reSeArcH Article
epigenetic reduced representation bisulfite sequencing (RRBS), and
16S rRNA gene amplicon sequencing (16S), which were matched with
human exome sequencing, serological profiles, and RRBS from blood.
All data are available at https://ibdmdb.org/.
Multi-omic gut microbiome changes in IBD
Consistent with prior studies^4 ,^5 , although subsets of IBD (CD in par-
ticular) contributed to the second axis of taxonomy-based principal
coordinates (Fig. 1d, Extended Data Fig. 2a), inter-individual variation
accounted for the majority of variance for all measurement types^5 ,^9 ,^10
(Fig. 1e, f, Extended Data Fig. 2a). Even relatively large effects, such
as disease status or physiological and technical factors, explained a
smaller proportion of variation (Fig. 1f); this was true across measure-
ment types, although these captured distinct aspects of IBD dysbiosis
(see below).
Most measurement types captured correlated changes among
and within subjects, cross-sectionally and longitudinally (Fig. 1e).
Functional profiles, measured from MGX, MTX, and MPX, were
the most tightly coupled (Fig. 1e), although some individual feature-
wise correlations were weak (Spearman’s correlation MGX–MTX
0.44 ± 0.10 (mean ± s.d.), MGX–MPX 0.14 ± 0.083, and MTX–
MPX 0.18 ± 0.096l; Extended Data Fig. 2b). Unexpectedly, charac-
terized enzymes tended to be only weakly correlated with their known
substrates or products (Supplementary Fig. 1). Although our dietary
characterization was obtained through a very broad-level food fre-
quency questionnaire, it provides an initial characterization of longitu-
dinal diet–microbiome coupling in a substantial population over many
months; diet accounted for a small but significant 3% (false discovery
rate (FDR) P = 7.4 × 10 −^4 ) of taxonomic variation between subjects,
and 0.7% (FDR P = 4.3 × 10 −^4 ) of variation longitudinally.
Simple cross-sectional differences between individuals with IBD and
those without (Supplementary Tables 1–14) were most apparent in the
metabolome (Figs. 1f, 2a, Extended Data Fig. 2a, c, d, see Methods).
Overall, metabolite pools were less diverse in individuals with IBD, par-
alleling previous observations for microbial diversity (Supplementary
Table 2); this might be caused by poor nutrient absorption, greater
water or blood content in the bowels, and shorter bowel transit times
in individuals with active IBD^11. The smaller number of compounds
that were more abundant in patients with IBD included polyunsatu-
rated fatty acids such as adrenate and arachidonate. Pantothenate and
nicotinate (vitamins B5 and B3, respectively) were particularly depleted
in the gut during IBD; this is notable because these are not typically
among the B vitamins that are deficient in the serum of patients with
IBD^12 , although low nicotinate levels have been detected during active
CD^13. Both vitamins are required to produce cofactors used in lipid
metabolism^14 , and nicotinate has anti-inflammatory and anti-apoptoticadfbcePCo2 (6.7%)PCo1 (8.0%)Diagnosis
CDUC
Non-IBD1.0
0.5
01.0
0.5
0.0Rel. abd.
Gini–Simpson Firmicutes BacteroidetesMGH (38)Cedars-Sinai (33)Cincinnati Children’s (33)MGH Pediatrics (17)Emory (11)Total: 132(^00)
5
10
15
20
10203030506070
25
No. of subjects
Age (y)
33 34
PaediatricAdult
15
23 1314
UC
CD
Non-IBD
Population
AFR
AMR
EAS
EUR
SAS
Source
1kG
HMP2
PC2
PC1
Stool
Biopsy
Blood Genetics
Serological pro les
Bisul te sequences
Bisul te sequences
n = 1,785
n = 651
n = 529
Global samples
Sample and
measurement types
Metagenome sequences
Metabolite pro les
Virome sequences
16S rRNA gene pro les
Transcriptome sequences
Faecal calprotectin
Proteomic pro les
Metatranscriptome sequences 1,638 835
(^546450)
(^703652)
178
(^252221)
21092
228
Measurement type n
36
MetagenomicsMetatranscriptomics Proteomics Viromics16S rRNA
week
(^452)
Microbial
Host GenomicsTranscriptomics Bisul te sequencesSerological pro les Diet surveysFaecal calprotectin
02 81624
Metabolomics
**
68.6%33.1%18.1%9.8% 0.7%
51.6% 40.0%17.7%5.4% 0.5%
