Cytokine & Growth Factor Reviews Vol. 7, No. 2, pp. 161 173, 1996

Copyright ,(? 1996 Published by Elsevier Science Ltd. All rights reserved

P I I : S1359-6101(96)00021-4 Printed in Great Britain.

E L S E V I E R 1359-6101;96 $32.110 +1).00

SURVEY

Inhibition of Insulin Receptor Signaling by TNF: Potential Role in Obesity and Non-Insulin-Dependent Diabetes Mellitus

Edward Y. Skolnik* and Jerom Marcusohn

Adipocytes produce a variety of molecules that are capable of functioning in both a paracrine and autocrine fashion. Tumor necrosis factor (TNF) is one of the proteins produced by adipocytes that has been shown to regulate adipocyte function. Interestingly, adipocyte expression of TNF increases with increasing adipocyte mass and expression of TNF is increased in adipocytes isolated from several genetic models of rodent obesity and from obese humans. This finding has led to the idea that TNF produced by adipocytes functions as a local "adipostat" to limit fat accumulation. Increased production of TNF by adipocytes, however, may contribute to insulin resistance in obesity and in non-insulin-dependent diabetes mellitus (NIDDM). TNF has been shown to inhibit insulin- simulated tyrosine phosphorylation of both the insulin receptor (IR) and insulin receptor substrate (IRS)-I and to stimulate downregulation of the insulin-sensitive glucose transporter, GLUT4, in adipocytes. These findings raise the possibility that pharmacological inhibition of TNF may provide a novel therapeutic target to treat patients with NIDDM. Copyright 1996 Published by Elsevier Sci ...... Ltd

Key words: Inhibition- Insulin receptor. Signaling. TNF" Obesity" Diabetes.

Obesity is among the most common metabolic disorder in cose transport, GLUT4, from an intracellular vesicle to Western society, affecting more than 30% of the adult the plasma membrane [6, 7]. The importance of obesity population. The negative impact of obesity on health is as a risk factor for the development of clinical NIDDM evident by the increased morbidity and mortality observed (elevated blood glucose levels) is supported by the fin- in this patient population. The increased morbidity caused dings that over 80% of NIDDM patients are obese and by obesity likely arises from both obesity itself and the that weight reduction in this patient population is associ- metabolic consequences that result from the increase in ated with decreased insulin resistance and improved adipocyte mass [1]. With respect to the metabolic derran- blood glucose control. However, while it has long been gements that contribute to the increase in morbidity and known that the propensity to develop diabetes is strongly mortality, obesity is associated with atherogenic lipid pro- linked not only to genetic factors but also to obesity, the files and, perhaps most importantly, insulin resistance and underlying molecular mechanisms responsible for insulin non-insulin-dependent diabetes mellitus (NIDDM) (also resistance in the majority of patients with NIDDM, as known as type 2 diabetes mellitus) [1]. well as the mechanism(s) whereby obesity contributes to

NIDDM affects about 5% of the population and is insulin resistance in this patient population are still characterized by increased blood glucose levels which poorly defined. This review focuses on the role of TNF arise primarily from peripheral resistance to insulin's in mediating insulin resistance in NIDDM. action in fat and muscle [2-5]. Binding of insulin to its receptor on fat and muscle regulates postprandial blood THE ADIPOCYTE AS AN ENDOCRINE ORGAN glucose levels by stimulating the redistribution of a glu-

Although adipocyte hyperplasia and hypertrophy are central to the obese phenotype, there has been little evi- dence until recently that the adipocyte actively par-

*To whom correspondence should be addressed at: New York University MedicalCenter, SkirballInstitute, 540 First Avenue, ticipates in the maintenance of the body weight "set New York, NY 10016, U.S.A. Tel.: 212-263-7458; Fax: 212- point" [8]. The prevailing view had been that the adipo- 263-5711. cyte is a passive organ responsible only for storing tri-

IAI

162 E.Y. Skolnik and J. Marcusohn

glycerides. The recent cloning of the gene responsible termed tumor necrosis factor-alpha (TNF) and tumor for obesity in the ob/ob homozygous mouse has now necrosis factor-beta [also known as lymphotoxin-7 (LT)]. provided conclusive evidence that the adipocyte par- TNF and LT are highly homologous at the protein level ticipates in a feedback loop to regulate food intake, ther- and both molecules engage identical receptors on target mogenesis and energy expenditure [9]. In response to an cells [27, 33]. It is not surprising, therefore, that these increase in adipocyte mass, the protein product of the ob molecules have been found to mediate essentially ident- gene, leptin, is secreted by adipocytes [9-12]. Leptin then ical responses. The major difference between these mol- functions as an endocrine signaling molecule to limit ecules is that, whereas LT is secreted mostly by activated body weight by acting on a receptor in the hypothalmus T- and NK-cells, TNF is expressed by a greater variety (and elsewhere)to decrease food intake and increase ther- of cells [20, 27]. This review focuses on TNF since the mogenesis [11, 13-18]. expression of LT has not been found to be perturbed in

Leptin is unlikely to be the only molecule secreted obese patients. by adipocytes that directly or indirectly regulates body TNF is synthesized as a 26-kDa prohormone which weight. Several other molecules, such as adipsin, angio- undergoes cleavage to yield a 17-kDa soluble TNF mol- tensinogen and TNF have been found to be secreted by ecule [27, 34]. Both the 26-kDa and 17-kDa forms of adipocytes [8, 19, 20]. However, with the exception of TNF are capable of mediating biological responses. The TNF the biological functions ofthese molecules in adipo- difference between these two forms is that the 26-kDa cyte regulation and food intake is still not known. TNF form of TNF contains a transmembrane domain and is regulates a variety of biological activities in the adipocyte, expressed as an integral membrane protein while the 17- Treatment of adipocytes with TNF leads to decreased kDa product functions as a soluble TNF molecule. Sol- lipoprotein lipase (LPL) activity and stimulation of lip- uble TNF, and presumably its transmembrane form, olysis [21-23]. In addition, TNF induces insulin resistance form non-covalent trimers which function as a unit to in adipocytes [20, 24, 25]. It has been hypothesized that engage three receptor molecules leading to receptor these and other actions of TNF function to limit lipid oligomerization and activation (see below)[35]. accumulation in adipocytes [20]. Thus, whereas ob func- TNF mediates its biological effects by binding to two tions as a systematic "adipostat" to regulate adipocyte different receptors on the surface of cells [33, 36]. These size, TNF may function as one of the local "adipostats" receptors have been termed p60 and p80 to coincide with to regulate adipocyte cell size. their molecular weights. The p60 and pS0 TNF receptors

TNF is primarily produced by immune cells and are part of a growing receptor superfamily that is char- mediates a variety of beneficial effects. These include the acterized by multiple cystine-rich domains in the extra- stimulation of immune and inflammatory responses and cellular amino-terminal domain [37]. In contrast to the the involution of transplantable tumors [26, 27]. similarity in the extracellular domains, the cytoplasmic However, soon after its discovery, it was clear that pro- domains of receptors in this group are diverse, implying duction of TNF was not always beneficial for the host. that different receptors couple to different cytoplasmic Overproduction or inappropriate secretion of TNF has signaling molecules. However, although similarities been associated with several deleterious outcomes. Cer- between the cytoplasmic domain of the p60 and p80 ami and co-workers identified the same molecule, which TNF receptors have not been found, recent studies have they termed cachectin, as a potential endogenous identified domains that are conserved among some family mediator causing wasting in infected animals [28]. More- members that function to couple to similar signaling mol- over, overproduction of TNF has been found to be one ecules [38, 39]. of the mediators of circulatory shock and tissue injury in Both TNF receptors are expressed on most cells [33]. sepsis and the CNS manifestation of malaria infection Studies using agonistic antibodies, mutant forms of TNF [27, 29, 30]. TNF secreted by adipocytes also is likely capable of binding only the p60 or p80 receptor, as well to produce either beneficial or deleterious consequences, as mice containing targeted disruption of either the p60 depending on the context in which it is produced. On the or p80 receptors have indicated that activation of the p60 one hand, insulin resistance induced by TNF may be a receptor is sufficient to mediate most biological responses risk factor for the development of atherosclerosis and by TNF [33, 36, 4(~42]. The role of the p80 receptor in hypertension. However, the finding that insulin resist- TNF signaling has been more difficult to resolve. ance, by itself, may function to prevent weight gain sug- Although activation of the p80 receptor has been clearly gests that induction of insulin resistance may also be shown to contribute to signaling by the p60 receptor, beneficial and serve to limit fat accumulation [31, 32]. In activation of the p80 receptor alone has been shown to modern society, the negative impact of insulin resistance activate biological responses under only certain con- arising from increased production of TNF by "overfilled" ditions [33, 36, 43]. This led to the initial idea that under adipocytes appears to outweigh its positive impact, most circumstances, the pS0 receptor functions as an

accessory molecule to facilitate signaling by the p60 TNF AND ITS RECEPTOR receptor [36, 44]. The "ligand passing" model proposes

that ligand binding to the p80 receptor does not directly TNF has been used to describe two related cytokines signal the cell, but rather functions solely to increase the

encoded by distinct genes. These molecules have been local concentration of TNF at the cell surface, thereby

Inhibition of IR Signaling by TNF 163

making more TNF available to bind the p60 receptor ondary changes that contributes to the increase in insulin which is responsible for transducing the TNF signal, resistance.

