Diabetes and Suppressors of Cytokine Signaling Proteins

The SOCS proteins were identified in 1997 and were characterized as a family of proteins capable of inhibiting Janus kinase (JAK)–signal transducers and activators of transcription (STAT) (JAK-STAT) signaling in various tissues (5–7). Eight members of the SOCS family have been identified, SOCS-1–7 and cytokine-inducible SH2-containing protein (CIS) (8). They all contain a conserved COOH-terminal region of ∼40 amino acids termed the SOCS box (Fig. 1) (5). They have a central SH2 domain, while the NH₂-terminal region is of variable length with no recognizable motif (8). A kinase inhibitory region (KIR) consisting of 12 amino acids is found immediately NH₂-terminal to the SH2 domain in SOCS-1 and SOCS-3 (9,10).

In general, the constitutive level of SOCS protein expression in cells is low, but SOCS protein expression is highly inducible, often in a transient manner, upon stimulation with cytokines both in vitro and in vivo (Fig. 2A) (11). IL-1β, IFN-γ, and TNF-α can induce SOCS expression in the β-cell (12,13). Cytokine-induced SOCS expression is regulated via activation of STAT proteins. STAT-binding elements have been identified in the promoters of CIS (14), SOCS-1 (15), and SOCS-3 (16). Mutations of these elements reduce SOCS expression, and expression of dominant-negative variants of the STAT proteins blocks cytokine-induced SOCS expression (7,16,17).

SOCS-mediated downregulation of cytokine-induced JAK-STAT signaling involves different mechanisms (Fig. 2B). Via its SH2 domain, SOCS-1 binds directly to the JAK and inhibits kinase activity (10). SOCS-3 also inhibits JAK activity, but in contrast to SOCS-1, this requires binding between the SH2 domain of SOCS-3 and the phosphorylated receptor (9). CIS inhibits cytokine signaling by binding to phosphorylated tyrosine residues on the cytokine receptor, thereby masking potential docking sites for downstream signaling molecules such as the STAT proteins (18). Finally, SOCS proteins can inhibit signaling by coupling of signaling proteins to degradation via the proteasomal machinery (19).

In the context of diabetes, SOCS-1 and -3 are the most relevant SOCS members, and their effects are discussed below. SOCS-2 influences growth hormone effects, and gigantism is seen in mice lacking SOCS-2 (20). Overexpression and knock-out studies have revealed sparse information on CIS and SOCS-4–7; however, it seems as though both SOCS-6 and -7 are involved in suppression of insulin signaling (21–23).

Since the SOCS proteins were discovered based on their ability to suppress JAK-STAT signaling, we and others investigated whether these proteins were able to suppress signaling induced by cytokines in β-cells, as this might prove to be a new target of β-cell protection. IFN-γ signaling through activation of the JAK-STAT pathway has previously been shown to be one of the classical targets of SOCS-mediated cytokine inhibition, and indeed, we found that SOCS-3 could inhibit IFN-γ signaling in the β-cell line INS-1 (24). Similarly, SOCS-1 has been shown to suppress IFN-γ signaling in the β-cell (25,26). In addition to IFN-γ signaling, SOCS-3 also suppresses IL-1β signaling in the β-cell, an unexpected finding, as this cytokine induces activation of mainly nuclear factor-κB (NF-κB) and mitogen-activated protein (MAP) kinases, demonstrating a novel mode of action of the SOCS proteins (27).

NF-κB is generally involved in cell-survival pathways, and indeed the exposure of β-cells to IL-1β activates both cell protective and deleterious mechanisms. However, in the β-cell the protective response is probably not sufficient to overcome the destructive mechanisms, and eventually apoptosis is induced (1). Numerous IL-1β–induced NF-κB–dependent genes, both proapoptotic such as inducible nitric oxide synthase (iNOS) and antiapoptotic such as manganese superoxide dismutase (MnSOD), are downregulated upon SOCS-3 overexpression in the β-cell (28), suggesting that SOCS-3 influences NF-κB activity. SOCS-3 inhibits IL-1β–induced NF-κB DNA binding as well as IL-1β–induced activation of NF-κB–dependent gene transcription. In addition, SOCS-3 prevents IL-1β–induced IκB degradation (28). Despite the inhibitory effect on both pro- and antiapoptotic pathways, SOCS expression protects the β-cells against the toxic effect of IL-1β.

