Insulin Constitutively Secreted by β-Cells Is Necessary for Glucose-Stimulated Insulin Secretion

Type V collagenase (for pancreatic islet isolation), soybean trypsin inhibitor, phenol red, Histopaque 1077, and fetal bovine serum (FBS) were purchased from Sigma (St. Louis, MO). RPMI-1640 media with glutamine (Cat. No. 10-040-CV) was from Mediatech Cellgro (Herndon, VA). Anti-insulin (H-86, Santa Cruz Biotechnology) is a rabbit polyclonal antibody whose epitope includes residues 25–110 of human proinsulin. H-86 immunoprecipitates insulin of mouse, rat, or human origin. Rabbit anti-mouse IgG (anti-IgG) was purchased from Jackson Laboratories (Bar Harbor, ME). Rat insulin enzyme-linked immunosorbent assay (ELISA) and ultra-sensitive rat insulin ELISA kits were obtained from Crystal Chemicals (Chicago, IL). These kits are 100% cross-reactive with mouse insulin and do not detect C-peptide. Calibration curves to mouse insulin (Crystal Chemicals) were prepared.

Hanks’ balanced salt solution (HBSS; pH 7.4) contained KCl 5.4 mmol/l, Na₂HPO₄ 0.3 mmol/l, KH₂PO₄ 0.4 mmol/l, NaHCO₃ 4.2 mmol/l, CaCl₂ 1.3 mmol/l, MgCl₂ 0.5 mmol/l, MgSO₄ 0.6 mmol/l, NaCl 137 mmol/l, glucose 5.6 mmol/l, and phenol red 20 mg/dl. HBSS-BSA was HBSS containing 1% (wt/vol) BSA. Krebs-Ringer bicarbonate buffer (KRBB; pH 7.4) contained CaCl₂ 2.4 mmol/l, NaCl 120 mmol/l, MgSO₄ 1.2 mmol/l, KCl 5.4 mmol/l, KH₂PO₄ 1.2 mmol/l, HEPES 20 mmol/l, and phenol red 20 mg/dl. KRBB-BSA was KRBB buffer with 0.1% (wt/vol) BSA. Insulin secretion was monitored in KRBB-BSA with 3, 11, or 25 mmol/l glucose or with 20 mmol/l arginine. Acid-ethanol was 77% (vol/vol) ethanol and 1% (vol/vol) 12N HCl.

CD1 albino mice (Charles River), ages 3–4 months, were fed ad libitum and kept on a 12-h dark-light cycle. All protocols for animal use and euthanasia were reviewed and approved by the Animal Care Committee of the Faculty of Medicine, University of Calgary. Mice were anesthetized with sodium pentobarbital (MTC Pharmaceuticals, Cambridge, ON), 115 mg/kg i.p. Then 2 ml collagenase solution (1.5 mg/ml collagenase, 2 mg/ml soybean trypsin inhibitor in HBSS-BSA) was injected into the pancreas through the bile duct (30). The pancreas was excised and incubated at 37°C for 16 min. Islets, acinar cells, and ductal tissue were resuspended in 20 ml HBSS-BSA and subsequently washed three times with 20 ml HBSS-BSA (centrifuged at 1,200 rpm [Jouan Cr 3i] for 1.5 min, decanted, and resuspended). After being filtered through nylon mesh, pancreatic digests were centrifuged and pellets were resuspended in 10 ml Histopaque 1077. Next, 10 ml HBSS were overlaid on Histopaque layers. After being centrifuged for 25 min at 2,400 rpm, 10 ml of the interface was collected, diluted with 20 ml HBSS-BSA, and washed three times as described above. Islets were collected under a Wild M5A microscope (Heerbrugg, Switzerland) and cultured in RPMI-1640 with glutamine, 10% FBS, 3 mmol/l glucose, penicillin 100 units/ml, and streptomycin 100 pg/ml for 16–40 h, a period that allows islets to equilibrate to in vitro conditions and to become enriched over acinar tissue (31).

