Intercellular Communication in the Islet of Langerhans in Health and Disease

Abstract

Blood glucose homeostasis requires proper function of pancreatic islets, which secrete insulin, glucagon and somatostatin from the β-, α- and δ-cells respectively. Each islet cell type is equipped with intrinsic mechanisms for glucose sensing and secretory actions, but these intrinsic mechanisms alone cannot explain the observed secretory profiles from intact islets. Regulation of secretion involves interconnected mechanisms among and between islet cell types. Islet cells lose their normal functional signatures and secretory behaviors upon dispersal as compared to intact islets and in vivo. In dispersed islet cells, the glucose response of insulin secretion is attenuated from that seen from whole islets, coordinated oscillations in membrane potential and intracellular Ca²⁺ activity, as well as the two-phase insulin secretion profile are missing, and glucagon secretion displays higher basal secretion profile and a reverse glucose-dependent response from that of intact islets. These observations highlight the critical roles of intercellular communication within the pancreatic islet, and how these communication pathways are crucial for proper hormonal and non-hormonal secretion and glucose homeostasis. Further, misregulated secretions of islet secretory products that arise from defective intercellular islet communication are implicated in diabetes. Intercellular communication within the islet environment comprises multiple mechanisms, including electrical synapses from gap junctional coupling, paracrine interactions among neighboring cells, and direct cell-to-cell contacts in the form of juxtacrine signaling. In this review, we describe the various mechanisms that contribute to proper islet function for each islet cell type and how intercellular islet communications are coordinated among the same and different islet cell types.

Introduction

The islet of Langerhans is a micro-organ that ranges from 60 to 200 μm in diameter and contains ~1,000 to 10,000 cells that play a key role in glucose homeostasis through regulated secretion of insulin from β-cells (234, 268), glucagon from α-cells (137, 170, 420), and somatostatin from δ-cells (327). A fourth cell type, the PP-cells, secrete pancreatic polypeptide, which is another hormone thought to act mainly on the exocrine pancreas and the liver (387). The cells in the islet are highly-coupled through multiple routes, including gap junctions that form electrical synapses, receptors for secreted products of neighboring cells that form chemical synapses, as well as juxtacrine interactions between plasma membrane attached ligands and receptors. How cells in the islet communicate to act in a coordinated fashion to control blood glucose homeostasis has been the focus of research for many years. Insulin and glucagon function as hormones with actions throughout the body, while δ-cell somatostatin is thought to act only locally since it is very short lived in the circulation (<1 min). Of the two main isoforms of somatostatin (14 and 28 amino acids long), the pancreatic δ-cells secrete the somatostatin-14 variant through Ca²⁺-dependent exocytosis (29). These three major secretory products as well as other transmitters all play important signaling roles within the islet (53). A role for pancreatic polypeptide in islet cell communication has not been elucidated. The importance of communication within the islet can easily be illustrated by comparing the function of the cells within an intact whole islet versus isolated cells. The α-, β-, and δ-cells all require the islet milieu for proper function, since they lose their normal characteristics when dispersed from the islet (22, 23, 142, 310). Further, it has become clear that intercellular communication in the islet environment is not only critical for proper hormone secretion, but that abnormal secretion of islet hormones resulting from communication defects causes or contributes to disease progression in diabetes.

Over the physiological blood glucose range, insulin levels track in opposition to glucagon, and the subsequent actions of these two hormones are opposite in almost all respects (159). During hypoglycemia, insulin secretion is shut off while maximal glucagon levels stimulate glucose output from the liver. As glucose increases, glucagon secretion is largely suppressed and insulin begins to rise (183). Both hormones are regulated by somatostatin secreted from the δ-cell (53), and the secretory dynamics of all three of these molecules are mis-regulated in type 1 (134) or type 2 diabetes (352). The therapeutic success of insulin as a treatment for diabetes has led to a focus on β-cell research (25) although we now know that aberrant glucagon secretion exacerbates hyperglycemia (285, 393). In diabetes, glucagon secretion is not inhibited by glucose and hyperglucagonemia is observed at all glucose levels (394). Inhibiting glucagon secretion or blocking glucagon action during diabetes can restore euglycemia in the absence of insulin (192, 217, 393, 412, 436), although it is unknown how this happens (111, 303). However, it should be noted that several drug treatments that block glucagon action by targeting glucagon receptors have shown adverse side effects during clinical trials (69, 143, 193). We also understand that somatostatin from the δ-cell inhibits the secretion of both insulin and glucagon during hyperglycemia. It was shown over 40 years ago that somatostatin infusion into patients with both type 1 and type 2 diabetes improved glycemic control and lowered insulin requirements. This study led to the proposal that a long-acting somatostatin analog could be a helpful addition to insulin therapy (122, 124). Since islet cell function is intertwined with the signaling between α-, β-, and δ-cells, cellular communication within the islet plays an important role in hormone secretion (54).

One additional islet cell secretion behavior seen is that apparently all of the islets in the pancreas are also in communication, which result in pulsatile insulin secretion, as measured in mice (155) and humans (364). This synchronization across the organ may be important for glucose tolerance (23) and proper regulation of blood flow as a function of blood glucose levels (357). While the mechanisms behind this islet to islet communication are beyond the scope of this article, it is likely that the molecular mechanism(s) would be shared with intra-islet communication pathways, although mediated by neural rather than paracrine pathways. One model that has been put forward suggests that cholinergic stimulation plays a role (440), which will also be discussed below in terms of islet α-cell dynamics.

In this review, we will first describe the architecture of the islet, and detail its main cell types. While each of these cell types has its own characteristic phenotype, they work together as a unit, and at times a syncytium. Thus, we next present the set of mechanisms that these cells utilize to communicate and coordinate their function. Finally, we cover the current understanding of the specific mechanisms used for the various communications among cells of the same type and between cells of different types.

The Islet of Langerhans

The islet appears to be the functional unit of pancreatic hormone secretion, and cellular proximity and communication play key roles, since dissociated cells have long been observed to exhibit different behaviors than those within intact islets (Fig. 1). First, the insulin secretory response to glucose from whole islets is ~10 fold greater than that from an equal number of isolated β-cells (218). Second, coherent oscillations of the membrane potential and intracellular free calcium activity, and the normal two-phase insulin secretion profile are glucose-response phenomena that do not occur in isolated β-cells (10, 12, 325). Third, glucagon secretion in vivo and from intact islets is inhibited by elevated glucose, but with isolated α-cells, glucagon secretion increases with glucose, similar to what is observed for insulin secretion (215, 310). Finally, human islet transplants are able to reverse the Type I diabetes phenotype while β-cell transplants do not (332). These points emphasize the importance of understanding the mechanistic pathways in the islet cells, as well as the intercellular interactions within the intact islet.

Figure 1:

Comparison of insulin and glucagon secretory responses to glucose in intact islets and isolated cells. Green, red and blue cells represent β-, α- and δ-cells respectively.

The characteristic secretion patterns of insulin, glucagon, and somatostatin observed in response to changing glucose levels with isolated islets accurately recapitulate the patterns seen in vivo (155, 364, 365), but nearly all of those patterns are changed or lost when the islets are dispersed into individual cells (218, 247). Data from pseudoislets built with various combinations of the constituent islet cell types show that all three of the α-, β-, and δ-cells are required to mimic the function of intact islets (310, 410). This has led to a consensus model for islet function based on a highly coupled three-hormone signaling network. In this tri-partite cell regulation model, the secretory products of one cell type regulate the other cell types and vice versa (Fig. 2). As detailed in this review article, the hormones and other chemical factors that are released by each cell type play a role in overall islet function. For example, insulin secreted from the β-cells impinges upon insulin receptors in the α-cells (107, 308) and δ-cells (222) as well as on the β-cells themselves (11, 314). These paracrine signals may be communicated between cells via the bloodstream or through interstitial space, which is likely considering that isolated islets and pseudoislets maintain most of the functional outputs seen from islets in vivo. Receptors for these cellular release products are likely to be concentrated on primary cilia that are found on every islet cell, and which protrude throughout the islet (169). As described below, cilia may act in a way that enhances non-nearest neighbor interactions. While each of these three cell types is needed for normal islet function, the requirements on specific cell ratios appear to be quite flexible. The relative amount of the different cell types in an islet vary by species as well as the islet’s size (44) and location within the pancreas (375). In mice and rats, the β-cells make up 65 to 75% of the islet, with α-cells contributing about 10 to 20% and the remainder consisting mainly of δ-cells and PP-cells along with a few rare cell types that secrete ghrelin and perhaps other hormones (137). In humans, the relative number of non-β-cells is increased with α-cells making up to 35% and beta cells only contributing less than 60% (294) (Fig. 3). In pig and dog islets, the cell type distribution appears to be close to that seen in humans (97, 390). The reason behind these differences remains unknown. The hormone secretion responses of pseudoislets to glucose remain similar over a wide range of cellular compositions (204, 410), so the differences in islet type ratios are perhaps driven by the physiological requirements for hormone action in different organisms.

Figure 2:

Overview of the intercellular communication pathways within the islet. β-cells and δ-cells sense glucose directly through glucose transporters and are coupled via gap junctions. α-cells undergo a series of paracrine regulation by insulin receptors (IR), 5-HT_1F receptors and SSTR2 which are activated by β- and δ-cell secretory products. α-cells are also regulated by EphA-ephrinA juxtacrine signaling through direct cell-to-cell contacts with β-cells. Secretion of insulin, glucagon and somatostatin by β-, α- and δ-cells, respectively, is coupled to intracellular Ca²⁺, ATP/ADP and cAMP concentrations.

Figure 3:

Image of an intact human islet immunostained with insulin (green), glucagon (red) and somatostatin (yellow) for the identification of β-, α- and δ-cells respectively.

The cells in the islet appear to form a heterogeneous population, which requires intercellular communication in order to result in robust functional responses. β-cell heterogeneity has been implicated as important for proper islet function (24, 27, 100, 293). The relationship between cellular communication and heterogeneity in islet function, with a focus on gap junctional coupling between β-cells, is detailed below. It was postulated that proper β-cell heterogeneity is altered during T1D, either as a result of or a cause of disrupted intercellular cooperation, leading to the loss of islet function (28, 35). Changes in intercellular cooperation and the underlying heterogeneities may interact in ways that exacerbate disease initiation and progression (27). We know there are heterogeneities in α-cells, although the role of this heterogeneity has been only minimally explored (214). In mouse islets, α-cell heterogeneity is reflected as an anticorrelation between the presence of Ca²⁺ activity and absence of Zn²⁺ uptake in one population of α-cells, and vice versa in another population (213). A perhaps related heterogeneity was found in human islets, where measurements of secretory granule Zn²⁺ uncovered two populations of α-cells in terms of their glucagon content (125). To this time, we have no functional data to hypothesize about possible heterogeneity in δ-cell and PP-cell functions.

Communication Pathways in the Islet

Before diving into the signaling networks that govern the individual cell types and the specific communication pathways that have been discovered to play roles among and between the cell types, we will provide some background about the various mechanisms that are general among all islet cell types. These mechanisms are not unique to the islet and are utilized by tissues throughout the body, but each has unique roles to play within the islet milieu. There are three classes of communication mechanisms: electrical synapses carried through gap junctions between cells, chemical synapses or paracrine signaling between adjacent or nearby cells, and juxtacrine signaling based on direct cell-to-cell contact. All of these mechanisms are used by islet cells, with some better understood than others. Importantly, these mechanisms engender at least some of their effects through contacts dependent on primary cilia, vasculature, and nerves. In this section, we will describe the three communication mechanisms in general, and provide some background about how they depend on the cilia, vasculature, and nerve fibers to manifest their effects.

Cell-cell Communication through Gap Junctions

It has been known for almost 40 years that gap junction coupling plays a role in the islet (62, 103, 256). The discovery of synchronous pulses of Ca²⁺ activity (339) suggested that gap junction coupling was a powerful component. Once wave speeds were measured (26, 318), it became clear that only gap junction coupling could provide the connectivity needed to generate such synchronicity over the islet in such a short time (~100 msec). After many years of research (249), genetic deletion of connexin-36 (62, 351) was shown to disrupt synchronous Ca²⁺ activity oscillations and yield abnormal glucose stimulation of insulin secretion (GSIS) (26, 307), leading to the conclusion that this gap junction variant plays the dominant role in islet β-cell electrical connectivity (Fig. 4). Six connexin sub-units assemble on the plasma membrane to form the hexameric connexon. Connexons from one cell dock with those from a neighboring cell to form gap junctions that are typically clustered in a plaque, an area where the interstitial space is reduced. Most of these connexons are made up of a single connexin sub-type, and they can form functional gap junctions with neighboring connexons of the same type, called homotypic channels. Many connexin sub-types can form functional gap junctions with other connexons formed from sub-types, called heterotypic channels. Connexons can also be formed from different connexin sub-units, but that is not thought to be a major driver of islet cell communication. There are 23 members of the human connexin family (denoted CxX), 18 of which have orthologs in the mouse genome (105). While the best evidence of these proteins is from the studies of their genetic deletion in mice, several connexin genes are found to be expressed by RNA-sequencing in the human islet (222), although across many studies, it becomes clear that expression levels are much lower for all connexins in human as opposed to mouse islets, especially Cx36 (348, 425). Thus, how gap junctional coupling works in human islets remains an open question. Still, the synchronous pulses of Ca²⁺ activity across regions of human islets are consistent with the islet physiology seen in mouse.

Figure 4:

Formation of functional homotypic Cx36 gap junctions between β-cells and non-functional heterotypic Cx36-Cx46 gap junctions between β- and α-cells.

The formed gap junctions can be functional or non-functional (208). Functional gap junctions act basically as a hole between the two cells, although each type has a specific limitation on the size and charges of molecules that can pass through. The limits on size and preferences of charge can be quite complicated, and only a few of the details have worked out for most combinations (149). For example, adenosine preferentially passes through gap junctions formed from Cx32, while ATP does not. This preference is reversed by several hundred fold through gap junction channels formed from Cx43 (130). Non-functional gap junctions still may act as cell adhesion molecules, and in some cases they can become functional by specific cellular signaling pathways.

