Nicotinic acid-adenine dinucleotide phosphate-sensitive calcium stores initiate insulin signaling in human beta cells

Abstract

Recent studies suggest a role for autocrine insulin signaling in beta cells, but the mechanism and function of insulin-stimulated Ca²⁺ signals is uncharacterized. We examined Ca²⁺-dependent insulin signaling in human beta cells. Two hundred nanomolar insulin elevated [Ca²⁺]_c to 284 ± 27 nM above baseline in ≈30% of Fura-4F-loaded cells. Insulin evoked multiple Ca²⁺ signal waveforms, 60% of which included oscillations. Although the amplitude of Ca²⁺ signals was dose-dependent between 0.002 and 2,000 nM, the percentage of cells responding was highest at 0.2 nM insulin, suggesting the interaction of stimulatory and inhibitory pathways. Ca²⁺-free solutions did not affect the initiation of insulin-stimulated Ca²⁺ signals, but abolished the second phase of plateaus/oscillations. Likewise, inositol 1,4,5-trisphosphate (IP₃) receptor antagonists xestospongin C and caffeine selectively blocked the second phase, but not the initiation of insulin signaling. Thapsigargin and 2,5-di-tert-butylhydroquinone (BHQ) blocked insulin signaling, implicating sarcoplasmic/endoplasmic Ca²⁺-ATPase (SERCA)-containing Ca²⁺ stores. Insulin-stimulated Ca²⁺ signals were insensitive to ryanodine. Injection of the CD38-derived Ca²⁺ mobilizing metabolite, nicotinic acid-adenine dinucleotide phosphate (NAADP), at nanomolar concentrations, evoked oscillatory Ca²⁺ signals that could be initiated in the presence of ryanodine, xestospongin C, and Ca²⁺-free solutions. Desensitizing concentrations of NAADP abolished insulin-stimulated Ca²⁺ signals. Insulin-stimulated Ca²⁺ signals led to a Ca²⁺-dependent increase in cellular insulin contents, but not secretion. These data reveal the complexity of insulin signal transduction and function in human beta cells and demonstrate functional NAADP-sensitive Ca²⁺ stores in a human primary cultured cell type.

Discovery of insulin receptors on pancreatic beta cells and the characterization of beta cell-specific insulin receptor null mice with a diabetes-like phenotype suggest a physiological role for autocrine insulin signaling (1). Although results from in vivo and whole islet experiments suggested a negligible or inhibitory effect of insulin on insulin release, recent experiments with dispersed rodent beta cells suggested a stimulatory effect on calcium signaling, insulin expression, and exocytosis (2–7). Unlike glucose signaling, insulin-evoked Ca²⁺ signals in mouse beta cells required intracellular Ca²⁺ stores sensitive to the sarcoplasmic/endoplasmic Ca²⁺-ATPase (SERCA) inhibitor thapsigargin (4). Insulin action was not blocked by a phospholipase C inhibitor, suggesting indirectly that inositol 1,4,5-trisphosphate (IP₃)-sensitive Ca²⁺ stores were not involved (5). The mechanisms of autocrine insulin feedback are unknown in human beta cells.

Ca²⁺ signals control multiple functions in secretory cells, and at least three distinct biochemical classes of Ca²⁺ stores coexist (8, 9). Aside from the phospholipase C/IP₃ pathway that is commonly activated by G-protein-coupled receptors, Ca²⁺ can be mobilized through ryanodine receptors, activated by Ca²⁺ or cyclic ADP-ribose (cADPr). A third class of Ca²⁺ store, mobilized by nicotinic acid adenine dinucleotide phosphate (NAADP), functions in oocytes, Jurkat T lymphocytes, and mouse pancreatic acini (8, 10, 11). The production of NAADP and cADPr are catalyzed by CD38 and related ADP-ribosyl cyclases (12, 13). CD38 is found in many cell types, including human beta cells. Glucose-stimulated Ca²⁺ mobilization and insulin release (in vivo and in vitro) were enhanced by CD38 overexpression and reduced by CD38 knockout (14, 15). Beta cells with poor glucose-stimulated insulin production/release (ob/ob, GK, RINm5F) have less CD38 (16, 17). CD38 autoantibodies, found in ≈14% of diabetic patients, evoked Ca²⁺ signals and insulin release from human islet cells (18).

