Effects of Intravesicular H+ and Extracellular H+ and Zn2+ on Insulin Secretion in Pancreatic Beta Cells*

The effects of extracellular Zn2+ and pH and intravesicular pH on insulin and 5-hydroxytryptamine (5-HT) secretion from pancreatic beta cells were investigated. Insulin and 5-HT secretion from single cells was detected by amperometry as a series of current spikes corresponding to detection of multimolecular packets secreted by exocytosis. Spike width was used as a measure of the kinetics of clearance from the cell and the area of spikes as a measure of amount released. Changes in extracellular pH from 6.9 to 7.9 caused insulin spikes to become narrower with no change in area, whereas the same treatments had no effect on 5-HT secretion. Treatment of cells with Bafilomycin A1 orN-ethylmaleimide, both of which are expected to increase intravesicular pH by inhibiting V-type H+-ATPase, had no effect on 5-HT secretion but caused insulin spikes to become more narrow. These results indicate that exposure to high pH, whether intravesicular or extracellular, accelerates release of insulin during exocytosis without affecting the amount of insulin released. Increasing extracellular Zn2+ concentration from 0 to 25 μm increased the width and decreased the area of insulin spikes without affecting 5-HT secretion. Zn2+ effects were likely exerted through a common-ion effect on Zn2+-insulin dissociation. It was concluded that intravesicular storage conditions and extracellular ions can affect free insulin concentration in the vicinity of beta cells during secretion.

Insulin, produced and stored in pancreatic beta cells, is released by exocytosis in response to external stimuli, such as elevated glucose concentration. A greater understanding of exocytosis in beta cells is of intense interest because of the possible role of defective insulin secretion in type II diabetes (1,2). Secretion by exocytosis can be considered to occur in several steps including vesicle docking, vesicle priming, vesicle fusion, and finally extrusion or clearance of vesicular contents into the extracellular medium (3). In this study, we examined the postfusion clearance of insulin from beta cell secretory vesicles during exocytosis.
Clearance of insulin from secretory vesicles is intimately related to the storage of insulin within a vesicle. Insulin is generally believed to be stored inside secretory vesicles as a solid hexamer bound with two Zn 2ϩ ions per hexamer (4 -6). This understanding is supported by a broad spectrum of evidence. It is known that in the presence of Zn 2ϩ , insulin will form tetramers and hexamers that bind Zn 2ϩ in an insulin: Zn 2ϩ ratio of 4:1 and 6:2, respectively (5), and that vesicles contain Zn 2ϩ in 1-1.5-fold excess of that necessary to form Zn 2ϩ -insulin hexamers (7). Zn 2ϩ -insulin hexamers are insoluble below pH 7 (5,6,8,9), and beta cell secretory vesicles possess a V-type H ϩ -ATPase that maintains the vesicle interior at or below pH 6 (10,11). Electron micrographs and immunohistology also suggest the presence of solid insulin inside vesicles (12,13). During exocytosis, secretory vesicles fuse with the plasma membrane, exposing the vesicular interior to the extracellular milieu, allowing release of stored insulin granules. It is apparent that insulin must dissolve to escape the vesicle. In addition, insulin must dissociate from Zn 2ϩ because insulin monomer is the biologically active form (14,15). Little is known about the mechanism or time scale of dissolution and dissociation, although one study has demonstrated that insulin is free from Zn 2ϩ within 60 s of release (4). Additional investigation of this phenomenon has been hindered by the lack of methods for monitoring insulin secretion with sufficient temporal resolution to observe such effects.
Recently, amperometry with microelectrodes has been applied to the study of exocytosis at a variety of cell types, including adrenal chromaffin cells (16 -19), PC12 cells (20), mast cells (21), neurons (22), and melanotrophs (23,24). Secretion is detected as a series of current spikes that correspond to detection of concentration pulses generated by exocytosis. Analysis of current spikes can reveal subtle details about secretion. The area of current spikes (measured in coulombs) can be used to quantitate the moles of detected hormone (or transmitter) released per vesicle using Faraday's Law (19). The dynamics of vesicular release are reflected in the shape of the current spikes. Specifically, the width of spikes can be used as a measure of the rate of clearance from the vesicle because the width is determined by any slow kinetic step involved in clearing hormone from the vesicle and diffusional broadening of the secreted packet of molecules (25)(26)(27)(28). Thus, the exquisite temporal resolution and sensitivity of amperometry for measuring actual secretory products has made it well suited for probing postfusion events in exocytosis (25-27, 29, 30).
