Monomethylated-adenines potentiate glucose-induced insulin production and secretion via inhibition of phosphodiesterase activity in rat pancreatic islets

Abstract

Monomethyladenines have effects on DNA repair, G-protein-coupled receptor antagonism and autophagy. In islet ß-cells, 3-methyladenine (3-MA) has been implicated in DNA-repair and autophagy, but its mechanism of action is unclear. Here, the effect of monomethylated adenines was examined in rat islets. 3-MA, N6-methyladenine (N6-MA) and 9-methyladenine (9-MA), but not 1- or 7-monomethylated adenines, specifically potentiated glucose-induced insulin secretion (3-4 fold; p ≤ 0.05) and proinsulin biosynthesis (∼2-fold; p ≤ 0.05). Using 3-MA as a ‘model’ monomethyladenine, it was found that 3-MA augmented [cAMP]_i accumulation (2-3 fold; p ≤ 0.05) in islets within 5 minutes. The 3-, N6- and 9-MA also enhanced glucose-induced phosphorylation of the cAMP/protein kinase-A (PKA) substrate cAMP-response element binding protein (CREB). Treatment of islets with pertussis or cholera toxin indicated 3-MA mediated elevation of [cAMP]_i was not mediated via G-protein-coupled receptors. Also, 3-MA did not compete with 9-cyclopentyladenine (9-CPA) for adenylate cyclase inhibition, but did for the pan-inhibitor of phosphodiesterase (PDE), 3-isobutyl-1-methylxanthine (IBMX). Competitive inhibition experiments with PDE-isoform specific inhibitors suggested 3-MA to have a preference for PDE4 in islet ß-cells, but this was likely reflective of PDE4 being the most abundant PDE isoform in ß-cells. In vitro enzyme assays indicated that 3-, N6- and 9-MA were capable of inhibiting most PDE isoforms found in ß-cells. Thus, in addition to known inhibition of phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3′K)/m Target of Rapamycin (mTOR) signaling, 3-MA also acts as a pan-phosphodiesterase inhibitor in pancreatic ß-cells to elevate [cAMP]_i and then potentiate glucose-induced insulin secretion and production in parallel.

Abbreviations

3-MA

3-methyladenine

N6-MA

N6-methyladenine

9-MA

9-methyladenine

1-MA

1-methyladenine

7-MA

7-methyladenine

PKA

protein kinase-A

CREB

cAMP-response element binding protein

9-CPA

9-cylopentyladenine

PDE

phosphodiesterase

IBMX

3-isobutyl-1-methylxanthine

GPR-40

G-protein coupled receptor-40

PLC

phospholipase-C

DAG

diacylglycerol

PKC

protein kinase-C

GLP-1

glucagon-like peptide-1

GIP

gastric inhibitory peptide

MMPX

8-Methoxymethyl-3-isobutyl-1-methyl xanthine

EHNA

Erythro-9-(2-hydroxy-3-nonyl) adenine

KRBH

Krebs-Ringer bicarbonate buffer

PMSF

phenylmethylsulfonyl fluoride

TLCK

Tosyl-lysyl chloromethylketone

RIA

radioimmunoassay

BCA

bicinchoninic acid

LC3

Microtubule-associated protein-1 light chain-3

mTOR

mammalian Target of Rapamycin

PI3′K

phosphatidylinositol-4,5-bisphosphate 3-kinase

PDK-1

Phosphoinositide dependent kinase-1

AMPK

AMP-activated protein kinase.

Introduction

The onset of common type 2 diabetes is marked by failure of the functional pancreatic ß-cell mass to meet the demand for insulin.¹ Under normal circumstances, the pancreatic ß-cell has an optimal intracellular storage pool of insulin mostly maintained by a balance between insulin production and insulin secretion. When insulin secretion is stimulated, as in response to elevated glucose concentration, there is a synchronized upregulation of proinsulin biosynthesis. This serves to rapidly restock the depletion of insulin stores lost from the ß-cell via exocytosis.² This upregulation of proinsulin biosynthesis is specifically mediated at the translational level and uniquely directed at a minor subset of ß-cell proteins, all of which are localized to insulin secretory granules (also known as ß-granules) including the proinsulin processing endopeptidases. This coordinated translational upregulation of ß–granule proteins is the major control of ß-granule biogenesis and ensures that an increased demand for proinsulin processing is achieved.^2,3 Not all ß-granules undergo exocytosis, and if retained in the ß-cell for ∼5 d are targeted for internal degradation by autophagy.² Normally, this autophagic deposal of ß-granules plays an additional janitorial role in maintaining optimal insulin secretory stores in the ß-cell.

Glucose metabolism is required for both glucose-induced insulin production and secretion.² However, the mechanisms behind the translational control of proinsulin biosynthesis and regulated insulin secretion are quite distinct.² For insulin secretion, an increase in the ATP:ADP ratio as a result of increased mitochondrial oxidative metabolism closes ATP sensitive K⁺-channels in the plasma membrane causing decreased cellular K⁺ efflux, depolarization of the ß-cell, and influx of Ca²⁺ via voltage sensitive L-type Ca²⁺-channels.⁴ The subsequent rise in cytosolic [Ca²⁺]_i is a required trigger for insulin exocytosis.⁴ However, unlike regulated insulin secretion, glucose-induced proinsulin biosynthesis is Ca²⁺-independent and regulated by alternative metabolic secondary signals emanating from ß-cell mitochondria.⁵ Although ß-cell autophagy can be influenced by glucose levels, the precise mechanism behind this control is unknown.⁶ When proinsulin biosynthesis and insulin secretion are upregulated, autophagic degradation of insulin stores is generally decreased,^2,7 but also by undetermined mechanisms.