Many connexin isoforms show some minimal gene expression by RNA-sequencing of mouse(89, 210) and human (222) islet cells. As discussed above, the functional isoform in mouse β-cells has been shown to be Cx36 (351). The δ-cell appears to express Cx40.1, which has also been found to have some expression in the β-cell. Gap junctions have been reported to function between β-cells and δ-cells (53), but it is not known whether these Cx40.1 sub-units make heterotypic gap junction channels with Cx36 (209), or if the β-cell to δ-cell connections are through homotypic Cx40.1 gap junctions. In contrast to the β- and δ-cells, the α-cell appears to express only Cx46 (222), which cannot form heterotypic gap junctions with either Cx36 or Cx40.1 (209). The PP-cells express Cx45, which is also expressed by the ductal cells, but is not expected to form functional channels with any of the other islet-specific connexin sub-types (209).

Paracrine Signaling Between and Among Islet Cell Types

Paracrine signaling is a broad category of communication mechanisms defined as the release of a signaling molecule from one cell that affects another cell. This signaling can take the form of a classical synapse where the molecule is directed to a specific neighbor. However, the distance range can be anything up to hormonal type signals where the molecule may impact any other cell in the islet. Given the islet structure, where cilia can reach several cells away (see section 3.d.) (114, 169) and δ-cells may take a neuron-like shape with processes reaching away from the cell body (327), it is not surprising that a large number of paracrine signals have been discovered to act within the islet. The descriptions of individual secretory products of each islet cell type and how they affect the other cells in the islet comprises a large part of this review, but as an introduction, these are summarized in Figure 5.

Figure 5:

Overview of paracrine interactions between islet cells.

The predominant cell type in the islet, the β-cells, also secrete the largest number of known paracrine factors, beginning with insulin which has been shown to signal to other islet cells (150, 192), as well as the β-cells themselves. The β-cell supplies the paracrine signaling factors of Zn²⁺, serotonin, γ-aminobutyric acid (GABA), glutamate, γ-hydroxybutyrate (GHB), dopamine, amylin (or islet amyloid polypeptide – IAPP), and purinergic signals. These have been shown to affect the α-cell (137, 144, 170, 432), but some or all of them may also play a role in modulating secretion from the δ- and/or PP-cells (327, 402). Glucagon from the α-cell has been shown to signal to β-cells (380) and δ-cells (9), and the glucagon signaling to the β-cell has been shown to couple through both glucagon and GLP-1 receptors (380). In the δ-cell, on the other hand, this signaling is expected to be transduced only through the highly expressed glucagon receptor (222). Under some circumstances, the α-cell can alternatively splice pro-glucagon to produce GLP-1 as an enhancer of β-cell function (224). The α-cell has also been shown to secrete acetylcholine that signals primarily to the β-cell (321), but also to the δ-cell (260). Glutamate is known to be an autocrine feedback regulator of glucagon secretion from the α-cell (266), but it also may signal to the δ-cell (266). Somatostatin secreted from the δ-cell can down-regulate both insulin and glucagon secretion (327). Other potential paracrine mediators have not been ascribed to δ-cell secretion. Finally, while pancreatic polypeptide is thought to be an enhancer of secretory activity in the islet (361), the exact mechanisms that might underlie such an effect remain unknown.

While the definition of paracrine signaling is generally assumed to be from one cell to another, autocrine signaling can be considered a special case of paracrine signaling. In autocrine signaling, the secretory product from one cell can directly stimulate or inhibit further activity from that cell itself or one of its neighbors of the same cell type. In the islet, there are many examples such as insulin regulation of insulin secretion from β-cells (11, 314, 315), dopamine from β-cells that is co-secreted with insulin and inhibits further insulin secretion (330, 358, 396), and glutamate secretion from α-cells that stimulates glucagon secretion (58).

Communication via Juxtacrine Pathways

Besides electrical (gap junctions) and chemical (paracrine) synapses that form between islet cells, there are also a number of types of direct cell-to-cell contact (juxtacrine) signaling. One is the Eph/ephrin signaling system where a cell attached ephrin ligand interacts with an Eph receptor from a neighboring cell. This system can signal in both the forward direction through the Eph receptor as well as in the reverse direction through the ephrin ligand (47). As is the situation with connexins, there are numerous EphA and EphB receptors as well as multiple ephrin-A and ephrin-B ligands. It was originally thought that the A and B represented two non-interacting classes, but that is now known not to be true, and it seems that the more experiments reported, the more promiscuous these receptors and ligands are found to be. Ephrin signaling has been shown to be important for proper islet hormone secretion (173, 205, 310), and several Eph receptors and ephrin ligands are expressed in islets. EphA5 and ephrin-A5 are highly expressed in β-cells, EphA4 and EphA7 are expressed in α-cells, EphA6, ephrin-A1, and ephrin-B3 are expressed in δ-cells, and EphB3 is expressed in PP-cells (95, 222).

Other juxtacrine signals have been examined only in the context of the β-cell. The cell surface receptor Fas and its associated ligand FasL has been reported to regulate β-cell turnover as well as insulin secretion. Various cell-adhesion molecules are also expressed in pancreatic islets, primarily E- and N-cadherins (78) and neural cell adhesion molecules (NCAMs) (72). These molecules play a major role in holding the islets together, but the loss of E-cadherin or NCAM attenuates insulin secretion (274, 428). Finally, tight junctions involving claudins and occludins have also been observed in islets (279), but we have developed little understanding of their potential functions in the islet. Details about these pathways will be provided below in terms of the specific cell-type interactions.

Primary Cilia as a Signaling Hub in the Islet

Another way that α-, β-, and δ-cells in the islet may communicate is via their primary cilia, which are required for normal glucose regulation of islet hormone secretion (61, 354). For example, it has been reported that β-cell cilia define cell polarity and are required for insulin production (115), and a number of human conditions, including type 2 diabetes, have been linked to ciliary dysfunction (120, 406), which suggests that cilia represent a novel therapeutic target. This pathway adds another dimension to islet cell regulation, as cilia are known to not only play a role in chemo- and mechano-sensation, but also coordinate paracrine signaling pathways.

Long considered a vestigial organelle of dubious functional importance, cilia are found in almost all islet endocrine and exocrine cells (13, 90, 120, 226, 429). The primary cilium, an extended cellular organelle that protrudes out of the cell body, plays an integral role in islet cell communication. This single hairlike-cellular protrusion consisting of a microtubule-based axoneme and a lipid bilayer is present in all β-, α- and δ-islet cell types of various organisms ranging from zebrafish, mice, rabbit to human (169, 429). Enriched in transmembrane cell surface receptors including G protein-coupled receptors (GPCRs) (30, 202, 344), various receptor tyrosine kinases (e.g. insulin receptors and somatostatin receptor 3), and transforming growth factor β receptors, the primary cilium acts as a signaling antenna to sense extracellular cues and orchestrate multiple signaling pathways to regulate cellular and physiological processes such as embryonic development, cell differentiation and metabolic homeostasis (70, 71, 176, 359, 423). In the pancreatic islet β-cell, the primary cilium localizes at the apical domain, which is structurally defined by β-cell polarity as one of the three distinct domains orientated furthest away from the vasculature, and projects towards the extracellular luminal interstitial space between adjacent islet cells (114). The other two domains are the lateral domain that is localized between islet cell-to-cell contacts and is enriched in glucose transporter 2 (GLUT2) for glucose sensing and uptake, and the basal domain which is in direct contact with the vasculature and identified as the main site for insulin secretion and thereafter targeted delivery towards the blood stream (228). Such cell polarity dependent spatial compartmentalization of distinct domains in β-cells, specifically the apical localization of the primary cilia away from the vasculature, is speculated to have functional significance in modulating local paracrine signaling. This is proposed to act by segregating insulin detection at the primary cilia to circulating insulin within islet interstitial spaces (~50 – 300 pM at physiological glucose levels by mathematical modeling), away from targeted insulin secretion towards the vasculature (228, 411).

Of late, there has been emerging evidence of the critical role played by the primary cilia in islet cell communication and function. In a study that utilized the adenoviral shRNA knockdowns of Oral-facial-digital syndrome (shOfd1), which results in the loss of cilia, and Bardet-Biedl-Syndrome 4(shBbs4), which disrupts ciliary function, in MIN6m9 cultured cells and isolated murine islets, Gerdes et al. reported an impairment of the first phase insulin secretion and aberrant insulin signaling due to diminished localization of insulin receptors to the cilia (120). They have also established the misregulation of ciliary genes and loss of cilia in β-cells as a phenotype of type 2 diabetes in rat models, as has been demonstrated in a complementary study on mice model and human islets (201). The role of β-cell primary cilia in islet cell communication was further highlighted in the group’s subsequent study on a β-cell-specific inducible cilia knockout (BICKO) mice model which resulted in an aberrant phosphorylation and downstream signaling of the EphA/EphrinA juxtacrine signaling components and a subsequent impairment of insulin secretion and glucose intolerance (407). Hughes et al. generated an Ins1-Cre β-cell cilia knockout (βCKO) mice model which extensively revealed the functional role of β-cell primary cilia not only in the regulation of glucose-stimulated insulin secretion, but also for proper glucagon and somatostatin secretion in α- and δ-cells (i.e. islet cross talk), accurate sensing and processing of paracrine signals in the islet interstitial space, proper cellular synchronicity in insulin secretion and β-cell Ca²⁺ dynamics, and the regulation of energy metabolism and glucose homeostasis (169). Collectively, this evidence suggests a pivotal role of the primary cilia in core islet functions via intercellular islet communication. Although islet α-and δ-cells also contain a single primary cilium (226, 429), little is known about their mechanistic functions in islets as most literature focus on the primary cilia of β-cells. Therefore, it would be of great interest to decipher the overall islet cell connectivity and function among the primary cilia of all islet cell types.

Communication via the Vasculature and Blood Flood

The cells within the islet are closely packed. They are not only in physical contact with each other, which offers many potential routes of communication, but they are also juxtaposed to the vasculature (41, 42, 181) and nerves (5). Because of the perceived importance of blood flow in islet function, (42), detailed structural and dynamic information about blood flow has long been sought (179). More recently, it was shown that insulin secretion was targeted preferentially to the vasculature (228), which would be expected given the requirements for insulin throughout the body, and this observation underscored the importance of islet blood flow for proper physiology. The flow of blood plasma and cells in living animals has been assessed using a number of strategies, and successive improvements in instrumentation have allowed an ever increasing level of detail about in vivo blood cell flow to be obtained. These methods include approaches such as Doppler imaging, two-photon microscopy, and ultrasound (131, 200, 371).

Hormonal communication between islet cells has long been considered as a function of the blood flow (42). Blood flow patterns are observed typically moving first past the beta cells and continuing to the other cell types (180, 272, 373), which has led to models considering insulin’s effects on the α- and δ-cells, while minimizing the possibilities of glucagon and somatostatin action on the β-cell. In opposition to this model, though, hormones within the islet are not restricted to traveling through the bloodstream, but they can also affect neighboring cells by diffusion through the interstitial space (59). Unlike secretion into the bloodstream (228), interstitial secretion is difficult to visualize and measure. Still, this kind of paracrine communication between islet cells is an attractive hypothesis because the hormone concentrations in the blood stream are likely to be too low to yield the effects seen on isolated islets (180). Because of the high flow rates in the islet vasculature, even maximal concentrations of plasma insulin in the region of the islet are expected to be sub-μM. In experiments with isolated islets, the concentration of insulin required to elicit functional effects on islet cell types can be 10- to 100-fold greater than what is expected to be the vascular concentration in vivo. However, the small size of the interstitial space suggests that insulin concentrations in this region could potentially reach the 100 μM range depending on how fast hormones are cleared from that space. This communication model is also supported by data showing that cilia grow into the interstitial region and not towards the blood vessels (114). Given that many of the key communication receptors are located on cilia, as described above, and the fact that they are not directed to the vasculature, supports the hypothesis that islet cell paracrine communication is primarily via the interstitial space rather than the vasculature.

While it remains an experimentally untested concept, the possibility of interstitial signaling appears to be supported within the known physical parameters of the islet and its cells. We can estimate that if 1% of the islet volume consists of interstitial space, the insulin concentration in that space resulting from a single granule release would be, depending on islet size, 10–100 nM (using numbers from (160)). It is likely that multiple granules could be released into this space, especially given the large number of β-cells within the islet and the large number of granules per β-cell. Further, without the presence of flow, this concentration would be expected to persist for some time, so it could easily be in the concentration range of 1–10 μM and possibly even higher. The specific case described here is for insulin, but one can expect that interstitial communication could be used by any intra-islet paracrine or autocrine factor. In fact, interstitial transmission of signals may be even more important for secretory products that are thought to be mainly downstream of β-cells via the vasculature, such as those coming from α- and δ-cells (181, 272, 373).

Communication via the Nervous System

The fact that islet hormone secretion can change without any signals arriving through the blood stream, for example during the cephalic phase (400), indicates that inputs from the central nervous system can modulate islet hormone secretion. The islet is innervated by multiple nerves including ones from the sympathetic, parasympathetic, and sensory systems (126, 152, 381). It is not fully understood how innervation affects glucose homeostasis, and the role of the central nervous system in diabetes progression remains controversial (118, 391). One part of this controversy stems from the differences in innervation that have been reported between rodents and humans (321, 322), although even these differences lack consensus (168, 383, 384). Still, the key functional properties of the nervous system action are conserved across species in terms of islet hormone secretion (43, 118, 126, 152). In terms of disease progression, it is agreed that islet innervation decreases with age (299), and that innervation is decreased during human type 1 diabetes (264), as well as in rodent models of autoimmune diabetes (265). As mentioned in the introduction, there is a synchronization of insulin pulses across the pancreas, and it is likely that the nervous system plays a central role in this coordination. One model that has been put forward suggests that cholinergic stimulation plays a role (440), which could represent a major functional role for nerves independent of the central nervous system.

