The 70 kDa heat-shock protein (HSP70) family is important for a dynamic range of cellular processes that include protection against cell stress, modulation of cell signalling, gene expression, protein synthesis, protein folding and inflammation. Within this family, the inducible 72 kDa and the cognate 73 kDa forms are found at the highest level. HSP70 has dual functions depending on location. For example, intracellular HSP70 (iHSP70) is anti-inflammatory whereas extracellular HSP70 (eHSP70) has a pro-inflammatory function, resulting in local and systemic inflammation. We have recently identified a divergence in the levels of eHSP70 and iHSP70 in subjects with diabetes compared with healthy subjects and also reported that eHSP70 was correlated with insulin resistance and pancreatic β-cell dysfunction/death. In the present review, we describe possible mechanisms by which HSP70 participates in cell function/dysfunction, including the activation of NADPH oxidase isoforms leading to oxidative stress, focusing on the possible role of HSPs and signalling in pancreatic islet α- and β-cell physiological function in health and Type 2 diabetes mellitus.
More than 10% of the world's population is now overweight or obese . The long-term impact of obesity, which is commonly associated with insulin resistance, glucose intolerance and dyslipidemia, can lead to the development of pancreatic β-cell failure and overt Type 2 diabetes mellitus (T2DM). Significant advances in our understanding of the molecular pathways leading to insulin resistance and T2DM have occurred in the last two decades. However, this has not resulted in the development of a new generation of novel and effective treatment options. It is now widely accepted that obesity is linked to a state of chronic inflammation , especially adipose-tissue-associated inflammation. The secretion of pro-inflammatory cytokines such as tumour necrosis factor-α (TNFα) and interleukin-1β (IL-1β) from resident and activated macrophages and their release from adipose tissue leads to activation of specific signal transducing pathways in insulin-sensitive tissues and organs, including c-Jun N-terminal kinases (JNKs) and inhibitor of κB kinases (IKKs) in adipose tissue, skeletal muscle and liver [3,4]. In contrast with the specific phosphorylation targets following physiological insulin receptor occupation, JNK and IKK phosphorylate the key signal transduction protein insulin receptor substrate 1 (IRS-1) on Ser307, reducing its ability to act as a substrate for the activated insulin receptor . In addition, a pro-inflammatory cytokine stimulated increase in the level of the key reactive oxygen species (ROS)-producing enzyme, NADPH oxidase (NOX), in endothelial cells, skeletal muscle and pancreatic islets (including β-cells), is critical for the development of insulin resistance  and β-cell dysfunction associated with T2DM .
During the evolutionary development of mammals, rapid triggering of inflammatory responses was conserved in order to protect the whole organism against pathogens and to initiate the repair of tissue injuries. The processes controlling the resolution of inflammation are also conserved and stimulate local and/or systemic elevation of temperature. As a consequence, the extraordinarily conserved heat-shock response begins with the activation of a transcriptional program based on the activation of heat-shock transcription factor 1 (HSF1) . HSF1 activation exacerbates production of the anti-inflammatory and cytoprotective heat-shock proteins (HSPs) whose chief representative is the 70 kDa HSP family (HSP70). Small heat-shock proteins induced by fever, such as HSP27, also contribute to cytoprotection [8,9]. Although pro-inflammatory cytokines are associated with impaired insulin signalling, in contrast HSPs can act as anti-inflammatory agents within the pro-inflammatory milieu thus protecting cells against elevation in temperature, oxidative stress and side effects of exacerbated inflammation. Achieving appropriate levels of intracellular HSP70 (iHSP70) results in stress tolerance and protection against otherwise lethal insults . The induction of iHSP70 expression therefore may lead to cellular protection against metabolic or inflammatory damage. Indeed, regardless of the mechanism used to induce elevations in iHSP70, protection against diet- or obesity-induced hyperglycaemia, hyperinsulinemia, glucose intolerance and insulin resistance has been reported [3,11–17].
As briefly introduced above, the induction of HSP gene expression is regulated by the interaction between the HSFs (mainly HSF1) and the regulatory heat-shock elements (HSEs, specific sequences of DNA located in the promoter region of the HSP gene) . At rest HSF1 is inactive in a monomeric state bonded with the cytosolic HSP70s, located in the cytosol. Under stress conditions (i.e. any shift from cellular homoeostasis), particularly in the presence of denatured proteins, HSP70 releases HSF1 and subsequently binds to denatured proteins, acting as chaperones (aiding protein refolding) and releasing HSF1. Serine phosphorylation and trimerization of HSF1 leads to enhanced HSF1 DNA-binding affinity with respect to the cis-acting regulation HSEs in target genes . Interestingly, activated HSF1 trimers initiate the transcription not only of iHSP70, but also of genes closely related to inflammation, such as the inducible cyclo-oxygenase-2 (COX-2) . What might appear paradoxical (because COX-2 generates pro-inflammatory prostaglandins, e.g. PGE2 and PGD2) is not. Indeed, these prostaglandins are precursors of cyclopentenone prostaglandins (e.g. PGA2 and 15-deoxy-Δ12,14-PGJ2), which are strongly anti-inflammatory, leading to activation of HSF1 and attenuating the transcription of nuclear factor κB (NF-κB)-dependent pro-inflammatory genes . Therefore stress-activated HSF1 stimulates a loop of positive feedback that provides a robust anti-inflammatory response.
