Rooks et al. 2013
1
SUPPLEMENTARY FIGURE LEGENDS
2
Figure S1. 16S rRNA analysis pipeline for investigating the influence of perturbations
3
on gut-associated microbial communities. Computational analysis pipeline used to
4
characterize 16S rRNA gene sequences of gut-associated microbial communities. See
5
Supplementary Methods for details of the computational tools used in this study. QIIME =
6
Quantitative Insights Into Microbial Ecology; PICRUSt = Phylogenetic Investigation of
7
Communities by Reconstruction of Unobserved States; HUMAnN = The Human Microbiome
8
Project Unified Metabolic Analysis Network; LEfSe = Linear Discriminant Analysis with Effect
9
Size.
10
11
Figure S2. PCoA of weighted Unifrac distances of gut microbial communities (related
12
to Figure 1c). PCoA plots of the weighted UniFrac distances of gut microbial communities
13
from stool collected at baseline (pre-intervention) and upon treatment completion (post-
14
intervention). The first two principal coordinates (PC) from the PCoA are plotted. Symbols
15
represent data from individual mice, color-coded by the indicated metadata.
16
17
Figure S3. PCoA and LEfSe analyses to assess the influence of early-life exposures
18
on antibiotic-driven microbial clade responses (related to Figure 2b-c). a) PCoA plots
19
of the weighted UniFrac distances of gut microbial communities from stool collected upon
20
completion of treatment with metronidazole (n=10) or vancomycin (n=10). The first two PCs
21
from the PCoA are plotted. Symbols represent data from individual mice, color-coded by the
22
indicated metadata. For caging, M = metronidazole-treated and V = vancomycin-treated. b)
23
Cladogram and corresponding histogram of the LDA scores for differentially abundant
24
microbial clades in stool from progeny of the breeding pairs (BP) indicated in Figure 2c prior
25
to metronidazole treatment. c) Cladogram and corresponding histogram of the LDA scores
26
for differentially abundant microbial clades in stool from progeny of the breeding pairs
27
indicated in Figure 2c prior to vancomycin treatment. For cladograms, white circles delineate
28
non-significant clades.
1
Rooks et al. 2013
29
30
Figure S4. LEfSe rank plots of differentially abundant microbial clades in gut
31
microbiomes associated with active colitis versus remission following antibiotic
32
treatment (related to Figure 2d). a) LDA scores for differentially abundant microbial clades
33
in stool from mice treated with gentamicin (gent; n=10), metronidazole (metro; n=10), or
34
vancomycin (vanco; n=10) using an all-against-all multi-class comparison. b) LDA scores for
35
differentially abundant microbial clades in stool from mice in remission that received
36
gentamicin or metronidazole (n=20) and mice with active colitis that were untreated (sham)
37
or received vancomycin (n=22).
38
39
Figure S5. LEfSe rank plots of differentially abundant microbial clades in gut
40
microbiomes exposed to immunomodulatories (related to Figure 3b-c). LDA scores for
41
differentially abundant microbial clades in stool from: a) immunomodulatory-treated (anti-
42
TNF-α or TRegs; n=20) versus sham (untreated; n=12) mice, and b) anti-TNF-α (n=10) versus
43
TReg (n=10) treated mice.
44
45
Figure S6. LEfSe rank plots of differentially abundant microbial clades in gut
46
microbiomes following a daily dietary intervention (related to Figure 4a-b). LDA scores
47
for differentially abundant microbial clades in stool from mice: a) before and after
48
administration of a fermented milk product (n=10); b) before and after administration of a
49
non-fermented milk control (n=10); and c) after administration of a fermented milk product
50
(n=10) versus a non-fermented milk control (n=10).
51
52
Figure S7. PCoA of weighted Unifrac distances of stool and mesenteric lymph node
53
(MLN) microbial communities of mice administered a daily dietary intervention
54
(related to Figure 4c). PCoA plots of the weighted UniFrac distances of microbial
55
communities from stool and MLNs collected upon completion of a dietary intervention with a
56
fermented milk product (FMP) or non-fermented milk control (MC). The first two PCs from
2
Rooks et al. 2013
57
the PCoA are plotted. Symbols represent stool (FMP, n=10; MC, n=10) or pooled MLN
58
(FMP, n=21; MC, n=16; 5 MLNs/mouse) samples from individual mice, color-coded by the
59
indicated metadata.
60
61
Figure S8. LEfSe rank plot of differentially abundant microbial clades between stool
62
and MLN microbial communities of mice receiving a daily dietary intervention (related
63
to Figure 4e). LDA scores for differentially abundant microbial clades in stool (n=20) versus
64
MLNs (n=37 mice; 5 MLNs/mouse) of mice receiving either a fermented milk product or non-
65
fermented milk control.
66
67
Figure S9. LEfSe rank plot of differentially abundant microbial clades between MLN
68
microbial communities of mice receiving a daily dietary intervention (related to Figure
69
4f). LDA scores for differentially abundant microbial clades in MLNs from mice (5
70
MLNs/mouse) receiving a fermented milk product (n=21) versus a non-fermented milk
71
control (n=16).
72
73
Figure S10. LEfSe rank plot of differentially abundant microbial clades in gut
74
microbiomes associated with active colitis and treatment-induced remission (related
75
to Figure 5a). LDA scores for differentially abundant microbial clades in stool from mice with
76
active colitis (n=31) versus remission (n=51) upon intervention completion (see Figure 1b).
