Ribose utilization by the human commensal Bifidobacterium breve UCC2003.

Growth of Bifidobacterium breve UCC2003 on ribose leads to the transcriptional induction of the rbsACBDK gene cluster. Generation and phenotypic analysis of an rbsA insertion mutant established that the rbs gene cluster is essential for ribose utilization, and that its transcription is likely regulated by a LacI-type regulator encoded by rbsR, located immediately upstream of rbsA. Gel mobility shift assays using purified RbsR(His) indicate that the promoter upstream of rbsABCDK is negatively controlled by RbsR(His) binding to an 18 bp inverted repeat and that RbsR(His) binding activity is modulated by D-ribose. The rbsK gene of the rbs operon of B. breve UCC2003 was shown to specify a ribokinase (EC 2.7.1.15), which specifically directs its phosphorylating activity towards D-ribose, converting this pentose sugar to ribose-5-phosphate.

Introduction

Bifidobacteria are Gram‐positive, pleomorphic, anaerobic bacteria, representing one of the dominant components of the intestinal microbiota of humans and other mammals (Picard ; Turroni ). Members of this genus are believed to play a positive role in host health through protection against infectious diseases, their beneficial influence on the immune system and active role in the maintenance of gut function (Tuohy ; Guarner, 2006; Parracho ; Turroni ). Their reported beneficial effects furthermore include prevention of diarrhoea, symptom alleviation of inflammatory bowel disease and irritable bowel syndrome, reduction of the risk of colorectal cancer and easing lactose maldigestion (Tuohy ; Leahy ; Gibson, 2008; de Vrese and Schrezenmeir, 2008).

Various carbohydrates have been shown or are claimed to selectively stimulate the growth and/or activity of (beneficial) (bifido)bacteria, a property for which the term prebiotic was coined (Gibson, 2008; Macfarlane ; Ramirez‐Farias ). A prebiotic was originally defined as ‘a selectively fermented ingredient that allows specific changes, both in the composition and/or activity in the gastrointestinal microflora that confers benefits upon host wellbeing and health’ (Gibson ). Fructo‐oligosaccharides (FOS) and galacto‐oligosaccharides (GOS) are the most commonly used prebiotics on the market in Europe and Japan. Soybean oligosaccharides (SOS), isomalto‐oligosaccharides (IMO) and xylo‐oligosaccharides (XOS) represent promising prebiotic compounds that have not yet been widely marketed (Gibson, 2008). Interestingly, from available genome sequences it has become clear that intestinal bifidobacteria dedicate a significant portion of their coding capacity (approximately 8% of all annotated genes) to the metabolism of a wide variety of carbohydrates (Schell ; Ventura ). Over 50 different bifidobacterial carbohydrases have been mentioned in the literature to date (recently reviewed by van den Broek ). Using Bifidobacterium breve UCC2003 as a model to study bifidobacterial carbohydrate metabolism, we previously characterized an operon encoding a β‐fructofuranosidase (Ryan ), an extracellular amylopullulanase which hydrolyses α‐1,4 and α‐1,6 glucosidic linkages in starch and related polysaccharides (Ryan ; O'Connell‐Motherway ), a galA gene involved in the degradation of potato galactan (O'Connell‐Motherway ) and two novel α‐glucosidases exhibiting hydrolytic activities towards panose, isomaltose, isomaltotriose and trehalose (Pokusaeva ). Moreover, a phosphoenolpyruvate phosphotransferase (PEP‐PTS) system involved in fructose metabolism was identified and studied in this bacterium (Maze ).

d‐ribose, a component of DNA (present as deoxyribose), RNA and many cofactors, may be found in the human gut (Walker ) and therefore could represent a carbon source for components of the gastrointestinal microbiota. In order to expand our knowledge on bifidobacterial carbohydrate metabolism, we characterized ribose utilization in B. breve UCC2003. The genetic elements involved in ribose uptake from the environment and its subsequent phosphorylation have been described previously for several bacteria, including Escherichia coli, Bacillus subtilis, Lactobacillus sakei and Corynebacterium glutamicum (Lopilato ; Mauzy and Hermodson, 1992; Woodson and Devine, 1994; Sigrell ; Stentz and Zagorec, 1999; Tourneux ; Chuvikovsky ; Ogbunude ; Nentwich ). In the current study, we describe the identification of the ribose utilization (rbs) operon in B. breve UCC2003. We present the biochemical characterization of the rbsK gene, whose protein product is responsible for the phosphorylation of d‐ribose. Furthermore, we present data that imply the LacI‐type regulator RbsR in the regulation of the rbs promoter in a d‐ribose‐dependent manner.

