TIMP3 Overexpression in Macrophages Protects From Insulin Resistance, Adipose Inflammation, and Nonalcoholic Fatty Liver Disease in Mice.

1.5 MB·English

by  Menghini, Rossella

ORIGINALARTICLE
TIMP3 Overexpression in Macrophages Protects From
Insulin Resistance, Adipose Inflammation, and
Nonalcoholic Fatty Liver Disease in Mice
Rossella Menghini,1 Viviana Casagrande,1 Stefano Menini,2 Arianna Marino,1 Valeria Marzano,1,3
Marta L. Hribal,4 Paolo Gentileschi,5 Davide Lauro,1 Orazio Schillaci,6,7 Giuseppe Pugliese,2
Paolo Sbraccia,1 Andrea Urbani,1,3 Renato Lauro,1 and Massimo Federici1
modulationofinflammatorycomponentswithintheadipose
The tissue inhibitor of metalloproteinase (TIMP)3, a stromal
protein that restrains the activity of proteases and receptors, tissue, including monocytes/macrophages and T cells, has
is reduced in inflammatory metabolic disorders such as type 2 been shown to improve not only the obesity-induced in-
diabetes mellitus (T2DM) and atherosclerosis. We overexpressed flammatory responses but also their metabolic complica-
Timp3 in mouse macrophages (MacT3) to analyze its potential tionsinexperimentalmodels(7–9).Nevertheless,ithasnot
antidiabetic and antiatheroscleroticeffects. Transgenic mice with been fully elucidated whether inflammation per se leads to
myeloid cells targeting overexpression of TIMP3 were generated the development of diabetes and which molecules and
and fed a high-fatdiet for 20weeks.Physical and metabolic phe-
mechanisms might be involved.
notypesweredetermined.Inflammatorymarkers, lipid accumula- Werecentlyreportedthatdeficiencyoftissueinhibitorof
tion,andinsulinsensitivityweremeasuredinwhiteadiposetissue
metalloproteinase (TIMP)3 acts at the interface of inflam-
(WAT),liver,andskeletalmuscle.Inamodelofinsulinresistance,
mationandinsulinresistance(10).TIMP3isasecreted24–
MacT3miceweremoreglucosetolerantandinsulinsensitivethan
wild-type mice in both in vitro and in vivo tests. Molecular and 27-kDa proteinbelonging tothe familyofTIMPs thatbinds
biochemicalanalysesrevealedthatincreasedexpressionofTIMP3 to the extracellular matrix and inhibits metalloproteases
restrainedmetabolicinflammationandstress-relatedpathways,in- suchasmatrixmetalloproteinase-2/9andsheddasessuchas
cludingJunNH -terminalkinaseandp38kinaseactivation,inWAT tumor necrosis factor (TNF)-aconverting enzyme (also
2
and liver. TIMP3 overexpression in macrophages resulted in re- known as ADAM17), participating in the modulation of
ducedactivationofoxidativestresssignalsrelatedtolipidperox- inflammation, cellular migration, andproliferation(11,12).
idation,proteincarbonylation,andnitrationinWATandliver.Our
We and others have previously reported that the TIMP3/
datashowthatmacrophage-specificoverexpressionofTIMP3pro-
tectsfrommetabolicinflammationandrelatedmetabolicdisorders ADAM17 pathway is involved in the control of glucose ho-
meostasis and adipose, hepatic, and vascular inflammation
such as insulin resistance, glucose intolerance, and nonalcoholic
steatohepatitis.Diabetes61:454–462,2012 inbothgeneticandnutritionalmodelsofobesityinmice,as
well as in patients with obesity-related T2DM and athero-
sclerosis (13–18). Adam17-null mice consistently show a
O hypermetabolic phenotype, and heterozygous Adam17+/2
besity is a chronic disorder characterized by mice are protected from diet-induced obesity and in-
majormetaboliccomplications,includinginsulin flammation (19,20). Furthermore, data from our and other
resistance and steatohepatitis, often coexisting laboratorieshaverevealedthatobesityischaracterizedbya
inpatientswithtype2diabetesmellitus(T2DM) deficiencyofTIMP3inwhiteadiposetissue(WAT)andliver
(1–3). Recent studies suggest that low-grade inflammation (14,21), suggesting that downregulation of TIMP3 in meta-
might promote the onset of insulin resistance and hepatic bolic tissues may contribute to the deregulation of inflam-
steatosis (4,5). The triggers and the mechanisms provoking matorypathways.TounderstandtheextenttowhichTIMP3
the accumulation of inflammatory cells and the release could restrain the negativesynergy between unchecked
of cytokines within adipose tissue and liver are still un- inflammation and nutrient overload, we generated a new
defined, although free fatty acids have been suggested to mousemodelwithacell-targetedoverexpressionofTIMP3
playamajorpartinthisprocess(6).Ontheotherhand,the in myeloid cells (MacT3) in order to directly overexpress
TIMP3 protein in key inflammatory sites (22). The aim of
thisstrategywastoverifythehypothesisthatTIMP3could
Fromthe1DepartmentofInternalMedicine,UniversityofRome“TorVergata,” rescue the metabolic alterations associated with obesity.
Rome,Italy;the2DepartmentofClinicalandMolecularMedicine,“Sapienza”
University, Rome, Italy; the 3Laboratory of Proteomics, EBRI/Santa Lucia
Foundation,Rome,Italy;the4DepartmentofMedicalandSurgicalSciences, RESEARCHDESIGNANDMETHODS
UniversityofMagna Graecia,Catanzaro, Italy; the 5Department of Surgery,
UniversityofRome“TorVergata,”Rome,Italy;the6DepartmentofDiagnostic Mousemodelandmetabolicanalysis.ThemouseTimp3genewasampli-
Imaging,UniversityofRome“TorVergata,”Rome,Italy;andthe7Istitutodi fiedbyPCRwithcDNAobtainedfrommousemuscleasatemplate.Thegene
RicoveroeCuraaCarattereScientificoNeuromed,Pozzilli,Italy. wasthenclonedinavectorcontainingCD68promoter/enhancer.Forgener-
Correspondingauthor:MassimoFederici,[email protected]. ation of transgenic mice, the linearized construct was microinjected into
Received9May2011andaccepted21November2011. mousepronucleiusingstandardmethodsaspreviouslydescribed(23,24).Off-
DOI:10.2337/db11-0613 springderivedfromtheinjectionsweregenotypedbyPCRandconfirmedby
This article contains Supplementary Data online at http://diabetes
Southernblotting.Forhematopoieticcelllineagedistribution,peripheralblood
.diabetesjournals.org/lookup/suppl/doi:10.2337/db11-0613/-/DC1.
wascollectedfromthetailveinandanalyzedonabloodcellanalyzer(Simply
R.M.andV.C.contributedequallytothisstudy.
