cAMP-Activated Protein Kinase-Independent Potentiation of Insulin Secretion by cAMP Is Impaired in SUR1 Null Islets

The generation of Sur1KO mice has been described previously (6).

Human GIP, human GLP-1-(7–36)amide, and exendin-4 were purchased from Sigma Chemical (St. Louis, MO).

The methods of islet isolation and culture have been described elsewhere (6). Briefly, animals were anesthetized using sodium pentobarbital (65 mg/kg). Islets were isolated after digestion of the pancreas by intraductal injection of 1 mg/ml collagenase P (Boehringer-Mannheim, Indianapolis, IN) dissolved in Krebs-Ringer bicarbonate buffer (KRB) solution as described elsewhere (9,10). For batch experiments, the islets were incubated overnight in RPMI 1640 medium supplemented with 10% fetal bovine serum and test concentrations of glucose equilibrated with 5% CO₂/95% air, pH 7.4, at 37°C. Islets were preincubated in media containing 2.8 mmol/l glucose before incubation in 0.75 ml of medium containing test concentrations of glucose. Perifusion was done following Komatsu et al. (11) with modifications. One-hundred islets in a column of Bio-Gel P-10 (Bio-Rad, Richmond, CA) were continuously perifused at 37°C with Hepes-NaHCO₃ KRB (equilibrated with 5% CO₂/95% air, pH 7.4) at a flow rate of 0.75 ml/min. After preincubation for 30 min in 2.8 mmol/l glucose, solutions were changed as indicated in the figures. Test substances were added to the basal media without adjustment of the final osmolarity and were present from 10 through 90 min. Insulin assays were done using radioimmunoassay kits from Linco Research (St. Charles, MO) with rat insulin as a standard. cAMP was measured by radioimmunoassay using kits from Amersham Pharmacia Biotech (Piscataway, NJ).

cAMP-activated protein kinase (PKA) activity was assayed in permeabilized islets following Maller et al. (12). Briefly, islets were permeabilized at 4°C with digitonin (25 mg/l) in (mmol/l) 20 HEPES, 137 NaCl, 5.4 KCl, 0.3 NaHPO₄, 0.4 KH₂PO₄, 10 MgCl₂, 25 Na β-glycerol phosphate, 5 EGTA, 5.6 mmol/l glucose, and 20 μmol/l Na₃VO₄, pH 7.2, and incubated with 10 μmol/l ATP plus 2.5 μCi [γ-³²P]ATP and 0.1 μmol/l “Kemptide” (Leu-Arg-Arg-Ala-Ser-Leu-Gly) in the absence or presence of 5 μmol/l cAMP, 5 μmol/l cAMP plus 10 nmol/l protein kinase inhibitor (PKI), the classic Walsh heat-stable PKI (13), or 5 μmol/l cAMP plus H-89 (N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide), a synthetic PKA inhibitor (14). After a 10-min incubation at 30°C, the phosphorylated peptide was isolated on DEAE-cellulose, washed three times with ice-cold 75 mmol/l H₃PO₄, and counted in a scintillation counter.

Islets were isolated into RPMI medium supplemented with 1% BSA and 25 mmol/l HEPES, pH 7.4, and then cultured overnight in RPMI supplemented with 10% fetal bovine serum, 10 mmol/l HEPES, pH 7.4, 2 mmol/l glutamine, 1 mmol/l pyruvate, and 50 μmol/l β-mercaptoethanol. Islets were washed once with RPMI containing 2.8 mmol/l glucose and 1% BSA and then washed three times with a KRB solution buffered with 25 mmol/l HEPES, pH 7.4, supplemented with 1% BSA and 2.8 mmol/l glucose (10-min incubation per wash). The washed islets were preincubated for 40 min with 3-isobutyl-1-methylxantine (IBMX; 500 μmol/l) or IBMX plus 5 or 12.5 μmol/l H-89 or KT5720 before being stimulated for 20 min by addition of glucose to 16.7 mmol/l. The islets were solubilized with boiling SDS sample buffer, subject to electrophoresis, and transferred to nitrocellulose. Antiphosphorylated CREB and anti-CREB antibodies (Cell Signaling Technology, Beverly, MA) were used following the manufacturers recommendations, detected using an horseradish peroxidase-conjugated second-step anti-mouse antibody (Sigma Chemical, St. Louis, MO), and visualized by chemiluminescence.