We have extended amperometry to detection of insulin exocytosed from pancreatic beta cells (31,32). For this application, a carbon-fiber electrode modified with ruthenium oxide/cyanoruthenate (Ru-O/CN-Ru), a catalyst that promotes oxidation of insulin, is used for detection. This method has allowed the first direct measurement of exocytosis of insulin from single beta cells with high temporal resolution (32). In using amperometry to study postfusion events in insulin release, we have found it useful to measure 5-HT 1 release as a control. 5-HT can be "loaded" into beta cells by incubating the cells in medium containing both 5-HT and 5-hydroxytryptaphan. Substantial evidence now exists that demonstrates that 5-HT is loaded primarily into beta cell secretory vesicles and that 5-HT is co-secreted with insulin by exocytosis (34 -36). 2 5-HT can be detected at bare carbon electrodes and insulin can be detected only at Ru-O/CN-Ru-modified electrodes. Thus, it is possible to measure, in separate experiments, secretion of 5-HT from loaded cells and secretion of insulin from cells that have not been allowed to accumulate 5-HT.
In an initial study on postfusion clearance of insulin, we found that exposure to a pH gradient between the vesicle and the extracellular environment is required to achieve rapid release of free insulin from solid Zn 2ϩ -insulin granules (37). In the present study, we extended this investigation to a variety of extracellular pH values. In addition, we examined the effect of intravesicular pH and extracellular Zn 2ϩ on insulin secretion. The study reveals strong effects of the extracellular and intravesicular environments on the rate and amount of free insulin released during exocytosis.

MATERIALS AND METHODS
Electrode Preparation and Testing-Carbon fiber microelectrodes were constructed as described previously (38,39). Finished electrodes consisted of a 9 m carbon fiber (P-55S, Amoco Performance Products) sealed with epoxy (Miller Stephenson, Danbury, CT) in the tip of a glass pipette. The total electrode diameter at the tip was ϳ30 m, and the electrode was polished at a 30 -45°angle using a pipette beveler (BV-10, Sutter Instruments). For detection of insulin, electrodes were chemically modified as described previously to produce a mixed valent Ru-O/CN-Ru film (32), whereas 5-HT detection was performed with bare carbon fiber microelectrodes (34). 2 Tests of the electrode response to insulin and 5-HT were performed by positioning the tips of electrodes approximately 10 m from micropipette tips (outer diameter, 30 m) under a Krebs-Ringer buffer solution on the stage of a microscope. Solutions were pressure-ejected from the tips of the pipettes at 9 p.s.i. for 5 s while the amperometric current was recorded.
Data Collection and Analysis-Amperometry was performed using a battery to apply potential to a sodium-saturated calomel electrode. Currents were monitored at the working electrode using an AI-403 current amplifier and Cyberamp 320 signal conditioner (Axon Instruments, Foster City, CA). For detection of 5-HT, the potential at the working electrode was 0.65 V, whereas for detection of insulin, the potential was 0.85 V. All voltages are versus a sodium-saturated calomel electrode. When using the modified electrode, the potential was held at 0.40 V between recordings to improve electrode stability as described elsewhere (32). Data were low pass filtered at 300 Hz for insulin measurements and at 1000 Hz for 5-HT. These filter settings were found to not affect the peak widths relative to higher settings. Data were collected at a rate that was three to five times the filter frequency using a personal computer (Gateway 2000 P5-166) via a data acquisition board (DigiData 1200B, Axon Instruments). The area and half-width of current spikes were calculated using software provided by Prof. R. Mark Wightman (University of North Carolina). For data analysis, spikes were used only if the signal:noise ratio was greater than 10. All means are reported Ϯ 1 S.E. Statistical differences between means were evaluated using a two-tailed Student's t test.
