There are two types of secretion: exocrine and endocrine. In endocrine secretion, the secreted molecules end up in the blood and they reach their target cells throughout the body via the circulation. By contrast, exocrine secretion does not involve the circulation and the products are released directly into the outside world. Most of the pancreas serves the exocrine function of secreting digestive enzymes into the gut. Less than 1% of the pancreatic tissue is devoted to an endocrine function. The endocrine tissue of the pancreas is organized as cell clusters, called the islets of Langerhans, which are dispersed throughout the pancreatic exocrine tissue and receive a rich vascular (blood vessel) supply ( Figure 1). A pancreatic islet comprises three main cell types. Pancreatic α cells (15%) occupy the islet periphery and secrete glucagon in response to low blood glucose. Glucagon opposes the actions of insulin, thereby increasing circulating glucose levels. Pancreatic δ cells, the least abundant cell type (5%), are dispersed throughout the islet and secrete somatostatin, which has important paracrine effects that suppress insulin and glucagon secretion. The insulin-secreting β cells are the most abundant cell type (80%) and comprise the islet core.

During development, the pancreas arises as an off-branching of early gut tissues, and develops as a set of branching tubules which give rise to clusters of endocrine and exocrine cells. Studies have shown that the cytokine TGF-β plays a major role in the development of pancreatic β cells during development of the organ [ 1, 2], and a paper by Smart et al. in this issue of PLoS Biology [ 3] demonstrates that TGF-β signalling is also critical in the maintenance of β cell functional identity in the adult. Smart and her colleagues were able to show that loss of TGF-β signalling in these cells causes reversion of these cells to an immature differentiated state and resulted in diabetes. Therefore, TGF-β is important for maintaining the functional characteristics of β cells.

Metabolism of glucose is essential for insulin secretion, and inhibition of mitochondrial metabolism blocks insulin secretion [ 15]. Mechanisms of β cell glucose metabolism and metabolic signal generation have been recently reviewed [ 16]. The breakdown of glucose results in the generation of ATP, one of the key molecules fueling cellular reactions. An increased ATP:ADP ratio represents the critical link between mitochondrial metabolism and insulin secretion through its ability to close ATP-dependent K ⁺ (K _ATP) channels and depolarise the cell [ 17] ( Figure 1). K _ATP channels are composed of four pore-forming subunits (K _ir6.2 in β cells) and four accessory sulfonylurea receptor subunits (SUR1 in β cells). The latter are the target of the anti-diabetic sulphonylurea drugs which stimulate insulin secretion by mimicking the effect of glucose to close K _ATP channels. Polymorphism in K _ATP subunits contribute to diabetes susceptibility by altering the biophysical properties of the channels [ 6].

These oscillations reflect a balance between activation of VDCCs (depolarization) and K ⁺ channel activity (repolarization) [ 10]. The depolarizing component predominates at the beginning of the burst, but the resultant influx of Ca ²⁺ during the plateau leads to a progressive Ca ²⁺-induced increase in K ⁺ channel activity. This occurs both via a direct effect on small conductance Ca ²⁺-activated K ⁺ (SK) channels [ 28], and via an indirect effect on K _ATP channels by lowering of the cytoplasmic ATP:ADP ratio due to increased Ca ²⁺ ATPase activity [ 29]. The increase in K ⁺ channel activity eventually becomes large enough to repolarize the β cell, ending the burst. In this scenario, the slow pacemaker depolarization between two successive bursts results from the gradual restoration of [Ca ²⁺] _i and the ATP:ADP ratio until SK and K _ATP channels are again closed and the background depolarizing conductance becomes sufficiently large to trigger a new burst of action potentials.