30.7%54.1% 7.5%6.3% 0.1%
24.7%28.3%3.9% 7.6% 0.6%
7.0%7.1%6.6%7.6% 2.5%
15.2%18.5%11.2%8.6%1.7%
0.4%0.5%1.5%0.4%1.2%2.2%
3.0%3.3%1.4%0.0%4.1%0.3%0.2%
N/AN/A
N/AN/A
N/AN/A
N/AN/A
N/AN/A
N/AN/A
N/A
Diet
Biopsy HTX
Biopsy 16S
Metabolites
KOs (protein)
KOs (RNA)
KOs (DNA)
Taxonomy
TaxonomyKOs (DNA)KOs (RNA)KOs (protein)Metabolite
s
Biopsy
(16S)
Biopsy (HTX)Diet
Intraindividual Feature
Interindividual
305
457
348
350
527
418
592
424
553
791
175
321
210
268
461
357
579
282
491
785
0250 500 750 1,000
Time points
Strictness
Strict
±2 weeks
±4 weeks
least MGX and MTXTime points with at
MGXMTXVXMPXMBXFC
1.3%
0.5%
0.6%
1.8%
1.1%
3.2%
0.6%
0.6%
1.9%
4.8%
8.0%
7.7%
3.0%
67.7%
69.0%
0.8%
0.4%
0.7%
1.7%
1.0%
3.3%
1.1%
0.5%
1.6%
3.7%
6.2%
6.9%
1.4%
57.5%
58.7%
1.1%
0.4%
0.3%
1.4%
0.6%
3.5%
0.4%
0.2%
1.2%
4.6%
4.4%
3.7%
1.5%
46.3%
47.2%
1.3%
0.3%
0.3%
1.5%
1.4%
1.4%
1.2%
0.2%
1.3%
6.2%
5.1%
7.0%
1.7%
39.0%
42.0%
1.8%
0.7%
0.7%
2.0%
0.8%
4.5%
0.4%
0.4%
3.5%
4.3%
6.6%
3.1%
0.9%
53.3%
54.9%
2.1%
1.5%
1.4%
3.7%
2.9%
7.0%
0.8%
0.6%
3.1%
11.6%
18.8%
6.0%
3.9%
30.5%
1.4%
0.4%
0.7%
0.8%
0.5%
2.8%
32.4%
3.5%
2.9%
4.5%
4.3%
1.3%
6.2%
46.7%
4.5%
1.7%
1.2%
2.3%
0.5%
9.6%
0.4%
0.2%
2.4%
6.2%
5.8%
1.4%
0.8%
66.6%
67.0%
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
**
**
**
**
**
All
Subject [88–129]
Dysbiotic (UC) [1]
Dysbiotic (CD) [1]
Disease loc. (UC) [2]
Disease loc. (CD) [3]
Disease [2]
In ammation [1]
Biopsy location [2]
Immunosupp. [1]
Antibiotics [1]
Recruitment site [3–4]
Bowel surgery [1]
Race [3]
BMI [1]
Sex [1]
Age [1]
Taxonom
y [1,595]
KOs (DNA) [1,
595]
KOs (RNA
) [818]
Proteins [450]Metabolites
[546]
Biopsy
(16S) [146]
Biopsy
(HTX) [247]
Diet [1,
584]
Per cent of variance in taxonom
y
explained by antibiotics
Feature [DOF]
0
0.25
0.50
0.75
MGH 1.00
CC
Emory CSMC
MGH Ped.
Fig. 1 | Multi-omics of the IBD microbiome in the IBDMDB study.
a, Overview of cohort characteristics. We followed 132 participants (with
CD, with UC, or without IBD (control)) for one year each. Principal
component analysis (PCA) of SNP profiles shows that the resulting
IBDMDB cohort is mostly of European ancestry as compared to the 1000
Genomes (1kG) reference (see Methods). b, Sampling strategy. The study
yielded host and microbial data from colon biopsy (baseline), blood
(approximately quarterly), and stool (every two weeks), assessing global
time points for all subjects and dense time courses for a subset. Raw, non-
quality-controlled sample counts are shown. c, Overlap of multi-omic
measurements from the same sample (strict) or from near-concordant
time points (with differences of up to 2 or 4 weeks; see Methods).
d, Principal coordinates analysis (PCoA) based on species-level Bray–
Curtis dissimilarity; most variation is driven by a tradeoff between phylum
Bacteroidetes versus Firmicutes. Samples from individuals with IBD
(CD in particular) had weakly lower Gini–Simpson alpha diversity
(Wald test P = 0.26 and 0.014 for UC and CD compared with non-IBD,
respectively). e, Mantel tests quantifying variance explained (square of
Mantel statistic) between measurement type pairs, with differences across
subjects (inter-individual) or within subjects over time (intra-individual;
see Methods); results show tight coupling across measurement types.
Sample sizes in f. f, PERMANOVA shows that inter-individual variation
is largest for all measurement types, with even relatively large effects
(for example, antibiotics or IBD phenotype) capturing less variation
(see Methods). Stratified tests (CD/UC) consider only samples within the
indicated phenotype (note that sample counts decrease for these, resulting
in larger expected covariation by chance). Stars show FDR-corrected
statistical significance (FDR P ≤ 0.05, P ≤ 0.01, ***P ≤ 0.001).
Variance is estimated for each feature independently (Methods). ‘All’ refers
to a model with all metadata. Total n for each measurement type is shown
in square brackets, distributed across up to 132 subjects (Extended Data
Fig. 1a, see Methods).
656 | NAtUre | VOl 569 | 30 MAY 2019