Role of Endoplasmic Reticulum Stress and c-Jun NH₂-Terminal Kinase Pathways in Inflammation and Origin of Obesity and Diabetes

Inflammation is the body’s response to deal with a broad range of injuries; however, this response does not assure a satisfactory outcome, particularly in the long term. This certainly appears to be the case in glucose homeostasis in obesity and type 2 diabetes. In contrast to the classic paradigm, the inflammatory responses in obesity are low grade and occur over an exceedingly extended time frame, thus increasing the risk of harm to multiple systems including, but not limited to, those involved in glucose homeostasis. In the past decade, several key lines of evidence emerged to establish firmly that inflammation lies at the heart of obesity and type 2 diabetes. First, a molecular link between inflammation and obesity was identified by the discovery of tumor necrosis factor (TNF)-α as an inflammatory cytokine overexpressed in the adipose tissues from rodent models of obesity, and this link disrupts insulin action in cells and whole animals (1,2). TNF-α is also overproduced in adipose and other tissues, such as muscle in obese humans (3–5). Administration of recombinant TNF-α to cultured cells or to whole animals impairs insulin action, whereas biochemical, pharmacological, or genetic blockade of TNF-α action results in improved insulin sensitivity (3). Thus, particularly in rodents, it is clear that adipose tissue overproduction of TNF-α is an important feature of obesity and contributes significantly to insulin resistance. Subsequently, it became clear that, in addition to TNF-α, several other inflammatory mediators are also dysregulated in obesity and impair insulin action both in experimental models and in humans. A comprehensive list of these with compelling evidence linking them to metabolic homeostasis can be found in another recent review (6). Second, clinical and epidemiological studies have linked inflammatory mediators to the development of insulin resistance and diabetes (7,8). For example, high levels of TNF-α, interleukin-6, C-reactive protein, plasma activator inhibitor 1, and many other inflammatory mediators were observed in obese individuals, those with insulin resistance and type 2 diabetes, and their first-degree relatives (4,6,9–12). Third, it became clear that the presence of chronic inflammatory diseases, such as rheumatoid arthritis or hepatitis, significantly increases the risk of development of insulin resistance and/or type 2 diabetes, thus providing an additional and important link between chronic inflammatory states and impaired insulin action in humans (13–15).

Recently, it was also proposed that infiltration of macrophages into adipose tissue, a phenomenon reminiscent of classic inflammation, might contribute to the inflammatory responses of obesity (16,17). Because this macrophage infiltration occurs rather late in the disease process, whether or to what extent it also contributes to insulin resistance or other metabolic abnormalities in obesity is unclear. It is worth noting, however, that there is a strikingly high degree of overlapping features between adipocytes and macrophages, which illustrates another interesting aspect of the integration and coregulation of metabolic and immune responses. This paradigm is best exemplified by the impact of fatty acid binding proteins and peroxisome proliferator–activated receptor (PPAR)-γ–LXR (liver X receptor) pathways on insulin action, type 2 diabetes, and atherosclerosis through their related actions on metabolic and inflammatory responses in adipocyte and macrophages, respectively (18–20). Finally, in addition to the inflammatory cytokines regulating insulin action and metabolic homeostasis, hormones that are highly specifically produced by adipocytes, such as adiponectin, resistin, and visfatin, also act at the interface of metabolic and immune systems (21–25). Taken together, there is now a general consensus regarding the inflammatory origin of obesity and diabetes; however, the exact role of each target cell that might be critical in the different pathological outcomes and the underlying mechanistic details, especially regarding the initiating signals, are still not completely understood.

In addition, because many inflammatory mediators are simultaneously regulated in obesity and type 2 diabetes, it is unlikely that a single inflammatory mediator alone could explain the whole of the disease pathogenesis. Hence, an important consideration in designing therapeutics is the continuing need to understand more central mechanistic events that give rise to obesity-related inflammatory responses. In the recent years, there have been groundbreaking discoveries to uncover these core mechanistic pathways.

