TS

RVOR

*oTSD

BfllsmrittaaTmrtfiTinImIptprfflsecCtprg

GASTROENTEROLOGY 2009;136:663–672

issue Inhibitor of Metalloproteinase 3 Deficiency Causes Hepaticteatosis and Adipose Tissue Inflammation in Mice

OSSELLA MENGHINI,* STEFANO MENINI,‡ ROBERTA AMORUSO,* LOREDANA FIORENTINO,* VIVIANA CASAGRANDE,*ALERIA MARZANO,*,§ FEDERICA TORNEI,* PIERFRANCESCO BERTUCCI,� CARLA IACOBINI,‡ MATTEO SERINO,*TTAVIA PORZIO,*,� MARTA L. HRIBAL,¶ FRANCO FOLLI,# RAMA KHOKHA,** ANDREA URBANI,§,‡‡,§§

ENATO LAURO,*,� GIUSEPPE PUGLIESE,‡ and MASSIMO FEDERICI*,� �

Department of Internal Medicine, University of Rome “Tor Vergata,” Rome, Italy; ‡Department of Clinical Sciences, “La Sapienza” University, Rome, Italy; §Laboratoryf Proteomics, European Brain Research Institute/Santa Lucia Foundation, Rome, Italy; �Department of Laboratory Medicine, � �Center for Atherosclerosis, “Policlinicoor Vergata” University Hospital, Rome, Italy; ¶University of Magna Graecia, Catanzaro, Italy; #Division of Diabetes, University of Texas Health and Science Center,an Antonio, Texas; **Ontario Cancer Institute, University of Toronto, Toronto, Ontario, Canada; ‡‡Centro Studi sull’Invecchiamento, University Foundation “G.

’Annunzio,” Chieti, Italy; and the §§Department of Biomedical Science, University “G. D’Annunzio,” Chieti, Italy

Oipfmaktbi

flsmnepr

smtt(tl(p

rkfapvt

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

See Mazhar SM et al on page 135 in CGH.

ackground & Aims: Obesity-driven, low-grade in-ammation affects systemic metabolic function and can

ead to insulin resistance, hepatic steatosis, and athero-clerosis. Decreased expression of tissue inhibitor of

etalloproteinase 3 (Timp3) is a catalyst for insulinesistance and inflammation. Timp3 is a natural inhib-tor of matrix metalloproteinases, tumor necrosis fac-or-�–converting enzyme (TACE), and vascular endo-helial growth factor receptor 2, and therefore couldffect signaling processes involved in inflammation andngiogenesis. Methods: We assessed the effects ofimp3 on inflammation, tissue remodeling, and inter-ediary metabolism in mice, under conditions of envi-

onmental stress (high-fat diet), genetic predispositiono insulin resistance (insulin receptor [Insr] haploinsuf-ciency), and varying levels of inflammation (Timp3 orace deficiencies). Metabolic tests, immunohistochem-

stry, real-time polymerase chain reaction, and immu-oblotting were used to compare data from wild-type,

nsr�/�, Timp3�/�, Insr�/�Timp3�/�, and Insr�/�Tace�/�

ice placed on high-fat diets for 10 weeks. Results:nsr�/�Timp3�/� mice showed a higher degree of adi-ose and hepatic inflammation compared with wild-

ype, Insr�/�, Timp3�/�, and Insr�/�Tace�/� mice. Inarticular, the Insr�/�Timp3�/� mice developed mac-ovesicular steatosis and features of severe nonalcoholicatty liver disease, including lobular and periportal in-ammation, hepatocellular ballooning, and perisinu-oidal fibrosis. These were associated with increasedxpression of inflammatory and steatosis markers, in-luding suppressor of cytokine signaling 3 and stearoyloA desaturase 1, in both liver and adipose tissue. In-

erestingly, Insr�/�Tace�/� mice had a nearly oppositehenotype. Conclusions: Timp3, possibly through itsegulation of TACE, appears to have a role in the patho-

enesis of fatty liver disease associated with obesity.

besity, a chronic disorder with increased incidenceand prevalence in Western countries, is believed to

nduce a systemic low-grade inflammatory state predis-osing to diabetes, vascular diseases, and cancer, arisingrom deregulated release of free fatty acids and inflam-

atory cytokines from adipose tissue.1 Free fatty acidsnd proinflammatory cytokines activate serine-threonineinases such as c-Jun-N-terminal kinase (JNK) and pro-ein kinase C �, which perturb glucose homeostasis bylunting insulin-receptor– dependent signaling pathways

n peripheral tissues including liver.1–3

One major metabolic consequence of increased lipid ef-ux to the liver and obesity-driven inflammation is hepaticteatosis, which is sustained by alteration in both inflam-

atory and metabolic pathways.4–6 Pathways and mecha-isms linking obesity to hepatic steatosis are not completelylucidated but involve expression of monocyte chemotacticrotein-1 (MCP-1) and its receptor C-C motif chemokineeceptor-2.7,8

We recently identified a modifier gene for insulin re-istance and inflammation in mice, tissue inhibitor of

etalloproteinase 3 (Timp3).9 Timp3 acts pericellularlyo modulate several cell membrane enzymes and recep-ors.10,11 Among Timp3 substrates, tumor necrosis factorTNF)-�– converting enzyme (TACE, also named a disin-egrin and metalloproteinase domain 17 or ADAM17) regu-ates TNF receptors, epidermal growth factor receptorEGFR), and interleukin 6 receptor signaling pathways,utting Timp3 at the cross-road of inflammatory and

Abbreviations used in this paper: EGFR, epidermal growth factoreceptor; HFD, high-fat diet; Insr, insulin receptor; JNK, Jun-N-terminalinase; MCP-1, monocyte chemotactic protein-1; NAFLD, nonalcoholicatty liver disease; NO, nitric oxide; PPAR�, peroxisome proliferator-ctivated receptor �; SCD1, stearoyl CoA desaturase 1; SOCS-3, sup-ressor of cytokine signaling 3; TACE, tumor necrosis factor-�–con-erting enzyme; Timp3, tissue inhibitor of metalloproteinase 3; TNF,umor necrosis factor; WAT, white adipose tissue; WT, wild-type.

© 2009 by the AGA Institute0016-5085/09/$36.00

doi:10.1053/j.gastro.2008.10.079

fiotrgfi(oH

stsvaa(

wmMofotwBmwsklSMaMuoBfMSim

mpwkMWE

STtSaITIe

tewtavmn

aOipCtmrfCtpntaBsl

mewactIbg

5

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

664 MENGHINI ET AL GASTROENTEROLOGY Vol. 136, No. 2

brosis signals.10 However, via its inhibitory propertiesn matrix metalloproteinase-2 and matrix metallopro-einase-9 as well as vascular endothelial growth factoreceptor 2, Timp3 may affect tissue remodeling and an-iogenesis.11–14 Moreover, we have shown that TACE de-ciency confers partial protection from high-fat diet

HFD)-induced insulin resistance,12 with positive effectsn liver ability to resist inflammatory effects increased byFD, particularly TNF-�–mediated insulin resistance.Here, we test and exploit the effects of the signaling

witch properties of Timp3 connecting inflammation,issue remodeling, and intermediary metabolism. Wehow that combination of genetic manipulation and en-ironmental stress (HFD) impacts on inflammation, ste-tosis, and fibrosis, resulting in a phenotype with char-cteristics similar to nonalcoholic fatty liver diseaseNAFLD).4 – 6,13,14