The connection between inflammation and the heat-shock protein pathways highlights a much more intricate network at gene regulatory level. For instance, the promoter region of the TNFα gene contains an HSF1-binding site that represses TNFα transcription. The loss of this repressor results in sustained expression of TNFα . This finding explains why HSF1 knockout is associated with a chronic increase in TNFα levels and increased susceptibility to endotoxin challenge [22,23]. Counter-regulation of such interplay has also been reported, indicating that TNFα may transiently repress HSF1 activation . Moreover, JNK1 was demonstrated to phosphorylate HSF1 in its regulatory domain, causing suppression of HSF1-transcribing activity, whereas HSP70 prevents Bax activation by inhibiting the JNK/Bim pathway [23,25]. Any sub-lethal exposure to a specific stressor can rapidly induce the appearance of HSP70 (mainly HSP72) in the cytoplasm  (see Figure 1).
The activation of the heat-shock response after non-lethal stress
Another key player involved in the stress response and regulation of HSP70 synthesis is the NAD+-dependent deacetylase sirtuin-1 (SIRT1). Multiple studies have described the role of SIRT1 in the activation of HSF1, and thus molecular chaperones including heat-shock protein 70 that regulate stability and function of cellular proteins. It has been shown that activation of SIRT1 prolongs HSF1 binding to the promoter regions of heat-shock genes by maintaining HSF1 in a deacetylated and DNA-binding competent state (see Figure 1) , so enhancing the transcription of molecular chaperones such as HSP70 and HSP25 [27,28]. The importance of SIRT1 for the chaperone machinery is clearly demonstrated by SIRT1 knockdown, which attenuates the heat-shock response . Conversely, it has recently and convincingly been demonstrated that whole-body heat-shock treatment in HFD (high fat diet)-fed rats reverses insulin resistance-induced vascular defects and increases SIRT1 expression/activity .
Little is known of the role of iHSP70 in maintaining pancreatic islet function, despite the above considerations, but we speculate, that iHSP70 acts to protect pancreatic β- and α-cells during metabolic and inflammatory challenge. Regarding eHSP70 challenge, recent data has demonstrated that the addition of eHSP70 (HSP72, in this case) to β-cells in vitro resulted in a reduction in β-cell viability and functional integrity . In the present review we have described the essential roles of HSPs in modulating pancreatic β-cell function based on current knowledge, and have presented testable speculation as to potential roles in pancreatic α-cells.
β-CELL FAILURE IN T2DM AND THE IMPORTANCE OF iHSP70
Pancreatic β-cells are vulnerable to metabolic and oxidative stress, eventually becoming unresponsive to glucose and fatty acid stimulation and subsequently dying via apoptosis. The mechanisms involved include an increase in generation of ROS and reactive nitrogen species (RNS) , activation of oxidative stress responses, activation of endoplasmic reticulum (ER) stress and a reduction in energy generation capacity . On the other hand, HSPs are associated with protection against oxidative stress and ER stress , although insulin secretion may be impaired once a threshold level of stress has been passed.
When compared with other cell types, the striking vulnerability of pancreatic β-cells towards oxidative damage (resulting from chronic inflammation, and/or the chronic increase in macro-nutrient availability including glucose and lipids), is due to lower gene expression and activity of catalase and glutathione peroxidase (GPx)  as well as other antioxidant enzymes. This may result in a lower resistance to exogenous peroxide , and under inflammatory conditions lower resistance to NOX activation [6,36]. However, β-cells may (under appropriate conditions) alter their redox state in response to increased glucose, decreasing the ratio of GSSG to GSH (the GSSG/GSH ratio) either by decreasing the GSSG or by increasing GSH content, indicating that glucose metabolism improves the ability to maintain glutathione in the reduced state under oxidative stress conditions probably via NADPH production . However, during metabolic overload, glucose and other metabolites may negatively affect the cell function by the induction of glucolipotoxicity . Elevated levels of glucose are capable of generating excessive levels of ROS in β-cells and this process is essential to the hypothesis that glucose-induced oxidative stress is a key player in glucose toxicity . Potential pathways that may contribute to elevated ROS include stimulation of the NOX complex, oxidative phosphorylation, glycosylation and the glucosamine pathways [6,39,40]. Key evidence that this sequence of events occurs in β-cells was provided by Ihara et al. . As oxidative stress causes protein oxidation and misfolding as well as the activation of the inflammatory pathways, the activation of iHSP70 expression (especially HSP72) is essential to protect the cell against glucotoxicity.
