PAPER

Inflammatory pathways and insulin action

GS Hotamisligil1*

1Department of Genetics and Complex Diseases, Harvard School of Public Health, Boston, MA, USA

Obesity and type 2 diabetes are associated with a state of abnormal inflammatory response. While this correlation has also beenrecognized in the clinical setting, its molecular basis and physiological significance are not yet fully understood. Studies in recentyears have provided important insights into this curious phenomenon. The state of chronic inflammation typical of obesity andtype 2 diabetes occurs at metabolically relevant sites, such as the liver, muscle, and most interestingly, adipose tissues. Thebiological relevance of the activation of inflammatory pathways became evident upon the demonstration that interference withthese pathways improve or alleviate insulin resistance. The abnormal production of tumor necrosis factor alpha (TNF-a) inobesity is a paradigm for the metabolic significance of this inflammatory response. When TNF-a activity is blocked in obesity,either biochemically or genetically, the result is improved insulin sensitivity. Studies have since focused on the identification ofadditional inflammatory mediators critical in metabolic control and on understanding the molecular mechanisms by whichinflammatory pathways are coupled to metabolic control. Recent years have seen a critical progress in this respect by theidentification of several downstream mediators and signaling pathways, which provide the crosstalk between inflammatory andmetabolic signaling. These include the discovery of c-Jun N-terminal kinase (JNK) and Ikappabkinase (IkK) as critical regulators ofinsulin action activated by TNF-a and other inflammatory and stress signals, and the identification of potential targets. Here, therole of the JNK pathway in insulin receptor signaling, the impact of blocking this pathway in obesity and the mechanismsunderlying JNK-induced insulin resistance will be discussed.

International Journal of Obesity (2003) 27, S53–S55. doi:10.1038/sj.ijo.0802502

Keywords: JNK; TNF-a; IRS-1; signaling pathways; metabolism; diabetes; inflammation

Obesity is associated with a state of chronic inflammation

characterized by abnormal production of proinflammatory

cytokines and acute phase reactants.1 Several cell types, such

as adipocytes and macrophages, are involved in this

abnormal cytokine production. This seemingly surprising

ability of fat cells to mount an immune response is likely to

be a function of its common features with macrophages2 and

critical in systemic insulin action.3

Insulin signaling is a very complex process that involves

multiple pathways and cascades of phosphorylation events.

For the purposes of this short review, these pathways will not

be described; however, readers could refer to excellent recent

reviews on these pathways.4,5 Over the past decade, it is well

established that insulin receptor signaling is defective at

multiple levels in type 2 diabetes both in experimental

models and humans and this defective signaling is central to

pathogenesis of type 2 diabetes. One such pathway that

involves phosphorylation and signaling via the members of

the insulin-receptor substrate (IRS) family of proteins will be

the focus of the present discussion, since it is particularly

relevant to insulin-mediated glucose metabolism and cross-

talk with inflammatory pathways.4

Phosphorylation of tyrosine (tyr) residues of IRS-1 upon

activation of the insulin receptor constitutes a critical step in

the transmission of the insulin signal to downstream

effectors and biological outcomes. Several laboratories have

demonstrated that insulin-stimulated tyr phosphorylation of

IRS-1 as well as other proximal substrates of the insulin

receptor is reduced in obesity, and such reduction is believed

to be at the core of the development of insulin resistance.5

We and others have shown that the proinflammatory

cytokine tumor necrosis factor alpha (TNF-a), which pro-

motes insulin resistance in several insulin-responding cells

and tissues, inhibits tyr phosphorylation of IRS-1 resulting in

a state reminiscent of type 2 diabetes and obesity.6 Interest-

ingly, the same phenomenon is observed in the presence of

an increase in intracellular fatty acids,7 which also promote

insulin resistance, showing that both cytokines and fatty

acids may act via common pathways to block insulin action.

Under these conditions, cytokines and fatty acids stimulate

phosphorylation of IRS-1 on its serine (ser) residues.8–10 In

contrast to tyr phosphorylation, which results in the

transmission of the insulin signal, ser phosphorylation

*Correspondence: Dr GS Hotamisligil, Department of Genetics and

Complex Diseases, Harvard School of Public Health, 665 Huntington

Av., Boston, MA 02115, USA.

E-mail: ghotamis@hsph.harvard.edu

International Journal of Obesity (2003) 27, S53–S55& 2003 Nature Publishing Group All rights reserved 0307-0565/03 $25.00

www.nature.com/ijo

blunts insulin signaling.8–10 In experimental systems, when

ser phosphorylation is prevented, the deleterious action of

TNF-a and possibly fatty acids on insulin signaling might be

prevented as well, further indicating the critical importance

of this modification in insulin-resistant states.

