Type 1 Diabetes: Chapter 2

Chapter 2 - The Pancreatic Beta-Cell
George S. Eisenbarth, 8/06 update of 2002 chapter by Kirstine Juhl and John Hutton

Introduction
The two most common forms of diabetes in man have very different etiologies and different clinical presentation (1) . Nevertheless the underlying loss of b-cell function has similar consequences in terms of glycemic control and the emergence of devastating long-term complications. In type 1 autoimmune diabetes the loss of b-cells is usually absolute as reflected in the dependence on insulin therapy and the absence of circulating C-peptide. Its sudden onset belies the fact that the underlying loss of b-cell mass is the culmination of many years of gradual and progressive loss of b-cells in the face of autoimmune attack which is first evident with the appearance of autoantibodies to islet proteins in the preceding 1-10 years (see other chapters) (2-6) In the NOD mouse the infiltration of the islets with immune and inflammatory cells that initiates the disease first appears in the islet periphery, affects a subpopulation of islets and is possibly benign or at least kept in check by the presence of regulatory T cells (7-10) . The invasive insulitis seen in NOD mice closer to disease onset may reflect a change in the balance of destructive and protective responses in favor of the former. The histological changes in man are comparatively mild and may reflect the slower progression of the disease or possibly a different immune process. The islet tends to be viewed as the source of autoantigen that supports or initiates the immune attack and ultimately the victim of the crime. However, much more needs to be learned about the reaction of the islet to cytokine mediators of the immune response and about how the b-cell manages to survive so long or replenish its population from progenitor cells in the pancreas. Since the mechanism of autoimmune destruction by effector cells may be mediated by CD4+ cells, and thus indirect, there is also the question of whether the b-cell is uniquely susceptible to oxygen and nitrogen free radicals or cytokine mediators of cell death which may account for the fact that other islet cells exposed to same molecules survive while the b-cell dies. Although we have scant knowledge of b-cell biochemistry and physiology in this regard there is a wealth of information regarding the physiological and pathophysiological responses of the islet to nutrient secretagogues and pharmacological agents. The focus of the following review reflects this balance, however it also aims to emphasize how the b-cell differs from its neighbors and from other endocrine tissues and how it may participate in its own demise in type 1 diabetes. The review also illustrates the challenges faced by investigators wishing to genetically engineer non-b cells for cellular therapy of type 1 diabetes or wanting to introduce specific genes into the b-cell population to afford it greater protection from autoimmune attack.
Physiology of the islet of Langerhans
The endocrine pancreas is arranged in clusters of secretory cells the Islets of Langerhans scattered throughout the exocrine glandular tissue (Fig. 1). In man, the pancreas contains around one million islets that comprise 1-2% of the total mass of the gland. The islets are separated from the exocrine tissue by a capsule made up of connective tissue fibers and by a recent report glial like cells (11) and human islets vary in size from less than 50 up to several thousand cells. Four different endocrine cell types are contained in the islets; b-cells, which produce insulin and constitute 60-80% of the endocrine cell mass, glucagon secreting a-cells (10-20%), somatostatin producing d-cells (~5%) and pancreatic polypeptide secreting PP-cells (<1%) (12) . The cells within islets are critically dependent upon a series of transcription factors during embryonic development and of note, in knockouts of the homeodomain protein nkx2.2, almost all islet cells are replaced with ghrelin producing cells (13) .

Figure 1. The pancreas is located close to the duodenum. The enlarged part of the pancreas shows the exocrine acinar cells next to the islet endocrine cells. The islets consist of four different cell types, which secrete different peptide hormones. The majority of the islets are b-cells that secrete insulin. The a-cells secrete glucagon, d-cells secrete somatostatin and PP cells secrete pancreatic polypeptide.

