www.sciencemag.org/cgi/content/full/science.

1208344/DC1

Supporting Online Material for

Linking Long-Term Dietary Patterns with Gut Microbial Enterotypes

Gary D. Wu,* Jun Chen, Christian Hoffmann, Kyle Bittinger, Ying-Yu Chen, Sue A. Keilbaugh,
Meenakshi Bewtra, Dan Knights, William A. Walters, Rob Knight, Rohini Sinha, Erin Gilroy,
Kernika Gupta, Robert Baldassano, Lisa Nessel, Hongzhe Li, Frederic D. Bushman,*
James D. Lewis*

*To whom correspondence should be addressed. E-mail: [email protected] (G.D.W.);

[email protected] (J.D.L.); [email protected] (F.D.B.)

Published 1 September 2011 on Science Express

DOI: 10.1126/science.1208344

This PDF file includes:

Materials and Methods
Figs. S1 to S5
References (14–20)

Other supporting material for this report includes the following:

Tables S1 to S11
Materials and Methods

Human subjects.

Healthy volunteers between the ages of 18 and 40 were recruited to participate in

the controlled feeding experiment. To be eligible, participants were required to be

free from any chronic gastrointestinal disease, cardiac disease, diabetes mellitus

or immunodeficiency diseases, to have a normal bowel frequency (minimum

once every 2 days, maximum 3 times per day), to have body mass index (BMI)

between 18.5 and 35. Participants could not have taken antibiotics within 6

months prior to enrollment, proton pump inhibitors, H2 receptor antagonists,

tricyclic antidepressants, narcotics, anticholinergic medications, laxatives, or anti-

diarrhea medications within 4 weeks of enrollment, or NSAIDs, dietary

supplements, or antacids within 2 weeks prior to enrollment. Similar eligibility

criteria were applied for our cross-sectional study except that the age range for

participation was 2 years to 50 years and participants were required to have

been weaned from nursing. All participants provided informed consent, or assent

in the case of minors. Legal guardians provided informed consent for minors.

Participants completed three 24-hour dietary recalls during the week before

collection of stool samples to assess recent dietary composition. The third recall

was performed for the day preceding collection of the first stool sample or

beginning the inpatient stays for the controlled feeding experiments. Dietary

  2  
recalls were conducted by trained bionutritionists and nutrient intake was

computed with the Nutrition Data System program (University of Minnesota). In

addition, each participant or their guardian completed a food frequency

questionnaire appropriate for the participant’s age that assesses usual dietary

composition over the preceding year. The complete list of nutrients studied is in

Table S3 (Recall) and Table S4 (FFQ).

In the controlled feeding experiment, healthy volunteers were randomly assigned

to either a high fiber/low fat diet or a low fiber/high fat diet. Each participant

consumed identical meals for 10 consecutive days. The composition of the two

study diets was identical. Only the portion sizes were modified to adjust the

distribution of fat, carbohydrate, and protein in the diet. Total calories in the high

fat were 38% from fat, 35% from carbohydrates, and 27% from protein. In the low

fat diet, total calories were 13% from fat, 69% from carbohydrates, and 18% from

protein. Portion sizes were also calculated based on the expected caloric

requirements for the participant. Stool samples were collected daily and

immediately frozen at -80⁰C. Sigmoidoscopy without bowel cleansing was

performed on the first and last day of the inpatient stay. Mucosal pinch biopsies

were obtained from the rectum using large cup forceps and samples were flash

frozen in liquid nitrogen. Participants consumed 24 x-ray-opaque markers at the

time of entry into the research center. Abdominal x-rays were taken 1, 3, and 5

days later to quantify whole gut transit during this time period. Participants were

not allowed to leave the clinical research unit without being accompanied by a

  3  
member of the research team. Participants were instructed to consume all food

provided to them within each 24-hour period. Water, tea and coffee were

provided ad lib but no sweetener or milk products could be added. The

characteristics of the CAFE subjects are summarized in table S5, and the

compositions of their diets are summarized in table S6.

DNA preparation

Stool or biopsy samples were stored at -80oC prior to use. DNA was purified

using the MoBio PowerSoil kit according to the manufacturer's instructions with

addition of a high temperature heating step to improve lysis. DNA samples were

amplified using V1-V2 region primers targeting bacterial 16S genes and

sequenced using 454/Roche Titanium technology. DNA sequence reads from

this study are available from the Sequence Read Archive (CaFE: SRX037803,

SRX021237, SRX021236, SRX020772, SRX020771, SRX020588, SRX020587,

SRX020379, SRX020378 (metagenomic). COMBO: SRX020773, SRX020770).

Sample metadata (compliant with the MIMARKS standard) are available as

tables S7 and S8 below.

Quantification of absolute levels of bacterial and human DNA.

We analyzed the total abundance of bacterial and human DNA in a subset of our

samples using quantitative PCR assays for bacterial 16S rDNA gene segments

  4  
and the human beta-tubulin gene. We found on average 5x108 16S rDNA copies

and ~300 beta-tubulin copies per microgram of total DNA in the 95 samples

tested. Given some simple assumptions (mean of five 16S gene copies per

bacterium and 5 Mb per bacterial genome), we calculate that bacterial DNA

accounts for the majority of the mass of DNA in our stool samples, and the

contribution of human DNA is several orders of magnitude less. The estimated

number of bacteria detected using the 16S Q-PCR assay per gram of stool (wet

weight) ranged from 2X108 to 7X1010, a range that overlaps with earlier studies.

