articles

Pathwaytodiabetesthroughattenuationofpancreatic betacellglycosylationandglucosetransport
Kazuaki Ohtsubo13, Mark Z Chen4, Jerrold M Olefsky4 & Jamey D Marth1,2
A connection between diet, obesity and diabetes exists in multiple species and is the basis of an escalating human health problem. The factors responsible provoke both insulin resistance and pancreatic beta cell dysfunction but remain to be fully identified. We report a combination of molecular events in human and mouse pancreatic beta cells, induced by elevated levels of free fatty acids or by administration of a high-fat diet with associated obesity, that comprise a pathogenic pathway to diabetes. Elevated concentrations of free fatty acids caused nuclear exclusion and reduced expression of the transcription factors FOXA2 and HNF1A in beta cells. This resulted in a deficit of GnT-4a glycosyltransferase expression in beta cells that produced signs of metabolic disease, including hyperglycemia, impaired glucose tolerance, hyperinsulinemia, hepatic steatosis and diminished insulin action in muscle and adipose tissues. Protection from disease was conferred by enforced beta cellspecific GnT-4a protein glycosylation and involved the maintenance of glucose transporter expression and the preservation of glucose transport. We observed that this pathogenic process was active in human islet cells obtained from donors with type 2 diabetes; thus, illuminating a pathway to disease implicated in the diet- and obesity-associated component of type 2 diabetes mellitus. Type2diabetesisadiseaseofimpairedglucosehomeostasisand insulinactionwithanetiologythatencompassesgenetic,cellular andenvironmentalfactors,2.Humantype2diabetesaffectshundredsofmillionsofpeople,andcasesarepredictedtodoublewithin 20years3.Thisescalationinfrequencyhasbeencloselyassociated with a widespread change in the human diet in the presence of o besity. Although only a fraction of overweight and obese individualsarediabetic,themajorityofnewlydiagnosedcasesoftype 2diabetesreflectthisgrowingsegmentofthehumanpopulation4. Ahigh-fatorWestern-styledietleadingtoobesityisevidentlya predisposingfactorindiseasesusceptibilityandonset5.Moreover,a linkbetweenobesityanddiabetesisnotrestrictedtohumans.When providedtovariousanimalspecies,ahigh-fatdietleadstoobesity andinducesdiseasesignsthatmodelhumantype2diabetes68.Yet itisunclearhowdietandobesitytriggerdiabeticpathophysiology. Geneticvariationseemstocontributetoonlyasmallfractionof diseasecases9. Insulinresistanceisametabolichallmarkoftype2diabetes.Insulin actioncanbeimpairedbytheinheritanceofgeneticmutationsthat disrupt essential components in the insulin action pathways0. Althoughthisetiologymayrepresentonlyasmallproportionoftype2 diabetescases,insulinresistancecontributessubstantiallyandbroadly tothepathophysiologyofdiabetesinsubjectswiththisdisease,2. Theidentificationofmolecularpathwaysbywhichorgan-specificand systemicinsulinresistancedevelopinresponsetodietandobesity isimportantforunderstandingtheetiologyoftype2diabetesand learninghowtopreventthisdisease.
1Center

2011 Nature America, Inc. All rights reserved.

Pancreaticbetacelldysfunctionisalsoadiagnosticdeterminantof type2diabetesandincludesdefectiveinsulinsecretionexemplified bythelossofglucose-stimulatedinsulinsecretion(GSIS).Normally, betacellssenseelevationsofbloodglucosebyexpressingcellsurface glucosetransportersthatenableconcentration-dependentglucose transportacrosstheplasmamembrane.Thisisfollowedbyglucokinaseactionandultimatelycalcium-dependentinsulinsecretion. LossofGSIShasbeenlinkedtoimpairedbetacellglucosetransporter (Glut)expressionamonganimalmodels,andfailureofGSIShasbeen proposedtofurtherprovokethepathogeniconsetoftype2diabetes inhumans324. A model of type 2 diabetes has been reported in mice lacking the GnT-4a glycosyltransferase encoded by the Mgat4a gene25. GnT-4ageneratesthecore-4GlcNAclinkageamongtheN-glycan structuresofsomeglycoproteins26.GnT-4afunctionpromotesbeta cellsurfaceresidencyoftheSlc2a2-encodedglucosetransporter-2 (Glut-2)glycoproteinneededforglucoseuptakeandGSIS,andits deficiency is induced by high-fat diet administration concurrent withtheonsetofdiabetes25.Theseobservationssuggestthatbetacell GnT-4aglycosylationandglucosetransporthaveimportantrolesin thepathogenesisofdiabetes. WeundertookstudiestoidentifyfactorscontrollingGnT-4afunctionandglucosetransporterexpressionandtodeterminewhether deficientGnT-4aproteinglycosylationandglucosetransportcause disease pathogenesis in the mouse model. We further examined humanisletsfromnormalandtype2diabetesdonorsforthepresence ofasimilarregulatoryandpathogenicprocess.Ourstudiesreveal

for Nanomedicine, Sanford-Burnham Medical Research Institute, University of CaliforniaSanta Barbara, Santa Barbara, California, USA. 2Department of Molecular, Cellular and Developmental Biology, University of CaliforniaSanta Barbara, Santa Barbara, California, USA. 3Disease Glycomics Team, RIKEN Advanced Science Institute, Saitama, Japan. 4Department of Medicine, University of CaliforniaSan Diego, La Jolla, California, USA. Correspondence should be addressed to J.D.M. (jmarth@sanfordburnham.org). Received 7 April; accepted 7 June; published online 14 August 2011; doi:10.1038/nm.2414

nature medicine advance online publication

articles

In pu t N con or m trol Ac al H ser 4 a In Ab pu tc N or ont m ro Ac al s l er H 4 a Ab AcH4 bound (% standard diet)

ut N con or m tro Fo al s l xa er a In 2 A pu b tc N or ont m ro Fo al s l xa er a 2 Ab Foxa2 bound (% standard diet)

a
RNA abundance (%)
120 100 80 60 40 20 0

Standard diet (SD) High-fat diet (HFD)

& $

b
Mgat4a Slc2a2 Ins2

140 120 100 80 60 40 20 0

**

c
Mgat4a Slc2a2 Ins2

125 100 75 50 25 0

***

***

In p

t4 a

In s

M ga

t4 a

a2

M ga

d
ut N con or m trol H al s nf 1 era In a A pu b t N con or m tro H al l nf se 1a ra Ab Hnf1a bound (% standard diet) In p
Mgat4a Slc2a2 Ins2

Nuclear locaization (%)

120 100 80 60 40 20 0

***

***

e
SD

Foxa2
+PI +PI +PI

Hnf1a
100 80 60 40 20 0

Sl c2

***

HFD 5 m 5 m 07-633 M-20 5 m

t4 a

a2

2011 Nature America, Inc. All rights reserved.

Figure 1 Dietary regulation of *** 175 Control Foxa2 Hnf1a Foxa2 & Hnf1a Mgat4a and Slc2a2 gene expression 150 siRNA siRNA siRNA siRNA by Foxa2 and Hnf1a in mouse 125 1.0 *** pancreatic islet cells. (a) Abundance 100 of mRNA produced from Mgat4a, 0.8 75 Slc2a2 and Ins2 genes in mouse 0.6 50 pancreatic islet cells (>90% beta 0.4 25 cells) isolated from 18- to 0 22-week-old WT mice on standard 0.2 (SD) and high-fat diet (HFD) dietary Foxa2 Pdx1 Hnf4a Arnt Hnf1a 0 Foxa2 Hnf1a Slc2a2 Mgat4a Actb regimens. (bd) ChIP and real-time Actb PCR (rtPCR) analysis of acetylated histone H4 (AcH4) bound to promoter regions of the Mgat4a, Slc2a2 and Ins2 genes (b) and binding to these regions by Foxa2 (c) and Hnf1a (d). ND, not detected. (e) In situ localization of Foxa2 and Hnf1a proteins in beta cells from mouse islet tissue in response to diet. Results represent analyses of >24 fields of view consisting of >100 beta cells. Two different antibodies that detect Foxa2 (07-633 and M-20) were used. Image analyses quantified percentage of cellular protein localized to nuclear region. (f) Total islet cell abundance of Foxa2, Pdx1, Hnf4a, Arnt and Hnf1a proteins determined using antibodies specific for each factor. (g) Pancreatic beta cells from mice receiving SD transfected with siRNAs to knock down Foxa2 and Hnf1a, or with a siRNA control, were cultured for 72 h followed by mRNA abundance measurements using rtPCR. Mice for study were normal C57BL/6J mice 1822 weeks old before initial experimentation. Data in af are means s.e.m. of triplicate experiments per mouse from 12 mice consisting of 6 separate littermate pairs. &P = 0.0007; $P = 0.0049; *P = 0.0421; **P = 0.0418; ***P < 0.0001 (Students t test, af; Bonferroni test after analysis of variance (ANOVA), g).