A central component in the IL-1β–induced signaling pathway is the transforming growth factor-β–activated kinase (TAK-1), as this enzyme is required for activation of the MAP kinase and NF-κB signaling pathways induced by IL-1β (Fig. 3). In addition to NF-κB, SOCS-3 also inhibits IL-1β–induced activation of the MAP kinases JNK and p38 (27). SOCS-3 inhibits IL-1β signaling at the level of the TAK-1 kinase by interfering with the interaction between TAK-1 and TRAF-6, an interaction essential for TAK-1 activation (Fig. 3) (27).

Taken together, expression of SOCS-3 in β-cells leads to a prevention of β-cell destruction upon cytokine exposure due to reduced NF-κB and MAP kinase activation and a reduction in NO production (and probably caspase activation), thereby preventing cell necrosis and/or apoptosis. Array analysis suggested that SOCS-3 inhibits IL-1β–induced activation of genes involved in the immune/inflammatory response such as intracellular adhesion molecule and proteasome and complement components and chemokines (28) along with genes involved in apoptosis, for example the oncogene c-myc, which has been shown to cause apoptosis of β-cells in RIP-II/c-myc transgenic mice (28,29).

Because SOCS proteins can protect β-cells in vitro, it is obvious to hypothesize a protective effect of SOCS expression against diabetes development in vivo, and SOCS-1 has been exploited in this particular context. The T-cell receptor transgenic NOD8.3 mouse is a simple diabetes model in which CD8⁺ T-cell mediated β-cell death can be studied (30). When SOCS-1 is overexpressed in β-cells in these mice (RIP-SOCS-1/NOD8.3), diabetes development is completely prevented. Inflammation of the islets is not affected in this model, indicating that SOCS-1 inhibits the effector pathways activated by the 8.3 T-cells in β-cells, e.g., TNF-α–and IFN-γ–induced Fas expression (31). NOD mice with β-cell–specific overexpression of SOCS-1 have a reduced incidence of diabetes (26,31), correlating with a decreased IFN-γ–induced STAT-1 activation in the SOCS-1–expressing cells (26). These data support that cytokines are involved in the pathogenic process causing type 1 diabetes. However, type 1 diabetes was not completely prevented in the model, illustrating that SOCS-1 overexpression alone is not sufficient to avert diabetes. This might be explained either because the expression level of SOCS-1 was too low to completely abolish the toxic cytokine effects or because mechanisms not influenced by SOCS expression are also involved in the pathogenesis of diabetes. As mentioned, SOCS-1 was shown to inhibit IFN-γ signaling in the transgenic β-cells, whereas an inhibition of other inflammatory cytokines such as IL-1β and TNF-α was not investigated. Redundant effects caused by these cytokines and noncytokine–mediated β-cell killings, such as the T-cell–mediated perforin pathway, probably explain why diabetes was not fully prevented. We have shown that SOCS-3 can inhibit both IL-1β and IFN-γ–mediated β-cell death (24). Moreover, array data has shown that SOCS-3 can suppress IL-1β–induced chemokine expression in β-cells (28), suggesting SOCS-3 as another β-cell protector, since SOCS-3 can inhibit the signaling pathways of several proinflammatory cytokines and perhaps prevent the recruitment of inflammatory cells, an effect that was not observed in the SOCS-1 transgenic islets. Generation of a NOD-RIP-SOCS-3 mouse will help clarify the ability of SOCS-3 as an effective protector of the β-cell.