Results

Growth of Bifidobacterium strains on glucose and/or ribose

In order to investigate if bifidobacteria are capable of ribose metabolism, growth in modified Rogosa medium (mMRS) supplemented with glucose or ribose was assessed for 36 different bifidobacterial strains (Table S1, in Supporting information), representing 11 bifidobacterial species, by measuring the OD600 following 24 h of anaerobic growth at 37°C ( Fig. 1). All bifidobacterial strains grew well on glucose, reaching OD600 values in excess of 1. In contrast, just 18 out of the 36 strains tested were able to reach an OD600 higher than 1.0 when grown on ribose as the sole carbon source, seven of which belong to the B. breve species, including B. breve UCC2003. Eight strains were shown to grow rather poorly on ribose with final OD600 values ranging from 0.4 to 0.9, while the remaining eleven bifidobacterial strains could not or very poorly metabolize ribose (OD600 < 0.4). Only particular bifidobacterial strains are thus capable of utilizing ribose, which means that this sugar can be considered as a selective growth substrate for such strains.

Figure 1

Final optical density (OD600) values obtained following 24 h growth of 36 different bifidobacterial strains in modified MRS containing 0.01 g ml−1 glucose or ribose as the sole carbon source. The results are mean values obtained from three separate experiments.

Genome response of B. breve UCC2003 to growth on ribose

In order to investigate differences in global gene expression upon growth of B. breve UCC2003 on ribose, or a combination of ribose and glucose, as compared with glucose, DNA microarray experiments were performed. Total RNA was isolated from B. breve UCC2003 cultures grown on ribose, a combination of ribose and glucose, or glucose (see Experimental procedures in Supporting information).

Analysis of the DNA microarray data obtained from two independent biological replicates revealed that the expression of the adjacent rbsA, rbsC, rbsB, rbsD and rbsK genes was significantly upregulated (fold change > 3.0, P < 0.001) in B. breve UCC2003 cultures grown on ribose or a combination of ribose and glucose, whereas rbsR was downregulated, relative to cultures grown on glucose (Table 1). These results implicate the rbsRACBDK gene cluster in ribose metabolism in B. breve UCC2003. Furthermore, various genes encoding components of the PEP‐PTS system (Maze ) and genes encoding a number of carbohydrate ABC transport systems were significantly downregulated when B. breve UCC2003 was grown in ribose (results not shown).

Table 1

Effect of d‐ribose on the transcriptome of B. breve UCC2003.

Locus tag_gene	Putative function	Ribosea	Ribose + glucosea	qRT‐PCRb
Bbr_1415_rbsK	Ribokinase	8.70	3.30	36.30
Bbr_1416_rbsD	Ribose mutarotase	6.55	3.54	31.88
Bbr_1417_rbsB	d‐ribose‐binding protein	8.09	4.82	21.20
Bbr_1418_rbsC	Ribose transport system permease protein	12.15	4.33	22.30
Bbr_1419_rbsA	Ribose transport system ATP‐binding protein	11.37	4.64	18.24
Bbr_1420_rbsR	Transcriptional regulator, LacI family	−1.16	−1.25	−3.09

Genes for which transcription was changed when test cultures were grown on ribose or combination of ribose and glucose and reference cultures were grown on glucose. The numbers represent the expression ratios of ribose versus glucose or a combination of ribose and glucose versus glucose‐grown cells. Positive values indicate upregulation; negative values indicate downregulation. Expression ratios presented in bold have a Bayesian P‐value < 0.001 according to the Cyber t‐test (Long ).

