Overexpression of Alpha2A-Adrenergic Receptors Contributes to Type 2 Diabetes

See allHide authors and affiliations

Science  08 Jan 2010:
Vol. 327, Issue 5962, pp. 217-220
DOI: 10.1126/science.1176827


Several common genetic variations have been associated with type 2 diabetes, but the exact disease mechanisms are still poorly elucidated. Using congenic strains from the diabetic Goto-Kakizaki rat, we identified a 1.4-megabase genomic locus that was linked to impaired insulin granule docking at the plasma membrane and reduced β cell exocytosis. In this locus, Adra2a, encoding the alpha2A-adrenergic receptor [alpha(2A)AR], was significantly overexpressed. Alpha(2A)AR mediates adrenergic suppression of insulin secretion. Pharmacological receptor antagonism, silencing of receptor expression, or blockade of downstream effectors rescued insulin secretion in congenic islets. Furthermore, we identified a single-nucleotide polymorphism in the human ADRA2A gene for which risk allele carriers exhibited overexpression of alpha(2A)AR, reduced insulin secretion, and increased type 2 diabetes risk. Human pancreatic islets from risk allele carriers exhibited reduced granule docking and secreted less insulin in response to glucose; both effects were counteracted by pharmacological alpha(2A)AR antagonists.

Reduced secretory capacity of pancreatic β cells and insulin resistance constitute the central features of type 2 diabetes (T2D). Recently, a more comprehensive picture of the common genetic variations that predispose to T2D has emerged (13), but the cellular disease mechanisms remain largely undefined. One strategy to dissect complex polygenic disorders like T2D is to use inbred animal models. The Goto-Kakizaki (GK) rat displays many of the hallmarks of T2D and is well characterized genetically (4). The major diabetes susceptibility locus in the GK rat is the 52-Mb locus Niddm1 on rat chromosome 1 (5). A 16-Mb portion of Niddm1, Niddm1i, confers defective insulin secretion without insulin resistance (6) and is homologous to a region on human chromosome 10 that is associated with T2D (7) and includes TCF7L2, the strongest candidate gene for T2D to date (8).

In congenic strains harboring different parts of GK-derived Niddm1i on the genetic background of normoglycemic F344 rats, two distinct regions within Niddm1i have been identified that confer impaired glucose metabolism and aberrant β cell exocytosis, respectively (9). The latter “dysexocytotic” 4.5-Mb locus with 26 known genes is fully contained in the congenic strain N1I12, which was further dissected here by generation of the congenic strains N1I5 and N1I11 that have reduced extent of GK genotype in the locus (Fig. 1A) (10).

Fig. 1

Characterization of congenic rat strains. (A) Solid bars show known GK-derived genetic segments; open ends designate intervals containing the recombinant end points. (B) Insulin secretion from control, N1I11, and N1I5 islets (n = 5 to 9 batches per strain). (C) Increases in cell capacitance (ΔC) in N1I11 and N1I5 β cells after 10 depolarizations. The histogram shows total exocytosis evoked by the train stimulus (ΔCTOT) (n = 12, 8, 11, and 21 cells from control, N1I11, N1I5, and N1I12, respectively. (D) ΔC evoked by intracellular infusion of a Ca2+-containing solution in N1I11 (blue) and N1I5 (red) β cells). The arrow indicates the establishment of the standard whole-cell configuration. (E) Maximal change in fluorescence (F) ratio (ΔF ratio) relative to baseline after the elevation of glucose to 20 mM or KCl to 60 mM (n = 9 to 11 islets for glucose, and n = 5 islets for K+). (F) The percentage of rat β cell granules docked at the plasma membrane (n = 25 cells per strain). (G) Islet mRNA expression of indicated genes normalized to Hprt (n = 8 to 9 rats per group). (H) Immunoblots of total protein from rat islets using polyclonal alpha(2A)AR antisera. (I) Glucose and insulin levels during IPGTT with or without yohimbine or clonidine as indicated (n = 7 to 20 rats per group). Time scale expanded at 0 to 30 min. *P < 0.05; **P < 0.01; ***P < 0.001.

