Muscarinic Stimulation of Pancreatic Insulin and Glucagon Release Is Abolished in M₃ Muscarinic Acetylcholine Receptor–Deficient Mice

M₃ muscarinic receptor−deficient mice were generated as previously described (29). All mice used for the present study (WT, M3^+/−, and M3^−/− mice) were littermates generated by intermating heterozygous M₃ receptor mutant mice (M3^+/− mice; genetic background: 129SvEv/C57BL/6J). Unless indicated otherwise, all experiments were carried out with adult male mice ages 3–7 months at the time of testing.

Mice were housed four to five per cage in a room with a 12-h light/dark cycle (lights on at 6:00 a.m.) and given ad libitum access to food and water. All manipulations were approved by the Animal Care and Use Committee of the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland.

Total RNA was extracted from mouse pancreatic islets and total brain using the total RNA isolation kit from Invitrogen. Extracted RNA samples were treated with 4 units of RNAse-free DNase (Ambion) at 37°C for 1 h to remove residual genomic DNA. The RNA was then reversed transcribed with an oligo-dT₁₆ primer and MuLV RT using the GeneAmp RNA PCR kit, as described by the manufacturer (Applied Biosystems). The reverse transcription step was omitted in control samples to test for the presence of contaminating genomic DNA. The reverse-transcribed products were screened for the presence of M₁−M₅ cDNA by PCR, using the GeneAmp RNA PCR kit (Applied Biosystems) and an Eppendorf Mastercycler thermal cycler (40 cycles of 94°C at 1 min, 56°C at 2 min, and 72°C at 3 min). PCRs were carried out in a final volume of 50 μl containing 10 μl of the RT reaction product (corresponding to ∼0.5–1 μg RNA), 10 mmol/l Tris-HCl (pH 8.3), 50 mmol/l KCl, 2 mmol/l MgCl₂, 1 mmol/l of each dNTP, 100 ng each of the sense and corresponding antisense primers, and 1.25 units of AmpliTaq DNA polymerase. The identity of the PCR products was confirmed by restriction analysis (data not shown). The RT-PCR products were separated on 1.5% agarose gels.

The sizes of the expected RT-PCR products were M₁ (497 bp), M₂ (480 bp), M₃ (498 bp), M₄ (474 bp), and M₅ (485 bp). Subtype-specific primers were designed based on the mouse M₁−M₅ muscarinic receptor sequences (GenBank data accession numbers: M₁, J04192; M₂, AF264049; M₃, F264050; M₄, X63473; and M₅, AF264051). The following primers were used: 1) M₁: 5′-TCTGCTCATCAGCTTTGACCG-3′ (forward), 5′-CATCCTCTTCCTCTTCTTCTTTCC-3′ (reverse); 2) M₂: 5′-TGTCAGCAATGCCTCCGTTATG-3′ (forward), 5′-GCCTTGCCATTCTGGATCTTG-3′ (reverse); 3) M₃: 5′-GGTGTGATGATTGGTCTGGCTTG-3′ (forward), 5′-GGAAGCAGAGTTTTCCAGGGAG-3′ (reverse); 4) M₄: 5′-TCAAGAGCCCTCTGATGAAGCC-3′ (forward), 5′-AGATTGTCCGAGTCACTTTGCG-3′ (reverse); and 5) M₅: 5′-GCTGACCTCCAAGGTTCCGATTC-3′ (forward), 5′-CCGTCAGCTTTTACCACCAAT C-3′ (reverse).

Mice were killed by cervical dislocation, and their abdominal wall was opened. The pancreas was distended by injecting 3–6 ml of dissociation solution (Hanks’ balanced salt solution containing 0.23 mg/ml liberase enzyme; Roche Diagnostics) via the common bile duct. The organ was then removed from the abdomen and incubated with an additional 1.5 ml of dissociation solution for 23 min at 37°C. Islets of Langerhans were then hand picked from the digested pancreatic tissue using a dissection microscope.