MethodsAnimal Models and Analytic ProceduresInsulin receptor (Insr)�/�Timp3�/� and Insr�/�Tace�/�

ere obtained breeding Insr�/�, Timp3�/�, and TACE�/�

ice on a C57/BL6 background, as previously described.9,12

etabolic testing procedures have been described previ-usly.9,12 Briefly, for glucose tolerance tests animals wereasted for 16 hours and injected with 2 g/kg body weightf glucose into the peritoneal cavity; insulin toleranceests were performed by injection of 0.75 U/kg bodyeight of human regular insulin (Novo Nordisk/ASagsvaerd, Denmark) into the peritoneal cavity of ani-als fasted for 6 hours. Blood glucose concentrationsere determined by using an automated Onetouch Life-

can Glucometer (Milpitas, CA). Hormones and adipo-ines levels were measured using commercial kits: insu-

in (Mercodia, Uppsala, Sweden), adiponectin (R&Dystems), and leptin (R&D Systems, Minneapolis,N). Aspartate aminotransferase (AST) and alanine

minotransferase (ALT) levels were measured using aodular P analyzer (Roche SpA, Monza, Italy). Individ-

ally caged mice from all groups were fed an HFD (60%f calories from fat, code D12492; Research Diets, Newrunswick, NJ) or chow (standard chow, 10% calories

rom fat, code 4RF18; GLP Mucedola Srl, Settimoilanese, Italy) for 10 weeks after weaning as indicated.

tudies were performed only in male mice. Animal stud-es were approved by the University of Tor Vergata Ani-

al Care and Use Committee.

Western BlotsWestern blots were performed on total tissue ho-

ogenates and nuclear and cytosolic extracts prepared asreviously described.9,12,15,16 The following antibodiesere used: anti-phospho Thr172 AMP-activated proteininase (AMPK) (Cell Signaling Technology Inc, Danvers,A) and total AMPK (UBI Upstate Biotechnology Inc,altham, MA), anti-phosphoTyr1068 EGFR and total

GFR (Santa Cruz, Santa Cruz, CA), anti-phosphoSer180/ M

er181 IkB kinase (IKK)�/� and total IKK (Cell Signalingechnology Inc), anti-phosphoThr183/Tyr185 JNK and

otal JNK (Cell Signaling Technology Inc), anti-phosphoer473 Akt and total Akt (Cell Signaling Technology Inc),nti-phosphoSer256 FoxO1 (Cell Signaling Technologync) and total FoxO1 (Santa Cruz), FoxA2 (Santa Cruz),imp3 (Santa Cruz), Socs-3 (Santa Cruz), p65 (Abcam

nc, Cambridge, MA), histone 1 (Santa Cruz), topoisom-rase I (Santa Cruz), and tubulin (Santa Cruz).

Insulin Signaling StudiesInsulin signaling studies with phospho-specific an-

ibodies were performed as previously described.9,12 Briefly,xperiments were performed in overnight-fasted, 8- to 12-eek-old mice. Animals were anesthetized by the intraperi-

oneal administration of sodium pentobarbital (65 mg/kg),nd human insulin (5 U) was injected through the inferiorena cava. Liver and epididymal fat were removed 1 and 4inutes after insulin injection, respectively, and homoge-

ized as previously described.9,12

Gene Expression AnalysisTotal RNA was isolated from fresh adipose tissues

nd liver using Trizol reagents (Invitrogen Corp, Eugene,R). A total of 2 �g total RNA was reverse-transcribed

nto complementary DNA (cDNA) using the High Ca-acity cDNA Archive kit (Applied Biosystems, Fosterity, CA). Quantitative real-time polymerase chain reac-

ion was performed on individual samples from adultice using an ABI PRISM 7700 System and TaqMan

eagents (Applied Biosystems). Each reaction was per-ormed in triplicate using standard reaction conditions.alculations were performed by a comparative cycle

hreshold method: the starting copy number of test sam-les was determined in comparison with the known copyumber of the calibrator sample (delta-delta Cyclehreshold). The relative gene copy number was calculateds 2�ddCtas previously described.9,12 The list of Appliediosystems primers for the 62 genes studied is listed inupplementary Table 1 (see supplementary material on-ine at www.gastrojournal.org).

Adipose Tissue Histologic AnalysisEpididymal fat was obtained from 4-month-old

ice, specimens were fixed in 10% paraformaldehyde, andmbedded in paraffin. Consecutive sections (10 �m) thenere mounted on slides and stained with H&E. MCP-1nd F4/80 immunostaining were performed using Ab-am antibodies following the manufacturer’s instruc-ions. Adipose cell size was calculated with Nationalnstitutes of Health (Bethesda, MD) Image 1.62 softwarey manual tracing of at least 500 adipocytes for eachenotype.

Histology and Quantification of Liver LesionsFormalin-fixed liver tissue was processed, and

-�m–thick paraffin sections were stained with H&E and

asson’s trichrome for histologic analysis. For Oil Red

Owatobae3absw

Sdtel1bss

tsfOcDqiE

a1tt

bs

FrIIctptd

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

February 2009 TIMP3 AND HEPATIC STEATOSIS 665

, staining liver was embedded in OCT. A pathologistho was blinded to the experimental conditions evalu-ted the severity of NAFLD, based on the amount andypes of fat (macrovesicular and microvesicular), extentf inflammation, presence of cell degeneration (acidophilodies, ballooning, and Mallory’s hyaline) or necrosis,nd degree of fibrosis.13 The extent of inflammation wasvaluated on H&E-stained sections and scored from 0 to

as follows: 0, no inflammation; 1, mild; 2, moderate;nd 3, severe, whereas the degree of fibrosis was assessedy digital morphometry on samples stained with Mas-on’s trichrome. MCP-1 and F4/80 immunostainingas performed as described earlier.

Histology and Quantification of Arterial LesionsMice were anesthetized with isoflurane (IsoVet;

chering–Plough SpA, Milano, Italy), euthanized by cervicalislocation, and bled via cardiac puncture. The heart withhe proximal aorta was fixed in 4% paraformaldehyde andmbedded in paraffin. For assessment of atheroscleroticesions we used the method of Paigen et al.17 Briefly, serial0-�m–thick sections were cut from the apex toward thease of the heart. Sections containing the aortic origin,tarting where the aortic leaflets first become visible, weretained with Oil Red O, Weigert–Van Gieson’s elastic stain

igure 1. Effects of short-termegular diet on WT, Timp3�/�,nsr�/�, Insr�/�Timp3�/�, andnsr�/�Tace�/�mice. (A) Glu-ose, insulin, adipokine, andransaminases levels. (B) Intra-eritoneal glucose and insulinolerance tests; n � 5 per group,

ata are mean � SD.

o highlight the medial boundaries of atherosclerotic le-ions, or H&E for routine morphology. Analysis was per-ormed on every other section (n � 10 per mouse). Oil Red

was performed in OCT embedded tissue. Photomicros-opy was performed on a Leittz Dialux microscope (Leitzialux Labequipe Ltd, Markham, Ontario, Canada) and

uantification of atherosclerotic lesions was performed us-ng computer-assisted image analysis (Optimas 6.5; Bioscan,dmunds, WA).

Statistical AnalysisResults of the experimental studies are mean � SD,

s indicated. Statistical analyses were performed using-way analysis of variance (ANOVA) or 2-way ANOVA, orhe unpaired Student t test as indicated. P values of lesshan .05 were considered statistically significant.