Glucose-stimulated insulin secretion (GSIS), measured in batch-incubated islets, was normal in N1I11 islets. By contrast, N1I5 islets had a 35% reduction in GSIS compared with N1I11 (P < 0.001) and control islets (P < 0.001) (Fig. 1B). Pancreatic β cells release insulin through Ca2+-induced exocytosis, which can be monitored as increases in cell capacitance. The exocytotic response to a depolarization train stimulus amounted to 134 ± 18 fF in N1I11 β cells, similar to control cells (120 ± 11 fF) (Fig. 1C). However, N1I5 β cells displayed a ~50% reduction of exocytosis (69 ± 9 fF; P < 0.01 versus N1I11), equal to N1I12 β cells, demonstrating that the full dysexocytotic phenotype of N1I12 rats is retained in the N1I5 strain. Also, when exocytosis was evoked by intracellular dialysis of a Ca2+ buffer (free [Ca2+]i ~ 1.5 μM), it was significantly reduced in N1I5 (Fig. 1D and fig. S1), suggesting a late-stage defect, distal to elevation of cytosolic [Ca2+]i. This was corroborated by ratiometric fura-2 measurements of intracellular [Ca2+]i that produced similar results in both strains when the islets were stimulated with either 5 or 20 mM glucose or high K+ (60 mM) (Fig. 1E and fig. S2).

Insulin-containing secretory granules exist in different functional pools and are recruited from a large reserve to dock the plasma membrane, where they become release-competent (11, 12). Interestingly, insulin granule distribution differed between N1I5 and N1I11 β cells. In N1I11 cells, 3.3 ± 0.3% of the insulin granules were docked, compared with only 1.8 ± 0.2% in N1I5 cells (Fig. 1F and fig. S3). These observations suggest major differences in the secretory machineries of the two strains. These are not secondary to hyperglycemia, as the animals investigated had no overt diabetes. Instead, the impaired β cell exocytosis in N1I5 seems to be caused by the additional 1.4-Mb GK-derived genetic segment (see Fig. 1A).

The segment contains five known protein-coding genes: Pdcd4, Lysmd3, Shoc2, Adra2a, and ENSRNOG00000036577 (fig. S4). Expression analysis revealed a 59% up-regulation of Adra2a mRNA in pancreatic islets from N1I5 compared with N1I11 (P < 0.01) but no differences for the other genes (Fig. 1G). This was paralleled by a 90% increase in alpha2A-adrenergic receptor [alpha(2A)AR] protein in both islets (P < 0.001) (Fig. 1H and fig. S5A) and brain (P < 0.01) (fig. S5B) in N1I5 relative to N1I11 (see also fig. S6).

The alpha(2A)AR is known to mediate adrenaline-mediated suppression of insulin secretion. Accordingly, Adra2a knockout mice present with enhanced insulin secretion (13), and animals with β cell–specific overexpression of Adra2a are glucose-intolerant (14), but the receptor has not previously been implicated in the pathogenesis of type 2 diabetes. We hypothesized that naturally occurring genetic variations in the GK genome could cause glucose intolerance by alpha(2A)AR overexpression in N1I5 and investigated this possibility by in vivo intraperitoneal glucose tolerance tests (IPGTT) (Fig. 1I). Already in the fasting state, N1I5 rats had reduced plasma insulin compared with N1I11 (P < 0.05). Five minutes after a challenge, N1I5 rats displayed significantly elevated glucose levels compared with N1I11 (P < 0.05), which were paralleled by a pronounced reduction in insulin (P < 0.05 at 15 min). There was no difference in insulin sensitivity between the strains (k = 0.9 ± 0.1 and 1.0 ± 0.2 min–1 for N1I11 and N1I5, respectively). Treatment with the alpha(2A)AR agonist clonidine reduced insulin levels and impaired glucose tolerance in both strains. Interestingly, treatment with the antagonist yohimbine largely obliterated the differences between the strains and significantly lowered plasma glucose concentrations in N1I5 (P < 0.001 at 5 min) while increasing plasma insulin levels by as much as 156% (P < 0.01 at 15 min).

These findings echoed those observed in vitro. Insulin release was reduced in N1I5 islets at 8.3, 16.7, and 20 mM glucose (Fig. 2A). In the presence of yohimbine, GSIS was similar in N1I5 and N1I11 at all glucose concentrations. At 20 mM glucose, yohimbine increased GSIS by 30% in N1I11 islets, whereas secretion was enhanced by as much as 90% in N1I5 (P < 0.001). The strong inhibitory effect of clonidine on GSIS was also demonstrated in both strains (see also fig. S7). Next, silencing of Adra2a by RNA interference (Fig. 2B and fig. S8) (10) prevented clonidine-mediated suppression of GSIS (Fig. 2C). Interestingly, GSIS was enhanced by more than 55% in N1I5 islets after Adra2a silencing, to levels similar to those observed in N1I11 islets under the same condition. The reversal of the secretory defect was evident also at the single β cell level (P < 0.05) (Fig. 2D).