In the overwhelming majority of the individuals suffering from obesity-induced insulin resistance or type 2 diabetes, the principal defect in insulin action lies distal to the insulin receptor. Because of space limitations, insulin receptor signaling pathways will not be covered in detail here and the readers are referred to excellent recent reviews on insulin action and receptor signaling (26–28). Interestingly, inhibition of downstream components of the insulin receptor signaling pathways is also a primary mechanism by which inflammatory cascades are mechanistically linked to insulin resistance. Numerous studies have described that inhibitory phosphorylation of insulin receptor substrate (IRS)-1 on serine residues is a key event in generation of impaired insulin action upon exposure of cells to TNF-α or elevated free fatty acids or in the context of obesity and systemic insulin resistance in whole animals and humans (29–33). Serine phosphorylation of IRS-1 reduces both tyrosine phosphorylation of IRS-1 in response to insulin and the ability of IRS-1 to associate with insulin resistance and thereby suppresses downstream signaling and insulin action (30,34,35). The search for enzymatic mediators of this key modification has resulted in uncovering critical insights into mechanistic bases of insulin resistance. These studies have yielded several serine/threonine kinases that are activated by inflammatory or stressful stimuli and contribute to inhibition of insulin signaling (6). These include c-Jun NH₂-terminal kinase (JNK), inhibitor of nuclear factor-κB kinase (IKK), and protein kinase C-θ. Again, the activation of these “inflammatory” kinases in obesity and their involvement in insulin action illustrates the close relationship between metabolic and immune pathways. In particular, IKK and JNK control major inflammatory response pathways, and they are activated by a variety of stress signals and mediate the innate immune responses by Toll-like receptor signaling upon exposure to pathogens and their products (36). These signaling pathways can also affect T-cell function and adaptive immunity. Hence, the detailed mechanistic studies have also reinforced the close integration of immune and metabolic response pathways and their critical importance in the pathogenesis of type 2 diabetes.

The three members of the JNK group of serine/threonine kinases, JNK-1, -2, and -3, belong to the mitogen-activated protein kinase family and regulate multiple activities in development and cell function, in large part through their ability to regulate transcription by phosphorylating activator protein-1 (AP-1) complex proteins, including c-Jun and JunB (37). JNK-1 and JNK-2 are expressed in a wide range of tissues, whereas JNK-3 is restricted to brain, pancreatic islets, heart, and testis. Recent studies have demonstrated that JNK is a central player in the modulation of insulin action and a critical component of the pathogenesis of obesity, fatty liver disease, and type 2 diabetes (32). In particular, JNK1 appears to be the predominant isoform present in muscle, liver, and adipose tissues, and its activity is strikingly elevated in obesity at these sites (32). JNK can also directly phosphorylate IRS-1 at several sites, including serine 307 (31). There is also increased IRS-1 serine phosphorylation in obesity, which depends on the elevated JNK activity, and this modification leads to reduced insulin-stimulated IRS-1 tyrosine phosphorylation. In obese animals, deletion of the JNK-1 gene results in reversal of the obesity-induced increase in JNK activity and IRS-1 serine phosphorylation, whereas JNK-2 deficiency alone does not alter these significantly. There is also dramatic protection against insulin resistance and defective insulin receptor signaling in JNK-1–deficient obese animals (32). Again, none of these changes could be observed in the JNK-2–deficient animals, suggesting that this isoform alone is not sufficient to mediate the metabolic effects and JNK-1 is the principal isoform mediating obesity-induced JNK activity and consequently insulin resistance. However, it is possible that in the setting of JNK1 haploinsufficiency, which has little or no impact on systemic insulin action, JNK2 deficiency could influence insulin sensitivity, since JNK-2 can modulate JNK-1 action (38). JNK-interacting proteins (JIPs) are scaffolding molecules for JNK isoforms and can play an important role in determining kinase activity of all JNK isoforms. For example, a JIP-1 loss of function model creates a phenotype very similar to JNK-1 deficiency with reduced JNK activity and increased insulin sensitivity in mice (39). Interestingly, a point mutation in the human homolog of JIP-1, which leads to increased JNK activity, has been identified in a rare Mendelian form of diabetes in humans, providing crucial genetic proof-of-principle supporting the role of JNK in the pathogenesis of type 2 diabetes in humans (40).

