A physiological function of insulin is stimulation of cellular glucose uptake mostly into adipocytes, e.g., after a meal. While this helps to acutely maintain euglycemia, it can chronically lead to excessive lipid storage, i.e., obesity. Therefore, several studies have addressed the effects of β3-AR agonists on insulin release and circulating insulin levels, and on the interaction of insulin and β3-ARs in the control of cellular glucose uptake.
3.1. Insulin release
An early study reported that BRL 26,830 increased plasma insulin concentrations in fasted rats and improved glucose disposal after a glucose load in non-diabetic rats and mice [
15]. Subsequent studies by others reported dose-dependent increases of blood insulin and glucagon concentrations along with a lowering of blood glucose in fasted mice; propranolol at 10-50 mg/kg attenuated these responses whereas metoprolol and ICI 118,551 at 50 mg/kg mimicked the propranolol response only partly [
56]. No increase of blood insulin concentrations was observed in mice with streptozotocin-induced T1DM, implying that these effects required intact pancreatic β cells. Within the same report, BRL 26,830 also increased blood insulin and glucagon in dogs; while glucose levels were not affected, those of FFA increased markedly, pointing to a site of action not in the pancreas but rather in adipose tissue (AT) in canines.
In cultured rat pancreatic islet cells, neither BRL 26,830 nor the related compound BRL 28,410 stimulated insulin release in the presence of 2.8 or 5.6 mM glucose. The same group also studied the β
3-AR agonist CL 316,243 in
in situ perfused mouse pancreas [
57]. The agonist concentration-dependently stimulated insulin secretion, which as partially inhibited by propranolol and ICI 118,551, but not by metoprolol; based on an only incomplete inhibition by even high concentrations of propranolol (200 µM, expected to saturate β
1- and β
2-ARs), the investigators proposed that a major part of this effect was mediated by β
3-ARs.
Experiments in the rat insulinoma cell line RIN 1040-38 (a model for pancreatic β cells) found β
3-AR expression and increased insulin release in the presence of BRL 37,344 and CL 316,243 [
58]. However, this release had two characteristics that complicate interpretation of the data. Firstly, both agonists had bell-shaped concentration-response curves with maximal effects at 1-10 nM but a lack of effect at 100-100 nM. Second, the effect was transient with a peak after 30 min and return to control levels after 60 min. If the cells were transfected with the human wild-type β
3-AR, the bell-shaped concentration-response curve remained; upon transfection with the 64Arg variant of the β
3-AR, responses to CL 316,243 were not detected, again pointing to this variant being hypofunctional.
CL 316,243 increased pancreatic islet blood flow and plasma insulin concentration in rats while not affecting overall pancreatic blood flow [
59]. This was prevented by a high dose of bupranolol (general β-AR antagonist also inhibiting β
3-ARs) but not by nadolol (not inhibiting β
3-ARs), implying involvement of a β
3-AR. Based on these findings, the authors proposed that insulin release by a β
3-AR agonist may occur at least partly secondary to vasodilation of microvessels in the islet of Langerhans. Of note, while vasodilation is typically attributed to β
2-ARs, it can occur via β
3-ARs in some vascular beds [
60].
A study with three selective β
3-AR agonists for 14 days in db/db mice, including solabegron that has been tested clinically in overactive bladder patients [
61], reported dose-dependent reductions of plasma insulin concentrations [
62], which may be secondary to overall metabolic improvements as shown by concomitant reductions of glucose and HbA1c.
The involvement of β
3-AR in insulin release, food intake, and oxygen consumption induced by CL 316,243 in rat WAT was confirmed by experiments in genetically modified mouse models with either transgenic expression of the β
3-AR in AT or in β
3-AR knockout mice [
63].
Taken together these data indicate that β3-AR agonists can promote insulin release from the pancreas of rats and mice upon acute administration, but the underlying cellular and molecular mechanisms remain unclear and may be indirect, i.e., secondary to vasodilation. Moreover, these acute effects were not observed in dogs and, at least in hyperglycemic mice, turn into the opposite upon chronic administration.
