INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Pancreatic islet -cell secretion of insulin stimulated by glucose is dependent upon elevations in intracellular calcium concentration ([Ca²⁺]_i)¹ (1). Together, two voltage-dependent processes have been proposed to constitute the glucose-stimulated rise in [Ca²⁺]_i: the influx of extracellular Ca²⁺ through Ca²⁺ channels (2-4) and the release of intracellular Ca²⁺ from the endoplasmic reticulum (ER) (5-7). Glucose initiates changes in membrane potential via an increase in cytosolic ATP, derived from the glycolytic reduction of NAD⁺, that results in block of ATP-sensitive K⁺ channels (KATP) (8, 9). Consequently, the membrane potential slowly depolarizes from approximately 65 mV, whereupon trains of action potentials occur (the active phase), interrupted by quiescent periods of hyperpolarization (the silent phase), both forming the so-called electrical bursting behavior characteristic of glucose-stimulated islets (10-12). Closely coupled with this bursting activity are oscillations in [Ca²⁺]_i that ultimately determine the oscillatory nature of insulin secretion (5, 11-14). The mechanism underlying these oscillations in membrane potential, [Ca²⁺]_i, and insulin secretion remains uncertain, but most likely stems from a cyclical activation of an ion conductance governed indirectly by metabolism, changes in [Ca²⁺]_i, or some hitherto uncharacterized intracellular mediator (15, 16).

Confounding an early resolution of this puzzle is the profusion of membrane currents present in -cells. Thus, in addition to ATP-sensitive K⁺ current (I_KATP) and voltage-dependent Ca²⁺ (I_Ca) and Na⁺ currents, delayed rectifier K⁺ (I_KDR), Ca²⁺-activated K⁺ (I_KCa), and G-protein-coupled inward rectifier K⁺ currents have been described, all of which could contribute to the regulation of the -cell membrane potential by glucose (17-20). More recently, a nonselective cation current whose conductance is regulated by the Ca²⁺ content of the ER (I_CRAN; Ca²⁺ release-activated nonselective cation current) was found in mouse -cells (21). This current could be activated indirectly by ER Ca²⁺ store depletion or directly by maitotoxin (MTX) (12, 21). Carried primarily by Na⁺, I_CRAN produces membrane depolarization and indirectly, an elevation in [Ca²⁺]_i, as a consequence of activation of voltage-dependent I_Ca (21).

Although membrane potential and [Ca²⁺]_i oscillations have been readily recorded from whole mouse islets and -cell clusters (11-14, 20), single mouse -cells normally do not respond to glucose stimulation with regular oscillatory activity (16). We have previously reported that a stable transgenically derived murine insulinoma cell line (TC3-neo) also responds to glucose with large amplitude oscillations in [Ca²⁺]_i when in the presence of 10-20 mM tetraethylammonium (TEA), a blocker of delayed rectifier K⁺ channels (Kv) (20). We have thus utilized this cell line to characterize and identify the membrane conductance that underlies glucose-stimulated oscillatory activity.

In this article, we provide evidence that activation of I_CRAN in TC3-neo cells regulates glucose-stimulated [Ca²⁺]_i oscillations and insulin secretion. Furthermore, to characterize the molecular identity of I_CRAN, we detected in insulin-secreting insulinoma cells and mouse islets the expression of multiple trp (transient receptor potential) genes that are known to encode intracellular Ca²⁺ store release-activated channels in other mammalian and insect tissues (22-28). Our findings suggest that the interaction between intracellular Ca²⁺ stores and plasma membrane conductances is an important mechanism that controls glucose-dependent stimulus response coupling in -cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

TC3 cells were cultured as described elsewhere (20, 29). A clonal subline (TC3-neo) was isolated after transfection with pSV2-neo by electroporation (20) and maintained in the continued presence of 1 mg/ml G418 (Life Technologies, Inc.). Cells were seeded for 4-6 days in Dulbecco's modified Eagle's medium supplemented with 15% normal horse serum, 2.5% fetal calf serum, 50 units/ml penicillin, 50 µg/ml streptomycin. Eighteen hours prior to experiments, medium was changed to RPMI 1640 medium containing 15% dialyzed horse serum, 2.5% dialyzed fetal calf serum, 1 mM glucose, and antibiotics as above. MIN6 cells were cultured essentially as described (30). Islets of Langerhans were isolated from 8-10-week-old C57BL/KsJ (+/+) mice (Jackson Laboratories) as described (5, 9, 12).