Based on the estimated dissociation constant (K_D) for GLP-1 (∼200 pmol/l) (15) and plasma levels of GLP-1-(7–3)amide in mice after challenge with gastric glucose (∼15 pmol/l) (16), we used saturating concentrations (10 nmol/l) of GLP-1-(7–36)amide, GIP, and exendin-4, a GLP-1 receptor agonist (17), to show that these peptides stimulate insulin secretion from wild-type but not Sur1KO islets (Fig. 1A and B). Stimulation of wild-type islets requires elevated glucose and is abolished by the l-type Ca²⁺ channel blocker, nifedipine. The incretins comparably increase the steady-state islet cAMP content, indicating that the lack of response of Sur1KO islets is not due to altered coupling of incretin receptors with adenylyl cyclases (Fig. 1C). One-way ANOVA indicated that there were no significant differences between the wild-type and Sur1KO values within a group or with glucose or nifedipine concentration; therefore, we averaged the wild-type and Sur1KO values within a group (Fig. 1D).

To manipulate the cAMP content further, we inhibited phosphodiesterases with IBMX and stimulated adenylyl cyclases with forskolin. The intracellular content of cAMP ([cAMP]_i) is elevated similarly in both wild-type and Sur1KO islets after incubation (30 min) with forskolin, IBMX, or both agents (Fig. 2A). One-way ANOVA indicates there are no significant differences between the conditions within the four treatment groups, i.e., no treatment, plus IBMX, plus forskolin, or plus both agents; therefore, we averaged the values for the four islet groups. The grouped data are 86.79 ± 33.02, 254.87 ± 116.25, 2,471.9 ± 948.87, and 10,905 ± 3,167.3 fmol of cAMP/10 islets ± SD for no treatment versus treatments with IBMX, forskolin, or IBMX plus forskolin, respectively. ANOVA on this grouped data indicates that the approximate threefold difference with and without the addition of IBMX does not reach statistical significance (P > 0.05), whereas the differences between the control (no treatment) and other treatments are highly significant (P < 0.001).

Figure 2B shows the result of elevating [cAMP]_i on glucose-dependent insulin release from wild-type versus Sur1KO islets. The data in Fig. 2A and B were obtained from the same islets and are replotted in Fig. 2C, giving insulin release as a function of cAMP content. Increased [cAMP]_i has little effect on insulin release from wild-type islets in low glucose (2.8 mmol/l), whereas in high glucose (16.7 mmol/l), even the threefold average increase in [cAMP]_i after treatment with IBMX potentiates insulin secretion ∼15-fold above basal. In depolarized Sur1KO islets (6) with elevated [Ca²⁺]_i (18,19), superphysiologic concentrations of cAMP, >100-fold above basal values, significantly increased insulin release even in low glucose. Insulin release from Sur1KO islets in low glucose was about six times the basal rate of release or nearly twice as great as that obtained from wild-type islets in high glucose alone. In high glucose, Sur1KO islets exhibit a restricted stimulatory response at all [cAMP]_i values relative to that observed from wild-type islets (Fig. 2B and C). The effect is most evident at low [cAMP]_i values where wild-type islets display their full response. The difference in insulin release from Sur1KO islets in low and high glucose, both with elevated [Ca²⁺]_i, is consistent with the known Ca²⁺-dependent stimulation of insulin secretion by products of glucose metabolism (20–23, reviewed in 24) and long-chain acyl CoA (25, reviewed in 26). Consistent with earlier reports (27,28), increasing glucose metabolism did not raise [cAMP]_i in wild-type or Sur1KO islets significantly. The data show that there is no significant difference in the ability of wild-type versus Sur1KO islets to generate cAMP and suggest that the impairment in Sur1KO animals is in a cAMP-effector pathway.

Perifusion experiments were done at different [cAMP]_i to examine the kinetics of glucose-induced insulin secretion and confirm the loss of cAMP responsiveness in Sur1KO islets. To compare the responses of wild-type versus Sur1KO islets directly, we calculated the stimulated glucose-induced insulin secretion (GIIS) by subtracting the secretory values induced by high glucose alone from the values obtained when the stimulating agent was present. The procedure is illustrated in Fig. 3A using IBMX-potentiated glucose-induced insulin secretion as an example. Fig. 3B–E compare stimulated GIIS from Sur1KO with wild-type islets in which [cAMP]_i was varied with IBMX, dibutyryl cAMP, forskolin, or forskolin plus IBMX, respectively. At all [cAMP]_i values, the response of Sur1KO islets is impaired versus wild-type islets. The differential response of Sur1KO versus wild-type islets in high versus low glucose is most evident in Fig. 3E, where [cAMP]_i is elevated ∼100-fold (Fig. 2) and significantly stimulates insulin release in 2.8 mmol/l glucose. The results show that superphysiologic [cAMP]_i together with elevated [Ca²⁺]_i can drive insulin release, even when the rate of glucose metabolism is low.

