Cell Metabolism

Article

Interleukin-6 Signaling in Liver-ParenchymalCells Suppresses Hepatic Inflammationand Improves Systemic Insulin ActionF. Thomas Wunderlich,1,6 Peter Strohle,1,6 A. Christine Konner,1 Sabine Gruber,1 Sulay Tovar,1 Hella S. Bronneke,2

Lisa Juntti-Berggren,3 Luo-Sheng Li,3 Nico van Rooijen,4 Claude Libert,5 Per-Olof Berggren,3 and Jens C. Bruning1,*1Institute for Genetics, University of Cologne, Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases

(CECAD), Center of Molecular Medicine Cologne (CMMC), 2nd Department for Internal Medicine University Hospital Cologne andMax PlanckInstitute for the Biology of Ageing, D-50674 Cologne, Germany2Phenotyping Facility of the Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD),

D-50674 Cologne, Germany3Rolf Luft Research Center for Diabetes and Endocrinology, Karolinska Institute, SE-171 77 Stockholm, Sweden4Department of Molecular Cell Biology, Vrije Universiteit, N-1081 Amsterdam, The Netherlands5Department for Molecular Biomedical Research, Flanders Institute for Biotechnology, University Ghent, B-9052 Ghent, Belgium6These authors contributed equally to this work

*Correspondence: jens.bruening@uni-koeln.deDOI 10.1016/j.cmet.2010.06.011

SUMMARY

The contribution of interleukin (IL)-6 signaling inobesity-induced inflammation remains controversial.To specifically define the role of hepatic IL-6 sig-naling in insulin action and resistance, we havegenerated mice with hepatocyte-specific IL-6 recep-tor (IL-6R) a deficiency (IL-6RaL-KO mice). Theseanimals showed no alterations in body weight andfat content but exhibited a reduction in insulinsensitivity and glucose tolerance. Impaired glucosemetabolism originated from attenuated insulin-stim-ulated glucose transport in skeletal muscle and fat.Surprisingly, hepatic IL-6Ra-disruption caused anexaggerated inflammatory response during euglyce-mic hyperinsulinemic clamp analysis, as revealedby increased expression of IL-6, TNF-a, and IL-10,as well as enhanced activation of inflammatory sig-naling such as phosphorylation of IkBa. Neutraliza-tion of TNF-a or ablation of Kupffer cells restoredglucose tolerance in IL-6RaL-KO mice. Thus, ourresults reveal an unexpected role for hepatic IL-6signaling to limit hepatic inflammation and to protectfrom local and systemic insulin resistance.

INTRODUCTION

Obesity-associated insulin resistance develops in the presence

of activated inflammatory signaling in white adipose tissue

(WAT), liver, skeletal muscle, and even in the central nervous

system (CNS) (Cai et al., 2005; Hotamisligil et al., 1993; Kleinrid-

ders et al., 2009; Plomgaard et al., 2005; Zhang et al., 2008).

During obesity development, infiltration of immune cells such

as macrophages and T-lymphocytes into WAT leads to

increased secretion of tumor necrosis factor (TNF)-a and inter-

Cell Me

leukin-6 (IL-6) (Nishimura et al., 2009; Xu et al., 2003). Whereas

activation of TNF-a-inducible kinases in insulin target tissues

impairs insulin signal transduction (Aguirre et al., 2000; Arkan

et al., 2005; Cai et al., 2005; Hirosumi et al., 2002), the function

of IL-6 in this context is not yet fully elucidated and is controver-

sially debated (Pedersen and Febbraio, 2007).

IL-6 serum levels are increased in obesity (Bastard et al.,

2000), and IL-6 injection can cause hepatic insulin resistance

mediated by IL-6-induced expression of suppressor of cytokine

signaling (SOCS) 3 (Klover et al., 2003). Moreover, IL-6 infusion

during euglycemic hyperinsulinemic clamp analysis caused

reduced glucose infusion rates, indicating that IL-6 indeed

causes insulin resistance (Kim et al., 2004). Similarly, blocking

IL-6 action via application of neutralizing antibodies improved

glucose tolerance and insulin sensitivity in mice with liver-

specific overactivation of NFkB signaling (Cai et al., 2005). Strik-

ingly, though these studies are in line with IL-6 causing insulin

resistance and increased IL-6 concentrations and action con-

tributing to the development of obesity-associated insulin resis-

tance, numerous studies and independent approaches indicate

that IL-6may also improve glucosemetabolism. Acute IL-6 stim-

ulation leads to activation of AMP-dependent kinase (AMPK) by

a yet undetermined mechanism and thus results in increased

glucose uptake and b oxidation (Carey et al., 2006). Furthermore,

IL-6 injection increased whole-body insulin sensitivity and

improved glucose tolerance in a more recent study (Holmes

et al., 2008). Of interest, central insulin action results in hepatic

IL-6 release, leading to the activation of signal transducer and

activator of transcription (Stat) 3-dependent signaling in hepato-

cytes to inhibit expression of genes required for hepatic glu-

cose production, such as glucose-6-phosphatase (G6Pase)

and phosphoenolpyruvate carboxykinase (Inoue et al., 2006).

It has been demonstrated that this effect depends on insulin

action in agouti-related peptide-expressing neurons (Konner

et al., 2007), which are essential for control of energy homeo-

stasis, food intake, and glucose metabolism (Gropp et al.,

2005; Plum et al., 2006). Moreover, IL-6 knockout mice devel-

oped mature onset obesity under some conditions (Wallenius

tabolism 12, 237–249, September 8, 2010 ª2010 Elsevier Inc. 237

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

et al., 2002), whereas other investigators did not find an effect of

IL-6 on body weight control (Chida et al., 2006; Di Gregorio et al.,

2004). However, more recently, it was demonstrated that over-

expression of human IL-6 in mice improves insulin sensitivity

(Sadagurski et al., 2010).

Taken together, the role of IL-6 action in obesity-associated

insulin resistance remains highly controversial, and the effect

of IL-6 signaling in individual tissues with respect to control of

glucose and energy homeostasis has not been addressed so

far. Here, we show that mice with hepatic deficiency of IL-6

receptor a (IL-6Ra) develop insulin resistance not only in liver,

but also in skeletal muscle andWAT. Systemic insulin resistance

in IL-6RaL-KO mice develops in the presence of an exaggerated

inflammatory response in liver and muscle. Depletion of TNF-a

or liver-resident Kupffer cells in themutantmice restores glucose

homeostasis and insulin sensitivity to normal. Thus, our results

reveal a function for hepatic IL-6 action to limit local but also

systemic inflammation originating from Kupffer cell activation

to prevent insulin resistance.

RESULTS

Generation of IL-6RaL-KO MiceTo determine the specific role of hepatocyte-autonomous IL-6

signaling in control of glucose homeostasis, i.e., insulin action

and the development of insulin resistance, we aimed to generate

mice with a conditionally inactivatable IL-6Ra allele. To this end,

we generated a targeting vector, in which exons 2 and 3 of the

IL-6Ra gene were flanked by loxP sites (Figure 1A). After electro-

poration of this vector in C57/BL6-derived Bruce-4 embryonic

stem (ES) cells, correctly targeted ES cell clones were identified

by Southern blot analysis (Figure 1B). Homologous recombinant

ES cell clones, which still contained a FRT site-flanked neomycin

resistance gene, were used to generate mice carrying the

IL-6RaFL-neo allele (Figure 1A). These mice were crossed with

mice expressing the Flp recombinase in the germline (Rodrıguez

et al., 2000) to remove the neomycin resistance gene and to yield

IL-6RaFL mice (Figure 1A). IL-6RaFL mice were crossed with

ALFP-Cre mice (Kellendonk et al., 2000), and double heterozy-

gous animals were further intercrossed to yield IL-6RaFL/FL

ALFP-Cre, i.e., hepatocyte-specific IL-6Ra knockout mice on

a C57/BL6 background (IL-6RaL-KO).