Smaller proportions of bacterial DNA were detected in biopsy samples,

consistent with a larger contribution of human DNA (data not shown). No

differences in absolute bacterial abundance were detected between diets.

Statistical analysis.

Initial analysis was carried out using the QIIME pipeline

(14)(http://qiime.sourceforge.net/) which implemented taxonomic assignment with

RDP(15), and distances-based analysis using UniFrac(16). We used the default

parameter settings in the QIIME pipeline, including Lane masking (17), and a

single even-depth rarefaction analysis (depth = 2408), chosen to exclude only the

lowest-depth sample, to control for sequencing effort (clustering analyses were

not, however, sensitive to sequencing depth). The analysis produced weighted

and unweighted UniFrac distance matrices and a table of per-sample genus

counts.

  5  
 We then performed clustering by partitioning around medoids (PAM)(18) using

Jensen-Shannon divergence (JSD) of the normalized genus counts. Weighted

UniFrac distance, Euclidean distance and Bray-Curtis distance of the normalized

genus counts were also compared. The optimal number of clusters was chosen

by the maximum average silhouette width, known as the silhouette coefficient

(SC)(19). The quality of those clusters was assessed by the same measure,

following the accepted interpretation that SC values above 0.5 indicate a

reasonable clustering structure (20).

Lane masking was used in generating alignments for the analysis described

above, which involves masking hypervariable regions to avoid artifacts due to

convergent sequence evolution. Results without Lane masking were compared

and found to be generally parallel, showing strong support for two clusters

dominated with Bacteroides and Prevotella, and weak support in one case for a

third cluster. Results were compared in fig. S2.

Overall associations between dietary / demographic variables and microbiome

compositions were assessed using PERMANOVA based on weighted or

unweighted UniFrac distances. Associations between dietary nutrients and

individual taxa proportions were assessed by Spearman's rank correlation test

while associations between enterotypes and individual taxa proportions or dietary

nutrients were assessed by Wilcoxon rank sum test. False discovery rate (FDR)

control was used to account for multiple comparisons when evaluating these

  6  
associations. Nutrient intake was normalized using the residual method to

standardize for caloric intake. Quantities of nutrients were standardized over all

98 samples to have mean=0 and standard deviation=1.

For the shotgun metagenomic analysis, reads were aligned using BLAST and

annotated using KEGG. We used a two-sample t test to compare the changes

for each functional category between diets. All category counts were converted

to relative abundance before performing the analysis. Columns were normalized

to mean 0 and standard deviation 1. P values for shotgun metagenomic analysis

were not corrected for multiple comparisons. Most statistical analysis was

conducted using the statistical software package R.

  7  
  

Additional analysis

Association of gut microbial composition with BMI

We investigated the relationship of microbiome data to demographic data using

PERMANOVA (table S1). Among the variables tested, BMI was among the most

strongly associated with microbiome composition (p=0.001 unweighted analysis;

p=0.145 weighted analysis). Additional demographic variables achieving

significance included sex, race and consumption of yogurt and alcohol.

Presence of pets, appendectomy history and drinking of tap water achieved

marginal significance. Since weighted UniFrac analysis puts relatively more

weight on common taxa and unweighted analysis puts more weight on rare taxa,

some differences were observed between these two types of analysis, as

expected. In general, quantitative and qualitative metrics provide different

insights into relationships between community structure and external factors that

may be important(16).

We compared short-term and long-term diet using two types of questionnaires. A

validated food frequency questionnaire ("FFQ") queried usual (long term) diet.

Three 24 hour dietary recalls within 7 days of sample collection were used to

assess recent food intake ("Recall"). The dietary patterns measured between the

two were partially correlated, so we first used principal component analysis to

  8  
summarize the diets. The 154 nutrients for Recall and 214 nutrients for FFQ

were used to generate five principal components, which explained 55% and 58%

of the total variation, respectively. We then performed PERMANOVA to test the

overall association between these 5 PCs and microbiome variation, where the

microbiome data were summarized as weighted or unweighted pairwise UniFrac

distances. The composition of both the recent diets (p<0.001 unweighted;

p=0.047 weighted) and usual diets (p=0.011 unweighted; p=0.003 weighted)

were associated with microbiome composition.

To test the independent association of diet and BMI on the composition of the gut

microbiome, multivariate models were constructed including BMI, sex, race and

the principle components for recent diet. Adjusting for sex and race had minimal

impact on the associations of recent diet and BMI with the microbiome. In a

model including only BMI and diet, the strength of the associations was less

strong, suggesting that at least part of the association of BMI with microbiome

composition was due to the influence of diet and vice versa (table S9). Similar

results were observed when BMI and total fats or saturated fatty acid intake was

included in the model.

Analysis of effects of red wine consumption

At FDR 25%, we identified five anthocyanidins of various forms and total

anthocyanidins to be associated with microbiome composition (unweighted

  9  
UniFrac). In an American diet, the major sources of anthocyanidins are red wine

and some fruits. These anthocyanidins showed weak to moderate correlations

with red wine consumption (Spearman correlation 0.23 to 0.62). Red wine

consumption is also associated with overall microbiome composition (p = 0.003;

PERMANOVA, unweighted analysis). To determine whether anthocyanidins

and red wine consumption are independently associated with the overall

microbiome composition, we tested the association of anthocyanidins and red

wine consumption with microbiome composition after adjustment for each other.