Sl c2

ga

In s

SD

HFD

Foxa2

Protein abundance (% standard diet)

asequenceofmoleculareventsthatlinksbetacellspecificGnT-4a glycosylationtothecontrolofglycemia,insulinemiaandglucosetoleranceaswellashepaticsteatosisandsystemicinsulinresistance.Our findingsindicatethepresenceofaconservedpathogenicpathwayto diabetesinmiceandhumansthatislinkedtoadeficiencyofbetacell proteinglycosylationandglucosetransport. RESULTS Control of Mgat4a and Slc2a2 expression by Foxa2 and Hnf1a Wild-type(WT)C57BL/6Jmicefedahigh-fatdietbecamedeficient inMgat4aandSlc2a2(alsoknownasGlut2)RNAabundanceinpancreaticisletcells,within34weeksandcontinuingthereafter,comparedwithcohortsfedthestandarddiet(Fig. 1a;ref.25).Wedetected reducedacetylationofhistoneH4inthepromoterregionsoftheMgat4a andSlc2a2genes,suggestingdiminishedtranscriptionofthesealleles (Fig. 1b).ThroughpromoterregionDNAsequenceanalysesofmouse genesMgat4aandSlc2a2,theorthologoushumangenesMGAT4Aand SLC2A1(alsoknownasGLUT1)andthehumanSLC2A2gene,weidentifiedpotentialbindingsitesofmultipletranscriptionfactorsincluding mouseandhumanFOXA2andHNFA(Supplementary Fig. 1). Throughisletcellchromatinimmunoprecipitation(ChIP)analyses frommicereceivingthestandarddiet,wefoundthatFoxa2andHnfa proteinswereboundtothecorrespondingpromoterregionsofthe
2

Mgat4aandSlc2a2genes.Thisbindingwassignificantlyreducedin isletsfrommicethathadreceivedthehigh-fatdiet(Fig. 1c,d)coincidentwithamarkedreductioninnuclearlocalizationofFoxa2and Hnfafrom>80%to3%and29%,respectively,andconcurrentwith increasedcytoplasmiclocalizationofbothfactors(Fig. 1e).Inisletcells, ourmeasurementsofproteinabundancefurthershowedthatFoxa2 wasunalteredwhereasHnfawasreducedby50%(Fig. 1f).Together, nuclearexclusionandreducedexpressionofHnfacombinedtoyield anoverallnucleardeficiencyof>80%,similartothatofFoxa2. WemeasuredFoxa2andHnfatransactivationoftheMgat4aand Slc2a2genesusingsiRNAsdesignedtospecificallyknockdownthese twotranscriptionfactorsinculturedWTmouseisletcells.BothFoxa2 andHnfawerefoundtotransactivatetheMgat4aandSlc2a2genes, andtheknockdownofbothfactorsdiminishedMgat4aandSlc2a2 RNAinmicefedthestandarddiettosimilarlevelsmeasuredafter high-fatdietfeeding(Fig. 1g).Foxa2andHnfaeachcontributedto Mgat4aandSlc2a2transactivationinisletsofnormalmicefedthe standarddiet,andthistransactivationwasimpairedinisletsofmice fedthehigh-fatdiet. Palmitic acid recapitulates beta cell dysfunction IsletcellswereisolatedfromWTmicefedthestandarddietandculturedfor48hwithorwithoutpalmiticacid,alipidusedtomodelthe

RNA expression (relative to untreated)

advance online publication

nature medicine

In s

Mgat4a Slc2a2

Ins2

Sl c2

SD

HFD

***

Hnf1a

SD

HFD

ND ND

a2

articles a
Foxa2 +PI Hnf1a +PI Nuclear locaization (%) 100 80 60 40 20 0 Foxa2 Hnf1a +PA

No addition +PA +NAC

FOXA2

HNF1A

+PI

+PI
Nuclear locaization (%) 100 80 60 40 20 0

No addition 5 m +PA

****

***

5 m

****

****

No addition 5 m +PA
5 m

5 m

5 m

5 m

+PA +NAC

5 m

5 m PA NAC 1.2 1.0 0.8 0.6 0.4 0.2 0 1.2 1.0 0.8 0.6 0.4 0.2 0 1.2 1.0 0.8 0.6 0.4 0.2 0 1.2 1.0 0.8 0.6 0.4 0.2 0 +

5 m + + +

FOXA2

HNF1A

In pu N tc or on FO ma tro X ls l In A2 era pu A N tc b or on FO ma tro XA l se l 2 ra Ab FOXA2 bound (% of control)

b
Mgat4a

d
MGAT4A SLC2A1 SLC2A2

Relative expression levels

** *

*
100 75 50 25 0

*
NA +PA

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

**

** *

** *

INS No addition +PA

ND ND

Relative mRNA abundance

4A

C 2A

C 2A

IN S

0 h 48 h 0 h 48 h 0 h 48 h 0 h 48 h MGAT4A SLC2A1 SLC2A2 INS

2011 Nature America, Inc. All rights reserved.

In p N ut c or o H ma ntro N l l In F1A ser pu A a N t co b or n m H al trol N s F1 e ra Ab

125 HNF1A bound ( % of control) 100 75 50 25 0

****

Insulin release (ng ml1 5 min1 5 104 cells1)

e
MGAT4A SLC2A1

Slc2a2

G AT

SL

SL

Foxa2

4A

**

**

SLC2A2 INS No addition +PA

2A

2A

AT

SL

Hnf1a

Figure 2 Effect of palmitic acid on normal mouse and human islet cells. (a) Subcellular localization of Foxa2 and Hnf1a proteins within beta cells from islet tissue measured in response to palmitic acid (PA) and N-acetylcysteine (NAC). Propidium iodide (PI) staining of the nucleus (red) is also included where indicated. (b) mRNA expression of Mgat4a, Slc2a2, Foxa2 and Hnf1a measured by rtPCR from WT islet cells isolated and cultured in medium containing 5 mM glucose for 48 h with or without PA and NAC. Islets in a and b were isolated from normal C57BL/6J mice maintained on standard diet. (c) In situ localization of FOXA2 and HNF1A proteins in beta cells from normal human islet tissue measured with or without PA. (d,e) ChIP and rtPCR measurement of FOXA2 (d) and HNF1A (e) proteins bound to promoter regions of human MGAT4A, SLC2A1, SLC2A2 and INS genes. ND, not detected; NA, no addition. (f) Abundance of mRNA from MGAT4A, SLC2A1, SLC2A2 and genes in human islet cells. (g) Glucose-stimulated insulin secretion assayed in islet cell cultures containing medium bearing indicated concentrations of glucose. Results in a and c represent analyses of >24 fields of view consisting of >100 beta cells. Data are means s.e.m. of triplicate experiments per mouse from six mice fed standard diet and four normal human islet donors. *P = 0.010.049; **P = 0.0050.009; ***P = 0.0004; ****P < 0.0001 (Students t test).

diabetogeniceffectofincreasedfreefattyacids.Palmiticacidinduced nuclearexclusionofFoxa2andHnfa,whereasco-incubationwith theantioxidantN-acetylcysteineblockedthiseffect(Fig. 2a).Palmitic acid treatment further caused the downregulation of Mgat4a and Slc2a2 gene expression, coincident with diminished Hnfa RNA, whereas co-incubation with N-acetylcysteine preserved normal expression(Fig. 2b). Wesimilarlystudiedhumanisletsfromhealthynondiabeticdonors tomeasuretranscriptionfactorbinding,geneexpressioninresponse topalmiticacid,andGSISactivity.Intheabsenceofexogenouspalmiticacidaddition,FOXA2andHNFAproteinscolocalizedpredominantlywithnuclearmarkers,andtheadditionofpalmiticacid causednuclearexclusionofFOXA2andHNFAwithacorrespondingincreaseincytoplasmiclocalization(Fig. 2c).Bothtranscription factorswereboundtothepromoterregionsofthehumanMGAT4A, SLC2A1 and SLC2A2 genes in untreated islet cell cultures from healthydonors.TheadditionofpalmiticaciddiminishedFOXA2 andHNFAbindingandattenuatedmRNAexpressionofMGAT4A, SLC2A1andSLC2A2,coincidentwiththelossofGSIS(Fig. 2dg). Thesimilaritiesweobservedintheresponsesofnormalmouseand

humanisletcellstopalmiticacidsuggestedthatisletsfromhuman type2diabetesdonorsmayshowdefectscomparabletothoseofmice rendereddiabeticbyhigh-fatdietadministration. Human islet cell dysfunction in type 2 diabetes Inanalysesofsixnormalhumandonorislets,wefoundthatFOXA2 andHNFApredominantlylocalizedwiththecellnucleus,whereas weobservedtheirnuclearexclusionby>70%intheisletcellsoftwo independenthumandonorswithtype2diabetes(Fig. 3a).Inaddition,expressionofSLC2A1andSLC2A2mRNAsencodingthehuman GLUTandGLUT2glucosetransporterswasdecreasedby7090% inhumantype2diabetesislets,andMGAT4AmRNAexpressionwas reducedby60%(Fig. 3b).TherelativelymodestdecreaseofMGAT4A RNAabundancetranslatedintoa0-to50-foldreductionoftheGnT 4aglycanproductmeasuredas-4GlcNAcglycanlinkagespresentat thecellsurfaceusingDSAlectinbinding(Fig. 3c).GLUTisthepredominantglucosetransporterexpressedinnormalhumanisletcells27; however,wedetectedbothGLUTandGLUT2glycoproteins.Inour findings,isletcellsfromdonorswithtype2diabetesweredeficient incellsurfaceexpressionofbothGLUTandGLUT2glycoproteins,
3

nature medicine advance online publication

SL

IN

22.5 20.0 17.5 15.0 12.5 10.0 7.5 5.0 2.5 0

2.8 16.8 2.8 16.8 mM glucose in culture 0h 48 h

articles a
+PI Normal 5 m 5 m FOXA2 +PI HNF1A Nuclear locaization (%) Normal T2D 100 **** **** 80 60 40 20 0 FOXA2 HNF1A