To investigate deletion efficiency and specificity of the loxP-

flanked IL-6Ra allele in these mice, we performed Southern

blot analyses on genomic DNA isolated from liver, WAT, skeletal

muscle, and tails of IL-6RaL-KO and control mice. This analysis

revealed efficient Cre-mediated recombination of the conditional

IL-6Ra allele exclusively in liver of IL-6RaL-KO mice, but not in

other organs (Figure 1C). Moreover, to directly determine hepatic

IL-6 responsiveness, we performed IL-6 injections into the vena

cava of anesthetized control and IL-6RaL-KO mice and analyzed

tyrosine phosphorylation of Stat-3. Whereas IL-6 injection

resulted in robust phosphorylation of Stat-3 in liver of con-

trol mice, this response was dramatically blunted in liver of

IL-6RaL-KO mice (Figure 1D). The residual IL-6-stimulated

Stat-3 phosphorylation in liver of IL-6RaL-KO mice most likely

arises from IL-6 responsiveness of the liver-resident Kupffer

cells, which are not affected by the ALFP-Cre transgene. Finally,

determination of IL-6Ra gene expression from livers of control

238 Cell Metabolism 12, 237–249, September 8, 2010 ª2010 Elsevie

and IL-6RaL-KO mice by real-time PCR demonstrated nearly

complete absence of IL-6Ra mRNA expression in IL-6RaL-KO

mice (Figure 1E).

The IL-6Ra chain exists in a membrane-bound and in

a cleaved, soluble form (sIL-6Ra), the latter of which also renders

cell types IL-6 responsive that do not express IL-6Ra (Mackie-

wicz et al., 1995). To investigate the contribution of hepato-

cyte-derived IL-6Ra to the bioavailability of sIL-6Ra, we deter-

mined serum sIL-6Ra concentrations in control and IL-6RaL-KO

mice (Figure 1F). However, this analysis revealed similar abun-

dance of the sIL-6Ra in both control and IL-6RaL-KO mice, indi-

cating that hepatocyte-specific inactivation of IL-6Ra does not

affect circulating sIL-6Ra concentrations. Moreover, ablation of

hepatic IL-6Ra expression had no effect on liver weight and his-

tomorphological appearance, as revealed by hematoxylin and

eosin (H&E) staining (Figures 1G and 1H). In summary, these

experiments unequivocally demonstrate the successful genera-

tion of hepatocyte-specific IL-6Ra knockout mice.

IL-6RaL-KO Mice Exhibit Unaltered Energy HomeostasisConventional IL-6 knockout mice—at least in some studies—

present with mature onset obesity; however, it is not clear in

which organ lack of IL-6 signaling may lead to the development

of obesity (Wallenius et al., 2002). To determine whether IL-6

signaling in liver affects energy homeostasis, we first analyzed

body weight of control mice and IL-6RaL-KO mice from 3 to

19 weeks of age. However, these experiments revealed no dif-

ference in body weight of IL-6RaL-KO mice when compared to

controls (Figure 2A). Similarly, body fat content and epigonadal

fat pad mass remained unchanged in IL-6RaL-KO compared to

control mice (Figure 2B). Moreover, as an indirect measure of

fat content, serum leptin concentrations were comparable

between control and IL-6RaL-KO mice (Figure 2C). In line with

these observations, IL-6RaL-KO mice ingested similar amounts

of food as compared to controls (Figure 2D). Furthermore, O2

consumption and CO2 production were not altered by hepato-

cyte-specific disruption of IL-6Ra, as assessed by indirect calo-

rimetrie (Figures 2E and 2F). Moreover, the respiratory exchange

ratio was comparable between control and IL-6RaL-KO mice

(Figure 2G). Finally, IL-6RaL-KO mice exhibited no detectable

alterations in locomotor activity (Figure 2H). Taken together,

hepatocyte-specific disruption of IL-6Ra signaling does not

result in alterations of body weight, adiposity, and energy

homeostasis.

Impaired Glucose Homeostasis in IL-6RaL-KO MiceNext, we assessed key parameters of glucose homeostasis in

control and IL-6RaL-KO mice. This analysis revealed unaltered

blood glucose concentrations in IL-6RaL-KO mice compared

to controls (Figure 3A). Strikingly, however, when we subjected

IL-6RaL-KO mice to glucose tolerance tests (GTT), these mice

exhibited significantly impaired glucose tolerance compared

to their control littermates (Figure 3B). To distinguish whether

impaired glucose tolerance in IL-6RaL-KO mice arises from

decreased glucose-stimulated insulin secretion or as a result

of insulin resistance, we next measured serum insulin concen-

trations during the GTT (Figure 3C). However, glucose-stimu-

lated insulin secretion was not significantly altered in IL-6RaL-KO

mice compared to controls. Also, pancreatic islet mass was

r Inc.

Figure 1. Generation of IL-6RaL-KO Mice(A) Targeting strategy for the generation of the conditional IL-6Ra allele. In Southern blot analysis on EcoRI (RI)-digested DNA using probe A, theWT IL-6Ra gene

(WT) comprising exons 2, 3, 4, 5, and 6 produces an 11.5 kb band. The IL-6Ra targeting vector (TV) was generated by inserting the short arm of homology (SAH),

the loxP-flanked exons 2 and 3, and the long arm of homology (LAH) into the GK12TK plasmid using common cloning techniques. Homologous recombination

creates the FLneo allele, which produces an 8.2 kb fragment in Southern blot analysis using probe A on RI-digested DNA. Flp-mediated recombination of the

FRT-flanked neo seletion marker generates the 6.3 kb FL allele, whereas Cre-mediated recombination of loxP-flanked exons 2 and 3 results in the 4.3 kb D allele

using probe B in Southern blot analysis of BglI-digested DNA.

(B) A9 and C3 ES cell clones were identified and verified by Southern blot using probe A.

(C) Southern blot analysis of genomic DNA isolated from liver (L), white adipose tissue (W), skeletal muscle (M), and tail (T) of WT and IL-6RaL-KO mice.

(D) WT and IL-6RaL-KO mice were injected with 50 ng IL-6 into the vena cava. Subsequently, livers were dissected and subjected to western blot analysis using

P-Stat-3 and Stat-3 antibodies, respectively.

(E) Relative expression of IL-6Ra mRNA in livers of WT and IL-6RaL-KO mice was determined by real-time PCR. n = 6/genotype.

(F) Serum concentrations of sIL-6Ra at week 8 were examined by ELISA in control and IL-6RaL-KO mice. n = 10/genotype.

(G) Liver weight of WT and IL-6RaL-KO mice was determined at week 20. n = 15/genotype.

(H) Cryostat sections of livers from WT and IL-6RaL-KO mice were stained with hematoxylin and eosin to analyze overall liver morphology.

Displayed values are means ± SEM. *p % 0.05, **p % 0.01, and ***p % 0.001 in Student’s t test.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

similar in the two groups of mice (Figures 3D and 3E). To further

address b cell function in these mice, we next examined

glucose-stimulated insulin secretion in isolated islets from

control and IL-6RaL-KO mice in vitro (Figure 3F). Of interest, iso-

lated islets of IL-6RaL-KO mice incubated in medium supple-

Cell Me

mented with low glucose exhibited an increased basal insulin

secretion compared to control islets, whereas the relative ability

of higher glucose concentrations to promote insulin secretion

was significantly reduced in islets from IL-6RaL-KO mice

(Figure 3F).

tabolism 12, 237–249, September 8, 2010 ª2010 Elsevier Inc. 239

Figure 2. IL-6RaL-KO Mice Show Unaltered

Adiposity and Energy Expenditure

(A) Body weight of control and IL-6RaL-KO mice was deter-

mined weekly from 3 to 19 weeks of age. n = 12–15/

genotype.

(B) Epigonadal fat pad weight and total body fat content of

WT and IL-6RaL-KO mice. Body fat was examined using

nuclear magnetic resonance. n = 12–15/genotype.

(C) Leptin concentrations at 8 and 16 weeks of control and

IL-6RaL-KO mice were determined by ELISA. n = 10/

genotype.

(D) Daily food intake of control and IL-6RaL-KO mice was

measured for 48 hr at 10 weeks of age. n = 8/genotype.

(E–H) (E) VO2 consumption and (F) VCO2 production, (G)

respiratory exchange ratio, and (H) locomotor activity

were examined in control and IL-6RaL-KO mice using indi-

rect calorimetric cages at 10 weeks of age. n = 8/

genotype.