Possibly due to the relatively weak correlation between anthocyanidin and red

wine consumption, the strength of the associations were minimally attenuated in

the adjusted analyses (table S10). Thus, red wine consumption and

anthocyanidin consumption appear to be independently associated with the

microbiome composition.

Application of published enterotype clustering methodology

To reproduce a previously published enterotype clustering methodology (4), we

performed clustering by PAM using the square root of the Jensen-Shannon

divergence, and chose the number of clusters by the Cali ksi-Harabasz (CH)

index of the relative clustering quality as defined in the original publication of the

method (20). The CH index indicated that three clusters were preferred, but the

quality score for three clusters (SC=0.17) indicated no substantial structure. We

also applied the CH Index to clustering using several alternative distance

  10  
measures (Bray-Curtis, Euclidean, Jensen-Shannon, weighted UniFrac, and

weighted normalized UniFrac). Interestingly, in all but one case (weighted

UniFrac) the CH index chose three as the optimal number of clusters, even

though the silhouette scores were substantially higher for two clusters. No

reasonable support (SC ≥ .5) for three clusters was found using any distance

measure.

Prevotella-Bacteroides gradient analysis

The enterotype clustering is driven primarily by the ratio of the two dominant

genera, Prevotella to Bacteroides; this ratio defines a clear gradient across the

putative COMBO enterotypes (fig. S5; note that 69 samples with no Prevotella

were excluded), emphasizing that the boundary between enterotypes is not

sharply defined. When we removed these genera, the structure was

undetectable (17 clusters, SC=0.115). Also, these genera compose between

12% and 83% of the relative abundance in the communities (mean +/- s.d. = 0.46

+/- 0.17), and the only distance measures that produced reasonable support for

clustering, JSD and Euclidean distance, are measures that emphasize

differences in the largest components of a distribution. These findings highlight

the importance of these two genera.

  11  
Supplementary Figures

fig. S1. Associations between bacterial taxa and food groups. A) Heat map as in

Fig. 1, but showing lower level phylogenetic assignments of bacterial taxa (right

of the figure panel). Rows correspond to bacterial taxa, columns correspond to

nutrients or nutrient classes. Colors represent the taxa-nutrient Spearman’s

correlations, where red color indicate positive association, blue color negative

association, and * indicates the significant association at the false discovery rate

(FDR) of 25%. The “F” or “R” following the label for each nutrient indicates the

origin of the data from either FFQ or recall, respectively. Clustering was carried

out using neighbor joining. B) A view of two-way clustering of Spearman’s

correlations between nutrients and microbiome taxa, where bacterial lineages are

grouped by Phylum.

fig. S2. Enterotype clustering under different data processing and clustering

methods. Clustering in the COMBO data was probed using several methods with

and without Lane masking (in addition to the analysis in Figure 2 showing

Jensen-Shannon clustering on data processed with Lane masking). For each

silhouette scores provide a measure of the relative strength of clustering.

fig. S3. Bacterial taxa associated with each enterotype in the COMBO data. A)

Proportional representation. The proportion of each genus is standardized using

  12  
z-score transformation to reflect the relative abundance. B) Absolute

representation.

fig. S4. Analysis of CAFE samples using shot-gun metagenomics. A)

Comparison of Phylum level proportions for the ten CAFE1 subjects determined

using 16S tag sequencing (left) or shot-gun metagenomic sequencing (right). A

total of 423 sequences from the metagenomic dataset were identified as

containing an rRNA gene region and used in the analysis. The identified rDNA

containing sequences, as well as the amplified sequences, were classified using

RDP classifier (bootstrap cutoff 80%). B) Analysis of gene functional categories

(KEGG) that changed in reciprocal directions depending on the diet. Red and

blue bars indicate samples obtained from subjects fed a high fat/low fiber and low

fat/high fiber diet, respectively. P values are not corrected for multiple

comparisons. C) Domain level lineages detected in shot-gun metagenomic

analysis of the controlled feeding study. Proportions of reads in KEGG

categories detected in the controlled feeding samples are in table S11.

fig. S5. Prevotella-Bacteroides gradient.

The log ratio of the relative abundance of Prevotella to Bacteroides in a given

sample is plotted against the first principal coordinate of that sample based on

Jensen-Shannon divergence (Fig. 2). Samples are colored to match their

enterotype assignments (green: Bacteroides predominant enterotype; red:

Prevotella predominant). This plot excludes those subjects with zero Prevotella

  13  
abundance due to the undefined log ratio (69 of the subjects). The clear gradient

across the enterotype boundary emphasizes that the boundary is not sharply

defined.

  14  
Supplementary Tables

table S1. Demographic variables and their association with microbiome

composition. P-values are not adjusted for multiple comparisons.

table S2. Association of nutrients and enterotype partitioning. The order of the

nutrients is the same as in Figure 2C. Mean values of the normalized nutrient

intake are given for both enterotypes with standard deviation indicated in the

bracket. Unadjusted P values from Wilcoxon rank sum test as well as the FDR

adjusted P values (Q values) are also listed.

table S3. Complete list of the dietary categories analyzed for Recall

questionnaire and subject data.

table S4. Complete list of the dietary categories analyzed for the FFQ

questionnaire and subject data.

table S5: Characteristics of participants in CAFE study.

table S6: Comparison of usual, recent and assigned diets of CAFE participants.