ResultsMetabolic Effect of the Combination of GeneticPredisposition to Insulin Resistance WithIncreased TACE Activity and HFDDevelopment and progression of common meta-

olic disorders such as type 2 diabetes and atherosclero-is is dependent on the interaction between environmen-

tpWatmicImnsHtflf

iistTasw

sIhiplTIcH

wIIobptTrw

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

666 MENGHINI ET AL GASTROENTEROLOGY Vol. 136, No. 2

al factors (eg, high calorie intake) and geneticredisposition to insulin resistance and inflammation.e recently observed that the interaction between Insr

nd Timp3, adding up insulin resistance and inflamma-ion, results in glucose intolerance in an age-dependent

anner.9,12 Therefore, we inquired whether the samenteraction may accelerate the development of obesityomplications. At 10 weeks after weaning wild-type (WT),nsr�/�, Timp3�/�, Insr�/�Timp3�/�, and Insr�/�Tace�/�

ice, fed a standard diet, do not show statistically sig-ificant differences in terms of glucose tolerance, insulinensitivity, adipokines, and liver function (Figure 1).9,12

ence, we chose this time point to test the hypothesishat Insr/Timp3 interaction may favor obesity-related in-ammation induced by a diet containing 60% Kcal from

at (HFD).Under the HFD regimen, fasting and fed glucose and

nsulin levels, HOMA IR, and intraperitoneal glucose andnsulin tolerance tests showed that Insr�/�Timp3�/�wereignificantly more glucose intolerant and insulin resis-ant compared with WT, Insr�/�, Timp3�/�, and Insr�/�

ace�/� littermates (Figure 2A–C). However, both Insr�/�

nd Timp3�/� showed mildly increased fed insulin levels,uggesting a state of compensatory hyperinsulinemia,hich maintains their glucose homeostasis at a level

Figure 2. Timp3 deficiency exacerbates metabolic defects caused by iand fed glucose and insulin (*P � .05, ***P � .001 for Insr�/�Timp3�/�

with WT and Insr�/�Tace�/�; 1-way ANOVA, n � 9 per group). (B) intrapcompared with other groups; 2-way ANOVA, n � 9 per group). (C) HOANOVA, n � 9 per group). (D) Adipokines and transaminases (*P � .05,

group). Data are mean � SD.

imilar to WT and Insr�/�Tace�/� (Figure 2A and B).nsr�/�Timp3�/� have significantly lower adiponectin andigher AST, ALT, and leptin levels, suggesting alterations

n liver and white adipose tissue (WAT) functions com-ared with WT, Insr�/�, Timp3�/�, and Insr�/�Tace�/�

ittermates (Figure 2D). The observation that Insr�/�

ace�/� showed lower leptin levels compared withnsr�/�Timp3�/� mice suggests that Tace deficiency canonfer some degree of leptin sensitivity in mice fed anFD (Figure 2D).

Effect of the Combination of GeneticPredisposition to Insulin Resistance WithIncreased TACE Activity and HFD in WATIn WAT, we found that the mean adipocyte area

as higher in Insr�/�Timp3�/� compared with WT,nsr�/�, Timp3�/�, and Insr�/�Tace�/� (P � .01, Figure 3).mmunohistochemical analysis also revealed the presencef more severe inflammatory infiltration, pericellular fi-rosis, MCP-1 expression, and accumulation of macro-hage cells, as estimated by F4/80 immunohistochemis-ry, in Insr�/�Timp3�/� compared with WT, Insr�/�,imp3�/�, and Insr�/�Tace�/�(Figure 3A). Western blotsevealed that insulin-stimulated AKT phosphorylationas decreased markedly in Insr�/�Timp3�/� compared

tion of HFD with predisposition to genetic insulin resistance. (A) Fastingpared with other groups; #P � .05 for Insr�/�and Timp3�/� comparedneal glucose and insulin tolerance test (***P � .001 for Insr�/�Timp3�/�

IR (*P � .05 for Insr�/�Timp3�/� compared with other groups, 1-wayy ANOVA, for Insr�/�Timp3�/� compared with other groups, n � 9 per

nteraccomeritoMA1-wa

wmbI4slTgws(lvi

sHml

praasa(

pc(tlasopcTbpl

Fm(mfFa ed w

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

February 2009 TIMP3 AND HEPATIC STEATOSIS 667

ith WT, Insr�/�, Timp3�/�, and Insr�/�Tace�/� litter-ates (Figure 4A). IKK and Jnk1/2 phosphorylation were

oth increased in Insr�/�Timp3�/� compared with WT,nsr�/�, Timp3�/�, and Insr�/�Tace�/� littermates (FigureA). Quantitative polymerase chain reaction analysis of aelection of 40 genes relevant for glucose/lipid metabo-ism and inflammation (Figure 4B, and supplementaryable 2; see supplementary material online at www.astrojournal.org) revealed that Insr�/�Timp3�/� comparedith WT, Insr�/�, Timp3�/�, and Insr�/�Tace�/� showed

ignificantly increased expression of macrophage markersCD68), markers of macrophage M1 state (CCR2, inter-eukin-6), cytokines (MCP-1, CXCL16), and genes in-olved in leptin resistance (suppressor of cytokine signal-ng 3 [SOCS-3], stearoyl CoA desaturase 1 [SCD-1]).

Combination of Genetic Insulin ResistanceWith Increased TACE Activity and HFDInduces Inflammation and Hepatic SteatosisBecause of the role of TNF and the adipokine

ystem in NAFLD,14 we investigated liver histology underFD conditions. We found that only Insr�/�Timp3�/�

ice developed mixed microvesicular and macrovesicu-

igure 3. Interaction between predisposition to genetic insulin resiorphologic characteristics. (A) WAT tissue morphology from WT (left co

100�); Insr�/�Timp3�/� showed increased inflammatory infiltrate andagnifications); trichrome (100�), Insr�/�Timp3�/� showed increased

rom the dotted area (250�); MCP-1 immunostaining (250�) is presen4/80 immunostaining (400�), Insr�/�Timp3�/� showed increased mdipocyte area (**P � .01, 1-way ANOVA, for Insr�/�Timp3�/� compar

ar steatosis, lobular and periportal inflammation, he- 7

atocellular ballooning, and perisinusoidal fibrosis,eminiscent of grade 3 NAFLD,18 with increased MCP-1nd F4/80 expression (Figure 5). Conversely, WT, Insr�/�,nd Timp3�/� mice were affected only by microvesicularteatosis but no inflammation as in grade 1 (Figure 5),nd Insr�/�Tace�/� mice were between grades 0 and 113

Figure 5).Insulin-stimulated AktSer473 and FoxO1Ser256 phos-

horylation were decreased markedly in Insr�/�Timp3�/�

ompared with WT, Insr�/�, Timp3�/�, and Insr�/�Tace�/�

Figure 6). On the contrary, we found by Western blot-ing nuclear vs cytosolic liver extracts that FoxA2 wasocalized mainly in the nucleus, both in basal conditionsnd after insulin stimulation (supplementary Figure 1;ee supplementary material online at www.gastrojournal.rg). Western blot analysis showed significantly increasedhosphorylation of JNK, IKK, and EGFR in Insr�/�Timp3�/�

ompared with WT, Insr�/�, Timp3�/�, and Insr�/�

ace�/�, which may take part in inflammatory and fi-rotic processes, and a significant decrease in AMPKhosphorylation, which may contribute to increased

ipid content via reduction of fatty acid oxidation (Figure

e and HFD with Timp3 deficiency promotes inflammation in WAT:), Insr�/�, Timp3�/�Insr�/�Timp3�/�, and Insr�/�Tace�/�: H&E stainingellular fibrosis as suggested by a particular of the dotted area (higheren staining and dilated interstitial spaces as suggested by a particularsr�/�Timp3�/� but not in WT, Insr�/�, Timp3�/�, and Insr�/�Tace�/�;hage accumulation as suggested by a particular (1000�). (B) Meanith other groups). Data are mean � SD.

stanclumnperic

collagt in Inacrop

). Unbalanced inflammation in Insr�/�Timp3�/� com-

pcif

t6flcigecFC(m

dImmdTIRm

maepiecthtcifirtpmmtsWs

FtIioTm

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

668 MENGHINI ET AL GASTROENTEROLOGY Vol. 136, No. 2

ared with WT, Insr�/�, Timp3�/�, and Insr�/�Tace�/� isonfirmed both by increased expression of Socs-3 and byncreased nuclear localization of p65 subunit of nuclearactor-� B transcription factor (Figure 7).