With respect to insulin resistance, activation of the Current evidence indicates that the primary as well as p60 TNF receptor is also sufficient to mediate insulin secondary defect(s) accounting for insulin resistance in resistance in insulin-responsive tissues[24, 45, 110]. How- NIDDM occur downstream of insulin binding to its ever, while "ligand passing" may be one function of the receptor. Thus, before discussing TNF's role in mediating p80 receptor, recent experimentalevidence has unequivo- insulin resistance, we shall first review what is known cally demonstrated that p80 receptor directly signals cells about how the insulin receptor (I R) normally signals cells and that activation of the p80 receptor also contributes (Figure 1). The early intracellular signaling events linking to the inhibition of insulin signaling by TNF [34, 43, 46, the IR to GLUT4 translocation and glucose uptake in 110]. These findings, coupled with the demonstration that fat and muscle are now beginning to be understood. The the p80 receptor is upregulated in fat and skeletal muscle first step in insulin signaling involves binding of insulin to in the obese rodent model of insulin resistance [47], sug- a cell surface receptor containing tyrosine kinase activity gests that the p80 receptor may play an important role in [52]. The IR is a heterotetrameric protein composed of mediating the insulin resistance induced by TNF. two identical alpha and beta subunits. The alpha subunit

is extracellular and binds the ligand, whereas the beta subunit is mostly cytosolic and contains the receptor tyro-

INSULIN SIGNALING AND THE PATHOGENESIS sine kinase activity. Ligand binding to the alpha subunit OF NIDDM results in autophosphorylation of the IR beta subunit and

activation of the intrinsic tyrosine domain. The tyrosine Insulin maintains blood glucose homeostasis by both kinase is critical for the IR to signal cells; any impairment

stimulating glucose uptake into insulin-sensitive fat and of the tyrosine kinase leads to a decreased ability of the muscle cells and by suppressing glucose production by receptor to stimulate cell growth and metabolic functions, hepatocytes [2, 5, 48]. Insulin mediates glucose uptake such a glucose uptake and glycogen synthesis [52-54]. into peripheral tissues by stimulating the translocation Once activated, the IR associates with and tyrosine of a specialized glucose transporter, GLUT4, from an phosphorylates a variety ofcytosolic signaling molecules intracellular compartment to the plasma membrane [6, in the cell. IR substrates (IRS) 1 and 2 are two of the 7]. Following insertion of GLUT4 into the plasma mem- targets of the I R and comprise a family of 180-190-kDa brane, GLUT4, then functions to transport glucose into cytoplasmic proteins that are tyrosine phosphorylated by the cell. Only adipose tissue, skeletal muscle and cardiac the IR and the insulinqike growth factor (IGF-I) receptor tissue express GLUT4 and, as a result, these tissues are [52, 55]. IRS-1 and 2 contain over 20 potential tyrosine primarily responsibleforinsulin-stimulatedregulationof phosphorylation sites and over 30 potential serine/ blood glucose levels in the post-absorptive state. The fact threonine phosphorylation sites. Many of the tyrosine that skeletal muscle accounts for the removal of > 80% phosphorylation sites on IRS-1,2 are contained within of insulin-stimulated glucose from blood indicates that consensus SH2 binding motifs. SH2 domains are found insulin resistance in muscle must occur for hyperglycemia in a variety of distinct signaling molecules and mediate to develop. Both longitudinal and cross-sectional studies, protein-protein interaction by binding phosphotyrosine as well as studies employing the euglycemicinsulin clamp, moieties in the context of short amino acid sequences have convincingly demonstrated that an increase in per- [56, 57]. Binding of signaling molecules, via their SH2 ipheral resistance to insulin's action in muscle and fat domains, to tyrosine phosphorylated IRS-I,2 couples the initiates the preclinical process of NIDDM [2, 5, 48]. IR to downstream signaling pathways [52]. These findings, coupled with the demonstration that there PI3-kinase is one of the SH2 domain-containing sig- is a strong hereditary component that accounts for haling molecules activated by binding IRS-I,2 that is NIDDM [49, 50], have led to the conclusion that the likely to be critical for coupling the IR to glucose uptake primary or inherited lesion in NIDDM is impaired sen- and GLUT4 translocation [58-63]. Once activated, PI3- sitivity of peripheral tissues to insulin. While secretion of kinase phosphorylates phosphatidylinositol (PI) PI-4,5- insulin by the beta cell of the pancreas initially increases P2 to generate PI-3,4,5-P3 [64]. PI-3,4,5-P~ functions as to maintain euglycemia, over time the beta cell becomes an early second messenger linking the IR to GLUT4 insufficient to maintain euglycemia, and clinical NIDDM translocation and glucose transport, although, the mech- (increased blood glucose levels) develops. The transition anism whereby PI3-kinase mediates this function is still from a compensated state of insulin resistance to clinical not known. The importance of IRS-I and PI3-kinase disease is characterized by a variety of additional changes in insulin-stimulated GLUT4 translocation and glucose which may either be inherited or arise as a result of the uptake is supported by the findings that IRS-1 knockout diabetic environment [2, 5, 48, 49, 5l]. These changes mice are insulin-resistant and that inhibition of PI3- include; (1) a decrease in beta cell function resulting in a kinase, using pharmacological inhibitors or by expression reduced rate of insulin secretion; (2) increased glucose of dominant/negative PI3-kinase molecules, blocks insu- production by the liver; and (3) an increase in peripheral lin-stimulated GLUT4 translocation and glucose uptake resistance to insulin in fat and muscle. Increased pro- in adipocytes [65 69]. duction of TNF by adipocytes may be one of the sec- The genetic or primary component(s) accounting for

164 E.Y. Skolnik and J. Marcusohn

INSULIN

Skeletal Muscle Adipocytes

Figure 1. Insulin receptor signaling. Ligand binding to the IR leads to receptor autophosphorylation which functions to activate the tyrosine kinase catalytic domain and to create binding sites for the PTB domains oflRS-1,2 and SHC. Binding of IRS-1,2 and SHC to the IR juxtaposes these molecules adjacent to tyrosine kinase, thereby enabling them to become tyrosine-phosphorylated. Tyrosine-phosphorylated IRS-1,2 and SHC then bind and activate SH2-containing signaling molecules. Binding of the SH2 domain of GRB2 to SHC results in the activation of the RAS-MAP kinase pathway, whereas binding of the SH2 domains of P85-associated PI3-kinase to IRS1,2 results in the activation of PI3-kinase. PI3-kinase is responsible for coupling the IR to GLUT4 translocation in insulin-responsive tissues such as skeletal muscle and adipocytes.

insulin resistance in the "common type" of NIDDM are sine phosphorylation of IRS- 1; (3) a decrease in GLUT4 still not known. Studies have focused on molecular protein expression in adipocytes; and (4) a decrease in defects in two insulin-stimulated pathways that may be insulin-stimulated GLUT4 translocation and insulin- primarily affected in patients with a genetic pre- stimulated non-oxidative glucose disposal. disposition to develop NIDDM. These pathways are both The nature of these secondary changes that contribute involved in glucose disposal. The first pathway includes to insulin resistance has been the focus of several studies. the activation of GLUT4 translocation and glucose One of the secondary changes that contributes to insulin uptake by insulin [3]. The second pathway involves the resistance is hyperglycemia [48]. High glucose levels activation of glycogen synthase by insulin. Abnormalities impair glucose transport by interfering at several steps in in the non-oxidative pathways of glucose disposal, i.e. the insulin-signaling cascade and have led to the concept glycogen synthesis, have been found in first degree pre- of the "glucose toxicity" model [3, 48]. Several other diabetic relatives of patients with NIDDM and insulin candidate molecules that may contribute to the increase resistance, leading investigators to propose that defects in insulin resistance have also been described [5, 48, 70]. in intracellular glucose metabolism, rather than GLUT4 TNF is one of the best characterized of these molecules. translocation, are primarily responsible for decreased glucose uptake in NIDDM [49]. The early intracellular TNF AND INSULIN RESISTANCE responses activated by the IR are central to the activation of both pathways. However, studies conducted so far The first association of TNF, as well as other cytokines have failed to identify mutations in any candidate genes with insulin resistance, involved studies of patients with in a high percentage of patients with NIDDM [5, 48, 49 several disease states. It has long been known to phys- 52, 70]. icians that infection and other stresses result in increased

