J Biol Chem, Vol. 274, Issue 32, 22337-22344, August 6, 1999

The Calcium/Calmodulin-dependent Phosphodiesterase PDE1C Down-regulates Glucose-induced Insulin Secretion^*

Ping Han, John Werber, Manju Surana^§, Norman Fleischer^§, and Tamar Michaeli^¶

From the Departments of  Developmental and Molecular Biology and ^§ Medicine, Albert Einstein College of Medicine, Bronx, New York 10461

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

To understand the role cAMP phosphodiesterases (PDEs) play in the regulation of insulin secretion, we analyzed cyclic nucleotide PDEs of a pancreatic -cell line and used family and isozyme-specific PDE inhibitors to identify the PDEs that counteract glucose-stimulated insulin secretion. We demonstrate the presence of soluble PDE1C, PDE4A and 4D, a cGMP-specific PDE, and of particulate PDE3, activities in TC3 insulinoma cells. Selective inhibition of PDE1C, but not of PDE4, augmented glucose-stimulated insulin secretion in a dose-dependent fashion thus demonstrating that PDE1C is the major PDE counteracting glucose-dependent insulin secretion from TC3 cells. In pancreatic islets, inhibition of both PDE1C and PDE3 augmented glucose-dependent insulin secretion. The PDE1C of TC3 cells is a novel isozyme possessing a K_m of 0.47 µM for cAMP and 0.25 µM for cGMP. The PDE1C isozyme of TC3 cells is sensitive to 8-methoxymethyl isobutylmethylxanthine and zaprinast (IC₅₀ = 7.5 and 4.5 µM, respectively) and resistant to vinpocetine (IC₅₀ > 100 µM). Increased responsiveness of PDE1C activity to calcium/calmodulin is evident upon exposure of cells to glucose. Enhanced cAMP degradation by PDE1C, due to increases in its responsiveness to calcium/calmodulin and in intracellular calcium, constitutes a glucose-dependent feedback mechanism for the control of insulin secretion.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The second messengers cAMP and cGMP mediate diverse physiological responses to hormones, neurotransmitters, and light. Rates of cyclic nucleotide synthesis by cyclases and of their degradation by phosphodiesterases (PDEs)¹ regulate their cellular concentrations (reviewed in Refs. 1 and 2). Cyclic nucleotide PDEs have been distinguished into nine families based on their substrate affinity and specificity, and their selective sensitivity to cofactors and inhibitory drugs. Cyclic nucleotide PDE families are: 1) PDE1, Ca²⁺/calmodulin stimulated PDEs; 2) PDE2, cGMP stimulated PDEs; 3) PDE3, cGMP inhibited PDEs; 4) PDE4, cAMP-specific PDEs; 5) PDE5, cGMP-specific PDEs; 6) PDE6, photoreceptor PDEs; 7) PDE7, higher affinity cAMP specific PDEs; 8) PDE8, cAMP-specific IBMX-resistant PDEs (3-5); and 9) PDE9, cGMP-specific IBMX-resistant PDEs (6, 7). All mammalian PDEs contain a related C-terminal domain with ~30% sequence identity between families, and N-terminal regulatory domains containing cofactor or cGMP-binding sites, localization and other regulatory sequence motifs. Tissue- and cell-specific gene expression, and a variable splicing pattern, all contribute to the unique and complex composition of cyclic nucleotide PDEs in mammalian cells that normally contain activities derived from several families of PDEs (1, 2). PDE inhibitors that do not affect adenosine uptake and exhibit high selectivity between PDE families, and in some cases between PDE isozymes, are powerful tools for identification of PDEs involved in diverse physiological responses (8-10).