As mentioned, expression of the SOCS proteins is induced by cytokines, and the physiological role of the SOCS proteins is most likely to prevent uncontrolled cytokine signaling in the cell by negative feedback. In rat islets and NIT-1 insulinoma cells, IFN-γ induces expression of CIS, SOCS-1, and SOCS-2 mRNA and protein (12). SOCS-3 mRNA is induced by IL-1β and IFN-γ in primary rat islets (unpublished observation), while a combination of IL-1β, IFN-γ, and TNF-α induces SOCS-1, -2, and -3 expression in human islets (13). Thus, in addition to induced transcription of proapoptotic genes, the inflammatory cytokines also induce expression of cytokine-protective genes, such as the SOCS genes, and the balance between pro- and antiapoptotic pathways subsequently determines the fate of the β-cell. This balance might in fact represent one of the explanations why the β-cell is more vulnerable to inflammatory cytokines than its neighboring cells. It could be speculated that the induction of SOCS expression is delayed in the β-cell when compared with other cell types, and the level of cytokine-induced SOCS expression in the β-cell may be insufficient to overcome the destructive effect of the cytokines.

The protective effect of SOCS proteins in β-cells could be exploited, as overexpressing SOCS proteins in pancreatic islets or β-cells could be used for transplantation of diabetic patients, thereby preventing either recurrence of disease or allograft rejection. In β-cells, apart from inhibiting the toxic effects induced by inflammatory cytokines, SOCS-3 has been shown to suppress growth hormone and insulin signaling (32,33). Growth hormone is known to induce β-cell proliferation as well as induce insulin gene expression (34,35), and it is a concern that SOCS expression abrogates these effects. However, GLP-1 (glucagon-like peptide-1) or serum-induced β-cell proliferation is not affected by SOCS-3 (32). Furthermore, the replication rate of adult β-cells is very low (36,37), and the mitogenic effect of growth hormone on β-cells is more pronounced in neonatal islets than in adult islets (38). As transplantation of β-cells would normally be performed using adult islets, the beneficial effects of SOCS-3 against cytotoxic cytokines would most likely prevail. In RIP-SOCS-3 tg mice, it was observed that β-cell–specific overexpression of SOCS-3 had a negative effect on the β-cell mass in female mice, but apparently the reduced β-cell mass did not influence their glucose metabolism, which may be explained by an increased insulin secreting capacity of the β-cells (39). These data further suggest that the potential inhibitory effect of SOCS-3 on insulin signaling in β-cells does not intervene with normal β-cell function. SOCS-1–expressing islets delay allograft rejection (40), indicating that the SOCS proteins may have a clinical potential in transplantation strategies.

If endogenous SOCS protein expression in β-cells is insufficient to prevent destruction by cytotoxic cytokines or if overexpression of SOCS proteins in general can protect β-cells, therapeutic approaches manipulating SOCS protein expression may be of great value to diabetic patients and perhaps to people suffering from other inflammatory diseases. However, it should be stated that if in the future the SOCS proteins are going to be used in the clinic, a major challenge will be to obtain an SOCS effect in specific target cells or tissues, such as the β-cells. This will be absolutely necessary in order to avoid side effects due to disturbances of beneficial cytokine effects throughout the body, the most important being their role in the inflammatory response against pathogens. NOD mice with β-cell–specific SOCS-1 expression have a reduced incidence of diabetes but show an enhanced susceptibility to virus-induced diabetes because of a decreased IFN-γ response (26,41), illustrating the important balance between beneficial and nonbeneficial cytokine signaling. Moreover, it is important to avoid conditions in which massive SOCS expression may induce other pathologic conditions (examples discussed below).

The insulin receptor belongs to the tyrosine kinase receptor family and has intrinsic kinase activity, leading to autophosphorylation as well as phosphorylation of intracellular substrates such as insulin receptor substrate (IRS)-1 and -2 (Fig. 4). Subsequently, these phophorylated proteins recruit other signaling proteins, leading to activation of different signaling cascades (42).

Insulin resistance is associated with an increased level of circulation cytokines (e.g., IL-6, growth hormone, TNF-α, IFN-γ, IL-1β, and leptin) (42,43). A reduced level of IRS tyrosine phosphorylation has been observed both in animal models of type 2 diabetes and in type 2 diabetic patients (44). Also, TNF-α has been found to increase serine phosphorylation of IRS-1, which decreases the activity of this protein due to conformational changes, making it unable to interact with the insulin receptor (45).