Measurement of [Ca²⁺]_i-- TC3-neo cells were loaded with Fura-2 by a 25-min incubation at 37 °C in Krebs-Ringer buffer containing (in mM): 119 NaCl, 4.7 KCl, 1.8 CaCl₂, 1.2 MgSO₄, 1.2 KH₂PO₄, 25 NaHCO₃, 0 glucose, supplemented with 5 µM acetoxymethyl ester of Fura-2 (Molecular Probes Inc.). In some experiments, external NaCl was replaced with equimolar amounts of N-methyl-D-glucamine-Cl. [Ca²⁺]_i was expressed as the ratio change in fluorescence (340/380) as detailed in full elsewhere (9, 20).

Patch Clamp Current Recordings-- Single TC3-neo cells were voltage-clamped using perforated patch clamp techniques. Patch electrodes contained 80 mM potassium aspartate, 50 mM KCl, 5 mM NaCl, 5 mM MgCl₂, 10 mM HEPES/KOH (pH 7.2), and 100-120 µg/ml nystatin (9, 20). The perfusion medium was a Krebs-Ringer buffer solution as detailed above.

Insulin Secretion Measurements-- TC3-neo cells were plated at a density of 25 × 10⁴/cm² and cultured overnight in RPMI 1640 medium as described above. Insulin secretion measurements were made in the presence of 0 or 1 mM glucose in the presence or absence of TEA, maitotoxin, or thapsigargin, as described previously (20). The concentration of insulin was determined using an SPA assay kit (Amersham Pharmacia Biotech) and calibrated using rat insulin (Novo) as standard.

Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR) Identification and Amplification of Probes for trp-related Gene Expression-- Mouse brain, TC3-neo, and MIN6 poly(A)⁺ RNA were prepared using an oligo(dT) binding method (Quickprep micro mRNA kit, Amersham Pharmacia Biotech) and first strand cDNA was reverse-transcribed using random primers (Superscript II transcriptase, Life Technologies, Inc.). Transcripts of the six known members of the murine trp (mtrp) family (accession number in parentheses) were amplified by polymerase chain reaction (PCR) using the following oligonucleotide primers: mtrp-1 (GenBank>U40980) 5'-GATTTTGGGAAATTTCTGGGAATG-3' (sense) and 5'-TTTATCCTCATGACTTGCTATCA-3' (antisense); mtrp-2 (U40981) 5'-GACATGATCCGGTTCATGTTC-3' (sense) and 5'-CATCAGCATCATCCTCGATCT-3' (antisense); mtrp-3 (U40982) 5'-GACATATTCAAGTTCATGCGTTCTC-3' (sense) and 5'-ACATCACTGTCATCCTCGATCTC-3' (antisense); mtrp-4 (X90697) 5'-CTGCAGATATCTCTGGGAAGG-3' (sense) and 5'-GCTTTGTTCGAGCAAATTTCC-3' (antisense); mtrp-5 (U40984) 5'-TCTACTGCCTAGTACTACTGGC-3' (sense) and 5'-GTAGGAGTTATTCATCATGGCG-3' (antisense); mtrp-6 (U40969) 5'-GATATCTTCAAATTCATGGTCATA-3' (sense) and 5'-GTCCGCATCATCCTCAATTTC-3' (antisense). The PCR protocol used consisted of an initial denaturation at 94 °C for 3 min; followed by five cycles of 94 °C for 1 min, 64 °C for 30 s; followed directly by 25 cycles of 94 °C for 1 min, 60 °C for 1 min, and 72 °C for 30 s.

Northern Blot Analysis-- For RNA isolation, TC3 cells were grown to near confluence, washed three times with phosphate-buffered saline, and lysed in solution containing 4 M guanidine thiocyanate, 100 mM Tris-HCl (pH 7.5), 1% -mercaptoethanol, and 0.5% lauryl sarcosinate. Total RNA was purified by centrifugation through a 0.9-ml cushion of 5.7 M CsCl in a TLS-55 rotor at 53000 rpm for 4 h in the Optima TL ultracentrifuge (Beckman, Palo Alto, CA). Mouse brain RNA was isolated using TRIzol (Life Technologies, Inc.) according to the manufacturer's protocol and purified by centrifugation through CsCl as described above. RNA was size-fractionated in the presence of ethidium bromide (50 µg/ml) in a 1% agarose-formaldehyde gel at 100 V and transferred to Hybond N nylon membranes (Amersham Pharmacia Biotech). The probes used were the subcloned PCR products described in the previous section, except in the case of mtrp-4, for which a full-length mtrp-4 clone derived from a TC3 cDNA library was employed.² Probes were gel-purified and labeled with [-³²P]dCTP using Prime-It random primer labeling kit (Stratagene, La Jolla, CA). After overnight hybridization at 68 °C in ExpressHybe solution (CLONTECH) blots were washed to a final stringency of 0.1× SSC, 0.1% SDS at 60 °C and exposed to BioMax MS-1 film with BioMax intensifying screen (Eastman Kodak Corp.) for 18 h to 1 week at 80 °C.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