  15  
table S7: Metadata for COMBO samples, arranged according to the MIMARK

standard.

table S8: Metadata for CAFE samples, arranged according to the MIMARK

standard.

table S9: Multivariate models to test the interaction between BMI and diet on the

strength of association with microbiome composition.

table S10. Association of anthocyanidin consumption with the microbiome

composition after adjustment for red wine consumption and association of red

wine consumption with the microbiome composition after adjustment for

anthocyanidin. The * indicates the p value is for the association of red wine and

the microbiome composition.

table S11. Proportions of reads in KEGG categories detected in the controlled

feeding samples.

  16  
Supplementary References

14.   J.  G.  Caporaso  et  al.,  Nat  Methods  7,  335  (2010).  
15.   Q.  Wang,  G.  M.  Garrity,  J.  M.  Tiedje,  J.  R.  Cole,  Applied  and  Environmental  
Microbiology  73,  5261  (2007).  
16.   C.  A.  Lozupone,  M.  Hamady,  S.  T.  Kelley,  R.  Knight,  Applied  and  Environmental  
Microbiology  73,  1576  (2007).  
17.   D.  J.  Lane,  in  Nucleic  Acid  Techniques  in  Bacterial  Systematics,  E.  Stackebrandt,  
M.  Goodfellow,  Eds.  (John  Wiley  and  Sons,  Chichester,  1991),    pp.  115-­‐175.  
18.   L.  Kaufman,  P.  J.  Rousseeuw,  Finding  Groups  in  Data:  An  Introduction  to  
Cluster  Analysis.,    (1990).  
19.   P.  J.  Rousseeuw,  J.  Comput.  Appl.  Math.  20,  53  (1987).  
20.   T.  Calinski,  J.  Harabasz,  Communications  in  statistics  1,  1  (1974).  
 
 
 

  17  
Fig. S1
A
* * * * * * * * * * * * * * * * * * * * Bacteria.Proteobacteria.Betaproteobacteria.Burkholderiales
* * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Proteobacteria.Betaproteobacteria
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Proteobacteria
* * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Roseburia
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae
* * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Butyricicoccus
* * * * * * * * * * Bacteria.Proteobacteria.Betaproteobacteria.Burkholderiales.Alcaligenaceae.Parasutterella
* * Bacteria.Proteobacteria.Betaproteobacteria.Burkholderiales.Alcaligenaceae
* * Bacteria.Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Faecalibacterium
* * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Coprococcus
* * * * * * * Bacteria.Firmicutes
* * Bacteria.Firmicutes.Clostridia.Clostridiales
* * * * Bacteria.Firmicutes.Clostridia
* * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Phascolarctobacterium
* * * * * * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Butyricimonas
* * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Barnesiella
Bacteria.Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Veillonella
* * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Dorea
* * * * * * * * * Bacteria.Firmicutes.Erysipelotrichi.Erysipelotrichales
* * * * * * * * * Bacteria.Firmicutes.Erysipelotrichi
* * * * * * * * * Bacteria.Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae
* * * * * * * * * * * * * Bacteria.Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae.Coprobacillus
* * * * * * * * * Bacteria.Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae.Holdemania
* * * * * Bacteria.Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae.Turicibacter
* * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Peptostreptococcaceae
Bacteria.Firmicutes.Clostridia.Clostridiales.Clostridiaceae.Clostridium
Bacteria.Firmicutes.Clostridia.Clostridiales.Clostridiaceae
* * Bacteria.Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XIV.Blautia
* * Bacteria.Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XIV
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidaceae.Bacteroides
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidaceae
* * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales
* * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia
* * * * * * * * * Bacteria.Bacteroidetes
* * * * * * * * * Bacteria.Proteobacteria.Gammaproteobacteria
* * Bacteria.Actinobacteria.Actinobacteria.Coriobacteriales.Coriobacteriaceae.Collinsella
* * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Rikenellaceae.Alistipes
* * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Rikenellaceae
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Anaerotruncus
* * * * * Bacteria.Actinobacteria.Actinobacteria.Coriobacteriales.Coriobacteriaceae
* * * * * Bacteria.Actinobacteria.Actinobacteria.Coriobacteriales
* * * * * * * * * * * * * Bacteria.Actinobacteria.Actinobacteria
* * * * * * * * * * * * * Bacteria.Actinobacteria
* Bacteria.Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XI
* * * * * * * * Bacteria.TM7.TM7_genera_incertae_sedis
* * * * * * * * Bacteria.TM7
* * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Acidaminococcus
* * * * * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Megasphaera
* * * * * * * * * * * * * * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Odoribacter
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Parabacteroides
* * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XIII.Anaerovorax
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Actinobacteria.Actinobacteria.Actinomycetales
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Actinobacteria.Actinobacteria.Actinomycetales.Actinomyces
* * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Fusobacteria.Fusobacteria.Fusobacteriales
* * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Fusobacteria.Fusobacteria
* * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Fusobacteria
* * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Dialister
Bacteria.Firmicutes.Bacilli.Lactobacillales.Lactobacillaceae.Lactobacillus
Bacteria.Firmicutes.Bacilli.Lactobacillales.Lactobacillaceae
* * * * * * * Bacteria.Firmicutes.Bacilli.Lactobacillales
* * * * * * * * * * Bacteria.Firmicutes.Bacilli
* * * * * * * * Bacteria.Firmicutes.Bacilli.Lactobacillales.Streptococcaceae
* * * * * * * * * Bacteria.Firmicutes.Bacilli.Lactobacillales.Streptococcaceae.Streptococcus
* Bacteria.Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Megamonas
* * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Ruminococcus
* Bacteria.Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Subdoligranulum
* * * * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Oscillibacter
* * * * * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XIII
* * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Ruminococcaceae
* * * * * * * * Bacteria.Proteobacteria.Betaproteobacteria.Burkholderiales.Alcaligenaceae.Sutterella
* * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Veillonellaceae
* * Bacteria.Firmicutes.Clostridia.Clostridiales.Eubacteriaceae.Eubacterium
* * * * * * * * * * Bacteria.Firmicutes.Clostridia.Clostridiales.Eubacteriaceae
Bacteria.Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae.Catenibacterium
* * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Prevotellaceae.Paraprevotella
* * * * * * * * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Prevotellaceae.Prevotella
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Bacteria.Bacteroidetes.Bacteroidia.Bacteroidales.Prevotellaceae