Signaling and communication of the main islet cell types: β-cells, α-cells, δ-cells, and PP-cells

To understand the role of communication within the pancreatic islet, it is first important to understand the properties of the individual cell types that make up the islet. The islet is a complex multi-cellular system where many aspects of cellular function depend upon communication between cells of the same type and between the different cell types. For the pancreatic islet, these communication pathways are critical for the maintenance of whole body glucose homeostasis. The two key hormones secreted by the islet are insulin and glucagon, which function via actions throughout the body and lead to direct effects on glucose utilization. At the same time, the third “hormone” secreted by the islet, somatostatin, appears to function only locally since it is very short lived in the circulation (<1 min) (302). As described above, there are two main isoforms of somatostatin (14 and 28 amino acids long), and the δ-cells secrete somatostatin-14 (29, 327, 416). Within the islet, these three major secretory products as well as other secreted transmitters all play important signaling roles (53). A fourth cell type is also found in some islets, the PP-cells that secrete pancreatic polypeptide, and while those cells appear to be inhibited by somatostatin, the role of PP in communicating with other islet cells remains largely unknown. In this section, we will describe some central signaling features of each cell type including autocrine feedback loops. We will then detail the mechanisms that have been discovered by which each cell type affects and is affected by communication with other cell types.

β-cells

The β-cell is the only insulin secreting cell type in the body, and the multi-step process of glucose-stimulated insulin secretion (GSIS) is quite well understood (Fig. 6) (268, 325). The consensus model of GSIS involves the proximal events of glucose transport and phosphorylation, intermediate metabolism by glycolysis and citric acid cycle, and distal electrophysiological responses leading to insulin secretion. Glucokinase (GK) phosphorylates glucose and triggers glycolysis, which produces pyruvate and reduced nicotinamide adenine dinucleotide (NADH). Tricarboxylic acid cycle (TCA) metabolism in the mitochondria produces more NADH and also reduced flavin adenine dinucleotide (FADH₂) (317). Data from these studies have shown that both glycolysis and TCA metabolism are needed to generate the changes in [ATP]/[ADP] required for normal GSIS (135, 282, 316, 317). Closure of the K_ATP channels depolarizes the plasma membrane, thus activating Ca²⁺ channels and increasing intracellular Ca²⁺ activity (340), and in turn, stimulates insulin secretion (116).

Figure 6:

Intracellular and intercellular pathway mechanisms that regulate insulin secretion in the β-cell.

In addition to this standard model, many other metabolic pathways have been shown to cause insulin secretion or modulate GSIS (322). For example, the signaling by free fatty acids (430), or amino acids (439), or GLP-1, which may arise from the gut (14) or within the islet itself (80). Glucose-independent insulin secretory pathways have been clearly revealed by data from transgenic mice containing a β-cell specific disruption of the K_ATP channel either through a deletion of Kir6.2 (258) or Sur-1 (349, 356). These mice maintain normal blood glucose levels. Islets from these mice exhibit reduced responses to glucose, but do show normal or enhanced responses to molecules that signal directly through calcium, PLC-linked receptors, or cAMP (93). Since these pathways have been shown to influence GSIS, it is likely that they also act as part of the paracrine signaling pathways that allow communication among and between different islet cells.

A hallmark of β-cell and whole islet function is pulsatile behavior in membrane depolarization, Ca²⁺ activity, and insulin secretion, which occur at elevated glucose levels (234). Such pulsatile behaviors are exhibited synchronously throughout the islet (31). This strongly suggests that there is significant communication between the islet cells, and over two decades of data support the model where gap junctions provide the major component of this communication (26, 62, 248, 307). Importantly, this pulsatility is impaired in diabetic phenotypes (297). β-cells also exhibit oscillations in ATP/ADP (6, 82), mitochondrial membrane potential (191), oxygen consumption (196), and NAD(P)H (230). The oscillations of mitochondrial membrane potential and NAD(P)H are both small in comparison with the overall changes upon glucose stimulation. These small changes suggest that metabolic oscillations are a by-product of Ca²⁺-influx, rather than the driving factor behind the observed electrical oscillations, which are of much larger amplitude. Still, metabolic oscillations have been proposed to be a driving force behind the electro-physiological oscillations, especially under high glucose conditions where most K_ATP channels are closed and subtle changes in conductance can lead to significant effects on electrical oscillations (35, 253, 286).

The secretory products of the β-cell consist not only of insulin, but also Zn²⁺ (125, 300), GHB (220), GABA (414), dopamine (330, 397), serotonin (7), amylin (269), and purinergic signals (145, 177). While the exact details regarding the action of each of these constituent molecules remain unclear, the consensus appears to be that the net action of these products is to inhibit secretion from the other cell types. The β-cell also contains many receptors for these and other secretory products that facilitate autocrine or paracrine regulation of insulin secretion (331). Autocrine regulation has been shown to happen through insulin receptors (46, 263, 314), purinergic receptors (177), GABA receptors (48) and dopamine (358, 396). The main paracrine mediated pathways appear to be δ-cell somatostatin stimulation of somatostatin-R₂ (187) and α-cell acetylcholine stimulation of muscarinic receptors (321).

A close examination of the cell surface of β-cells demonstrates the importance of physical contact between islet cells, and extends beyond their electrical connectivity to juxtacrine communication pathways that can lead to diverse biological effects. In addition to connexins and gap junctions, the EphA/ephrin-A (296) system has been demonstrated to affect GSIS via bi-directional signaling through the β-cell expressed EphA5 receptors and ephrin-A5 ligands (205). In addition, Fas (238) proteins have been shown to enable juxtacrine signaling pathways in the islet. Further, structural adhesion through cell adhesion molecules (CAMs) (72, 77) have been reported to be important factors in proper β-cell communication required for the robust regulation of insulin secretion in the islet.

β-cell communication

The islet is comprised mainly of β-cells, and insulin is the dominant hormone secreted by the islet (234). Defects in insulin secretion lead directly to diabetes, while abnormal secretion of islet glucagon and somatostatin do not seem to have significant physiological impacts on their own (195). In addition, the therapeutic benefits of insulin for the treatment of diabetes have been truly outstanding for the last 100 years, which has led most islet research to focus on the insulin-secreting β-cells (25). Thus, we will first discuss the important communication pathway between β-cells via gap junctions. While the glucose-sensing mechanism is present in each β-cell, the cellular communication properties of the intact islet lead to important aspects in the dynamics of insulin secretion. First, intact islets show near-zero insulin secretion at low glucose levels (< 3 mM), in contrast to a small but measurable leak of insulin seen from dispersed β-cells (23, 319). Further, dispersed β-cells exhibit significant heterogeneities in signaling and secretion that are not seen in intact islets (22, 100, 293, 363). From the perspective of whole body physiology, this is one of the most critical aspects since sustained insulin action during hypoglycemia can have dire consequences, including death (237). Second, the whole islet insulin-secretory response to glucose follows a steep sigmoidal secretory pattern with further increased insulin secretion at high glucose levels (>11 mM). Again, these are favorable behaviors for the restoration of euglycemia after a carbohydrate challenge. Third, the temporal profile of islet insulin secretion following a glucose bolus also mimics the biphasic insulin response seen in vivo, where a peaked first phase of insulin secretion lasting around 5 minutes is followed by a sustained second phase of pulsed insulin secretion with a period of 3–8-minutes (32, 155, 364). These patterns are lost upon dispersion of islets into single cells (292, 293).

These two secretory phases and insulin pulsatility have been observed in humans, non-human primates, and rodents, and are likely common, at least among mammals. The physiological relevance of these phenomena remains poorly understood, but evidence suggest that they are important players in blood glucose homeostasis (155, 252, 259). Data support a role for insulin pulsatility in suppressing hepatic glucose production (246), and therapeutic intravenous insulin pulses have been proposed as an improved diabetes treatment (158, 259). The physiological importance of these patterns are underscored by the facts that both the first-phase of insulin secretion and amplitude of the second phase insulin pulses are reduced and eventually lost as type 2 diabetes mellitus (T2DM) progresses (297). The pulsatility of insulin that is seen in vivo is recapitulated in isolated islets, and has been shown to arise from oscillations in membrane potential (10, 64), intracellular Ca²⁺ activity (31, 194), and cAMP (101). Interactions between β-cells play a key role in islet function, and these interactions have been shown to underlie the two phases and pulsatility of insulin secretion (25). The evidence for functional coordination among β-cells was initially observed by electrophysiology (76, 103), and followed by further measurements of interstitial K⁺ (288, 289), insulin secretion (328), and Ca²⁺ activity (339).

As described above, individual β-cells each contain the mechanisms required for GSIS, although there are significant differences in the insulin secretion dynamics between isolated β-cells and intact islets (218). β-cell gap junctions were initially detected by electron microscopy (280) and functionally characterized using dye coupling (256) and electrophysiology (103). After 20 years of research (reviewed in (45)), the biophysical properties of β-cell gap junctions pointed to connexin 36 (Cx36) as the major pore-forming unit in the islet (262), and this was quickly confirmed using mice lacking the Cx36 gene (Cx36^−/−), whose islets do not show synchronous oscillations in Ca²⁺ activity nor pulsatile insulin release (307). The β-cells in Cx36^−/− islets show unsynchronized, apparently random oscillations of Ca²⁺ activity similar to those seen in dispersed β-cells (369) (Fig. 7). Furthermore, the coordinated waves of Ca²⁺ activity following glucose stimulation are also missing from Cx36^−/− islets (26). A striking feature of the Cx36^−/− mice is that they are glucose intolerant even though the total insulin released in Cx36^−/− animals is similar to that of wild type littermates (155). Again, this glucose intolerance in Cx36^−/− mice is similar to what is measured in prediabetic and diabetic phenotypes (298). Thus, it appears to be the temporal dynamics of insulin secretion, rather than just the total amount secreted, that is critical for whole-animal glycemic control. This indicates an important physiological role for pulsatile insulin secretion.

Figure 7:

Dependence of intracellular Ca²⁺ dynamics on gap junction coupling in heterozygous Cx36 knockout Cx36^+/− (~50% gap junction conductance; A and B) and homozygous Cx36 knockout Cx36^−/− (~0% gap junction conductance; C and D) intact mouse islets. A phase map of synchronized (colored cells) and unsynchronized or lack (gray cells) of Ca²⁺ oscillations and representative Ca²⁺ intensity traces in four cells of an islet are shown for each islet type. Gradual loss of synchronicity in intracellular Ca²⁺ oscillations is observed as gap junction conductance is reduced with increasing knockout of Cx36. Figure reproduced with permission from (26).

One interesting consequence of this coordination is that the physical properties of the electrical activity coupling through gap junctions constrain the minimum and maximum islet size (24). The amount of coupling defines the velocity of the wave of electrical excitation, and for islet coordination, this needs to balance with the excitation rates of individual cells. If the islets are too small (<60 μm or ~8 cells in diameter) or too big (>200 μm or ~20 cells in diameter) the coordination across the islet will be unstable and not be sustained (167). These conclusions from mathematical modeling are born out in observations of islet morphology across species, where islet sizes remain similar from zebrafish (245) to large mammals (443). Rather, it is the number of islets that scale with the size of the animal. Notably, islet size distribution has been shown to be influenced by biological and pathological processes such as pregnancy and diabetes, and mathematical modeling has been used to estimate such processes based on the changes in islet size distribution (184).

While the Ca²⁺ activity is severely abnormal in Cx36^−/− islets, measurements of secretion as a function of glucose from isolated islets yield similar insulin levels from Cx36^−/− and wild type islets, even though dispersed β-cells show significantly increased insulin release for both genotypes. These data suggest that additional cellular communication mechanisms are required for normal islet function (23). As detailed elsewhere in this review, several possible communication mechanisms have been hypothesized between β-cells in the islet, including paracrine signaling by nitric oxide (NO) (133) and purinergic signaling (145). As discussed below, EphA-ephrin-A juxtacrine signaling is also known to be needed for normal GSIS (205). The observed secretion data from Cx36^−/− islets (23) is consistent with an expected role of EphA-ephrin-A signaling between β-cells in regulating insulin secretion at low glucose levels (205). Regardless of the exact molecules involved, these data make it clear that multiple regulatory mechanisms are necessary for normal islet function.

These data indicated that gap junctions are not the only intercellular communication mechanism in the islet, but deletion of β-cell gap junctions still leads to a situation that mimics many type 2 diabetes phenotypes. In Cx36^−/− mice, the first-phase of insulin secretion is significantly reduced in vivo (as well as from the isolated islets) although the total insulin output remains near normal (155). These data suggest that first-phase secretion requires synchronous pulses of insulin coming from the islet. Second-phase insulin oscillations are also reduced in Cx36^−/− mice, which suggests a similar role of synchronous islet cell activity for these oscillations. In both phases, the data are similar to those seen in T2DM, where first-phase peak and second-phase oscillations in insulin secretion are reduced and eventually lost as the disease progresses (297). Despite this similarity, it remains unclear whether lost β-cell coupling is a cause or symptom of T2DM, although the consequences of Cx36 deletion are analogous to pathological observations of patients with T2D (252). One indication that gap junction coupling may be, at least in part, a cause rather than a symptom of type 2 diabetes is that coupling strength is lost in aging (421), and this correlates with increased risk of the disease.

