Hypothalamic and pituitary c-Jun N-terminal kinase 1signaling coordinately regulates glucose metabolismBengt F. Belgardta,1, Jan Mauera,1, F. Thomas Wunderlicha,1, Marianne B. Ernsta, Martin Pala, Gabriele Spohna,Hella S. Brönnekeb, Susanne Brodesserc, Brigitte Hampela, Astrid C. Schaussd, and Jens C. Brüninga,2

aDepartment of Mouse Genetics and Metabolism, Institute for Genetics, Cologne Excellence Cluster on Cellular Stress Responses in Aging Associated Diseases(CECAD) and Center of Molecular Medicine Cologne, University of Cologne and 2nd Department for Internal Medicine, University Hospital Cologne, MaxPlanck Institute for the Biology of Aging, D-50674 Cologne, Germany; bMouse Phenotyping Core Facility, CECAD, D-50674 Cologne, Germany; cLipidomicsCore Facility, CECAD and Institute for Medical Microbiology, Immunology, and Hygiene, D-50674 Cologne, Germany; and dMolecular Imaging Core Facility,CECAD, D-50674 Cologne, Germany

Communicated by Marc R. Montminy, The Salk Institute for Biological Studies, La Jolla, CA, February 13, 2010 (received for review December 22, 2009)

c-JunN-terminal kinase (JNK) 1-dependent signalingplaysa crucial rolein the development of obesity-associated insulin resistance. Here wedemonstrate that JNK activation not only occurs in peripheral tissues,but also in the hypothalamus and pituitary of obese mice. To resolvethe importance of JNK1 signaling in the hypothalamic/pituitarycircuitry, we have generated mice with a conditional inactivation ofJNK1 in nestin-expressing cells (JNK1ΔNES mice). JNK1ΔNES mice exhibitimproved insulin sensitivity both in the CNS and in peripheral tissues,improvedglucosemetabolism, aswell as protection fromhepatic stea-tosis and adipose tissue dysfunction uponhigh-fat feeding.Moreover,JNK1ΔNES mice also show reduced somatic growth in the presence ofreduced circulating growth hormone (GH) and insulin-like growth fac-tor 1 (IGF1) concentrations, as well as increased thyroid axis activity.Collectively, these experiments reveal an unexpected, critical role forhypothalamic/pituitary JNK1 signaling in the coordination of meta-bolic/endocrine homeostasis.

brain | diabetes | obesity | inflammation | insulin resistance

Obesity causes increased circulating concentrations of cyto-kines such as tumor necrosis factor (TNF) α and interleukin

(IL) 6 in animal models as well as humans (1, 2). These proin-flammatory mediators act on peripheral tissues, causing periph-eral insulin resistance via inflammatory and stress signalingcascades including activation of c-Jun N-terminal kinase (JNK)signaling (3). Thus, cytokine and fatty acid-induced activation ofJNK1 causes insulin resistance in vitro and in vivo (4, 5). JNKsare able to regulate transcription, survival, apoptosis, and othercellular events in response to diverse stimuli such as UV irra-diation or cytokine stimulation (6). Interestingly, JNK1 plays animportant role in obesity-associated pathologies, because con-ventional JNK1 but not JNK2 knockout mice are protected fromobesity-induced hyperglycemia, hyperinsulinemia, and insulinresistance (4, 7). Moreover, fat cell-specific disruption of JNK1provided evidence for an important role of adipose tissue JNK1in glucose metabolism (8).Glucose homeostasis and body weight are under control of

hypothalamic circuits regulating food intake, energy expenditure,and hepatic glucose production (9). The hypothalamus has beenidentified as one of themain targets both for insulin and leptin (10).Consistently, insulin and leptin action on hypothalamic neuronsubpopulations is necessary for normal body weight control andglucose homeostasis (11–15).Recent findings indicate that obesity leads to activation of stress

signaling cascades not only in peripheral, classical insulin targettissues such as skeletal muscle, liver, and adipose tissue, but also inthe central nervous system thereby causing neuronal insulin andleptin resistance (16). Similar to findings in peripheral tissues,obesity causes an increase in NFκB activity in the hypothalamus,and neuronal inactivation or pharmacological inhibition ofinhibitor of NFκB kinase (IKK) 2 signaling protects from insulinand leptin resistance (17, 18), in line with the notion that proin-

flammatory cytokines, fatty acids, or both activate the NFĸBpathway in neurons (18–20).However, the role of central JNK1 signaling in insulin and leptin

resistance is less clear. In rats, it has been shown that high-fat diet(HFD) feeding increases total hypothalamic JNK activity, andintracerebroventricular (ICV) application of a general JNK inhib-itor ameliorated hyperphagia and obesity (21). Moreover, conven-tional JNK1 knockout mice showed reduced adiposity, potentiallyindicating improved central leptin and/or insulin action in theseanimals (4). In light of thesefindings,weablatedJNK1specifically inthe CNS and pituitary cells. We report an unexpected role of neu-ronal and pituitary JNK1 as a regulator of hypothalamic andperipheral insulin sensitivity as well as somatic growth.