Insulin secretion from pancreatic -cells is governed by the interplay between nutritional secretagogues and regulatory hormonal or neural stimuli (11-13). Glucose, the major insulin secretagogue, triggers insulin release, in part, through calcium-dependent vesicular exocytosis (11, 12, 14-17). The glucose signal cascade leads to membrane depolarization and a calcium influx via the opening of L-type voltage-sensitive calcium channels, and also to other effects that include release of calcium from intracellular stores (13, 18, 19). Hormones and insulinotropic gut factors that stimulate cAMP synthesis strongly augment glucose-induced insulin secretion (11-14). Conversely, hormonal inhibition of insulin secretion involves reductions in cAMP levels (10, 20-23). In addition to the role cAMP plays in hormonal modulation of insulin secretion, basal cAMP levels appear to be required for glucose to induce insulin secretion (24). Potentiation of glucose-induced insulin secretion is evident not only upon treatment of insulin secreting cells with the insulinotropic gut factor GLP1 (glucagon-like peptide 1), but also upon treatment with reagents that stimulate cAMP signaling including membrane permeable cAMP analogs, activators of adenylyl cyclase, and PDE inhibitors (11, 20, 25, 26). Like GLP1, these cAMP elevating agents do not induce significant insulin secretion in the absence of glucose. Targets for cAMP action are PKA substrates such as the voltage-sensitive calcium channel, GLUT2, and potentially also ion channels to which cAMP binds directly (13, 27-30). In addition to the potentiation of glucose and calcium-dependent insulin secretion, cAMP-stimulated exocytosis via calcium-independent mechanisms is evident in patch clamped cells, and the contribution of this mechanism to insulin secretion under physiological conditions remains to be determined (29). A requirement for the localization of PKA to specific sites within pancreatic -cells via anchor proteins has been demonstrated for GLP-1 potentiation of insulin secretion (31).

The involvement of cyclic nucleotide PDEs in the regulation of insulin secretion is inferred from the stimulatory effects of the non-selective PDE inhibitor isobutylmethylxanthine (IBMX) on insulin secretion from insulin secreting cell lines, from islets, and from transgenic mice expressing a constitutively activated G_s mutant in their pancreatic -cells (11, 20, 21, 25). Cyclic nucleotide PDEs present in -cells were thus far investigated as total PDE activities of crude islet extracts and the presence of PDEs 3 and 4, and calcium-sensitive PDEs, in -cells has been inferred from these studies (32-36). The involvement of PDE3 in glucose-induced insulin secretion from pancreatic islets has been demonstrated in studies using selective PDE3 inhibitors (32, 33, 37, 38). The presence of PDE3B in pancreatic -cells and its involvement in insulin-like growth factor-1 and in leptin-mediated inhibition of insulin secretion has been demonstrated recently (10, 39). However, in cultured pancreatic -cells PDE3B does not appear to play a role in insulin secretion induced by glucose in the absence of hormone regulation (10).

This study details the chromatographic, biochemical, and pharmacological characterization of -cell PDEs, and the identification of a subset of -cell PDEs that are involved in counteracting glucose-induced insulin secretion. We demonstrate the presence of PDEs 1C, 4A, and 4D, and a cGMP-specific PDE, in soluble cell lysates, and of PDE3 in particulate cell lysates, of TC3 insulinoma cells. We also demonstrate the involvement of PDE1C in the regulation of glucose-dependent insulin secretion from TC3 cells and pancreatic islets.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Preparation of Cell Extracts and Mono Q-FPLC Fractionation-- Cells were washed twice with cold phosphate-buffered saline, scraped, and homogenized in a buffer containing 50 mM Tris, pH 7.5, 250 mM sucrose, 5 mM MgCl₂, 0.2 µg/ml aprotinin, leupeptin, and pepstatin, and 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride. Following a 30-min 150,000 × g centrifugation, the supernatant was loaded onto a Pharmacia Mono-Q anion exchange column and PDE activities were fractionated by FPLC along a two-step salt gradient. Column buffers were: A) 50 mM Tris, pH 7.5, and B) 50 mM Tris, pH 7.5, 0.5 M NaCl. The gradient consisted of a 10-min increase to 0.1 M NaCl (20% B), and a 90-min increase to 0.5 M NaCl (100% B). One-ml fractions were collected.

PDE Assays and Kinetic Measurements-- PDE assays were performed in duplicate as described (40). Briefly, 0.5-1 µg of protein extract, or 10 µl of Mono-Q fractions, were incubated for 15 min at 30 °C with appropriate concentrations of either 2,8-[³H]cAMP or 8-[³H]cGMP in a buffer containing 40 mM Tris, pH 7.5, 10 mM MgCl₂, and 0.1 unit of snake venome nucleotidase. Reactions were terminated by the addition of EDTA, AMP, and cAMP to final concentrations of 10, 1.25, and 1.25 mM, respectively, and transfer to 4 °C. Reaction mixtures were applied to AG-1-X8 anion exchange columns (Bio-Rad) and the hydrolyzed products were eluted with 50% ethanol. The ³H content of the eluate was determined using a liquid scintillation counter. Analysis and plotting of kinetic data was performed by using the computer program Kaleidagraph. The contributions of the high and low affinities cGMP PDEs of peak I to total PDE activity of this peak at a given substrate concentration were calculated using the formula: V = V₁ + V₂ = V_m1·S/(K_m1 + S) + V_m2·s(K_m1 + S).