Glucose produces a concentration-dependent increase in the duration of the bursts at the expense of the silent intervals until eventually, at glucose concentrations beyond 20 mM, uninterrupted action potential firing is observed. This may result from the higher rate of glucose metabolism at high concentrations of the sugar so that Ca ²⁺ influx is unable to lower ATP sufficiently to produce an increase in K ⁺ conductance large enough to trigger membrane repolarization. This model is supported by the ability of tolbutamide, a blocker of the K _ATP channel that has been used for more than 50 years to treat diabetes, to suppress β cell membrane potential oscillations that results in continuous firing [ 29, 30]. Thus, the role of K _ATP channels in the β cell extends beyond merely serving as the glucose-regulated resting conductance. They also contribute to the progressive stimulation of electrical activity and insulin release by supra-threshold glucose levels.

There is an interesting dependence of oscillatory electrical activity on islet integrity and the 10–15 s period typically observed in intact pancreatic islets is for the most part lost in isolated cells maintained in short-term tissue culture [ 30]. This has been attributed to changes in channel expression [ 30], loss of paracrine signalling [ 31], and requirement of cell coupling [ 32]. Indeed, β cells within the same pancreatic islet are electrically coupled [ 33, 34], such that the [Ca ²⁺ ] _i oscillations within different parts of the islet occur in phase ( Figure 2). This synchronization presumably accounts for the observation of pulsatile insulin secretion from individual pancreatic islets [ 26]. Pancreatic β cells contain the gap junction protein connexin-36, ablation of which leads to loss of oscillatory insulin secretion, whereas [Ca ²⁺] _i oscillations in the individual cells is maintained [ 35].

Whereas β cells are electrically coupled to each other, electrical coupling [ 36] and synchronization of the [Ca ²⁺ ] _i oscillations [ 37] between β cells and non β cells and between non β cells appears much weaker if it exists at all. This it at variance with some of the early data looking at the flow of an injected dye between cells which demonstrated the existence of both homotypic (i.e., β to β cell) and heterotypic (e.g., β to a cell) cell coupling [ 38, 39]. However, more recent observations using noninvasive techniques suggest that dye coupling may be less extensive than previously thought [ 40].

In this issue of PLoS Biology, Rocheleau et al. [ 41] have studied the functional significance of electrical coupling between β cells using a novel and ingenious approach. They have used genetically engineered mice in which the K _ATP channel is rendered non-functional—by replacement of specific amino acids—in only some of the pancreatic β cells. This mosaic expression of the transgene (Kir6.2[AAA]) results in functional K _ATP channel knockout in ~70% of the β cells. Somewhat surprisingly, intact islets from mice expressing the transgene exhibited an essentially normal glucose-dependent insulin secretion, when tested in vitro. Importantly, this required the integrity of the pancreatic islet since normal glucose regulation was lost upon dispersion of the islet into single cells. Insulin secretion from individual Kir6.2[AAA] islet cells occurred already at 1 mM glucose, which in normal cells is a non-stimulatory concentration. Moreover, insulin release was not further stimulated with increasing glucose concentrations. The observation that application of the gap junction inhibitor 18a-glycyrrhetinic acid to intact islets mimicked the effect of islet dispersion makes it likely that this difference results from electrical coupling that can only operate within the intact islet.

These data are consistent with the view that the islet functions as a syncytium (that is, an organ that in electrical terms behaves like one cell) where K _ATP channel activity in the individual cells determines the excitability of the entire organ. This is reminiscent of the channel sharing concept originally proposed by Sherman et al. [ 42] to explain the membrane potential oscillations in islets. Work on isolated cells, even when taken from the same animal, indicate a significant heterogeneity in the time courses and magnitude of their responses to glucose stimulation. It seems possible that this reflects a metabolic heterogeneity and that some cells metabolise glucose better than others. This metabolic heterogeneity will result in variable K _ATP channel activity in the individual cells. The report by Rochelau et al. is significant also in this context. They show that all cells within an intact islet respond to glucose in the same way although the K _ATP channel activity in the individual cells ranged between zero and 100% of the normal. The only deviation from normality was a slight shift (~2 mM) towards lower concentration in the glucose dose-response curve. Thus, a lowered K _ATP channel activity in the Kir6.2[AAA] expressing cells will increase excitability in their normal neighbours and vice versa.