Regulation of Exocytosis by Protein Kinases and Ca²⁺ in Pancreatic Duct Epithelial Cells

In neurons and some endocrine cells, Ca²⁺ plays a pivotal role as the final signal for rapid stimulus-evoked release of neurotransmitters and hormones. An elevation of intracellular Ca²⁺ will also trigger exocytosis in various nonexcitable cells (Morimoto et al. 1995; Coorssen et al. 1996). Nevertheless, although the machinery of exocytosis uses related universal molecular components, Ca²⁺ may not be the physiological signal for exocytosis in all nonneuronal cells. For example, in neutrophils, eosinophils, and mast cells, the intracellular signal for exocytosis and degranulation seems not to be Ca²⁺, protein kinase A, or protein kinase C (Neher and Almers 1986; Almers and Neher 1987; Scepek et al. 1998). On the other hand, the hormonal regulation of secretion in various gastrointestinal and airway epithelial cells is described as using cAMP or PKC, depending on the stimulating hormone (Forstner et al. 1993; Larivee et al. 1994; Klinkspoor et al. 1996; Oda et al. 1996; Abdullah et al. 1997; Urushidani and Forte 1997; Brown et al. 1998; Fujita-Yoshigaki 1998; reviewed in Hille et al. 1999). To compare this kind of exocytosis with that in neurons, we decided to investigate properties of exocytosis in cultured dog pancreatic duct epithelial cells using amperometry. In these cells, roles for cAMP and Ca²⁺ in mucin secretion are known (Oda et al. 1996; Nguyen et al. 1998). The principal questions were whether PKA, PKC, and Ca²⁺ all can trigger secretion in one cell type and whether they act independently.

Stock solutions of 20 mM forskolin, 20 mM H-89, 100 μM phorbol-12-myristate-13-acetate, and 500 μM bisindolylmaleimide I (BIS) were prepared in dimethyl sulfoxide. Stock solutions of 1 mM epinephrine, 1 mM vasoactive intestinal peptide (VIP), and 1 M cAMP analogues were made up in saline solution. The cAMP analogues included 8-Br-cAMP and the Rp and Sp optical enantiomers of 8-Br-cAMPS. Rp-8-Br-cAMPS, Sp-8-Br-cAMPS, H-89, PMA, and BIS were purchased from Calbiochem. VIP was from Bachem. Serotonin-HCl and dopamine-HCl were from RBI. Other chemicals were from Sigma-Aldrich.

The derivation and culture of pancreatic epithelial cells from the main pancreatic duct of the dog has been described (Oda et al. 1996). Although the cells are not a transformed cell line, they can be repeatedly subcultured. In brief, the cells were cultured in Vitrogen (Collagen)-coated Transwell inserts (3-μm pore size, 24-mm diameter; Corning Costar), and the inserts were placed above a confluent feeder layer of human gallbladder myofibroblasts. The cells were maintained in Eagle's minimum essential medium with 10% fetal bovine serum plus insulin-transferrin-sodium selenite medium supplement from Sigma Chemical Co. When confluent, the cells were treated with 2.5% trypsin/EDTA at 37°C for 30 min and passaged (10⁵ cells per well) to new coated inserts with fresh feeder layers.

For single-cell experiments, cells were plated on small coverglass chips (5 × 5 mm) coated with a thick layer of Vitrogen. The cells were fed with medium conditioned by gallbladder myofibroblasts. Experiments were performed with cells between 2 and 7 d after plating. Although the epithelial cells we use can form well differentiated monolayers with apical-basolateral polarity, our experiments were carried out on subconfluent epithelial cells that were presumably not completely differentiated or polarized.

We used carbon-fiber amperometry to detect exocytosis from single cells in real time (Kawagoe et al. 1993; Chow and von Rüden 1995). Amperometry provides high resolution to detect molecules released from single secretory vesicles and stability for long recordings. Since the method measures only exocytosis, it can detect infrequent exocytotic events without being disturbed by concurrent endocytosis. Amperometric measurements are normally limited to cells that package and secrete an endogenous oxidizable molecule such as catecholamines or serotonin; however, in some cases oxidizable molecules can be introduced artificially. Thus, Smith et al. 1995 and Zhou and Misler 1996 demonstrated the feasibility of loading serotonin exogenously via the serotonin transporter into pancreatic β cells by incubating overnight with 0.5–1-mM concentrations of serotonin. We have found that soaking a variety of cells in 50–100× higher concentrations of amines such as dopamine or serotonin will force the exogenous molecules to distribute passively into cytoplasm and acidic secretory vesicles (Billiard et al. 1997; Koh et al. 1997; Kim et al. 2000). This loading process does not need amine transporters in the plasma and vesicular membranes and therefore can be used in principle for most cell types, but the exogenous molecules probably end up in a range of types of intracellular acidic compartments.