Expression ratios of selected genes quantified by qRT‐PCR. cDNA templates were derived from RNA samples of B. breve UCC2003 culture grown on ribose or glucose as a negative control. A 2−ΔΔCt method was used to calculate relative changes in gene expression (Livak and Schmittgen, 2001). Expression ratios presented in bold have standard deviation values < 0.12 and a Mann–Whitney P‐value < 0.0022 according to the t‐test (Livak and Schmittgen, 2001).

To confirm the microarray results, quantitative RT‐PCR (qRT‐PCR) analysis was performed using primer pairs representing individual genes of the rbs gene cluster (Table S2 in Supporting information). cDNA templates were derived from RNA isolated from B. breve UCC2003 following growth on ribose, a combination of ribose and glucose, or glucose. As expected, the rbsACBDK genes were shown to be highly upregulated, while the rbsR gene was downregulated, consistent with the obtained microarrays results (Table 1).

Genetic organization of the putative ribose utilization operon

Our presumption, based on the microarray analysis, that the rbs gene cluster is involved in ribose utilization was supported by the high level of sequence similarity between rbsRACBDK and ribose‐specific metabolic genes from other bacteria. The genetic organization of the rbs gene cluster on the chromosome of B. breve UCC2003 and its comparison with similar clusters found in other bacteria are schematically displayed in Fig. 2. Due to its adjacent location (Fig. 2) and similarity to LacI‐type transcriptional repressors, the protein product of the rbsR gene was suspected to represent the regulator of the rbsACBDK operon of B. breve UCC2003. The rbsA, rbsC and rbsB genes, which specify the presumptive ribose uptake system, encode putative ATP‐binding, ribose permease and ribose‐binding proteins respectively. A putative ribose mutarotase, which is an enzyme involved in catalysing the conversion of the β‐pyran form into β‐furan of ribose (Kim ), and a putative ribokinase are encoded by the rbsD and rbsK genes respectively. The gene order of the B. breve UCC2003 rbs gene cluster differs from rbs clusters that have previously been characterized from E. coli (rbsDACBKR), B. subtilis (rbsRKDACB) and C. glutamicum (rbsRACBD..rbsK1..rbsK2) (Lopilato ; Woodson and Devine, 1994; Nentwich ). Furthermore, in some species, for example L. sakei, ribose uptake is not performed by an ABC transporter but by a secondary transport system (encoded by rbsU in L. sakei) (Stentz and Zagorec, 1999). Comparative genome analysis showed that the B. breve UCC2003 rbs operon is most similar to the similarly organized putative rbs gene cluster in the Bifidobacterium dentium ATCC27678 genome (Fig. 2). A predicted ribokinase‐encoding gene is present in the genomes of the three sequenced Bifidobacterium longum strains, B. breve DSM20213 (also known as ATCC15700) as well as Bifidobacterium adolescentis DSM20083 (also known as ATCC15703), but the ABC transporter, ribose mutarotase and repressor‐encoding genes (except for that of B. longum NCC2705) appear to be absent from the immediate vicinity of each of these rbsK genes. Remarkably, the ability of B. adolescentis DSM20083 to metabolize ribose as its sole carbon source indicates that it possesses a ribose uptake system, which is significantly different from that encoded by the B. breve UCC2003 genome.

Figure 2

Comparison of the rbs locus of B. breve UCC2003 with corresponding (putative or proven) ribose utilization loci from other bacteria. Each solid arrow indicates an open reading frame (ORF). The lengths of the arrows are proportional to the length of the predicted ORF and the gene locus name, which is indicative of its putative function, is indicated within the arrow. Orthologues are marked with the same colour while the amino acid identity of each predicted protein is indicated as a percentage relative to its equivalent protein encoded by B. breve UCC2003. The bent arrow indicates putative or proven promoter region; lollipop sign designates putative or proven rho‐independent terminator region.

