King’s Research Portal Document Version Peer reviewed version Link to publication record in King's Research Portal Citation for published version (APA): Beaumont, M., Goodrich, J., Jackson, M. A., Yet, I., Davenport, E. R., Vieira-Silva, S., ... Bell, J. T. (2016). Heritable components of the human fecal microbiome are associated with visceral fat. GENOME BIOLOGY. Citing this paper Please note that where the full-text provided on King's Research Portal is the Author Accepted Manuscript or Post-Print version this may differ from the final Published version. If citing, it is advised that you check and use the publisher's definitive version for pagination, volume/issue, and date of publication details. And where the final published version is provided on the Research Portal, if citing you are again advised to check the publisher's website for any subsequent corrections. General rights Copyright and moral rights for the publications made accessible in the Research Portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognize and abide by the legal requirements associated with these rights. •Users may download and print one copy of any publication from the Research Portal for the purpose of private study or research. •You may not further distribute the material or use it for any profit-making activity or commercial gain •You may freely distribute the URL identifying the publication in the Research Portal Take down policy If you believe that this document breaches copyright please contact librarypure@kcl.ac.uk providing details, and we will remove access to the work immediately and investigate your claim. Download date: 01. Oct. 2016 1 Heritablecomponentsofthehumanfecalmicrobiomeareassociatedwithvisceralfat 2 MichelleBeaumont1,JuliaK.Goodrich2,3,MatthewA.Jackson1,IdilYet1,EmilyR.Davenport3,Sara 3 Vieira-Silva,9,10,JustineDebelius4,7,TessPallister1,MassimoMangino1,JeroenRaes9,10,RobKnight4,5,6,7, 4 AndrewG.Clark2,RuthE.Ley2,38,TimD.Spector1*,JordanaT.Bell1* 5 1 Dept of Twin Research & Genetic Epidemiology, King’s College London, U.K. 6 2 Dept of Microbiology, Cornell University, Ithaca, NY 14853 US 7 3 Dept of Molecular Biology & Genetics, Cornell University, Ithaca, NY 14853, USA 8 4 Dept of Chemistry and Biochemistry, University of Colorado, Boulder, CO 80309, USA 9 5 Biofrontiers Institute, University of Colorado, Boulder, CO 80309, USA 10 6 Howard Hughes Medical Institute, Boulder, CO 80309, USA 11 7 Current address: Departments of Pediatrics and Computer Science & Engineering, University of California San 12 Diego, La Jolla CA 92093, USA 13 8 14 9 15 10 16 17 *Equalcontribution 18 Corresponding: 19 JordanaT.Bell 20 DepartmentofTwinResearch&GeneticEpidemiology 21 King’sCollegeLondon 22 StThomas’Hospital 23 3rdFloor,SouthWing,BlockD Dept of Microbiome Science, Max Planck Institute for Developmental Biology, Tübingen, Germany KULeuven–UniversityofLeuven,DepartmentofMicrobiologyandImmunology,Leuven,Belgium. VIBlabforBioinformaticsand(eco-)systemsbiology,Belgium. 1 24 London 25 UnitedKingdom 26 SE17EH 27 Tel:(+44)02071881506 28 Email:jordana.bell@kcl.ac.uk 29 TimD.Spector 30 DepartmentofTwinResearch&GeneticEpidemiology 31 King’sCollegeLondon 32 StThomas’Hospital 33 3rdFloor,SouthWing,BlockD 34 London 35 UnitedKingdom 36 SE17EH 37 Tel:(+44)02071886739 38 Email:tim.spector@kcl.ac.uk 39 40 41 42 43 44 2 45 Abstract 46 Variationinthehumanfecalmicrobiotahaspreviouslybeenassociatedwithbodymassindex(BMI). 47 Althoughobesityisaglobalhealthburden,theaccumulationofabdominalvisceralfatisthespecific 48 cardio-metabolicdiseaseriskfactor.Here,weexplorelinksbetweenthefecalmicrobiotaandabdominal 49 adiposityusingbodycompositionasmeasuredbyDualX-RayAbsorptiometry(DXA)inalargesampleof 50 twinsfromtheTwinsUKcohort,comparingfecal16SrRNAdiversityprofileswithsixadipositymeasures. 51 Results 52 Weprofilesixadipositymeasuresin3,666twinsandestimatetheirheritability,findingnovelevidence 53 forstronggeneticeffectsunderlyingvisceralfatandandroid:gynoidratio.Weconfirmtheassociationof 54 lowerdiversityofthefecalmicrobiomewithobesityandadipositymeasures,andthencomparethe 55 associationbetweenfecalmicrobialcompositionandtheadiposityphenotypesinadiscoverysub- 56 sampleoftwins.Weidentifyassociationsbetweentherelativeabundancesoffecalmicrobial 57 operationaltaxonomicunits(OTUs)andabdominaladipositymeasures.Mostoftheseresultsinvolve 58 visceralfatassociations,withthestrongestassociationsbetweenvisceralfatandOscillospiramembers. 59 UsingBMIasasurrogatephenotype,wepursuereplicationinindependentsamplesfromthree 60 population-basedcohortsincludingAmericanGut,FlemishGutFloraProject,andtheextendedTwinsUK 61 cohort.Meta-analysisacrossthereplicationsamplesindicatethat8OTUsreplicateatastringent 62 thresholdacrossallcohorts,while49OTUsachievenominalsignificanceinatleastonereplication 63 sample.Heritabilityanalysisoftheadiposity-associatedmicrobialOTUspromptedustoassesshost 64 genetic-microbeinteractionsatobesity-associatedhumancandidateloci.Weobservesignificant 65 associationsofadiposity-OTUabundanceswithhostgeneticvariantsintheFHIT,TDRG1andELAVL4 3 66 genes,suggestingapotentialroleforhostgenestomediatethelinkbetweenthefecalmicrobiomeand 67 obesity. 68 Conclusion 69 Ourresultsprovidenovelinsightsintotheroleofthefecalmicrobiotaincardio-metabolicdiseasewith 70 clearpotentialforpreventionandnoveltherapies. 71 Keywords 72 Fecalmicrobiome;Obesity;VisceralFat;Heritability;Geneticassociation;Twins 4 73 Background 74 Obesityhasrapidlybecomeaglobalpublichealthproblemwithobesity-relateddiseasenowoneofthe 75 leading causes of preventable death worldwide[1]. Although overall obesity poses a global health 76 epidemic,itistheaccumulationofexcessabdominalfatthatisacriticalriskfactorforcardiovascular 77 andmetabolicdisease[2].Changesindietandsedentarylifestylecanpartlyexplaintheriseinobesity, 78 and family and twin studies also show a genetic influence, with obesity heritability estimates of 0.60- 79 0.70[3-6].Genome-wideassociationstudies(GWAS)haveidentifiedgeneticriskfactors[7-9],butgenetic 80 variantsdetectedtodateexplainlessthan3%oftheheritabilityofobesity,withpredictionabilityofup 81 to20%,suggestingaroleforothermechanisms[10]. 82 Recent insights show that the gut microbiota may play a crucial role in obesity and cardio-metabolic 83 disease risk. Many studies have linked different aspects of the fecal microbiome to obesity[11-17]. 84 However, in most cases the causal mechanisms leading to these associations are unclear, although a 85 numberoftheorieshavebeensuggested,includingalterationsinenergyharvestfromfood[18]andan 86 increaseinpotentialinflammatorymicrobes[19].Therearealsoinconsistenciesinthetaxaassociated 87 withobesity[16]thatmaybeexplainedinpartbystudydesigndifferencessuchascontrolofdietand 88 sequencingplatforms,butcouldalsobeduetodifferencesincollectivebacterialgenefunctionrather 89 thanthespeciescommunitycomposition[20].Anothercomplicatingfactorthatvariesamongstudiesis 90 thequantificationofobesity.WhilemosthumanstudiesconsiderBMIasthemeasureofobesity[11,21, 91 22],mousestudiestypicallyuseepididymalfatweight[23]orDual-EnergyX-rayAbsorptiometry(DXA)- 92 derived measures of total body fat[18, 24]. BMI is an imprecise measure of adiposity and measures 93 overallmasswithoutdistinctionbetweenleanandfatmass[25].Estimatesofvisceralfat,however,have 94 stronger associations with obesity-related cardio-metabolic diseases, such as type-2 diabetes and 5 95 cardiovasculardisease[26-28],buthavetypicallybeendifficulttomeasureinhumansandhaveyettobe 96 linkedwithvariationinthehumanfecalmicrobiome. 97 Previousstudieshaveattemptedtotackleheritabilityofattributesofthemicrobiome.Zoetendaletal. 98 [29] found that monozygotic twins had more similar microbiomes than marital partners or unrelated 99 individuals,suggestingeitheraroleforhostgenotypeingutmicrobiomecolonisationormother-to-child 100 transmissionofmicrobes.Inaddition,astudyofMethanobrevibactercarriageconcordancerateintwins 101 showedhigherconcordanceinmonozygotic(MZ)twins[30].ArecentstudybyGoodrichetal.[31]was 102 the first large-scale analysis to report heritability of the human fecal microbiota, with the relative 103 abundanceof16SrRNAgenesequencesbelongingtothefamilyChristensenellaceaeshowingthemost 104 varianceattributedtohostgeneticeffects.Christensenellaceaewasalsoenrichedinabundanceinthe 105 microbiomes of low-BMI individuals. Here, we build upon these findings to explore the association 106 between the human fecal microbiome and abdominal adiposity, as the main risk factor for cardio- 107 metabolicdiseaserisk.WeobtainedDXA-basedmeasuresofabdominaladiposity,specifically,visceral 108 fatmass,subcutaneousfatmass,andpreviouslyreportedtrunkfatmeasures[32],aswellasbodyfat 109 distribution,inalargerdatasetoftwins,includingasubsetofthetwinsampleprofiledbyGoodrichetal. 110 (2016)[33] and a subset of the twin sample profiled by Jackson et al. [34]. We show that heritable 111 components of the human fecal microbiome [31, 33] are significantly associated with visceral fat, 112 confirming the key role of the microbiome in cardio-metabolic disease risk. We further identify a link 113 betweenfecalmicrobiomeprofiles,visceralfatandsubcutaneousfatwithgeneticvariantsincandidate 114 obesitycandidategenes,providingpotentialinsightsintomechanismstorelatethefecalmicrobiometo 115 cardio-metabolicdiseaserisk. 116 117 6 118 Results 119 Measuresofadipositywereobtainedfromanunselectedsampleof3,666predominantlyfemaletwins 120 fromtheTwinsUKcohort(TUK-D),whichincluded1,044monozygotic(MZ)and789dizygotic(DZ)twin 121 pairs(averageage63years(range32-87);96.4%female).Fecalmicrobiomeprofileswereavailablefor 122 1,313oftheseindividuals(496MZ,594DZand223unrelatedindividuals;averageage63years(range 123 32-87); 96.4% female). Demographics for these samples can be found in Table 1 and Supplementary 124 Table7.Fecalmicrobiomeprofilesincluded601previouslypublishedprofilesfromGoodrichetal.[35], 125 anadditional671profilesrecentlyreportedwithinJacksonetal.[34]and41additionaltwinprofiles[33]. 126 All fecal samples underwent 16S rRNA profiling (V4 region) gene sequencing on the Illumina MiSeq 127 platform,providing2,135operationaltaxonomicunits(OTUs)at97%sequenceidentity. 128 AdiposityandVisceralfatheritability 129 Westudiedsixadipositymeasuresintotal,andtheseincludedthreemeasuresofabdominaladiposity 130 (visceralfatmass(VFM),subcutaneousfatmass(SFM),percentagetrunkfat(pTF)),twomeasuresof 131 bodyfatdistribution(android:gynoidratio(AGR)andwaist-hipratio(WHR)),andonemeasureofoverall 132 obesity,BMI.AdipositywasestimatedusingDXA-derivedmeasures,whichhavebeenshowntobe 133 reliablealternatives[36-39]totraditionalCTandMRIscan-basedmeasuresofadiposity.Themajorityof 134 theseadipositymeasureshavebeenpreviouslyexploredintheTwinsUKcohort,howeverVFMandAGR 135 arenewlyobtainedphenotypes.ThenewmeasureofVFMwashighlycorrelatedwithotherabdominal 136 andoveralladipositymeasures,includingBMI(Figure1A).Twin-basedheritabilityanalysisofVFM 137 showedevidenceofasignificantadditivegeneticcomponent,orheritability(h2),contributingto0.70 138 (95%CI=0.58–0.74)ofthetotalvarianceinVFM.TheVFMheritabilityestimateremainedhighafter 139 adjustmentforBMI(0.64,seeAdditionalFile5).Weobtainedcomparableestimatesfortheheritability 140 ofSFM(h2=0.72(95%CI=0.60–0.77)),pTF(h2=0.66(95%CI=0.55–0.77),AGR(h2=0.65(95%CI=0.55 7 141 –0.76)),BMI(h2=0.75(95%CI=0.68–0.80)andaslightlylowerestimateforWHR(h2=0.32(95%CI= 142 0.24–0.40)),inlinewithpreviousstudies[40-42](Figure1B). 143 Thetwinfecalmicrobiomeanditsheritability 144 Thehumanfecalmicrobiomeinthedatasetof1,313twinscomprisedFirmicutesasthemostdominant 145 phylum (51%), followed by Bacteroidetes (39%) and Proteobacteria (4%). These estimates are 146 comparable to previously published results [35] and reflect a typical western fecal microbiome. Using 147 twins, we then explored evidence for heritability in the gut microbial profiles (relative abundances of 148 OTUs),extendingtheresultsofGoodrichetal.