Insulin and glucagon secretion were measured after static incubation of pancreatic islets (23) prepared from WT and M₃ receptor mutant mice. Islets were preincubated for 60 min in a modified Krebs solution containing (in mmol/l) 5.6 glucose, 120 NaCl, 5 KCl, 1.1 MgCl₂, 2.6 CaCl₂, 25 NaHCO₃, and 0.5% BSA. The solution was equilibrated with 95% O₂/5% CO₂ and maintained at 37°C at a pH of 7.4 throughout the experiments. Batches of 10–12 islets were then transferred into tubes containing the same solution supplemented with various amounts of glucose and oxotremorine-M (Oxo-M; Sigma). The tubes were incubated for 1 h in a chamber maintained at 37°C and equilibrated with 95% O₂/5% CO₂. The medium was then withdrawn and frozen for the analysis of insulin and glucagon concentrations. The remaining islet insulin and glucagon content was extracted by sonication in acid/ethanol (30). The amount of insulin or glucagon secreted during the incubation period was normalized to the total insulin or glucagon content of the islets in each tube. Insulin concentrations were determined by enzyme-linked immunosorbent assay (ELISA; Alpco). Glucagon concentrations were measured via radioimmunoassay (RIA; Linco).

After an overnight (8–10 h) fast, mice received an oral load of glucose (2 mg/g body wt) via oral gavage. Blood samples were collected via retro-orbital sinus puncture before (t = 0 min) and 15, 30, 60, and 120 min after glucose administration. Blood glucose levels were determined using an automated blood glucose reader (Glucometer Elite Sensor; Bayer).

At the same time points, blood collected via the retro-orbital sinus puncture was transferred into serum separator tubes (0.5 ml, Capojet; Terumo Medical). Serum was obtained by centrifugation (10 min at 4°C) at 1,200g using an Eppendorf bench top centrifuge. Serum insulin concentrations were determined via ELISA (Alpco) using mouse insulin as a standard.

In a separate experiment, we also determined plasma glucagon levels at defined time points after oral administration of glucose (2 mg/g) to mice that had been fasted for 8–10 h overnight. Plasma glucagon concentrations were measured via RIA (Linco).

Mice were killed by cervical dislocation, and whole pancreata were dissected and immediately frozen in liquid nitrogen. Protein extracts were prepared using the acid-ethanol method (26). In brief, pancreata weighing 200–300 mg were homogenized in 1 ml acid-ethanol (95% ethanol and 10.2 N HCl at a ratio of 50:1) by an Ultrasonic Homogenizer (Biologic) for 2 min applying 20 pulses. After an overnight incubation at 4°C, the extracts were centrifuged at 650g for 30 min at 4°C. Insulin concentrations in pancreatic extracts were measured via ELISA (Alpco), following the manufacturer’s instructions. The protein content of the pancreatic extracts was determined using the Bio-Rad protein assay kit.

Data are expressed as means ± SE for the indicated number of observations. P values were calculated using one-way ANOVA followed by appropriate post hoc tests (Fisher’s least significant differences method for islet insulin and glucagon measurements and Bonferroni method for all other studies).

To examine which muscarinic receptor subtypes are expressed in pancreatic islets of the mouse, we subjected total RNA prepared from WT mouse pancreatic islets to RT-PCR amplification using M₁−M₅ mouse muscarinic receptor−specific primers. Because all five muscarinic receptors are known to be expressed in the brain (31), mouse brain total RNA served as a positive control. As expected, all five muscarinic receptors were found to be expressed in the WT mouse brain (Fig. 1A). In contrast, only M₁ and M₃ receptor cDNA could be detected in samples from WT mouse islets (Fig. 1B). Figure 1C shows the absence of M₃ receptor transcripts in pancreatic islets prepared from M3^−/− mice.