In contrast to normal mouse islets, the murine insulinoma cell line, TC3-neo, does not usually respond to a step increase in glucose concentration with regular oscillatory increases in [Ca²⁺]_i. Instead, a slow rise in [Ca²⁺]_i occurs with occasional intermittent spikes (Fig. 1A). Exposure of TC3-neo cells to 20 mM TEA, a blocker of delayed rectifier K⁺ channels, however, permitted the generation by glucose of large regular oscillations in [Ca²⁺]_i of a type normally seen in mouse islets exposed to nutrients (5, 13, 20) (Fig. 1A). In the absence of glucose, TEA was without effect (n = 5). This glucose dependence suggests that the governing ionic conductance is carried by I_KATP as in the case of primary -cells, i.e. I_KATP sets the resting membrane potential (17, 18). Indeed, activation of I_KATP with 250 µM diazoxide fully suppressed the TEA-activated [Ca²⁺]_i oscillations (Fig. 1B). The [Ca²⁺]_i oscillations were also immediately abolished by exposure to either low Ca²⁺-containing external solutions (50 µM EGTA, no added Ca²⁺) or 1 µM nitrendipine, indicating their dependence on depolarization-triggered Ca²⁺ influx through voltage-dependent Ca²⁺ channels (VDCC) (Fig. 1C).

View larger version (33K): [in this window] [in a new window]   Fig. 1.   TC3-neo cells display voltage-dependent, TEA-activated [Ca2+]i oscillations. A, changes in [Ca2+]i in a single TC3-neo cell, expressed as ratiometric changes in Fura-2 fluorescence (F340/F380), following exposure to a maximally stimulating concentration of glucose (1 mM; GLU, open bar). The subsequent addition of tetraethylammonium (20 mM; TEA, filled bar) activated large amplitude oscillations in [Ca2+]i. B, induction of membrane hyperpolarization with diazoxide (250 µM; DZ, hatched bar), an activator of ATP-sensitive K+ channels abolished the oscillations in [Ca2+]i induced by glucose and TEA. C, the [Ca2+]i oscillations induced by glucose and TEA were abolished by maneuvers designed to suppress Ca2+ influx. Both reduction in the extracellular Ca2+ concentration using solutions containing no added Ca2+ and 50 µM EGTA (EGTA, open bar) and blockade of L-type Ca2+ channels using nitrendipine (1 µM; NIT, open bar) reversibly suppressed oscillatory activity.

It has been previously suggested that glucose-induced electrical and [Ca²⁺]_i oscillations result from oscillations in metabolism (15, 31, 32). However, the experiments shown in Fig. 2 (A and B) would indicate that direct modulation of plasma membrane conductances alone is insufficient to induce Ca²⁺ oscillations. In the absence of glucose, exposure of TC3-neo cells to 100 µM tolbutamide to block I_KATP induced [Ca²⁺]_i oscillations only in the presence of TEA that were indistinguishable from those generated by glucose (Fig. 2A). Furthermore, membrane depolarization induced by 20 mM KCl was able to similarly substitute for glucose in permitting TEA to induce [Ca²⁺]_i oscillations (Fig. 2B). Thus, a necessity for the conductance regulating the [Ca²⁺]_i oscillations to be coupled to oscillations in KATP activity or glucose metabolism seems excluded.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   TEA-induced [Ca2+]i oscillations in TC3-neo cells can be supported by depolarizing secretagogues in addition to glucose. Pre-exposure to 100 µM tolbutamide (A) or 20 mM KCl (B) induces monophasic rises in [Ca2+]i similar to those of glucose. Addition of 20 mM TEA in both cases activated large amplitude [Ca2+]i oscillations (compare with Fig. 1, A-C).

We were, however, able to measure marked alterations in the [Ca²⁺]_i oscillations stimulated by glucose following exposure to thapsigargin, a selective inhibitor of the ER Ca²⁺ ATPases (33, 34). Although thapsigargin caused no detectable alteration in [Ca²⁺]_i in the absence of glucose, addition of thapsigargin following exposure to glucose and TEA converted the resulting [Ca²⁺]_i oscillations into a sustained rise in [Ca²⁺]_i (Fig. 3A). Previous studies in mouse pancreatic islets and -cells have indicated that the effect of maneuvers to deplete intracellular Ca²⁺ stores, as exemplified by the action of thapsigargin, result from the activation of I_CRAN (12, 21, 34).