ResultsHigh-Fat Feeding Causes Neuronal and Pituitary JNK Activation. Toaddress the role of central JNK1 signaling in regulation of energyhomeostasis, we analyzed the activation of the JNK signalingcascade in hypothalami of diet-induced obesemice.HypothalamicJNKactivity, as assessed by the ability of immunoprecipitated JNKto phosphorylate c-Jun in vitro, was significantly increased uponexposure to HFD (Fig. 1 A and B). Obesity is associated withnumerous endocrine abnormalities affecting the somatotrophic,thyroid, and glucocorticoid axes (22–24). Thus, we asked whetherobesity not only promotes JNK activation in the CNS, but also inthe pituitary. Indeed, we detected significantly increased JNKactivity in pituitaries of HFD-fed mice, indicating that JNK-dependent signaling may play a role in pituitary regulation ofperipheral metabolism (Fig. 1 C and D).

Generation of JNK1ΔNES Mice. Tounravel the functional role of JNK1in the CNS and pituitary, we generated mice with targeted ablationof JNK1. It has recently been demonstrated that in addition toneurons, the Nestin gene is also expressed in pituitary stem cells,allowingdeletionof target genes inall cell typesof theadultpituitaryusing Nestin-Cre transgenic mice (25). Hence, after generation ofmice with a loxP-flanked JNK1 allele (Fig. S1C), we crossed theseanimals with mice expressing the Cre recombinase under control oftheNestin promoter to subsequently generatemicehomozygous forthe loxP-flanked JNK1 allele and positive for the Nestin-Cretransgene (genotype JNK1fl/fl, Nestin-Cre+, i.e., JNK1ΔNES mice).Littermates negative for Cre were used as controls (JNK1fl/fl, called

Author contributions: B.F.B., J.M., F.T.W., and J.C.B. designed research; B.F.B., J.M., F.T.W.,M.B.E., M.P., H.S.B., S.B., B.H., and A.C.S. performed research; F.T.W. and G.S. contributednew reagents/analytic tools; B.F.B. and J.M. analyzed data; and B.F.B., J.M., and J.C.B.wrote the paper.

The authors declare no conflict of interest.1B.F.B., J.M., and F.T.W. contributed equally to this work.2To whom correspondence should be addressed. E-mail: jens.bruening@uni-koeln.de.

This article contains supporting information online at www.pnas.org/cgi/content/full/1001796107/DCSupplemental.

6028–6033 | PNAS | March 30, 2010 | vol. 107 | no. 13 www.pnas.org/cgi/doi/10.1073/pnas.1001796107

controls henceforth). As noted in similar studies, we assured thatneither presence of the Nestin-Cre transgene nor loxP sites per seimpaired control of body weight (Fig. S2A) (19).Immunoblot analyses of JNK protein expression revealed

unchanged JNK1 expression in peripheral tissues, whereas JNK1expression was largely reduced in the brain of JNK1ΔNES micecompared to controls (Fig. S2B). Thus, hypothalamic expression ofJNK1was decreased bymore than 95% in JNK1ΔNESmice, whereasexpression of JNK2 or JNK3 remained unaltered (Fig. S2C).Importantly, we also detected pituitary JNK1 deletion in JNK1ΔNES

mice using PCR-based analysis (Fig. S2D). Notably, CNS JNK1ablationdidnot affect anxiety,motor control, ormemory retrieval inbehavioral tests (Fig. S3 A–F).

Reduced Body Weight but Unchanged Adiposity in JNK1ΔNES Mice. Toinvestigate the potential role of neuronal/pituitary JNK1 in controlof energy homeostasis and metabolism, we analyzed male controland JNK1ΔNESmice exposed to normal diet (ND) and high-fat diet(HFD). Beginning at the age of 4 weeks, body weight of JNK1ΔNES

mice on either diet was decreased compared to control animals(Fig. 1E and F). Next, wemeasured epigonadal fat pad weight andbody composition of control and JNK1ΔNES mice. Surprisingly,body composition was unchanged in JNK1ΔNES mice compared tocontrol mice (Fig. S4A). Accordingly, although absolute weight ofepigonadal fat pads was decreased in JNK1ΔNES mice (Fig. S4B),the relation of fat pad to body weight was unchanged betweengenotypes (Fig. S4C). Consistent with unaltered fat mass inJNK1ΔNESmice, we could not detect significant alterations in foodintake (Fig. S4D), physical activity (Fig. S4E), or respiratoryquotient (Fig. S4E) in JNK1ΔNES mice. We have previously dem-onstrated that JNK1ΔNES mice are not protected from obesity-induced leptin resistance (20). In line with unchanged leptin sen-

sitivity also under ND conditions, intracerebroventricular (icv)injection of leptin reduced body weight and food intake to thesame extent in lean control and JNK1ΔNES mice, indicating thathomeostatic control by leptin is not changed upon central JNK1ablation (Fig. 1G and Fig. S4F). Taken together, central JNK1ablation affects body weight, but does not affect body compositionor leptin action in mice.

Impaired Somatic Growth in JNK1ΔNES Mice.Asbody compositionwasunchanged, but bodyweightwas significantly decreased in JNK1ΔNES

mice compared to control mice, we sought to determine whethersomatic growth differed between these two groups of mice. Impor-tantly, somatic growth and insulin sensitivity are interlinked, asreduced somatic growth induced either genetically or by caloricrestriction (CR) improves glucose homeostasis and insulin sensitivityin mice, rats, apes, and humans, whereas caloric overabundance willlead to increased somatic growth (26–30). Strikingly, JNK1ΔNESmiceshowed significantly decreased naso-anal length on both diets (Fig.2A). Although high-fat diet caused a significant increase in length(increase of 0.4 cm,P=0.01) in control animals compared to controlanimals on normal diet, this diet-induced elongation was blunted inJNK1ΔNES mice (increase of 0.2 cm, not significant) (Fig. 2A).In light of these findings, we examined somatotrophic regulation

in both groups of mice. Somatic growth is controlled via growthhormone releasing hormone (GHRH), which is secreted by hypo-thalamic neurons, and acts on somatotrophs in the pituitary toproduce and release growth hormone (GH).GH in turn acts onGHreceptors on the liver, inducing insulin-like growth factor 1 (IGF1)transcription and release, which acts on chondrocytes to increaselongitudinal growth (for review see ref. 31). Strikingly, serum IGF1levels were decreased by 50% in JNK1ΔNES mice (Fig. 2B), as washepatic IGF1 mRNA expression (Fig. S4G). Despite considerable