[35]inthelargersampleof1,313twinsusingthesame 149 methods for OTU heritability. Altogether, OTU heritability in this dataset ranged between 0 and 0.42, 150 andtheaverageestimateoverallOTUswas0.07.ThemostheritablemicrobewasanOTUclassifiedas 151 Clostridum perfringens (h2=0.42 (95%CI 0.23-0.51)) (Additional File 4: Supplementary Table 1). The 152 family, Christensenellaceae was the most heritable family reported in Goodrich et al.[35], and while 153 OTUsrepresentingChristensenellawerenotthetoprankedinthislargerdataset,ChristensenellaOTU 154 heritabilityremainedhighwithoneOTU,GreengenesOTU176318,showingaheritabilityof0.31(95%CI 155 0.21-0.41). The microbial heritability estimates presented here are overall consistent with the original 156 microbialtwin-basedheritabilityfindingsfromGoodrichetal.[31],andwithrecentextendedheritability 157 estimatesfromtheextendedTwinsUKcohort[33].Forexample,atthegenuslevelheritabilityestimates 158 acrossstudiesshowacorrelationof0.67betweenthisTUK-DdatasetandGoodrichetal.[31](Additional 159 File 1: Figure A) and 0.77 between TUK-D and Goodrich et al.[33](Additional File 1: Figure B). 160 Approximately 6% of the dataset (122 OTUs) had evidence for at least moderate heritability (h2>0.2), 161 andthesewerepresentinatleast25%ofindividuals(AdditionalFile4:SupplementaryTable1). 162 Fecalmicrobiomediversityisstronglylinkedtoobesityandcentraladiposity 8 163 Microbialdiversity(alpha-diversity)inobeseindividualshasbeenreportedtobelowerthanthatoflean 164 individuals[15,31,43].HerewecomparedestimatesofShannonDiversityforthesubjects’fecal 165 microbiomeswithalladipositymeasuresusingalinearmixedeffectsmodel,adjustingfordiet(see 166 Methods),age,sexandfamilyrelatedness.Asinpreviousreports[15,31,43,44],weobserveda 167 significantnegativeassociationbetweenShannondiversityandalladiposityphenotypes(Figure2, 168 AdditionalFile2).VFMshowedthemostsignificantassociationwithalphadiversity(beta=-0.14,se= 169 0.27,P=4.13x10-7)andWHRshowedtheleastsignificantassociation(beta=-0.05,se=0.031,P=0.097). 170 AllmeasuresbutWHRweresignificantlyassociatedwithdiversityandtherefore,alphadiversity 171 measuresinoursamplearenotonlynegativelyassociatedwithobesity,butarealsosignificantlylower 172 inindividualswithgreaterabdominaladiposityandvisceralfat. 173 Fecalmicrobiomeprofilesassociatewithcentraladiposityacrosstwins 174 WeinvestigatedtheassociationofeachOTUwithalladipositytraits,includingBMI,acrossindividuals. 175 Oftheapproximately12,000OTU-phenotypeassociationsconsidered,3,217werenominallysignificant 176 and 149 OTU results surpassed Bonferroni correction (P = 3.90x10-6). The 149 significant microbial- 177 adiposityassociationsinvolved97uniqueOTUs(AdditionalFile4:SupplementaryTable2),andthesefell 178 within either the Firmicutes or Bacteroidetes phyla, and most within the Ruminococcaceae family. 179 Visceral fat (VFM) associations made up the highest proportion of significant results surpassing the 180 Bonferronithreshold(45%,Figure3A).ThepeakresultwasanOTUclassifiedasOscillospira(Greengenes 181 OTU 372146), which was associated with VFM (P = 1.93x10-12). Ruminococcaceae OTUs featured 182 prominently in the top significant results, along with a number of other OTUs within the 183 Lachnospiraceae family. Given the importance of VFM and AGR in cardiovascular risk, we were 184 interested to determine potential microbial markers of cardiovascular risk. OTUs within Oscillospira, 185 Lachnospira and Ruminococcus all showed negative associations with VFM and AGR, suggesting a 9 186 potential protective role for these bacteria in cardiovascular risk (Figure 3, Additional File 4: 187 SupplementaryTable2).BlautiaOTUsshowedapositiveassociationwithVFMandAGRandmaybea 188 microbialmarkercandidateforcardiovascularrisk(Figure3A,AdditionalFile4:SupplementaryTable2). 189 Theseresultssupportthecrucialroleofthemicrobiometowardsvisceralfatasamarkerofadiposity 190 andcardio-metabolicdiseaserisk. 191 Due to the large number of OTUs in the top-ranked association results belonging to the same genera 192 and families, we also explored the peak results with respect to collapsed taxonomies, whereby we 193 combinedsequencesfromOTUswiththesametaxonomy.Wefirsttestedtheassociationbetweenthe8 194 genera, to which the 97 adiposity-significant OTUs were assigned, and adiposity. Altogether, 11 195 associationspassBonferronicorrection(P=0.001)and27genus-adiposityassociationswerenominally 196 significant. These comprised 6 of the genera and included Blautia, Oscillospira, Lachnospira, 197 Ruminococcus (within both Lachnospiraceae and Ruminococcaceae) and Clostridium. The two most 198 significantassociationswereobtainedbetweenOscillospiraandVFM(P=3.29x10-07,AdditionalFile4: 199 SupplementaryTable3)andBlautiaandVFM(P=3.65x10-06).Weconsideredasimilarapproachatthe 200 family level where the most significant bacterial-adiposity association was a negative relationship 201 between Christensenellaceae and VFM (P = 1.48x10-10) supporting prior findings from Goodrich et al 202 [31]. 203 Wethenexploredfecalmicrobiomeassociationswithobesitywithinasubsampleof247MZtwinpairs 204 to identify potential environmental or stochastic effects. MZ twins are matched for sex, age, have 205 nearly identical genomes, and very similar early life environments. Therefore, microbiome differences 206 observed within MZ twin pairs are likely to be a result of environmental and stochastic influences, or 207 effectsthataresecondarytothephenotype.WeestimateddifferencesinfecalOTUrelativeabundances 208 within 247 MZ twin pairs and compared these to differences in adiposity, for each of the 6 adiposity 10 209 phenotypes using a Pearson correlation. Although no results surpassed Bonferroni significance for 210 multipletestingacrossthe6phenotypes,potentiallyinpartduetothesmallersamplesize,manyofthe 211 observed effects were consistent with those observed across 1,313 individuals (Additional File 4: 212 Supplementary Table 2). The peak association in MZ twin pairs was observed between an unknown 213 ClostridialesOTU(GreengenesOTU331113)andAGR(Pearsoncoefficient=-0.24,P=9.68x10-05). 214 Weperformedtwoadditionalfollow-upanalysesoftheassociationbetweenfecalmicrobiomeOTUs 215 andadiposityphenotypes.Becauseofthehighcorrelationacrossmultipleobesitymeasures,we 216 exploredtheassociationbetweenfecalmicrobiomeOTUsandadipositymeasuresindependentofBMI. 217 FollowingadjustmentforBMI,133ofthe149ofsignificantassociationsremainednominallysignificant 218 withthesamedirectionofeffect(AdditionalFile4:SupplementaryTable2).Becauseofthestrong 219 associationthatweobservedbetweenalphadiversityandadiposity,wealsowantedtoassessifthe 220 strongestadiposity-OTUassociationswerewithOTUsthatweremarkersofdiversity,orwhetherthese 221 taxaassociatedwithadipositywereindependentofspeciesrichness.Tothisendwerepeatedthe 222 adiposity-OTUanalysesatthe149significantOTU-phenotypeassociationsnowincludingalphadiversity 223 asacovariateinthelinearmodelaspreviouslydescribed[34].Allofthereportedsignificantassociations 224 remainednominallysignificantafteradjustmentforalphadiversity. 225 Replicationofmicrobial-obesityassociations 226 Wepursuedreplicationofthe97significantOTUsassociatedwithvisceralfatin4,286independent 227 samplesfromthreeadditionalpopulation-basedreplicationsamples,includingsamplesfromthe 228 AmericanGut,FlemishGutFloraProject(FGFP),andextendedTwinsUKcohorts.Duetothelackof 229 cohortswithbothvisceralfatmeasurementsandgutmicrobiomedataavailable,BMIwasusedasa 230 surrogatephenotypeforvisceralfatintheseanalyses.Ineachreplicationcohort,individualsselected 231 wereofEuropeandescent,overtheageof20,andhadBMIrangingbetween18.5and30units, 11 232 resultingin2,338individualsfromtheAmericanGutcohort(USA),917individualsfromtheFGFPcohort 233 (Belgium),and1,031individualsfromtheextendedTwinsUKcohort(UK)whowerenotincludedinthe 234 discoveryTwinsUKsample.Toaccountforthedifferencebetweenthe16SrRNAgenesequences 235 betweenTwinsUKdiscoverysampleandthereplicationdatasets,OTUsineachreplicationdatasetwere 236 pickedusingclosedreferenceinQIIME[45]at97%usingUCLUST[46]againsttherepresentative 237 sequencesforthe97BonferronisignificantOTUsassociatedwithadipositymeasures,tomaximize 238 similarityindatasetsduringreplication.Furtherprocessinganddownstreamanalysestookintoaccount 239 technicalandlifestylecohort-specificcovariatestomatchascloselyaspossiblethediscoverysample 240 andaccountfordifferencesacrosscohorts(seeMethods). 241 The97OTU-BMIassociationsweretestedwithineachreplicationsample(AdditionalFile3),andthe 242 resultswerecombinedinameta-analysisacrossthethreereplicationsamples(AdditionalFile4: 243 SupplementaryTable4).AtastringentBonferronisignificancethreshold(P=0.05/97),8OTUsrobustly 244 replicatedwiththesamedirectionofeffectacrossthethreereplicationcohorts(Figure3b,Additional 245 File4:SupplementaryTable4),excludingresultswithevidenceforheterogeneity.Furthermore,13OTUs 246 showedevidenceforassociationwithBMIthatwasstrongerinthemeta-analysisacrossallfour 247 population-basedcohorts,comparedtothediscoveryTwinsUKsamplealone(AdditionalFile4: 248 SupplementaryTable4).Atamorerelaxedsignificancethreshold(nominalsignificanceandsame 249 directionofeffectinatleast1cohort),49OTUsshowedthesamedirectionofeffectandnominally 250 significantevidenceforassociationwithBMIinatleastoneofthereplicationsamples(AdditionalFile4: 251 SupplementaryTable4).The8OTUswithrobustevidenceforreplicationincludedmembersofthe 252 LachnospiraceaeandRuminococacceae,whilethe49OTUsincludednotjustmembersofthe 253 LachnospiraceaeandRuminococcaceae,butalsoChristensenellaceae,Clostridiales,Bacteroidaceaeand 254 Rikenellaceae. 12 255 Microbialfunctionalalterationsinobesity 256 Giventhestrongassociationsobservedbetweenfecalmicrobiomevariationandobesityphenotypes,we 257 nextwishedtoassesspotentialfunctionaldifferencesthatmaybetheresultofafecalmicrobiome 258 dysbiosisinobesity.Tothisend,weaimedtoinferKEGGfunctionsofthefecalmicrobesbyusing 259 PICRUSttopredictmetagenomesforeachsamplebasedontheclosed-reference16SrRNAgene 260 sequences.ThePICRUStanalysesresultedinaltogether233KEGGpathwaysand6,909KEGG 261 Orthologies(KOs)withinoursample.Wenextaimedtodeterminetherelationshipbetweentheinferred 262 KEGGfunctionsandadiposity.WeadjustedtheKEGGpathwayscoresobtainedforeachindividualin 263 oursamplefortechnicalcovariates(seeMethods),andperformedassociationanalysisofthese 264 functionswiththeadipositymeasuresusingalinearmixedeffectsmodelaspreviouslydescribed.We 265 alsoassessedKOdifferentialabundanceinhighandlowvisceralfatindividualsusingSTAMP[47]. 266 Ofthe233KEGGpathwayassociationswithsixadipositymeasures,13associationssurpassed 267 Bonferronicorrection(P=3.6x10-5)and218associationsweresignificantafterFDRcorrection(FDR5%). 268 Fourofthe13Bonferroni-significantassociationswerewithKEGGfunctionsrelatedtometabolism,as 269 wellas98oftheFDR5%significantresults(Figure4A).Functionswithincarbohydratemetabolismwere 270 positivelyassociatedwithadiposity,inparticular,glyoxylateanddicarboxylatemetabolism,whichhada 271 Bonferroni-significantassociationwithVFM(P=1.19x10-06),andFDR5%significantassociationswith 272 theremainingadipositymeasures.Fivegroupswithintheglyoxylateanddicarboxylatemetabolism 273 pathwayweresignificantlydifferentiallyabundantinhighvisceralfatandlowvisceralfatindividuals 274 (Figure4B).Twoofthese5groupsremainedsignificantfollowingBonferronicorrection,andthesewere 275 K03779(ttdA,Q=2.87x10-3)andK03780(ttdB,Q=1.79x10-3),bothincreasedinsubjectswithhigh 276 visceralfat.Theremaining3pathwaysthatsurpassedBonferronicorrectionwereobtainedbetweenpTF 277 anddioxindegradation,prenyltransferasesandN-glycanbiosynthesis. 