Most mammalian islets have a b-cell rich core surrounded by a mantle of a-, d- and PP-cells with some controversy as to applicability to human islets. Afferent arterioles enter the b-cell rich medulla where the arterioles split into a branching system of capillaries that traverse the b-cell mass before reaching the mantle of a- and d-cells. Thus, the b-cells are the first endocrine islet cell type to sense blood-borne metabolic changes. Insulin is known to inhibit glucagon secretion and glucagon to stimulate insulin secretion. Somatostatin will inhibit the secretion of both insulin (Somatostatin's inhibition of insulin secretion is reported to result from its inhibition of glucose metabolism of the beta cell (14) ) and glucagon, hence the anatomical relationship of the islet cells is critical to the function of the organ. The islets are innervated by autonomic nerve fibers, containing both parasympathetic and sympathetic nerves that enter the islets and terminate in proximity to the endocrine cells.
The Beta Cell as a Neuron-like Cell
Though islet beta cells are derived from endodorem and not ectoderm they share many properties with neurons that probably relate to their function in terms of secreting a peptide hormone stored in dense core secretory granules and having microvesicles with "neurotransmitters". The neuron-like properties include presence of microvesicles containing molecules such as GABA gamma-Aminobutyric acid (GABA) (15), an inhibitory neurotransmitter, expression of enzymes such as Glutamic acid decarboxylase (GAD, both GAD65 and GAD67), IA-2 (ICA512) (16) , IA-2beta (phogrin) (17), surface expression of complex neuronal/Oligodendrocyte gangliosides such as GQ gangliosides (e.g. target of monoclonal A2B5) and the targets of monoclonals 3G5 (18) and R2D6 (19), and expression of Type 2 monoamine vesicular transporters(VMAT2) (20). These shared neuronal/neurendocrine properties are of importance in considering the pathophysiology of autoimmune beta cell destruction (e.g. Though GAD65 autoantibodies are characteristic of type 1 diabetes of man, only beta cells and not the many other cell types that express GAD65 are destroyed in type 1A diabetes and in contrast in Stiff Man Syndrome neurologic disease predominates in association with very high levels of GAD65 autoantibodies) and recently for the in vivo detection of islets, potentially for determining beta cell mass. A recent report has utilized [(11)C]Dihydrotetrabenazine ([(11)C]DTBZ) that binds specifically to VMAT2, with PET scanning to estimate beta cell mass in the BB type 1 diabetes rat (20) . It is likely that many of the technologies developed to interrogate the brain (e.g. MRI spectroscopy) will be applicable to studies of islet beta cells, if the goal is determination of approximate mass rather than direct visualization of the relatively small islets, given the shared biochemistry of neurons and islets.
The Biosynthesis of Insulin
The islet b-cell is the principal cell in the adult mammal able to transcribe the insulin gene. The insulin receptor by comparison is widely distributed even on cells that are not thought to be insulin responsive and in the case not even exposed to significant concentrations of the hormone. This may reflect the evolution of the molecule from primarily a neurotransmitter to an endocrine function (21) .
Insulin mRNA is translated to a pre-pro-insulin peptide, which is rapidly processed in the rough endoplasmic reticulum to pro-insulin by removal of the N-term signal peptide. Proinsulin contains the A and B chains of insulin linked by the C peptide, a structure that helps to align the disulfide bridges that are generated between the A and B chain. Proinsulin is transported through the Golgi apparatus and thereafter further processed in the maturing granule to insulin by excision of the C peptide by the endopeptidases, PC1 and PC2 (prohormone convertases), and carboxypeptidase H (for more details see chapter with Hutton). The bioactive insulin molecule consists of an A and B chain (21 and 30 amino acids, respectively), linked intramolecularly by disulfide bridges. Most mammalian insulin bind Zn co-ordinately forming hexamers and can form large crystalline arrays which promote condensation and aggregation of the hormone in the acidic core of the secretory granule. Proinsulin although forming Zn hexamers is prevented from further aggregation by the highly charged C-peptide, a physical property that may be important in retaining its solubility in the proximal elements of the secretory pathway. In the mature secretory granule C peptide exists in equimolar amounts with insulin in the granule and is co-secreted thus making it a useful marker for b-cell function in diabetics receiving insulin therapy.
Transcription of the insulin gene is modulated by nutrient secretagogues, a variety of cytokines and by insulin itself. Insulin signal transduction enhances transcription of the insulin gene through phosphorylation of insulin receptor substrate (IRS)-1 and -2, and activation of phosphatidylinositol-3-kinase (PI3K) and it has been postulated that insulin plays a feed forward role in insulin biosynthesis ensuring that the stores of insulin are always replenished (22) . Exogenous insulin is also shown to hyperpolarize b-cell plasma membrane in mice by insulin-induced activation of the KATP channels possible through the PI3K thereby turning off the islet insulin secretion (23) . Glucose is also shown to stimulate the production of proinsulin through rapid activation of translation of a pre-formed pool of mRNA (24). Given the high rates of proinsulin synthesis and the need for proper folding of proinsulin within the endoplasmic reticulum it has been hypothesized that pancreatic beta cells are particularly susceptible to apoptosis induced by endoplasmic reticulum(ER) stress (25) . Wolfram syndrome (DIDMOAD: Diabetes Insipidus/Diabetes Mellitus/Optic ptrophy/Nerve Deafness) results from mutations of the WFS1 gene which is a transmembrane protein of the ER regulated by the unfolded protein response and with a role in ER calcium transport (25) .
The proinflammatory cytokines IL-1b, TNFa and INFg modulate transcription of the insulin gene. Culturing human b-cells 48 or 72 h with single cytokines does not affect the cellular content in insulin or proinsulin whereas the combination of several cytokines (IL-1β + IFNγ) disproportionally elevates medium proinsulin levels (26) . This is caused by a conserved proinsulin synthesis and a slower hormone conversion possibly contributed to lower expression of PC1 and PC2 convertases. These cytokines mediates signal transduction involving binding to specific receptors, through MAPK and SAPK and mobilization of diverse transcription factors: NFkB, AP-1 and STAT-1 involved in b-cell apoptosis (27;28) .
Physiological regulation of islet hormone secretion
The blood glucose level is maintained within a narrow range around 5 to 7 mM in the fasting state mainly by the combined and reciprocal action of insulin and glucagon (29) . Essentially, when the blood glucose concentration is elevated after a meal, the b-cells are stimulated to secrete insulin. Conversely, when blood glucose levels are low the a-cells are stimulated to secrete glucagon, which leads to glycogenolysis and gluconeogenesis in the liver and therefore release of glucose to the blood. Between meals, when the plasma glucose decreases, the release of the neurotransmitter norepinephrine and the neuropeptide galanin from the sympathetic nerves will directly activate glucagon release and inhibit insulin release (30). During a meal or the postprandial state, the parasympathetic nerves potentiate glucose-induced insulin release from islet b-cells. This is achieved by the release of the neurotransmitter acetylcholine and the neuropeptides pituitary adenylate cyclase activating polypeptide (PACAP), vasoactive intestinal polypeptide (VIP), GLP-1 (31-34) and GIP (30). The innervation of the islet and its microanatomy is probably important in the orchestration of these responses.
Somatostatin inhibits both insulin and glucagon secretion and potentially serves as a modulator of islet hormone release (35). However, it is unclear whether it has a physiological role given the vascular anatomy of the islet in which the secreted products from the a- and d-cell enter the blood circulation without reaching the b-cells. The physiological role of islet PP is unknown, but appears not to affect the secretion of the other islet hormones except a possible inhibition of islet insulin release (36-38) . Rare cells within islets secrete ghrelin, with ghrelin primarily produced in the gastrointestinal tract and ghrelin under experimental conditions can influence insulin secretion and is reported to increase IA-2 beta but not IA-2 message in islets and inhibition of IA-2 beta with RNAi decreased ghrelin inhibition of insulin secretion (16;39) .
Much of the physiological regulation of insulin secretion can be reproduced in vitro and a wealth of information has been generated from the study of isolated islets, which appear to behave as autonomous organs. Such studies have led to the classification of insulin secretagogues into three groups, initiators, potentiators and inhibitors. Glucose is the most important initiator in monogastric animals, but other carbohydrates, amino acids and fatty acids alone can also initiate insulin secretion. Pharmacological agents like the clinically useful sulphonylureas tolbutamide and glibenclamide also initiate insulin secretion in a glucose-independent manner. Arginine initiates insulin secretion by depolarizing the plasma membrane because of its transport in a positive charge form. It triggers insulin release by increasing [Ca2+]i through a KATP channel-independent pathway (40;41) . Potentiators of insulin secretion do not initiate insulin secretion by themselves but enhance stimulation of secretion in the presence of an initiator. Examples of potentiators are the intestinal glucoincretin hormones glucagon-like peptide 1 (GLP-1) (42) and gastric insulinotropic polypeptide (GIP), which provide fine regulators of insulin secretion in the post prandial state along with neurotransmitters. Other neurotransmitters and hormones such as somatostatin, galanin, adrenalin and even endocannabinoids (through CB2 receptors (43) reduce insulin secretion (Fig. 2).

Figure 2. Examples of potentiators, initiators and inhibitors of insulin secretion from the pancreatic b-cell. Abbreviations used: GIP, gastric insulinotropic polypeptide; VIP, vasoactive intestinal polypeptide; GLP-1, glucagon-like peptide 1.

