one of five experiments. B, changes in secretory rate; means ±s.e.m. of all five experiments.

In the absence of extracellular Cl⁻, 10 nM secretin still caused pH_L to increase rapidly, reaching 8.27 ± 0.09 after 8 min (Fig. 5A). The secretory rate also increased by 1.54 ± 0.22 nl min⁻¹ mm⁻² (Fig. 5B). Thus, the absence of Cl⁻ did not affect the final value of pH_L or the increase in fluid secretory rate following secretin stimulation. The HCO₃⁻ flux during the final minute of this period was 0.38 ± 0.04 nmol min⁻¹ mm⁻². The latter value is somewhat smaller than that obtained during stimulation in the presence of Cl⁻. However, the effective concentration of HCO₃⁻ in the secreted fluid in the absence of Cl⁻ was estimated to be 167 ± 16 mM, confirming that the calculated HCO₃⁻ flux probably reflected a true secretory flux, accompanied by water in near-isosmotic proportions, with relatively little, if any, contribution from anion exchange.

The low rate of spontaneous secretion that was observed when the bath solution was switched to the HCO₃⁻/CO₂ solution in the absence of Cl⁻ continued for 20 min in the longer experiment shown in Fig. 6B. While the luminal pH continued to rise slowly during this period (Fig. 6A), there was a slow decline in the secretory rate (from 0.48 ± 0.10 nl min⁻¹ mm⁻² in the first 5 min to 0.32 ± 0.07 nl min⁻¹ mm⁻² in the last 5 min; P≈ 0.05) as also observed in the presence of Cl⁻ (Fig. 4B). The secretory response to 10 nM secretin, however, was undiminished compared with the shorter experiments illustrated in Fig. 5.

Figure 6. Time course of luminal pH and secretory rate during spontaneous secretion in the absence of Cl⁻.

Under the same experimental conditions as in Fig. 5, changes in luminal pH and fluid secretory rate were monitored for 20 min in the HCO₃⁻/ CO₂ bath solution before secretin (10 nM) was applied. A, changes in luminal pH; one of four experiments. B, changes in secretory rate; means ±s.e.m. of all four experiments.

Figure 7A (○) shows the concentration-response curve for the effects of secretin on the fluid secretory rate. The concentration of secretin required for a half-maximal increase in fluid secretion was approximately 0.3 nM (log[secretin]= (−9.49 ± 0.45), which was not significantly different from the value obtained in the presence of Cl⁻. Stimulation with forskolin (1 μM) also increased the fluid secretory rate in the absence of extracellular Cl⁻, to 2.21 ± 0.39 nl min⁻¹ mm⁻² (n = 5; Fig. 7B), although this was not significantly different from the rate observed in the presence of Cl⁻. These results indicate that both secretin and forskolin can evoke the secretion of a HCO₃⁻-rich fluid in the nominal absence of Cl⁻ from the bath and with, at most, a very low concentration of Cl⁻ in the lumen (as judged by our previous measurements of the luminal Cl⁻ concentration under comparable conditions; Ishiguro et al. 1996b).

Luminal pH and fluid secretory rate with high luminal HCO₃⁻

Although the HCO₃⁻ concentration in the lumen of guinea-pig interlobular ducts in vivo is not known, the concentration of HCO₃⁻ in the secretion collected from the main pancreatic duct of anaesthetized guinea-pigs can reach 150 mM during secretin stimulation (Padfield et al. 1989). Therefore, to study the flux of HCO₃⁻ and water into the lumen under more physiological conditions, in the following experiments the duct lumen was filled with a solution initially containing 125 mM HCO₃⁻ and 23 mM Cl⁻ (pH 8.2 when equilibrated with 5 % CO₂). Figure 8 shows the changes in pH_L and fluid secretory rate under such conditions. Prior to the application of secretin, there was a slight decrease in pH_L (from 8.12 ± 0.05 to 8.04 ± 0.03, P < 0.05) accompanied by a sustained secretion of fluid at a rate of 1.03 ± 0.20 nl min⁻¹ mm⁻² (n = 4). The calculated HCO₃⁻ flux in the final minute of this control period was very small (0.02 ± 0.01 nmol min⁻¹ mm⁻²) and the estimated concentration of HCO₃⁻ in the secreted fluid was only 24 ± 9 mM. This suggests either that the principal anion in the spontaneously secreted fluid was Cl⁻ rather than HCO₃⁻, or that spontaneous HCO₃⁻ secretion was accompanied by a net exchange of luminal HCO₃⁻ for cytosolic Cl⁻.