Cell Culture, Treatments, and Measurements of Insulin Secretion-- TC3 cells were cultured as described and early passages (22-40) were maintained (20). For measurements of insulin secretion cells were plated onto 24-well culture dishes 3-4 days, and refed 16 h prior to the assay. On the experiment day, cells were washed in Hepes-buffered Krebs-Ringer solution, and incubated in this glucose-free solution for 1 h. Subsequently, either no glucose or 16.7 mM glucose and, when relevant, PDE inhibitors were added to the cells in Hepes-buffered Krebs-Ringer solution and the cells were incubated for 2 additional hours. Each condition was assayed in triplicate. Secreted insulin found in the supernatant after centrifugation, and endogenous insulin content of the cells after acid extraction, were determined by a radioimmunoassay as described (20). Average insulin content was 2% of total cellular protein.

PDE inhibitors used are: IBMX, a non-selective inhibitor; 8-methoxymethyl-isobutylmethylxanthine (8MM-IBMX), a PDE1 selective inhibitor; zaprinast, a PDE1/5/6 selective inhibitor; rolipram and RO20-1724, PDE4 selective inhibitors; milrinone and trequinsin, PDE3 selective inhibitors (1, 41, 42). All inhibitors applied to TC3 cells in glucose-free Hepes-buffered Krebs-Ringer solution did not induce significant insulin release.

Islet Preparations, Treatments, and Measurements of Insulin Secretion-- Mouse pancreatic islets were prepared from normal C3H male mice via collagenase digestion of pancreatic tissue, separation on a Ficoll gradient and manual picking (43). Groups of 10 islets were first preincubated in Hepes-buffered Krebs-Ringer solution containing 1.67 mM glucose for 1 h and subsequently for 2 h in the presence of either 1.67 mM glucose alone or 16.7 mM glucose along with PDE inhibitors. Following the incubation with glucose and various PDE inhibitors, insulin released to the incubation media, and islet insulin content were measured by a radioimmunoassay.

Cyclic AMP Measurements-- Cells were washed twice with cold phosphate-buffered saline, scraped after the addition of butanol-saturated 1 M formic acid and incubated on ice for 10 min. After a 10-min spin at 10,000 × g, the supernatant was harvested, dried by vacuum centrifugation, and resuspended in water. Cyclic AMP content was measured by a radioimmunoassay after acetylation in the assay buffer (50 mM sodium acetate, pH 4.75). Samples and standards were incubated overnight with ¹²⁵I-cAMP (adenosine 3',5'-cyclic phosphoric acid, 2'-O-succinyl-[¹²⁵I]iodotyrosine methyl ester, NEN Life Science Products Inc.), goat anti-cAMP antibodies and rabbit -globulins. Following ammonium sulfate precipitation, the antibody-¹²⁵I-cAMP immune complexes were counted in a -counter.

Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) Analysis-- RT-PCR analysis was performed on 5 µg of RNA prepared from TC3 cells using Trizol (Life Technologies, Inc.). Controls lacking reverse transcriptase were included in the reactions. To determine expression of PDE1C the following oligonucleotides were used: for RT, oligo-dT; for PCR amplification, JWPDE1C-5 5'-ACAGGGCAGAGGAGATCAAGTTT; and JWPDE1C-3, 5'-CTTTTCGCCTGCCTTTTCTCCTT. The 408-base pair PCR product was cloned and its DNA sequence was determined to be identical to the published mouse PDE1C sequence.

