Stimulus-Response Coupling in Insulin-secreting HIT Cells EFFECTS OF ON AND

The hamster islet B cell line HIT retains the ability to secrete insulin in response to glucose and several receptor agonists. We used HIT cells to study the initial signaling events in glucose or receptor agonist-stimu-lated insulin secretion. Glucose stimulated insulin re- from HIT cells in a dose-dependent manner with a half-maximal effect seen already at 1 mM. was also stimulated by carbachol in a glucose- dependent manner. depolarized the HIT cell membrane potential as assessed with the fluorescent probe and raised intracellular Ca2+ as fura-measurements. a Mn2+ fura- quenching

The hamster islet B cell line HIT retains the ability to secrete insulin in response to glucose and several receptor agonists.
We used HIT cells to study the initial signaling events in glucose or receptor agonist-stimulated insulin secretion. Glucose stimulated insulin release from HIT cells in a dose-dependent manner with a half-maximal effect seen already at 1 mM. Insulin release was also stimulated by carbachol in a glucosedependent manner.
Glucose depolarized the HIT cell membrane potential as assessed with the fluorescent probe bisoxonol and raised intracellular Ca2+ as revealed by fura-measurements.
Using a Mn2+ furaquenching technique, we could show that the rise in intracellular Ca2+ was due to Ca2+ influx following opening of voltage-gated Ca2+ channels. Glucose is thought to increase the diacylglycerol (DAG) content of insulin-secreting cells. However, although HIT cells respond to glucose in terms of insulin secretion, membrane depolarization, and Ca2+ rise, the hexose was unable to increase the proportion of protein kinase C activity associated with membranes. In eontrast, the membrane-associated protein kinase C activity increased in HIT cells exposed to the two receptor agonists carbachol and bombesin. Bombesin was shown to generate DAG with the expected fatty acid composition of activators of phospholipase C. Glucose, in contrast, only caused minor increases in DAG containing myristic and palmitic acid without affecting total DAG mass. The failure to detect stimulation of protein kinase C by glucose could be due to both the limited amount and to the different fatty acid composition of the metabolically generated DAG. The latter was in part supported by experiments performed on protein kinase C partially purified from HIT cells. Indeed, 1,2dipalmitoylglycerol, presumed to be the main DAG species generated by glucose, was only one-third as active as 1,2-dioleoylglycerol and 1-stearoyl-2-arachidonylglycerol in stimulating the isolated enzyme at physiological Ca2+ concentration. It is therefore unlikely that DAG and protein kinase C play a major role in glucose-stimulated insulin secretion.
Insulin secretion from the pancreatic B cell is influenced by a variety of physiological factors. These include the elevation of glucose concentration in the circulation and the * This work was supported by Swiss National Science Foundation Grants 3.214-085 and 32.25665.88 (to C. B. W.) and 31.26513.89 (to J. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. release of acetylcholine from nerve endings within the pancreatic islets (1). It is well established that glucose depolarizes the B cell membrane potential by closing ATP-sensitive K' channels, in turn causing the gating of voltage-dependent Ca*+ channels (2). The resulting rise of cytosolic Ca2+ ([Ca"+]i)' can be explained entirely by Ca*+ influx and precedes the stimulation of insulin secretion (3). In contrast, receptor agonists such as cholinergic substances and bombesin lead to a [Ca*+]i rise mainly due to mobilization of Ca*+ from intracellular stores (4,5). This effect is mediated by the second messenger inositol 1,4,5-trisphosphate, which is generated, together with 1,2-diacylglycerol (DAG), by the hydrolysis of phosphatidylinositol 4,5-bisphosphate (6). The increase of DAG in the plasma membrane is thought to activate protein kinase C a Ca"+-and phospholipid-dependent enzyme (7, 8). Stimulation of pancreatic islets with glucose resulted in a similar increase of total mass of DAG as that evoked by carbachol in two studies (9, lo), whereas no increase was found in a third (11). Although glucose promotes phosphati-dylinositol4,5-bisphosphate hydrolysis (12,13), the DAG was believed to be formed mainly by de lzouo synthesis via glycolytic intermediates and phosphatidic acid (9,11,14-16). However, a recent study measuring total islet mass of DAG by an enzymatic method did not find any change due to glucose, despite stimulation of the de nouo synthesis pathway (11). Activators of protein kinase C, such as phorbol esters and synthetic diacylglycerols, have been shown to stimulate insulin release (17)(18)(19), but the involvement of this enzyme during the stimulation of intact B cells with physiological secretagogues remains controversial.