Figure 8. Luminal pH and secretory rate with high luminal HCO₃⁻.

Ducts filled with a BCECF-dextran solution containing 125 mM HCO₃⁻ and 23 mM Cl⁻ (pH 8.2) were superfused with the standard HCO₃⁻-buffered solution and 10 nM secretin was applied after 5 min. A, changes in luminal pH; one of four experiments. B, changes in fluid secretory rate; means ±s.e.m. of all four experiments.

Upon application of 10 nM secretin, pH_L rose to 8.21 ± 0.04 and the fluid secretory rate increased to 1.89 ± 0.33 nl min⁻¹ mm⁻². The calculated HCO₃⁻ flux at the end of this period was 0.32 ± 0.07 nmol min⁻¹ mm⁻² and the concentration of HCO₃⁻ in the secreted fluid was estimated to be 149 ± 13 mM, indicating a minimal contribution from anion exchange. It also appears that, under these relatively physiological conditions, the composition of the secreted fluid changed upon stimulation with secretin from being Cl⁻-rich to HCO₃⁻-rich.

Inhibition of basolateral membrane transporters and metabolism

The methodology described above allows us to continuously monitor fluid and HCO₃⁻ secretion in isolated interlobular pancreatic ducts. Using this methodology, we have also studied the effects of transport inhibitors added either basolaterally (i.e. to the bathing fluid) or to the luminal fluid.

In pancreatic ducts of rat and cat, Na⁺,K⁺-ATPase activity is restricted to the basolateral membrane (Bundgaard et al. 1981; Madden & Sarras, 1987). Ouabain, a specific inhibitor of Na⁺,K⁺-ATPase strongly inhibits fluid secretion in perfused cat pancreas (Case & Scratcherd, 1974) and in isolated rabbit pancreas (Kuijpers et al. 1984). Consequently, HCO₃⁻ secretion in pancreatic duct cells is thought to require the indirect participation of Na⁺,K⁺-ATPase in order to generate a sodium gradient which favours the operation of Na⁺-H⁺ exchange and Na⁺-HCO₃⁻ cotransport.

In the experiment shown in Fig. 9A, 1 mM ouabain was added to the bath solution during stimulation with 10 nM secretin in the presence of Cl⁻ (the same conditions as in Fig. 3). After 7 min, ouabain had reduced the secretory rate by 74 %: from 3.50 ± 0.15 to 0.92 ± 0.11 nl min⁻¹ mm⁻² (n = 4). In the experiments shown in Fig. 9B, intracellular ATP was depleted by adding both cyanide, an inhibitor of cytochrome oxidase, and iodoacetate, a glycolytic inhibitor, to the perfusate during stimulation with 10 nM secretin in the absence of Cl⁻ (the same conditions as in Fig. 5). The combination of 2 mM cyanide and 0.1 mM iodoacetate caused the secretory rate to fall by 83 %: from 2.18 ± 0.31 to 0.36 ± 0.04 nl min⁻¹ mm⁻² (n = 4). These results show that secretin-evoked fluid secretion across the pancreatic duct epithelium requires the activity of Na⁺,K⁺-ATPase and intracellular ATP as an energy source.

Figure 9. Effects of ouabain and ATP depletion on fluid secretion.

A, the effect of 1 mM ouabain on fluid secretory rate. Ducts filled with a Cl⁻-containing BCECF-dextran solution were superfused with the standard HCO₃⁻-buffered solution and then stimulated with 10 nM secretin; mean ±s.e.m. of four experiments. B, the effect of 2 mM cyanide and 0.1 mM iodoacetate on secretory rate. Ducts filled with a Cl⁻-free BCECF-dextran solution were superfused with a Cl⁻-free HCO₃⁻-buffered solution and stimulated with 10 nM secretin; means ±s.e.m. of four experiments.

