Evidence for contribution by increased cytoplasmic Na+ to the insulinotropic action of PACAP38 in HIT-T15 cells.

Pituitary adenylate cyclase-activating polypeptide (PACAP) is localized to pancreatic nerve terminals and stimulates insulin secretion. The insulinotropic effect of PACAP38 in insulin-producing HIT-T15 cells is accompanied by increases in cellular cAMP and cytoplasmic Ca2+ ([Ca2+]cyt). As also intracellular Na+ is important for insulin secretion after glucose and other cAMP forming peptides, we examined the Na+ dependence of the insulinotropic effect of PACAP38 in HIT-T15 cells. We found that PACAP38 (100 nM)-induced insulin secretion was diminished by approximately 50% by removal of extracellular Na+ (replaced by equimolar N-methyl-D-glucamine). In contrast, removal of Na+ did not diminish the formation of cellular cAMP (measured by radioimmunoassay) or the increase in [Ca2+]cyt (measured in FURA-2AM-loaded cell suspensions) induced by PACAP38. Furthermore, PACAP-38 increased the cytoplasmic Na+ ([Na+]cyt) in single HIT-T15 cells as measured by the fluorophore sodium-binding benzofran isophthalate. This increase was reduced by removal of extracellular Na+ and by inhibition of protein kinase A by H-89. We conclude that the insulinotropic action of PACAP38 is Na+-dependent. We propose that PACAP38 opens plasma membrane Na+ channels by an action partially mediated by cAMP and protein kinase A, and the subsequent raise in [Na+]cyt elicits insulin secretion by an as yet unsolved mechanism.

Pituitary adenylate cyclase-activating polypeptide (PACAP) 1 was originally isolated from ovine hypothalamus where it was found to stimulate adenylate cyclase with a 1000 times greater potency than vasoactive intestinal peptide (VIP) (1). PACAP shows a high structural homology with VIP and was therefore placed in the glucagon/VIP family of peptides (1). The peptide exists in two forms. The most abundant form in all tissues is PACAP38, which consists of 38 amino acid residues, and the other form is PACAP27, corresponding to the 27 N-terminal amino acid residues of PACAP38 (2).
PACAP has been demonstrated to be a ubiquitously distributed neuropeptide throughout the body (3). In the pancreas, PACAP is localized to nerves innervating the exocrine parenchyma, blood vessels, islets of Langerhans as well as to intrapancreatic ganglia (4,5), which suggests that the neuropeptide is involved in the neural regulation of pancreatic function. We have shown recently that two types of the presently three known PACAP receptor subtypes are expressed in insulinproducing tissues, the PACAP type 1 and the VIP2/PACAP receptors, which further supports a role for PACAP in regulating islet function (5). It is well established that PACAP potently stimulates insulin secretion, as has been demonstrated in vitro in insulin-producing clonal cells (6,7), in isolated mouse and rat islets (5,8), and in perfused rat pancreas (9,10), as well as in vivo in mice (11) and humans (12). The potent insulinotropic action of PACAP has been thought to be mediated by raised formation of cellular cAMP, since PACAP stimulates cAMP formation in insulin-producing tissues (13,14) and since cAMP through activation of protein kinase A (PKA) is known to stimulate the exocytosis of insulin containing granules (15). However, we showed previously that PACAP38 (100 nM) induces insulin secretion to a greater extent than the adenylate cyclaseactivating agent forskolin (0.25 M), even though at these doses PACAP38 and forskolin induce formation of cAMP to the same extent (14). This implies that formation of cAMP cannot fully explain the insulinotropic effect of PACAP, which led us to speculate that also one or several other signaling mechanisms contribute to its insulinotropic action. Since PACAP38 also increases cytoplasmic Ca 2ϩ ([Ca 2ϩ ] cyt ) in insulin-producing HIT-T15 cells (14), such an action might contribute to the action of PACAP on insulin secretion, since Ca 2ϩ accentuates the exocytosis of granules in insulin producing cells (16). However, in addition whether other signaling mediators for PACAP exist in insulin-producing cells remains to be established.