Islet Pathophysiology in type 1 diabetes
Type 1 diabetes is characterized by the appearance of circulating antibodies targeted at b-cell proteins, which appear many years before the onset of clinical disease. Proteins such as insulin, GAD, IA2, IA-2 beta (phogrin) and CPH (carboxy-peptidase H) are specifically targeted but as yet it is not settled if any one (or unknown targets) of these molecules is primary or dominant in the autoimmune response. In the NOD mouse model there is accumulating evidence that insulin may be a primary autoantigen with knockouts or alterations of specific insulin sequences preventing diabetes, while knockouts of GAD65, IA-2, IA-2 beta (and even both IA-2s) do not change the course of development of NOD diabetes (44-46). A recent study of lymphocytes from pancreatic lymph node of patients with type 1 diabetes also implicated insulin (47). Insulin is the only molecule in this group that is specific to the b-cell though the others have the common feature of being associated with regulated pathway of secretion. Insulin and CPH are secreted molecules and to some extent remain on the cell surface following exocytosis. By contrast the dominant epitopes seen by antibodies reactive to GAD, IA2 and phogrin are intracellular. It is presently unknown whether the b-cell is targeted directly or indirectly, whether the antigens are presented by the living cell or through death or antigen shedding and why the a-cell survives in spite of sharing many autoantigens with the b-cells and being bathed in the same milieu of soluble immune effector molecules. One hypothesis is that the b-cell participates in its own destruction by increased presentation of antigens in response to glucose, a situation that worsens as b-cell mass decreases and the remaining cells compensate by increased transcription and translation of secretory pathway proteins (48) . b and a-cells have a differential sensitivity to cytokines in that inhibition of glucose-stimulated insulin release by IL-1 is reversible, whereas the effect on glucose-modulated glucagon release is not (49) . The b-cells are more sensitive to cytokines than the other three endocrine cell types in the islets. Cytokine-induced free radicals in b-cells such as NO catalyzed by the inducible nitric oxide synthase (iNOS) may be involved in b-cell-specific destruction in type 1 diabetes (50). NO is an important mediator but not the sole mediator of cytokine-induced cytotoxicity in b-cells as shown by the fact that inhibition of NO production in human islets did not prevent cytokine-induced apoptosis, a-cells do not express NO in response to cytokines. b-cells but not a-cells induce the expression of heat shock protein 70 (hsp70), heme oxygenase and MnSOD (manganese superoxide dismutase) upon exposure to IL-1 (51). Native b-cells have been shown to possess low scavenging potential for oxygen-derived free radicals and overexpression in b-cells of scavenger or antioxidant enzymes such as MnSOD, catalase and glutathione peroxidase has been shown to increase their survival after exposure to NO and reactive oxygen species (50) .Production of oxygen free radicals mediated by macrophages can damage b-cells directly resulting in type 1 diabetes in NOD mice (52). Superoxide dismutase and catalase protected isolated b-cells against alloxan-induced diabetes in vivo indicating a role for superoxide radicals and hydrogen peroxide in the toxicity of alloxan (53).
Studies with isolated islets have provided models of how mediators of the immune response may damage the islet in the autoimmune response in type 1 diabetes. The macrophage cytokine IL-1b induces an initial phase of functional stimulation in rodent pancreatic islets, which is followed after 4-7 hours by a progressive inhibition of (glucose-induced) insulin release and eventually overt damage to the b-cell caused by production of NO (54;55). The two cytokines, TNFa and INFg potentiate the IL-1b-induced production of toxic NO and oxygen free radicals which inhibits insulin secretion (56;57). This has led to the concept that IL-1b in combination with INFg and TNFa plays an important role for b-cell dysfunction and death. The general process of cytokine-induced b-cell 'de-differentiation' with impairment of some of the most differentiated functions of b-cells, such as the preferential mitochondrial metabolism of glucose and the biosynthesis and release of insulin is paralleled by the activation of proteins related to cell survival, such as heat shock proteins and antioxidant enzymes (58) . There are important species differences in b-cell response to IL-1. IL-1 has stimulatory effects on human islets with no formation of NO compared to rodent islets, which produces NO when exposed to IL-1. Human islets cultured in the presence of IL-1b alone exhibit a more prolonged stimulatory phase, which may last up to 48 hours (55) . But again, with prolonged exposure of human islets to IL-1b + TNFa + INFg an inhibition of human islet function is also observed. This difference between species is probably due to better capacity of human islets to scavenge oxygen free radicals and to the higher content of hsp70, catalase and superoxide dismutase in these cells (59) . Hsp70 has a direct anti-apoptotic role by inhibiting protein aggregation, decreasing formation of oxygen free radicals and blocking effector caspases and it could also decrease necrosis by preventing cellular ATP depletion.
The cytokine-induced b-cell de-differentiation is probably associated to cytokine-induced down-regulation of pancreas duodenum homeobox gene-1 (PDX-1) and Isl-1. PDX-1 is a crucial gene for b-cell development and for the maintenance of its differentiated phenotype (see chapter with Jensen). The decreased expression of PDX-1 together with the inhibition of Isl-1 expression could contribute to the observed decreased expression of insulin and glucokinase mRNA which depend on PDX-1 for their regulated transcription. While IL-1b inhibits PDX-1 and Isl-1 expression and up-regulate c-myc it does not trigger b-cell apoptosis (59).
Plasma type II secreted phospholipase A2 (PLA2) may be closely involved in the pathophysiology of acute pancreatitis. The serum levels of type II PLA2 has been shown to correlate with the severity of the disease (60) . Also, an inhibitor of type II PLA2 protects the pancreas against tissue damage when pancreatitis is induced in vitro (61) . Interestingly PLA2 enzyme isoforms are also produced by inflammatory cells and by the islets themselves (see below). As in the case of the soluble immune effector molecules we have little information on the actual concentrations to which the b-cells are exposed and the actual in vivo response in terms of regulation of receptors and signaling pathways.
The Stimulus-Secretion Coupling of the Pancreatic b-Cell
The pancreatic b-cell is unusual from the biochemical standpoint in that it has a high rate of glycolytic and oxidative metabolism and that its glucose metabolism is largely insulin-independent. Most-strikingly it responds to extracellular glucose concentrations by increased metabolic flux even at concentrations of the sugar as high as 25 mM. The latter provides a unique mechanism that couples the availability of metabolizable nutrients to the control of secretory, translational, and transcriptional processes without the mediation of ligand-specific cell surface receptors. It may also be the Achilles heel of the cell in that intracellular proteins are exposed to higher concentrations of glucose and reactive metabolites than would otherwise occur in a typical cell that restricts glucose entry and its metabolism to meet its energetic demands. Within the secretory granule insulin is stored complexed with zinc and the beta cell has mechanisms to take up zinc (62) and transport it into the secrtory granule (ZnT-8[expressed only in beta cells], a cation diffusion facilitator) (63). Two other transport systems that are abundant in islet beta cells and important for islet physiology are GPR40 (G protein-coupled receptor 40 a free fatty acid receptor) and 4F2 coupled amino acid transporters. GPR40 mediates the insulin secretory stimulation of fatty acids such as linoleic acid reportedbly through cAMP and protein kinase A activity, and inhibition of delayed rectifying voltage gated K+ channels (64-66). The 4F2 heavy chain (CD98 heavy chain) is a component of heterodimeric amino acid transporters that is highly represented on the surface of islet beta cells (67). Of note this molecule is also an "activation" antigen and present at high levels on malignant cells (68;69) .
The b-Cell Glucose Sensor
The b-cell glucose sensor consists of the combination of two molecules, which have a restricted tissue distribution, namely Glut2 and glucokinase (GK). Glucose enters the b-cell through the glucose transporter Glut2 (Km ~17 mM) and is quickly phosphorylated by GK to glucose 6 phosphate (G6P) (Fig. 3) (70). Glut2 expression changes in diabetes and hyperglycemic states, which seems to underline its importance in the response of the b-cell, however there is great species variation in the expression of this molecule and little evidence per se that glucose entry is rate limiting to glucose metabolism in the b-cell. GK on the other hand, although present in other tissues such as the liver and kidney appears to be driven by a pancreatic specific promoter (71-73). Its high expression and accompanying downregulation of low Km forms of hexokinases (74-76) create a situation where kinetics of the enzyme dictates the kinetics of the generation of G6P in a cellular context. There has been considerable debate as to whether the b-cell has a specific glucose 6 phosphatase (G6Pase) and whether it participates with GK in a regulatory substrate cycling activity (77-79). The b-cell expresses islet G6Pase related protein (IGRP) a b-cell specific homolog of the liver G6Pase catalytic subunit (55% identity) but the molecule is apparently not catalytically active (80). Of note IGRP is a major target of CD8 T lymphocytes of the NOD mouse (47). Other biochemical properties of the b-cell which have been implicated in its ability to sense metabolic fuels include the absence of lactate dehydrogenase and thus an inability to re-oxidize cytoplasmically generated NADH (81) and the enhanced expression of the mitochondrial glycerophosphate shuttle (82;83) which may be involved in the transport of reducing equivalents in to and out of the mitochondria. Anaplerosis (the refilling of Krebs cycle intermediates) is shown to be implicated in the KATP-independent pathways of insulin secretion perhaps via malonyl-CoA formation and lipid esterification processes or through a pyruvate/malate shuttle generating NADPH (84). Anaplerosis requires conversion of pyruvate into oxaloacetate by pyruvate carboxylase an enzyme that is abundant in islet tissue (85) . Consistent with the view that anaplerosis is an essential component of b-cell signaling is the fact that the dose dependence of insulin release highly correlates with the accumulation of citrate, malate, and citrate-derived malonyl-CoA in the cell (84). The fact that the association of rise in citrate and a-ketoglutarate and initiation of insulin secretion does not require Ca2+ further indicates that anaplerosis is an early event of b-cell activation(46). A net result of this biochemical machinery is that rates of glucose metabolism are strongly correlated to changes in the redox state of pyridine nucleotides (particularly NADPH (86;87) and the ratio of ATP/ADP in the cell both which may act as intracellular signals for ionic and other events in the cell (Fig. 3) (88) .
The b-cell is electrically excitable and when stimulated with glucose it shows oscillations in membrane potential characterized by bursts of 5-15 sec duration. Increasing glucose lengthen the burst and shortens the interval leading to continuous spiking over a base of DY <-40 mV (89;90). The b-cell resting membrane potential of about -70 mV is mainly determined by the activity of ATP-sensitive K+ channels (KATP) (91). When the ATP/ADP ratio increases as a result of glucose metabolism, the KATP channels in the plasma membrane close (92-98). This causes the cell membrane to depolarize due to the presence of an inwardly directed cation current that is dominant in the absence of a reduced KATP-conductance. The nature of this current is currently unknown but may be a member of the trp family that carries Na+ and Ca2+ ions (99). The resulting membrane depolarization leads to opening of the voltage-dependent L-type Ca2+ channels in the plasma membrane. The resultant Ca2+ influx elevates the intracellular free Ca2+ concentration [Ca2+]i and initiates exocytosis of insulin-containing secretory granules (Fig. 3).