Except where noted, all steps were performed at room temperature (20°C). RPMI-equilibrated islets of similar size were suspended in 10 ml KRBB buffer in a 6-cm culture plate and then transferred to a second 10-ml KRBB buffer plate. Sequential transfer of islets from the RPMI-calf serum medium to the two KRBB plates diluted 10⁴- to 10⁵-fold factors that may be in the culture media (Fig. 1). This wash procedure increases the recovery of islets over centrifugation wash steps and is less stressful to islets.

Unless otherwise noted in the legend to Fig. 1, the following steps were subsequently used. Islets were incubated for 30 min at 4°C in 10 ml KRBB-BSA and 3 mmol/l glucose (control), or in 10 ml KRBB-BSA, 3 mmol/l glucose, and 1 μg/ml anti-insulin or 1 μg/ml anti-IgG. Control and antibody-treated islets were washed. Islets (n = 5–10) were picked and placed in wells of a 96-well plate containing 200 μl insulin secretion media. To ensure suspension of secreted insulin, plates were rotated (Labline 3D Rotator). At 1, 2, 5, and 10 min, 15 or 25 μl aliquots were removed from wells. At 30 min, the remaining secretion media were removed and acid-ethanol (200 μl) was added. The 96-well plates were kept at –20°C for 16 h, after which well contents were collected. All samples were stored at –20°C until being assayed for insulin content (rat insulin ELISA or ultra-sensitive rat insulin ELISA kits).

Paired and unpaired Student’s t tests (Sigmaplot) were used to test for significance of differences between samples.

Figure 1 schematically illustrates the experimental procedure. Washed islets were treated (incubated for 30 min at 4°C) in one of three ways: with KRBB-BSA and 3 mmol/l glucose (control); with KRBB-BSA, 3 mmol/l glucose, and 1 μg/ml anti-insulin; or with KRBB-BSA, 3 mmol/l glucose, and 1 μg/ml anti-IgG. After the wash steps of treated islets, 5–10 islets were transferred to wells containing 200 μl KRBB-BSA buffer containing 3, 11, or 25 mmol/l glucose or 20 mmol/l arginine (29). Aliquots were removed at timed intervals. The insulin content of collected aliquots and acid-ethanol extracts of islets was determined immunologically. To compare data among different secretion media, islet treatments, and assays on different days, insulin secretion was expressed as the percent of islet insulin content.

Untreated islets secreted insulin in an anticipated manner (Fig. 2A). When islets were added directly to 25 mmol/l glucose after the wash step of an overnight culture (Fig. 1), there was a rapid rise in insulin secretion after 1–2 min (first phase), little or no change in the next 3 min, and then a slower rise for the next 15 min (second phase). Control-treated islets (Fig. 2B) or anti-IgG−treated islets (Fig. 2D) also secreted insulin in response to 25 mmol/l glucose, but the onset appeared to have a 5-min lag. In contrast, islets preexposed to anti-insulin did not exhibit 25 mmol/l glucose-stimulated insulin secretion during the 30-min insulin secretion assay (Fig. 2C). The latter islets did respond to 20 mmol/l arginine (Fig. 2E), similar to arginine-stimulated insulin secretion in anti-IgG−treated islets (Fig. 2F). Further, arginine-stimulated insulin secretion did not have a lag phase, as was seen with glucose-stimulated insulin secretion (Fig. 2B and 2D). Finally, control–, anti-IgG−, and anti-insulin−treated islets contained statistically equivalent amounts of insulin (Fig. 3), indicating that islet insulin content was not affected by the treatment procedure.

Control-treated islets (5 islets/well) exposed to 11 or 25 mmol/l glucose demonstrated similar secretion patterns and were equivalent in the percent of islet insulin content secreted at each time point (P > 0.1) (Fig. 4A). At 25 mmol/l glucose, increasing the number of islets per well to 10 did not change the percent of islet insulin content secreted at any time point (P > 0.1) (Fig. 4A). However, the amount of insulin secreted and the rates of insulin secretion at measured time points were different (Fig. 4B); whereas 5 control-treated islets secreted insulin at a slow rate for 5 min, 10 control-treated islets demonstrated a much sharper increase in insulin secretion in the same time period. Between 10 and 30 min, insulin secretion rates at 25 mmol/l glucose were independent of islet number.