In addition, glucose toxicity triggered by chronic exposure to high levels of glucose may trigger β-cell apoptosis, via release of cytochrome c from the mitochondria, and activation of specific caspases resulting in DNA fragmentation, in line with known events relating to apoptosis by the mitochondrial pathway . Pancreatic β-cells may also be damaged by islet amyloid polypeptide (IAPP), also known as amylin . This peptide is normally co-released with insulin. Metabolic overload (e.g. excess amount of carbohydrates and fat) may lead to an elevated amount of amylin being secreted , resulting in local formation of various types of oligomers, leading to aggregation, and possible age-related accumulation, so damaging and reducing the regenerative potential of the β-cell and additionally elevating levels of apoptosis . IAPP aggregates decrease cell viability, activating ER stress, suppressing proteasome activity and increasing levels of HSP90 .
Endogenous chaperones could play important protective roles by decreasing amyloid deposition, as reported for the ER-resident chaperone glucose-regulated protein 78 (GRP78), iHSP70 and HSP40, in a concentration-dependent fashion. It is possible that amyloidogenicity can be inhibited by HSP expression . In addition IAPP can interact with immune cells, mainly resident macrophages, activating the nod-like receptor protein 3 (NLRP3) inflammasome, so promoting the synthesis of pro-inflammatory cytokines such as IL-1β, and inducing islet inflammation and subsequent glucose intolerance .
Intracellular HSPs are essential for β-cells, to protect them against damage. During islet transplantation, for example, iHSP70 may be critical for β-cell protection . Elevated expression of iHSP70 in islets protects the β-cell by reducing cytokine-induced cell death . In heat-shocked rat islets, the glucose-stimulated insulin secretion (GSIS) level was unaffected by the presence of IL-1β, whereas in control islets (without previous heat shock), insulin release was decreased, suggesting that oxidative stress is partially reduced by previous heat stress . iHSP70 expression is important for the protective effect, since heat shock induces resistance in pancreatic islet cells against the diabetogenic agent streptozotocin, oxygen-free radicals and nitric oxide (NO•) . Additionally, in the rat insulinoma cell line RINm5F, heat-shock treatment and/or transfection with the human HSP72 gene (HSPA1A, resulting in elevated iHSP70) resulted in reduced cell lysis induced by either NO• or oxygen-free radicals . Interestingly, human β-cell lines have enhanced resistance against NO• and ROS-induced necrosis, due to increased capacity for the expression of iHSP70. When iHSP70 expression was suppressed in these cells, the protection from NO• was abolished, suggesting that iHSP70 can attenuate NO• injury . In support of the protective role of HSPs, when human and rat islets were heat shocked, IL-1β-dependent iNOS expression, IκB degradation and NF-κB nuclear translocation were all reduced .
Several other reports have demonstrated the importance of iHSP70 for maintenance of pancreatic islet function during the inflammatory insult that is associated with diabetes. The first important observation was demonstrated by Welsh et al.  when they described that human islets are more resistant to chemical or cytokine-induced injury than rodent islets, possibly due to a greater capacity to express iHSP70. Interestingly, in response to IL-1β exposure, islets initiate a HSP70-mediated counter-regulatory anti-inflammatory response, to protect against the cytotoxicity of this cytokine [52,53]. The protective effect of HSP70 against IL-1β cytotoxicity was further supported when liposomal delivery of purified HSP70 into rat pancreatic islets resulted in protection against IL-1β-induced β-cell dysfunction . In addition, in islets derived from shb (Src homology-2 domain-containing protein B) knockout mice, increased HSP70 expression attenuated cytokine-induced cell death .
During the early development of T2DM, when pancreatic β-cells counter insulin resistance (by producing and secreting large amounts of insulin), ER stress may occur, causing β-cell dysfunction. Transgenic mice overproducing an alternative HSP from the 70 kDa family, GRP78 (a chaperone with numerous roles in the ER) in pancreatic β-cells, are protected against high-fat fed glucose intolerance (as the islets maintained their normal structure and function) . Indeed, in humans, the expression of GRP78 protein can be used as a marker for ER stress in β-cells. The response is influenced by polymorphisms of the GRP78 gene. Patients with the −415AA/−180GG genotype have lower levels of fasting plasma glucose and haemoglobin A1c (HbA1C) than the patients with −415GG/−180deldel and −415AG/−180Gdel, suggesting that the first yield a higher activity and more GRP78 protein, inhibiting β-cell apoptosis and maintaining glucose homoeostasis . Furthermore, iHSP70 was also shown to protect against ER stress in pancreatic β-cells . Lipopolysaccharide (LPS)-induced ER stress (as a model for systemic inflammation) was attenuated by acute whole-body hyperthermia (42°C for 15 min), a treatment dependent on iHSP70 expression .
iHSP70 and other iHSPs are therefore essential for normal β-cell function, thus any inhibition/reduction in the cell capacity to express iHSP70 will result in vulnerability to stress, damage, dysfunction and ultimately cell death. A possible consequence of the latter is the up-regulation and overactivation of NOX1 .