In this study, we tested the hypothesis that NAADP-sensitive Ca²⁺ stores regulate beta cell function, in the context of autocrine insulin signaling. We report that insulin generates complex Ca²⁺ signals by mobilizing novel intracellular Ca²⁺ stores in primary cultures of dispersed human islet cells. NAADP-sensitive Ca²⁺ stores initiate insulin signaling, whereas subsequent oscillatory behavior is sensitive to the removal of extracellular Ca²⁺ and to putative IP₃ receptor inhibitors. These prolonged insulin-stimulated Ca²⁺ signals regulate cellular insulin levels, but do not measurably stimulate overall secretion.

Materials and Methods

Drugs and Solutions.

Reagents were from Sigma unless otherwise stated. Stocks (<1,000-fold) of BHQ (2,5-di-tert-butylhydroquinone), CPA (cyclopiazonic acid), xestospongin C, BAPTA-AM (1,2-bis-(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester; Calbiochem), and ryanodine were dissolved in DMSO. Thapsigargin stock was dissolved in EtOH. Recombinant insulin, caffeine, IP₃, NADP, and NAADP were dissolved directly into solutions. Ca²⁺ imaging was performed in Ringer's solution containing (in mM): 5.5 KCl, 2 CaCl₂, 1 MgCl₂, 20 Hepes, 141 NaCl, and 3 glucose. High glucose solutions, KCl and Ca²⁺-free solutions were made by equimolar substitution of NaCl.

Tissue Culture.

Human islets were provided by the Juvenile Diabetes Research Foundation Islet Isolation Core (Washington University School of Medicine, St. Louis). Islets were gently dispersed after three 1-min washes with Ca²⁺/Mg²⁺-free Eagle's minimal essential media (MEM; Mediatech, Herndon, VA), followed by 30-s exposure to trypsin/EDTA (GIBCO) solution diluted 1:5 in MEM. Cells were plated on coverslips in RPMI 1640 [containing 5.5 mM glucose and 100 units/ml penicillin, 100 μg/ml streptomycin (pH 7.4) with NaOH] and supplemented with 10% FCS. Cultures were maintained at 37°C, 5% CO₂ and saturated humidity. Cells were studied 2–4 days after dispersion. Experiments were replicated on cells from two to five donors.

Ca²⁺ Imaging.

Islet cells were incubated with 1 μM Fura-4F-AM (Molecular Probes) in RPMI 1640 for 30 min and rinsed in Ringer's solution for 30 min. Coverslips, in a narrow 32°C chamber, were continuously perifused with preheated solutions to maximize control over the contents of the bath. Recordings were made from 6–25 cells by using a monochromator, a charge-coupled device (CCD) camera (TILL Photonics, Planegg, Germany) and an IX70 microscope (×20 or ×40 objectives; Olympus, Tokyo). Ratios (340/380) were calibrated by exposing cells to 10 μM ionomycin for >30 min (R_MAX), then 20 μM EGTA (R_MIN). Ca²⁺ responses were defined as having 1-min bins with a mean [Ca²⁺]_c that was more than two standard deviations over the mean [Ca²⁺]_c of the 5-min pretreatment period. Ca²⁺ responses listed in the text were quantified as maximal amplitude above baseline. Unless indicated otherwise, the number of replicates reported in the figure legends and text represents the total number of islet cells in imaging experiments.

Microinjection.

Fura-4F-AM-loaded islet cells were injected by using an Eppendorf 5126. Microinjection pressure was 100 hPa (60 hPa compensation, duration 0.5 s). Either Eppendorf Femtotips II or pipettes pulled in house were filled with filtered “intracellular solution” consisting of (in mM): 144 KCl, 2 MgCl₂, 2 Na₂ATP, 5 EGTA, and 20 Hepes (pH 7.3 with KOH). No differences in the responses were seen when water was used as a vehicle. Unsuccessful injections (≈15%) showed immediate cell swelling and decrease in fluorescence and were discarded.

Hormone Release and RIA.