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Stimulation of [Ca2+]i oscillations in TC3-neo cells by ER Ca2+ store depletion-activated nonselective cation channels. A, depletion of intracellular Ca2+ stores by exposure to the ER Ca2+-ATPase inhibitor thapsigargin (500 nM; TG, lower bar) converts [Ca2+]i oscillatory activity induced by glucose (1 mM; GLU, open bar) and TEA (20 mM; middle bar) into a sustained rise. In mouse islets, similar maneuvers cause activation of a depolarizing nonselective cation current carried principally by Na+ (21). B, reduction of extracellular Na+ and its equimolar replacement by 119 mM amount of the non-permeating cation N-methyl-D-glucamine (NMDG, lower bar) reversibly ablated the [Ca2+]i oscillations induced by glucose (1 mM; open bar) and TEA (20 mM; middle bar). C, exposure to the Ca2+ store-activated channel blocker SKF 96365 (60 µM; SKF, hatched bar) also ablated the [Ca2+]i oscillations triggered by glucose (1 mM; open bar) and TEA (20 mM; hatched bar). D, the suppressive effect of SKF 96365 on oscillatory activity is not related to block of classical Ca2+ entry pathways. The rise induced by exposure to KCl (20 mM; KCl, open bar) was unaffected by SKF 96365 (60 µM; SKF, filled bar), but completely suppressed by the L-type Ca2+ channel blocker nitrendipine (1 µM; NIT, hatched bar).

The principal permeating ion carrying the depolarizing current through I_CRAN in mouse -cells is Na⁺ (21). To test whether I_CRAN was involved in regulating the glucose-dependent oscillations in TC3-neo cells, we therefore examined the effects of Na⁺ substitution. Reduction of external Na⁺ by substitution with the non-permeant cation N-methyl-D-glucamine largely suppressed the [Ca²⁺]_i oscillations induced in TC3-neo cells by glucose and TEA (Fig. 3B). Similarly, application of SKF 96365, a putative blocker of intracellular Ca²⁺ store release-activated channels (35) reversibly abolished the glucose-stimulated oscillations in [Ca²⁺]_i (Fig. 3C). Nonspecific suppressive effects of SKF 96365 on [Ca²⁺]_i were ruled out by the observation that KCl-induced elevations in [Ca²⁺]_i were unimpaired (Fig. 3D).

Application of 100 pM MTX, an agent that directly activates nonselective cation channels in HL60 cells (36) and I_CRAN in mouse -cells (21), caused an immediate sustained rise in [Ca²⁺]_i, further indicating the prominent role of this channel in regulating [Ca²⁺]_i in TC3 cells (Fig. 4A). Similar to its effect on glucose-stimulated [Ca²⁺]_i oscillations, application of SKF 96365 also suppressed the MTX-dependent elevation in [Ca²⁺]_i in TC3-neo cells (Fig. 4B). Finally, diazoxide application was able to completely reverse the MTX-induced [Ca²⁺]_i rise (Fig. 4C), indicating that I_CRAN was a relatively small current that could be overridden by concomitant activation of I_KATP.

View larger version (11K): [in this window] [in a new window]   Fig. 4.   Maitotoxin causes a depolarization-dependent increase in [Ca2+]i in TC3-neo cells related to opening of ER Ca2+ store depletion-activated nonselective cation channels. A, exposure to 50 pM of the dinoflagellate toxin MTX caused a sustained elevation in [Ca2+]i in TC3-neo cells. In mouse pancreatic -cells, a similar rise in [Ca2+]i was attributed to activation of nonselective cation channels, leading to a depolarization-driven increase in Ca2+ entry through L-type Ca2+ channels (21). B, the maitotoxin-induced rise in [Ca2+]i (50 pM; MTX, open bar) was reversibly attenuated by the Ca2+ store-operated ion channel blocker SKF 96365 (60 µM; SKF, filled bar). C, the maitotoxin-induced elevation in [Ca2+]i could also be reversibly suppressed by application of the ATP-sensitive K+ channel activator diazoxide (250 µM; DZ, filled bar), indicating that the rise in [Ca2+]i was likely secondary to depolarization-dependent activation of L-type Ca2+ channels.

To confirm the actual presence of a nonselective cation channel in TC3-neo cells, we carried out perforated patch recordings. Application of 1-100 pM MTX caused the activation of a non-inactivating current with a reversal potential of 10 mV (Fig. 5A), properties identical to I_CRAN observed in mouse -cells (21); furthermore, both SKF 96365 and external Na⁺ removal were able to reversibly attenuate the MTX-activated current (Fig. 5, A and B). These findings suggest that the increase in [Ca²⁺]_i induced by MTX, and the ablation of the glucose-stimulated [Ca²⁺]_i oscillations by SKF 96365 or external Na⁺ removal, were secondary to activation and block of a Ca²⁺ store-operated, nonselective cation channel, respectively.

