Bacterial strain, media, and growth conditions
B. xylanisolvens XB1AT (DSM 18836T) was grown anaerobically at 37 °C in BX medium containing clarified rumen fluid  and 5 g/L of complex substrates (insoluble oat-spelt xylan or OSX, Serva, France; washed twice in distilled water and autoclaved to remove free sugars, and soluble wheat arabinoxylan or WAX, Megazyme, France) or sugars (glucose or xylose). The media were prepared, dispensed and inoculated by using strictly anaerobic techniques in Balch tubes. A 2.5 % (v/v) inoculum of culture pre-adapted on each substrate was used for inoculation. Bacterial growth on glucose, xylose and WAX was followed by optical density of the culture at 600 nm (OD600nm) recorded directly in Balch tubes using a Jenway 6320D spectrophotometer. Because OSX interfered with OD600 measurements, growth on this substrate was monitored by estimating the amount of bacterial proteins in the culture using the Bradford Protein Assay . Three independent cultures were performed for each substrate condition i.e. glucose, xylose, WAX and OSX for subsequent transcriptomic and/or proteomic analyses (Table 1).
OSX and WAX composition
The analysis of OSX was carried out as previously described  in order to determine its composition in monosaccharides, methylester groups and starch. Uronic acids were measured spectrophotometrically by the m-hydroxydiphenyl assay using galacturonic acid as external standard. The difference in response of glucuronic acid and galacturonic acid in the presence and absence of tetraborate was used to quantify them in OSX (Additional file 1: Table S4) . The composition of OSX used in this study is comparable with a glucuronoxylan poorly substituted with arabinose (Xyl/GlcA/Ara 84/8/1); OSX also contains small amounts of starch (0.9 %) and possibly pectin, although pectin has never been quantified in OSX [28–30].
The WAX substrate used in this study corresponded to medium viscosity WAX (lot 40302b, Megazyme) and its composition was provided by the supplier. It is 95 % pure with a Xyl/Ara sugar ratio of 62/38, and contains 3.7 % proteins in addition to traces of starch (0.09 %) and β-glucan (0.1 %).
Preparation of enriched fractions of mRNAs
Total RNAs were isolated from cultures harvested at mid- and late-log phase using a modified guanidinium–phenol–chloroform procedure previously described for rumen fluid . Briefly, bacterial cultures (4 tubes × 8 ml) were centrifuged for 15 min at 3,000 g at 4 °C. The pellets were resuspended in 9 ml of a RNA-E solution containing solution D , water saturated phenol, sodium acetate 0.2 M pH 4.0 and 2-mercaptoethanol (1:1:0.1:0.007). Cells were then disrupted by bead beating for 1 min with 0.1 g zirconia beads (0.1 mm) followed by a 2-min incubation at 60 °C. These two steps were then repeated. After addition of 3.75 ml of chloroform, the samples were briefly mixed, incubated for 15 min on ice and centrifuged (12,000 g, 20 min, 10 °C). The RNAs contained in the aqueous supernatants (approximately 6 ml) were precipitated with 0.25 volume isopropanol and washed with 1 volume 75 % cold ethanol in DEPC-treated water. Total RNAs were solubilized in 100 μl of DEPC-treated water. Genomic DNA was removed using the Turbo DNA-Free DNAse (Ambion, France) for 30 min at 37 °C. RNAs were quantified using a ND-2000 NanoDrop spectrophotometer (Nanodrop Technologies, France). Enriched fractions of mRNAs were prepared using the MicrobExpress™ Bacterial mRNA Purification kit (Ambion, France). The high RNA quality and the reduction in 16S and 23S rRNA in enriched fractions of mRNAs were confirmed using an Agilent 2100 Bioanalyser (Agilent technologies, France).
cDNA libraries were prepared with 100 ng of mRNA-enriched fractions following the protocols of the Illumina TruSeq Stranded Total RNA Library prep kit. The final libraries had an average fragment size of ∼ 250 bp and were quantified by qPCR before being sequenced with an Illumina HiSeq 2000 instrument on a single lane in paired end reads. The data are available in GEO datasets at NCBI (http://www.ncbi.nlm.nih.gov/) under the accession number GSE74379. Approximately 8 to 9 million paired reads per sample were obtained (Additional file 1: Table S1). Quality filtering and adapter trimming were performed with Trimmomatic v0.30  using Illumina TruSeq3 adapter sequences for adapter clipping.
The B. xylanisolvens XB1AT genome (GenBank accession NC_021017.1) was indexed using novoindex, and reads aligned with novoalign v. 3.00.05 (http://www.novocraft.com) against the indexed genome. For downstream gene expression analysis only the aligned R1 reads were used; these were extracted from the paired-end alignment file using samtools v0.1.19  using the bitwise flag for the first read for a read pair.
Gene counts were determined for the aligned data using featureCounts v. 1.4.3-p1  and the NCBI GFF3 feature file (obtained from the NCBI FTP site Dec. 2013).
Differential gene expression was performed using R 3.0.0 using edgeR/limma [36, 37]. Samples were assessed for potential outliers based on counts using normalized counts (RPKM = Reads per Kilobase per Million) that would affect downstream analyses. From this analysis we determined that samples mRNA 1, 2, 3 and 10 were problematic and thus removed them from the analysis (Additional file 1: Table S1). Counts per million (CPM) mapped reads were calculated per gene; genes with more 1 CPM in three or more samples were retained in the final edgeR analysis. Samples were normalized using edgeR’s TMM normalization. A simple generalized linear model was generated using the aforementioned filtered data from all remaining samples, and simple contrasts based on carbon source and growth stage were used to determine genes differentially expressed under the conditions shown. A separate coordinate analysis was performed using Rockhopper v. 1.30  on all mRNA samples (excluding mRNA 1, 2, 3 and 10) again using data obtained from NCBI GFF3 feature file as mentioned above. Results from this analysis were primarily used to find potential groups of genes that may be expressed as a single transcriptional units or operons.
In silico prediction of operons from genomic sequences
Putative promoters and terminators were searched within intergenic sequences (>100 bp) using different tools (BPROM, PPP, Arnold) available at http://molbiol-tools.ca/Promoters.htm. Operon prediction was carried out using FGENESB, which is based on distances between ORFs and frequencies of different genes neighboring each other in known bacterial genomes, as well as on promoter and terminator predictions (http://www.softberry.com/berry.phtml?topic=fgenesb&group=programs&subgroup=gfindb).
Reverse transcription (RT) followed by PCR or quantitative PCR (qPCR)
Total RNAs (1 μg) were reverse-transcribed into cDNAs using random hexamer primers (Invitrogen, France) and 200 U SuperscriptII Rnase H− reverse transcriptase (Invitrogen, France) according to the procedure supplied with the enzyme. For each RNA sample, a negative RT (no addition of reverse transcriptase) was performed and used as a negative control in subsequent PCR and qPCRs.
The presence of a polycistronic mRNA transcribed from PUL 43 of B. xylanisolvens XB1AT was determined by performing PCR using cDNAs prepared from bacterial cultures on xylan and primer pairs designed to amplify the intergenic regions between two consecutive ORFs within the locus i.e. forward primers targeting the 3’ end of one ORF and reverse primers targeting the 5’ end of the following ORF (Additional file 1: Table S5). PCR was carried out with the HotMaster Taq DNA polymerase (5PRIME, Deutschland) following the instructions given for the enzyme.
The relative expression of PUL target genes in the OSX culture condition (mid and late-log phases) versus the sugar conditions was performed by quantitative PCR using a Mastercycler ep Realplex 2S (Eppendorf, France) and Quantifast SYBR Green PCR mastermix (Qiagen, France) using the supplier’s instructions. The designed specific primers are listed in the Additional file 1: Table S6. The fold change in gene expression (OSX versus glucose or xylose) was calculated from 3 biological replicates (+ two technical replicates) according to Livak and Schmittgen  using the 16S rRNA reference gene for normalization. Log2 fold-change at mid- and late-log phase were considered as significantly different at p < 0.01 (Student’s t-test).
Preparation of bacterial soluble proteins
Bacterial soluble proteins (= cell-associated proteins) were prepared from xylose and OSX cultures (500 ml) harvested at late-log phase. Bacterial cell pellets were first washed 3 times in distilled water containing a protease inhibitor (Complete EDTA free protease inhibitor cocktail, Roche, France) and re-suspended in 1/25 volume of the same solution. Cells were broken by one passage at 2000 bars using a One Shot cell Disruptor (CellD, France). The obtained suspension was treated with 270 U endonucleases (Dnase + RNase, Sigma, France) for 30 min at ambient temperature. Unbroken cells were removed by centrifugation (23,000 g, 40 min, 4 °C). The supernatant was subsequently submitted to ultracentrifugation (100,000 g, 1 h, 4 °C) to obtain a final supernatant containing bacterial soluble proteins that were aliquoted and frozen. Before analysis, an aliquot of protein solution was mixed with 3 volumes of ice-cold acetone and kept at -20 °C for at least 2 h. After centrifugation (13,000 g, 30 min, 4 °C), the protein pellet was dried under vacuum for 5 min and solubilized in isoelectric focusing (IEF) buffer (7 M urea, 2 M thiourea, 4 % CHAPS, 0.2 % triton X100). The amount of proteins solubilised in IEF buffer was determined with the PlusOne™ 2-D Quant kit (Amersham, France) according to the supplier’s instructions.
Two-dimensional gel electrophoresis (2-DE) and Image analysis
2-DE was performed with IPG strips of 17 cm with a linear gradient of pH 4-7 (Bio-Rad, France). They were submitted to passive (9 h) and active (9 h, 50 V) rehydration with 300 μl IEF buffer containing 0.4 % DTT, 0.15 (v/v) % Biolyte pH 4–6, 0.15 (v/v) % Biolyte pH 5–7 and 100 μg proteins. IPG strip rehydration and IEF were conducted in a focusing tray using the Bio-Rad PROTEAN® IEF System at a temperature of 20 °C. Focusing conditions were 30 min at 200 V, 1 h at 1 kV, linear voltage ramp to 10 kV for 6 h followed by a plateau at 10 kV until total 54 kV.h was reached. Focused IPG strips were stored at –20 °C in glass tubes. Prior to the second dimension, strips were equilibrated twice for 30 min in an equilibration solution (50 mM Tris–HCl pH 8.8; 6 M urea; 30 % v/v glycerol, 2 % w/v SDS) containing 130 mM DTT for the first step and 135 mM iodoacetamide and trace of bromophenol blue for the second step. The second dimension (SDS-PAGE) was carried out with 12 % polyacrylamide gels in a PROTEAN® II XL Cell (Bio-Rad, France). Proteins were visualized by silver-staining according to Bradford et al. .
Silver stained 2-DE gels were scanned with a GS800 imaging densitometer (Bio-Rad, France) and spot detection and image analysis were performed with the PD-Quest software (version 7.1, Bio-Rad). Three reproducible gels from each independent culture for each bacterial growth condition (OSX, xylose) were selected and included for image analysis.
Statistical image analysis was performed using Progenesis Samespots software, version 4.1 (Nonlinear Dynamics). For each independent analysis, gels were aligned on a reference gel (xylose condition). After background subtraction, spot autodetection and quantification were performed by the software. Each spot was assigned a relative value corresponding to a spot volume. To compare the OSX to the xylose condition, an experimental design was set up with these two conditions, where corresponding gels were added. Differential spot intensity was considered significant at p < 0.01 using ANOVA (Analysis of Variance) procedure.
Identification of proteins by mass spectrometry
Spots of interest were excised in the 100 μg loaded silver-stained gels and subjected to the following treatments. First, the spots were unstained in a 30 mM potassium ferricyanid-100 mM sodium thiosulfate solution for 2 min and rinsed twice with MilliQ water for 15 min. They were then washed in 25 mM ammonium bicarbonate (pH 8.5)–5 % acetonitrile for 30 min and twice in 25 mM ammonium bicarbonate–50 % acetonitrile for 30 min each. The spots were then dehydrated with 100 % acetonitrile. The dried gels were reswelled in 25 mM ammonium bicarbonate containing 20 ng/μL− 1 trypsin. Digestion was performed at 37 °C for at least 5 h. The resulting peptides were extracted with 100 % acetonitrile. After 15 min at 37 °C, each sample was mixed with saturated cyano-4-hydroxycinnamic acid onto the matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) target. Using MALDI-TOF mass spectrometry (MS) (Voyager DE-Pro, Applied BioSystems) and Voyager software for data collection and analysis, positive-ion MALDI mass spectra were recorded in the reflectron mode. Identifications by nano-liquid chromatography (LC) coupled to electrospray ionization (ESI) and tandem mass spectrometry (MS/MS) (LTQVelos, Thermo Scientific) were performed when identification from MALDI-TOF MS failed. Monoisotopic peptide masses were assigned and used for database searches with Mascot v2.2.0. Interrogations were performed against a home database containing distinct entries corresponding to the predicted mature proteins in B. xylanisolvens XB1AT. The following parameters were considered for the searches: a maximum ion mass tolerance of 25 or 50 ppm, possible modification of cysteines by carbamidomethylation, as well as partial oxidation of methionine.
Construction of a HTCS mutant within PUL 43
An insertion mutation was created into the sensor/regulator HTCS gene (BXY_29350) of PUL 43. An internal fragment corresponding to the sensor domain of the protein (841 bp) encoded by BXY_29350 was cloned into the pGERM suicide vector (Additional file 1: Table S7 and Figure S4). The resulting construct was transformed into Escherichia coli WM3064 and transferred to B. xylanisolvens XB1AT by conjugation as previously described . Plasmid insertion into the target gene was then verified by PCR using primers targeting junction regions between pGERM and PUL 43 HTCS gene (Additional file 1: Table S7 and Figure S4).
Rumen fluid used to prepare growth media was collected in the experimental slaughterhouse at INRA, Saint-Genes-Champanelle, France, from animals slaughtered in accordance with the guidelines of the local Ethics Committee and current INRA ethical guidelines for animal welfare (Permit number: 63345001).
Consent to publish
Availability of data and materials
The data sets supporting the results of this article are included within the article and its additional supplementary file. The transcriptome data are available in GEO datasets at NCBI (http://www.ncbi.nlm.nih.gov/) under the accession number GSE74379.