In addition to the genetic loss of function models, several recent studies confirmed the critical impact of JNK in insulin resistance even after the establishment of disease in rodents. For example, adenovirus-mediated modulation of JNK activity with the expression of a dominant-negative JNK isoform in the livers of obese animals results in increased systemic insulin sensitivity and improved glucose homeostasis (41). Conversely, exogenous expression of JNK in the adult liver results in severe insulin resistance in mice (41). In these animals where JNK activity is blocked, there is also reduced IRS-1 serine phosphorylation and increased insulin receptor signaling, in perfect agreement with the genetic results obtained in JNK-deficient models. Finally, administration of small molecule or peptide inhibitors of JNK activity has resulted in significant increases in systemic insulin sensitivity in obese insulin-resistant animals (42,43). Inhibition of JNK is also effective against short-term lipid infusion-induced insulin resistance in rats (C. Gorgun, J. Kim, G.S.H., unpublished data). The importance of these results is twofold. In addition to illustrating the strong therapeutic potential of JNK inhibition against type 2 diabetes, these studies also demonstrate the fact that the impact of JNK activity on systemic insulin action is independent of its effects on body weight, which is also regulated in JNK-deficient genetic models where JNK activity is blocked throughout the life of the animal (32).

Finally, it is likely that JNK activity modulates islet function and/or survival in numerous ways. First, there is convincing evidence for the involvement of JNK in islet cell inflammation and death mediated by cytokines (44–46). Second, JNK activation might generate a state of β-cell dysfunction and defective insulin production, thereby contributing to the development of frank diabetes (47). Third, administration of SP600125, a synthetic inhibitor of JNK, results in improved glucose-stimulated insulin production in isolated islets in the db/db model of obesity and diabetes (42). Hence, there is a strong possibility that JNK might integrate defects in insulin secretion with peripheral insulin resistance in type 2 diabetes through its actions in pancreatic β-cells as well as peripheral sites of insulin action. If this is the case, it is also likely that JNK might be important in the pathogenesis of type 1 diabetes, and recent studies have provided evidence to support a role for the JNK-2 isoform in this disease (48). Taken together, there is very strong evidence that abnormal JNK activation is a critical event in the deterioration of glucose homeostasis and offers a very promising target for therapeutic strategies.

Two other inflammatory kinases that play important roles in counteracting insulin action are IKK and, particularly in response to lipid metabolites, protein kinase C. Because of space limitations, these will not be discussed in detail, but readers are referred to recent excellent reviews in these areas (49,50). It is highly likely that there are interactions between the signaling and action of these key mediators, which might involve regulation of the activity of each other. For example, it has already been suggested that protein kinase C-θ can act through IKKβ to phosphorylate IRS-1 (49). Lipid exposure can modulate the activity of all three enzymes (51). Both JNK and IKKβ are required for TNF-induced insulin resistance in cells, and suppression of either can increase insulin sensitivity (52). All three enzymes can increase IRS-1 serine phosphorylation and alter inflammatory gene expression profiles (51,53). At high doses of salicylate exposure, both JNK and IKKβ activity could be suppressed (52). The action of both JNK and IKKβ appears to play an important role in their impact on systemic insulin action at similar target sites (32,51,54–56). Hence, it will be extremely interesting to examine the potential integration or synergy between these pathways, particularly with JNK and IKK, regarding the relevant target sites of insulin action, underlying mechanisms, and systemic metabolic outcomes.

In addition to these serine/threonine kinase cascades, several other pathways also contribute to inflammation-induced insulin resistance, including activation of inducible nitric oxide synthase, production of reactive oxygen species in adipose tissue, regulation of suppressor of cytokine signaling (SOCS) proteins, and alterations in AMP-K and mTOR pathways in obesity (57–61). It is also highly likely that many of these mediators share common regulatory cues and signaling pathways; however, their integration mechanisms are yet to be fully uncovered. Regrettably, these highly promising areas and additional mechanisms involving these molecules are not discussed in detail here because of space limitations.