View larger version (11K): [in this window] [in a new window]   Fig. 5.   TC3-neo cells possess maitotoxin-sensitive nonselective cation channels A. Perforated patch recordings of membrane current from a single TC3-neo cell. Using a 100-ms ramp protocol (100 mV to 60 mV, shown in inset below panel), traces shown are control (), and following exposure to 50 pM MTX (), which caused an increase in membrane current recorded over a range of potentials. The reversal potential for this current was 12 mV, as predicted for a nonselective cation channel and similar to a store operated nonselective cation channel reported in mouse pancreatic -cells (21). Application of 40 µM SKF 96365 () caused a marked suppression of the MTX-activated current, which was reversed on washout of the store-operated channel blocker (). B, in the presence of 50 pM MTX (), reduction of extracellular Na+ and its equimolar replacement by 119 mM of the non-permeating cation N-methyl-D-glucamine () caused a substantial diminution of the MTX-activated current, which was reversed on re-introduction of normal Na+-containing solutions and MTX ().

We also measured the effects of modulation of I_CRAN on insulin secretion. As we described previously (20), in the presence of 20 mM TEA, 1 mM glucose gave a robust stimulation of insulin secretion (Fig. 6). We investigated the effect on insulin secretion of two mechanisms that would lead to the activation of I_CRAN. Activation of the store-operated channel by inhibition of the ER Ca²⁺ ATPase with thapsigargin greatly potentiated glucose-stimulated insulin secretion in the presence of TEA (Fig. 6). Likewise, direct activation of I_CRAN with 50 pM MTX also markedly increased glucose-stimulated insulin secretion (Fig. 6). Thus, indirect or direct activation of I_CRAN leads to an amplification of glucose-stimulated insulin secretion, presumably by enhancing depolarization-driven influx of Ca²⁺ (21).

View larger version (9K): [in this window] [in a new window]   Fig. 6.   Activation of nonselective cation channels modulates glucose-stimulated insulin secretion. Whereas stimulatory concentrations of glucose caused only a minor stimulation of insulin secretion (1 mM glucose, filled bar) compared with basal insulin secretion output (0 mM glucose, open bar), TEA (20 mM) potentiated glucose-stimulated insulin secretion 10-fold. Addition of thapsigargin (1 µM; TG) further augmented the insulin secretion. Exposure to MTX (50 pM)in the presence of 1 mM glucose induced a similar amount of insulin secretion as that produced by the combined addition of TEA and thapsigargin.

To characterize the molecular identity of I_CRAN, we used RT-PCR and Southern blots of TC3-neo and MIN6 insulinoma cells and mouse islets to identify transcript expression and prepare probes for Northern blot analysis. Primers (see "Materials and Methods") were designed to amplify the six known mouse trp genes (26, 27). Products from all six trp genes were detected in TC3-neo cells and mouse brain. In MIN-6 cells, trp-1, trp-2, trp-4, and trp-6 were expressed, whereas trp-3 and trp-5 were not reproducibly observed. In mouse islets all of the trp gene products were evident except trp-5. RT-PCR experiments with mock cDNA synthesis in the absence of reverse transcriptase were done in parallel, and no bands were observed, excluding amplification of genomic DNA (data not shown). All six trp PCR products from the brain and TC3 amplifications were cloned and sequenced, revealing identity to those reported previously (26, 27). We then characterized the expression of trp genes using Northern blots comparing total RNA prepared from mouse brain and TC3 cells (Fig. 7). This showed that, despite amplification of most of the trp genes by RT-PCR, only mtrp-4 message was readily detectable in TC3 cells, while five of six trp transcripts (the exception being trp-2, a putative pseudogene; Ref. 24) were visualized in brain RNA. Prolonged exposures (>1 week) of several blots revealed very faint bands in TC3 RNA corresponding to trp-1 and trp-3 (data not shown).

View larger version (40K): [in this window] [in a new window]   Fig. 7.   Analysis of mammalian trp expression in insulin-secreting cells. Northern blots comparing trp expression in RNA isolated from mouse brain (lane 1) and TC3-neo cells (lane 2). Panels 1-6 indicate the trp gene probe employed (i.e. trp1-6). Note that the trp-2 probe (panel 2, arrowhead) gave a faint band at about 7.5 kilobase pairs in mouse brain. The location of the trp-6 transcript is indicated (panel 6, arrowhead). trp-4 expression was nearly equivalent in the mouse brain and TC3-neo RNA (panel 4). Lane 3 in panel 4 contains purified brain ribosomal RNA as a control. Each blot shown is representative of at least four experiments.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We have employed a -cell model, the TC3-neo cell, to study the underlying mechanisms that regulate glucose-induced oscillations of [Ca²⁺]_i and insulin secretion. Previous studies in insulinoma cells and islets have demonstrated a close coupling between glucose-stimulated oscillations in membrane potential and the consequent oscillations in [Ca²⁺]_i and insulin secretion (14, 15, 20). However, precise identification of the signaling events underlying oscillations has remained elusive. In this study, we have demonstrated the importance of a nonselective cation current activated following depletion of intracellular Ca²⁺ stores in the regulation of glucose-dependent oscillations of [Ca²⁺]_i and show that activation of I_CRAN potentiates insulin secretion. Our findings suggest that this ionic conductance generated by the discharge and refilling of intracellular calcium stores plays a critical role in regulating glucose-induced oscillatory signaling in -cells.