Figure 3. Stimulus-secretion coupling in the pancreatic b-cell. Glucose is transported across the plasma membrane through the GLUT2 transporter. Glucose is metabolized through the glycolysis and Krebs (TCA) cycle. The increased ATP/ADP ratio leads to closure of the ATP-sensitive K+ channel in the plasma membrane, membrane depolarization and opening of the voltage-dependent Ca2+ channels. The resulting Ca2+ influx stimulates release of the insulin-containing granules by exocytosis. [Note: Trp is the transient receptor potential gene family that first was cloned from Drosophila where it function in photoreception, but which seems to include Ca2+ store-operated channels and inositol 1,4,5-trisphosphate-activated channels.]

Besides the effect of glucose on the KATP channels glucose is also shown to affect the stimulus-secretion coupling independently of the KATP channels thereby amplifying the pathway of glucose-induced insulin secretion (100) . This has been shown by the fact that glucose dose-dependently increases insulin secretion when the cell is depolarized and KATP channels cannot be closed (diazoxide plus high K+) (101). Under conditions when the KATP channels are closed by sulfonylureas and therefore the b-cell plasma membrane is depolarized and insulin secretion is increased, glucose is shown to be able to further amplify insulin secretion dose-dependently. This KATP channel-independent (amplifying) effect of glucose on insulin secretion is Ca2+-dependent but not mediated by any further rise in [Ca2+]i (101). The amplifying pathway serves to augment the secretory response induced by the triggering signal and creates dose-dependent secretory response in the 5-25 mM concentration range.
Glucose may also regulate insulin secretion by a Ca2+-independent mechanism (102-104) . Under stringent Ca2+-deprived conditions when protein kinases A and C are activated simultaneously, glucose stimulates insulin release in the absence of an elevation of [Ca2+]i. This Ca2+-independent amplification of glucose-induced insulin release may require GTP late in stimulus-secretion coupling (105;106) . This has been interpreted in terms of two different mechanism of exocytosis, one driven by Ca2+ and the other by GTP. Others argue that glucose-induced insulin release is mainly regulated by the two Ca2+-dependent pathways (rise in b-cell [Ca2+]i and increase in Ca2+ efficacy) without significant contribution of a Ca2+-independent mechanism. In the following, some of the key molecular components in the b-cell stimulus-secretion coupling will be described.
The ATP-Sensitive K+ Channel Complex
KATP channels are expressed in different cell types including islets, heart, muscles and ventromedial hypothalamus (VMH) and serve to couple cell metabolism to membrane excitability. They are composed of a pore-forming complex consisting of subunits, a specific K+ channel (Kir6.2) surrounded by regulatory sulphonylurea (SUR) binding subunits. See reviews (107-109) .
SUR (Mw 140 kDa) is a member of the ABC-transporter superfamily and Kir6.2 (390 amino acids, 35 kDa) a member of the inward rectifier K+ channel superfamily. Four subunits of Kir6.2 and SUR1 each constitute the functional channel complex in islets (110;111) . KATP channel activity is modulated by a range of compounds and intracellular metabolites. Channel activity is reduced by Mg-ATP and sulphonylureas, the most commonly used class of compounds in the treatment of type 2 diabetes. ADP and diazoxide conversely activate KATP channel activity. Changes in the b-cell ATP/ADP ratio are an important physiological regulator of channel activity with an increase in the blood glucose concentration leading to an enhanced ATP/ADP ratio and consequently channel closure. Recently, KATP channels have also been detected in a- and d-cells (112-115). Whereas the functional consequence of channel closure in the d-cell is similar to that in the b-cell (increased hormone release), inhibition of channel activity in the a-cell leads to inhibition of glucagon release (115).
The physiologic importantance of this channel in man has become abundantly clear in the past several years with the discovery that approximately ½ of neonatal diabetes results from activating mutations of the Kir6.2 molecule. There is a hierarchy of disease dependent on the severity of the functional effect of the mutations leading to transient neonatal diabetes, permanent neonatal diabetes, and finally a syndrome with mental retardation and neonatal diabetes (DEND syndrome: developmental delay, epilepsy, and neonatal diabetes (116) ) (108;109;117-119). Paternal inactivating mutations of the genes for Kir6.2 and SUR1 combined with loss of the corresponding maternal chromosomal (11p15) region in focal areas of the pancreas leads to focal islet lesions,while autosomal inheritance of mutations results in diffuse pancreatic lesions, and both cause hyperinsulinemic hypoglycemia (120;121). Another gain of function mutation that apparently causes hypoglycemia through its influence on the K ATP channel, are mutations of mitochondrial glutamate dehydrogenase (122). It is thought that glutamate dehydrogenase mutations lead to increased ATP in islet beta cells, closing the K sensitive ATP channel and thus inappropriate insulin secretion and hypoglycemia. Abnormalities of mitochondria are increasingly recognized as central to beta cell abnormalities and in particular there are a series of mitochondrial mutations that lead to diabetes and more recently reports that mitochondrial "stress" and abnormalities related to mitochondrial "uncoupling" are important for insulin secretion defects in type 2 diabetes (123;124) .
The b-Cell Ca2+ Channels
Membrane depolarization, resulting from KATP channel closure leads to elevation of cytosolic Ca2+ via voltage-gated Ca2+ channels in the plasma. The influx of Ca2+ in turn regulates several steps in exocytosis, such as the size of the readily releasable vesicle pool, the fusion event, and expansion of the fusion pore (125). The Ca2+ concentration required for initiating insulin secretion in different experimental systems has been estimated in the range of 10-30 mM (126;127). The resting level is in the submicromolar range.
Pancreatic b-cells are found to express N-, P/Q- and L-type Ca2+ channels, with the major contribution to Ca2+ influx mediated by the L-type channel (large and long-lasting) (91). The L-type channels are characterized by the requirement of a strong depolarization to activate. They inactivate slowly, which is seen in a patch-clamp experiment during a voltage-clamp depolarization to 0 mV where channel inactivation is maximal (Fig. 4) (128). The L-type channels are by themselves inactivated by Ca2+ ions at the inner side of the membrane, providing a negative feedback mechanism that limits Ca2+ entry into the cell (129). There is also a voltage-dependent component in the inactivation The b-cell L-type Ca2+ channel activity is blocked by dihydropyridines e.g. nifedipine and stimulated by the cyclic AMP/protein kinase A pathway. Elevated cAMP levels lead to a reduced rate of Ca2+ channel inactivation whereas the peak current is only moderately increased (130).