THE ROLE OF NOX IN PANCREATIC ISLET DYSFUNCTION AND POSSIBLE MODULATION BY HSP70
It is now widely accepted that ROS contribute to β-cell dysfunction and damage caused by glucolipotoxicity and islet inflammation in T2DM. Recent studies have demonstrated that a key source of ROS (in addition to the mitochondrial electron transport chain) in both pancreatic β-cells and in insulin-responsive cells are various isoforms of NOX . NOX-associated ROS may alter parameters of signal transduction, insulin secretion, insulin action, cell proliferation or, in situations of inflammation and/or exposure to high levels of glucose/fatty acids, cell death. Thus NOX may be a useful target for intervention strategies based on minimizing the negative impact of glucolipotoxicity in diabetes .
The NOX enzyme complex is able to transfer electrons from NADPH to molecular oxygen to generate O2•−. NOX activation is more widely associated with efficient killing of pathogens by phagocytes, such as macrophages, monocytes, dendritic cells (DCs) and neutrophils, that utilize ROS generated by NOX2 within the phagosomal membrane . It is now evident that NOX isoforms (NOX1, NOX2, NOX4) are expressed in pancreatic β-cells [59,61], where their function is related to regulation of insulin secretion and cell integrity .
After the initial demonstration that pancreatic β-cells express NOX components , the effects of this enzyme activation/inhibition has been extensively studied and many therapies involving the manipulation of this protein have been suggested . As ROS are involved in intracellular signalling, the phagocyte-like NOX activation by glucose may play an important role for β-cell function. The glucose-dependent increase and oscillations in intracellular Ca2+ concentration associated with the mechanism of GSIS can stimulate mitochondrial generation of ROS, whereas Ca2+, via protein kinase C activation, may enhance NOX-dependent generation of ROS . Excessive levels of ROS may lead, of course, not only to direct damage to cells by oxidizing DNA, proteins and lipids, but indirectly damage cells by activating a variety of stress-sensitive intracellular signalling pathways such as NF-κB, p38 mitogen-activated protein kinase (MAPK), JNK/stress-activated protein kinase (SAPK), hexosamine and others . Activation of these pathways results in the increased expression of numerous gene products that may cause cellular damage and play a major role in the aetiology of the late complications of diabetes . Indeed, NOX2 in pancreatic β-cells on overactivation can be a negative modulator of the secretory response, reducing cAMP/protein kinase A (PKA) signalling secondary to ROS generation . However, physiological activation (low levels) of this enzyme seems to be essential for normal β-cell function and insulin secretion . Production of ROS for short periods is associated with an increase in GSIS, but excessive or sustained production of ROS is negatively correlated with the insulin secretory process . For example, fatty acid modulation of ROS production by pancreatic β-cells can occur by one or more of several possible mechanisms, such as the control of mitochondrial respiratory complexes and electron transport, induction of uncoupling proteins and NOX activation . Indeed, NOX is physiologically activated by several secretagogues such as glucose, L-arginine, fatty acids and KCl . The use of NOX pharmacological inhibitors attenuates insulin secretion, indicating that NOX activity is essential for GSIS in normal conditions. Thus NOX may play a key role in normal β-cell physiology, but under specific conditions, such as inflammation and chronic presence of high levels of nutrients, may also contribute to β-cell demise.
In pancreatic β-cells, chronic high glucose concentrations are associated with reduced insulin expression, elevated cell stress and apoptosis via oxidative damage delivered from NOX2 activation . However, metabolic overload in vivo is not necessarily associated with NOX2-derived oxidative stress and decreased function of pancreatic β-cells since it was demonstrated that pancreatic islets isolated from rats fed a high-fat diet for 13 weeks had increased pancreatic islet functionality, associated with high levels of glucose metabolism and GSIS, but also low levels of NOX2 expression and ROS production . Moreover, human islets isolated from T2DM patients demonstrated increased mRNA levels of the p22 subunit of NOX .
It is possible that a reduction in islet NOX2 expression and thus ROS production, in vivo, may constitute an adaptive response of the pancreatic β-cell to handle high levels of metabolic fuels. However, during longer periods of metabolic overload, this adaptive response may fail to operate and the system would then be subject to a positive feedback loop, enhancing NOX2 expression, oxidative stress, impaired insulin secretion and overt T2DM . It may be more than a coincidence that the chaperone machinery and the levels of iHSP70 also fail during metabolic overload and obesity [2,31]. Thus there may be a connection between the levels of iHSP70 and NOX activity/levels.
Indeed, as recently demonstrated, depletion of iHSP70 in vascular smooth muscle cells resulted in higher NOX4 expression and increased NOX activity . The importance of iHSP70 within these cells was demonstrated following heat shock (thermal treatment: daily immersion of cells in water at 41°C for 15 min) which significantly attenuated infiltration of inflammatory cells in the adventitia (tunica externa) and suppressed neointimal thickening in cuff-injured arteries, processes associated with the enhancement of HSP70 expression and suppression of oxidative stress . Therefore we propose that a fall in the iHSP70/eHSP70 ratio results in overactivity/expression of NOX in pancreatic islets, and their imbalance is associated with β-cell dysfunction and death in T2DM.