In spite of our lack of knowledge about the primary insulin resistance; infection triggers clinical diabetes in defect(s) accounting for NIDDM, a variety of abnor- prediabetic patients and leads to increased insulin malities in insulin signaling have been identified in pat- requirements in diabetic patients [71]. The finding that ients once clinical disease is present. These abnormalities several of these pathological states are also associated further impair signaling by the IR, thereby leading to a with increased levels of circulating TNF, together with worsening of the insulin resistance and hyperglycemia if the finding that chronic administration of TNF to ani- increased insulin secretion from the pancreas is unable to mals induces insulin resistance, supported the idea that compensate for the increased insulin resistance. These TNF contributes to insulin resistance in these diseases abnormalities in IR signaling are thought to be secondary [71, 73]. to either the diabetic state or to environmental factors Increased production of TNF by inflammatory cells, because these changes improve, at least partly, after the such as macrophages, is responsible for high circulation institution of better glycemic control, weight reduction, levels of TNF in the diseases described above [21,27, 63]. and exercise [2, 5, 48]. Some of the changes that have Even before TNF was identified as a secretory product been identified include: (1) a decreased number of IRs; of adipocytes, experiments, performed in cultured cells (2) a decrease in IR tyrosine kinase activity resulting in and whole animals, demonstrated that TNF modulates decreased IR autophosphorylation and decreased tyro- adipocyte function and phenotype. TNF (cachegtin) was

Inhibition of IR Signaling by TNF 165

originally isolated, based on its ability to suppress lipo- pression of GLUT4 and C/EBP-c< genes by TNF was protein lipase in cultured adipocytes [21]. In addition, rapid with maximal suppression of both genes occurring TNF was shown to stimulate lipolysis in fat cells, to after only 4 h of TNF treatment [76]. prevent the differentiation of preadipocytes into adipo- The finding that expression of several adipocyte spec- cytes in cell culture and, when applied at high doses to ific genes is downregulated by TNF in these studies raised adipocytes in vitro, to mediate the reversal of the adipo- the possibility that partial dedifferentiation of adipocytes cyte phenotype [21]. is responsible for the decrease in GLUT4. In addition

With this background, Spiegelman and co-workers whiledownregulation of GLUT4byTNF could certainly provided the first insights into a possible role for TNF in account for insulin resistance, decreased GLUT4 unlikely mediating insulin resistance in rodent models of obesity accounts for the insulin resistance observed in obesity and in obese patients with NIDDM. These investigators and NIDDM. Decreased GLUT4 is found in adipose found that TNF mRNA was elevated five-to-ten-fold in tissue, but not in skeletal muscle from obese patients with adipose tissue isolated from several rodent models of NIDDM [3]. The finding that skeletalmuscle accounts for obesity and insulin resistance compared with lean con- > 80% of the insulin-stimulated glucose uptake indicates trols [25]. The increase in TNF mRNA was also associ- that TNF would also have to induce downregulation of ated with about a two-fold increase in TNF protein GLUT4 in skeletal muscle to induce insulin resistance in measured in explanted adipose tissue from obese adipo- vivo. In addition, more recent evidence has suggested cytes. Moreover, these investigators directly linked TNF that TNF mediates insulin resistance in adipocytes by to at least some of the insulin resistance observed in these mechanisms that are independent of GLUT4 down- obese rodent models. Neutralization of TNF following regulation [25, 45, 74, 77]. First, chronic treatment of the 3 days of treatment of zucker fatty rats (fa/fa) with a 3T3-F442A and 3T3-L1 adipocyte cell lines with doses recombination soluble TNF receptor-Ig (TNFR-Ig) of TNF that were much lower than those usedin previous fusion protein led to a reduction in the insulin resistance studies inhibited insulin-stimulated glucose uptake, in this animal model [25, 74]. although, at this dose of TNF, GLUT4 protein was not

More recent studies have extended some of the findings decreased [45]. Second, neutralization of TNF in fa/fa in obese rodent models of insulin resistance to obese non- rats using a soluble TNFR-Ig fusion protein led to diabetic humans. These studies have confirmed that TNF improved insulin sensitivity, although GLUT4 levels is expressed in human adipocytes and the expression of were not normalized by this treatment [25, 74]. Thus, adipocyte TNF mRNA and protein, for the most part these findings indicate that TNF induces insulin resist- increases with increasing obesity [32, 75]. Moreover, ance by a mechanism(s) that is independent of GLUT4 adipocyte TNF mRNA was found to decrease with downregulation in vivo. weight reduction in the subjects studied. Thus, these pre- liminary studies indicate that at least with regards to the expression of TNF by obese adipocytes, TNF is regulated similarly in human and rodent adipocytes. Inhibition of insulin-stimulated phosphorylation

IR tyrosine kinase activity and insulin-stimulated tyro- sine phosphorylation of IRS-I are decreased in all three MECHANISM WHEREBY TNF INDUCES INSULIN

RESISTANCE insulin-responsive tissues, including adipocytes, skeletal muscle and liver isolated from patients with NIDDM and

Downregulation of GLUT4 from animal models of genetic and acquired forms of obesity and insulin resistance [5, 48]. In addition, the

The cellular and molecular mechanisms linking TNF decrease in IRS-1 phosphorylation is associated with to insulin resistance are now beginning to be understood, decreased insulin-stimulated activation of PI3-kinase [5], Even before it was recognized that adipocytes secrete The findings that these changes frequently improve fol- TNF, several studies demonstrated that TNF down- lowing weight reduction and better glycemic controlindi- regulates GLUT4in 3T3-LI adipocytes and in L6myo- cate that they are likely secondary to the diabetic blasts in cell cultures [24, 76]. Pekala and co-workers environment [48, 78]. Thus, if TNF is an important showed that treatment of 3T3-L1 adipocytes with TNF mediator of insulin resistance in vivo, treatment of insulin- for 3 days reduced GLUT4 protein by 80% [24, 76]. sensitive tissues with TNF should either inhibit the IR Moreover, downregulation of GLUT4 correlated with tyrosine kinase and/or inhibit insulin-stimulated tyrosine TNF-induced insulin resistance in 3T3-L1 adipocytes, phosphorylation ofIRS-1. Studiesin both hepatoma and TNF inhibited GLUT4 by both destabilizing GLUT4 adipocyte cell lines, as well as in rodent models of insulin mRNA and by decreasing GLUT4 transcription [76]. resistance in vivo, indicate that TNF is capable of media- TNF inhibition of GLUT4 transcription was mediated, ting both these effects. at least partially, by TNF-induced downregulation of Treatment of the Fao hepatoma cell line with TNF the transcription factor C/EBP-c~; C.EBP-c~ is thought to resulted in about a 60% decrease in insulin-stimulated control the expression of several adipocyte specific genes, phosphorylation of IRS-1, while treatment of 3T3-L1 including GLUT4. Interestingly, transcriptional sup- and 3T3-F442A adipocyte cell lines with TNF led to

166 E.Y. Skolnik and J. Marcusohn

decreased insulin-stimulated tyrosine phosphorylation of Enhanced production of other molecules which are both the IR and IRS-1 [45, 77, 79, 80]. Interestingly, responsible for the increase in insulin resistance although hepatoma cells exhibited decreased insulin- stimulated phosphorylation after only 15 min of TNF TNF may also mediate insulin resistance via indirect treatment, adipocytes required at least 3 days of treat- mechanisms. For example, TNF-stimulated lipolysis in ment with TNF to mediate this effect. The impairment in the adipocyte may also contribute to insulin resistance in insulin-stimulated phosphorylation by TNF in adipo- vivo by TNF. Several studies have demonstrated that cytes also correlated with decreased insulin-stimulated TNF leads to the stimulation of lipase activity in adipo- glucose uptake. Moreover, low doses of TNF (25 pM vs cytes [23]. Increase in adipocyte lipolysis could then lead 5 nM in the previous studies) were found to inhibit insulin to increased plasma levels of fatty acids, which, by pro- signaling in the 3T3-F442A and 3T3-L1 adipocyte cell viding an alternative source of energy for skeletal muscle, lines in these studies. At these low doses, TNF inhibited would lead to decreased insulin-stimulated glucose insulin signaling without causing downregulation of uptake [83]. GLUT4. Thus, these findings suggest that, at physio- logically relevant levels, TNF mediates insulin resistance TNF INHIBITS INSULIN-STIMULATED in adipocytes primarily by blocking insulin-stimulated PHOSPHORYLATION BY ENHANCING SERINE tyrosine phosphorylation. PHOSPHORYLATION OF IRS-1