Total Alpha Tocopherol Equivalents R

Proanthocyanidin, dimers F

Daidzein R
Natural Folate food folate R

Catechin, flavan−3−ol F

Alcohol F

Genistein R

Glycitein R

Pinitol R

Phenylalanine, Aspartame F

Cyanidin, anthocyanidin F

Pelargonidin, anthocyanidin F

Proanthocyanidin, 4−6mers F

Proanthocyanidin, trimers F

Proanthocyanidin, polymers F

Proanthocyanidin, 7−10mers F

Folate Equivalents, suppl. & fort. foods F

Vitamin E w/o vit. suppl. F

Acesulfame Potassium R

Natural Alpha Tocopherol R

Polyunsaturated to Saturated Fat Ratio R

Mannitol R

Beta Tocopherol R

PUFA linoleic acid R

Retinol Equivalents of Vit A F

Retinol F

Total Folate post 1998 F

Eicosenoic fatty acid F

Gamma Linolenic fatty acid F

Gamma linolenic fatty acid F

Glycemic Load bread reference R

Total Dietary Fiber R

AOAC fiber F

Free Choline, choline−contrib. metabolite F

Natural Food Folate F

Synthetic Alpha Tocopherol R

Pantothenic Acid R

Total Vitamin A Activity I R

Total Vitamin A Activity II R

Pantothenic Acid w/o suppl. F

Beta Carotene R

Beta Carotene Equivalents R

Total Vitamin A Activity IU R

Vitamin K R

Lutein Zeaxanthin R

Alpha Carotene R

Aspartic Acid, Aspartame F

Aspartamefort. foods F

Caffeine F

Caffeine R

Peonidin, anthocyanidin F

Malvidin, anthocyanidin F

Petunidin, anthocyanidin F

Total anthocyanidins F

Delphinidin, anthocyanidin F
Naringenin, flavanone F

Riboflavin B2 w/o vit. pills F

Erythritol R

PUFA parinaric acid R

Glycemic Load glucose reference R

Energy kcal R

Vitamin E, Food Fortification F

Eriodictyol, flavonone F

PUFA EPA R

Total Choline, no betaine F

Palmitoleic fatty acid F

Choline, Sphingomyelin F

Total Monounsaturated Fatty Acids MUFA R

Total Fat R
Insoluble Dietary Fiber R

Pectins R

Magnesium R

Potassium R

Oxalic Acid R

Vegetable Protein R
Copper R

Manganese R

Phytic Acid R

Ash R

Magnesium F

Magnesium w/o suppl. F

Free Choline w/o suppl. F

Potassium F

Potassium w/o suppl. F

Biochanin A R

Vitamin E IU R

Vitamin E mg R

Vitamin C R

Soluble Dietary Fiber R

Added Germ from wheats F

Cholesterol R

Arginine F

Asparate F

Glycine F

Alanine F

Histidine F

Threonine F

Methionine F

Lysine F

Leucine F

Valine F

Tyrosine F

Isoleucine F

Protein F

Phenylalanine F

Serine F

Glutamate F

Proline F

Phospherous F

Phosphorous w/o suppl. F

Choline, Phosphocholine F

Phosphorus R

Calcium R

Dairy Protein F

Dairy Calcium F

MUFA oleic acid R

SFA caprylic acid R

SFA caproic acid R

Total Conjugated Linoleic Acid CLA 18 2 R

SFA stearic acid R

SFA palmitic acid R

Percent Calories from SFA R

Total Saturated Fatty Acids SFA R

Palmitelaidic trans fatty acid F

Tryptophan F

Choline w/o suppl. F

Sum of Betaine & Choline F

Choline, Phosphatidylcholine F

Choline, Phosphatidylcholine w/o suppl. F

Taurine F

Cholesterol F

Animal Protein F

Hydroxyproline F

Sucralose R

Galactose R

Vitamin D w/o vit. pills F

Choline, Glycerophosphocholine F

Cystine F

Calcium F

Calcium w/o vit. pills F

SFA capric acid R

SFA myristic acid R

SFA butyric acid R

CLA trans 10 cis 12 R

CLA cis 9 trans 11 R

Sodium F

Animal fat F

Palmitic fatty acid F

Total Trans Fatty Acids R

Stearic fatty acid F

c9,t11 conjug diene isomer 18:2 Linoleic F

Saturated fat F

Total Trans/Cis Trans Linoleic F

Trans Oleic fatty acid F

TRANS 18 1 R

TRANS 18 2 R

Dihydrophylloquinone Vitamin K1 F

Total Trans F

Total Carbohydrate R

Sucrose R

Total Sugars R

Total Sugars F
Added Sugars R

Glycemic Index F

Maltose F

Fructose F

Glucose F

Carbohydrates F

Sucrose F
Average Z Score

−0.4 −0.2 0 0.2 0.4

Value

B * *
*
*
* *
Actinobacteria
Actinobacteria.Actinobacteria
Actinobacteria.Actinobacteria.Actinomycetales
Actinobacteria.Actinobacteria.Coriobacteriales

* * * * Actinobacteria.Actinobacteria.Actinomycetales.Actinomyces
Actinobacteria.Actinobacteria.Coriobacteriales.Coriobacteriaceae
Actinobacteria.Actinobacteria.Coriobacteriales.Coriobacteriaceae.Collinsella

* * * * * * * * * Proteobacteria

