Report

Regulation of Pancreatic β Cell Mass by Neuronal Signals from the Liver

See allHide authors and affiliations

Science  21 Nov 2008:
Vol. 322, Issue 5905, pp. 1250-1254
DOI: 10.1126/science.1163971

Abstract

Metabolic regulation in mammals requires communication between multiple organs and tissues. The rise in the incidence of obesity and associated metabolic disorders, including type 2 diabetes, has renewed interest in interorgan communication. We used mouse models to explore the mechanism whereby obesity enhances pancreatic β cell mass, pathophysiological compensation for insulin resistance. We found that hepatic activation of extracellular regulated kinase (ERK) signaling induced pancreatic β cell proliferation through a neuronal-mediated relay of metabolic signals. This metabolic relay from the liver to the pancreas is involved in obesity-induced islet expansion. In mouse models of insulin-deficient diabetes, liver-selective activation of ERK signaling increased β cell mass and normalized serum glucose levels. Thus, interorgan metabolic relay systems may serve as valuable targets in regenerative treatments for diabetes.

Obesity is a major public health concern in most industrialized countries (1). The development of insulin resistance in obese individuals can promote pancreatic β cell proliferation, a compensatory response that leads to increased insulin secretion (2). This in turn can lead to hyperinsulinemia, often observed in type 2 diabetes and metabolic syndrome. The mechanism(s) by which obesity-induced insulin resistance alters pancreatic β cell mass are poorly understood.

Metabolic communication between organs is essential for maintaining systemic glucose and energy homeostasis. In addition to humoral factors such as hormones and cytokines (3, 4), neuronal signals, both afferent (5, 6) and efferent (7), play important roles in such interorgan metabolic communication (8). Disruption of insulin signaling in the liver (9), but not in muscle (10) or adipose tissue (11), induces pancreatic β cell hyperplasia and hyperinsulinemia, which suggests that the liver plays important roles in regulating pancreatic β cell mass.

To identify possible mechanisms underlying the compensatory responses of pancreatic β cells to obesity-induced insulin resistance, we studied proteins that are up-regulated or activated in the livers of mouse obesity models. One of these proteins is extracellular regulated kinase (ERK). We confirmed that ERK phosphorylation is enhanced in the livers of leptin-deficient (ob/ob) and high-fat-diet–induced obese (HF) mice (fig. S1A) (12), two murine obesity models that exhibit islet hyperplasia in response to insulin resistance.

Activation of mitogen-activated protein kinase/ERK kinase (MEK) results in ERK phosphorylation (13). To elucidate the metabolic roles of hepatic ERK activation, we expressed the constitutively active mutant of MEK-1 (CAM) in the liver (14). To distinguish endogenous from exogenous MEK1, we expressed the Xenopus homo-log of MEK1. Mice administered an adenovirus encoding the LacZ gene were used as controls. Systemic infusion of recombinant adenoviruses resulted in expression of transgenes primarily in the liver, particularly hepatocytes (6) (fig. S1B), with no detectable expression in other organs, including the gastrointestinal tract (fig. S1C). Hepatic ERK phosphorylation, which is dependent on adenoviral titers (fig. S1D), was strongly enhanced on day 3 but had returned to the control level by day 9 after adenoviral administration (Fig. 1A). Hepatic ERK phosphorylation levels of CAM mice on day 3 were at most 2.1 times as high as those in the murine obesity model (fig. S1A). Hepatic lipid accumulation was markedly enhanced on day 3 but had also returned to the control level by day 14 (fig. S1E). No tumor formation was observed in the livers of CAM mice on day 44 (fig. S1F).

Fig. 1.

Activation of hepatic ERK pathway in mice enhances glucose-stimulated insulin secretion and pancreatic β cell proliferation. (A) Time courses of hepatic ERK phosphorylation after injection of 1 × 108 plaque-forming units (PFU) per mouse of recombinant adenovirus containing CAM. (B) Blood glucose (left) and plasma insulin (right) levels during glucose tolerance tests performed on day 3 after adenoviral administration. (C) Blood glucose levels after intraperitoneal insulin injection on day 3 after adenoviral administration. Data are presented as percentages of the blood glucose levels immediately before insulin loading. In (B) and (C), open and closed circles indicate LacZ and CAM mice, respectively. (D and E) Time course of islet masses (D) and pancreatic insulin content (E). In (E), white, gray, and black circles indicate LacZ, CAM and untreated mice, respectively. (F) BrdU staining of pancreases. Representative images on day 3 after adenoviral administration are shown in the two left panels. Scale bar, 100 μm. Time course of BrdU-positive cell ratios within (left) and outside of (right) the islets. In (D) and (F), open and closed bars indicate LacZ and CAM mice, respectively. (G) Double staining of pancreases from CAM mice with BrdU (brown) and insulin (red) on day 3 after adenoviral administration. A representative image is shown. Scale bar, 100 μm. (H) Expression of exogenous (Xenopus) MEK1 (upper panel) and endogenous (mouse) MEK1 (lower panel) in pancreatic islets of LacZ and CAM mice on day 3 after adenoviral treatments. After 40 polymerase chain reaction cycles, the samples were subjected to gel electrophoresis. Data are presented as means ± SEM. *, P < 0.05, **, P < 0.01 versus LacZ mice, assessed by unpaired t test.