Analysis by quantitative polymerase chain reaction ofhe messenger RNA (mRNA) expression of a selection of2 relevant genes in glucose/lipid metabolism and in-ammation revealed that liver from Insr�/�Timp3�/�

ompared with WT, Insr�/�, Timp3�/�, and Insr�/�Tace�/�

s characterized by significantly different expressions ofenes involved in lipid metabolism (peroxisome prolif-rator-activated receptor � [PPAR�], CD36, SCD-1), glu-ose metabolism (Akt2, PGC1a, Nur77, Nurr1, NOR1,bp1), and inflammation (SOCS-3, CD68, CX3CR1,XCL16, TNF�, inducible nitric oxide [NO] synthase)

Figure 8, and supplementary Table 2; see supplementaryaterial online at www.gastrojournal.org).Because fatty liver disease is emerging as an indepen-

ent risk factor for atherosclerosis,18 we inquired whethernsr�/�Timp3�/� were more prone to lipid deposition in

acrovascular tissues such as aorta. Histologic andRNA analysis confirmed the presence of increased lipid

eposition and inflammation only in aorta from Insr�/�

imp3�/� compared with WT, Insr�/�, Timp3�/�, andnsr�/�Tace�/� WT or Insr�/�Tace�/� mice (supplementaryesults section and supplementary Figure 2; see supple-

igure 4. Interaction between predisposition to genetic insulin resistanissue: molecular aspects. (A) Insulin-stimulated phosphorylation of AKTnsr�/�Tace�/� fed an HFD (**P � .01, 1-way ANOVA, for Insr�/�Timp3�

n Insr�/�Timp3�/� compared with other groups (**P � .01, 1-way ANOf MCP-1, CCR2, CD68, interleukin-6, CXCL16, RbP4, ADRP, SCD-1imp3�/�, and Insr�/�Tace�/� (*P � .05, **P � .01, 1-way ANOVA, foean � SD.

entary material online at www.gastrojournal.org). a

DiscussionTimp3 is an extracellular negative regulator of

etalloproteinases 2 and 9, of sheddases such as TACE,nd of cell-surface receptors such as EGFR and vascularndothelial growth factor receptor 2,9 –12 potentially im-acting processes such as inflammation, tissue remodel-

ng, fibrosis, insulin sensitivity, and angiogenesis, all rel-vant for the onset and the progression of obesityomplications.9 –12,19 –25 Because Timp3 modulates the ac-ivation of inflammatory pathways that impair metabolicomeostasis in the Insr-haploinsufficient mice,9 we inves-igated whether and how Insr/Timp3 concomitant defi-iencies cooperate during the systemic response to diet-nduced obesity caused by a short-term increased dietaryat regimen. In the Insr-haploinsufficient mouse model,n which the onset of diabetes is dependent from aging-elated genetic modifiers, including Timp3,9 the imposi-ion of a short-term HFD results only in a modest com-ensatory hyperinsulinemic state that is sufficient toaintain glucose homeostasis. Thus, the absence of overtetabolic toxicity explains why Insr�/�mice at this early

ime point of HFD do not show alterations in insulinignaling at the level of Akt and FoxO1 as compared with

T littermates. Likewise, Timp3 deficiency alone is notufficient to deteriorate glucose homeostasis and insulin

d HFD with Timp3 deficiency promotes inflammation in white adiposeduced in Insr�/�Timp3�/� compared with WT, Insr�/�, Timp3�/�, andmpared with other groups); JNK and IKK phosphorylation are increasedr Insr�/�Timp3�/� compared with other groups). (B) mRNA expressionSOCS-3 is increased in Insr�/�Timp3�/� compared with WT, Insr�/�,/�Timp3�/� compared with other groups); n � 5 per group. Data are

ce anis re

/� coVA, fo, andr Insr�

ction in comparison with WT mice, when both are

sHTic

chekw

ae

TstflaoarwssmwrflsmrpnmIgdbbdcfisTsads

FihTsdsslpa

FoIWD

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

February 2009 TIMP3 AND HEPATIC STEATOSIS 669

ubjected to a short-term environmental insult such asFD. Nonetheless, similarly to Insr haploinsufficiency,

imp3 deficiency combined with HFD causes a modestncrease in fed insulin levels, suggestive of a state ofompensatory hyperinsulinemia.

By contrast, the synergy between high-fat intake withoncomitant absence of Timp3 and Insr resulted on oneand in an exacerbated inflammatory response, as madevident by the accumulation of inflammatory cells, cyto-ines, tissue fibrosis, and up-regulation of signaling path-ays involved in inflammation in both liver and WAT,

igure 5. Timp3 deficiency induces fatty liver disease in mice combin-ng defects from HFD and predisposition to genetic insulin resistance:istologic aspects. Liver tissue H&E staining (250�) from WT, Insr�/�,imp3�/�, Insr�/�Timp3�/�, and Insr�/�Tace�/�; Insr�/�Timp3�/�

howed mixed microvesicular and macrovesicular steatosis; arrow in-icates ballooning degeneration in Insr�/�Timp3�/� mice. Oil Red Otaining (20�), trichrome (400�), MCP-1 (250�), and F4/80 (250�)taining; Insr�/�Timp3�/� mice showed perisinusoidal fibrosis (dotted

ine) and increased MCP-1 and F4/80 expression. CV, central vein; P,ortal region. Representative images are shown, 5 mice per group werenalyzed.

igure 6. Insulin-signaling in livers from WT, Timp3�/�, Insr�/�, Insr�/�Tf AKT Ser 473 is reduced in Insr�/�Timp3�/� compared with WT, Insr�

nsr�/�Timp3�/� compared with other groups). Insulin-stimulated phosT, Insr�/�, Timp3�/�, and Insr�/�Tace�/� fed an HFD (*P � .05, 1-way

ata are mean � SD.

nd on the other side in hyperglycemia, hyperinsulin-mia, and hepatic steatosis.