Earlier studies in human pituitary adenoma cells have shown that PACAP induces growth hormone secretion with a mechanism that is inhibited by tetrodotoxin, a voltage-gated Na ϩ channel blocker, and that PACAP increases tetrodotoxinsensitive Na ϩ channel currents in such cells (17). This would suggest that also increased uptake of Na ϩ is a mechanism for actions induced by PACAP. Extracellular Na ϩ has been shown previously to be required for glucose-induced insulin secretion in ␤-cells (18,19), and, furthermore, the muscarinic agonist, acetylcholine, has been shown to stimulate insulin secretion in a Na ϩ -dependent manner (20). Moreover, earlier studies from our laboratory have shown that glucagon-like peptide-1 (GLP-1), which, like PACAP, activates adenylate cyclase (21), stimulates insulin secretion in a Na ϩ -dependent manner (22).
In this study, we have examined the possible contribution by Na ϩ on influences of PACAP in insulin secretory cells, by studying the Na ϩ dependence of the effect of PACAP38 on insulin secretion, on cAMP formation, and on [Ca 2ϩ ] cyt in insulin-producing clonal hamster insulinoma HIT-T15 cells.
Since pronounced Na ϩ -dependent effects were found on insulin secretion, suggesting that Na ϩ indeed is of importance for the insulinotropic action of PACAP, we proceeded and used the fluorophore Na ϩ -binding benzofran isophthalate (SBFI) to study whether PACAP38 also affects the cytoplasmic concentration of Na ϩ ([Na ϩ ] cyt ) in these cells.
Cell Culture-HIT-T15 cells, i.e. the clonal hamster ␤-cell line, were cultured at ϩ37°C in 5% CO 2 , 95% air in RPMI 1640 medium supplemented with 10% FCS, 100 units/ml penicillin G, 0.1 mg/ml streptomycin, and 2.5 g/ml amphotericin B. Passages were performed every 7 days, and the medium was changed every 3-4 days. Cells of the passages 72-82 were used.
Insulin Secretion-Cells were seeded on 24-well plates (0.5 million cells/well) and cultured for 48 h (about 80% confluence). They were then washed twice in a Hepes medium (125 mM NaCl, 5.9 mM KCl, 1.28 mM CaCl 2 , 1.2 mM MgCl 2 , 25 mM Hepes, 0.1% human serum albumin, pH 7.36) and incubated at 37°C in the Hepes medium in a volume of 200 l at 10 mM glucose with or without PACAP38. In the Na ϩ -free medium, NaCl was replaced with an equimolar concentration of N-methyl-Dglucamine (NMDG). After the end of incubation, 150 l of the medium were collected and centrifuged at 350 ϫ g for 5 min. Aliquots of 50 l in duplicate were then stored at Ϫ20°C until analysis of insulin by radioimmunoassay, using guinea pig anti-porcine insulin, mono-125 I-insulin, and, as standard, rat insulin. Free and bound radioactivity were separated by the double antibody technique.
Cellular Cyclic AMP Content-HIT-T15 cells were seeded on fourwell plates (0.5 million cells/well) and cultured for 48 h as above. The cells were then washed twice in the Hepes medium and incubated at ϩ37°C in a volume of 200 l in presence of 10 mM glucose and 0.1 mM isobutylmethylxanthine with or without addition of PACAP38 (100 nM) or forskolin (0.25 M). In the Na ϩ -free medium, NaCl was replaced by NMDG. The incubation was stopped after 2 min with addition of icecold ethanol (final concentration: 65%), and the cells were scraped off with a rubber policeman. After being washed twice in 65% ice-cold ethanol, the extracts were centrifuged at 2000 ϫ g at ϩ4°C for 15 min, transferred to fresh test-tubes, evaporated at ϩ60°C under a stream of nitrogen, and then stored at Ϫ20°C until analysis for protein content by the Lowry method (23) and for cAMP by radioimmunoassay, using a rabbit anti-succinyl-AMP serum, cyclic 2-succinyl-3-125 I-methyl ester as tracer, and cAMP as standard. Free and bound radioactivity were separated by the double antibody technique.
