Cloning, expression and characterization of a β-d-xylosidase from Lactobacillus rossiae DSM 15814T

Background Among the oligosaccharides that may positively affect the gut microbiota, xylo-oligosaccharides (XOS) and arabinoxylan oligosaccharides (AXOS) possess promising functional properties. Ingestion of XOS has been reported to contribute to anti-oxidant, anti-bacterial, immune-modulatory and anti-diabetic activities. Because of the structural complexity and chemical heterogeneity, complete degradation of xylan-containing plant polymers requires the synergistic activity of several enzymes. Endo-xylanases and β-d-xylosidases, collectively termed xylanases, represent the two key enzymes responsible for the sequential hydrolysis of xylan. Xylanase cocktails are used on an industrial scale for biotechnological purposes. Lactobacillus rossiae DSM 15814T can utilize an extensive set of carbon sources, an ability that is likely to contribute to its adaptive ability. In this study, the capacity of this strain to utilize XOS, xylan, d-xylose and l-arabinose was investigated. Results Genomic and transcriptomic analyses revealed the presence of two gene clusters, designated xyl and ara, encoding proteins predicted to be responsible for XOS uptake and hydrolysis and d-xylose utilization, and l-arabinose metabolism, respectively. The deduced amino acid sequence of one of the genes of the xyl gene cluster, LROS_1108 (designated here as xylA), shows high similarity to (predicted) β-d-xylosidases encoded by various lactic acid bacteria, and belongs to glycosyl hydrolase family 43. Heterologously expressed XylA was shown to completely hydrolyse XOS to xylose and showed optimal activity at pH 6.0 and 40 °C. Furthermore, β-d-xylosidase activity of L. rossiae DSM 15814T was also measured under sourdough conditions. Conclusions This study highlights the ability of L. rossiae DSM 15814T to utilize XOS, which is a very useful trait when selecting starters with specific metabolic performances for sourdough fermentation or as probiotics. Electronic supplementary material The online version of this article (doi:10.1186/s12934-016-0473-z) contains supplementary material, which is available to authorized users.


Background
In recent years the use of prebiotics, in particular oligosaccharides, to modulate the gut microbiota composition and associated metabolic activities, together aimed at improving gut health, has enjoyed considerable scientific and commercial interest [1,2]. Colonic fermentation of particular oligosaccharides into short chain fatty acids (SCFA) is believed to increase the number and metabolic activity of certain beneficial bacterial populations [1,3]. Among the oligosaccharides that may positively alter the composition of the gut microbiota, xylo-oligosaccharides (XOS) and arabinoxylan oligosaccharides (AXOS) possess promising functional properties as they can be specifically fermented by intestinal commensals such as bifidobacteria and lactobacilli [4][5][6]. XOS resist hydrolysis by enzymes and/or the low pH present in human saliva, gastric and pancreatic juices, and are not absorbed during transit through the small intestine, thus reaching the colon where they are available as fermentable substrates for certain members of the