In a previous paper (Ishiguro et al. 1996a) we proposed (on the basis of a variety of evidence) that the accumulation of HCO₃⁻ across the basolateral membrane of the guinea-pig pancreatic duct cell was achieved largely (about 75 %) through the activity of a Na⁺-HCO₃⁻ cotransporter which could be inhibited by H₂DIDS. In order to evaluate the contribution of the cotransporter to fluid secretion, we have now examined the effect of basolateral application of H₂DIDS on secretory rate. These experiments were carried out in the absence of Cl⁻ using the same protocol as those shown in Fig. 5. During stimulation with 10 nM secretin, 500 μM H₂DIDS was added to the bath solution for a period of 4 min (Fig. 10A). After 2 min of H₂DIDS application, the secretory rate had decreased by 56 ± 12 % (n = 5, P < 0.01). However, after 4 min of application, the secretory rate appeared to have recovered partially and the reduction was only 26 ± 18 %, which was not significantly different from the value prior to the application of H₂DIDS. Assuming that inhibition of the Na⁺-HCO₃⁻ cotransporter was complete, this suggests that a H₂DIDS-insensitive transport mechanism, most probably Na⁺-H⁺ exchange, was supporting HCO₃⁻ accumulation across the basolateral membrane under these conditions.

Figure 10. Effects of basolateral inhibitors on fluid secretion.

Ducts filled with a Cl⁻-free BCECF-dextran solution were superfused with a Cl⁻-free HCO₃⁻-buffered solution and then stimulated with 10 nM secretin. A, the effect of basolateral H₂DIDS (500 μM) on secretory rate; mean ±s.e.m. of five experiments. B, the effect of basolateral MIA (10 μM) and H₂DIDS (500 μM) on secretory rate; means ±s.e.m. of four experiments.

In another experiment (Fig. 10B), 10 μM MIA, a specific and potent inhibitor of Na⁺-H⁺ exchange (Dixon et al. 1987), was added to the bath to inhibit the basolateral exchanger. MIA had rather less effect on secretory rate than H₂DIDS and only reduced the secretory rate by 18 ± 4 % (n = 4, P < 0.05). However, when 500 μM H₂DIDS was also added to the bath, the combined effect of MIA and H₂DIDS was to completely block fluid secretion. Furthermore, when H₂DIDS but not MIA was removed from the bath, the secretory rate showed a rapid and substantial recovery. These results support our hypothesis that the H₂DIDS-sensitive Na⁺-HCO₃⁻ cotransporter makes a larger contribution to the basolateral accumulation of HCO₃⁻ in secretin-stimulated ducts than the MIA-sensitive Na⁺-H⁺ exchanger.

De Ondarza & Hootman (1997) have recently demonstrated the participation of a vacuolar-type H⁺-ATPase in the recovery of intracellular pH following an acid load in guinea-pig pancreatic duct cells. This transporter, working in concert with carbonic anhydrase, could theoretically contribute to the accumulation of intracellular HCO₃⁻ and thus to HCO₃⁻ and fluid secretion. In order to assess this possibility, we examined the effect of bafilomycin A₁, a specific inhibitor of vacuolar H⁺-ATPases, on secretory rate. When applied at a concentration of 1 μM during stimulation with 10 nM secretin (in the presence of Cl⁻), the secretory rate was 3.41 ± 0.22 nl min⁻¹ mm⁻² (n = 4). This was not significantly different from the rate prior to the addition of bafilomycin A₁ (3.45 ± 0.21 nl min⁻¹ mm⁻², n = 4) and it suggests that the contribution of this transporter to secretion is probably small.

Inhibition of luminal membrane transporters

To study the luminal transport mechanisms involved in HCO₃⁻ secretion, we also introduced transport inhibitors into the luminal space. In our preliminary attempt to identify these mechanisms (Ishiguro et al. 1996b), three anion channel blockers were tested for their ability to inhibit the increase in pH_L following stimulation with forskolin. In Cl⁻-free conditions we found that luminal glibenclamide, H₂DIDS and 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB) all failed to inhibit the increase in pH_L following stimulation. These results suggested the existence of a novel HCO₃⁻ efflux pathway at the luminal membrane. However, it was also evident that, when Cl⁻ was present in the lumen, alkalinization occurred spontaneously even in the absence of stimulation. We suggested that this spontaneous component of HCO₃⁻ secretion might be mediated by Cl⁻-HCO₃⁻ exchange; if so, stilbenes such as H₂DIDS would be expected to block it. To test this prediction, we have now investigated the effect of H₂DIDS on luminal alkalinization and secretory rate.

In the presence of Cl⁻, H₂DIDS (500 μM) injected into the duct lumen markedly inhibited the increase in pH_L when the bath solution was switched to the HCO₃⁻-buffered solution (Fig. 11A). Luminal pH increased only modestly to 7.34 ± 0.05 (n = 5) after 4 min compared with 8.00 ± 0.05 (n = 4; P < 0.001) in the absence of H₂DIDS. Over the same time period, luminal H₂DIDS reduced the secretory rate by 59 ± 21 % compared with the controls (P < 0.05, Fig. 11B). These results indicate that, in the presence of Cl⁻, a significant fraction of the spontaneous HCO₃⁻ efflux and the flow of fluid to the lumen is sensitive to H₂DIDS and therefore probably involves Cl⁻-HCO₃⁻exchange.