As mentioned, since the discovery of the SOCS proteins, it has been clear that their expression is induced by cytokines, after which the classical target of the SOCS proteins is to inhibit the JAK-STAT signaling pathway (5–7). However, several reports have established that the SOCS proteins can also inhibit other kinds of signaling pathways, among these being insulin signaling and insulin-like growth factor signaling (46). Based on these findings and on the fact that SOCS-1 knock-out mice have a low blood glucose level and increased insulin signaling (47), the SOCS proteins have been suggested to represent the link between elevated levels of cytokines and insulin resistance.

Phosphorylation of the main signaling components activated by IRS-1 and - 2 is inhibited by SOCS expression both in vitro and in vivo. SOCS-1 and -3 interact with the insulin receptor, but a direct inhibition of autophosphorylation of the insulin receptor has not been reported. On the other hand, SOCS-3 binds phosphorylated Tyr⁹⁶⁰ on the insulin receptor, which is important for IRS-1 binding, proposing one mechanism by which the SOCS proteins inhibit insulin signaling (Fig. 4) (48,49). Phosphorylated Tyr⁹⁶⁰ serves as docking site for signaling molecules activated by insulin, among these STAT-5B, and insulin-induced STAT-5B activation is inhibited by SOCS-3 (48). SOCS-3 inhibits IRS-1 phosphorylation induced by insulin in COS-7 cells and also inhibits binding between IRS-1 and the down-stream phosphatidylinositol-3 kinase (50). Inhibition by SOCS-1 is probably mediated through binding to the kinase domain of the insulin receptor, preventing further phosphorylation (49). Finally, targeting of IRS-1 and -2 for ubiquitin-dependent degradation via the proteasomal machinery has been suggested as another mechanism explaining SOCS-1–and SOCS-3–mediated inhibition of insulin signaling (51). Thus, when mutations were introduced in the conserved SOCS box of SOCS-1, its interaction with the elongin BC ubiquitin-ligase complex was abrogated and ubiquitination and degradation of IRS-1 and -2 prevented (51). Hepatic levels of IRS-1 and -2 are reduced in mice with hepatic overexpression of SOCS-1; moreover, these animals have an impaired glucose tolerance. In contrast, when an SOCS-1 protein with a deletion in the SOCS box region was expressed, IRS levels were unaffected and the animals had a normal glucose response (51), further illustrating the importance of SOCS-1 in insulin resistance.

Elevated levels of SOCS-3 expression were found in the adipose tissue of ob/ob and db/db fat mice when compared with lean control animals, supporting the hypothesis of SOCS-3 as a mediator of insulin resistance (50). Moreover, SOCS-1 and -3 liver mRNA levels are increased in different insulin-resistant mouse models such as db/db mice, ob/ob mice, and mice on high-fat diet (HFD) (52). TNF-α causes an increased level of SOCS-3 in the adipose tissue of OF1 mice (50), and ob/ob TNF-α receptor–deficient mice are more insulin sensitive than wild-type ob/ob mice (53). Since the level of SOCS-3 was reduced so much in adipose tissue of these mice, TNF-α–induced SOCS-3 expression could be a plausible explanation of insulin resistance in obese animals (50).

The involvement of SOCS-3 and -1 in insulin resistance has been further established by studying their influence on insulin signaling in mouse liver. Adenoviral-mediated hepatic overexpression of the SOCS proteins induced a state of insulin resistance in C57BL/6 mice (52), demonstrated by reduced phosphatidylinositol-3 kinase activation, reduced expression, and phosphorylation of IRS-1 and -2 in the liver upon insulin treatment (49). Further, when insulin-resistant db/db mice were treated with antisense oligonucleotides directed against SOCS-1 and -3, the otherwise repressed phosphorylation of IRS-1 and -2 was partially restored, and insulin sensitivity was greatly improved, particularly after reducing SOCS-3 expression in the liver (52).