For loading, cells were incubated for 40 min at room temperature in a solution containing (mM): 70 serotonin or dopamine, 68 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 10 d-glucose, and 10 HEPES, pH 7.3 with NaOH. The cells were then transferred to culture medium and kept at 37°C, 5% CO₂ for up to 4 h. Experiments were performed in a monoamine-free saline solution containing (mM): 138 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 10 d-glucose, and 10 HEPES, pH 7.3 with NaOH.

Vesicular release of catecholamine was monitored as pulses of electric current generated by oxidation of the molecules at the tip of a carbon-fiber electrode polarized at +600 mV. Carbon-fiber microelectrodes were fabricated from 5–11-μm carbon fibers (PAN T650 or P25; Amoco Performance Products) and polypropylene 10 μl micropipettor tips as described by Koh and Hille 1999. Application of +600 mV to a fresh electrode in the bath elicited an initial transient polarization current. Measurements were begun after this electrode current fell below 10 pA. The amperometric currents were recorded with EPC 5 or EPC 9 (HEKA Elektronik) patch-clamp amplifiers, filtered at 10 kHz (−3 dB cut off frequency) using a four-pole Bessel filter (Ithaco Co.), digitized, and stored on video tape. For analysis, the recordings were replayed, filtered, digitized, and stored by a computer. The final filter frequency was 100 Hz except where indicated. The sampling rate was four or five times faster than the filter frequency. In simultaneous recordings of [Ca²⁺]_i and amperometry, illumination of the carbon-fiber probe with ultraviolet light produced small current artifacts that were digitally corrected when larger than the baseline noise level (see Fig. 2, Fig. 4, and Fig. 7).

Solutions were exchanged by a local perfusion system that allows complete exchange of medium bathing the cells within 2 s.

The intracellular Ca²⁺ concentration ([Ca²⁺]_i) was measured with the Ca²⁺-sensitive fluorescent dye indo-1 loaded into cells such as the membrane-permeable AM ester (Molecular Probes, Inc.). The cells were incubated for 30 min in saline solution with 1 μM indo-1 AM at room temperature. The excitation wavelength was 365 nm (100 W mercury lamp), and fluorescence signals were recorded at 405 and 500 nm, using a pair of photon-counting photomultiplier tubes. The sample rate was 0.5 Hz for photometry alone and 0.1 Hz for simultaneous photometry and amperometry. Background fluorescence measured from a cell-free area was corrected. The [Ca²⁺]_i was calculated using the equation: [Ca²⁺]_i = K* (R − R_min)/(R_max − R), where R is the 405/500-nm fluorescence ratio and R_min and R_max are the ratios for Ca²⁺-free and -bound dye, respectively (Grynkiewicz et al. 1985). R_min, R_max, and K* were measured on cells perfused with Na⁺-rich external solutions containing 20 μM ionomycin plus 50 mM EGTA or 20 mM Ca²⁺, or 20 mM EGTA and 15 mM Ca²⁺ for >15 min. In addition, 5 μM thapsigargin and 2 μM carbonyl cyanide m-chlorophenylhydrazone were included in all calibration solutions to block active clearance of Ca²⁺ from the cytoplasm. The calculated free [Ca²⁺] in the 20 mM EGTA and 15 mM CaCl₂ solution was 251 nM. Values for R_min, R_max, and K* were 0.44, 4.46, and 1,770 nM, respectively (n = 8–13 cells for each measurement).

For measuring [Ca²⁺]_i levels higher than several micromolar, some cells were loaded with the low-affinity dye, mag-indo-1. Loading of the dye was performed as for indo-1, except for a shorter loading time (5–10 min). The calibration solutions were the same, except the mid-point solution contained 10 mM nitrilotriacetic acid and 5 mM Ca²⁺ with a calculated free [Ca²⁺] of 100 μM. Values for R_min, R_max, and K* for mag-indo-1 dye were 0.19, 1.90, and 327 μM, respectively (n = 7–18 cells).

Amperometric recordings were semiautomatically analyzed using software written in Igor (WaveMetrics). Some recordings with a small number of amperometric signals were plotted on a fast chart recorder and events were counted manually. The rate of exocytosis was defined as the number of amperometric spikes per 30-s time bin. To evaluate relative exocytosis, the rates of exocytosis in control and test conditions were averaged for 2 min. As the maximal exocytosis is reached at different times in different test solutions, the 2-min analysis period was taken after various time delays: 3 min for forskolin or PMA on 1,2-bis-(2-aminophenoxy)ethane-N,N,N,N-tetraacetic acid (BAPTA)–loaded cells (see Fig. 4, and Fig. 6,Fig. 7,Fig. 8), Rp-8-Br-cAMPS (see Fig. 6) or BIS (see Fig. 8) treatments, 2 min for forskolin, PMA or Sp-8-Br-cAMPS on cells not loaded with BAPTA (see Fig. 5, Fig. 6, and Fig. 8), and 1 min for 2 mM Ca²⁺ on ionomycin- (see Fig. 2) and hyperosmotic solution-treated (see Fig. 10) cells. During treatments with hyperosmotic solution, only the last minute of the recordings was analyzed. Relative exocytosis was defined by the ratio of exocytosis rates after and before treatments.