To get an integrated view of the role of Timp3 andACE in metabolic disease, we performed gene expres-

ion studies on different sets of genes that are involved inhe regulation of glucose metabolism, lipogenesis, in-ammation, and mitochondrial function. In WAT, thebsence of Timp3 and Insr resulted in the accumulationf F4/80-positive cells and increased expression of MCP-1nd CCR2, both recently shown to be involved in theecruitment of M1 activated macrophages.7,8 However,e did not find differences in the WAT-specific expres-

ion of several other M1 and M2 markers in our modelsuch as Mgl1, Mgl2, and inducible NO synthase (supple-

entary Table 2; see supplementary material online atww.gastrojournal.org), suggesting that Timp3 down-

egulation may be part of a general mechanism for in-ammatory cell accumulation in injured tissues. This isupported by the nearly opposite phenotype shown by

ice heterozygous for both Insr and Tace, signifying thategulation of TNF-� shedding may initiate paracrineropagation of inflammatory and insulin resistance sig-als, which then spread via other TNF-independentechanisms, such as interleukin 6, CXCL16, and Rbp4.25

n the liver, the combination of diet-induced defects withenetic predisposition to insulin resistance and Timp3eficiency resulted in a complex phenotype closely resem-ling more severe (grades 2 and 3) NAFLD, as evidencedy the presence of macrovesicular steatosis, ballooningegeneration, and an increase in transaminases levels, inombination with up-regulation of inflammatory andbrosis signals including activation of signaling cluesuch as p65 nuclear translocation, JNK, IKK, and EGFR.he metabolic derangement, hyperglycemia, and steato-

is, observed by combining diet-induced obesity to Insrnd Timp3 deficiency, can be the result of simultaneousefects such as altered regulation of AKT and its down-tream effectors FoxO1 and FoxA2.26 –28 Interestingly, we

�/�, and Insr�/�Tace�/� fed an HFD. Insulin-stimulated phosphorylationmp3�/�, and Insr�/�Tace�/� fed an HFD (**P � .01, 1-way ANOVA, forlation of FoxO1Ser256 is reduced in Insr�/�Timp3�/� compared with

VA, for Insr�/�Timp3�/� compared with other groups), n � 5 per group.

imp3/�, TiphoryANO

omrfimlarspecTSacliatAnbdiiltTds

lA

sTmkituIcsiaHpimtpJWlIptaoat

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

670 MENGHINI ET AL GASTROENTEROLOGY Vol. 136, No. 2

bserved deficient insulin-dependent control of FoxO1 inice lacking Insr and Timp3, supporting its role in

egulation of hepatic steatosis.29 On the contrary, weound that FoxA2 was localized in the nucleus, support-ng previous findings from other laboratories.30,31 Other

echanisms are involved to explain increased hepaticipogenesis and gluconeogesis are the up-regulation ofdiprotein genes by PPARgamma, as well as the up-egulation of NR4A transcription factors and their sub-trates such as Fbp1.31–35 Interestingly, it was shownreviously that Timp3 deficiency can induce PPAR� over-xpression in a model of liver regeneration.19 Anotherommon defect observed in both liver and WAT ofimp3-deficient mice was the increased expression ofOCS-3 and SCD-1, an enzyme involved in leptin-medi-ted weight loss.36 It has been shown that SCD-1 defi-iency or inhibition protects from diet-induced obesity,iver steatosis, and hyperglycemia via several mechanisms,ncluding regulation of AMPK phosphorylation in livernd other unidentified mechanisms in WAT.35– 40 Consis-ently, we found that in liver of Insr�/�Timp3�/� mice,MPK phosphorylation was down-regulated. The combi-ation of increased SOCS-3 and SCD-1 raises the possi-ility that leptin resistance may be exacerbated by Timp3eficiency. SOCS-3 is known to induce leptin resistance

n the central nervous system and in peripheral tissuesncluding WAT and liver,41 which ensues in the loss ofeptin repression of SCD-1 promoter.37– 40 Another genehat we found up-regulated in both WT and Insr�/�

imp3�/� compared with Insr�/�Tace�/� mice is adiposeifferentiation related protein (supplementary Table 2;

ee supplementary material online at www.gastrojourna- e

.org). Recently, it has been suggested that inhibition ofDRP is a target for NAFLD treatment.42

Our gene expression study in WAT revealed that genesignificantly up-regulated or down-regulated in the Insr�/�

imp3�/�mice generally are linked to the control ofacrophage function, fatty acid metabolism, and adipo-

ine resistance.7,8 In liver, we found similar genes point-ng to innate immunity activation as a major trigger forhe phenotype observed in Insr�/�Timp3�/�mice. A stim-lating result, however, was the finding that combined

nsr/Timp3 deficiency does not alter expression of geneslassically involved in gluconeogenesis and lipogenesisuch as PEPCK, LXRs, SREBP, or ChREBP. Althoughncreased PPAR� expression may in part explain the ste-tosis observed in liver of Insr�/�Timp3�/� mice fed anFD,33,34 the molecular mechanisms underlying the hy-erglycemic state may involve several factors. Because

nsulin signaling is more defective in Insr�/�Timp3�/�

ice than in other groups, it is plausible that inflamma-ory mechanisms impairing activation of intracellularathways are involved. By this point of view, increased

NK and IKK signaling may explain insulin resistance inAT and liver.1– 6 However, the observation of increased

evels of the NR4A family of transcription factors innsr�/�Timp3�/� mice fed an HFD is another intriguinglausible explanation because it may link the inflamma-ory signals dependent from excessive TACE activation tolteration of glucose metabolism. NR4A differently fromther nuclear receptors generally primarily are controlledt the level of protein expression and/or posttransla-ional modifications, but they are induced by multiple

Figure 7. Timp3 deficiency ex-acerbates activation of inflam-matory pathways linked to ste-atosis in mice combining defectsfrom HFD and predisposition togenetic insulin resistance. Insr�/�

Timp3�/� compared with WT,Insr�/�, Timp3�/�, and Insr�/�

Tace�/� fed an HFD showed de-creased phosphorylation ofAMPK, increased phosphoryla-tion of JNK, IKK�/�, EGFR, andSOCS-3 expression and in-creased nuclear translocationp65 subunit of nuclear factor � B(*P � .05, ***P � .001, 1-wayANOVA, for Insr�/�Timp3�/�

compared with other groups;n � 3 per group).

xtracellular signals in a cell- and tissue-specific man-

nvgfthmThfalr

Nadrdfces

bg

liPmi

aG1

1

1

1

1

1

1

1

1

FdmNcICcIc

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

February 2009 TIMP3 AND HEPATIC STEATOSIS 671

er.35 Timp3 regulated factors such as TNF-�, EGF, andascular endothelial growth factor are activators of NR4Aene expression.43– 46 Induction of NR4A transcriptionactors increases hepatic gluconeogenesis via Fbp1 ratherhan PEPCK,35 providing a further explanation for theyperglycemic phenotype observed in Insr�/�Timp3�/�

ice fed an HFD. From our data it appears that theimp3/Tace dyad may be involved in the generation ofepatic signals leading to increased NR4A transcription

actor expression. Because inflammatory mediators, cyclicdenosine monophosphate, fasting, and glucagon regu-ate NR4A expression,35,46 the role of Timp3 and TACEequires further investigation.

Some recent clinical studies have proposed thatAFLD may play a role as an independent risk factor for

therosclerosis.21 Although this study was not targeted toefine the role of fatty liver in the context of atheroscle-osis, we observed that Insr�/�Timp3�/� mice tend toevelop fatty streaks in the aortic root. Moreover, aortarom the Insr�/�Timp3�/� mice were more inflamed inomparison with WT and Insr�/�Tace�/� mice. We cannotxclude, however, that these findings are secondary to theevere hyperglycemic state observed in Insr�/�Timp3�/� mice.

In conclusion, we propose that loss of Timp3 in com-ination with predisposing genetic and dietary back-

igure 8. Expression of genes linked to steatosis in mice combiningefects from HFD and predisposition to genetic insulin resistance. (A)RNA expression of metabolic genes PPAR�, CD36, PGC1�, Nur77,urr1, NOR1, Fbp1, and SCD-1 are increased whereas Akt2 is de-reased in Insr�/�Timp3�/� compared with WT, Insr�/�, Timp3�/�, and

nsr�/�Tace�/�. (B) mRNA expression of inflammatory genes SOCS-3,D68, CX3CR1, CXCL16, TNF�, and inducible NO synthase are in-reased in Insr�/�Timp3�/� compared with WT, Insr�/�, Timp3�/�, and

nsr�/�Tace�/� (*P � .05, **P � .01, 1-way ANOVA for Insr�/�Timp3�/�

ompared with other groups); n � 5 per group. Data are mean � SD.

rounds compromises metabolic adaptation in WAT and

iver in part via up-regulation of SOCS-3 and SCD-1 andn part involving up-regulation of nuclear receptorsPAR� and NR4A. Based on these findings in animalodels we propose that the role of Timp3 should be

nvestigated in human patients affected by NAFLD.

