Open Access Highly Accessed Research

Dysfunction of the intestinal microbiome in inflammatory bowel disease and treatment

Xochitl C Morgan1, Timothy L Tickle12, Harry Sokol345, Dirk Gevers2, Kathryn L Devaney34, Doyle V Ward2, Joshua A Reyes1, Samir A Shah6, Neal LeLeiko6, Scott B Snapper7, Athos Bousvaros7, Joshua Korzenik37, Bruce E Sands8, Ramnik J Xavier234 and Curtis Huttenhower12*

Author Affiliations

1 Department of Biostatistics, Harvard School of Public Health, Boston, MA 02115, USA

2 Broad Institute of Massachusetts Institute of Technology and Harvard University, Cambridge, MA 02142, USA

3 Gastrointestinal Unit and Center for the Study of Inflammatory Bowel Disease, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA

4 Center for Computational and Integrative Biology, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA

5 Current address: Department of Gastroenterology, AP-HP, Hôpital Saint-Antoine and UPMC University of Paris, Paris, 75012, France

6 Division of Pediatric Gastroenterology, Hasbro Children's Hospital, The Warren Alpert School of Medicine at Brown University, Providence, RI 02903, USA

7 Gastrointestinal Unit, Children's Hospital and Brigham and Women's Hospital, Harvard Medical School, Boston, MA 02115, USA

8 Department of Gastroenterology, Mount Sinai School of Medicine, New York, NY 10029, USA

For all author emails, please log on.

Genome Biology 2012, 13:R79  doi:10.1186/gb-2012-13-9-r79

Published: 26 September 2012

Additional files

Additional file 1:

Taxa significantly associated with IBD status or subject metadata using a boosted general linear model. A multivariate analysis was performed to associate each microbial clade with a sparse selection of disease status and clinical metadata (selected through boosting; see Materials and methods). All clades and metadata in these associations are given with nominal P-values from the multivariate linear model and with Benjamini and Hochberg (BH) corrected false discovery rate (q-values) up to a threshold of 0.25. In this and all other supplemental tables, blank spaces indicate values that were not significant but are shown for comparison with related significant data.

Format: XLS Size: 49KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 2:

Effects of biogeography on gut microbiome composition differentiates stool and biopsy communities. The composition of phyla stratified by biopsy location or fecal sample origin mainly differentiates stool and biopsy communities. Sample count per location is indicated in parentheses. Biopsy locations (above) do not substantially differ in composition, while biopsies compared to stool (below) differ significantly in all phyla.

Format: PDF Size: 19KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 3:

Univariate analysis of associations between microbial composition and biopsy location. A univariate analysis for associations between taxa and biopsy sites was conducted using LEfSe [102] considering the six regions annotated for these samples: 1) terminal ileum (TI), 2) cecum, 3) left colon, 4) transverse colon, 5) right colon, and 6) sigmoid colon and rectum. (a) Relatively few clades were strongly associated with biopsy locations, and these tended to mirror expected intestinal pH and the clades described here as particularly affected by disease-linked inflammation. (b-g) Abundant major clades, including the Firmicutes (b), showed extremely modest variations with intestinal region, driven by specific members depleted in low-pH regions, including Roseburia (c) (high in the left and sigmoid colon), Ruminococcaceae (d), and to a lesser degree Alistipes (e). Clades enriched in low pH regions included Fusobacterium (f) (high in TI and right colon) and Enterobacteriales (g) (particularly in TI).

Format: PDF Size: 2.9MB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 4:

Locations of patient biopsies. Distribution of biopsy samples available for this study as classified by the OSCCAR and PRISM cohort collection protocol.

Format: DOCX Size: 12KB Download file

Open Data

Additional file 5:

Univariate analysis of associations between microbial composition and gender. A univariate test for associations of subject gender with microbial clades was conducted using LEfSe [102], resulting in few and weak associations concordant with previous studies [5]. Here, Clostridium and the Streptococcaceae were weakly associated with gender at P < 0.05, but did not remain significant at P < 0.1.

Format: PDF Size: 1.4MB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 6:

Bifidobacterium genus abundance decreases significantly with age. The association of Bifidobacterium abundance with disease status and clinical metadata (including age) was determined to be significant in these data using a sparse general linear model. Clade abundances were transformed with the arcsine square-root transformation for proportional data (y-axis). Size of effect, standard deviation, P-value (p) and Benjamini and Hochberg false discovery rate (q) are shown in parentheses, and the line of best fit in green.