Another physiological consequence of whole islet synchronization is the regulation of islet blood flow, which has been shown to increase upon blood glucose elevation independent of the exocrine tissue blood flow (271, 272). It was shown that this increase in blood flow is absent in islets within Cx36^−/− mice (357), which suggests that whole islet synchronous electrical activity is needed to couple into the vasculature and prompt increased flow. Islet blood flow has been shown to depend on pericyte activity around pre-capillary sphincters (8). This raises the intriguing possibility of electrical coupling between islet β-cells and the vasculature. Given recent discoveries that exocrine and endocrine blood flows may be intertwined (102), it will be even more important to understand the mechanisms underlying the observed islet-specific changes in blood flow (271, 357).

Considerable effort has gone towards understanding the role that β-cell heterogeneity plays in islet function. It has long been understood that dispersed β-cells exhibit heterogeneous behavior in glucose transport (388), insulin biosynthesis (292), glucose metabolism (22), Ca²⁺ activity (60), and most importantly in their functional output, insulin secretion (163, 336, 399). Gene expression studies have discovered sub-populations of β-cells that may underlie these observed heterogeneities (4, 33, 38). However, measurements of these parameters indicate that most of them are largely eliminated within the intact islet. A consensus model has developed where β-cells with higher sensitivity to glucose, for example from differences in glucose metabolism or channel activity, are stimulated first, and that they bring along the cells with lower excitability (60, 254, 293, 422). To test these models, though, it is challenging to measure cellular excitability within intact islets under normal conditions because of the strong gap junctional coupling and other communication pathways (23, 26, 369). This challenge can be met through the creation of well-defined heterogeneous β-cell populations in the islet, which has been done in three ways. The first actively introduced heterogeneity was obtained using the non-uniform infection of adenoviral Cre-mediated incomplete deletion of glucokinase (217) or, using gene deletion to introduce a variegated expression of a dominant negative Kir6.2 subunit that leads to non-functional K_ATP channels in a portion of the islet β-cells (207). Both of these first two approaches manipulated key glucose-sensing molecules that exhibit naturally occurring heterogeneities that are key for proper islet function (59). The third approach to generate heterogeneous β-cell populations in the islet was to use microfluidics to fabricate a non-uniform glucose-stimulation pattern throughout the islet (24, 320). More recently, a fourth approach based on the use of optogenetics to stimulate or hyperpolarize a single β-cell or a localized cluster of β-cells was used to create precise heterogeneities within the islet (186).

Using the first approach, the primary glucose sensor, glucokinase, was genetically-deleted from approximately 30% of β-cells, which created a heterogeneous distribution of metabolic responses within the islet (295). Despite this heterogeneity, overall islet electrical activity and insulin secretion profiles from these islets were unchanged from wild type islets. These data are consistent with our understanding of how cell coupling leads to a homogeneous response from a heterogeneous population of β-cells. This is also consistent with the demonstration that incretin action plays a role in maintaining coordinated islet activity during lipotoxicity in mice and humans, although the molecular determinants of that effect remain to be defined (164).

The second approach to create two distinct populations of β-cells in an islet utilized the mosaic expression of a dominant-negative Kir6.2[AAA] transgene, where the K⁺ channel forming subunit of the K_ATP channel becomes nonfunctional (207). The loss of K_ATP channel function eliminated glucose-stimulation of Ca²⁺ influx and insulin secretion. In dispersed cells, this causes glucose-independent hyperexcitability on a cell-by-cell level so that β-cells exhibited pulses of Ca²⁺ activity transients even at very low glucose levels (2 mM). Intact islets from the Kir6.2[AAA] mice showed a mosaic pattern with 70% of β-cells expressing the mutated gene and the remaining cells showed normal K_ATP channel function. The data from these islets showed coordinated synchronous [Ca²⁺]_i oscillations at stimulatory glucose levels. These data were in agreement with the islet syncytium hypothesis that proposed cells within the islet create a uniform membrane potential by ‘sharing’ K_ATP channels through gap junctions (81, 355). Importantly, no β-cells in the islet exhibited Ca²⁺ activity transients at nonstimulatory glucose concentrations (319), which shows that even a small percentage (~30%) of functional β-cells is sufficient to polarize the entire islet at sub-stimulatory glucose levels. Functionally, this effect prevents inadvertent insulin secretion at low glucose levels from any emergent highly excitable β-cell in an islet, which could have dire consequences for the animal or patient during hypoglycemia. The islet configuration accomplishes this safeguard through gap junction coupling that clamps the membrane potential of all neighboring β-cells to what appears to be an average value (293, 319). The gap junctional coupling appears to average the stimulatory glucose level over the entire islet, and thus overcoming the effect of individual cell heterogeneity (369).

The third approach was designed around a novel two-sided microfluidic device that created a concentration gradient of stimulant across the islet. Using glucose as the stimulant, the gradient could be measured using fluorescently-labeled glucose and NAD(P)H intensities (320). Experiments performed with this device confirmed that β-cells are not metabolically coupled in the islet (295). Looking at electrical coupling using Ca²⁺ activity as a surrogate marker, however, showed a surprising sharp line of demarcation between β-cells with normal excited Ca²⁺ activity oscillations on the elevated glucose side of the gradient and those clamped at basal Ca²⁺ activity without a gradient Ca²⁺ activity between them. This line could be translated across the islet by changing the glucose gradient (24). These data suggest that β-cells within an islet are controlled largely by their own glucose sensing apparatus, and that gap junctional coupling is not sufficient to drive Ca²⁺ activity into β-cells where the glucose concentration is below the stimulation threshold (~ 7 mM glucose). Still, the cells are strongly-enough coupled to coordinate the Ca²⁺ activity throughout the regions above this threshold. These data show that the islet response must be understood in terms of both the local glucose environment that dictates the number of open K_ATP channels for each β-cell and the overall electrical syncytium created by gap junctions. The two-sided microfluidic device creates a glucose gradient so that a β-cell located just below the stimulatory glucose level would be expected to have only a small number of open K_ATP channels, yet it still resists the coupling force of its depolarized neighbors. At the same time, though, that cell is also coupled to other cells that are experiencing even lower glucose levels, which would be expected to provide additional resistance to depolarization. Many aspects of the islet behavior using this microfluidic device could be accurately described by existing mathematical models (254, 287). However, predicting the existence and precise location of the sharp line of oscillation activation require inclusion of a glucose-dependent component of gap junction conductance (24).

The advent of optogenetics (83) permitted the stimulation of specific β-cells or regions of β-cells. Elegant experiments combining optogenetics and live cell imaging of islets led to the concept of “hub cells” (186), more recently dubbed “leader cells” (335), that play an important pacemaking role in controlling islet hormone secretion. This concept has become controversial, and it remains unclear whether the resulting model accurately reflects the actual islet dynamics. The situation is complicated by the use of “hubs” to describe both cells that are highly connected and cells that can silence the islet when inactivated. While, a cell could fit both definitions, it is certainly possible to have one without the other. From a mathematical modeling perspective, it appears that the hub cell model is not needed to explain the data (422). Rather, the percolation model (26) expansion of standard β-cell electrophysiological models (34) works fine to explain the observed phenomena (23, 24, 186), and recent consideration of the biophysical properties of β-cells suggests that hubs are physically improbable (341). The percolation model continues to explain well how overall connectivity and Ca²⁺-wave velocities change as a function of gap junctional coupling strength (24, 26). A central prediction of the percolation model is the presence of β-cells with only two connections, given that the average β-cell is connected to 4.6 other beta cells (26, 110). Combined with the significant spatial heterogeneity within the heavily vascularized and innervated islet, it is expected that many β-cells would have only 2 connections. These architectural features of the islet are critical factors that led to the development of a percolation model (25, 100, 422). The β-cells with exactly 2 connections to other β-cells would, even though they are not highly connect, appear as “hubs” in some experimental situations, and their inactivation would be expected to disrupt whole islet synchronization.

The connexins expressed in the β-cell not only form homotypic gap junctions with other β-cells, but can also form heterotypic gap junctions with other islet cell types. As described below, there is no evidence for functional gap junctions between β- and α-cells, but one report using optogenetic activation of β-cells indicates electrical coupling between β-cells and δ-cells via gap junctions (53).

The potential importance for juxtacrine regulatory mechanisms was suggested by experiments with Cx36^−/− mice (23). One pathway, EphA/Ephrin-A signaling, has been shown to be important for proper insulin secretion in islets. Several Eph receptors and ephrin ligands are expressed in islets, with EphA5 and Ephrin-A5 highly expressed in β-cells (95, 205, 222). Stimulation of EphA5 forward signaling suppresses insulin signaling, while stimulation of Ephrin-A5 reverse signaling stimulates insulin secretion (205). The suppression of insulin secretion by EphA5 forward signaling was postulated to occur through the increased polymerization of F-actin filaments upon the inhibition of Rac1 activity, a Rho-GTPase involved in actin cytoskeleton organization and endocytosis (205, 242). On the contrary, Ephrin-A5 reverse signaling enhanced Rac1 activity and decreased F-actin density, leading to increased insulin secretion. At low glucose conditions, EphA5 activity dominates, resulting in the net suppression of insulin in the islet. At high glucose conditions, the PTP-mediated dephosphorylation of EphA5 inhibits downstream signaling, causing Ephrin-A5 reverse signaling to dominate and stimulate insulin secretion. In keeping with this model, more recent studies demonstrated that the pharmacological inhibition of Eph receptors in islets resulted in a selective increase in insulin secretion at high glucose conditions as well as improved glucose tolerance in mice (128, 178). Although the EphA/Ephrin-A system provides an important homeostatic mechanism of β-cell communication via cell-cell contact, further study is necessary to establish its relevance in diabetes.

The cell surface receptor Fas and its associated ligand FasL have also been reported to affect β-cell communication. In addition to playing a central role in β-cell apoptosis (94, 238, 270), the modulation of Fas signaling by FLICE-inhibitory protein (FLIP) has been shown to affect glucose sensitivity and insulin secretion in mice by controlling the expression of PDX-1, a β-cell specific transcription factor associated with T2D (148, 236). These studies demonstrate the dual function of the Fas pathway as a regulator of β-cell turnover as well as a regulator of insulin secretion. However, how the β-cells utilize this pathway to communicate at the intercellular level remains a mystery.

Various CAMs are expressed in pancreatic islets, primarily E- and N-cadherins (78) and NCAMs (72). Cadherins form cell surface homodimers and play important roles in the formation of adherens junctions as well as cellular signaling through catenin and other signaling pathways. In islets, loss of E- and N-cadherin function resulted in the inability of β-cells to aggregate and disrupted proper islet development in mice (78). Beyond the simple role in holding islets together, CAMs play a direct part in regulating GSIS, as loss of E-cadherin (408, 428) and NCAM function (274) attenuated GSIS in rodent islets. One possible pathway in which cadherins regulate GSIS is through the interaction with β-catenin (73, 368). At a whole-islet level, evidence suggests that CAMs play an important role in maintaining proper islet architecture through differential expression levels in the different islet cell types, with NCAMs preferentially expressed in non β-cells (72). Rodent islets dispersed in vitro were able to spontaneously reaggregate into native-like pseudoislets (147, 261) with restored hormonal activity (310). Disruption of NCAM resulted in the inability of proper reaggregation (72, 108), with α-cells distributed throughout the islet. Although the general function of NCAM is independent of Ca²⁺, there is evidence suggesting that this architectural arrangement may be driven at least in part by Ca²⁺. This is supported by the observation that reaggregation of β-cells in vitro is dependent on the presence of Ca²⁺, whereas non β-cell reaggregation was relatively unaffected (329). Cadherins have also been implicated to have a role in controlling β-cell proliferation and apoptosis. In vivo knockout of E-cadherin in β-cells resulted in higher cyclin D2 levels and increased β-cell mass in mice (408). However, despite increased β-cell mass, E-cadherin KO mice exhibited reduced GSIS. It is interesting to note that these islets also exhibited reduced connexin 36 levels. Dispersed human β-cells displayed lower levels of apoptotic cells when allowed to bind to E- and N-cadherin coated surfaces (281). Taken together, these results suggest that the islet utilizes CAMs as a method of contact communication to regulate both the proper formation and maintenance of islet structure and function. In the context of diabetes, mice with high-fat diet-induced diabetes had significantly lower levels of E- and N-cadherin junctions in β-cells and displayed increased β-cell mass yet reduced GSIS (109). Further investigation on the effect of CAMs on the pathophysiology of diabetes will be necessary to explain these observations.

Tight junctions involving claudins and occludins have also been observed in islets. Despite being first documented over 40 years ago in human islets (279), our understanding of their function in the islet is extremely limited. Tight junctions have been found to be associated with gap junctions (250), and they were also found to increase in number upon insulin secretion and as a function of glucose concentration (278, 350), suggesting a role in regulating domain accessibility on the plasma membrane. More recent studies have reported an increase in expression of claudins and occludins in the islet during pregnancy (313, 345) and aging (74, 221). In particular, Cldn4 was found to be downregulated in T2D db/db mouse islets, and genetic knockout resulted in impaired glucose tolerance.

It is important to keep in mind that the different pathways presented above are not independent from each other, but in fact act as an interconnected system. The downstream signaling of Eph receptors and CAMs through Rho GTPases has been well characterized in other cellular systems as well as in islet cells. Ephs/Ephrins have also been shown to be involved in apoptosis signaling in different cell types (188), including the caspase-8 pathway that Fas signals through. PDX-1 was identified as an activator of E-cadherin expression in β-cells (243). Understanding the complexity presented by the various contact-mediated signaling pathways islet cells utilize to communicate, as well as the communication between the pathways themselves, will likely be a crucial for the treatment of islet dysfunction and diabetes.

α-cells

The discovery of insulin and its utilization for the treatment of diabetes ranks among the most important medical advances of the 20^th century, and its therapeutic benefits have led most islet research to focus on the β-cells. In contrast, the actions of glucagon secreted by the α-cells were thought to be counterproductive for the treatment of diabetes, although the physiological importance of the glucagon-secreting α-cells is becoming clear (393). During hypoglycemia, the α-cells secrete glucagon that stimulates glucose output from the liver, whereas under normoglycemic conditions, glucagon release is suppressed (183). Glucagon affects glucose production by the liver, but also potentiates insulin secretion, thus directly affecting overall islet function (395). While there are no recognized diseases resulting directly from glucagon secretion defects, there are clearly α-cell differences seen in diabetes phenotypes (98, 99, 111), and there is also evidence that α-cell malfunction is associated with diabetes (91, 172).