INTRACELLULAR VERSUS EXTRACELLULAR HSP70 IN DIABETES
One key feature of intracellular against extracellular HSP70 level (i.e. iHSP70/eHSP70 ratio) is the capacity to modulate NF-κB translocation . iHSP70 has direct anti-inflammatory and protective effects including promotion of protein synthesis, membrane integrity and mitochondrial function after injury, as well as inhibition of apoptosis . iHSP70 decreases NF-κB activation . Thus, apart from being a molecular chaperone which is employed to reduce the formation of protein aggregates and reverse protein denaturation, iHSP70 is able to associate with the complex formed by NF-κB with IκB, stabilizing this complex, thus impeding NF-κB translocation to the nucleus . Attenuation or partial attenuation of the powerful anti-inflammatory effect of iHSP70 is likely to be a key component in low-grade inflammatory diseases such as diabetes.
Diabetic subjects have a reduced capacity to induce iHSP70 and HSF-1 expression in cells and tissues  and this appears to precede the appearance of major metabolic defects associated with T2DM. We have recently verified the same in severely obese patients in which low levels/activity of proteins associated with the HSF1/HSP70 axis in adipose tissue are correlated with high JNK activity and with the development of non-alcoholic fat liver disease (NAFLD) and insulin resistance . Conversely, increments in iHSP70 levels tend to reduce insulin resistance in obese rats by reducing JNK and IKK activation .
Although iHSP70 has anti-inflammatory effects, when released to the extracellular environment (eHSP70), this protein exerts counter-regulatory effects, inducing inflammation and immune activation . Interestingly, in T2DM, there are divergences between the intracellular and extracellular HSP70 content compared with non-diabetic individuals, as the content of iHSP70 is decreased whereas the levels of eHSP70 are elevated .
eHSP70 may bind to cell receptors such as Toll-like receptor 2 (TLR2) and TLR4, activating innate immune responses which may lead to adaptive immune responses . TLR signalling may activate NF-κB and JNK by a pathway related to the IL-1-receptor-associated kinase (IRAK) family of protein kinases . eHSP70-dependent signalling from TLR is increased in obese and T2DM subjects, an effect that can explain the high basal rate of MAPK phosphorylation and NF-κB activation found in these patients [80–83]. The above findings help to explain why inhibition or absence of TLR4 confers protection against insulin resistance in skeletal muscle , adipose tissue and liver [85,86]. Interestingly, among the NOX isoforms, NOX4 was shown to be responsible for LPS-induced H2O2 generation in TLR4-overexpressing HEK (human embryonic kidney)-293T cells, and to interact with TLR4 to regulate NF-κB activation . Since eHSP70 may be a TLR ligand, increased levels of plasma eHSP70 may increase NOX activation during inflammation and T2DM. Although iHSP70 may attenuate NOX activity, the plasma levels of eHSP70 may activate this enzyme. Therefore, the iHSP70/eHSP70 ratio may be critical to the activity of this enzyme.
Recently, we demonstrated that high plasma levels of eHSP70 are correlated with insulin resistance in humans and that eHSP70 may lead, chronically, to pancreatic β-cell and islet dysfunction and death . Thus we now speculate on the role and function of HSP70 in pancreatic islets. Although iHSP70 induction has a protective role in β-cells, eHSP70 may provoke cellular dysfunction following β-cell damage dependent on the time of exposure, resulting in apoptosis and death after long-term exposure.
THE ROLE OF eHSP70 IN ISLET PHYSIOLOGY: AN INTEGRATIVE HYPOTHESIS
What is the consequence of elevated plasma eHSP70 in diabetes?
Elevation of the plasma levels of eHSP70 is likely to occur during pro-inflammatory conditions, such as obesity and diabetes . Adipose tissue expansion is a known modulator of cytokine profile and inflammatory status . Although the source of eHSP70 associated with obesity, insulin resistance and diabetes has not been fully determined, it is possible that lymphocytes are involved . The connection between adipose tissue expansion and eHSP70 release may be related to changes in leptin production (see Figure 1). This relationship has recently been demonstrated by Krause et al. , as they reported a positive correlation between eHSP70 and the leptin/adiponectin ratio in older people with excessive abdominal fat accumulation. The extensive storage of fat in the adipose tissue results in a large change in the levels of adipocytokines i.e. increased leptin and lower adiponectin release . The chronically increased level of leptin can affect many cells in the body, including those of the hypothalamus, inducing changes in the autonomic output, specifically increasing sympathetic nervous system (SNS) activity . Indeed, chronic intravenous leptin infusion in rats (at concentrations found in obesity), have been shown to increase arterial blood pressure and heart rate, apparently through increments in sympathetic activity . Thus, when the SNS is chronically activated in obesity , all cells containing adrenoreceptors, including lymphocytes and hepatocytes, will respond to the released catecholamines . Interestingly, hepatosplanchnic tissue seems to be the main source of eHSP70 during exercise  in a mechanism mediated by the activation of α-1-adrenoreceptors . In addition to hepatocytes, immune cells are also a source of eHSP70, e.g. circulating lymphocytes are considered the major eHSP70 source among mononuclear cells [95,96] and the release of this protein may mediate effects through the activation of adrenoreceptors [88,94]. Hence obesity may induce increments in the plasma eHSP70 level because adipose tissue expansion leads to chronic leptin release and activation of SNS (Figure 1). The elevation of the sympathetic activity results in the activation of hepatocytes and blood lymphocytes, inducing them to release eHSP70 into the bloodstream. Evidence exists that circulating lymphocytes are the major source of eHSP70 in the circulatory system. Indeed, lymphocytes release secretory vesicles containing eHSP70 (exosomes) that can be assembled and released after elevation of intracellular Ca2+ [97,98] or adrenaline/noradrenalin stimulation . However, due to variation in lymphocyte subtype, number and activation state, dependent on the severity of obesity and diabetes, the release kinetics can be different. Once delivered, eHSP70 may affect the function of many cells in the body, however, in the present review we will focus only on pancreatic islet cells (α- and β-cells).