The following oligonucleotides were used for PCR amplification to determine the expression of PDE4A, JWPDE4A-5, 5'-AGCCATGGAACAGTCAAAGGTCAA; and JWPDE4A-3, 5'-TCAGGAGGGCCAGGAGTCGT. To determine the expression of PDE4D, JWPDE4D-5, 5'-GAGGGCCGGCAGGGACAGAC; and JWPDE4D-3, 5'-GGGGGTGGGGTGGGTGAGAGG. Amplification products 436 and 470 base pairs long were obtained for PDE4A and D, respectively.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cyclic Nucleotide Phosphodiesterases of TC3 Cells-- To identify -cell PDEs involved in the regulation of insulin secretion, we first set to identify the cyclic nucleotide PDE families and isozymes expressed in TC3 insulinoma cells. For these purposes, we fractionated soluble TC3 cell extracts by Mono-Q FPLC anion exchange chromatography and analyzed their cyclic nucleotide PDE activities profile (Fig. 1). The PDE profile was comprised of four peaks of PDE activities: a single peak (peak I) containing both cAMP and cGMP PDE activities, and three cAMP-specific PDE activity peaks (peaks II-IV). Both cAMP and cGMP PDE activities of peak I were stimulated by calcium and calmodulin at low substrate concentrations, demonstrating the presence of a high affinity isozyme of the dual specificity, calcium/calmodulin-dependent PDE1 in this peak (Table I). However, while the cAMP PDE activity of peak I was highly sensitive to the PDE1 selective inhibitor 8MM-IBMX, the cGMP PDE activity of peak I was partially resistant to 8MM-IBMX, raising the possibility that additional cGMP PDE activities are present in this peak. PDE activities of peaks II-IV were abolished by 2.5 µM rolipram, a highly selective PDE4 inhibitor, thus demonstrating the presence of multiple PDE4 isozymes in TC3 cells. Expression of both PDE4A and 4D in these cells was demonstrated by RT-PCR analysis (see "Materials and Methods"). Milrinone-sensitive PDE3 activity constituted the majority of the particulate PDE activity (>70%, not shown). These observations demonstrate the presence of soluble PDE1 and -4 isozymes, of a soluble cGMP-specific PDE, and of a particulate PDE3, in TC3 cells.

View larger version (23K): [in this window] [in a new window]   Fig. 1.   Fractionation of soluble cyclic nucleotide PDE activities of TC3 cells. Ten mg of soluble extract of TC3 cells were fractionated by Mono-Q FPLC. The elution profile of cAMP PDE activities (large circles) and cGMP PDE activities (small circles) assayed at 1 µM substrate is presented. The 0-0.5 M NaCl gradient is depicted. Roman numbers indicate PDE peak numbers.

View this table: [in this window] [in a new window]   Table I Biochemical and pharmacologic properties of fractionated PDE activities of TC3 cells

Kinetic and Pharmacologic Analyzes of PDE Activities of Peak I-- To identify PDE1 isozymes and other PDE activities of peak I, we examined the kinetic and pharmacological properties of PDEs present in peak I (Fig. 2, Table II). Kinetics of cAMP PDE activities of peak I demonstrated the presence of a single high affinity cAMP PDE activity possessing a K_m of 0.47 µM for cAMP (Fig. 2A). The kinetics of cGMP PDE activities of peak I demonstrated the presence of two cGMP PDE activities (Fig. 2B). Two kinetic curves of cGMP PDE activities resolved as the best fit for the obtained data included activities possessing K_m of 0.25 µM and of 57.5 µM for cGMP. Based on these kinetic parameters, it is calculated that 95% (at 0.1 µM) and 50% (at 10 µM) of the cGMP PDE activity of peak I are derived from the high affinity PDE of this peak (V = V₁ + V₂, see "Materials and Methods"). Accordingly, 8MM-IBMX inhibited 82% of the cGMP PDE activity of peak I at 0.1 µM substrate, and only 57% of the cGMP PDE activity of peak I at 10 µM substrate. As 8MM-IBMX inhibition reflected the calculated content of the high affinity cGMP PDE activity of peak I, it appears that the low affinity cGMP PDE activity of peak I is 8MM-IBMX resistant and is not derived from the PDE1 isozyme present in this peak (Table I). Taken together, these observations indicate that peak I contains a high-affinity dual specificity PDE1 activity that is stimulatable by calcium/calmodulin and is sensitive to 8MM-IBMX, and a low affinity cGMP PDE activity.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Kinetic analysis of PDE activities of peak I. Double-reciprocal Lineweaver-Burk plots and Scatchard plots (inset) derived from kinetic curves are shown. Cyclic AMP and cGMP PDE assays were performed as described under "Materials and Methods." Each data point represents measurements of initial rates at a suitable enzyme dilution. A plot of a typical measurement out of three determinations is depicted. A, cyclic AMP concentrations ranging from 2 [times ]108-105 M were used to determine the kinetic curve. The calculated Km is 0.47 µM cAMP, and the calculated Vmax is 120 pmol/min·mg. B, cyclic GMP concentrations ranging from 107 to 2 × 105 M were used to determine the kinetic curve. Two independent kinetic curves were derived as the best fit for the measured data using the computer program Kaleidagraph. The derived Km values for the two cGMP PDE activities are: 0.25 and 57.5 µM cGMP, and the derived Vmax values are 60 and 400 pmol/min/mg, respectively.