Activation of protein kinase C (PKC) pathways by 500 nmol/l 12-O-tetradecanoylphorbol 13-acetate (TPA) stimulates insulin release from both wild-type and Sur1KO islets. In high glucose, the rates of release from wild-type versus Sur1KO islets were indistinguishable (Fig. 4A), suggesting saturation of the stimulatory effect. In low glucose, TPA significantly potentiates secretion from Sur1KO versus wild-type islets. Carbamylcholine (100 μmol/l) potentiates insulin release from wild-type islets (Fig. 4B); stimulation is greatest in high glucose but is significant even in low glucose. Insulin secretion from Sur1KO islets is markedly stimulated by carbamylcholine in both low and high glucose (Fig. 4C). Comparison of the release stimulated by glucose alone versus glucose plus carbamylcholine shows that there is a quantitatively greater effect on insulin release from Sur1KO islets, ∼3.7 ± 0.2-fold more than the control versus ∼2.0 ± 0.1-fold more for wild type, due mainly to the lower rate of secretion from the knockout islets. The results are consistent with the idea that the pathways by which PKC stimulates insulin secretion remain functional in Sur1KO islets.

It is not well understood where cAMP acts in the exocytotic pathway. Some have suggested that phosphorylation via PKA is critical for the action of cAMP in β-cells (29–32), but an early study (28) and recent reports indicate PKA-independent pathways are important in islets (33–36) and in the regulation of other endocrine systems (37). We tested whether inhibitors of PKA would block phosphorylation in wild-type islets and thus mimic the incretin refractory phenotype of Sur1KO islets. Three PKA inhibitors were used: H-89 (14), the reportedly more specific KT5720 (38), and (Rp)-Adenosine 3′,5′-monophosphorothioate (Rp-cAMPS) (39,40), which inhibits PKA and is resistant to phosphodiesterase but is expected to compete with cAMP for binding sites on Epac (exchange protein directly activated by cAMP)/cAMP-GEF (guanine-nucleotide exchange factor) proteins (41,42) without their activation (43). Neither H-89 or KT5720 markedly altered the potentiating effect of GLP-1 (Fig. 5) at 5 μmol/l, a concentration ∼100-fold greater than the constants for inhibition (K_is) of PKA (48 and 60 nmol/l for H-89 [14] and KT5720 [38]), respectively. Rp-cAMPS (100 μmol/l; K_i for PKA ∼11 μmol/l) did significantly inhibit potentiation of insulin release by elevated [cAMP]_i, reducing secretion to ∼30% of the control.

Figure 6A and B summarize dose-response curves for H-89 and KT5720 on insulin release from wild-type islets. Neither H-89 (25 μmol/l) nor KT5720 (25 μmol/l) produced a statistically significant effect when all the data were analyzed by ANOVA. Preincubation with H-89 did not significantly affect the potentiation of glucose-induced insulin release by different concentrations of IBMX (Fig. 6C); the estimated half-maximal activation values for IBMX were 9.5 ± 3.3 vs. 13.8 ± 1.2 μmol/l in the presence and absence of 5 μmol/l H-89, respectively. Perifusion experiments (Fig. 6D) show preincubation of islets with H-89 (5 μmol/l) does not markedly affect the kinetics of insulin release.

Digitonin (25 mg/l)-permeabilized islets incubated with γ[³²P] ATP and a PKA substrate, Leu-Arg-Arg-Ala-Ser-Leu-Gly (12), were used to estimate PKA activity. The results show that both wild-type and Sur1KO islets have comparable levels of kinase activity, which is stimulated 8- to 10-fold by 5 μmol/l cAMP, is inhibited by nanomolar concentrations of PKI (13) and by H-89 (Fig. 7A).

