Global transcriptome response in Lactobacillus sakei during growth on ribose

Background Lactobacillus sakei is valuable in the fermentation of meat products and exhibits properties that allow for better preservation of meat and fish. On these substrates, glucose and ribose are the main carbon sources available for growth. We used a whole-genome microarray based on the genome sequence of L. sakei strain 23K to investigate the global transcriptome response of three L. sakei strains when grown on ribose compared with glucose. Results The function of the common regulated genes was mostly related to carbohydrate metabolism and transport. Decreased transcription of genes encoding enzymes involved in glucose metabolism and the L-lactate dehydrogenase was observed, but most of the genes showing differential expression were up-regulated. Especially transcription of genes directly involved in ribose catabolism, the phosphoketolase pathway, and in alternative fates of pyruvate increased. Interestingly, the methylglyoxal synthase gene, which encodes an enzyme unique for L. sakei among lactobacilli, was up-regulated. Ribose catabolism seems closely linked with catabolism of nucleosides. The deoxyribonucleoside synthesis operon transcriptional regulator gene was strongly up-regulated, as well as two gene clusters involved in nucleoside catabolism. One of the clusters included a ribokinase gene. Moreover, hprK encoding the HPr kinase/phosphatase, which plays a major role in the regulation of carbon metabolism and sugar transport, was up-regulated, as were genes encoding the general PTS enzyme I and the mannose-specific enzyme II complex (EIIman). Putative catabolite-responsive element (cre) sites were found in proximity to the promoter of several genes and operons affected by the change of carbon source. This could indicate regulation by a catabolite control protein A (CcpA)-mediated carbon catabolite repression (CCR) mechanism, possibly with the EIIman being indirectly involved. Conclusions Our data shows that the ribose uptake and catabolic machinery in L. sakei is highly regulated at the transcription level. A global regulation mechanism seems to permit a fine tuning of the expression of enzymes that control efficient exploitation of available carbon sources.