Open Access
Microbial Cell Factories *Correspondence: raffaella.dicagno@uniba.it 1 Department of Soil, Plant and Food Science, University of Bari Aldo Moro, Via G. Amendola 165/A, 70126 Bari, Italy Full list of author information is available at the end of the article resident microbiota [7]. It has previously been reported that ingestion of XOS contributes to anti-oxidant, antibacterial, immune-modulatory and anti-diabetic activities [8,9]. Arabinoxylans represent 50 % of dietary fibers present in wheat bran [10,11]. In particular, the highly polymerized arabinoxylan from bran may be hydrolyzed by specific bacteria possessing arabinoxylan-degrading enzymes [6].
In nature, a variety of microorganisms, including bacteria and fungi, encode xylanases and β-xylosidases [14]. These enzymes are either cell wall-associated, secreted into the extracellular environment or present in the cytoplasm. Based on amino acid sequence similarities xylanases and β-xylosidases are classified into glycoside hydrolase (GH) families 1, 3,30,39,43,51,52,54,116 or 120. In particular, β-xylosidases belonging to the GH43 family, which are predominantly encoded by bacteria, hydrolyze the non-reducing ends of the xylooligomers using an inverting mechanism [14]. Despite the recognized potential of XOS as a prebiotic to target beneficial components of the human gut microbiota, in particular lactobacilli and bifidobacteria, very little is known about the enzymes used by the former group of bacteria to hydrolyze these complex substrates. To the best of our knowledge, β-d-xylosidase from Lactobacillus brevis is the only report on the characterization of such an enzyme in the Lactobacillus genus [9,19].
Recently, the genomic annotation and comparative analysis of L. rossiae DSM 15814 T revealed the predicted presence of a number of extracellular or cell wall-associated polysaccharide-degrading enzymes, represented by putative cyclomaltodextrinase (E.C. 3 [20]. Furthermore, genes involved in the degradation of arabinose and xylose-containing poly-and/or oligo-saccharides were predicted. L. rossiae is an obligatory hetero-fermentative lactic acid bacterium, which has been isolated from the gastrointestinal tract of humans [21] and animals [22], wheat sourdoughs [23][24][25], legumes [26], spelt flour [27] and pineapple [28]. L. rossiae was found to be a promising probiotic candidate thanks to its ability to survive under simulated gastric and intestinal conditions, and to stimulate immune-mediators by peripheral blood mononuclear cells [29]. Exposure to gastric and intestinal fluids is the main environmental stress that decreases viability of ingested probiotics [30,31]. More in depth knowledge on XOS metabolism by L. rossiae is important from a biotechnological perspective to facilitate the selection of strains with specific metabolic performances to be used as starters for sourdough bread making, aimed at improving rheology and nutritional properties, or as probiotic for human applications.
In the current study we used a transcriptome approach to identify L. rossiae DSM 15814 T genes that were upregulated when this strain was grown on XOS-, xylose-or arabinose. One of the identified genes, xylA, was selected for further characterization and was cloned in Lc. lactis subsp. cremoris NZ9000 and the encoded recombinant enzyme was purified and characterized.