We and others have been investigating the molecular basis

and functional consequences of IRS-1 ser phosphorylation

and the responsible kinase(s). The inhibition of insulin

signaling by TNF-a and fatty acids in adipocytes and

hepatocytes involves activation of a kinase utilizing IRS-1

as a substrate.8,9 Once serine is phosphorylated, IRS-1

becomes a poor substrate for the insulin receptor, and

preventing such serine phosphorylation restores IRS-1-

mediated insulin receptor signaling in vitro. Of physiological

significance is the fact that in obesity and type 2 diabetes,

such an inhibitory phosphorylation of IRS-1 exists, in the

adipocyte and also in other metabolically important tissues.6

However, the identification of this kinase(s) and of its site(s)

of action on IRS-1 has been difficult due to the presence of

multiple phosphorylation sites on IRS-1, the existence in

cells of multiple distinct kinases that can act on IRS in vitro,4

and by the fact that TNF-a-mediated phosphorylation of IRS-

1 is a relatively weak phenomenon and therefore uneasy to

quantify.8,9,11 Despite such constraints, recent work has shed

light on IRS-1 phosphorylation sites as well as kinases

relevant to insulin resistance. Among others, biochemical

analysis has led to the identification of several serine

phosphorylation sites on IRS-1, of which ser-307 appears to

be of major importance and targeted by c-Jun N-terminal

kinase (JNK) in vitro.10 In cell-based systems and whole

animals, JNK serves as a sensing juncture for cellular stress

and the inflammatory status.12 JNK also phosphorylates IRS-

1 at the ser-307 site robustly. It is likely that other sites are

also targeted by this kinase, some simultaneously with serine

307, some depending on the stimulus and context. Phos-

phorylation of IRS-1 at ser-307 inhibits tyrosine phosphor-

ylation, thereby inactivating further transmission of the

insulin signal. Serine phosphorylation can also be involved

in the targeting of IRS-1 for proteasome-mediated acute

degradation under certain but not all conditions. Of

relevance is the fact that JNK activity is elevated in the

tissues of several obesity models, including obesity resulting

from high-fat feeding and the genetic deficiency of leptin.12

In these mice, there is also elevation of ser-307 phosphoryla-

tion of IRS-1 and decreased tyrosine phosphorylation upon

insulin stimulation.

These observations indicated that JNK might be a key

molecule leading to insulin resistance and type 2 diabetes.

This was tested and validated in models where obesity is

generated in genetic deficiency of JNK isoforms, JNK1 and

JNK2.12 The JNK1 and JNK2 isoforms are expressed in most

tissues at different levels and ratios, whereas JNK3 is

primarily present in the central nervous system.13 In

experiments with complete deficiency of JNK1 or JNK2,

most of the obesity-related increase in JNK activity and

the metabolic consequences are attributable to the JNK1

isoform.12 JNK1 knockout mice gain less weight when fed

either a regular low-fat diet or a high-fat diet. Compared to

wild-type mice, adipose tissue of the JNK1�/� mice is

morphologically normal, with smaller adipocytes, and the

liver contains dramatically less fat. Overall, there is a 10%

reduction in total body adipose mass in JNK1�/� compared

to wild-type controls with no alterations in the lean body

mass.12 Other tissues do not appear to be affected by lack of

JNK1. Interestingly, the volumes of subcutaneous and

visceral but not gonadal fat depots are reduced in JNK1�/�mice. Plasma levels of adiponectin are increased while those

of resistin are reduced in JNK1-deficient mice on a high-fat

diet.12 JNK1 deficiency also dramatically enhances insulin

sensitivity and confers protection against diet-induced

insulin resistance as well as insulin resistance associated

with the ob/ob model of genetic obesity,12 although to a

lesser extent in the latter. It may seem counterintuitive at

first glance that the improvement in insulin sensitivity,

which should increase the anabolic action of insulin, leads

instead to reduced fat accretion. The reasons for such an

effect remain elusive at this time but suggest that the impact

of the JNK pathway on insulin action is not dependent on its

potentially distinct effects on whole-body adiposity. The

physiological relevance of the JNK1 isoform is further

illustrated by the fact that it is specifically increasedFin

liver, adipose tissue and skeletal muscleFin obesity resulting

in increased total JNK activity and increased IRS-1 serine

phosphorylation at these sites.12 In JNK1-deficient mice,

obesity-related ser-307 phosphorylation of IRS-1 is substan-

tially reduced demonstrating that JNK1 is the predominant

ser kinase activity in obesity acting on IRS-1.12 Both TNF-aand fatty acids increase JNK1 activity as well as ser

phosphorylation of IRS-1, and ser-307 phosphorylation is

reduced in JNK1 knockout mice. It must be mentioned that

JNK1 is not likely to be the sole serine kinase involved in the

development of insulin resistance. Several studies have

indeed underlined the importance of other kinases such as

protein kinase c (PKC) and IkK.14,15 The interactions

between these various kinases is complex and understanding

of these networks and mechanisms remain incomplete.16

The above discussion has focused on findings in cultured

cells and murine models, but there is evidence that JNK

might be highly relevant to human health. Indeed, human

mutations in the gene coding for JNK-binding protein, a

natural inhibitor of JNK activity, causes type 2 diabetes.17 It

is therefore not unreasonable to suggest that increased JNK

activity more generally plays a role in the development

of insulin resistance and type 2 diabetes in humans. If

JNK inhibitors turn out to be effective in improving

insulin sensitivity as much as gene targeting in experimental

models, they may offer attractive possibilities of deve-

loping effective means for the treatment of human insulin

resistance.