Similar to the responses elicited in mouse primary -cell clusters and islets, glucose stimulation of insulin-secreting TC3-neo cells results in synchronous oscillations in membrane potential and [Ca²⁺]_i (20). However, activation of the oscillations required the additional presence of TEA. In this respect, the TC3 cells more closely resemble rat islets, which also do not respond to glucose with large amplitude oscillations in [Ca²⁺]_i, but can be induced to do so by TEA (20). This probably reflects subtle differences between the expression levels of K⁺ channels between -cells of different species. Recently, we determined that the glucose-dependent effects of TEA on TC3-neo cells stemmed from inhibition of delayed rectifier K⁺ channels (K_DR) (20). TC3-neo cells have a similar complement of K_DR subtypes to normal rodent -cells, and express Shab (Kv2.1)- and Shaw (Kv3.2)-related transcripts (18, 20). Under conditions where I_KDR is blocked, TC3-neo cells display glucose-dependent oscillatory responses of the types seen in intact mouse islets or -cell clusters. This suggests that the TC3-neo cell line will prove useful as a relevant model to examine the ion conductances underlying glucose-stimulated oscillatory behavior in islets.

Several models have been advanced to explain islet bursting activity induced by glucose. One of the earliest proposed that Ca²⁺ influx during the active phase caused a slow rise in [Ca²⁺]_i, which activated I_KCa that in turn repolarized the membrane potential and inactivated I_Ca (37). Our data are not consistent with this model, since oscillations were observed in the presence of 10-20 mM TEA (20), conditions under which I_KCa should be completely suppressed by TEA (K_d, 140 µM) (17). A more recent model suggests that cyclic inhibition of I_KATP, or a direct coupling between cellular fuel metabolism and electrical activity in -cells acts as the -cell membrane potential oscillator (15, 32). Our findings that in the presence of complete block of I_KATP with tolbutamide, [Ca²⁺]_i oscillations identical in nature to those induced by glucose are produced would rule out a requirement for oscillations in I_KATP, and by inference metabolism, being responsible. Although cyclic variations in KATP activity or glucose metabolism were not a prerequisite for the induction and maintenance of [Ca²⁺]_i oscillations, glucose was nonetheless an important activator of oscillatory behavior. This permissive effect of glucose and tolbutamide is most likely related to blockade of the repolarizing influence of I_KATP, the predominant ionic current in pancreatic -cells. In support of this contention, direct overriding of I_KATP using depolarizing concentrations of KCl was similarly able to substitute for both glucose or tolbutamide in permitting the induction of [Ca²⁺]_i oscillations. As will be discussed further, the inhibition of I_KATP allows smaller conductance(s) to exert significant effects on the membrane potential and, indirectly, on [Ca²⁺]_i.

In addition to stimulating the influx of Ca²⁺ as a result of depolarization-induced opening of VDCCs, it has been recently suggested that glucose mobilizes intracellular Ca²⁺ from the ER (5-7). As a corollary of this triggered release, glucose has been demonstrated by a number of groups to stimulate the sequestration of Ca²⁺ in the ER (34, 38, 39). The state of filling of the ER with Ca²⁺ in turn seems to regulate a plasma membrane nonselective cation channel in pancreatic -cells, akin to the "capacitative" Ca²⁺ entry system described in non-excitable cells (40). This pathway permits the refilling of depleted intracellular Ca²⁺ stores by activation of a voltage-independent Ca²⁺ current called I_CRAC (Ca²⁺ release-activated Ca²⁺ current), that directly supplies Ca²⁺ to the depleted organelle (41). By contrast, in -cells I_CRAN enhances the entry of Ca²⁺ through I_Ca by causing a sustained membrane depolarization (21). We have confirmed the presence of I_CRAN in TC3 cells using patch clamp techniques, whose properties are indistinguishable from the current previously described in primary mouse -cells (21). Thus, the current is activated by exposure to MTX and the principal permeating ion seems to be Na⁺; the reversal potential of around 10 mV confirms the nonselective nature of the channel.