Figure 11. Effects of luminal H₂DIDS on fluid secretion.

A and B, ducts filled with a Cl⁻-containing BCECF-dextran solution with or without H₂DIDS (500 μM) were superfused with the standard HCO₃⁻-buffered solution and then stimulated with 10 nM secretin. A, the effect of luminal H₂DIDS (500 μM) on luminal pH. Control data are represented by the thin line, and data from ducts containing H₂DIDS by the thick line. Mean traces of five and four experiments respectively. B, the effect of luminal H₂DIDS (500 μM) on secretory rate. ▪, control data; □, data from ducts containing H₂DIDS. The secretin-stimulated component of the secretory rate is the difference between the secretin-evoked and spontaneous secretory rates. Means ±s.e.m. of five and four experiments respectively. * Statistical differences (P < 0.05). C and D, ducts filled with a Cl⁻-free BCECF-dextran solution with or without H₂DIDS (500 μM) were superfused with the Cl⁻-free HCO₃⁻-buffered solution and stimulated with 10 nM secretin. C, the effect of luminal H₂DIDS (500 μM) on luminal pH. Control data are represented by the thin line, and data from ducts containing H₂DIDS by the thick line. Mean traces of five and four experiments respectively. D, the effect of luminal H₂DIDS (500 μM) on secretory rate. ▪, control data; □, data from ducts containing H₂DIDS. Means ±s.e.m. of five and four experiments respectively.

When the ducts containing H₂DIDS were then stimulated with secretin, the rise in pH_L became more rapid and after 8 min pH_L had approached a steady-state value that was not significantly lower than in the secretin-stimulated control ducts (Fig. 11A). Furthermore, the increase in fluid secretion evoked by secretin (Fig. 11B) was only reduced by 33 ± 16 % (P < 0.05) by luminal H₂DIDS.

In the absence of Cl⁻, neither the rise in pH_L nor the secretory rate was affected by luminal H₂DIDS under resting or secretin-stimulated conditions (Fig. 11C and D). Furthermore the pattern of changes in pH_L observed in Cl⁻-free conditions (with or without H₂DIDS) was similar to those observed in Cl⁻-containing conditions when H₂DIDS was present in the lumen (thick line, Fig. 11A). These results support our hypothesis that, while spontaneous secretion is Cl⁻-dependent and inhibited by H₂DIDS, and therefore probably involves Cl⁻-HCO₃⁻ exchange, the secretion of HCO₃⁻-rich fluid evoked by secretin in the nominal absence of Cl⁻ is not inhibited by H₂DIDS and must therefore be mediated by a novel channel or transporter.

DISCUSSION

Measurement of luminal pH and secretory rate

The rate of fluid secretion by the pancreatic duct epithelium (uncontaminated by acinar fluid) was first measured by Argent et al. (1986) in interlobular ducts isolated from rat pancreas. Using a micropuncture technique, they collected total fluid output over 1 h periods and determined the mean secretory rate under a variety of experimental conditions. To examine ductal HCO₃⁻ secretion in guinea-pig pancreatic ducts, we modified the micropuncture technique and used it to inject pH-sensitive BCECF-dextran into the luminal space thereby allowing changes in luminal pH to be measured (Ishiguro et al. 1996b). The limitations of this approach were that (a) it was not possible to quantify HCO₃⁻ secretion since changes in luminal volume were not taken into account, and (b) we could not determine whether the flux of HCO₃⁻ into the lumen was due to net secretion or due to exchange with luminal Cl⁻ via a Cl⁻-HCO₃⁻ exchanger. To overcome these problems, in the present study we have used fluorescence images of BCECF-dextran in the luminal space to monitor simultaneously changes in luminal volume (and hence fluid secretion) and luminal pH (which reflects the HCO₃⁻ flux).

Validity of the experimental model

When isolated pancreatic ducts were stimulated with secretin or forskolin for 15 min, the luminal volume usually increased by 40–50 % in total. The fluid secretory rate during this period was quite constant, indicating that the accompanying rise in luminal hydrostatic pressure - which would tend to oppose duct swelling - was negligible. We routinely stopped the experiments after 15 min because, by that stage, the ionic composition of the luminal solution would probably differ significantly from that of the solution initially injected. However, in the longer experiments shown in Fig. 4, where secretin was applied after 20 min of spontaneous secretion, the response to secretin was undiminished despite the substantial accumulation of luminal fluid during the control period. Other studies, in which changes in luminal volume have been induced osmotically, have shown that the volume of isolated ducts can double without any significant restriction due to limited compliance (M. C. Steward & R. M. Case, in preparation).