Cytoplasmic Ca 2ϩ -[Ca 2ϩ ] cyt was determined in FURA-2AM-loaded HIT-T15 cells as described previously (14). In brief, cells were grown for 4 -7 days in RPMI medium supplemented as above. After trypsination, cells recovered for 2 h in 10 ml of RPMI medium supplemented with 10% FCS at ϩ37°C in 5% CO 2 and were thereafter loaded with FURA-2AM (1 M) for 45 min. The cells were then washed either in a Hepes medium as described above or in a Hepes medium in which NaCl was replaced by NMDG. After equilibration for 20 min, 2 ml of the cell suspension (0.5 million cells/ml) were transferred to a cuvette for measurement of [Ca 2ϩ ] cyt in a Perkin-Elmer LS-50 spectrophotofluorometer at ϩ37°C. PACAP38 was added at defined time points and remained in the cuvette until the end of the experiment. Excitation wavelengths were 340 and 380 nm, and the emission wavelength was 510 nm. Fluorescence maximum was obtained by adding 0.03% Triton X-100 and fluorescence minimum by adding EGTA in excess at the end of experiments performed in the absence of albumin. [Ca 2ϩ ] cyt was calculated according to Grynkiewicz et al. (24).
Cytoplasmic Na ϩ -[Na ϩ ] cyt was determined in SBFI-AM-loaded cells using the protocol of Davies et al. (25). Cells were seeded on round ( 25 mm) sterile glass coverslips and cultured for 48 -72 h in RPMI 1640 medium supplemented as above. Cells attached to the coverslips were washed and loaded for 2 h with SBFI (7 M) in the presence of 0.02% Pluronic F-127 R (Sigma) in RPMI 1640 supplemented with 10% FCS at ϩ37°C in 5% CO 2 . After loading, the cells were washed again in the Hepes medium or in a Hepes medium in which NaCl was replaced with NMDG. The coverslips were then mounted in a specially designed temperature-controlled (ϩ37°C) open superifusion chamber (volume: 110 l). The coverslip with the attached cells constituted the bottom of the chamber, which was placed over a 100ϫ Fluor objective (Nikon, Tokyo, Japan) on the stage of an inverted microscope (Nikon, DIA-PHOT-TMD) with a 75-watt xenon lamp. Cannulas connected to peristaltic pumps, regulating both inflow and outflow, were fixed to the temperature-controlled chamber, and the cells were superifused at a flow rate of 0.9 ml/min in Hepes buffer (with NaCl or NMDG) supplemented with 3.3 mM glucose and 0.05% bovine serum albumin. The chamber, the stage of the inverted microscope, the peristaltic pump regulating the inflow to the chamber, as well as the experimental solutions were contained in a climate box maintained at ϩ37°C. The time for equilibration in the superifusion chamber when changing the experimental solutions of the superifusate (80 s) were taken into account when indicating the switch of experimental solutions in the graphs presented. The fluorescence of SBFI was recorded with dual wavelength excitation spectrophotofluorometry using a Nikon P1 photometer (Nikon, Tokyo, Japan) modified with a 1000-⍀ resistor to increase the sensitivity of the recordings. The excitation wavelengths were 350 and 380 nm, and the emission wavelength was 510 nm, respectively. The filter changer and the data collection were governed by software designed by Bergström Instrument AB (Solna, Sweden). The values are reported as the ratio between fluorescence at 350 versus 380 nm, which is proportional to [Na ϩ ] cyt (26).