(cid:1)2012bytheAmericanDiabetesAssociation.Readersmayusethisarticleas Cell;BPCBioSed,Rome,Italy)todeterminecellcounts.
longastheworkisproperlycited,theuseiseducationalandnotforprofit, Forthediet-inducedobesitymodel,micewerefedahigh-fatdiet(HFD)(60%
and the work is not altered. See http://creativecommons.org/licenses/by of calories from fat; Research Diets, New Brunswick, NJ) or a normal diet
-nc-nd/3.0/fordetails. (standardchow,10%caloriesfromfat,GLP;Mucedolas.r.l.,SettimoMilanese,
454 DIABETES,VOL.61,FEBRUARY2012 diabetes.diabetesjournals.org
R.MENGHINIANDASSOCIATES
Italy)for20weeksafterweaningasindicated.Metabolictestshavepreviously valuesforsofttissuewerecalibratedfromafive-pointlinearfitlineusingmix-
been described (13). Hormones and adipokine levels were measured using turescontainingvariousratiosoftwoliquids,rangingfrom0.78mg/mL(100%
commercialkits:insulin(Mercodia,Uppsala,Sweden)andadiponectinand ethanol;Sigma-Aldrich)to1.26mg/mL(100%glycerol;J.T.Baker,Phillipsburg,
leptin (R&D Systems, Minneapolis, MN). Aspartate aminotransferase and NJ).Densityvaluesforbonetissuewerecalibratedviaacommerciallyavailable
alanineaminotransferaselevelsweremeasuredwithaModularPanalyzer phantomcontaininghydroxyapatiterodsofvariousdensities(GEHealthcare).
(RocheS.p.A,Monza,Italy). Adiposetissuehistologicalanalysis.Histologicalproceduresandantibodies
Energybalance.IndirectcalorimetrywasperformedusingLabMaster(TSE have previously been described (13). The total number and cross-sectional
Systems,BadHomburg,Germany).Allmicewereacclimatizedfor24hbefore areas of adipocytes were obtained by computer-assisted image analysis
measurementsweretakenevery15minfor24h.VO2isexpressedasmilliliters (Optimas 6.5; Bioscan, Edmunds, WA). Ten random fields per section were
ofO consumedperkilogramofbodyweightperminute.Carbondioxidecon- analyzed (size of field 0.78 mm2), and values ,100 mm2 were automatically
2
sumption(VCO2)isexpressedasmillilitersofCO2producedperkilogramofbody excluded from analysis because they were assumed to represent artifacts
weightperminute.TherespiratoryquotientisdefinedasVCO2/VO2anddoesnot resultingfromtheimageconversionprocess.Thedatawerefirstaveragedper
have any units. Activity was expressed by movement-induced beam inter- sectionandthenperanimal.Forthedoubleimmunofluorescencelabelingstudy,
ruptionsperminuteinthecalorimeter.Dailyfoodintakewasmonitoredby sections were incubated with rat monoclonal anti-CD68 antibody and rabbit
weighingfoodhoppers. polyclonal anti-CD3 antibody (Abcam, Cambridge, MA). The stained sections
Geneexpressionanalysis.TotalRNAwasisolatedfromWAT,muscle,and wereexaminedunderafluorescencemicroscope(OlympusBX51equippedwith
liver,usingTRIzolreagents(Invitrogen,Eugene,OR).Atotalof2mgtotalRNA OlympusC-3030digitalcamera;Olympus,Tokyo,Japan).
wasreversetranscribedintocDNAusingtheHighCapacitycDNAArchivekit Vasculardensity.WATsections(n=8)immunolabeledwiththepolyclonal
(AppliedBiosystems,FosterCity,CA).Quantitativereal-timePCRwasperformed goatanti-mouseCD31antibody(PECAM-1;SantaCruzBiotechnology,Santa
on individual samples using an ABI PRISM 7700 System and TaqMan Cruz,CA)wereviewedat3250.Twentyrandomfieldspersectionwerean-
reagents (Applied Biosystems). Each reaction was performed in triplicate alyzed(sizeoffield0.125mm2).Imageswereconvertedinabinaryformat,and
usingstandardreactionconditions:1cycleat50°Cfor2min,1cycleat95°Cfor thenumberofvesselsperfieldwascountedandexpressedasnumberofvessels
10min,and40cycleseachat95°Cfor15sand60°Cfor1min.Calculationswere permillimetersquared.Averagebloodvesselsizewascalculatedbydividingthe
performedviaacomparativecyclethresholdmethod(13). totalstainedareabythenumberofvessels.
Westernblots.Westernblotswereperformedontotaltissuehomogenatesand Histologyandquantificationofliverlesions.Histologicalprocedureshave
macrophage extracts prepared as previously described (25). The following previouslybeendescribed(14).F4/80,N´-(carboxymethyl)lysine(CML),and
antibodieswereused:TIMP3,actin,anti–phospho-Ser473AktandtotalAkt, nitroTyrimmunostainingwereperformedusingantibodies(Abcam),following
anti–phospho-Thr183/Tyr185 Jun NH-terminal kinase (JNK) and total JNK, themanufacturer’sinstructions.
2
andanti–phosphoThr180/Tyr182p38andtotalp38(CellSignalingTechnology, Isolation of adipocytes and stromal vascular fraction from WAT. To
Danvers,MA).Arepresentativeimageofeightmicepereachgroupisshown. determinetherelativeexpressionofTIMP3inadipocytesandstromalvascular
ADAM17 activityandTNF-alevels.ADAM17activitywas determined as fraction(SVF),WATwassubjectedtocollagenasedigestion(1mg/mLcolla-
previouslydescribed(13).TNF-alevelsweremeasuredincellmediumusinga genasetype1;Sigma-Aldrich)inKrebs-Ringerbuffer,withshakingat150rpm
commercialELISAkit(R&DSystems)inaccordancewiththemanufacturer’s for30minat37°C.Afterdigestion,adipocyteswereallowedtoseparateby
instructions. flotation and the infranatant solution was centrifuged for 5 min at 300g to
Pancreatahistologicalanalysisandmorphometry.Pancreatawerefixed pellettheSVF.TheadipocytefractionwaswashedthreetimeswiththeKrebs-
overnightin4%paraformaldehydeandembeddedinparaffinusingstandard Ringerbuffer.Subsequently,RNAwasisolatedfromadipocytesandtheSVF
techniques.Sectionswerestainedwithhematoxylin-eosin.Immunostainingfor fractions and analyzed by real time PCR. The profile of adiponectin mRNA
insulinwascarriedoutwiththeHistomouse-SPkit(ZymedLaboratories,South expressionprovidesevidenceofthepurityofthefractionpreparations.