Displayed values are means ± SEM. *p % 0.05,

**p % 0.01, and ***p % 0.001 versus control.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

Moreover, consistent with the impairment in glucose toler-

ance, insulin’s ability to lower blood glucose concentrations

during an insulin tolerance test (ITT) was significantly blunted

in IL-6RaL-KO compared to control mice, clearly indicating the

presence of insulin resistance in these mice (Figure 3G). Thus,

disruption of IL-6 signaling specifically in hepatocytes leads to

the development of insulin resistance and, in parallel, impairs

the ability of pancreatic b cells to mount an appropriate compen-

satory insulin secretory response, leading to impaired glucose

tolerance in the absence of alterations in body weight and

adiposity.

Activation of Hepatic Glycogen Synthesisand Glycogenolysis in IL-6RaL-KO MiceTo further define the tissue-specific alterations of glucose

homeostasis in IL-6RaL-KO mice, we performed euglycemic

hyperinsulinemic clamp studies in these animals. During these

experiments, comparable glycemia was obtained in control and

IL-6RaL-KO mice (Figure 4A). However, the glucose infusion

rate, necessary to maintain euglycemia, was dramatically

240 Cell Metabolism 12, 237–249, September 8, 2010 ª2010 Elsevie

s

s

e

h

m

w

tr

p

s

a

fi

m

B

a

to

ti

u

in

e

in

r Inc.

reduced in IL-6RaL-KO compared to controlmice

(Figure 4B), consistent with insulin resistance in

mice with liver-parenchymal disruption of IL-6

signaling as already detected during the ITTs.

Thus, whole-body glucose disposal during the

clamp was significantly reduced in the steady

state in IL-6RaL-KO mice (Figure 4C). When we

next assessed hepatic glucose production

(HGP) in control and IL-6RaL-KO mice, the rate

of HGP did not significantly differ between

the two groups of mice, neither during the basal

fasting condition nor during the hyperinsuline-

mic steady state at the end of the clamp

(Figure 4D). Because expression of key enzymes

of gluconeogenesis had been reported to

depend on hepatic IL-6 signaling (Inoue et al.,

2006), we next investigated expression of

G6Pase mRNA in liver under basal and hyperin-

ulinemic steady-state conditions. Consistent with a comparable

uppression of HGP in control and IL-6RaL-KO mice, relative

xpression of G6Pase was efficiently suppressed during the

yperinsulinemic clamp period in both control and IL-6RaL-KO

ice (Figure 4E). Nevertheless, steady-state G6Pase expression

as significantly higher in IL-6RaL-KO mice compared to con-

ols (Figure 4E), indicating that complete insulin-induced sup-

ression of G6Pase expression indeed requires functional IL-6

ignaling in liver-parenchymal cells. Consistently, western blot

nalyses of liver lysates from control and IL-6RaL-KO mice con-

rmed elevated expression of G6Pase in liver of IL-6RaL-KO

ice (Figure 4F). In line with this finding, IL-6 injection in C57/

L6 control mice reduced G6Pase expression significantly 2

nd 4 hr after injection (Figure S1A available online). Taken

gether, insulin’s ability to suppress hepatic glucose produc-

on—at least at the chosen insulin concentrations—remained

naltered in IL-6RaL-KO mice compared to controls, whereas

sulin failed to completely suppress G6Pase mRNA and protein

xpression in these animals, indicating the presence of a partial

sulin resistance in liver of these animals.

Figure 3. IL-6RaL-KO Mice Show Impaired

Insulin Sensitivity

(A) Blood glucose concentrations of fasted and

random fed control and IL-6RaL-KO mice. n = 9–13/

genotype.

(B) Glucose tolerance test of control and IL-6RaL-KO

mice was performed at 11 weeks of age. n = 9–13/

genotype.

(C) Serum insulin concentrations during the glucose

tolerance tests were examined by ELISA from control

and IL-6RaL-KO mice. n = 9–13/genotype.

(D) Representative H&E-stained section of pancreata

from control and IL-6RaL-KO mice to visualize pancre-

atic islets.

(E) Quantitation of average pancreatic islet mass was

determined on serial sections of pancreata isolated

from control and IL-6RaL-KO mice. n = 8/genotype.

(F) Static insulin secretion in islets from IL-6RaL-KO

mice (KO) compared to controls (WT). We incubated

6–9 islets at low (3 mM) or high glucose (11 mM) for

30 min. Data are scatter plot for insulin secretion

(ng/ng DNA) in 4 groups of batch incubations: WT

low glucose, 6 batch incubations from 3 mice; WT

high glucose, 10 batch incubations from 5 mice; KO

low glucose, 8 batch incubations from 5 mice; KO

high glucose, 13 batch incubations from 5 mice.

Figure on the right shows mean ± SEM for fold change

of insulin secretion after stimulation with high glucose.

(G) Insulin tolerance test of control and IL-6RaL-KO

mice was performed at 12 weeks of age. n = 9–13/

genotype.

Displayed values are means ± SEM. *p % 0.05,

**p % 0.01, and ***p % 0.001 versus control.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

Hepatic glucose production is counteracted by hepatic

glycogen storage. A key step in this process is phosphorylation

of glucose by glucokinase. Surprisingly, whereas insulin induced

glucokinase expression in control mice by 4-fold, this response

was enhanced to an 8-fold stimulation in livers of IL-6RaL-KO

mice (Figure 4G). Consistent with exaggerated glucokinase

gene expression in the liver of IL-6RaL-KO mice, IL-6 injection

acutely suppressed glucokinase expression in control mice

(Figure S1B). To investigate the effect of accelerated glucoki-

nase expression and impaired G6Pase suppression during the

clamp in IL-6RaL-KO mice, we compared glycogen concentra-

tions in liver of control and IL-6RaL-KO mice in the fasted state

as well as at the end of the steady-state phase during hyperinsu-

linemic euglycemic clamps (Figure 4H). After an overnight

fast, hepatic glycogen content was similar in liver of control

and IL-6RaL-KO mice (Figure 4H). However, whereas hepatic

glycogen stores remained unaltered in control mice in the steady

state compared to basal, hepatic glycogen content was sig-

nificantly reduced to 32 mg/g in IL-6RaL-KO mice, pointing to

impaired insulin-induced suppression of glycogenolysis in the

Cell Metabolism 12, 237–24

abscence of hepatic IL-6 signaling (Fig-

ure 4H). We also determined the hepatic

glycogen synthesis rate in the steady state

of control and IL-6RaL-KO mice bymeasuring3H-glucose incorporation into glycogen.

Importantly, this analysis revealed increased

glycogen synthesis in IL-6RaL-KO mice (Figure 4I). Thus, our

experiments provide direct evidence for partly counteracting

actions of IL-6 on glucose metabolism in liver by simultaneously

suppressing enzymes required for gluconeogenesis, glycogen-

olysis, and glycogen synthesis. Nevertheless, as a consequence

of this, the overall effect of IL-6 disruption on hepatic glucose

metabolism appears largely balanced.

IL-6RaL-KO Mice Exhibit Insulin Resistance in SkeletalMuscle and WATGiven the overall minor alterations of hepatic glucose metabo-

lism in the presence of hepatic IL-6 resistance, assessment

of insulin-stimulated AKT phosphorylation during the clamp

revealed the presence of partial hepatic insulin resistance in

the IL-6RaL-KO compared to control mice also at a molecular

level (Figure 5A). Strikingly, insulin resistance in IL-6RaL-KO

micewas not only observed locally in the liver, but also in skeletal

muscle and WAT, where insulin-stimulated glucose uptake

was significantly reduced by more than 50% compared to

controls, further supporting the development of systemic insulin

9, September 8, 2010 ª2010 Elsevier Inc. 241

Figure 4. IL-6RaL-KO Mice Exhibit Largely

Unaltered Hepatic Glucose Production

(A) Blood glucose levels during euglyemic hyperin-

sulinemic clamp studies of WT and IL-6RaL-KO

mice. n = 12/genotype.