Figure 4. Ca2+ current in a mouse b-cell is mediated by L-type Ca2+ channels. The peak amplitude and the integrated Ca2+ current are depicted resulting from a membrane depolarization to 0 mV during a voltage-clamp experiment.

Molecular Motors and SNARES
When the Ca2+ channels in the plasma membrane open and Ca2+ flows in the cell, the onset of exocytosis follows less than 50 ms later. This latency is shorter than the time required for Ca2+ to equilibrate in the cytosol, which suggests that the granules are located in the vicinity of the Ca2+ channels and thus sensitive to local [Ca2+]i changes. Indeed, areas with a high density of Ca2+ channels are found to co-localize with the granules and represent 'hot spots' of secretion(78). A mouse b-cell contains about 13,000 secretory granules (131) but only a small fraction (50-75 granules (132) ) in the readily releasable pool (RRP) are available for immediately release (133;134) . The remainder referred to as the reserve pool need to be mobilized into the RRP before they can undergo exocytosis. The number of granules released seems to be dependent on the phosphorylation state of key proteins. Activation of protein kinase A and C (PKA and PKC) or inhibition of protein phosphatases is required for a large exocytotic response (135). The effects of these kinases are additive, which suggests that phosphorylation of different regulatory proteins are involved. Glucose-stimulated insulin secretion in vivo is typically biphasic and characterized by a steep transient first phase is followed by a gradually developing second phase (Fig. 5). In one model it is explained as the first phase represents release of granules in RRP, triggered by Ca2+ influx, and the second phase most likely reflects the ATP-dependent recruitment and release of granules from the reserve pool (133). The electrophysiological response is also biphasic so this conclusion is open to debate.

Figure 5. Glucose-induced insulin secretion is characterized by a rapid first phase and a slower second phase insulin released to the bloodstream. Abbreviation used: iv, intra venous.

Mobilization of granules may involve physical translocation within the b-cell and/or chemical modification (136) and docking of granules near the plasma membrane. Docked granules need to be primed in order to enter the RRP. The insulin granules are transported from the reserve pool to the plasma membrane initially along microtubules and then along the microfilament networks of the cytoskeleton associated with the plasma membrane (137). An ATP-dependent motor protein such as kinesin is thought to provide the necessary motive force for transportation. However, the exact mechanism and its regulation involved in the transport of granules towards the plasma membrane remains largely unknown (138).
The processes by which granules in close proximity to the plasma membrane bind and fuse to it can be divided into three phases docking, priming and fusion based on electrophysiological and secretion studies performed on various neurosecretory tissues and in combination with biochemical and molecular genetic analysis.
Docking describes the initial reversible interaction of granules with the plasma membrane and constitutes a pool of unprimed granules in close proximity to the exocytotic site. There is little morphological evidence of the existence of granule docking, but the presence of this pool in the b-cell is suggested from the speed by which RRP is replenished following addition of ATP in the cytosol. The next priming step may involves ATP-hydrolysis mediated in part by the ATPase N-ethylmaleimide-sensitive factor (NSF), which is activated by the soluble NSF-attachment factor (a-SNAP) (139). NSF functions as a molecular chaperone to activate SNAp REceptor (SNARE) proteins destined for the fusion complex (Fig. 6) (138). Another possible ATP-dependent event in priming is the synthesis of phosphatidylinositol (4,5)-bisphosphate (PIP2), which is thought to be important in the recruitment and modulation of proteins implicated in the fusion apparatus - namely the Ca2+ binding synaptotagmin and calcium-dependent activator protein for secretion (CAPS) (125;140) . Multiple different molecules influence insulin granule exocytosis. The transcription factor HNF-1 when mutated is a cause of a severe form of MODY (Maturity Onset Diabetes of Youth) and in addition to its many effects upon islets and insulin secretrion, the molecule collectrin is regulated downstream of HNF-1, and collectrin under or overexpression leads to decreased or enhanced insulin secretion (141). Insulin secretion is deficient in a mouse with a mutant Akt/PKB kinase and this protein regulates exocytosis in a model system (142) .

Figure 6. Docking, priming and fusion of insulin-containing granules. Granules from the reserve pool needs to be mobilized in order to enter RRP prior to release. The insert illustrates the three primary stages of insulin exocytosis at the plasma membrane, docking, priming and fusion. The granule is tethered at the plasma membrane during docking. Thereafter the a-SNAP activates NSF, which hydrolyses ATP thereby activating the SNARE proteins. A second ATP-dependent step in the priming process could be PI-4kinase phosphorylation in the generation of PIP2 (after PI-5kinase phosphorylation). Synaptotagmin is proposed to act as a Ca2+ sensor and a fusion clamp preventing fusion until a Ca2+ signal arises. When the granule is primed, Ca2+ is the only requirement for granule fusion with the plasma membrane.