San Francisco, California) according to themanufacturer’s instructions. Six Isolationofhepatocytesandnonparenchymalcellsfromliver.Micewere
sectionsforeachpancreaswereimagedat320magnificationandquantified anesthetized, andlivertissues wereperfused in situviathe superiorvena
for surface areas. Areas corresponding to the b-cells were measured by di- cavawithaperfusionbuffer(13Hanks’balancedsaltsolution),followedby
vidingtheareaofinsulin-positiveimmunoreactivitybywholepancreasarea. a digestion buffer (13 Hanks’ balanced salt solution, supplemented with
2,4-Dinitrophenyldrazine assay. Proteins were derivatized on carbonyl 0.05% collagenase [Type IV; Sigma Chemical, St. Louis, MO], 1.25 mmol/L
groupsby10mmol/L2,4-dinitrophenyldrazine(Sigma-Aldrich)in2Nchloridric CaCl2, 4 mmol/L MgSO4, and 10 mmol/L HEPES). The resulting cell sus-
acid and were precipitated overnight with an ice-cold solution containing pensionwasfilteredthroughasterile100-mmnylonmeshandcentrifugedat
50%ethanol,25%methanol,and25%acetoneat–20°C.Afterfourwasheswith 50gtoselectivelysedimenthepatocytesfromnonparenchymalcells(NPCs).
thesameice-coldsolution,thefinalpelletwasdriedatroomtemperatureand The pellet of hepatocytes was resuspended and subsequently washed two
redissolved in Laemmli sample buffer. After SDS-PAGE separation, proteins moretimesat50g.TheNPCsinthefirstandsecondsupernatantsfromthelow-
were electroblotted onto nitrocellulose membrane and immunodetection was speedcentrifugationswerepelletedbyhigh-speedcentrifugation(1,300g),fol-
performedusing anti–2,4-dinitrophenylkeyhole limpethemocyaninrabbitIgG lowed by resuspension in a small volume before isopycnic sedimentation in
(MolecularProbes,Eugene,OR)inordertoevaluatecarbonylatedproteins. Percollaspreviouslydescribed(26).Cellviability(.90%)wasdeterminedby
Theimmunoreactivebandsweredetectedusinggoatanti-rabbithorseradish trypanblueexclusion.TheprofileofalbuminmRNAexpressionprovidesevi-
peroxidase–conjugatedIgG(Bio-RadLaboratories,Hercules,CA)andachemilu- denceforthepurityofthefractionpreparations.
minescencedetectionmethod(ECL;GEHealthcare,Buckinghamshire,U.K.). Statistical analysis. Results of the experimental studies are expressed as
Lipid peroxidation. Lipid peroxidation products were determined by thio- means6SD.Statisticalanalyseswereperformedusingtheone-wayANOVA
barbituricacidreactivesubstances(TBARSs)inproteinextractsusingcolorimetric orunpairedStudentttestasindicated.ValuesofP,0.05wereconsidered
assayaccordingtothemanufacturer’sinstructions(LPO-586Bioxytech;OXIS statisticallysignificant.
International,FosterCity,CA).
In vivo positron emission tomography imaging. The positron emission
tomography(PET)protocolincludedalocalizerscan,ahigh-resolution(0.63 RESULTS
mmslicethickness)whole-bodycomputedtomographystudy,andtwo15-min
Metabolic effect of diet-induced obesity on MacT3.
whole-body PET acquisitions starting at 5 and 35 min after tracer adminis-
tration.Forassessmentoffluorodeoxyglucose(FDG)distribution,transverse We generated transgenic mice (MacT3) overexpressing
and dorsal slices were obtained, as well as whole-body three-dimensional TIMP3 under the control of the monocyte/macrophage
maximum-intensity projections. Measurements of FDG distribution were lineage-specificpromoterCD68inordertoincreaseTIMP3
obtained on dorsal planes using small regions of interest positioned on the expressiondirectlyinmetabolictissueswheremyeloidcells
brain,heart,liver,andlimb.Foreachstructure,measurementswereobtained
are gradually recruited during obesity (22). Three positive
fromfourdifferentslicesandaveraged.FDGuptakeratiosforheart,liver,and
founder lines, which gave germ line transmission, were
forelimb were calculated using brain uptake as a reference structure that
metabolizes glucose independent of insulin. Maximum standardized uptake expandedtoestablishtheMacT3transgeniccolonies(data
valueswerecalculatedusingaleanbodymassalgorithm. not shown). Because the overexpression of TIMP3 was
Computedtomographyscan.Forinvivoscans,micewereanesthetizedby1% comparable in the three lines and they showed similar
isofluraneinhalationandpositionedwithbothlegsfullyextended.Theentire features after a preliminary characterization, we used one
torsoofeachmousewasscannedatanisotropicvoxelsizeof76microns(45-kV,
of them for our studies. MacT3 mice fed a normal diet
133-mA, and 300-ms integration time) with a GE Explore Locus scanner
did not differ from wild-type (WT) littermates in terms
(GEHeathcareItalia,Milano,Italy).Two-dimensionalgrayscaleimageslices
werereconstructedintoathree-dimensionaltomographicimage.Density of fasting and fed glucose insulin levels, glucose and
diabetes.diabetesjournals.org DIABETES,VOL.61,FEBRUARY2012 455
TIMP3PROTECTSFROMMETABOLICINFLAMMATION
insulin tolerance, adiponectin, or leptin secretion (Sup- mice (Fig. 1D). Indirect calorimetry showed that MacT3
plementary Fig. 1A–F, available in the Supplementary mice exhibited a slight significant increase in VO2 and
Data). VCO2, activities, and food intake (Fig. 1E and F). To in-
For evaluation of the role of TIMP3 overexpression in vestigate whether the improvement in glucose homeo-
thedevelopmentofobesity,MacT3andWTmicewerefed stasis was due, at least in part, to an increase in glucose
an HFD for 20 weeks. MacT3 mice on an HFD did not transport, we conducted a PET scan to directly measure
showsignificantdifferencesinwhitebloodcelllineagesor FDG uptake after insulin infusion in both anterior and
monocyte counts compared with WT littermates (Supple- posterior legs, and we observed that the standardized
mentaryFig.2A).MacT3micedidnotdifferinweightfrom uptake value was statistically higher in MacT3 mice
WT mice while consuming the HFD (Fig. 1A) but showed compared with WT mice (Fig. 2A). MacT3 mice also
improved fasting and fed glucose and insulin levels (Fig. showed an increased insulin-induced phosphorylation of
1B). Similarly, MacT3 mice showed improved glucose Ser473 Akt, which is a major regulator of glucose uptake
tolerance and insulin sensitivity compared with WT in muscle (Fig. 2B).