Among the three known PDE1 genes, PDE1A-C, PDE1C encoded isozymes possess high affinity cAMP and cGMP PDE activities with K_m values in the range of peak I, while PDE1A and PDE1B encoded isozymes possess low affinity cAMP PDE activity with K_m values ranging from 25 to 120 µM (1, 44, 45). RT-PCR analysis with oligonucleotides derived from a region common to all five known PDE1C splice variants, and determination of the DNA sequence of the amplified fragment, demonstrated the presence of PDE1C mRNA in TC3 cells (see "Materials and Methods"). The PDE1 activity of peak I is sensitive to zaprinast and resistant to vinpocetine (Table II) as is the case for PDE1C but not for PDE1A and -1B (44, 45). Thus, based on the kinetic and pharmacological properties of the cAMP PDE and of the high affinity cGMP PDE activities of peak I, and on RT-PCR analysis, the calcium/calmodulin sensitive activity present in peak I is derived from PDE1C.

Cyclic Nucleotide PDEs That Counteract Glucose-induced Insulin Secretion from TC3 Cells-- To identify PDEs involved in counteracting glucose-induced insulin secretion we used membrane-permeable, family-selective PDE inhibitors. These inhibitors do not affect adenosine uptake (1, 41, 42, 46). As inhibitor IC₅₀ values differ among isozymes and to minimize nonspecific effects of high inhibitor concentrations, we determined inhibitor IC₅₀ values for TC3 PDE isozymes and applied the inhibitors to the cells at concentrations 10-100-fold above the established IC₅₀ values. IC₅₀ values for peak I (PDE1C) are presented in Table II, and for the PDE4 selective inhibitor, rolipram are: 50 nM (peak II) and 1.5 nM (peak IV). Insulin secretion was measured by a radioimmunoassay following a 2-h incubation in the presence of glucose and family-selective PDE inhibitors (Fig. 3). Controls included the addition of each inhibitor and of the solvent, dimethyl sulfoxide, in the absence of glucose (not shown). These analyses demonstrated that the PDE1 selective inhibitor 8MM-IBMX strongly augmented insulin secretion in the presence of glucose. Inhibition of PDE1 by 8MM-IBMX exhibited a dose-dependent augmentation of insulin secretion that was significant even at 20-fold excess over its IC₅₀ value. In this respect the effects of this PDE1-selective inhibitor were equivalent to those of the non-selective inhibitor IBMX whose IC₅₀ values for the involved PDE families are in the low micromolar range (1, 41, 47). An additional PDE1 inhibitor, zaprinast, when applied at 0.5 and 1 mM, also had stimulatory effects on insulin secretion equivalent to 64 and 84%, respectively, of the effects of IBMX. Thus, it appears that PDE1 inhibits glucose-dependent insulin secretion from TC3 cells.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Stimulation of insulin release from TC3 cells by glucose and cyclic nucleotide PDE inhibitors. Insulin release was measured by a radioimmunoassay after a 2-h incubation with 16.7 mM glucose and various cyclic nucleotide PDE inhibitors at the depicted concentrations. Control incubations with inhibitors in the absence of glucose did not result in significant insulin secretion (see "Materials and Methods"). Error bars represent S.E. Differences between all measurements and glucose-stimulated insulin secretion were statistically significant (p < 0.01). * indicates p < 0.005 in comparison to 150 µM 8MM-IBMX. Assays were performed in triplicate. Experiments were performed 7 times for no glucose, glucose, glucose + IBMX, and glucose + 8MM-IBMX; 4-7 times for glucose + rolipram; and 4 times for glucose + milrinone. IC50 values for the tested inhibitors are: 7.5 µM 8MM-IBMX for inhibition of PDE1C; 1.5 and 50 nM rolipram for inhibition of PDE4 isozymes; and 0.3 µM milrinone for inhibition of PDE3 (41). Average insulin content was 2% of total cellular protein content.