Relative expression

2.5 2.0 1.5 1.0 0.5 0

Normal

T2D

c Autofluorescence/normal d
T2D1/T2D2 Cell count Cell count

Autofluorescence/2Ab Normal/T2D1/T2D2

****

****

DSA lectin

GLUT-1 Mean fluorescence 300 200 100 0 GLUT-1 800 600 400 200 0

GLUT-2 GLUT-2

SL C 2A 1 SL C 2A 2

AT 4A

T2D1 5 m 5 m

Insulin release 4 (ng ml1 5 min1 5 10 cells1)

Mean fluorescence

e Autofluorescence/normal
T2D1/T2D2 Cell count

IN S

T2D2 5 m 5 m

N A or F m T2 al D T2 1 D 2

2-NBDG

400 300 200 100 0

17.5 15.0 12.5 10.0 7.5 5.0 2.5 0

Normal T2D1 T2D2

*** ****

*** ***

2.8 mM

16.8 mM Sucrose Insulin release 4 (ng ml1 5 min1 5 10 cells1) 10 8 6 4 2 0 0 mM 2.8 mM 16.8 mM 25.0 20.0 15.0 10.0 5.0 0 2.8 mM 16.8 mM LacNAc (LacNAc)3

Cell count

2-NBDG uptake (nmol 2 min1 1 105 cells1)

Ratio to no addition

Blot: GLUT-1/-2 DSA L-PHA TL MAL I SNA RCA I

IP: GLUT-1 GLUT-2

+/ 2 6

h
4

AF/2Ab/GLUT Sucrose

LacNAc

(LacNAc)3

GLUT-1

2011 Nature America, Inc. All rights reserved.

Asn

GLUT-2

1.2 0.8 0.4 0 1.2 0.8 0.4 0

No addition 1.0 0.8 0.6 0.4 0.2 0

AF N 2A or b m T2 al D T2 1 D 2

Insulin release 4 (ng ml1 5 min1 5 10 cells1)

Mean fluorescence

j
SLC2A1

ECA

Cell count

**

*
1.5 1.0 0.5 0

*
SLC2A2

MGAT4A

Cell count

1.5 1.0 0.5 0

1.2 0.8 0.4 0

Autofluorescence/ siRNA control/siGLUT1 siGLUT2/siGLUT1&2 INS 1.0 0.5 0

200 150 100 50 0

** *** ***

LU LU T2 T1 &2

T1

LU T LU 2 T1 &2

T1

T1

T2

T1

&2

T2

tro

tro

tro

tro

&2

LU

si

si

si

si

si

si

si

si

si

si

si

si

LU

2-NBDG

si C A si ont F G ro s LU l si iGL T1 G U LU T T1 2 &2

Figure 3 Analyses of human islets from normal donors and donors with type 2 diabetes. (a) Subcellular localization of FOXA2 and HNF1A proteins in human beta cells from normal donors and donors with type 2 diabetes (T2D). Results represent analyses of six normal islet samples and two T2D islet samples. Propidium iodide (PI) staining of the nucleus (red) is included where indicated. (b) Abundance of mRNA produced from MGAT4A, SLC2A1, SLC2A2 and INS genes in the designated human islet cells. (c) Islet cell surface abundance of the DSA lectin-binding glycan produced by the GnT-4a glycosyltransferase. (d) Islet cell surface expression of human GLUT-1 and GLUT-2 glucose transporters. (e) Glucose transport activity of the indicated human islets measured using the fluorescent glucose analog 2-NBDG. (f) Glucose-stimulated insulin secretion assayed in islet cell cultures containing medium bearing the indicated concentrations of glucose. (g) Left, GLUT-1 and GLUT-2 glycoproteins were immunoprecipitated (IP) from normal human islet cell extracts followed by electrophoresis and analyses with the indicated antibodies and lectins. Right, deduced tetra-antennary N-glycan structure residing on both human islet cell GLUT-1 and GLUT-2 bearing undersialylated glycan branch termini (+/) . Gray circle, core 1-4GlcNAc linkage produced by GnT-4a. (h) Normal human islet cells were cultured with or without the indicated glycans (10 mM) for 2 h before analyses of cell surface GLUT-1 and GLUT-2 expression by flow cytometry. (i) Fluorescent glucose analog (2-NBDG) transport (left) and GSIS activity (right) measured among islet cells treated in h. The results in h and i represent analyses of three islet cell samples from normal human donors. (j) siRNA knockdown of GLUT1 and GLUT2 mRNA in normal human islet cell cultures measured at 72 h (left four graphs). Glucose analog 2-NBDG transport and GSIS activity were also measured at 72 h. Data are expressed as means s.e.m. from six normal human islet donors and two human donors with T2D, unless otherwise stated. *P = 0.010.04; **P = 0.0020.005; ***P = 0.00020.0005; ****P < 0.0001 (Students t test).

withan8090%reductiononaverage(Fig. 3d).Furthermore,human type2diabetesisletcellshadverylowglucosetransportactivityand lackedtheGSISresponse(Fig. 3e,f). Human GnT-4a glycosylation and glucose transporter expression DownregulationofhumanisletcellsurfaceGLUTandGLUT2glycoproteinsassociatedwiththelossofGSISintype2diabetesmayinvolve thelackofsufficientGnT-4aactivity,ashasbeenobservedinthemouse25. Theresultsoflectin-bindinganalysesofhumanGLUT-andGLUT-2 N-glycansfromhealthydonorswereconsistentwiththepresenceofan undersialylatedtetra-antennarystructurebearingthecore-4GlcNAc glycan linkage produced by GnT-4a (Fig. 3g). This tetra-antennary N-glycanexistsontheGlut-2glycoproteinofmousebetacellsandhas beenproposedtocontributetoformationofalectinligandthatcontrolstherateofglucosetransporterendocytosisfromthecellsurface andsubsequentdegradation25.Supportforthismodelinhumanswas gained after the addition of the lectin ligand N-acetyllactosamine to normalisletcellcultures,whichledtoarapiddeclineincellsurface expressionofbothGLUT-andGLUT-2glycoproteins,indicatingthat
4

botharestabilizedatthecellsurfacebyalectin-ligandinteractionbearing similarbindingspecificity(Fig. 3h).Decreasesincellsurfaceglucose transporterexpressionwereproportionaltoreductionsinglucosetransportandGSIS(Fig. 3i). WecomparedtherolesofGLUT-andGLUT-2inglucosetransportandGSISinnormalhumanisletcellculturesbyapplyingsiRNAs designed to knock down RNA expression from the corresponding genes. We measured a significant decrease in mRNA expression involvingthecorrespondingglucosetransporters(Fig. 3j).However, onlytheadditionofbothSLC2A1andSLC2A2siRNAsledtotheeliminationofGSIS,andthisalsocoincidedwiththelargestimpairmentof glucosetransport(Fig. 3j).Thesefindingsindicatethatendogenous expressionofeitherGLUT-orGLUT-2issufficienttomaintainGSIS inhumanbetacells,andarealsoconsistentwithstudiesreporting that expression of either human SLC2A1 or SLC2A2 genes can re-establish GSIS activity in Glut-2 deficient mouse beta cells28. Ourfindingsfurthersupporttheviewthatdiminishedcellsurface e xpressionofGLUT-andGLUT-2observedinhumanisletsfrom donorswithtype2diabetesisresponsibleforthelossofGSIS.

advance online publication

C o si ntr G ol si LU G T2 si L si GL UT N U 1 eg T1 a /2 si Co G n si LUT t si GL 2 G U LU T1 T1 /2

LU

LU

LU

LU

LU

LU

T1

on

on

on

on

T1

si

si

si

si

si

nature medicine

AF N 2A or b m T2 al D T2 1 D 2

***

articles a
RIP2 promoter
MGAT4A cDNA

pA signal

978 980

Islets Brain Heart Lung Liver Kidney S.int Spleen RT + + + + + + + + MGAT4A Actb MGAT4A Actb

c
GLUT-2 DSA Actb

WT

Tg 978

Tg 980

WT

Tg 978

Tg 980

WT (SD)
SD HFD 8 7 6 5 4 3 2 1 0

WT (HFD) Tg 978 (HFD) Tg 980 (HFD)

b
SD

Wild type

MGAT4A 978

MGAT4A 980

Tg 978 (SD) Tg 980 (SD)

d
Glucose (mg dl )
1

300

40

Insulin (ng ml )

Body weight (g)

0 4 8 12 Time (week) 16

250 200 150 100 0 4 8 12 Time (week) 16

35 30 25 20 15 0 4 8 12 Time (week) 16

HFD 5 m 5 m 5 m

Glut-2 / insulin / DAPI

Insulin (ng ml min1 5 105 cells1)

e
Glucose (mg dl )
1

600 500

f
Insulin (ng ml )
1

2011 Nature America, Inc. All rights reserved.

Figure 4 Enforced beta cellspecific GnT-4a 2.5 400 2.0 glycosylation prevents loss of Glut-2 expression 1.5 300 and inhibits onset of disease signs including 1.0 200 0.5 hyperglycemia and failure of GSIS. (a) Human 0 100 MGAT4A cDNA was incorporated into a transgene 0 30 60 90 120 0 4 8 12 16 20 0 60 120 180 240 0 30 60 90 120 Time (min) Time (min) Time (min) Time (min) vector that conferred beta cellspecific expression in multiple tissues of two separate founder lines, 978 and 980. Transgene expression was detected using vector-specific primers. (b) Pancreatic beta cell histology from WT and MGAT4A transgenic littermates that received either standard diet (SD) or high-fat diet (HFD) for the preceding 10 weeks. Antibody binding and visualization of GLUT-2 (green), insulin (red) and DNA (DAPI-blue). (c) GLUT-2 protein abundance and glycosylation in beta cells from WT or MGAT4A (Tg) littermates as in b were analyzed by blotting GLUT-2 immunoprecipitates with antibody to GLUT-2 or DSA lectin. Single analysis shown represents three independent experiments with different littermates. (d) Blood glucose (left), blood insulin (center) and body weight (right) were measured (unfasted) every 2 weeks for up to 16 weeks of HFD administration (e,f). In fasted mice receiving SD or the HFD for 10 weeks, glucose was injected into the intraperitoneal space before analysis of glucose tolerance (e) and insulin release (f). (g) GSIS activity was analyzed ex vivo by perifusion in islet cells isolated from mice that received either SD or HFD for the preceding 4 weeks. Glucose concentration was increased from 2.8 mM to 16.8 mM at the time indicated. Data are mean s.e.m. of three independent studies of beta cells from distinct littermates. Data from WT beta cells, red line. (h) Insulin secretion response to l-arginine injection measured in fasting WT and MGAT4A Tg mice receiving SD. Data are means s.e.m. in triplicate experiments per mouse, from six or seven mice of each genotype unless otherwise stated.