mice, as assessed by the intraperitoneal glucose tolerance Effect of diet-induced obesity on adiposity in MacT3.
test and intraperitoneal insulin tolerance test (Fig. 1C). In To evaluate how an HFD impacts adipose tissue morphol-
contrast, we did not observe any differences in glucose- ogy and functions in MacT3 and WT mice, we analyzed
stimulatedinsulinsecretion,pancreaticisletnumbers,or adipose tissue depots in vivo through computed tomogra-
gross morphology in MacT3 mice compared with WT phy scans, which revealed that MacT3 mice had a slightly
littermates fed an HFD (Supplementary Fig. 2B–E). Lev- decreased visceral adipose tissue fraction, despite a simi-
elsofadiponectin,transaminases,TNF-a,andinterleukin laroverall adiposity(Fig.3A). Epididymal and perirenal
(IL)-6 were improved in MacT3 mice compared with WT fat pads were also smaller in MacT3 mice (Fig. 3B and
FIG.1.MacT3miceareprotectedagainstHFD-inducedinsulinresistance.WTandMacT3micewerefedanHFDfor20weeks.A:Bodyweightcurve
duringHFD(n=9pergroup).B:Fastingandfedglucoseandinsulinlevels(n=9pergroup;*P<0.05,Studentttest).C:Intraperitonealglucose
tolerancetestandintraperitonealinsulintolerancetest(n=9pergroup;*P<0.05,one-wayANOVA).D:Adiponectin,leptin,TNF-a,IL-6,and
transaminaselevels(n=9pergroup;*P<0.05,Studentttest.Dataaremeans6SD).E:VO2infree-fedMacT3miceoveraperiodof24h(line
chart)comparedwithVO2inWTmiceafter20weeksofHFD.F:Mean24-hvaluesforVO2,VCO2,respiratoryquotient(RER),locomotoractivity,and
foodintake.(EandF:n=6pergroup.Studentttest,*P<0.05forMacT3HFDvs.WTHFDmice.Dataaremeans6SD.)
456 DIABETES,VOL.61,FEBRUARY2012 diabetes.diabetesjournals.org
R.MENGHINIANDASSOCIATES
FIG.2.MacT3miceshowimprovedinsulin-stimulatedglucosetransportinvivo.A:AnteriorimagesfromtheFDGPETscansofWTandMacT3mice
fedanHFDfor20weeks.Arrowsindicatesiteswithdifferentglucoseuptakeinforelimbsandhindlimbs.Maximumstandardizeduptakevalues
(SUV )indicatinganincreasedglucoseuptakeinMacT3vs.WTmice.(n=5pergroup;*P<0.05,Studentttest.Dataaremeans6SD.)B:
max
Skeletalmuscletissuefromovernight-fastedWTandMacT3micefedanHFDfor20weekswasdissected5minafterinsulininjection,andAkt
phosphorylation(p)wasanalyzedbyWesternblot.(n=8pergroup;**P<0.01,Studentttest.Dataaremeans6SD.)Arepresentativeimageof
fourmicepergroupisshown.a.u.,arbitraryunit.(Ahigh-qualitycolorrepresentationofthisfigureisavailableintheonlineissue.)
Supplementary Fig. 3A). WAT from MacT3 mice showed (Fig. 4B) observed in MacT3 mice. MacT3 mice showed
significantly increased TIMP3, both at mRNA and protein reducedactivationofp38andJNK(Fig.4C),whichallowed
levels, and decreased ADAM17 activity (Fig. 3C). By ana- for increased phosphorylation of Akt kinase, the mediator
lyzingisolatedtissuefractionsseparately,weobservedthat of insulin metabolic effects, both in the fed state and after
expressionofTIMP3inadipocyteissignificantlylowerthan insulin stimulation (Fig. 4D).
in SVF in WT mice (Fig. 3D). In addition, TIMP3 level was Effect of diet-induced obesity on hepatic steatosis in
furtherincreasedinSVFobtainedfromWATofMacT3mice MacT3.WerecentlyobservedthatTIMP3deficiencyledto
compared with WT. Adiponectin mRNA expression was steatohepatitis under an HFD regimen (14). In contrast,
analyzedtoconfirmthepurityofthetwodifferentfractions underthesameconditionsMacT3miceshowedsignificantly
(Fig. 3D). reduced liver weights (Fig. 5A), higher levels of TIMP3 ex-
Histological analysis of HFD WAT sections showed a pression, and reduced ADAM17 activity (Fig. 5B–D). In ad-
higher density of smaller adipocytes (Fig. 3E and Sup- dition,analysisofisolatedhepaticcells(hepatocytesand
plementary Fig. 3B). The insulin-sensitizing and anti- NPCs) showed that TIMP3 mRNA was significantly over-
inflammatory hormone adiponectin, and oxidative stress expressedinnonparenchymalcells(whichincludeKupffer
markers such as TBARSs and the presence of protein car- cells)fromMacT3mouseliverscomparedwithexpression
bonylation(2,4-dinitrophenyldrazine),wasreducedinMacT3 in WT mouse livers (Fig. 5E). Albumin mRNA expression
mice compared with levels in WT mice (Fig. 3F and was analyzed in order to confirm the purity of the two dif-
Supplementary Fig. 3C and D). Immunohistological analy- ferent fractions (Fig. 5E).
ses of inflammatory markers for macrophages (F4/80), Histologicalanalysisrevealedthatobesityinliversfrom
T cells (CD3), and capillary density were conducted by WTmiceinducedamacrovesicularsteatosissimilartothe
CD31 staining and revealed a lower number of tissue mac- grade-2stageinhumansandcharacterizedbyfeaturessuch
rophages and T cells and slightly reduced adipose tissue asballooningdegeneration;incontrast,MacT3mouselivers
vascularization in MacT3 mice compared with WT mice werelesssteatotic(betweengrade0andgrade1)(Fig.5F).