As expected, the muscarinic agonist carbachol activates protein kinase C in islets as measured either by the association of the enzyme with membranes (20) or by phosphorylation of a protein substrate (21). Translocation also occurred in the insulin-secreting cell line RINr (22). With respect to carbohydrate stimuli, the situation is less clear. In islets, glucose failed to elicit protein kinase C translocation (20, 23), while causing a small phosphorylation of a protein kinase C substrate (21). In RINm5F cells, which do not recognize glucose as a secretagogue (24), D-glyceraldehyde caused a marked redistribution of the enzyme (25).
The purpose of the present study was to examine whether both Ca'+-mobilizing receptor agonists and glucose activate protein kinase C as a step in the initiation of insulin secretion. We employed the hamster B cell line HIT-T15, which has been shown to respond to various secretagogues including glucose (26,27 were used for the incubation, the tubes were rapidly placed on ice and the cells were then pelleted for 10 s in an Eppendorf microcentrifuge, followed by the addition of HB. The homogenate was centrifuged for 1 h at 100,000 x g, resulting in the cytosol and a 100,000 X g pellet. The latter was resuspended by sonication in 1 ml of HB supplemented with 1% Triton X-100 and centrifuged again for

Insulin secretion from HIT cells was assessed in static incubations
following a period of glucose deprivation as described by Meglasson et al. (27). The results in Fig. 1 demonstrate that glucose stimulates insulin secretion from HIT cells in a dose-dependent manner with a half-maximal effect at approximately 1 mM. The preincubation in the absence of glucose in these experiments was 60 min, but similar results were obtained after 30 min. The addition of 100 PM carbachol enhanced insulin secretion in a glucose-dependent manner with an optimal effect seen at 3 mM glucose (Fig. 1). This glucose dependency is similar to that found in pancreatic islets (1). In contrast, the sensitivity to glucose alone is exaggerated in HIT cells.
Glucose depolarized the HIT cell membrane potential as assessed with the fluorescent probe bisoxonol. The effect was apparent at 1 mM and maximal at 10 mM (Fig. 2.4) After 3 h of spinner culture, the cells were preincubated for 1 h in a glucose-free modified Krebs-Ringer bicarbonate buffer. They were then exposed to different glucose concentrations in the absence or presence of carbachol for 15 min. The results are given as mean f S.E. (15 mM) did not depolarize the cells further (not shown). Although it is known that glucose opens voltagedependent Ca'+ channels, the contribution of Ca2+ influx to the extent of the depolarization is not clear. It can be seen in Fig. 2C that the effect of glucose on membrane potential was reduced by approximately 70% when the extracellular Ca2+ concentration was reduced acutely by the addition of 2 mM EGTA to Ca'+-containing buffer. Preincubation of the cells for 30 min in nominally Ca*+-free buffer, followed by the acute addition of 0.4 mM EGTA, resulted in a marked reduction of the effect of both glucose and K+ (Fig. 20). Similar results were obtained when the Ca" channel blocker nifedipine was added before glucose (Fig. 2E). This suggests that part of the depolarizing current in HIT cells stimulated with glucose is due to Ca2+ influx. Experiments were then performed to show Ca2+ entry. To this end, Mn*+ was added to the extracellular medium in order to enable the monitoring of Mn*+ influx into HIT cells. It has been shown previously that Mn*+ quenches the fluorescence of quin 2 and fura-and that the decrease of fluorescence in dye-loaded cells reflects Mn2+ influx through Ca*' conductance channels (30-32). As can be seen in Fig.  3A, the addition of Mn2+ (300 MM) to fura-2-loaded HIT cells caused a rapid decrease in fluorescence due to the quenching of extracellular fura-2. Thereafter, the fluorescence steadily decreased, reflecting the slow uptake of Mn2+ by the cells. The addition of 10 mM glucose resulted in an accelerated uptake of Mr?+, which continued until voltage-dependent Ca*+ channels were blocked with 20 pM verapamil.