(B) Glucose infusion rate to obtain steady-state

blood glucose levels in euglycemic hyperinsuline-

mic clamp studies. n = 12/genotype.

(C) Glucose disposal rate of basal and steady state

was calculated as described in the Experimental

Procedures. n = 12/genotype.

(D) Hepatic glucose production of basal and

steady state was calculated as described in the

Experimental Procedures. n = 12/genotype.

(E) Relative expression of G6Pase in liver of fasted

(basal) and steady-state (clamp) control and

IL-6RaL-KO mice. n = 6/genotype.

(F) Western blot of liver lysates from control and

IL-6RaL-KO mice under basal or steady-state con-

ditions using G6Pase and AKT antibodies.

(G) Relative expression of glucokinase in livers of

the indicated mice during basal and steady-state

conditions using real-time PCR. n = 6/genotype.

(H) Glycogen content was measured from KOH-

digested livers of control and IL-6RaL-KO mice

under basal or steady-state conditions using

colorimetric glucose determination after enzy-

matic breakdown of glycogen into glucose.

n = 12/genotype.

(I) Hepatic glycogen synthesis rate of control and

IL-6RaL-KO mice was examined by dividing the3H-glucose incooperation into glycogen to the

SA during the time of the clamp. n = 12/genotype.

Displayed values are means ± SEM. *p % 0.05,

**p % 0.01, and ***p % 0.001 versus control.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

resistance in IL-6RaL-KO mice despite the fact that IL-6 signaling

was exclusively inactivated in hepatocytes of these mice (Fig-

ure 5B). Consistent with largely attenuated insulin-stimulated

glucose transport in skeletal muscle of IL-6RaL-KO mice, insulin

almost completely failed to stimulate AKT phosphorylation in

skeletal muscle of these animals (Figures 5C and 5D). Taken

together, hepatic IL-6 resistance causes a minor degree of

insulin resistance in liver but results in a more pronounced

impairment of insulin action in skeletal muscle and WAT.

Exaggerated Inflammatory Action in IL-6RaL-KO MiceTo further define the molecular basis of insulin resistance both in

liver and in skeletal muscle of IL-6RaL-KO mice, while assuming

that inflammatory signaling plays a central role in the induction

of insulin resistance, we next compared the expression of

inflammatory cytokines in liver of control and IL-6RaL-KO mice.

Whereas basal hepatic expression of IL-6, TNF-a, and IL-10

did not differ between the two groups of mice, the insulin-

induced expression of these genes was profoundly increased

in liver of IL-6RaL-KO mice (Figure 6A). Consistent with increased

242 Cell Metabolism 12, 237–249, September 8, 2010 ª2010 Elsevie

ti

te

p

o

ti

s

e

in

IK

T

th

a

m

la

IL

d

th

a

r Inc.

expression of proinflammatory cytokines,

also phosphorylation of inhibitor of NFkB

(IkB) a was clearly enhanced in liver of

IL-6RaL-KO mice compared to controls

(Figures 6B and 6C). Of interest, activa-

on of Stat-3 phosphorylation during the clamp remained unal-

red in the liver of IL-6RaL-KO mice despite the lack of liver-

arenchymal IL-6 signaling, indicating redundant mechanisms

f Stat-3 phosphorylation in response to in vivo insulin stimula-

on such as IL-10 signaling (Figures 6B and 6C).

Strikingly, besides the activation of inflammatory gene expres-

ion and IKK signaling in liver of IL-6RaL-KO mice, also insulin-

voked expression of inflammatory genes was largely enhanced

skeletal muscle of these animals, concomitant with increased

K activation (Figures 6D–6F). Moreover, expression of IL-6,

NF-a, and IL-10 was also significantly increased in WAT during

e steady-state compared to basal conditions of both control

nd IL-6RaL-KO mice (Figure 6G).

Ultimately, to address whether these local changes in inflam-

atory gene expression also translate into alterations of circu-

ting cytokine concentrations, we quantified the amount of

-6, TNF-a, and IL-10 in sera of control and IL-6RaL-KO mice

uring basal and steady-state conditions (Figure 6H). Whereas

e abundance of circulating IL-6, TNF-a, and IL-10 was similar

t the basal state in both groups of mice, strikingly, IL-6

Figure 5. IL-6RaL-KO Mice Exhibit Insulin

Resistantance in Liver, Skeletal Muscle

and WAT

(A) Western blot of liver lysates from control and

IL-6RaL-KO mice under basal and steady-state

conditions in euglycemic hyperinsulinemic clamps

using P-AKT and AKT antibodies.

(B) 14C glucose uptake in euglycemic hyperinsuli-

nemic clamps of control and IL-6RaL-KO mice.

n = 12/genotype.

(C) Western blot of skeletal muscle lysates from

control and IL-6RaL-KO mice under basal and

steady-state conditions using P-AKT and AKT

antibodies.

(D) Quantification of insulin-stimulated AKT phos-

phorylation in liver and muscle under basal and

steady-state conditions using lysates from the

indicated mice. n = 6/genotype.

Displayed values are means ± SEM. *p % 0.05,

**p % 0.01, and ***p % 0.001 versus control.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

concentrations in serum of control mice did rise significantly

under steady-state conditions compared to the fasted state.

However, IL-6 concentrations in serum of IL-6RaL-KO mice

increased even more during the steady state (Figure 6H). Fur-

thermore, during euglycemic hyperinsulinemic clamps, serum

TNF-a levels were 2.5-fold increased in IL-6RaL-KO mice com-

pared to controls (Figure 6H). Also, IL-10 levels increased during

the clamp but to similar extents in control and IL-6RaL-KO mice

(Figure 6H). Taken together, disruption of hepatic IL-6 signaling

exaggerates not only inflammatory gene expression and sig-

naling in liver, but also in skeletal muscle to cause increased

circulating concentrations of IL-6 and TNF-a, thus potentially

promoting systemic insulin resistance.

Attenuation of Inflammation Normalizes GlucoseMetabolism in IL-6RaL-KO MiceTo directly address whether the increased inflammatory tone,

specifically the increase in circulating TNF-a concentrations in

the presence of liver-parenchymal IL-6Ra deficiency, contrib-

utes to the development of systemic insulin resistance and

impaired glucose tolerance in vivo, we analyzed the effect of

neutralizing circulating TNF-a in control and in IL-6RaL-KO

mice. To this end, we administered an anti-TNF-a antiserum

that was previously described for its potency to diminish

TNF-a signaling in mice (Dejager and Libert, 2008). Prior to

anti-TNFa treatment, IL-6RaL-KO mice demonstrated impaired

glucose tolerance compared to control mice to a similar degree

as described above (Figures 3B and 7A). Strikingly, depletion of

TNF-a resulted in a normalization of glucose clearance upon

intraperitoneal injection of a glucose bolus in IL-6RaL-KO mice,

whereas, in control mice, anti-TNF-a treatment had no signifi-

cant effect on glucose tolerance (Figure 7B). These experiments

directly demonstrate that elevation of circulating TNF-a concen-

trations contributes to the impairment of glucose metabolism in

these mice.

Next, we aimed to address whether liver-resident macro-

phages, i.e., Kupffer cells, are responsible for the exaggerated

Cell Me

release of TNF-a, thereby causing development of systemic

insulin resistance in IL-6RaL-KO mice. Therefore, we assessed

the effect of clodronate liposome administration, which effi-

ciently depletes Kupffer cells (Van Rooijen and Sanders, 1996),

on glucose metabolism in control and IL-6RaL-KO mice. Prior to

clodronate liposome treatment, livers of control and IL-6Ra-

deficient mice exhibited substantial immunoreactivity for Mac-2

as a marker of liver-resident macrophages (Figure 7C). In con-

trast, clodronate liposome administration virtually abolished

Mac-2 immunoreactivity 2 days after injection in liver of both

control and IL-6RaL-KO mice. (Figure 7D). Again consistent with

our previous experiments, IL-6RaL-KO mice showed impaired

glucose tolerance prior to clodronate liposome-induced deple-

tion of Kupffer cells (Figure 7E). Strikingly, Kupffer cell depletion

selectively improved glucose tolerance in IL-6RaL-KO mice to

control values, whereas no alteration in glucose clearance was

observed in control mice following clodronate liposome treat-

ment (Figure 7F). These experiments directly demonstrate that

activation of liver-resident macrophages causes insulin resis-

tance in IL-6RaL-KO mice.

