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