When verapamil was added prior to glucose, the rate of Mn2+ entry was not altered by the stimulus (data not shown). Theoretically, the increased dye quenching could be explained by increased leakage of fura-from the glucose-stimulated cells. We examined this possibility by limiting fura-efflux with the anion channel blocker sulfinpyrazone (29). The results in Fig. 3B obtained in the presence of 200 pM sulfinpyrazone are qualitatively similar to those shown in Fig. 3A. This effect is also observed in nominally Ca'+-free medium, where it is possible to reduce the Mn*+ concentration to 100 pM due to the competition of Mn*+ and Ca*+ for the same ion channel (Fig. 3C). Thus, it appears that glucose stimulates Mn*' influx through a verapamil-sensitive conductance pathway. Changes in [Ca2+lr were monitored in fura-2-loaded HIT cells preincubated for 30 min in glucose-free medium at room temperature to limit fura-leakage, which is a temperaturedependent process. Under these conditions, glucose raised [Ca2+li from 137 f 9 nM to 222 f 28 nM (n = 7). A representative trace of the seven experiments is shown in Fig. 4A. The rise in [Ca'+ll was attenuated by verapamil, added either after (Fig. 4A) or before (Fig. 4B) glucose. In the latter case, the values were 128 + 8 and 139 f 13 nM (n = 4) before and after the addition of glucose, respectively. Similar results were obtained when 5 pM nifedipine was used to block voltagedependent Ca2+ channels (data not shown). As expected, the Ca*+ ionophore ionomycin still caused a [Ca'+]i rise in the presence of verapamil (Fig. 4B). The action of glucose on [Ca*+], was then tested in the presence of 200 pM sulfinpyrazone, which was also present during the 30-min preincubation period at 37 "C. Again, glucose raised [Ca"+]i from 204 f 9 to 318 f 19 nM (n = 6), which is illustrated in Fig. 4C. The addition of 5 pM nifedipine caused an immediate return of [Ca"+]; to basal levels (Fig. 4C). In the absence of extracellular Ca*+, glucose failed to raise [Ca*+]i (not shown). Taken together, these results demonstrate that glucose raises [Ca2+11 by stimulating Ca*+ influx following the opening of verapamiland nifedipine-sensitive Ca'+ channels. As demonstrated previously in RINm5F cells (4), in mouse islet cells (36), in single HIT cells (37), and in B cells (38), carbachol enhanced [Ca*+], in suspensions of HIT cells (Fig.  40). The agonist increased [Ca2+li from 151 f 24 to 225 t 37 nM (n = 3). Thus, it appears that HIT cells respond to glucose and carbachol with both a [Ca"], rise and insulin secretion in a manner to be expected of these two secretagogues.