Prevalence of rbs genes in B. breve genomes

In order to investigate whether homologues of the rbs genes are present in other B. breve strains, comparative genome hybridization analysis was performed using microarrays based on the B. breve UCC2003 genome and genomic DNA from eight B. breve strains (Table 2). Bifidobacterium breve UCC2005, NCFB2258, NCIMB8815 and UCC2004 were shown to possess homologous DNA representing all genes of the rbs cluster. The presumed presence of the rbs gene cluster in these strains correlates with their ability to grow on ribose as a sole carbon source (Fig. 1). Bifidobacterium breve JCM7017 and JCM7019 appear to harbour only a homologue of the rbsC gene, while the B. breve NCFB2257 and NCTC11815 genomes did not seem to contain sequences with significant identity to any of the rbs genes (Table 2). The CGH analysis is in agreement with growth profiles of B. breve JCM7017, NCFB2257 and NCTC11815, which were shown to grow very poorly on ribose. In contrast, B. breve JCM7019 is very well capable of metabolizing ribose, suggesting that this strain may have an alternative, non‐homologous system for ribose utilization.

Table 2

Prevalence of rbs genes in B. breve genomes.

Gene	B. breve strain number
	UCC2005	JCM7017	JCM7019	NCFB2257	NCFB2258	NCIMB8815	NCTC11815	UCC2004
rbsR	+	−	−	−	+	+	−	+
rbsA	+	−	−	−	+	+	−	+
rbsB	+	−	−	−	+	+	−	+
rbsC	+	+	+	−	+	+	−	+
rbsD	+	−	−	−	+	+	−	+
rbsK	+	−	−	−	+	+	−	+

The plus sign (+) indicates the presence of the gene; the minus sign (−) the absence of the gene. Genes were considered absent when the fold ratio was < 0.2 and P < 0.001.

Complementation of ribose metabolism

In order to provide further evidence for the presumed involvement of the B. breve UCC2003 rbs operon in ribose metabolism we constructed a pWSK29‐derivative plasmid, designated pWRbsK, which harbours the rbsK gene of rbs operon from B. breve UCC2003. This construct and the pWSK29 negative control were introduced into E. coli JW3731‐1, which is a mutant strain that carries a deletion in the ribokinase‐encoding gene, making it unable to grow in a medium that contains ribose as its sole carbon source. The resultant pWRbsK‐containing E. coli JW3731‐1 strain, in contrast to its pWSK29‐containing equivalent, was shown to complement the growth defect of E. coli JW3731‐1 on a medium containing ribose as the sole carbon and energy source (Fig. S1 in Supporting information). These results demonstrate that the rbsK‐encoding gene from B. breve UCC2003 is responsible for phosphorylation of ribose.

Disruption of the rbsA gene in B. breve UCC2003

In order to establish if disruption of the rbsACBK operon in B. breve UCC2003 would result in loss of this strain's ability to metabolize ribose, an rbsA insertion mutant was generated (Fig. S2A in Supporting information). To verify the expected ribose‐negative phenotype of the B. breve UCC2003::rbsA insertion mutant strain, both the wild type and the insertion mutant strain were analysed for their ability to grow in mMRS supplemented with ribose or glucose (positive control) as the sole carbon source (Fig. S2B in Supporting information).

As expected, and in contrast to the wild type, the B. breve UCC2003::rbsA insertion mutant was shown to be incapable of growth on ribose as the sole carbon source. Although the rbsA disruption in B. breve UCC2003::rbsA is likely to have a polar effect on the transcription of the downstream genes of the rbs operon, it nevertheless demonstrates that (elements of) the rbs gene cluster is uniquely necessary for ribose metabolism in B. breve UCC2003.

Transcriptional analysis of the rbs gene cluster

In B. breve UCC2003, the rbsA, rbsC, rbsB, rbsD and rbsK genes (Fig. 3B) are presumed to be expressed as a single transcript, based on their similar expression patterns as determined from microarray analysis and RT‐PCR experiments, which showed that amplification products were generated from cDNA (generated from total RNA of ribose‐grown B. breve UCC2003) using various primer combinations that spanned the individual genes of the rbs gene cluster (data not shown). Furthermore, the only predicted Rho‐independent terminator structure in the DNA region that harbours the very closely organized rbsACBDK genes is present downstream of the rbsK gene (Fig. 3B). The transcriptional start site of the presumed rbsABCDK operon was determined by primer extension analysis (see Experimental procedures in Supporting information) and it was shown to be located 125 bp upstream of the predicted rbsA start codon (Fig. 3A, I and II) and 7 bp downstream of sequences resembling consensus −10 and −35 sequences of a vegetative promoter.

Figure 3

A. Schematic representation of the B. breve UCC2003 rbs promoter region (I). Boldface type and underlining indicate the −10 and −35 hexamers as deduced from the primer extension results (II) and ribosomal binding site (RBS); the transcriptional start site (TSS) is indicated by an asterisk; arrows under sequence in bold indicate the inverted repeat sequence that represents the RbsR binding sequence. The R6 fragment (see legend to B) is indicated as a broken thick black line underneath the relevant sequence.
B. Representation of the B. breve UCC2003 rbs operon and DNA fragments used in electrophoretic mobility shift assays (EMSAs). Plus and minus signs indicate whether or not RbsR was able to bind to the particular DNA fragment respectively.
C. EMSA showing RbsRHis interaction with DNA fragments encompassing fragment R1 (I); R1 fragment with the addition of d‐ribose‐5‐phosphate (10 µM) (II); or R1 fragment upon addition of d‐ribose (10 µM) (III). In each panel lane X represents a binding reaction to which no protein was added, while the remaining lanes represent binding reactions with the R1 DNA probe incubated with increasing amounts of RbsRHis[concentrations ranging from 0.04 nM (for I) (or 0.08 nM for II and III) to 0.01 µM]. Each successive lane, from left to right, corresponds to a doubling in the concentration of RbsRHis.

Determination of RbsRHis DNA binding site

The helix–turn–helix motif of the N‐terminus of RbsR is characteristic for members of the LacI‐type superfamily of transcriptional factors (Nguyen and Saier, 1995), which are usually responsible for transcriptional regulation of genes involved in carbohydrate metabolism (Perez‐Rueda and Collado‐Vides, 2000). In order to determine if RbsR is capable of direct interaction with the rbs promoter, we first cloned the rbsR gene in the nisin‐inducible vector pNZ8048 with the introduction of a His‐tag‐encoding sequence to facilitate purification (see Experimental procedures in Supporting information). The purified RbsRHis protein was then used to perform electrophoretic mobility shift assays (EMSAs) (Fig. 3C), which clearly demonstrated that this protein was able to form a complex with IRD800‐labelled DNA fragments encompassing the rbsABCDK promoter region (Fig. 3B and C, I). The results showed that a 56 bp region, which was commonly present in DNA fragments R1, R2 and R3, contained a presumed RbsR‐specific operator sequence. In support of this notion, we showed that fragments R4 and R5, which do not contain this RbsR DNA‐binding sequence, were not bound by RbsRHis (data not shown). These results were further validated by an EMSA performed with a shorter, 56 bp fragment, R6, which contained this operator sequence, and which was shown to be bound by RbsRHis (results are not shown). Introduction of two point mutations in the putative RbsR‐binding motif (a G→T and a C→T mutation in positions +7 and +22, respectively, relative to the transcription start site which was taken as +1, see Fig. 3A, I), corresponding to positions that were previously shown to play an important role in the binding affinity of other LacI‐type regulatory proteins (Kim and Chambliss, 1997), was shown to prevent binding of RbsRHis (data not shown). As with other LacI‐type regulators, it was assumed that the RbsR binding activity will depend on the presence or absence of a particular sugar ligand, which would be identical or related to the carbohydrate that induces the expression of the rbs operon (Weickert and Adhya, 1992). In order to identify the RbsR ligand molecule, EMSAs were performed upon adding (10 mM final concentration) d‐ribose or ribose‐5‐phosphate to the binding reaction. The results obtained clearly showed that the binding ability of RbsRHis for the R1 fragment decreased about fourfold in the presence of d‐ribose (approximately 50% of the R1 fragment was bound in the presence of 2.5 nM of protein), while ribose‐5‐phosphate did not affect RbsRHis binding to the proposed operator sequence (0.625 nM of protein was required for an approximate 50% shift of the R1 target) (Fig. 3C, II and III). EMSAs of other intergenic regions within rbs operon and its surroundings did not result in any shift (data not shown) suggesting the presence of a single operator sequence that is regulated by RbsR in B. breve UCC2003.