Dispersed cells were cultured for 3 days on NUNCLON culture plates (Nalge). Cells were washed for 30 min, then incubated as above for 90 min in Kreb's Ringer's buffer containing (in mM): 115 NaCl, 24 NaHCO₂, 5 KCl, 2 CaCl₂, 1 MgCl₂, and 5 g/liter RIA grade albumin, after which the supernatant was removed and frozen for subsequent RIA. BAPTA-AM was included in the 30-min wash for some experiments. Cellular hormone content was measured after lysing cells by freeze/thaw cycles and sonication in distilled H₂O containing 10 micro-trypsin inhibitory units (μTIU)/ml aprotinin (Sigma). Human insulin and c-peptide radioimmunoassays were conducted by the Washington University School of Medicine RIA facility.

Statistical Analysis.

The effects of treatments on the frequency of responses to insulin were evaluated by χ² test. Statistical analyses on parametric data were performed by using Student's unpaired t test or one-way ANOVA [followed by Fisher's probable least-squares difference (PLSD) post hoc test]. Differences were considered significant when P < 0.05. Results are presented as mean ± SEM.

Results and Discussion

Human Beta Cells Generate Complex Insulin-Evoked Ca²⁺ Signals.

We imaged a large number of cells (n = 335) to establish the occurrence, and characterize the extended time course, of insulin-stimulated Ca²⁺ signals in human pancreatic islet cells. On treatment with 200 nM insulin for 15 min, 31% of cells responded with significant Ca²⁺ signals (mean amplitude 284 ± 24 nM above baseline). The fraction of human islet cells responding to 200 nM insulin is comparable to previous findings in mouse islet cultures; there, 43% of tolbutamide-sensitive cells responded to a 30-s pulse of 100 nM insulin (4, 5).

Multiple Ca²⁺ signal waveforms were observed (Fig. 1 A–E). The first type was characterized by two or more large bursts superimposed on a prolonged plateau. The second type of Ca²⁺ signal consisted of higher-frequency oscillations on a sustained plateau. A third type consisted of a large monophasic response without oscillations. Less common was a fourth type of response that included short spikes with a negligible plateau phase, and a fifth type exhibiting a single transient burst that returned to the preexposure baseline. Thus, salient features of insulin-stimulated Ca²⁺ signals are oscillatory behavior (70%) and a plateau that persists during wash out (≈80%). The significance and mechanism of these oscillations is not clear. Surprisingly, removal of insulin evoked a large transient Ca²⁺ signal in ≈15% of nonresponding cells (not shown). Perhaps cells responded to low insulin levels during the process of wash out.

Fig 1.

Insulin-stimulated Ca²⁺ signals in human islet cells. (A–E) Representative traces from 101 of 335 cells responding to 200 nM insulin (INS, open bars) are shown. The relative frequency of each wave form, of the 30% of insulin-responsive cells, is indicated. (F) The percentage of cells responding at each insulin concentration (0.002–20,000 nM). (G) The maximal amplitude of Ca²⁺ signals for various insulin levels (n = 23, 18, 335, 29, 20, 85, 35, and 25; increasing [insulin]) compared with control solution changes (n = 10).

Next, we examined the dose-response relationship of autocrine insulin signaling. The maximal amplitude of the Ca²⁺ signals above baseline increased from picomolar concentrations of insulin to maximal levels at 2–2,000 nM insulin, but were reduced at 20,000 nM (Fig. 1G). Interestingly, the percentage of islet cells responding did not follow a similar pattern. This dose-response profile had a minor peak at 200 nM insulin and a major peak at 0.2 nM insulin, where 80% of cells responded (Fig. 1F). Ca²⁺ signals in 0.2 nM insulin were generally smaller and monophasic. This complex concentration-response profile can be explained by an interaction between positive and negative regulators of beta cell [Ca²⁺]_c with distinct concentration dependencies. Indeed, insulin (1–600 nM) reduced beta cell electrical activity dose-dependently by activating K_ATP channels in a mouse beta cells and islets bathed in high glucose (19). Alternatively, the smaller Ca²⁺ signals seen at 20,000 nM could be caused by the buildup of a second messenger that has strong desensitizing properties.