Isolation of Canine Islets and in Vitro Culture-Unless otherwise stated, all chemicals for islet and cell culture were obtained from Life Technologies. Pancreatic islets were isolated from mongrel dogs using controlled collagenase (Boehringer Mannheim) perfusion via the duct, automated dissociation, and discontinuous Euro-Ficoll purification using the COBE 2991 blood cell processor as described previously (40,41). Following isolation, groups of islets were placed in modified CMRL 1066 tissue culture solution (G-100) containing 10% fetal calf serum, 25 mM HEPES, 100 units/ml penicillin, and 100 g/ml streptomycin and shipped overnight. Upon arrival, islets were immediately dispersed into single cells using a previously described procedure (32). Cells were plated onto tissue culture dishes (Corning) and incubated at 37°C in 5% CO 2 in G-100, pH 7.4.
Accumulation of 5-Hydroxytryptamine into Beta Cells-For experiments requiring measurement of 5-HT secretion, dispersed beta cells were allowed to incubate in G-100 containing 0.5 mM 5-HT and 1 mM 5-hydroxytryptophan for 16 h at 37°C in 5% CO 2 , pH 7.40 (37). 2 Cells were used for secretion experiments immediately after loading.
Single Cell Experiments-Single cell measurements were performed similar to those described elsewhere (19,23,31 (42), and 3.0 mM D-glucose and maintained at 37°C and 5% CO 2 on the stage of the microscope by a microincubator (Medical Systems, Inc.). Buffer pH was adjusted by varying the bicarbonate concentration to achieve the desired pH after bubbling with 5% CO 2 . Extracellular Zn 2ϩ concentration was adjusted by adding ZnCl 2 to the desired concentration. Ionic strength was held constant for all solutions. When appropriate, cells were incubated in G-100 containing either 50 M N-ethylmaleimide or 1 M Bafilomycin A 1 for 30 min. Amperometry was performed at isolated single cells by positioning the sensing tip of the microelectrode 1 m from the cell using a micromanipulator (PCS-250, Burleigh Instruments). Stimulant solution (200 M tolbutamide in bicarbonate buffer) was applied to individual cells by pressure ejection for 10 s from a micropipette positioned ϳ30 m from the cell.
Simulation of Spike Shape-A random-walk model was used to simulate diffusion-limited current spikes obtained from exocytosis (43). In the model, the cell was considered a hemisphere with a radius of 5 m, and the electrode was centered directly over the apex of the cell, with a 1.0-m gap between the cell and the electrode. An exocytosis event was modeled as a point source of 50,000 molecules diffusing from a location on the cell surface to the electrode. Thus, exocytosis events originating at the base of the cell travel farther to the electrode than exocytosis events originating at the apex of the cell, where the distance to the electrode was minimized. Molecules encountering the electrode were counted as detected and were removed from the system. Molecules encountering the cell membrane or the insulating portions of the electrode were reflected back to their previous position. The simulation was ended when all molecules were detected or after 3000 steps had been made in the walk. Longer random walks did not affect the results. For the model, 5-HT and insulin were considered to have diffusion coefficients of 6.0 ϫ 10 Ϫ6 and 2.0 ϫ 10 Ϫ6 cm 2 /s, respectively (31).

RESULTS
Effect of Zn 2ϩ on Insulin Detection-We have previously used the Ru-O/CN-Ru electrode to detect insulin and insulin secretion (31,32). An issue that has not been addressed but that is important in the interpretation of these data is the effect of Zn 2ϩ complex formation on insulin detection. Fig. 1  trates current recorded during injection of Krebs-Ringer buffer solutions that contained 30 M insulin (A) or 30 M insulin along with 15 M Zn 2ϩ , which promotes the formation of Zn 2ϩinsulin complexes (B). As the data show, addition of Zn 2ϩ abolished the signal obtained from detection of insulin. The presence of Zn 2ϩ did not affect the detection of other, non-Zn 2ϩ binding compounds, such as 5-HT ( Fig. 1, C and D). These results strongly suggest that the electrode detects free insulin and not insulin complexed with Zn 2ϩ .