Since in other cell systems the mobilization of intracellular Ca²⁺ stores is often oscillatory in nature (42), this suggests that a conductance activated by the state of filling of intracellular Ca²⁺ stores (e.g. I_CRAN) would similarly oscillate and could therefore underlie the membrane potential oscillations seen in islets and -cells exposed to glucose. It was therefore of great interest to observe that pharmacological manipulation of I_CRAN resulted in alterations in [Ca²⁺]_i oscillations in TC3-neo cells that were consistent with such a role for the channel. Thus, depletion of intracellular Ca²⁺ stores with thapsigargin to induce activation of I_CRAN resulted in a conversion of oscillatory alterations in [Ca²⁺]_i to a sustained rise. That this was due to continuous entry of Ca²⁺ through I_Ca, activated as a consequence of I_CRAN-dependent membrane depolarization, was demonstrated by the ablative effects of the L-type Ca²⁺ channel inhibitor nitrendipine.³ Consistent with its effects on glucose-stimulated alterations in [Ca²⁺]_i, thapsigargin potentiated the insulin secretion response of TC3-neo cells elicited by glucose and TEA. Direct activation of I_CRAN following exposure to MTX induced a similar, nitrendipine-sensitive sustained elevation in [Ca²⁺]_i. MTX also stimulated insulin secretion in the presence of glucose alone, to the same extent as that produced by the combined administration of glucose, TEA and thapsigargin. Interestingly, in the absence of glucose, MTX was without effect on insulin secretion,³ consistent with the proposal that the MTX-activated current is of small magnitude and unable to override I_KATP, which would be activated under these conditions. Suppression of current flow through I_CRAN had the opposite effect on the glucose-stimulated [Ca²⁺]_i oscillations in TC3-neo cells. Thus, SKF 96365, a blocker of intracellular Ca²⁺ store release-activated channels (35) completely suppressed glucose-induced [Ca²⁺]_i oscillations. Although this compound has also been shown to block L-type calcium channels at high concentrations, the demonstration that KCl-induced depolarization raised [Ca²⁺]_i in the presence of SKF 96365 suggests that voltage-dependent calcium channels were not inhibited. Furthermore, reduction of extracellular Na⁺, the principal permeating ion through I_CRAN, similarly ablated the glucose stimulated oscillations.

We initially characterized the putative molecular identity of I_CRAN using RT-PCR with Southern blot analysis, and identified trp genes (trp1-6) expressed differentially in two mouse insulinoma cells and islets of Langerhans. These findings are in partial agreement with recent studies in which the expression of mouse trp-1 was reported in MIN6 cells (43). However, upon Northern blot analysis, trp-4 was the only detectable trp transcript expressed in TC3 insulinoma cells, whereas five of the six genes were readily identified in mouse brain (Fig. 7).

Drosophila trp and trp-like (trpl) genes encode Ca²⁺ store-operated Ca²⁺ channels and inositol 1,4,5-trisphosphate-activated nonselective cation channels, respectively (22, 23). Recent evidence has suggested that mammalian trp family channels may have several activation mechanisms, with functional channels encoded by multiple subunits (44). Mammalian trp genes are unlikely to direct the expression of I_CRAC, as none of the expressed trp genes to date are able to reproduce the Ca²⁺ selectivity or single channel conductance (0.03 picosiemens) of the native channel (44). For example, htrp-1, also termed hTrpC1, encodes a nonselective channel activated by depletion of Ca²⁺ stores with a single channel conductance of 16 picosiemens (45). Both TrpC1 and TrpC3 can increase calcium entry following depletion of intracellular stores (27). When transfected into murine L(tk) cells, a mixture of plasmids containing antisense cDNAs for six mouse trp genes eliminated the thapsigargin-induced increase in [Ca²⁺]_i (27), and a similar transfection with a 1.2-kilobase pair antisense fragment to mtrp4 alone also eliminated receptor-stimulated calcium entry (46). mtrp6 was likewise found to increase Ca²⁺ entry when stimulated by activation of a cotransfected muscarinic receptor, an effect that was blocked by SKF 96365, an agent we employed here to also block store-operated currents (47). The similarity of I_CRAN to trp family channels is further supported by the finding of a single channel conductance of 25 picosiemens for the MTX-sensitive nonselective cation channel activated by ER Ca²⁺ store depletion in fibroblasts, within the range reported for htrp-1 (48). Our findings thus support the likelihood that trp-related genes encode not only the current we have termed I_CRAN, but also the Na⁺-dependent inward currents that cause rises in [Ca²⁺]_i in -cells following Ca²⁺ store depletion with carbachol, acetylcholine, pituitary adenylate cyclase activating polypeptide, and glucagon-like peptide-1 (49-52).