Pancreatic islet α-cells play an important role in regulating blood glucose homeostasis by secreting hormonal glucagon at low glucose levels (hypoglycemia) to stimulate hepatic glucose production from glycogen stores in the liver, thus restoring euglycemia. As a counterregulatory hormone of insulin, glucose regulation of glucagon secretion by α-cells tracks in opposition to insulin, where glucagon secretion is stimulated at low glucose levels (< 4 mM) and inhibited at high glucose levels (> 7 mM), and their hormonal behaviors are opposite in almost every respect (158, 183, 393, 394). It is of interest to note that islet architecture and the composition of α-cells within islets significantly differs across species, from rodents (~15–20%) to humans (~40%), indicating a possible evolutionary adaption (376). Although the more commonly known cause of both type 1 (T1D) and type 2 (T2D) diabetes is the misregulation of insulin due to the destruction of β-cells or compromised β-cell functionality, many studies have shown that uncontrolled hypersecretion of glucagon in α-cells, resulting in hyperglucagonemia at all glucose levels, exacerbates hyperglycemia (55, 129, 134, 154, 352, 394). Still, the role of α-cells and glucagon secretion in diabetes remains controversial. One extreme view is that hyperglucagonemia (rather than insulin deficiency) is the primary determinant of hyperglycemia; if the glucagon excess could be corrected, the lack of insulin would not be deleterious, at least with regard to glucose control (393). This model does not explain how a pancreatectomy (which eliminates both α- and β-cells) causes severe hyperglycemia (111), and it ignores the well-documented effects of insulin on muscle, which lack glucagon receptors (303). Still, the data supporting this view cannot be ignored, and in light of these data, much effort has been invested in the treatment of diabetes to restore euglycemia either by suppressing glucagon secretion or antagonizing glucagon receptor action (154, 192, 217, 393, 413, 433). These glucagon therapeutics effectively restore euglycemia in the absence of insulin, albeit little is known about their modes of action. Moreover, clinical trials based on glucagon receptor antagonism often display undesirable side effects, such as increased plasma lipids, liver fats and enzymes, and elevated blood pressure, which render them unsafe for treating diabetes in their current status (69, 143, 193).

Owing to the insulinocentric dogma that had persisted over the last century since the discovery of insulin as a treatment for diabetes in 1921 (17), the mechanisms for insulin secretion in β-cells have been well characterized. In contrast, the mechanisms that regulate glucagon secretion in α-cells remain controversial to date, with many still carrying the notion that it plays a minor regulatory role in glucose homeostasis (138). It was only in recent years as glucagon garnered more attention in the field as a crucial regulator of normal glucose homeostasis that studies of regulation of glucagon secretion in α-cell escalated. The various models that describe regulation of glucagon secretion can be categorized into three main groups: i) intrinsic α-cell mechanism, ii) paracrine interactions, and iii) juxtacrine interactions (Fig. 8). Note that these models are not mutually exclusive and can work synergistically to control glucagon exocytosis. It is worth noting that glucose is not the only physiological signal that influences glucagon secretion. In addition, some amino acids like glutamate and arginine bind to receptors on the α-cell, and appear to directly stimulate the release of glucagon (220, 439). Here we will describe the α-cell intrinsic models of how glucose might regulate glucagon secretion, and the models based on paracrine and juxtacrine islet cell communication pathways will be discussed later.

Figure 8:

Intracellular and intercellular pathway mechanisms that regulate glucagon secretion in the α-cell.

There are two postulated models for intrinsic α-cell regulation of glucose-mediated glucagon secretion. The first is based on α-cell plasma membrane excitability and is sometimes referred to as the “electrophysiological model”, and the second is based on signaling through cyclic adenosine monophosphate (cAMP). The electrophysiological model was built on similarities with the electrical excitability of β-cells and the signaling pathways that drive β-cell glucose-stimulated insulin secretion (GSIS) (51). α- and β-cells contain comparable signaling pathways: glucose transporters, the glycolytic enzyme glucokinase, ATP-sensitive K⁺ (K_ATP) channels, high-voltage-gated Ca²⁺ channels, and secretory granules (40, 136, 156, 157). The functional commonalities between α- and β-cells begin with their metabolic response to changes in glucose levels, where the α-cell response is similar to that of β-cells in direction, but not in amplitude (215). The ratio of anaerobic:aerobic glycolysis is also reportedly higher in α-cells (346). Another study indicates that not only is [ATP] 2-fold higher (6.5 mM) in α-cells than in β-cells, but also that the [ATP]/[ADP] ratio in α-cells does not change following increase in glucose from 1 to 10 mM (86). These data would suggest that electrophysiological changes in the α-cell arise from non-metabolic pathways, perhaps by direct G-protein modulation of channel activity. This would argue against an intrinsic α-cell mechanism, although other studies have indicated a minimal glucose-dependent change in the α-cell [ATP]/[ADP] ratio (308).

Unlike the well-defined GSIS mechanism in β-cells where glucose metabolism increases intracellular ATP to ADP ratio, resulting in membrane depolarization by the closure of K_ATP channels, activation of voltage-gated Ca²⁺ channels, thus triggering Ca²⁺ influx and subsequent insulin exocytosis (161, 268), regulation of glucagon secretion in α-cells is less straightforward and has multiple conflicting reports. One such hypothesis postulates that at low glucose levels, the amplitudes of the action potentials (AP) of α-cells reach an appropriate membrane voltage level that allows for the opening of N-type Ca²⁺ channels (235) and P/Q-type Ca²⁺ channels, identified as the main voltage-gated Ca²⁺ channel which contributes to glucagon secretion (442), leading to increased glucagon release (268). Notably, under cAMP-elevating conditions by adrenaline (via β-adrenergic receptors) or forskolin stimulation in rat pancreatic α-cells, L-type Ca²⁺ channels are the predominant Ca²⁺ channels that facilitate Ca²⁺ influx and glucagon secretion (136). Such membrane AP amplitude is maintained presumably by the hyperpolarizing effects of voltage-gated K⁺ channels (K_V) and calcium-activated K⁺ channels (K_Ca) (88, 370), which prevents the depolarization-dependent inactivation of α-cell AP firing, thus activating tetrodotoxin-sensitive voltage-gated Na⁺ channels (Na_V channels) that discharge high amplitude APs (18, 132, 141). As glucose level rises, membrane depolarization occurs in a similar pathway as β-cells, with the blocking of K_ATP channels by increased ATP to ADP ratio. Contrary to β-cells, membrane depolarization in α-cells leads to the voltage-dependent inactivation of Na_V channels which lowers the overall membrane AP amplitude and voltage level, inactivating the N- and P/Q-type Ca²⁺ channels, suppressing Ca²⁺ influx and response, and consequently inducing the inhibition of glucagon secretion (141, 235, 442). This hypothesis, however, was disputed by a study on isolated pancreatic islets of K_ATP channel knockout (Kir6.2^−/−) mice which still maintained regulation of glucagon secretion albeit the loss of K_ATP channel functionality (257). In a paradoxical concept, membrane hyperpolarization, instead of depolarization, was postulated as the stimulus for regulation of glucagon secretion at high glucose levels. Such membrane hyperpolarization in α-cells was observed in intact mouse and rat islets upon glucose stimulation (239). Rising intracellular ATP levels generated from glucose metabolism can cause a myriad of outcomes such as sequestration of Ca²⁺ into the endoplasmic reticulum (ER) by the activation of sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA), and the opening of a 2-pore domain K⁺ channel, TWIK-related acid-sensitive K⁺ channel 1 (TASK1), both of which leads to membrane hyperpolarization, store-operated Ca²⁺ channel inactivation (144, 225, 404) and subsequent inhibition of glucagon secretion (75). Despite their differing yet potentially synergistic mechanisms (415), both hypotheses rely on the same consequence of elevated Ca²⁺ activity at low glucose levels and reduced Ca²⁺ influx at high glucose levels as the basis of activation and suppression of glucagon secretion, respectively. Several labs have demonstrated that unlike electrically coupled β-cells which display highly synchronous Ca²⁺ oscillations with glucose stimulation that are tightly coordinated with insulin secretion, α-cells are decoupled and have highly heterogeneous Ca²⁺ response to glucose (214, 215, 223). There appears to be multiple populations of α-cells within an islet, as characterized by their Ca²⁺ activities, where some exhibit reduced Ca²⁺ oscillations while others show unchanged or increased Ca²⁺ activity upon glucose stimulation. Hence, regulation of glucagon secretion likely occurs in a Ca²⁺-independent pathway since Ca²⁺ activity remains elevated at high glucose levels in some cells, decoupled from glucagon secretion. With α-cell Ca²⁺ influx being independent of glucose concentrations and persisting throughout the physiological glucose range, it likely performs an auxiliary role for glucagon exocytosis, given that it is still necessary for glucagon secretion, but does not directly coordinate glucagon secretion. Furthermore, published data show that α-cells dispersed from murine islets exhibit higher basal glucagon secretion and an opposite glucagon secretion response to glucose as compared to intact islets, suggesting that communication between the islet cells in the form of paracrine and juxtacrine interactions is required for proper regulation of glucagon secretion (215, 310). Data from Gromada and co-workers corroborate these observations that regulation of glucagon secretion within dispersed rat α-cells also track in opposition to that of intact islets, where glucose stimulate rather than inhibit glucagon secretion (275). This is further supported by a recent study which reported non-physiological regulation of glucagon secretion in dispersed human α-cells while physiological glucagon secretion dynamics were maintained in α-cells within intact human islets (276).

The other α-cell intrinsic mechanism postulated to explain glucose-inhibition of glucagon secretion involves the second messenger cAMP and its signaling pathways. Based on direct imaging of cAMP reporters, biochemical and electrophysiological approaches coupled with pharmacological perturbations on cAMP and its downstream signaling effectors, the role of cAMP on glucagon secretion, both dependent and independent of intracellular Ca²⁺ activity, were investigated (79, 136, 216, 435). Several studies demonstrated that raising intracellular cAMP levels by pharmacological activation of adenylyl cyclase (by adrenaline, forskolin, etc.) or direct addition of cell-permeable cAMP analogs in α-cells trigger an increase in intracellular Ca²⁺ levels and activity, which in turn enhance glucagon secretion (79, 136, 185, 225). At low glucose levels, the main drivers of Ca²⁺-dependent glucagon exocytosis are N-type Ca²⁺ channels (79, 136, 216, 235). Importantly, cAMP was shown to orchestrate a switch in the Ca²⁺ channel-dependent glucagon exocytosis in α-cells of intact mouse islets from N-type to L-type Ca²⁺ channels at hypoglycemic levels (79, 136, 216). This was attributed to the activation of protein kinase A (PKA) by a small increase in intracellular cAMP levels which causes the phosphorylation and inhibition of N-type Ca²⁺ channel activity, while the activation of low-affinity cAMP sensor Epac2 by a large increase in cAMP concentration promotes the influx of Ca²⁺ through L-type Ca²⁺ channels. Treatments with GLP-1, where there is low GLP-1 receptor density in α-cells, low concentration of forskolin (1 – 10 nM) or adrenaline (50 pM – 500 nM), recapitulate the PKA-dependent inhibitory effect of low elevations in cAMP levels on N-type Ca²⁺ channels, thus lowering glucagon secretion (79). This was further supported by the activation of the low density of GLP-1 receptors in α-cells with GLP-1, leading to small elevations in cAMP levels and subsequent PKA-dependent inhibition of P/Q-type Ca²⁺ channels and reduced glucagon secretion (304). On the other hand, treatments with high concentrations of forskolin (0.1 – 10 μM) or adrenaline (5 μM), where there is high density of β-adrenoceptors in α-cells, mirrors the stimulatory effect of high elevations in cAMP levels which activates Epac2 and drive Ca²⁺ influx through L-type Ca²⁺ channels, resulting in increased glucagon exocytosis (79). In addition to voltage-gated Ca²⁺ channels, cAMP elevation increases intracellular Ca²⁺ levels in mouse α-cells by activating Ca²⁺ release from Ca²⁺ stores in the endoplasmic reticulum (ER), thus inducing a store-operated current that depolarizes and activates the α-cell (185, 225). Furthermore, cAMP was shown to amplify glucagon exocytosis by mobilizing readily releasable pool of granules (136). Notably, as mentioned above, glucose has the opposite effect of hyperpolarizing the membrane by filling up intracellular Ca²⁺ stores which inactivates store-operated Ca²⁺ channels and inhibit glucagon secretion. In a recent study, cAMP was implicated as a key messenger molecule that is directly involved in the α-cell intrinsic mechanism of regulation of glucagon secretion (435). The authors measured subplasmalemmal cAMP concentrations, [cAMP]_pm, in live intact mouse and human islets labeled with fluorescent cAMP biosensor adenoviruses via total internal reflection fluorescence (TIRF) and confocal imaging. [cAMP]_pm tracks in parallel with glucagon secretion upon a series of dynamic glucose challenges where hypoglycemic glucose levels (1 – 3 mM) triggers an increase in [cAMP]_pm and glucagon levels while hyperglycemic conditions (> 7 mM glucose) lower [cAMP]_pm and glucagon levels. Such glucose-induced response between [cAMP]_pm and glucagon secretion is shown to be uncorrelated with [Ca²⁺]_pm and independent of paracrine insulin and somatostatin interactions as established by various treatments utilizing external insulin and somatostatin stimulations, and their respective insulin and somatostatin receptor antagonists. This was further substantiated by similar glucose modulation of [cAMP]_pm observed in dispersed α-cells as compared to intact islets. Moreover, glucose inhibition of glucagon secretion was abolished by preventing the degradation of cAMP with phosphodiesterase (PDE) inhibitor 3-isobutyl-1-methylxanthine (IBMX) whereas lowering of [cAMP]_pm using a PKA inhibitor resembles hyperglycemic inhibition of glucagon release. Although the evidence indicates that regulation of glucagon secretion is a result of direct modulation of cAMP in α-cell, the actual mechanism of how glucose controls cAMP levels remain unknown. Contrary to the proposed α-cell intrinsic regulation of glucagon secretion mechanism of cAMP, the importance of paracrine insulin and somatostatin signaling on lowering cAMP was highlighted, which is more pronounced at high glucose concentrations, to mediate glucagon secretion (107). Hence, it is likely that the α-cell intrinsic mechanism of regulation of glucagon secretion by cAMP modulation dominates at hypoglycemic conditions while paracrine regulation glucagon secretion is of paramount importance during hyperglycemia.