* * * * * Proteobacteria.Betaproteobacteria
Proteobacteria.Gammaproteobacteria

* * Proteobacteria.Betaproteobacteria.Burkholderiales
Proteobacteria.Betaproteobacteria.Burkholderiales.Alcaligenaceae
Proteobacteria.Betaproteobacteria.Burkholderiales.Alcaligenaceae.Parasutterella
Proteobacteria.Betaproteobacteria.Burkholderiales.Alcaligenaceae.Sutterella

* * * * * Bacteroidetes

* * * * * * Bacteroidetes.Bacteroidia

* * * * * * Bacteroidetes.Bacteroidia.Bacteroidales

* * * * * Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidaceae

* * * * * * * * * * Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae

* * * Bacteroidetes.Bacteroidia.Bacteroidales.Prevotellaceae

* Bacteroidetes.Bacteroidia.Bacteroidales.Rikenellaceae

* * * * * Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidaceae.Bacteroides

* * Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Barnesiella

* Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Butyricimonas

* * * * * Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Odoribacter

* * * * * * * * * * * Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Parabacteroides
Bacteroidetes.Bacteroidia.Bacteroidales.Prevotellaceae.Paraprevotella

* * Bacteroidetes.Bacteroidia.Bacteroidales.Prevotellaceae.Prevotella

* Bacteroidetes.Bacteroidia.Bacteroidales.Rikenellaceae.Alistipes

* * * * Firmicutes
Firmicutes.Bacilli

* * * Firmicutes.Clostridia

* * * Firmicutes.Erysipelotrichi
Firmicutes.Bacilli.Lactobacillales

* * Firmicutes.Clostridia.Clostridiales

* * * Firmicutes.Erysipelotrichi.Erysipelotrichales
Firmicutes.Bacilli.Lactobacillales.Lactobacillaceae
Firmicutes.Bacilli.Lactobacillales.Streptococcaceae
Firmicutes.Clostridia.Clostridiales.Clostridiaceae

* * Firmicutes.Clostridia.Clostridiales.Eubacteriaceae
Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XI
Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XIII
Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XIV

* * * * * Firmicutes.Clostridia.Clostridiales.Lachnospiraceae
Firmicutes.Clostridia.Clostridiales.Peptostreptococcaceae
Firmicutes.Clostridia.Clostridiales.Ruminococcaceae
Firmicutes.Clostridia.Clostridiales.Veillonellaceae

* * * Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae
Firmicutes.Bacilli.Lactobacillales.Lactobacillaceae.Lactobacillus
Firmicutes.Bacilli.Lactobacillales.Streptococcaceae.Streptococcus
Firmicutes.Clostridia.Clostridiales.Clostridiaceae.Clostridium
Firmicutes.Clostridia.Clostridiales.Eubacteriaceae.Eubacterium

* * * Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XIII.Anaerovorax
Firmicutes.Clostridia.Clostridiales.Incertae_Sedis_XIV.Blautia

* * Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Coprococcus
Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Dorea

* * * Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Roseburia

* * * Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Anaerotruncus

* * * * * * * * * Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Butyricicoccus
Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Faecalibacterium

* Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Oscillibacter

* * Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Ruminococcus

* Firmicutes.Clostridia.Clostridiales.Ruminococcaceae.Subdoligranulum

* * * * Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Acidaminococcus
Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Dialister
Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Megamonas

* Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Megasphaera

* * * * * Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Phascolarctobacterium
Firmicutes.Clostridia.Clostridiales.Veillonellaceae.Veillonella
Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae.Catenibacterium

* * Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae.Coprobacillus

* * * Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae.Holdemania
Firmicutes.Erysipelotrichi.Erysipelotrichales.Erysipelotrichaceae.Turicibacter
SFA myristic acid