Background
The Lactobacillus sakei species belongs to the lactic acid bacteria (LAB), a group of Gram-positive organisms with a low G+C content which produce lactic acid as the main end product of carbohydrate fermentation. This trait has, throughout history, made LAB suitable for production of food. Acidification suppresses the growth and survival of undesirable spoilage bacteria and human pathogens. L. sakei is naturally associated with the meat and fish environment, and is important in the meat industry where it is used as starter culture for sausage fermentation [1,2]. The bacterium shows great potential as a protective culture and biopreservative to extend storage life and ensure microbial safety of meat and fish products [3][4][5][6]. The genome sequence of L. sakei strain 23K has revealed a metabolic repertoire which reflects the bacterium's adaption to meat products and the ability to flexibly use meat components [7]. Only a few carbohydrates are available in meat and fish, and L. sakei can utilize mainly glucose and ribose for growth, a utilization biased in favour of glucose [7][8][9]. The species has been observed as a transient member of the human gastrointestinal tract (GIT) [10,11], and ribose may be described as a commonly accessible carbon source in the gut environment [12]. Transit through the GIT of axenic mice gave mutant strains which grow faster on ribose compared with glucose [13].
Glucose is primarily transported and phosphorylated by the phosphoenolpyruvate (PEP)-dependent carbohydrate phosphotransferase system (PTS). A phosphorylation cascade is driven from PEP through the general components enzyme I (EI) and the histidine protein (HPr), then via the mannose-specific enzyme II complex (EII man ) to the incoming sugar. Moreover, glucose is fermented through glycolysis leading to lactate [7,8,14]. Ribose transport and subsequent phosphorylation are induced by the ribose itself and mediated by a ribose transporter (RbsU), a Dribose pyranase (RbsD), and a ribokinase (RbsK) encoded by rbsUDK, respectively. These genes form an operon with rbsR which encodes the local repressor RbsR [15,16]. The phosphoketolase pathway (PKP) is used for pentose fermentation ending with lactate and other end products [8,17]. L. sakei also has the ability to catabolize arginine, which is abundant in meat, and to catabolize the nucleosides inosine and adenine, a property which is uncommon among lactobacilli [7,18].
By proteomics, we recently identified proteins involved in ribose catabolism and the PKP to be over-expressed during growth on ribose compared with glucose, while several glycolytic enzymes were less expressed. Moreover, also enzymes involved in pyruvate-and glycerol/glycerolipid metabolism were over-expressed on ribose [19]. Bacteria often use carbon catabolite repression (CCR) in order to control hierarchical utilization of different carbon sources. In low G+C content Gram-positive bacteria, the dominant CCR pathway is mediated by the three main components: (1) catabolite control protein A (CcpA) transcriptional regulator; (2) the histidine protein (HPr); and (3) catabolite-responsive element (cre) DNA sites located in proximity to catabolic genes and operons, which are bound by CcpA [20][21][22][23]. The HPr protein has diverse regulatory functions in carbon metabolism depending on its phosphorylation state. In response to high throughput through glycolysis, the enzyme is phosphorylated at Ser46 by HPr kinase/phosphorylase (HPrK/P). This gives P-Ser-HPr which can bind to CcpA and convert it into its DNAbinding-competent conformation. However, when the concentration of glycolytic intermediates drop, the HPrK/ P dephosphorylates P-Ser-HPr [20,[22][23][24]. Under low glucose concentrations, HPr is phosphorylated by E1 of the PTS at His15 to give P-His-HPr, which has a catalytic function in the PTS and regulatory functions by phosphorylation of catabolic enzymes and transcriptional regulators with a PTS regulation domain (PRD). Several P-EIIBs also phosphorylate different types of non-PTS proteins and regulate their activities [20][21][22]. Evidence for regulatory processes resembling glucose repression was shown both during lactose utilization [25] and catabolism of arginine [26,27] in L. sakei. A cre site has been reported upstream of the rbs operon [28], thus CcpA could likely be acting on the rbs operon as well as other catabolic genes and operons in this bacterium.
In the present study, we use a microarray representing the L. sakei 23K genome and an additional set of sequenced L. sakei genes, to investigate the global transcriptome response of three L. sakei strains when grown on ribose compared with glucose. Moreover, we predict the frequency of cre sites presumed to be involved in CCR in the L. sakei 23K genome sequence. Our objective was to identify differentially expressed genes between growth on the two sugars, and to increase the understanding of how the primary metabolism is regulated.

Methods
Bacterial strains, media and growth conditions L. sakei 23K is a plasmid-cured sausage isolate [29], and its complete genome sequence has been published [7]. L. sakei LS 25 is a commercial starter culture strain for salami sausage [30]. L. sakei MF1053 originates from fermented fish (Norwegian "rakfisk") [9]. The strains were maintained at -80°C in MRS broth (Oxoid) supplemented with 20% glycerol. Growth experiments were performed in a defined medium for lactobacilli [31] supplemented with 0.5% glucose (DMLG) or 0.5% ribose + 0.02% glucose (DMLRg) as described previously [19]. Samples were extracted at three different days from independent DMLG and DMLRg cultures from each strain grown at 30°C to mid-exponential phase (OD 600 = 0.5-0.6) for a total of three sample sets (parallels).

Microarrays
The microarrays used have been described by Nyquist et al. [32], and a description is available at http://migale.jouy. inra.fr/sakei/Supplement.html/. 70-mer oligonucleotide probes representing the L. sakei strain 23K genome and an additional set of sequenced L. sakei genes were printed in three copies onto epoxy glass slides (Corning).

RNA extraction
Total RNA extraction was performed using the RNeasy Protect Mini Prep Kit (Qiagen) as described by Rud et al. [33]. The concentration and purity of the total RNA was analysed using NanoDrop ND-1000 (NanoDrop Technologies), and the quality using Agilent 2100 Bioanalyzer (Agilent Technologies). Sample criteria for further use in the transcriptome analysis were A 260 /A 280 ratio superior to 1.9 and 23S/16S RNA ratio superior to 1.6.
cDNA synthesis, labeling, and hybridization cDNA was synthesized and labeled with the Fairplay III Microarray Labeling Kit (Stratagene, Agilent Technologies) as described previously [34]. After labeling, unincorporated dyes were removed from the samples using the QIAQuick PCR purification kit (Qiagen). The following prehybridization, hybridization, washing, and drying of the arrays were performed in a Tecan HS 400 Pro hybridization station (Tecan) as described by Nyquist et al. [32]. For studying the carbon effects, samples from DMLG and DMLRg were co-hybridized for each of the three strains. Separate hybridizations were performed for each strain on all three biological parallels. In order to remove potential biases associated with labelling and subsequent scanning, a replicate hybridization was performed for each strain for one of the three parallels, where the Cy3 and Cy5 dyes (GE Healthcare) used during cDNA synthesis were swapped. The hybridized arrays were scanned at wavelengths 532 nm (Cy3) and 635 nm (Cy5) with a Tecan scanner LS (Tecan). GenePix Pro 6.0 (Molecular Devices) was used for image analysis, and spots were excluded based on slide or morphology abnormalities.

Microarray data analysis
Downstream analysis was done by the Limma package http://www.bioconductor.org in the R computing environment http://www.r-project.org. Pre-processing and normalization followed a standard procedure using methods described by Smyth & Speed [35], and testing for differential expressed genes were done by using a linear mixed model as described by Smyth [36]. A mixed-model approach was chosen to adequately describe betweenarray variation and still utilize probe-replicates (three replicates of each probe in each array). An empirical Bayes smoothing of gene-wise variances was conducted according to Smyth et al. [37], and for each gene the p-value was adjusted to control the false discovery rate (FDR), hence all p-values displayed are FDR-adjusted (often referred to as q-values in the literature).

Validation of microarray data by qRT-PCR analysis
The microarray results were validated on selected regulated genes for the LS 25 strain by quantitative real-time reverse transcriptase PCR (qRT-PCR) performed as described previously [38]. Primers and probes (Additional file 1, Table S3) were designed using Primer Express 3.0 (Applied Biosystems). Relative gene expression was calculated by the ΔC T method, using the DNA gyrase subunit alpha gene (gyrA) as the endogenous reference gene.

Microarray accession numbers
The microarray data have been deposited in the Array Express database http://www.ebi.ac.uk/arrayexpress/ under the accession numbers A-MEXP-1166 (array design) and E-MEXP-2892 (experiment).

Sequence analysis
A prediction of cre sites in the L. sakei 23K genome sequence (GeneBank acc. no. CR936503.1), both strands, was performed based on the consensus sequence TGWNANCGNTNWCA (W = A/T, N = A/T/G/C), confirmed in Gram-positive bacteria [39]. We made a search with the consensus sequence described by the regular expression T-G-[AT]-X-A-X-C-G-X-T-X-[AT]-C-A, allowing up to two mismatches in the conserved positions except for the two center position, highlighted in boldface. All computations were done in R http://www.r-project.org.

Results and Discussion
Selection of L. sakei strains and growth conditions We have previously investigated L. sakei strain variation [9], and used proteomics to study the bacterium's primary metabolism [19], providing us with a basis for choosing strains with interesting differences for further studies. The starter culture strain LS 25 showed the fastest growth rates in a variety of media, and together with strain MF1053 from fish, it fermented the highest number of carbohydrates [9]. The LS 25 strain belongs to the L. sakei subsp. sakei, whereas the 23K and MF1053 strains belong to L. sakei subsp. carnosus [9,19]. By identification of differentially expressed proteins caused by the change of carbon source from glucose to ribose, LS 25 seemed to down-regulate the glycolytic pathway more efficiently than other strains during growth on ribose [19]. For these reasons, LS 25 and MF1053 were chosen in addition to 23K for which the microarray is based on. Nyquist et al. [32] recently investigated the genomes of various L. sakei strains compared to the sequenced strain 23K by comparative genome hybridization (CGH) using the same microarray as in the present study. A large part of the 23K genes belongs to a common gene pool invariant in the species, and the status for each gene on the array is known for all the three strains [32].
As glucose is the preferred sugar, L. sakei grows faster when glucose is utilized as the sole carbon source compared with ribose [8,9,15]. However, glucose stimulates ribose uptake and a possible co-metabolism of these two sugars present in meat and fish has been suggested, a possibility that give the organism an advantage in competition with other microbiota [15,16,40]. To obtain comparable 2-DE gels between samples issued from bacteria grown on the two carbohydrates in our recent proteomic analysis, growth on ribose was enhanced by adding small amounts of glucose [19]. For the present transcriptome analysis we therefore chose the same growth conditions.

Global gene expression patterns
A microarray representing the L. sakei 23K genome and an additional set of sequenced L. sakei genes was used for studying the effect of carbon source on the transcriptome of L. sakei strains 23K, MF1053 and LS 25. Genes displaying a significant differential expression with a log 2 ratio > 0.5 or < -0.5 were classified into functional categories according to the L. sakei 23K genome database http://migale.jouy.inra.fr/sakei/genome-server and are listed in Table 1. The 23K strain showed differential  expression for 364 genes within these limits, MF1053 and  LS 25 for 223 and 316 genes, respectively. Among these,  88, 47 and 82, respectively, were genes belonging to the category of genes of 'unknown' function. Eighty three genes, the expression of which varied depending on the carbon source, were common to the three strains, among which 52 were up-regulated and 31 down-regulated during growth on ribose ( Figure 1). The function of these common regulated genes was mostly related to carbohydrate transport and metabolism (34 genes, Table 1). The reliability of the microarray results was assessed by qRT-PCR analysis using selected regulated genes in the LS 25 strain. As shown in Table S4 in the additional material (Additional file 1), the qRT-PCR results were in agreement with the data obtained by the microarrays.
Several of the up-regulated genes are located in operons, an organisation believed to provide the advantage of coordinated regulation. In addition, in order to discriminate genes induced by growth on ribose from those repressed by glucose (submitted to CCR mediated by CcpA), a search of the complete genome sequence of L. sakei 23K [7] was undertaken, with the aim to identify putative cre sites. The search revealed 1962 hits, most of which did not have any biological significance considering their unsuitable location in relation to promoters. Relief of CcpAmediated CCR likely occur for many of the up-regulated genes in the category of carbohydrate transport and metabolism. Putative cre sites were identified in their promoter region, as well as for some genes involved in nucleoside and amino acid transport and metabolism (Table 2). In the other gene categories, the presences of putative cre sites were rare. With regard to gene product, the L. sakei genome shares high level of conservation with Lactobacillus plantarum [7], and high similarity of catabolic operon organization. The role of CcpA in CCR in L. plantarum has been established, and was shown to mediate regulation of the pox genes encoding pyruvate oxidases [41,42]. During growth on ribose, L. plantarum induces a similar set of genes as observed in the present study, and putative cre sites were identified in the upstream region of several genes involved [33].