To study the potential role of the M₃ muscarinic receptor subtype in augmenting pancreatic insulin release, we carried out a series of in vitro insulin release studies using isolated islets prepared from WT, M3^+/−, and M3^−/− mice. Oxo-M, a non–subtype-selective muscarinic agonist, was used as a hydrolytically stable muscarinic stimulant throughout all experiments.

In the absence of Oxo-M, static incubation of pancreatic islets from WT mice with a basal concentration of glucose (5.6 mmol/l) led to the release of only very small amounts of insulin (Fig. 2). The magnitude of this basal secretory response was not significantly affected by the absence of M₃ receptors or the presence of Oxo-M (0.5 or 20 μmol/l) (Fig. 2).

In contrast, incubation of pancreatic islets from WT and M₃ receptor mutant mice with a high concentration of glucose (16.7 mmol/l) led to a significant increase (∼10- to 15-fold above basal levels determined in the presence of 5.6 mmol/l glucose) in insulin release (Fig. 2). In islets from WT mice (16.7 mmol/l glucose), the addition of Oxo-M (0.5 or 20 μmol/l) led to a pronounced potentiation of glucose-dependent insulin release (Fig. 2). Strikingly, the Oxo-M−mediated potentiation of glucose-dependent insulin release was totally abolished in islets prepared from M3^−/− mice and significantly reduced (by ∼30–50%) in islets from M3^+/− mice (Fig. 2).

Total pancreatic insulin content did not differ significantly between 6-month-old WT and M₃ receptor mutant mice (276 ± 40 [WT] vs. 228 ± 15 [M3^+/−] vs. 341 ± 36 [M3^−/−] pg insulin/μg pancreatic protein; n = 6). Moreover, mean wet weights of the pancreata (expressed as a percentage of total body weight) did not differ significantly among the three genotypes (0.92 ± 0.06 [WT] vs. 0.80 ± 0.05 [M3^+/−] vs. 0.95 ± 0.08% [M3^−/−]; n = 6). Preliminary immunohistochemical studies showed that the number and size distribution of pancreatic islets were similar in WT and M₃ receptor mutant mice (data not shown).

To examine whether M₃ receptor mutant mice showed altered blood glucose and insulin levels, we continuously monitored these parameters in freely fed WT, M3^+/−, and M3^−/− mice over a 6-month period. As shown in Fig. 3A, M3^+/− mice showed blood glucose levels that, at most time points, were not significantly different from the corresponding levels obtained with the WT control mice. In contrast, M3^−/− mice showed a significant reduction in blood glucose levels (by ∼20–50%) starting at age 3 months (Fig. 3A). This hypoglycemia persisted in M3^−/− mice that were >1 year old (data not shown). The serum insulin levels of 1-month-old WT, M3^+/−, and M3^−/− mice did not differ significantly among the three genotypes (Fig. 3B). In contrast, 2- to 6-month-old M3^+/− and M3^−/− mice showed significant reductions (by approximately two- to sixfold) in serum insulin levels as compared with their WT littermates (Fig. 3B). As we reported previously (29), M3^−/− mice showed pronounced reductions in body weight (by ∼25%) throughout the 6-month observation period (Fig. 3C). This difference in body weight persisted throughout the entire life of the M3^−/− mice (data not shown). In contrast, M3^+/− mice showed body weights that were similar to those of their WT littermates (Fig. 3C).

To determine whether the lack of M₃ receptors was associated with changes in glucose tolerance in vivo, WT, M3^+/−, and M3^−/− mice were subjected to an oral glucose tolerance test (OGTT; 2 mg glucose/g body wt). As shown in Fig. 4A, M3^+/− mice exhibited a similar temporal pattern of changes in blood glucose levels as their WT littermates. In contrast, M3^−/− mice showed a significant reduction in blood glucose levels compared with WT mice (P < 0.05 at 15 min after glucose administration), indicating that M3^−/− mice cleared glucose more efficiently from the circulation than the WT control animals.