Age [weeks]

0

5

10

15

20

25

30

35

40

4 5 6 7 8 9 10 11 12 13 14 15 16

JNK1 fl/fl

JNK1 NES

**

*** ***

Bo

dy w

eig

ht

[g]

E

p-c

-ju

n/J

NK

% o

f N

D

50

100

150

200

250

0

ND HFD

*

JNK input

p-c-Jun

ND HFD

Pituitary

*

p-c

-ju

n/J

NK

% o

f N

D

0

50

100

150

200

ND HFD

JNK input

p-c-Jun

ND HFD

Hypothalamus

G

0

20

40

60

80

100

120

- + - + Leptin

Fo

od

in

tak

e[%

aC

SF

]

* *

B C A D

Bo

dy w

eig

ht

[g]

0

5

10

15

20

25

30

35

40

45

50

4 5 6 7 8 9 10 11 12 13 14 15 16

Age [weeks]

***

JNK1 fl/fl

JNK1 NES F

Fig. 1. JNK1ΔNES mice show reduced body weight but unchanged leptin sensitivity. (A) High-fat diet (HFD) feeding increases hypothalamic JNK activation.C57bl/6 mice were fed either ND or HFD for 8 weeks, and hypothalami were microdissected. Total JNK activity in this tissue was measured by performing JNKkinase assays, and phosphorylation of the JNK target c-Jun was detected by immunoblot. JNK1/3 loading was used as input control. A representativeimmunoblot from one ND and one HFD animal is shown. (B) Quantification of HFD-induced JNK activation in the hypothalamus. The ratio of p-c-Jun to JNK1/3input was quantified in Western blots of hypothalami from 8 ND and 8 HFD-fed animals as shown in A. (C) High-fat diet (HFD) feeding increases pituitary JNKactivation. C57bl/6 mice were fed either ND or HFD for 8 weeks, and pituitaries were isolated. Total JNK activity in this tissue was measured by performing JNKkinase assays, and phosphorylation of the JNK target c-Jun was detected by immunoblot. JNK1/3 loading was used as input control. A representativeimmunoblot from one ND and one HFD animal is shown. (D) Quantification of HFD-induced JNK activation in the pituitary. The ratio of p-c-Jun to JNK1/3input was quantified in Western blots of pituitaries from 4 ND and 4 HFD-fed animals as shown in C. (E) Average body weight of JNK1fl/fl (▫) and JNK1ΔNES (▪)mice on normal diet (n = 12 per group). (F) Average body weight of JNK1fl/fl (▫) and JNK1ΔNES (▪) mice on high-fat diet (n = 12 per group). (G) Twenty-four-hour food intake after intracerebroventricular leptin treatment in JNK1fl/fl (▫) (n = 5) and JNK1ΔNES (▪) mice (n = 4) mice on normal diet at the age of 12 weeks.Mice were injected with either vehicle or 2 μg leptin immediately before onset of dark phase, and food intake was measured 24 h later. Displayed values aremeans ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Belgardt et al. PNAS | March 30, 2010 | vol. 107 | no. 13 | 6029

NEU

ROSC

IENCE

variation of GH concentrations in control mice, presumably due tothe pulsatile secretion pattern of GH (32), JNK1ΔNES mice showeddecreased circulating GH concentrations (Fig. S4H).However, hypothalamic mRNA expression of both GHRH

and somatostatin, which can inhibit GH release, was unchanged,indicating that JNK1-mediated regulation of the somatotrophicaxis does not occur in the hypothalamus (Fig. S4I). Thus, weassessed expression of GH mRNA directly in the pituitary andfound decreased GH mRNA expression in pituitaries ofJNK1ΔNES mice (Fig. 2C). However, immunohistochemicalanalysis of GH expression revealed unaltered pituitary structure,indicating that cell loss does not account for the decrease in GHmRNA (Fig. 2D).On the other hand—consistent with unchanged GHRH expres-

sion but decreased GH production—GHRH receptor (GHRHR)expression was decreased in pituitaries from JNK1ΔNES mice com-pared to control mice, indicating that decreasedGHRHR signalingmay account for the decreases in circulating GH and subsequentlyIGF1 (Fig. 2E). Taken together, JNK1ΔNES mice show decreasedpituitary GHRHR expression, and subsequently reduced GH andIGF1 serum levels, and ultimately impaired somatic growth,revealing an unexpected critical role for pituitary JNK1 signaling incontrol of somatic growth and body length.

JNK1ΔNES Mice Show Increased Central Insulin Sensitivity. We nextasked whether central insulin action may be altered in JNK1ΔNES

mice (18). Intracerebroventricular administration of insulin at adosage that had no effect on food intake and body weight in controlanimals reduced food intake and triggered significant weight loss inJNK1ΔNES mice fed ND (Fig. 3 A and B). Insulin’s anorexigeniceffect was also retained under HFD conditions in JNK1ΔNES mice(Fig. 3 C and D), indicating that central JNK1 ablation improvescentral insulin sensitivity under ND and HFD conditions. Impor-tantly, improvedhypothalamic insulinaction in theabsenceof JNK1was additionally confirmed by increased AKT activation (phos-

phorylation) upon icv. insulin treatment in JNK1ΔNES mice (Fig.3 E and F).