eHSP70 and pancreatic islets: receptors, signalling and functional effects
eHSP70 is a known ligand for TLR2 and TLR4 in a variety of cells . The binding of eHSP70 to TLR2/4 can lead to the activation of pro-inflammatory pathways, via myeloid differentiation primary response gene 88 (My88) and Toll-IL-1 receptor protein (TIRAP) that signal downstream respectively, to the activation of NF-κB via IKK, and of JNK via MAPK/extracellular-signal-regulated kinase kinase kinase 4/7 (MEKK4/7; also known as MAPK kinase kinase 4/7) [100–102]. As previously demonstrated, activation of TLR2 and TLR4 is linked with increased cytokine gene expression in human islets , while inhibition results in improvement in insulin secretion from rodent islets . In addition, both TLR2- and TLR4-deficient mice are protected from the metabolic consequences of a high-fat diet [85,104–108]. Thus TLR2 and TLR4 activation can lead to islet dysfunction that occurs during the time course of chronic inflammatory disease. Specifically, TLR4 may bind ligands such as eHSP70, leading to activation of NOX, increasing ROS and causing cell dysfunction.
Recent studies have described the effects of eHSP70 on pancreatic β-cells, indeed our group was the first to demonstrate that the chronic exposure of clonal rodent and human β-cells to eHSP70 can induce metabolic dysfunction and loss of cell integrity . β-Cell dysfunction induced by eHSP70 could be a result of impaired signalling, gene or protein expression pathways and/or impaired mitochondrial function within the cell. For example, elevated proton leak (as measured directly by Seahorse Bioscience XFe96 Flux Analyzer) stimulated by the activation of uncoupling proteins (UCPs); the formation of ROS and RNS by activation of inflammatory pathways and, finally, by reducing the intracellular synthesis of iHSP70 (see Figure 1). In our model, a reduction in iHSP70 levels in pancreatic β-cells is induced by chronic inflammation which can lead to release and thus elevation in eHSP70 for chronic periods. In this case, the binding of eHSP70 to TLR2 and TLR4 would lead to: (i) activation of NF-κB and JNK downstream pathways followed by (ii) inhibition of Akt signalling that allows for (iii) the activation of glycogen synthase kinase-3β (GSK-3β), which is an inhibitor of HSF1 activation. Lower levels of HSF-1 will reduce the cell capacity for the synthesis of iHSP70. In fact, in type III obese individuals (who present systemic markers of chronic inflammation, insulin resistance and T2DM), the HSF1/iHSP70 axis has been found to be severely impaired in both fat and liver . The inability of the β-cell to express constitutive and/or inducible iHSP70 can lead to the known fragility of this cell against stress agents such as free radicals, including ROS and RNS, which can be elevated by glucolipotoxicity and inflammation. In addition, as β-cells adapt to higher demands for insulin (involving increased secretion) in response to chronic hyperglycaemia, the large increase in protein synthesis can lead to the unfolded protein response (UPR) and ER stress . Under normal conditions, when iHSP70 expression is normal, the chaperone supply for the appropriate folding of newly synthesized proteins is sufficient. However, due to the inflammation-induced inability to express iHSP70, the UPR is induced, causing β-cell dysfunction, perpetuation of inflammation and an increased risk of apoptosis .
GSK-3β activation is crucial for delivery of intracellular signals in the β-cell [109,110]. GSK-3β overexpression impairs β-cell proliferation and decreases expansion, whereas GSK3 inhibitors stimulate β-cell growth and protection against glucolipotoxicity, as well as inhibiting apoptosis by reducing DNA fragmentation and caspase activity . GSK-3β deficiency preserves β-cell mass in insulin-resistant mice, maintaining proliferation rates and reducing apoptosis , whereas the hyperactivity of this enzyme increases apoptosis . JNK inhibition preserves the pancreatic islet mass, by allowing for elevated Akt phosphorylation and promoting a reduction of GSK-3β activity , due to increased GSK-3β phosphorylation (Ser9) (see Figure 2), so protecting the β-cell against GSK-3β and JNK-mediated cell apoptosis . In other cells, the connection between TLR, Akt and GSK-3β activity with cell survival has been described at the molecular level. TLR-mediated apoptosisis is dependent on the activation of GSK-3β in kidney cells, where excessive apoptosis was attenuated by inhibition of the latter enzyme . Also, TLR4 activation inhibits enterocyte proliferation by decreasing Akt phosphorylation while increasing the expression and activity of GSK-3β .