(Fig. 4A). Analysis using immunofluorescence microscopy Compared with WT mice, MacT3 mice revealed signifi-
confirmed that CD68 and CD3 signals did not colocalize cantly reduced F4/80 staining and reduced oxidative
(Supplementary Fig. 4). These results were also supported stress, as suggested by markers such as N´-CML and
by the lower mRNA expression of inflammatory markers nitroTyr staining (Fig. 5F). Analysis of TBARS levels in
diabetes.diabetesjournals.org DIABETES,VOL.61,FEBRUARY2012 457
TIMP3PROTECTSFROMMETABOLICINFLAMMATION
FIG.3.MacT3miceareprotectedagainstHFD-inducedchangesinadiposetissue.A:Representativecomputedtomographyscansectionindicating
adecreasedvisceraladiposetissuefractioninMacT3vs.WTmicefedanHFDfor20weeks,confirmedbyvisceraladiposetissuequantification.
(n=4pergroup;*P<0.05,Studentttest.Dataaremeans6SD.)B:Epididymalfatpadswereharvestedandweighed.(n=5pergroup;*P<0.05,
Studentttest.Dataaremeans6SD.)C:TIMP3mRNAandproteinlevelswereincreasedandADAM17activitywasreducedinadiposetissueof
MacT3micefedanHFDcomparedwithWTmice.Arepresentativeimageoffourmicepergroupisshown.(n=8pergroup;*P<0.05,Student
ttest.Dataaremeans6SD.)D:TIMP3andadiponectin(consideredadipocytemarkers)andmRNAexpressioninmatureadipocytes(AFs)and
SVF.(n=5pergroup;**P<0.01and***P<0.001,Studentttest.Dataaremeans6SD.)E:Representativesectionsofadiposetissuestainedwith
hematoxylin-eosin(H/E)showinganincreasedadipocytedensityandareductioninmeanadipocyteareainMacT3micevs.WTmice.(n=5per
group;**P<0.01and*P<0.05,Studentttest.Dataaremeans6SD.)F:WATexpressionofleptinandadiponectinmRNA.ExpressionofmRNA
wasdeterminedbyreal-timePCRandnormalizedto18SRNA.(n=5pergroup;*P<0.05,Studentttest.Dataaremeans6SD.)a.u.,arbitraryunit.
(Ahigh-qualitydigitalrepresentationofthisfigureisavailableintheonlineissue.)
liver homogenates confirmed reduced oxidative stress in sufficient for containing the development of insulin re-
MacT3 mice compared with WT mice (Fig. 6A). mRNA sistance and NAFLD in vivo.
expression levels of oxidative stress genes (catalase, In the context of diet-induced obesity, TIMP3 improved
SOD2,Gp47,andGP67),andinflammatory(suppressorof insulin sensitivity and systemic glucose uptake in vivo.
cytokinesignaling3andIL-6)andanti-inflammatory(IL-10) AlthoughtheoverexpressionofTIMP3didnotaffectbody
markers supported the histological evidence for reduced weight, it was involved in the adipose tissue architecture.
inflammation in the livers of MacT3 mice (Fig. 6B). As in In fact, at the microscopic level the adipose tissue archi-
adipose tissue, MacT3 mouse livers exhibited significantly tecture in MacT3 mice was characterized by the presence
lower activation of the stress kinase JNK and improved of small adipocytes of higher density, suggesting an in-
phosphorylation of Ser473 Akt both in the fed state and crease in metabolic activity (28). At the molecular level,
upon insulin stimulation (Fig. 6C). TIMP3overexpressionresultedinacontinuedsecretionof
adiponectin, an anti-inflammatory and insulin-sensitizer
hormone known to positively affect glucose and lipid me-
DISCUSSION tabolism, as well as liver inflammation (29,30). This was
Visceralobesityandtheconsequentlow-gradeinflammatory accompanied by a reduced inflammatory infiltrate and de-
state are major determinants for the onset of insulin re- creased IL-6 expression and secretion in vivo. Analysis of
sistance and its metabolic sequels, T2DM and nonalcoholic signaling readouts showed that in MacT3 mice, the activa-
fattyliverdisease(NAFLD)(1,27).Inthisstudy,wehave tionoftheinsulinreceptorpathwaywaspreserved,tosome
shown that a transgenic approach aimed to increase the extent,throughreductionofthenegativeeffectproducedby
levels of TIMP3 at the sites of metabolic inflammation is stresskinasessuchasJNKandp38.Furthermore,computed
458 DIABETES,VOL.61,FEBRUARY2012 diabetes.diabetesjournals.org
R.MENGHINIANDASSOCIATES
FIG.4.ProtectiveeffectoftransgenicexpressionofTIMP3inmacrophagesoninflammationinWAT.A:RepresentativesectionsofWATfromWT
andMacT3micefedanHFDfor20weeksstainedwithF4/80,CD3,andCD31,revealingareductioninmacrophageandT-cellinfiltrationandin
vesseldensity.(n=5micepergroup;*P<0.05,Studentttest.Dataaremeans6SD.)B:WATexpressionofgenesinvolvedininflammation.