Interestingly, depletion of Kupffer cells also enhanced

glucose-induced insulin concentrations in IL-6RaL-KO mice.

Whereas glucose-stimulated insulin secretion was not sig-

nificantly affected by Kupffer cell depletion in control mice,

particularly the early phase of glucose-stimulated insulin secre-

tion was significantly elevated in IL-6RaL-KO mice following

clodronate liposome treatment (Figures 7G and 7H). These

experiments indicate that, potentially, the elevated systemic

inflammatory tone in these mice not only causes peripheral

insulin resistance, but at the same time also prevents an appro-

priate compensatory increase in insulin secretion. In summary,

the present study reveals a crosstalk between hepatocytes

and liver-resident Kupffer cells in control of inflammatory cyto-

kine expression. Our results clearly demonstrate that IL-6

signaling in hepatocytes is essential to downmodulate TNF-a

expression in Kupffer cells, whose desinhibition results in the

development of systemic insulin resistance and potentially also

modulates pancreatic b cell function.

tabolism 12, 237–249, September 8, 2010 ª2010 Elsevier Inc. 243

Figure 6. Exaggerated Inflammation in IL-6RaL-KO Mice

(A) Relative expression of IL-6, TNF-a, and IL-10 in liver of control and IL-6RaL-KO mice under basal and steady-state conditions determined by real-time PCR.

n = 6/genotype.

(B) Representatitive western blot analysis of liver lysates from control and IL-6RaL-KO mice under basal and steady-state conditions using P-Stat-3, P-IkBa, IkBa,

and AKT antibodies.

(C) Quantification of hepatic Stat-3 and IkBa phosphorylation under basal and steady-state conditions relative to AKT and IkBa expression, using Quantity

One software. n = 6. Values obtained from mice in basal conditions were set to 100%.

(D) Relative expression of IL-6, TNF-a, and IL-10 in skeletal muscle under basal and steady-state conditions of control and IL-6RaL-KO mice determined by real-

time PCR. n = 6/genotype.

(E) Representatitve western blot analysis of muscle lysates from control and IL-6RaL-KO mice under basal and steady-state conditions using P-Stat-3, P-IkBa,

IkBa, and AKT antibodies.

(F) Quantification of skeletal muscle Stat-3 and IkBa phosphorylation under basal and steady-state conditions relative to AKT and IkBa expression, usingQuantity

One software. n = 6. Values obtained from mice in basal conditions were set to 100%.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

244 Cell Metabolism 12, 237–249, September 8, 2010 ª2010 Elsevier Inc.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

DISCUSSION

Obesity-associated insulin resistance correlates with increased

concentrations of inflammatory cytokines in circulation, both in

obese rodent models and humans (Hotamisligil et al., 1997;

Olszanecka-Glinianowicz et al., 2004). Whereas increased

TNF-a expression inhibits insulin signaling and thus is a major

causal contributing factor in the development of insulin resis-

tance (Hotamisligil et al., 1994a, 1994b), the role of elevated IL-6

concentrations in obesity-induced insulin resistance is still

matter of debate (Pedersen and Febbraio, 2007). Whereas IL-6

levels are high during exercise (Pedersen et al., 2004), a situation

in which glucose uptake is critical for the energy supply of skel-

etal muscle, chronically increased IL-6 levels during the course

of obesity induce SOCS3 expression in liver, which in turn can

impair insulin signaling (Sabio et al., 2008; Ueki et al., 2004).

Besides the known effect of exercise to increase IL-6 concen-

trations (Penkowa et al., 2003), we demonstrate in the current

study that, during euglycemic hyperinsulinemic clamps, IL-6

expression is increased in liver, skeletal muscle, and WAT,

resulting in increased IL-6 concentrations in circulation. Of

interest, during the clamp, not only IL-6 expression is increased,

but also that of TNF-a and IL-10. Consistent with increased cyto-

kine expression, we could also detect activation of downstream

signaling events of these signals such as phosphorylation of IkBa

and Stat-3 in liver and muscle under euglycemic hyperinsuline-

mic clamp conditions. Thus, euglycemic hyperinsulinemic clamp

conditions seem to promote a proinflammatory tone.

The notion that endogenous IL-6 concentrations increase

during euglycemic hyperinsulinemic clamps may also explain

theminor effects observed when IL-6 was infused during clamps

to investigate the acute effect of IL-6 on glucose homeostasis

(Carey et al., 2006; Kim et al., 2004; Yuen et al., 2009).

Importantly, our study reveals an overall anti-inflammatory role

for IL-6 signaling in the liver. This is consistent with a study in

which IL-6 was shown to act as negative feedback regulator of

inflammatory signaling pathways in macrophages (Xing et al.,

1998). Here, the authors examined IL-6-deficient macrophages

and found that, in response to lipopolysaccharide stimulation,

they exhibit increased expression of TNF-a, indicating that IL-6

can function rather anti- than proinflammatory to downregulate

inflammatory signaling cascades. However, in our model, abro-

gation of IL-6 signaling in liver-parenchymal cells promotes an

activation of inflammatory cytokine expression in liver-resident

macrophages, i.e., Kupffer cells, indicating a crosstalk of an

IL-6-dependent signal in hepatocytes to the neighboring Kupffer

cells to limit inflammation. Whatever the nature of this signal is,

the functional significance of Kupffer cell activation in the mani-

festation of the phenotype on the IL-6RaL-KO mice is clearly

evident from the fact that Kupffer cell ablation improves impaired

glucose metabolism in these mice.

Of interest, Kupffer cell activation in the IL-6RaL-KO mice not

only causes a local inflammatory response in liver, but is also par-

(G) Relative expression of IL-6, TNF-a, and IL-10 in WAT under basal and steady-

n = 6/genotype.

(H) Serum concentrations of IL-6, TNF-a, and IL-10 under basal and steady-state c

array. n = 10/genotype.

Displayed values are means ± SEM. *p % 0.05 and **p % 0.01 versus control.

Cell Me

alleled by increased expression of IL-6, TNF-a, and IL-10 gene

expression in skeletal muscle andWAT, presumably as a conse-

quence of increased Kupffer cell-derived TNF-a. In line with this

notion, abrogation of TNF-a action completely reverses the

impaired glucose tolerance in the IL-6RaL-KO animals. However,

we cannot exclude that increased systemic IL-6 concentrations

in thesemice and subsequent activation of IL-6 signaling in other

organs than liver is responsible for the development of insulin

resistance, as only hepatocytes are IL-6 resistant in our model.

Nevertheless, besides the dramatic effect of TNF-a neutraliza-

tion, also unaltered Stat-3 phosphorylation in skeletal muscle of

IL-6RaL-KO mice argues against a direct effect of increased IL-6

action on skeletal muscle to cause insulin resistance.

Besides the effect of systemic inflammation to cause insulin

resistance, our experiments also indicate that activation of proin-

flammatory signaling can simultaneously impair b cell function,

consistent with the notion that inflammatory cytokines, particu-

larly TNF-a, can inhibit insulin release from the b cells (Southern

et al., 1990; Zhang and Kim, 1995).

In addition to the insights into the role of IL-6 action in liver to

control local and systemic inflammation, the present study

moreover revealed the complex effect of this cytokine on liver

glucose metabolism. Recently, insulin signaling in the CNS was

shown to inhibit hepatic glucose production via a mechanism

including stimulation of IL-6 expression from Kupffer cells, in

turn acting on hepatocytes to downregulate G6Pase expression

and to reduce hepatic glucose output (Inoue et al., 2006). Here,

we confirm the inhibitory action of IL-6 on G6Pase expression,

as G6Pase mRNA and protein expression are significantly

increased during hyperinsulinemic euglycemic clamps in liver

of IL-6RaL-KO mice compared to controls. However, G6Pase

expression in these animals is still largely suppressed compared

to the fasted state. Together with unaltered Stat-3 phosphoryla-

tion during the clamps in IL-6RaL-KO mice, our results thus indi-

cate the existence of additional and/or compensatory IL-6-inde-

pendent pathways responsible for insulin-induced Stat-3

phosphorylation in liver of these mice.