13 278 Hostgeneticinfluencesonmicrobiome-obesityassociations 279 Twin-based heritability estimates supported a strong genetic component for visceral fat and our 280 heritability analyses of the fecal microbiome in this sample showed wide variability in heritability 281 between fecal microbial taxa (0–0.42), with members of Firmicutes and Actinobacteria being most 282 heritable.Theaverageheritabilityoftheoverallfecalmicrobiomesamplewas0.07,whiletheaverage 283 sharedenvironmentcomponentwas0.046andtheaverageuniqueenvironmentwas0.93. Thepeak97 284 BonferronisignificantOTUsassociatedwithabdominaladiposityinoursamplehadamedianheritability 285 of0.16andanaverageOTUheritabilityof0.16.Theaverageheritabilityofthe97adiposity-associated 286 OTUswassignificantlygreaterthantheoverallaverageheritabilityoverallOTUsat0.07(Wilcoxonrank 287 test,P=<2.2x10-16).Inadditiontheaverageuniqueenvironmentalcomponentofthe97issignificantly 288 lower than the overall average for all OTUs (0.79 vs 0.93, Wilcoxon rank test, P < 2.2x10-16). These 289 estimatessuggesthostgeneticsimpactsboththefecalmicrobiomeandadiposity. 290 To explore the hypothesis that host genetics may influence the observed microbial-adiposity 291 associations,weperformedcandidategeneanalysiscomparinghostgeneticvariantsathumanobesity 292 candidatelociwiththeadiposity-associatedfecalmicrobiomeprofiles.WeselectedSNPswithinhuman 293 loci previously associated with obesity as reported by Locke et al. [8], using common genetic variants 294 within 97 50-kb regions, centred around the peak BMI-associated GWAS SNP in each region. At a 295 Bonferroni-correctedP-valuethreshold(P=5.31x10-06)takingintoaccountthetotalnumberofgenomic 296 regions and adiposity-OTUs considered, OTU associations with genetic variants in 3 genomic regions 297 surpassedmultipletesting.Thestrongestassociationbetweenhostgenotypeandadiposity-associated 298 OTUs was observed between an OTU within the Clostridiales order (Greengenes OTU 181702) and a 299 host genetic variant within an intron of the FHIT gene (rs74331972 with OTU 181702 P = 2.49x10-06, 300 Figure5).FHITencodesthefragilehistidinetriadproteinandisatumoursuppressorgenethathasbeen 14 301 linkedtocancersofthedigestivetract.AlthoughthemostsignificantFHITassociationwasobtainedwith 302 OTU181702(h2=0.13),whichweidentifiedassignificantlyassociatedwithSFMandVFM(P=1.18x10- 303 06 304 SFM-associated OTU (rs74331972 with OTU 287790, P = 5.38x10-05). The second ranked significant 305 geneticassociationwasobtainedbetweenvariantsneargeneTDRG1(peakSNPrs1433723,P=4.32x10- 306 06 307 significantlyassociatedwithVFM(P=4.97x10-07,Figure5).Thefinalsignificantgeneticassociationwas 308 observed at a variant in an intron of the gene ELAVL4 (rs2480677, P = 4.95x10-06, Figure 5) with an 309 unknownBlautiaOTU,194733(h2=0.02),whichwehadidentifiedassignificantlyassociatedwithVFM 310 (P=1.27x10-07),SFM(P=2.26x10-06),andAGR(P=3.11x10-07)inthepeak149adiposity-OTUresults. 311 Whenweconsideredthegenetic-OTUassociationresultsatalessconservativesignificancethreshold(P 312 = 5x10-4), there were a total of 412 suggestive OTU-genetic associations located within or near 48 313 uniquegenes,includingobesitygenessuchasFTO,RPTORandTMEM18. 314 Functionalcharacterizationofhostgeneticvariantsassociatedwithadiposity-OTUmarkersinthegut 315 We next focused on the three significant host genetic associations in FHIT, TDRG1 and ELAVL4 with 316 adiposity-associated OTUs. The transcriptomic profiles for these genes from the GTEx resource[48] 317 indicatedthatallthreegenesareexpressedintissuesthatformpartofthegastrointestinaltract.FHITis 318 expressed across a wide range of tissues, including stomach, colon (transverse and sigmoid), small 319 intestine (terminal ileum), esophagus (muscularis and mucosa) and oesophagus-gastroesophageal 320 junction.TDRG1isexpressedathighestlevelsintestis,inmultiplebraintissues,aswellasoesophagus 321 (mucosa), while ELAVL4 is expressed at highest levels in multiple brain tissues, testis, pituitary gland, 322 colon(sigmoidandtransverse),pancreas,smallintestine(terminalileum),oesophagus(muscularis)and 323 oesophagus-gastroesophagealjunction. and 1.27x10-06 respectively), the same genetic variant was also associated with another VFM- and ) with an open reference unknown Clostridiales OTU (h2 = 0.14), which we had identified as 15 324 We next explored the functional impact of the SNPs in these three genes that we identified as most 325 significantly associated with adiposity-related fecal OTUs. The GTEx expression quantitative trait locus 326 (eQTL)analysisresultsindicatethatrs1433723inTDRG1isaneQTLforTDRG1expressionspecificallyin 327 oesophagus mucosa in the GTEx dataset. We then tested the associations of this variant on TDRG1 328 gene-expressionandDNAmethylationlevelsinadiposebiopsiesavailablefor542individualsfromthe 329 TwinsUKcohort[49,50].Wefoundthatrs1433723alsoissignificantlyassociatedwithDNAmethylation 330 levelsinTDRG1(peakassociation: rs1433723withTDRG1cg10553343,P=1.53x10-17;AdditionalFile4: 331 SupplementaryTable5),butnotwithgeneexpressionprofilesinthesesamplesinadiposetissue[49]. 332 The remaining two associated SNPs with adiposity-OTUs (rs74331972 in FHIT and rs2480677 in 333 ELAVL4) were not associated with the corresponding gene’s expression levels in GTEx across multiple 334 tissues. However, both variants showed modest effects on DNA methylation levels in adipose tissue 335 (rs74331972on FHIT cg15570148,P=3.4x10-4;andrs2480677onELAVL4cg00322486,P=1.4x10-3), 336 butnotongeneexpressionprofilesinadiposetissueinourdataset. 337 338 Discussion 339 Here,inthelargestmicrobiota-obesitystudytodateusingdetailedadiposityandvisceralfatmeasures, 340 wehaveshownthatfecalmicrobialdiversityandspecificmembersofthehumanfecalmicrobiotaare 341 strongly associated with obesity-related phenotypes, specifically abdominal adiposity. The majority of 342 microbialassociationswereobtainedwithvisceralfat,akeymetabolicdiseaseriskfactor,whichwealso 343 show is strongly heritable in our extended sample of over 3,000 twins. In addition to obtaining novel 344 heritabilityestimatesforvisceralfat,weshowthatandroid:gynoidratioishighlyheritableinthissame 345 cohort,andconfirmhighheritabilityestimatesfortheremainingadiposityphenotypes.UsingBMIasa 16 346 measure of obesity, we robustly replicate 8 obesity-associated fecal microbes in three independent 347 samplesfromtheAmericanGut,FGFP,andextendedTwinsUKdataset.Wealsodemonstratethathost 348 genes have an effect on components of the fecal microbiota, including the 97 adiposity-associated 349 OTUs,althoughthemechanismremainsunclear.Giventheimpactofhostgeneticsonbothobesityand 350 fecalmicrobes,andourfindingsofstrongassociationbetweenadiposityandfecalmicrobiomevariation, 351 our results therefore support the hypothesis that heritable microbes play a role in determining 352 components of obesity relevant to cardio-metabolic disease and may be one potential source 353 contributingtomissingheritabilityinobesity. 354 Heritabilityandimportanceofvisceralfat 355 Visceralfat,thetypeofadiposetissuewiththemostimportantimplicationsformetabolichealth[2]was 356 highlyheritable(0.70)andshowedthemostsignificantassociationswiththefecalmicrobiota.Previous 357 studies,forexamplefromtheFramingham[40],Quebec[42]andHeritage[41]family-basedcohortsused 358 CT scans to estimate visceral adiposity and found heritability estimates of visceral fat to be between 359 0.36 and 0.55, which is lower but comparable to our estimates in twins. Given the strong correlation 360 betweenmultipleadiposityphenotypes,wesoughttoassessifthehighheritabilityobservedinvisceral 361 fatisduetohighheritabilityofBMI.HeritabilityofvisceralfatremainedhighafteradjustmentforBMI 362 (0.64,seeAdditionalFile5),suggestingthatthehighheritabilitywereporthereforVFMisindependent 363 toitsassociationwithBMI. 364 Severalmicrobiomestudieshavelinkedincreasedoverallabdominaladiposity,forexampleusingbody 365 fat distribution measures such as waist-to-hip ratio, to fecal microbiome profiles[15, 51]. However, 366 althoughspecificprobioticintakehasbeenreportedtolowervisceralfat[52],todatetherehasnotyet 367 been a systematic comparison of visceral fat and variation in the fecal microbiome across human 368 subjects. Our results are therefore the first, to our knowledge, to link visceral fat with changes in the 17 369 fecalmicrobiotavariationinhumans.Thefindingssuggestthatvisceralfatmassismoreimportantfor 370 differences in the obese microbiome, rather than overall body mass. This is also demonstrated in the 371 recentClarkeetal.study[53],whereeliterugbyunionplayers,whoduetotheirgreatermusclemass 372 weredefinedasoverweightandobese(BMI=29± 3),hadamorediversemicrobiotathanthatofboth 373 lowBMIandhighBMIcontrols.Wealsoreportanovelandroid:gynoidratioheritabilityestimateof0.65, 374 higherthanpreviousfamily-basedestimatesof0.43[54].Android:gynoidratio,likevisceralfat,isalsoa 375 risk factor for cardio-metabolic disease, and accordingly our results show consistent direction of 376 associationeffectsforOTUsassociatedwithbothphenotypes. 377 Visceralfathaswidespreadassociationswiththehumanfecalmicrobiome 378 Due to its importance in cardio-metabolic disease risk, direct measures of visceral fat are much more 379 informative than BMI in assessing the metabolic consequences of obesity. Most microbiome obesity 380 studies to date have used BMI as a biomarker, with some mouse studies using epididymal fat. We 381 therefore present a novel dual approach, using visceral fat to find novel metabolic associations in a 382 sampleoftwins,andBMItoconfirmassociationsacrossindependentsamples.Thepowerofthisdesign 383 was apparent in the number of peak associations observed with visceral fat. All of the reported 384 adiposity-OTUs reported in this study were significantly associated with visceral fat, while only 7 of 385 theseOTUsweresignificantlyassociatedwithBMI,suggestingthatmicrobialstudiesthatonlyuseBMI 386 asameasurementofobesitymaybelimited. 387 Altogether, we identified 97 OTUs that were strongly significantly associated with the adiposity 388 phenotypes,andalloftheseweresignificantlyassociatedwithvisceralfat.Althoughthe97OTUswere 389 not the most abundant OTUs within the gut, OTUs within the Firmicutes phylum showed the most 390 significantadiposityassociations,perhapsinpartreflectingthedominantfrequencyofFirmicutesinthe 391 humangut.However,thedirectionofassociationdifferedfordifferentgenerawithintheFirmicutes.For 18 392 example, Oscillospira OTUs often showed protective associations with VFM while Blautia OTUs 393 commonlyshowedadverseassociationsthatcouldbeusedaspotentialmarkersofcardiovascularrisk. 394 Our findings also confirm the importance of fecal microbiome diversity in obesity, as we are able to 395 replicate strong association between adiposity phenotypes and alpha diversity. Furthermore, the 396 majorityoftheobserved97VFMassociationsareindependentofalphadiversityandBMI,suggesting 397 thattheOTUssignificantinthisanalysisarenotjustmarkersofdiversity,butformrealassociationswith 398 adiposityandcardio-metabolicdiseaserisk. 399 To explore the factors underlying the variation in the 97 adiposity-associated OTUs we used twin 400 modelling and determined that these 97 OTUs showed significantly greater average heritability (0.16) 401 compared to all OTUs in the larger dataset. The average shared environment component was also 402 higher, but the average unique environment component was lower than the overall dataset average. 