Other features of insulin-induced Ca²⁺ signals are presented in Fig. 2. First, successive exposures to insulin did not evoke repeated Ca²⁺ signals (Fig. 2A), suggesting the involvement of a depletable or desensitizable component. Second, we wanted to be reasonably certain that beta cells respond to insulin (60–80% of islet cells contain insulin). Indeed, insulin-stimulated Ca²⁺ signals occurred predominantly in larger cells, which have been shown to correspond to beta cells in many studies. In addition, where insulin-stimulated Ca²⁺ signals were relatively modest, 15 mM glucose and 15 mM arginine evoked subsequent Ca²⁺ signals (Fig. 2B), demonstrating that insulin stimulates beta cells. However, when insulin-stimulated Ca²⁺ signals were large and prolonged, additional increases in [Ca²⁺]_c were not evoked by 15 mM glucose (n = 25, not shown). Insulin did not increase [Ca²⁺]_c in small cells exhibiting oscillations in 3 mM glucose indicative of alpha cells or delta cells (n = 7; ref 20; Fig. 2C). In previous studies, mouse islet cells were treated with tolbutamide to identify putative beta cells (ref. 4, but see ref. 21). We avoided this approach because such stimulation alters beta cell Ca²⁺ homeostasis and bath insulin levels.

Fig 2.

Islet cells with the physiological characteristics of beta cells, but not alpha or delta cells, respond to single insulin treatments. (A) Repeated insulin treatments did not reliably evoke multiple Ca²⁺ signals (n = 21). (B) Insulin-stimulated Ca²⁺ signals could be localized to beta cells sensitive to sequential application of 15 mM glucose and 15 mM arginine (n = 41). (C) Cells displaying obvious alpha cell or delta cell behavior did not respond to insulin (n = 7).

Novel Intracellular Ca²⁺ Stores Mediate Insulin Signaling.

To determine the mechanism of insulin-stimulated Ca²⁺ signals, we used inhibitors of various different Ca²⁺ signaling pathways. Response rates and amplitudes in the presence of both insulin and inhibitors were compared with the responses of cells treated with insulin alone (parallel controls) because sequentially repeatable responses were not reliably observed with long insulin treatments. As was the case in cells stimulated with 200 nM insulin under control conditions, 30% of the cells initiated responses in nominally Ca²⁺-free solution, suggesting that these Ca²⁺ signals originated from intracellular Ca²⁺ stores (Fig. 3A; 189 ± 22 nM). However, the absence of plateaus/bursts indicated that the second phase of the insulin-stimulated Ca²⁺ signals required influx of extracellular Ca²⁺. Cells with elevated Ca²⁺ in normal extracellular [Ca²⁺] (≈200 nM; 20%) showed an immediate reduction in [Ca²⁺]_c on exposure to low Ca²⁺ Ringers (Fig. 3B), indicating that extracellular Ca²⁺ influx is constitutively active in a some cells and that changes in extracellular Ca²⁺ were rapid.

Fig 3.

Insulin-stimulated Ca²⁺ signals are initiated by intracellular Ca²⁺ stores. (A) Exposure to Ringer's solution without Ca²⁺ salts (NOCA) did not inhibit the first phase, but did block the second phase (n = 64). (B) NOCA Ringer's solution reduces [Ca²⁺]_c in some cells with high baseline values. (C and D) Insulin-stimulated Ca²⁺ signals were abolished in 1 μM thapsigargin (n = 82) (C) or 10 μM BHQ (n = 64) (D). (E) CPA evoked robust Ca²⁺ signals. Only 15% of all cells responded to insulin in the presence of CPA, representing approximately half the usual number of responsive cells (n = 71).

Next, we directly confirmed the involvement of intracellular Ca²⁺ stores in insulin signaling by blocking SERCA pumps, which refill many agonist-sensitive Ca²⁺ stores, with three structurally different inhibitors: thapsigargin, BHQ, and CPA (22). Insulin-stimulated Ca²⁺ signals were virtually abolished in 1 μM thapsigargin (Fig. 3C; 3% of cells displayed diminutive responses), whereas 30 mM KCl-induced Ca²⁺ signals were unaffected. No cells responded to insulin in the presence of 10 μM BHQ (Fig. 3D). Surprisingly, exposure to 10 μM CPA alone produced larger elevations in [Ca²⁺]_c when compared with thapsigargin and BHQ, suggesting the possibility of distinct molecular actions (J.D.J., unpublished results). In the presence of CPA, insulin generated normal-sized Ca²⁺ signals (286 ± 57 nM) in 15% of cells (Fig. 3E), suggesting that a subpopulation of cells requires CPA-sensitive Ca²⁺ stores for insulin signaling. Thus, insulin releases Ca²⁺ stores refilled by thapsigargin/BHQ-sensitive SERCA.