One of the major areas that has been quite difficult to address has been the mechanism(s) underlying the origin of the stress and inflammatory responses associated with obesity. How does a cell interpret the presence of metabolic overload and start transmitting stress signals such as activation of JNK and IKK pathways to trigger a prolonged inflammatory response? In our recent studies, we have produced some insights into this issue and demonstrated that the endoplasmic reticulum (ER) might be a key site where the metabolic signals are sensed, integrated, and transmitted in the form of stress signals, including activation of JNK, IKK, and possibly other pathways (62). Accordingly, this model offers a new paradigm, where inflammatory pathways are activated by metabolic stresses originating from inside of the cell and propagated by extracellular signaling molecules or inflammatory mediators as described above. In particular, obesity-induced activation of the JNK pathway appears to be controlled, at least in part, by this mechanism. As such, ER can offer mechanistic models to integrate the molecular pathways shown to take part in the development of insulin resistance and type 2 diabetes.

The ER is a critical organelle where all secreted and transmembrane proteins are synthesized, folded into their correct three-dimensional structures, modified, and transported to their final cellular destinations. Under conditions in which the client traffic is increased or mutant proteins arrive at the ER, the organelle experiences stress and activates an adaptive mechanism called “unfolded protein response” to reestablish equilibrium (63). The principal events during unfolded protein response include transcriptional activation of chaperones to aid protein folding, selective attenuation of new protein synthesis (which allows ER to cope with the existing proteins), and activation of proteosome-mediated protein degradation. In addition to increased client load, changes in nutrient, energy, and Ca²⁺ availability and exposure to pathogens and pathogen components and hypoxic signals can also induce ER stress (63–65). Thus, it is possible to envision ER as one of the earliest sites evolved as a nutrient- and pathogen-sensing device in the cells. In our very recent studies, we have postulated that obesity is associated with many conditions that can lead to ER stress (62). These include an increase in protein and lipid synthesis, lipid accumulation, perturbations in intracellular nutrient and energy fluxes, and severe changes in tissue architecture. The latter is particularly the case for expanding adipocytes and adipose tissue. In testing this postulate, we observed elevation of several biochemical indicators of ER stress in liver and adipose tissues of obese animals compared with their lean counterparts (62). These observations held true for both dietary (high-fat diet–induced) and genetic (leptin-deficient, ob/ob mice) models of obesity and included phosphorylation of PERK, activation of JNK, and, depending on the stage of the disease, increased chaperone expression (62). Subsequently, these observations were independently confirmed by Nakatani et al. (66), who observed increased (immunoglobulin heavy chain–binding protein) Bip expression in the liver tissues of a genetic model of obesity (leptin receptor–deficient db/db mice).

We next tested whether experimental manipulation of ER stress responses could lead to modification of insulin receptor signaling and insulin action. In cultured cells, induction of ER stress by tunicamycin or thapsigargin exposure or by deletion of XBP-1 (x-box–binding protein), a critical transcription factor regulating unfolded protein response, led to generation of ER stress, increased IRS-1 serine phosphorylation, and suppressed insulin-induced insulin receptor signal transduction (62). This ER stress–induced cellular insulin resistance was mediated by inositol-requiring enzyme-1 (IRE-1)–dependent activation of JNK and reversed by blocking the activity of either kinase (62). Conversely, reduction of ER stress by overexpression of a spliced and transcriptionally active form of XBP-1 rendered cells refractive to ER stress–induced insulin resistance. Existence of a similar causal relationship between ER stress and insulin action was also demonstrated in whole animals. Mice with XBP-1 haploinsufficiency in the Balb/C genetic background, which is normally resistant to insulin resistance and diabetes, were predisposed to high-fat diet–induced ER stress, insulin resistance, and hyperglycemia compared with wild-type controls. In liver and adipose tissues of these animals, an increase in ER stress markers including elevation in JNK activity and IRS-1 serine phosphorylation was evident. Consequently, XBP^+/− animals exhibited significantly impaired insulin receptor signaling in adipose and liver tissues (62). In the absence of a high-fat diet or obesity, there was no indication of alterations in systemic insulin sensitivity or insulin receptor signaling capacity in liver or adipose tissues of lean XBP-1^+/− animals (62). Taken together, these studies demonstrated that ER stress responses are closely linked to insulin receptor signaling and insulin action in both cultured cells and whole animals, and modification of these responses have a significant impact on systemic insulin sensitivity.