Fig. 2

Analysis of alpha(2A)AR signaling in rat pancreatic islets. (A) Insulin secretion from N1I11 and N1I5 islets at different glucose concentrations with or without clonidine or yohimbine as indicated (n = 3 to 8 batches per group). (B) Immunoblots of total protein from N1I5 islets transfected with small interfering RNA (siRNA) active against Adra2a or inactive siRNA. The histogram shows average alpha(2A)AR signal normalized for β-actin. Data from four blots. (C) Insulin secretion at 20 mM glucose from islets that are nontransfected (N.T.) or transfected with inactive or active siRNA [n = 6 to 13 batches per transfection condition for N1I11, N1I5, or N1I5 with 1 μM clonidine (C)]. (D) Depolarization-evoked capacitance increase in N1I5 β cells transfected with inactive or active siRNA. Average total exocytosis (ΔCTOT) from 7 to 9 cells per group. (E) Insulin release from N1I5 islets at 20 mM glucose with active siRNA, deltamethrin, or clonidine as indicated (n = 4 to 6 batches per group). *P < 0.05; **P < 0.01; ***P < 0.001.

The alpha(2A)AR couples to inhibitory heterotrimeric GTP–binding proteins (Gi proteins). Inactivation of Gi by pertussis toxin reversed the exocytotic defect in both islets and single β cells from N1I5 (fig. S9). Pertussis toxin was effective even in the absence of adrenaline, suggestive of tonic activity in the alpha(2A)AR/Gi signaling system. This is in line with previous findings in Noc2 knockout mice (15). Gi proteins inhibit exocytosis distal to the elevation of [Ca2+]i (16) by decreasing cyclic adenosine monophosphate (cAMP) production, but activation of the protein phosphatase 2B/calcineurin has also been proposed to contribute to the effect (17, 18). Interestingly, in all capacitance recordings, cytosolic cAMP was clamped at 0.1 mM, and impaired cAMP production accordingly cannot be the sole explanation for reduced exocytosis in N1I5. In fact, cAMP levels did not differ between the strains at 2.8 or 20 mM glucose (fig. S10). However, 5 min after addition of clonidine, cAMP was significantly lower in N1I5 islets compared with N1I11, as expected if alpha(2A)AR is overexpressed. In the absence of receptor stimulation, additional mechanisms for suppression of insulin secretion must be in operation. The previously suggested role of calcineurin was corroborated by incubation with the calcineurin inhibitor deltamethrin that increased GSIS in N1I5 islets by almost the same magnitude (45%; P < 0.05) as silencing of the receptor (Fig. 2E), while the inhibitor was ineffective in Adra2a-silenced islets. A near-identical enhancement of GSIS was obtained using FK506, another blocker of calcineurin, while the inactive deltamethrin analog perimethrin did not improve GSIS (fig. S11). Interestingly, calcineurin, in concert with protein kinases A and C, has been suggested to affect functional granule distribution and exocytosis (19, 20). Taken together, our data demonstrate that genetically encoded overexpression of Adra2a impairs β cell exocytosis at a late stage and that inappropriately activated calcineurin plays a prominent role by interfering with granule recruitment.

The Niddm1i locus is species-conserved, and we therefore genotyped 19 single-nucleotide polymorphisms (SNPs) covering 1 Mb up- or downstream of the human ADRA2A gene in 935 individuals well characterized for insulin secretion (fig. S12 and table S1) (10). Notably, the minor allele (A) of rs553668, located in the 3′ UTR region of ADRA2A, was associated with impaired insulin secretion (additive model). The impact was prominent, with a significant effect both on acute insulin response and late-phase insulin secretion at 20 to 60 min during the intravenous glucose tolerance test (IVGTT) (Fig. 3A and Table 1). Some of the other SNPs were also associated with impaired insulin secretion (table S2). Based on this, five SNPs, rs553668, rs7911129, rs1971596, rs602618, and rs2203616, were replicated in a larger cohort with 4935 individuals, which verified the importance of genetic variation in rs553668 for insulin secretion capacity. The SNP was associated with reduced fasting insulin as well as decreased insulin secretion at 30 and 120 min in response to oral glucose (Table 1). Moreover, in a case-control material with 3740 nondiabetics and 2830 diabetics, rs553668 was associated with increased risk of T2D [recessive effect; odds ratio (OR) 1.42, confidence interval (CI) 1.01 to 1.99, P = 0.04]. When focusing on individuals with low body mass index (BMI) (<24) or low C-peptide levels (<0.6), the increased T2D risk was evident also among heterozygous subjects [OR 1.31 (CI 1.03 to 1.68), P = 0.02, and OR 1.28 (CI 1.02 to 1.61), P = 0.03, respectively, using an additive model]. The results for rs553668 were significant also when correcting for the genotype for TCF7L2 rs7903146, located 1.9 Mb downstream of ADRA2A. ADRA2A rs553668 has previously been associated with increased stress sensitivity and blood pressure in humans (2123). In our cohort, it associated with lower blood pressure, which is in agreement with current views on the physiology of alpha(2A)AR (24) (Table 1).