Having established that intracellular Ca²⁺ stores mediate insulin signaling in human beta cells, we examined which intracellular Ca²⁺ mobilization pathway(s) are involved. It is established that beta cells contain IP₃-sensitive Ca²⁺ stores, which mediate the signaling of some agonists and play a role in the generation of Ca²⁺ oscillations (23). In the presence of 1 μM xestospongin C, a putative IP₃ receptor inhibitor (ref. 24, see below), the initiation of insulin signaling was still observed (Fig. 4A; 248 ± 27 nM; 40% responded), but the plateau was reversibly inhibited (Fig. 4B).

Fig 4.

Insulin signaling is not initiated by IP₃- or ryanodine-sensitive Ca²⁺ stores. (A) Xestospongin C did not block the first phase of insulin-stimulated Ca²⁺ signals (n = 20). The second phase was absent or consisted only of baseline spiking. (B) Xestospongin C, applied after the initiation of insulin-stimulated Ca²⁺ signals, inhibited oscillations (n = 4 insulin-responding cells). (C and D) Neither phase of the Ca²⁺ signals evoked by insulin is sensitive to 10 μM ryanodine (Ry; n = 52), regardless of preexposure to 30 mM KCl (n = 14). (E) Caffeine did not inhibit the initiation of insulin-stimulated Ca²⁺ signals (n = 44).

Human beta cells also possess Ca²⁺ stores mobilized via RyR2 activation (25). Indeed, 1–10 nM ryanodine, which locks RyR in an open subconductance state, evoked Ca²⁺ signals in beta cells with a maximal amplitude of 117 ± 24 nM, confirming the presence of functional ryanodine-sensitive Ca²⁺ stores in our preparations (n = 21; not shown). Notwithstanding, 10 μM ryanodine did not block either phase of insulin-stimulated Ca²⁺ signals (Fig. 4C; 233 ± 30 nM; 35% responded). Ryanodine binds to RyR in their open conformation, making ryanodine action use-dependent in some cells (26). A protocol with a 30 mM KCl pretreatment to activate RyR-mediated Ca²⁺-induced Ca²⁺ release still failed to inhibit insulin signaling (Fig. 4D; 285 ± 56 nM; 43% responded). Caffeine, which can empty RyR-containing Ca²⁺ stores or inhibit IP₃R-dependent Ca²⁺ release depending on the cell type (26), did not affect the initiation of insulin-evoked Ca²⁺ signals (167 ± 13 nM; 27% responded, Fig. 4E), but inhibited the plateau phase. Caffeine evoked only modest Ca²⁺ release in 14% of cells. Together, these data strongly suggest that IP₃R- and RyR-linked pathways, including those downstream of cyclic ADP-ribose accumulation, do not initiate insulin signaling in human beta cells.

NAADP-Sensitive Ca²⁺ Stores Mediate Insulin Signaling.

In mammalian cells, NAADP is the most potent intracellular Ca²⁺-releasing metabolite known (10) and is involved in Jurkat T lymphocyte activation (11) and cholecystokinin (CCK) signaling in pancreatic acinar cells (8). Complex Ca²⁺ signals were evoked in large cells injected with NAADP (but not β-NADP or buffer), with a concentration-response profile peaking at 100 nM (Fig. 5), similar to what has been previously reported in T cells (11). Injection of NAADP was sufficient to reproduce several characteristics of insulin-evoked Ca²⁺ signals. As in T lymphocytes (11), bursts occurred at intervals of several minutes. Our findings contrast with the observations of others (27), where permeabilized ob/ob mouse beta cells did not respond to 10 nM–10 μM NAADP. If NAADP-sensitive Ca²⁺ stores are immediately below the plasma membrane, as they are in some cell types, they may have been destroyed with permeabilization. Using injection, our data strongly suggest that human beta cells possess NAADP-sensitive Ca²⁺ stores, with properties similar to other cell types.

Fig 5.