ExpressionofmRNAwasdeterminedbyreal-timePCRandnormalizedto18SRNA.(n=5pergroup;*P<0.05and**P<0.01,Studentttest.Data
aremeans6SD.)C:WATfromWTandMacT3micefedanHFDfor20weekswaslysedandanalyzedforphosphorylationofJNKandp38.D:WAT
fromWTandMacT3micefedanHFDfor20weekswaslysedandanalyzedforphosphorylationofAktbyWesternblot(upperpanel).Adipose
tissuefromovernight-fastedWTandMacT3micefedanHFDfor20weekswasdissected5minafterinsulininjection;insulin-stimulatedAkt
phosphorylation(p)wasanalyzedbyWesternblot(lowerpanel).(CandD:ForallWesternblots,n=8pergroup;arepresentativeimageoffour
micepergroupisshown.*P<0.05and**P<0.01,Studentttest.Dataaremeans6SD.)a.u.,arbitraryunit.(Ahigh-qualitydigitalrepresentation
ofthisfigureisavailableintheonlineissue.)
tomography scan and histological analyses revealed a re- release of systemic inflammatory factors, such as IL-6. The
duction of the visceral adipose tissue fraction and micro- currentstudyshowedthatinmicefedaregulardiet,TIMP3
vesseldensityinMacT3mice.Previousstudieshaveshown was physiologically more expressed in the SVC fraction
that TIMP3 acts to stabilize vessels during angiogenesis than in mature adipocytes, suggesting that the effect of
andlimitsthedevelopmentofanuncontrolledangiogenic TIMP3onadipocytesmaydependonthecomplexdynamics
response (31). Because antiangiogenic factors such as of inflammatory cells within the expanding adipose tissue
angiostatin demonstrated activity against obesity and its during obesity. Still controversial is whether TIMP3 is in-
complications (32), we can conjecture that part of the volved in regulation of adipogenesis. In fact, some authors
TIMP3 effect on systemic metabolism may be due to its havesuggestedthatTIMP3mightlimitadipogenesis(21,33),
ability to stabilize the vessels inside adipose tissue. There- while others have suggested that TIMP3 could favor adi-
fore,itispossiblethattheactionofTIMP3onadiposetissue pogenesis via the limitation of the TNF-a converting
is sufficient for maintenance of metabolic control, thus enzyme–dependent shedding of Pref-1, a molecule that,
avoiding the deterioration of liver metabolism, which is when released from the membrane, is known to block the
acentralcheckpointforbothlipoproteinsynthesisandthe differentiation of the preadipocyte cell (34).
diabetes.diabetesjournals.org DIABETES,VOL.61,FEBRUARY2012 459
TIMP3PROTECTSFROMMETABOLICINFLAMMATION
FIG.5.ReducedhepaticsteatosisinHFD-fedMacT3mice.A:LiversofMacT3andWTmicefedanHFDfor20weekswereharvestedandweighed.
Resultsareexpressedingrams.(n=10pergroup;*P<0.05,Studentttest.Dataaremeans6SD.)B:TIMP3mRNA.CandD:TIMP3proteinlevels
(C)andADAM17activity(D)inliversofMacT3micecomparedwithWTmice.(n=8pergroup;*P<0.05,Studentttest.Dataaremeans6SD.)E:
TIMP3andalbumin(hepatocyte[HEP]marker)mRNAexpressioninpurifiedhepatocytesandNPCsfromMacT3andWTmouselivers.(n=5per
group;**P<0.01and***P<0.001,Studentttest.Dataaremeans6SD).F:Representativesectionsofliverstainedwithhematoxylin-eosin(H/
E),OilRedO,F4/80,CML,ornitroTyr(N-Tyr).ArrowindicatesalipogranulomainWTmice.CV,centralvein.(n=5pergroup;**P<0.01and*P<
0.05,Studentttest.Dataaremeans6SD.)a.u.,arbitraryunit.(Ahigh-qualitydigitalrepresentationofthisfigureisavailableintheonlineissue.)
Although our data supporta major direct effectof TIMP3 hepatic inflammation and fibrosis, especially in synergy
in adipose tissue, we cannot exclude the possibility that the with nutrient overload (13,14). Whether the TIMP3/Fas
effects of TIMP3 on the liver parenchyma may also be in- connection is relevant for transition from fatty liver dis-
volvedintheMacT3phenotype.Infact,ithasbeensuggested ease to steatohepatitis requires further studies, although
that proinflammatory macrophages in the liver are charac- FashasbeendemonstratedtoberelevanttoNAFLDviaits
terizedby reduced TIMP3 expression (35). We observed interaction with Met (37), and Met signaling may be af-
that hepatocytes expressed more TIMP3 than did cells in fected by metalloprotease regulation (38).
thenonparenchymalfraction.InMacT3mice,weobtained PreviousstudieshaveshownindividualeffectsofTIMPs
a specific increase of TIMP3 in the fraction containing on separate components of the metabolic syndrome or its
Kupfferandhepaticstellatecells,suggestingthattheeffect complications, such as insulin resistance and obesity
in the liver is independent of the cellular source, which is (13,39–44). However, among TIMPs only TIMP3 was found
reasonable given that TIMP3 is released outside the cells tobereducedintheadiposetissueofexperimentalmodels
and acts in the extracellular space. of obesity and insulin resistance when genes regulating
Recently,ithasbeensuggestedthatlossofTIMP3could matrix biology were analyzed (44).
cooperate with epidermal growth factor receptor activa- Ourstudyhasclearstrengthsandlimitations.Infact,our
tiontodictatetheresistancetoFas-inducedapoptosisand approach was designed to test the effect of TIMP3 at the
hepatitis (36). The counterintuitive effect of apoptosis pre- site of metabolic inflammation independent of the physi-
vention is the unchecked activation of proliferative signals ological cellular sources. The coupling of TIMP3 to the
mediated by release of epidermal growth factor receptor CD68 promoter allowed us to test the effect of TIMP3 in
ligands and TNF-a shedding, both of which are dependent atime-restrainedmanner becausethetransgenicTIMP3is
on ADAM17. Therefore, while loss of TIMP3 could protect transcribed and released only when the CD68 promoter is
fromacuteinsultstotheliver,itmayparadoxicallypromote activated. Therefore, it is intriguing to hypothesize that
460 DIABETES,VOL.61,FEBRUARY2012 diabetes.diabetesjournals.org
R.MENGHINIANDASSOCIATES
FIG.6.Reducedhepaticoxidative(Ox)stressandimprovedinsulinsignalinginHFD-fedMacT3mice.A:TBARSassaydetectinglipidperoxidation
products(foldchange).(n=5pergroup;*P<0.05,Studentttest.Dataaremeans6SD.)B:mRNAexpressionofindicatedmarkersinlivertissue
ofWTandMacT3micewasanalyzedbyreal-timePCR,expressedrelativetoWT,andnormalizedtoexpressionof18S.CAT,catalase.(n=5per
group;*P<0.05,Studentttest.Dataaremeans6SD).C:LiversfromWTandMacT3micefedanHFDfor20weekswerelysedandanalyzedfor
phosphorylation(p)ofAktandJNKbyWesternblot.Forinsulin-stimulatedAktphosphorylation,liversfromovernight-fastedmiceweredissected
5minafterinsulininjectionandanalyzedbyWesternblot.(n=8pergroup.Arepresentativeimageoffourmicepergroupisshown.*P<0.05,
Studentttest.Dataaremeans6SD.)a.u.,arbitraryunit.(Ahigh-qualitycolorrepresentationofthisfigureisavailableintheonlineissue.)