Strikingly, IL-6 not only controls gluconeogenesis, but also

negatively regulates glucokinase expression, thus promoting

breakdown of glycogen. Thus, lack of IL-6 signaling promotes

glycogen synthesis, as directly demonstrated under clamp con-

ditions, resulting in only a minor net effect on hepatic glucose

output in IL-6RaL-KO mice. Thus, we provide in vivo evidence

for partly antagonizing actions of IL-6 in hepatic glucose metab-

olism by simultaneously suppressing enzymes essential for

gluconeogenesis, glycogenolysis, and glycogen synthesis.

Given these complex effects of IL-6 signaling on hepatic

metabolism and the initiation of a systemic inflammatory

response upon inhibition of IL-6 signaling, caution should be

warranted to potential diabetogenic side effects of newly evolv-

ing therapies aiming to interfere with IL-6 signaling to treat

chronic inflammatory diseases, such as rheumatoid arthritis

(Choy et al., 2002; Nishimoto et al., 2004).

state conditions of control and IL-6RaL-KO mice determined by real-time PCR.

onditions in control and IL-6RaL-KOmice was examined using cytometric bead

tabolism 12, 237–249, September 8, 2010 ª2010 Elsevier Inc. 245

Figure 7. Depletion of TNF-a or Kupffer Cells Normalizes Glucose Homeostasis in IL-6RaL-KO Mice

(A) Glucose tolerance test of control and IL-6RaL-KO mice was performed at 9 weeks of age. n = 10/genotype.

(B) Glucose tolerance test of the same control and IL-6RaL-KO animals as in (A) was performed at 10 weeks of age after anti-TNF-a treatment. n = 10/genotype.

Animals received a 100 ml intraperitoneal injection of an anti-TNF-a antiserum 1 hr prior to the glucose challenge.

(C) Representative Mac-2 immunohistochemistry of liver sections of control and IL-6RaL-KO mice.

(D) Representative Mac-2 immunohistochemistry of liver sections of control and IL-6RaL-KO mice, which received a single 1 ml injection of clodronate liposomes

2 days before.

(E) Glucose tolerance test of control and IL-6RaL-KO mice was performed at 10 weeks of age. n = 10/genotype.

(F) Glucose tolerance test of the same control and IL-6RaL-KO animals as in (E) was performed at 11 weeks of age. n = 10/genotype. Two days before the GTT,

animals received a single 1 ml i.p. injection of clodronate liposomes to deplete the Kupffer cells.

(G) Glucose-induced insulin secretion during the GTT of clodronate liposome-treated control and IL-6RaL-KO animals as in (F). Serum insulin concentrations were

determined by ELISA.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

246 Cell Metabolism 12, 237–249, September 8, 2010 ª2010 Elsevier Inc.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

EXPERIMENTAL PROCEDURES

Animal Care

Care of all animals was within institutional animal care committee guidelines,

and all animal procedures were approved by local government authorities

(Bezirksregierung Koln, Cologne, Germany) and were in accordance with

NIH guidelines. Mice were housed in groups of 3–5 at 22�C–24�C using

a 12 hr light/12 hr dark cycle with lights on at 7 AM. Animals were fed regular

chow food (NCD; Teklad Global Rodent 2018; Harlan) containing 53.5%

carbohydrates, 18.5% protein, and 5.5% fat (12% of calories from fat).

Animals had ad libitum access to water at all times, and food was only with-

drawn when required for experimental procedures. At 20 weeks of age,

animals were sacrificed using CO2.

Body Weight and Body Composition

Body weight was measured weekly. Body fat content was measured in vivo by

nuclear magnetic resonance using the minispec mq 7.5 (Bruker).

Euglycemic Hyperinsulinemic Clamp Studies

Catheter implantation was performed as previously described (Fisher and

Kahn, 2003). Only mice that had regained at least 90% of their preoperative

bodyweight after 6 days of recovery were included in the experimental groups.

After starvation for 16 hr, awake animals were placed in restrainers for the

duration of the clamp experiment. After a D-[3-3H]Glucose (Amersham Biosci-

ences) tracer solution bolus infusion (5mCi), the tracer was infused continu-

ously (0.05 mCi/min) for the duration of the experiment. At the end of the

50 min basal period, a blood sample (50 ml) was collected for determination

of the basal parameters. Insulin (human regular insulin; NovoNordisc Pharma-

ceuticals) solution containing 0.1% BSA (Sigma) was infused at a fixed rate

(4 mU/g/min) following a bolus infusion (40 mU/g). Blood glucose levels were

determined every 10 min (B-Glucose Analyzer; Hemocue), and physiological

blood glucose levels (between 120 and 145 mg/dl) were maintained by adjust-

ing a 20% glucose infusion (DeltaSelect). Approximately 60 min before steady

state was achieved, a bolus of 2-Deoxy-D-[1-14C]Glucose (10 mCi, Amersham)

was infused. Steady state was ascertained when glucose measurements were

constant for at least 30 min at a fixed glucose infusion rate and was achieved

within 100 to 130 min. During the clamp experiment, blood samples (5 ml) were

collected after the infusion of the 2-Deoxy-D-[1-14C]Glucose at the time points

0, 5, 15, 25, 35min etc. until reaching the steady state. During the steady state,

blood samples (50 ml) for the measurement of steady-state parameters were

collected. All blood samples were kept at room temperature until centrifuga-

tion; serum samples were stored at �20�C.At the end of the experiment, mice were killed by cervical dislocation, and

brain, liver, white adipose tissue (WAT), and skeletal muscle were dissected

and stored at �80�C.Plasma [3-3H] Glucose radioactivity of basal and steady statewasmeasured

as previously described (Fisher and Kahn, 2003). Plasma Deoxy-[1-14C]

Glucose radioactivity was directly measured in the liquid scintillation counter.

WAT and skeletal muscle lysates were processed through ion exchange

chromatography columns (Poly-PrepR Prefilled Chromatography Columns,

AGR1-X8 formate resin, 200–400 mesh dry; Bio Rad Laboratories) to separate

2-Deoxy-D-[1-14C]Glucose (2DG) from 2-Deoxy-D-[1-14C]Glucose-6-Phos-

phate (2DG6P).

Glucose turnover rate (mg 3 kg�1 3 min�1) was calculated as previously

described (Fisher and Kahn, 2003). In vivo glucose uptake forWAT and skeletal

muscle (nmol 3 g�1 3 min�1) was calculated based on the accumulation of

2DG6P in the respective tissue and the disappearance rate of 2DG from

plasma as described previously (Ferre et al., 1985).

Glucose and Insulin Tolerance Test

Glucose tolerance tests were performed with 16 hr fasted animals. After deter-

mination of fasted blood glucose levels, each animal received an intraperito-

neal injection of 20% glucose (10 ml/kg) (DeltaSelect). Blood glucose levels

were detected after 15, 30, 60, and 120 min. To deplete TNF-a in control

(H) Comparison of serum insulin concentrations of nontreated or clodronate liposo

challenge taken from Figures 3C and 7G.

Displayed values are means ± SEM. *p % 0.05 and **p % 0.01 versus control.

Cell Me

and mutant mice, mice received a single 100 ml i.p. injection of the anti-

TNF-a anti-serum (Dejager and Libert, 2008) 2 hr before the glucose tolerance

test. Kupffer cells were depleted by i.p. injection of 1 ml clodronate liposomes

(Van Rooijen and Sanders, 1996) in control and IL-6RaL-KO mice 2 days before

the glucose tolerance test. Clodronate was a gift from Roche.

Insulin tolerance tests were performed with fasted mice. Animals were in-

jected with 0.75 U/kg body weight of human regular insulin (Novo Nordisk)

into the peritoneal cavity. Blood glucose levels were detected after 0, 15, 30,

and 60 min.