JNK1ΔNES Mice Are Protected from Diet-Induced Glucose Intoleranceand Insulin Resistance. We next addressed whether elevated CNSinsulin sensitivity as well as reduced GH levels altered peripheralglucose homeostasis and insulin sensitivity. Indeed, random fed andfasted blood glucose concentrations were significantly reduced inlean (Fig. S5A) and obese JNK1ΔNES mice compared to controls(Fig. 4A). Concomitantly, these mice performed significantly betterin glucose tolerance tests (Fig. 4B), exhibited decreased seruminsulin concentrations (Fig. 4C) and a significantly improved per-formance in insulin tolerance tests (ITTs) (Fig. S5B and Fig. 4D).To identify the tissues responsible for the improved glucose

homeostasis,wedirectlydetermined liver insulin sensitivity, becausehepatic glucose production is under both direct (hepatic) andindirect (CNS) control of insulin action (9, 11). Intraperitonealinsulin injections led to enhanced AKT phosphorylation in obeseJNK1ΔNESmice (Fig. 4E andF), indicating that JNK1ΔNESmice areprotected from diet-induced hepatic insulin resistance. Moreover,in line with previous reports that CNS insulin action up-regulateshepatic IL6 expression to reduce gluconeogenesis, we detected 1.6-fold elevated IL6 expression in livers of JNK1ΔNES mice comparedto controls (Fig. S5C) (11). Besides enhanced insulin sensitivity, wealso noted reduced hepatic glucose production as measured bypyruvate tolerance test, possibly due to reduced GH stimulation(Fig. S5D). Collectively, central JNK1 ablation enhances hypo-thalamic and hepatic insulin sensitivity.

JNK1ΔNES Mice Show Enhanced Thyroid Activation. We examinedwhether other pituitary functions besides the somatrophic axiswere affected in JNK1ΔNES mice. Although corticosterone con-centrations were unchanged between control and JNK1ΔNES mice(Fig. S6A), serumT3 levels were surprisingly increased in the latter(Fig. 5A). In line with increased T3 levels, we detected increasedO2 consumption in JNK1ΔNES mice (Fig. S6B), and additionally

B

D

A

fed fasted 0

50

100

150

200

250

300

350

400

450

500

Se

rum

IG

F-1

[n

g/m

l]

*** ***

0

0.2

0.4

0.6

0.8

1

1.2

GH

*

Re

l. p

itu

ita

rye

xp

res

sio

n

GH staining

JNK1 fl/fl

PL

AL AL

PL

JNK1 NES

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Re

l. p

itu

ita

rye

xp

res

sio

n

ND HFD

***

*** ***

GHRHR

E

ND HFD 8.5

9.0

9.5

10.0

10.5

11.0

11.5 ***

***

Len

gth

[cm

]

C

JNK1 fl/fl

JNK1 NES

*

Fig. 2. JNK1ΔNES mice show decreased activation of the somatotrophic axis. (A) Naso-anal length of JNK1fl/fl (▫) (n = 9 vs. 9) and JNK1ΔNES (▪) (n = 9 vs. 9) onnormal or high-fat diet at the age of 16 weeks. (B) Serum IGF1 concentrations of JNK1fl/fl (▫) (n = 10) and JNK1ΔNES (▪) (n = 10) on normal diet either randomfed or fasted at the age of 10 weeks. (C) Pituitary expression of GH of JNK1fl/fl (▫) (n = 6) and JNK1ΔNES (▪) mice (n = 6) at the age of 16 weeks as measured byreal-time–PCR. (D) Immunohistochemistry for GH from pituitary sections of JNK1fl/fl (▫) and JNK1ΔNES (▪) mice at the age of 16 weeks (green, GH; blue, DAPI).At least 3 mice of each genotype were analyzed. (Original magnification, ×100.) (Scale bar, 80 μm.) PL, posterior lobe; AL, anterior lobe. (E) Pituitaryexpression of GHRHR of JNK1fl/fl (▫) (n = 6) and JNK1ΔNES (▪) mice (n = 6) fed either normal chow diet or high-fat diet at the age of 16 weeks as measured byreal-time PCR. Displayed values are means ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

6030 | www.pnas.org/cgi/doi/10.1073/pnas.1001796107 Belgardt et al.

lipid load of brown adipocytes in interscapular brown adiposetissue (BAT) of JNK1ΔNESmice appeared less (Fig. 5B) andUCP1mRNA expression tended to be increased (Fig. S6C).We next asked whether hypothalamic mRNA expression of thy-

rotropin-releasing hormone (TRH) was also increased. Real-timeanalysis revealed no consistent change in TRH mRNA expressionon either diet, again pointing to a pituitary autonomous dysregula-

tion of thyroid control in these mice (Fig. S6D). In line with thefinding of increased circulating T3 concentrations, we observed amore than 3-fold increase in thyroid-stimulating hormone β (TSHβ)mRNAexpression in JNK1ΔNESmice (Fig. 5C).AlsoTRH receptor(TRHR) expression was increased in pituitaries of JNK1ΔNES mice(Fig. 5D). Immunohistological analysis of TSHβ positive cells didnot show significant structural alterations inpituitaries of JNK1ΔNES

D

B

E

0

20

40

60

80

100

120

140 C

- + - + Insulin

Fo

od

in

tak

e[%

aC

SF

]

*

*

*

-2.5

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

Bo

dy w

eig

ht

ch

an

ge

[% o

f B

W p

re-i

nj.]