Other distinctions between the regulation of proinsulin biosynthesis and insulin secretion arise in considering the effect of other nutrients. Both exogenous fatty acids and products of endogenous ß–cell nutrient metabolism, particularly malonyl-CoA and long chain fatty acyl-CoA, generate secondary signals that potentiate glucose-induced insulin secretion.⁸ Exogenous fatty acids can also bind a G_q-linked receptor, GPR-40,⁹ to activate PKC and mobilize [Ca²⁺]_i from intracellular stores that then trigger insulin secretion. In contrast, fatty acids do not stimulate proinsulin biosynthesis, and under conditions of chronic hyperlipidemia this results in depletion of intracellular stores of insulin in the ß-cell.¹⁰ Neurotransmitters also influence insulin secretion. Acetylcholine, via parasympathetic innervation to pancreatic islets and G_q-linked muscarinic-M₃ receptors on the ß-cells, potentiates glucose-induced insulin secretion.^11,12 But acetylcholine, like fatty acids, does not control insulin production in ß-cells. In contrast, neurotransmitters operating via sympathetic neurons, such as epinephrine, markedly inhibit both glucose-induced insulin secretion and production.^11,13 There are other common regulators of both insulin production and secretion. For example, incretin hormones, such as glucagon-like peptide-1 (GLP-1) and gastric inhibitory peptide (GIP), via their G_αs-protein coupled receptors, as well as other reagents that instigate cAMP-signaling in ß-cells, potentiate both glucose-induced insulin secretion and production in parallel.¹⁴

Pharmacological augmentation of insulin secretion and ß-cell function can prove to be an effective treatment for type 2 diabetes. For example, GLP-1 analogs, in potentiating glucose-induced insulin production and secretion in parallel, assist the ß-cell in compensating for insulin resistance and better uphold insulin secretory capacity.^14,15 However, other pharmacological agents that also target the ß-cell, such as sulfonylureas, only stimulate insulin secretion without promoting a compensatory increase in proinsulin biosynthesis.¹⁵ This reduces insulin secretory stores in the ß-cell eventually leading to treatment failure by compromising ß-cell function.¹⁶ Thus, it is essential that ß-cell based pharmacological therapies preserve ß-cell functions, especially a parallel regulation of insulin secretion and production that is the predominant mechanism to maintain ß-granule stores of insulin.^15,16

Nonetheless, the third mechanistic component contributing to maintenance of ß-cell intracellular stores, autophagy, might also be considered a therapeutic target to maintain ß-cell insulin secretory stores.^2,7 For instance, the so-called ‘classic’ inhibitor of autophagy, 3-methyladenine (3-MA) can inhibit autophagosome formation without influencing intracellular [ATP] or protein synthesis.¹⁷ Here we have investigated the mechanism of action of monomethyladenines on primary pancreatic islet ß-cells. Surprisingly, 3-MA, N6-methyladenine (N6-MA) and 9-methyladenine (9-MA) markedly potentiated glucose-induced insulin secretion and proinsulin biosynthesis in parallel, via specific inhibition of a phosphodiesterase activity. This emphasizes the importance of cAMP-signaling in maintaining an optimal insulin secretory capacity in ß-cells, and that cAMP/protein kinase-A (PKA) signaling may negatively control autophagy in certain cell types.

Results

Specific methyladenines potentiate glucose stimulated insulin secretion

The effect of monomethylated adenine compounds on primary pancreatic islet ß-cells is essentially unknown. Isolated rat islets exposed to 1 mM monomethyladenine derivatives, 3-MA, N6-MA, and 9-MA, but not 1-MA, 7-MA, adenosine, guanine, cytosine, or thymine significantly potentiated 17 mM glucose-induced insulin secretion beyond that of adenine alone (p ≤ 0.05, Fig. 1A). Subsequent dose-response experiments with 3-MA, N6-MA, and 9-MA at 17 mM glucose indicated a similar half-maximal effective concentration (EC₅₀) of 400 ± 15 µM (n = 9). Potentiation of insulin secretion by monomethyladenines is glucose-dependent, since 3-MA, N6-MA or 9-MA did not increase insulin secretion at basal glucose concentrations between 3–6 mM, but did potentiate insulin secretion at stimulatory glucose concentrations ≥ 8 mM, reaching a maximum at 11 mM (p ≤ 0.05, Fig. 1B). Isolated islets perifused at a rate of 1 ml/min with 1 mM 3-MA or the glucagon-like peptide-1 (GLP-1; 10 nM) analog, exendin-4, indicated potentiation of 17 mM glucose-induced insulin secretion for both the 1st and 2nd phases of stimulated insulin secretion (p ≤ 0.05, Fig. 1C).

Figure 1.