Beyond the regulation of glucagon secretion into the blood stream in response to glucose, the α-cell also provides paracrine signals within the islet. Notably, glucagon itself functions as a paracrine signal to the δ-cells. The glucagon receptor is highly expressed by the δ-cell (222), and activation of this receptor is expected to stimulate somatostatin secretion during hypoglycemia. Indeed, this was observed in ex vivo perfused pancreas studies (283). Glucagon has also been shown to stimulate further glucagon secretion directly through an autocrine positive-feedback loop by activating glucagon receptors, stimulating cAMP production and thereby increasing glucagon release (232). In addition to glucagon, α-cells have also been shown to secrete the alternatively spliced proglucagon product, GLP-1, especially after stimulation of GPR142. This islet-generated GLP-1 can significantly improve β-cell function (224). Finally, the α-cell also secretes acetylcholine, which has been shown to signal to β-cells (321, 323, 420).

As discussed, α-cells have a positive feedback loop with glucagon stimulating further secretion, but they also release glutamate as a positive autocrine signal (58). While glutamate secretion can have significant effects in isolated islets, the action of autocrine glutamate signaling in vivo may be limited as circulating glutamate can also stimulate glucagon secretion. High levels of amino acids in the blood plasma can arise after a high protein meal. In this case, enhanced glucagon secretion may be useful to prevent hypoglycemia, especially following a low carbohydrate meal. This action would be similar to what has been proposed for circulating L-dopa, which has the effect of inhibiting insulin secretion (397).

The α-cell does not express the same Cx36 gap junction component found in the β-cell, but instead expresses Cx46 (222), which does not form heterotypic gap junctions with the β-cell Cx36 (209). Most data show that there are no coordinated oscillations among α-cells or between α- and β-cells (213, 215), although some α-cells have been reported to show synchronous pulses of Ca²⁺ activity under hyperglycemic conditions (215, 223). Taken together, though, the current consensus is that α-cells do not communicate via gap junctions.

Less is known about potential juxtacrine pathways involving the α-cell. While the α-cells express EphA receptors, ephrin ligands do not appear to be expressed in α-cells. EphA4 appears to be the main Eph receptor expressed in both human and mouse α-cells (95, 210), although EphA7 is also expressed in mouse (173) and its transcription has been shown in human α-cells (222).

α-cell communication

The α-cell is closely related to the β-cell, but in terms of hormone secretion, they exhibit opposite actions in response to glucose – while insulin secretion increases with glucose, glucagon secretion decreases (394). While the basic molecular pathways that lead to glucose-stimulated insulin secretion are well-understood, there is still not a consensus around the corresponding mechanisms underlying the α-cell response to glucose (138, 144). α-cell function remains poorly understood (323, 420), but it is generally assumed that α-cell regulation requires the juxtaposed presence of either β-cells, δ-cells, or both. The need for juxtaposed β-cells is evidenced by the loss of glucose inhibition of glucagon secretion as β-cell function is lost in either Type 1 (134) or during advanced stages of Type 2 diabetes (352). Similarly, normal glucagon secretion patterns from islets is disrupted when the secretory product of islet δ-cells, somatostatin, is deleted (151). However, somatostatin alone does not inhibit glucagon secretion from dispersed α-cells (107, 215), consistent with the finding that GIGS is lost in dispersed α-cells (215). Thus, cellular communication in the islet is expected to play a central role in the regulation of glucagon secretion. Here we describe the models and data supporting both paracrine and juxtacrine communication pathways in the islet for the regulation of glucagon secretion from the α-cells.

As discussed above, the intrinsic α-cell mechanism is insufficient to account for regulation of glucagon secretion independently. An intricate interplay and crosstalk among multiple intercellular pathways, along with the intrinsic mechanisms, is likely necessary to direct regulation of glucagon secretion. Paracrine signaling is one such intercellular mechanism in the pancreatic islet that has been highlighted to influence glucagon secretion in α-cells. It follows that neighboring islet cell types, β- and δ-cells, secrete paracrine factors into the islet extracellular space to interact with α-cells via their respective membrane receptors. The known paracrine signaling molecules secreted by islet β-cells which interact with α-cells include insulin, zinc (Zn²⁺), γ-aminobutyric acid (GABA), serotonin, glutamate and γ-hydroxybutyrate (GHB) (138, 144). Somatostatin, unlike insulin and glucagon, is not a hormone but a paracrine factor secreted by δ-cells in the pancreatic islet to inhibit insulin and glucagon secretion from β- and α-cells respectively. The effects of each paracrine factor on regulating glucagon secretion in α-cells have been reviewed (138, 144). Another study demonstrated that islet cell type specific α/β pseudo-islets failed to inhibit glucagon secretion at high glucose concentration without the supplement of exogenous somatostatin (310). Likewise, proper intact islet-like glucose suppression of glucagon secretion can only be recapitulated in co-cultured purified α- and δ-cells with the addition of exogenous insulin, highlighting the importance of paracrine modulation of glucagon secretion. In this section, we examine the intercellular paracrine signaling mechanisms of the various paracrine factors in the coordination of glucagon secretion in pancreatic islet α-cells.

It was demonstrated that the microcirculation within islets flows from β- to α-cells, which results in a scenario where α-cells lie downstream of β-cells, reinforcing the paracrine regulatory role of β-cell secretory products on α-cell glucagon secretion (42, 272, 338, 372–374). Among the various β-cell secretory products, insulin is the most prominent paracrine molecule that inhibits glucagon secretion in α-cells upon its triggered release by high glucose. This was supported by numerous studies including i) anti-insulin serum treatment which abolished the effects of glucose inhibition of glucagon secretion in isolated perfused rat pancreas (244), isolated rat islets (113), and isolated perfused human pancreas (56); ii) inhibition of glucagon release induced by addition of exogenous insulin in αTC-6 cells, In-R1-G9 cells (189, 198), αTC1–9 cells, intact mouse islets (308) and isolated rat islets (113); iii) loss of regulation of glucagon secretion upon insulin receptor antagonism by small interference RNA (siRNA) knockdown of insulin receptors in intact mouse islets and αTC-6 cells (87), insulin receptor antagonist S961 treatment in intact mouse and human islets (107, 276), and α-cell-specific insulin receptor knockout (αIRKO) mice model (192). Paracrine insulin signaling in α-cells is activated by insulin receptors that are highly expressed in pancreatic α-cells (113, 189, 198). Insulin signaling pathways are highly diverse (337) and several mechanisms have been proposed to regulate glucagon secretion in α-cells (16). Insulin treatment in dispersed mice α-cells negatively regulated glucagon secretion by decreasing the sensitivity of their K_ATP channels to ATP inhibition, causing K_ATP channel activation, in a phosphatidylinositol-3-kinase (PI3K)-dependent manner (219). An opposing study in rat α-cells revealed that insulin transiently increase K_ATP channel activity, leading to reduction of electrical activity and suppression of glucagon secretion, in a PI3K-independent manner (113). This discrepancy could be a result of different animal models used in both studies. The glucagon suppressive action of insulin via the PI3K signaling pathway was further substantiated by Kaneko et al. on In-R1-G9 clonal α-cells where insulin inhibition of glucagon secretion and glucagon gene expression were mediated by the recruitment and activation of PI3K in the plasma membrane (189). Further, Ravier and Rutter established that insulin suppressed Ca²⁺ oscillations in dissociated α-cells from mouse islets and inhibited glucagon secretion in αTC1–9 cells and intact mouse islets (308). Such insulin-mediated glucagon inhibition was reversed upon PI3K inhibition by wortmannin. Similar observation of PI3K-mediated glucagon suppression was made by Chen and Östenson in isolated rat islets (65). Consistent with a PI3K-dependent mechanism, it was shown that insulin secreted by β-cells at high glucose in intact mouse islets inhibits glucagon secretion in α-cells by binding to α-cell insulin receptors to activate PI3K, which phosphorylates Akt and thereafter result in phosphodiesterase 3B (PDE3B) phosphorylation, leading to degradation of cAMP and reduced cAMP/PKA signaling (107). It is worth noting that in the study, adequate glucagon inhibition was achieved only with the dual inhibitory actions of insulin and somatostatin to lower cAMP levels. Such insulin receptor signaling pathway, namely PDE3B degradation of cAMP through PI3K activation, was observed and characterized extensively in the antilipolytic action of insulin on mice, rat and human adipocytes (199, 277, 301, 324, 424). Furthermore, the involvement of Akt phosphorylation downstream of PI3K signaling in inhibiting glucagon secretion was substantiated by reduced insulin-mediated Akt phosphorylation in siRNA knockdown of insulin receptors in αTC-6 cells (87). Insulin-driven activation of the PI3K/Akt signaling pathway was also suggested to suppress glucagon secretion in α-cells during hyperglycemia by translocating A-type GABA (GABA_A) receptors to the plasma membrane to amplify their responsiveness to activation by GABA, another secretory product of β-cells co-secreted with insulin, leading to membrane hyperpolarization and eventual inhibition of glucagon release (426). In addition to inhibiting glucagon secretion, insulin was demonstrated to decrease glucagon gene expression in α-cells in a PI3K-dependent manner (189, 290, 291). It was also suggested that insulin plays a role in triggering a signal for glucagon release in α-cells at hypoglycemic levels by a “switch-off” process where pre-exposure of α-cells to intra-islet insulin released by β-cells at high glucose levels and its discontinuation at low glucose conditions thereafter modulates α-cell glucagon response to hypoglycemia (166, 444).

Another candidate β-cell secretory paracrine factor is the metal ion Zn²⁺ which co-crystalizes and co-secretes with insulin during hyperglycemia-induced granule exocytosis (67, 92, 174, 311). Several studies have shown the inhibitory effect of Zn²⁺ on glucagon secretion in intact mouse islets and flow-sorted mouse α-cells (215), isolated rat α-cells (113, 275), perfused rat pancreas (175), intact rat islets (113), and αTC-6 cells (146). One of the earliest works that highlight the inhibitory effects of Zn²⁺ on glucagon secretion in α-cells was performed by Ishihara et al. in perfused rat pancreas where micro-molar concentrations (30 μM) of ZnCl₂ inhibited pyruvate-stimulated glucagon secretion while calcium ethylenediaminetetraacetic acid (Ca²⁺ EDTA), a divalent metal ion chelator, removed the suppressive effect of monomethyl-succinate on glucagon release (175). The same group further extended their study to delineate the possible mechanism of action of Zn²⁺ on glucagon inhibition in α-cells (113). They demonstrated that Zn²⁺ inhibited glucagon secretion in intact rat islets at low glucose levels and glucose- and pyruvate-stimulated isolated rat α-cells by directly activating α-cell K_ATP channels and consequently inhibiting α-cell electrical activity. The involvement of K_ATP channel activation as a mechanism for glucagon suppression in α-cells by Zn²⁺ was further substantiated by the group that postulated the “switch-off” hypothesis for insulin control of glucagon release (362, 445). Instead of insulin, like their previous studies suggested, Zn²⁺ was implicated as the paracrine factor that triggers the “switch-off” signal for glucagon secretion at low glucose concentrations upon pre-exposure of α-cells to intra-islet Zn²⁺, co-secreted with insulin by β-cells, at high glucose levels and the subsequent cessation of Zn²⁺ and glucose. The authors utilized streptozotocin (STZ)-treated diabetic wild-type mouse islets to remove the effects of endogenous insulin and Zn²⁺ on the glucagon response of α-cells upon addition of exogenous Zn²⁺. They elucidated the “switch-off” mechanism of Zn²⁺ in modulating glucagon response as the direct opening of α-cell K_ATP channels and consequent closure of L-type Ca²⁺ channels at high glucose levels to suppress glucagon release and the discontinuation of Zn²⁺ release by β-cells under glucose-deprived conditions to trigger a signal to close K_ATP channels and activate L-type Ca²⁺ channels to stimulate glucagon secretion. These experiments used a series of drug treatments and sulfonylurea receptor 1 knockout (SUR1KO) mouse islets to influence K_ATP channel activation/inactivation (diazoxide/tolbutamide) and functionality (SUR1KO), and L-type Ca²⁺ channel inhibition (nifedipine). Although the aforementioned evidence point towards K_ATP channel activation and subsequent inhibition of α-cell electrical activity at high glucose concentration as the mechanism of Zn²⁺ paracrine inhibition of glucagon secretion in α-cells, contradictory results have also been reported by other groups. The Wheeler group determined that the suppression of glucagon secretion by Zn²⁺ was independent of α-cell K_ATP channel activity in dispersed mouse α-cells and αTC-6 cells (146). Rather, they propose that the transportation and accumulation mechanisms of Zn²⁺ via Ca²⁺ channels, plasma membrane Zn²⁺ transporters, and cellular redox state in α-cells could play a role in the suppressive action of Zn²⁺ against glucagon release. Supporting results from Leung et al. also revealed that Zn²⁺ failed to activate K_ATP channels in the presence of 1 mM intracellular ATP in dispersed mouse α-cells (219). The role of Ca²⁺ channels in Zn²⁺-mediated glucagon inhibition was validated in a study using a Zn²⁺ chelating agent and voltage-gated R-type Ca²⁺ channel (Ca_V2.3)-deficient mice where functional Ca_V2.3 channels were required for proper Zn²⁺ paracrine inhibition of glucagon secretion, possibly by the blockage of Ca_V2.3 channels by Zn²⁺ (96). Despite the large body of studies that highlight the paracrine inhibitory role of Zn²⁺ in α-cell glucagon secretion, several studies dispute the inhibitory action of Zn²⁺ on α-cell glucagon secretion. Ravier and Rutter demonstrated that 30 μM of ZnCl₂ did not influence glucagon secretion in both isolated mice islets and clonal αTC1–9 cells (308). In another controversial study, Zn²⁺ was shown to stimulate glucagon secretion in human islets (305). Therefore to date, the role of Zn²⁺ as a paracrine inhibitor of glucagon secretion remains disputed.