*

- + - + Insulin

BW gain

BW loss

A

- + - + Insulin 0

20

40

60

80

100

120

Fo

od

in

tak

e[%

aC

SF

]

**

**

- + + + - + + + Insulin 0

100

200

300

400

500 **

p-A

KT

/AK

T%

of

JN

K1

fl/f

l

p-AKT

AKT

-actin

JNK1 fl/fl JNK1 NES

3

2

1

0

1

2

3

4

5

6 **

**

Bo

dy w

eig

ht

ch

an

ge

[% o

f B

W p

re-i

nj.]

- + - + Insulin

BW gain

BW loss

F

JNK1 fl/fl

JNK1 NES

Fig. 3. JNK1ΔNES mice show increased hypothalamic insulin sensitivity. (A) Twenty-four-hour food intake after intracerebroventricular insulin treatment inJNK1fl/fl (▫) (n = 9) and JNK1ΔNES (▪) (n = 8) mice on normal diet at the age of 12 weeks. Mice were injected with either vehicle or 2 mU insulin immediatelybefore onset of dark phase, and food intake was measured 24 h later. (B) Body weight loss 24 h after intracerebroventricular insulin treatment in JNK1fl/fl (▫)(n = 9) and JNK1ΔNES (▪) (n = 8) mice on normal diet at the age of 12 weeks. Mice were injected with either vehicle or 2 mU insulin immediately before onset ofdark phase, and body weight was measured 24 h later. Shown is percent change of body weight after insulin injection compared to vehicle injection. (C)Twenty-four-hour food intake after intracerebroventricular insulin treatment in JNK1fl/fl (▫) (n = 5) and JNK1ΔNES (▪) (n = 5) mice on high-fat diet at the age of10 weeks. Mice were injected with either vehicle or 4 mU insulin immediately before onset of dark phase, and food intake was measured 24 h later. (D) Bodyweight change 24 h after intracerebroventricular insulin treatment in JNK1fl/fl (▫) (n = 5) and JNK1ΔNES (▪) (n = 5) mice on high-fat diet at the age of 10 weeks.Mice were injected with either vehicle or 4 mU insulin immediately before onset of dark phase, and body weight was measured 24 h later. Shown is percentchange of body weight after insulin injection compared to ACSF injection. (E) Hypothalamic AKT activation upon icv insulin treatment is improved in JNK1ΔNES

mice. JNK1fl/fl and JNK1ΔNES mice on high-fat diet at the age of 10 weeks were fasted for 48 h, injected with ACSF or 4 mU insulin, and killed 20 min later.Immunoblot was performed for phosphorylated (activated) AKT, total AKT, and β-actin protein content. (F) Quantification of insulin-induced AKT phos-phorylation compared to total AKT content shown in Fig. 3E. N = 3 per group. Displayed values are means ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

0

20

40

60

80

100

120

* ** **

Re

l. b

loo

dg

luc

os

e[%

]

0 15 30 60

rel. Time [min]

rel. Time [min]

0

100

200

300

400

500

600

0 15 30 60 120

*

* Blo

od

glu

co

se

[mg

/dl]

- + + + - + + + Insulin

p-AKT

JNK1 fl/fl JNK1 NES

IR-

Liver

- + - + Insulin

0

2

4

6

8

10

12

14 *

He

pa

tic

AK

T a

cti

va

tio

n

(p-A

KT

/IR

)

Se

rum

in

su

lin

[ng

/ml]

0

1

2

3

4

5

6

7

*

p=0.06

HFD ND

W24 A

0

50

100

150

200

250

***

W8

*

W14 W16

*

***

Blo

od

glu

co

se

[mg

/dl]

fasted fed

B

JNK1 fl/fl

JNK1 NES

C

D E F

Fig. 4. JNK1ΔNES mice show improved glucose homeostasis and elevated insulin sensitivity. (A) Fasted and fed blood glucose concentration of JNK1fl/fl (▫) andJNK1ΔNES (▪) mice on high-fat diet at the indicated ages (n = 10–25 per group). (B) Intraperitoneal glucose tolerance test in JNK1fl/fl (▫) and JNK1ΔNES (▪) mice onhigh-fat diet at theageof14weeks (n=10pergroup). (C) Serum insulin concentrations in JNK1fl/fl (▫) and JNK1ΔNES (▪)miceonnormal andhigh-fatdiet at theageof 8–10weeks (n=8per group). (D) Intraperitoneal insulin tolerance test in JNK1fl/fl (▫) and JNK1ΔNES (▪)mice onhigh-fat diet at the age of 15weeks (n=9–10pergroup). (E) Insulin-stimulated hepatic AKT-phosphorylation in JNK1fl/fl and JNK1ΔNES mice on high-fat diet at the age of 12 weeks (n = 3 per group). (F) Quan-tification of insulin-stimulated hepatic AKT phosphorylation as shown in Fig. 4E. Displayed values are means ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Belgardt et al. PNAS | March 30, 2010 | vol. 107 | no. 13 | 6031

NEU

ROSC

IENCE

mice, indicating that TRHR expression, but not a change in TSHβcell number, underlies the increased expression of TSHβ mRNA(Fig. S6E).To further study the pituitary cell-autonomous regulation of

TRHR expression, we employed a rat pituitary cell line (GH4C1),which has been previously used to study TRHR expression (33).Although incubationwith either a PI3Kor anERK inhibitor did nothaveaneffect onTRHRexpression, incubationwith a JNKinhibitorsignificantly increased TRHRexpression by 70% (Fig. 5E). Thus, inline with the increase in pituitary TRHR expression of JNK1ΔNES

mice, this finding highlights the importance of JNK-dependent sig-naling in pituitary, cell-autonomous regulation of the thyroid axis.Taken together, JNK1ΔNES mice show increased thyroid actioncaused by an increase in pituitary TSHβ and TRHR expression.