Tail vein insulin concentrations in fasted and intraperitoneal glucose−injected mice were ∼200 and ∼1,500 pg/ml², respectively. These concentrations are sufficient to initiate insulin’s metabolic responses in isolated rat adipocytes (27). Capillaries surrounding islets, however, are near the site of insulin secretion (28), and concentrations of insulin in the vicinity of β-cells are likely higher. It would appear that mouse islets are continuously exposed to stimulating levels of insulin. Our data show that β-cells need this insulin to elicit glucose-stimulated insulin secretion.

Islets secrete insulin constitutively and in a regulated manner (32). Experimentally, the objective was to deplete constitutively secreted insulin from islet suspensions and deplete β-cell insulin receptors of their bound insulin. Anti-insulin H-86 (Santa Cruz Biotechnology) recognizes residues 25–110 of human insulin and immunoprecipitates mouse insulin. The regions of insulin recognized by the insulin receptor are within the final 10 amino acids of the B chain and the first five amino acids of the A chain (33); that is, the insulin receptor and H-86 recognize the same regions of insulin, and antibody binding of insulin is restricted to its unbound form. Because binding of insulin to its receptor is entropy driven (34), lower temperatures will facilitate insulin dissociation. The temperature of islet treatment was 4°C. The t_1/2 to displace insulin from its receptor is 30 min (34) and is the time period used to treat islets. Finally, anti-insulin was ∼7 × 10⁴-fold in excess over insulin receptors (i.e., anti-insulin was ∼6.7 nmol/l and insulin receptor was ∼0.1 nmol/l). This concentration difference is sufficient for H-86 to successfully compete with insulin receptors for unbound insulin. In summary, antibody H-86, its concentration, and the time and temperature used to treat islet suspensions should deplete β-cells of receptor-bound and extracellular insulin.

Insulin secretion from H-86–treated islets (Fig. 2C) was similar to the basal rate (3 mmol/l glucose) from control-treated islets (Fig. 2B) and was independent of glucose concentration (3 or 25 mmol/l). In contrast, 25 mmol/l glucose increased insulin secretion in control-treated (Fig. 2B) and anti-IgG−treated islets (Fig. 2D). The anti-insulin−mediated loss of stimulus-secretion coupling is specific for glucose because 20 mmol/l arginine stimulated insulin secretion equally well in anti-IgG−treated (Fig. 2F) and anti-insulin−treated islets (Fig. 2E). Hence, islets depleted of receptor-bound and constitutively secreted insulin did not respond to glucose up to 30 min after anti-insulin treatment.

Surprisingly, control-treated (Figs. 2B and 4A) and anti-IgG–treated islets (Fig. 2D) lagged 5 min in their response to 25 mmol/l glucose, something not seen with untreated islets (Fig. 2A). Previously, a 5-min lag was observed in insulin secretion when rat islets, immobilized on a nylon membrane, were first perifused (1 ml/min) with the KRBB 3 mmol/l glucose buffer for 30–40 min before perifusion with 8 or 15 mmol/l glucose (15). The absence of islet sensitivity for 5 min is specific for the glucose stimulus because anti-IgG–treated islets showed no lag in their response to 20 mmol/l arginine (Fig. 2F). The loss in glucose sensitivity was shortened if the number of islets in the secretion assays (islets/well) was doubled (Fig. 4B), so that 10 control-treated islets stimulated with 25 mmol/l glucose demonstrated an insulin secretion time profile reminiscent of that of untreated islets (Figs. 2A and 4B). These data reveal that islets secrete a factor required by β-cells to be glucose sensitive and that perifusion (15) or incubation in KRBB-BSA plus 3 mmol/l glucose for 30 min at 4°C removes this factor. Further, recovery of glucose sensitivity occurred when the islet solution regained this factor.