In WT mice, the increase in blood glucose levels was accompanied by a significant increase in serum insulin levels that reached peak values 15 min after glucose administration (Fig. 4B). Interestingly, the glucose-induced spike in serum insulin levels observed 15 min after glucose administration was significantly blunted in M3^−/− mice (P < 0.05) (Fig. 4B). M3^+/− mice also showed a reduction in serum insulin levels at this time point, which, however, failed to reach statistical significance (P = 0.08 vs. WT) (Fig. 4B).

The incubation of isolated pancreatic islets with muscarinic agonists not only facilitates insulin secretion but also stimulates the release of glucagon (5,6,26–28). To assess the potential role of the M₃ muscarinic receptor subtype in this activity, we carried out in vitro glucagon-release studies using isolated islets prepared from WT, M3^+/−, and M3^−/− mice. In the absence of Oxo-M and in the presence of 16.7 mmol/l glucose, the amount of glucagon released from M3^+/− and M3^−/− preparations did not differ significantly from the corresponding WT value.

The addition of Oxo-M (20 μmol/l) led to a significant increase (P < 0.05) in glucagon release from WT and M3^+/− islets, as compared with the corresponding preparations that were not treated with Oxo-M (Fig. 5). In the presence of Oxo-M, M3^−/− islets secreted significantly less (P < 0.01) glucagon than WT islets, clearly demonstrating a role for M₃ receptors in cholinergic control of glucagon secretion. Moreover, the amount of glucagon released by M3^−/− islets in the presence of Oxo-M did not differ significantly (P > 0.05) from the amount released by M3^−/− islets in the absence of Oxo-M.

We next measured plasma glucagon levels in freely fed WT, M3^+/−, and M3^−/− mice. Whereas M3^+/− mice showed similar plasma glucagon levels as WT mice, M3^−/− mice displayed significantly reduced plasma glucagon levels (∼30–35% reduction vs. WT mice; P < 0.05) (Fig. 6). Similar findings were obtained when plasma glucagon levels were determined in fasted mice (116 ± 16 [WT] vs. 114 ± 22 [M3^+/−] vs. 52 ± 10 [M3^−/−] pmol/l; n = 6–8). Oral administration of glucose (2 mg/g) to these fasted animals resulted in decreased plasma glucagon levels in WT, M3^+/−, and M3^−/− mice, measured 15 or 30 min after glucose administration. However, these decreases did not reach statistical significance (P > 0.05) for any of the three genotypes (data not shown).

The stimulatory effects of acetylcholine on pancreatic insulin release are known to be mediated by activation of muscarinic receptors (4–6). RT-PCR analysis showed that mouse pancreatic islets express M₁ and M₃ muscarinic receptors (Fig. 1). Similar findings were obtained with rat pancreatic islets (23,25). Whereas previous studies (24,25) reported the expression of additional muscarinic receptor subtypes (M₄ or M₅) in islet preparations or in β-cell–derived cell lines, we did not detect M₂, M₄, or M₅ receptor mRNA in mouse pancreatic islets. Because the majority of the cells in rat or mouse islets represent β-cells, it is highly likely that both M₁ and M₃ receptors are expressed by the insulin-secreting β-cells.

To better understand the role of the muscarinic cholinergic system in β-cell function, it is essential to identify the muscarinic receptor subtype(s) mediating stimulation of insulin release. Classic pharmacological studies using different “subtype-preferring” muscarinic antagonists have suggested that the M₃ receptor subtype (previously also referred to as “glandular M₂ receptor subtype”) (32) plays a key role in the control of insulin secretion (26,27,32). However, the proper interpretation of these experiments is complicated by the limited subtype selectivity of the muscarinic antagonists used in these studies. For example, virtually all antagonists that have a high affinity for M₃ receptors (e.g., 4-DAMP, derivatives of sila-hexocyclium) also exhibit a high affinity for M₁ and M₅ receptors (33,34). Moreover, it is especially difficult to predict the simultaneous involvement of two or more muscarinic receptor subtypes (e.g., M₁ and M₃ receptors) in a specific functional response by using the currently available subtype-preferring muscarinic antagonists.