3-MA, N6-MA, and 9-MA specifically potentiate glucose-induced insulin secretion. Batches of isolated rat islets were incubated for 1 hour at various glucose concentrations between 3–17 mM ± 1 mM monomethylated adenines, 1-MA, 3-MA, N6-MA, 7-MA and 9-MA, adenosine, adenine, cytosine, guanine or thymine, as indicated. Insulin secreted as a percentage of the islet insulin content was determined as described in the ‘Experimental Procedures’ section. Panel A – shows a static incubation of rat islets incubated at 17 mM glucose and monomethylated adenines or nucleobases as indicated (1 mM). Results are shown as a percentage of the islet insulin content secreted expressed as a mean ± SE (n ≥ 4), where * indicates significant difference from the equivalent no addition control (p ≤ 0.001). Panel B – shows the glucose dose response of statically incubated rat islets with no addition (black circles), 3-MA (white circles), N6-MA (white squares), or 9-MA (white diamonds). Results are shown as a percentage of the islet insulin content secreted expressed as a mean ± SE (n ≥ 3), where * indicates significant difference (p ≤ 0.005) of the islets treated with a monomethyladenine from the equivalent glucose alone. Panel C – shows rat islets perifused at the indicated glucose concentrations in the absence (black circles) or the presence of 1 mM 3-MA (white circles) or 10 nM exendin-4 (Ex-4; white squares). Results are shown as a percentage of the islet insulin content secreted expressed as a mean ± SE (n ≥ 3).

Specific methyladenines potentiate glucose-induced proinsulin biosynthesis

Glucose regulation of insulin secretion is usually paralleled by a complementary increase in proinsulin biosynthesis to replenish insulin stores lost via regulated exocytosis.² A 1 h incubation of isolated rat islets at intermediate 8 mM or stimulatory 17 mM glucose significantly increased proinsulin biosynthesis above basal 3 mM glucose (∼4-fold; p ≤ 0.05, Fig. 2A), as previously observed.² At 8 mM and 17 mM glucose, 3-MA (1 mM) or N6-MA (1 mM) significantly potentiated proinsulin biosynthesis above 8 mM and 17 mM glucose alone (∼2-fold; p ≤ 0.05, Fig. 2A), with 9-MA showing a similar potentiating effect (data not shown). Neither 3-MA, N6-MA nor 9-MA significantly affected proinsulin biosynthesis at basal 3 mM glucose indicating this to be a glucose-dependent effect. Total protein synthesis was modestly increased at 8 mM and 17 mM glucose, (∼2-fold; p ≤ 0.05, Fig. 2B) but not to the same extent as the specific effect on proinsulin biosynthesis. Neither 3-MA nor N6-MA had an effect on total protein synthesis in rat islets, consistent with previous findings in rat hepatocytes.¹⁷

Figure 2.

Proinsulin biosynthesis is significantly increased by 3-MA and N6-MA. Isolated rat islets were pulse-radiolabeled with [³H]-leucine as described in “Experimental Procedures” by incubation at 3 (white bars), 8 (gray bars), or 17 (black bars) mM glucose ± 1 mM 3-MA or N6-MA. Panel A – shows a representative gel autoradiograph of the analysis with densitometric quantification of a series of experiments shown below. Results are represented as a mean ± SE (n ≥ 4) of the fold-increase over basal 3 mM glucose, where * indicates significant difference (p ≤ 0.05) of the islets treated with a monomethyladenine from the equivalent glucose alone. Panel B – Isolated rat islet total [³H]-protein synthesis as analyzed by trichloroacetic acid precipitation as described in “Experimental Procedures”. Results are expressed as a mean ± SE (n ≥ 4).

Specific methyladenines increase glucose-stimulated cAMP accumulation and PKA activation

The glucose-dependent effect of 3-MA, N6-MA and 9-MA to potentiate insulin secretion and proinsulin biosynthesis in parallel was reminiscent of a cAMP-dependent mediated effect on islet ß-cells.¹⁵ As such, we investigated whether 3-MA, as a model ‘effector’ methyladenine, could influence glucose-induced increases in [cAMP]_i in isolated rat islets. Within 5 minutes, a stimulatory 17 mM glucose concentration instigated an ∼2-3-fold increase in [cAMP]_i above that at basal 3 mM glucose (p ≤ 0.05, Fig. 3A). In the presence of 17 mM glucose, 3-MA further increased [cAMP]_i ∼8 fold higher than that at basal 3 mM glucose (p ≤ 0.05, Fig. 3A). As a positive control, [cAMP]_i accumulation at 17 mM glucose in the presence of the generic pan-phosphodiesterase inhibitor IBMX (100 µM), and adenylate cyclase activator forskolin (100 µM) was observed that was ∼14-fold above basal 3 mM glucose (p ≤ 0.05, Fig. 3A). In the ß-cell, elevated [cAMP]_i activates protein kinase-A (PKA), which in turn phosphorylates protein substrates such as the transcription factor CREB.¹⁸ Complementary to the elevation of [cAMP]_i, an increased CREB phosphorylation was observed in islets in the presence of 1 mM 3-MA, N6-MA or 9-MA above that of 17 mM glucose alone (Fig. 3B). These data indicate that PKA signaling was activated by stimulatory glucose and further augmented by 3-MA, N6-MA, and 9-MA, consistent with these compounds causing an elevation of [cAMP]_i in primary islet ß-cells.

Figure 3.

3-MA significantly increases glucose-induced cAMP accumulation and CREB phosphorylation. Panel A - isolated rat islets were incubated for 5 minutes at 3 mM glucose or 17 mM ± 1 mM 3-MA, 100µM IBMX/100 µM forskolin. Islet lysates were then analyzed for cAMP content by ELISA and insulin by RIA, as described in “Experimental Procedures”. Results are shown as cAMP concentration per islet insulin content and expressed as a mean ± SE (n ≥ 3) where * indicates significant difference (p ≤ 0.005) between 2 incubation conditions. Panel B - Isolated islets were incubated for 0 min, 15 min, 1 h or 6 h at 17 mM glucose ± 1 mM 3-MA, N6-MA or 9-MA. Islets were lysed and lysates subjected to immunoblot analysis for phospho(S¹³³)-CREB, total CREB, and total Akt. Representative images of 6 independent experiments are shown.

3-MA action is not mediated through K_ATP channels, heterotrimeric G-protein signaling or adenylate cyclase activity

In further experiments, 3-MA was used as a ‘model’ methyladenine for N6-MA and 9-MA that appeared to have similar action on ß-cells. Glucose can transiently elevate [cAMP]_i in ß-cells via closure of the plasma membrane K_ATP channel which then leads to increased cytosolic [Ca⁺²]_i and subsequent activation of Ca⁺² dependent adenylate cyclase isoforms I & VIII.¹⁹ The K_ATP-channel opener diazoxide (0.2 mM) significantly inhibited glucose-induced and 3-MA potentiated insulin secretion (p ≤ 0.05, Fig. 4A). The K_ATP-channel closer glyburide (1 µM) significantly enhanced basal insulin secretion at 3 mM glucose and also further increased 17 mM glucose-induced as well as 3-MA potentiation of 17 mM glucose-induced insulin secretion (p ≤ 0.05, Fig. 4A). Taken together, these data suggest that 3-MA acts independently of K_ATP-channel mediated depolarization and consequent Ca⁺² influx. The GLP-1 receptor agonist, exendin-4 (10 nM), significantly potentiated 17 mM glucose-stimulated insulin secretion 2-3 fold (p ≤ 0.05, Fig. 4A), and in the presence of 3-MA this was further increased an additional 2-fold (p ≤ 0.05, Fig. 4A). This suggests that the GLP-1 receptor, and immediate downstream signaling, are not necessarily a direct target of 3-MA. In contrast, the pan-phosphodiesterase inhibitor IBMX (100 µM) significantly potentiated 17 mM glucose-induced insulin secretion at a level equivalent to that with exendin-4 plus 3-MA (p ≤ 0.05, Fig. 4A). However, 3-MA had no additional effect on the potentiation of glucose-induced insulin secretion by IBMX (Fig. 4A). This suggested that 3-MA may be acting as a PDE inhibitor to increase ß-cell [cAMP]_i .

Figure 4.

3-MA action is independent of adenylate cyclases and the KATP-channel but is competed for by the pan-phosphodiesterase inhibitor IBMX. Panel A - isolated rat islets were incubated for 1 h at 3 mM glucose (white bars), 17 mM glucose (gray bars) or 17 mM glucose + 1 mM 3-MA (black bars) in the absence or the presence of 10 nM exendin-4, 100 µM IBMX, 0.2 mM diazoxide or 1 µM glyburide. Islets were then lysed and insulin secretion and islet insulin content analyzed by RIA as described in “Experimental Procedures”. Results are shown as a percentage of islet insulin content secreted, and expressed as a mean ± SE (n ≥ 3) where * indicates significant difference (p ≤ 0.001) versus 3 mM glucose alone, ** indicates significant difference (p ≤ 0.001) vs. 17 mM glucose alone, and *** indicates significant difference (p ≤ 0.05) versus 17 mM glucose + 1 mM 3-MA. Panel B - isolated islets were incubated for 1 h at 3 mM glucose (white bars), 17 mM glucose (gray bars) or 17 mM glucose + 1 mM 3-MA (black bars) in the absence or the presence of 3 μg/mL pertussis toxin (PTX), 5 μg/mL cholera toxin (CTX) or 100 µM 9-cyclopentyladenine (9-CPA). Islets were then lysed, and insulin secretion and islet insulin content analyzed by RIA. Results are shown as a percentage of islet insulin content secreted and expressed as a mean ± SE (n ≥ 3) where * indicates significant difference (p ≤ 0.05) vs. 17 mM glucose alone, and ** indicates significant difference (p ≤ 0.05) versus 17 mM + 1 mM 3-MA.

Nonetheless, it was investigated whether heterotrimeric G-protein mediated activation of adenylate cyclase was targeted by 3-MA. Both the Gα_i inhibitor pertussis toxin (3 μg/mL) and the Gα_s activator cholera toxin (5 μg/mL) potentiated 17 mM glucose-induced insulin secretion, and in the presence of 3-MA this was further increased (p ≤ 0.05, Fig. 4B). The lack of a competitive effect of 3-MA on pertussis or cholera toxins suggested that 3-MA was not acting through these G-protein coupled signaling mechanisms. Furthermore, the adenylate cyclase inhibitor 9-cyclopentyladenine (9-CPA; 100 µM) had no effect on 17 mM glucose-induced insulin secretion but significantly decreased 3-MA potentiated insulin secretion (p ≤ 0.05, Fig. 4B). Again, the lack of a competitive effect of 3-MA on 9-CPA indicated that 3-MA was acting independently of adenylate cyclase.

3-Methyladenine inhibition of phosphodiesterase isoform activity in vitro

PDE1C, PDE3B, PDE4C, PDE 8B, PDE10A and PDE11A are the most commonly expressed PDE isoforms in pancreatic islet ß-cells.²⁰ Recombinant active protein of these PDE isoforms were obtained and assayed for activity in vitro in the presence of increasing concentration of methlyadenines and then the IC₅₀ calculated (Table 1). Neither 1-MA nor 7-MA inhibited any PDE isoform activity complementary to their having no effect on insulin secretion (Fig. 1A). As a positive control, IBMX (100 µM) was used as a pan-phosphodiesterase inhibitor and gave >60% inhibition of all these PDE isoforms. 3-MA was a weak inhibitor for PDE1C, but was a more effective inhibitor of the other PDE isoforms tested having an IC₅₀ in the 1–2 mM range (Table 1). N6-MA was a more effective inhibitor of PDE1C than 3-MA, but had a greater preference for PDE10A (Table 1). 9-MA, like 3-MA, was a relatively weak inhibitor of PDE1C, but had a preference for PDE4C (Table 1). Nonetheless, 3-MA, N6-MA and 9-MA were capable of inhibiting all PDE isoforms in the mM concentration range (Table 1) that is commonly used to inhibit autophagy,¹⁷ and should be considered as a class of pan-phosphodiesterase inhibitors similar to IBMX.

Table 1.

IC₅₀ values for selected methyladenines' inhibition of phosphodiesterase isoforms in vitro. Purified catalytic subunits of phosphodiesterase (PDE) isoforms that are commonly expressed in pancreatic islet ß-cells were incubated in vitro in increasing concentrations of 3-MA, N6-MA or 9-MA (0–10 mM) and assayed for PDE activity as described (see ‘Experimental Section’). The IC₅₀ was then calculated for each PDE isoform. A mean IC₅₀ (mM) ± SE (n ≥ 5) is shown.

PDE isoform	3-MA	N6-MA	9-MA
1C	10.7 ± 0.03	1.46 ± 0.02	4.53 ± 0.01
3B	1.00 ± 0.02	3.05 ± 0.03	1.27 ± 0.02
4C	0.97 ± 0.02	2.47 ± 0.02	0.41 ± 0.01
8B	1.61 ± 0.01	4.66 ± 0.03	3.46 ± 0.01
10A	1.46 ± 0.01	0.24 ± 0.01	1.39 ± 0.018
11A	0.89 ± 0.01	0.76 ± 0.01	0.80 ± 0.01

3-MA preferentially inhibits certain PDE isoforms in islet ß-cells

3-MA, N6-MA and 9-MA specifically increase [cAMP]_i in primary islet ß-cells to potentiate glucose-induced insulin secretion and production, most likely by acting as PDE inhibitors. To substantiate this idea it was examined whether 3-MA, as a model effector methyladenine, had any preferential specificity for PDE isoforms in isolated islets. We conducted a series of competitive experiments with other various PDE isoform specific inhibitors at concentrations that would inhibit >90% of their targeted PDE.²⁰ EHNA (5 µM) and sildenafil (1.5 nM), inhibitors specific to PDE2 and PDE5 respectively, did not potentiate 17 mM glucose-stimulated insulin secretion and had no effect on 3-MA potentiated insulin secretion (p ≤ 0.05, Fig. 5) suggesting that PDE2 & PDE5 are unlikely involved in 3-MA induced [cAMP]_i accumulation in islet β-cells that is complementary to their relatively low expression.²⁰ In contrast, incubation with MMPX (26 µM), trequinsin (0.25 nM), and papaverine (250 nM), inhibitors specific to PDE1, PDE3, and PDE10 respectively, significantly potentiated 17 mM glucose-induced insulin secretion, which was further increased in the presence of 3-MA (p ≤ 0.05, Fig. 5). This suggested that PDE1, PDE3, and PDE10 isoforms could contribute to [cAMP]_i accumulation in β-cells for potentiation of glucose-stimulated insulin secretion, but since their effect was not competed for by 3-MA are not necessarily 3-MA targets for PDE inhibition in vivo. However, incubation with the PDE4 specific inhibitors roflumilast (3 nM) and L-186,141 (1.5 nM) potentiated 17 mM glucose stimulated insulin secretion (p ≤ 0.05, Fig. 5) but competed for 3-MA induced potentiation of glucose-induced insulin secretion (Fig. 5). As such, 3-MA may act as a preferential PDE4 inhibitor in islet β-cells, but this is more likely reflective of PDE4 being the mostly commonly expressed PDE isoform in these cells.²⁰ Taken together, these experiments indicate that PDE1, PDE3, PDE4 and PDE10, but not PDE2 and PDE5, are important modulators of 17 mM glucose-stimulated insulin secretion in line with their relative expression levels in β-cells.²⁰

Figure 5.

3-MA competes with PDE4 to increase glucose-stimulated insulin secretion. Isolated rat islets were incubated for 1 h at 3 mM glucose (white bars), 17 mM glucose (gray bars) or 17 mM glucose + 1 mM 3-MA (black bars) in the absence or the presence of 1% (v/v) DMSO, 26 µM MMPX, 1.5 nM sildenafil, 0.25 nM trequinsin, 1.5 nM L-186,141, 5 µM EHNA, 250 nM papaverine or 3.0 nM roflumilast. Then, islets were lysed and insulin secretion and islet insulin content analyzed by RIA as described in “Experimental Procedures”. Results are shown as a percentage of islet insulin content secreted and expressed as a mean ± SE (n ≥ 3) where * indicates significant difference (p ≤ 0.05) vs. 17 mM glucose alone control, and ** indicates significant difference (p ≤ 0.05) versus 17 mM glucose + 1 mM 3-MA control. The PDE isoform preferentially inhibited by each inhibitor is indicated.

Effects of 3-MA on PI3′K/mTOR versus PKA signaling in isolated rat islets

The mTOR kinase is a negative regulator of autophagy, and as 3-MA is known to be an inhibitor of PI3′K/mTOR signaling in many cell types it is presumed that that is the means by which it inhibits autophagy.^21,22 But, our data also implicate 3-MA as a pan-PDE inhibitor in islet ß-cells. As such, we investigated the effect of 3-MA on glucose induced PI3′K/mTOR and PKA signaling in parallel in isolated rat islets (Fig. 6). Using p70^S6K phosphorylation as a read-out for mTORC1 activity, 3-MA had no effect on glucose-induced p70^S6K phosphorylation whereas the specific inhibitor of mTOR, rapamycin, inhibited glucose-induced p70^S6K phosphorylation (Fig. 6). CREB phosphorylation was used as a glucose-induced PKA phosphorylation substrate in ß-cells. Basal phosphorylation was increased by rapamycin as previously observed,²³ but it did not affect glucose-induced CREB phosphorylation (Fig. 6). In contrast, 3-MA enhanced both basal and glucose-induced CREB phosphorylation (Fig. 6), consistent with it being a PDE inhibitor that in turn increases cAMP/PKA signaling. Rapamycin inhibited glucose-induced Akt phosphorylation at Ser⁴⁷³, consistent with inhibiting mTOR activity in the context of mTORC2, but not at Akt Thr³⁰⁸ as this is a PDK-1 phosphorylation site²⁴ (Fig. 6). However, 3-MA inhibited phosphorylation at both Akt Thr³⁰⁸ and Ser⁴⁷³, consistent with it being a PI3′K inhibitor.²¹ Thus, our study collectively indicated the 3-MA is both a PI3′K and pan-PDE inhibitor.

Figure 6.

3-MA activates upstream effectors of PKA and mTORC1 but blunts PI3′K signaling. Overnight cultured isolated rat islets were preincubated for 1 h at 3 mM glucose in KRBH and then transferred to KRBH containing either 3 mM or 17 mM glucose with or without 200 nM rapamycin and 2 mM 3-MA for 20 minutes. Then, islets were lysed and protein content was determined to yield lysates of equivalent amounts. Forty μg of protein lysate was used per well, and proteins were resolved on 7.5% SDS-PAGE gels, transferred to nitrocellulose membranes, and blotted for phospho-p70^S6K, phospho-CREB, phospho-Akt⁴⁷³ and phospho-Akt³⁰⁸. Membranes were stripped and reblotted for total p-70^S6K, total CREB, and total Akt as loading controls for respective phosphoprotein analyses.

Discussion

Monomethylated adenines are involved in various biological functions in cells including DNA repair,²⁵ G-protein coupled receptor antagonism,²⁶ and inhibition of autophagy.¹⁷ Here, it was found that 3-MA, N6-MA and 9-MA, but not 1-MA or 7-MA, potentiated glucose-dependent insulin secretion and proinsulin biosynthesis via increasing intracellular [cAMP]_i. This is reminiscent of the effect of incretin hormones, GLP-1 and GIP, on ß-cells that work via specific Gα_s-protein linked receptors to activate adenylate cyclase and elevate [cAMP]_i¹⁴. However, it is a combination of changes in adenylate cyclase activity, which synthesize cAMP, vs. phosphodiesterase (PDEs) activity, which catalyze its degradation to inactive 5′-AMP, that control fluctuations of intracellular [cAMP]_i in ß-cells.²⁷ Using 3-MA as a model effector methyladenine with similar activity to N6-MA and 9-MA, it was found that 3-MA rapidly increased [cAMP]_i in ß-cells. This resulted in downstream activation of PKA, since 3-MA, N6-MA and 9-MA all markedly enhanced glucose-induced phosphorylation of CREB in islets. However, 3-MA did not appear to elevate [cAMP]_i in islet -cells via augmenting Gα_s heterotrimeric G-protein signaling, alleviating inhibition of Gα_i heterotrimeric G-protein signaling, or activating adenylate cyclase. Rather, 3-MA instigated an inhibition of PDE activity to keep ß-cell [cAMP]_i elevated via preventing its degradation.