403 Wethereforeinferthathostgeneticsinparticular,aswellasearlysharedenvironmentalfactors,playan 404 important role in the variation of obesity-associated microbes. However, although the estimate 405 (including measurement error) is approximately 0.80, the average unique environment component in 406 the97OTUswaslowerthantheoveralldatasetaverage.Thisisconsistentwithfindingsfromprevious 407 work[31]thatenvironmentalfactorsarekeydriversofthemicrobiome,andtheystronglycontributeto 408 variationattheobesity-microbiomeassociations. 409 AsourresultsconsistedofOTUassociationswithinthesamefamiliesandgeneraofbacteria,wealso 410 exploredassociationsofadiposityatthecollapsedtaxonomiclevel(familyandgenus)forthetopOTU 411 associations.LachnospiraceaeandRuminococcaceaecontinuetodisplayopposingdirectionsofeffectat 412 the family level. We also see that the heritable family Christensenellaceae is strongly associated with 413 visceral fat, as previously reported for BMI in Goodrich et al.[31]. Christensenellaceae was previously 414 foundtobeprotectiveofobesitybothinmiceandinhumantwinsinourpreviouswork[31].Weextend 19 415 thefindingshere,showingstrongprotectiveassociationsbetweenChristensenellaceaeandvisceralfat, 416 in particular suggesting that individuals with Christensenellaceae have less cardiovascular risk than 417 thosewithout.Furtherworktoelucidatethemechanismofprotectionisrequired. 418 PartialreplicationofadiposityassociationsusingBMIinindependentcohorts 419 Wepursuedreplicationofthefecalmicrobialadiposityassociationsinalargesampleof4,286Caucasian 420 individuals from three population-based cohorts. Due to the lack of cohorts with both visceral fat 421 measurementsandgutmicrobiomedataavailable,BMIwasusedasasurrogatemeasureforvisceralfat 422 andthereforethefindingsinthissectiondonotcapturestrictreplication,butrathervalidationofthe 423 visceralfatfecalmicrobialassociations.AtastringentBonferronithreshold8OTUsreplicatedacrossall 424 cohorts, and at a more relaxed threshold 49 OTUs replicated at nominal significance in at least one 425 cohort with the same direction of effect as in the discovery sample. The replication results included 426 OTUs classified within Lachnospiraceae and Ruminococacceae. A higher relative abundance of 427 Firmicuteshaspreviouslybeenassociatedwithobesityinsomebutnotallstudies,andseveralofthese 428 studiesshowanincreaseinFirmicutesinobesesubjects[12,55,56].Forexample,Ruminococcusgnavus 429 hasbeenshowntobesignificantlyenrichedinlowmicrobialgenecountindividualsthatwereproneto 430 obesityinonestudy[15].However,thephylumFirmicutescontainsover270generawithmanydifferent 431 anddiversefunctions.Becausechangesmayoccuratfiner-grainedtaxonomicresolution,andbecause 432 the baseline abundance of different genera differs among human populations, opposing selection 433 processes at different phylogenetic levels may in part explain differing results in studies determining 434 microbiota differences in obesity[57, 58], including recent meta-analyses across multiple obesity 435 microbiotadatasets[16,59].Thisdifferenceindirectionaleffectsingeneraofthesamefamilyhasbeen 436 noted in animal models. Mice consuming a Western diet have previously been reported to have 437 increased levels of Eubacterium dolichum[60], while another study has shown that mice consuming a 20 438 high-fat diet have decreased levels of Allobaculum OTUs[61]. Both of these microbes are members of 439 theErysipelotrichaceaefamily,andshowoppositedirectionsofeffectinourdata. 440 We observe a small number of robust associations across all cohorts, but over half of the findings 441 validatedinatleastoneofthereplicationcohorts.Thereasonsforthepartialreplicationofourresults 442 areunclearandarelikelyinparttocapturedifferencesinprotocols,genetics,geography,lifestyleand 443 diet, which are difficult to account for in full during the analysis. The American Gut dataset used the 444 protocolsdevelopedfortheEarthMicrobiomeProject[62]whicharelargelythesameasthoseusedin 445 Goodrich et al.[31] except that the American Gut data set is filtered to remove Gammaprotobacteria 446 sequences increased following transit. FGFP protocols are most similar to TUK-D and TUK-R however, 447 with similar sample collection methods and sequencing protocols. Analysis of the American Gut data 448 adjustedformorecovariatesthanTUK-R,TUK-DandFGFP,andthecohortitselfincludedadifferentage 449 range(average50,21-94)andgenderratio(53%female).TheagerangeandgenderratiooftheFGFP 450 cohort was similar to AG (average age = 51 (21-85), 55% female) while TUK-R was expectedly more 451 similartoTUK-D(averageage=57.3(20-89),80%female).Additionally,theAmericanGutsampleswere 452 primarilyfromtheUnitedStates(84%),unlikethediscoveryandreplicationTwinsUKsamplesfromthe 453 UnitedKingdom,andtheBelgianFGFPsamples.Themajorityofthe97OTUsshowthesamedirection 454 ofassociationintheTwinsUKdiscovery,TwinsUKreplication,andFGFPsamples,butthisisnotthecase 455 in the American Gut sample results (Additional File 3). This observation may partly reflect potential 456 differencesinOTUassociationsindifferentpopulationsandgeographicallocations,specificallybetween 457 European and American samples. Previous studies have shown some differences at the OTU-level 458 between countries[44, 63]. Additionally, the sample collection methods also differed between the 459 European studies with greater quantities collected in TUK-D, TUK-R and FGFP compared to swabs in 460 American Gut, although neither used fixatives. These factors highlight the on-going difficulty in 21 461 replicatingOTUlevelresultsbetweenstudiesandcohorts.Additionalstandardisationofbothtechnical 462 and analytical methods will help distinguish features common across populations, those sensitive to 463 technicalinfluence,andthoseuniquetoaparticularregionorgroup. 464 Microbialmetabolismisalteredinobeseindividuals 465 Ourfindingsofstrongassociationbetweenparticularmicrobesalongwiththeirdirectionaleffectin 466 obesity,canprovideinsightsintofunctionalimpacts.Ithasbeensuggestedthatthedysbiosisof 467 microbialspeciesinobesityisnotasimportantastheresultingfunctionaldysbiosis[64].Usingthe 468 PICRUStmethodtopredictmetagenomesandfunctions,wefindanenrichmentofmetabolism-related 469 KEGGfunctionsassociatedwithadiposity,inparticularcarbohydratemetabolismandN-glycan 470 biosynthesis.Weseeastrongpositiveassociationbetweenvisceralfatandglyoxylateanddicarboxylate 471 metabolism,particularlywiththegenesttdAandttdB.Theglyoxylatecycle,apathwaythatuntil 472 recentlywasthoughttobeabsentinmostanimals[65],isabletometabolisefatty-acidsintoglucose, 473 thuscontributingtoinsulinresistanceintheeventoffatty-acidabundance[66].Morerecently 474 glyoxylatehasbeenhighlightedasabiomarkeroftypeIIdiabetes,evenasmuchasthreeyearspriorto 475 diagnosisofdiabetes[67].Hereweshowapotentialmechanisticimpactofthemicrobialdysbiosis 476 observedinindividualswithincreasedvisceralfat.Wecannotassesswhethertheheritablemicrobesare 477 drivingthedifferencesobservedinthesepathways,however,itwouldbeinterestingtodeterminethe 478 heritabilityoftheinferredmicrobialfunctions,particularlyinfuturestudieswithmetagenomicdatasets. 479 Hostgeneticsinfluencesthehumanfecalmicrobiomeanditslinktocentraladiposity 480 Our new OTU heritability estimates in this larger twin sample remain largely similar to those in the 481 original Goodrich et al.[31] twin-based heritability analysis results, although peak heritable OTUs are 482 slightly different. Comparison of the heritability results at the genus level show that consistent 22 483 estimateswereobtainedacrossthetwodatasets(r2=0.67,AdditionalFile1:FigureA),andthesewere 484 also consistent with a recent heritability analysis of the extended TwinsUK dataset[33] (r2 = 0.77, 485 AdditionalFile1:FigureB).ThemostheritabletaxoninourdatasetisanOTUclassifiedasClostridium 486 perfringens(h2=0.42).TheheritabilityestimateforthemostheritableOTUreportedintheGoodrichet 487 al.[31] study, OTU 176318 assigned to Christensenellaceae (original heritability estimated at 0.36) 488 remainedsimilarintheextendeddataset,withanewheritabilityof0.31.ThemostheritableOTUsin 489 ourdatasetwererelativelyfrequent,beingpresentinatleast25%ofindividuals.Apotentialmechanism 490 that underlies fecal microbial heritability is the ability of the host genome to impact the gut 491 environment, for example, by altering acidity of the gut or by host control of miRNAs that can enter 492 bacterialcellsandaffectbacterialgrowth[68].Therefore,thelikelymechanismofmicrobialheritability 493 ishostgeneticmanipulationofmicrobes[69]. 494 TheOTUsthatweidentifiedassignificantlyassociatedwithadiposityphenotypesdisplayedsignificantly 495 greaterestimatesofheritabilitycomparedtoallOTUsprofiledinthegut,andthereforepromptedusto 496 test if specific human genetic variants may explain the observed microbial heritability at these 97 497 adiposity-associatedOTUs.Toexplorehowhumangenescouldinfluencethesemicrobesinthecontext 498 of obesity, we pursued host genetic association analyses of the adiposity-associated OTUs using a 499 candidate gene analysis approach at host obesity GWAS loci. The most significant genetic association 500 was obtained with a variant in the gene FHIT, which encodes the fragile histidine triad protein. 501 Significantly reduced FHIT gene expression levels have been reported in cardiac tissue from obese 502 compared to lean subjects[70]. Although this genetic variant is not an eQTL itself, it does show 503 moderate evidence for genetic impacts on DNA methylation levels in adipose tissue, specifically, at a 504 CpG-site in a CpG-island shore in the promoter of the FHIT gene. Therefore, it may have an indirect 505 impact on FHIT adipose tissue gene expression in obesity through epigenetic regulation of FHIT 23 506 expression. FHIT also appears to act as a tumour suppressor in several types of cancer[71], and its 507 abnormal function has been linked to cancers of the digestive tract[72]. The FHIT-associated OTU is 508 strongly associated with abdominal adiposity in our data, including SFM and VFM, suggesting that it 509 playsaroleincardio-metabolicdiseaseriskinthehost. 510 Human genetic variants in or near two other genes, ELAVL4 and TDRG1, were also significantly 511 associated with adiposity-OTUs. Although these genes currently lack a clear biological link to obesity, 512 they are expressed in tissues that are part of the GI tract. Furthermore, rs1433723 in TDRG1 is an 513 interestingcandidateforfollow-upasitexhibitsstronggeneticinfluencesonTDRG1geneexpressionin 514 parts of the GI tract (oesophagus mucosa). This genetic variant also impacts TDRG1 DNA methylation 515 levelsinadiposetissue,specificallyataCpG-siteinanintra-genicCpGislandinthegenebody,which 516 could mark an alternative transcription start site. GTEx summary expression profiles of this gene 517 correspondinglyshowexonbiasinexpressionlevelsacrossdifferenttissues,consistentwithalternative 518 transcriptusage. 519 ThefoursignificantlyassociatedhostgeneticvariantsinFHIT,ELAVL4,andTDRG1werenotthesameas 520 theleadSNPsforeachcandidateobesitylocusreportedinLockeetal.[8],andareinlowtomoderateLD 521 with the corresponding lead SNPs (R2 = 0.014-0.25). This could be attributed to several reasons, 522 includingsamplesizeandimputationdifferences.Comparedtooursampleof1,313twins,thesample 523 size in Locke et al.[8] is over 300,000 individuals, resulting in excellent power to detect phenotypic 524 associationswithparticularleadSNPs.Therewerealsodifferencesbetweentheimputationstrategies 525 usedbythetwostudies;whereasinLockeetal.[8]SNPswereimputedtoHapMap2referencepanel, 526 our dataset was imputed to the 1000 Genomes reference panel. These and other factors may impact 527 theassociationoftheleadSNPfromLockeetal.[8]withOTUsinthisdataset. 