To circumvent these difficulties, we carried out in vitro insulin release studies using pancreatic islets prepared from mutant mice in which the M₃ muscarinic receptor gene had been inactivated by gene-targeting techniques (29). Consistent with previously published results (16–21), the muscarinic agonist, Oxo-M, had little effect on basal insulin release measured in the presence of a low concentration of glucose (5.6 mmol/l), using isolated islets from WT, M3^+/−, and M3^−/− mice (Fig. 2). On the other hand, in the presence of a stimulatory glucose concentration (16.7 mmol/l), the addition of Oxo-M to islets prepared from WT mice resulted in a pronounced potentiation of insulin output (Fig. 2). Strikingly, this insulinotropic activity of Oxo-M was completely abolished in islets prepared from M3^−/− mice (Fig. 2). Pancreata from M3^−/− mice contained normal amounts of total insulin, and preliminary studies showed that the number and size of pancreatic islets were similar in WT and M3^−/− mice (data not shown). These observations demonstrated in an unambiguous fashion that muscarinic receptor−mediated stimulation of pancreatic insulin release is mediated by the M₃ receptor subtype and that M₁ or other muscarinic receptor subtypes do not contribute to this response to a significant extent. The potential functional role of islet M₁ receptors remains to be elucidated.

Interestingly, heterozygous M₃ receptor mutant mice (M3^+/− mice), in which the density of M₃ receptors is reduced by ∼50% (29), also showed significant impairments in muscarinic agonist−mediated potentiation of glucose-dependent insulin release (reduction in maximum secretory responses by ∼30–50%). This observation indicated that >50% of the islet M₃ muscarinic receptors must be occupied to achieve maximum insulinotropic activity, at least in the mouse model. In contrast to many other muscarinic responses, M₃ muscarinic receptor−mediated augmentation of insulin release is therefore characterized by a very low degree of receptor reserve.

The M₃ muscarinic receptor, like the M₁ and M₅ receptor subtypes, is known to selectively couple to G proteins of the G_q family (22,33). Consistent with this finding, muscarinic stimulation of pancreatic β-cells leads to a series of biochemical events that are usually associated with the activation of G_q-type G proteins, including the activation of phospholipase C, protein kinase C, and phospholipase A2 (4–6). Stimulation of these signaling cascades eventually results in elevated intracellular calcium levels and an increase in the efficiency of calcium-dependent exocytosis of insulin-containing storage vesicles (4–6). These latter two mechanisms are predicted to be primarily responsible for the insulinotropic effects of M₃ receptor activation.

Our findings suggest that drugs that can selectively activate M₃ muscarinic receptors may be of potential therapeutic benefit in the treatment of type 2 diabetes. Such agents appear particularly attractive as they would exert their full insulinotropic effects only in the presence of stimulatory concentrations of glucose, an effect not seen with other antidiabetic drugs such as the commonly used sulfonylureas (3). Unfortunately, muscarinic agonists that can activate M₃ muscarinic receptors (or other muscarinic receptor subtypes) with a high degree of subtype selectivity are not available at present (22,33). The development of such agents therefore represents an important goal for medicinal chemists.

Consistent with the results of the in vitro insulin release studies, M3^+/− and M3^−/− mice showed reduced serum insulin levels in vivo (by approximately two- to sixfold, as compared with their WT littermates) (Fig. 3B). Moreover, M3^−/− mice displayed a significant decrease in serum insulin levels 15 min after an oral glucose load (2 mg/g) (Fig. 4B). However, the M3^+/− and M3^−/− mice did not develop hyperglycemia or impaired glucose tolerance in vivo. In fact, M3^−/− mice showed significantly reduced blood glucose levels (Fig. 3A) and even displayed a significant increase in glucose tolerance in the OGTT (Fig. 4A), indicative of an increase in insulin sensitivity. Consistent with this observation, M3^−/− mice showed a more pronounced and prolonged hypoglycemia response than WT control mice in an insulin tolerance test (29).