Kinetics of single quantal events were fitted with the sum of two exponential functions using events without a foot. Rise times (10–90% of peak amplitude) and fall times (90–10% of peak amplitude) were extracted from the fitted curves. Half-width was measured as time spent above 50% of peak amplitude.

All numerical values are given as mean ± SEM. The significance of differences between mean values of two groups was tested by Student's t test. The Kolmogorov-Smirnov test (see Fig. 5) was performed using software written in Igor. A probability of P ≤ 0.05 was considered significant.

We begin by describing the amperometric signature of exocytosis in epithelial cells for comparison with better-characterized systems. Touching a polarized carbon-fiber electrode gently to a single cell previously loaded in 70 mM dopamine or serotonin revealed spontaneous amperometric current spikes (Fig. 1A and Fig. B). They reflect rapid oxidation of many thousands of oxidizable molecules released abruptly at the cell surface. In tests on five different batches of cells, signals were not observed without preloading the amine, indicating that the spikes represented the exogenous oxidizable amine. It was necessary to incubate epithelial cells in the 70-mM loading medium for longer than 30 min to detect the oxidation currents. Attempting to load with 1.5 mM of dopamine or serotonin for 4 h elicited much smaller peaks (not shown). Rates of spontaneous exocytosis ranged from 2 to 53 events per min (mean: 17.2 ± 4.9 events/min, n = 11 cells). The rate typically decreased over 4–5 h after loading was stopped, presumably due to a loss of vesicular amines through spontaneous exocytosis and by leak of cytoplasmic amines to the culture medium.

As shown in Fig. 1 B, each spike had the rapid onset and decay that are characteristic of exocytotic release from single secretory vesicles in other systems (Wightman et al. 1995). To analyze the kinetics of single events, amperometric recordings were filtered at 500 Hz frequency (−3 dB cut off). The average half-width of events was 5.0 ± 0.3 ms (118 events from 10 experiments). The average 10–90% rise and fall times were 1.13 ± 0.06 and 6.7 ± 0.3 ms, respectively. As the measured 10–90% rise time of a square pulse filtered the same way was 0.70 ms, the true rise time of amperometric events should be much shorter than the measured value. The short overall time course of the events is consistent with a diffusion distance of <1 μm between the amperometric probe and sites of release (Chow and von Rüden 1995).

Preceding large events with a good signal-to-noise ratio (Fig. 1 B), one could occasionally see a small step of release preceding the main event, named the “foot” by Chow et al. 1992. In chromaffin cells and mast cells, the foot current represents leakage of the secretory product through a flickering fusion pore before full dilation of the structure connecting a vesicle to the plasma membrane (Alvarez de Toledo et al. 1993; Zhou et al. 1996). The charge and peak amplitude distribution histograms of single amperometric events were broad, with respective means of 30.3 fC and 0.7 pA in this particular experiment using a serotonin-loaded cell (Fig. 1C and Fig. D). The peak amplitude of amperometric events from cells loaded with 70 mM dopamine or serotonin varied from <1 pA to several tens of picoamperes (mean amplitude in nine cells, 0.93 ± 0.11 pA, at a filter frequency of 100 Hz). The time integral of the current in single events had a mean charge of 16.5 ± 2.5 fC (n = 15 cells) or ∼103,000 electronic charges per event. For a two-electron oxidation of dopamine, this would correspond to ∼51,500 molecules per spike. The similarities with amperometric studies in other cells argue that these amperometric spikes represent quantal exocytosis of loaded amines and that release begins with formation of a fusion pore (see discussion). However, one difference here is the extreme range of sizes of the amperometric events. Since our loading and recording methods did not isolate a specific class of secretory vesicles (Kim et al. 2000), the broad range of event sizes detected probably reflected various degrees of filling of exocytotic vesicles of varying sizes.

As for neurons and many endocrine cells, exocytosis in the pancreatic duct epithelial cells could be stimulated by elevating [Ca²⁺]_i. We used the Ca²⁺ ionophore, ionomycin, to bring Ca²⁺ into the cytoplasm. Dopamine-loaded cells were incubated in Ca²⁺-free solution containing 5 μM ionomycin for 5–10 min. Subsequent application of a solution containing 2 mM Ca²⁺ induced vigorous exocytosis (Fig. 2A and Fig. B). The significant rise of the current baseline during the Ca²⁺ application very likely reflected spillover of released dopamine from regions not directly under the electrode—on the same cell and on neighboring cells. The mean rate of exocytosis increased to 9.6 ± 2.4 (n = 16) times the control rates. The barrage of exocytotic events seen at the end of the Ca²⁺ application in Fig. 2 A was not seen in other similar experiments. It may reflect the sudden rise of [Ca²⁺]_i late in the Ca²⁺ application (Fig. 2 C).