γ-aminobutyric acid (GABA) is another secretory product of β-cells that is shown to be a potential paracrine signaling molecule for the inhibition of glucagon secretion in pancreatic α-cells at high glucose levels. It is a well-known major inhibitory neurotransmitter that blocks neuronal signals in the central nervous system by activating GABA_A and/or GABA_B receptors (66). In the pancreatic β-cell, GABA molecules are packaged into synaptic-like microvesicles (SLMV), distinct from insulin secretory vesicles (309, 366, 385, 386), where they undergo insulin vesicle-like glucose-stimulated exocytosis with a high recycling rate of their readily releasable pool upon kiss-and-run exocytosis at the plasma membrane (233). In addition to SLMV, subsequent studies reveal that GABA molecules are also found in a subpopulation of insulin-containing vesicles in β-cells (48, 49). However, a recent study conducted on human and rat pancreatic islets disputed the vesicular localization of GABA in β-cells (251). The authors demonstrated that around 99% of the human β-cells displayed cytosolic expression of GABA while less than 1% had vesicular localization of GABA within insulin granules. Cytosolic GABA in human β-cells were shown to undergo pulsatile secretion that is facilitated by the volume regulatory anion channel, Swell1. Secreted GABA molecules then localize in the intra-islet environment and likely target and activate GABA_A receptors in α-cells to inhibit glucagon secretion. GABA_A receptors have been detected in rat α-cells (419), guinea pig α-cells (326), human pancreatic islet α-cells (382), mouse pancreatic α-cells and αTC1–9 cells (15). The inhibitory action of GABA on glucagon secretion has been illustrated by a considerable number of studies which include the suppression of arginine-stimulated glucagon secretion in isolated guinea pig (326) and mouse islets (127) by exogenous GABA, glucagon inhibition in αTC-6 cells (low glucose) (117), isolated rat α-cells (low and high glucose) (275), intact mouse islets and αTC1–9 cells (low glucose) (15, 215) with exogenously-added GABA, and the loss of glucagon inhibition upon GABA_A receptor antagonism in isolated rat islets (419), αTC-6 cells, intact mouse islets, and αTC1–9 cells (15, 419). The mechanism of GABA paracrine inhibition of glucagon release was shown in pancreatic α2-cells and isolated rat α-cells to involve the activation of GABA_A receptor Cl⁻ channels, leading to inward Cl⁻ currents which hyperpolarize the α2-cell, lower electrical activity and thereby decrease glucagon secretion (326). This was supported by antagonizing endogenous GABA_A receptors in isolated rat α-cells using specific GABA_A receptor antagonist SR95531 which blocked inward Cl⁻ currents elicited by 1 mM GABA and increased glucagon secretion. Furthermore, as mentioned earlier, paracrine inhibition of glucagon secretion by GABA in clonal In-R1-G9 and αTC-6 cells was enhanced by insulin-driven PI3K/Akt activation at high glucose levels which translocate GABA_A receptors to the plasma membrane and augment their responsiveness to GABA activation, leading to membrane hyperpolarization (426). Although GABA was shown to regulate glucagon secretion in α-cells, one study showed that GABA had only a modest inhibitory effect on glucagon secretion (~20%) and that GABA antagonist bicuculline failed to remove GABA inhibitory effect on glucagon secretion in isolated mouse islets, likely due to the nonspecific inhibition of glucagon secretion by bicuculline, highlighting its potential side effects (127).

The metabolite of GABA, γ-hydroxybutyrate (GHB), has also been identified as a β-cell paracrine factor that is shown to regulate α-cell glucagon secretion. GHB is produced in β-cells through a GABA shunt pathway which converts GABA to succinate semialdehyde (SSA) via GABA transaminase and the subsequent conversion of SSA to GHB via SSA reductase during glucose stimulation (220). Intra-islet GHB then acts on its GHB receptors on α-cells to inhibit glucagon secretion. This was validated by a variety of experiments including the loss of glucose inhibition of glucagon secretion in intact human islets upon blocking the GABA transaminase with vigabatrin, the inhibition of amino acid-stimulated glucagon release upon treatment with GHB receptor agonist 3-chloropropanoic acid (3-CPA), and the reversal of glucose-inhibited amino acid-stimulated glucagon release with GHB receptor antagonist NCS-382 (220). However, the Tengholm group reported a contradictory outcome in mouse islets where GHB failed to inhibit glucagon secretion at low and high glucose levels, did not affect [Ca²⁺]_pm and [cAMP]_pm dynamics, and GHB agonist, antagonist and blocker had little influence on glucagon secretion (434). In fact, they demonstrated that GHB had the tendency to stimulate, rather than inhibit, glucagon secretion at low glucose levels in human islets.

Intra-islet glutamate, a major neurotransmitter in the central nervous system, has been demonstrated to play both inhibitory and stimulatory roles in autocrine/paracrine regulation of glucagon secretion in α-cells. It is produced in both β- and α-cells and co-secreted in SLMVs (similar to GABA) and glucagon-containing secretory vesicles respectively (153, 182). As a positive feedback autocrine/paracrine effector, glutamate acts to stimulate glucagon secretion in perfused isolated rat pancreas in low glucose conditions and isolated mouse islets by activating α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) or kainate receptors, which possibly led to membrane depolarization, influx of Ca²⁺ through activated voltage-gated Ca²⁺ channels, and increased glucagon exocytosis in α-cells (36, 58, 68). On the contrary, glutamate was reported to mediate the inhibition of glucagon secretion in rat islets by activating class III metabotropic glutamate receptors, mGlu4 and/or mGlu8, which triggers the reduction of intracellular cAMP levels (389, 392). With the opposing roles of glutamate on the regulation of glucagon secretion, it is thus crucial to delineate and discriminate the extent of glutamate activation of both ionotropic and metabotropic glutamate receptors in α-cells at different glucose conditions.

Serotonin, another β-cell secretory product (312) that is co-secreted with insulin (7, 106, 273), has been identified as a possible paracrine regulator of α-cell glucagon exocytosis. Although the role of serotonin as a β-cell autocrine signaling molecule for the stimulation of insulin secretion and maintenance of GSIS has been well established (21, 197, 284), less is known about its paracrine signaling properties on α-cells. It was shown that serotonin inhibits glucagon secretion by activating membrane-bound G-protein coupled 5-HT_1F receptors on α-cells and signaling via G_i/o proteins to inhibit adenylyl cyclase, thus lowering intracellular cAMP levels (3, 7). The presence of 5-HT_1F receptors was identified in α-cells of intact human islets and pancreatic sections by quantitative real-time PCR, immunohistochemistry and in situ hybridization. A number of studies validated the inhibitory role of serotonin in glucagon secretion in α-cells. The effect of serotonin on glucagon levels were studied by orally administering serotonin antagonists or serotonin synthesis blocker in human subjects and detecting plasma glucagon levels (240). The serotonin antagonists, cryproheptadine and methysergide, and serotonin synthesis blocker, para-chlorophenylalanine (PCPA), enhanced arginine-induced and hypoglycemia-induced (elicited by administering insulin) glucagon secretion. In a follow-up study, the authors tested the effect of exogenously-added serotonin and its precursor 5-hydroxytryptophan on glucagon release in intact mouse islets (241). Serotonin inhibited both basal (3 mM glucose) and arginine-stimulated glucagon secretion while 5-hydroxytryptophan only managed to inhibit arginine-stimulated glucagon secretion. Serotonin inhibition of glucagon secretion was also observed in healthy human islets at both low and high glucose concentrations (20). The Adeghate group demonstrated that serotonin reduced glucagon secretion in healthy rat pancreatic tissue fragments (2). Almaça and colleagues established in human islets that exogenous serotonin significantly inhibits glucagon secretion at both low and high glucose levels, inhibition of endogenous serotonin levels with reserpine or PCPA leads to increased glucagon secretion at high glucose level and amplified glucagon secretion upon glucose reduction (11 mM to 1 mM), and increasing endogenous serotonin levels with fluvoxamine inhibited glucose reduction-stimulated glucagon secretion (7). They also showed that 5-HT_1F receptor agonist LY344864 decreased plasma glucagon levels and exacerbated insulin-induced hypoglycemia in C57BL/6 mice, as well as promoted the recovery of glucose homeostasis in STZ-induced diabetic mice to normoglycemia.

Amylin secreted from β-cells (269) has also been shown to be a potent inhibitor of glucagon secretion, with even greater suppression of glucagon than seen with glucose. However, this inhibition of α-cell function appears to be extrinsic to the islet, as it is seen in animals and patients, but not in isolated islets or perfused pancreata (432).

The δ-cell secretory product, somatostatin, has been well established as a paracrine inhibitor of glucagon secretion in pancreatic α-cells of rodent, dog, baboon and human (63, 107, 123, 151, 187, 203, 215, 334, 379). It is required for the regulation of glucose-inhibition of glucagon secretion in α-cells (427). Paracrine signaling of somatostatin in α-cells occurs through the activation of α-cell somatostatin receptors (SSTR) by intra-islet somatostatin which triggers various downstream signaling mechanisms to inhibit glucagon secretion. The major SSTR expressed in mice, rat and human α-cells is type 2 somatostatin receptor (SSTR2) (171, 187, 211, 231). Using SSTR2 knockout mice islets and a selective SSTR2 agonist, the Strowski group demonstrated that somatostatin inhibition of glucagon secretion is primarily modulated by SSTR2 in α-cells (379). They further investigated the therapeutic properties of the selective SSTR2 agonist on type 2 diabetes treatment in diabetic mice and showed that the selective agonist reduced glucagon and glucose levels (377). The Cejvan group further supported the role of SSTR2 as the main SSTR involved in somatostatin inhibition of glucagon release in isolated rat islets using a SSTR2 selective antagonist (63). Kailey et al. also established SSTR2 as the functionally dominant SSTR in human α-cells (187). Early studies suggest that somatostatin inhibits glucagon secretion by modifying cAMP production pathway in perfused rat pancreas (123, 347). They also reported that glucagon secretion is significantly more sensitive to somatostatin inhibition than insulin secretion. In another study, Yoshimoto and colleagues proposed a mechanism where somatostatin activates SSTRs on mice islet α-cells to generate intracellular GTP, further activating G-protein-gated K⁺ channels, leading to hyperpolarization of the α-cell and inhibition of glucagon exocytosis (431). Gromada et al. substantiated the finding by demonstrating that the G-protein subtype G_i2 associate SSTRs to low-conductance K⁺ channels and the binding of somatostatin to SSTRs triggers the activation of the low-conductance K⁺ channels, causing hyperpolarization and reduced α-cell electrical activity (140). The same group revealed another mechanism of somatostatin inhibition of glucagon secretion in which localized G_i2-dependent activation of calcineurin at regions where SSTRs associate with L-type Ca²⁺ channels, the main voltage-gated Ca²⁺ channel that drives glucagon release at hypoglycemic levels, result in the depriming of secretory granules near the L-type Ca²⁺ channels in rat α-cells (139). Our group has also investigated the mechanism of action of somatostatin on glucagon inhibition in intact murine islets and determined, using SSTR2 specific antagonist and pertussis toxin treatment, that somatostatin regulates glucose-inhibition of glucagon secretion by the activation of the G_αi subunit of SSTR2 which inhibits adenylyl cyclase and reduce cAMP production (107). With the collective actions of the library of paracrine factors affecting glucagon secretion, it is imperative to delineate their synergistic effects along with the intrinsic mechanisms of GRGS in α-cells to have a better understanding of the overall islet physiology.

While the activation of α-cells appears to be dominated by their paracrine inputs, these cells also function as a source of paracrine signals that have been shown to be important for proper islet function. Glucagon has been shown to signal to both β-cells (380) and δ-cells (9). Glucagon signaling to the β-cell has been shown to couple through both glucagon and GLP-1 receptors (380). It is interesting to consider that signaling through GLP-1 receptors to stimulate β-cells by α-cell activity may preface an important adaptive mechanism. This signaling pathway could be present to allow efficient regulation of β-cell function under circumstances when the α-cell can alternatively splice pro-glucagon to produce GLP-1 (224). In this case, the α-cell secreted GLP-1 acts as a powerful enhancer of β-cell function. In the δ-cell, on the other hand, glucagon is expected to signal only through the highly expressed glucagon receptor (222). The second paracrine factor that the α-cell has been shown to secrete is acetylcholine. Acetylcholine from the α-cell signals primarily to the β-cell (321), but has also been shown to affect the δ-cell (260). Finally, the neurotransmitter glutamate is known to function as an autocrine feedback regulator of glucagon secretion from the α-cell (58), and while it has not been shown to play an important role in communication with other cell types, it could also signal to the δ-cells, as they are known to express ionotropic glutamate receptors (266).