TIMP3 could restrain the inflammatory phenotype of a ACKNOWLEDGMENTS
macrophage in a cell-autonomous manner, thus present- This study was supported by the following grants: Tele-
ing a new target for the control of inflammation in met- thonGGP08065,FondazioneRoma2008,JuvenileDiabetes
abolic inflammatory disorders. While CD68 is generally Research Foundation Regular Research 1-2007-665, FP7-
considered a selective marker for myeloid cells, various Health-241913FLORINASHandEURHYTHDIA,andMinis-
studies have reported that the expression of CD68 was try of Health RF 2007 (all to M.F.).
notrestrictedtothemacrophagelineage(45).Therefore, No potential conflicts of interest relevant to this article
althoughCD68expressioninnonmyeloidcells,including were reported.
fibroblasts, was detected at a variable but lower level R.M. performed experiments, analyzed data, drafted the
compared with monocytes and macrophages, we cannot manuscript, and wrote the final version of the manuscript.
ruleoutaminorcontributionofthesecellstothephenotype V.C., A.M., P.G., and V.M. performed experiments. S.M.
observed in MacT3 mice. Moreover, this approach limited researched data and reviewed and edited the manuscript.
the opportunity to identify the circumstances regulating M.L.H. performed experiments, analyzed data, and drafted
physiological TIMP3 release in adipose tissue and liver. In the manuscript. A.U. analyzed data. D.L. and P.S. contrib-
conclusion, to our knowledge our study is the first to de- uted to the discussion and reviewed and edited the
scribeaprotectiveeffectofTIMP3oninsulinresistanceand manuscript. O.S. performed experiments and analyzed
NAFLD, thus providing the conceptual basis for further ex- data.G.P.andR.L.contributedtothediscussionandreviewed
ploitationofaTIMP3-basedtherapyformetabolicdisorders. and edited the manuscript. M.F. performed experiments,
diabetes.diabetesjournals.org DIABETES,VOL.61,FEBRUARY2012 461
TIMP3PROTECTSFROMMETABOLICINFLAMMATION
analyzed data, drafted the manuscript, wrote the final 22. LangR,RutschmanRL,GreavesDR,MurrayPJ.Autocrinedeactivationof
version of the manuscript, conceived the study, and is the macrophagesintransgenicmiceconstitutivelyoverexpressingIL-10under
guarantor of this work and, as such, had full access to all controlofthehumanCD68promoter.JImmunol2002;168:3402–3411
23. Hogan B, Beddington R, Costantini F, Lacy E. Manipulating the Mouse
the data in the study and takes responsibility for the
Embryo.2nded.ColdSpringHarbor,NY,ColdSpringHarborLaboratory
integrityofthedataandtheaccuracy ofthedataanalysis.
Press,1994
24. Casagrande V, Menghini R, Menini S, et al. Overexpression of tissue in-
hibitorofmetalloproteinase3inmacrophagesreducesatherosclerosisin
REFERENCES low-densitylipoproteinreceptorknockoutmice.ArteriosclerThrombVasc
1. CaldwellSH,OelsnerDH,IezzoniJC,HespenheideEE,BattleEH,Driscoll Biol2012;32:74–81
CJ. Cryptogenic cirrhosis: clinical characterization and risk factors for 25. MenghiniR,MarchettiV,CardelliniM,etal.PhosphorylationofGATA2by
underlyingdisease.Hepatology1999;29:664–669 Akt increases adipose tissue differentiation and reduces adipose tissue-
2. PoonawalaA,NairSP,ThuluvathPJ.Prevalenceofobesityanddiabetesin relatedinflammation:anovelpathwaylinkingobesitytoatherosclerosis.
patientswithcryptogeniccirrhosis:acase-controlstudy.Hepatology2000; Circulation2005;111:1946–1953
32:689–692 26. Smedsrød B, Pertoft H. Preparation of pure hepatocytes and reticuloen-
3. Hanley AJ, Williams K, FestaA, etal.;insulin resistanceatherosclerosis dothelialcellsinhighyieldfromasingleratliverbymeansofPercollcen-
study.Elevationsinmarkersofliverinjuryandriskoftype2diabetes:the trifugationandselectiveadherence.JLeukocBiol1985;38:213–230
insulinresistanceatherosclerosisstudy.Diabetes2004;53:2623–2632 27. TilgH,MoschenAR.Evolutionofinflammationinnonalcoholicfattyliver
4. TilgH,HotamisligilGS.Nonalcoholicfattyliverdisease:cytokine-adipokine disease:themultipleparallelhitshypothesis.Hepatology2010;52:1836–1846
interplay and regulation of insulin resistance. Gastroenterology 2006; 28. Varlamov O, Somwar R, Cornea A, Kievit P, Grove KL, Roberts CT Jr.