Cytoplasmic pH-Cytoplasmic pH (pH cyt ) was measured in BCECF-AM-loaded HIT-T15 cell suspensions using a protocol of Trebilcock et al. (27), which was modified according to our technique of measuring [Ca 2ϩ ] cyt as described above. After trypsination and recovery, the cells were loaded with 4 M BCECF-AM for 45 min at ϩ37°C in 5% CO 2 . Experiments were carried out in a Hepes buffer as described above, containing 125 mM NaCl, or in a Hepes medium in which NaCl was replaced by 125 mM NMDG. The excitation wavelength was 500 nm and emission wavelength was 530 nm. Calibration was achieved by using an additional cell suspension handled the same as the experimental cell suspension (27). The cells in both preparations were lysed with 0.03% Triton X-100, and thereafter small aliquots of HCl were added to the suspensions. Fluorescence was measured in the experimental suspension, and pH in the calibration suspension, and the two curves were then plotted against each other.
Statistics-The results are reported as means Ϯ S.E. HIT-T15 cells exhibit a large variation between insulin secretion depending on passage numbers, even within the narrow range of 70 -81 (28). Therefore, besides reporting data as absolute values, the results are also normalized for each individual experiment and expressed as percent of controls run in the same experiment and assay. Statistical evaluation of differences between groups were performed by one-way analysis of variance followed by Bonferroni post-hoc test or by two-tailed Student's t test for unpaired comparisons, except regarding [Ca 2ϩ ] cyt and [Na ϩ ] cyt when statistical evaluation was performed by the Mann-Whitney U test. p Ͻ 0.05 was considered significant.

RESULTS AND DISCUSSION
Insulin Secretion-In the presence of extracellular Na ϩ , PACAP38 increased medium insulin from 1230 Ϯ 131 pmol/ liter to 6410 Ϯ 687 pmol/liter (p Ͻ 0.001), whereas in the absence of extracellular Na ϩ , medium insulin was increased from 1840 Ϯ 107 pmol/liter to only 4640 Ϯ 431 pmol/liter (p Ͻ 0.001) by the peptide, representing a reduction of the insulinotropic effect of PACAP38 by 45 Ϯ 4.0% after removal of extracellular Na ϩ (Fig. 1). Hence, the results show that the Na ϩ dependence for the action of PACAP is not restricted to pituitary cells (17), and, furthermore, that PACAP resembles glucose, acetylcholine, and the cAMP-forming peptide, GLP-1, exhibiting partial Na ϩ dependence for insulinotropic action (18 -20, 22). The Na ϩ dependence of the action of PACAP38 might be executed by Na ϩ being important for the generation of cAMP or for the increase in [Ca 2ϩ ] cyt , since PACAP38 increases cAMP formation and [Ca 2ϩ ] cyt in HIT-T15 cells (14). The potential site of the Na ϩ dependence was therefore further examined by determining cAMP and [Ca 2ϩ ] cyt in HIT-T15 cells after PACAP38 activation.
Cyclic AMP-To investigate whether removal of extracellular Na ϩ affects the PACAP38-induced formation of cellular cAMP, the cells were incubated at 10 mM glucose with the addition of 0.1 mM isobutylmethylxanthine for 2 min in the presence or absence of extracellular Na ϩ . Fig. 2 shows that PACAP increased cellular cAMP from 10.9 Ϯ 0.9 fmol/g protein to 41.4 Ϯ 2.6 fmol/g protein in the presence of Na ϩ (p Ͻ 0.001). This effect was not reduced by removing Na ϩ from the medium (13.3 Ϯ 0.4 fmol/g protein without PACAP38 versus 40.2 Ϯ 3.4 fmol/g protein with PACAP38; p Ͻ 0.001). Similarly, the effect of forskolin (0.25 M) on cellular cAMP formation was not affected by removal of extracellular Na ϩ , since cellular cAMP content was 50.1 Ϯ 5.7 fmol/g protein in the presence of extracellular Na ϩ and 43.5 Ϯ 1.7 fmol/g protein in a Na ϩ -free medium after stimulation by forskolin (not significant, Fig. 2). These results thus show that the activation of adenylate cyclase by PACAP38 or forskolin in HIT-T15 cells is a process not dependent on Na ϩ . This is in contrast to previous studies in parotid glands showing that the binding of G s -protein to the catalytic unit of adenylate cyclase is Na ϩ -dependent (29), which suggests that the Na ϩ dependence of G s -protein binding to adenylate cyclase is different in different cell systems or that the PACAP-activated G s is different from other G s -proteins in this respect. A third possibility is that a minimal amount of Na ϩ remains intracellularly despite incubating the cells in a medium devoid of Na ϩ during experiments, and this is sufficient to facilitate the binding of the G s -protein to adenylate cyclase. In any case, our results show that the Na ϩ dependence of the insulinotropic effect of PACAP38 is located further downstream from adenylate cyclase of the intracellular pathway in HIT-T15 cells or executed by signaling mechanisms of PACAP not involving cAMP.