Supplementary Data

Note: To access the supplementary materialccompanying this article, visit the online version ofastroenterology at www.gastrojournal.org, and at doi:0.1053/j.gastro.2008.10.079.

References

1. Hotamisligil GS. Inflammation and metabolic disorders. Nature2006;444:860–867.

2. Qatanani M, Lazar MA. Mechanisms of obesity-associated insulinresistance: many choices on the menu. Genes Dev 2007;21:1443–1455.

3. Shoelson SE, Lee J, Goldfine AB. Inflammation and insulin resis-tance. J Clin Invest 2006;116:1793–1801.

4. Sanyal AJ. Mechanisms of disease: pathogenesis of nonalcoholicfatty liver disease. Nat Clin Pract Gastroenterol Hepatol 2005;2:46–53.

5. Perlemuter G, Bigorgne A, Cassard-Doulcier AM, et al. Nonalco-holic fatty liver disease: from pathogenesis to patient care. NatClin Pract Endocrinol Metab 2007;3:458–469.

6. Ahima RS. Insulin resistance: cause or consequence of nonalco-holic steatohepatitis? Gastroenterology 2007;132:444–446.

7. Kanda H, Tateya S, Tamori Y, et al. MCP-1 contributes to macro-phage infiltration into adipose tissue, insulin resistance, andhepatic steatosis in obesity. J Clin Invest 2006;116:1494–1505.

8. Weisberg SP, Hunter D, Huber R, et al. CCR2 modulates inflam-matory and metabolic effects of high-fat feeding. J Clin Invest2006;116:115–124.

9. Federici M, Hribal ML, Menghini R, et al. Timp3 deficiency ininsulin receptor-haploinsufficient mice promotes diabetes andvascular inflammation via increased TNF-alpha. J Clin Invest2005;115:3494–3505.

0. Murphy G, Murthy A, Khokha R. Clipping, shedding and RIPpingkeep immunity on cue. Trends Immunol 2008;29:75–82.

1. Qi JH, Ebrahem Q, Moore N, et al. A novel function for tissueinhibitor of metalloproteinases-3 (TIMP3): inhibition of angiogen-esis by blockage of VEGF binding to VEGF receptor-2. Nat Med2003;9:407–415.

2. Serino M, Menghini R, Fiorentino L, et al. Mice heterozygous fortumor necrosis factor-alpha converting enzyme are protectedfrom obesity-induced insulin resistance and diabetes. Diabetes2007;56:2541–2546.

3. Neuschwander-Tetri BA, Caldwell SH. Nonalcoholic steatohepati-tis: summary of an AASLD Single Topic Conference. Hepatology2003;37:1202–1219.

4. Tilg H, Hotamisligil GS. Nonalcoholic fatty liver disease: cytokine-adipokine interplay and regulation of insulin resistance. Gastro-enterology 2006;131:934–945.

5. Menghini R, Marchetti V, Cardellini M, et al. Phosphorylation ofGATA2 by Akt increases adipose tissue differentiation and reducesadipose tissue-related inflammation: a novel pathway linking obesityto atherosclerosis. Circulation 2005;111:1946–1953.

6. Marchetti V, Menghini R, Rizza S, et al. Benfotiamine counteractsglucose toxicity effects on endothelial progenitor cell differentia-tion via Akt/FoxO signaling. Diabetes 2006;55:2231–2237.

7. Paigen B, Morrow A, Holmes PA, et al. Quantitative assessmentof atherosclerotic lesions in mice. Atherosclerosis 1987;68:

231–240.

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

I07

(

bGv

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

672 MENGHINI ET AL GASTROENTEROLOGY Vol. 136, No. 2

8. Brea A, Mosquera D, Martin E, et al. Nonalcoholic fatty liverdisease is associated with carotid atherosclerosis: a case-con-trol study. Arterioscler Thromb Vasc Biol 2005;25:1045–1050.

9. Mohammed FF, Smookler DS, Taylor SE, et al. Abnormal TNFactivity in Timp3�/� mice leads to chronic hepatic inflammationand failure of liver regeneration. Nat Genet 2004;36:969–977.

0. Lautrette A, Li S, Alili R, et al. Angiotensin II and EGF receptorcross-talk in chronic kidney diseases: a new therapeutic ap-proach. Nat Med 2005;11:867–874.

1. Saunders WB, Bohnsack BL, Faske JB, et al. Coregulation ofvascular tube stabilization by endothelial cell TIMP-2 and pericyteTIMP-3. J Cell Biol 2006;175:179–191.

2. Federici M, Menghini R, Mauriello A, et al. Insulin-dependentactivation of endothelial nitric oxide synthase is impaired byO-linked glycosylation modification of signaling proteins in humancoronary endothelial cells. Circulation 2002;106:466–472.

3. Browning JD, Horton JD. Molecular mediators of hepatic steato-sis and liver injury. J Clin Invest 2004;114:147–152.

4. Cao Y. Angiogenesis modulates adipogenesis and obesity. J ClinInvest 2007;117:2362–2368.

5. Gustafson B, Hammarstedt A, Andersson CX, et al. Inflamedadipose tissue: a culprit underlying the metabolic syndromeand atherosclerosis. Arterioscler Thromb Vasc Biol 2007;27:2276–2283.

6. Matsumoto M, Han S, Kitamura T, et al. Dual role of transcriptionfactor FoxO1 in controlling hepatic insulin sensitivity and lipidmetabolism. J Clin Invest 2006;116:2464–2472.

7. Kamagate A, Qu S, Perdomo G, et al. FoxO1 mediates insulin-dependent regulation of hepatic VLDL production in mice. J ClinInvest 2008;118:2347–2364.

8. Wolfrum C, Asilmaz E, Luca E, et al. Foxa2 regulates lipid metab-olism and ketogenesis in the liver during fasting and in diabetes.Nature 2004;432:1027–1032.

9. Valenti L, Rametta R, Dongiovanni P, et al. Increased expressionand activity of the transcription factor FOXO1 in nonalcoholicsteatohepatitis. Diabetes 2008;57:1355–1362.

0. Zhang L, Rubins NE, Ahima RS, et al. Foxa2 integrates thetranscriptional response of the hepatocyte to fasting. Cell Metab2005;2:141–148.

1. Steneberg P, Rubins N, Bartoov-Shifman R, et al. The FFA recep-tor GPR40 links hyperinsulinemia, hepatic steatosis, and im-paired glucose homeostasis in mouse. Cell Metab 2005;1:245–258.

2. Yu S, Matsusue K, Kashireddy P, et al. Adipocyte-specific geneexpression and adipogenic steatosis in the mouse liver dueto peroxisome proliferator-activated receptor gamma1 (PPAR-gamma1) overexpression. J Biol Chem 2003;278:498–505.

3. Zhang YL, Hernandez-Ono A, Siri P, et al. Aberrant hepatic ex-pression of PPARgamma2 stimulates hepatic lipogenesis in amouse model of obesity, insulin resistance, dyslipidemia, andhepatic steatosis. J Biol Chem 2006;281:37603–37615.