131:934–945 Single-cellanalysisofinsulin-regulatedfattyaciduptakeinadipocytes.Am
5. PickupJC.Inflammationandactivatedinnateimmunityinthepathogen- JPhysiolEndocrinolMetab2010;299:E486–E496
esisoftype2diabetes.DiabetesCare2004;27:813–823 29. Adachi M, Brenner DA. High molecular weight adiponectin inhibits pro-
6. Schenk S, Saberi M, Olefsky JM. Insulin sensitivity: modulation by nu- liferationofhepaticstellatecellsviaactivationofadenosinemonophosphate-
trientsandinflammation.JClinInvest2008;118:2992–3002 activatedproteinkinase.Hepatology2008;47:677–685
7. Uysal KT, Wiesbrock SM, Marino MW, Hotamisligil GS. Protection from 30. Marra F, Bertolani C. Adipokines in liver diseases. Hepatology 2009;50:
obesity-induced insulin resistance in mice lacking TNF-alpha function. 957–969
Nature1997;389:610–614 31. SaundersWB,BohnsackBL,FaskeJB,etal.Coregulationofvasculartube
8. Cai D, Yuan M, Frantz DF, et al. Local and systemic insulin resistance stabilization by endothelial cell TIMP-2 and pericyte TIMP-3. J Cell Biol
resulting from hepatic activation of IKK-beta and NF-kappaB. Nat Med 2006;175:179–191
2005;11:183–190 32. RupnickMA,Panigrahy D,ZhangCY,etal.Adiposetissue masscanbe
9. WinerS,ChanY,PaltserG,etal.Normalizationofobesity-associatedin- regulatedthroughthevasculature.ProcNatlAcadSciUSA2002;99:10730–
sulinresistancethroughimmunotherapy.NatMed2009;15:921–929 10735
10. FedericiM,HribalML,MenghiniR,etal.Timp3deficiencyininsulinreceptor- 33. BernotD,BarruetE,Poggi M,BonardoB,Alessi MC,Peiretti F.Down-
haploinsufficientmicepromotesdiabetesandvascularinflammationviain- regulationoftissueinhibitorofmetalloproteinase-3(TIMP-3)expressionis
creasedTNF-alpha.JClinInvest2005;115:3494–3505 necessaryforadipocytedifferentiation.JBiolChem2010;285:6508–6514
11. Murphy G, Murthy A, Khokha R. Clipping, shedding and RIPping keep 34. WangY,SulHS.Ectodomainsheddingofpreadipocytefactor1(Pref-1)by
immunityoncue.TrendsImmunol2008;29:75–82 tumornecrosisfactoralphaconvertingenzyme(TACE)andinhibitionof
12. DavisGE,SengerDR.Extracellularmatrixmediatesamolecularbalance adipocytedifferentiation.MolCellBiol2006;26:5421–5435
betweenvascularmorphogenesisandregression.CurrOpinHematol2008; 35. Österreicher CH, Penz-Österreicher M, Grivennikov SI, et al. Fibroblast-
15:197–203 specificprotein1identifiesaninflammatorysubpopulationofmacrophages
13. Menghini R, Menini S, Amoruso R, et al. Tissue inhibitor of metal- intheliver.ProcNatlAcadSciUSA2011;108:308–313
loproteinase3deficiencycauseshepaticsteatosisandadiposetissuein- 36. MurthyA,DefamieV,SmooklerDS,etal.EctodomainsheddingofEGFR
flammationinmice.Gastroenterology2009;136:663–672,e4 ligandsandTNFR1dictateshepatocyteapoptosisduringfulminanthepa-
14. FiorentinoL,VivantiA,CavaleraM,etal.Increasedtumornecrosisfactor titisinmice.JClinInvest2010;120:2731–2744
alpha-converting enzyme activity induces insulin resistance and hep- 37. ZouC,MaJ,WangX,etal.LackofFasantagonismbyMetinhumanfatty
atosteatosisinmice.Hepatology2010;51:103–110 liverdisease.NatMed2007;13:1078–1085
15. MonroyA,KamathS,ChavezAO,etal.ImpairedregulationoftheTNF-alpha 38. FoveauB,AncotF,LeroyC,etal.Down-regulationofthemetreceptor
convertingenzyme/tissueinhibitorofmetalloproteinase3proteolyticsystem tyrosine kinase by presenilin-dependent regulated intramembrane pro-
inskeletalmuscleofobesetype2diabeticpatients:anewmechanismof teolysis.MolBiolCell2009;20:2495–2507
insulinresistanceinhumans.Diabetologia2009;52:2169–2181 39. GerinI,LouisGW,ZhangX,etal.Hyperphagiaandobesityinfemalemice
16. Cardellini M, Menghini R, Martelli E, et al. TIMP3 is reduced in athero- lacking tissue inhibitor of metalloproteinase-1. Endocrinology 2009;150:
sclerotic plaques from subjects with type 2 diabetes and increased by 1697–1704
SirT1.Diabetes2009;58:2396–2401 40. JiangH,ZhuH,ChenX,etal.TIMP-1transgenicmicerecoverfromdiabetes
17. EwensKG,GeorgeRA,SharmaK,ZiyadehFN,SpielmanRS.Assessmentof inducedbymultiplelow-dosestreptozotocin.Diabetes2007;56:49–56
115candidategenesfordiabeticnephropathybytransmission/disequilibrium 41. BodenG,SongW,PashkoL,KresgeK.Invivoeffectsofinsulinandfreefatty
test.Diabetes2005;54:3305–3318 acidsonmatrixmetalloproteinasesinrataorta.Diabetes2008;57:476–483
18. BarthJL,YuY,SongW,etal.Oxidised,glycatedLDLselectivelyinfluences 42. DuvalC,ThissenU,KeshtkarS,etal.Adiposetissuedysfunctionsignals
tissue inhibitor of metalloproteinase-3 gene expression and protein pro- progression of hepatic steatosis towards nonalcoholic steatohepatitis in
ductioninhumanretinalcapillarypericytes.Diabetologia2007;50:2200–2208 C57BL/6mice.Diabetes2010;59:3181–3191
19. GellingRW,YanW,Al-NooriS,etal.DeficiencyofTNFalphaconverting 43. Jaworski DM, Sideleva O, Stradecki HM, et al. Sexually dimorphic diet-
enzyme (TACE/ADAM17) causes a lean, hypermetabolic phenotype in inducedinsulinresistanceinobesetissueinhibitorofmetalloproteinase-2
mice.Endocrinology2008;149:6053–6064 (TIMP-2)-deficientmice.Endocrinology2011;152:1300–1313
20. SerinoM,MenghiniR,FiorentinoL,etal.Miceheterozygousfortumorne- 44. Demeulemeester D, Scroyen I, Voros G, et al. Overexpression of tissue in-
crosisfactor-alphaconvertingenzymeareprotectedfromobesity-induced hibitorofmatrixmetalloproteinases-1(TIMP-1)inmicedoesnotaffectadi-
insulinresistanceanddiabetes.Diabetes2007;56:2541–2546 pogenesisoradiposetissuedevelopment.ThrombHaemost2006;95:1019–1024
21. ChaveyC,MariB,MonthouelMN,etal.Matrixmetalloproteinasesaredif- 45. Kunz-SchughartLA,WeberA,RehliM,etal.The“classical”macrophage
ferentiallyexpressedinadiposetissueduringobesityandmodulateadipocyte marker CD68 is strongly expressed in primary human fibroblasts. Verh
differentiation.JBiolChem2003;278:11888–11896 DtschGesPathol2003;87:215–223[inGerman]
462 DIABETES,VOL.61,FEBRUARY2012 diabetes.diabetesjournals.org