In humans, a link between elevated levels of inflammatory cytokines, SOCS expression, and insulin resistance has also been suggested. Elevated IL-6 correlates with SOCS-3 expression in skeletal muscle of type 2 diabetic patients when compared with control subjects. Moreover, IL-6 induced SOCS-3 expression and consequently inhibited insulin signaling in human differentiated myotubes grown in vitro (54). SOCS-3 expression was not elevated in muscle from nondiabetic obese individuals, despite an enhanced IL-6 expression and insulin resistance in these subjects. SOCS-3 expression in skeletal muscle of type 1 diabetic patients is not elevated, indicating that the high SOCS expression in type 2 diabetic patients cannot be explained by hyperglycemia. However, high glucose concentrations enhanced IL-6–induced SOCS-3 mRNA expression in cultured human muscle cells, suggesting that the increased SOCS-3 expression in type 2 diabetics may be explained by the combination of high glucose and IL-6 levels in the blood of these individuals (54). Though, it cannot be excluded that the high level of SOCS-3 expression in muscle of type 2 diabetic patients may also be induced by other cytokines or hormones.

To summarize, both in vitro and in vivo data support the hypothesis that high levels of inflammatory cytokines lead to increased expression of SOCS-1 and -3 in insulin-sensitive tissues, which induce insulin resistance via inhibition of the insulin signaling pathway (Fig. 4). Based on these findings, SOCS-1 and -3 are obvious candidate genes coding for the development of type 2 diabetes. However, mutation analysis of the human SOCS-3 gene revealed conflicting results. In one study, a SOCS-3 promoter polymorphism was associated with increased whole-body insulin sensitivity (55), thereby supporting the theory of SOCS-3 as a mediator of insulin resistance, whereas a correlation between variants in the SOCS-3 gene and insulin resistance could not be detected in a U.K. population of female twins (56).

Leptin is an adipocyte-derived cytokine controlling food intake, energy balance, and neuroendocrine function via actions in the hypothalamus. Leptin has significant effects on insulin sensitivity. For example, in conditions of lipodystrophia, characterized by the absence of adipose tissue and consequently leptin, severe insulin resistance is seen. However, the insulin sensitivity is restored upon leptin infusion (57). Like most cytokines, leptin signals through the classical JAK-STAT pathway (Fig. 2). It binds the long isoform of the leptin receptor, which recruits JAK-2 upon activation. Subsequently, JAK-2 is activated and facilitates tyrosine phosphorylation of JAK-2 itself and of Tyr⁹⁸⁵ and Tyr^1,138 of the leptin receptor (58). Phosphorylated Tyr⁹⁸⁵ binds the SH-2 domain containing tyrosine phosphatase, and Tyr^1,138 leads to STAT-3 activation (3), thereby mediating effects of leptin in the cell.

Ob/ob mice lack leptin expression, db/db mice lack a functional leptin receptor, and both of these models are characterized by an obese phenotype and insulin resistance (58). In obese humans and rodent models of obesity, leptin resistance is often observed, characterized by a high circulating level of leptin in the blood that nevertheless fails to mediate its normal effects, such as reduction in food intake and increased energy expenditure, thereby resulting in weight gain (3). The mechanism behind leptin resistance is not clearly understood, but defects in intracellular leptin signaling may represent one explanation. As leptin signals through the JAK-STAT pathway, it has been investigated whether the SOCS proteins are able to suppress leptin signaling, thereby representing an explanation of leptin resistance. SOCS-3 inhibits leptin signaling in mammalian cell lines (59–61), and leptin signaling is enhanced when SOCS-3 is knocked down in RNA interference experiments (61). Leptin-induced STAT-3 activation mediates activation of the SOCS-3 promoter and thereby production of SOCS-3 protein in leptin-sensitive tissues such as the hypothalamus, indicating that SOCS-3 is a classical feedback mechanism regulating leptin signaling (59,61–64). The molecular mechanism behind SOCS-3–mediated inhibition of leptin signaling involves binding of SOCS-3 to phosphorylated tyrosine residues on the activated JAK-2 and leptin receptor, thereby preventing activation of downstream signaling components (60,61).

