Rational design of a microbial consortium of mucosal sugar utilizers reduces Clostridiodes difficile colonization
Ethics statement
All animal experiments were performed at the Max F. Perutz Laboratories of the University of Vienna, Austria. All experiments were discussed and approved by the University of Veterinary Medicine, Vienna, Austria, and conducted in accordance with protocols approved by the Federal Ministry for Education, Science and Research of the Republic of Austria under the license number BMWF-66.006/0001-WF/V/3b/2016. Animals were randomized for interventions but researchers processing the samples and analyzing the data were aware which intervention group corresponded to which cohort of animals.
Mouse colon incubations
Three adult (6-8 weeks old) C57BL/6N mice bred at the Max F. Perutz Laboratories, University of Vienna, under SPF conditions were sacrificed per experiment, and their colon was harvested anaerobically (85% N2, 10% CO2, 5% H2) in an anaerobic tent (Coy Laboratory Products, USA). Contents from each colon were suspended in 7.8 mL of 50% D2O-containing PBS and homogenized by vortexing. Similar conditions have been successfully applied in the past to monitor activity of individual cells in gut communities without causing major changes in the activity of individual community members34. The homogenate was left to settle for 10 min, and the supernatant was then distributed into glass vials and supplemented with different concentrations of mucosal sugar monosaccharides, glucose, mucin or nothing (no-amendment control) (all amendment chemicals were from Sigma-Aldrich, except D(+)-galactose which was purchased from Carl Roth GmbH) (Fig. 1a,c). After incubation for 6 h at 37 °C, glycerol was added (to achieve a final concentration of 20% (v/v) of glycerol in the microcosms) and the vials were crimp-sealed with rubber stoppers and stored at −80 °C until further processing. Prior to glycerol addition, subsamples of the biomass were collected, pelleted and supernatants stored at −80 °C for HILIC LC-MS/MS measurements. Pellets were washed with PBS to remove D2O and were fixed in 3% formaldehyde for 2 h at 4 °C and stored in 50% PBS/50% ethanol solution at −20 °C until further use. A total of three biological replicates were established using starting material pooled from three animals each (experiments MonoA, MonoB and MonoC). For the MonoA and MonoB experiments, microcosms were established for all the different concentrations of monosaccharides tested (Fig. 1c), while for MonoC only the highest concentrations of monosaccharides tested in incubations MonoA and MonoB were supplemented. Note that analysis of mucin-amended sorted fractions has been published elsewhere35. Since mucin contains all the monosaccharides included in this study, it constitutes an important control, and therefore we processed the sequencing data from mucin sorts in parallel with our samples and included it in our analyses (Fig. 2).
Mass spectrometric analysis of mucosal monossaccharides
Hydrophilic interaction chromatography (HILIC) LC-MS/MS was used for the measurement of mucosal monosaccharides in microcosm supernatants. Frozen samples were thawed at room temperature and centrifuged for 10 min at 18.000 × g and 4 °C. Supernatants were then diluted 1:50 with acetonitrile:water (1:1; v/v) and a volume of 3 µl was injected onto the chromatographic column. The UHPLC system (UltiMate 3000, Thermo Scientific) was coupled to a triple quadrupole mass spectrometer (TSQ Vantage, Thermo Scientific) by an electrospray ionization interface. Hydrophilic interaction chromatographic separation was realized on a Luna aminopropyl column (3 µm, 150 × 2 mm; Phenomenex, Torrance, CA) at a flow rate of 0.25 ml/min. Eluent A consisted of 95% water and 5% acetonitrile with 20 mM ammonium acetate and 40 mM ammonium hydroxide as additives and eluent B of 95% acetonitrile and 5% water. A multi-step gradient was optimized as follows: 100% B until minute 2, then linearly decreased to 80% B until minute 20 and further to 0% B until minute 25. The column was kept at 0% B for 4 min before it was equilibrated for 5 min at the initial conditions. The column temperature was maintained at 40 °C. The mass spectrometer was operated in multiple reaction monitoring (MRM) mode. Electrospray ionization (ESI) was optimized as follows: spray voltage 2800 V (positive mode) and 3000 V (negative mode); vaporizer temperature 250 °C; sheath gas pressure 30 Arb; ion sweep gas pressure 2 Arb; auxiliary gas pressure 10 Arb; capillary gas temperature 260 °C. Mass spectrometric parameters were optimized by direct injection and are reported together with the retention times of individual sugars in Supplementary Table 6. Spiking experiments and regular quality control checks were conducted to evaluate and ensure the systems’ proper performance.
Confocal Raman microspectroscopy and spectral processing of fixed samples
Formaldehyde-fixed samples were spotted on aluminum-coated slides (Al136; EMF Corporation) and washed by dipping into ice-cold Milli-Q (MQ) water (Millipore) to remove traces of buffer components. Individual cells were observed under a 100×/0.75 NA microscope objective. Single microbial cell spectra were acquired using a LabRAM HR800 confocal Raman microscope (Horiba Jobin-Yvon) equipped with a 532-nm neodymium-yttrium aluminum garnet (Nd:YAG) laser and either 300 grooves/mm diffraction grating. Spectra were acquired in the range of 400–3200 cm−1 for 30 s with 2.18 mW laser power. Raman spectra were background-corrected using the sensitive nonlinear iterative peak algorithm, and afterwards normalized to the sum of its absolute spectral intensity34. For quantification of the degree of D substitution in CH bonds (%CD), the bands assigned to C–D (2040–2,300 cm−1) and C–H (2,800–3,100 cm−1) were calculated using integration of the specified region34.
Raman-activated cell sorting
For RACS of D-labeled cells, 100 μl of glycerol-preserved microcosms containing non-fixed cells were pelleted, washed once with MQ water containing 0.3 M glycerol and finally resuspended in 0.5 ml of 0.3 M glycerol in MQ water. Cell sorting was performed in a fully automated manner using a Raman microspectroscope (LabRAM HR800, Horiba Scientific, France) combined with optical tweezers and a polydimethylsiloxane (PDMS) microfluidic sorter. The optical tweezers (1,064 nm Nd:YAG laser at 500 mW) and Raman (532 nm Nd:YAG laser at 45 mW or 80 mW; see below) laser were focused at the same position of the interface between the sample and sheath streams using a single objective (63x, 1.2 NA water-immersion, Zeiss). The in-house software based on the graphical user interface (GUI; written in MATLAB) detected the single-cell capture and its deuterium labeling status by calculating the cell index (PC = I1,620-1,670/Ifluid,1,620-1,670; where I is the integrated intensity between the indicated wavenumbers) and the labeling index (PL = I2040-2300/I1850-1900), respectively. We did not detect a significant change in the C–D peak region (2040–2300 cm−1) due the presence of 0.3 M of glycerol in the sorting fluid (added to minimize the osmotic stress when the sample was re-suspended for the RACS) (Supplementary Fig. 2a). Other spectral regions (e.g., 2700 cm−1) were slightly affected, but the sorting algorithm employed and the parameters described above take these small changes into account: the cell index PC (I1620–1670/Ifluid,1620–1670) used to detect single-cell capture was calculated by comparing the Raman intensity of cells measured in real-time to that of the working fluid measured in the calibration (conducted before the actual sorting). The threshold value for PL (I2040–2300/I1850–1900) was chosen based on the measurement of the control sample (i.e., sample incubated in non-D2O-containing medium). We used two software versions, each of which uses 45 mW (version 1) and 80 mW (version 2) Raman laser powers, respectively. The second version operates with higher power based on the addition of a laser shutter that blocks the Raman laser while the cells are being translocated, reducing the laser-induced damage on the cell. This version allows shorter acquisition times to be employed, and therefore higher throughput of the platform. The laser power for each version was chosen based on visual inspection of captured cells as described35. For the NeuAc and GlcNAc-amendment sorts (version 1, since version 2 was not yet available), PC value was calculated from cell spectra acquired for 2 s at the “capture location”, while the PL value was calculated from spectra obtained with a 5 s exposure time at the “evaluation location”. Fucose, GalNAc, and galactose-supplemented sorts were performed with version 2 of the platform, which in the meantime became available, significantly reducing sorting times. For these sorts both PC and PL values were simultaneously measured at the “capture location” with a 0.3 s exposure time. Only the D-labeled cells were translocated to the ‘evaluation location’ and immediately released. In order to determine the threshold PL above which a cell from the microcosms should be considered D-labeled (and therefore selected and sorted), cells from glucose-supplemented microcosms incubated in the absence or presence of D (0% versus 50% D2O in the microcosms) were run on the platform prior to sorting. The threshold PL number can vary across microcosms due to different microbial compositions and/or physiological status of cells present in the starting material, as well as due to different laser powers employed. Therefore we determined the PL threshold separately for both MonoA and MonoB incubations using both 45 and 80 mW laser power. Nevertheless, we reached a PL threshold of 6.19 for all sets of conditions and incubations tested (Supplementary Fig. 2b). We speculate this was due to the identical conditions used in both incubations and the fact that both communities have a similar microbial composition (Fig. 1e). To test the sorting accuracy of the platform on our samples, the negative control (H2O, glucose-supplemented microcosm) was re-run in the platform and sorted using the adopted threshold (PL = 6.19) (Supplementary Table 1). As expected, no cells were considered labeled by the platform under these conditions. Sorted fractions were nevertheless collected and sequenced as controls.
Preparation of 16S rRNA gene amplicon libraries and 16S rRNA gene sequence analyses
DNA extracted from the mouse colon microcosms or from mouse fecal pellets using a phenol-chloroform bead-beating protocol52 was used as a template for PCR. PCR amplification was performed with a two-step barcoding approach53. In the first-step PCR, the 16S rRNA gene of most bacteria was targeted using oligonucleotide primers (Supplementary Table 7) containing head adaptors (5′-GCTATGCGCGAGCTGC-3′) in order to be barcoded in a second step PCR53. Barcode primers consisted of the 16 bp head sequence and a sample-specific 8 bp barcode from a previously published list at the 5′ end. The barcoded amplicons were purified with the ZR-96 DNA Clean-up Kit (Zymo Research, USA) and quantified using the Quant-iT PicoGreen dsDNA Assay (Invitrogen, USA). An equimolar library was constructed by pooling samples, and the resulting library was sent for sequencing on an Illumina MiSeq platform at Microsynth AG (Balgach, Switzerland). Paired-end reads were quality-filtered and processed using QIIME 153,54. Reads were then clustered into operational taxonomic units (OTUs) of 97% sequence identity and screened for chimeras using UPARSE implemented in USEARCH v8.1.186155. OTUs were classified using the RDPclassifier v2.1256 as implemented in Mothur v1.39.557 using the Silva database v13258. Sequencing libraries were rarefied and analyzed using the vegan package (v2.4-3) of the software R (https://www.r-project.org/, R 3.4.0).
Sequencing of mouse gut isolates
Bacteroides sp. Isolate FP24 was isolated from YCFA agar plates (DSMZ medium 1611- YCFA MEDIUM (modified)) by plating ten-fold dilution series of a microcosms supplemented with 2 mg/ml of NeuAc (experiment MonoB)). Escherichia sp. isolate FP11 and Anaerotruncus sp. isolate FP23 were isolated from C. difficile minimal medium17 agar plates supplemented with 0.25% NeuAc or 0.25% GlcNAc by plating ten-fold dilution series of a microcosms supplemented with 2 mg/ml of NeuAc (experiment MonoB) or of a microcosms supplemented with 5 mg/ml of GlcNAc (experiment MonoA), respectively. Colonies were re-streak on the same medium plates until complete purity. Pure colonies were grown overnight in 10 mL of BHI medium (Brain heart infusion at 37 g per liter of medium) supplemented with: yeast extract, 5 g; Na2CO3, 42 mg; cysteine, 50 mg; vitamin K1, 1 mg; hemin, 10 µg. DNA was extracted from pelleted biomass using the QIAGEN DNAeasy Tissue and Blood kit (Qiagen, Austin, TX, USA) according to the manufacturer´s instructions. Sequencing libraries were prepared using the NEBNext® Ultra™ II FS DNA kit (Illumina) and sequenced in an Illumina MiSeq platform with 300-bp paired-end sequencing chemistry (Joint Microbiome Facility, University of Vienna and Medical University of Vienna, Austria). Reads were quality trimmed with the bbduk option of BBmap (v 34.00) at phrad score 15. Quality-trimmed reads were assembled with SPAdes (v 3.11.1)59. For isolate FP11, assembled reads were subsequently, iteratively (n = 6) reassembled with SPAdes using contigs of >1 kb from the previous assembly as “trusted contigs” for input and iterating kmers from 11 to 121 in steps of 10. CheckM (v1.0.6) assessment60 of these genomes is summarized in Supplementary Data 1.
Mini-metagenome sequencing and genomic analyses
Labeled RACS cells were collected into PCR tubes, lysed and subjected to whole-genome amplification using the Repli-g Single Cell Kit (QIAGEN), according to the manufacturer’s instructions. Shotgun libraries were generated using the amplified DNA from WGA reactions (sorted fractions) or DNA isolated using the phenol-chloroform method (initial microcosms) as a template and Nextera XT (Illumina) reagents. Libraries were sequenced with a HiSeq 3000 (Illumina) in 2 × 150 bp mode at the Biomedical Sequencing Facility, Medical University of Vienna, Austria. The sequence reads were quality trimmed and filtered using BBMap v34.00 (https://sourceforge.net/projects/bbmap/). The remaining reads were assembled de novo using SPAdes 3.11.159 in single-cell mode (k-mer sizes: 21, 35, 55). Binning of the assembled reads into metagenome-assembled genomes (MAGs) was performed with MetaBAT 2 (v2.12.1)61 using the following parameters: minContig 2000, minCV 1.0, minCVSum 1.0, maxP 95%, minS 60, and maxEdges 200. The quality and contamination of all MAGs were checked with CheckM 1.0.660 (Supplementary Data 1). MAGs >200 kb obtained from all samples were compared and de-replicated using dRep 1.4.362. Automatic genome annotation of contigs >2 kb within each de-replicated MAG was performed with RAST 2.063. Taxonomic classification of each MAG was obtained using GTDB-Tk64 (v0.1.3, gtdb.ecogenomic.org/).
The relative abundance of each MAG on the initial microcosms was calculated based on metagenomic coverage. Filtered reads from each sequenced microcosm were mapped competitively against all retrieved MAGs using BBMap (https://sourceforge.net/projects/bbmap/). Read coverage was normalized by genome size and relative abundances of each genome in each sample were calculated based on the formula: covA = (bpA/gA)/(bpT/gT), where covA is the relative abundance of MAG A on a particular sample, bpA is the number of base pairs from reads matching MAG A, gA is the genome length of MAG A, bpT is the total number of base pairs from reads matching all MAGs recovered from that particular sample and gT is the sum of all MAGs genome lengths.
For determination of the presence of encoded enzymes for catabolism of mucin monosaccharides among MAGs, predicted protein sequences from recovered MAGs were subject to local BLASTP analyses65, against a custom database. The database was composed of all enzymes involved initial hydrolysis and catabolism of mucosal sugar monosaccharides (Supplementary Data 2), which were previously curated from a total of 395 human gut bacteria15. A strict e-value threshold of 10−50 was used for all BLASTP analyses. During initial setup of the analysis pipeline, functional assignments of proteins that gave positive BLASTP hits were manually verified by examining annotations from RAST 2.0 and by performing BLASTP analyses against the NCBI-nr database (NCBIBlast 2.2.26).
To verify the enrichment of a selected dataset of mucin-degrading enzymes35 in the assemblies derived from sorted fractions, BLASTX analyses of scaffolds from each fraction as well as from the initial microcosms metagenomes (unsorted) were performed against the selected mucin-degrading enzyme sequences15 (Supplementary Table 3). An e-value threshold of 10−50 was also used for all BLASTX analyses.
Phylogenomic analyses
A concatenated marker alignment of 34 single-copy genes was generated for all MAGs using CheckM 1.0.660 and the resulting alignment was used to calculate a tree with the approximate maximum-likelihood algorithm of FastTree 2.1.1066. Phylogenomic trees were visualized and formatted using iTOL v4 (https://itol.embl.de/). In order to identify the closest relative for each MAG, the query MAG and close reference genomes (based on the generated phylogenomic tree) were compared using dRep 1.4.362. Compared genomes with a whole-genome-based average nucleotide identity (ANIm) >99%39 were considered to be the same population genome.
High-resolution mass spectrometric analyses
Glycerol-preserved biomass (150 μL) from microcosm incubations was pelleted and suspended in 50 μL of lysis buffer (1% sodium dodecyl sulfate (SDS), 10 mM TRIS base, pH 7.5). Protein lysates were subjected to SDS polyacrylamide gel electrophoresis followed by in-gel tryptic digestion. Proteins were stained with colloidal Coomassie Brilliant Blue G-250 (Roth, Kassel, Germany) and detained with Aqua dest. Whole protein bands were cut into gel pieces and in-gel-digestion with trypsin 30 µL (0.005 µg/µL) was performed overnight. Extracted peptides where dried and resolved in 0.1% formic acid and purified by ZipTip® treatment (EMD Millipore, Billerica, MA, USA).
In total, 5 µg of peptides were injected into nanoHPLC (UltiMate 3000 RSLCnano, Dionex, Thermo Fisher Scientific), followed by separation on a C18-reverse phase trapping column (C18 PepMap100, 300 µm × 5 mm, particle size 5 µm, nano viper, Thermo Fischer Scientific), followed by separation on a C18-reverse phase analytical column (Acclaim PepMap® 100, 75 µm × 25 cm, particle size 3 µm, nanoViper, Thermo Fischer Scientific). Mass spectrometric analysis of eluted peptides where performed on a Q Exactive HF mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) coupled with a TriVersa NanoMate (Advion, Ltd., Harlow, UK) source in LC chip coupling mode. LC Gradient, ionization mode and mass spectrometry mode were performed as described before67. Briefly, peptide lysate were injected into a Nano-HPLC and trapped in a C18-reverse phase column (Acclaim PepMap® 100, 75 µm × 2 cm, particle size 3 µM, nanoViper, Thermo Fisher) for 5 min. Peptide separation was followed by a two-step gradient in 90 min from 4 to 30% of B (B: 80% acetonitrile, 0.1% formic acid in MS-grade water) and then 30 min from 30 to 55% of B. The temperature of the separation column was set to 35 °C and the flow rate was 0.3 µL/min. The eluted peptides were ionized and measured. The MS was set to a full MS/dd-MS2 mode scan with positive polarity. The full MS scan was adjusted to 120,000 resolution, the automatic gain control (AGC) target of 3 × 106 ions, maximum injection time for MS of 80 s and a scan range of 350 to 1550 m/z. The dd-MS2 scan was set to a resolution of 15,000 with the AGC target of 2 × 105 ions, a maximum injection time for 120 ms, TopN 20, isolation window of 1.6 m/z, scan range of 200 to 2000 m/z and a dynamic exclusion of 30 s.
Raw data files were converted into mzML files and searched with MS-GF + against a database obtained from microcosm metagenomes composed of 276,284 predicted protein-encoding sequences. The following parameters were used for peptide identification: enzyme specificity was set to trypsin with one missed cleavage allowed using 10 ppm peptide ion tolerance and 0.05 Da MS/MS tolerance. Oxidation (methionine) and carbamidomethylation (cysteine) were selected as modifications. False discovery rates (FDR) were determined with the node Percolator68. Proteins were considered as identified when at least one unique peptide passed a FDR of 5%.
The MetaProSIP toolshed69 embedded in the Galaxy framework70 (v2.3.2, http://galaxyproject.org/) was used to identify the incorporation of stable isotopes into peptides. MetaProSIP calculates the relative isotope abundance (RIA) on detected isotopic mass traces (m/z tolerance of ±10 ppm, intensity threshold of 1000, and an isotopic trace correlation threshold of 0.7).
In vitro growth experiments
A. muciniphila strain Muc (DSM 22959), Ruthenibacterium lactatiformans strain 585-1 (DSM 100348) and Alistipes timonensis strain JC136 (DSM 25383) were obtained from DSMZ. Muribaculum intestinale strain YL27 (DSM 28989) was kindly provided by Prof. Bärbel Stecher (Max-von-Pettenkofer Institute, LMU Munich, Germany). Bacteroides sp. FP24 was isolated from YCFA agar plates (DSMZ medium 1611-YCFA MEDIUM (modified)). All strains were grown in reduced A II medium71 consisting of (per liter of medium): BHI, 18.5 g; yeast extract, 5 g; trypticase soy broth, 15 g; K2HPO4, 2.5 g; hemin, 10 µg; glucose, 0.5 g; Na2CO3, 42 mg; cysteine, 50 mg; menadione, 5 µg; fetal calf serum (complement-inactivated), 3% (vol/vol). For A. muciniphila cultivation, the growth medium was supplemented with 0.025% (w/v) of mucin. C. difficile was grown in BHI medium (37 g per liter of medium) supplemented with: yeast extract, 5 g; Na2CO3, 42 mg; cysteine, 50 mg; vitamin K1, 1 mg; hemin, 10 µg. All strains were grown at 37 °C under anaerobic conditions until stationary phase, and then serially diluted and plated into media agar plates in order to determine the number of viable cells present in 1 ml of stationary phase-culture. For mixed-growth experiments, the culture volume equivalent to 1 × 106 CFU of each strain was pelleted, cells were washed with PBS, mixed in equal proportions and finally resuspended in 100 μl of PBS. This bacterial mixture containing a total of 5 × 106 BacMix cells was then used to inoculate 2.5 ml of A II medium (diluted two fold in 2× PBS) supplemented or not with 0.25% (10 mM) carbon source (0.125% or 4 mM of NeuAc and 0.125% or 6 mM of GlcNAc). After 12 h, the same tube was inoculated with 1 × 106 C. difficile CFU and bacterial growth was followed by measuring the OD at 600 nm every hour until stationary phase. At three distinct points of the C. difficile growth curve—lag (t12, right after C. difficile addition), mid-exponential (t18) and early stationary phase (t21)—a sample aliquot was collected and ten-fold dilutions were plated in a C. difficile selective medium72. This selective medium (CCFA) includes antibiotics such as cycloserine and cefoxitn at concentrations that are inhibitory to most gut organisms, except for C. difficile, allowing to determine total C. difficile counts. A second aliquot was immediately pelleted and the pellet was stored at −80 °C for RNA extraction.
Quantitative PCR of C. difficile 16S rRNA gene copy number density
DNA was extracted from 100 mg of mouse fecal pellet using the QIAGEN DNAeasy Tissue and Blood kit (Qiagen, Austin, TX, USA) according to the manufacturer´s instructions, with an additional step of mechanical cell disruption by bead beating (30 s at 6.5 m/s) right after addition of kit lysis buffer AL. Extracted DNA (2 μl) was subjected to quantitative PCR using 0.2 μM of primers specifically targeting the C. difficile 16S rRNA gene73 (Supplementary Table 7) and 1× SYBR green Master Mix (Bio-Rad) in a total reaction volume of 20 μl. Standard curves were generated from DNAs extracted from fecal pellets of SPF (uninfected) mice spiked in with different known numbers of C. difficile cells (102, 103, 104, 105, 106, 107, and 108) as described in Kubota et al., 2014. Amplification and detection were performed using a CFX96™ Real-Time PCR Detection System (Bio-Rad) using the following cycling conditions: 95 °C for 5 min, followed by 40 cycles of 95 °C for 15 s, 56 °C for 20 s, and 72 °C for 30 s. To determine the specificity of PCR reactions, melt curve analysis was carried out after amplification by slow cooling from 95 to 60 °C, with fluorescence collection at 0.3 °C intervals and a hold of 10 s at each decrement. Only assays with amplification efficiencies above 80% were considered for analysis.
RNA extraction and quantitative real-time PCR
Total nucleic acids (TNA) were extracted from mouse fecal pellets or from in vitro cultures using a phenol-chloroform bead-beating protocol52. RNA was purified from DNAse-treated TNA fractions using the GeneJET Cleanup and Concentration micro kit (Thermo Fisher Scientific). cDNA was synthesized from 0.5 μg of total RNA with 1 μl of random hexamer oligonucleotide primers. Samples were heated for 5 min at 70 °C. After a slow cooling, 2 μl of deoxynucleoside triphosphates (dNTP; 2.5 mM each), 40 units of recombinant ribonuclease inhibitor (RNaseOUT) and 4 μl of reverse transcription (RT) buffer were added and cDNAs were synthesized for 2 h at 50 °C using 200 units SuperScript™ III Reverse Transcriptase (all reagents used in cDNA synthesis were from Thermo Fisher Scientific). Real-time quantitative PCR was performed in a 20-μl reaction volume containing 2 μl of cDNA, 1x SYBR green Master Mix (Bio-Rad) and 0.2 μM of gene-specific C. difficile primers targeting the following genes: DNA polymerase III PolC-type dnaF74, nanA, nanT17 and nagB (this work; Supplementary Table 7). Amplification and detection were performed as described above. In each sample, the quantity of cDNAs of a gene was normalized to the quantity of cDNAs of the C. difficile DNA polymerase lII gene74 (dnaF). The relative change in gene expression was recorded as the ratio of normalized target concentrations (threshold cycle [ΔΔCT] method75). Fold changes were normalized to in vitro growths in C. difficile minimal medium containing 0.5% glucose17. To determine the specificity of PCR reactions, melt curve analysis was carried out after amplification by slow cooling from 95 to 60 °C, with fluorescence collection at 0.3 °C intervals and a hold of 10 s at each decrement. Only assays with amplification efficiencies above 80% were considered for analysis.
Murine in vivo adoptive transfer experiments
Female C57BL/6N 6-8 weeks old mice (n = 33 total) were purchase from Janvier Labs. Animals were kept in isolated, ventilated cages under specific pathogen-free conditions at the animal facility of the Max F. Perutz Laboratories, University of Vienna, Austria, with controlled temperature of 21 ± 1 °C and humidity of 50 ± 10%, in a 12-h light/dark cycle. Mice received a standard diet (V1124-300; Ssniff, Soest, Germany) and autoclaved water ad libitum. Mice were administered antibiotics (0.25 mg/ml clindamycin (Sigma-Aldrich) for six days in drinking water) and subsequently assigned randomly to one of two groups. One day following antibiotic cessation, mice from each group were split into 3 cages (to minimize the cage effect) and each mouse received either 5,000,000 CFU of a 5-bacteria suspension (BacMix, containing equal numbers of A. muciniphila strain Muc (DSM 22959), Ruthenibacterium lactatiformans strain 585-1 (DSM 100348), Alistipes timonensis strain JC136 (DSM 25383), Muribaculum intestinale strain YL27 (DSM 28989), and Bacteroides sp. isolate FP24) or vehicle (PBS) by gavage (Fig. 5a). At the time of BacMix and BacMixC administration, the mouse diet was switched from a standard diet (V1124-300; Ssniff, Soest, Germany) to a isocaloric polysaccharide-deficient chow76 with sucrose but no cellulose or starch (Ssniff, Soest, Germany). For the BacMixC adoptive transfer, each mouse (n = 10 per group) received 5,000,000 CFU of a 3-bacteria suspension (BacMixC, containing equal numbers of Anaerotruncus sp. isolate FP23; Lactobacillus hominis strain DSM 23910 and of Escherichia sp. isolate FP11) or vehicle (PBS) by gavage (Supplementary Fig. 7). One day after BacMix or BacMixC administration, mice were challenged with 1,000,000 CFU of C. difficile strain 630 deltaErm77. A. muciniphila, R. lactatiformans and M. intestinale were grown in reduced A II medium (supplemented with 0.025% mucin for A. muciniphila). Anaerotruncus sp. isolate FP23, A. timonensis and Bacteroides sp. isolate FP24 were grown in PYG (DSMZ medium 104). Lactobacillus hominis and Escherichia sp. isolate FP11 were grown in YCFA medium (DSMZ medium 1611). C. difficile was grown in BHI medium (37 g per liter of medium). All bacteria were grown under anaerobic conditions (5% H2, 10% CO2, rest N2) at 37 °C and resuspended in anaerobic PBS prior to administration to animals. C. difficile titers were quantified in fecal samples obtained from mice 24, 48, 72, and 120 h after infection by overnight cultivation in C. difficile selective agar plates72. Animals were monitored throughout the entire experiment and weight loss was recorded.
Measurement of mucus thickness and goblet cell volume
Segments of mouse colon (approximately 10 mm long) were fixed in 2% PFA in phosphate-buffered saline (PBS) for 12 h at 4 °C. Samples were washed with 1× PBS and then stored in 70% ethanol at 4 °C until embedding. For embedding, samples were immersed in 3 changes of xylene for 1 h each, then immersed in 3 changes of molten paraffin wax (Paraplast, Electron Microscopy Sciences) at 56–58 °C for one hour each. Blocks were allowed to harden at room temperature. Sections were cut to 4 μm thickness using a Leica microtome (Leica Microsystems) and were floated on a water bath at 40–45 °C, then transferred to slides, dried, and stored at room temperature. In preparation for staining, slides were de-paraffinized by heating at 60 °C for 30 min followed by immersion in 2 changes of xylene for 5 min each, then in 2 changes of 100% ethanol for 1 minute each, then rehydrated through 80%, and 70% ethanol for 1 minute each. Slides were then dipped in water, drained, air-dried and a drop of Alcian Blue stain (Sigma Aldrich) applied on top. Samples were incubated with Alcian Blue for 20 min at room temperature and then washed in water to remove the excess of stain. Samples were mounted in Vectashield Hardset™ Antifade Mounting Medium (Vector Laboratories) and visualized using a Leica Confocal scanning laser microscope (Leica TCS SP8X, Germany). For measurements of the width of the mucus layer and determination of goblet cell volume per crypt, ImageJ software (https://imagej.nih.gov/ij/, 1.48 v) was used.
Histology and histopathological scoring
Mouse colons were flushed with cold PBS to remove all contents, and the entire colon was rolled into Swiss rolls. The tissues were fixed in 2% PFA for 12 h at 4 °C and then washed in 1x PBS and transferred to 70% ethanol until embedding. Swiss rolls were embedded in the same manner as described above for segments of mouse colon. Paraffin-embedded, PFA fixed tissues were sectioned at 4.5 μm. Tissue sections were de-paraffinized and hematoxylin and eosin stained (Mayer’s Hematoxylin, Thermo Scientific; Eosin 1%, Morphisto, Germany). Histopathological analyses were performed using a semi-quantitative scoring system78 that evaluated the severity of crypt damage and cellular infiltration, epithelial erosion and tissue thickening using a severity score from 0 to 3 (0 = intact, 1 = mild, 2 = moderate, 3 = severe), and those scores were multiplied by a score for percent involvement (0 = 0%, 1 = 1–25%, 2 = 26–50%, 3 = 50–100%). A trained and blinded scientist performed the scoring. Representative images were acquired using an Olympus CKX53 microscope and Olympus SC50 camera.
Quantification of C. difficile toxin TcdB
Levels of TcdB in mouse colon contents were quantified relative to a standard curve of purified TcdB using an ELISA assay kit (“Separate detection of C. difficile toxins A and B”, TGC Biomics) according to the manufacturer’s instructions. For each mouse, approximately 10 mg of colon content were used in the assay. The limit of detection for the assay in our conditions was determined to be 5.14 ng of TcdB per gram of colon content (Supplementary Fig. 8c). One of the mice from the BacMix group had toxin levels below the detection limit and was therefore excluded from analysis.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article. More