77
78
Figure S11. LEfSe rank plot of differentially abundant microbial genes in gut
79
microbiomes associated with active colitis and treatment-induced remission (related
80
to Figure 6a). LDA scores for differentially abundant PICRUSt predicted microbial genes
81
(specified as KEGG Orthology groups), pathways, and classified functional categories (bold)
82
defined by the KEGG BRITE hierarchy in stool from mice with active colitis (n=31) versus
83
remission (n=51) upon intervention completion (see Figure 1b).
84
3
Rooks et al. 2013
85
Table S1. Summary of pyrosequenced samples. This table details the number of mice in
86
each experimental group; the post-intervention histologic colitis scores (mean ± SEM); the
87
number of mice with active colitis versus remission upon treatment completion; and the
88
average number of quality-filtered 16S rRNA gene sequences and binned operational
89
taxonomic units (OTUs; selected at 97% sequence identity) for samples in each pre- or post-
90
intervention group (mean ± SEM). Colitis scores >2 indicate active colitis and scores ≤2
91
remission.
92
93
Table S2. Summary of PICRUSt analysis of post-intervention samples. Number and
94
percentage of quality-filtered 16S rRNA gene sequences that mapped to Greengenes
95
reference OTUs at 97% sequence identity; the resulting number of binned reference-based
96
OTUs; the calculated weighted NSTI score representing the metagenome prediction
97
accuracy (lower values indicate a more accurate prediction); and the number of inferred
98
genes (specified by KEGG Orthology groups) for stool samples from each post-intervention
99
group (mean ± SEM).
100
101
Table S3. Summary of corrected false discovery rate (FDR) q-values associated with
102
LEfSe identified microbial features. All microbial clades and functions identified as
103
significantly, differentially abundant with specific treatments or host disease status are given
104
with effect sizes (LDA score) and p-values from LEfSe and with Benjamini and Hochberg
105
corrected FDR q-values up to a threshold of 0.25.
4
Rooks et al. 2013
Table S1.
Control
Treatment
Sham (n=12)
Time Point
Post
Colitis Score
Antibiotics
Gentamicin (n=10)
Pre
Immunomodulatories
Metronidazole (n=10)
Post
Pre
Vancomycin (n=10)
Post
Pre
Anti-TNF-α (n=10)
Post
Pre
Dietary Interventions
T-regulatory cells (n=10)
Post
Pre
Milk Control (n=10)
Post
Pre
Post
Pre
Post
1.7 ± 1.77
3.4 ± 1.65
6.0 ± 0.43
0.0 ± 0.0
0.1 ± 0.32
5.3 ± 0.67
0.4 ± 0.52
0.5 ± 0.56
1.7 ± 1.77
Disease Status
Colitis (n=12)
Remission (n=10)
Remission (n=10)
Colitis (n=10)
Remission (n=10)
Remission (n=10)
Remission (n=4)
Colitis (n=6)
Sequences
6 172 ± 5 338
7 801 ± 6 157
7 974 ± 4 333
7 968 ± 2 424
7 672 ± 2 842
732 ± 395
702 ± 336
138 ± 20
912 ± 236
363 ± 325
OTUs
4 282 ± 4 509 8 735 ± 2 407 10 754 ± 3 944
633 ± 292
156 ± 30
1 120 ± 214
Fermented Milk (n=10)
Remission (n=7)
Colitis (n=3)
8 276 ± 6 050
5 771 ± 1 998
8 700 ± 3 864
4 194 ± 1 353
4 100 ± 1 143
3 874 ± 1 450
3 913 ± 1 196
814 ± 374
787 ± 139
939 ± 306
369 ± 78
362 ± 53
357 ± 56
365 ± 78
Table S2.
Control
Antibiotics
Immunomodulatories
Dietary Interventions
Treatment
Sham (n=12)
Gentamicin (n=10)
Metronidazole (n=10)
Vancomycin (n=10)
Anti-TNF-α (n=10)
T-regulatory cells (n=10)
Milk Control (n=10)
Fermented Milk (n=10)
Mapped Sequences
(% Total)
5 084 ± 1 344
(80.6 ± 2.2)
6 896 ± 1 207
(86.0 ± 0.7)
6 661 ± 733
(86.8 ± 2.4)
8 106 ± 718
(92.7 ± 1.4)
7 022 ± 1 639
(84.9 ± 1.2)
7 175 ± 1 055
(82.0 ± 1.6)
3 954 ± 355
(96.3 ± 0.5)
3 707 ± 365
(94.5 ± 0.8)
197 ± 19
46 ± 3
85 ± 15
43 ± 5
221 ± 22
227 ± 18
196 ± 9
193 ± 10
0.115 ± 0.004
0.071 ± 0.006
0.091 ± 0.011
0.047 ± 0.002
0.109 ± 0.007
0.115 ± 0.004
0.131 ± 0.003
0.139 ± 0.006
2 634 000 ± 641 717
2 047 000 ± 361 900
1 953 000 ± 242 140
3 243 000 ± 437 144
3 479 000 ± 873 417
3 300 000 ± 534 338
2 065 000 ± 272 946
2 218 000 ± 273 132
Reference-based OTUs
Weighted NSTI
KOs
Table S3 is provided as a separate excel sheet.
5