Fig. 3

Association of ADRA2A rs553668 with insulin secretion in humans. (A) Effects of rs553668 genotype on insulin secretion during IVGTT in 799 individuals. Data are means ± SEM. P values were obtained using an additive model. (B) Immunoblots of total protein from human islets from eight individuals using polyclonal alpha(2A)AR antisera. The histogram shows average alpha(2A)AR signal normalized for β-actin from four blots from a total of 11 GG, 7 GA, and 1 AA carriers. (C) Islet ADRA2A mRNA expression in 24 GG, 7 GA, and 1 AA carriers. P < 0.05 for GG versus GA/AA, or P < 0.05 for linear regression of expression versus number of risk alleles. (D) Islet insulin secretion at 2.8 or 20 mM glucose with or without alpha(2A)AR antagonist. (E) Electron micrographs of human islet sections from GG and GA carriers at 20 mM glucose without alpha(2A)AR antagonist showing β cells with insulin granules, recognized by the central dense core and surrounding halo. The border to adjacent cells is indicated. The histogram shows the distribution of insulin granules located in 0.2-μm concentric shells in the first 2 μm below the plasma membrane after incubation at 20 mM glucose with or without alpha(2A)AR antagonist. Total granule numbers per section did not differ between GG (104 ± 9) and GA (93 ± 15) [n for (D) and (E) is specified in (10)]. *P < 0.05; **P < 0.01; ***P < 0.001.

Table 1

Effects of ADRA2A rs553668 on metabolic parameters in the human study populations. Serum insulin (S-insulin) levels are expressed as the corrected insulin response (mU × l/mmol2), and plasma glucose (P-glucose) is in mmol/l. Parameters analyzed by linear regression. Data are means ± SD. HOMA-IR, homeostasis model assessment of insulin resistance; ISI, insulin sensitivity index.

View this table:

ADRA2A overexpression in islets from carriers of the risk A allele for rs553668 was verified on both the transcript and the protein level (Fig. 3, B and C). Pancreatic islets from risk allele carriers exhibited a 30% decrease in insulin secretion at basal glucose (2.8 mM; P < 0.05) and a 40% reduction when stimulated by 20 mM glucose (P < 0.05) (Fig. 3D). Intriguingly, risk carriers also exhibited a reduced number of docked insulin granules (Fig. 3E). These defects were corrected by alpha(2A)AR antagonism, which normalized insulin granule distribution and insulin secretion to levels identical to those in non–risk carriers. These results unanimously suggest that impaired insulin secretion in rs553668 risk allele carriers is the consequence of hyperactive alpha(2A)AR signaling.

In conclusion, the present data establish the exact mechanism of reduced insulin secretion and increased T2D risk associated with ADRA2A rs553668. Alpha(2A)AR has also been implicated in the control of blood pressure and adipocyte function (24, 25). It is tempting to speculate that ADRA2A could be a common culprit for several components of the metabolic syndrome and T2D. The present findings open up a route for specific diagnosis and therapy tailored to the individual patient.

Supporting Online Material

Materials and Methods

Figs. S1 to S12

Tables S1 to S3


  • * These authors contributed equally to this work.

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank I. Lundquist and T. Andersson for help with the in vivo and fura-2 data, and B.-M. Nilsson and M. Neptin for technical assistance. Supported by project grants from the Swedish Research Council, the European Foundation for the Study of Diabetes, the Novo Nordisk and Albert Påhlsson foundations (E.R), Kungliga Fysiografiska Sällskapet (A.R.), The Nordic Centre of Excellence in Disease Genetics (NCoEDG) (R.J.), a Linnaeus grant to the Lund University Diabetes Centre, and the Knut and Alice Wallenberg Foundation. Access to human pancreatic islets was granted by J. Taneera in collaboration with the Nordic Network for Clinical Islet Transplantation (NNCIT), O. Korsgren, Uppsala University.
View Abstract

Navigate This Article