Spike Width Analysis- Fig. 2 compares high resolution recordings of spikes resulting from detection of insulin and 5-HT. Several differences in spike shape were observed for detection of 5-HT and insulin. The most marked difference in shape of the spikes was that the average width at half-height of 5-HT spikes was 5.3 Ϯ 0.4 ms, which is significantly lower (p Ͻ 0.0005) than the 30 Ϯ 2.1 ms obtained for insulin spikes. As mentioned under "Introduction," the width at half-height of a current spike is a measure of the time required for detected material to extrude from a vesicle, diffuse to the electrode, and be detected. Thus, some of the difference in width could be attributed to the difference in diffusion coefficient for 5-HT and insulin. The random walk model, which predicts widths based solely on diffusion, indicates that the insulin spikes should be three times the width of 5-HT spikes, yet we observed a 6-fold difference. The model also predicts that the widest diffusionlimited spikes, which would occur for a release site that is the maximum possible distance from the electrode, should be 22 ms for 5-HT and 58 ms for insulin. Experimentally, we found that none of the 5-HT spikes were broader than this maximum value. In contrast, 45% of the insulin spikes were over the maximum diffusion-limited width. Taken together, these results indicate that release of 5-HT is indistinguishable from diffusion control and that a factor other than diffusion contributes to slowing down the kinetics of release of insulin.
Another difference in shape was that 14% of the 5-HT spikes had a small current increase or "foot" prior to the rapid increase to the spike apex, as shown in Fig. 2C. In contrast, such features were not observed on spikes due to detection of insulin at physiological pH. Prespike features like those seen for 5-HT detection have also been observed in the detection of catecholamine release from adrenal chromaffin cells (16,25) and 5-HT release from mast cells (21). This shape has been attributed to leakage of material from a fusion pore that forms prior to complete opening of the vesicle (16,21,25). The lack of a foot for detection of insulin suggests that insulin does not leak from a fusion pore in quantities that are detectable.
Effects of Extracellular pH on Insulin Extrusion-To evaluate the effect of extracellular pH on exocytosis, single cell recordings of insulin secretion were made following tolbutamide stimulation when the extracellular pH was 6.40, 6.90, 7.05, 7.20, 7.40, and 7.90. Similar experiments were performed for detection of 5-HT secretion from loaded cells at pH 6.40, 7.10, and 7.40. Fig. 3 summarizes the spike widths and areas that were recorded in these different experiments. As mentioned under "Introduction," the area of isolated current spikes is considered a direct measure of the amount of product (insulin or 5-HT in this case) detected from a single vesicle (19). Spike area for insulin detection is not significantly changed by increasing extracellular pH above 6.9 (Fig. 3A); however, no spikes were detected at pH 6.4. In contrast, the width at half- height of spikes decreased with increasing pH for insulin detection. The constant area combined with decreasing width caused an increase in amplitude with increasing pH for insulin detection. In contrast, the area and width at half-height for 5-HT current spike detection was unaffected at any pH tested.
Effects of Intravesicular pH on Insulin and 5-HT Extrusion-To investigate the effect of the vesicular proton pump and lowered vesicular pH on insulin and 5-HT secretion, the effect of Bafilomycin A 1 and N-ethylmaleimide, compounds known to inhibit V-type H ϩ -ATPase activity and raise vesicular pH to cytoplasmic pH (44,45), on spike shapes was evaluated. Fig. 4 compares typical current traces recorded at untreated cells and cells incubated with 1 M Bafilomycin A 1 or 50 M N-ethylmaleimide. As observed from an inspection of these plots, the number of spikes per stimulation, duration of secretory activity, and area of the spikes are not significantly different for any of these treatments.
Although overall secretory activity for insulin was unaffected by ATPase inhibition, insulin spike shape was affected as illustrated in Fig. 5, A-C, and summarized in Fig. 5D. The spike width was significantly decreased (p Ͻ 0.001) by treatment with both agents. Furthermore, after treatment with Bafilomycin A 1 , 10% of the detected spikes exhibited a foot, and after treatment with N-ethylmaleimide 18% of spikes had this feature. Spike samples shown in Fig. 5, B and C, are typical of those that had a foot. As stated above, no spikes obtained from untreated cells had such prespike features. In contrast, the spike width for 5-HT was unaffected (Fig. 5D), and there was no effect on the percentage of 5-HT spikes with feet.