24 528 Ourheritabilityandhostgeneticassociationresultsshowthathumangeneticfactorshavearolebothin 529 determining obesity and in influencing the host fecal microbial composition. The strong associations 530 that we detect between fecal microbes and obesity phenotypes suggest that host genetics may 531 influence these microbial-obesity associations. Using a candidate gene approach we identified host 532 genetic variants in three obesity human loci, which show associations with adiposity-associated fecal 533 microbesinourdataset.Thesevariantsarepromisingcandidatesforfurtherfollowupstudiestoassess 534 theirpotentialroleinobesity-microbialinteractions.Althoughhumanobesityishighlyheritable,human 535 geneticvariantsdetectedtodateexplainonlyasmallproportionoftheheritabilityinobesity.Givenour 536 observationthattheaveragemicrobialheritabilityisgreateratthemostassociatedadiposity-OTUs,our 537 resultsarethereforeconsistentwiththehypothesisthataproportionoftheheritabilityinobesitymay 538 beexplainedbyheritablefecalmicrobes. 539 LimitationsandConsiderations 540 Due to the cross-sectional and observational nature of the study we are unable to determine causal 541 relationships between the fecal microbiota, host genetics and visceral adiposity. Furthermore, to 542 determinemorerobustgeneticassociations,werequireamuchlargersamplesizethanisrepresented 543 here.Lackofmetagenomicdatalimitsourfunctionalinterpretationofthemicrobialdysbiosisobserved 544 in obesity, although predictions from PICRUSt do provide some interesting insights for further 545 investigation. Some research has shown that in individuals that are dieting, there is an increase in 546 Bacteroidetesandconversely,individualsthatareover-eatingshowanabundanceofFirmicutes[55].It 547 isthereforeimportanttoknowwhethertheindividualsinthestudywerecalorierestrictedThisisone 548 limitationofthepresentstudyandfutureworkshouldfocusspecificallyonhowtheinteractionofdiet 549 affects visceral fat associations with the fecal microbiome. Other considerations that should be taken 550 intoaccountinthefuturearestoolconsistency[73],antibioticandotherdruguse[74,75]. 25 551 Duetothenoveltyofvisceralfatmeasurement,andlackofadipositymeasurementsotherthanBMIin 552 othermicrobiomedatasets,welackedtruereplicationinindependentcohortswhichwouldlikelyhave 553 improved our results as would have having a better balance of genders in discovery and replication 554 samples. 555 Conclusion 556 We have found strong associations between fecal microbial profiles with total and visceral fat, in the 557 largest fecal microbiota-obesity study to date using multiple measures of human adiposity, some of 558 which are robustly replicated. We identify novel and confirm previously established fecal microbe 559 associationswithoverallobesity,andadditionally,ourstudycanalsohelpdistinguishprotectiveandrisk 560 microbes in human cardiovascular and metabolic disease risk, by using visceral fat as risk phenotype. 561 We obtain novel high heritability estimates for visceral fat and android:gynoid ratio in a large twin 562 sample, and confirm previously reported heritability estimates for other adiposity measures. 563 Furthermore, we identify promising host genetic variants that may influence the interaction between 564 the human fecal microbiome and obesity and its metabolic consequences. Our findings support the 565 hypothesis that heritable microbes play a major role in the components of adiposity that are most 566 relevanttocardio-metabolicdiseaserisk. 567 568 Methods 569 Subjectsandsamplecollection 570 AllsubjectsincludedinthestudywerehealthyvolunteersfromtheTwinsUKAdultTwinRegistry[50,76]. 571 Adiposityphenotypedatawerecollectedontheextendedsampleof3,666twins(Table1).These 26 572 included1,044monozygotic(MZ)and789dizygotic(DZ)predominantlyfemaletwinpairs.Thesample 573 waspredominantlyofEuropeandescentandtheaverageagewas63(Table1).Fecalmicrobiomedata 574 wereobtainedforasubsetof1,313individualsfromthesample3,666twins.Themicrobiomedataset 575 included496MZ,594DZand223unrelatedindividuals,whowerealsopredominantlyfemale, 576 European,andtheaverageagewasalso63(Table1andAdditionalFile4:SupplementaryTable7).Fecal 577 samplesfromallindividualsfollowedthesamecollectionprotocolaspreviouslyreported[31,33,34]. 578 Briefly,fecalsampleswererefrigeratedorkeptonicefor1-2dayspriortoarrivingatthelaboratoryat 579 theDepartmentforTwinResearch,King’sCollegeLondon,atwhichpointtheywereimmediatelystored 580 forupto8weeksat-80°CbeforeDNAextraction.FrozensampleswereshippedtoCornellUniversityfor 581 DNAextraction,PCRamplification,andsequencing.Thestudywasapprovedbythelocalresearchethics 582 committee,andsignedandwrittenconsentwasobtainedfromallparticipants. 583 Fecalmicrobiomeprofiles 584 Thismanuscriptexplores16Sgutmicrobialprofilesfromatotalof1,313twinstoolsamples.Ofthese, 585 601 16S fecal microbiome profiles were previously described by Goodrich et al[35], and 671 profiles 586 were recently published in Jackson et al (2015)[34], and additional 16S microbial profiles were 587 generatedinafurther41twinstoolsamples[33],givingusaltogether1,313individualsforwhomboth 588 16S gut microbial data and adiposity measures were available. The sequence data for the discovery 589 TwinsUKfecalmicrobialsequencedatasetusedinthisstudyisavailablefromtheEuropeanNucleotide 590 Archive(ERP006339,ERP006342,ERP015317). 591 Briefly, all samples underwent the same laboratory protocol and data processing steps, following the 592 qualitycontrolprocedureoutlinedbyGoodrichetal.[35].DNAextractedfromfecalsamplesunderwent 593 amplification of the V4 region of the 16S rRNA gene using the 515F and 806R primers, followed by 594 250bp paired-end sequencing on the Illumina MiSeq platform. Processing of sequencing and OTU 27 595 pickingwascarriedoutaspreviouslydescribed[34].Inbrief,paired-endsequencesweremergedwithat 596 least a 200bp overlap, and those over 275bp in length were filtered from the dataset. The remaining 597 sequences were analysed using QIIME 1.7.0 (Quantitative Insights Into Microbial Ecology)[45]. 598 Sequences containing ambiguous or low quality reads (Phred score ≤ 25) and uncorrectable barcodes 599 were removed from the dataset and open-reference OTU picking was performed against the 600 GreengenesMay2013database.OTUsnotfoundinatleast25%ofindividualswerethendiscardedand 601 their counts were converted to relative abundances within samples, followed by the addition of a 602 pseudocount(10-6)toremovezerocounts.Mixedeffectsmodelswerefittedtothesedatatocontrol 603 for technical and batch effects. The regression included sequencing run, depth of sample sequencing, 604 andtechnicianwhoextractedandloadedtheDNAforsequencing,aswellassamplecollectionmethod 605 (bypostorinperson)aspredictorswithOTUabundancesastheresponse.Residualsfromthesemodels 606 were then used in subsequent downstream analyses. The adiposity-OTU and OTU candidate gene 607 downstreamanalysesalsoincludedfurtheradjustmentforcovariatessuchasage,gender,andfivelong- 608 term (10-year) summary broad dietary profiles (see phenotype Methods section). The final dataset 609 contained2,135OTUs. 610 Inestimatingcollapsedtaxonomyquantifications,OTUsfromthecompleteset(includingthosein<25% 611 of individuals) were collapsed into genera and families based on shared taxonomic assignment. Taxa 612 found in fewer than 10 individuals were then discarded and the counts converted to relative 613 abundances. We applied the OTU quality control regression framework to correct for technical 614 covariates and batch effects as described above, and then used the resulting residuals in subsequent 615 downstreamanalyses. 616 Phenotypedata 28 617 Obesity phenotype data were collected during each participant’s annual clinic visit. We explored 6 618 obesity related phenotypes in this work for a complete dataset of 3,666 phenotyped twins. The 619 adiposity phenotypes included a number of measures from total body DXA (Dual-Energy X-ray 620 Absorptiometry)whole-bodyscanning.DuringtheDXAprocedureparticipantswereaskedtolieflatand 621 straight while a full body scan took place, as previously reported[77] taking measures for percentage 622 trunk fat and visceral fat mass (g)[32]. Visceral fat mass was calculated from one cross-section of the 623 wholebodyatL4-L5,thetypicallocationofaCTslice.Otherphenotypescollectedweresubcutaneous 624 fat mass, percentage trunk fat, BMI, android:gynoid ratio and waist:hip ratio. Subjects were asked to 625 removetheirshoesandheight(cm)wasmeasuredusingastadiometer.Weight(kg)wasmeasuredon 626 digital scales. Waist circumference was measured using a tape, halfway between the lower border of 627 theribsandtheiliaccrestinahorizontalplane.Hipcircumferencewasmeasuredatthewidestpoint 628 overthebuttocks. 629 Covariatesforphenotypicanalysesincludedageandgenderinall3,666individualsanddietaryprofiles 630 in1,313individuals.Wedidnothavedetaileddietaryinformationatthetimeoffecalsampledonation 631 for this dataset, however we had available previously collected and reported dietary profiles [78]. 632 Teucheretal.collectedfoodfrequencyquestionnairesandperformedprincipalcomponentanalysison 633 these data. The proportion of variance explained by the first five principal components was 22% and 634 theycorrespondedbroadlytothefollowingapproximatediets:FruitandVeg,TraditionalEnglish,High 635 Alcohol, Dieting and Low Meat (see [78]). We used these five variables as long-term stable dietary 636 profilesinthedownstreammicrobiomeassociationanalyses. 637 StatisticalAnalysis 638 Toassesstheevidenceforassociationbetweenfecalmicrobiomecompositionandobesity-relatedand 639 metabolicphenotypes,weperformedtwoprincipalanalyses.First,wecomparedmicrobiomeOTUsand 29 640 phenotypesbyfittingalinearmixedeffectsmodel(LMER)usingtheRpackagelme4[79]acrossall1,313 641 individuals. In this model the phenotype was the response variable and the OTU was a fixed effect 642 predictor.Additionalfactorsincludedfamilyandzygositytakenintoaccountasrandomeffects,andsex, 643 age, and dietary profiles (principle components) considered as fixed effects. Each phenotype was 644 normalised to a standard Normal distribution prior to analysis. To assess the significance of the 645 associations,wecomparedthefullregressionmodeldescribedabove,toanullmodelthatexcludedthe 646 OTU predictor using an analysis of variance (ANOVA) test in R. We report associations that passed 647 nominalsignificance(P=0.05),aswellasthosethatpassedaBonferronithresholdformultipletesting 648 (P=3.90x10-06) 649 Heritability,theproportionoftotalvarianceinatraitattributedtogenetics,wasassessedusingtheACE 650 model. Under the assumption that the dominance effects are negligible, the ACE model can estimate 651 theadditivegenetic(A),commonenvironment(C)anduniqueenvironment(E)componentsofthetrait 652 variance. Narrow-sense heritabilities were estimated from the proportion of the total phenotypic 653 variance explained by estimated additive genetic effect. To estimate parameters of the ACE model, a 654 maximum-likelihood method was applied under multivariate normality assumptions using OpenMx 655 software[80],astructuralequationmodellingpackageinR.Wealsoprovide95%confidenceintervals 656 forbothphenotypeandOTUheritabilityestimatesinthisstudy. 657 FunctionalanalysiswasperformedusingPICRUSTv1.0.0.CountsofKEGGfunctionswereobtainedthen 658 transformed into relative abundances. The values were then transformed using the Hellinger 659 transformation prior to adjustment for the following technical covariates: sequencing run, technician 660 whoperformedtheDNAextractionandtechnicianwholoadedtheplate.Thefunctionalresidualsfrom 661 this adjustment were then fit into a linear mixed effects regression as predictor variables, where 662 adiposityphenotypesweretheresponse.Furthercovariatesintheregressionincludedage,sex,zygosity 30 663 and family structure as described above. Differential abundance analysis was performed on the 664 residuals using STAMP[47] whereby a two-sided Welch’s t-test was used to test the difference in KO 665 countsbetweenhighandlowvisceralfatgroups.Benjamini-HochbergFDRcorrection[81]wasappliedto 666 the association p-values from the linear mixed effects regression and the differential abundance 667 analysis. 