Fusion of granules
Conserved protein families involved in fusion include the SNAREs, the Rab GTPases, and the Sec1/munc18-related proteins (also referred to as SM proteins) (143). Granuphilin is reported to be in dense core vesicles of beta cells and binds to GTP-bound Rab27a (144). Rab27a, on the granule membrane is involved in the regulation of exocytosis of secretory granules. Granuphilin also directly interacts with plasma membrane bound SNARE protens (144). Insulin granules docked to the membrane are reduced in granuphilin deficient beta cells (145) despite granuphilin null mice having augemented insulin secretion (145). The SNARE proteins were originally identified as synaptic proteins, but are now generally accepted as universally involved in the core machinery in all intracellular fusion events. The primed granules are ready to fuse with the plasma membrane in an ATP-independent process when the intracellular Ca2+ concentration [Ca2+]i is sufficiently elevated.
Synaptobrevin (VAMP-2, synaptic vesicle-associated membrane protein) is located on the b-cell granule as a v-SNARE, and syntaxin 1 and SNAP-25 (synaptosome-associated protein of 25 kDa) are located on the plasma membrane as t-SNARE (Fig. 6) (146;147) . Synaptobrevin, SNAP-25 and syntaxin 1 assemble in a 1:1:1 stoichiometry to form a complex that drives membrane fusion. Helical portions in the SNARE complex bundle together (two helical domains from SNAP-25 and one from both syntaxin 1 and synaptobrevin) and form an extended coiled-coil rod-like structure (see note), which brings the granule into close proximity with the plasma membrane, thereby overcoming the free energy barrier for fusion (Fig. 6) (148) .
[Note: Coiled-coil structure is composed of several a-helices that interact through hydrophobic residues and stabilizing electrostatic interactions between the side chains. This results in a very stable structure resembling a strong rope.]
It has been postulated that munc18 functions in a late stage in the fusion process in chromaffin cells, where its dissociation from syntaxin 1 determines the kinetics of postfusion events (149) . However, a munc18 null mutation leads to a selective defect in the docking of LDCVs at the target membrane and overexpression of munc18 leads to an increase in number of fusion-competent vesicles without affecting the kinetics of vesicle fusion. This suggests that munc18 is also important in the docking step before trans-SNARE complexes have assembled.
The essential role of these proteins in neurosecretion is illustrated by the potent inhibition of insulin secretion by various botulinum and tetanus toxins that specifically cleave the different SNARE proteins. Studies with botulinum neurotoxin A which cleaves VAMP-2 and SNAP-25 demonstrate that VAMP-2 is absolutely necessary for insulin exocytosis whereas the action of SNAP-25 could be partially reversed by higher levels of Ca2+ or cAMP potentiation (150) . Botulinum neurotoxin C1 mainly cleaves syntaxin and reduces K+-induced insulin release by 95% but glucose stimulated insulin release only by 25% pointing to further complexity in the exocytotic mechanism (151) .
The fusion event is highly Ca2+-dependent (>10 mM), probably due to regulatory proteins that function as Ca2+ sensors. Synaptotagmin is the best characterized exocytotic Ca2+ sensor with two Ca2+-binding regions, the C2 domains (see note) C2A and C2B, each sharing homology with the C2 regulatory domain of protein kinase C (PKC). The C2A domain is responsible for the Ca2+-dependent insertion of synaptotagmin into the b-cell granule membrane at low micromolar (5 mM) free Ca2+ and the binding of synaptotagmin to the core complex protein syntaxin 1 (138;152) . CAPS is another Ca2+-binding protein thought to be important for Ca2+-triggered fusion from neuroendocrine cells, including the b-cell (138) . It contains a pleckstrin homology domain (see note) that exhibits binding specificity towards PIP2, which makes it a specific target for phospholipid mediators formed during priming. The lipid specificity of CAPS is switched at elevated Ca2+ concentrations and is thought to help disrupt the plasma membrane bilayer to permit fusion (153) . Synaptotagmin is also found to have an inhibitory function in exocytosis. It has been suggested to lock the fusion core complex by preventing a-SNAP from binding until a stimulatory Ca2+ signal arises (154;155) . Other components of the fusion process include syntaphilin that acts as a syntaxin clamp in regulating assembly of the SNARE complex during membrane fusion events (156) and NSF that may be necessary to dissociate SNARE complexes formed within a single membrane in favor of complexes between membranes (trans-SNARE complex) (157;158) .
[Note: A C2 domain contains about 120-140 amino acids and has been found in single or multiple copies in more than 60 proteins. It was first discovered in PKC as a Ca2+-dependent, phospholipid-binding domain, but now several other C2 domains have been shown to mediate protein-protein interactions and Ca2+-independent binding to proteins and phospholipids 173,174.
Pleckstrin homology domain mediates membrane association of kinases with their target molecules, in some cases by interaction with the headgroup of a phosphoinositide lipid 174.]
Rab GTPases represent a large family of homologous Ras-like GTP-binding proteins that direct the vectorial movement of secretory vesicles. Four isoforms of Rab3 (Rab3A, -B, -C, and -D) have been identified so far. Rab3A is associated with dense-core insulin-containing secretory granules and thought to play a negative role in the fusion process by preventing Ca2+-dependent exocytosis (159) . Rab3A activation is shown to inhibit Ca2+-evoked exocytosis in b-cells by possibly binding to calmodulin at low stimulatory [Ca2+]i (160) . Calmodulin is a key Ca2+ sensing protein and this may be a way for Ca2+ to act in concert with GTP in the insulin exocytotic mechanism (provides a downstream connection between an intracellular secondary messenger (Ca2+) and the exocytotic machinery of the pancreatic b-cell). It is postulated (161) that it is a way to direct calmodulin to the cytoplasmic face of a b-granule membrane. When [Ca2+]i increases, calmodulin binds CaMKII more readily than Rab3A and this activates CaMKII to phosphorylate effector exocytotic molecules and enhance either the transport of a b-granule to the plasma membrane and/or the mechanism of insulin exocytosis(109).
Ca2+ plays another essential role in triggering endocytosis following secretion. This may be controlled by the Ca2+/calmodulin-dependent serine/threonine protein phosphatase calcineurin, which is postulated to be a calcium sensor for endocytosis in synaptosomes. Calcineurin is activated by calcium and thereby leads to dephosphorylation of proteins involved in endocytosis like dynamin, amphiphysin 1, amphiphysin 2, and synaptojanin (162;163) .
Signaling inside the b-cell
Metabolizable secretagogues and extracellular molecules such as hormones, neurotransmitter that act by binding to receptors on the b-cell plasma membrane appear to mediate their effects through a common set of second messengers. These effects include 1) modulation of adenylate cyclase-cAMP system and activation of protein kinase A (PKA) and recent evidence that cAMP may influence the ATP-sensitive Potassium channel by signaling through Epacs (Exchange Proteins Activated by Cyclic AMP) (164), 2) generation of inositol 1,4,5-triphosphate (IP3) and sn-1,2-diacylglycerol (DAG) resulting in activation of the PKC and finally 3) changes in intracellular free Ca2+ ions that function through Ca2+-dependent regulatory proteins such as calmodulin and PKC. Phosphoinositide signaling appears to also be important for insulin exocytosis with recent evidence using RNAi technology of effect through phospholipase D1, Ca2+ dependent activator protein for secretion 1, and Munc-18-interacting protein 1 (165) .
Hormones binding to G-protein coupled receptors on the cell surface signal to a heterotrimeric seven transmembrane G-protein (inhibitory G-protein, Gi or stimulatory G-protein, Gs), which convey the signal to effector systems such as adenylate cyclase. When a stimulatory G-protein (Gs) is activated (e.g. glucagon stimulation of liver cells) the adenylate cyclase, which resides on the inner plasma membrane, converts ATP to the second messenger cyclic AMP (cAMP) that activates the cAMP-dependent serine/threonine kinase PKA. PKA phosphorylates several proteins at their serine or threonine residues causing a change in either enzyme activity or protein structure. An example is the activation of glycogen mobilization by activating glycogen phosphorylase.
cAMP has a short half-life in the cytoplasm and it is quickly hydrolyzed to AMP by the action of cAMP phosphodiesterases which is important for shutting down the stimulation by the second messenger. cAMP also potentiates Ca2+-dependent insulin secretion (130;166;167). Besides acting on the Ca2+ channels, the major action of cAMP on insulin release is by a direct effect on the secretory machinery itself (130;168). This is thought to be a result of an increased Ca2+ sensitivity of exocytosis by cAMP via an extension of the distance over which the Ca2+ channels can recruit the secretory granules for release (130). It has been suggested that cAMP acts by increasing the release probability of granules already in RRP (PKA-independent) and by accelerating granule mobilization thereby refilling the RRP (PKA-dependent) (166). A recent study has reported that Epac-selective cAMP analogs, but not PKA-selective analogs influence K ATP channels. Epac 1 and 2 bind to SUR1 of the ATP K+ channel, and thus cAMP may stimulate insulin secretion through a novel Epac mediated mechanism (164). Other G-protein receptors (inhibitory and stimulatory G, Go; stimulatory G, Gq) are coupled to phospholipase C (PLC). Acetylcholine stimulates exocytosis by activating the cholinergic muscarinic receptor that acts via a stimulating G-protein (Gq), which is bound to the Ca2+-dependent membrane-bound PLC. PLC hydrolyses PIP2 in the membrane to yield the two second messengers IP3 and DAG. IP3 diffuses to the endoplasmic reticulum (ER) where it binds to and opens a Ca2+ channel in the ER membrane thereby allowing Ca2+ to exit from the ER stores into the cytosol resulting in increased [Ca2+]i. The membrane bound DAG stimulates protein kinase C (PKC) by lowering its requirement for Ca2+. DAG is also produced when PLC is activated following increases in [Ca2+]i, thus PKC will be activated under all conditions in which intracellular Ca2+ is elevated (168) . PKC is shown to modulate Ca2+-dependent insulin secretion (130;135). However, glucose-stimulated insulin secretion is unaffected in PKC-depleted islets, indicating that this enzyme is not essential for glucose stimulated insulin secretion. Rather, PKC appears to be involved in the stimulation of insulin secretion by potentiators of release e.g. acetylcholine and carbamylcholine by sensitizing the secretory machinery to Ca2+ (135) .
Glucose-induced insulin secretion is associated with inhibition of free fatty acid (FFA) oxidation, increased esterification and complex lipid formation by pancreatic b-cells. Increases in the mass of DAG, triglyceride and phosphatidic acid are observed. Exogenous FFA also potentiate glucose-stimulated secretion of insulin, possibly by providing additional acyl groups for long chain-CoA formation or complex lipid synthesis. FFA does not stimulate insulin secretion in the absence of glucose probably due to their rapid entry into the mitochondria when malonyl-CoA levels are low. 116 LC-CoA opens the ATP-sensitive K+ channel whereas the corresponding FFA closes the channel. Long-chain fatty acids stimulate PKC, but the saturated ones are ineffective (169) or much less effective than the unsaturated ones (170) .
Phospholipase A2 Superfamily and Pancreatic b-Cell Function
The phospholipase A2 family plays an important role in the inflammatory response seen in IDDM by generating inflammatory molecules from lipids substrate. Phospholipase A2 (PLA2) constitutes a growing superfamily of enzymes that hydrolyze membrane phospholipids at the sn-2 position generating lipid second messengers as well as structural modulations of cellular membranes (Fig. 7). The fatty acid released from sn-2 is in most cases arachidonic acid.