These studies in combination with the expression of β
3-AR mRNA and protein in the human pancreas, specifically in the islets of Langerhans [
58], have prompted limited investigations in human subjects. Eight healthy subjects received single oral doses of BRL 35,135 (8 mg) or salbutamol (8 mg) after pre-treatment with placebo, bisoprolol (5 mg) or nadolol (20 mg) [
64]. Both agonists lowered serum potassium concentrations, a known β
2-AR response, and increased serum glucose, insulin, and lactate. All three metabolic responses were blocked by nadolol but not bisoprolol, indicating that they occurred via β
2-AR Interestingly, BRL 35,135 but not salbutamol increased serum FFA and glycerol concentrations (similar to findings in dogs [
56]), but that also appeared to be a β
2-AR effect. A recent study administered a high dose of mirabegron (100 mg q.d.) to 14 healthy women of various ethnicities for a period of 4 weeks [
13]. Insulin responses in a glucose tolerance test were similar prior to and after 27 days of treatment. Taken together these limited data do not support a relevant β
3-AR-mediated insulin release in humans, which is similar to dogs but contrasts findings in rats and mice.
3.2. Cellular glucose uptake
Despite the controversial data concerning the role of β
3-AR in mediating insulin release, several lines of evidence have indicated the possible involvement of β
3-AR in enhancing glucose tolerance and uptake. El Hadri et al. attributed this phenomenon to the complex interaction between feeding/fasting status and the expression of β
3-AR in AT [
65]. Nevertheless, insulin stimulated glucose transport in rat adipocytes with a rank order implying a β
3-AR involvement [
66,
67,
68,
69]. Several studies supported this claim. For instance, using CL 316,243 in transgenic mouse model with either β
3-AR specific expression in AT or with β
3-AR knockout mice, suggested that insulin release, food intake and oxygen consumption induced by CL 316,243 is mediated by β
3-AR in WAT [
63]. Moreover, CL 316,243 seemed to increase glucose uptake in a tissue dependent manner, and variations were observed in different models of diabetes. CL 316,243 was effective in improving glucose uptake in BAT but not in inguinal WAT (iWAT) in T2DM mice. The opposite was observed in T1DM animals [
70]. One study included different organs to examine the effect of BRL 37,344 in male Sprague Dawley rats. As expected, BRL 37,344 improved glucose uptake in skeletal muscle, heart and diaphragm, in addition to BAT and WAT [
71]. Other than that, a one week treatment with CL 316,243 (1mg/kg/day) induced glucose uptake in Wistar rat WAT but not in guinea pigs, alongside upregulation of GLUT4 mRNA expression in scWAT and BAT of treated rats [
72]. These data indicate a variation of β
3-AR role and response in glucose uptake across species.
Functional evidence unravelling β
3-AR induced glucose uptake in humans is limited, however, it is speculated that β
3-AR agonists such as mirabegron improve glucose tolerance and uptake [
73]. One study indicated that in obese subjects, mirabegron improved glucose tolerance and insulin sensitivity [
74]. Another clinical trial indicated that a supratherapeutic dose of 200 mg of mirabegron in healthy subjects stimulated BAT glucose uptake and resting metabolic rate [
75]. Other researchers proposed that glucose uptake in BAT of both in human subjects and
in vitro is modulated by GLUT4 and uncoupling protein 1 (UCP1) is diurnal bound [
76]. On another note, cold induced BAT activation increased glucose uptake in the supraclavicular and paraspinal regions [
77], but it remains unclear whether β
3-AR are involved in this effect.
In conclusion, β3-AR agonists consistently seemed to improve glucose uptake in various organs in human and animal models. This effect was more prominent in BAT and was mostly associated with a better metabolic status. However, given the limited presence of BAT in adult humans, it remains unclear how much this contributes to systemic glucose handling. Yet, more functional studies must be done, especially in humans to further support these conclusions.