7 6 5 4 3 2 1 0

g
1

6 5 4 3 2 1 0

2.8 mM 16.8 mM

h
Insulin (ng ml )
1

3.5 3.0

Beta cell GnT-4a protects against diabetes We evaluated the impact of diminished GnT-4a glycosylation on g lucosetransporterexpressionandtheonsetofdiseasesignsinthe high-fatdietinducedmousemodeloftype2diabetes.Transgenic
Figure 5 Beta cellspecific GnT-4a protein glycosylation promotes systemic insulin sensitivity and inhibits development of hepatic steatosis. (a) Insulin challenge response measured in 1618 week old littermates of indicated genotypes after 10 weeks of standard diet (SD) or high-fat diet (HFD) administration. (b) Phosphorylation of Akt-1 at Thr308 a nd IRS-1 at Ser307 in equivalent cellular protein preparations from adipose and muscle tissue of mice as in a after 2 min of insulin perfusion through inferior vena cava. (c) Euglycemic and hyperinsulinemic clamp studies after 10 weeks on HFD compared measurements of glucose infusion rate, glucose disposal rate, insulin-stimulated glucose disposal rate, and suppression of hepatic glucose production in response to insulin. The analyses included nine WT and seven Tg mice. (d,e) Liver tissue observed macroscopically (d) and by histological analysis stained with H&E (e). *P = 0.0017; **P = 0.0235; ***P = 0.0401 (single-tail t test). Data are means s.e.m. from three or more littermates unless otherwise indicated.

miceweregeneratedbearingconstitutiveexpressionofthehuman MGAT4A gene in beta cells and not in other cell types or tissues s urveyed,includingthebrain(Fig. 4a).InWTmicereceivingthe high-fatdiet,betacellGlut-2ismostlyinternalizedintoendosomes

Glucose level (ratio to time point 0)

Adipose

30

60 Time (min)

120

WT (SD) Tg 978 (SD) Tg 980 (SD)

WT (HFD) Tg 978 (HFD) Tg 980 (HFD)

IRS-1 P-Ser307 IRS-1

Muscle

1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4

b
978 WT Tg 980 WT Tg 978 WT Tg 980 WT Tg Insulin + + + + Akt-1 P-Thr308 Akt-1 Insulin + + + + Akt-1 P-Thr308 Akt-1 IRS-1 P-Ser307 IRS-1

30 20 10 0

40 30 20 10 0 WT Tg Tg

20 15 10 5 0 WT Tg

Hepatic glucose production (percentage suppression)

c
Glucose infusion rate 1 1 mg kg min

40

Glucose disposal rate 1 1 (mg kg min )

50

**
Insulin-stimulated glucose disposal rate 1 1 (mg kg min )

25

***

80 60 40 20 0

WT WT

Tg

WT

Tg Tg

WT

10 mm

10 mm

20 m

20 m

nature medicine advance online publication

articles
andlysosomes,ashasbeenobserved25.Incontrast,Glut-2protein localizationwasfullyretainedatthebetacellsurfaceinMGAT4A transgeniclittermates(Fig. 4b).ThepreservationofbetacellsurfaceGlut-2expressioninMGAT4Atransgenicmiceonahigh-fat dietcoincidedwithpreservationofGnT-4aactivityevidencedbythe retentionofDSAlectinbinding(Fig. 4c). ComparedtoWTmicewithhyperglycemiaandhyperinsulinemia, MGAT4Atransgeniclittermatesmaintainedmuchlowerconcentrations
hMGAT4a

RIP2 promoter
GLUT2 cDNA

pA signal

b
Cell count

2ndAb

c
2-NBDG uptake (nmol 2 min1 5 104 cells1)

926

Islets Brain Heart Lung Liver Kidney S.int Spleen RT + + + + + + + + SLC2A2 Actb SLC2A2 Actb Wild type SLC2A2 926

Wild type

2.0 1.6 1.2 0.8 0.4 0

2-NBDG uptake (nmol 2 min1 5 104 cells1)

hGLUT2

** *
1.2

0.8

930

0.4

Glut-2 SLC2A2 930

d
SD

WT 926 930 SLC2A2 Tg 926

WT 978 980 MGAT4A Tg 926 Tg 930

e
GLUT-2 DSA

WT

Tg 930

WT

2011 Nature America, Inc. All rights reserved.

HFD 5 m 5 m Glut-2 / Insulin / DAPI 3.0 2.5 2.0 1.5 1.0 0.5 0 5 m

Actb SD HFD

f
Relative Glut-2 expression

Relative Glut-2 glycosylation

****** ******
SD

**** **** * *
HFD
Insulin 1 5 1 min 5 10 cells )

WT MGAT4A Tg 978 MGAT4A Tg 980 SLC2A2 Tg 926 SLC2A2 Tg 930

250 200 150 100 40 0 12 4 8 Time (week) 16

**** **** * **
SD HFD 7 6 5 4 3 2 1 0

0.75 0.50 0.25 0

Insulin (ng ml1)

1.50 1.25 1.00

Glucose (mg dl )

**** ****

g
1

300

8 7 6 5 4 3 2 1 0

h
600
Glucose (mg dl )

12 4 8 Time (week)

16

Body weight (g)

6 5 4 3 2 1 0

2.8 mM

16.8 mM

35 30 25 20 15 0 12 4 8 Time (week) 16

500 400 300 200 100 0 30 60 Time (min) 120

Insulin (ng ml1)

WT (SD) Tg 926 (SD) Tg 930 (SD)

WT (HFD) Tg 926 (HFD) Tg 930 (HFD)

30 60 Time (min)

120

(ng ml

8 12 16 Time (min)

20

** 30 20 40 40 Figure 6 Enforced beta ** 3.0 cellspecific expression of GnT-4a 30 15 30 2.5 substrate GLUT-2 mitigates diet- and 20 2.0 obesity-induced diabetes. (a) Human 20 10 20 1.5 SLC2A2 cDNA was incorporated into 1.0 a transgene vector that conferred 10 10 5 10 0.5 beta cellspecific expression in multiple tissues of two separate founder 0 0 60 120 240 0 0 0 0 lines, 926 and 930. (b) Beta cell surface WT Tg WT Tg WT Tg WT Tg Time (min) expression of GLUT-2 analyzed by flow cytometry from WT, MGAT4A Tg (978), and SLC2A2 Tg (926) mice receiving standard diet (SD). (c) Glucose analog 2-NBDG transport into cultured islet cells isolated from WT mice and Tg littermates fed SD. (d) Pancreatic beta cell histology of WT and MGAT4A Tg littermates that received either SD or high-fat diet (HFD) for the preceding 10 weeks. Antibody binding and visualization of Glut-2 (green), insulin (red) and DNA (blue). (e) Islet cell Glut-2 and GLUT-2 immunoprecipitates were analyzed for Glut-2 abundance and DSA lectin binding from WT or SLC2A2 Tg littermates administered indicated dietary stimuli. Single analysis shown represents three independent experiments with different littermates. (f) Histogram of GLUT-2 protein abundance (left) and GLUT-2 glycosylation (right) was calculated from results in e. (g) Blood glucose (left), blood insulin (right) and body weight (bottom) were measured in unfasted mice every 2 weeks for up to 16 weeks of indicated diet administration. (h) In fasted mice that had received SD or HFD for 10 weeks, glucose tolerance tests measured blood glucose (left) and insulin release (middle). GSIS activity was analyzed ex vivo by perifusion (bottom) in islet cells isolated from mice that received either SD or HFD for preceding 4 weeks. Glucose concentration was increased from 2.8 mM to 16.8 mM at time indicated. Data from WT beta cells, red line. (i) Insulin secretion response to l-arginine injection in fasting WT and SLC2A2 Tg mice receiving SD. (j) Euglycemic and hyperinsulinemic clamp studies after 10 weeks on HFD compared measurements of glucose infusion rate, glucose disposal rate, insulin-stimulated glucose disposal rate and hepatic suppression of gluconeogenesis. The clamp studies included eight WT and eight SLC2A2 Tg littermates. Data plotted throughout are means s.e.m. At least six or seven mice including littermates were studied in each experiment unless otherwise stated. *P = 0.010.04; **P = 0.0010.09; ***P = 0.00010.0009; ****P < 0.0001 (Students t test, c, f and g; single-tail t test, j).
Hepatic glucose production (percentage suppression) Glucose disposal rate (mg kg1 min1) Glucose infusion rate (mg kg1 min1) Insulin-stimulated glucose disposal rate (mg kg1 min1) Insulin (ng ml1)