Growth on XOS, xylan, d-xylose or l-arabinose
When maltose was used as a sole carbon and energy source in growth medium (see "Methods" section), L. rossiae was shown to increase its viable count from ca. 7.4 ± 0.1 to 9.4 ± 0.3 log CFU ml −1 . The stationary phase of growth was reached after approximately 10 h, with a lag phase and μmax of 2.9 ± 0.2 h and 0.52 ± 0.03 log CFU ml −1 h −1 , respectively. In the presence XOS, d-xylose or l-arabinose, L. rossiae was shown to exhibit similar growth kinetics. The increase of cell viability ranged from 7.3 ± 0.1 to 9.5 ± 0.3 log CFU ml −1 , the values of λ varied from 2.7 ± 0.3 (l-arabinose) to 2.5 ± 0.1 h (XOS), and those of μmax from 0.27 ± 0.02 (l-arabinose and XOS) to 0.31 ± 0.03 Log CFU ml −1 h −1 (d-xylose). L. rossiae was not shown to exhibit appreciable growth in the presence of xylan, rye arabinoxylan, wheat arabinoxylan, arabinan, arabinogalactan and xyloglucan, as the sole carbon source (data not shown).
Genome response of L. rossiae DSM 15814 T to growth on XOS In order to investigate which genes are expressed when L. rossiae DSM 15814 T is grown in presence of XOS, d-xylose, l-arabinose or maltose (as a reference) as the sole carbon source, global gene expression was determined by RNAseq analysis. Compared to growth on maltose as the sole carbon source, various adjacent genes (designated xylE, xylA, xynT, xylT, xylI, xylK and xylR; and here referred to as the xyl cluster) were shown to exhibit transcriptional increases that ranged from 8.6 to 250 fold, or from 11.4 to 259.3 fold, when the strain was grown on XOS ( Fig. 1a) or d-xylose, respectively. Furthermore, the co-located araA, araD, araB, araR and araRS genes, which encompass the ara gene cluster, predicted to encode enzymes for l-arabinose utilization, exhibited an increase in their transcription from 0.9 to 156 fold when l-arabinose was used as the only carbon source (Fig. 1b) (see below for details on putative functions).
Genetic organization of the ara and xyl custers, and comparative analysis of the xylA-encoded β-d-xylosidase As mentioned above, transcriptome analysis of L. rossiae DSM 15814 T revealed the presence of two gene clusters, xyl and ara, that are linked to XOS/d-xylose and l-arabinose metabolism, respectively (Fig. 2). The xyl cluster encompasses seven genes, which are schematically outlined in Fig. 2a, and which are predicted (based on BLAST-mediated similarity searches, Additional file 1: Table S1) to encode a xylose isomerase (E.C. 5.3.1.5; LROS_1111, designated here as xylI), a xylulose kinase (E.C 2.7.1.17; LROS_1112, designated here as xylK) and a ROK family transcriptional regulator (LROS_1113, designated here as xylR). Upstream of, but divergently oriented from the three genes mentioned above are genes encoding a predicted aldose 1-epimerase (E.C. 5. LROS_1107, designated here as xylE) and a putative β-dxylosidase (EC3.2.1.37; LROS_1108, designated as xylA), followed by xynT (a putative xyloside transporter-encoding gene; LROS_1109) and xylT (predicted to specify a d-xylose proton symporter; LROS_1110). Thus, the genes in this xyl cluster represent proteins that may be involved in XOS metabolism (Kyoto Encyclopedia of Genes and Genomes database; Fig. 2). Among the genes of the xyl cluster, xylA may have a key role in XOS hydrolysis as its encoded protein product is predicted to convert XOS into d-xylose. Alignment of the L. rossiae DSM 15814 T xylA-encoded β-d-xylosidase with similar protein sequences from other lactic acid bacteria (NCBI website) produced the phylogenetic tree displayed in Fig. 3. This analysis shows that the β-d-xylosidase from L. rossiae DSM 15814 T belongs to a cluster consisting of mostly predicted β-d-xylosidases encoded by Leuconostoc, Weissella and Pediococcus species, and Lc. lactis.

XylA expression in Lc. lactis subsp. cremoris NZ9000 and characterization of β-d-xylosidase activity
To demonstrate if xylA encodes, as predicted, a β-dxylosidase capable of XOS hydrolysis, this gene was cloned into the protein expression vector pNZ8048, and placed under the transcriptional control of the inducible P nisA promoter (see "Methods" section). In order to verify the hydrolytic activity of XylA, purified protein and crude cell extract (CCE) from Nisaplin-induced Lc. lactis subsp. cremoris, harbouring the pNZ8048-xylA were individually incubated for 24 h. Following incubation, CCE from Lc. lactis, harbouring the pNZ8048-xylA, showed hydrolytic activity towards XOS (Fig. 4). As expected, using identical experimental conditions, CCE from Lc. lactis, harbouring the empty pNZ8048 (negative control) did not exhibit measurable activity. In agreement with the comparative genome sequence analysis of L. rossiae, these results demonstrate that xylA specifies a β-d-xylosidase responsible for the observed XOSdegrading activity. Unfortunately, for unknown reasons the XylA protein completely lost its hydrolytic activity upon purification from the CCE (results not shown), and therefore further characterization of this enzyme was performed using XylA-containing CCE.
The effect of the pH on the activity of the recombinant His-tagged XylA was determined in Na-acetate (pH 3.0-6.0), phosphate (pH 6.0-7.0) and Tris-HCl (pH 7.0-9.0) buffer (Additional file 1: Figure S1). The enzyme was highly active (>80 %) in the range of pH of 5.0-7.0, displaying an optimum at pH 5.7. Above pH 7.0, enzyme activity rapidly decreased, being completely lost at pH 9.0. XylA enzyme activity was optimal at 40 °C, and it decreased thereafter (Additional file 1: Figure S1).
Using L. rossiae DSM 15814 T cells, enzyme activity was also determined under conditions that mimicked sourdough. After 24 h of fermentation, the xylosidase activity measured in sourdough was 4.4 ± 0.1 U, corresponding to 70 % of the maximum activity found in phosphate buffer. Following growth of L. rossiae DSM 15814 T in wheat flour hydrolysate medium (WFH), the CCE was shown to exhibit an activity of 5.6 ± 0.2 U.