Notably, CAM-adenovirus administration induced insulin hypersecretion. CAM mice exhibited better glucose tolerance, with markedly higher serum insulin levels at 15 min after a glucose load during glucose tolerance testing (Fig. 1B) but no significant alterations in insulin sensitivity (Fig. 1C). Enhanced glucose-stimulated insulin secretion was also observed in isolated pancreatic islets from CAM mice (fig. S2A). Hepatic expression levels of gluconeogenic enzymes were decreased in CAM mice (fig. S2B), which may account for the lowering of fasting blood glucose levels. In addition, pancreatic islet masses in CAM mice increased gradually, by a factor of 1.9 on day 15 (Fig. 1D and fig. S2C), with no significant differences in the body weights of LacZ and CAM mice (fig. S2D). The pancreatic insulin content of CAM mice also increased, rising to more than double the control level on day 16 (Fig. 1E), although these insulinotropic effects were attenuated after about 6 weeks (fig. S3, A and B). Hepatic activation of the p38 MAPK pathway—another MAPK pathway induced by administration of adenovirus encoding the constitutively active mutant of MAPK kinase 6 (CAMKK6) (fig. S4A)—did not cause insulin hypersecretion (fig. S4B) or increase pancreatic insulin content (fig. S4C).

To examine the mechanisms underlying the increased pancreatic insulin content in CAM mice, we performed bromodeoxyuridine (BrdU) staining. BrdU-positive cells were dramatically increased specifically in pancreatic islets in CAM mice, by a factor of 4.7, on day 3 after adenoviral treatment (Fig. 1F). Furthermore, nearly all (97.6%) BrdU-positive islet cells were also positive for insulin (Fig. 1G), indicating selective proliferation of pancreatic β cells in CAM mice.

The ERK pathway in pancreatic β cells is required for mitogenic responses (15); however, exogenous Xenopus MEK1 expression was undetectable in CAM mouse islets (Fig. 1H), making it unlikely that the β cell proliferation observed in CAM mice is due to direct infection of pancreatic β cells by the CAM-adenovirus.

We hypothesized that interorgan metabolic communication from the liver to pancreatic islets is the mechanism underlying insulin hypersecretion and selective proliferation of pancreatic β cells in CAM mice. Efferent vagal signals to the pancreas modulate insulin secretion (16) and pancreatic islet mass (17, 18). To examine the possible role of efferent vagal signals, we performed pancreatic vagotomy (PV) or a sham operation on the mice, followed by adenoviral administration 7 days later. PV almost completely abolished the CAM-induced glucose-lowering effects (Fig. 2A) and enhancement of glucose-stimulated insulin secretion (Fig. 2B), as well as the increases in pancreatic insulin content (Fig. 2C) and BrdU-positive islet cells (Fig. 2D) with no significant body weight alterations (fig. S5A). PV did not decrease glucose-stimulated insulin secretion, pancreatic insulin contents, or BrdU-positive islet cell numbers in LacZ mice (Fig. 2, B to D). These results strongly suggest that vagal nerves innervating the pancreas are involved in insulin hypersecretion and pancreatic β cell proliferation in CAM mice.

Fig. 2.