Glucose-Induced Insulin Secretion of Isolated Pancreatic Islets

The medium used for both isolation and the experiments was a HEPES buffer

containing (in mM): 125 NaCl, 5.9 KCl, 1.2 MgCl2, 1.28 CaCl2, and 3 glucose

(pH 7.4). Bovine serum albumin was added to the medium at a concentration

of 1 mg/ml. Pancreatic islets were isolated by a collagenase technique (Lern-

mark, 1974), and the islets were maintained overnight in RPMI 1640 culture

medium supplemented with 100 ug/ml streptomycin, 100 units of penicillin,

2 mM glutamine, and 10% FCS at 37�C in a humidified atmosphere of 5%

CO2 in air. Islets were preincubated in buffer containing 3 mM glucose for

30 min at 37�C. Batches of 6–9 islets were incubated for 30 min at 37�Cwith 3 or 11 mM glucose. The incubation was stopped by chilling the samples

on ice, samples were taken for analysis of insulin, and the islets were collected

for measurement of dsDNA. Insulin was measured using the TPX plate reader

(ArcDia International Oy Ltd, Turku, Finland) (Hanninen et al., 2000), and Pico-

Green dsDNA quantification reagent kit (Molecular Probes) was used to quan-

tify dsDNA in the islets.

Statistical Methods

Data were analyzed for statistical significance using a two-tailed unpaired

Student’s t test.

SUPPLEMENTAL INFORMATION

Supplemental Information includes Supplemental Experimental Procedures

and one figure and can be found with this article online at doi:10.1016/

j.cmet.2010.06.011.

ACKNOWLEDGMENTS

We thank Gisela Schmall and Tanja Rayle for excellent secretarial assistance

and Gabriele Ringeisen, Brigitte Hampel, Anke Lietzau, and Sigrid Irlenbusch

for excellent technical assistance. Moreover, we cordially thank Gunther

Schutz for providing the Alfp-Cre transgenic animals. This work was supported

by grants from the CMMC, the Novartis Foundation (to J.C.B.), and the

German Research Foundation (DFG) (grants 1492-7-1 and SFB 635 to

J.C.B. and SFB 832 to J.C.B. and F.T.W.).

Received: September 14, 2009

Revised: April 7, 2010

Accepted: June 1, 2010

Published: September 7, 2010

REFERENCES

Aguirre, V., Uchida, T., Yenush, L., Davis, R., andWhite, M.F. (2000). The c-Jun

NH(2)-terminal kinase promotes insulin resistance during association with

insulin receptor substrate-1 and phosphorylation of Ser(307). J. Biol. Chem.

275, 9047–9054.

Arkan, M.C., Hevener, A.L., Greten, F.R., Maeda, S., Li, Z.W., Long, J.M.,

Wynshaw-Boris, A., Poli, G., Olefsky, J., and Karin, M. (2005). IKK-beta links

inflammation to obesity-induced insulin resistance. Nat. Med. 11, 191–198.

Bastard, J.P., Jardel, C., Bruckert, E., Blondy, P., Capeau, J., Laville, M., Vidal,

H., and Hainque, B. (2000). Elevated levels of interleukin 6 are reduced in

me-treated control and IL-6RaL-KO animals 0, 15, and 30 min after the glucose

tabolism 12, 237–249, September 8, 2010 ª2010 Elsevier Inc. 247

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

serum and subcutaneous adipose tissue of obese women after weight loss. J.

Clin. Endocrinol. Metab. 85, 3338–3342.

Cai, D., Yuan, M., Frantz, D.F., Melendez, P.A., Hansen, L., Lee, J., and Shoe-

lson, S.E. (2005). Local and systemic insulin resistance resulting from hepatic

activation of IKK-beta and NF-kappaB. Nat. Med. 11, 183–190.

Carey, A.L., Steinberg, G.R., Macaulay, S.L., Thomas, W.G., Holmes, A.G.,

Ramm, G., Prelovsek, O., Hohnen-Behrens, C., Watt, M.J., James, D.E.,

et al. (2006). Interleukin-6 increases insulin-stimulated glucose disposal in hu-

mans and glucose uptake and fatty acid oxidation in vitro via AMP-activated

protein kinase. Diabetes 55, 2688–2697.

Chida, D., Osaka, T., Hashimoto, O., and Iwakura, Y. (2006). Combined inter-

leukin-6 and interleukin-1 deficiency causes obesity in young mice. Diabetes

55, 971–977.

Choy, E.H., Isenberg, D.A., Garrood, T., Farrow, S., Ioannou, Y., Bird, H.,

Cheung, N., Williams, B., Hazleman, B., Price, R., et al. (2002). Therapeutic

benefit of blocking interleukin-6 activity with an anti-interleukin-6 receptor

monoclonal antibody in rheumatoid arthritis: a randomized, double-blind,

placebo-controlled, dose-escalation trial. Arthritis Rheum. 46, 3143–3150.

Dejager, L., and Libert, C. (2008). Tumor necrosis factor alpha mediates the

lethal hepatotoxic effects of poly(I:C) in D-galactosamine-sensitized mice.

Cytokine 42, 55–61.

Di Gregorio, G.B., Hensley, L., Lu, T., Ranganathan, G., and Kern, P.A. (2004).

Lipid and carbohydrate metabolism inmice with a targetedmutation in the IL-6

gene: absence of development of age-related obesity. Am. J. Physiol. Endocri-

nol. Metab. 287, E182–E187.

Ferre, P., Leturque, A., Burnol, A.F., Penicaud, L., and Girard, J. (1985).

A method to quantify glucose utilization in vivo in skeletal muscle and white

adipose tissue of the anaesthetized rat. Biochem. J. 228, 103–110.

Fisher, S.J., and Kahn, C.R. (2003). Insulin signaling is required for insulin’s

direct and indirect action on hepatic glucose production. J. Clin. Invest. 111,

463–468.

Gropp, E., Shanabrough, M., Borok, E., Xu, A.W., Janoschek, R., Buch, T.,

Plum, L., Balthasar, N., Hampel, B., Waisman, A., et al. (2005). Agouti-related

peptide-expressing neurons are mandatory for feeding. Nat. Neurosci. 8,

1289–1291.

Hanninen, P., Soini, A., Meltola, N., Soini, J., Soukka, J., and Soini, E. (2000).

A new microvolume technique for bioaffinity assays using two-photon excita-

tion. Nat. Biotechnol. 18, 548–550.

Hirosumi, J., Tuncman, G., Chang, L., Gorgun, C.Z., Uysal, K.T., Maeda, K.,

Karin, M., and Hotamisligil, G.S. (2002). A central role for JNK in obesity and

insulin resistance. Nature 420, 333–336.

Holmes, A.G., Mesa, J.L., Neill, B.A., Chung, J., Carey, A.L., Steinberg, G.R.,

Kemp, B.E., Southgate, R.J., Lancaster, G.I., Bruce, C.R., et al. (2008). Pro-

longed interleukin-6 administration enhances glucose tolerance and increases

skeletal muscle PPARalpha and UCP2 expression in rats. J. Endocrinol. 198,

367–374.

Hotamisligil, G.S., Shargill, N.S., and Spiegelman, B.M. (1993). Adipose

expression of tumor necrosis factor-alpha: direct role in obesity-linked insulin

resistance. Science 259, 87–91.

Hotamisligil, G.S., Budavari, A., Murray, D., and Spiegelman, B.M. (1994a).

Reduced tyrosine kinase activity of the insulin receptor in obesity-diabetes.

Central role of tumor necrosis factor-alpha. J. Clin. Invest. 94, 1543–1549.

Hotamisligil, G.S., Murray, D.L., Choy, L.N., and Spiegelman, B.M. (1994b).

Tumor necrosis factor alpha inhibits signaling from the insulin receptor.

Proc. Natl. Acad. Sci. USA 91, 4854–4858.

Hotamisligil, G.S., Arner, P., Atkinson, R.L., and Spiegelman, B.M. (1997).

Differential regulation of the p80 tumor necrosis factor receptor in human

obesity and insulin resistance. Diabetes 46, 451–455.

Inoue, H., Ogawa, W., Asakawa, A., Okamoto, Y., Nishizawa, A., Matsumoto,

M., Teshigawara, K., Matsuki, Y., Watanabe, E., Hiramatsu, R., et al. (2006).