Discussion
The relatively large genome (genome size ~ 2.8 Mb) of L. rossiae DSM 15814 T is believed to reflect the metabolic and adaptive versatility of the species, mirroring an impressive potential to colonize diverse environments [20]. L. rossiae DSM 15814 T was shown to grow well on modified Rogosa medium with XOS or its constituent pentose sugar xylose, or arabinose as the only carbon source. Growth of lactobacilli on XOS is not widely documented: only Lactobacillus fermentum (syn. L. cellobiosus) [32] and Lactobacillus acidophilus [33] have been reported to exhibit moderate growth on XOS, though this less vigorous compared to the growth capacity on this sugar shown by certain Bifidobacterium spp. [5,34]. Furthermore, a gut-derived strain of Lactobacillus paracasei was shown to be positively stimulated by XOS [35]. L. rossiae DSM 15814 T is predicted to specify a number of extracellular or cell wall-associated polysaccharide-degrading enzymes, as well as putative enzymatic pathways for the metabolism of arabinose and xylosecontaining poly-and/or oligosaccharides [20]. In this study a β-d-xylosidase was identified in L. rossiae DSM 15814 T and successfully expressed in Lc. lactis subsp. cremoris NZ9000. β-d-xylosidase represents one of the key xylanolytic enzymes in supplying carbon and energy to a variety of organisms [36,37]. Sequential hydrolysis of xylans leads to the release of xylose, which is then transformed to the common metabolic intermediate xylulose 5-phosphate [38]. In L. rossiae DSM 15814 T the genes involved in the xylose and xylo-oligosaccharides utilization pathway appear to be clustered in a single operon designated here as xyl. The deduced amino acid sequence of the LROS_1108 gene belonging to the xyl cluster and designated here as xylA, showed high similarity to (predicted) β-d-xylosidases encoded by various lactic acid bacteria, a glycosyl hydrolase family 43 enzyme [14]. Recombinantly produced XylA was shown to catalyse the complete hydrolysis of XOS to xylose. The presence of a highly specific xylosidase is expected to play a crucial role in facilitating the utilization of XOS. As shown by transcriptome analysis, all xyl cluster genes were variously up-regulated when L. rossiae DSM 15814 T was grown in the presence of XOS and D-xylose, compared to cells grown when maltose was the sole carbon source. Furthermore, the up-regulation of the ara gene cluster, predicted to encode enzymes for l-arabinose utilization, was observed when, as expected, this sugar was present as the sole carbon source. Although these findings provide insights into the genes required for growth on XOS, our data are insufficient to reconstruct the xylanolytic machinery of L. rossiae DSM 15814 T . Nonetheless, β-xylosidase is generally associated with other xylanolytic enzymes in order to impart an efficient conversion of xylo-oligosaccharides into xylose [36], and the involvement of other, perhaps secreted xylosidases cannot be ruled out.
The β-xylosidase of L. rossiae DSM 15814 T was shown to exhibit pH and temperature optima of ca. 6.0 and 40 °C, respectively. It was highly active (>80 %) in the range of pH of 5.0-7.0. β-xylosidases from other bacteria are usually active and stable at neutral pH [42] and a temperature range of 35-45 °C. The β-xylosidase purified from L. brevis NCDC01 was shown to exhibit pH and temperature optima of 6.0 and 40 °C [9], while the β-xylosidases of Bacillus thermantarcticus [43] and Bifidobacterium adolescentis were characterized by optimal activity at 5.0-7.0 and 50-70 °C [44]. Based on enzyme properties, the β-xylosidase of L. rossiae was found to be similar to that purified from L. brevis. L. rossiae is a novel species [23] that is able to adapt to different environments [21,[23][24][25]28], and therefore further insights are desirable to exploit its biotechnological features and understand its metabolic versatility.
β-xylosidase activity appears to be produced by L. rossiae DSM 15814 T during sourdough fermentation, even though at a level slightly lower than when the strain is cultivated in WFH.
Cocktails of xylanases including β-d-xylosidase are used to hydrolyze glycosidic linkages from xylan and/ or arabinoxylan liberating XOS in a first step, which are subsequently degraded releasing smaller water-soluble fragments, positively correlated with bread properties such as oven spring and volume [45]. Although L. rossiae was shown to be incapable of utilizing xylan and/or arabinoxylan, its β-xylosidase activity might contribute to the subsequent degradation of the XOS released by other xylanases.

Conclusions
The activity of the β-d-xylosidase encoded by L. rossiae DSM 15814 T allows this bacterium to effectively fulfil the role of a xylo-oligosaccharide (XOS)-metabolizing cell factory. This study highlights some biochemical traits of L. rossiae that may be exploited for biotechnological purposes such as its use as a starter for the sourdough process or as a potential probiotic. The XOS metabolic trait has no doubt implications for the environmental adaptation by L. rossiae DSM 15814 T , and functional genomic studies are underway to better understand XOS metabolism by L. rossiae.
Fermentations were carried out at 30 °C for 24 h. Growth kinetic data was determined by plating on SDB agar, and data were modeled according to the Gompertz equation as modified by Zwietering et al. [46]: y = k + A exp{−exp[(μ max e/A)(λ−t) + 1]}, where k is the initial level of the dependent variable to be modeled (log CFU ml −1 h −1 ), A is the difference in cell density between inoculation and the stationary phase, μ max is the maximum growth rate (expressed as Δ log CFU ml −1 h −1 ), λ is the length of the lag phase (expressed in hours), and t is the time. The experimental data was modeled through the non-linear regression procedure of the statistics package Statistic for Windows (Statsoft, Tulsa, Oklahoma, USA).

DNA manipulations and plasmid construction
Chromosomal DNA was isolated from L. rossiae DSM15814 T as previously described [48]. Plasmid DNA was isolated from Escherichia coli with the Roche High Pure plasmid isolation kit (Roche Diagnostics, Basel, Switzerland). An initial lysis step was performed with 30 mg ml −1 of lysozyme for 30 min at 37 °C prior to plasmid isolation from E. coli. The single-stranded oligonucleotide primers used in this study were synthesized by Eurofins (Ebersberg, Germany) ( Table 1). Gene functional annotations were made by BLAST. The primers used for cloning (1108CHisPstF/1108CHisXbaR) were designed so as to incorporate two different endonuclease enzyme sites (PstI and XbaI) in order to facilitate subsequent cloning in the expression vector pNZ8048 and an in-frame His6-encoding sequence into reverse primer  [49]. PCR fragments were purified employing the Roche High Pure PCR purification kit (Roche Diagnostics). Introduction of plasmid DNA into competent Lc. lactis by electroporation was carried out as previously reported [50]. The fragment encompassing xylA and the nisin-inducible translational fusion plasmid pNZ8048 were both digested with PstI and XbaI, and then ligated [51]. The ligation mixtures were introduced into Lc. lactis subsp. cremoris NZ9000 by electroporation, and transformants were selected based on chloramphenicol resistance. The plasmid contents of a number of chloramphenicol-resistant transformants were screened by restriction analysis, and the integrity of the plasmid, designated here as pNZ8048-xylA, retrieved from positively identified clones was verified by sequencing (MWG Biotech AG, Ebersberg, Germany). The β-D-xylosidase sequence of XylA from L. rossiae DSM 15814 T , as well as deduced β-d-xylosidase sequences from lactic acid bacteria belonging to Lactobacillus, Leuconostoc, Weissella and Lactococcus genera (GenBank database), were used for comparative analysis. Sequence alignments were carried out using the MultiAlign program and Clustal W [52]. Phylogenetic tree construction was performed using the neighbor-joining program from the Clustal X software (National Center for Biotechnology Information) and visualized with the TreeView program.

Heterologous XylA production
Nisin-inducible gene expression was performed as described previously [53]. When a gene is placed downstream of the inducible promoter P nisA of plasmid pNZ8048 [47], transcription of that gene will be induced following the addition of sub-inhibitory concentrations of nisin [54]. In brief, 400 ml of GM17 broth supplemented with 200 μl of chloramphenicol (10 mg ml −1 ) was inoculated with a 2 % inoculum of Lc. lactis strains harboring pNZ8048.xylA, followed by incubation at 30 °C until an OD 600 of 0.2 was reached. At this point, protein expression was induced by the addition of nisin (Nisaplin ™ , DuPont; 40 ng ml −1 ), followed by continued incubation for a further 4 h. Un-induced control was incubated as above without the addition of Nisaplin ™ . Cells were harvested by centrifugation (7000×g, 10 min), resuspended into buffer (50 mM Tris-HCl pH 8, 300 mM NaCl and 20 mM CaCl 2 ), and stored at −20 °C overnight until further use. The thawed cell suspension was incubated in lysis buffer 50 mM Tris-HCl pH 8, 300 mM NaCl, 50 mM CaCl 2 and 25 mg ml −1 lysozyme (Sigma-Aldrich, Ireland) at room temperature for 30 min, then sonicated in an ice bath at 200 W for 2.5 min (alternating between 30 s of sonication and 10 s cooling), and centrifuged (14,000×g for 25 min at 4 °C). The supernatant, containing the soluble protein fraction, and pellet (insoluble protein fraction and cell debris) were stored at 4 °C until further analysis. Aforementioned fractions were analyzed by SDS-polyacrylamide gel electrophoresis, followed by fixation and staining with Coomassie brilliant blue R-250. The apparent molecular weight of the protein was estimated by comparison with rainbow-pre-stained, low-molecular-weight protein markers (New England Biolabs, Herefordshire, United Kingdom). Protein concentration was determined by the Bradford method [55]. Protein purification of the crude cell extract (CCE) from Nisaplin-induced Lc. lactis cells containing pNZ8048-xylA was performed using Ni-NTA matrices according to the manufacturers' instructions (Qiagen).

Assessment of XOS hydrolysis by means of high-performance anion exchange chromatographypulsed amperometric detection (HPAEC-PAD)
To assess the β-d-xylosidase activity of XylA, 50 μl of purified XylA protein or CCE (6.27 mg ml −1 ) was incubated at 37 °C for 24 h with 0.5 % (wt/vol) XOS in 0.1 M 3-(N-morpholino) propanesulfonic acid 4-morpholinepropanesulfonic acid (MOPS) buffer (pH 7.0). The reaction was stopped by heat treatment at 99 °C for 2 min and the mixture was centrifuged at 12,000×g for 5 min. The hydrolysis of XOS was assessed by HPAEC-PAD analysis employing a Dionex ICS-3000 system (Dionex, Sunnyvale, CA, USA) as described by Watson et al. [56]. The end-product d-xylose was identified by comparison with the retention time of the standard (5 min).

β-d-xylosidase assay under in vitro sourdough conditions
The enzyme properties of His-tagged XylA were determined by assaying the amount of p-nitrophenol (pNP) released from p-nitrophenyl-β-d-xylopyranoside (pNPX) substrate at 37.5 ± 1 °C, as described by Lasrado and Gudipati [9]. The reaction mixture, consisting of 5 mM pNPX in 50 mM phosphate buffer (900 µl, pH 5.7), was incubated with 50 µl (6.27 mg ml −1 ) XylA-containing CCE for 10 min. The reaction was stopped by the addition of 100 µl of saturated solution of sodium tetraborate and the amount of pNP released was determined by measuring the absorbance at 410 nm. Since L. rossiae was isolated from sourdough, enzyme activity was also assayed in sourdough and in WFH as follows. First, 300 g of dough (DY, 160) was prepared with 187.5 g of flour and 112.5 g of water, containing L. rossiae cells at a final cell density of approximately log 7 CFU g −1 . The dough was supplemented with pNPX at a final concentration of 5 mM and incubated at 30 °C for 24 h. The amount of pNP released was determined by measuring the absorbance at 410 nm prior and after the incubation. The pNP was extracted using phosphate buffer (pH 6.0). Four milliliters of buffer was added to 1.6 g of dough, homogenized for 180 s and centrifuged for 10 min at 10,000×g at 4 °C. The activity was expressed as the increase of absorbance per min per mg of protein. WFH was prepared as described by Gobbetti [57]. The medium was inoculated with L. rossiae at a final cell density of approximately log 7 CFU ml −1 . After 24 h of fermentation at 30 °C, the CCE was obtained as described for Nisaplininduced cells and the enzyme activity was determined as described by Lasrado and Gudipati [9].

Effect of pH and temperature on β-d-xylosidase activity
The optimum pH for the XylA-mediated β-d-xylosidase activity was determined by incubating 50 μl CCE (6.2 mg ml −1 ) from Nisaplin ™ -induced Lc. lactis NZ9000 cells containing plasmid pNZ8048-xylA at 37 °C for 10 min with pNPX. The different values of pH were obtained using 0.05 M Na-acetate (3.0-6.0), phosphate (6.0-7.0) and Tris-HCl (7.0-9.0) buffer. The optimum temperature for XylA activity was determined at pH 6.0 (50 mM phosphate buffer), in the range of 20-70 °C using pNPX as a substrate (see previous paragraph for a detailed description of the employed assay).

RNA-seq analysis and phylogenetic study
Approximately 10 μg of total RNA was obtained from L. rossiae DSM15814 T cells grown in modified Rogosa medium, containing XOS, d-xylose, l-arabinose or maltose as the only carbon source. Ribosomal RNA was removed using the Ribo-Zero rRNA removal kit according to supplier's instructions (Epicentre, Madison, WI, USA). The rRNA depletion step was verified by a 2200 TapeStation (Agilent technologies, USA). Then, 100 ng of rRNA-depleted RNA was fragmented using RNaseIII (Life Technologies, USA) followed by size evaluation using a 2200 TapeStation (Agilent technologies, USA). A whole transcriptome library was constructed using the Ion Total-RNA Seq Kit v2 (Life Technologies, USA). Barcoded libraries were quantified by qRT-PCR and each library template was amplified on Ion Sphere Particles using Ion One Touch 200 Template Kit v2 (Life Technologies, USA). Samples were loaded into 316 Chips and sequenced on the PGM (Life Technologies, USA). Sequencing reads were depleted of adapters, quality filtered (with overall quality, quality window and length filters) and aligned to the L. rossiae DSM 15814 T reference genome through BWA [58]. The sequence data have been submitted to the Sequence Read Archive database of the National Center for Biotechnology Information under accession no.
SRP072553. Counts of reads that corresponded to ORFs were performed using HTSeq (http://www-huber.embl.de/ users/anders/HTSeq/doc/overview.html) and analysis of the count data was performed using the R package DESeq [59].

Nucleotide sequence analysis
Sequence data were obtained from the Artemis-mediated genome annotations of L. rossiae DSM15814 T [20]. Database searches were performed using the non-redundant sequence database accessible at the National Center for Biotechnology Information website (http://www.ncbi.nlm. nih.gov) with BLAST [60]. Sequence analysis was performed employing Seqbuilder and Seqman programs of the DNASTAR software package (DNASTAR, Madison, WI).