Figure 7. Illustration of cleavage sites for phospholipase A2 in the glycerol backbone of a phospholipid. Phospholipase A2 cuts the phospholipids in the sn-2 position resulting in formation of a fatty acid and a lysophospholipid, which can be further metabolized to eicosanoids and PAF, respectively. Abbreviations used: PAF, platelet-activating factor (modified from (171).

Several different types of PLA2 have been identified. They include: 1) cytosolic PLA2 (cPLA2), 2) secreted PLA2 (sPLA2), 3) Ca2+-independent PLA2 (iPLA2) and 4) platelet-activating factor acetylhydrolase (PAF-AH). These groups are further subdivided according to structure, molecular weight, substrate specificity, cellular localization, and Ca2+ dependency of the enzyme (172) .
The cytosolic forms of PLA2 (group IV PLA2) are high molecular weight interfacial PLA2 enzymes (see note). They show a preference for phosphatidylcholine (PC) with arachidonic acid (AA) in the sn-2 position and are thought to play an important role in providing free AA for eicosaniod biosynthesis (173-175) . Type IV knock-out mice (IV cPLA2a-/- mice) show deficient inflammatory response (176) and peritoneal macrophages from IV cPLA2a-/- mice fail to produce prostaglandins and leucotrienes after stimulation (177) .
[Note: Interfacial PLA2 enzymes need to bind to an aggregated surface in order to access the phospholipid substrates.]
Group I, IIs (IIA, IIC, IID, IIE, IIF), III, V, IX, X and XI PLA2 comprise a group of low molecular weight PLA2 (13-18 kDa) interfacial enzymes which are either secreted or extracellular PLA2 (sPLA2). These enzymes have been implicated in various physiological and pathological functions, including lipid digestion, lipid mediator generation, cell proliferation, exocytosis, antibacterial defense, inflammatory diseases and cancer (178) .
The Ca2+-independent PLA2 (iPLA2) is a group of cytosolic PLA2 of 85-88 kDa whose major cellular function is the mediation of phospholipid remodelling and homeostasis (179;180) . This enzyme seems to regulate the incorporation of AA and other fatty acids into membrane phospholipids and as such may be a limiting factor for eicosanoid biosynthesis though it has been questioned in the case of the rat pancreatic islet group VIA enzyme (181) .
Platelet-activating factor acetylhydrolase (PAF-AH) is another group of Ca2+-independent PLA2's (182) . All of them have restricted substrate specificity and are selective for PAF (see note) and for phospholipids with a short (or intermediate) acyl chain at the sn-2 position that protects normal membrane lipids from hydrolysis by PAF-AH. A second group of substrates is oxidatively fragmented phospholipids. The main function of PAF-AH is to act as scavenger of bioactive phospholipids especially to regulate the level of PAF (183) .
[Note: PAF (1-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine), a bioactive phospholipid involved in asthma. It is a mediator of a wide range of immune and allergic reactions. It is synthesized in a regulated pathway and activates inflammatory cells. Lysophospholipid can be converted to PAF.]
Different types of PLA2 in islets and pancreatic b-cells
Pancreatic islets contain at least five different types of PLA2 (Table 2). Among the secreted form, both type IB and IIA are found with type IB more abundant. Type IB has been localized to the secretory granules of b-cells and stimulation with insulin secretagogues results in the co-secretion of insulin and the enzyme (184) . Type IB also appears to be regulated by fasting and ambient glucose (184-186) . Furthermore, mRNA for a sPLA2 membrane receptor is found in rat pancreatic islets.

By use of different inhibitors of PLA2, it has been demonstrated that Ca2+-dependent PLA2 may participate in acetylcholine-induced activation of the granule movement in pancreatic b-cells (187) . It is proposed that the type IV cPLA2 couples the secretagogue induced increase in intracellular Ca2+ to the liberation of AA and the subsequent release of insulin (188) . Both exogenous sPLA2 and products of its action have also been shown to suppress KATP channel activity in excised and cell-attached patches of plasma membranes from HIT insulinoma cells (189) .
In glucose-stimulated islets, hydrolysis of AA from membrane phospholipids appears in part to be mediated by a PLA2 that is catalytically active in the absence of Ca2+ and inactivated by the iPLA2 inhibitor HELSS (haloenollactone suicide substrate)(138). However, iPLA2 is not required for AA incorporation or phospholipid remodelling in b-cells, suggesting that iPLA2 plays another role in these cells (181) . iPLA2 in rat and human b-cells is implicated as a putative glucose sensor, which can couple alterations in glycolytic metabolism to the generation of biologically active eicosanoids and thereby facilitate glucose-induced insulin secretion (190) . iPLA2 is activated by ATP and other nucleotides and it has been postulated that it may respond to the increase in the ATP/ADP ratio after glucose stimulation, however this function has been questioned.
Besides playing an important role in the stimulus-secretion coupling in the b-cell different types of PLA2 may be involved in inflammatory responses. Type II sPLA2 and type IV cPLA2 mRNA is upregulated in areas with histologic damage in pancreatitis suggesting that these isoforms might contribute to the morphologic changes that occur (191) . In addition, several proinflammatory cytokines such as IL-1, IL-6 and TNFa have been shown to induce transcription of the type IIA sPLA2 gene in a variety of cells. TNF priming of human neutrophils (PMN) caused marked increase in fMLP stimulated AA release in parallel to enhanced activity and secretion of sPLA2 and activity of cPLA2 (192) . Type I sPLA2, on the contrary, is non-toxic to acinar cells in both severe and mild pancreatitis. This may indicate that type II but not type I sPLA2 have pathophysiologic importance in severe acute pancreatitis associated with systemic inflammatory response syndrome.
Arachidonic acid in cell signaling
Arachidonic acid is a major constituent of islet phospholipids comprising about 34-39% of the sn-2 fatty acyl mass in islet glycerophospholipids (190;193;194) compared to 13% in rat brain and heart and 1% in rat liver (195) . The intracellular level of free AA in islets is normally in the low micromolar range, but rises to a level of 50-200 mM during glucose stimulation (196;197) . Such changes could result by the action of PLA2, but also by the action of diacylglycerol lipase and monoacylglycerol lipase on diacylglycerol (DAG), generated from the action of phospholipase C (Fig. 10) (198;199) . Phosphatidic acid (PA), generated by phospholipase D hydrolysis of PL, is also a potential source of AA (198) .