DiscussionOur study identifies JNK1 as a specific inhibitor of hypothalamicinsulin action and thus together with earlier findings in adiposetissue-specific JNK1-deficient mice partially explains the improvedglucose homeostasis and systemic insulin sensitivity in conventionalJNK1 knockout mice (4, 8). Moreover, JNK1ΔNES mice are pro-tected from obesity-associated hepatic steatosis and adipose tissueinflammation (Fig. S7), further underlining the comprehensiveprotection from obesity-associated pathologies.Moreover, the current study reveals unique important insights

into the role of JNK1 in pituitary control of somatic growth andthyroid function. We show that similar to the hypothalamus andperipheral tissues, JNK activation in the pituitary is increasedupon HFD feeding. This finding has profound effects on somaticgrowth, energy expenditure, and glucose metabolism, as demon-strated in JNK1ΔNES mice. Here, we show that JNK1 deficiency insomatotroph pituitary cells reduces GHRH receptor and sub-sequently GH mRNA expression, translating into significantly

decreased circulating GH and IGF1 serum concentrations ulti-mately resulting in mildly reduced somatic growth.As indicated in multiple studies of dwarfism and caloric restric-

tion, mice with reduced GH and IGF1 levels are insulin sensitive,demonstrate decreased serum glucose concentrations, and inter-estingly, show increased life span and decreased aging-associatedpathologies (26–29). Notably, caloric restriction does not furtherimprove insulin sensitivity (and life span) in growth hormonereceptor-deficient mice, indicating that caloric restriction elicits itsbeneficial effects by dampening of the somatotrophic axis (34). It isconceivable that JNK activation in times of nutritional surplusevokeshormonal changes resulting in somatic growth, i.e., increasedGH expression. In line with this model, overfeeding and/or obesity,inducedbymutations in genes essential forweight control, increasesandaccelerates somatic growth in rodents andhumans, andweshowthat incontrol animals,HFDfeeding significantly increases pituitaryexpression of GHRHR, offering a unique mechanism for obesity-associated overgrowth (24, 30, 35). Taken together, pituitary JNK1deficiency in control of somatotrophic function can cooperate withthe insulin-sensitizing effect of JNK1 deficiency in the CNS toimprove peripheral glucose metabolism (Fig. 5G).Furthermore,our study reveals a critical role for JNK1 inpituitary

control of thyroid function, as demonstrated by increased T3 serumconcentrations, pituitary TSHβ expression, and energy expenditurein JNK1ΔNES mice especially under HFD conditions. Althoughcorrected for leanmass, the energy expenditure in obese JNK1ΔNES

mice may be confounded by decreased body length and mass.However, the well-documented hyperthyroidism along with alteredBAT morphology indicate a true increase in energy expenditure inthese mice. Additionally, we show that JNK inhibition increasedTRHR expression in vitro, further demonstrating the pituitary-autonomous role of JNK1 signaling in control of thyroid regulation.Because control of TRHRandGHRHR transcription is only partlyunderstood, and at this point we cannot exclude an additional role

TRHR

0

1

2

3

4

***

* *

TSHND HFD

Re

lati

ve

pit

. e

xp

res

sio

n

0

0.5

1

1.5

2

2.5

ND HFD

*

**

Re

lati

ve

pit

. e

xp

res

sio

n

0

Fre

e T

3 [

ng

/ml]

1

2

3

4

5

6 ***

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

control

ERK inhibitor*

JNK inhibitor

PI3K inhibitor

Re

lati

ve

TR

HR

ex

pre

ss

ion

JNK1fl/fl

JNK1 NESBAT

BAT

A B C

D E

Fig. 5. JNK1ΔNES mice show increased activation of the thyrotropic axis. (A) Serum-free T3 concentration of JNK1fl/fl (▫) (n = 8) and JNK1ΔNES (▪) mice (n = 8)on normal diet at the age of 10 weeks. ***, P ≤ 0.001. (B) Representative H&E staining of brown adipose tissue of control and JNK1ΔNES mice on high-fat dietat the age of 16 weeks. BAT sections of 3 mice per genotype were analyzed. (C) Pituitary expression of TSHβ of JNK1fl/fl (▫) (n = 6) and JNK1ΔNES (▪) mice (n = 6)on normal chow or high-fat diet at the age of 16 weeks as measured by real-time PCR. *, P ≤ 0.05; ***, P ≤ 0.001. (D) Pituitary expression of TRHR of JNK1fl/fl

(▫) (n = 6) and JNK1ΔNES (▪) mice (n = 6) on normal chow or high-fat diet at the age of 16 weeks as measured by real-time PCR. *, P ≤ 0.05; **, P ≤ 0.01. (E)Expression of TRHR in the rat pituitary cell line GH4C1. Expression of TRHR was measured after 16 h incubation with either 0.1% DMSO (control), SP600125(JNK inhibitor), LY294002 (PI3K inhibitor), or PD98059 (ERK inhibitor). For each sample within an experiment, triplicate values were averaged and then themeans of the real-time–PCR results from three independent experiments were compared. *, P ≤ 0.05.