Cytoplasmic Ca 2ϩ -It is known that PACAP38 increases [Ca 2ϩ ] cyt in HIT-T15 cells (14). In isolated rat islets, uptake of extracellular Ca 2ϩ induced by GLP-1 is abolished by removal of extracellular Na ϩ (22). This would infer that a site of the Na ϩ dependence for GLP-1, and therefore perhaps also for PACAP38, resides in the mechanism of increased [Ca 2ϩ ] cyt . We therefore examined the Na ϩ dependence of the increase in [Ca 2ϩ ] cyt induced by PACAP38 in FURA-2-AM-loaded cell suspensions. We found, as seen in Fig. 3, that PACAP38 increased the cytoplasmic concentration of Ca 2ϩ both in the presence and in the absence of extracellular Na ϩ . Furthermore, in the absence of extracellular Na ϩ , the increase in [Ca 2ϩ ] cyt in response to PACAP38 was greater than in the control cells incubated in the presence of extracellular Na ϩ . Thus, at 300 s after addition of PACAP38, [Ca 2ϩ ] cyt had increased by 200 Ϯ 8 nmol/liter in the absence of extracellular Na ϩ versus 52.0 Ϯ 7 nmol/liter in the presence of extracellular Na ϩ (p Ͻ 0.001; Fig. 3, A and B). Therefore, in contrast to previous results that the Ca 2ϩ uptake in isolated rat islets in response to GLP-1 was reduced by removal of extracellular Na ϩ (22), PACAP38 induced an exaggerated increase in [Ca 2ϩ ] cyt after omission of Na ϩ from the medium. Hence, the Na ϩ dependence of the insulinotropic action of PACAP38 does not reside in impaired increase in [Ca 2ϩ ] cyt . The finding that the cytoplasmic concentration of Ca 2ϩ increased after activation by PACAP38 also in the absence of extracellular Na ϩ suggests that the opening of Ca 2ϩ channels is not dependent on depolarization of the cell by the influx of Na ϩ , but is instead dependent on other mechanisms. One such mechanism could involve opening of PKA-dependent Ca 2ϩ channels (15). We have shown previously that in HIT-T15 cells, the PACAP38-induced increase in [Ca 2ϩ ] cyt is abolished when adenylate cyclase is activated by forskolin prior to introduction of PACAP38, suggesting that cAMP in fact mediates the opening of membranous Ca 2ϩ channels (14). Another possibility involves opening of Ca 2ϩ channels directly coupled to a G-protein activated by PACAP38. Such a Ca 2ϩ channel, which is directly opened after activation of a G-protein, has been described previously to be coupled to at least one of the PACAP receptors expressed in the pancreatic endocrine tissue (30). Further studies are required to examine this possibility. The potentiated PACAP38-induced increase in [Ca 2ϩ ] cyt in the absence of extracellular Na ϩ is interpreted to reflect that the countertransport through Na ϩ -Ca 2ϩ exchange ion channels is prevented by removal of extracellular Na ϩ . Such countertransport is of importance for the cellular Ca 2ϩ homeostasis (31), and when prevented by removal of extracellular Na ϩ , the increase in [Ca 2ϩ ] cyt by PACAP38 is accentuated.