We wanted to estimate the dependence of exocytosis on the mean [Ca²⁺]_i achieved with ionomycin treatment. Initial experiments with indo-1 had shown that application of 2 mM Ca²⁺ in the presence of ionomycin raised [Ca²⁺]_i into the range where the response of indo-1 saturates. Therefore, we switched to mag-indo-1-AM, a dye with much lower Ca²⁺ affinity. In various cells exposed to 2 mM Ca²⁺, [Ca²⁺]_i increased monotonically, after a delay of ∼1 min (Fig. 2C and Fig. D). The average [Ca²⁺]_i reported by this dye at the end of the 3-min Ca²⁺ stimulation was 313 ± 74 μM (n = 8). Although [Ca²⁺]_i rose for several minutes, exocytosis developed early, before a large change in the mag-indo-1 signal was observed, indicating that exocytosis can be efficiently induced at low micromolar concentrations of Ca²⁺ (Fig. 2 D). To improve our estimate, we examined exocytosis in ionomycin with several lower Ca²⁺ concentrations in the bathing solution (Fig. 3A and Fig. B). The 1 mM Ca²⁺ was about as effective as 2 mM, and 300 μM elicited a half-maximal rate of exocytosis. After these 3-min incubations in 5 μM ionomycin, the [Ca²⁺]_i was definitely not proportional to extracellular [Ca²⁺]. With 1 mM Ca²⁺ outside, the [Ca²⁺]_i rose to 107 ± 25 μM (n = 9), whereas with 300 and 100 μM outside, it reached only 419 ± 41 nM (n = 5) and 176 ± 13 nM (n = 6), respectively. (The first value was measured with mag-indo-1 and the latter two with indo-1.) Apparently, cellular Ca²⁺ clearance mechanisms remained relatively effective with the lower bath [Ca²⁺], and they seemed to become overwhelmed if [Ca²⁺]_i rose above 5 μM for many seconds. This result led us to use higher concentrations of ionomycin, longer incubations, and Ca²⁺ clearance inhibitors when calibrating the dyes (see methods). In any case, exocytosis from the pancreatic epithelial cells is quite sensitive to Ca²⁺ since it reaches significant levels already with [Ca²⁺]_i as low as 300–400 nM (Fig. 3 C).

Secretion from many gastrointestinal and airway epithelia can be stimulated by hormones and drugs that elevate intracellular levels of cAMP or stimulate phospholipase C. In cultured dog pancreatic epithelial cells, mucin secretion is induced by treatments that elevate cAMP (Oda et al. 1996). We therefore determined whether the exocytosis detected by amperometry is similarly enhanced. However, to establish whether any stimulation observed was secondary to an elevation of [Ca²⁺]_i, we first monitored [Ca²⁺]_i using indo-1 as an indicator (Table). The conditions tested included addition of forskolin, a direct activator of adenylyl cyclase, of epinephrine or VIP, whose receptors activate adenylyl cyclase (Oda et al. 1996), and of membrane-permeant 8-Br-cAMP. Each treatment often induced a modest transient elevation of [Ca²⁺]_i, although not always. Therefore, in the following two sections, when asking if there is a Ca²⁺-independent component of exocytosis with these treatments, we buffered changes of [Ca²⁺]_i by preincubating cells for 1 h in 50 μM BAPTA-AM, the membrane-permeant form of the Ca²⁺ chelator BAPTA (Table).

Fig. 4 A illustrates a typical long-lasting amperometric recording before and after activation of adenylyl cyclase by forskolin in a BAPTA-loaded cell. After 3 min of control recording, 20 μM forskolin was applied and the frequency of amperometric spikes gradually increased. The number of events in every 30-s interval is plotted in Fig. 4 B. The frequency rose from near 6/min in control to 32/min after forskolin treatment, a fivefold increase. The activation was first apparent at 30–60 s, and grew even after the forskolin was removed. The simultaneous [Ca²⁺]_i measurement (Fig. 4 C) confirmed that preloading with BAPTA had effectively eliminated any [Ca²⁺]_i transient during the forskolin treatment.

The time course of individual quantal events was not changed by forskolin treatment. Thus, the average half-width of events was 4.9 ± 0.2 ms (204 events from 10 experiments, at a filter frequency of 500 Hz; P = 0.89 compared with the value of control events). The average 10–90% rise- and fall-times were 1.13 ± 0.04 and 6.7 ± 0.3 ms, respectively. The mean current amplitude and charge of quantal events were slightly increased by forskolin treatment (1.11 ± 0.07 pA (P = 0.04) and 17.9 ± 2.1 fC (P = 0.39), respectively, n = 15 cells). Forskolin increased the rate of release of quanta of all sizes, but selectively enhanced the release of larger quanta, as can be seen in the histograms of Fig. 5 that compare the fraction of events in each charge range. (A significance of P = 0.002 was calculated with the Kolmogorov-Smirnov test using a cumulative histogram of the same data.)