We also demonstrated previously that the mass of peripheral fat deposits is significantly reduced in M3^−/− mice and that this phenotype was linked to a reduction in food intake (29). It is well known that lean animals (individuals) generally exhibit an increase in insulin sensitivity (35). Muscarinic receptors located in insulin-sensitive tissues, such as liver, skeletal muscle, and fat, do not seem to play a significant role in regulating glucose utilization or other metabolic functions in these tissues (36). It is therefore likely that the increased insulin sensitivity displayed by the M3^−/− mice in vivo was primarily due to the reduction in body fat mass (29). This observation may explain why the lack of cholinergic stimulation of insulin release observed with islets from M3^−/− mice in vitro did not result in hyperglycemia and/or impaired glucose tolerance in vivo. The increase in insulin sensitivity displayed by the M3^−/− mice may represent another major factor contributing to the hypoinsulinemia associated with the lack of M₃ receptors.

Interestingly, although the M3^−/− mice were able to clear glucose from the circulation more efficiently than WT mice (OGTT), glucose clearance was unaltered in M3^+/− mice (Fig. 4A). In contrast to M3^−/− mice, which showed a significantly reduced body weight, the body weight of M3^+/− mice did not differ significantly from that of their WT littermates (Fig. 3C). These findings therefore support the concept that the increased insulin sensitivity displayed by the M3^−/− mice is caused primarily by the reduction in body weight/body fat mass (29).

In a recent study (29), we provided evidence that the absence of hypothalamic M₃ receptors may be responsible, at least partially, for the hypophagia displayed by the M3^−/− mice. This hypophagia phenotype may represent a major factor contributing to the reduction in body weight and blood glucose levels associated with the absence of M₃ receptors (29).

The activation of vagal nerves not only promotes insulin release but also stimulates the release of glucagon from pancreatic α-cells (5,6,28). In vitro and in vivo experiments suggest that this activity is mediated by muscarinic receptors predicted to be located on pancreatic α-cells (5,6,26–28). In the present study, we demonstrated that the muscarinic agonist, Oxo-M, failed to stimulate the release of glucagon from isolated islets prepared from M₃ receptor−deficient mice (M3^−/− mice) (Fig. 5), indicating that this response is mediated by the M₃ receptor subtype in WT animals. In agreement with this finding, previous in vitro and in vivo studies using muscarinic antagonists of limited receptor subtype selectivity have also suggested that the M₃ subtype plays a key role in muscarinic receptor−mediated glucagon release (26,27).

Consistent with the in vitro glucagon release data, we found that plasma glucagon levels were significantly decreased (by ∼30–50%) in both fed and fasted M3^−/− mice. Glucagon counteracts the hypoglycemic effects of insulin, primarily by stimulating hepatic glucose production. It is therefore likely that the reduced glucagon plasma levels of the M3^−/− mice contribute to the hypoglycemia displayed by these mutant animals.

Because hypoglycemia usually triggers an increase in circulating glucagon levels, it is possible that our findings underestimated the importance of pancreatic M₃ receptors in mediating glucagon secretion. Moreover, because hypoglycemia normally leads to reduced insulin secretion, the reduced blood glucose levels displayed by the M3^−/− mice may also contribute to the hypoinsulinemia associated with the lack of M₃ receptors.

In conclusion, we have demonstrated that muscarinic stimulation of pancreatic insulin and glucagon release is mediated by the M₃ muscarinic receptor subtype. Consistent with this observation, M₃ receptor−deficient mice showed pronounced reductions in plasma insulin and glucagon levels. Given the lack of muscarinic ligands that can selectively block or inhibit specific muscarinic receptor subtypes with a high degree of selectivity, these findings highlight the usefulness of muscarinic receptor mutant mice as a novel tool for dissecting the metabolic roles of the M₁−M₅ muscarinic receptors.

This work was supported in part by a grant from the American Diabetes Association (to D.M.).