In the following experiments, we examined whether both receptor agonists and glucose activate protein kinase C in HIT cells to estimate the involvement of this enzyme in insulin secretion. Muscarinic agonists lead to the activation of phospholipase C and the generation of DAG in pancreatic islets (9,lO). Protein kinase C attaches to cellular membranes Suspensions of HIT cells were loaded with the fluorescent Ca*' indicator fura-and preincubated in the absence of glucose for 30-40 min (see "Experimental Procedures"). A and B were performed in Ca'+-containing buffer and C in nominally Ca*+-free buffer. In B and C, 200 pM sulfmpyrazone was present in the cuvette. The initial rapid decrease in fluorescence is due to quenching of extracellular fura-2. The traces are representative of at least three separate experiments. when it is activated, permitting the assessment of the degree of enzyme activation by measuring the subcellular distribution after exposure of intact cells to secretagogues. Using the homogenization protocol described under "Experimental Procedures," about the same amount of protein was obtained in both cytosolic and microsomal fractions. Under resting conditions, 84 f 1% (n = 5) of the protein kinase C activity was found to be soluble, while 16 + 1% (n = 5) was recovered in the microsomal extract. As shown in Fig. 5, in cells exposed to 100 pM carbachol, there was a rapid increase in the amount of protein kinase C associated with the cellular membranes and a corresponding decrease in the cytosolic fraction. Maximal redistribution of protein kinase C was reached after 2 min of incubation in the presence of the agonist and was still seen after 10 min. In contrast, no changes in the subcellular localization of the enzyme were found in the controls even after 10 min of incubation (not shown). In order to ascertain if exposure of HIT cells to glucose could evoke protein kinase C activation, these cells were HIT cells in monolayer culture were exposed to carbachol(lO0 pM) in the modified Krebs-Ringer bicarbonate buffer containing 2.8 mM glucose. The cell incubation was stopped and subcellular fractions prepared as described under "Experimental Procedures." Protein kinase C (PKC) activity was measured in the cytosolic and microsomal fraction after partial purification of the enzyme by polyacrylamide gel electrophoresis. The results are given as mean + SE. of three to five independent observations. preincubated for 30 min in the absence of glucose and then stimulated for 10 min with 10 mM of hexose. As seen in Table  I, incubation of the cells in the presence of glucose did not change the subcellular localization of protein kinase C. Similar results were obtained when HIT cells were treated with glucose for 5 or 15 min (not shown). In contrast, in paired experiments, carbachol (100 pM) after 10 min of treatment caused an increase in protein kinase C activity associated with the microsomal fraction (Table I). In order to exclude that the failure to detect protein kinase C redistribution in glucose-stimulated cells was due to the method used, the protocol was varied in two ways. First, protein kinase C was assayed in the presence of phosphatidylserine and phorbolmyristateacetate and with histone H III-S as substrate instead of protamine (see "Experimental Procedures"). This approach yielded quantitatively similar results (data not shown). Second, the cells were incubated under conditions identical with  (Table I). Both approaches failed to reveal an effect of glucose on protein kinase C subcellular distribution. Bombesin, another activator of phospholipase C, has been shown to raise [Ca2'li, increase DAG turnover, and stimulate insulin secretion in suspensions of HIT cells (5). In the present study, bombesin (100 nM) activated protein kinase C in suspensions of HIT cells, after both 2 and 10 min of stimulation (Table I). Thus, under conditions where glucose, like carbachol and bombesin, raises [Ca'+], and stimulates insulin secretion, only the receptor agonists cause a detectable increase in protein kinase C activity associated with the membrane fraction.
The absence of protein kinase C activation by glucose could be due either to the failure to generate DAG in HIT cells, or, if indeed generated, the DAG species might be inefficient in stimulating protein kinase C. The fatty acid composition of DAG in bombesin-or glucose-stimulated HIT cells was measured by gas-liquid chromatography (see "Experimental Procedures"). After 30 min of preincubation in the absence of glucose, the cells were incubated either for 2 min with or without bombesin (100 nM) or for 10 min with or without glucose (10 mM). Under resting conditions, the main fatty acid species detected in DAG were palmitic, oleic, and stearic acids. These results are similar to our previous studies in rat islets except for a much lower content of arachidonic acid in HIT cells (9). Bombesin caused a small but significant increase in palmitic (lo%), oleic (22%), stearic (22%), and arachidonic acids (50%) (see Fig. 6, top). Glucose stimulation, in contrast, only caused a slight increase in DAG containing myristic (20%) and palmitic acids (15%) (Fig. 6, bottom). Thus, in HIT cells, a receptor agonist causes an increase of DAG with the composition expected for DAG originating from inositol-containing phospholipids (6,9). Glucose, on the other hand, if at all, only generates DAG enriched in saturated fatty acids. Estimation of the total DAG mass from the data in Fig.  6 yields the following values (in picomoles/106 cells, mean + S.E., n = 4): control 124 + 6, bombesin 142 + 6 @ < 0.025 Student's paired t test); control 137 f 8, glucose 146 f 9 (p > 0.05 paired t test). The effect of the two types of stimuli on the total DAG mass was also examined using an enzymatic method (35). At 10 min, bombesin