To demonstrate that TNF production by adipocytes Treatment of Fao hepatoma cells and adipocytes with contributes to insulin resistance in vivo, these same inves-

TNF results in serine phosphorylation of IRS-1 [77, 80]. tigators determined whether neutralization of TNF by a

From a variety of different studies, it is now clear that soluble TNFR-Ig fusion protein improved insulin sen- serine/threonine phosphorylation of IRS-1, and possibly sitivity and insulin-stimulated phosphorylation in fa/fa the IR itself, negatively regulates insulin signaling, rats [25, 74]. fa and its murine homologue db have now although the mechanism whereby these phosphorylations been shown to encode the ob (leptin) receptor [16, 17, inhibit insulin signaling is still poorly defined (Figure 2). 81]. Mutations in the ob receptor in fatty rats and db Previous studies have shown inhibition of IR signaling mice lead to defective ob receptor signaling, fa/fa rats are after treatment of cells with the serine/threonine phos- a good genetic model of obesity and insulin resistance, phatase inhibitors okadaic acid and calyculin or by over- and marked increases in TNF mRNA can be detected in expressing some isoforms of the serine/threonine protein adipocytes from these animals by 7-8 weeks of life [25]. kinase, protein kinase C [84, 85]. Interestingly, okadaic Treatment of fa/fa rats with soluble TNFR-IgG fusion acid inhibited insulin-signaling by stimulating the serine/ protein for 3 days led to decreased insulin resistance as threonine phosphorylation of IRS-1 [85]. While treat- demonstrated by the improvement in plasma glucose, ment of cells with okadaic acid did not impair the func- insulin and free fatty acid levels [25, 74]. The decrease tion of the IR tyrosine kinase, okadaic acid markedly in insulin resistance after treatment was associated with impaired the ability of the IR to tyrosine phosphorylate improved insulin-stimulated phosphorylation of the IR IRS-1 in vivo and in reconstitution experiments in vitro.

and IRS-1 in skeletal muscle and fat to levels that were These findings suggested that serine/threonine phos- similar to lean control animal. In contrast to the improve- phorylation induced a conformational change in IRS- 1 ment in insulin-stimulated phosphorylation, treatment that may have prevented IRS-1 from interacting with the with TNFR-Ig did not affect the GLUT4 protein levels IR. This suggestion is consistent with the finding that in either skeletal muscle or fat. Thus, two conclusions binding of IRS-1 to the IR is required for IRS-1 phos- can be drawn from these experiments. First, TNF con- phorylation by the IR both in vivo and in vitro [52]. IRS- tributes to the insulin resistance in vivo in at least one 1 and IRS-2 contain a domain that is distinct from SH2 generic rodent model of obesity and insulin resistance, domains that bind phosphotyrosine, termed phos- Second, these findings suggest that inhibition of insulin- photyrosine binding (PTB), which is responsible for the stimulated tyrosine phosphorylation in both skeletal association of IRS-1 and 2 as well as another signaling muscle and fat, rather than downregulation of GLUT4, molecule SHC with the autophosphorylated IR [55, 86- is the primary mechanism whereby TNF mediates insulin 90] (Figure 1). Thus, serine/threonine phosphorylation of resistance in vivo. IRS-1,2 may inhibit tyrosine phosphorylation of IRS-1,2

Surprisingly, treatment of fa/fa rats with soluble by interfering with the binding of IRS-1,2's PTB domain TNFR-Ig did not improve insulin-stimulated phos- to the activated IR. phorylation in the liver [74]. These results could be inter- A recent study, designed to understand the mechanism preted to indicate that TNF does not mediate insulin whereby TNF inhibits insulin signaling, has also dem- resistance in the liver of these animals. However, in con- onstrated that serine/threonine phosphorylation of IRS- trast to other studies [82], these investigators did not 1 is central to the inhibitory effect of TNF on insulin observe a significant decrease in insulin-stimulated phos- signaling [77]. However, the findings in this report suggest phorylation in livers isolated from untreated fa/fa rats that serine phosphorylated IRS-1, may not only uncouple compared to control lean animals after normalizing for an active IR from IRS-1, but may also directly inhibit the number of IRs. the IR tyrosine kinase. In contrast to the finding in oka-

Inhibition of IR Signaling by TNF 167

TNFR

ISer, ne,..reon,ne IPhosphorylation I of IRS-1

Adipocytes j

I? Inhibit tyrosine phosphorylation 1 i i I Inhibit IR kinase I Iof IRS1,2 by inhibiting binding of /

IIRS-1,2 PTB domain to activated IRJ Figure 2. Inhibition of IR signaling by TNF. TNF inhibits insulin signaling by stimulating the serine phosphorylation of IRS-l. Serine phosphorylated IRS-1 inhibits signaling by the IR by inhibiting the tyrosine phosphorylation of IRS-I by the IR. In addition, TNF may also decrease the expression of GLUT4 in adipocytes.

daic acid treated adipocytes, IR tyrosine kinase activity did not inhibit the IR tyrosine kinase & vitro [85, 91]. It assayed in vitro is decreased in adipocytes treated with is possible that the discrepancies observed are due to the TNF [74, 77, 85]. Interestingly, the decrease in IR tyrosine fact that serine/threonine phosphorylation of IRS- 1 leads kinase activity is mediated by IRS-1 that is serine phos- to different outcomes depending upon the sites phos- phorylated as result of TNF treatment. This conclusion phorylated on IRS-I. For example, TNF may activate was based on the observation that TNF did not inhibit distinct signaling pathways in liver and fat which result the IR tyrosine kinase in 32D cells which lack endogenous in the serine phosphorylation of different sites on IRS-I IRS-1,2 [77]. However, in contrast to the findings in the in each cell line. The finding that liver and fat must be parental 32D cells, TNF inhibited the IR in these same treated with TNF for different lengths of time before cells following transfection with a plasmid encoding the TNF inhibits insulin signaling in each cell line is con- IRS-1 cDNA. Moreover, IRS-1 that was serine sistent with the idea that TNF activates distinct signaling phosphorylated inhibited the IR tyrosine kinase pathways in these two ceils. activity in vitro. The IR tyrosine kinase activity was In contrast to the findings by Hotamisligil et al. [74, inhibited in vitro by IRS-1 that was isolated either from 77], a recent study has suggested that downregulation adipocytes stimulated in cell culture with TNF or from of GLUT4 by TNF in L1-3T3 adipocytes, rather than IRS-1 isolated from skeletal muscle or adipocytes of fa/fa impaired insulin-stimulated phosphorylation, is the pri- rats. The inhibition of the IR kinase depended upon mary mechanism whereby TNF inhibits glucose trans- serine phosphorylation of IRS-I because dephos- port in this cell line [92]. These investigators found that phorylation of IRS-1 with calf intestinal alkaline phos- treatment of LI-3T3 adipocytes with low doses of TNF phatase reversed the inhibitory effects of IRS- 1. Although led to a marked reduction in GLUT4 protein levels after 3 it is still not clear whether serine phosphorylated IRS-I days of TNF treatment. However, the signaling pathway directly or indirectly inhibits the IR tyrosine kinase via connecting the IR to GLUT4 translocation appeared the association with another molecule, these findings sug- intact in these cells since insulin was still capable of gests a second possible mechanism whereby serine phos- stimulating GLUT4 translocation in these cells. The phorylated IRS-1 inhibits insulin signaling, decrease in glucose uptake was attributed to the decrease

While it is clear that TNF inhibits insulin-stimulated in total pool of GLUT4 available for translocation rather phosphorylation in a variety of different cells, not all than changes in insulin-stimulated tyrosine phos- studies have demonstrated an inhibitory effect of TNF phorylation. At present, the reason for the discrepancy on IR tyrosine kinase activity. Treatment of the Fao in these results is not clear. hepatoma cell line with TNF led to decreased tyrosine phosphorylation of IRS-1 without inhibiting the IR tyro- SIGNALING PATHWAYS ACTIVATED BY TNF sine kinase [80]. In addition, whereas serine phos- THAT MEDIATE SERINE PHOSPHORYLATION IRS- phorylated IRS-1 isolated from TNF-treated adipocytes 1 AND INSULIN RESISTANCE inhibited the IR tyrosine kinase, serine/threonine phos- phorylated IRS-I isolated from okadaic acid treated The signaling pathway(s) activated by the TNF recep- adipocytes or from cells overexpressing protein kinase C tor that mediates insulin resistance are not yet known.