Figure 8. Arachidonic acid can be made in different ways, but also degraded in several ways. AA can be reesterified into membranes or metabolized by the lipoxygenase pathway to leucotrienes, by the cyclooxygenase pathway to prostaglandins and thromboxanes or by the cytochrome P450 monooxygenase enzyme to epoxyeicosatrienoic acid. Abbreviations used: PAPH, phosphatidic acid phosphohydrolase.

Inhibition of AA release prevents insulin secretion (200) . Increases in the availability of AA (or its metabolites) by itself, however is probably insufficient to initiate insulin release, instead it seems that AA potentiates the action of other secretagogue (201) . In different cell types, AA has been associated with several specific cellular processes such as regulation of PKC and PLC and modulation of intracellular Ca2+ transients (202-205) }. In b-cells, it has been suggested that both Ca2+-dependent and Ca2+-independent components are involved in PLA2-induced insulin secretion. The Ca2+-independent component could be a result of AA or a lipoxygenase product affecting late events in stimulus-secretion coupling (206;207) . The AA-induced Ca2+-dependent event is biphasic with an early rise probably reflecting release of Ca2+ from intracellular stores and a sustained phase, which could reflect extracellular Ca2+ entry possible mediated by voltage-sensitive Ca2+ channels (208) .
Role of arachidonic acid metabolites in cell signaling
When AA is generated, it can be reesterified into phospholipids by acyltransferases or metabolized by the cyclooxygenase (COX), lipoxygenase (LPX) or cytochrome P450 epoxygenase pathways generating prostaglandins and thromboxanes, leukotrienes and lipoxins, or epoxyeicosatrienoic acids (209) .
Prostaglandins made by the cyclooxygenase-pathway acting on AA have been demonstrated, in cell-attached patches on a b-cell line, to be involved in insulin secretion by increasing KATP channel activity (189) . E Prostaglandins inhibit insulin secretion and a pertussis toxin insensitive G alpha i family member G alpha z apparently mediates this effect (210) . The lipoxygenase metabolites of AA are potential mediators of insulin secretion. During nutrient-induced insulin secretion 12-hydroxyeicosatetraenoic acid (12-HETE; a product of 12-lipoxygenase metabolized AA) is augmented (201) and a lipoxygenase inhibitor (nordihydroguaiaretic acid, NDGA) inhibits glucose-induced insulin secretion (211) . Products of the lipoxygenase pathway have also been demonstrated to play a role in apoptosis induced by Ca2+ store depletion in b-cells by using the inhibitor NDGA (212) .
The cytokine IL-1 also augments islet production of 12-HETE and PGE2 (213) . This is not caused by an increased production of AA but by suppression of reesterification of AA into islet phospholipids. This might occur as a result the IL-1-induced production of NO and inhibition of ATP-dependent conversion of the fatty acid to a coenzyme A thioester intermediate (213). 12-HPETE, a precursor of 12-HETE, has been shown to induce apoptosis of cells with reduced levels of glutathione peroxidase. Thus the effects of IL-1 to increase substrate flux through the lipoxygenase pathway and to augment NO production may interact cooperatively to inflict injury on b-cells, and at the same time stimulate insulin release in the short term.
Possible effects of lysophospholipids on b-cell signaling
PLA2 hydrolysis of membrane phospholipids also generates a lysophospholipid (lysoPL). Several types of lysoPL's exist in the islet membrane: lysophosphatidylcholine (lysoPC), -ethanolamine (lysoPE), -serine (lysoPS) and -inositol (lysoPI). LysoPL's and phosphatidic acid (PA) are shown to promote insulin release when the cellular content of either lipid increases(156). LysoPL's affect stimulus-secretion coupling at a number of points(162,163) (214;215). They have been observed to inhibit directly the KATP channel activity in the pancreas (216). LysoPC and lysophosphatidylglycerol can also effectively mobilize Ca2+ from intracellular stores (215). LysoPAF has been demonstrated to circumvent the inhibition of glucose-induced insulin release caused by phospholipase inhibitors (217) .
Pathophysiogical role of lipid metabolites
Inflammatory response in the pancreas not only plays an important role in the destruction of islet tissue in autoimmune diabetes (see chapter by Mandrup Poulsen), but also in the ability of the pancreatic b-cell to recover from such insult.
Pancreatitis induced by cellophane wrapping (CW), surgery, coxsackievirus B4, cerulein treatment or with other chemicals (218-220) for example can lead to islet regeneration. Locally produced INFg causes lymphocyte infiltration and islet cell destruction and while IFN-g gene linked to an insulin promoter produces diabetes new cells are formed from duct epithelial cells and they differentiate into endocrine cells (221) . The proinflammatory cytokines, IL-1, TNF and perhaps IL-6 is shown to induce PLA2 expression and release (222). The proinflammatory action of TNF depends in part on activation of PLA2 (223) . In humans with acute pancreatitis, there is a correlation between elevated serum PLA2 levels and inflammation and lipid metabolites are also potential mediators in these events. In this regard it will be of interest to see whether selective inhibition of PLA2 can be used as a therapeutic approach against type 1 diabetes.
Micro RNA miR-375
There may be as many as 1,000 different micro RNAs in the human genome, Micro RNAs following production in the nucleus by the enzyme Drosha are exported to the cytoplasm, and processed by the enzye DICER. They then are incorporated into the RNA-induced silencing complex (RISC) and function to either silence by binding to messenger RNA or by directing the cleavage of mRNA. It is estimated that there are 68 miRNAs in beta cells. The the function of one islet specific miRNA has recently been elucidated. MiRNA-375 is reported to inhibit insulin secretion by inhibiting the production of the protein myotrophin and thereby decreasing granule exocytosis (224).

Concluding remarks
Despite considerable recent progress in defining the molecular physiology of insulin secretion there remain many unanswered questions. It is also important to recognize that some current techniques may introduce complexity, such as the recent report that the RIP-Cre trasnsgene itself produces mice with impaired glucose tolerance (225). Nevertheless given our current knowledge it is clear that the pancreatic b-cell is an excitable cell that is critically dependent upon oxidative metabolism (226) and dependent on the influx of Ca2+ through gated channels and its subsequent removal from the cell by intracellular sequestration or plasma membrane pumping leaves it in a precarious position when faced with agents that disrupt Ca2+ homeostasis in the cell. In a similar way, its susceptibility to drugs such as the DNA alkylating agent streptozotocin or the free radical generator alloxan may be the consequence of the mechanism of sugar transport, which can carry these agents into the cell. Such phenomenon may explain in part why so many of the agents that one associates with pancreatic b-cell damage and death turn out to be specific secretagogues at least in the short-term. The challenge is to discover which of these interactions are part of the normal responsiveness of the cell to metabolizable substrates, peptides and neurotransmitters and which are truly deleterious and thus a legitimate target for pharmacological intervention. It is likely that the concepts that there are "good" or a "bad" cytokines in the context of autoimmune diabetes is no more valid than that of glucose being "good" or "bad" in the context of glucotoxicity and type 2 diabetes. The short-term effects of secretagogues on islet function have been extensively documented but there is also the need to consider the broader perspective of how the cell responds in the longer term to inflammatory stimuli or physiological changes in energy homeostasis such as accompany pregnancy and starvation (Fig. 9).

Figure 9. Yin and Yan representation of the pancreatic b-cell in terms of induction of either cell death or secretory response to secretagogues such as inflammatory cytokines and PLA2.

For comments, corrections or to contribute teaching slides, please contact Dr. Eisenbarth at: george.eisenbarth@ucdenver.edu