668 HostGenomicAnalyses 669 Hostgenotypedatawereavailablein1,059individualsfromthemicrobiomedataset(seeAdditionalFile 670 5). Briefly, genotyping in TwinsUK was performed with a combination of Illumina HumanHap300, 671 HumanHap610Q, 1M-Duo and 1.2MDuo 1M chips, and genotypes were called as previously 672 described[50]. Imputation was performed using the IMPUTE software package (v2)[82]using as 673 reference panel the 1,000 Genomes haplotypes (based on SHAPEIT2) Phase I integrated variant set 674 release(v3)(September2013)(seeAdditionalFile5forfurtherdetails). 675 InthecandidategeneanalysisthelistofhumancandidateobesityGWASSNPassociationswasobtained 676 fromLockeetal.[8].ThereportedSNPsinLockeetal.weretakenastheleadSNPandcandidategene 677 regionswereextendedtoincludeadditionalSNPswithina25kbregioneithersideoftheleadSNP.SNPs 678 wereincludedindownstreamanalysisiftheyhadaMAFof5%andaninfoscoreofmorethan0.4inour 679 imputedsample.Overall,therewere8,876SNPsacross97uniquegenomicregionsthatwereincluded 680 indownstreamassociationanalyses.Hostgeneticassociationanalysiswasperformedusingthesoftware 681 GEMMA(Genome-wideEfficientMixedModelAssociation)[83]atthevariantsincludedinthe97human 682 obesity candidate loci. GEMMA implements a univariate linear mixed model to perform association 683 tests,usingakinshipmatrixtotakeintoaccounttwinrelatedness. 31 684 DNA methylation profiles were obtained in 542 female Caucasian twins using the Infinium 685 HumanMethylation450BeadChipassay(Illumina450k).TheDNAmethylationIllumina450kdatasetwas 686 obtainedfromadiposetissuebiopsiesinthetwins,aspreviouslydescribed[50](ArrayExpressE-MTAB- 687 1866). DNA methylation levels were first normalized using BMIQ[84], and adjusted for covariates 688 includingage,smoking,alcohol,zygosity,family,plate,bisulphite-sequencing(BS)conversionefficiency, 689 and BS conversion concentration (see Additional File 5). The resulting residuals were normalized and 690 methylation QTL analysis was performed using Matrix eQTL[85]. The analysis considered genetic 691 association under the additive model between genetic variants at rs74331972, rs1433723, rs2480677 692 and467,928DNAmethylationprobesthatpassedqualitycontrol. 693 Gene expression Quantitative Trait Locus (eQTL) results were available from a large-scale study of 694 humangeneexpressioninmultipletissuesamplesincludingsubcutaneousfat,lymphoblastoidcelllines 695 andwholeskin,derivedfrom856monozygotic(MZ)anddizygotic(DZ)femaletwinsfromtheTwinsUK 696 cohort, as part of the MuTHER project[49]. We interrogated the candidate SNPs for eQTL results in 697 adipose tissue (subcutaneous fat) using the Genevar software[86]. The functional impact of the 698 candidateSNPswasalsoexploredusingGTExresultsacrossmultipletissues,usingGTExAnalysisRelease 699 V6(dbGaPAccessionphs000424.v6.p1)[48]. 700 Replicationofobesity-microbialassociationsinindependentcohorts 701 AmericanGutCohort 702 Replicationanalyseswerepursuedin2,338individualsfromtheAmericanGutproject(seeAdditional 703 File5).AmericanGut(AG)participantswereselectedfromsequencerounds1-21(EBI:ERP012803)as 704 Caucasianovertheageof20withaBMIbetween18.5and30,andlivingintheUS,UK,Australiaor 705 Canada(AdditionalFile4:SupplementaryTable7).SequencingprotocolsoftheAmericanGutsamples 32 706 havepreviouslybeendescribed[62,87,88].Additionally,thefecalsamplehadtohaveatleast1,000 707 sequenceswhenpickedclosedreferenceusingSortMeRNA[89]againsttheAugust2013releaseof 708 Greengenes[90].AmericanGutsampleswerepre-processedtoremovecandidateovergrowth 709 sequences,asdescribedelsewhere[91].OTUswerepickedusingclosedreferenceinQIIME[45]at97% 710 usingUCLUST[46]againstthe97replicationOTUs,and96OTUsweresuccessfullyidentified.Regression 711 wasperformedinStatsmodels[92]usingathree-stepmodel.OTUswerepowertransformedandoffset 712 by1.Thepowertransformwasregressedtocontrolfortechnicalcovariates(seeAdditionalFile5). 713 Residualsfromthiswereregressedagainstlifestylecovariates,includingage,lastantibioticuse,IBD 714 diagnosis,flossingfrequencyandcountry.NormalizedBMIwasregressedagainsttheresidualfromthe 715 secondregression,withsexasacovariate.IntheAGBMI-OTUassociationresultsweobservedthat26 716 associationshadthesamedirectionofeffectwhile18werenominallysignificantandhadthesame 717 directioneffectastheTwinsUKdiscoverysample. 718 FGFPCohort 719 Replicationanalyseswerepursuedin917individualsfromtheBelgianFlemishGutFloraProject(FGFP) 720 (seeAdditionalFile5),selectedasCaucasiansovertheageof20withaBMIbetween18.5and30[17]. 721 PriortoOTUpicking,sequencingdepthwasdownsizedto10000readspersample.OTUswerepickedin 722 QIIME[45]at97%usingUCLUST[46]byclosedreferencepickingagainstthe97replicationOTUs,and96 723 OTUsweresuccessfullydetectedinFGFP.OTUabundanceswerepowertransformedandoffsetby1. 724 Thepowertransformwasregressedtocontrolforage,gender,alcoholaverageconsumptioninthe 725 weekpriortosamplingandsmoking(yes/no).NormalizedBMIwasthenregressedagainsttheresidual 726 fromthefirstregression,withgender,dietaryrestrictions(None/Vegetarian,Vegan,Macrobiotic/Other) 727 ascovariates. 33 728 IntheFGFPBMI-OTUassociationresultsweobservedthat83associationshadthesamedirectionof 729 effectwhile34werenominallysignificantandhadthesamedirectioneffectastheTwinsUKdiscovery 730 sample. 731 TwinsUKReplicationDataset 732 TheextendedTwinsUKfecalmicrobiomedatasetwasrecentlydescribed[33],andincludedasetof 733 1,031individuals(notoverlappingwiththe1,313TwinsUKdiscoverysamplehere)forwhombothfecal 734 microbialprofilesandBMI,butnotvisceralfat,wereavailable.Wethereforeincludedtheseadditional 735 dataasathirdreplicationsample(TwinsUK(TUK-R)),whereallindividualswereofEuropeandescent 736 overtheageof20withaBMIbetween18.5and30.Readdatawasclosed-referenceclusteredusingthe 737 representativesequencesforthereplicationOTUsasareference,usingUCLUST[46]at97%inQIIME 738 v1.9.0[45].CountsforeachOTUineachindividualwereconvertedtorelativeabundancesbydividingby 739 thetotalnumberofreadsinthesample.Acountof0.000001wasaddedtotherelativeabundancesto 740 accountforzerosandthesewerethenlogtransformed.Thetransformedcountswerethenresidualised 741 toadjustforsequencingrun,collectionmethod,personwholoadedtheplateandpersonwho 742 performedtheDNAextraction.Lifestylecovariatesincludedindownstreamanalysesmatchedthose 743 describedfortheTwinsUKdiscoverysample.Altogether,76associationsshowedthesamedirectionof 744 effectand7associationswerenominallysignificantintheTwinsUKreplicationsample. 745 746 Meta-analysisacrosscohorts 747 Random-effectsmeta-analysiswasfirstperformedacrossthethreereplicationsamples(AmericanGut, 748 FGFP,andTwinsUKreplication).WeconsideredOTU-BMIresultstoreplicateiftheyshowedthesame 749 directionofassociationwithBMIinthemeta-analysisasinthediscoveryTwinsUKsample,andifthe 34 750 significancelevelinthemeta-analysispassedBonferroniadjustmentformultipletesting(P=5.15x10-4). 751 Wealsoperformedameta-analysisacrossallfourpopulationsamples,toidentifyadditionalOTUsat 752 whichtheevidenceforassociationwithBMIimprovedoverthediscoveryTwinsUKP-value,evenifthese 753 didnotreachtheBonferronisignificancecut-offforreplication.Ineachmeta-analysis,weassessed 754 evidenceforheterogeneityusingCochran’sQ-statisticandtheI2-statistic[93],andonlyconsidered 755 resultswithnostrongevidenceforheterogeneity(Cochran’sQP>0.05andI2<0.75).Themeta-analysis 756 wasperformedinR3.1.2usingtheRpackage‘metafor’[94]usingasinputbetacoefficientsand 757 standarderrorsfromeachindependentcohort. 758 AvailabilityofDataandMaterials 759 Individual-level16Ssequencedatafor2294sampleswithinthisstudyareavailablethroughthe 760 EuropeanBioinformaticsInstitute(EBI)datarepositoryunderaccessionnumbersERP006339, 761 ERP015317andERP006342(Goodrichetal.2016).Allremaining16S,genotypeandphenotypedatain 762 thisstudyareavailableuponrequestthroughapplicationtotheTwinsUKdataaccesscommittee. 763 Informationondataaccessandhowtoapplyisavailableat 764 http://www.twinsuk.ac.uk/dataaccess/submission-procedure/ 765 Ethics 766 Ethicalapprovalforthephenotype,genotypeandmicrobiotastudieswithinTwinsUKwereprovidedby 767 theNRESCommitteeLondon–Westminster,London,UK.Writtenconsentwasobtainedfromall 768 participants.ResearchwascarriedoutinaccordancewiththeHelsinkideclaration. 769 Funding 35 770 This work was funded by NIH RO1 DK093595, a David and Lucile Packard Foundation Fellowship, the 771 Arnold and Mabel Beckman Foundation, the Cornell Center for Comparative Population Genomics, a 772 NationalScienceFoundationGraduateFellowship,theWellcomeTrustandtheEuropeanCommunity’s 773 SeventhFrameworkProgramme(FP7/2007-2013),andtheEuropeanResearchCouncil(projectnumber 774 250157). The study also received support from the National Institute for Health Research (NIHR) 775 BioResource Clinical Research Facility and Biomedical Research Centre based at Guy’s and St Thomas’ 776 NHSFoundationTrustandKing’sCollegeLondon. 777 Acknowledgements 778 WeacknowledgethetwinsandTwinsUKfordataaccess.TwinsUKwasfundedbytheWellcomeTrust; 779 European Community’s Seventh Framework Programme (FP7/2007-2013) and also receives support 780 from the National Institute for Health Research (NIHR)-funded BioResource, Clinical Research Facility 781 and Biomedical Research Centre based at Guy's and St Thomas' NHS Foundation Trust in partnership 782 withKing'sCollegeLondon.SNPgenotypingwasperformedbyTheWellcomeTrustSangerInstituteand 783 National Eye Institute via NIH/CIDR. In addition, the study received support from the MRC, grant 784 numberMR/N030125/1. 785 ConflictOfInterest 786 Theauthorsdeclarenoconflictofinterest. 787 788 AuthorContributions 36 789 J.T.B,T.D.S,A.G.C.,andR.E.Ldesignedthestudy.T.D.SandJ.T.Boversawsamplecollection.M.B.,J.K.G., 790 M.A.J.,T.P.,E.R.D.andI.E.performedthebioinformaticsandstatisticalanalyses.R.K.providedthedata 791 forAmericanGutreplicationandJ.Dperformedthereplicationanalyses.J.R.providedthedataforthe 792 Flemish Gut Flora Project and S.V-S. performed the replication analyses. M.B and J.T.B. wrote the 793 manuscript.Allauthorscontributedtothepreparationandrevisionofthemanuscript. 794 795 References 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. BarnessLA,OpitzJM,Gilbert-BarnessE:Obesity:Genetic,molecular,andenvironmentalaspects. AmericanJournalofMedicalGeneticsPartA2007,143A:3016-3034. DespresJ-P:BodyFatDistributionandRiskofCardiovascularDiseaseAnUpdate.Circulation2012, 126:1301-1313. MaesHHM,NealeMC,EavesLJ:Geneticandenvironmentalfactorsinrelativebodyweightandhuman adiposity.BehaviorGenetics1997,27:325-351. AllisonDB,KaprioJ,KorkeilaM,KoskenvuoM,NealeMC,HayakawaK:Theheritabilityofbodymass indexamonganinternationalsampleofmonozygotictwinsrearedapart.InternationalJournalof Obesity1996,20:501-506. O'RahillyS,FarooqiIS:HumanObesity:AHeritableNeurobehavioralDisorderThatIsHighlySensitiveto EnvironmentalConditions.Diabetes2008,57:2905-2910. SamarasK,SpectorTD,NguyenTV,BaanK,CampbellLV,KellyPJ:Independentgeneticfactorsdetermine theamountanddistributionoffatinwomenafterthemenopause.JournalofClinicalEndocrinology& Metabolism1997,82:781-785. PoirierP,GilesTD,BrayGA,HongYL,SternJS,Pi-SunyerFX,EckelRH:Obesityandcardiovasculardisease -Pathophysiology,evaluation,andeffectofweightloss.ArteriosclerosisThrombosisandVascular Biology2006,26. LockeAE,KahaliB,BerndtSI,JusticeAE,PersTH,DayFR,PowellC,VedantamS,BuchkovichML,YangJ,et al:Geneticstudiesofbodymassindexyieldnewinsightsforobesitybiology.Nature2015,518:197-206. ShunginD,WinklerTW,Croteau-ChonkaDC,FerreiraT,LockeAE,MagiR,StrawbridgeRJ,PersTH,Fischer K,JusticeAE,etal:Newgeneticlocilinkadiposeandinsulinbiologytobodyfatdistribution.Nature 2015,518:187-196. YangW,KellyT,HeJ:Geneticepidemiologyofobesity.EpidemiologicReviews2007,29. ClaessonMJ,JefferyIB,CondeS,PowerSE,O'ConnorEM,CusackS,HarrisHMB,CoakleyM, LakshminarayananB,O'SullivanO,etal:Gutmicrobiotacompositioncorrelateswithdietandhealthin theelderly.Nature2012,488. LeyRE,BackhedF,TurnbaughP,LozuponeCA,KnightRD,GordonJI:Obesityaltersgutmicrobialecology. ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica2005,102. ArmougomF,HenryM,VialettesB,RaccahD,RaoultD:MonitoringBacterialCommunityofHumanGut MicrobiotaRevealsanIncreaseinLactobacillusinObesePatientsandMethanogensinAnorexic Patients.PlosOne2009,4. EverardA,BelzerC,GeurtsL,OuwerkerkJP,DruartC,BindelsLB,GuiotY,DerrienM,MuccioliGG, DelzenneNM,etal:Cross-talkbetweenAkkermansiamuciniphilaandintestinalepitheliumcontrols 37 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. diet-inducedobesity.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica 2013,110:9066-9071. LeChatelierE,NielsenT,QinJ,PriftiE,HildebrandF,FalonyG,AlmeidaM,ArumugamM,BattoJ-M, KennedyS,etal:Richnessofhumangutmicrobiomecorrelateswithmetabolicmarkers.Nature2013, 500:541-+. WaltersWA,XuZ,KnightR:Meta-analysesofhumangutmicrobesassociatedwithobesityandIBD. FebsLetters2014,588:4223-4233. FalonyG,JoossensM,Vieira-SilvaS,WangJ,DarziY,FaustK,KurilshikovA,BonderMJ,Valles-ColomerM, VandeputteD,etal:Population-levelanalysisofgutmicrobiomevariation.Science2016,352:560-564. TurnbaughPJ,LeyRE,MahowaldMA,MagriniV,MardisER,GordonJI:Anobesity-associatedgut microbiomewithincreasedcapacityforenergyharvest.Nature2006,444:1027-1031. DingS,ChiMM,ScullBP,RigbyR,SchwerbrockNMJ,MagnessS,JobinC,LundPK:High-FatDiet:Bacteria InteractionsPromoteIntestinalInflammationWhichPrecedesandCorrelateswithObesityandInsulin ResistanceinMouse.PlosOne2010,5. TilgH,KaserA:Gutmicrobiome,obesity,andmetabolicdysfunction.JournalofClinicalInvestigation 2011,121. TurnbaughPJ,HamadyM,YatsunenkoT,CantarelBL,DuncanA,LeyRE,SoginML,JonesWJ,RoeBA, AffourtitJP,etal:Acoregutmicrobiomeinobeseandleantwins.Nature2009,457:480-484. ArumugamM,RaesJ,PelletierE,LePaslierD,YamadaT,MendeDR,FernandesGR,TapJ,BrulsT,BattoJM,etal:Enterotypesofthehumangutmicrobiome.Nature2011,473:174-180. Vijay-KumarM,AitkenJD,CarvalhoFA,CullenderTC,MwangiS,SrinivasanS,SitaramanSV,KnightR,Ley RE,GewirtzAT:MetabolicSyndromeandAlteredGutMicrobiotainMiceLackingToll-LikeReceptor5. Science2010,328:228-231. TurnbaughPJ,BaeckhedF,FultonL,GordonJI:Diet-inducedobesityislinkedtomarkedbutreversible alterationsinthemousedistalgutmicrobiome.CellHost&Microbe2008,3:213-223. Romero-CorralA,SomersVK,Sierra-JohnsonJ,ThomasRJ,Collazo-ClavellML,KorinekJ,AllisonTG,Batsis JA,Sert-KuniyoshiFH,Lopez-JimenezF:Accuracyofbodymassindexindiagnosingobesityintheadult generalpopulation.InternationalJournalofObesity2008,32:959-966. WajchenbergBL:Subcutaneousandvisceraladiposetissue:Theirrelationtothemetabolicsyndrome. EndocrineReviews2000,21:697-738. FontanaL,EagonJC,TrujilloME,SchererPE,KleinS:Visceralfatadipokinesecretionisassociatedwith systemicinflammationinobesehumans.Diabetes2007,56:1010-1013. SaitoT,MurataM,OtaniT,TamemotoH,KawakamiM,IshikawaS-e:Associationofsubcutaneousand visceralfatmasswithserumconcentrationsofadipokinesinsubjectswithtype2diabetesmellitus. EndocrineJournal2012,59:39-45. ZoetendalEG,AkkermansADL,Akkermans-vanVlietWM,deVisserJAGM,deVosWM:Thehost genotypeaffectsthebacterialcommunityinthehumangastrointestinaltract.MicrobialEcologyin HealthandDisease2001,13:129-134. HansenEE,LozuponeCA,ReyFE,WuM,GurugeJL,NarraA,GoodfellowJ,ZaneveldJR,McDonaldDT, GoodrichJA,etal:Pan-genomeofthedominanthumangut-associatedarchaeon,Methanobrevibacter smithii,studiedintwins.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesof America2011,108:4599-4606. GoodrichJK,WatersJL,PooleAC,SutterJL,KorenO,BlekhmanR,BeaumontM,VanTreurenW,KnightR, BellJT,etal:HumanGeneticsShapetheGutMicrobiome.Cell2014,159:789-799. LivshitsG,KatoB,WilsonS,SpectorT:Linkageofgenestototalleanbodymassinnormalwomen. JournalofClinicalEndocrinology&Metabolism2007,92:3171-3176. GoodrichJK,DavenportER,BeaumontM,JacksonMA,KnightR,OberC,SpectorTD,BellJT,ClarkAG,Ley RE:GeneticDeterminantsoftheGutMicrobiomeinUKTwins.CellHost&Microbe2016,19:731-743. JacksonMA,GoodrichJK,MaxanM-E,FreedbergDE,AbramsJA,PooleAC,SutterJL,WelterD,LeyRE, BellJT,etal:Protonpumpinhibitorsalterthecompositionofthegutmicrobiota.Gut2015. 38 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. GoodrichJuliaK,WatersJillianL,PooleAngelaC,SutterJessicaL,KorenO,BlekhmanR,BeaumontM, VanTreurenW,KnightR,BellJordanaT,etal:HumanGeneticsShapetheGutMicrobiome.Cell2014, 159:789-799. SnijderMB,VisserM,DekkerJM,SeidellJC,FuerstT,TylavskyF,CauleyJ,LangT,NevittM,HarrisTB:The predictionofvisceralfatbydual-energyX-rayabsorptiometryintheelderly:acomparisonwith computedtomographyandanthropometry.InternationalJournalofObesity2002,26:984-993. BertinE,MarcusC,RuizJC,EschardJP,LeuteneggerM:MeasurementofvisceraladiposetissuebyDXA combinedwithanthropometryinobesehumans.InternationalJournalofObesity2000,24:263-270. KaulS,RothneyMP,PetersDM,WackerWK,DavisCE,ShapiroMD,ErgunDL:Dual-EnergyX-Ray AbsorptiometryforQuantificationofVisceralFat.Obesity2012,20:1313-1318. DirekK,CeceljaM,AstleW,ChowienczykP,SpectorTD,FalchiM,AndrewT:Therelationshipbetween DXA-basedandanthropometricmeasuresofvisceralfatandmorbidityinwomen.BmcCardiovascular Disorders2013,13. FoxCS,MassaroJM,HoffmannU,PouKM,Maurovich-HorvatP,LiuC-Y,VasanRS,MurabitoJM,MeigsJB, CupplesLA,etal:Abdominalvisceralandsubcutaneousadiposetissuecompartments-Associationwith metabolicriskfactorsintheFraminghamHeartStudy.Circulation2007,116:39-48. RiceT,DawEW,GagnonJ,BouchardC,LeonAS,SkinnerJS,WilmoreJH,RaoDC:Familialresemblance forbodycompositionmeasures:TheHERITAGEFamilyStudy.ObesityResearch1997,5:557-562. ChaputJ-P,PérusseL,DesprésJ-P,TremblayA,BouchardC:FindingsfromtheQuebecFamilyStudyon theEtiologyofObesity:GeneticsandEnvironmentalHighlights.CurrentObesityReports2014,3:54-66. TurnbaughPJ,HamadyM,YatsunenkoT,CantarelBL,DuncanA,LeyRE,SoginML,JonesWJ,RoeBA, AffourtitJP,etal:Acoregutmicrobiomeinobeseandleantwins.Nature2009,457:480-U487. LeeS,SungJ,LeeJ,KoG:ComparisonoftheGutMicrobiotasofHealthyAdultTwinsLivinginSouth KoreaandtheUnitedStates.AppliedandEnvironmentalMicrobiology2011,77:7433-7437. CaporasoJG,KuczynskiJ,StombaughJ,BittingerK,BushmanFD,CostelloEK,FiererN,PenaAG,Goodrich JK,GordonJI,etal:QIIMEallowsanalysisofhigh-throughputcommunitysequencingdata.Nature Methods2010,7:335-336. EdgarRC:SearchandclusteringordersofmagnitudefasterthanBLAST.Bioinformatics2010,26:24602461. ParksDH,BeikoRG:Identifyingbiologicallyrelevantdifferencesbetweenmetagenomiccommunities. Bioinformatics2010,26:715-721. LonsdaleJ,ThomasJ,SalvatoreM,PhillipsR,LoE,ShadS,HaszR,WaltersG,GarciaF,YoungN,etal:The Genotype-TissueExpression(GTEx)project.NatureGenetics2013,45:580-585. GrundbergE,SmallKS,HedmanAK,NicaAC,BuilA,KeildsonS,BellJT,YangT-P,MeduriE,BarrettA,etal: Mappingcis-andtrans-regulatoryeffectsacrossmultipletissuesintwins.NatureGenetics2012, 44:1084-+. GrundbergE,MeduriE,SandlingJK,HedmanAK,KeildsonS,BuilA,BuscheS,YuanW,NisbetJ,Sekowska M,etal:GlobalAnalysisofDNAMethylationVariationinAdiposeTissuefromTwinsRevealsLinksto Disease-AssociatedVariantsinDistalRegulatoryElements.AmericanJournalofHumanGenetics2013, 93:876-890. DaoMC,EverardA,Aron-WisnewskyJ,SokolovskaN,PriftiE,VergerEO,KayserBD,LevenezF,ChillouxJ, HoylesL,etal:Akkermansiamuciniphilaandimprovedmetabolichealthduringadietaryintervention inobesity:relationshipwithgutmicrobiomerichnessandecology.Gut2016,65:426-436. KadookaY,SatoM,ImaizumiK,OgawaA,IkuyamaK,AkaiY,OkanoM,KagoshimaM,TsuchidaT: Regulationofabdominaladipositybyprobiotics(LactobacillusgasseriSBT2055)inadultswithobese tendenciesinarandomizedcontrolledtrial.EuropeanJournalofClinicalNutrition2010,64:636-643. ClarkeSF,MurphyEF,O'SullivanO,LuceyAJ,HumphreysM,HoganA,HayesP,O'ReillyM,JefferyIB, Wood-MartinR,etal:Exerciseandassociateddietaryextremesimpactongutmicrobialdiversity.Gut 2014,63:1913-1920. 39 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. ZillikensMC,YazdanpanahM,PardoLM,RivadeneiraF,AulchenkoYS,OostraBA,UitterlindenAG,Pols HAP,vanDuijnCM:Sex-specificgeneticeffectsinfluencevariationinbodycomposition.Diabetologia 2008,51:2233-2241. LeyRE,TurnbaughPJ,KleinS,GordonJI:Microbialecology-Humangutmicrobesassociatedwith obesity.Nature2006,444:1022-1023. IsmailNA,RagabSH,AbdElBakyA,ShoeibARS,AlhosaryY,FekryD:FrequencyofFirmicutesand BacteroidetesingutmicrobiotainobeseandnormalweightEgyptianchildrenandadults.Archivesof MedicalScience2011,7. BervoetsL,VanHoorenbeeckK,KortlevenI,VanNotenC,HensN,VaelC,GoossensH,DesagerKN, VankerckhovenV:Differencesingutmicrobiotacompositionbetweenobeseandleanchildren:acrosssectionalstudy.GutPathogens2013,5. PayneAN,ChassardC,ZimmermannM,MuellerP,StincaS,LacroixC:Themetabolicactivityofgut microbiotainobesechildrenisincreasedcomparedwithnormal-weightchildrenandexhibitsmore exhaustivesubstrateutilization.Nutrition&Diabetes2011,1. FinucaneMM,SharptonTJ,LaurentTJ,PollardKS:ATaxonomicSignatureofObesityintheMicrobiome? GettingtotheGutsoftheMatter.PlosOne2014,9. TurnbaughPJ,RidauraVK,FaithJJ,ReyFE,KnightR,GordonJI:TheEffectofDietontheHumanGut Microbiome:AMetagenomicAnalysisinHumanizedGnotobioticMice.ScienceTranslationalMedicine 2009,1. RavussinY,KorenO,SporA,LeDucC,GutmanR,StombaughJ,KnightR,LeyRE,LeibelRL:Responsesof GutMicrobiotatoDietCompositionandWeightLossinLeanandObeseMice.Obesity2012,20:738747. GilbertJA,JanssonJK,KnightR:TheEarthMicrobiomeproject:successesandaspirations.BmcBiology 2014,12. KemppainenKM,ArdissoneAN,Davis-RichardsonAG,FagenJR,GanoKA,Leon-NoveloLG,VehikK, CasellaG,SimellO,ZieglerAG,etal:EarlyChildhoodGutMicrobiomesShowStrongGeographic DifferencesAmongSubjectsatHighRiskforType1Diabetes.DiabetesCare2015,38:329-332. HuttenhowerC,GeversD,KnightR,AbubuckerS,BadgerJH,ChinwallaAT,CreasyHH,EarlAM,FitzGerald MG,FultonRS,etal:Structure,functionanddiversityofthehealthyhumanmicrobiome.Nature2012, 486:207-214. KondrashovFA,KooninEV,MorgunovIG,FinogenovaTV,KondrashovaMN:Evolutionofglyoxylatecycle enzymesinMetazoa:evidenceofmultiplehorizontaltransfereventsandpseudogeneformation. BiologyDirect2006,1. SongS:Cantheglyoxylatepathwaycontributetofat-inducedhepaticinsulinresistance?Medical Hypotheses2000,54:739-747. NikiforovaVJ,GiesbertzP,WiemerJ,BethanB,LooserR,LiebenbergV,NoppingerPR,DanielH,ReinD: Glyoxylate,aNewMarkerMetaboliteofType2Diabetes.JournalofDiabetesResearch2014. LiuS,daCunhaAP,RezendeRM,CialicR,WeiZ,BryL,ComstockLE,GandhiR,WeinerHL:TheHost ShapestheGutMicrobiotaviaFecalMicroRNA.Cellhost&microbe2016,19:32-43. vanOpstalEJ,BordensteinSR:Rethinkingheritabilityofthemicrobiome.Science2015,349:1172-1173. Philip-CoudercP,PathakA,SmihF,DambrinC,HarmanceyR,BuysS,GalinierM,MassabuauP,RoncalliJ, SenardJM,RouetP:Uncomplicatedhumanobesityisassociatedwithaspecificcardiactranscriptome: involvementoftheWntpathway.FasebJournal2004,18:1539-+. ZanesiN,PekarskyY,CroceCM:AmousemodelofthefragilegeneFHIT:Fromcarcinogenesistogene therapyandcancerprevention.MutationResearch-FundamentalandMolecularMechanismsof Mutagenesis2005,591:103-109. OhtaM,InoueH,CotticelliMG,KasturyK,BaffaR,PalazzoJ,SiprashviliZ,MoriM,McCueP,DruckT,etal: TheFHITgene,spanningthechromosome3p14.2fragilesiteacidrenalcarcinoma-associatedt(3;8) breakpoint,isabnormalindigestivetractcancers.Cell1996,84:587-597. 