168 E.Y. Skolnik and J. Marcusohn

However, as discussed above, the mechanism whereby ceramide induces insulin resistance in these cells [109, TNF stimulates serinephosphorylation of lRS-1 is likely 110]. Ceramide has been shown to activate a serine/ to be central in confering insulin resistance. TNF could threonine protein kinase termed ceramide activated increase serine phosphorylation of IRS-I by either acti- protein kinase (CAP) [94]. Although CAP kinase is not vating a kinase and/or inhibiting a phosphatase. The yet cloned, in vitro reconstitution experiments have sug- finding that ceils treated with the phosphatase inhibitor gested that this kinase couples TNF to the RAS-RAF- okadaic acid exhibit both an increase in serine/threonine MAP kinase signaling pathway in HL60 cells [101]. How- phosphorylated IRS-1 and insulin resistance indicates ever, while MAP kinase (also known as ERK) can phos- that an active kinase that is able to phosphorylate IRS-1 phorylate IRS-1 in vitro activation of MAP kinase is is present in unstimulated cells and that this kinase must unlikely to be responsible for the insulin resistance be kept in check by a phosphatase [85]. However, as of mediated by TNF [52]; other receptors that activate MAP now, it is still not known whether TNF stimulates IRS-1 kinase, such as PDGF, do not induce insulin resistance. phosphorylation by regulating a kinase or a phosphatase. Nevertheless, the finding that CAP kinase activates a well Moreover, the different length of time in which adipo- known kinase cascade indicates that CAP kinase is a cytes and hepatocytes must be treated with TNF before candidate kinase that may directly or indirectly stimulate they manifest insulin resistance, suggests that TNF serine/threonine phosphorylation of IRS-1. Two mediates insulin resistance by distinct mechanisms in additional pieces of evidence suggest that ceramide may different cell lines [45, 80]. The finding that insulin resist- mediate insulin resistance caused by TNF. First, several ance in Fao hepatoma cells occurs after only 15 min other cytokines have been shown to mediate insulin resist- of TNF treatment indicates that TNF mediates insulin ance in adipocytes in culture. Interestingly, these cyto- resistance in this cell via post-translational modification kines, which include interleukin (IL)-I and interferon 7, of an existing protein. However, the extended length of have been shown to stimulate the production ofceramide time that adipocytes must be treated before they manifest in cells [45, 93]. Second, ceramide regulates the tran- insulin resistance suggests that TNF mediates insulin scription of a variety of proteins in cells [94, 96]. These resistance in adipocytes by regulating the transcription findings raise the possibility that ceramide activates a of a gene that either directly or indirectly stimulates IRS- signaling pathway, which regulates the transcription of a 1 phosphorylation, still yet to be defined gene that is responsible for the

Unlike the IR, which contains intrinsic tyrosine kinase increase in IRS-1 phosphorylation and insulin resistance activity, neither the p60 not the p80 TNF receptors con- in insulin-responsive tissues, such as adipocytes. Treat- rains intrinsic catalytic activity, and thus it has been ment of insulin-responsive tissues with cell permeable difficult to determine the mechanism whereby these recep- analogs of ceramide should shed light on the possible role tors signal cells. As of now, the TNF receptors have been of ceramide in mediating insulin resistance by TNF. shown to utilize two distinct mechanisms to couple to TNF also activates PC-PLC which generates DAG by proximal cytoplasmic signaling molecules (Figure 3). stimulating hydrolysisofphosphatidylcholine[95]. DAG First, the p60 TNF receptor has been shown to activate is likely an important second messenger responsible for at least two lipid second messengers. TNF activates mediating cellular responses by TNF. DAG is a well- sphingomylinase, leading to the generation of the lipid established activator of the serine/threonine kinase pro- second messenger ceramide and also a phos- tein kinase C (PKC). In addition, DAG activates an phatidylcholine-specific phospholipase C (PC-PLC)lead- acidic sphingomyelinase, which has been implicated in ing to the generation of diacylglycerol (DAG) [93-96]. mediating the activation of NF-xB by TNF [95, 96]. PKC Second, both TNF receptors have now been shown to has received a great deal of interest as a potential negative associate with a variety of signaling molecules in cells, regulator of insulin-stimulated phosphorylation. Acti- [97-99]. Oligomerization of the receptors is thought to vation of endogenous PKC, using phorbol esters or over- activate these molecules, although the mechanism expression of some PKC isoforms inhibits insulin- whereby these molecules are activated is still poorly stimulated phosphorylation in cells [84, 91, 102]. Some defined [100]. studies, although not all, have shown that activation of

Ceramide is a.component of all sphingolipids and over protein kinase C leads to serine/threonine phos- the past several years it has become evident that the phorylation of the IR resulting in decreased IR kinase p60 TNF receptor activates both an acidic and neutral activity [52, 91, 102]. In addition, overexpression of sphingomylinase, which hydrolyses sphingolipid to gen- PKC~ has been shown to stimulate serine/threonine erate the lipid second messenger ceramide [93, 94 96]. phosphorylation of IRS-1 in CHO cells [91]. However, Using cell permeable analogs of ceramide, ceramide has IRS-1 isolated from CHO cells overexpressing PKC~ is been linked to the regulation of a variety of biological phosphorylated normally by the IR in vitro [91]. This is responses. Some of these responses include the stimu- in contrast to the decreased in vitro phosphorylation of lation of apoptosis, monocyte differentiation and acti- serine/threonine phosphorylated IRS-1 isolated from vation of NF-xB [93, 94]. Ceramide may also mediate okadaic acid-treated or TNF-treated adipocytes [77, 85]. some of the insulin resistance induced by the p60 TNF Moreover, staurosporin, a non-specific inhibitor of pro- receptor; treatment of hepatoma cells and 3T3-L1 adipo- tein kinase C, did not abrogate the ability of TNF to cytes with sphingomylinase or cell-permeable analogs of inhibit insulin-stimulated phosphorylation in Fao hepa-

Inhibition of IR Signaling by TNF 169

p 8 0 ~ ~ TNFRI I [ I TNFR I I I I

( ~ ) . neutral IPC - P L C [ ~ PC I [ ceramld-e I ~SMase • •

[ ceramine I

I III I*= I I TRAF2 NF-•B activation Others TRADD ? Other effects

other TNFR associated proteins

Figure 3. TNF receptor signaling. See text for details. Ligand binding of the TNF receptor leads to receptor oligomerization and activation of several intracellular signaling pathways. The p60 TNF receptor activates sphingomylinase, leading to the generation of the lipid second messenger ceramide and also a phosphatidylcholine-specific phospholipase C (PC-PLC) leading to the generation of diacylglycerol (DAG). In addition, both TNF receptors have now been shown to associate with a variety of signaling molecules in cells. The signaling pathway, activated by TNF that mediates insulin resistance, is still not known.

tocytes [80]. Thus, these findings suggest that PKC is proteins, TRAF2, activates NFkB when overexpressed unlikely to mediate insulin resistance caused by TNF. in cells [98]. These findings raise the intriguing possibility

A variety of other serine/threonine kinases are acti- that one of the proximal signaling molecules responsible vated by the p60 TNF receptor. These kinases include for stimulating insulin resistance by TNF will be found the stress activated protein kinase (SAP, also known as to directly associate with the TNF receptor. Struc- N-terminal Jun kinase), p38, protein kinase A and /% ture/function studies, in which mutant TNF receptors are casein kinase [103-105]. SAP kinase and p38 are members expressed in insulin-responsive tissues, should allow the of the MAP kinase family and are strongly activated in identification of the cytoplasmic portion of the TNF cells treated with TNF and ceramide [111]. However, receptor responsible for mediating insulin resistance. SAP kinase is also unlikely to mediate the inhibition of These studies should be a valuable first step to help insulin-stimulated phosphorylation in TNF-treatedcells; uncover signaling molecules that are responsible for overexpression of activated p21 Rac in 3T3-L1 adipo- stimulating insulin resistance by TNF. cytes results in constitutive activation of SAP kinase, yet insulin-stimulated phosphorylation and glucose uptake HOW IS THE TNF mRNA REGULATED IN are normal [106]. ADIPOCYTES?