Cytoplasmic Na ϩ -The results above imply that Na ϩ is involved in the insulinotropic effect of PACAP38 in HIT-T15 cells, although the exact site of this involvement is still an open question. Indirect actions through cAMP or Ca 2ϩ seem less likely, since neither the formation of cAMP nor the increase in [Ca 2ϩ ] cyt after PACAP38 was reduced by removal of Na ϩ , although the insulin secretory response to PACAP38 was impaired. This pattern of effects could be executed by activation of a channel increasing the uptake of Na ϩ , with a subsequent increase in cytoplasmic concentration of Na ϩ yielding a direct secretory action of intracellular Na ϩ . To examine this possibility, we measured [Na ϩ ] cyt by using the fluorophore SBFI in HIT-T15 cells. Measurement of [Na ϩ ] cyt in insulin-producing cells by using SBFI has been performed previously by several groups showing that glucose, glyceraldehyde, and acetylcholine increase the [Na ϩ ] cyt , which might contribute to the insulinotropic action of these secretagogues (25,26,32,33). In our hands, measurement of [Na ϩ ] cyt in suspensions of HIT-T15 cells by using SBFI did not yield reliable results, despite extensive trials in our laboratory (data not shown). We therefore proceeded to measure [Na ϩ ] cyt in single cells, which proved successful. As is seen in Fig. 4A, PACAP38 increased [Na ϩ ] cyt in a medium containing both extracellular Na ϩ and Ca 2ϩ . At 400 s after introduction of PACAP38, the ratio of [Na ϩ ] cyt fluorescence had increased from 0.68 Ϯ 0.02 in controls to 0.80 Ϯ 0.03 (p ϭ 0.018) after stimulation by PACAP38. Fig. 4B shows a typical trace of such an experiment. In contrast, when extracellular Na ϩ was replaced with NMDG (125 mM), the effect of PACAP38 was abolished (Fig. 4C). This implies that PACAP38 activates a Na ϩ channel mediating the uptake of extracellular Na ϩ , thereby increasing the intracellular Na ϩ concentration. The PACAP38-induced increase in [Na ϩ ] cyt was abolished also when the cells were treated with 20 M H89 (Fig.  4C), which is a specific PKA inhibitor (34). These results imply that PACAP38 causes an uptake of Na ϩ into the cells and that this uptake is mediated by PKA and therefore probably executed through a PKA-sensitive Na ϩ channel. These findings support the important notion by Leech et al. (35) that PACAP stimulates an inward current in HIT-T15 cells, which is mainly caused by an influx of Na ϩ into the cells and which is mediated by cAMP. This is also consistent with a previous study showing that cAMP activates a cation channel in the rat insulinoma cell line CRI-G1 (36).