Inhibition of PDE4 by the selective inhibitor rolipram had partial stimulatory effects on glucose-dependent insulin secretion. However, in contrast to the dose-dependent effects of the PDE1 inhibitor 8MM-IBMX, the effects of rolipram were not stimulated by a 10³-fold increase in its concentrations. Maximal rolipram concentrations reached a 300-fold inhibitor excess over the high IC₅₀ value we measured for rolipram and the PDE4 isozyme present in peak II (50 nM). Thus, the effects of rolipram on insulin secretion appear limited and restricted to the highly rolipram-sensitive PDE4 isozyme present in peak IV (IC₅₀ = 1.5 nM). As observed in cultured -cells, inhibition of PDE3 with the selective inhibitor milrinone did not augment glucose-induced insulin secretion from TC3 cells (10).

As a measure for the effectiveness and selectivity of the inhibitors, we assessed the intracellular cAMP concentrations 30 min following the exposure of TC3 cells to glucose and PDE inhibitors (Table III). Intracellular cAMP concentrations were stimulated more than 12-fold by the non-selective inhibitor IBMX. The PDE1 and PDE4 selective inhibitors, 8MM-IBMX and rolipram, respectively, stimulated intracellular cAMP concentrations 2-3-fold. As their stimulatory effects on cAMP content are limited in comparison to those of IBMX, 8MM-IBMX and rolipram appear to inhibit a subset of TC3 cell PDEs. In agreement with selective inhibition of a subset of TC3 cell PDEs, the combination of the three selective PDE inhibitors stimulated intracellular cAMP to the levels stimulated by IBMX (not shown). Milrinone did not stimulate cAMP concentrations significantly, and consequently, the contribution of PDE3 to the regulation of insulin secretion from TC3 cells was not assessed. Thus, it appears that although the inhibition of both PDE1 and PDE4 lead to similar increases in intracellular cAMP concentrations, only the inhibition of PDE1 lead to significant increases in insulin secretion. The selective effects of PDE1 inhibition on insulin secretion were confirmed by use of additional family-selective inhibitors for PDE1 (zaprinast), PDE3 (trequinsin), and PDE4 (RO20-1724). Taken together these observations indicate that PDE1, but not PDE4, is an effective inhibitor of glucose-dependent insulin secretion from TC3 cells.

View this table: [in this window] [in a new window]   Table III Glucose and PDE inhibitor effects on intracellular cAMP content in TC3 cells

Cyclic Nucleotide PDEs That Counteract Glucose-induced Insulin Secretion from Pancreatic Islets-- To identify PDEs involved in counteracting insulin secretion from pancreatic islets, we examined the effects of family selective PDE inhibitors on insulin secretion from mouse pancreatic islets. Islet extracts contained 8MM-IBMX-sensitive PDE1 activity, rolipram-sensitive PDE4 activity, and milrinone-sensitive PDE3 activity, and selective inhibition of these PDEs was used to assess their effects on insulin secretion from islets (Fig. 4). For these purposes, islets were first incubated in low glucose, and subsequently the insulin secreted during a 2-h incubation in the presence of glucose and family-selective PDE inhibitors was measured. In islets, the PDE1 inhibitor 8MM-IBMX and the PDE3 inhibitor milrinone had similar stimulatory effects on glucose-induced insulin secretion. The PDE4 inhibitor rolipram, however, did not elicit a significant increase in insulin secretion. A combination of all three selective inhibitors had strong synergistic effects on insulin secretion demonstrating the additive effects of selective inhibition of PDEs 1, 3, and 4. Thus, it appears that PDE1 and PDE3, but not PDE4, inhibit glucose-dependent insulin secretion from islets. These observations suggest that islets and TC3 cells differ in their responses to PDE3 inhibitors, and that in both systems PDE4 does not affect, while PDE1 inhibits, glucose-stimulated insulin secretion.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Stimulation of insulin release from pancreatic islets by glucose and cyclic nucleotide PDE inhibitors. Groups of 10 islets were incubated with 1.67 mM glucose for 1 h prior to the addition of low or high glucose and the tested PDE inhibitors. Insulin release was measured by a radioimmunoassay after a 2-h incubation with 16.7 mM glucose and the indicated PDE inhibitors at the depicted concentrations. Experiments were performed two times in triplicate. Error bars represent S.E. Differences between all measurements and glucose-stimulated insulin secretion were statistically significant (p < 0.05).