Islets are made up of α-, β-, γ-, and PP-cells. Although α-, γ-, and PP-cells are few in number, they represent the outer layer of cells. It is their secreted factors that would be removed first during perifusion or 30 min of the control treatment (Fig. 1). Glucagon from α-cells increases glucose-stimulated insulin secretion, but somatostatin from γ-cells inhibits insulin secretion (35). Removal of glucagon may lower a glucose-stimulated response, but it would not prevent it, as was seen in the first 5 min with control-treated islets. On the other hand, insulin (3), IGF-I (10), and NGF (11) synthesized in β-cells, the major cell type in islets, may sensitize islets to glucose. Because untreated islets are cultured in RPMI-FCS and FCS may contain insulin, IGF-1, and NGF, their respective receptors may be occupied. During 30 min of control or anti-IgG treatment, or during perifusion, some ligand dissociation from receptors is likely; recovery of glucose sensitivity would depend on reassociation of one or more of these peptides.

Adult β-cells express NGF receptors, but NGF itself has not been detected (36). Except during recovery from pancreatectomy, β-cell expression of IFG-I is low (37) and mouse islet IFG-I receptor mRNA content is ∼50% of its insulin receptor mRNA content (H.J.G., R. Kulkarni, C.R. Kahn, unpublished observations). Thus, of the three β-cell−secreted peptides, insulin is the most likely to resensitize β-cells to glucose. For example, during insulin secretion assays (Fig. 1), islets were exposed to constitutively secreted insulin. Figure 4B indicates that in one such assay, five control-treated islets were exposed to 0.3 nmol/l insulin for 5 min. Subsequently, 25 mmol/l glucose was able to increase insulin secretion. In three other experiments (n = 3 for each experiment), in which five control-treated islets were assayed for 25 mmol/l glucose−stimulated insulin secretion, the islets experienced 0.12 ± 03 nmol/l insulin for 5 min before their insulin secretion increased. In contrast, when control-treated islets were exposed to ∼1 nmol/l insulin for ∼1 min, glucose-stimulated insulin secretion was seen immediately (Fig. 4B).

Control treatment or perifusion would remove extracellular insulin, and anti-insulin treatment would also remove receptor-bound insulin. Thus, to observe a recovery of glucose sensitivity in anti-insulin–treated islets, one may anticipate a need for additional insulin and insulin exposure time. During secretion assays of anti-insulin–treated islets, five islets were exposed to 0.06 ± 0.03 nmol/l insulin (two experiments, n = 3) during the 5-min lag period. This insulin concentration and time exposure were not sufficient to detect 25 mmol/l glucose−stimulated insulin secretion during 30-min secretion assays (Fig. 2C). In a third anti-insulin−treated islet experiment (n = 3; data not shown), islet number and size were greater and the level of insulin secretion reached 0.3–0.4 nmol/l in the first 10 min. At 30 min, the extracellular insulin concentration reached ∼0.9 nmol/l or approximately twice the basal insulin secretion. Taken together, these data are consistent with insulin being a β-cell secretion product responsible for rendering islets glucose sensitive.

First-phase insulin secretion is the insulin expelled from plasma membrane−tethered β-granules (2,7), and glucose metabolism activates vesicle-plasma membrane fusion (38,39). Second-phase insulin secretion includes steps to recruit nontethered β-granules to the plasma membrane. β-IRKO mice do not exhibit first-phase glucose-stimulated insulin secretion, but they do retain second-phase insulin secretion (9). An absence of insulin signaling, as in β-IRKO mice, may prevent tethering of first-phase β-granules, activation of tethered granules, or vesicle-membrane fusion. The amino acid arginine also stimulates two phases of insulin secretion. In β-IRKO mice, arginine stimulates both phases, suggesting that an appropriate number of first-phase β-granules are tethered in β-IRKO mice. It follows, then, that insulin signaling contributes to the activation process and/or the fusion process (Fig. 5).

An appropriate [Ca²⁺]_c is critical in the fusion of β-granules with plasma membranes (2,40). Because arginine is able to stimulate first-phase insulin secretion independent of insulin receptors (9), either arginine initiates signaling steps not initiated by glucose or arginine increases intracellular Ca²⁺ at appropriate sites that can be reproduced by glucose and insulin but not by glucose alone.