Functional NAADP-sensitive Ca²⁺ stores exist in human beta cells. (A) Large islet cells injected with various concentrations of NAADP exhibited oscillatory [Ca²⁺]_c elevations. (B) Maximal amplitude above baseline of Ca²⁺ signals evoked in individual cells injected with 10 different pipet concentrations of NAADP (n = 6, 3, 8, 16, 12, 5, 4, 2, 4, and 3; increasing [NAADP]). (C) β-NADP (10 μM) did not evoke Ca²⁺ signals (n = 8).

We further tested the involvement of NAADP-sensitive Ca²⁺ stores in insulin signaling by desensitizing the NAADP receptor with micromolar concentrations of NAADP (8, 10, 11). Injection of 10 mM NAADP blocked all insulin-stimulated Ca²⁺ signals (Fig. 6A). Cells injected with buffer alone responded to insulin (Fig. 6B; 181 ± 39 nM; 24% responded). Subsequent depolarization-evoked Ca²⁺ influx was not affected by the injection procedure, indicating that general Ca²⁺ homeostasis remained intact. Taken together, these data demonstrate that, in human beta cells, NAADP-sensitive Ca²⁺ stores are required to initiate insulin signaling. These findings complement other work supporting a role for CD38 in beta cell physiology (14, 15, 18). However, previous studies have focused on cADPr as the mediator of CD38 signaling. Although a rise in cADPr is coincidental with glucose stimulation, evidence suggests that cADPr is not sufficient for insulin release and that glucose-stimulated insulin release is not blocked by a cADPr inhibitor (28). Our data suggest that the cADPr/RyR pathway is not involved in autocrine insulin signaling in beta cells.

Fig 6.

NAADP-sensitive Ca²⁺ stores mediate insulin signaling. (A) Injection of 10 mM NAADP to desensitize the NAADP-receptor blocked all insulin-stimulated Ca²⁺ signals (n = 28). (B) Insulin evoked normal Ca²⁺ signals after injection with buffer (n = 24).

CD38 is found on the plasma membrane and in multiple organelles, including the nucleus (29–31). CD38 is known to favor the production of NAADP over cADPr at acidic pHs or in the presence of cAMP and may require endocytosis, suggesting its subcellular location may be important (32, 33). The exact cellular location of NAADP-sensitive Ca²⁺ stores is also unknown, but they are pharmacologically distinct from Ca²⁺ stores sensitive to IP₃ or cADPr/ryanodine in other cell types. NAADP initiated Ca²⁺ signals in human beta cells preincubated for ≈2 min in Ca²⁺-free solutions, but did not elicit secondary oscillations or plateaus (Fig. 7A). The second phase of NAADP-induced Ca²⁺ signals was blocked by removing Ca²⁺ (Fig. 7B). NAADP-stimulated Ca²⁺ signals were insensitive to RyR blockade. Likewise, Ca²⁺ release could be initiated by NAADP, but not IP₃, in the presence of xestospongin C (Fig. 7 D–F). These results indicate that NAADP acts on Ca²⁺ stores relatively distinct from those modulated by ryanodine and xestospongin in this cell type with a pharmacological profile similar to that of insulin. Interestingly, NAADP-stimulated Ca²⁺ release can be inhibited by voltage-gated Ca²⁺ channel antagonists (34), which are potent suppressors of Ca²⁺ signaling in human beta cells (35).

Fig 7.

Properties of NAADP-induced Ca²⁺ signaling. (A) Ca²⁺ signals could be initiated by 100 nM NAADP in nominally Ca²⁺-free conditions (NOCA, n = 4). (B) Removal of Ca²⁺ abrogated a second phase of Ca²⁺ signaling in cells previously stimulated by NAADP (n = 2). (C) Neither the initiation nor the maintenance of NAADP-evoked Ca²⁺ signals was sensitive to 10 μM ryanodine (Ry; n = 5). (D) Xestospongin C (Xesto) did not block the initiation if NAADP-stimulated Ca²⁺ release (n = 3). (E) Microinjection of 100 μM IP₃ evoked oscillatory Ca²⁺ signals (n = 9). Often, 1 μM xestospongin C induced a slow elevation in Ca²⁺ in cells previously stimulated with IP₃ (n = 3). (F) IP₃ (100 μM) did not evoke Ca²⁺ signals in the presence of xestospongin C (n = 5).