4. Pei L, Waki H, Vaitheesvaran B, et al. NR4A orphan nuclearreceptors are transcriptional regulators of hepatic glucose me-

tabolism. Nat Med 2006;12:1048–1055. 2

5. Cohen P, Miyazaki M, Socci ND, et al. Role for stearoyl-CoAdesaturase-1 in leptin-mediated weight loss. Science 2002;297:240–243.

6. Dobrzyn P, Dobrzyn A, Miyazaki M, et al. Stearoyl-CoA desaturase1 deficiency increases fatty acid oxidation by activating AMP-activated protein kinase in liver. Proc Natl Acad Sci U S A 2004;101:6409–6414.

7. Ntambi JM, Miyazaki M, Stoehr JP, et al. Loss of stearoyl-CoAdesaturase-1 function protects mice against adiposity. Proc NatlAcad Sci U S A 2002;99:11482–11486.

8. Flowers MT, Miyazaki M, Liu X, et al. Probing the role of stearoyl-CoA desaturase-1 in hepatic insulin resistance. J Clin Invest2006;116:1478–1481.

9. Jiang G, Li Z, Liu F, et al. Prevention of obesity in mice byantisense oligonucleotide inhibitors of stearoyl-CoA desatu-rase-1. J Clin Invest 2005;115:1030–1038.

0. Gutierrez-Juarez R, Pocai A, Mulas C, et al. Critical role ofstearoyl-CoA desaturase-1 (SCD1) in the onset of diet-inducedhepatic insulin resistance. J Clin Invest 2006;116:1686–1695.

1. Howard JK, Cave BJ, Oksanen LJ, et al. Enhanced leptin sensi-tivity and attenuation of diet-induced obesity in mice with haplo-insufficiency of Socs3. Nat Med 2004;10:734–738.

2. Imai Y, Varela GM, Jackson MB, et al. Reduction of hepatoste-atosis and lipid levels by an adipose differentiation-related pro-tein antisense oligonucleotide. Gastroenterology 2007;132:1947–1954.

3. Suzuki S, Suzuki N, Mirtsos C, et al. Nur77 as a survival factor intumor necrosis factor signaling. Proc Natl Acad Sci U S A 2003;100:8276–8280.

4. Kolluri SK, Bruey-Sedano N, Cao X, et al. Mitogenic effect oforphan receptor TR3 and its regulation by MEKK1 in lung cancercells. Mol Cell Biol 2003;23:8651–8667.

5. Zeng H, Qin L, Zhao D, et al. Orphan nuclear receptor TR3/Nur77regulates VEGF-A-induced angiogenesis through its transcrip-tional activity. J Exp Med 2006;203:719–729.

6. Pei L, Castrillo A, Tontonoz P. Regulation of macrophage inflam-matory gene expression by the orphan nuclear receptor Nur77.Mol Endocrinol 2006;20:786–794.

Received March 11, 2008. Accepted October 30, 2008.Address requests for reprints to: Massimo Federici, MD, Department of

nternal Medicine, University of Rome “Tor Vergata,” Via Montpellier 1,0133 Rome, Italy. e-mail: federicm@uniroma2.it; fax: (39) 06-2596890.The authors thank Domenico Accili for the Insr�/� mice and Amgen

Thousand Oaks, CA) for the Tace�/� mice.The authors disclose the following: This work was supported in part

y the Società Italiana di Diabetologia Research Grant 2007, Telethonrant GGP08065, Italian Ministry of University grant PRIN 2006, Ju-enile Diabetes Research grant 1-2007-665 (all to M.F.), and ASI grant

006 (to R.L.).

iqtlTf

pporTisaifp

February 2009 TIMP3 AND HEPATIC STEATOSIS 672.e1

Supplementary Materials

Results

Effect of HFD in Vascular Tissues ofInsr�/�Timp3�/�

Because fatty liver disease is emerging as anndependent risk factor for atherosclerosis,18 we in-uired whether Insr�/�Timp3�/� mice were more proneo lipid deposition in macrovascular and microvascu-ar tissues such as aorta and kidney. WT, Insr�/�,imp3�/�, and Insr�/�Tace�/� mice were totally negative

or any gross sign of early atherosclerosis and macro- I

hage infiltration as measured by F4/80 staining (sup-lementary Figure 2A, n � 5 per group). Instead, webserved the presence of several fatty streaks and mac-ophage infiltration at the aortic root in Insr�/�

imp3�/� HFD mice as measured by Oil Red O stain-ng, F4/80 staining, and plaque morphometry in H&Etaining (mean plaque area 16,190.5 � 543.8 �m2/orta cross-section) (supplementary Figure 2B). Stain-ng with antibodies specific for MCP-1 and for receptoror advanced glycated end-products also markedly wasositive in Insr�/�Timp3�/� HFD mice compared with

nsr�/�Tace�/� and WT mice.

S

CCCCCCCCITSSSVILMMCN

LLPPPSCCNNNFFPH

672.e2 MENGHINI ET AL GASTROENTEROLOGY Vol. 136, No. 2

upplementary Table 1. List of Applied Biosystems Primers

Gene:inflammation

Assay ID(Applied Biosystem

D68 Mm 00839636_gD14 Mm 00438094_gCL2 (MCP1) Mm 00441242_mCL7 (MCP3) Mm 00443113_mCR2 Mm 99999051_gX3CR1 Mm 02620111_sX3CL1 Mm 00436454_mXCL16 Mm 00469712_m

nterleukin 6 Mm 00446190_mNF� Mm 99999068_mOCS-1 Mm 00782550_sOCS-2 Mm 00850544_gOCS-3 Mm 00545913_sCAM Mm 00449197_mCAM Mm 00516023_mY6 Mm 00812620_mgl1 Mm 00546124_mgl2 Mm 00460844_msf1R Mm 00432689_mO synthase 2 (inducible NO synthase) Mm 00440485_m

Transcription factors

XR� Mm 00443454_m1XR� Mm 00437262_m1PAR� Mm 00440939_m1PAR Mm 00440945_m1PAR Mm 00803186_g1rebf1 (Srebp) Mm 00550338_m1EBP� Mm 00514283_s1EBP� Mm 00843434_s1r4a1 (Nur77) Mm 00439358_m1r4a2 (Nurr1) Mm 00443086_m1r4a3 (NOR1) Mm 00450074_m1OXO1 Mm 00490672_m1OXA2 Mm 00839704_mHpargc1� (PGC1�) Mm 00447183_m1ousekeeping geneAct1b Mm 00607939_s1

s) Gene: metabolismASSAY ID

(Applied Biosystems)

1 SCD1 Mm 00772290_m11 Fasn (FAS) Mm 00662319_m11 CD36 Mm 00432403_m11 FABP4 Mm 00445880_m1

H SLC27A1 (FATP1) Mm 00449511_m11 Cpt1a Mm 00550438_m11 Adfp (Adrp1) Mm 00475794_m11 Leptin Mm 00434759_m11 Adipoq (Adiponectin) Mm 00456425_m11 Rbp4 Mm 00803266_m11 Retn (Resistin) Mm 00445641_m11 Dlk1 (Pref1) Mm 00494477_m11 IRS1 Mm 00439720_s11 G6pc (G6Pase) Mm 00839363_m11 Fbp1 Mm 00490181_m11 Fbp2 Mm 00484280_m11 Eno3 Mm 00468267_m11 PCK1 (PEPCK) Mm 00440636_m11 Mlxipl (ChREBP) Mm 00498811_m11 Slc2a2 (Glut2) Mm 00446224_m1