Cytoplasmic pH-A possible confounding factor in studying effects of removal of Na ϩ from the extracellular medium on insulin secretion is the potential influence through changes in pH cyt . Omission of Na ϩ from the medium abolishes the Na ϩ -H ϩ exchange, thus preventing H ϩ to be transported out from the cytoplasm, which has been shown previously to be of importance for the regulation of pH cyt in the ␤-cell (37). Therefore, removal of extracellular Na ϩ might reduce the pH cyt , which might then affect exocytosis. The inhibitory effect of removal of extracellular Na ϩ on PACAP38-induced insulin secretion might thus partially be explained by alteration in pH cyt . To examine whether removal of Na ϩ actually affects pH cyt in HIT-T15 cells, we measured pH cyt in cell suspensions by using the fluorophore BCECF-AM. We found that removal of extracellular Na ϩ expectedly decreased pH cyt (Fig. 5). Thus, at time ] cyt 300 s after introduction of PACAP38 is shown in a medium with extracellular Na ϩ (n ϭ 10; open bar), after addition of PACAP38 in a medium in which Na ϩ had been replaced with NMDG (n ϭ 4; gray bar), and after addition of PACAP38 in a medium containing the specific PKA inhibitor, H89 (20 M; n ϭ 3; hatched bar). Asterisks indicate the probability level of random difference versus experiments run with extracellular Na ϩ . **, p Ͻ 0.01; ***, p Ͻ 0.001. 0, pH cyt was 7.41 Ϯ 0.04 in the presence of extracellular Na ϩ and 7.02 Ϯ 0.11 (p ϭ 0.035) when Na ϩ in the medium was replaced with NMDG (125 mM). pH cyt was stable throughout the study period, and therefore this difference in pH cyt , in the presence versus in the absence of extracellular Na ϩ , persisted throughout the study period (Fig. 5). Introduction of PACAP38 did not alter pH cyt , neither in the presence nor in the absence of extracellular Na ϩ . To test whether this reduction in pH cyt contributes to the reduced insulin secretion seen after PACAP38 in a Na ϩ -free medium, HIT-T15 cells were incubated with or without PACAP38 in media of different extracellular pH. The cellular buffering of changes in extracellular pH (38) was compensated by decreasing the extracellular pH to 6.8 to examine the influence of pH cyt of 7.0. We found that lowering of extracellular pH from 7.36 to 6.8 did not affect insulin secretion stimulated by 10 mM glucose alone or by 10 mM glucose together with 100 nM PACAP38 (Fig. 6). Thus, the inhibitory effect of removal of Na ϩ on PACAP38-induced insulin secretion cannot be explained by the accompanying change in pH cyt .
Summary and Conclusions-Our results thus show that the insulinotropic effect of PACAP38 is partially (ϳ50%) dependent on extracellular Na ϩ . PACAP38 therefore resembles GLP-1 in this respect (22), and the Na ϩ dependence of actions of PACAP is not restricted to pituitary cells (17). The failure of removal of extracellular Na ϩ to completely abolish the insulinotropic action of PACAP38 infers that the signaling pathway mediating the insulinotropic effect of PACAP38 may be partly activated in the absence of extracellular Na ϩ or, alternatively, that several intracellular pathways might be involved, whereof at least one is strictly Na ϩ -sensitive. Examining these possibilities, we found that the Na ϩ dependence does not involve the formation of cAMP or the increase in [Ca 2ϩ ] cyt and that PACAP38 increases [Na ϩ ] cyt by a mechanism abolished by inhibition of PKA. This shows that neither activation of adenylate cyclase nor a rise in [Ca 2ϩ ] cyt can explain the Na ϩ dependence of PACAP and suggests that cAMP and raised [Ca 2ϩ ] cyt can not solely explain PACAP38-induced insulin secretion. Instead, our results are integratively interpreted to indicate that PACAP38 Na ϩ -independently stimulates formation of cAMP, which activates PKA, which, in turn, opens both Ca 2ϩ and Na ϩ channels. The subsequent influx of Na ϩ raises the cytoplasmic level of Na ϩ , which then contributes to the PACAP-induced insulin secretion. This contribution could be mediated by depolarization induced by increased intracellular Na ϩ , which, in turn, could augment Ca 2ϩ uptake by opening of voltage sensitive Ca 2ϩ channels. A tentative remaining possibility is that Na ϩ per se is of importance for the exocytotic mechanism. However, although exocytosis of secretory granules in ␤-cells has been shown to be mediated by several proteins (39), the Na ϩ dependence of the action of these proteins remains to be studied. Finally, it should be emphasized that in several excitable tissues, such as neuronal cells, influx of Na ϩ through voltagegated ion channels plays a major role in cell activation by inducing depolarization (40). Our present results therefore strengthen the similarity between insulin producing cells and other excitable cells also in this respect.