In addition to generating the lipid second messenger described above, both the p60 and p80 TNF receptors It is still not known what signals the "overfilled" adipo- have been shown to associate with signaling molecules cyte to increase production of TNF mRNA. The finding that can mediate several TNF-stimulated biological that TNF mRNA is increased in adipocytes isolated from responses. The cytoplasmic domain of the p60 TNF a variety of different genetic models of obesity suggests receptor contains a domain of about 100 amino acids that the signal to increase TNF mRNA may be intrinsic that has been termed the death domain [38]. This domain to the adipocyte [25]. Specifically, the adipocyte may is responsible for both receptor oligomerization and the directly regulate its own production of TNF mRNA association of the p60 TNF receptor with a variety of based upon the amount of stored triglycerides. Alter- signaling molecules which are capable of stimulating natively, obesity itself may lead to the secretion of a apoptosis when overexpressed in cells [100]. Similarly, common factor(s), which in turn stimulates the adipocyte several proteins that associate with cytoplasmic domain to produce TNF mRNA. Leptin, however, cannot be the of the p80 receptor have been identified [46]. One of these secreted factor responsible for increasing adipocyte TNF

170 E.Y. Skolnik and J. Marcusohn

mRNA. This is based on the fact that adipocyte TN F proposed that adipocytes found in skeletal muscle of mRNA is increased in homozygous ob mice which lack obese subjects are responsible for the increased pro- leptin and that exogenous administration of leptin to duction of T N F in muscle. However, as of yet, increased these animals corrects the insulin resistance [14]. expression of T N F by adipocytes infiltrating skeletal

muscle has not been shown. It remains a possibility that T N F produced by adipocytes indirectly stimulates insulin

FUTURE QUESTIONS resistance in skeletal muscle by stimulating the pro- duction of a circulating factor, which is then responsible

The above findings raise the intriguing possibility that for inhibiting insulin-stimulated phosphorylation in T N F comprises one of the central molecules responsible muscle. Future studies will no doubt more directly test for mediating insulin resistance in obesity and NIDDM. TNF's role in both mediating insulin resistance and in However, many questions remain to be answered before functioning as a local "adipostat" to limit fat accumu- the role of T N F as an endogenous mediator of insulin lation. These studies will likely include the crossing of resistance in these diseases becomes clear. One critical insulin resistance rodent models of obesity with mice question that remains unanswered is whether the findings containing targeted disruption of the p60 and p80 TNF in genetic models of rodent obesity and insulin resistance receptors or of TNF. The role of TN F in mediating can be extrapolated to N I D D M in humans. At least with insulin resistance in humans with N I D D M will likely be respect to TNF m R NA by adipocytes, both humans and more difficult to resolve. Nevertheless, these findings raise rodents are similar; adipocytes isolated from obese pat- the intriguing possibility that treatment of N ID D M may, ients express increased T N F mRNA and protein [32, one day, include therapies designed to block TN F pro- 75]. However, other findings in humans are not entirely duction by adipocytes and/or TNF's effects on insulin- consistent with experimental results obtained either from sensitive target cells. adipocytes treated in cell culture with T N F or from rod- ent models of obesity and insulin resistance. Defects in REFERENCES IR tyrosine kinase activity, for the most part, have not been observed in skeletal muscle or fat isolated from 1. Flier JS. Obesity. In Kahn CRandWeirGC, eds. Diabetes

Mellitus. Lea and Febiger, Malverne, PA, 1994, 351-362. obese insulin resistant patients prior to the onset ofclini- 2. DeFronzo RA, Bonadonna RC, Ferrannini E. Patho- cal diabetes, despite the fact that T N F mRNA is genesis of NIDDM. A balanced overview. Diabetes Care increased in these patients [48, 107]. Moreover, a recent 1992, 15, 318-368. report has demonstrated a decrease in protein levels of 3. Garvey WT. Glucose transport and NIDDM. Diabetes

the IR, IRS-1 and P85 associated with PI3-kinase in Care 1992, 15, 396-417. skeletal muscle isolated from obese insulin-resistant sub- 4. Haring HU, Mehnert H. Pathogenesis of Type 2 diabetes

mellitus: candidates for a signal transmitter defect causing jects [108]. These findings led these investigators to con- insulin resistance of the skeletal muscle. Diabetologia 1993, clude that decreased levels of these proteins, rather than 36, 176-182. impairment in insulin-stimulated phosphorylation of 5. Kahn CR. Insulin action, diabetogenes, and the cause of

type II diabetes. Diabetes 1994, 43, 1066-1084. IRS-1, is primarily responsible for insulin resistance in 6. Birnbaum MJ. The insulin-sensitive glucose transporter. these patients. As of yet, T N F has not been shown to Int Rev Cytol 1992, 13, 239-97. mediate a decrease in these protein levels in skeletal 7. James DE, Piper RC. Targeting of mammalian glucose muscle. These findings, together with the demonstration transporters. J Cell Sci 1993, 104, 607-612. that insulin resistance must be accompanied with pan- 8. Flier JS. The adipocyte: Storage depot or node on the

energy information superhighway? Cell 1995, 80, 15-18. creatic beta cell failure for clinical diabetes to develop, 9. Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, suggests that TNF-independent events are likely critical Friedman JM. Positional cloning of the mouse obese gene for N I D D M to occur [51]. and its human homologue. Nature 1994, 372, 425-431.

In addition, very little is known about the mechanism 10. Maffei M. Increased expression in adipocytes of ob RNA whereby TNF mRNA expression is regulated in adipo- in mice with lesions of the hypothalmus and with cytes or the mechanism whereby T N F mediates insulin mutations in the db locus. Proc Natl Acad Sci USA 1995,

92, 6957-6960. resistance either in cell culture or in obese rodent models. 11. Hallaas JL, Gajiwala KS, Maffei Met al. Weight reducing For example, neither the signaling pathways activated by effects of the plasma protein encoded by the obese gene. T N F that are responsible for serine phosphorylating IRS- Science 1995, 269, 543-546. 1 nor the mechanism whereby serine-phosphorylated 12. Frederich RC, Lollmann B, Hamann Aet al. Expression IRS-1 inhibits the IR tyrosine kinase are known. The of ob mRNA and its encoded protein in rodents. J Clin

Invest 1995, 96, 1658 1663. mechanism whereby TNF produced by adipocytes blocks 13. McGarry JD. Does leptin lighten the problem of obesity? glucose transport in skeletal muscle is also not known. Curr Biol 1995, 5, 1342 1344. The fact that increased circulating plasma levels of TN F 14. Pellymounter MA, Cullen MJ, Baker MB et al. Effects of are observed in only a small percentage of obese fatty the obese gene product on body regulation in ob/ob mice. rats and not at all in obese humans suggests that TN F Science 1995, 269, 540-543. must be produced locally within skeletal muscles to 15. Campfield LA, Smith F J, Guisez Y, Devos R, Burn P.

Recombinant mouse ob protein: evidence for peripheral mediate its effects. Production of TNF by skeletal muscle signal linking adiposity and central neural networks. Sci- has not been easy to demonstrate. Therefore, it has been ence 1995, 269, 546~59.

Inhibition of I R Signaling by TNF 171

16. Chen H, Charlat O, Tartaglia LA et al. Evidence that the 38. Tartaglia LA, Ayres TM, Wong GHW, Goeddel DV. A diabetes gene encodes the leptin receptor: Identification of novel domain within the 55 kd TNF receptor signals cell a mutation in the leptin receptor gene in db/db mice Cell death. Cell 1993, 74, 845-853. 1996, 84, 491495. 39, Itoh, N, Nagata S. A novel protein domain required for

17, Lee GH, Proenca R, Montez JM et al. Abnormal splicing apoptosis. J Biol Chem 1993, 268, 10932-10937. of leptin receptor in diabetic mice. Nature 1996, 379, 632 40. Rothe J, Lesslauer W, Lotscher H et al. Mice lacking the 635. tumor necrosis factor receptor 1 are resistant to TNF-

18. Tartaglia LA, Dembski M, Weng X et al. Identification mediated toxicity but highly susceptible to infection by and expression cloning of a leptin receptor, OB-R. Cell Listeria monocytogenes. Nature 1993, 364, 798. 1995, 83, 1263-1271. 41. Erickson S, de Sauvage FJ, Kikly K et al. Decreased

19. Speigelman BM, Choy L, Hotamisligil GS, Graves RA, sensitivity to tumor-necrosis factor but normal T-cell Tontoniz P. Regulation of adipocyte gene expression in development in TNF receptor-2-deficient mice. Nature differentiation and syndromes of obesity/diabetes. J Biol 1994, 372, 561~563. Chem 1993, 268, 6823-6826. 42. Pfeffer K, Matsuyama T, Kundig TM et al. Mice deficient

20. Spiegelman BM, Hotamisligil GS. Through thick and for the 55 kd tumor necrosis factor receptor are resistant thin: Wasting, obesity, and TNF-e. Cell 1993, 73, 625- to endotoxic shock, yet succumb to L. monocytogenes 627. infection. Cell 1993, 73, 457467.

21. Beutler B, Cerami A. Tumor necrosis, cachexia, shock, 43. Grell M, Scheurich P, Meager A, Pfizenmaier K. TR60 and inflammation: a common mediator. Ann Rev Biochem and TR80 tumor necrosis factor-receptors can inde- 1988, 57, 505 518. pendently mediate cytolysis. Lympho Cyto Res 1993, 12,

22. Fong Y, Lowry SF. Tumor necrosis factor in the patho- 143-148. physiology of infection and sepsis. Clin Immunol Immu- 44. Tartaglia LA, Pennica D, Goeddel DV. Ligand passing: nopath. 1990, fig, 157-170. the 75-kDa tumor necrosis factor (TNF) receptor recruits

23. Grunfeld C, Feingold KR. The metabolic effects of tumor TNF for signaling by the 55-kDa TNF receptor. J Biol necrosis factor and other cytokines. Biotherapy 1991, 3, Chem 1993, 268, 18542 18548. 143 158. 45. Hotamisligil GS, Murray DL, Choy LN, Spiegelman BM.