Identification and characterization of RbsK sequence and purified RbsKHis from B. breve UCC2003

The rbsK gene is 1002 bp in length and encodes a protein of 333 amino acids (molecular weight ∼34.4 kDa). blastp and multiple sequence alignments showed that RbsK is similar to putative and characterized ribokinases encoded by various different microorganisms, with the highest level of identity – 95%, to a hypothetical protein encoded by B. breve DSM20213, and 91% and 90% identity to putative PfkB family sugar kinases from B. longum NCC2705 and B. longum ssp. infantis CCUG52486 respectively (Table S3 in Supporting information). Overall, the deduced RbsK protein from B. breve UCC2003 exhibits varying levels of similarity (ranging from 53% to 69% identity) to 16 putative ribokinases, encoded by 13 different bifidobacterial genomes and all belonging to the ribokinase family (EC number 2.7.1.15). clustal w alignment of putative or proven RbsK enzymes from a variety of different bacteria revealed conserved amino acids known to be involved in substrate and ATP binding (see Fig. S3 in Supporting information) (Sigrell ).

In order to characterize the B. breve UCC2003 RbsK ribokinase, we purified and biochemically characterized a His‐tagged version of this protein, designated RbsKHis (see Experimental procedures). The molecular mass of RbsKHis was estimated by SDS‐PAGE to be approximately 38 kDa (see Fig. S4 in Supporting information), which is in good agreement with the deduced molecular mass of 36.2 kDa.

Using ribose and ATP as substrates the expected ribokinase activity of the rbsK gene product was confirmed by NMR (see Experimental procedures), which also allowed characterization of the phosphorylated product of the bifidobacterial ribokinase. The obtained 1H‐NMR spectrum of the reaction displayed two signals (5.43 and 5.27 ppm, Fig. 4A), in the region where the anomeric protons of sugars typically appear. These two resonances partially overlap those of ribose at 5.42 and 5.29 ppm corresponding to the α‐furanosic and β‐furanosic conformations, respectively (data not shown), whereas no resonances corresponding to the pyranosic forms were detected (β‐pyranoribose, 4.88 ppm; α‐pyranoribose, 4.82 ppm). The proton‐decoupled 31P‐NMR spectrum showed two partially overlapping resonances (4.49 and 4.47 ppm, Fig. 4B, magnification) in the phosphomonoester region. These resonances appeared as triplets in a proton‐coupled 31P‐NMR spectrum, indicating that the phosphate group is in the vicinity of two protons as in carbon 5 (CH2). Altogether, these data led us conclude that the product of the ribokinase reaction was ribose‐5‐phosphate, which was confirmed by comparing 1H‐ and 31P‐spectra with those derived from commercially obtained ribose‐5‐phosphate.

Figure 4

1H‐NMR (A) and 31P‐NMR (B) spectra of the ribokinase reaction. The sample was prepared in 2H2O as described in Experimental procedures and the spectra were acquired at 33°C. Proton and phosphorus spectra were processed with 0.1 and 1 Hz line broadenings. Rib5P, ribose‐5‐phosphate; the peaks indicated with an asterisk represent ribose. The inset (in B) shows magnification of the region between 4.49 and 4.47 ppm.

In order to investigate substrate specificity of the recombinant RbsKHis from B. breve UCC2003 possible phosphorylated products were analysed by 31P‐NMR and/or by monitoring ATP hydrolysis using an ATP Bioluminoscent assay kit (see Experimental procedures for details). The B. breve UCC2003 ribokinase was shown to specifically phosphorylate d‐(‐)‐ribose, but failed to phosphorylate other tested pentoses, hexoses or more complex carbon sources under the conditions used in this study (see Table S4 in Supporting information).