**

advance online publication

nature medicine

articles
of blood glucose and insulin, whereas both groups became obese (Fig. 4dandSupplementary Table 1).FastedMGAT4Atransgenic mice had markedly greater glucose tolerance (Fig. 4e), whereas insulinemia was decreased and GSIS was preserved (Fig. 4f). Ex vivoperifusionstudiesofisletcellsalsoshowedthepreservation ofGSISactivityaffordedbyconstitutivebetacellGnT-4aexpression (Fig. 4g),whereasarginine-inducedinsulinreleasewasunaffectedby theMGAT4Atransgene(Fig. 4h).CirculatingfreefattyacidandtriglycerideconcentrationsinMGAT4Atransgenicmiceweremarkedly reducedincomparisontoWTlittermates(Supplementary Table 1), indicatingthatenforcedbetacellGnT-4aproteinglycosylationdiminished multipleandsystemicdiseasesigns. Beta cell GnT-4a increases insulin sensitivity and prevents steatosis Insulintolerancetestsinmiceadministeredthehigh-fatdietindicated greaterglucoseclearanceinMGAT4AtransgenicmicethaninWTmice (Fig. 5a).Wefurtheranalyzedsignaturesofinsulinactioninadipose andmuscletissue,involvingthephosphorylationofAkt-Thr308and IRS-Ser307(refs.29,30).Thepresenceofprotein-linkedphosphate wasconsiderablygreateronAkt-Thr308andsubstantiallyloweron IRS-Ser307inbothadiposeandmuscletissueofMGAT4Atransgenic micecomparedwithWTmice,consistentwiththelikelihoodofhigher systemicinsulinsensitivity(Fig. 5b).Wethereforefurtheranalyzed insulinactionusingeuglycemic-hyperinsulinemicclamps(Fig. 5c). Theglucoseinfusionrate,indicativeofwhole-bodyinsulinsensitivity, wassignificantlygreaterinMGAT4AtransgenicmicethanintheirWT littermates.Furthermore,theglucosedisposalrate,reflectingcombined tissueabilitytouptakeglucose,andtheinsulin-stimulatedglucosedisposalrate,indicatingmuscle-specificinsulinsensitivity,weresimilarly higherinMGAT4Atransgenicmice,indicatinggreatermuscleinsulin sensitivity.Insulinsuppressionofhepaticglucoseproductiontodetect inhibitionoflivergluconeogenesisbyinsulinwasvariablyincreased, insomecohortsmoreprofoundlythaninothers. HepaticsteatosiswasevidentinWTmicethatreceivedthehigh-fat diet.Incontrast,theliversofMGAT4Atransgenicmiceadministered thehigh-fatdietnotablylackedsignsofsteatosis(Fig. 5d,e).Together, thesefindingsindicatethatenforcedbetacellGnT-4aglycosylation impartedasystemicinfluencesubstantiallyimprovinginsulinaction in adipose and muscle tissue and conferring protection from the developmentofsteatosis. Diminished beta cell glucose transport in diabetes Disease protection afforded by GnT-4a may involve maintenance of beta cell glucose sensing. To test this hypothesis, we generated t ransgenic mice bearing constitutive beta cellspecific expression ofthehumanSLC2A2gene(Fig. 6a).TheSLC2A2transgeneledto greaterbetacellGLUT-2glycoproteinexpressionandglucosetransportascomparedwithbetacellsfromWTandMGAT4Atransgenic mice that received the standard diet (Fig. 6b,c). Both SLC2A2 t ransgeniclinesshowedretentionofglucosetransporterglycoprotein e xpression at the beta cell plasma membrane after high-fat diet administration,althoughwealsonotedanaccompanyingincrease incytosoliclocalization(Fig. 6d).Inaddition,GLUT-2glycoprotein a bundancewasdiminishedinbetacellsofSLC2A2transgenicmice afterhigh-fatdiettreatment,andcomparedwithMGAT4Atransgenic mice.Nevertheless,overallexpressionwasgreaterthanthatinWT c ounterpartsandtheGLUT-2moleculesremainingwereenrichedin glycoproteinsmodifiedbyGnT-4aactivity(Fig. 6e). WecomparedresultsandfoundthattheSLC2A2transgeneconferred an intermediate degree of improvement between WT and MGAT4Atransgenicmice.Forexample,glucosetransporterprotein abundancedecreasedinSLC2A2transgenicmiceafterhigh-fatdiet administration,incontrastwithamuchlowerlevelobservedinWT litteramatesandthecompleteretentionofendogenousGlut-2expressionobservedinMGAT4Atransgenicmice(Fig. 6f).Inanalysesof glucose transporter glycosylation, the high expression of GLUT-2 proteinproducedbytheSLC2A2transgeneseemedtosurpassthe capacityofendogenousGnT-4aactivitytoglycosylatetheincreased amount of Glut-2 (Fig. 6f). Glucose transporter glycosylation by GnT-4aremainedoptimalinisletcellsofMGAT4Atransgenicmice receivingthehigh-fatdiet,whereascorrespondingSLC2A2transgenic betacellsshowedadecreaseinGLUT-2glycosylationamongmicefed thehigh-fatdiet,consistentwiththedownmodulationofendogenous Mgat4aexpression,diminishedGnT-4aactivityanddecreasedGLUT-2 abundance(Fig. 6f). TheSLC2A2transgenelikewiseprovidedanintermediatereduction inhyperglycemiaandhyperinsulinemia,whereasthedevelopmentof obesitywasunaffected(Fig. 6g).Nevertheless,anintraperitonealglucosetolerancetestindicatedmarkedlygreaterglucosetoleranceand retentionofGSIS(Fig. 6h).WefurthernotedpreservationofGSIS inex vivoisletcellperifusionstudies(Fig. 6h),andinsulinreleasein responsetoargininewasunalteredbytheSLC2A2transgene(Fig. 6i). Inaddition,hyperinsulinemic-euglycemicclampstudiesofSLC2A2 transgenic mice indicated significantly greater systemic insulin s ensitivity(Fig. 6j).Uponexaminationofliversections,however,we foundnosignificantprotectionfromliversteatosis(datanotshown), incontrastwithourfindingsinMGAT4Atransgenicmice.TheintermediateandlesserimpactofenforcedGLUT-2expressioninbeta cells, as compared with GnT-4a expression, is consistent with the assignmentofGnT-4aactivityasthelimitingfactorinoptimalglucose transporterglycosylationandretentionatthebetacellsurface. Thesefindings,spanningbothmouseandhuman,canbeincorporatedintoamolecularandpathogenicpathwaythatincludesakeyrole ofpancreaticbetacellGnT-4aglycosylationandglucosetransportin theoriginandseverityofdiseasesignsincludinginsulinresistancethat aretogetherdiagnosticoftype2diabetes(Supplementary Fig. 2). DISCUSSION Genetics,dietandobesitycontributetothecurrentepidemicofhuman type 2 diabetes. A substantial proportion of the human population seemspredisposedtoacombinationofthesefactors,perhapsanalogous todifferentmousestrainsthatareresistantorsusceptibletodiet-and obesity-induceddiabetes3.Ourfindingsshowthattheextentofdisruptionofbetacellglycosylationandglucosetransportbydietandobesity directlycontributestodiseaseonsetandseverityinsusceptiblemice, illuminatingapathogenicpathwaythatencompasseslipotoxicityand glucotoxicity32.Thispathwayseemstobeconservedinnormalhuman isletcellsandisactivatedinisletsfromdonorswithtype2diabetes. Apathogenictippingpointinthispathwaymayoccurwhenelevated freefattyacid(FFA)concentrationsimpairtheexpressionandfunctionofFOXA2andHNFAtranscriptionfactorssufficientlyinbeta cellstodepleteGnT-4aglycosylationandglucosetransporterexpression.Theresultingdysfunctionofbetacellsleadstoimpairedglucose toleranceandfailureofGSISandfurthercontributestohyperglycemia, hepaticsteatosisandsystemicinsulinresistance.Preservationofbeta cellGnT-4aglycosylationandglucosetransporterexpressionbreaks thispathogeniccycleanditslinktodietandobesity. Thenuclearexclusionandfunctionalattenuationoftranscription factorsFoxa2andHnfainducedbyhigh-fatdietinmicewassimilarlyobservedinisletcellsfromhumandonorswithtype2diabetes
7