Role of hepatic STAT3 in brain-insulin action on hepatic glucose production.

Cell Metab. 3, 267–275.

Kellendonk, C., Opherk, C., Anlag, K., Schutz, G., and Tronche, F. (2000).

Hepatocyte-specific expression of Cre recombinase. Genesis 26, 151–153.

248 Cell Metabolism 12, 237–249, September 8, 2010 ª2010 Elsevie

Kim, H.J., Higashimori, T., Park, S.Y., Choi, H., Dong, J., Kim, Y.J., Noh, H.L.,

Cho, Y.R., Cline, G., Kim, Y.B., and Kim, J.K. (2004). Differential effects of inter-

leukin-6 and -10 on skeletal muscle and liver insulin action in vivo. Diabetes 53,

1060–1067.

Kleinridders, A., Schenten, D., Konner, A.C., Belgardt, B.F., Mauer, J.,

Okamura, T., Wunderlich, F.T., Medzhitov, R., and Bruning, J.C. (2009).

MyD88 signaling in the CNS is required for development of fatty acid-induced

leptin resistance in diet-induced obesity. Cell Metab. 10, 249–259.

Klover, P.J., Zimmers, T.A., Koniaris, L.G., and Mooney, R.A. (2003). Chronic

exposure to interleukin-6 causes hepatic insulin resistance in mice. Diabetes

52, 2784–2789.

Konner, A.C., Janoschek, R., Plum, L., Jordan, S.D., Rother, E., Ma, X., Xu, C.,

Enriori, P., Hampel, B., Barsh, G.S., et al. (2007). Insulin action in AgRP-

expressing neurons is required for suppression of hepatic glucose production.

Cell Metab. 5, 438–449.

Lernmark, A. (1974). The preparation of, and studies on, free cell suspensions

from mouse pancreatic islets. Diabetologia 10, 431–438.

Mackiewicz, A., Wiznerowicz, M., Roeb, E., Karczewska, A., Nowak, J., Hein-

rich, P.C., and Rose-John, S. (1995). Soluble interleukin 6 receptor is biologi-

cally active in vivo. Cytokine 7, 142–149.

Nishimoto, N., Yoshizaki, K., Miyasaka, N., Yamamoto, K., Kawai, S., Takeu-

chi, T., Hashimoto, J., Azuma, J., and Kishimoto, T. (2004). Treatment of rheu-

matoid arthritis with humanized anti-interleukin-6 receptor antibody: a multi-

center, double-blind, placebo-controlled trial. Arthritis Rheum. 50, 1761–1769.

Nishimura, S., Manabe, I., Nagasaki, M., Eto, K., Yamashita, H., Ohsugi, M.,

Otsu, M., Hara, K., Ueki, K., Sugiura, S., et al. (2009). CD8+ effector T cells

contribute to macrophage recruitment and adipose tissue inflammation in

obesity. Nat. Med. 15, 914–920.

Olszanecka-Glinianowicz, M., Zahorska-Markiewicz, B., Janowska, J., and

Zurakowski, A. (2004). Serum concentrations of nitric oxide, tumor necrosis

factor (TNF)-alpha and TNF soluble receptors in women with overweight and

obesity. Metabolism 53, 1268–1273.

Pedersen, B.K., and Febbraio, M.A. (2007). Point: Interleukin-6 does have

a beneficial role in insulin sensitivity and glucose homeostasis. J. Appl. Physiol.

102, 814–816.

Pedersen, B.K., Steensberg, A., Fischer, C., Keller, C., Keller, P., Plomgaard,

P., Wolsk-Petersen, E., and Febbraio, M. (2004). The metabolic role of IL-6

produced during exercise: is IL-6 an exercise factor? Proc. Nutr. Soc. 63,

263–267.

Penkowa, M., Keller, C., Keller, P., Jauffred, S., and Pedersen, B.K. (2003).

Immunohistochemical detection of interleukin-6 in human skeletal muscle

fibers following exercise. FASEB J. 17, 2166–2168.

Plomgaard, P., Bouzakri, K., Krogh-Madsen, R., Mittendorfer, B., Zierath, J.R.,

and Pedersen, B.K. (2005). Tumor necrosis factor-alpha induces skeletal

muscle insulin resistance in healthy human subjects via inhibition of Akt

substrate 160 phosphorylation. Diabetes 54, 2939–2945.

Plum, L., Belgardt, B.F., and Bruning, J.C. (2006). Central insulin action in

energy and glucose homeostasis. J. Clin. Invest. 116, 1761–1766.

Rodrıguez, C.I., Buchholz, F., Galloway, J., Sequerra, R., Kasper, J., Ayala, R.,

Stewart, A.F., and Dymecki, S.M. (2000). High-efficiency deleter mice show

that FLPe is an alternative to Cre-loxP. Nat. Genet. 25, 139–140.

Sabio, G., Das, M., Mora, A., Zhang, Z., Jun, J.Y., Ko, H.J., Barrett, T., Kim,

J.K., and Davis, R.J. (2008). A stress signaling pathway in adipose tissue regu-

lates hepatic insulin resistance. Science 322, 1539–1543.

Sadagurski, M., Norquay, L., Farhang, J., D’Aquino, K., Copps, K., and White,

M.F. (2010). Human IL6 enhances leptin action in mice. Diabetologia 53,

525–535.

Southern, C., Schulster, D., and Green, I.C. (1990). Inhibition of insulin secre-

tion by interleukin-1 beta and tumour necrosis factor-alpha via an L-arginine-

dependent nitric oxide generating mechanism. FEBS Lett. 276, 42–44.

Ueki, K., Kondo, T., and Kahn, C.R. (2004). Suppressor of cytokine signaling 1

(SOCS-1) and SOCS-3 cause insulin resistance through inhibition of tyrosine

phosphorylation of insulin receptor substrate proteins by discrete mecha-

nisms. Mol. Cell. Biol. 24, 5434–5446.

r Inc.

Cell Metabolism

Hepatic IL-6 Signaling Improves Insulin Action

Van Rooijen, N., and Sanders, A. (1996). Kupffer cell depletion by liposome-

delivered drugs: comparative activity of intracellular clodronate, propamidine,

and ethylenediaminetetraacetic acid. Hepatology 23, 1239–1243.

Wallenius, V., Wallenius, K., Ahren, B., Rudling, M., Carlsten, H., Dickson, S.L.,

Ohlsson, C., and Jansson, J.O. (2002). Interleukin-6-deficient mice develop

mature-onset obesity. Nat. Med. 8, 75–79.

Xing, Z., Gauldie, J., Cox, G., Baumann, H., Jordana,M., Lei, X.F., and Achong,

M.K. (1998). IL-6 is an antiinflammatory cytokine required for controlling local

or systemic acute inflammatory responses. J. Clin. Invest. 101, 311–320.

Xu, H., Barnes, G.T., Yang, Q., Tan, G., Yang, D., Chou, C.J., Sole, J., Nichols,

A., Ross, J.S., Tartaglia, L.A., and Chen, H. (2003). Chronic inflammation in fat

Cell Me

plays a crucial role in the development of obesity-related insulin resistance.

J. Clin. Invest. 112, 1821–1830.

Yuen, D.Y., Dwyer, R.M., Matthews, V.B., Zhang, L., Drew, B.G., Neill, B.,

Kingwell, B.A., Clark, M.G., Rattigan, S., and Febbraio, M.A. (2009). Inter-

leukin-6 attenuates insulin-mediated increases in endothelial cell signaling

but augments skeletal muscle insulin action via differential effects on tumor

necrosis factor-alpha expression. Diabetes 58, 1086–1095.

Zhang, S., and Kim, K.H. (1995). TNF-alpha inhibits glucose-induced insulin

secretion in a pancreatic beta-cell line (INS-1). FEBS Lett. 377, 237–239.

Zhang, X., Zhang, G., Zhang, H., Karin, M., Bai, H., and Cai, D. (2008). Hypo-

thalamic IKKbeta/NF-kappaB and ER stress link overnutrition to energy imbal-

ance and obesity. Cell 135, 61–73.

tabolism 12, 237–249, September 8, 2010 ª2010 Elsevier Inc. 249