Fig. 8 depicts a model in which I_CRAN regulates glucose-stimulated oscillations of membrane potential and [Ca²⁺]_i. This model provides an explanation for the oscillatory behavior of TC3-neo cells in the presence of glucose, and may also be applicable to islet responses following nutrient stimulation. An experimental model centered around I_CRAN modulation has also recently been advocated to explain the effects of carbachol on [Ca²⁺]_i and membrane potential (49). In conclusion, our data suggest that I_CRAN modulation plays a critical role in controlling nutrient-dependent electrical bursting behavior in islets, and also demonstrates that alterations in I_CRAN activity modulate insulin secretion. Since defective ER Ca²⁺ sequestration has been demonstrated in several animal models of non-insulin-dependent diabetes mellitus (53, 54), it is possible that alterations in I_CRAN regulation may play a role in aberrant insulin secretion responses in some forms of non-insulin-dependent diabetes mellitus.

View larger version (24K): [in this window] [in a new window]   Fig. 8.   Proposed model for the role of an ER Ca2+ store depletion-activated nonselective cation channel in regulating [Ca2+]i oscillations in membrane potential and [Ca2+]i in insulin-secreting cells. In the upper panel is shown experimentally observed alterations in [Ca2+]i in a single TC3-neo cell following a step increase in glucose from 0 to 1 mM (filled bar), and the subsequent effect of exposure to 20 mM TEA (open bar). The vertical calibration marks represent the F340/F380 fluorescence ratio for the Fura-2 [Ca2+]i estimations. The lower panels represent a schematic -cell, comprising (from left to right) an extracellular space, the plasma membrane, an intracellular space, the endoplasmic reticulum membrane, and an intracellular Ca2+ storage compartment. Shown are the predicted changes in plasma membrane ion channel and ER Ca2+ pump activity occurring during the corresponding segments I-IV. The size of the blue shaded area in the storage compartment represents the content of the store. The solid circles in the plasma membrane and ER represent Ca2+ the glucose transporter (GLUT2) and the thapsigargin-sensitive ER Ca2+ ATPase, respectively. The yellow boxes represent membrane ion channels. Going counterclockwise from GLUT2, they are: the ATP-sensitive K+ channel (KATP), the delayed rectifier K+ channel (Kv), the voltage-dependent Ca2+ channel (VDCC), and the calcium store release-activated nonselective cation channel (CRNSC). Their respective permeating cations, K+, Ca2+, and Na+ are shown in the vicinity of the channel entrance vestibule. The large arrows originating in the intracellular Ca2+ storage compartment represent Ca2+ release and the putative retrograde signal emanating from the ER and regulating CRNSC (solid and unfilled, respectively). The influx of Ca2+ through VDCC is determined by the balance between the magnitude (represented by the size of arrow) of outward repolarizing current through KATP and Kv, versus the inward depolarizing current through CRNSC. Thus, Ca2+ influx occurs only when the current flow through CRNSC exceeds that through KATP and Kv. In the absence of glucose, outward K+ current flow through KATP more than offsets the inward current flowing through CRNSC open as a consequence of the partially depleted intracellular Ca2+ stores (panel I). Following exposure to stimulatory concentrations of glucose, cellular ATP begins to rise. This increase initially provides energy substrate to activate Ca2+ pumps located in the ER, thereby loading intracellular Ca2+ stores. As ATP continues to rise, the current flow through KATP diminishes, allowing some membrane depolarization, driven by Na+ entry through CRNSC (panel II), resulting in a partial activation of VDCC and a modest increase in [Ca2+]i. This depolarization, however, also activates the voltage-dependent Kv channels, and the outward flow of K+ acts to prevent further membrane depolarization. Block of Kv channels (with TEA) (panel III) reduces this repolarizing influence. The resultant unfettered depolarization, either directly or indirectly through inositol 1,4,5-trisphosphate (2), eventually triggers release of intracellular Ca2+ stores. The now empty intracellular Ca2+ store induces full activation of CRNSC the consequence of which is an accelerated rate of depolarization, which recruits more VDCCs leading to an augmented influx of Ca2+. Ca2+ influx continues until the ER Ca2+ stores are filled, at which time the store depletion signal switches off and CRNSCs deactivate. Due to residual outflow of K+ through partially inhibited KATP and the closing of CRNSC the membrane hyperpolarizes, thereby deactivating VDCC (panel IV). The assumption as to the hyperpolarizing action of residual KATP is based on the experimental observation that increasing concentrations of tolbutamide or combination of tolbutamide and glucose cause conversion of [Ca2+]i oscillations into a sustained rise (see Footnote 3). A subsequent release of Ca2+ initiates a cycle of ER Ca2+ store release and refilling, which results in cyclical activation and inactivation of CRNSC and thereby indirectly regulates the degree of calcium influx through VDCC (panels III and IV).