2011 Nature America, Inc. All rights reserved.

nature medicine advance online publication

articles
aswellasinmouseandhumanisletcellculturesaugmentedwiththe FFApalmiticacid.ThelinkageofpalmiticacidabundancetodiminishedGnT-4aactivityandglucosetransportcanfurtherexplainhow elevatedFFAconcentrationscontributetobetacelldysfunctionand theonsetofdiabetes.FFAsinducetheactivationofoneormorebeta cellGproteincoupledFFAreceptorsuchasGPR40,GPR9and GPR20(ref.33).AlthoughFFAsseemtopromotebetacellfunction insomecontexts,thechronicelevationofFFAsincreasesmitochondrialoxidationandreactiveoxygenspeciesandhasbeenobserved in beta cell cultures with diminished GSIS and reduced glucose transporter expression34,35. Our observations that the antioxidant N-acetylcysteine inhibits FFA-induced attenuation of Foxa2 and Hnfa function are consistent with those findings. Nuclearc ytoplasmic shuttling of Foxa2 has been observed in hepatocytes, where attenuation of Foxa2 contributes to the onset of steatosis and insulin resistance36. In addition, Foxa2 inactivation has been biochemicallylinkedtophosphorylationatThr56,independentof nuclearexclusion,andinvolvingtheAktkinase,whichcanbeactivatedinresponsetoinsulinorstimulithatinduceoxidativestress inbetacells37,38.WhereasbothFoxa2andHnfacontributetothe expressionofGnT-4aandtheglucosetransportersinbetacells,a completeabsenceofFoxa2engineeredbygeneticablationcanalone causefailureofGSIS39,40.OurfindingssupporttheroleofFoxa2and Hnfaaskeymetabolicregulatorsofenergyhomeostasispathways thatincludebetacellresponsestonutritionalcues. TheconnectionsestablishedamongbetacellHNFAfunctionand theexpressionofGnT-4aandtheglucosetransporterssuggestthatthe pathogenesisofdiet-andobesity-associatedtype2diabetesmayoccur byasimilarmechanismtothatoperatinginmatureonsetdiabetesof theyoungsubtype3(MODY3),themostcommonformofhuman MODY4.HnfatransactivatesthemouseSlc2a2gene42andwehave foundthatexpressionofthehumanMGAT4Ageneisalsomaintained byHNFAfunction.ThepresenceofmultipleHNFA-bindingsites inthepromoterregionsofhumanMGAT4A,SLC2A1andSLC2A2 genes is consistent with the possibility that singular or combined decreasesinhumanGnT-4aandglucosetransporterexpressionin betacellsmayberesponsibleforthelossofGSISinMODY3. Thealteredexpressionofhundredsofgeneshasbeenreportedin comparisonsofhumanisletcellsfromnormaldonorsanddonors withtype2diabetes43.Inthosestudiesandinoursreportedhere,the MGAT4Ageneissubstantiallyandsimilarlydownregulated,although thedecreaseisrelativelymodestandmightnotgarnerattentionwhen theexpressionofothergenesisalteredbytenfoldormore.Yetthis 60%reductionofMGAT4ARNAexpressionwasassociatedwitha 0- to 50-fold decrease of core -4GlcNAc N-glycan linkages p roduced by GnT-4a activity. This outcome may reflect various mechanismsthatcanregulateglycosyltransferases,includingaltered intracellular localization, competition with endogenous acceptor s ubstratesandvariableaccesstodonorsubstrates44. MultiplecelltypesexpressGnT-4awithhighexpressioninthepancreasofnormalrodentsandhumans.Nevertheless,betacellspecific transgene expression markedly diminished signs of high-fat diet induceddiabetesinthepresenceofendogenousGnT-4aexpression.In betacells,multipleglycoproteinsubstratesofGnT-4aexist,including the insulin-like growth factor- receptor and insulin receptor- s ubunit.However,theturnoverofthesesubstrateswasunaffectedby thedeficiencyofGnT-4aactivityandcore-4GlcNAcglycanlinkages25.Theglucosetransportersofbetacellsarethusuniqueamong glycoproteinsanalyzedinrequiringGnT-4aglycosylationtosustain anextendedhalf-lifeatthecellsurface,suggestingastabilizingroleof
8

lectin-ligandbindingthatmayinvolvethegalectins45,46.Thisspecificityofactionmayreflectacombinatorialroleofproteinandglycan sequencesindeterminingbiologicalfunction,analogoustoenzymatic phosphorylation,whichdictatesdifferentfunctionaloutcomesondifferentproteins44. TheimpactofenforcedbetacellspecificexpressionofMGAT4A andSLC2A2onmetabolicabnormalitiesincludingGSISandinsulin resistancewasnotableinthisstudy.Constitutivebetacellexpression ofGnT-4aorGlut-2preservedconsiderablesystemicinsulinsensitivity,indicatingthatbetacellfunctioninfluencesinsulinactiononthese peripheraltargettissues.Betacellsmayaccomplishthisbyoneor moremechanisms.TheretentionofbetacellGSISmayachieveamore effectivedeliveryofinsulin,preciselytimedtoglucoseexcursionsto meettissueandbodyneeds.Withoutsuchanaturaldeliverysystem, chronicinsulinexposurebysomecurrenttreatmentmodalitiesmay desensitizeinsulin-signalingpathwaysorprovideimproperlytimed signals.Forexample,theoptimizationofinsulinadministrationcan betternormalizebloodglucoseconcentrationandincreaseperipheral insulinsensitivityinsubjectswithtype2diabetes,andthedegreeof glycemiccontrolachievedbyinsulintreatmentinsubjectswithtype diabetesisamajordeterminantofthedegreeofinsulinsensitivity retained in peripheral tissues in vivo4749. The marked reduction inhyperglycemiawe observedfrom enforced expressionofeither GnT-4aorGlut-2indicateshoweffectivepreservationofbetacell glucosetransportcanbeinimprovingglycemia.Inturn,normalizing glucose homeostasis could also promote peripheral insulin action byattenuatingtheeffectsofglucolipotoxicity,whichimpairinsulin signaling. For instance, the severity of hyperglycemia is a factor inproteinkinaseCmediatedinhibitionofinsulinsignaling50. OurfindingsdonotaddresswhetherretentionofGSISisprotective intheonsetofdiabetes;however,lossofGSISisamarkerofbetacell dysfunctionintype2diabetes,andretentionofGSISisassociatedwith diseaseprotection.Thebetacellcouldproduceandsecretemultiple glycoproteinfactorsthatdependuponGnT-4aglycosylationfornormal activityandhalf-lifeandthat,alongwithinsulin,directlycontribute toglucosehomeostasisandperipheralinsulinaction.Nevertheless,we predictthattheseveredecreaseinglucosetransportandintracellular glucoseavailabilitycausesmetabolicalterationsinbetacellsthatmay connect our findings to further downstream events that contribute todiseaseetiology.Forexample,thereductionofglucosetransporter expressionmayreachathresholdatwhichglucosetransportinsteadof glucokinaseactivitybecomesratelimitingintheformationofglucose6-phosphateneededformaintainingvariousbetacellfunctions. ProtectionfromdiseaseconferredbyGnT-4aandbetacellglucose transportwasconsistentwiththeassignmentofbetacellGnT-4aglycosylationasalimitingfactor.BothGnT-4aandtheglucosetransporters arehighlyregulatedinpancreaticbetacells.Theirexpressionisrapidly lostinprimarybetacellcultures,andimmortalizedbetacelllineslose GSISactivity.Dietarystimulileadingtoobesitytriggerthisprocesswith theinactivationoftranscriptionfactorsthatnormallysustainbetacell GnT-4aproteinglycosylationandglucosetransport,therebypromotingtheonsetofadiseasepathwayimplicatedinthediet-andobesityassociatedcomponentoftype2diabetesmellitus.Themoleculesthat impingeuponthispathwaytosustainbetacellGnT-4aactivityand glucosetransporterexpressionmaysuggestnewtherapeutictargetsto achieveeffectivepreventionandtreatmentofdiabetes. METHODS Methods and any associated references are available in the online v ersionofthepaperathttp://www.nature.com/naturemedicine/.