The effects of forskolin on cells with and without BAPTA loading are summarized in Fig. 6 A. The amperometrically detected exocytosis rate in forskolin was 3.4 ± 1.2 (n = 8) relative to controls in BAPTA-loaded cells and 3.4 ± 0.6 (n = 23) in BAPTA-free cells. The basically identical effects of forskolin with or without buffering of the small transient rise of [Ca²⁺]_i (Table) indicated that forskolin-stimulated exocytosis was mainly activated by cAMP. In parallel, we tested other agents known to increase intracellular cAMP levels and mucin secretion in these pancreatic epithelial cells (Oda et al. 1996). As anticipated, the amperometrically detected exocytosis rates were increased by 1 μM epinephrine (5.1 ± 1.1, n = 8), 1 μM VIP (4.4 ± 2.0, n = 4), and 1 mM 8-Br-cAMP (4.0 ± 0.9, n = 7) (Fig. 6 A). In these experiments, cells were not loaded with BAPTA, so some of the increase of exocytosis with epinephrine and 8-Br-cAMP could have been due to a [Ca²⁺]_i rise in addition to their cAMP effect.

Does the stimulation of exocytosis require activation of cAMP-dependent protein kinase (PKA)? As a test, we compared the actions of the Rp and Sp enantiomers of 8-Br-cAMPS (Fig. 6 B). The Rp isomer is a specific inhibitor of PKA, and the Sp isomer is an activator. Application of Rp-8-Br-cAMPS did not change the rate of exocytosis in unstimulated cells (1.06 ± 0.15, n = 6), but it blocked the effect of forskolin applied 5 min later (0.80 ± 0.11, n = 6). In contrast, application of Sp-8-Br-cAMPS increased exocytosis by itself (2.7 ± 0.3, n = 9). These actions further support the hypothesis that stimulation of exocytosis is mediated through cAMP and PKA.

We then tested the phorbol ester PMA to ask whether activating PKC also increased amperometrically detected exocytosis. The elevation of [Ca²+]_i by PMA is minimal (Table). Nevertheless, we preloaded one batch of cells with BAPTA and tested the actions of 100 nM PMA. The phorbol ester increased exocytosis (Fig. 7A and Fig. B) without any accompanying change of [Ca²⁺]_i (Fig. 7 C; and Table), but the potentiation of exocytosis was smaller than that by cAMP. On average, this concentration of PMA elevated relative exocytosis to 2.7 ± 0.6 (n = 5) in cells preloaded with BAPTA and to 2.4 ± 0.4 (n = 13) without BAPTA (Fig. 8 A). These values are not statistically different (P = 0.67). Next, we did controls to check whether PMA acted via PKC to increase exocytosis. A structurally related but inactive compound, 4-α-phorbol-12,13-didecanoate (4-α-phorbol) slightly increased exocytosis (1.37 ± 0.27, n = 5) at a concentration of 100 nM; this increase, however, was not statistically significant (P = 0.12). To reduce possible nonspecific effects, the concentration of the active and control phorbol esters was decreased to 10 nM (Fig. 8 B). Exocytosis was still significantly evoked by PMA at 10 nM (2.2 ± 0.4, n = 4) but not by 4-α-phorbol (0.95 ± 0.21, n = 4). The PKC-selective inhibitor, BIS (500 nM), slightly reduced exocytosis by itself (0.68 ± 0.07, n = 10). After a 5-min incubation with BIS, PMA was not able to evoke exocytosis (0.44 ± 0.07, n = 9). The further decrease of exocytosis during the combined application of BIS and PMA may be due to a continued slow development of the effect of BIS.

Using simultaneous measurements of [Ca²⁺]_i and exocytosis, we have demonstrated that PKA and PKC can increase exocytosis when [Ca²⁺]_i rises are blocked with BAPTA. Now we asked if the reverse is true. Can Ca²⁺ induce exocytosis when protein kinases are blocked? In these experiments, we use BIS to block PKC as before, and H-89 to block PKA. (The expense of using Rp-8-Br-cAMPS would have been too great.) To check the efficacy of H-89 in blocking PKA, we asked whether pretreatment with 10 μM H-89 for 5 min reduced relative exocytosis induced by 10 μM forskolin. In untreated cells, forskolin raised relative exocytosis from 3.31 ± 0.85 (n = 3), and in H-89–treated cells, forskolin was ineffectual (1.15 ± 0.22, n = 7, relative to H-89 alone).

Both kinase inhibitors depressed relative exocytosis in control experiments. It was reduced to 0.64 ± 0.05 by H-89 (n = 27) and to 0.76 ± 0.06 (n = 24) by BIS (Fig. 9 A). The depression implies either that there is some background activation of the kinases in unstimulated cells or that the inhibitors have nonspecific actions at other steps in exocytosis. Therefore, in the statistical tests below, we use the H-89- or BIS-inhibited measurement as the control value for calculating the effect of subsequently added Ca²⁺.