increased the DAG mass from 56 f 6 to 70 f 3 pmol/106 cells (p < 0.05), whereas the value for the cells incubated in parallel with 10 mM glucose was 57 + 3. After 30 min of glucose stimulation, the DAG level was 58 + 2 pmol/106 cells relative to 59 f 8 in controls. Expressed per milligram of cellular protein, the latter value becomes 432 pmol/mg, which is similar to the amount observed in a recent publication on HIT cells (39). Next, we tested DAG of varying fatty acid compositions for their efficacy in activating protein kinase C partially purified from HIT cells. These cells only contain the ol isoform of the enzyme (8) as demonstrated by Western blotting using monoclonal antibodies directed against the a, ~3, and y isoforms (data not shown). The experiments were performed at two different Ca2+ concentrations (2 and 25 PM). At 25 pM, Ca2+ 1,2-dipalmitoylglycerol caused activation of protein kinase C similar to that caused by 1,2-dioleoylglycerol and l-stearoyl-2-arachidonylglycerol, the main species generated from inositol-containing phospholipids (6) ( Table II). 1,2-Distearoylglycerol was slightly less active while l-monooleoylglycerol was inactive. At this Ca2+ concentration, phosphatidylserine alone stimulated enzyme activity 7-fold. In contrast, at 2 PM Ca*', phosphatidylserine alone did not significantly alter protein kinase C activity. Under these conditions, 1,2-dipalmitoylglycerol caused a doubling of the enzyme activity that, however, was only one-third of the effect seen with 1,2dioleoylglycerol and 1-stearoyl-2-arachidonylglycerol (Table  II). Again, 1-monooleoylglycerol was inactive. Other experiments performed at various Ca2+ concentrations confirmed these results. Thus, the fatty acid composition, as well as differences in the amount of DAG generated, could be responsible for the difference in protein kinase C activation between glucose and receptor agonists acting on phospholipase C.

DISCUSSION
In this study, we employed HIT cells, a B cell line that, in contrast to RINm5F cells, secretes insulin in response to glucose (26,27,39,40). In order to demonstrate that HIT cells behave like islet cells, it was important to ascertain that glucose depolarizes HIT cells and promotes the subsequent opening of voltage-dependent Ca'+ channels. Using the membrane potential probe bisoxonol, we could show that glucose indeed depolarizes these cells. The effect was already seen at 1 mM glucose, a concentration causing half-maximal insulin secretion (27). It should be noted that HIT cells display an exaggerated sensitivity to glucose compared to pancreatic islets, where the half-maximal glucose concentration is about 8 mM (1). The bisoxonol measurements suggest that part of the glucose effect is due to Ca2+ influx, since the depolarization was attenuated in Ca'+-free medium. As the effect of K+ was also diminished by Ca2+ removal, it is possible that the depolarizing current is in part carried by Ca*+ because of the abundance of voltage-dependent Ca*+ channels in this cell line (41). By using the Mn*+-quench technique previously described for platelets (30), neutrophils (31), and endothelial cells (32), it was possible to monitor Mn*+ influx through Ca*+ conductance pathways. Glucose stimulated Mn*+ influx and the effect was blocked by the Ca*+ channel antagonists verapamil and nifedipine. This is a direct demonstration of the opening of Ca*+ channels by glucose. The effect was observed both in the presence and absence of extracellular Ca*+, suggesting that glucose-mediated gating of Ca*+ channels is similar under both conditions. As expected, glucose causes the rise in [Ca*+]i, which, in this system, remained elevated for a long period of time. The effect was entirely due to Ca*+ influx since it was inhibited by Ca*+ channel blockers and was abolished in the absence of external Ca*+. At 37 "C, HIT cells display fura-leakage that, in other cell types, has been shown to be attenuated by the anion channel blockers probenecid and sulfinpyrazone (29, 42). Application of sulfinpyrazone largely inhibited the loss of fura-and confirms the conclusion that glucose raises [Ca*+]i by stimulating Ca2+ influx. Glucose and glyceraldehyde have been demonstrated to depolarize insulin-secreting cells by closing ATP-sensitive K' channels (2). This has also been shown for glucose in HIT cells (43). Consequently, the HIT cells thus appeared to be a valid model for the investigation as to whether glucose stimulation is accompanied by protein kinase C activation. We chose to assess protein kinase C activation by monitoring the association of the enzyme with cellular membranes. A similar methodology has been successfully applied to the investigation of agonist stimulation in many cell types (44,45). Following stimulation of intact cells with receptor agonists, an increased proportion of the enzyme becomes associ-ated with membranes. This intercalation depends on the receptor-mediated formation of DAG and on the rise of [CP]i (46). Stimulation of HIT cells with the muscarinic agonist carbachol, which causes a transient [Ca*+]i rise, led to an approximately 50% increase of membrane-associated protein kinase C activity. The effect was first observed after 1 min and maintained for 10 min. Similar findings have been reported in other cell systems (47). Such an effect of carbachol has also been reported for RINr cells (22), another insulinsecreting cell line, and for isolated islets (20). Bombesin, another activator of phospholipase C in HIT cells (5), also promoted membrane association of protein kinase C (Table  I). Thus, the binding of carbachol and bombesin to their respective receptors on insulin-secreting cells causes not only the formation of 1,4,&trisphosphate, the subsequent rise in Ca*+, and the generation of DAG, but also the activation of protein kinase C.