PDE1 Activity Is Up-regulated by Glucose-- PDE1 activity is regulated by intracellular calcium levels and by phosphorylation (1). It was therefore of interest to determine whether feedback regulation of glucose-induced insulin secretion involves the stimulation of PDE1C activity by glucose. To determine whether PDE1C activity is elevated upon glucose feeding, we compared the high affinity soluble cGMP PDE activity of glucose fed cells to that of glucose-starved cells (Table IV). As soluble TC3 cell extracts contain several high affinity cAMP PDE activities of PDE1C and PDE4 isozymes and a single high affinity cGMP PDE activity of PDE1C, PDE1C activity assays were performed with cGMP substrate at concentrations calculated to be composed of >95% of the high affinity activity of PDE1C (0.1 µM). This analysis demonstrated that the PDE1C activity of glucose-starved cells was consistently less responsive to the calcium/calmodulin provided in the assay mixture in comparison to the PDE1C activity of glucose-fed cells. Thus, the calcium/calmodulin stimulatable activity of PDE1C appeared to be elevated upon exposure to glucose. The elevations in stimulatable PDE1C activity detected in protein extracts of glucose-fed cells may reflect increased PDE1C expression and/or post-translational modifications that sensitize its response to calcium/calmodulin. In vivo, elevations in intracellular calcium induced by glucose, as well as increased responsiveness to calcium/calmodulin, simultaneously enhance PDE1C activity upon exposure of cells to glucose.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The study details the characterization of cyclic nucleotide PDEs of TC3 cells and the identification of PDE1C as a regulator of glucose-induced insulin secretion. Chromatographic fractionation, coupled with biochemical and pharmacological characterization, and with RT-PCR analysis, established the presence of PDEs 1C, 4A, 4D, and of a cGMP-specific PDE, in soluble extracts of TC3 cells. By use of family selective PDE inhibitors in concentrations that do not exceed a 100-fold excess over IC₅₀ values we determined for the isozymes expressed in TC3 cells, we observe the involvement of PDE1C, and the limited involvement of PDE4 isozymes, in the regulation of glucose-stimulated insulin secretion from TC3 cells and pancreatic islets. As demonstrated previously, PDE3 inhibits glucose-stimulated insulin secretion from pancreatic islets (32, 33). PDE1C activity is elevated upon exposure of cells to glucose, constituting a feedback control mechanism of glucose-induced insulin secretion via increased cAMP degradation by PDE1C.

Fractionation of cellular PDE activities permits the identification of active PDEs and provides an estimate of their relative contribution to cellular PDE activity. While PDE isozymes can be identified by molecular approaches such as RT-PCR, fractionation of cellular PDE activities allows the determination of kinetic and pharmacological properties of the specific isozymes expressed in the cells. The fractionation of TC3 cell PDEs we undertook allowed the identification not only of TC3 cell PDE isozymes but also of isozyme-effective inhibitors and the applicable range of concentrations for specific inhibition of a given PDE isozyme within the cell. The fractionation of TC3 cell PDEs proved critical for detecting the involvement of PDE1C in the regulation of insulin secretion, as its inhibitor profile identified relevant and effective inhibitors that were not examined in pancreatic -cells (32, 33). Previous studies in pancreatic islets suggest the involvement of PDE3, but not of PDE4, in the regulation of insulin secretion (32, 33). However, perhaps due to differences in its abundance or to its regulation in islets and in cultured pancreatic -cells, PDE3B does not appear to counteract glucose-induced insulin secretion but to mediate the inhibitory effects of insulin-like growth factor-1 and leptin on insulin secretion in cultured - and insulinoma cells (10, 39). Thus, our analysis is in agreement with the currently held notion that in cultured pancreatic -cells PDE3 and -4 are not major PDEs that counteract glucose-induced insulin secretion. Our analysis, however, identifies PDE1C as a PDE that counteracts glucose-induced insulin secretion both in TC3 cells and in mouse pancreatic islets.