24. Stephens JM, Pekala PH. Transcriptional repression of Tumor necrosis factor e inhibits signaling from the insulin the GLUT4 and C/EBP genes in 3T3-L1 adipocytes by receptor. Proc Natl Acad Sci USA 1994, 91, 48544858. tumor necrosis factor-e. J Biol Chem 1991, 266, 21839 46. Rothe M, Wong SC, Henzel WJ, Goeddel DV. A novel 21845. family of putative signal transducers associated with the

25. Hotamisligil GS, Shargill NS, Spiegleman BM. Adipose cytoplasmic domain of the 75 kDa tumor necrosis factor expression of tumor necrosis factor-e: Direct role in obes- receptor. Cell 1994, 78, 681-692. ity-linked insulin resistance. Science 1993,259, 87-91. 47. Hofmann C, Lorenz K Braithwaite SS et al. Altered gene

26. Fiers W. Tumor necrosis factor: characterization at the expression for tumor necrosis factor-alpha and its recep- molecular, cellular and in vivo levels. F E B S Lett 1991 285, 199 212. ' tor: during drug and dietary modulation of insulin resist-

27. Beutler B. TNF, Immunity and Inflammatory Disease: ance. Endocrinology 1994, 134, 264-270. Lessons of the Past Decade. J Invest Med 1995, 43, 227- 48. Olefsky JM, Nolan JJ. Insulin resistance and non-insulin- 235. dependent diabetes: cellular and molecular mechanisms.

Am J Clin Nutr 1995, 61,980S-986S. 28. Beutler B, Greenwald D, Hulmes JD et al. Identity of 49. Beck-Nielsen H, Groop LC. Metabolic and genetic tumor necrosis factor and the macrophage-secreted factor

characterization of prediabetic states. J C/in Invest 1994, cachectin. Nature 1985, 316, 552-554. 29. Berendt AR, Simmons DE, Tansey J, Newbold Cl, Marsh 94, 1714-1721.

K. Intracellular adhesion molecule-1 is an endothelial cell 50. Granner DK, O'Brien RM. Molecular physiology and adhesion receptor for plasmodium falciparum. Nature genetics of NIDDM. Diabetes Care 1992, 15, 369-388. 1989, 341, 57-59. 51. Polansky KS, Sturis J, Bell GI. Non-insulin-dependent

30. Tracey K J, Vlassara H, Cerami A. Cachectin/TNF. Lan- diabetes mellitus-a genetically programmed failure of the cet 1989, 1, 1122-1126. beta cell to compensate for insulin resistance. N Engl J

3 I. Eckel RH. Insulin resistance: an adaption for weight main- Med 1996, 334, 777 783. tenance. Lancet 1992, 340, 1452 1453. 52. White MF, Kahn CR. The insulin signaling system. J Biol

32. Kern PA, Saghizadeh M, Ong JM, Bosch RJ, Deem R, Chem 1994, 269, 14 . Simsolo RB. The expression of tumor necrosis factor in 53. Quon M J, Guere-Millo M, Zarnowski MJ et al. Tyrosine human adipose tissue: Regulation by obesity, weight loss, kinase-deficient mutant human insulin receptors and relationship to lipoprotein lipase. J Clin Invest 1995, (Met l153-11e) overexpressed in transfected rat adipose 95,2111 2119. cells fails to mediate translocation of epitope-tagged

33. Vandenabeele P, Declerq W, Beyaert R, Fiers W. Two GLUT4. P r o c N a t l A c a d S c i USA 1994, 91, 5587--5591. tumor necrosis factor receptors: structure and function. 54. Ebina Y, Araki R, Taira M e t al. Replacement of lysine Tremts Cell Biol 1995, 5, 392 399. residue 1030 in the putative ATP-binding region of the

34. Grell M, Douni E, Wajant H et al. The transmembrane insulin receptor abolishes insulin and antibody stimulated form of tumor necrosis factor is the prime activating ligand glucose uptake and receptor tyrosine kinase~ Proc Natl of the 80 kDa tumor necrosis factor receptor. Cell 1995, Acad Sci USA 1987, 84, 704-708. 83, 793-802. 55. Sun XJ, Wang L-M, Zhang Y e t al. Role of IRS-2 in

35. Banner DW, D'Arcy A, Janes W et al. Crystal structure insulin and cytokine signaling. Nature 1995, 377, 173-177. of the soluble human 55 kd TNF receptor-human TNFfl 56. Pawson T. Protein modules and signaling networks. Nat- complex: Implications for TNF receptor activation. Cell ure 1995, 373, 573 580. 1993, 73, 431445. 57. Schlessinger J. SH2/SH3 signaling molecules. Curr Opin

36. Tartaglia LA, Goeddel DV. Two TNF receptors. Immunol Genet Dez" 1994, 4, 25 30. Today 1992, 13, 151 153. 58. Escobedo JA, Navankasattusas S, Kavanaugh WM, Mil-

37. Smith CA, Farrah T, Goodwin RG. The TNF receptor fray D, Fried VA, Williams LT. cDNA cloning of a novel superfamily of cellular and viral proteins: Activation, 85 kd protein that has SH2 domains and regulates binding costimulation, and death. Cell 1994, 76, 959-962. of Pl3-kinase to the PDGF receptor. Cell 1991,65, 75-~82.

172 E.Y. Skolnik and J. Marcusohn

59. Carpenter CL, Duckworth BC, Auger KR, Cohen BS, Reversibility of defective adipocyte insulin receptor kinase Schatthausen BS, Cantley LC. Purification and charac- activity in non-insulin dependent diabetes mellitus: effect terization of PI3-kinase from rat liver. J Biol Chem 1990, of weight loss. J Clin Invest 1998, 82, 1398-1406. 265, 19704-19711. 79. Feinstein R, Kanety H, Papa MZ, Lunenfeld B, Karasik

60. Skolnik EY, Margolis B, Mohammadi M e t al. Cloning A. Tumor necrosis factor-e suppresses insulin-induced of PI3-kinase associated p85 using a novel method for tyrosine phosphorylation of insulin receptor and its sub- expression/cloning of target proteins for receptor tyrosine strates. J Biol Chem 1993, 268, 26055-26058. kinases. Cell 1991, 65, 83-90. 80. Kanety H, Feinstein R, Papa MZ, Hemi R, Karasik A.

61. Otsu M, Hiles I, Gout Ie t al. Characterization of two 85 Tumor necrosis factor ct-induced phosphorylation ofinsu- kD proteins that associate with receptor tyrosine kinases, lin receptor substrate-1 (IRS-1): Possible mechanism for middle T, complexes and PI3 kinase. Cell 1991, 65, 91- suppression of insulin-stimulated tyrosine phosphory- 104. lation of IRS-1. J Biol Chem 1995, 270, 23780-23784.

62. Hiles I, Otsu M, Volinia S, Fry MJ et al. PI3-kinase: 81. Chua SC, Chung WK, Wu-Peng X et al. Phenotypes of structure and expression of the 110 kd catalytic subunit, mouse diabetes and rat fatty due to mutations in the OB Cell 1992, 70, 419-429. (leptin) receptor. Science 1996, 271, 994-996.

63. Backer JM, Myers MG Jr, Shoelson SE et al. Phos- 82. Hurrel DG, Pedersen O, Kahn CR. Alterations in the phatidylinositol 3-kinase is activated by association with hepatic insulin receptor kinase in genetic and acquired IRS-1. EMBO J 1992, 11, 3469-3479. obesity in rats. Endocrinology 1989, 125, 2454-2462.

64. Cantley LC, Auger KR, Carpenter C et al. Oncogenes and 83. Randle PJ, Garland PB, Hales CN, Newsholme EA. The signal transduction. Cell 1991, 64, 281-302. glucose fatty-acid cycle: its role in insulin sensitivity and

65. Araki E, Lipes MA, Patti ME et al. Alternative signalling the metabolic disturbances of diabetes mellitus. Lancet pathway in mice with targeted disruption of the IRS-1 1963, i, 785-789. gene. Nature 1994, 372, 186-190. 84. Chin JE, Liu F, Roth RA. Activation of protein kinase

66. Tamemoto H, Kadowaki T, Tobe K et al. Insulin resist- Ce inhibits insulin-stimulated tyrosine phosphorylation ance and growth retardation in mice lacking insulin recep- of insulin receptor substrate 1. Molec Endocr 1994, 8, 51- tot substrate-1. Nature 1994, 372, 182-185. 58.