CLA trans 10 cis 12

Total Trans Fatty Acids

TRANS 18 1

TRANS 18 2

SFA stearic acid

SFA palmitic acid

Percent Calories from PUFA

Percent Calories from SFA

Pectins

Total Saturated Fatty Acids SFA

BMI

Insoluble Dietary Fiber

Soluble Dietary Fiber
Total Dietary Fiber

Color Key

−0.4 −0.2 0 0.2 0.4

Value
 Fig. S2

A Clustering based on Weighted UniFrac distance (Lane mask) B Clustering based on Euclidean distance (Lane mask) C Clustering based on Bray−Curtis distance (Lane mask)
PAM clustering assessment PAM clustering assessment PAM clustering assessment

0.6
Average silhouette width

Average silhouette width

Average silhouette width

0.4
0.30

0.5
●
● ● ●
● ●
● ● ●
● ● ●
● ●
● ●●● ● ●
● ● ● ● ●

0.3
● ● ● ● ● ●
● ● ●

0.4
●
● ● ●● ●
● ● ●●
●
● ● ● ●● ●
● ● ●
● ● ● ● ● ●
0.20

● ●
2
● ● ● ● ● ●
● ●
● ● ● ● ● ● ● ●
● ● ● ●

PC2
● ●● ● ● ●
1
● ● ● ● ● ●
●● ● ●●

PC2
2
●●
●

PC2
● ●●

0.3
● ● ●●● ●●●
● ● ● ● ● ●
● ●●●
● ● ● ● ● ●

0.2
● ● ● ●
1
● ● ● ● ●●
●●● ● ●●
●
●
●
● ●
● ●
●●● ●●
●●● ● ● ●● ●
● ● ● ●

1
●
2
● ● ●●
●●●● ●
● ● ● ●● ●
● ● ● ●
●●

0.2
●
●
● ● ● ●● ● ● ●● ● ● ●●
● ●
0.10

●●● ● ●●●● ● ●●
● ● ●● ● ● ● ● ●●
● ● ● ● ●●●
●
● ● ●
●
●●●● ● ●
●
● ●● ● ●● ● ●● ●
● ●● ● ●

0.1
●
●●
●● ● ● ● ●
●
●●

0.1
● ● ● ●
●● ● ● ● ● ●
● ●
●
●
0.00

0.0

0.0
best 5 10 15 20 best 5 10 15 20 best 5 10 15 20
2 k (# clusters) 2 k (# clusters) 2 k (# clusters)
PC1 PC1 PC1

Bacteroides Prevotella Ruminococcus Bacteroides Prevotella Ruminococcus Bacteroides Prevotella Ruminococcus

0.7

0.7

0.7
0.8

0.8

0.8
0.06

0.06

0.06
0.6

0.6

0.6
0.6

0.5

0.6

0.5

0.6

0.5
Proportion

Proportion

Proportion

Proportion

Proportion

Proportion

Proportion

Proportion

Proportion
0.04

0.04

0.04
0.4

0.4

0.4
0.4

0.4

0.4
0.3

0.3

0.3
0.02

0.02

0.02
0.2

0.2

0.2
0.2

0.2

0.2
0.1

0.1

0.1
0.00

0.00
0.0

0.00
0.0

0.0
0.0

0.0

0.0
1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype

D Clustering based on Weighted UniFrac distance (No lane mask)

E Clustering based on Euclidean distance (No lane mask) F Clustering based on Bray−Curtis distance (No lane mask) G ClusterLQJEDVHGRQ-HQVHQí6KDQQRQGLVWDQFH1RODQHPDVN
PAM clustering assessment PAM clustering assessment PAM clustering assessment PAM clustering assessment
0.20

0.6
Average silhouette width

0.4
Average silhouette width

Average silhouette width

Average silhouette width

0.5
ƽ

0.5
● ● ●
●
0.15

● ● ƽ
● ●

0.4
● ● ƽ

0.3
● ● ●

2
● ● ● ƽ
● ● ● ●
●
● ● ● ● ●
● ● ●

0.4
● ●● ● ● ● ƽ ƽ
● ●● ƽ ƽ
●
2
● ● ● ●
●● ƽ
● ● ● ●● ● ƽ ƽ
ƽƽ
●

0.3
● ●
●●
●●● ● ● ƽ ƽ ƽ
ƽ ƽ ƽ ƽ
ƽ
● ●
●● ●● ƽ ƽ
PC2

● ● ● ●●
2
0.10

ƽ ƽ ƽ
● ● ●● ● ● ƽ ƽ

PC2
● ● ƽ

PC2
● ƽ ƽ ƽƽ ƽ
● ● ●● ●

0.3
0.2
●
● ƽ ƽ

PC2
●● ● ● ƽ ƽ ƽ ƽ
1
● ƽ ƽƽ
2
●●
●● ● ƽ ƽƽ ƽ
● ●●
●
1●
● ● ƽƽƽƽ
ƽ ƽ
ƽ ƽ
●
●●● ● ● ●●● ● ƽ ƽ
ƽ ƽ ƽ
ƽƽƽ ƽ ƽ ƽ
1
● ● ƽ
● ƽ ƽƽ

0.2
● ● ƽ ƽ
● ●
●●● ● ● ● ƽ ƽƽ ƽ
● ● ● ● ● ● ƽ ƽ
●
●
●
●●
● ƽ ƽ ƽƽƽ ƽ
● ● ●● ● ƽ ƽ ƽ
● ● ● ƽ ƽ