40 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. VandeputteD,FalonyG,Vieira-SilvaS,TitoRY,JoossensM,RaesJ:Stoolconsistencyisstrongly associatedwithgutmicrobiotarichnessandcomposition,enterotypesandbacterialgrowthrates.Gut 2016,65:57-62. BarcM-C,Charrin-SarnelC,RochetV,BourliouxF,SandreC,BoureauH,DoreJ,CollignonA:Molecular analysisofthedigestivemicrobiotainagnotobioticmousemodelduringantibiotictreatment: InfluenceofSaccharomycesboulardii.Anaerobe2008,14. PreidisGA,VersalovicJ:TargetingtheHumanMicrobiomeWithAntibiotics,Probiotics,andPrebiotics: GastroenterologyEnterstheMetagenomicsEra.Gastroenterology2009,136:2015-2031. MoayyeriA,HammondCJ,ValdesAM,SpectorTD:CohortProfile:TwinsUKandHealthyAgeingTwin Study.InternationalJournalofEpidemiology2013,42:76-85. MoayyeriA,HammondCJ,HartDJ,SpectorTD:Effectsofageongeneticinfluenceonbonelossover17 yearsinwomen:TheHealthyAgeingTwinStudy(HATS).JournalofBoneandMineralResearch2012, 27:2170-2178. TeucherB,SkinnerJ,SkidmorePML,CassidyA,Fairweather-TaitSJ,HooperL,RoeMA,FoxallR,Oyston SL,CherkasLF,etal:DietarypatternsandheritabilityoffoodchoiceinaUKfemaletwincohort.Twin ResearchandHumanGenetics2007,10:734-748. BatesD,MaechlerM,BolkerBM,WalkerSC:FittingLinearMixed-EffectsModelsUsinglme4.Journalof StatisticalSoftware2015,67:1-48. BokerS,NealeM,MaesH,WildeM,SpiegelM,BrickT,SpiesJ,EstabrookR,KennyS,BatesT,etal: OpenMx:AnOpenSourceExtendedStructuralEquationModelingFramework.Psychometrika2011, 76:306-317. BenjaminiY,HochbergY:CONTROLLINGTHEFALSEDISCOVERYRATE-APRACTICALANDPOWERFUL APPROACHTOMULTIPLETESTING.JournaloftheRoyalStatisticalSocietySeriesB-Methodological1995, 57:289-300. HowieBN,DonnellyP,MarchiniJ:AFlexibleandAccurateGenotypeImputationMethodfortheNext GenerationofGenome-WideAssociationStudies.PlosGenetics2009,5. ZhouX,StephensM:Genome-wideefficientmixed-modelanalysisforassociationstudies.Nature Genetics2012,44:821-U136. TeschendorffAE,MarabitaF,LechnerM,BartlettT,TegnerJ,Gomez-CabreroD,BeckS:Abeta-mixture quantilenormalizationmethodforcorrectingprobedesignbiasinIlluminaInfinium450kDNA methylationdata.Bioinformatics2013,29:189-196. ShabalinAA:MatrixeQTL:ultrafasteQTLanalysisvialargematrixoperations.Bioinformatics2012, 28:1353-1358. YangT-P,BeazleyC,MontgomerySB,DimasAS,Gutierrez-ArcelusM,StrangerBE,DeloukasP, DermitzakisET:Genevar:adatabaseandJavaapplicationfortheanalysisandvisualizationofSNP-gene associationsineQTLstudies.Bioinformatics2010,26:2474-2476. CaporasoJG,LauberCL,WaltersWA,Berg-LyonsD,HuntleyJ,FiererN,OwensSM,BetleyJ,FraserL, BauerM,etal:Ultra-high-throughputmicrobialcommunityanalysisontheIlluminaHiSeqandMiSeq platforms.IsmeJournal2012,6:1621-1624. WaltersW,HydeER,Berg-LyonsD,AckermannG,HumphreyG,ParadaA,GilbertJA,JanssonJK,Caporaso JG,FuhrmanJA,etal:ImprovedBacterial16SrRNAGene(V4andV4-5)andFungalInternalTranscribed SpacerMarkerGenePrimersforMicrobialCommunitySurveys.mSystems2015,1. KopylovaE,NoeL,TouzetH:SortMeRNA:fastandaccuratefilteringofribosomalRNAsin metatranscriptomicdata.Bioinformatics2012,28:3211-3217. McDonaldD,PriceMN,GoodrichJ,NawrockiEP,DeSantisTZ,ProbstA,AndersenGL,KnightR, HugenholtzP:AnimprovedGreengenestaxonomywithexplicitranksforecologicalandevolutionary analysesofbacteriaandarchaea.IsmeJournal2012,6:610-618. AmericanGutAnalysisiPythonNotebook[<spanstyle="font-size:12.0pt][mso-bidi-language:ARSA">http://nbviewer.jupyter.org/github/biocore/American-Gut/blob/master/ipynb/module2_v1.0.ipynb] SeaboldJS,PerktoldJ:Statsmodels:EconometricandStatisticalModelingwithPython.In9thPythonin ScienceConference.2010 41 1028 1029 1030 1031 1032 1033 93. 1034 1035 1036 1037 94. 95. HigginsJPT,ThompsonSG:Quantifyingheterogeneityinameta-analysis.StatisticsinMedicine2002, 21:1539-1558. ViechtbauerW:ConductingMeta-AnalysesinRwiththemetaforPackage.JournalofStatisticalSoftware 2010,36:1-48. LetunicI,BorkP:InteractiveTreeOfLifev2:onlineannotationanddisplayofphylogenetictreesmade easy.NucleicAcidsResearch2011,39:W475-W478. 42 FigureLegends Figure1.Distributionandheritabilityofadiposityphenotypes.A)Scatterplotmatrixshowingthe distributionandcorrelationbetweensixadipositymeasuresin3,666twins.Thedistributionof eachphenotype(priortonormalisation)isshownalongthediagonal.Thelowerpanelshows scatterplotsforeachpairofadiposityphenotypesandtheupperpaneldenotesthecoefficientsof determination.B)HeritabilityofsixadipositymeasuresintheTwinsUKcohort,aswellasvisceral fatmeasuresin3independentcohorts:Framingham[40],Quebec[42]andHeritage[41].Thetotal varianceofeachadiposityphenotypeisdecomposedintovariancecomponentsattributedto additivegenetics(A)ornarrow-senseheritability(h2),commonenvironment(C)andunique environment(E). Figure2.Alphadiversityofthefecalmicrobiomeinindividualswithhighandlowfatcontent. Foreachphenotype,individualswhoweremorethan1.5standarddeviationsfromthemeanof thephenotypewereassignedtohighandlowphenotypegroupsrespectively.Alphadiversity measures(usingShannondiversity)werecomparedbetweenthehighandlowphenotypegroups (Wilcoxontest*=0.05**=0.001***=0.0001). Figure3.Associationsbetweenfecalmicrobiome16SOTUsandvisceralfatintheTwinsUK andreplicationdatasets.A)TheinnercircledenotesthephylogenetictreeofOTUs,produced usingiTOL[95]basedonGreenGenesMay2013treefilteredfortheOTUsinthesample.Tree leavesarecolouredaccordingtothedirectionofassociationwithvisceralfat,whereblue indicatesanegativeassociation,whileredindicatesapositiveassociation.Theoutercircle denotesthesignificanceofeachOTU-visceralfatassociation,wherePvaluesareplottedas– 43 log10(Pvalue)andtheredlineshowstheBonferronisignificancethreshold.Thefigure highlightsthemost-associatedOTUinthesample(OTU372146),aswellasthetwoclosedreferenceOTUsthatweresignificantlyassociatedwithhostgeneticvariantsingenesFHIT (OTU181702)andELAVL4(194733).Inadditionitalsohighlightstheheritable ChristensenellaceaeOTU176318.Thefigurealsodenotesthetreebranchescontaining membersoftheClostridiales,Bacteroides,andChristensenellaceaetoaccompanyresultsand discussioninthemaintext.B)Forestplotofbetacoefficientswithconfidenceintervalsof8 OTUsthatreplicatedrobustlyinameta-analysisof3independentcohorts(TUK-R,AG& FGFP). Figure4.Microbialfunctionalanalysisinobesity.A)MicrobialPICRUSt-predictedKEGGfunctions relevanttometabolisminthetwindataset,andtheirassociationwiththesixadipositymeasures. TheheatmapdenotesthedirectionofassociationbetweeneachmicrobialPICRUSt-predicted KEGGfunctionandadipositymeasures,whereblueindicatesanegativeassociation,whilered indicatesapositiveassociation.Bonferroni-significantassociationsarehighlighted(*).B)FiveKO genesthataredifferentiallyabundantbetweenhighandlowvisceralfatindividualsinglyoxylate anddicarboxylatemetabolism,astestedbyatwo-sidedWelch’st-test.FDR-adjustedp-valuesare reportedattherightoftheimageandstarsindicateBonferroni-significantassociations.Figurewas producedusingSTAMP[47]. Figure5.PeakgeneticassociationsbetweenobesityhumangeneticvariantsandadiposityassociatedOTUsinthetwinfecalmicrobiome.A)AssociationbetweenOTU181702andFHITSNP rs74331972.TheboxplotindicateschangeinOTU181702abundancewithgenotypeatSNP rs74331972.TheLocusZoomplotdenotesthestrengthofassociationofOTU181702withSNP rs74331972,aswellastheSNPsinthesurroundingregion.B)Associationbetweenopen 44 referenceOTU25576andTDRG1SNPrs1433723.Theboxplotindicateschangeinopenreference OTU25576abundancewithgenotypeatSNPrs1433723.TheLocusZoomplotdenotesthe strengthofassociationofopenreferenceOTU25576withSNPrs1433723,aswellastheSNPsin thesurroundingregion.C)AssociationbetweenOTU194733andELAVL4SNPrs2480677.The boxplotindicateschangeinOTU194733abundancewithgenotypeatSNPrs2480677.The LocusZoomplotdenotesthestrengthofassociationofOTU194733withSNPrs2480677,aswellas theSNPsinthesurroundingregion.SNPsinallLocusZoomplotsarecolouredaccordingtotheir strengthofLDwiththepeakSNPplotted. 45 Tables Table1.DescriptionoftheTwinsUKdiscoverysample. FullDataseta MicrobiomeSubsetb 3666 1313 Female 3296 1266 Male 370 47 MZ 2088 496 DZ 1578 594 0 223 Age(mean(range)) 63(32-87) 63(32-87) BMI(mean(range)) 26.1(15.7-49.9) 26.1(16.2-45.9) European 3511 1295 Other 102 9 Unknown 53 8 NoofSamples Sex Zygosity Unrelated Ethnicity a Summariesareshownforboththeextendeddatasetof3,666twinsusedinthephenotype heritabilityanalyses,andforthemicrobiomesamplesubsetof1,313individuals.bAdditionalFile 4:SupplementaryTable7includesextendeddescriptionsofthemicrobiomedatasubset. 46 AdditionalFiles AdditionalFile1:Genus-levelheritabilityestimatesbetweentheTUKdiscoverydatasetand Goodrich.A)Scatterplotshowingthegenus-levelheritabilitybetweenTUK-DandGoodrichetal (2014)[2](r2=0.67).B)Scatterplotshowingthegenus-levelheritabilitybetweenTUK-Dand Goodrichetal(2016)(r2=0.76).(PDF) AdditionalFile2:Comparisonofalphadiversityandhumanadiposityphenotypes.Alphadiversity ismeasuredusingtheShannonmetric.Foreachadiposityphenotypeconsideredinthisstudywe presenttheR2andstrengthofassociation,andatrendlineisshowninred.(PDF) AdditionalFile3:ConcordanceofOTU-BMIassociationresultsatthe97OTUsbetweenTwinsUK discoverysampleandthethreeindependentreplicationcohorts.Pointsmarkedinredshowno consistenteffectbetweenthestudieswhilebluedenotesconsistentdirectionofassociation.Blue pointsthatareemptycirclesarenotsignificant,whilebluefilledcirclesindicatenominally significantresults.A)ScatterplotbetweenTwinsUKdiscoverysample(TUK-D)andAmericanGut (AG).B)ScatterplotbetweenTUK-DandFGFP.C)ScatterplotbetweenTUK-DandTwinsUK replicationsample(TUK-R).(PDF) AdditionalFile4:SupplementaryTables.SupplementaryTable1:OTUsshowingatleastmoderate (A>0.2)heritabilityintheextendedTwinsUK16Sdataset.Eachvariancecomponent,additive genetics(A,orheritability),sharedenvironment(C)anduniqueenvironment(E)areshownalong withwithupperandlower95%confidenceintervals.AveragerelativeabundanceoftheOTUsin theoveralldatasetisalsoshown.SupplementaryTable2:149Bonferonni-significantOTUadiposityassociations.OTU-adiposityresultsareobtainedfromlinearmixedeffectsregression models,andincludetheoriginalresults(OTU-Adiposity)andresultsafteradjustmentforBMI (OTU-AdiposityBMI-adjusted).WealsoprovidetheOTUheritabilityandaverageOTUabundance inthecohort.ThelasttwocolumnsshowtheOTUresultsfromthewithinMZtwin-pairdifference analyses(MZDiffBeta),andwhethertheMZresultswereconcordantwiththelinearmixedeffects results.Reportedphenotypesarevisceralfat(VFM),subcutaneousfat(SFM),android:gynoidratio (AGR),BMI,waist:hipratio(WHR)and%trunkfat(pTF).SupplementaryTable3.Genus-level 47 associationswithadiposity,wherethecollapsedtaxonomyOTUgenuswassignificantlyassociated withadiposity.Reportedphenotypesarevisceralfat(VFM),subcutaneousfat(SFM), android:gynoidratio(AGR),andBMI.SupplementaryTable4:ReplicationofTwinsUK(TUK-D) resultsin3independentcohorts,theAmericanGut(AG),theFlemishGutFloraProject(FGFP)and theexpandedTwinsUKdataset(TUK-R).TheTableliststhe97OTUsforming149significant adiposityassociationsinTwinsUK,andtheirassociationwithBMIinboththereplicationand discoverysamples.Inadditionthetableoutlinestheresultsoftwometa-analysesthatwere performedacrossthestudies,thefirstofjusttheindependentcohorts,andthesecondincluding thediscoverycohort.Finally,thelastsectionofthetableshowswhichOTUswerereplicatedinat leastoneofthereplicationcohorts.SupplementaryTable5:Humangenomicanalysesat adiposity-relatedcandidatehostgenesFHIT,TDRG1andELAVL4.Thetableshowssignificant geneticassociationsbetweentheadiposity-relatedcandidatehostgeneticvariantsandadiposityassociatedOTUs.Thefinalthreecolumnssummarizetheresultsoftheassociationbetweenthe hostgeneticvariantsandDNAmethylationatCpGsitestargettedbytheIllumina450Karray.DNA methylationassociationwasperformedattheleadOTU-associatedSNPineachlocus,therefore rs1433722wasnottested(NA).SupplementaryTable6:Listofthe97BMI-associatedloci reportedinLockeetal.(2015)thatwereusedforanalysisinthisstudy.Thelastcolumndenotes theBMIGWASp-valuesfromLockeetal.(2015).SupplementaryTable7:Demographicsofthe TwinsUKmicrobiomediscoverysampleandtheAmericanGutreplicationsample.Supplementary Table8:ENAaccessionIDsforsamplesinthisstudycurrentlyavailableonlineandbasicmetadata. (XLSX) AdditionalFile5:Supplementarymethodsincludingexpandedmethodsforanalysesanddata collectionperformedwithinthediscoverycohort.(DOCX) 48