To show that H-89 and KT5720 were able to inhibit PKA in intact islet preparations, we assessed their ability to block cAMP-stimulated phosphorylation of CREB in intact wild-type islets under the same conditions used to test their effect on insulin release. Although influenced by a variety of signals, including Ca²⁺, growth factors, and cellular stress (see 37,44 for reviews), phosphorylation of CREB is used as a reporter of cAMP-dependent kinase activity. H-89 (Fig. 7A) and KT5720 inhibited the increase in islet CREB phosphorylation observed when [cAMP]_i was increased by IBMX treatment. The results are expressed as the ratio of phosphorylated CREB to total CREB, estimated using specific antibodies (Fig. 7C). Basal phosphorylation of CREB due to the activity of other kinases is approximately one-third the level stimulated by addition of IBMX. H-89 and KT5720 (5 μmol/l) completely blocked the stimulatory effect of IBMX on CREB phosphorylation consistent with their hydrophobic nature and estimated K_D values in the nanomolar range (14,38). The results indicate that both compounds can enter islets and inhibit PKA(s) and confirm a report in insulinoma cells and rat islets (45).

Mastoparan is a cationic, amphiphilic, 14 amino acid peptide that stimulates insulin release (46) reportedly at a late stage of exocytosis (47,48), an effect that is attenuated by pertussis toxin (46), independent of increased [Ca²⁺]_i (47,49), augmented by nutrients, and dependent on GTP (48). Mastoparan prompts insulin release from Sur1KO islets in the absence of external Ca²⁺, and this effect is augmented by glucose, 3.5 ± 0.8-fold for wild-type vs. 1.6 ± 0.9-fold for Sur1KO islets (Fig. 8). The apparent lower rates of insulin release from Sur1KO islets are consistent with their lower insulin content. The insulin contents (ng/50 islets; mean ± SD) for males are 5,192.2 ± 344.3 vs. 2,904.2 ± 556.1 and for females are 4,774.5 ± 572.7 vs. 2,174.4 ± 498.3, for wild-type versus Sur1KO islets, respectively. The values for fractional secretion for the three mastoparan challenges show that the normalized rates of secretion are essentially the same. The fractional values for each challenge are 1) 0.6 ± 0.1 vs. 1.4 ± 0.6, 2) 2.0 ± 0.3 vs. 2.3 ± 1.0, 3) 2.0 ± 0.5 vs. 2.4 ± 1.0, given as percentage of total insulin content ±SD for wild-type versus Sur1KO islets, respectively. The augmentation by glucose appears to be abolished by prior treatment with mastoparan, because a third exposure to mastoparan in the absence of glucose gives an equivalent bolus of insulin to that obtained with 16.7 mmol/l glucose. The results suggest there is no general secretory defect in Sur1KO β-cells and are consistent with the idea that the cAMP-dependent step is upstream of the GTP-dependent Ca²⁺-independent step targeted by mastoparan (Fig. 8).

We have shown that isolated islets from Sur1KO mice, lacking β-cell/neuronal-type K_ATP channels, have an impaired insulin secretory response to cAMP. One physiologic consequence of this biochemical defect is that although the cAMP content of Sur1KO islets increases upon stimulation with incretins, the rate of insulin release is not affected, consistent with a defect in cAMP sensing. Our data suggest the cAMP response in wild-type islets is nonlinear; a small increase in [cAMP]_i (<50% above basal values) is sufficient to increase insulin release two- to threefold when the glucose concentration is high. Blocking phosphodiesterase activity with IBMX elevates [cAMP]_i about threefold above basal values, which is sufficient to saturate the potentiating effect of cAMP on glucose-induced insulin release from wild-type islets. Superphysiologic [cAMP]_i, ∼100-fold higher than basal levels, obtained with IBMX plus forskolin, has no stimulatory effect on insulin release from wild-type islets in low glucose, but does strongly potentiate insulin release from Sur1KO islets, which have elevated [Ca²⁺]_i, even in low glucose. The impaired response to incretins does not appear to result from a generalized defect in exocytosis, since insulin release is comparably stimulated from wild-type and Sur1KO islets by activators of PKC pathways and by mastoparan, which is reported to activate exocytosis at a late, GTP-dependent step (47,48).