Unlike the limited effects of PDE4 inhibitors, inhibition of PDE1C exhibits a dose-dependent stimulation of insulin secretion that can account for the stimulatory effects of non-selective inhibition of PDEs in TC3 cells. In pancreatic islets, PDE1C inhibition has strong stimulatory effects on insulin secretion though not equivalent to effects of non-selective PDE inhibition. The combined inhibition of PDEs 1, 3, and 4 was as potent as non-selective PDE inhibition, suggesting that both PDE1 and PDE3 inhibit glucose-induced insulin secretion from islets. Differences in PDE abundance, or in regulation patterns between pancreatic islets and cultured -cells, may account for the quantitative differences of PDE1C effects on insulin secretion. The involvement of PDE1C in the regulation of insulin secretion, and the stimulation of its activity by glucose, suggest the existence of glucose-dependent feedback control loop of insulin secretion (Fig. 5). Exposure of pancreatic -cells to glucose leads to increases in intracellular calcium concentrations and in vesicular exocytosis of insulin. Calcium has been proposed to activate the calcium/calmodulin-dependent adenylyl cyclase of -cells, a process that can potentiate insulin secretion. However, calcium also stimulates the calcium/calmodulin-dependent PDE1C. While threshold calcium levels required for the activation of adenylyl cyclase and PDE1C are not known, we demonstrate in this study that the responsiveness of PDE1C to calcium is stimulated by glucose. In the model depicted in Fig. 5 we propose that the glucose and calcium-dependent stimulation of PDE1C leads to reductions in intracellular cAMP concentrations that limit insulin secretion, thus establishing a feedback mechanism that down-regulates glucose-induced insulin secretion (48, 49).

View larger version (11K): [in this window] [in a new window]   Fig. 5.   Model of glucose- and calcium-dependent feedback regulation of cAMP in TC3 cells. Glucose triggers signaling cascades that lead to the activation of the voltage-sensitive calcium channel. Cyclic AMP augments the glucose-induced calcium influx and the exocytosis of vesicular insulin. PDE activity of the calcium/calmodulin-dependent PDE1C is stimulated both by elevations in calcium and by additional glucose-induced increases in its responsiveness to calcium/calmodulin. Increased PDE1C activity reduces intracellular cAMP concentrations and limits insulin secretion.

The PDE1C of TC3 cells appears to be a novel isozyme distinguished by its predominant cAMP PDE activity and inhibitor sensitivity profile. Five different splice variants of PDE1C have been identified thus far, all sharing relatively high affinity and equipotent cAMP and cGMP PDE activities (44, 45, 50). These PDE1C isozymes exhibit differential sensitivity to several PDE inhibitors. Provided the tissue distribution of the PDE1C splice variant expressed in pancreatic -cells is limited, it may prove to be a novel drug target for intervention with the progression of non-insulin-dependent diabetes melittus. As demonstrated in transgenic mice expressing a constitutively active G_s mutant in their pancreatic -cells, inhibition of -cell PDE1C will be particularly powerful in conjunction with cAMP elevating agents such as GLP1 (11, 20, 21, 25). Efforts to clone cDNAs encoding the PDE1C isozyme of pancreatic -cells and determine its tissue distribution pattern are underway.

Our analysis indicates that TC3 cells contain multiple high affinity cAMP PDEs, that are both cAMP-specific (PDE4) and dual specificity PDEs (PDE1C and 3), and a low affinity cGMP-specific PDE. Among these cAMP PDEs, PDE1C appears to counteract glucose-induced insulin secretion effectively as its selective inhibition leads to a dose-dependent augmentation of insulin secretion equivalent to the one observed upon use of the non-selective PDE inhibitor IBMX. In the absence of the intricate paracrine environment of pancreatic islets, selective inhibition of PDE3B in cultured pancreatic -cells by milrinone does not augment glucose-induced insulin secretion but counteracts insulin-like growth factor-1 and leptin inhibition of insulin secretion (10, 39). Thus, while PDE3B mediates hormone-dependent inhibition of insulin secretion in insulinomas, PDE1C appears to mediate feedback regulation glucose-dependent insulin secretion in the absence of added extracellular hormonal and neural signals. Thus, specific PDEs among the multiple PDEs present in TC3 cells appear to play different and specialized roles in -cell physiology and regulate insulin secretion under different physiological conditions.

	ACKNOWLEDGEMENTS

We thank S. Efrat and C. Rubin for useful discussions and support of this work, G. Orr, R. Stanley, Y. G. Yeung for assistance with FPLC and Drs. Harper and Brooker for anti-cAMP antibodies.

	FOOTNOTES

^* This work was supported in part by Diabetes Research and Training Center Grant DK20541 and an American Diabetes Association Career Development Award (to T. M.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^¶ To whom reprint requests should be addressed. Tel.: 718-430-2138; Fax: 718-430-8696; E-mail: michaeli@aecom.yu.edu.

	ABBREVIATIONS

The abbreviations used are: PDE, phosphodiesterase; IBMX, isobutylmethylxanthine; FPLC, fast protein liquid chromatography; RT-PCR, reverse transcriptase-polymerase chain reaction; 8MM-IBMX, 8-methoxymethyl isobutylmethylxanthine.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES