SUPPLEMENTARY INFORMATION FOR

Gut bacteria-Host metabolic interplay during conventionalisation of the mouse germfree colon

Sahar EL Aidy1,2 , Muriel Derrien1,2†, Claire A. Merrifield3, Mark V. Boekschoten4, Florence Levenez5,

Joël Doré5, Jan Dekker1,6, Elaine Holmes3, Sandrine P. Claus7, Erwin G. Zoetendal1,2, Peter van

Baarlen8, and Michiel Kleerebezem1,2,8,9

The supplementary file includes detailed description of the materials and methods enclosed in the main text, two supplementary tables, two supplementary figures and references.

SUPPLEMENTARY METHODS

Transcriptome Analysis

RNA quantity and quality was assessed on the total RNA obtained from colon spectrophotometrically

(ND-1000, NanoDrop Technologies, Wilmington, USA) and with 6000 Nano chips (Bioanalyzer

2100; Agilent), respectively. RNA was judged as being suitable for array hybridisation only if samples showed intact bands corresponding to the 18S and 28S ribosomal RNA subunits, displayed no chromosomal peaks or RNA degradation products, and had a RIN (RNA integrity number) above 8.0.

The Ambion WT Expression kit (Life Technologies, P/N 4411974) in conjunction with the Affymetrix

GeneChip WT Terminal Labelling kit (Affymetrix, Santa Clara, CA; P/N 900671) was used for the preparation of labelled cDNA from 100ng of total RNA without rRNA reduction. Labelled samples were hybridised on Affymetrix GeneChip Mouse Gene 1.1 ST arrays, provided in plate format.

Hybridisation, washing and scanning of the array plates was performed on an Affymetrix GeneTitan

Instrument, according to the manufacturer’s recommendations. Detailed protocols can be found in the

Affymetrix WT Terminal Labelling and Hybridisation User Manual (P/N 702808 revision 4), and are also available upon request.

Quality control (QC) of the datasets obtained from the scanned Affymetrix arrays was performed using Bioconductor (Gentleman RC, 2004) packages integrated in an on-line pipeline (Lin et al.,

2011). Various advanced quality metrics, diagnostic plots, pseudo-images and classification methods were applied to ascertain only excellent quality arrays were used in the subsequent analyses (Heber

1 and Sick, 2006). An extensive description of the applied criteria is available upon request. The more than 825.000 probes on the Mouse Gene 1.1 ST array were redefined according to Dai et al. (Dai et al., 2005) utilising current genome information. In this study, probes were reorganised based on the

Entrez Gene database, build 37, version 1 (remapped CDF v13). Normalised expression estimates were obtained from the raw intensity values using the Robust Multiarray Analysis (RMA) pre- processing algorithm available in the Bioconductor library affyPLM using default settings (Bolstad et al., 2004). Next an empirical Bayes method, called ComBat (Johnson et al., 2007), was used to correct for the systematic error (batch effect) introduced during labelling.

Differentially expressed probe sets were identified using linear models, applying moderated t-statistics that implemented empirical Bayes regularization of standard errors (Storey and Tibshirani, 2003). The moderated t-test statistic has the same interpretation as an ordinary t-test statistic, except that the standard errors have been moderated across genes, i.e. shrunk to a common value, using a Bayesian model. To adjust for both the degree of independence of variances relative to the degree of identity and the relationship between variance and signal intensity, the moderated t-statistic was extended by a

Bayesian hierarchical model to define an intensity-based moderated T-statistic (IBMT) (Sartor et al.,

2006). IBMT improves the efficiency of the empirical Bayes moderated t-statistics and thereby achieves greater power while correctly estimating the true proportion of false positives. P-values were corrected for multiple testing using a false discovery rate (FDR) method proposed by Storey et al.

(Storey and Tibshirani, 2003). Only probe sets with a fold-change (FC) of at least 1.2 (up/down) and

FDR < 0.05 were considered to be significantly regulated.

Biological interpretation of expression datasets

Several complementary methods were applied to relate changes in gene expression to functional changes. Hierarchical clustering of modulated genes in the mucosa that were unique for each time- point (or several time-points) was performed by MeV (Multi experiment Viewer) (Saeed et al., 2006).

Temporal and spatial comparative analysis were carried out using STEM (Short Time-series

Expression Miner), which differentiates between real and random gene expression patterns from short

(up to 8 time-points) time-series microarray experiments (Ashburner et al., 2000). The statistical

2 significance of the profiles generated by STEM was calculated via a permutation test (n=1000) corrected using FDR <0.001 (Ernst and Bar-Joseph, 2006).

Biological interaction networks among regulated genes activated in response to conventionalisation were identified using Ingenuity Pathways Analysis (IPA) (Ingenuity Systems). IPA utilises a large expert-curated repository of molecular interactions, regulatory events, gene-to-phenotype associations, and chemical knowledge, mainly obtained from peer-reviewed scientific publications, that provides the building blocks for network construction. IPA annotations follow the GO annotation principle, but are derived from a knowledge base of more than 1 million protein–protein interactions. The IPA output includes modulated signalling pathways with statistical assessment of the significance of their modulation being based on a Fisher’s Exact Test. Our IPA analyses included comparison of differentially regulated genes in the colon in days 1, 2, 4, 8, and 16 in each case relative to expression observed in the control group (germfree=day 0). The input was all differentially regulated genes (p value ≤ 0.001, FC ≥ 1.2 and intensity ≥ 20) of colon after 1, 2, 4, 8, and 16 days post- conventionalisation.

Microbial profiling of colonic luminal contents

Luminal contents of the colon were analysed by Mouse Intestinal Tract Chip (MITChip), a diagnostic

16S rRNA array that consists of 3,580 unique probes especially designed to profile mouse intestine microbiota (Geurts et al., 2011) in analogy to the human intestinal tract (HIT)Chip (Rajilić-Stojanović et al., 2009). In short, 20 ng of colonic DNA extract was used to amplify the 16S rRNA genes with the primers T7prom-Bact-27-for (5’-

TGAATTGTAATACGACTCACTATAGGGGTTTGATCCTGGCTCAG-3’) and Uni-1492-rev (5’-

CGGCTACCTTGTTACGAC-3’). Subsequently, an in vitro transcription and labeling with Cy3 and

Cy5 dyes was performed. Fragmentation of Cy3/Cy5-labeled target mixes was followed by hybridisation on the arrays at 62.5˚C for 16 h in a rotation oven (Agilent Technologies, Amstelveen,

The Netherlands). The slides were washed and dried before scanning. Signal intensity data were obtained from the microarray images using Agilent Feature Extraction software, version 9.1

(http://www.agilent.com). MITChip analyses were performed using a set of R based scripts (http://r-

3 project.org) in combination with a custom designed relational database, which runs under the MySQL database management system (http://www.mysql.com).

Quantitative PCR analyses included quantification of total bacteria using 16S rRNA-specific primers and gene-specific qPCR, targeting bacterial groups with the following capacity: sulfate reduction, methane production, or butyrate production. QPCR was performed using an IQ5 Cycler apparatus

(Bio-Rad, Veenendaal, The Netherlands). Reactions were performed in triplicate in a single run.

Samples were analysed in a 25 µl reaction mixture consisting of 12.5 µl Bio-Rad master mix SYBR

Green (50 mM KCl, 20 mM Tris-HCl, pH 8.4, 0.2 mM of each dNTP, 0.625 U iTaq DNA polymerase,

3 mM MgCl2, 10 nM fluorescein), 0.1 µM of each primer (Supplementary Table S2) and 5 µl of template colonic DNA diluted 100 or 1000 times. Standard curves of 16S rRNA PCR product from pure cultures were created using serial 10-fold dilution of the purified PCR product corresponding to

108 to 100 16 S rRNA or specific gene. Standards included Lactobacillus casei (total bacteria),

Methanobrevibacter arboriphilus (methanogen), Desulfovibrio spp. (sulfate reducer) and

Faecalibacterium prausnitzii (butyrate producer). The following qPCR conditions were used: 95°C for

10 min, followed by 35 cycles of denaturation at 95°C for 15 sec, annealing temperature for 20 sec, extension at 72°C for 30 sec and a final extension step at 72°C for 5 min. A melting curve was determined at the end of each run to verify the specificity of the PCR amplicons. Data analysis was performed using the Bio-Rad software supplied with the IQ5 Cycler apparatus.

Tissue and urine metabolite profiling

All spectra were acquired on an Avance II 600MHz NMR spectrometer (Bruker Biospin, Rheinstetten,

Germany) equipped with a Bruker 5mm TXI triple resonance probe operating at 600.13 MHz, and using a BACS auto-sampler. Spectral acquisition of the aqueous phase was performed using the first increment of a standard nuclear Overhauser enhancement spectroscopy (NOESY) 1-dimensional 1H

NMR experiment ((noesypr1D)(RD-90°-t1-90°-tm-90°-acquisition), where RD is a relaxation delay, t1 is a short delay of 2 μs and tm is a mixing time of 100 ms. Water suppression was applied during the

RD and a 90° pulse set at 10 μs. For each spectrum, a total of 256 scans (16 dummy scans) were collected into 64k data points over a spectral width of 20 ppm. Data were analysed by applying an

4 exponential window function with a line broadening of 0.3 Hz prior to Fourier transformation to all

1D NMR spectra. The resultant spectra were phased, baseline corrected and calibrated to lactate (δ

1.33) ( for tissue samples) and TSP (δ 0.00) (for urine samples), manually using Topspin (2.0a, Bruker

BioSpin 2006). The spectra were subsequently imported into MatLab (R20010aSV, The MathsWorks inc.) where the region containing the water resonance (δ 4.6-5.2) was removed. The data were then normalised to the probabilistic quotient (Dieterle et al, 2006) to minimise the effects of inter-sample variation due to phenomena such as disparate sample volume, although every effort was made to keep this constant during sample preparation. Permutation testing of the Y matrix for PLS-based models was conducted using an in-house algorithm (Barnes, 2009) to determine whether the Q2Y (predictive ability) of the model was significantly different from the Q2Y calculated from 100 random permutations of Y. Variables with a correlation coefficient |r| > 0.45 were reported in Figure 4 to ascertain an overview of temporal metabolic changes. Assignments were made using additional two- dimensional NMR experiments and reference to databases of spectra of authentic compounds.

For the integration of the metabonomic and transcriptomic data using and O2PLS model as previously described (Li et al., 2008), the O2PLS model was calculated for pairwise datasets (metabolome, microbiota and transcriptome) and displayed. The O2PLS model computed with 1 predictive component, 3 orthogonal components and a 7-fold cross-validation procedure had a good predictive ability, as reflected by the following parameters: R2Y=0.35 and Q2Y=0.20. The significant variables were then selected based on their correlation with the scores of the model (p < 0.01).

5 SUPPLEMENTARY TABLES, FIGURES AND LEGENDS

I-Table S1. Relative contribution of significant (KruskalWallis) level 2 groups (genera-like level)

between early (days 1 and 2) and later (days 4, 8 and 16) time-points exceeding 0.1%. Light grey

filling represent early colonisers while dark grey represent late colonisers

Taxonomic level Conventionalisation days (average relative contribution ± SEM) Day 1 Day 2 Day 4 Day 8 Day 16 Actinobacteria Catenibacterium 1.24 ± 0.28 0.80 ± 0.17 0.27 ± 0.10 0.36 ± 0.11 0.25 ± 0.05 Bacilli Enterococcus 3.96 ± 1.78 2.10 ± 0.53 0.44 ± 0.17 0.01 ± 0.00 0.01 ± 0.00 Lactobacillus acidophilus et rel. 0.02 ± 0.00 0.03 ± 0.01 0.01 ± 0.00 0.01 ± 0.00 0.12 ± 0.11 Lactobacillus plantarum et rel. 0.10 ± 0.05 0.06 ± 0.03 0.03 ± 0.01 0.02 ± 0.00 0.47 ± 0.45 Lactobacillus salivarius et rel. 4.20 ± 0.92 2.40 ± 1.21 0.45 ± 0.32 0.85 ± 0.56 0.37 ± 0.23 Staphylococcus aureus et rel. 0.66 ± 0.34 0.26 ± 0.07 0.05 ± 0.02 0.01 ± 0.00 0.01 ± 0.00 Bacteroidetes Bacteroides distasonis et rel. 1.54 ± 0.59 2.85 ± 0.48 1.57 ± 0.46 0.90 ± 0.42 0.43 ± 0.23 Bacteroides fragilis et rel. 6.49 ± 1.45 8.18 ± 1.37 7.05 ± 1.74 1.98 ± 0.56 1.90 ± 0.60 Bacteroides plebeius et rel. 0.93 ± 0.21 0.99 ± 0.18 0.88 ± 0.27 0.37 ± 0.16 0.31 ± 0.09 Bacteroides vulgatus et rel. 2.04 ± 0.53 2.25 ± 0.32 2.98 ± 0.84 1.41 ± 0.62 1.05 ± 0.27 Porphyr. asaccharolytica et rel. 1.49 ± 0.40 0.86 ± 0.24 0.44 ± 0.09 0.41 ± 0.12 0.28 ± 0.10 Clostridium cluster II Unclassified Clostridiales II 0.52 ± 0.12 0.69 ± 0.21 0.84 ± 0.30 1.10 ± 0.19 1.55 ± 0.17 Clostridium cluster IV Clostridium leptum et rel. 0.05 ± 0.02 0.05 ± 0.01 0.05 ± 0.01 0.23 ± 0.04 0.30 ± 0.12 Sporobacter termitidis et rel. 0.53 ± 0.05 0.72 ± 0.19 1.76 ± 0.43 3.56 ± 0.59 4.34 ± 0.42 Clostridium cluster IX Peptococcus niger et rel. 0.03 ± 0.01 0.03 ± 0.00 0.08 ± 0.03 0.19 ± 0.06 0.30 ± 0.10 Clostridium cluster XI Anaerovorax et rel. 0.02 ± 0.00 0.24 ± 0.18 0.74 ± 0.54 0.98 ± 0.59 0.12 ± 0.04 Clostridium difficile et rel. 0.73 ± 0.19 2.03 ± 0.97 2.16 ± 0.91 2.62 ± 0.56 3.55 ± 0.24 Clostridium cluster XIVa Bryantella et rel. 0.15 ± 0.02 0.24 ± 0.07 1.04 ± 0.45 3.22 ± 0.94 2.64 ± 0.60 Butyrivibrio crossotus et rel. 0.19 ± 0.04 0.43 ± 0.17 0.82 ± 0.25 1.20 ± 0.24 1.66 ± 0.15 Clostridium sphenoides et 0.39 ± 0.06 0.39 ± 0.08 0.73 ± 0.23 1.47 ± 0.42 1.79 ± 0.29

6 rel. Clostridium symbosium et rel. 0.01 ± 0.00 0.01 ± 0.00 0.06 ± 0.03 0.07 ± 0.03 0.17 ± 0.10 Dorea et rel. 0.10 ± 0.02 0.13 ± 0.03 1.07 ± 0.31 3.13 ± 0.79 2.25 ± 0.43 Eubacterium plexicaudatum et rel. 0.89 ± 0.21 0.77 ± 0.22 1.95 ± 0.66 3.81 ± 0.94 4.17 ± 0.24 Lachnobacillus bovis et rel. 0.05 ± 0.01 0.05 ± 0.01 0.28 ± 0.10 0.83 ± 0.19 0.84 ± 0.11 Lachnospira pectinoschiza et rel. 0.06 ± 0.01 0.06 ± 0.01 0.20 ± 0.03 1.06 ± 0.41 0.68 ± 0.20 Unclassified Clostridiales XIVa 2.23 ± 0.37 2.34 ± 0.42 4.56 ± 1.40 9.54 ± 1.79 11.50 ± 0.56 Clostridium cluster XVI Allobaculum et rel. 0.81 ± 0.17 1.70 ± 0.69 0.41 ± 0.14 1.70 ± 1.08 0.32 ± 0.13 Unclassified Clostridiales XVI 5.49 ± 0.74 3.39 ± 0.91 1.52 ± 0.45 2.14 ± 0.70 1.44 ± 0.37 Clostridium cluster XVII Coprobacillus catenoformis et rel. 1.81 ± 0.25 1.36 ± 0.30 0.58 ± 0.16 0.86 ± 0.27 0.50 ± 0.12 Clostridium cluster XVIII Clostridium ramosum et rel. 7.68 ± 0.93 5.18 ± 1.23 2.07 ± 0.66 2.92 ± 1.03 1.74 ± 0.44 Deferribacteres Mucispirillum schaedleri et rel. 1.14 ± 0.13 1.03 ± 0.22 2.73 ± 0.50 0.60 ± 0.09 3.72 ± 3.04 Fibrobacteres Fibrobacter succinogenes et rel. 0.46 ± 0.12 0.28 ± 0.07 0.09 ± 0.03 0.21 ± 0.03 0.21 ± 0.09 Fusobacteria Fusobacterium 0.15 ± 0.03 0.16 ± 0.03 0.37 ± 0.16 0.66 ± 0.12 0.77 ± 0.07 Mollicutes Acholeplasma et rel. 0.12 ± 0.03 0.07 ± 0.01 0.02 ± 0.00 0.02 ± 0.01 0.01 ± 0.00 Solobacterium moorei et rel. 1.41 ± 0.30 0.88 ± 0.21 0.29 ± 0.10 0.42 ± 0.13 0.28 ± 0.07 Unclassified Mollicutes 3.92 ± 0.94 2.52 ± 0.63 2.48 ± 0.81 1.69 ± 0.32 1.78 ± 0.31 Proteobacteria (alpha) Sphingomonas 0.15 ± 0.02 0.16 ± 0.02 1.93 ± 1.28 0.30 ± 0.05 0.74 ± 0.49 Labrys methylaminiphilus et rel. 0.02 ± 0.00 0.22 ± 0.13 0.38 ± 0.17 0.60 ± 0.12 0.57 ± 0.10 Proteobacteria (beta) Sutterella wadsorthia et rel. 1.54 ± 0.24 2.13 ± 0.56 0.72 ± 0.18 1.11 ± 0.46 0.72 ± 0.27 Vibrio 0.21 ± 0.06 0.39 ± 0.09 0.33 ± 0.10 0.52 ± 0.11 0.53 ± 0.08 Proteobacteria (delta) Bilophila 0.01 ± 0.00 0.01 ± 0.00 0.38 ± 0.14 0.19 ± 0.12 0.14 ± 0.07 Desulfovibrio 0.01 ± 0.00 0.01 ± 0.00 0.11 ± 0.04 0.08 ± 0.03 0.08 ± 0.03 Proteobacteria (epsilon) Helicobacter 0.80 ± 0.17 0.94 ± 0.14 7.42 ± 4.46 0.52 ± 0.08 3.33 ± 2.82 Proteobacteria (gamma)

7 Pasteurella 0.10 ± 0.03 0.16 ± 0.04 0.25 ± 0.09 0.47 ± 0.09 0.48 ± 0.07 TM7 Unclassified TM7 0.02 ± 0.00 0.04 ± 0.02 0.35 ± 0.13 0.67 ± 0.13 0.81 ± 0.02

8 II-Table S2. Primers used for the quantification of butyrate producers (ButyrilCoA gene), sulfate reducers (dsr gene) and Methanogens (McrA gene), by qPCR.

Primer Primer sequence (5’-3’) Annealing References name temperatur e

Methanogens mlas F GGTGGTGTMGGDTTCAC 55°C (Steinberg and Regan, mcrA-R MCARTA 2008) CGTTCATBGCGTAGTTVGGRTAG T Sulfate RH1dsr-F GCCGTTACTGTGACCAGCC 60°C (Ben-Dov et al, 2007) reducers RH3-dsr-R GGTGGAGCCGTGCATGTT Butyrate BcoATscr- GCIGAICATTTCACITGGAAYWSIT 53°C (Louis and Flint, 2007) producers F GGCAYATG BcoATscr- CCTGCCTTTGCAATRTCIACRAAN R GC

9 Day 1 Day 2 Day 4 Day 8 Day 16

Figure S1. Dynamics of the establishment of the gut microbiota in conventionalized mice in the large intestine (cecum and colon; n=6 per location), represented by the relative contribution of phylum-like levels (levels 0) at the different days of sampling.

10 Figure S2. Significantly altered signalling pathways in response to microbial colonisation. Cellular pathways significantly modulated in the colon during 16 days of conventionalisation as calculated via a one-tailed Fisher’s Exact test in IPA and represented as -log (p value); -log values exceeding 1.30 were significant (p <0.05), (n=6 mice/time-point).

11 Figure S3. Transcriptome-metabolite correlations

Heat maps of (A) the 209 genes of which the expression was positively correlated with the concentrations of myo- and scyllo-inositol, (B) 187 genes of which expression was positively correlated with the concentrations of glutamate, alanine and glycine. The colours represent the changes in gene expression level at the indicated time-points.

12 References:

Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. (2000). Gene Ontology: tool for the unification of biology. Nat Genet 25: 25-29.

Ben-Dov E, Brenner A, Kushmaro A. (2007). Quantification of sulfate-reducing bacteria in industrial wastewater, by real-time polymerase chain reaction (PCR) using dsrA and apsA genes. Microb Ecol

54: 439-451.

Barnes PJ. (2009). Intrinsic asthma: not so different from allergic asthma but driven by superantigens?

Clin Exp Allergy 39: 1145-1151.

Bolstad BM, Collin F, Simpson KM, Irizarry RA, Speed TP. (2004). Experimental design and low- level analysis of microarray data. Int Rev Neurobiol 60: 25-58.

Dai M, Wang P, Boyd AD, Kostov G, Athey B, Jones EG, et al. (2005). Evolving gene/transcript definitions significantly alter the interpretation of GeneChip data. Nucleic Acids Res 33: e175.

Dieterle F, Ross A, Schlotterbeck G, Senn H. (2006). Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in 1H NMR metabonomics. Anal chem 78: 4281-4290.

Ernst J and Bar-Joseph Z. (2006). STEM: a tool for the analysis of short time series gene expression data. BMC Bioinformatics 7: 191.

Geurts L, Lazarevic V, Derrien M, Everard A, Van Roye M, Knauf C, et al. (2011). Altered gut microbiota and endocannabinoid system tone in obese and diabetic leptin-resistant mice: impact on apelin regulation in adipose tissue. Front Microbiol 2:149.

Heber S and Sick B. (2006). Quality assessment of affymetrix GeneChip data. OMICS 10: 358-368.

Johnson WE, Li C, Rabinovic A. (2007). Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8: 118-127.

Li M, Wang B, Zhang M, Rantalainen M, Wang S, Zhou H, Zhang Y, et al. (2008). Symbiotic gut microbes modulate human metabolic phenotypes. Proc Natl Acad Sci U S A 105: 2117-2122.

Lin K, Kools H, de Groot PJ, Gavai AK, Basnet RK, Cheng F, et al. (2011). MADMAX-Management and analysis database for multiple ~omics experiments. J Integr Bioinform 8: 160.

13 Louis P, Flint HJ. (2007). Development of a semiquantitative degenerate real-time PCR-based assay for estimation of numbers of butyryl-coenzyme A (CoA) CoA transferase genes in complex bacterial samples. Appl Environ Microbiol 73: 2009-2012.

Rajilić-Stojanović M, Heilig HG, Molenaar D, Kajander K, Surakka A, Smidt H, de Vos WM. (2009).

Development and application of the human intestinal tract chip, a phylogenetic microarray: analysis of universally conserved phylotypes in the abundant microbiota of young and elderly adults. Environ

Microbiol 11: 1736-1751.

Saeed AI, Bhagabati NK, Braisted JC, Liang W, Sharov V, Howe EA, et al. (2006). TM4 microarray software suite. Methods Enzymol 411: 134-193.

Sartor MA, Tomlinson CR, Wesselkamper SC, Sivaganesan S, Leikauf GD, Medvedovic M. (2006).

Intensity-based hierarchical Bayes method improves testing for differentially expressed genes in microarray experiments. Bmc Bioinformatics 7: 538.

Steinberg LM, Regan JM. (2008). Phylogenetic Comparison of the Methanogenic Communities from an Acidic, Oligotrophic Fen and an Anaerobic Digester Treating Municipal Wastewater Sludge. Appl

Environ Microbiol 74 : 6663-6671.

Storey JD ,Tibshirani R. (2003). Statistical significance for genomewide studies. Proc Natl Acad Sci U

S A 100: 9440-9445.

14