The receptor occupation is transduced by a G-protein that stimulates phospholipase C (48). In contrast, the mechanism by which carbohydrates stimulate insulin secretion is less well understood. Both glucose and glyceraldehyde depolarize the plasma membrane potential of insulin-secreting cells by closing ATP-sensitive K' channels (2, 43). It has been proposed that channel closure is due to increases in ATP or in the ATP/ADP ratio following the metabolism of carbohydrates (2). Because of the high cytosolic ATP levels, additional coupling factors have been proposed for the control of the ATP-sensitive K' channel. Activators of protein kinase C have been shown to depolarize insulin-secreting RINmSF cells by closing ATP-sensitive K+ channels, and to mimic the effect of glyceraldehyde in this cell system (19). As carbohydrates have been shown to promote de nouo synthesis of DAG (9,11,(14)(15)(16)19), this mechanism provided an attractive alternative for the coupling of glucose metabolism to membrane depolarization and insulin secretion. In the present study, glucose only caused a minor increase in the DAG enriched in myristic and palmitic acid without significantly altering total DAG mass. This contrasts with our previous findings (9) and those of Wolf et al. (10) in rat islets demonstrating a sizeable increase in palmitic acid containing DAG following glucose stimulation. A rise in total DAG mass was also reported in HIT cells after 30 min of glucose stimulation (39). Recently, however, Wolf et al. (11) reported that glucose exposure up to 20 min failed to enhance DAG mass in both rat and human islets. The latter study (11) also provides evidence suggesting that de nouo synthesis of DAG from glucose contributes only slightly to islet DAG mass. The reason for these discrepant results is unclear. Nonetheless, our experiments demonstrate that, in HIT cells, glucose can trigger the ionic events, including the rise in [Ca'+];, and subsequent insulin secretion in the face of minor changes in DAG. Considering the main target for DAG, protein kinase C, it could be that the small amount of DAG generated still activates the enzyme. However, in contrast to the receptor agonists bombesin and carbachol, glucose was unable to cause membrane-association of protein kinase C. Similar results have been reported for islets (20,23). In view of the moderate increase in DAG mass in bombesin-stimulated cells, the absence of detectable protein kinase C translocation after glucose exposure could be due not only to the small amount of DAG generated in response to carbohydrates, but also to its different fatty acid composition in receptor agonist-and glucose-stimulated cells (Fig. 6) (9, 10). Indeed, paimitic acidenriched DAG (glucose) was found to be less efficient in activating protein kinase C than DAG containing unsaturated fatty acids in position 2 (receptor agonists). These findings agree with previous observations on the characteristics of protein kinase C from brain (49).
Thus, more and more evidence is accumulating against the involvement of protein kinase C in glucose-stimulated insulin secretion. In view of the present findings and those of Wolf et al. (ll), DAG does not seem to be the coupling factor linking glucose metabolism to membrane ion fluxes and secretory events. Other lipidic compounds whose concentration increases more dramatically during glucose stimulation could constitute such a link (39).