6032 | www.pnas.org/cgi/doi/10.1073/pnas.1001796107 Belgardt et al.

for alteredhypothalamicTRH/GHRHrelease, futureworkmust beaimed at identifying the mechanisms by which JNK1 controls reg-ulation of both.Collectively, the pleiotropic effects of CNS and pituitary JNK1

deficiency result in a phenotype that has previously been connectedto healthy aging (increased insulin sensitivity accompanied byreduced glucose, insulin, and GH levels). Thus, further analysis ofJNK1ΔNES mice with respect to control of life span may allow newinsights into the connections between metabolism, growth,and aging.

MethodsIntracerebroventricular Leptin and Insulin Stimulation. Cannulas were implan-ted as previously described (36), except that the lateral ventricle was targetedusing coordinates located using a Brain Atlas (coordinates were bregma 1.0mm lateral, 0.2mmcaudal, and 2.0mmventral).Micewere allowed to recoverfor 1 week after surgery. For baseline measurements, mice were injected with2 μL artificial cerebrospinal fluid (ACSF) immediately before onset of darkphase. Food intake was measured 24 h later. After a 1-day break, ACSF injec-tion was repeated. After an additional 1-day break, either 2 μg mouse leptin(Sigma-Aldrich), 2mU (ND), or 4mU(HFD)porcine insulin (Sigma-Aldrich)wereinjected and food intake measured 24 h later. Leptin and insulin were dis-solved according to manufacturer’s instructions to generate stock solutions,and fresh aliquots were dissolved in ACSF immediately before injection.Injected volume was always 2 μL.

Western Blotting and JNK Assay. Indicated tissues were dissected andhomogenized in homogenization buffer with a polytron homogenizer (IKAWerke), and Western blot analyses were performed by standard methodswith antibodies raised against insulin receptor subunit β (Santa Cruz, sc-711),β-actin (Sigma, no. 4970), JNK1/3 (Santa Cruz, sc-474), phospho-S473-AKT,

and pan-AKT (Cell Signaling) (37). SAPK/JNK assays were performed fol-lowing the manufacturer’s guidelines (no. 9810; Cell Signaling).

Body Composition. Body fat content was measured in vivo by NMR using aminispec mq7.5 (Bruker Optik) as previously described (38).

Glucose, Insulin, and Pyruvate Tolerance Tests. Glucose, insulin, and pyruvatetolerance tests were performed as previously described (39).

Intraperitoneal Insulin Stimulation. Insulin stimulated hepatic AKT phos-phorylation was determined as previously described (8).

Immunohistochemistry. ForGHstainings,pituitarieswereextracted,putrapidlyin tissue freezing medium, and sectioned on a cryostat. Stainings were per-formed as previously reported (12). The GH antibody (A0570) was purchasedfrom DAKO.

Statistical Methods. Data were analyzed for statistical significance by two-tailed unpaired Student’s t test unless indicated otherwise.

ACKNOWLEDGMENTS. We thank G. Schmall and Tanja Rayle for excellentsecretarial assistance and Pia Scholl, Sonja Becker, and Jens Alber foroutstanding technical assistance. This work was supported by grants fromthe Center of Molecular Medicine Cologne (TV2) and the European Union(LSHM-CT-2003-503041) to J.C.B., the Fritz Thyssen Stiftung (Az. 10.04.1.153/Az. 10.06.2.175) to J.C.B., the European Foundation for the Study of Diabe-tes/Lilly European Diabetes Research Programme to J.C.B., and the GermanResearch Foundation (Deutsche Forschungsgemeinschaft) (Br. 1492/7-1) to J.C.B., and research leading to these results has received funding from theEuropean Community’s Seventh Framework Programme (FP7/2007-2013),acronym “TOBI,” under grant agreement no. 201608 (to J.C.B.).

1. Hotamisligil GS, Shargill NS, Spiegelman BM (1993) Adipose expression of tumornecrosis factor-alpha: Direct role in obesity-linked insulin resistance. Science 259:87–91.

2. Vgontzas AN, et al. (1997) Elevation of plasma cytokines in disorders of excessivedaytime sleepiness: Role of sleep disturbance and obesity. J Clin Endocrinol Metab 82:1313–1316.

3. Hotamisligil GS (2006) Inflammation and metabolic disorders. Nature 444:860–867.4. Hirosumi J, et al. (2002) A central role for JNK in obesity and insulin resistance. Nature

420:333–336.5. Nguyen MT, et al. (2005) JNK and tumor necrosis factor-alpha mediate free fatty acid-

induced insulin resistance in 3T3-L1 adipocytes. J Biol Chem 280:35361–35371.6. Bogoyevitch MA (2006) The isoform-specific functions of the c-Jun N-terminal kinases

(JNKs): Differences revealed by gene targeting. Bioessays 28:923–934.7. Hotamisligil GS (2008) Inflammation and endoplasmic reticulum stress in obesity and

diabetes. Int J Obes (2005) 32 (Suppl 7):S52–S54.8. Sabio G, et al. (2008) A stress signaling pathway in adipose tissue regulates hepatic

insulin resistance. Science 322:1539–1543.9. Belgardt BF, Okamura T, Bruning JC (2009) Hormone and glucose signaling in POMC

and AgRP neurons. J Physiol 587:5305–5314.10. Schwartz MW (2002) Neuronal pathways regulating food intake and body adiposity.