Similar to the communication between β-cells, Ephs/ephrins and NCAMs have been reported to be involved in the communication between β-cells and α-cells. Tonic stimulation of EphA4 and EphA7 on α-cells by ephrin-A5 has been shown to be important in maintaining proper glucose inhibition of glucagon secretion through the modulation of F-actin density (173). As inhibition of F-actin with cytochalasin D restored the secretion of glucagon at low glucose conditions, F-actin is also likely involved in the NCAM regulatory pathway. More recently, tight junctions have also been observed specifically between β-cells and α-cells (170). However, very little is known about their role in α-cell regulation. Despite studies suggesting the increasing importance of these regulatory pathways arising from the clustered architecture of the islet, most of their function still remains an enigma.

δ-cells

Among the top three major islet cell types, pancreatic δ-cells are the least studied by islet biologists and have only recently garnered attention in the field. Pancreatic δ-cells comprise < 10% of the cells in pancreatic islets across species (57, 376). δ-cells are sparsely distributed within the islet of Langerhans, with most of them located at the islet cortex in mouse islets while localized at both the islet periphery and center in non-human primate and human islets (50, 57). They adopt a neuron-like morphology with long cytoplasmic neurite-like processes that can extend beyond 20 μm in length (327). Such cytoplasmic processes were shown to be in contact with multiple α- and β-cells via electron microscopy, reinforcing their extensive paracrine influence on neighboring islet cells despite being small in number. A recent study demonstrated that δ-cells also possess filopodia-like structures which contain secretory machinery to enable efficient paracrine regulation of β-cell activity (9). The δ-cell plays a critical role by controlling the actions of α- and β-cells, thus regulating islet cell function and maintaining proper blood glucose homeostasis. It should be noted, though, that δ-cell is not the major regulator of islet cell function since β-cell only pseudo-islets still mimic proper GSIS as seen in intact islets (310, 410). Somatostatin secreted by δ-cells function as a major inhibitory intra-islet paracrine factor to inhibit both insulin and glucagon secretion in pancreatic β- and α-cells, respectively. The paracrine inhibitory actions of somatostatin on insulin and glucagon release have been extensively discussed in the previous sections. In δ-cells, pre-prosomatostatin undergoes a cleavage step to form prosomatostatin which is post-translationally processed at the carboxy-terminal to produce somatostatin. The dominant isoform of somatostatin released from pancreatic δ-cells is somatostatin 14 while that of the gastrointestinal tract is somatostatin 28 (327). Elevated somatostatin signaling in diabetic animals has been implicated as the cause of reduced counter-regulatory glucagon secretion during insulin-induced hypoglycemia (190, 437, 438). This was supported by other findings where somatostatin secretion was found to be higher in diabetic animals than control animals at low glucose concentrations (1, 417). Moreover, the effects of oversecretion of somatostatin during insulin-induced hypoglycemia in diabetic rats were rectified by SSTR2 antagonist treatments (190, 437, 438). Contrarily, exogenous somatostatin or somatostatin analogs were tested and advocated as a supplement to insulin therapy, due to its glucagon suppressive properties, for more effective glycemic control in patients with diabetes mellitus (84, 121, 124, 306). These findings indicate that the precise control of somatostatin release from δ-cells and its subsequent intra-islet signaling in α- and β-cells within the islet architecture is critical to maintain proper glycemic control.

Similar to β-cell insulin secretion, somatostatin secretion in δ-cell is also stimulated by glucose, which increases with rising glucose levels (409). Contrary to insulin secretion which requires a higher glucose concentration (>7 mM) for initiation, somatostatin secretion is triggered at glucose levels as low as 3 mM and reaches a half-maximal at 5 – 6 mM glucose concentration in mouse islets (327). The glucose dependence curve of somatostatin secretion is slightly shifted to the right for human islets as compared to mouse islets. Glucose stimulation of somatostatin secretion in δ-cells mirrors that of GSIS in β-cells to a large extent (Fig. 9). Much like β-cells, electrically active δ-cells express the same type of K_ATP channels which remain open under low glucose conditions, thus sustaining a hyperpolarized membrane potential that prevents somatostatin secretion. Upon glucose metabolism in δ-cells, the intracellular ATP to ADP ratio increases, leading to closure of the K_ATP channels and subsequent membrane depolarization that triggers the firing of APs, influx of Ca²⁺, and the eventual secretion of somatostatin (104, 151, 162, 333). The resultant firing of APs caused by such membrane depolarization is attributed to the opening of T-type Ca²⁺ channels, which further depolarizes the cell and causes the activation of voltage-gated Na⁺ channels (4, 52). Consequently, the existing action potentials trigger the activation of voltage-gated L-type and P/Q-type Ca²⁺ channels which result in the influx of Ca²⁺ and thereafter somatostatin secretion. Ca²⁺ influx through these voltage-gated Ca²⁺ channels translates into intracellular [Ca²⁺] oscillations that have been demonstrated in dispersed mouse δ-cells and intact mouse islets at glucose levels of 3 mM and onwards (37, 267). Such glucose-induced intracellular [Ca²⁺] oscillations in the δ-cells have also been associated with the mobilization of intracellular Ca²⁺ stores caused by the activation of Ca²⁺-induced Ca²⁺ release through ryanodine receptors in the sarcoplasmic/endoplasmic reticulum, which is likely enhanced by increasing intracellular cAMP levels (85, 212, 441). A recent study on human and mouse islets revealed that Ca²⁺-induced Ca²⁺ release from the ER of δ-cells is regulated by the two-pore domain K⁺ channel TALK-1, which conducts K⁺ countercurrents across the ER membrane to facilitate ER Ca²⁺ leak, leading to reduced Ca²⁺-induced Ca²⁺ release and somatostatin secretion (405).

Figure 9:

Intracellular and intercellular pathway mechanisms that regulate somatostatin secretion in the δ-cell.

Little is known about the juxtacrine or electrical synapse coupling of δ-cells within the islet. Similar to β-cells, the δ-cells express both ephrins and EphA receptors. The δ-cell appears to highly express a wide range of ephrin ligand including ephrin-A1, ephrin-A5, ephrin-B3, and to a lesser level, ephrin-A3 (222). The broad affinities of these ligands for EphA receptors suggest that they could affect signaling in all of the other cell types. In terms of receptors, only the EphA6 receptor is reported to be highly expressed in the δ-cells (222). This would be consistent with a unique series of different EphA receptors expressed in the various islet cell types: EphA5 in β-cells, EphA4 in α-cells, and EphA6 in δ-cells. In particular, EphA6 is known to have a critical difference from these other islet cell EphA receptors in that it uniquely contains a valine at the hydrophobic pocket next to its ATP-binding site. This sequence change is expected to result in substrate specificity differences (360), which might be important in differentiating δ-cell juxtacrine interactions from those of the other islet cell types. As for gap junctions, no connexin genes are reported to be highly expressed in δ-cells, although Cx40.1 appears to be expressed in both β- and δ-cells (222), which may underlie the direct communication between these two cell types (53).

δ-cell communication

Besides the intrinsic mechanism of glucose-induced somatostatin secretion in δ-cells, intra-islet paracrine factors secreted from neighboring islet α- and β-cells are also involved in the modulation of somatostatin secretion. The few studies on the paracrine regulation of somatostatin by insulin show conflicting results. Hauge-Evans and colleagues did not observe significant effects of exogenous insulin or insulin receptor antagonism on basal and glucose-stimulated somatostatin secretion in intact mouse islets (150). On the other hand, Gerber et al. showed that exogenous insulin managed to inhibit glucose- and arginine-stimulated somatostatin secretion in isolated perfused rat pancreas (119). Lastly, exogenous insulin was shown to stimulate somatostatin secretion in isolated perfused chicken pancreas at low glucose concentration (165). These contradictory findings could stem from the different types of conditions and samples used in each study. Another β-cell secretory product, urocortin 3, which is co-secreted with insulin, was demonstrated to stimulate somatostatin secretion in intact mouse islets by activating the corticotropin-releasing factor receptor 2 (CRHR2), which increases cAMP production (398). GABA, a secretory product of β-cell, was also suggested to regulate somatostatin secretion in a similar manner as α-cell by activating GABA_A receptors which depolarizes the cell and stimulate somatostatin exocytosis (48). The α-cell paracrine factor, glucagon, was also reported to stimulate somatostatin secretion by binding to δ-cell glucagon receptors which likely activate adenylyl cyclase and raise cAMP production (283, 418). The glucagon receptor is highly expressed by the δ-cell (222), so increased glucagon secretion would be expected to stimulate somatostatin secretion. However, this does not fit with the physiology, where we know that somatostatin secretion is inhibited when α-cell glucagon secretion is expected to be maximal in the presence of low glucose concentrations (< 3 mM). This is yet one more example where the complex relationships between multiple communication pathways in the islet remain poorly understood. It is intriguing to speculate that the co-localization of α- and δ-cells might amplify the effects of glucagon on the δ-cell, and at least in part, play a role in stimulating somatostatin secretion at a lower glucose threshold than is seen for insulin secretion (327). In addition to paracrine interactions, optogenetic evidence indicates the presence of electrical coupling between β-cells and δ-cells via gap junctions which stimulates somatostatin secretion in δ-cells upon optogenetic activation of β-cells (53).

As discussed in the previous sections on the α- and β-cells, somatostatin is a known regulator of insulin and glucagon secretion from pancreatic islets, which gives it an important role in glucose homeostasis (229). Somatostatin acts through a family of inhibitory G-protein coupled receptors (GPCR), which consists of five subtypes, all of which can inhibit adenylyl cyclase and hence decrease cAMP levels (255). In the islet, these receptors modulate insulin and glucagon secretion, and in both α- and β-cells, activation of the somatostatin receptor inhibits exocytosis. Somatostatin signaling in the β-cells goes primarily through SSTR5 receptors, and in the α-cells mainly through SSTR2, as confirmed by the respective genetic deletions of SSTR2 (379) and SSTR5 (378). An extensive overview of δ-cell islet physiology was recently presented by Rorsman and Huising (327).

PP-cells

Pancreatic polypeptide (PP) is a peptide hormone (39), which is one of the least studied of the known islet secretory products. PP is secreted by PP cells in the islet, which have received little attention from researchers. These cells are found predominantly in islets from the head of the pancreas, and can be up to 5% of the cells in some islets (50, 227). The main function of PP as a hormone is thought to be the regulation of pancreatic secretion activities, both of the endocrine and exocrine cells, but PP also appears to play a role in regulating hepatic glycogen levels and gastrointestinal secretions (19, 343, 361). PP secretion is elicited during the cephalic phase (400), and well as during digestion (206, 342). In humans, hypoglycemia is known to stimulate PP secretion while hyperglycemia inhibits it, similar to what is seen with glucagon from α-cells. However, unlike in α-cells, these effects appear to be largely mediated by vagal efferent activity (401). Taken together, these data suggest that signals from the gut may play a central role in regulating PP secretion, and this is an area of current investigation. A few studies have shed some light on the components (222) and functional behaviors of PP-cells (367, 402). Still, very little is known about any potential role of pancreatic polypeptide or other PP-cell mechanisms in cellular communication underlying islet function.

Conclusion

The pancreatic islet is the functional unit that provides the hormones insulin and glucagon for whole body regulation of blood glucose levels. The most important evidence for the primacy of the islet in glucose homeostasis is the capability of islet transplants to correct diabetes in human patients (353). This capability is consistent with our established understanding that hormone secretion from the intact islet is different than it is from dispersed cells (218). As we have shown in this review, the islet is a highly complex micro-organ despite its modest list of components. This complexity arises from multiple communication pathways that instigate and modulate cellular functions of all islet cells, and new data continue to elucidate the details and interconnected networks of these pathways. The compact size of the islet (~100 μm diameter) and reasonable number of cells (~1,000) lends itself to detailed studies of cellular communication. In fact, concepts developed from islet data (25) can be used to inform cellular interactions in more complex tissues such as the brain (112).

This review highlights our understanding of cell-cell communication in the islet, which has largely been developed over the last 50 years. Even after such an extended period of successful research, recent insights continue to shape our understanding of the islet structure and dynamics. One striking example is the role of primary cilia in controlling islet function, which has only lately been investigated (114), but it is already becoming clear that these surface appendages play a key role in proper islet function (169). Of the various cell types in the islet, the insulin-secreting β-cells are the largest population and the best understood (325). However even for these cells, the role of cellular communication, and especially its role in overcoming β-cell heterogeneity, remains controversial (335, 341). Over the last ten years, there has been a significant upsurge in data and insights into the other islet cell types. The second most populous α-cells secrete the counter-regulatory hormone, glucagon, in a manner that is inverse to insulin secretion. While new data and concepts of α-cell functions proliferate (170, 323, 420), a consensus model of how these cells work within the islet milieu remains elusive. Given the clinical importance of hypoglycemic unawareness in T1D patients and the role that glucagon is thought to play in this pathology, improved understanding of how islet communication regulates α-cell function is a top research priority. The less populous islet cell types are also thought to play important roles. Despite recent insights into the behaviors of the δ-cells (327) and PP cells (387, 402), a complete picture of how these cells interact within the islet is still to be determined. As progress continues to be made in building functional islet cells from stem cells, it will be imperative to understand the important aspects of islet dynamics to optimize the development of stem-cell derived replacement islets for therapy (403).

Didactic Synopsis.

Major teaching points:

• Secretion of the hormones insulin and glucagon from the islet of Langerhans play a central role in blood glucose homeostasis, other molecules secreted by islet cells appear to function as local paracrine signals within the islet.

• The islet comprises several different cell types, which are all highly coupled within and between the cell types by gap junctions, paracrine signaling networks, and juxtacrine communication. This coupling is crucial for proper islet function.

• The islet cells are a heterogeneous population on their own, but the communication within the islet acts to create a more uniform syncytium that allows robust and reproducible responses to physiological challenges, such as hyperglycemia.

• The mechanisms of communication vary between the islet cell types. For example, the insulin secreting β-cells are highly coupled by gap junctions to each other and more weakly coupled to the somatostatin secreting δ-cells, but not at all to the glucagon secreting α-cells.