Slc2a4 (Glut4) Akt2 ACOX1 Mm 00436615_m1Mm 00545827_m1Mm 00443579_m1

Mitochondrial function

Err� Mm 00433143_m1Nrf1 Mm 00447996_m1Tfam Mm 00447485_m1UCP1 Mm 00494069_m1

S

I

M

T

S

M

L

a

February 2009 TIMP3 AND HEPATIC STEATOSIS 672.e3

upplementary Table 2. Other genes measured by qPCR

WAT WT Insr�/�/Timp3�/� Insr�/�/Tace�/�

nflammationCD14 3.02 � 2.86 2.30 � 0.83 2.10 � 2.18CCL7 (MCP3) 2.86 � 2.63 0.53 � 0.43 0.62 � 0.09CX3CR1 1.10 � 0.15 7.33 � 0.47 1.64 � 1.53TNF� 0.63 � 0.52 0.75 � 0.31 1.03 � 0.92Mgl1 33.29 � 45.66 48.06 � 53.30 7.26 � 9.20Mgl2 4.42 � 4.84 9.64 � 5.26 3.30 � 4.25NO synthase 2 (inducible NO synthase) 3.49 � 3.52 7.59 � 6.95 1.86 � 1.79etabolismCD36 2.00 � 1.41 0.92 � 0.81 0.71 � 0.78FATP1 2.99 � 2.82 0.54 � 0.17 2.15 � 1.47Leptin 0.53 � 0.66 1.93 � 1.54 0.52 � 0.27Adipoq (adiponectin) 2.23 � 1.74 5.72 � 5.50 1.85 � 0.83Dlk1 (Pref1) 1.32 � 0.45 1.66 � 1.47 5.88 � 7.23IRS1 3.35 � 3.32 1.34 � 0.61 1.02 � 0.11Slc2a4 (Glut4) 1.84 � 1.19 3.90 � 4.86 1.03 � 0.61Retn (Resistin) 1.59 � 0.83 2.31 � 2.10 0.66 � 0.15

ranscription factorsPPAR� 4.18 � 3.75 1.77 � 0.79 1.43 � 0.97PPAR� 1.30 � 0.42 6.31 � 3.04 0.80 � 0.13

rebf1 (Srebp) 7.86 � 9.70 7.88 � 9.22 4.13 � 1.09CEBP� 11.02 � 14.17 7.34 � 2.88 2.41 � 0.12Ppargc1� (PGC1�) 7.78 � 9.59 0.84 � 0.33 1.37 � 0.76itochondrial functionErr� 0.53 � 0.66 0.03 � 0.01 0.28 � 0.39Nrf1 0.78 � 0.31 0.28 � 0.08 0.35 � 0.08Tfam 0.56 � 0.62 0.04 � 0.03 0.08 � 0.08UCP1 0.90 � 0.14 0.04 � 0.01 1.03 � 1.66

iverInflammation

CD14 0.42 � 0.51 1.09 � 0.49 0.39 � 0.30CCL7 (MCP3) 1.36 � 0.31 0.92 � 0.71 0.55 � 0.28CCR2 0.76 � 0.36 0.56 � 0.21 2.36 � 2.28Interleukin 6 0.70 � 0.31 0.51 � 0.17 1.04 � 0.87SOCS-1 1.87 � 1.01 1.99 � 0.99 0.48 � 0.22SOCS-2 1.06 � 0.18 0.45 � 0.02 0.51 � 0.26VCAM1 0.59 � 0.37 3.72 � 2.89 0.91 � 0.26ICAM1 0.83 � 0.24 0.77 � 0.09 0.68 � 0.37Csf1R 0.81 � 0.48 0.80 � 0.02 0.43 � 0.10

MetabolismFasn (FAS) 0.88 � 0.14 1.29 � 0.17 1.51 � 0.65FABP4 1.13 � 0.19 2.11 � 1.34 2.46 � 2.30SLC27A1 (FATP1) 1.73 � 0.70 1.22 � 0.30 2.05 � 0.63Adfp (Adrp1) 0.52 � 0.36 1.38 � 0.36 0.62 � 0.23a

Adipoq (adiponectin) 5.87 � 7.51 2.69 � 0.50 2.16 � 0.36Eno3 0.73 � 0.38 0.85 � 0.36 0.80 � 0.14IRS1 3.73 � 2.71 1.74 � 0.68 1.49 � 0.47G6pc (G6Pase) 0.57 � 0.36 0.29 � 0.15 0.75 � 0.21Fbp2 1.45 � 1.10 2.51 � 1.20 2.91 � 1.41PCK1 (PEPCK) 1.51 � 0.44 0.77 � 0.18 1.24 � 0.36Mlxipl (ChREBP) 1.55 � 0.70 2.68 � 0.88 2.34 � 1.13Slc2a2 (Glut2) 1.73 � 1.24 2.14 � 0.50 3.02 � 1.59ACOX1 0.87 � 0.11 0.65 � 0.17 0.98 � 0.38

Transcription factorsLXR� 2.00 � 1.09 2.88 � 0.83 1.19 � 0.63LXR� 0.63 � 0.52 0.34 � 0.22 0.79 � 0.58PPAR� 1.09 � 0.58 2.75 � 0.34 2.25 � 1.09PPAR� 0.62 � 0.35 0.97 � 0.24 0.60 � 0.12Srebf1 (Srebp) 0.84 � 0.17 0.66 � 0.08 0.69 � 0.16CEBP� 0.56 � 0.47 1.37 � 0.03 0.88 � 0.63FOXO1 0.66 � 0.36 1.29 � 0.09 2.20 � 1.47FOXA2 1.53 � 0.61 0.94 � 0.09 1.38 � 0.09

P � .05 for Insr�/�Timp3�/� compared with WT and Insr�/�Tace�/�, 1-way ANOVA.

Spf2

S(wF�

IleI

672.e4 MENGHINI ET AL GASTROENTEROLOGY Vol. 136, No. 2

upplementary Figure 1. Effect of insulin stimulation on FoxA2 ex-ression and translocation in cytoplasmic and nuclear liver fractions

rom WT, Insr�/�, Timp3�/�, Insr�/�Timp3�/�, and Insr�/�Tace�/�; n �

per group. Data are mean � SD.

upplementary Figure 2. Interaction between genetic and HFD-induced insulin resistance with Timp3 deficiency promotes fatty streak formation.A) Aortic root histology from WT, Insr�/�, Timp3�/�, Insr�/�Timp3�/�, and Insr�/�Tace�/�. Oil Red O staining (50�): arrows indicate fatty streaks thatere observed only in Insr�/�Timp3-/; Weigert–Van Gieson staining (50�): arrows indicate fatty streaks that were observed only in Insr�/�Timp3-/;4/80 staining of aortic roots (magnification, 250�): arrows indicate sites of macrophage accumulation that were observed only in the Insr�/�Timp3�/

group; MCP-1 staining of aortic roots: staining is more intense in Insr�/�Timp3�/� compared with other groups (magnification, 400�); note innsr�/�Timp3�/�intense MCP-1 staining at both the endothelial and subendothelial layers, whereas WT show scarce staining only at the endothelialevel, no staining was observed in Insr�/�Tace�/�mice. RAGE staining of aortic roots (magnification, 250�): RAGE was expressed abundantly at thendothelium level in Insr�/�Timp3�/� mice. (B) Fatty streak area is increased in Insr�/�Timp3�/� compared with WT, Insr�/�, Timp3�/�, and

nsr�/�Tace�/� (P � .05, 1-way ANOVA). Representative images are shown; 5 mice per group were analyzed.