Effects of Extracellular Zn 2ϩ Concentration on Insulin and 5-HT Extrusion-In addition to the effects of pH, the effects of 0, 5, 15, and 25 M extracellular Zn 2ϩ on secretion measure-ments were examined. Fig. 6 illustrates typical secretion measurements of insulin and 5-HT under these different conditions. As extracellular Zn 2ϩ concentration increased, the area of insulin spikes decreased while the half-width increased, as summarized in Fig. 7. These combined effects caused spike amplitudes to decrease and result in detection of smaller numbers of spikes as illustrated in Fig. 6. The smaller number of spikes is presumed to result from fewer spikes being above the noise level because of the decreased amplitudes. This effect is maximized at 25 M Zn 2ϩ , at which no insulin spikes were detected (data not shown). In contrast, when measuring 5-HT release, the area and half-width of the spikes were not significantly different at any given Zn 2ϩ concentration (see Figs. 6 and 7). DISCUSSION Dynamics of Exocytosis Events-Although 5-HT is released simultaneously with insulin, it escapes from the vesicle at a significantly higher rate, as evidenced by the narrower spikes and the presence of feet, which suggest leakage out of vesicles during fusion pore formation. 5-HT release is indistinguishable from simple diffusion out of the vesicle, whereas insulin release is slower, suggesting that a factor other than diffusion controls postfusion release. These results may be explained by considering differences in storage of the two secretory products. Although the exact form that 5-HT is stored in the vesicles is unknown, it is reasonable to expect that it is stored in solution because 5-HT is highly soluble at the vesicular pH. In contrast, insulin is stored as a solid complex of Zn 2ϩ -insulin (4,5,12). In addition, the electrode only detects free insulin and not Zn 2ϩbound insulin, as illustrated in Fig. 1. Therefore, it seems likely from these considerations that the dissolution and dissociation of Zn 2ϩ -insulin complex controls the rate of insulin extrusion during exocytosis.
Effects of Extracellular and Intravesicular pH-In our initial work, we demonstrated that a change in extracellular pH from 7.4 to 6.4 could hinder insulin clearance to such an extent as to make it undetectable by amperometry while having no effect on 5-HT secretion (37). The data in Fig. 3, obtained over a wider pH range, show that decreasing extracellular pH from 7.9 to 6.9 increases the time required for insulin clearance (increasing spike width) but does not affect the amount of insulin released during an exocytosis event (no change in area). 5-HT secretion is unaffected in this pH range (Fig. 3), proving that this effect is specific to insulin extrusion and is not an effect on vesicular fusion. Further investigation has revealed that increasing vesicular pH with V-type H ϩ -ATPase inhibitors, which is expected to change the storage conditions for insulin, causes insulin to be more rapidly extruded (as demonstrated by narrower spikes) and even leak out of fusion pores (as demonstrated by the presence of feet). This effect suggests that some dissolution of the insulin granule occurs inside the vesicle when the vesicular pH is raised. Therefore, under normal conditions, low pH inside the vesicle maintains insulin in a solid state that does not escape as readily during fusion pore formation or vesicle fusion. These results further emphasize the importance of the pH gradient between the vesicle and extracellular media in rapidly changing insulin from a solid, storage state to a dissolved, releasable form. These results also highlight the difference between postfusion release of other hormones and insulin. For example, in catecholamine release from adrenal chromaffin cells, increasing the pH has the effect of increasing the time course of postfusion extrusion (25). Furthermore, the effects of pH on catecholamine release are not nearly as dramatic as those observed here. These differences can be attributed to differences in storage of the compounds. In the case of catecholamine, release requires the unraveling of a protein gel rather than dissolution of a solid-state hormone (25). Effect of Zn 2ϩ on Insulin Extrusion-In addition to dissolving, the insulin-Zn 2ϩ complex must ultimately dissociate after vesicle fusion because the active form of insulin is a monomer (14,15). We have postulated that availability of extracellular Zn 2ϩ may affect the release of free insulin from beta cells through a common-ion effect. Extracellular Zn 2ϩ concentrations in the range of 5-25 M markedly decreased area and increased width of insulin spikes without affecting 5-HT spikes. The lack of an effect on 5-HT detection demonstrates that the effect is specific to insulin extrusion and not to earlier steps in exocytosis. Unlike pH manipulations, extracellular Zn 2ϩ not only increased the time course of release of free insulin, but also decreased the amount of free insulin detected from an exocytosis event. Decreases in amount of free insulin detected occurred even though half-widths of detected spikes suggest that a release event is occurring at a fast enough rate to yield a detectable spike. A possible explanation for this result is that after vesicle fusion, the Zn 2ϩ -insulin hexamer dissolves but does not dissociate because of the high extracellular Zn 2ϩ concentration, allowing most of the insulin to escape from the detection area in the undetectable Zn 2ϩ -insulin complex form. Interestingly, the effects of Zn 2ϩ on insulin secretion are similar to those observed for catecholamine secretion from adrenal chromaffin cells (46). Although Zn 2ϩ decreased the quanta in both cases, the chemical mechanisms are different. Rather than a common ion effect on complex, the Zn 2ϩ effect on catecholamine secretion was due to the ability of Zn 2ϩ to crosslink the protein matrix in chromaffin cell vesicles and hinder the escape of the hormone (46).