The role of eHSP70 in islet physiology: an integrative hypothesis
Thus we hypothesise that, in pancreatic β-cells, the elevated levels of eHSP70 result in lower levels of the heat-shock response (including iHSP70 expression) via activating TLR2 and TLR4, NF-κB and JNK, leading to GSK-3β overactivity and inhibition of HSF1, culminating in lower levels of iHSP70 (see Figure 2). We also believe that, in parallel, a state of chronic low-grade inflammation may trigger NLRP3 inflammasome-based fat cell senescence that may also blunt the HSF1/iHSP70 axis .
The role of macrophages in the islet microenvironment and possible implications for α-cells
When islets are stressed (metabolic, inflammatory, ER and UPR) they can release several cytokines, including IL-1β, monocyte attractant protein 1 (MCP-1) and IL-8 . The last two are known inducers of monocyte infiltration and macrophage differentiation in the islet. The presence of activated macrophages induces a massive change in islet physiology, since these cells can respond to locally generated cytokines and to other molecules arriving from the bloodstream such as eHSP70. As described above, activation of TLRs on the membrane surface of macrophages can reduce β-cell insulin gene transcription and insulin secretion via release of IL-1 and IL-6, potentially contributing to β-cell dysfunction during T2DM. In addition to the cytokines and eHSP70, the accumulation of IAPP also induces the activation of the NLRP3 inflammasome  which, in turn, induces the release of more pro-inflammatory cytokines from macrophages such as IL-1β.
The role of these cytokines in β-cell dysfunction has been extensively reviewed [120–122], however, there is a lack of information regarding the effects of these cytokines in α-cell physiology. For example, it has recently been demonstrated that IL-6 promotes insulin secretion in pancreatic islets via a mechanism that is dependent on the release of glucagon-like peptide-1 (GLP-1) . Ellingsgaard et al.  have shown that administration of IL-6 (or elevated IL-6 concentrations in response to exercise) stimulates GLP-1 secretion from intestinal L-cells and pancreatic α-cells, so improving insulin secretion and glycaemia. Given that high-fat diet feeding increases systemic in vivo levels of IL-6, which is necessary for expansion of pancreatic α-cell mass and maintenance of fasting circulating glucagon levels, it is possible that high concentrations of IL-6 may directly regulate α- and β-cell function in the pancreatic islet [123,124] and additionally IL-6 can promote insulin secretion via GLP-1 release and subsequent β-cell activation in the presence of glucose. Thus it seems that IL-6 release by macrophages may be important for β-cell compensation driven by α-cell expansion and GLP-1 release. What is remarkable and curious regarding α-cells is that, in contrast with β-cells, they are still able to respond to different stimuli by secreting large amounts of its hormones (glucagon and GLP-1) even in an inflamed environment. Considering that α-cells also express TLRs, it is reasonable to think that the activation of TLR2/4 may lead (just as hypothesized for β-cells) to the activation of an inflammatory pathway that would culminate in overstimulation of GSK-3β, NLRP3 inflammasome and inhibition of the HSF-1/iHSP70 axis. However, potentially low chaperone capacity will limit α-cell expansion and protein synthesis would be compromised as these cells would undergo UPR and ER stress. It is possible that the α-cells are still able to produce a normal chaperone (iHSP70) response that allows these cells to maintain the large amount of protein synthesis required. Our hypothesis is that IL-6 released by islet macrophages may mediate this effect (Figure 2).
The mechanism(s) by which IL-6 exerts its diverse effects within target cells (such as skeletal muscle, hepatocytes, adipocytes, β- and perhaps α-cells) has been suggested to include activation of AMP-activated protein kinase (AMPK) . There is now accumulating evidence for a direct correlation between IL-6 concentration and AMPK activity in metabolically sensitive tissues [125,126]. Following that, changes in the ATP/ADP ratio act as a signal for the activation of different kinases, such as AMPK that is able to decrease GSK3β activity . Furthermore, IL-6 provokes an increase in suppressor of cytokine signalling-3 (SOCS3) expression, as reported in the liver and skeletal muscle cells, which resulted in significant changes in glucose metabolism [128,129]. Therefore, we suggest that the effects of IL-6 on α-cells may include: (i) activation of AMPK and UCP-2 (leading to low ATP levels and thus reducing glucagon and GLP-1 release); (ii) activation of SOCS, reducing the inflammatory response; and therefore (iii) increasing [via signal transducer and activator of transcription (STAT3)] the expression of iHSP70 and/or increasing AMPK activity, that will inhibit GSK-3β, then allowing HSF-1 to induce iHSP70 expression as reported for skeletal muscle cells. For example, during physical exercise, IL-6 can be expressed  and released  by the skeletal muscle and, within the extracellular space, can bind to the IL-6 receptor in an autocrine fashion . Interestingly, IL-6 has also been found to induce HSF-1 translocation to the nucleus, up-regulating the heat-shock HSP70 gene, protein expression and activity in human hepatic cells in a phosphoinositide 3-kinase (PI3K)-dependent manner (PI3K/Akt/GSK-3β) . On the other hand, the absence of IL6 is associated with decreased expression of iHSP70 in skeletal and cardiac muscle of mice challenged with LPS, although IL-6 seems not to be required for exercise-induced HSP production . In addition, recent work has shown that IL-6 acts on α-cells by binding to GR130 receptors and (via STAT3) thus stimulating glucagon synthesis . What is more, incubation of α-cells with IL-6 protects from cell death induced by a mixture of cytokines [IL-1β, TNFα and interferon γ (IFNγ)], consistent with previous observations under nutrient stress . Thus IL-6 may promote protection against inflammation in α-cells by the induction of iHSP70 (particularly, HSP72).
To sum up, chronically increased levels of eHSP70, as found in obesity and diabetes, may promote harmful effects in pancreatic β-cells, but not in α-cells due to the high capacity of the latter to induce iHSP70, thus protecting α-cells against the inflammatory insult. This characteristic is probably mediated by IL-6 and may explain the high capacity of secretion and expansion of α-cells even in an inflamed milieu. Finally, it is possible that the low levels/absence of NOX within α-cells may explain the protection associated with these cells when compared with β-cells with respect to their susceptibility to inflammation. Thus, even in the presence of high eHSP70 and pro-inflammatory cytokines, the α-cell does not generate as much ROS as β-cells. This, together with higher chaperone capacity in α-cells, would explain resistance to inflammatory and oxidative stress stimuli. However, the level of NOX (expression and activity) needs to be fully determined in these cells.
ARE THE EFFECTS OF eHSP70 ALWAYS NEGATIVE FOR β-CELLS?
It is known that chronic exposure of islet cells to eHSP70 may induce damage to pancreatic β-cells; however, acute exposure may have differential effects on these cells. For example, moderate physical exercise, which is known to induce health benefits (most of them attributed to the anti-inflammatory effect of exercise at low to moderate intensity) is also a known inducer of eHSP70 release . Since eHSP70, along with several cytokines including IL-6, are released during moderate exercise, it is both possible and reasonable to predict that the eHSPs may also induce benefits for heath, when released over short periods related to duration and intensity of exercise. The acute increase of eHSP70 is followed by a return to basal levels soon after termination of exercise . Therefore we suggest that there may be two different responses of β-cells to eHSP70, depending on the time of exposure and frequency. Accordingly, acute periods of short exposure may lead to the activation of Akt via exercise-related signalling  and the inhibition of GSK-3β (by Ser9 phosphorylation), allowing the normal production of iHSP70 that is essential for the function and viability of β-cells. On the other hand, prolonged exposure to eHSP70 may result in higher GSK-3β activity and reduced iHSP70 expression (Figure 3). In addition, chronic exposure of β-cells to eHSP70 (for example in a person who exercises regularly), may induce benefits by increasing chaperone activity and, perhaps, reducing the sensitivity of β-cells to TLR ligands, as demonstrated in other cells, a recognised anti-inflammatory effect of exercise [138–140].
The dual effect of eHSP70 on β-cells: time matters
CONCLUSIONS AND PERSPECTIVES
In the present review, we speculate that chronic elevation of eHSP70 is a key player in the development of islet inflammation and dysfunction associated with diabetes. Starting with adipose tissue expansion, hormonal and inflammatory responses provoke the release of several inflammatory cytokines which then initiate changes to cell and target tissue physiology. The increased SNS activity may be the major cause of the release of eHSP70 to the circulation, affecting cells expressing TLR2 and TLR4. In the pancreatic islets, β-cells will respond negatively due to low ability to express iHSP70 under inflammatory challenge in addition to the increase in level and activity of NOX, culminating in cell dysfunction, whereas α-cells are able to compensate and increase their secretory capacity due to the actions of IL-6, thus maintaining the chaperone machinery (iHSP70) and reducing the inflammatory damage in α-cells. The effects of IL-6 on the chaperone balance and the consequences for islet function is now under investigation in our laboratories.
AMP-activated protein kinase
regulatory heat-shock element
heat shock transcription factor
glucose-regulated protein 78
glucose-stimulated insulin secretion
islet amyloid polypeptide
inhibitor of κB kinase
c-Jun N-terminal kinase
nuclear factor κB
nod-like receptor protein
mitogen-activated protein kinase
reactive oxygen species
reactive nitrogen species
sympathetic nervous system
suppressor of cytokine signalling
signal transducer and activator of transcription
Type 2 diabetes mellitus
tumour necrosis factor-α
unfolded protein response
We thank the Curtin University School of Biomedical Sciences and Curtin Health Research Innovation Institute and the Federal University of Rio Grande do Sul, Department of Physiology, for supporting this work. We also thank the Brazilian National Council for Scientific and Technological Development (CNPq) for funding.
This work was supported by the School of Biomedical Sciences, Curtin University (Perth, Australia), and the Brazilian National Council for Scientific and Technological Development (CNPq) [grant numbers 402626/2012-5, 402364/2012-0, 402398/2013-2 and 372373/2013-5].