ER stress has been previously implicated in islet biology and function (63–65). The islet morphology of XBP-1^+/− animals was similar to wild-type controls, both exhibiting comparable levels of hypertrophy and insulin content (62). There was also no apparent defect in glucose-stimulated insulin secretion in XBP-1^+/− mice, even after exposure to a high-fat diet. However, because XBP^+/− mice developed peripheral insulin resistance and hyperglycemia, it could be inferred that their islet responses were insufficient to compensate for the degree of insulin resistance present; therefore, a role for XBP-1 in islet biology cannot be ruled out at this time. Subsequently, two independent groups also demonstrated that adenovirus-mediated or genetic depletion of an ER chaperone leads to ER stress insulin resistance and diabetes, whereas in vivo overexpression renders animals refractory to insulin resistance and diabetes (66,67). In all of these models, increased ER stress was associated with impaired insulin receptor signaling at target tissues. Taken together, these results demonstrated that ER stress is a central feature of obesity and a key mechanism of insulin resistance and diabetes.

These findings stimulate consideration of new models in understanding the pathogenesis of type 2 diabetes and may have important therapeutic implications (Fig. 1). First, it is quite possible that ER is the site where metabolic stress is translated into inflammatory responses, particularly the activation of JNK and other inflammatory mediators. More experiments are needed to test these possibilities. For example, it is not yet clear whether modification of ER stress modulates local or systemic inflammatory cytokine levels in obesity. Another important question is whether ER stress–induced insulin resistance and type 2 diabetes could be rescued by blocking JNK activity in vivo, as is the case in vitro. This will firmly establish the ER stress–JNK activation axis as a key mechanism in this model. Interestingly, ER stress also stimulates the activity of the nuclear factor κB pathway, which might represent a further link between JNK- and IKK-regulated pathways, the two important signaling cascades leading to insulin resistance. Therefore, it will be highly informative to test the role of the IKK–NF-κB pathway and inflammation in general in ER stress–induced insulin resistance. Also, studies are critical that are aimed at understanding the components of obesity that lead to ER stress activation and the metabolic responses and the related pathologies that are regulated by ER stress. Of particular interest are dyslipidemia and atherosclerosis, since cholesterol loading of macrophages has been shown to modify ER membrane properties and lead to an ER stress–mediated apoptotic response (68). The potential contribution of other lipid signals to triggering ER stress in obesity might open up new possibilities in understanding and preventing lipotoxicity in a variety of metabolically active tissues. Finally, there is need for development of physiologically relevant models of ER stress and highly sensitive reagents to monitor the biochemical indicators of ER stress and unfolded protein response and to study detailed mechanisms of metabolic regulation through this pathway during the course of obesity, insulin resistance, and type 2 diabetes.

Studies in the past decade clearly established that inflammation is an integral component of obesity and associated diseases. At the core of this mechanism lies the high degree of integration between nutrient- and pathogen-sensing pathways, the coordination of which is disrupted in obesity. Central to this obesity-induced inflammation is the activation of JNK and IKK and induction of ER stress. Along these lines, better decoding of the link between these response systems and modulation of inflammatory activity or ER stress pathways may offer multiple novel therapeutic and preventive strategies for metabolic diseases, particularly type 2 diabetes.

Because of space limitations, I regret the unintended omission of many important references from our colleagues. Studies in the Hotamisligil lab were supported by the National Institutes of Health, American Diabetes Association, Sandler Foundation, and Eli Lilly Pharmaceutical Company.

I would like to thank the members of the lab and our collaborators for their valuable contributions. The author holds equity in Syndexa Pharmaceutical Corporation, which develops therapeutics against diabetes.