Ribose catabolism and PKP
Confirming its major role in ribose transport and utilization in L. sakei, and in agreement with previous findings [16], our microarray data revealed a strong up-regulation (Table 1; log 2 = 2.8-4.3) of rbsUDK. The genes encoding an additional putative carbohydrate kinase belonging to the ribokinase family and a putative phosphoribosyl isomerase, lsa0254 and lsa0255, respectively, previously suggested to be involved in catabolism of ribose in L. sakei [7], were induced in all the strains (Table 1). Recent CGH studies revealed that some L. sakei strains which were able to grow on ribose did not harbour the rbsK gene, whereas lsa0254 was present in all strains investigated [32]. This second ribokinase could therefore function as the main ribokinase in some L. sakei strains. The rbsK sequence could also differ considerably from that of 23K in these strains. The PKP showed an obvious induction with an up-regulation (2.2-3.2) of the xpk gene encoding the key enzyme xylulose-5-phosphate phosphoketolase (Xpk). This enzyme connects the upper part of the PKP to the lower part of glycolysis by converting xylulose-5-phosphate into glyceraldehyde-3-phosphate and acetyl-phosphate. Acetyl-phosphate is then converted to acetate and ATP by acetate kinase (Ack). Supporting our results, previous proteomic analysis showed an over-expression of RbsK, RbsD and Xpk during growth on ribose [15,16,19]. The induction of ribose transport and phosphorylation, and increased phosphoketolase and acetate kinase activities were previously observed during growth on ribose [15]. Three genes encoding Ack are present in the 23K genome [7], as well as in MF1053 and LS 25 [32]. A preferential expression of different ack genes for the acetate kinase activity seem to exist. The ack2 gene was up-regulated in all the strains, while ack1 was up-regulated and ack3 downregulated in 23K and LS 25 (Table 1). An illustration of the metabolic pathways with genes affected by the change of carbon source from glucose to ribose in L. sakei is shown in Figure 2.
As a consequence of the pentose-induced PKP, genes involved in PKP-metabolism of glucose, such as gntZ, gntK and zwf, were down-regulated (Table 1, Figure 2). The glycolytic pathway was clearly repressed, supporting previous findings [15,19]. Among these genes were pfk (0.5-1.1) encoding 6-phosphofructokinase (Pfk), and fba (0.7-1.1) coding for fructose-bisphosphate aldolase, both acting at the initial steps of glycolysis. In addition, gpm3 encoding one of the five phosphoglycerate mutases present in the 23K genome, acting in the lower part of glycolysis, was also down-regulated (0.7-0.9). MF1053 down-regulated pyk (0.7) encoding pyruvate kinase (Pyk) that competes for PEP with the PTS (Figure 2). Its activity results in the production of pyruvate and ATP, and it is of major importance in glycolysis and energy production in the cell. MF1053 also showed a stronger downregulation of pfk than the other strains (Table 1). Similar to several other lactobacilli, pfk is transcribed together with pyk [43,44], and in many microorganisms the glycolytic flux depends on the activity of the two enzymes encoded from this operon [43,45]. At the protein level, we previously observed both Pfk and Pyk expressed at a lower level for all the three strains [19], however this was not confirmed at the level of gene expression for 23K and LS 25. We could also not confirm the lower Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0.5 (log 2 values > 1.0 or < -1.0 are shown in bold)  Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0. 5 Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0. 5 Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0. 5 Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0. 5 Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0.5 (log 2 values > 1.0 or < -1.0 are shown in bold) Secondary metabolites transport and metabolism Transport/binding proteins and lipoproteins LSA0046 lsa0046 Putative transport protein Putative drug transport protein -2.   Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0.  Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0. 5 Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0. 5 Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0. 5 Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0. 5  protein expression of glyceraldehyde-3-phosphate dehydrogenase, phosphoglycerate kinase and enolase previously seen in LS 25 [19]. The latter three enzymes are encoded from the central glycolytic operon (cggR-gappgk-tpi-eno) together with triose-phosphate isomerase and the putative central glycolytic genes regulator (CggR) [46]. Besides the cggR gene being down-regulated in MF1053 and LS 25, no change in gene expression was seen of these central glycolytic genes. Thus at the transcription level it is not obvious that the LS 25 strain down-regulate the glycolytic pathway more efficiently than the other strains, as previously suggested [19]. Interestingly, all the strains showed an induction (1.4-2.3) of mgsA encoding methylglyoxal synthase, which catalyzes the conversion of dihydroxyacetone-phosphate to methylglyoxal ( Figure 2). The presence of this gene is uncommon among LAB and so far a unique feature among the sequenced lactobacilli. The methylglyoxal pathway represents an energetically unfavourable bypass to the glycolysis. In E. coli, this bypass occurs as a response to phosphate starvation or uncontrolled carbohydrate metabolism, and enhanced ribose uptake was shown to lead to the accumulation of methylglyoxal [47,48]. As suggested by Chaillou et al. [7], such flexibility in the glycolytic process in L. sakei may reflect the requirement to deal with glucose starvation or to modulate carbon flux during cometabolism of alternative carbon sources. Breakdown of methylglyoxal is important as it is toxic to the cells [49]. An induction of the lsa1158 gene contiguous with mgsA was seen for 23K and MF1053. This gene encodes a hypothetical protein, also suggested as a putative oxidoreductase, which may reduce methylglyoxal to lactaldehyde [7]. However, no induction of the adhE (lsa0379) gene encoding an iron-containing aldehyde dehydrogenase suggested to further reduce lactaldehyde to L-lactate [7] was seen. By CGH [32]lsa1158 and adhE were present in all Table 1 Genes with significant differential expression in three L. sakei strains grown on ribose compared with glucose, FDR adjusted p-value less than 0.01 and log 2 of > 0.5 or < -0.5 (log 2 values > 1.0 or < -1.0 are shown in bold) (Continued) The microarray used has been described previously [32]. Asterix (*) relates the gene to Table 2. D and U refer to genes classified as 'divergent' and 'uncertain', respectively, by CGH analysis [32]. Genes encoding proteins with a change in expression according to McLeod et al. [19], are underlined. the L. sakei strains investigated, whereas mgsA was lacking in some strains, indicating that the MgsA function is not vital.