Next, tests were done on the effects of [Ca²⁺]_i elevation while kinases were inhibited. The first tests used 300 μM extracellular Ca²⁺, which raises [Ca²⁺]_i only to submicromolar concentrations (∼400 nM). The low Ca²⁺ stimulus applied to uninhibited, ionomycin-treated cells increased relative exocytosis to 3.01 ± 0.37 (Fig. 9 A). The same stimulus was not effectual on cells treated with 10 μM H-89 (0.71 ± 0.21, n = 8, relative exocytosis compared with H-89 alone) or with 500 nM BIS (1.08 ± 0.26, n = 16, relative exocytosis compared with BIS alone). Thus, protein kinase inhibitors block the effects of small [Ca²⁺]_i elevations. Then tests were repeated with 2 mM extracellular Ca²⁺, which raises [Ca²⁺]_i to 300 μM. The exocytosis induced by this stimulus was not significantly reduced by inhibition of PKA or PKC or by simultaneous inhibition of both kinases (Fig. 9 B).

In the beginning of the results, we emphasized similarities between the secretion in epithelial cells and that of excitable cells. There are some clear differences, however: insensitivity to hypertonic shocks, insensitivity to KCl depolarization, and occasional giant multi-vesicular secretory events.

Synapses release a small number of neurotransmitter quanta upon sudden stimulation with hypertonic solutions (Furshpan 1956; Rosenmund and Stevens 1996), and such stimuli are considered a reliable way to assay the size of the “readily releasable pool” of docked and primed neurotransmitter vesicles. In contrast, a saline solution supplemented with 250 mM sucrose strongly decreased the relative rate of amperometric events from our epithelial cells (Fig. 10, to 0.28 ± 0.08, n = 12). There was no apparent change of cell volume or visible detachment of the carbon electrode from the cell, and the effect of hypertonic solution was reversible (1.05 ± 0.18, n = 10).

In excitable cells, exocytosis is normally stimulated by membrane depolarization, which allows Ca²⁺ to enter cells via voltage-gated Ca²⁺ channels. As exocytosis from epithelial cells is also Ca²⁺ sensitive, it was interesting to test whether they can use a voltage-gated Ca²⁺ influx mechanism. The resting potential of the cultured epithelial cells was relatively low, −39 ± 2.1 mV (measured under perforated-patch conditions). It depolarized to −21 ± 3.1 mV (n = 11) when all the extracellular NaCl was replaced by KCl. However, the KCl depolarization did not change the rate of exocytosis (exocytosis relative to control = 0.97 ± 0.07, n = 9), indicating that, at least in this potential range, these cells lack voltage-gated Ca²⁺ entry mechanisms.

We occasionally observed large, long-lasting amperometric events. These complex events were characterized by multiple overlapping peaks, suggesting compound exocytosis with overlapping fusion of a large number of vesicles (Fig. 11). As an empirical criterion to detect such events, we used a sliding window of 50 adjacent sample points. A large event was scored if the total charge in this interval exceeded 20× the average charge of a single quantal event. The window was moved by 25 points each time to ensure the detection of the complex events. By this criterion, there were 12 giant events out of 2,714 amperometric events (one per 226 events) in seven experiments. The occurrence may be overestimated, as only recordings containing at least one obvious complex event were included in this analysis. The average charge of such complex amperometric events was 99 ± 25 (n = 7) times that of the quantal events. Such events were seen both during stimulated release and during spontaneous release in unstimulated cells.

The physiological ion transport and exocytosis in epithelia are regulated by hormones and neurotransmitters through stimulation of either adenylyl cyclase or phospholipase C. In most studies, it has not been possible to measure or regulate possible intracellular Ca²⁺ transients in response to the hormones. For hormones acting via phospholipase C, a Ca²⁺ transient as well as activation of PKC would be expected; all these pathways stimulate mucin secretion (Forstner et al. 1993; Davis and Abdullah 1997).

Secretion of mucin has previously been demonstrated in confluent monolayers of the cultured dog pancreatic epithelial cells used here. VIP and epinephrine raised the cAMP content of these differentiated cells approximately sixfold, and 4-h incubations with VIP, epinephrine, or dibutyryl cAMP increased mucin secretion to 1.5–3× control amounts (Oda et al. 1996). Similarly, treatment with ATP, which activates P2Y₂ purinergic receptors in these cells and stimulates phospholipase C, generated a [Ca²⁺]_i transient, and increased mucin secretion approximately fourfold (Nguyen et al. 1998). Calcium elevations alone, achieved with the ionophore A23187, stimulated mucin secretion threefold. Using single-cell biophysical methods, we have now shown that activation of PKA or PKC evoked Ca²⁺-independent exocytosis from nonconfluent monolayers of the same cells. A variety of stimuli increased exocytosis within 1–3 min. In our work, secretion rose to three to five times the control rate after cAMP stimuli (Fig. 4 and Fig. 6), two to three times after PMA (Fig. 7 and Fig. 8), and nine times after ionomycin (Fig. 2). Thus, the determinants of exocytosis assessed by amperometric measurements are comparable with the previously reported direct measurement of mucin secretion.