We present evidence for the sequential activation of multiple Ca²⁺ stores in beta cells, as in mouse pancreatic acinar cells where NAADP initiates Ca²⁺ release upstream of IP₃ receptors (8). Future work is required to determine the mechanism by which the insulin receptor substrate-1 (IRS-1)/phosphatidylinositol 3-kinase (PI3K) pathway releases intracellular Ca²⁺. It is possible that the first and second phase of insulin signaling, or the distinct Ca²⁺ waveforms, may have different physiological functions. In Ascidian oocytes, both cADPr- and NAADP-sensitive Ca²⁺ stores modulate membrane currents, but only cADPr regulates exocytosis (36). The subcellular location, amplitude, duration, and oscillation frequency of specific Ca²⁺ signals can lead to the differential regulation of cellular processes (37–39). Multiple Ca²⁺ stores, found in several organelles, add further complexity required to control the vast array of Ca²⁺-dependent cellular functions in endocrine cells (9).

Functional Consequences of Insulin Signaling.

Because a common feature of insulin-stimulated Ca²⁺ signals was their prolonged duration, we investigated the long-term actions of insulin on dispersed human islet cells by using cellular insulin content as an indicator of insulin production and secreted c-peptide as a marker of insulin exocytosis. We observed a marked increase in cellular insulin accumulation in cells treated with insulin for 90 min that was abolished by preincubating cells in BAPTA-AM, a membrane permeant Ca²⁺ buffer (Fig. 8A). The elevation in cellular c-peptide/insulin contents by insulin (0.2–200 nM) is time-dependent and reflects newly synthesized insulin (J.D.J., unpublished observations). These results complement previous studies describing a robust effect of insulin on insulin gene expression and protein synthesis (6, 40). Consistent with the role of insulin in insulin production and presence of known antiapoptotic components in the insulin signaling pathway (41, 42), islets of diabetic beta cell-specific insulin receptor knockout mice were 20–40% smaller and contained 35% less insulin (1). Interestingly, insulin activation of the pancreatic transcription factor PDX-1 is independent of K_ATP channel-mediated membrane excitability (43), suggesting a possible role for intracellular Ca²⁺ stores. Although both the production and secretion of hormones are dependent on calcium at several levels (9, 35, 44), insulin biosynthesis and secretion are not always coupled (45). In dispersed human islet cells, exogenous insulin did not stimulate c-peptide secretion in either resting or high glucose (Fig. 8B) in 90-min static incubations (or at 5 min, 30 min, or 6 h; J.D.J., unpublished observations). At some concentrations, secretion was significantly inhibited. Notably, high-resolution amperometry studies showed that the application of 100 nM insulin resulted in only modest exocytosis (≈10 insulin granules) in a minority of beta cells (4, 5). That insulin does not stimulate robust, prolonged hormone release agrees with the requirement for voltage-gated Ca²⁺ channel-mediated Ca²⁺ influx for exocytosis in human beta cells (35).

Fig 8.

Physiological outcomes of insulin signaling. (A) Insulin increased cellular insulin contents in dispersed islet cells during 90-min static incubations, through mechanisms sensitive to preincubation with 10 μM BAPTA-AM for 30 min (n = 6). Open stars indicate significant stimulation compared with control. Filled stars indicate significant inhibition by BAPTA. (B) Insulin does not stimulate c-peptide release from dispersed islet cells in static incubation (n = 18). Stars indicate significant difference from appropriate control.

In conclusion, we have characterized in beta cells a third class of intracellular Ca²⁺ store, sensitive to NAADP, which mediates autocrine insulin feedback. These Ca²⁺ stores comprise an early step in the generation of complex Ca²⁺ signals that are sustained through the influx of extracellular Ca²⁺ and xestospongin/caffeine-sensitive processes. These results provide a basis for the comparison of beta cell insulin signaling with insulin signaling in other tissues and with the signal transduction of other growth factors.

Abbreviations

• SERCA, sarcoplasmic/endoplasmic Ca²⁺-ATPase

• IP₃, inositol 1,4,5-trisphosphate

• cADPr, cyclic ADP-ribose

• NAADP, nicotinic acid-adenine dinucleotide phosphate

• BHQ, 2,5-di-tert-butylhydroquinone

• CPA, cyclopiazonic acid

• BAPTA-AM, 1,2-bis-(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester