Repurposing cAMP-Modulating Medications to Promote β-Cell Replication

Abstract

Loss of β-cell mass is a cardinal feature of diabetes. Consequently, developing medications to promote β-cell regeneration is a priority. cAMP is an intracellular second messenger that modulates β-cell replication. We investigated whether medications that increase cAMP stability or synthesis selectively stimulate β-cell growth. To identify cAMP-stabilizing medications that promote β-cell replication, we performed high-content screening of a phosphodiesterase (PDE) inhibitor library. PDE3, -4, and -10 inhibitors, including dipyridamole, were found to promote β-cell replication in an adenosine receptor-dependent manner. Dipyridamole's action is specific for β-cells and not α-cells. Next we demonstrated that norepinephrine (NE), a physiologic suppressor of cAMP synthesis in β-cells, impairs β-cell replication via activation of α₂-adrenergic receptors. Accordingly, mirtazapine, an α₂-adrenergic receptor antagonist and antidepressant, prevents NE-dependent suppression of β-cell replication. Interestingly, NE's growth-suppressive effect is modulated by endogenously expressed catecholamine-inactivating enzymes (catechol-O-methyltransferase and l-monoamine oxidase) and is dominant over the growth-promoting effects of PDE inhibitors. Treatment with dipyridamole and/or mirtazapine promote β-cell replication in mice, and treatment with dipyridamole is associated with reduced glucose levels in humans. This work provides new mechanistic insights into cAMP-dependent growth regulation of β-cells and highlights the potential of commonly prescribed medications to influence β-cell growth.

Diabetes is an increasingly common disorder of disrupted glucose homeostasis that exacts major health and economic costs to society. A cardinal pathologic feature of diabetes, both type 1 diabetes mellitus (T1DM) and T2DM, is the loss of β-cell mass and reduced insulin secretion. Indeed, an adaptive increase in β-cell mass is an important mechanism for avoiding hyperglycemia under physiologic conditions of increased insulin requirement, eg, pregnancy and insulin resistance (1–6). Although new β-cells may arise from a variety of sources including a poorly defined progenitor cell population or by the transdifferentiation of α-cells to β-cells, the predominant source of new β-cells in mice, and potentially humans, is previously existing β-cells (7–11). Consequently, there is substantial interest in finding methods to stimulate β-cell replication.

Inability to adequately increase β-cell mass is an important cause for the progressive hyperglycemia of T2DM. The role of impaired β-cell replication in the pathophysiology of T2DM is highlighted by the association of genetic variants in growth-associated loci, eg, CDKN2A/B and CCND2, with T2DM risk (12). Interestingly, a genetic variant that increases levels of CCND2, a replication-promoting gene product, reduces diabetes risk by 50% (13). Consistent with teleological expectations that predict insulin need to closely match insulin production capacity, glucose is a primary physiologic driver for β-cell replication (14–16). Additionally, a variety of signaling molecules including glucagon like peptide-1 (GLP-1), PTHrP, prolactin, adenosine, and osteocalcin promote β-cell replication (17–19). Importantly, all of these factors activate cAMP-dependent signaling pathways and increase intracellular levels of cAMP (20–23). Indeed, elevation of cAMP levels in β-cells leads to improved insulin secretion, enhanced survival, and increased replication (24–26). cAMP-dependent induction of β-cell proliferation is primarily mediated by activation of nuclear cAMP response element binding protein (CREB) by protein kinase A (PKA) and induction of insulin receptor substrate 2 expression (26–29). Prior work has suggested that CREB activity promotes β-cell growth; constitutive suppression of CREB activity in mouse β-cells (with a dominant-negative form of CREB or an inhibitory transcriptional cofactor [inducible cAMP early repressor I gamma]) impairs β-cell growth and glucose homeostasis, whereas constitutive activation of CREB signaling in mouse β-cells leads to a 2-fold increase in β-cell mass (25, 28, 29). Although constitutive CREB activation impaired glucose homeostasis and experiments with control transgenic animals (Rip-Cre) that are known to have impaired glucose tolerance were not shown, current data support the conclusion that cAMP-dependent signaling is an important control point of β-cell mass (30).

Given the central role cAMP plays in controlling β-cell replication, the development of methods for stimulating cAMP signaling within β-cells are of interest. One strategy for enhancing cAMP-dependent signaling is to increase cAMP stability. The stability of cAMP is primarily controlled by phosphodiesterases (PDEs) that catalyze the hydrolysis of cAMP and/or cGMP. In humans, there are 21 PDE genes that comprising 11 structurally related families (PDE1–11) (31). β-Cells express several PDE family members including PDE1, -3, -4, -7, -8,-10, and -11 (32–34). Despite the fact that nonselective PDE inhibitors (PDE-Is) such as 3-Isobutyl-1-methylxanthine are known to stimulate β-cell replication in vitro, an effort to assess the ability of highly selective PDE-Is to stimulate β-cell replication has not been undertaken (35). Of note, the therapeutic utility of nonselective PDE-Is is limited by the induction of hepatic glucose production and hyperglycemia (36). Thus, an important objective of this work is to characterize highly selective Food and Drug Administration (FDA)-approved PDE-Is that might be repurposed for the therapeutic stimulation of β-cell growth in vivo.

A second strategy for increasing cAMP-dependent signaling is to increase cAMP production by adenylyl cyclases. Several β-cell growth-promoting factors increase cAMP production by binding G protein-coupled receptors (GPCRs) that use an α_s-subunit to stimulate adenylyl cyclase activity. Indeed, loss of G_s protein activity in β-cells leads to loss of islet mass and diabetes (37). Hence, G_s-coupled receptor agonists such as GLP-1 may be useful for promoting β-cell replication. Alternatively, inhibitors of G_i-coupled receptors may also be useful for promoting β-cell replication by alleviating the inhibition of adenylyl cyclase activity. For example, cAMP production by β-cells is suppressed by the sympathetic nervous system, which innervates islets and activates α_2A-adrenergic receptors through the release of norepinephrine (NE) (38–40). Indeed, genetic variants of the α_2A-adrenergic receptor (ADRA2A) locus that increase α_2A-adrenergic receptor expression are associated with an increased risk for T2DM (41–43). Consequently, the mitogenic effect of endogenous factors that stimulate cAMP production may be impeded by NE-dependent inhibition of cAMP generation. Thus, the identification of FDA-approved medications that antagonize adrenergic-dependent inhibition of cAMP production may provide a method of facilitating β-cell regeneration in vivo. Herein we describe our efforts to identify medicines that may be repurposed to stimulate β-cell replication.

Materials and Methods

Islet isolation and β-cell replication screening protocol

All animal work was approved and carried out in accordance with our institutional animal care and use committee. Islets were isolated from male Sprague Dawley rats (250–300 g) as previously described (44). Freshly isolated islets were incubated (37°C, 5% CO₂) overnight in low-glucose (5.6mM) DMEM supplemented with 10% fetal bovine serum, penicillin/streptomycin, and GlutaMAX. The following morning, islets were trypsinized (0.25%) into cellular clusters of 1 to 3 cells, resuspended in the same medium, and plated into the wells of a 384-well plate (∼17 500 islet cells per well) that had been coated with the conditioned medium of 804G rat bladder carcinoma cells. The islet cells were allowed 48 hours to adhere at which time the medium was changed to Islet Media (Mediatech 99–786-CV) supplemented with 2% serum, 5mM glucose, penicillin/streptomycin, and GlutaMAX and the cells were treated with the compounds. For screening, compounds were tested at 10μM concentrations in duplicate. After 48 hours of compound treatment, cells were fixed with fresh 4% paraformaldehyde. Antigen retrieval was performed by heating the cells to 70°C in 95% formamide and 50mM citrate. Cells were then washed and permeabilized with PBS/0.3% Triton X-100. Staining was performed by overnight incubation with primary antibody in PBS at 4°C. For the primary screen, pancreatic and duodenal homeobox 1 (PDX-1) antibody (R&D Systems AF2419 at 1:300) was used to reveal β-cells and ki-67 antibody (BD Biosciences 556003 at 1:300) to visualize proliferating cells. Additional assays used insulin (Dako A0564 at 1:300) to identify β-cells or glucagon (Dako A0565 at 1:1000) to identify α-cells. Replication was assessed via automated image acquisition and analysis using a Cellomics ArrayScanVTI. The acquisition thresholds/parameters were established such that the computer-based calls of replication events were consistent with human-based calls.

Immunohistochemistry, chemicals, and PDE-inhibitory activity

A variety of antibodies were used: proliferating cell nuclear antigen (PCNA) (Santa Cruz Biotechnology sc-7907 at 1:50), vimentin (Millipore AB5733 at 1:1000), bromodeoxyuridine (BrdU) (Dako M074401–8 at 1:50), l-monoamine oxidase (MAO)-A/B (Santa Cruz Biotechnology sc-50333 at 1:50), phosphorylated histone 2A.X (γH2A.X) (Cell Signaling Technology 2577 at 1:500), and catechol-O-methyltransferase (COMT) (Santa Cruz Biotechnology sc-25844 at 1:100). Antigen retrieval was performed for staining of rat islet cultures (described above) and for human pancreatic sections by heating slides to 90°C for 10 minutes in 10mM sodium citrate (pH 6.0) solution. Normal human pancreatic paraffin-embedded sections were provided by the Network for Pancreatic Organ Donors with Diabietes. The chemicals used in these experiments are available from commercial vendors. Determination of the PDE-specific inhibitory activity of trequinsin, zardaverine, dipyridamole, cilostamide, tadalafil, and verdenafil was performed by BPS Bioscience with 24 purified recombinant human PDEs using the BPS Bioscience PDE assay kit (60300).

Measurement of in vivo β-cell and α-cell replication

Eight-week-old female C57BL/6J (The Jackson Laboratory; 0664) mice received daily ip injections of vehicle (10% ethanol), dipyridamole (2 mg/kg), mirtazapine (2 mg/kg), or dipyridamole (2 mg/kg) and mirtazapine for 7 days (10 μL/g body weight). After the first ip dose, animals were provided BrdU-containing water (0.8 mg/mL) in opaque bottles that were changed every 2 days. Mice were killed, and the pancreata were harvested on day 8. Paraffin sections were prepared, and immunofluorescent staining (insulin, BrdU, and 4′,6-Diamidino-2-phenylindole dihydrochloride [DAPI]) or (insulin, BrdU, and DAPI) was performed. To prevent bias, manual imaging of pancreatic sections and quantitation of the percentage of β-cells and α-cells that coexpressed BrdU were performed by investigators who were blinded to the treatment cohort. A minimum of 2000 β-cells or 1000 α-cells from nonconsecutive sections (>50 μm apart) were used to determine replication rates for each animal. Each treatment group comprised 5 animals (4 animals for dipyridamole-treated α-cell replication measurement). Exocrine cell replication (pancreatic non–β-cells) was approximated by counting all nuclei outside the islet structure but within the pancreatic parenchyma.

Glucose-stimulated insulin secretion

Rat islets were isolated and rested overnight in DMEM (5.6mM glucose with GlutaMAX). The following morning, 100 islets per condition were placed into a 40-μm cell strainer within the well of a 6-well plate containing 8 mL resting Krebs-Ringer Bicarbonate (KRB) buffer and incubated for 2 hours within the tissue culture incubator. Each condition was performed in triplicate at minimum. Islets were then serially transferred via the cell strainer to new wells that contained KRB buffer with variable glucose concentrations (2.5mM, 5mM, 15mM, and 25mM with or without compound or vehicle). Islets were maintained in each treatment condition within the tissue culture incubator for 1 hour before being transferred to the next treatment condition. After each incubation period, KRB buffer was collected and frozen. At completion of the experiment, each cell strainer was analyzed to ensure no islet loss. Insulin measurements were performed according to the manufacturer's instructions (Millipore EZRMI-13K). Normal human islets were provided by the National Disease Research Interchange at >90% purity and viability.

Clinical blood glucose measurements from electronic medical records

We performed a retrospective self-controlled case series comparing blood glucose of patients before and after dipyridamole and mirtazapine treatment. From the electronic medical record (EMR) data warehouse, Stanford Translational Research Integrated Database Environment (STRIDE), we extracted clinical laboratory tests including blood glucose (as part of a metabolic panel or by meter), fasting glucose and glycated hemoglobin. The use of STRIDE was in accordance with the Stanford Institutional Review Board.

We included patients (at least 18 years old) who initiated dipyridamole or mirtazapine after a 7-day washout period based on their prescription orders. We imposed a washout period to exclude prevalent users who had unknown initiation dates and might be exhibiting the long-term drug effects before cohort entry. The index date was the day of drug initiation. We considered only patients who were still using the drug when the laboratory testing was performed 1, 2, 3, 4, and 5 years (± 45 days) after the index date. For each patient, postdrug values were compared with his or her most recent predrug value by a nonparametric Wilcoxon paired test. Outlier laboratory values indicative of critical illness (eg, hyperosmolarity or severe hypoglycemia) were determined by Hampel's procedure and discarded (45).

Statistics

Error bars represent the SD of an experimental condition (n ≥ 3). The presence of statistically significant differences between treatment conditions was determined using the Student's 2-tailed t test where P ≤ .05 was taken to be significant. Experimental results were confirmed in independent experimentation in all cases except for the primary screening and in vivo replication experiments.

Results

Selective PDE-Is promote β-cell but not α-cell replication

The role of cAMP in enhancing β-cell replication is well-established (26). Therefore, we reasoned that PDE-Is, which prevent the breakdown of cAMP, might be used to enhance β-cell division. To test this hypothesis, we leveraged our recently established β-cell replication screening platform to measure the effect of 67 different PDE-Is on β-cell replication (Supplemental Table 1) (46). This platform uses high-content image analysis of dispersed rat islet cultures that are plated and compound-treated in a multiwell format. For primary screening, β-cell replication rates were estimated by measuring the frequency of ki-67 expression, a cell-cycle marker, by PDX-1⁺ cells. PDX-1 is a transcription factor predominantly expressed by mature rat β-cells and a fraction of δ-cells (47). For primary screening, compounds (10μM) that increased PDX-1⁺ replication by 2-fold above vehicle-treated wells were defined as hits. This experiment identified the ability of nonselective PDE-Is (3-Isobutyl-1-methylxanthine 3.6-fold, zardaverine 3.1-fold, trequinsin 6.2-fold), PDE3-Is (cilostamide 2.4-fold, milrinone 2.12-fold), and PDE4-Is (irsogladine 2.2-fold, glaucine 2.1-fold, etazolate 2.1-fold, CGH2466 3.2-fold, rolipram 2.7-fold, bay 19–8004 2.4-fold), as well as PDE5-I dipyridamole (2.2-fold) to promote β-cell replication (Figure 1A). For follow-up studies, we selected the FDA-approved drugs zardaverine and dipyridamole as well as the most efficacious compound (trequinsin). These compounds were used to generate dose-response curves (Figure 1B). All of the compounds demonstrated again the ability to promote β-cell replication.

Figure 1.

Select PDE-Is promote β-cell replication. A, The β-cell replication response of islet cell cultures treated with several PDE-Is (10μM) found to induce β-cell replication in primary screening. The fold induction of ki-67 expression by PDX-1–positive cells is shown. Data are normalized to the vehicle-treated control wells. B, Rat β-cell replication dose-response curves performed with primary screening hit compounds selected for follow-up studies (n = 4 per treatment condition). Statistically significant (P ≤ .05) induction of β-cell replication was observed for trequinsin (doses ≥0.25μM), zardaverine (doses ≥0.50μM), and dipyridamole (doses ≥2.5μM). The mean and SD of each treatment condition are shown.

Because our measurements of β-cell replication relied upon single markers of cell division (ki-67) and β-cell identity (PDX-1), we sought to confirm our findings with additional expression markers (48). We measured β-cell replication using a proliferating cell nuclear marker (PCNA) to substantiate our findings (Figure 2A). This experiment confirmed the ability of trequinsin (6.4-fold, P < .001), zardaverine (3.5-fold, P < .001), and dipyridamole (2.4-fold, P = .02) to promote β-cell replication. The concordant results of ki-67- and PCNA-based experiments confirm an enhanced replication rate in response to compound treatment. Representative images of the vehicle- and dipyridamole-treated islet cell cultures from this experiment display the anticipated distinct but overlapping expression patterns of ki-67 and PCNA (Figure 2B). Whereas ki-67 is expressed throughout the cell cycle (G1–G2/M), PCNA expression is present from late G1 to G2/M. Thus, all PCNA⁺ cells are ki-67⁺, but some ki-67⁺ cells are PCNA⁻. Next, we determined whether PDE-I–induced replication triggered a DNA damage response by quantifying the percentage of PDX⁺ cells that contained high levels of phosphorylated γH2A.X. Similar to previous studies, increased β-cell γH2A.X staining is observed in response to mitogenic stimuli (Supplemental Figure 1) (48, 49).

Figure 2.

A few PDE-Is selectively promote β-cell replication. A, The fold induction of β-cell replication in compound-treated vs vehicle-treated cells measured using PDX-1 expression to identify β-cells and PCNA expression to identify cellular replication events is shown. B, Representative images used to quantify β-cell replication are shown. β-Cells are identified by the expression of PDX-1 (blue), and replicating cells are identified by the expression of ki-67 (green) and/or PCNA (red). C, The β-cell replication response to treatment with a PDE-I is quantified using insulin expression to identify β-cells and ki-67 expression to identify dividing cells. D, The fold induction of α-cell replication, glucagon-expressing cells that coexpress ki-67, in response to vehicle or compound treatment is shown. Compounds concentrations for A–D were as follows: DMSO vehicle (0.1% vol/vol), zardaverine (10μM), trequinsin (2μM), dipyridamole (15μM), and forskolin (2μM) (n ≥ 5 per data point; *, P ≤ .02).

Islet cell cultures contain a mixture of endocrine cell types including α-, β-, and δ-cells. Because PDX-1 is expressed by somatostatin-expressing islet cells (δ-cells) as well as β-cells, we considered the possibility that we were observing a selective growth effect on δ-cells (47). Therefore, we used insulin staining as a second method to identify β-cells. An induction of insulin⁺ cell replication in response to trequinsin (5.2-fold, P = .001), zardaverine (3.0-fold, P = .009), and dipyridamole (4.4-fold, P < .001) confirms the ability of PDE-Is to promote the growth of rodent islet β-cells in vitro (Figure 2C)

Although the expression pattern of PDE family members in α-cells and the role of cAMP in governing α-cell replication are not established, we hypothesized that α-cells and β-cells might display distinct replication responses to PDE-Is. To test this hypothesis, we measured α-cell replication in response to PDE-I treatment (Figure 2D). α-Cell replication is promoted by elevations in cAMP levels through the activation of adenylyl cyclases with forskolin (3.5-fold, P < .015) and the nonselective PDE-I trequinsin (2.9-fold, P < .02). By contrast, zardaverine (1.65-fold, P = .15) and dipyridamole (0.8-fold, P = .69) did not significantly increase α-cell replication. These data uncover a role for cAMP in governing α-cell replication and confirm our hypothesis that specific PDE-Is may be used to selectively induce β-cell replication.

Dipyridamole induces a sustained, PDE5-independent β-cell replication response

Next, we measured the duration of the replication response to dipyridamole to ascertain whether PDE-Is induce a transient or a prolonged β-cell replication response. There is evidence that a cAMP-dependent inhibitory feedback loop is induced by prolonged growth stimulation (50). To test this premise, we measured the β-cell replication rate after treatment with dipyridamole for 24, 48, 72 and 96 hours (Figure 3A). Dipyridamole treatment increased β-cell replication at all time points compared with dimethylsulfoxide (DMSO) treatment: (24 hours: 1.4% vs 0.8%, P = .01; 48 hours: 3.4% vs 1.4%; 72 hours 3.3% vs 1.2%; 96 hours 5.2% vs 1.5%; P < .01 for all time points). Thus, the growth-promoting activity of dipyridamole is sustained despite chronic exposure.

Figure 3.

Dipyridamole induces sustained, PDE5-independent, β-cell replication. A, The replication rate of β-cells, the percentage of PDX-1–expressing cells that coexpressed ki-67, was determined after 24, 48, 72, and 96 hours of compound treatment (DMSO 0.1% vol/vol, dipyridamole [15μM] and Ex4 [20nM]). B and C, GSIS by isolated rat (B) and human (C) islets is shown. Islets were serially incubated at various glucose concentrations in the presence of DMSO (0.1% vol/vol), dipyridamole (15μM), Ex4 (20nM), or dipyridamole (15μM) plus Ex4 (20nM). The concentration-dependent effect of glucose on insulin secretion was determined by comparing the level of insulin secretion at 5mM, 15mM, and 25mM glucose with 2.5mM glucose (*, P ≤ .05). At a given glucose concentration, the impact of compound treatment on insulin secretion was assessed by comparing insulin secretion in the presence of compound vs vehicle (+, P ≤ .05) or the presence of dipyridamole and Ex4 vs Ex4 alone (x, P ≤ .05). Values represent mean values (n ≥ 3 independent secretion assays) with the SD shown. D, The β-cell replication response to treatment of islet cultures with vehicle (DMSO, 0.1% vol/vol), PDE5-Is (dipyridamole [15μM], vardenafil [15μM], and tadalafil [15μM]), and a soluble guanylyl cyclase inhibitor (bay 41–2272 [15μM]) is shown (*, P ≤ .01).

Contrary to expectations, exendin-4 (Ex4) did not increase β-cell replication in our assay (Figure 3A). Numerous attempts to induce β-cell replication with Ex4 treatment (used at various concentrations [0.1nM–100nM], in various media (RPMI, DMEM, and Islet Media), and at several glucose concentrations [5mM, 10mM, 15mM, and 25mM]) failed to have an effect (data not shown). To confirm the bioactivity of Ex4, we tested its impact on glucose-stimulated insulin secretion (GSIS) using isolated rat and human islets (Figure 3, B and C). As anticipated, glucose stimulated insulin release from vehicle-treated isolated rat islets: 5mM glucose 3.2-fold (P = .01), 15mM glucose 3.8-fold (P = .01) and 25mM glucose 6.8-fold (P = .001) compared with the 2.5mM glucose condition. Additionally, Ex4 showed the expected glucose-dependent enhancement of rat islet insulin secretion; no significant effect of Ex4 on insulin secretion was observed at 2.5mM or 5mM glucose, although at 15mM and 25mM glucose, an 8.5-fold (Ex4) vs 3.7-fold (DMSO) and 12.5-fold (Ex4) vs 6.8-fold (DMSO) increase in insulin secretion was observed (P = .03 and P = .001), respectively. Interestingly, dipyridamole increased rat islet insulin secretion compared with vehicle treatment at 15mM glucose (7.3- vs 3.7-fold, P = .04) and 25mM glucose (10.7- vs 6.8-fold, P = .02). Furthermore, the combination of dipyridamole and Ex4 increased insulin secretion compared with vehicle at 15mM glucose (11- vs 3.7-fold, P = .006) and 25mM glucose (14.6- vs 6.8-fold, P = .003). Finally, the combination of dipyridamole and Ex4 also enhanced insulin secretion compared with Ex4 alone at 25mM glucose (14.6- vs 12.5-fold, P = .03).

Based upon our findings that dipyridamole increases glucose-dependent and Ex4-dependent insulin secretion by rat islets, we tested the impact of dipyridamole on insulin secretion by normal human islets. As expected, glucose-induced insulin secretion at 15mM and 25mM glucose (1.6-fold and 2.6-fold, P = .02 and P < .001) compared with the 2.5mM glucose condition. Additionally, dipyridamole enhanced insulin secretion at 5mM glucose (1.3-fold vs 0.8-fold, P = .02), 15mM glucose (2.2-fold vs 1.6-fold, P = .04) and 25mM glucose (5.2-fold vs 2.6-fold, P = .008) compared with vehicle-treated islets at the same glucose concentration. Strikingly the combination of dipyridamole and Ex4 substantially enhanced the glucose-dependent effect of Ex4 on human islet insulin secretion at 5mM glucose (4-fold vs 1.9-fold, P = .01) 15mM glucose (5-fold vs 2.4-fold, P = .002) and 25mM glucose (22-fold vs 9.6-fold, P < .001). These results highlight the ability of dipyridamole to augment glucose-dependent and Ex4/glucose-dependent insulin secretion by human islets.

Based upon dipyridamole's well-characterized function as an inhibitor of PDE5, we hypothesized that this might be the mechanism by which it promotes β-cell replication. To test this hypothesis, we measured the ability of 2 highly potent PDE5-Is (vardenafil and tadalafil) and an activator of soluble cGMP cyclases (Bay 41–2272) to promote β-cell replication (Figure 3D). Contrary to our hypothesis, these drugs do not induce β-cell replication, whereas dipyridamole has the expected β-cell replication-promoting activity (12.6% vs 3.5%, P < .001; Figure 3D). Consequently, the growth-promoting effect of dipyridamole is unlikely to be mediated by PDE5 inhibition.

Dipyridamole induces β-cell replication by enhancing cAMP-dependent signaling through adenosine receptors

We hypothesized that dipyridamole might promote β-cell replication by inhibiting a PDE other than PDE5. As a first step toward testing this hypothesis, we measured the inhibitory activity of several PDE-Is against 20 recombinant human PDE enzymes (Figure 4A). The PDE-Is that promote β-cell replication (trequinsin, zardaverine, and cilostamide) primarily inhibit PDE3 and PDE4. However, dipyridamole primarily inhibits PDE5, -6, and -11 and, to a lesser extent, PDE2, -4, and -10. Among these, only PDE4, -10, and -11 are expressed by β-cells (33, 34). However, tadalafil, a potent inhibitor of PDE11, does not increase β-cell replication (Figure 3D). Therefore, dipyridamole's activity is unlikely to be mediated by PDE11. To assess a potential role for PDE10 in suppressing β-cell replication, we tested the ability of papaverine, a PDE10-selective inhibitor, to promote β-cell replication (Figure 4B) (51). Interestingly, papaverine does increase β-cell replication (1.8-fold, P < .001). Notably, this activity was absent at 10μM (our screening concentration) due to toxicity (loss of PDX⁺ cells; data not shown). The ability of PDE4-I and PDE10-I to promote β-cell replication indicates that dipyridamole's likely mechanism of action is through dual inhibition of PDE4 and PDE10. However, these experiments do not exclude a contribution from PDE-independent effects of dipyridamole.

Figure 4.

Dipyridamole-dependent induction of β-cell replication requires adenosine receptor, PKA, and VDCC activity. A, The PDE activity of 20 purified human PDEs was determined in the presence of various PDE-Is at 0.5μM and 5μM. The values shown indicate the percent inhibition. The color gradient reflects the degree of enzymatic inhibition where dark green indicates 100% inhibition and dark red indicates 0% inhibition. B, The β-cell replication response to papaverine, a PDE-10 inhibitor, is shown (*, P ≤ .05). C, The β-cell replication response to erythro-9-(2-hydroxy-3-nonyl)adenine, ozagrel, and dilazep is shown. D, β-cell replication measured after treatment with with DMSO (0.1% vol/vol), dipyridamole (dipy) (15μM), trequinsin (treq) (2μM), cilostamide (cilostam) (10μM), ABT-702 (15μM), or 5-iodotubercidin (5-IT) (1μM) in the absence (−CGS) or presence (+CGS) of the nonselective adenosine receptor antagonist CGS-15943 (10μM). Statistically significant increases in β-cell replication compared with DMSO (−)CGS are shown (+, P ≤ .01). Decreased β-cell replication in response to compound treatment in the absence vs the presence of CGS-15943 are indicated (*, P ≤ .01). E, The β-cell replication response to treatment of islet cultures with DMSO (0.1% vol/vol), dipyridamole (15μM), nitredipine (15μM), H89 (10μM), or combinations thereof is shown. A statistically significant difference in β-cell replication compared with DMSO (*, P ≤ .01) or dipyridamole (+, P ≤ .01) is indicated.

In addition to dipyridamole's activity as a PDE-I, it inhibits adenosine deaminase (ADA), thromboxane A₂ synthase, and adenosine reuptake (52). To further explore the mechanism of dipyridamole-induced β-cell replication, we tested the ability of similarly acting compounds to promote β-cell replication. Consequently, we tested the impact of erythro-9-(2-hydroxy-3-nonyl)adenine (an ADA and PDE2 inhibitor), ozagrel (a thromboxane A₂ synthase inhibitor), and dilazep (an adenosine reuptake inhibitor) on β-cell replication (Figure 4C). None of these compounds promote β-cell replication. We conclude that dipyridamole's β-cell replication-promoting activity is primarily a result of PDE4/10 inhibition.

The ribonucleoside adenosine stimulates β-cell replication by binding the G_αs-coupled adenosine 2a and 2b receptors (Adora2a and Adora2b) and increasing cAMP production (18). Because dipyridamole inhibits adenosine reuptake and augments adenosine signaling, we directly tested the impact of adenosine signaling on β-cell replication by treating islet cultures with the adenosine analog 5′-N-ethylcarboxamidoadenosine (NECA). Similar to our previous study, treatment with NECA did not increase β-cell replication (Supplemental Figure 2) (46). However, the inability of NECA to promote β-cell replication led us to consider whether adenosine receptor signaling activity in our experimental system was saturated, ie, sufficient to achieve maximal induction of β-cell replication. To test this hypothesis, we measured the impact of the nonselective adenosine receptor antagonist CGS 15943 on β-cell replication (Figure 4D). Indeed, the adenosine receptor antagonist decreased basal β-cell replication (1% vehicle-only vs 0.5% vehicle and CGS 15943, P < .001), dipyridamole-dependent β-cell replication (2.7% dipyridamole vs 0.6% dipyridamole and CGS 15943, P = .001), trequinsin-dependent β-cell replication (3.7% trequinsin vs 0.8% trequinsin and CGS 15943, P < .001), and cilostamide-dependent β-cell replication (2.0% cilostamide vs 0.5% cilostamide and CGS 15943, P < .001). By contrast, CGS 15943 did not impair the ability of the adenosine kinase inhibitors 5-iodotubercidin and ABT-702 to promote β-cell replication (Figure 4D) (46). Therefore, PDE-Is promote β-cell replication by augmenting endogenous adenosine receptor signaling.

Dipyridamole-dependent induction of β-cell replication requires PKA and voltage-dependent calcium channel activity

Elevated glucose levels and compounds such as glyburide that trigger insulin secretion promote β-cell replication (16, 53). Because dipyridamole augments insulin secretion and the cAMP/PKA signaling pathway play a prominent role in insulin release, we hypothesized that dipyridamole-dependent induction of β-cell replication might depend upon activation of the insulin secretion pathway (54). Therefore, we tested whether dipyridamole-induced replication requires calcium conductance and/or the PKA signaling pathway by measuring the impact of nitredipine (voltage-dependent calcium channel [VDCC] inhibitor) and H89 (PKA inhibitor) on β-cell replication (Figure 4E). The replication rate of β-cells increased from 2.8% in vehicle-treated cells to 8% in response to dipyridamole treatment (P < .001). However, the β-cell replication response to dipyridamole was mitigated by cotreatment with nitredipine (8% vs 2.4%, P < .001) or H89 (8% vs 0.7%, P < .001). Therefore, dipyridamole-induced β-cell replication requires VDCC-mediated calcium conductance and PKA signaling activity.

β-Cell replication is suppressed by α₂-adrenergic–mediated inhibition of cAMP generation

Next we tested whether pharmacologically enhancing cAMP production could be used to promote β-cell replication. Previous work established that cAMP production and insulin secretion are suppressed by NE via the α₂-adrenergic receptor (ADRA2A) (55–58). Consequently, we hypothesized that endogenous NE suppresses β-cell replication and that FDA-approved α₂-adrenergic receptor antagonists could be used to prevent this effect. As an initial experiment, we tested the ability of NE to suppress insulin secretion in an α₂-adrenergic receptor-dependent manner by treating isolated rat islets with NE in the absence and presence the α₂-adrenergic receptor antagonist mirtazapine (Figure 5A). Insulin secretion was increased by 9.5-fold in response to glucose treatment (2.5mM vs 25mM glucose, P < .001). However, NE reduced 15mM glucose-dependent insulin secretion by 50% (9.5-fold vehicle-treated vs 4.2-fold NE-treated, P < .001). Interestingly, the α₂-adrenergic receptor antagonist mirtazapine increased GSIS at 15mM glucose by approximately 33% (9.5-fold vehicle-treated vs 12.3-fold mirtazapine-treated, P = .005) and entirely prevented NE-dependent suppression of GSIS (4.2-fold NE-treated vs 14.4-fold NE and mirtazapine-treated, P = .001). Thus, NE-dependent suppression of GSIS is prevented by the antidepressant mirtazapine.

Figure 5.

NE-dependent suppression of β-cell replication is mediated by the G_i-coupled α₂-adrenergic receptor and modulated by endogenously expressed NE-metabolizing enzymes. A, GSIS from isolated rat islets measured in the presence of vehicle (DMSO), NE (0.4μM), mirtazapine (10μM), and the combination of NE (0.4μM) and mirtazapine (10μM). The mean (n = 3) insulin secretion and SD are shown. Insulin secretion at 25mM glucose is significantly (P ≤ .05) more than the insulin secretion at 5mM glucose for all conditions. A statistically significant effect of compound treatment at 25mM glucose is indicated (*, P ≤ .05). B, β-Cell replication rates in rat islet cultures treated with vehicle, mirtazapine (10μM), NE (2μM), or mirtazapine (10μM) and NE (2μM) are shown (*, P ≤ .05 compared with vehicle-treated). C, The effect of mirtazapine (10μM), NE (0.4μM), guanabenz (10μM), trequinsin (2μM), dipyridamole (15μM), and combinations thereof on the replication of primary β-cells in vitro are shown (*, P < .05). Statistical comparisons are made between the designated and vehicle-only treatment conditions unless otherwise indicated (+, P < .05). D, The effect of entacapone (10μM), clorgyline (10μM), mirtazapine (10μM), NE (0.4μM and 0.04μM), and combinations thereof on the replication of primary β-cells in vitro are shown. Statistical comparisons are made between the designated and vehicle-only treatment conditions (*, P < .05). E, Representative immunofluorescence staining of rat islet cultures for insulin (red) with COMT or MAO (green) and DAPI (blue).

Next we measured the effect of NE on β-cell replication (Figure 5B). Indeed, β-cell replication is inhibited by NE (4.4% vehicle-treated vs 0.7% NE-treated, P = .003) and the effect of NE on β-cell replication was reversed by blockade of α₂-adrenergic receptors with mirtazapine (0.7% NE-treated vs 4% NE- and mirtazapine-treated, P = .004). These results led us to hypothesize that α₂-adrenergic receptor-mediated suppression of β-cell replication might be dominant over PDE-I–dependent induction of β-cell replication; if α₂-adrenergic agonists inhibit cAMP generation, then PDE-Is don't have a pool of cAMP to stabilize. To test this hypothesis, β-cell replication was measured in islet cell cultures treated with vehicle, an α₂-adrenergic antagonist (mirtazapine), an α₂-adrenergic agonist (NE or guanabenz), a PDE-I (trequinsin or dipyridamole), or combinations of these compounds (Figure 5C). Consistent with our hypothesis, α₂-adrenergic agonists dramatically suppressed both basal β-cell replication (2.6% vehicle-treated vs 0.3% NE-treated [P < .001] or 0.25% guanabenz [P < .001]) and PDE-I–dependent β-cell replication (9.2% trequinsin-treated vs 3.3% trequinsin- and NE-treated [P < .001] or 3.25% trequinsin- and guanabenz-treated [P < .001]; 7.7% dipyridamole-treated vs 1.2% dipyridamole and NE-treated [P < .001] or 1.5% dipyridamole and guanabenz-treated [P < .0001]). Importantly, the suppressive effect of NE on PDE-I–induced β-cell replication is reversed by mirtazapine (3.3% trequinsin- and NE-treated vs 11.3% trequinsin-, NE-, and mirtazapine-treated [P < .001]; 1.2% dipyridamole- and NE-treated vs 3.3% dipyridamole-, NE-, and mirtazapine-treated [P = .001]) (Figure 5C and Supplemental Figure 3). These results demonstrate the ability of NE to suppress basal and PDE-dependent β-cell growth as well as the ability of an α₂-adrenergic antagonist to restore growth induction.

The ability of NE to impair β-cell function and replication led us to hypothesize that islets modulate the impact of NE by expressing NE-degrading enzymes such as COMT and MAO. To test this prediction, we measured β-cell replication in the presence of a COMT inhibitor (entacapone) or an MAO-A inhibitor (clorgyline) (Figure 5D). Although these inhibitors did not affect β-cell replication without addition of NE, they sensitized β-cells to the growth-inhibitory effects of NE. Whereas the treatment of islet cultures with NE [0.04μM] had no significant effect on β-cell replication (2.6% vehicle-treated vs 2.3% NE-treated [0.04μM] [P = .2]), concurrent treatment with a COMT or MAO-A inhibitor reduced β-cell replication by approximately 70% (2.3% NE-treated [0.04μM] vs 0.6% NE- [0.04μM] and entacapone-treated [P < .001]; 2.3% NE-treated [0.04μM] vs 1.1% NE- [0.04μM] and clorgyline-treated [P = .003]) (Figure 5D). Importantly, the ability of entacapone and clorgyline to potentiate the growth-suppressive effects of low-dose NE is prevented by concomitant addition of the α₂-adrenergic antagonist mirtazapine (0.6% NE- [0.04μM] and entacapone-treated vs 3.4% NE- [0.04μM], entacapone-, and mirtazapine-treated [P < .001]; 1.1% NE- [0.04μM] and clorgyline-treated vs 3.8% NE- [0.04μM], clorgyline-, and mirtazapine-treated [P < .001]). The ability of entacapone and clorgyline to sensitize β-cells to the growth-inhibitory effects of NE indicated that their enzymatic targets, COMT and MAO, are likely to be expressed by islet cells. Indeed, the expression of COMT and MAO was found in both insulin⁺ and insulin⁻ rat islet endocrine cells (Figure 5E). By contrast, human islets demonstrated expression of COMT but not MAO in insulin⁺ cells (Supplemental Figure 4). Consequently, the expression and regulation of NE-degrading enzymes is pertinent to the regulation of β-cell mass and function.

Dipyridamole and mirtazapine selectively induce β-cell replication in vivo

Next we assessed whether treatment with dipyridamole, mirtazapine or the combination promotes β-cell replication in vivo (Figure 6A). Treatment with either dipyridamole or mirtazapine increased β-cell replication (3.2% vehicle-treated vs 6.3% dipyridamole-treated [P = .002] and 7.5% mirtazapine-treated [P < .001]; Figure 6B) but had no significant effect on α-cell replication (Supplemental Figure 5) or non–islet-cell replication, primarily exocrine cells (0.8% vehicle-treated vs 1.1% dipyridamole-treated [P = .24] and 1.1% mirtazapine-treated [P = .27]; Figure 6C). Interestingly, combined treatment with mirtazapine and dipyridamole marginally increased β-cell replication beyond the effect of dipyridamole alone (6.3% dipyridamole-treated vs 8.0% dipyridamole- and mirtazapine-treated (P = .02)). These data confirm our hypothesis that pharmacologic modulation of cAMP-dependent signaling may be used to promote β-cell replication in vivo and highlight the β-cell selective replication effects of these medications.

Figure 6.

In vivo treatment of wild-type mice with dipyridamole or mirtazapine increases β-cell but not non–β-cell replication. A, Pancreatic sections were generated from wild-type mice that had been maintained in BrdU-containing water (7 days) and treated with vehicle, dipyridamole (2 mg/kg), mirtazapine (2 mg/kg), or both dipyridamole (2 mg/kg) and mirtazapine (2 mg/kg). Representative images of pancreatic sections taken from experimentally treated animals are shown. Sections were immunostained for insulin (red), BrdU (green), and DAPI (not shown). B, The percentage of β-cells (insulin-positive cells) that incorporated BrdU is shown (n = 5 animals per treatment condition). Statistical comparisons are made between the vehicle-treated and the designated compound-treated cohorts (*, P ≤ .05) unless otherwise indicated (+, P ≤ .05). C, The percentage of DAPI-positive cells that incorporated BrdU is shown. A statistically significant increase in BrdU incorporation by this cellular population was observed.

Treatment with dipyridamole may reduce blood glucose levels in humans

Treatment of adult human islet cultures with dipyridamole or mirtazapine failed to induce β-cell replication (data not shown). However, human β-cells may be more responsive to replication-promoting stimuli in vivo (1, 10, 11, 48). Therefore, we hypothesized that treatment of humans with dipyridamole or mirtazapine might have beneficial effects on glucose homeostasis, including enhanced β-cell replication. To test this hypothesis, we analyzed the effects of mirtazapine and dipyridamole treatment on random blood glucose levels (insufficient data were available for glycated hemoglobin or fasting glucose measurements) using a retrospective self-controlled case series drawn from the Stanford Hospital EMRs. This analysis indicated that treatment with dipyridamole significantly lowered the median blood glucose level at 2 years (−13.4 mg/dL, P < .001) and 4 years (−8.5 mg/dL, P < .05) of treatment (Figure 7). Although statistically significant decreases in glucose levels were not associated with dipyridamole treatment at all treatment intervals, the direction of changes was consistently downward. Taking all values over the 5-year treatment period (n = 610), the median glucose change was −3.0 mg/dL (P = .0025). Mirtazapine did not exhibit significant reduction in blood glucose.

Figure 7.

The median changes in blood glucose in patients before and after treatment with dipyridamole or mirtazapine for 1, 2 3, 4, and 5 years are shown (*, P < .05 by Wilcoxon's paired rank sum test).

Discussion

cAMP-dependent signaling plays an important role in controlling β-cell growth. Consequently, we undertook a systematic effort to identify medications that modulate intracellular cAMP levels and stimulate β-cell replication. This work identified and characterized the PDE-I dipyridamole and the α₂-adrenergic antagonist mirtazapine, 2 well-tolerated medicines, as in vitro and in vivo stimulators of rodent β-cell replication. Additionally, treatment with dipyridamole may reduce blood glucose levels in humans. The identification of these medications as modulators of β-cell growth and function may be an important step on the long road to our ultimate goal of pharmacologically controlling β-cell mass. Our experimental results and the studies that preceded ours provide the basis for a model of cAMP-dependent control of β-cell replication (Figure 8). This schematic highlights the cAMP-related control points for pharmacologic manipulation of β-cell mass identified and/or used in this work.

Figure 8.

Schematic model of cAMP-dependent β-cell replication. β-Cell replication is increased by adenosine binding to the adenosine A_2A,B receptors, which are coupled to Gα_S proteins (G_s) that activate adenylyl cyclases (AC) and increase cAMP levels. By contrast, β-cell replication is suppressed by catecholamines such as NE, which suppress cAMP production by activating α_2A-adrenergic receptors, C receptors that are Gα_I protein-coupled (G_i). NE-dependent growth suppression is prevented by mirtazapine, a nonselective α₂-adrenergic receptor antagonist. The potency of NE-dependent suppression of β-cell replication is modulated by endogenously expressed NE-degrading enzymes such as COMT. β-Cells express a variety of PDEs that degrade cAMP and limit the growth-promoting effects of adenosine. Inhibitors of PDE3, -4, and -10 (including dipyridamole) enhance β-cell replication. The growth-promoting activity of PDE-Is requires PKA and VDCC activity. A rise in intracellular calcium may potentiate cAMP production by increasing the activity of specific adenylyl cyclases.

Multiple cAMP-dependent signaling molecules, eg, GLP-1, GPCR 119 (GPR119) agonists, PTHrP, and osteocalcin, stimulate β-cell replication (19, 59–61). These GPCR ligands cause β-cell replication by activating cAMP-dependent signaling and inducing the expression of cell cycle-promoting proteins such as cyclins (cyclin A2, D1, and D3), S-phase kinase-associated protein 2 (skp2) and cdk2 and -4 that are required for β-cell mass expansion and by repressing the-negative cell cycle regulator p27 (27, 61–65). The basis of Ex4's (a GLP-1 analog) ability to promote insulin secretion but not β-cell replication in our experimental system is unclear and underscores the complexity of cAMP-dependent signaling: specific temporal, spatial, and costimulatory conditions may be required. Additionally, GLP-1–dependent induction of β-cell replication may be prevented by the expression of cAMP response element modulator-α and the dual-specificity phosphatase DUSP14, which have been shown to repress GLP-1–dependent β-cell replication (50). Thus, enhanced cAMP production is not always sufficient to induce β-cell replication.

β-Cell function (insulin secretion), survival, and replication are augmented by elevations in intracellular cAMP (28, 35, 66). However, β-cells express a variety of PDEs (PDE1, -3, -4, -7, -8, -10, and -11) that degrade cAMP and curtail its effects. The predominant PDE expressed by human and rodent β-cells is PDE3B (33). Consistent with this, we found several PDE3-Is (cilostamide, cilostazol, and milrinone) that stimulate β-cell replication. Although PDE3-Is are used clinically for treatment of symptoms associated with peripheral vascular disease (cilostazol) and heart failure (milrinone), PDE3B-deficient mice fail to suppress hepatic glucose production and display insulin resistance (67). Consequently, the utility of PDE3-Is for the treatment of diabetes is expected to be limited, and alternative strategies must be pursued (66).

Surprisingly, we found that the PDE5-I dipyridamole selectively stimulates rodent β-cell replication and enhances GSIS but does not inhibit PDE3B. Several lines of evidence support our conclusion that dipyridamole promotes β-cell replication by acting as a PDE-4/10 inhibitor. First, in vitro assays demonstrate that dipyridamole inhibits recombinant PDE4 and PDE10 activity. Second, PDE4 and PDE10 inhibitors promote β-cell replication. Third, compounds that mimic other functions of dipyridamole, which include inhibition of PDE5, ADA, thromboxane A₂ synthase, and adenosine reuptake, do not promote β-cell replication. Fourth, dipyridamole, like other replication-promoting PDE-Is, is dependent upon adenosine signaling for its replication-promoting activity. Fifth, dipyridamole's ability to promote β-cell replication is suppressed by NE-dependent inhibition of cAMP production. Taken together, these data strongly indicate that dipyridamole promotes β-cell replication by increasing cAMP stability through PDE4/10 inhibition. Interestingly, disruption of PDE10A protects mice from diet-induced obesity and insulin resistance, suggesting potential additional benefits of treatment with PDE10A inhibitors (68). It would be of interest to assess whether PDE10A-null animals display increased β-cell replication and mass.

Intracellular levels of cAMP depend upon the balance of its formation by adenylyl cyclases and its degradation by PDEs. A primary mechanism for controlling adenylyl cyclase activity is through GPCRs, which use G_αs or G_αI proteins to stimulate or inhibit cAMP production, respectively. Because PDEs function downstream of adenylyl cyclase, the ability of PDE-Is to promote β-cell replication implies the presence of G_αs protein activity in our experimental system. Interestingly, recent work identified the ability of adenosine to promote β-cell replication through the G_αs protein-coupled adenosine A_2A and/or A_2B receptors (18). These results prompted us to interrogate whether the replication-promoting activity of PDE-Is is dependent upon adenosine signaling. Indeed, the addition of CGS-15943, an antagonist of A_1,2A,2B adenosine receptors, reduced the basal β-cell replication rate and entirely prevented the induction of β-cell replication by PDE-Is. These results indicate that adenosine receptor agonist activity is required for cAMP-dependent β-cell replication in our experimental system. Consistent with our previous findings, β-cell replication induced by adenosine kinase inhibitors does not require adenosine signaling and acts by a distinct mechanism. These results indicate that PDE-I–dependent induction of β-cell replication in vivo may be limited by the requirement for concurrent stimulation of cAMP generation.

Our observation that cAMP-dependent β-cell replication requires activation of G_αs protein-coupled receptors raised the possibility that engagement of G_αI protein-coupled receptors would negatively regulate β-cell replication. Notably, the α_2A-adrenergic receptor (Adra2a) is among the most highly expressed G_αI GPCRs by β-cells (40). Indeed, agonists of the α_aA-adrenergic receptor, NE and guanabenz acetate, repress both basal and PDE-I–dependent β-cell replication. Importantly, we found that the antidepressant mirtazapine, which acts as a dual α₂-adrenergic and serotonin 5-HT receptor antagonist enhances the basal β-cell replication rate and rescues the NE-dependent suppression of β-cell replication. Mirtazapine's ability to promote basal β-cell replication suggests the presence of an endogenous α_2A-adrenergic receptor agonist or spontaneous G_αI protein activity in our system. The larger β-cell replication-promoting effect of mirtazapine in vivo (>2-fold) compared with in vitro (1.5-fold) may reflect tonic activity of sympathetic neurons that innervate islets. Interestingly, genetic variants that increase α_aA-adrenergic receptor expression are associated with an increased risk of developing diabetes (41). Although studies have established that α_aA-adrenergic receptor overexpression impairs insulin secretion, our experimental results indicate that the mechanism by which hyperactive adrenergic signaling contributes to the development of diabetes may be, in part, through inhibition of compensatory β-cell growth.

Based upon NE's profound effects on insulin secretion and β-cell replication, we hypothesized that β-cells express enzymes capable of metabolizing NE. Indeed, rat and human β-cells express the enzyme COMT, which degrades dopamine, epinephrine, and NE. Furthermore, pharmacologic inhibition of COMT sensitizes β-cells to the growth-suppressive effects of NE. The ability to metabolize NE may provide an endogenous mechanism for terminating the inhibitory signal generated by neuronal release of NE. Interestingly, the COMT locus displays substantial genetic heterogeneity. Several populations have minor alleles that reduce enzymatic activity in some tissues by 60% to 70% compared with the common allele (69). Although COMT has not been identified as a risk allele for T2DM through genome-wide association studies, it is of interest to determine whether these alleles are associated with reduced β-cell mass or insulin secretion capacity. Additionally, pharmacologic inhibition of COMT is used for the treatment of Parkinson's disease. To our knowledge, the impact of long-term treatment with COMT inhibitors on an individual's risk for developing diabetes and β-cell mass has not been studied. We hypothesize that impaired COMT activity predisposes individuals to diabetes by reducing β-cell mass and function.

Our goal is to develop a method for expanding human β-cell mass. We are particularly interested in exploring the potential utility of dipyridamole and mirtazapine for manipulating human β-cell growth. These compounds, which induced a replication response in free-living mice, may promote β-cell replication in humans. Unfortunately, promoting human β-cell replication is a major stumbling block for the field. Although adult human β-cells are largely resistant to replication, a variety of circumstances have confirmed their potential for growth (1). Recent studies of human β-cell replication have indicated that mitotic stimuli trigger a DNA damage response rather than cell growth, as indicated by γH2A.X staining (48, 49). Although PDE-Is increase rat β-cell γH2A.X staining, in contrast to human β-cells, this occurs primarily in the context of typical ki-67 staining (Figure 2B). Hence, the fate of ki-67–positive cells, apoptosis or survival, remains unclear because more β-cells in response to compound treatment were not directly measured. Interestingly, a recent publication by Avrahami et al (70) found that abrogation of the cyclin-dependent kinase inhibitor 1C (p57, kip2) expression, a highly expressed negative regulator of the cell cycle in human β-cells, fails to induce β-cell proliferation unless the β-cells are transplanted into diabetic mice. Similarly, we found that in vitro treatment of human islets with PDE-Is or mirtazapine does not stimulate β-cell replication. However, these compounds may effectively promote the replication of transplanted human β-cells.

The exceptional tolerability of dipyridamole and mirtazapine make them attractive candidates for use in the treatment of diabetes. Interestingly, short-term treatment of depressed nondiabetic patients with mirtazapine has shown both positive and neutral effects on glucose homeostasis in humans (71). However, the impact of chronic treatment with mirtazapine and dipyridamole on the metabolic profile of diabetic humans or animal models has not been assessed. Whether these medications are protective against diabetes remains to be tested. Although studying β-cell mass in humans is challenging, we hypothesized that treatment with these medications might increase β-cell mass and, consequently, improve glucose control. As a first step, we determined the effects of dipyridamole and mirtazapine on glucose levels through a retrospective self-controlled case series drawn from Stanford Hospital's EMRs. This analysis indicated that treatment with dipyridamole, but not mirtazapine, was associated with a significant decrease in blood glucose levels. Although the maximum effect of dipyridamole was observed at 2 years rather than more immediately, its effect may not be a result of increased β-cell replication, eg, increased insulin secretion.

The nature of our strategy for assessing the impact of dipyridamole and mirtazapine on blood glucose, eg, use of random glucose values, is expected to bias our outcome toward the null hypothesis. Consequently, the small but statistically significant reduction in glucose levels in response to dipyridamole is notable. Indeed, the impact of these medications might be larger than the effect we observed. However, our strategy does not control for prescribing practices. For instance, dipyridamole is most commonly prescribed in combination with aspirin for prevention of stroke and mirtazapine is prescribed for depression. These factors may exert a systematic bias that is not accounted for by our study design. To conclusively demonstrate that mirtazapine or dipyridamole lowers blood glucose levels, a prospective randomized controlled approach is necessary. Measuring the effects of these medications on prediabetic individuals is of particular interest. Although mirtazapine did not demonstrate an effect on glucose levels in our analysis, it is possible that individuals carrying the diabetes risk-associated Adra2a allele may be uniquely responsive. The potential use of dipyridamole, mirtazapine, or other similarly acting compounds for promoting human β-cell regeneration and improving glucose homeostasis is a fertile area for future investigation.

Additional material

Supplementary data supplied by authors.