Several observations in vivo now suggest SOCS-3 as a mediator of leptin resistance in obesity and type 2 diabetes. In the lethal yellow (A^y/a) mouse, which is an obese mouse model of leptin resistance, obesity correlates with an increased SOCS-3 expression in the leptin-sensitive sites in the hypothalamus, indicating that the increased leptin level seen in obesity leads to enhanced SOCS-3 expression, which subsequently inhibits leptin signaling and thereby confers leptin resistance (59). In another study, leptin resistance, measured by the level of STAT-3 phosphorylation in diet-induced obese mice, was likewise found to correlate with elevated SOCS-3 expression in the hypothalamus (65). SOCS-3 knock-out mice are embryonically lethal, but the haplo-insufficient SOCS-3 mouse has a marked reduction of SOCS-3 expression. Interestingly, these mice have increased leptin sensitivity when compared with their wild-type littermates, demonstrated by enhanced weight loss and enhanced tyrosine phosphorylation of STAT3 in the hypothalamus upon exogenous leptin administration (66). When mice are fed HFD, they develop obesity and leptin resistance. However, food consumption and weight gain were comparable in SOCS-3 haplo-insufficient mice, whether they were fed normal diet or HFD, and they ingested less food and gained less weight than their wild-type littermates (66). Wild-type mice on HFD developed insulin resistance, whereas this was prevented in the SOCS-3 haplo-insufficient mice. The exact mechanism behind this is not known, but since leptin is known to increase insulin sensitivity, the leptin resistance seen in the wild-type mice fed HFD may constitute one explanation of their insulin resistance. Interestingly, the SOCS-3 haplo-insufficient mice have normal insulin and blood glucose levels upon being fed HFD, implying that the increased leptin sensitivity in these mice is accompanied by normal insulin sensitivity (66). These results were reproduced in mice with conditional SOCS-3 knock-out in the brain. These mice also have increased leptin-induced STAT-3 phosphorylation in the hypothalamus, associated with increased weight loss, reduced food intake, and resistance to HFD-induced obesity (67). Likewise, HFD-induced insulin resistance was prevented in these mice. The data illustrate that mice with a low SOCS-3 expression level possess an obesity-resistant phenotype, including normal insulin sensitivity, and suggest SOCS-3 as an important inhibitor of leptin effects in vivo.

To summarize, SOCS-3 is a negative feedback inhibitor of leptin signaling. Leptin production is proportionate to body fat mass, and in conditions of obesity the enhanced level of leptin in the blood results in an elevated production of SOCS-3 in leptin-sensitive tissues. Subsequently, SOCS-3 terminates leptin signaling, thereby giving rise to leptin resistance. As elevated levels of SOCS-3 expression have been found in various mouse models of obesity and leptin resistance, suppression of SOCS-3 effects might be of therapeutic interest in order to treat or prevent leptin and insulin resistance as well as obesity (Table 1).

Since the discovery of the SOCS proteins, it has become clear that they are important negative regulators of cytokine signaling in various tissues. Numerous studies have contributed to a better understanding of the physiological role of the SOCS proteins, and SOCS proteins possess interesting properties in relation to both type 1 and type 2 diabetes. The SOCS proteins are able to suppress cytotoxic cytokine signaling, leading to β-cell death and making these proteins interesting targets in attempts to suppress the inflammatory response leading to type 1 diabetes. It has recently been shown that recombinant cell-penetrating SOCS-3 can protect mice from pathogen-induced acute inflammation by suppressing cytokine signaling (68); thus, this concept is testable in clinical islet transplantation. On the other hand, overindulgence of SOCS protein expression seems to be one of the mechanisms behind insulin and leptin resistance in type 2 diabetes, suggesting that inhibitors of SOCS action in insulin- and/or leptin-sensitive tissues might be valuable agents in the treatment of this disease (Table 1). Further studies on the role of SOCS proteins in diabetes might help elucidate important mechanisms involved in the pathogenesis of diabetes and thereby possibly support the SOCS proteins as interesting targets for the development of novel therapeutics for the prevention of diabetes and perhaps other inflammatory diseases.