Evidence indicates that insulin can elevate [Ca²⁺]_c by stimulating the release of stored Ca²⁺ from intracellular stores. Ca²⁺ can enter the cytosol from the endoplasmic reticulum (ER) or from β-granules via multiple Ca²⁺ transporters: inositol triphosphate (IP₃) receptors, ryanodine receptors (RyR), or nicotinic acid dinucleotide phosphate (NAADP) receptors (41,42). RyR in complex with FK506 binding protein (FKBP12) inhibits Ca²⁺ exit from intracellular stores. Cyclic-adenosine diphosphate ribose (cADPR) binds to FKBP12, leading to its dissociation from RyR and de-inhibition of the Ca²⁺ transporter (43). cADPR and NAADP, the ligand for NAADP receptors, are synthesized from ATP and are catalyzed by ADP ribosyl cyclase (CD38). In mice lacking CD38, or in humans with autoantibodies to CD38 or missense mutations in CD38, glucose-stimulated insulin secretion is impaired (44–46). This indicates that appropriate β-cell concentrations of cADPR and/or NAADP appear to be necessary for glucose-stimulated insulin secretion. Interestingly, insulin may regulate CD38 activity (47). Activation of caged NAADP in β-cells produces oscillations in [Ca²⁺]_c; NAADP is stimulatory at low concentrations and inhibitory at high concentrations. Insulin regulates [Ca²⁺]_c oscillations with a bell-shaped dosage dependency: stimulatory at low insulin concentrations and inhibitory at high insulin concentrations (47). Because insulin and NAADP are similarly dosage dependent on [Ca²⁺]_c, then insulin may regulate NAADP synthesis and hydrolysis, a property of CD38 (43) and insulin-regulated CD38 would affect cytosolic cADPR and NAADP levels, the Ca²⁺ transport activity of RyR and NAADP receptors, and, consequently, [Ca²⁺]_c (Fig. 5, site 1*). Further, with RyR and NAADP receptors located in β-granules (42), their activation would provide Ca²⁺ at sites essential for membrane fusion (40).

The passage of cytosolic Ca²⁺ back to the ER is with the Ca²⁺-transporter sarco-endoplasmic reticulum Ca²⁺ ATPase (SERCA) (48). Phosphorylated IRS-1, the product of activated insulin receptor tyrosine kinase (16), binds to SERCA and inhibits its Ca²⁺ transport activity (49) (Fig. 5, site 2*). In the presence of inhibited SERCA, a leaky IP₃ receptor and Ca²⁺-induced Ca²⁺ release through IP₃ receptors or RyR elevate [Ca²⁺]_c (41).

In addition to contributing to cytosolic Ca²⁺, insulin will change the phosphorylation states of cytosolic proteins; that is, insulin regulates the activity of several Ser/Thr kinases (16). A potential target of Ser/Thr kinases is the Rab3A effector, Rab3 interacting molecule (RIM). RIM is a scaffold protein whose binding domains for proteins of the β-granule−plasma membrane fusion process have phosphorylation sites within them. Because the phosphorylation state of RIM’s binding domains will affect its scaffold activity (50), then insulin’s role may include ensuring correct protein-protein associations for insulin exocytosis (Fig. 5, site 3*).

In conclusion, for β-cells to be responsive to glucose, the presence of a stimulatory level of insulin is required. Constitutively secreted insulin seems to be sufficient for this purpose. The necessity for insulin signaling appears to be at the fusion of β-granules with plasma membrane, but whether its role is in providing Ca²⁺ at fusion sites or in the regulation of protein associations in the membrane fusion process is not known. In addition to resolving the latter issues, further studies are needed to identify which intermediates of insulin signaling cascades react with intermediates of the glucose-stimulated cascade, and how such interactions allow first-phase insulin exocytosis.

This study was supported by the Cosmopolitan Foundation of Canada, Inc. S.S. was supported in part by funds from the Government of Canada Summer Career Placement Program.