Interestingly, purified RbsKHis enzyme was shown to be active at a rather broad pH spectrum, retaining 100% relative activity at pH 5.5–8.0 (results not shown).

Kinetic parameters (Vmax and Km) for ribose and ATP were determined using 31P‐NMR spectroscopy (see Experimental procedures for details). The ribose saturation curve of RbsKHis showed a hyperbolic kinetic response. The Km and Vmax values were determined to be 2.9 ± 0.4 mM and 231.6 ± 12 µmol min−1 mg−1 respectively. Curiously, the ATP saturation curve showed a sigmoidal profile. The deduced Km for ATP was 5.9 ± 0.2 mM, while the Hill coefficient was 1.4.

Discussion

d‐ribose is one of the most abundant carbohydrates found in nature. More specifically, it is thought to be present in the intestinal tract as it is a component of RNA which is one of the dominant macromolecules in every living cell being released following cell death and subsequent degradation (Walker ). It has been previously shown that, among others (for a review see Rediers ), E. coli genes responsible for ribose transport are specifically expressed in the gastrointestinal tract (GIT) of gnotobiotic mice (Alpert ). Furthermore, the genes predicted to encode ribose permease (rbsC) and ribokinase (rbsK3) of Lactobacillus plantarum have been shown to be induced in GIT of BALB/c mice (Bron ), while transcription of rbsR, which regulates ribose utilization, was detected in vivo in Klebsiella pneumoniae (Lai ) and Salmonella enterica serovar typhi (Daigle ). These reports strongly suggest that ribose is a commonly accessible carbon source in the gut environment.

Within the above context it is important to understand how natural inhabitants of the human gut have adapted to this particular environment through their ability to utilize available carbon sources. In ribose‐utilizing bifidobacteria it is expected that, once a ribose molecule is taken up and phosphorylated to ribose‐5‐phosphate, it can directly enter the ‘bifid shunt’ (de Vries ; Sanchez ), while it can also be used for the biosynthesis of nucleotides and histidine (Eriksen ).

This report describes the characterization of the ribose utilization cluster, rbsACBDK, and its presumed regulator, encoded by rbsR, of B. breve UCC2003. The involvement of rbsK in ribose metabolism was confirmed by phenotypic complementation of an rbsK‐negative E. coli mutant in a ribose‐containing medium. The B. breve UCC2003::rbsA insertion mutant confirmed that the rbs operon described in this study is the only or at least the predominant system required for ribose metabolism in B. breve UCC2003.

Transcription of the rbsACBDK operon was shown to be repressed when B. breve UCC2003 was grown in modified Rogosa medium containing glucose, while growth in medium containing ribose or a combination of ribose and glucose caused clear transcriptional upregulation of this operon. However, the expression level of the genes of the rbs operon was decreased by approximately twofold (Table 1) when the culture was grown in a combination of glucose and ribose in mMRS (relative to the expression levels obtained from a culture grown in ribose only), thereby suggesting that expression of the rbs operon was downregulated to some degree by catabolite repression in the presence of a highly metabolizable sugar such as glucose. PEP‐PTS‐regulated repression of ribose utilization has been observed in L. sakei (Stentz and Zagorec, 1999) and, although PEP‐PTS‐mediated carbon catabolite repression (CCR) system is common in bacteria such as E. coli and B. subtilis (for reviews see Deutscher ; Gorke and Stulke, 2008), B. breve UCC2003 does not appear to encode the key regulators present in these paradigm systems (S.C. Leahy, M. O'Connell‐Motherway, J.A. Moreno Muñoz, G.F. Fitzgerald, D.G. Higgins and D. van Sinderen, unpubl. data). In this context it is interesting to note that a glucose kinase from the high G+C Gram‐positive bacterium Streptomyces coelicolor was shown to be involved in CCR (Gorke and Stulke, 2008), while B. longum NCC2705 exhibits a lactose‐over‐glucose preference representing an example of reverse CCR, the mechanism of which is as yet not fully understood (Parche ). Our results suggest that B. breve UCC2003 possesses an unusually regulated CCR system thus warranting a more specific investigation to gain a better understanding of the mechanism involved.