Dissection of the pancreatic vagus, afferent blockade of the hepatic splanchnic nerve, or midbrain transection inhibits pancreatic β cell proliferation and insulin hypersecretion in CAM mice. (A and B, E and F, and I and J) Blood glucose [(A), (E), and (I)] and plasma insulin [(B), (F), and (J)] levels during glucose tolerance tests performed on day 3 after adenoviral administration, after sham operation (SO), pancreatic vagotomy (PV), and hepatic vagotomy (HV) [(A) and (B)], vehicle (VEH) and capsaicin (CAP) treatments [(E) and (F)], and SO and midbrain transection (MBT) [(I) and (J)]. (C, G, and K) Pancreatic insulin content of SO, PV and HV mice (C), VEH and CAP mice (G), and SO and MBT mice (K) on day 16 after adenoviral administration. (D and H) BrdU-positive cell ratios in whole islet cells in SO and PV mice (D) and VEH and CAP mice (H) on day 3 after adenoviral administration. Open bars/circles, LacZ mice; closed bars/circles, CAM mice. Data are presented as means ± SEM. *, P < 0.05; **, P < 0.01 versus LacZ mice; #, P < 0.05; ##, P < 0.01 versus SO-CAM [(A) to (D) and (I) to (K)] or VEH-CAM [(E) to (H)] mice, assessed by unpaired t test.

Thus, hepatic ERK activation is likely to transmit signals from the liver to the central nervous system (CNS), resulting in activation of the efferent vagus to the pancreas. To explore afferent signals from the liver to the CNS, we first performed hepatic vagotomy (HV). Contrary to the PV results, however, HV did not affect blood glucose levels (Fig. 2A), glucose-stimulated insulin secretion (Fig. 2B), or pancreatic insulin content (Fig. 2C) in CAM mice. Next, we blocked another type of afferent neuronal signal originating in the liver, the splanchnic nerve, which contains afferent fibers from the hepatobiliary system (19). Capsaicin application to the splanchnic nerve caused selective pharmacological deafferentation with no apparent effects on other nerves, including subdiaphragmatic vagal trunks (fig. S5B), and markedly blunted the glucose-lowering effects (Fig. 2E), glucose-stimulated insulin secretion (Fig. 2F), and increases in pancreatic insulin content (Fig. 2G) and BrdU-positive islet cells (Fig. 2H) in CAM mice. To exclude the possibility that these capsaicin effects are mediated by blockade of the celiac branch of the vagal nerve, we performed denervation of the celiac branch in CAM mice. After celiac vagus dissection, CAM-adenovirus administration still increased glucose-stimulated insulin secretion during glucose tolerance testing (fig. S6A) and pancreatic insulin content (fig. S6B). Collectively, these results suggest afferent signals from the liver to the CNS to be at least partially mediated by afferent splanchnic nerves.

To evaluate CNS involvement, we performed bilateral midbrain transection (MBT), which was confirmed by functional (fig. S7A) and histological (fig. S7B) analyses. MBT markedly blunted the glucose-lowering effects (Fig. 2I), glucose-stimulated insulin secretion (Fig. 2J), and increase in pancreatic insulin content (Fig. 2K) in CAM mice, suggesting CNS involvement in this neuronal pathway. Thus, the mechanism underlying these selective islet responses observed in CAM mice involves interorgan communication mediated by the peripheral and the central nervous system.

To determine whether this neuronal interorgan communication is involved in islet hyperplasia in obesity-induced insulin resistance, we inhibited hepatic hyperactivation of ERK signaling (fig. S1A) by expressing the dominant-negative mutant of MEK1 (DNM) in the livers of ob/ob and HF obesity mice. DNM-adenovirus administration suppressed hepatic ERK phosphorylation (Fig. 3A and fig. S8A) without affecting hepatic p38 MAPK phosphorylation (fig. S4D). In LacZ-adenovirus–treated control ob/ob and HF mice, pancreatic insulin content rose significantly, paralleling obesity development. In contrast, DNM-adenovirus administration blunted these rises in pancreatic insulin content in ob/ob (Fig. 3B) and HF mice (fig. S8B). These findings suggest that activation of the hepatic ERK pathway is involved in pancreatic islet expansion during obesity development.

Fig. 3.

The interorgan communication system originating in the liver is involved in compensatory islet expansion in response to insulin resistance associated with obesity. (A) Hepatic ERK phosphorylation in ob/ob mice on day 7 after administration of 5 × 108 PFU per mouse of recombinant adenovirus containing LacZ or the dominant-negative mutant of the MEK1 (DNM) gene. (B) Pancreatic insulin content of ob/ob mice on day 8 after LacZ (white bar) or DNM (gray bar) adenovirus administration. (C and D) Pancreatic insulin content before and after denervation experiments. Pancreatic insulin content of ob/ob mice on day 14 after SO (white bar) or PV (gray bar) (C) and on day 21 after VEH (white bar) or CAP (gray bar) treatment (D). In (B) to (D), pancreatic insulin content of 5-week-old ob/ob mice were used as baseline controls (black bars). *, P < 0.05; **, P < 0.01 versus LacZ-adenovirus–treated (B), SO (C) or VEH (D) ob/ob mice, assessed by unpaired t test. Data are presented as means ± SEM.