2011 Nature America, Inc. All rights reserved.

advance online publication

nature medicine

articles
Note: Supplementary information is available on the Nature Medicine website. ACKnOwleDgMents ThisresearchwasfundedprimarilybyUSNationalInstitutesofHealth(NIH) grantDK48247withadditionalsupportfromGM626andCA7932(J.D.M.). FurtherfundingwasobtainedfromDK03365,DK074868,T32DK007494, DK06349andtheEuniceKennedyShriverNationalInstituteofChildHealthand HumanDevelopmentNIHthroughcooperativeagreementofU54HD02303-25 aspartofthespecializedCooperativeCentersPrograminReproductionand InfertilityResearch(J.M.O.)andtheJapanDiabetesFoundationandSuntory InstituteforBioorganicResearch(SUNBORgrant;K.O.). AUtHOR COntRIBUtIOns K.O.conductedthemajorityofexperimentsandhelpedwritethemanuscript. M.Z.C.andJ.M.O.carriedoutthehyperinsulinemic-euglycemicclampstudies andhelpedwritethemanuscript.J.D.M.conceivedofandsupervisedtheproject andwrotethemanuscript. COMPetIng FInAnCIAl InteRests Theauthorsdeclarenocompetingfinancialinterests.
Published online at http://www.nature.com/naturemedicine/. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html.
1. Olefsky, J.M. & Courtney, C.H. Type 2 diabetes mellitus: etiology, pathogenesis and natural history. in Endocrinology 5th edn (eds. DeGroot, L.J. et al.) 10931117 (W.B. Saunders, 2005). 2. Weir, G.C. & Leahy, J.L. Pathogenesis of non-insulin dependent (type II) diabetes mellitus. in Joslins Diabetes Mellitus (eds. Joslin, E.P., Kahn, C.R. & Weir, G.C.), Ch. 14, 240264 (Lea and Febiger, 1994). 3. Smyth, S. & Heron, A. Diabetes and obesity: the twin epidemics. Nat. Med. 12, 7580 (2006). 4. Korner, J., Woods, S.C. & Woodworth, K.A. Regulation of energy homeostasis and heath consequences in obesity. Am. J. Med. 122, S12S18 (2009). 5. Parillo, M. & Ricardi, G. Diet composition and the risk of type 2 diabetes: epidemiological and clinical evidence. Br. J. Nutr. 92, 719 (2004). 6. Surwit, R.S., Kuhn, C.M., Cochrane, C., McCubbin, J.A. & Feinglos, M.N. Dietinduced type II diabetes in C57BL/6J mice. Diabetes 37, 11631167 (1988). 7. Kahn, S.E. The relative contribution of insulin resistance and beta-cell dysfunction to the pathophysiology of type 2 diabetes. Diabetologia 46, 319 (2003). 8. Winzell, M.S. & Ahren, B. The high-fat diet-fed mouse: a model for studying mechanisms and treatment of impaired glucose tolerance and type 2 diabetes. Diabetes 53 (suppl. 3) S215S219 (2004). 9. Leroith, D. & Accili, D. Mechanisms of disease: using genetically altered mice to study concepts of type 2 diabetes. Nat. Clin. Pract. Endocrinol. Metab. 4, 164172 (2008). 10. Saltiel, A.R. & Kahn, C.R. Insulin signaling and the regulation of glucose and lipid metabolism. Nature 414, 799806 (2001). 11. Schenk, S., Saberi, M. & Olefsky, J.M. Insulin sensitivity: modulation by nutrients and inflammation. J. Clin. Invest. 118, 29923002 (2008). 12. Reaven, G.M. The insulin resistance syndrome: definition and dietary approaches to treatment. Annu. Rev. Nutr. 25, 391406 (2005). 13. Johnson, J.H. et al. Underexpression of beta cell high Km glucose transporters in noninsulin-dependent diabetes. Science 250, 546549 (1990). 14. Orci, L. et al. Evidence that down-regulation of -cell glucose transporters in noninsulin-dependent diabetes may be the cause of diabetic hyperglycemia. Proc. Natl. Acad. Sci. USA 87, 99539957 (1990). 15. Orci, L. et al. Reduced -cell glucose transporter in new onset diabetic rats. J. Clin. Invest. 86, 16151622 (1990). 16. Thorens, B., Weir, G., Leahy, J.L., Lodish, H.F. & Bonner-Weir, S. Reduced expression of the liver/beta-cell glucose transporter isoform in glucose-insensitive pancreatic beta cells of diabetic rats. Proc. Natl. Acad. Sci. USA 87, 64926496 (1990). 17. Unger, R.H. Diabetic hyperglycemia: link to impaired glucose transport in pancreatic beta cells. Science 251, 12001205 (1991). 18. Valera, A. et al. Expression of GLUT2 antisense RNA in cells of transgenic mice leads to diabetes. J. Biol. Chem. 269, 2854328546 (1994). 19. Gremlich, S., Roduit, R. & Thorens, B. Dexamethasone induces posttranslational degradation of GLUT2 and inhibition of insulin secretion in isolated pancreatic cells. J. Biol. Chem. 272, 32163222 (1997). 20. Guillam, M.T. et al. Early diabetes and abnormal postnatal pancreatic islet development in mice lacking Glut-2. Nat. Genet. 17, 327330 (1997). 21. Guillam, M.-T., Dupraz, P. & Thorens, B. Glucose uptake, utilization, and signaling in GLUT2-null islets. Diabetes 49, 14851491 (2000). 22. Reimer, M.K. & Ahrn, B. Altered -cell distribution of pdx-1 and GLUT2 after a short-term challenge with a high-fat diet in C57BL/6J mice. Diabetes 51, S138S143 (2002). 23. Del Guerra, S. et al. Functional and molecular defects of pancreatic islets in human type 2 diabetes. Diabetes 54, 727735 (2005). 24. Kahn, S.E., Hull, R.L. & Utzschneider, K.M. Mechanisms linking obesity to insulin resistance and type 2 diabetes. Nature 444, 840846 (2006). 25. Ohtsubo, K. et al. Dietary and genetic control of glucose transporter 2 glycosylation promotes insulin secretion in suppressing diabetes. Cell 123, 13071321 (2005). 26. Minowa, M.T. et al. cDNA cloning and expression of bovine UDP-Nacetylglucosamine:1,3-D-mannoside 1,4-N-acetylglucosaminyltransferase IV. J. Biol. Chem. 273, 1155611562 (1998). 27. De Vos, A. et al. Human and rat beta cells differ in glucose transporter but not glucokinase gene expression. J. Clin. Invest. 96, 24892495 (1995). 28. Thorens, B., Guillam, M.T., Beermann, F., Burcelin, R. & Jaqueet, M. Transgenic overexpression of GLUT1 or GLUT2 in pancreatic beta cells rescues GLUT2-null mice from early death and restores normal glucose-stimulated insulin secretion. J. Biol. Chem. 275, 2375123758 (2000). 29. Alessi, D.R. & Cohen, P. Mechanism of activation and function of protein kinase B. Curr. Opin. Genet. Dev. 8, 5562 (1998). 30. White, M.F. IRS proteins and the common path to diabetes. Am. J. Physiol. Endocrinol. Metab. 283, E413E422 (2002). 31. Rossmeisl, M., Rim, J.S., Koza, R.A. & Kozak, L.P. Variation in type 2 diabetesrelated traits in mouse strains susceptible to diet-induced obesity. Diabetes 52, 19581966 (2003). 32. Poitout, V. & Robertson, R.P. Glucolipotoxicity: fuel excess and beta-cell dysfunction. Endocr. Rev. 29, 351366 (2008). 33. Kebede, M.A., Alquier, T., Latour, M.G. & Poitout, V. Lipid receptors and islet function: therapeutic implications? Diabetes Obes. Metab. 11 (suppl. 4) 1020 (2009). 34. Unger, R.H. Lipotoxicity in the pathogenesis of obesity-dependent NIDDM. Genetic and clinical implications. Diabetes 44, 863870 (1995). 35. Gremlich, S., Bonny, C., Waeber, G. & Thorens, B. Fatty acids decrease IDX-1 expression in rat pancreatic islets and reduce GLUT2, glucokinase, insulin, and somatostatin levels. J. Biol. Chem. 272, 3026130269 (1997b). 36. Wolfrum, C., Asilmaz, E., Luca, E., Friedman, J.M. & Stoffel, M. Foxa2 regulates lipid metabolism and ketogenesis in the liver during fasting and in diabetes. Nature 432, 10271032 (2004). 37. Howell, J.J. & Stoffel, M. Nuclear export-independent inhibition of Foxa2 by insulin. J. Biol. Chem. 284, 2481624824 (2009). 38. Chen, Y.W. et al. The role of phosphoinositide 3-kinase/Akt signaling in low-dose mercury-induced mouse pancreatic beta-cell dysfunction in vitro and in vivo. Diabetes 55, 16141624 (2006). 39. Lantz, K.A. et al. Foxa2 regulates multiple pathways of insulin secretion. J. Clin. Invest. 114, 512520 (2004). 40. Gao, N. et al. Foxa1 and Foxa2 maintain the metabolic and secretory features of the mature beta-cell. Mol. Endocrinol. 24, 15941604 (2010). 41. Mitchell, S.M. & Frayling, T.M. The role of transcription factors in maturity-onset diabetes of the young. Mol. Genet. Metab. 77, 3543 (2002). 42. Fukui, K. et al. The HNF-1 target collection controls insulin exocytosis by SNARE complex formation. Cell Metab. 2, 373384 (2005). 43. Gunton, J.E. et al. Loss of ARNT/HIF1beta mediates altered gene expression and pancreatic-islet dysfunction in human type 2 diabetes. Cell 122, 337349 (2005). 44. Ohtsubo, K. & Marth, J.D. Glycosylation in cellular mechanisms of health and disease. Cell 126, 855867 (2006). 45. Rapoport, E.M., Kurmyshkina, O.V. & Bovin, N.V. Mammalian galectins: structure, carbohydrate specificity, and functions. Biochemistry (Mosc.) 73, 393405 (2008). 46. Rabinovich, G.A., Toscano, M.A., Jackson, S.S. & Vasta, G.R. Functions of cell surface galectin-glycoprotein lattices. Curr. Opin. Struct. Biol. 17, 513520 (2007). 47. Scarlett, J.A., Gray, R.S., Griffin, J., Olefsky, J.M. & Kolterman, O.G. Insulin treatment reverses the insulin resistance of type II diabetes mellitus. Diabetes Care 5, 353363 (1982). 48. Revers, R.R., Kolterman, O.G., Scarlett, J.A., Gray, R.S. & Olefsky, J.M. Lack of in vivo insulin resistance in controlled insulin-dependent, type I, diabetic patients. J. Clin. Endocrinol. Metab. 58, 353358 (1984). 49. Garvey, W.T., Olefsky, J.M., Griffin, J., Hamman, R.F. & Kolterman, O.G. The effect of insulin treatment on insulin secretion and insulin action in type II diabetes mellitus. Diabetes 34, 222234 (1985). 50. Pillay, T.S., Xiao, S. & Olefsky, J.M. Glucose-induced phosphorylation of the insulin receptor. Functional effects and characterization of phosphorylation sites. J. Clin. Invest. 97, 613620 (1996).