The results presented here demonstrate that micromolar Zn 2ϩ concentrations can affect free insulin concentration in the immediate vicinity of a beta cell during secretion. The possible physiological significance of the Zn 2ϩ effect is not clear because the actual concentration of Zn 2ϩ in the interstitial space of an islet is not known; however, Zn 2ϩ concentration in serum is approximately 15-25 M (47-50). It has previously been demonstrated that in vivo insulin is available as free monomer within a few seconds of entering the portal vein, presumably because the large dilution of insulin drives dissociation (4); therefore, it seems unlikely that extracellular Zn 2ϩ could affect the endocrine function of free insulin.
Regulation of Insulin Extrusion during Exocytosis-Our observations show that free insulin concentrations in the vicinity of beta cells are affected by the intravesicular pH and extracellular pH and Zn 2ϩ concentrations. These results have been interpreted as effects on the rate of dissolution and dissociation as Zn 2ϩ -insulin hexamers during exocytosis. This interpretation ignores the possible role of vesicle matrix proteins, such as chromagranin, in storing and extruding free insulin. In adrenal chromaffin cells and mast cells, glycoproteins, such as chromagranin, in the vesicle matrix have been shown to be critical in storing and then extruding the secretory products (51)(52)(53)(54). These studies have shown that matrix proteins help release the secretory products in a pH-dependent mechanism that involves ion exchange and phase transition of the storage proteins (51)(52)(53)(54). Chromagranin is found in the secretory vesicles of beta cells (55)(56)(57) and therefore could play a similar role in extruding insulin. Our data cannot exclude such a possibility; however, it seems unlikely, given the differences in storage of insulin in beta cells and histamine or catecholamines in the other cells types. First, insulin is stored as a solid (4, 5, 6) and therefore is not associated with the chromagranin. Second, histochemical evidence shows that chromagranin in beta cell secretory vesicles is concentrated in the halo region separate from insulin in the dense core (57). Finally, in beta cell vesicles, insulin and C-peptide make up 80% of the protein content, and membrane-bound proteins make up 10%, leaving just 10% of the total protein as matrix proteins (58). Such a distribution would seem to preclude a dominant role for chromagranin in storing and dispersing insulin, especially given the importance of Zn 2ϩ -insulin complex formation in storage.
This study has revealed some of the chemical events involved in driving postfusion release of insulin from beta cells. The effects of H ϩ and Zn 2ϩ on insulin release that were observed are a natural manifestation of the complex mechanism of insulin synthesis and storage. The fact that the quanta of insulin release and the rate of insulin release can be altered also raises the possibility that insulin secretion can be regulated by the intravesicular and extracellular ionic environment. Such postfusion control of release has recently been proposed as a novel regulatory mechanism in a variety of cell types (52). Postfusion regulation of insulin release would seem to hold little relevance to the endocrine action of insulin given the large distance between the release site and the site of action. However, several lines of evidence support the notion of an autocrine and/or paracrine role for insulin within an islet. Specifically, insulin has been demonstrated to (a) bind to islet cells (59), (b) affect insulin secretion (61,62), and (c) activate insulin receptors in beta cell lines (60,33). Therefore, if the intravesicular or extracellular ionic environment does mediate postfusion regulation of insulin release, then its effects would likely be exerted at the level of autocrine or paracrine signaling of insulin.