Next, we examined the involvement of afferent splanchnic and efferent pancreatic vagal nerves in pancreatic islet expansion during obesity development. In pair-fed ob/ob mice, blockade of these neuronal signals blunted the normal rise in pancreatic insulin content (Fig. 3, C and D). Furthermore, pancreatic vagotomy suppressed glucose-stimulated insulin secretion in ob/ob mice, resulting in impairment of glucose tolerance (fig. S8C). Taken together, these results suggest that this interorgan communication system is physiologically involved in compensatory islet responses to insulin resistance associated with obesity.

To determine whether targeting of this interorgan communication system affects insulin-deficient (type 1) diabetes, we administered CAM-adenovirus to streptozotocin (STZ)–induced diabetic mice, a murine model of pharmacological β cell loss. Fasting blood glucose levels of CAM-adenovirus–treated STZ mice were dramatically improved (Fig. 4A). This was associated with an increase in the number of BrdU-positive islet cells (Fig. 4B) and an increase in pancreatic insulin content (Fig. 4C). We then administered CAM-adenovirus to Akita mice (20), a murine model of endoplasmic reticulum (ER) stress–induced β cell loss (21), because ER stress in β cells is involved in diabetes development (2224). In these mice as well, CAM-adenovirus treatment lowered blood glucose levels (Fig. 4D), enhanced proliferation of pancreatic islet cells (Fig. 4E), and increased pancreatic insulin content (Fig. 4F). In both these mouse models of insulin-deficient diabetes, CAM-adenovirus treatment greatly improved glucose tolerance by raising serum insulin levels (fig. S9, A and C) but did not significantly alter insulin sensitivity (fig. S9, B and D). In STZ-induced diabetic mice, the glucose-lowering effect induced by CAM persisted for at least 38 days (Fig. 4A), although in Akita mice it was gradually attenuated and was no longer significant by day 36 (Fig. 4D), possibly due to ER stress–induced apoptosis of regenerated β cells. Thus, manipulation of this interorgan communication system may lead to the development of novel therapeutic strategies for insulin-deficient diabetes.

Fig. 4.

Hepatic ERK activation induces pancreatic β cell proliferation and normalizes blood glucose levels in murine models of insulin-deficient diabetes. (A and D) Time course of fasting blood glucose levels in 1.5 × 108 PFU per mouse of LacZ- and CAM-adenovirus–treated STZ (A) and Akita (D) mice. (B and E) BrdU-positive cell ratios in whole islet cells in STZ-LacZ and STZ-CAM mice (B) and Akita-LacZ and Akita-CAM mice (E) on day 3 after adenoviral administration. (C and F) Pancreatic insulin content of STZ-LacZ and STZ-CAM mice (C) and Akita-LacZ and Akita-CAM mice (F) on day16 after adenoviral administration. (A) to (C): white bars/circles, STZ-LacZ mice; gray bars/circles, STZ-CAM mice. (D) to (F): dotted bars/circles, Akita-LacZ-mice; black bars/circles, Akita-CAM-mice. Data are presented as means ± SEM. *, P < 0.05; **, P < 0.01 versus [(A) to (C)] STZ-LacZ mice or [(D) to (F)] Akita-LacZ mice, assessed by unpaired t test.

We have identified a neuronal relay that induces proliferation of pancreatic β cells in response to insulin resistance, indicating that the CNS obtains information from peripheral organs and modulates pancreatic islet mass. Hepatic ERK activation is likely to play an important role in compensatory islet hyperplasia, although it is not yet clear how ERK signaling affects the neuronal pathway. The therapeutic effects we observed in two mouse models of insulin-deficient diabetes are especially noteworthy. Type 1 diabetes mellitus is characterized by progressive loss of pancreatic β cells, leading to a life-long insulin dependency. Recently, it was reported that β cell mass is also decreased in type 2 diabetes (25). Although substantial progress has been made with therapies that are based on transplantation of pancreatic islets (26), immune rejection and donor supply are still major challenges. In this context, therapeutic manipulation of the interorgan signaling mechanism described here may merit investigation as a potential strategy for regeneration of a patient's own β cells. Our results may open a new paradigm for regenerative medicine: regeneration of damaged tissues by targeting of interorgan communication systems, especially neural pathways.

Supporting Online Material

www.sciencemag.org/cgi/content/full/322/5905/1250/DC1

Materials and Methods

Figs. S1 to S9

References

References and Notes

View Abstract

Navigate This Article