Ann Endocrinol (Paris) 63:117–120.11. Könner AC, et al. (2007) Insulin action in AgRP-expressing neurons is required for

suppression of hepatic glucose production. Cell Metab 5:438–449.12. Belgardt BF, et al. (2008) PDK1 deficiency in POMC-expressing cells reveals FOXO1-

dependent and -independent pathways in control of energy homeostasis and stressresponse. Cell Metab 7:291–301.

13. Balthasar N, et al. (2004) Leptin receptor signaling in POMC neurons is required fornormal body weight homeostasis. Neuron 42:983–991.

14. Kitamura T, et al. (2006) Forkhead protein FoxO1mediates Agrp-dependent effects ofleptin on food intake. Nat Med 12:534–540.

15. Gropp E, et al. (2005) Agouti-related peptide-expressing neurons are mandatory forfeeding. Nat Neurosci 8:1289–1291.

16. Velloso LA, Araújo EP, de Souza CT (2008) Diet-induced inflammation of thehypothalamus in obesity. Neuroimmunomodulation 15:189–193.

17. Posey KA, et al. (2009) Hypothalamic proinflammatory lipid accumulation, inflammation,and insulin resistance in rats fed a high-fat diet. Am J Physiol Endocrinol Metab 296:E1003–E1012.

18. ZhangX, et al. (2008)Hypothalamic IKKbeta/NF-kappaBandER stress link overnutritionto energy imbalance and obesity. Cell 135:61–73.

19. Ozcan L, et al. (2009) Endoplasmic reticulum stress plays a central role in developmentof leptin resistance. Cell Metab 9:35–51.

20. Kleinridders A, et al. (2009) MyD88 signaling in the CNS is required for developmentof fatty acid-induced leptin resistance and diet-induced obesity. Cell Metab 10:249–259.

21. De Souza CT, et al. (2005) Consumption of a fat-rich diet activates a proinflammatoryresponse and induces insulin resistance in the hypothalamus. Endocrinology 146:4192–4199.

22. Jung R (1984) Endocrinological aspects of obesity. Clin Endocrinol Metab 13:597–612.23. Salehi M, Ferenczi A, Zumoff B (2005) Obesity and cortisol status. HormMetab Res 37:

193–197.24. VignoloM,NaselliA,DiBattistaE,MostertM,AicardiG (1988)Growthanddevelopment in

simple obesity. Eur J Pediatr 147:242–244.25. Gleiberman AS, et al. (2008) Genetic approaches identify adult pituitary stem cells.

Proc Natl Acad Sci USA 105:6332–6337.26. Clemmons DR (2006) Involvement of insulin-like growth factor-I in the control of

glucose homeostasis. Curr Opin Pharmacol 6:620–625.27. Coschigano KT, et al. (2003) Deletion, but not antagonism, of the mouse growth

hormone receptor results in severely decreased body weights, insulin, and insulin-likegrowth factor I levels and increased life span. Endocrinology 144:3799–3810.

28. Dominici FP, Hauck S, Argentino DP, Bartke A, Turyn D (2002) Increased insulinsensitivity and upregulation of insulin receptor, insulin receptor substrate (IRS)-1 andIRS-2 in liver of Ames dwarf mice. J Endocrinol 173:81–94.

29. Flurkey K, Papaconstantinou J, Miller RA, Harrison DE (2001) Lifespan extension anddelayed immune and collagen aging in mutant mice with defects in growth hormoneproduction. Proc Natl Acad Sci USA 98:6736–6741.

30. Kappeler L, et al. (2009) Early postnatal nutrition determines somatotropic function inmice. Endocrinology 150:314–323.

31. Giustina A, Mazziotti G, Canalis E (2008) Growth hormone, insulin-like growth factors,and the skeleton. Endocr Rev 29:535–559.

32. Jansson JO, Edén S, Isaksson O (1985) Sexual dimorphism in the control of growthhormone secretion. Endocr Rev 6:128–150.

33. Høvring PI, Matre V, Fjeldheim AK, Loseth OP, Gautvik KM (1999) Transcription of thehuman thyrotropin-releasing hormone receptor gene-analysis of basal promoterelements and glucocorticoid response elements. Biochem Biophys Res Commun 257:829–834.

34. Bonkowski MS, Rocha JS, Masternak MM, Al Regaiey KA, Bartke A (2006) Targeteddisruption of growth hormone receptor interferes with the beneficial actions ofcalorie restriction. Proc Natl Acad Sci USA 103:7901–7905.

35. Martin NM, et al. (2006) Abnormalities of the somatotrophic axis in the obese agoutimouse. Int J Obes (2005) 30:430–438.

36. Brown LM, Clegg DJ, Benoit SC, Woods SC (2006) Intraventricular insulin and leptinreduce food intake and body weight in C57BL/6J mice. Physiol Behav 89:687–691.

37. Brüning JC, et al. (1998) A muscle-specific insulin receptor knockout exhibits featuresof the metabolic syndrome of NIDDM without altering glucose tolerance. Mol Cell 2:559–569.

38. Mesaros A, et al. (2008) Activation of Stat3 signaling in AgRP neurons promoteslocomotor activity. Cell Metab 7:236–248.

39. Brüning JC, et al. (2000) Role of brain insulin receptor in control of body weightand reproduction. Science 289:2122–2125.

Belgardt et al. PNAS | March 30, 2010 | vol. 107 | no. 13 | 6033

NEU

ROSC

IENCE