PDE1, PDE3, PDE4, PDE10, and PDE11, are all significantly expressed in the pancreatic islet cell, suggesting that some of the PDE isoforms are likely involved in modulating glucose-stimulated insulin secretion and production.^20,28,29 In this study, in vitro PDE activity assays indicated that 3-MA, N6-MA and 9-MA were pan-phosphodiesterase inhibitors, perhaps with the exception of 3-MA and 9-MA inhibition of PDE1 and N6-MA inhibition of PDE8, which were rather weak (Table 1). In islet studies however, 3-MA has an apparent preference toward PDE4 inhibition in terms of potentiating glucose-induced insulin secretion (Fig. 5), but this was more likely reflective of PDE4 being the most common PDE isoform expressed in islet ß-cells.²⁰ For the moment, we cannot discount the potential role of PDE11 since no selective inhibitor currently exists for PDE11 to apply in 3-MA competition inhibition experiments (Fig. 5). However, it is likely that PDE11 is more highly expressed in non ß-cells of pancreatic islets rather than ß-cells.^20,29 As such, it is reasonable to believe that because of the high expression of PDE4 in ß-cells, it is the predominant target for 3-MA acting as a general PDE inhibitor to consequentially increase [cAMP]_i and potentiate glucose-induced insulin secretion and production. In this latter regard, 3-MA, N6-MA and 9-MA act on ß–cells similarly to GLP-1 analogs¹⁴ in activating cAMP/PKA signaling. It is tempting to think that 3-MA, N6-MA and 9-MA may be useful core-compounds to develop a new class of type 2 diabetes therapeutics that maintain ß-cell insulin secretory capacity. However, while PDE4 has already developed interest as a therapeutic target for treatment of diseases such as Alzheimer's, depression, chronic obstructive pulmonary disease, and asthma,^30,31 its expression in a wide variety of tissues including the pancreas, adipocytes, cardiovascular tissues, hepatocytes, and the brain,³² might make it a less desirable target for common obesity-linked type 2 diabetes. While increasing [cAMP]_i in ß-cells via PDE4 can boost glucose-induced insulin secretion and production,¹⁵ elevation of [cAMP]_i in insulin target cells, such as adipocytes and hepatocytes, at the same time may cause mobilization of fatty acids and gluconeogenesis that could dampen the effect of augmenting in vivo insulin production.^33,34

Another aspect of this study is to note 3-MA as a “classic inhibitor of autophagy” in eukaryotic cells when used at mM concentrations.¹⁷ In primary pancreatic ß-cells, there is an inverse relationship between autophagy and insulin production.^2,6 When insulin production and secretion are increased in ß-cells by elevated glucose concentrations, autophagic degradation of insulin stores is decreased. Conversely, when insulin production and secretion are relatively idle at basal glucose, or during prolonged fasting, intracellular autophagic degradation of insulin is enhanced.^2,6,35,36 As 3-MA markedly enhances glucose-induced insulin production, and considering the inverse relation that that has to autophagic degradation of insulin,^2,6 3-MA is unlikely to have an inhibitory effect on autophagy in ß-cells. Indeed, we have not been able to find 3-MA as an inhibitor of autophagy in primary islet ß-cells. But the data presented indicate 3-MA to be a pan-inhibitor of PDE activity, that in turn could imply that cAMP-signaling, particularly via PKA, may be a negative regulator of autophagy. In yeast, both mTOR and PKA signaling pathways regulate autophagy,³⁷ but PKA regulation of autophagy has yet to be clearly demonstrated in higher eukaryotic cells. This study may be a further indication of cAMP/PKA-signaling being able to negatively control autophagy in mammalian cells, presumably through PI3′K/mTOR activation³⁸ or direct PKA mediated phosphorylation of the autophagosome associated factor microtubule-associated protein-1 light chain-3 (LC3).³⁹ However, this requires substantiating experimentally especially in regard to identifying a PKA target in the complex autophagic machinery,⁴⁰ that when phosphorylated has an inhibitory role to block the autophagic process. These data also raise doubt for the notion that 3-MA inhibits autophagy exclusively via specific phosphatidylinositol 3′-kinase (PI3′K) inhibition.⁴¹ Nonetheless, finding 3-MA to be an effective PDE inhibitor might add novel mechanistic understanding for the control of autophagy and perhaps even lead to new therapeutic approaches aimed at decreasing autophagy in degenerative diseases,³⁸ including the loss of ß-cells in type 2 diabetes.²

Materials & Methods

Reagents

8-Methoxymethyl-3-isobutyl-1-methyl xanthine (MMPX) was obtained from Tocris Bioscience (Bristol, UK). Erythro-9-(2-hydroxy-3-nonyl) adenine (EHNA), L-186,141, and papaverine were generously supplied by Merck & Co. Inc. (Whitehouse Station, NJ). 1-MA and 9-MA were from Acros Organics/Thermo Fischer Scientific (Pittsburgh, PA), and 3-MA, N6-MA and 7-MA were obtained from Sigma Aldrich (St. Louis, MO). Roflumilast was from Santa Cruz Biotechnology (Santa Cruz, CA). L-[3,4,5–³H]Leucine (185 MBq/mL, 4 TBq/mmol) was from Perkin Elmer (Waltham, MA). Antibodies against CREB, phospho-CREB (Ser¹³³), Akt, phospho-Akt (Thr³⁰⁸), phospho-Akt (Ser⁴⁷³), phospho-p70^S6K and p70^S6K were all obtained from Cell Signaling (Danvers, MA), anti-bovine insulin serum was from Linco (St. Charles, MO). Recombinant PDE1C, PDE3B, PDE4C, PDE8B and PDE10A were purchased from SignalChem (Richmond, BC, Canada) and PDE11A from BPS Bioscience (San Diego, CA). Unless stated otherwise, all other reagents were from Sigma Aldrich (St. Louis, MO).

Islet isolation and in vitro insulin secretion analysis

Pancreatic rat islets were isolated from 2 month old adult male Wistar rats (Charles River Laboratories, Wilmington, MA) by collagenase digestion as outlined.⁴² All animal care, use, and experimental protocols were approved by the Institutional Animal Care and Use Committee of the University of Chicago. Insulin secretion from either static or perifusion incubated rat islets was performed as described,⁴³ with insulin measured by radioimmunoassay (Millipore, Billerica, MA).