Format: PDF Size: 90KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 7:

Escherichia/Shigella abundance is significantly decreased in mesalamine-treated subjects. The association of these genera (indistinguishable by 16S rRNA gene sequencing) with disease status and clinical metadata (including mesalamine treatment) was determined to be significant using a sparse general linear model (see Materials and methods). Clade abundances were transformed with the arcsine square root transformation for proportional data and are plotted along the y-axis as two notched box plots (samples without and with mesalamine use). Size of effect, standard deviation, P-value (p) and q-value (q) are shown in parentheses.

Format: PDF Size: 94KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 8:

Stratification of clades associated with IBD status by sample biogeography. Fifteen microbial clades were significantly associated specifically with IBD status (q < 0.25) using a multivariate linear model incorporating clinical metadata (see Materials and methods). Although this model putatively asserts that this association holds regardless of sample origin (biopsy or stool), we verified this by stratifying each clade's abundance by sample type, stool (1) or biopsy (0). Green coloring indicates that a clade's abundance was significantly reduced in IBD using the full model, red increased. These trends are uniformly preserved after explicit stratification by stool versus biopsy sample origins.

Format: PDF Size: 131KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 9:

Univariate associations of microbial composition with biopsy location. Results of a LEfSe analysis of the six location categories available for biopsies in this study, excluding two anatostamosis samples.

Format: XLSX Size: 10KB Download file

Open Data

Additional file 10:

Covariation of microbial community function in IBD with treatment, environment, biometrics, and disease subtype. Fecal and biopsy samples from 231 IBD patients and healthy controls are plotted as squares (iCD) or circles and colored by disease status. Axes show the first two components of overall variation as determined by multiple factor analysis (see Materials and methods). Clinical and environmental covariates are shown in bold, while individual microbial functions (Gene Ontology terms) are italicized. Covariation patterns are similar to those determined using microbial abundance (Figure 1).

Format: PDF Size: 57KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 11:

KEGG pathways significantly associated with IBD status or subject metadata using a boosted general linear model. A multivariate analysis was performed to associate each pathway with a sparse selection of disease status and clinical metadata (selected through boosting; see Materials and methods). All pathways and metadata in these associations are given with nominal P-values from the multivariate linear model and with Benjamini and Hochberg (BH) corrected false discovery rate (q-values) up to a threshold of 0.25.

Format: XLS Size: 59KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 12:

KEGG metabolic modules significantly associated with IBD status or subject metadata using a boosted general linear model. A multivariate analysis was performed to associate each metabolic module with a sparse selection of disease status and clinical metadata (selected through boosting; see Materials and methods). Each module and metadata in these associations is given with nominal p-values from the multivariate linear model and with Benjamini and Hochberg (BH) corrected false discovery rate (q-values) up to a threshold of 0.25.

Format: XLS Size: 70KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 13:

Gene Ontology terms significantly associated with IBD status or subject metadata using a boosted general linear model. A multivariate analysis was performed to associate each Gene Ontology term with a sparse selection of disease status and clinical metadata (selected through boosting; see Materials and methods). Each term and metadata in these associations is given with nominal P-values from the multivariate linear model and with Benjamini and Hochberg (BH) corrected false discovery rate (q-values) up to a threshold of 0.25.

Format: XLS Size: 163KB Download file

This file can be viewed with: Microsoft Excel Viewer

Open Data

Additional file 14:

Shotgun metagenomic sequencing validates predicted microbial metabolic trends in a subset of healthy and CD microbiomes. A subset of 11 stool samples for which microbial DNA was available were subjected to shallow metagenomic sequencing using the MiSeq platform (150-nucleotide paired-end reads) averaging 119 meganucleotides per sample. (a) Of the seven microbial metabolic modules highlighted in Figure 5, six retained the same over- or under-abundance trend predicted from 16S sequencing in this subset, with the seventh (cobalamin biosynthesis) falling below the limit of detection. (b) Six additional metabolic modules of interest with significant differences in the full IBD dataset retained the trend expected with CD in this subset, including depletion of glycolysis processes and enrichment for bacterial secretion systems.

Format: PDF Size: 61KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional file 15:

Correlation of microbial gene families estimated from 16S gene pyrosequencing and whole-genome shotgun sequencing data. Ancestral state reconstruction was used to infer metagenomes using 16S gene pyrosequencing of samples from multiple body sites from the Human Microbiome Project (see Materials and methods). The relative abundance of KOs inferred from 16S sequencing and measured from paired whole-community genome sequencing samples were correlated (Spearman rank correlation) and plotted per body site. Each box plot shows the distribution of the correlation of relative KO abundance from 16S and whole-genome sequencing; specific sample-pair correlations are plotted as dots. Median correlation for Human Microbiome Project stool samples is 0.75 for an average n = 75 per body site. As each correlation is calculated over approximately 5,400 KOs, correlation values above 0.59 are significant at a Bonferroni-corrected P < 0.05.

Format: PDF Size: 85KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data