Our assessment of the validity of this preparation as an experimental model for pancreatic duct fluid and HCO₃⁻ secretion is based on the following observations. First, fluid secretion was almost completely blocked by ouabain or by depletion of intracellular ATP. This shows that fluid secretion was achieved by primary or secondary active ion transport, and not by passive diffusion or osmotic imbalance. Secondly, secretin caused a concentration-dependent increase in fluid secretion which was mimicked by forskolin, an activator of adenylate cyclase. The concentration of secretin required for a half-maximal increase in fluid secretion was approximately 0.9 nM - similar to the K_m for cyclic AMP production (2 nM) evoked by secretin in the same preparation (Fayeed et al. 1995).

Luminal mechanisms for HCO₃⁻ secretion in unstimulated ducts

In the anaesthetized guinea-pig a significant amount of spontaneous fluid secretion is observed and the concentration of HCO₃⁻ in the juice (ca. 95 mM) is moderately high (Padfield et al. 1989). In our experiments on isolated duct segments, spontaneous fluid secretion was observed when HCO₃⁻/CO₂ was added to the perfusate but the rate depended upon the initial ionic composition of the luminal fluid. When the luminal fluid initially contained 148 mM Cl⁻, spontaneous fluid secretion was fairly rapid, being equivalent to 57 % of the maximal secretory rate obtained with secretin (Fig. 3). At the same time pH_L increased to a value above pH 8.0. This clearly indicates that spontaneous secretion is dependent on the presence of HCO₃⁻/CO₂ in the bath and that the secreted fluid is itself rich in HCO₃⁻ - consistent with the results of the in vivo studies. It is interesting to note, however, that in the isolated ducts the rate of spontaneous secretion in the presence of a high luminal Cl⁻ concentration is much higher, relative to the secretin-evoked response, than is observed in vivo where the luminal Cl⁻ concentration is much lower (Padfield et al. 1989). In agreement with this interpretation, we observed that the spontaneous secretion declines with time (Fig. 4), presumably as the initially high luminal Cl⁻ concentration falls, whereas the response to secretin is maintained.

When the Cl⁻ in the luminal fluid was replaced with glucuronate, both the secretory rate and the increase in pH_L were greatly reduced (Fig. 5). With a more physiological luminal fluid, initially containing 23 mM Cl⁻ and 125 mM HCO₃⁻, the situation was similar. We therefore conclude that, in our experimental model, the spontaneous secretion of HCO₃⁻-rich fluid is dependent on a moderately high luminal Cl⁻ concentration. Furthermore, it is almost completely blocked by luminal H₂DIDS (Fig. 11). This suggests that the luminal membrane mechanism responsible for spontaneous secretion may be analogous to that proposed for the rat pancreatic duct, namely a H₂DIDS-blockable Cl⁻/HCO₃⁻ exchanger linked to a Cl⁻ conductance (Gray et al. 1988; Novak & Greger, 1988b). In ducts containing 148 mM Cl⁻, the initially steep gradients for Cl⁻ and HCO₃⁻ across the luminal membrane accelerate the passive exchange of these ions thus accounting for the rapid increase in pH_L. The recycling of Cl⁻ across the luminal membrane via the Cl⁻ conductance, driven by the outward electrochemical gradient for Cl⁻, results in net solute transfer to the lumen and thus net fluid secretion.

Luminal mechanisms for HCO₃⁻ secretion in secretin-stimulated ducts

Although Cl⁻-free conditions inhibited responses in unstimulated ducts, the secretin-evoked increase in secretory rate and the final value of pH_L were hardly altered when Cl⁻ was replaced by glucuronate (Figs 3 and 5). This is in contrast to results from micropunctured rat ducts (Ashton et al. 1991) where secretin-evoked fluid secretion was reduced by 70 % when the Cl⁻ in the bath solution was replaced by gluconate.

The secretion of HCO₃⁻-rich fluid in response to secretin in the guinea-pig ducts was also unaffected by luminal H₂DIDS under these conditions (Fig. 11). Although the mechanism by which HCO₃⁻ is transported into the lumen in the absence of Cl⁻ is still unknown, our present data strongly suggest that this is the dominant pathway for luminal HCO₃⁻ transport in secretin-stimulated guinea-pig ducts.