We provide further evidence that stimulation by cAMP requires activation of PKA and does not require a [Ca²⁺]_i elevation, and that PKC also will stimulate exocytosis without a [Ca²⁺]_i elevation. Finally, high levels of [Ca²⁺]_i alone can evoke exocytosis without activation of PKA or PKC. However, exocytosis evoked by submicromolar Ca²⁺ was significantly reduced when the protein kinases are pharmacologically inhibited. Such results suggest that Ca²⁺ induces exocytosis by different pathways depending on its concentration, but a definitive conclusion awaits further study with other types of protein kinase inhibitors or cells lacking PKA or PKC expression.

The secretory mechanisms of excitable cells and epithelia have similarities and differences. Our study may be the first to apply amperometry to epithelial secretion. This allows an improvement of the time resolution of several orders of magnitude and resolution of single-vesicle release events. As in excitable cells, we find that the basic release event has a rise time shorter than 1 ms and is preceded, for some events, by a foot lasting several milliseconds. It appears, therefore, that, as in mast cells or chromaffin cells (Breckenridge and Almers 1987; Chow et al. 1992), the secretory vesicle passes through a transient phase with potentially reversible formation of a stereotyped fusion pore, which has a variable lifetime followed by sudden dilatation that allows maximal release of the vesicle contents. The amperometric spikes are relatively brief, having a falling time constant of ∼7 ms and half widths of 5 ms. This lies between the observed 5–15-ms half widths for catecholamine release from mammalian chromaffin cells (Wightman et al. 1995) and the 0.6- and 3.7-ms half widths of two classes of serotonin release from a leech synapse (Bruns and Jahn 1995).

What are the differences from excitable cells? As they apparently lack voltage-gated calcium channels, depolarization does not stimulate exocytosis in epithelial cells as it does in excitable cells. Additionally, hypertonic solutions depress exocytosis in our cells and in chromaffin cells (Holz and Senter 1986), unlike in neuronal synapses (Furshpan 1956; Rosenmund and Stevens 1996). Finally, the apparent speed of regulation of exocytosis seems much less than in a synapse. The effects of the stimuli take at least tens of seconds to develop in epithelia. In contrast, in a synapse, the rate of release can increase 1,000-fold during an action potential (Barrett and Stevens 1972; Goda and Stevens 1994), and the full stimulation rises and falls within <1 ms. This difference may suggest a completely different Ca²⁺ sensor acting at a different point in the vesicle delivery pathway. Judging from the effects of protein kinase inhibitors, the resting secretion in epithelial cells may reflect a small basal stimulation of the cAMP and PKC pathways, as is found in other systems.

All of these functional differences may point to well-known structural factors. In epithelia, although many secretory granules may be near the plasma membrane (Oda et al. 1996), there is no known active zone and the fusion machinery is certainly not intimately attached to voltage-gated Ca²⁺ channels as at a synapse. Even in chromaffin cells, where there are Ca²⁺ channels, a significant fraction of docked vesicles may not be in close proximity (Chow et al. 1992; Voets et al. 1999). It is possible that, unlike in synapses, epithelial secretion is not based on a pool of primed vesicles requiring only <1 ms for final fusion. There may remain many constitutive biochemical steps that separate the actions of kinases from the fusion events they ultimately accelerate. Also in epithelia and chromaffin cells, the size of the typical secretory granule is 5–30× larger than a neurotransmitter vesicle of synapses, and the granules contain newly synthesized proteins. Hence, the vesicles must be formed differently and have some different components. Indeed, the proteins of the secretory machinery in nonneuronal cells are believed to be homologous but not always identical to members of the membrane and soluble protein families used in neurons. While the characterization of proteins involved in exocytosis in pancreatic duct cells is sparse, in pancreatic acinus, the other major exocrine cell type in pancreas, the proteins rab3D (Valentijn et al. 1996), syntaxin 2 (Hansen et al. 1999), and SNAP-23 (Gaisano et al. 1997) may serve the functions in apical secretion that are served by rab3A, syntaxin 1, and SNAP-25 in synaptic exocytosis. The protein syntaxin 3, which is associated with the acinar granules, may serve compound exocytosis, the intracellular fusion of granules to each other (Hansen et al. 1999). Many of the above proteins can be phosphorylated by various protein kinases, and their phosphorylation has been implicated in changes of exocytosis (Turner et al. 1999).

Ca²⁺ is very clearly the immediate trigger of evoked transmitter release at synapses. Nevertheless, pharmacological activation of PKA or PKC increases evoked and spontaneous excitatory and inhibitory postsynaptic currents two- to fourfold, much as in epithelial cells. For example, phorbol esters acting via PKC increase the frequency but not the amplitude of spontaneous GABA release from interneurons and glutamate release from hippocampal neurons (Finch and Jackson 1990; Capogna et al. 1995). Activating PKA by forskolin has a similar effect on excitatory and inhibitory synapses (Chavez-Noriega and Stevens 1994; Capogna et al. 1995). Phorbol esters also stimulate secretion in anterior pituitary cells (Billiard et al. 1997).