Proinsulin biosynthesis was analyzed in isolated rat islets by [³H]leucine radiolabeling followed by proinsulin immunoprecipitation and alkaline-urea PAGE with subsequent fluorography analysis as described.⁴² A 5 µl aliquot of islet lysate was used for total protein synthesis analysis by trichloroacetic acid precipitation also as described.⁴²

Analysis of islet [cAMP]_i

About 25 isolated islets were preincubated for 75 min at 37°C in 0.4 ml of Krebs-Ringer Bicarbonate buffer containing 20 mM HEPES (pH 7.4) and 0.1% (w/v) BSA (KRBH) and basal 3 mM glucose, then incubated for 5 min in KRBH at either basal 3 mM glucose, or stimulatory 17 mM ± 1 mM 3-MA or 100 µM 1-methyl-3-isobutylxanthine (IBMX)/100 µM forskolin as indicated. After, islets were lysed by ultrasonication in 0.1 ml of lysis buffer (50 mmol/L HEPES, pH 8.0, 1% (v/v) SDS, 100 μM IBMX, and protease inhibitors (0.1% (w/v) phenylmethylsulfonyl fluoride (PMSF), 0.05% (w/v) E-64, 0.1% (w/v) pepstatin, 0.1% (w/v) Tosyl-lysyl chloromethylketone (TLCK), 0.1% (w/v) leupeptin). From the islet lysate, total insulin content was determined by radioimmunoassay and cAMP concentration by ELISA (Cell Biolabs Inc., San Diego, CA).

Phosphoprotein immunoblot analysis

Isolated rat islets were cultured overnight in RPMI-1640 medium containing 0.1% (w/v) BSA and 3 mM glucose; then, batches of 100 islets were incubated in 1 mL of RPMI-1640/0.1% (w/v) fatty acid-free BSA media containing 17 mM glucose ± 1 mM 3-MA, N6-MA, or 9-MA for 15 min – 6 h, as indicated. Then, islets were lysed by ultrasonication in lysis buffer (50 mM HEPES, pH 8.0, 1% (v/v) Triton X-100, protease inhibitors (0.1% (w/v) PMSF, 0.05% (w/v) E-64, 0.1% (w/v) pepstatin, 0.1% (w/v) TLCK, and 0.1% (w/v) leupeptin) and phosphatase inhibitors (4 mM EDTA, 2 mM Na₃VO₄, 10 mM Na₄P₂O₇ and 100 mM NaF). The lysate protein content was analyzed by bicinchoninic acid (BCA) assay (Pierce, Rockford, IL). Equivalent amounts of protein from each islet lysate were then analyzed by immunoblotting for total and phospho(Ser¹³³)-CREB; total, phospho(Thr³⁰⁸) and phospho(Ser⁴⁷³)-Akt; and total and phospho-p70S6K as described.⁴²

Phosphodiesterase activity assay

This protocol was adapted from an established method.⁴⁴ Essentially, recombinant PDE isoforms (1–2 mM) were incubated in 100 µl of the reaction mixture containing 5 mM MgCl₂, 3.75 mM 2-mercaptoethanol, 1 µM [³H]cAMP (1 nCi/µl) and 40mM Tris-HCl buffer (pH 7.4) either in the presence of a methyladenine (0.1–10 mM), IBMX (100 µM), or equivalent DMSO vehicle. The reaction mixture was then incubated at 30°C for between 0–120 min. At certain time points the reaction was terminated by placing at 100°C for 5 min, then cooled on ice to 4°C. Then, 2.5 µl of snake venom (Naja melanoleuca; 10mg/ml) was added and the mixture vortexed continuously for 5 min. After, 400 µl Dowex-1 resin slurry (Cl⁻ form; 50% (w/v) in water) was added and mixed again. This mixture was then centrifuged at 15,000 g for 5 min. After pelleting the [³H]cAMP, an aliquot (150 µl) of the supernatant was removed for counting of [³H]adenosine by liquid scintillation. A boiled denatured recombinant PDE enzyme preparation was used as a control to correct for [³H]cAMP not removed by the resin. PDE enzyme activity rates and the IC₅₀ of methyladenine to inhibit PDE activity were calculated.

Statistical analysis

Data are expressed as a mean ± SE of at least 3 independent experiments. Statistical significance between two groups was determined using an unpaired 2-tailed Students' t test where a p-value of ≤ 0.05 was considered statistically significant.

Disclosure of Potential Conflicts of Interest

This work was partly supported by grant from Merck & Company Inc., and some of the PDE inhibitors were provided by Merck & Company Inc. Otherwise there are no other potential conflicts of interest to disclose related to this study.

Acknowledgements

We thank the Merck & Company Inc. for generously supplying the PDE inhibitors papaverine, Erythro-9-(2-hydroxy-3-nonyl) adenine (EHNA), and L-186,141. We also thank Mr. Andrew Sprau's technical assistance in some of the early experiments of this study.

Funding

This work was supported by grants from the National Institutes of Health DK50610 (CJR), the DRC at the University of Chicago (DK020595) and an Investigator Initiated Study grant (#33602) from Merck & Company Inc., and the Kovler Diabetes Center at the University of Chicago. Brandon Boland was partly supported by a pre-doctoral training grant from the National Institutes of Health (5T32HL094282–03).