0.2
● ●
●● ● ● ●●
●● ● ƽ
● ●
●
● ● ●●●
●
● ● ●● ● ● ●
● ● ● ƽ ƽ
●●●●
0.05

● ● ● ƽ ƽ
● ● ● ● ●
● ● ƽ ƽ
3●
● ● ● ●
●● ● ●

0.1
● ● ● ● ●● ● ƽ
●●
● ● ●●● ● ● ● ●
0.1

1
● ● ● ●
● ●● ●● ●
●●●● ● ● ● ●
● ● ● ●●
● ● ● ●
● ● ● ● ●
●
● ● ● ● ƽ
● ●● ●● ● ●● ● ● ● ● ●
● ● ● ● ● ●

0.1
● ● ● ● ●●
● ● ● ● ●
● ●
●
0.0
0.00

0.0
ƽ

0.0
best 5 10 15 20
best 5 10 15 20 best best
5 10 15 20 5 10 15 20
3 k (# clusters) 2 k (# clusters) 2 2
PC1 PC1 k (# clusters) k (# clusters)
PC1 PC1

Bacteroides Prevotella Ruminococcus Bacteroides Prevotella Ruminococcus Bacteroides Prevotella Ruminococcus Bacteroides Prevotella Ruminococcus

0.7
0.7

0.7

0.7
0.8
0.8

0.8

0.8
0.6
0.6

0.6

0.6
0.06
0.06

0.06

0.06
0.5
0.6
0.5

0.5

0.5
0.6

0.6

0.6
Proportion

Proportion

Proportion
Proportion

Proportion

Proportion

Proportion

Proportion

Proportion

Proportion

Proportion

Proportion
0.4
0.4

0.04

0.4

0.4
0.04

0.04

0.04
0.4
0.4

0.3

0.4

0.4
0.3

0.3

0.3
0.2

0.02
0.2

0.02

0.2

0.02

0.2

0.02
0.2
0.2

0.2

0.2
0.1
0.1

0.1

0.1
0.00
0.0
0.00

0.00
0.0

0.00
0.0

0.0

0.0
0.0

0.0

0.0
1 2 3 1 2 3 1 2 3 1 2 1 2 1 2
1 2 1 2 1 2 1 2 1 2 1 2
Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype Enterotype
Fig. S3

A Relative Absolute
B average %
0.01
Bacteroidetes.Alistipes Bacteroidetes.Alistipes 0.02
0.04
0.08
0.16
0.32
Bacteroidetes.Bacteroides Bacteroidetes.Bacteroides
0.64
1.00
2.00
4.00
Bacteroidetes.Parabacteroides Bacteroidetes.Parabacteroides 8.00
10.0
20.0
30.0
40.0
Bacteroidetes.Paraprevotella Bacteroidetes.Paraprevotella 50.0
60.0
70.0
80.0
Bacteroidetes.Prevotella Bacteroidetes.Prevotella

Firmicutes.Catenibacterium Firmicutes.Catenibacterium

Average Z Score
Bacteroides
Enterotype

Enterotype

Bacteroides
Enterotype

Enterotype
Prevotella

Prevotella

−0.5 0 0.5
Value
Fig. S4
A 1.0
B B-2004-08-S1
B-2006-09-S1
B-2009-10-S1
0.8 B-2016-08-S1
B-2019-10-S1
Bacteria Phylum
Actinobacteria B-2005-10-S1
Archaea
Proportion

0.6 Bacteroidetes B-2008-10-S1

Cyanobacteria B-2011-10-S1
Deferribacteres
Firmicutes B-2012-10-S1
Fusobacteria
Lentisphaerae B-2020-10-S1
0.4 Proteobacteria
TM7

Bacterial secretion system, p = 0.010

Regulation of actin cytoskeleton, p = 0.017

Protein export, p = 0.022

Lipoic acid metabolism, p = 0.045

Fc gamma R mediated phagocytosis, p = 0.048

Endocytosis, p = 0.079

Styrene degradation, p = 0.086

Hedgehog signaling pathway, p = 0.094

Phosphatidylinositol signaling system, p = 0.008

Vibrio cholerae infection, p = 0.056

Steroid biosynthesis, p = 0.077

Glycerophospholipid metabolism, p = 0.094

Arachidonic acid metabolism, p = 0.109

Lysine biosynthesis, p = 0.112

Unclassified
Unclassified bacteria
Verrucomicrobia
0.2
3

2
0.0

Z−score
1
16S Tag Whole Genome Shotgun
0

−1

C High.Fat, 2004 High.Fat, 2006 High.Fat, 2009 High.Fat, 2016 High.Fat, 2019
100
80
60
40
Assignment
20
% Abundance

Archaea
0
Bacteria
Low.Fat, 2005 Low.Fat, 2008 Low.Fat, 2011 Low.Fat, 2012 Low.Fat, 2020
100 Eukaryota

80 Virus

60 Ambiguous

40
20
0
Begin End Begin End Begin End Begin End Begin End
 Fig. S5

●
●
Log Prevotella/Bacteroides Ratio
2

● ●● ●
●●
●●

●
0

●●
●●
●
●
−2

● ● ● ●

● ●
●
−4

●
●
●
−6

−0.4 −0.2 0.0 0.2

PC 1