Our loading method does not allow selective study of a specific class of secretory vesicle in the way that a biochemical or postsynaptic current assay would. With the high serotonin and dopamine concentrations used here, loading does not depend on monoamine transporters (Kim et al. 2000). Instead, we believe that intracellular vesicles load passively with the exogenous monoamine if their lumen is acidic. Some of these vesicles may be components of constitutive membrane traffic. On average, the spontaneously released vesicles were smaller than the population released under cAMP stimulation. Nevertheless, the degree of stimulation of amperometric events by our treatments parallels the stimulation of mucin secretion measured directly in differentiated monolayers (Oda et al. 1996; Nguyen et al. 1998).

We can compare the number of oxidizable molecules released per quantum in different cells. In forskolin-stimulated pancreatic duct cells, we see a broad distribution of sizes and a mean of ∼100,000 elementary charges, corresponding to ∼50,000 dopamine molecules per vesicle, but with 20% of the events < 7,000 molecules and 7% > 100,000 molecules (Fig. 5). In the same kind of assay with cells containing endogenous catecholamine or serotonin, the approximate number of molecules released per quantum is 7 × 10⁶ from chromaffin cells (Albillos et al. 1997), 35,000 from cultured sympathetic ganglion cells (Zhou and Misler 1995), 40,000 and 800,000 from two vesicle populations in somata of the snail Planorbis (Chen et al. 1995), and 5,000 and 80,000 from two populations of vesicles in leech synapses (Bruns and Jahn 1995). Should the larger amperometric events represent exocytosis of mucin granules, the dopamine concentration in these granule can be calculated to be 20 mM, assuming 100,000 dopamine molecules per granule and a granule diameter of 0.25 μm, as estimated from Oda et al. 1996. In chromaffin granules, the concentration of endogenously packaged catecholamine is near 700 mM (Albillos et al. 1997).

What are the giant complex events (Fig. 11)? The hypothesis that they represent the death of a cell, with release of all oxidizable contents, can be excluded by two observations. First, if the cell membrane is deliberately ruptured by advancing the amperometric probe by several micrometers, a much larger oxidation current (peak amplitude >50 pA) of longer duration (>10 s) was observed. This reflects a considerable quantity of cytoplasmic monoamine. Second, the more typical quantal events continue unabated with fast rise and fall times immediately after a complex event, suggesting that the cell is still functional and that its membrane remains within 1 μm of the probe.

A more likely explanation for these giant complex events is that a large number of quantal events are superimposed. This would account for the clear spike-like substructure of the complex event. However, it is unlikely that complex events are a random superposition of independent quantal events. We can estimate the chance of that occurring. Since the average frequency of spontaneous quantal events is one per 3 s, the probability of observing a quantum within a 500 ms time window is 1/6. Provided that amperometric events are randomly distributed, the probability of having 100 or more independent events occurring in a 500-ms time window is ∼(1/6)¹⁰⁰. The probability of seeing a complex event within 500 ms window was actually 1/1,360, calculated as the probability of observing any event (1/6) multiplied by fraction of complex events (1/226). Therefore, some concerted process must coordinate multivesicular events. It might be a form of compound exocytosis in which some vesicles are fusing to each other before the final fusion to the plasma membrane or it may represent a period in which the probability of fusion has been raised by several orders of magnitude by some intracellular signal. The number of molecules released in one giant event exceeds several million, which is comparable with the release from a typical chromaffin granule—except that it is spread out in time so that the amplitude is not 100× the mean quantal amplitude. Compound exocytosis and also apocrine-like secretion are commonly invoked in morphological studies of stimulated, mucin secreting epithelia (Phillips 1992; Newman et al. 1996). Compound exocytosis has also been described in eosinophils (Scepek and Lindau 1993). The unexpected finding is the large charge of these giant events and the scarcity of any events of charge intermediate between what we have called quantal and the giant events.

In conclusion, exocytosis in pancreatic duct epithelial cells shares several features with exocytosis in excitable cells. However, overall, the range of regulation, the speed of regulation, and the maximum rate of exocytosis are smaller. In the epithelium, the effects of kinases and Ca²⁺ are similar in magnitude, whereas in neurons the effect of Ca²⁺ is much stronger than that of kinases. Therefore, hormonal regulation via kinases becomes a viable way to regulate exocytosis from epithelia in the absence of any Ca²⁺ signal. Its molecular basis needs clarification.

We thank Drs. Ed Kaftan and Tao Xu for reading the manuscript and Drs. S.P. Lee and P. Verdugo for helpful discussions. We thank L. Miller and D. Anderson for technical assistance.

This work was supported by grants from the National Institutes of Health (AR17803), the Cystic Fibrosis Foundation (R565), and the Department of Veterans Affairs (Merit Review).