Although the magnitude of the secretin-evoked component of fluid secretion in the guinea-pig ducts (shown in Fig. 11) appeared to be unaffected by replacement of Cl⁻, we nonetheless obtained some evidence of a contribution from luminal Cl⁻-HCO₃⁻ exchange when Cl⁻ was present. Under these conditions, the secretin-stimulated increase in fluid secretion was partially but significantly inhibited (33 %) by luminal H₂DIDS (Fig. 11B). This suggests that the luminal anion exchanger may play a role in secretin-stimulated fluid secretion when Cl⁻ concentrations permit, but is not an essential component of the mechanism. This would be consistent with our own preliminary observations suggesting that secretin appears to activate Cl⁻-HCO₃⁻ exchange in guinea-pig pancreatic ducts (Ishiguro et al. 1995) - an observation which has also been reported in bile duct epithelial cells (Alvaro et al. 1993).

Accumulation of intracellular HCO₃⁻ across the basolateral membrane

The data in the present paper support our hypothesis (Ishiguro et al. 1996b) that intracellular accumulation of HCO₃⁻ in secretin-stimulated ducts is achieved largely by HCO₃⁻ uptake via a H₂DIDS-sensitive Na⁺-HCO₃⁻ cotransporter and to a lesser extent by H⁺ extrusion via an MIA-sensitive Na⁺-H⁺ exchanger. When added to the bath separately, H₂DIDS and MIA reduced the secretory rate during stimulation with secretin by 56 and 18 %, respectively. When applied simultaneously, however, they completely blocked secretion. The observation that the inhibitory effects of the two blockers are not additive suggests that inhibition of one pathway may lead to a compensatory increase in the activity of the other. This may simply be a consequence of sharing the same inward Na⁺ gradient as their primary driving force - a reduction in the Na⁺ influx through one transporter will increase the Na⁺ gradient driving the other one.

Although the existence of a vacuolar-type H⁺-ATPase has been demonstrated in rat, pig and also guinea-pig pancreatic duct cells (Zhao et al. 1994; Villanger et al. 1995; De Ondarza & Hootman, 1997), the lack of effect of bafilomycin A₁ on secretin-evoked fluid secretion, as compared with the complete inhibition that we observed when MIA and H₂DIDS were applied simultaneously, suggests that the contribution of the H⁺-ATPase is minimal in this species.

In summary, we have developed an experimental protocol for monitoring HCO₃⁻ and fluid secretion simultaneously in isolated interlobular duct segments from guinea-pig pancreas. The results obtained using this new methodology support our hypothesis that, in secretin-stimulated pancreatic duct cells, HCO₃⁻ uptake across the basolateral membrane is achieved largely by Na⁺-HCO₃⁻ cotransport, and that HCO₃⁻ efflux across the luminal membrane, and hence fluid secretion, may involve a novel, Cl⁻-independent mechanism.

APPENDIX

Calculation of the secretory HCO₃⁻ flux

The HCO₃ content of the duct lumen Q is given by:

where C is the luminal concentration of HCO₃ and V is the luminal volume. Since both C and V change with time, the rate of change of the HCO₃ content of the duct lumen is given by:

since

where J_HCO3 and J_v are the fluxes of bicarbonate and fluid volume expressed per unit area of epithelium (luminal membrane), we obtain:

To calculate dC/dt from the rate of change of luminal pH (dpH_L/dt) we use:

If we assume that the luminal concentration of H₂CO₃ is constant, i.e. that there is rapid exchange of CO₂ across the duct wall and sufficient carbonic anhydrase activity in the lumen, then the HCO₃/CO₂ buffering system acts as an open buffer with buffering capacity:

This is obtained by differentiating the Henderson- Hasselbalch equation:

where pK’ is the effective pK for the CO₂-H₂CO₃-HCO₃⁻ buffering system (6.1 at 37°C) and s is the solubility of CO₂ (3 × 10⁻⁵ mol l⁻¹ mmHg⁻¹ at 37°C). Rearranging this gives:

Differentiating at constant P_CO2, and neglecting the small amount of buffering capacity due to the luminal Hepes (which will be minimal at the higher pH_L values reached in these experiments), gives:

Thus:

Hence the flux of HCO₃ can be calculated using:

and the effective HCO₃ concentration of the secreted fluid is J_HCO3/J_v.