What biochemical mechanisms are impaired when SUR1 is missing is not known. SUR1 is the regulatory subunit of SUR1/Kir6.2, β-cell/neuronal-type K_ATP channels, but has not generally been considered to interact with other proteins or known to serve a function in exocytosis beyond modulating Ca²⁺ entry via regulation of membrane potential. K_ATP channels are reported to be inhibited by GLP-1 (50). Various mechanisms have been put forward to account for this inhibition, including phosphorylation of SUR1/Kir6.2 subunits by PKA (although phosphorylation by PKA [51] or by PKC [52] is also reported to increase the open probability of K_ATP channels), by direct binding of G-protein βγ subunits to SUR1 (53), by stimulation of glucose transport or metabolism (50), and by increasing the sensitivity of K_ATP channels to inhibitory ATP (54). We think the simple explanation that loss of a potential GLP-1 target, i.e., K_ATP channels, can account for the observed loss of incretin sensitivity is unlikely. If GLP-1 stimulates secretion by inhibiting K_ATP channel activity, then insulin secretion from Sur1KO islets, a model system for constitutively closed K_ATP channels, should exhibit sustained oversecretion equivalent to that stimulated by GLP-1, rather than the moderate, regulated secretion we observed.

How increased cAMP potentiates insulin secretion is not well understood. A number of reports have stressed the importance of PKA-dependent phosphorylation (29–32), but some early reports argued that basal and glucose-induced insulin secretion are PKA independent (28,55). Exocytosis has been reported to be activated directly by cAMP (33,56), and two recent reports (34,36) have described interactions between SUR1, Epac-2/cAMP-GEF II (41,42), Rab3A, and Rim, proteins involved in vesicle trafficking and fusion (57). Kashima et al. (36) report both PKA-dependent and -independent effects of increased [cAMP]_i on glucose-induced secretion from islets treated with antisense oligonucleotides. We observed that two PKA inhibitors, H-89 and KT5720, had no statistically significant effect on stimulation of wild-type insulin secretion by GLP-1 or by IBMX and forskolin. The reason(s) for the different responses in these studies is not clear. A third PKA inhibitor, Rp-cAMPS, did significantly inhibit glucose-induced insulin secretion. Our results are consistent with the possibility that SUR1, or SUR1 assembled with Kir6.2, can interact with a cAMP sensing pathway in β-cells, possibly Epac2/cAMP-GEFII, as suggested by Ozaki et al. (34), that does not use PKA as an effector. The results suggest the impaired reaction is upstream of the step(s) activated by the Ca²⁺-independent GTP-dependent reaction(s) activated by mastoparan.

What contribution the impaired secretory response to cAMP of Sur1KO islets makes to the normoglycemia of the knockout mice versus the profound hypoglycemia of HI neonates lacking functional K_ATP channels is not known. We considered several possibilities. First, that loss of the incretin response in Sur1KO islets disrupts the enteroinsular axis in the animals, reducing insulin secretion and glucose disposal, thus contributing to their normoglycemia. There is little insulin release after intraperitoneal injection of glucose into Kir6.2KO (5) and Sur1KO animals (6), and we anticipate that postprandial insulin levels in Sur1KO mice should be reduced versus wild type. However, although the data are controversial, GLP-1 has been reported to increase glucose uptake in peripheral tissues (compare 58,59 with 60–62) and thus may contribute to glucose disposal in Sur1KO mice. We are not aware of any data on the incretin responsiveness of hyperinsulinemic-hypoglycemic neonates. A second possibility is that there is a basic difference in cAMP sensing in mouse versus human islets, as has been suggested to explain the difference in kinetics of second-phase secretion of mouse versus rat and human islets (63). The available comparative data for knockout mouse versus HI islets are insufficient to address this question. Rates of insulin turnover have not been compared directly, but the insulin content of Sur1KO islets reported here is ∼60% of wild-type values versus ∼2.5% reported for HI islets (4), suggesting a higher turnover rate for the latter. Kaiser et al. (3) have also reported that cells derived from HI islets exhibit a high rate of glucose-independent insulin secretion, which is stimulated further by IBMX and forskolin. Similarly, Straub et al. (4) have shown that secretion from HI β-cells, determined by capacitance measurements, is potentiated by forskolin. However, in these studies, as in our experiments with IBMX and forskolin, [cAMP]_i is increased to superphysiologic levels where insulin secretion is significantly increased in the Sur1KO islets, and it is difficult to extrapolate to a physiologic condition. Finally, it is important to consider that SUR1-regulated K⁺ channels are present in nonpancreatic tissues, where their loss could contribute to glycemic control, for example, through counterregulatory responses that may well differ between rodents and humans.

This work was funded by National Institutes of Health grants DK50750 and DK52771 to J.B. and DK57671 to L.A.-B.S.U. is a recipient of a Heisenberg Fellowship and a travel award (UL140/3-1) from the Deutsche-Forschungs-Gemeinschaft.