The RbsR protein of B. breve UCC2003, which was shown to bind to the promoter region of the rbsABCDK operon, is presumed to act as an effective transcriptional repressor in the absence of the effector molecule, thereby preventing transcription. The RbsR‐specific binding site was shown to be located five nucleotides downstream of the rbs operon transcriptional start site. So therefore binding of RbsR is expected to prevent the RNA polymerase to proceed across this region, thus blocking transcription of the rbs operon. The B. breve UCC2003 RbsR recognition sequence is similar to RbsR binding sites from E. coli K‐12 (Mauzy and Hermodson, 1992), B. subtilis (Woodson and Devine, 1994) and L. sakei (Stentz and Zagorec, 1999) (data not shown). In addition, our RbsR binding results indicate that rbs transcription and therefore ribose metabolism are controlled by d‐ribose, as the affinity of RbsR for its DNA target decreases, while the transcription of rbsR and therefore the assumed amount of RbsR are also reduced in the presence of this sugar. This observation is consistent with results from a previous study of RbsR regulation in E. coli (Mauzy and Hermodson, 1992). In contrast, the binding activity RbsR from B. subtilis and C. glutamicum does not appear to be affected by d‐ribose (Strauch, 1995; Nentwich ).

Comparative sequence analysis of rbs gene clusters from other bacteria showed that the genetic organization of the B. breve UCC2003 rbs operon is similar to those found in E. coli and B. subtilis. Interestingly, blast searches showed that most of the sequenced bifidobacterial genomes have at least one copy of an apparent rbsK homologue, although no complete rbs operon appears to be present in any of the currently available bifidobacterial genomes (Fig. 2). An exception to this observation is B. dentium ATCC27678, whose genome contains a similarly organized rbs operon, although the rbsK homologue appears to be disconnected from the rest of the gene cluster, similar to that of the recently described rbs operon of C. glutamicum (Nentwich ). The results obtained from comparative sequence and genome hybridization analyses may indicate that rbs genes could have been lost in the chromosomes of some of the analysed strains that only retained parts of the rbs cluster. It could also mean that other bifidobacteria, such as B. adolescentis DSM20083, B. breve JCM7017, JCM7019 and NCFB2257, use a different ribose uptake system which has not yet been described or alternatively, these strains may encode a secondary transport‐type protein for ribose uptake which is not homologous to the known RbsU protein from L. sakei (Stentz and Zagorec, 1999).

The ribokinase, encoded by rbsK, of B. breve UCC2003 was shown to preferentially phosphorylate d‐ribose. The presence of ribokinase activity in bifidobacteria has not previously been reported, but this enzyme has been well studied in E. coli, Leishmania major and S. enterica (Tourneux ; Chuvikovsky ; Ogbunude ). Our results demonstrate that the phosphorylation product of RbsKHis from B. breve UCC2003 is d‐ribose‐5‐phosphate (Fig. 4) and that d‐ribose represents the only phosphorylation substrate (see Table S4 in Supporting information). This is consistent with a previous report on RbsK from L. major (Ogbunude ), but contrasts with characterized ribokinases from E. coli and S. enterica serovar typhi, which were shown to have broader specificity (Chuvikovsky ). Differences in primary structure between these enzymes may explain their different substrate preferences and kinetic parameters.

Previously, B. breve UCC2003 was shown to metabolize a number of different carbon sources (Ryan ; O'Connell‐Motherway ; Pokusaeva ). The results presented here provide the first in‐depth characterization of pentose metabolism in bifidobacteria and extends our understanding of carbohydrate utilization by our model bifidobacterial strain B. breve UCC2003, showing its adaptation to the available carbon sources present in the human and animal intestine.

Experimental procedures

The description of the experimental procedures can be found in Appendix S1 in Supporting information.