2011 Nature America, Inc. All rights reserved.

nature medicine advance online publication

ONLINE METHODS

2011 Nature America, Inc. All rights reserved.

Vector construction and transgenic mouse production.Theratinsulin2promoter(RIP2)wasclonedfromratgenomicDNA(Novagen)usingPCRwith primersRIP-FKpn(5-CGGGGTACCGGGCAGTACCAAATCAGGA-3)and RIP-RXho(5-TGGCTCGAGCTGAATCCCCACTAGCTTTA-3),andthenisolatedasa700-base-pair(bp)KpnI-XhoIDNAfragment.Thebovinegrowthhormonepolyadenylationsignalelement(pA)wasclonedbyPCRfrompcDNA3. (Invitrogen)withprimersBGH-FBam(5-CCGGGATCCAGCCTCGACTGT GCCTTCTA-3)andBGH-RXba(5-TCATCTAGAATAGAGCCCACCGCA TCCC-3),andthenisolatedasa220-bpBamHI-XbaIDNAfragment.These fragmentsweresequentiallyinsertedintothepBluescriptIISK+(Stratagene) cloningvector(pRIP2-BGH).ThehumanMGAT4AcDNAwasclonedbyrtPCR from human pancreas total RNA (Ambion) with primers hMGAT4ASal-F (5-TGAGTCGACTTGGCCTAGAGCCAGGAGTA-3)andhMGAT4AHindR(5-ATTAAGCTTCCAAGTGGAAATGACAATCG-3).ThehumanGLUT2 (SLC2A2)cDNAwasclonedbyrtPCRfromtotalRNAofHepG2cells(ATCC) withprimershGLUT2Sal-F(5-TCAGTCGACTGTGCCACACTCACACAA GA-3) and hGLUT2Hind-R (5-ATTAAGCTTCAGACGGTTCCCTTATT GTTT-3).ThecDNAfragments(.9kilobases(kb)and.7kb,respectively) weredigestedwithSalIandHindIII,andthenclonedintothepRIP2-BGH. The resulting vectors pRIP2-MGAT4A-BGH and pRIP2-GLUT2-BGH were sequencedforverificationanddigestedwithdigestedwitheitherKpnIandNotI orKpnIandXbaI,respectively,toisolatetransgenecistrons(~2.9kband~2.7kb, respectively)formicroinjection.Transgenicmicewereproducedusingzygotesof C57BL/6Jasdescribed5,andmousegenotypesweredeterminedfromtissueor bloodbyPCRusingtransgene-specificprimersBGH-R(5-CAGACAATGCGA TGCAATTTCCTCA-3),MGAT4A-F(5-GTTGCAGAAGGAATGGTGGATC CAA-3)andGLUT2-F(5-TCCAGTACATTGCGGACTTCTGTGG-3). Mouse care and maintenance.MicewerehousedinspecificpathogenfreeconditionsinaccordancewiththeSanford-BurnhamMedicalResearchInstituteandUC SantaBarbaraInstitutionalAnimalCareandUseCommitteecertifications.Mice wereprovidedeitherastandarddiet(6.4%protein,73.%carbohydratesand0.5% fatwith4.07kcalg;D2329,ResearchDiets)orahigh-fatdiet(6.4%protein, 25.5%carbohydratesand58.0%fatwith5.56kcalg;D233,ResearchDiets). Blood chemistry. Blood collection and the blood chemistry analyses were carriedoutasdescribed25. Blood glucose and insulin assays.Micewerefasted4hbeforebloodglucoseand insulinconcentrationsweredeterminedduringhigh-fatdietadministration.For theglucosetolerancetest,micewerefastedfor6hfollowedbyintraperitoneal glucoseinjection(.5gperkgbodyweight).Serumsampleswereobtainedat0, 30,60and20minaftertheinjectionandmeasurementsofglucoseandinsulin wasdetermined,thelatterusingmouseinsulinELISAassaywithmouseinsulin standard(CrystalChem).Intestinginsulintolerance,intraperitonealinjections of2Uperkgbodyweightofhumaninsulin(Carbiochem)werecarriedout.In argininetolerancetests,micewerefastedfor6hfollowedbyintraperitoneal injection(3gperkgbodyweight)ofL-arginine(Sigma). Hyperinsulinemic-euglycemic clamps.Mice02weeksoldwereplacedon thehigh-fatdiet.After0weeksonthehigh-fatdiet,micewereanesthetized withketamine(00mgperkg),acepromazine(0.5mgperkg)andxylazine (6mgperkg)viaintramuscularinjectionandthenunderwentcatheterization surgery.Twomicrourethanecatheters(Micro-Renthane)wereinsertedintothe rightjugularveinandsecuredtothetopoftheneckofthemouse.Micewere placedinindividualcagesandallowedtorecover57daftersurgerybeforethe hyperinsulinemic-euglycemicclampexperiment.Onthedayoftheclamp,mice werefastedfor4handsubsequentlyplacedinarestrainer(BraintreeScientific) foranadditional2h.Afteratotalof6hoffasting,bloodglucosewasassessed viatailnick.Asolutionof[3H]glucose-saline,4.6Ci 3Hml(NENLife ScienceProducts),wasinfusedfor90minat2lmin.Aftertheequilibration period,tailbloodwascollectedandusedtoassessbasaltracerconcentration. Afterbloodcollection,insulin(8.0mUkgmininsulinincomplexwith0% wt/volBSAand4.6Ciml[3H]glucose)wasinfusedatarateof2lmin. Adextrosesolution(454mgml)wasinfusedintothesecondcatheterata variableratetomaintaineuglycemia.After~20minofinsulininfusionand

maintenanceofbloodglucoseat~20mgdlforatleast30min,threesteadystatebloodsampleswerecollectedandusedformeasurementof[3H]glucose specificactivityanddeterminationofinsulinandFFAlevels. Islet cell preparation and culture.Mouseisletcellswereisolatedandprepared asdescribed25.Humanisletsfromdeceasedanonymousdonorswereobtained fromtheIntegratedIsletDistributionProgram,theJuvenileDiabetesResearch FoundationandPRODOLaboratories.Medicalhistoriesdidnotindicatetreatment withantidiabeticdrugs.Isletsacceptedforstudyweredeterminedbythesourceto be>90%viableandpure,andwerereceiveddafterpostmortemisolation.Islet puritywasfurthercheckeduponarrivalbydithizonestain,andinsulinantibody bindingindicatedthatallnormalandtype2diabeteshumanisletswerecomposed of6572%betacells.Forcellculture,isletsweredissociatedintosinglecellsand culturedwithRPMI640mediumsupplementedwith0%vol/volFCS,2mM L-glutamine,and00gmlpenicillin-streptomycin.ThepalmiticacidBSA complexwaspreparedasdescribed52.Isletcellswerepreculturedinbasalglucose culturemedium(RPMI640mediumsupplementedwith5mMglucoseand0% vol/volFCS)for24h.Thosereceiving0mMN-acetylcysteinewerepretreatedin culturesfor3hbeforepalmiticacidadditionandmediumreplacementwithbasal glucoseculturemediumcontaining%wt/volBSAor0.5mMpalmiticacidand %wt/volBSAwithorwithout0mMN-acetylcysteinefor48h.Themolarratio ofpalmiticacidtoBSAwas:3andcellviabilitywasunalteredinthesestudiesas determinedusingannexinV(BD).SamplesweresubjectedtoRNAexpression orhistologicalanalyses.TotalRNAwaspreparedandsubjectedtortPCR.Gene expressionwasnormalizedtomitochondrialribosomalproteinL4.Forhistological analyses,isletswerefixedwith4%vol/volparaformaldehydeandPBS,andembeddedinOCTcompound(Tissue-Tek)beforesectioning. Glucose stimulated insulin secretion assay.IntheperifusionGSISassay,205 isletcellswerepreincubatedinHEPES-bufferedKrebsRingersolutioncontaining 2.8mMglucosefor30minat37C,thenplacedina0.2msyringefilter(Sartorius). Thefilterwasconnectedtoaperistalticpumpandtheflowratewasadjustedto mlminfollowedbyequilibrationinKRBHcontaining2.8mMglucosefor 0minbeforetheglucoseconcentrationwasincreasedto6.8mM.Inthestatic GSISassay,isletcellswerepreincubatedwithKRBHcontaining2.8mMglucose at37Cfor30minandthenincubatedwitheither2.8mMor6.8mMglucosein KRBHat37Cfor5min.Mediumcollectedwasanalyzedbyanti-insulinELISA. Glycoprotein, glycan and transcriptional factor analyses.Immunoprecipitation wascarriedoutusingC-terminalepitoperecognizingantibodiestoGlut-(C-20, SantaCruzBiotechnology)andGlut-2(C-9,SantaCruzBiotechnology).ThedetectionofGlut-andGlut-2immunoprecipitateswascarriedoutusingN-terminal epitoperecognizingantibodiestoGlut-(N-20,SantaCruzBiotechnology)and Glut-2(AB342,Chemicon).GlycansonprecipitatedGlutglycoproteinswere a nalyzedbyincubationwithlectinsDSA,L-PHA,TL,MAL-I,SNA,RCA-IorECA asdescribed25.TranscriptionfactoranalysesincludedtheuseofantibodiestoFoxa2 (07-633,UpstateBiotechnology;M20,SantaCruzBiotechnology),PDX-(ab3243, Chemicon),HNF-4(H-7,SantaCruzBiotechnology),ARNT(29/HIF-b,BD TransductionLaboratories)andHNFA(ab96777,Abcam).Immunodetectionand chemiluminescentsignalquantificationwasdoneusinghorseradishperoxidase conjugatedsecondaryantibodies(AmershamPharmaciaBiotech)andtheEC3 BioImagingsystemwithLabWorks4.0software(UVPBioImaging). Statistical analysis.Datawereplottedasthemeans.e.m.ofthenumberof samplesanalyzedunlessotherwiseindicated.ANOVAandStudentsttestswere carriedoutusingPrism(GraphPadSoftware)withsignificancesetatP<0.05. PosthocanalysesinvolvingmultiplecomparisonswithBonferronicorrection werecarriedoutaftersignificancewasestablished. Additional methods.DetailedmethodologyisdescribedinSupplementary MethodsandinSupplementary Figures 3 and 4.
51. Shafi, R. et al. The O-GlcNAc transferase gene resides on the X chromosome and is essential for embryonic stem cell viability and mouse ontogeny. Proc. Natl. Acad. Sci. USA 97, 57355739 (2000). 52. Cousin, S.P. et al. Free fatty acidinduced inhibition of glucose and insulin-like growth factor Iinduced deoxyribonucleic acid synthesis in the pancreatic -cell line INS-1. Endocrinology 142, 229240 (2001).

nature medicine

doi:10.1038/nm.2414