67. Okada T, Kawano Y, Sakaibara T, Hazeki O, Ui M. 85. Tanti KJF, Gremeaux T, van Obberghen E, Le Marc- Essential role of PI3-kinase in insulin-induced glucose hand-Brustel Y. Serine/threonine phosphorylation of transport and antilipolysis in rat adipocytes. J Biol Chem insulin receptor substrate 1 modulates insulin receptor 1994, 269, 3568-3573. signaling. J Biol Chem 1994, 269, 6051-6057.

68. Cheatham B, Vlahos CJ, Cheatham L, Wang L, Blenis 86. Blaikie P, Immanuel D, Wu J, Li N, Yajnik V, Margolis J, Kahn CR. Phosphatidylinositol 3-kinase activation is B. A region in Shc distinct from the SH2 domain can bind required for insulin stimulation of pp70 $6 kinase, DNA tyrosine phosphorylated growth factor receptors. J Biol synthesis, and glucose transporter translocation. Molec Chem 1994, 269, 32031-32034. Cell Biol 1994, 14, 4902-4911. 87. Kavanaugh WM, Williams LT. An alternative to SH2

69. Quon M J, Chen H, BLI et al. Roles of 1-phos- domains for binding tyrosine-phosphorylated proteins. phatidylinositol 3-kinase and ras in regulating trans- Science 1994, 266, 1862-1865. location of GLUT4 in transfected rat adipose cells. Molec 88. O'Neill TJ, Craparo A, Gustafson TA. Characterization Cell Biol 1995, 15, 5403-5411. of an interation between insulin receptor substrate 1 and

70. Pillay TS, Langlois W J, Olefsky JM. The genetics of non- insulin receptor using the two-hybrid system. Molec Cell insulin dependent diabetes mellitus. Adv Genet 1995, 32, Biol 1994, 14, 6433-6442. 51-96. 89. Gustafson TA, He W, Craparo A, Schaub CD, O'Neill

71. Argiles JM, Lopez-Soriano J, Lopez-Soriano FJ. Cyto- TJ. Phosphotyrosine-dependent interaction of SHC and kines and diabetes: The final step? Involvement of TNF-e insulin receptor substrate 1 with the NPEY motif of the in both type I and II diabetes mellitus. Horm Metab Res insulin receptor via a novel non-SH2 domain. Molec Cell 1994, 26, 447-449. Biol 1995, 15, 2500-2508.

72. Lang CH, Dobrescu C. Sepsis-induced changes in in vivo 90. Isakoff SJ, Yu Y-P, Su Y-C et al. Interaction between the insulin action in diabetic rats. Am Physio 1989, 257, E301- phosphotyrosine binding domain of Shc and the insulin 308. E301-308. receptor is required for Shc phosphorylation by insulin in

73. Lang CH, Dobrescu C, Bagby GJ. Tumor necrosis factor vitro. J Biol Chem 1996, 271, 3959-3962. impairs insulin action on peripheral glucose disposal and 91. Danielsen AG, Liu F, Hosomi Y, Shii K, Roth RA. Acti- hepatic glucose output. Endocrinoloyy 1992, 130, 43-52. vation of protein kinase Ce inhibits signaling by members

74. Hotamisligil GS, Budavari A, Murray D, Spiefelman BM. of the insulin receptor family. J Biol Chem 1995, 270, Reduced tyrosine kinase activity of the insulin receptor in 21600-21605. obesity-diabetes: Central role of tumor necrosis factor-e. 92. Stephens JM, Lee J, Pilch PF. The primary cause of TNF- J Clin Invest 1994, 94, 1543-1549. induced insulin resistance in fat cells is GLUT4 down-

75. Hotamisligil GS, Arner P, Caro JF, Atkinson RL, Spi- regulation. Diabetes 1995, 44, SI 11A. egelman BM. Increased adipose tissue expression of tumor 93. Hannun Y, Obeid LM. Ceramide: an intracellular signal necrosis factor-e in human obesity and insulin resistance, for apoptosis. Trends Biochem Sci 1995, 120, 73-77. J Clin Invest 1995, 95, 240%2415. 94. Kolesnick R, Golde DW. The sphingomyelin pathway

76. Stephens JM, Pekala PH. Transcriptional repression of in tumor necrosis factor and interleukin-1 signaling. Cell the C/EBP-c~ and GLUT4 genes in 3T3-L1 adipocytes 1994, 77, 325-328. by tumor necrosis factor-e: Regulation is coordinate and 95. Schutze S, Machleidt T, Kronke M. The role of diacyl- independent of protein synthesis. J Biol Chem 1992, 267, glycerol and ceramide in tumor necrosis factor and inter- 13580-13584. leukin-I signal transduction. J Leukocyte Biol 1994, 56,

77. Hotamisligil GS, Peraldi P, Budavari A, Ellis R, White 533-541. MF, Speigelman BM. IRS-1 mediated inhibition of insulin 96. Wiegmann K, Schutze S, Machleidt T, White D, Kronke receptor tyrosine kinase activity in TNF-e- and obesity- M. Functional dichotomy of neural and acidic sphingo- induced insulin resistance. Science 1995, 271, 665-668. myelinases in tumor necrosis factor signaling. Cell 1994,

78. Friedenberg GR, Reichart D, Olefsky JM, Henry RR. 78, 1005-1015.

Inhibition of IR Signaling by TNF 173

97. Hsu H, Xiong J, Goeddel DV. The TNF receptor 1-associ- 106. Marcusohn J, Isakoff SI, Rose E, Symons M, Skolnik EY. ated protein TRADD signals cells death and NF-~B acti- The GTP-binding protein Rac does not couple PI3-kinase vation. Cell 1995, 81, 495-504. to insulin-stimulated glucose transport. Curr Biol 1995, 5,

98. Rothe M, Sarma V, VMD, Goeddel DV. TRAF2- 1296-1302. mediated activation of NF-kappa B by TNF receptor 2 107. Nolan LL, Friedenberg G, Henry R, Reichart D, Olefsky and CD40. Science 1995, 269, 1424-1427. JM. Role of skeletal muscle insulin receptor kinase in the

99. Rothe M, Pan M-G, Henzel WJ, Ayres TM, Goeddel DV. in vivo insulin resistance of noninsulin-dependent diabetes The TNFR2-TRAF signaling complex contains two novel mellitus and obesity. J Clin Endocr Metab 1994, 78, 471 -- proteins related to bacnlovirus inhibitor of apoptosis. Cell 477. 1995, 83, 1243-1252. 108. Goodyear LJ, Giorgino F, Sherman LA, Carey J, Smith

100. Cleveland JL, Ihle JN. Contenders in FasL/TNF death RJ, Dohm GL. Insulin receptor phosphorylation, insulin signaling. Cell 1995, 81,479-482. receptor substrate-1, and phosphatidylinositol 3-kinase

101. Yao B, Zhang Y, Delikat S, Mathias S, Basu S, Kolesnick activity are decreased in intact skeletal muscle strips from obese subjects. J Clin Invest 1995, 95, 2195-2204.

R. Phosphorylation of Rafby ceramide-activated protein 109. Kanety H, Hemi R, Papa MZ, Karasik A. Sphingo- kinase. Nature 1995, 378, 307-310.

102. Berti L, Mosthaf L, Kroder G e t al. Glucose-induced myelinase and ceramide suppress insulin-induced tyrosine phosphorylation of the insulin receptor substrate- 1. J Biol

translocation of protein kinase C isoforms in rat-1 fibro- Chem 1996, 271, 9895-9897. blasts is paralleled by inhibition of the insulin receptor 110. Peraldi P, Hotamisligil GS, Buurman WA, White MF, tyrosine kinase. J Bio! Chem 1994, 269, 3381-3386. Spiegelman BM. Tumor necrosis factor (TNF)~ inhibits

103. Davis RJ. MAPKs: new JNK expands the group. Trends insulin signaling through stimulation of the p55 TNF Biochem Sci 1994, 19, 470-473. receptor and activation of sphingomyelinase. J Biol Chem

104. Vilcek J, Lee TH. Tumor necrosis factor. J Biol Chem 1996, 271, 13018-13022. 1991, 266, 7313-7316. 111. Verheij M, Bose B, Lin XH, Yao B, Jarvis WD, Grant S,

105. Guesdon F, Waller RJ, Saklatvala J. Specific activation Birrer ML Szabo E, Zon LI, Kyriakis JM, Haimovitz- of beta-casein kinase by the inflammatory cytokines inter- Friedman A, Fuks Z, Kolesnick RN. Requirement for leukin 1 and tumor necrosis factor. Biochem J 1994, 3 0 4 , ceramide-initiated SAPK/JNK signalling in stress-induced 761-768. apoptosis. Nature 1996, 380, 75 79.