I should like to conclude this review with some thoughts

regarding the evolutionary reasons and advantages of the

remarkable overlap and correlations between the metabolic

Inflammation and insulin actionGS Hotamisligil

S54

International Journal of Obesity

and inflammatory signaling pathways. To this end, I will

allude to the fruit fly Drosophila melanogaster. This lowly and

distant cousin of ours bears a rather unique organ, called the

fat body, which functions at once as liver and hematopoie-

tic/immune system.18 This site also appears as the mamma-

lian homologue of adipose tissue.19 Remarkably, the GATA

family of transcription factors, which influences adipocyte

differentiation in mammals, is involved in the development

of the fat body of Drosophila.19 In the course of evolution,

the individualization of the hepatic, adipose and hemato-

poietic/immune organs occurred, while the function of a

common network of functional and molecular pathways

were retained in all. This might explain why there are so

many overlapping pathways in the hepatocyte, adipocyte,

macrophage and lymphocyte. The same molecules assume

both metabolic and inflammatory roles. Depending on the

cell type, the action of these overlapping pathways will

produce end points, such as insulin resistance (adipocyte),

atherosclerosis (macrophage) or hyperlipidemia (hepato-

cyte).

What could be the selective advantage of such a strong

overlap between metabolic and inflammatory signaling

pathways? One such advantage is that a strong immune

response confers higher odds of surviving infection, and

would therefore have been favored by natural selection. At

the same time however, in absence of starvation, a strong

immune response primes the system for an equally strong

metabolic response (eg insulin resistance) to stimuli such as a

high-energy and -fat intake. A survival advantage in hard

times hence becomes deleterious in times of plenty.

Acknowledgements

I thank Julie Gound and Gurol Tuncman for help in pre-

paring this manuscript. Gokhan S Hotamisligil is supported

by grants from the National Institutes of Health and

American Diabetes Association.

References

1 Sethi JK, Hotamisligil GS. Semin Cell Dev Biol 1999; 10: 19–29.2 Charriere G, Cousin B, Arnaud E, Andre M, Bacou F, Penicaud L,

Casteilla L. J Biol Chem 2003; 278: 9850–9855.3 Uysal KT, Wiesbrock SM, Marino MW, Hotamisligil GS. Nature

1997; 389: 610–614.4 White MF. Am J Physiol Endocrinol Metab 2002; 283: E413–E422.5 Saltiel AR, Kahn CR. Nature 2001; 414: 799–806.6 Hotamisligil GS In: LeRoith D, Taylor SI, Olefsky JM (eds).

Diabetes Mellitus, A Fundamental and Clinical Text. Lippincott-Raven: Philadelphia (in press).

7 Shulman GI. J Clin Invest 2000; 106: 171–176.8 Hotamisligil GH, Peraldi P, Budavari A, Ellis R, White MF,

Spiegelman BM. Science 1996; 271: 665–668.9 Kanety H, Hemi R, Papa MZ, Karasik A. J Biol Chem 1996; 271:

9895–9897.10 Aguirre V, Uchida T, Yenush L, Davis R, White MF. J Biol Chem

2000; 275: 9047–9054.11 Aguirre V, Werner ED, Giraud J, Lee YH, Shoelson SE, White MF. J

Biol Chem 2002; 277: 1531–1537.12 Hirosumi J, Tuncman G, Chang L, Gorgun CZ, Uysal KT, Maeda

K, Karin M, Hotamisligil GS. Nature 2002; 420: 333–336.13 Davis RJ. Cell 2000; 103: 239–252.14 Griffin ME, Marcucci MJ, Cline GW, Bell K, Barucci N, Lee D,

Goodyear LJ, Kraegen EW, White MF, Shulman GI. Diabetes 1999;48: 1270–1274.

15 Yuan M, Konstantopoulos N, Lee J, Hansen L, Li ZW, Karin M,Shoelson SE. Science 2001; 293: 1673–1677.

16 Jiang G, Dallas-Yang Q, Liu F, Moller DE, Zhang BB. J Biol Chem2003; 278: 180–186.

17 Waeber G, Delplanque J, Bonny C, Mooser V, Steinmann M,Widmann C, Maillard A, Miklossy J, Dina C, Hani EH, Vionnet N,Nicod P, Boutin P, Froguel P. Nat Genet 2000; 24: 291–295.

18 Rehorn KP, Thelen H, Michelson AM, Reuter R. Development 1996;122: 4023–4031.

19 Tong Q, Dalgin G, Xu H, Ting CN, Leiden JM, Hotamisligil GS.Science 2000; 290: 134–138.

Inflammation and insulin actionGS Hotamisligil

S55

International Journal of Obesity