Pyruvate metabolism
Pyruvate is important in both glycolysis and PKP. It can be converted into lactate by the NAD-dependent L-lactate dehydrogenase, which regenerates NAD + and maintains the redox balance. This enzyme is encoded by the ldhL gene which was down-regulated (0.7-1.4) in all three strains, in accordance with previous findings [50], and the down-regulation was strongest for the LS 25 strain. At the protein level, only LS 25 showed a lower expression of this enzyme during growth on ribose [19].
oxygen species and reacts readily with metal ions to yield hydroxyl radicals that damage DNA, proteins and membranes [53]. Remarkable differences in redox activities exist among Lactobacillus species and L. sakei is among those extensively well equipped to cope with changing oxygen conditions, as well as dealing effectively with toxic oxygen byproducts [7]. 23K up-regulated npr (1.0) encoding NADH peroxidase which decomposes low concentrations of H 2 O 2 to H 2 O and O 2 , and all the strains upregulated the sodA gene (1.7-3.4) encoding a superoxide dismutase which produces hydrogen peroxide from superoxide (O 2 -). Various oxidoreductases showed an up-regulation in all the strains (Table 1), indicating the need for the bacterium to maintain its redox balance.
The pdhABCD gene cluster encoding components of the pyruvate dehydrogenase enzyme complex (PDC) which transforms pyruvate into acetyl-CoA and CO 2 were among the strongly up-regulated (2.1-3.7) genes. The eutD gene encoding a phosphate acetyltransferase which further forms acetyl-phosphate from acetyl-CoA was also induced (1.0-2.0). Pyruvate can be transformed to acetolactate by acetolactate synthase and further to acetoin by acetolactate decarboxylase, before 2,3-butanediol may be formed by an acetoin recuctase (Figure 2). While the budC gene encoding the acetoin reductase showed a strong up-regulation in all three strains, the als-aldB operon was only strongly up-regulated in LS 25 (1.9). Pyruvate formate lyase produces acetyl-CoA and formate from pyruvate. Only in 23K, the pflAB genes encoding formate C-acetyltransferase and its activating enzyme involved in formate formation were strongly upregulated (4.0 and 1.7, respectively). This strain was the only one to strongly induce L-lactate oxidase encoding genes which are responsible for conversion of lactate to acetate when oxygen is present (Table 1). In 23K and LS 25, the ppdK gene coding for the pyruvate phosphate dikinase involved in regenerating PEP, was induced, as was also lsa0444 encoding a putative malate dehydrogenase that catalyzes the conversion of malate into oxaloacetate using NAD + and vice versa (Table 1).
During growth on ribose, L. sakei was shown to require thiamine (vitamine B1) [15]. The E1 component subunit α of the PDC, as well as Pox and Xpk, require thiamine pyrophosphate, the active form of thiamine, as a coenzyme [54]. This could explain the induction of the thiMDE operon and lsa0055 in LS 25, as well as lsa0980 in 23K, encoding enzymes involved in thiamine uptake and biosynthesis ( Table 1). The up-regulation of lsa1664 (1.1-1.6) encoding a putative dihydrofolate reductase involved in biosynthesis of riboflavin (vitamin B2) in all the strains could indicate a requirement for flavin nucleotides as enzyme cofactors. Riboflavin is the precursor for flavin mononucleotide (FMN) and flavin adenine dinucleotide (FAD) redox cofactors in flavoproteins, and the E3 component of PDC as well as glycerol-3-phosphate dehydrogenase encoded from the up-regulated glpD, are among enzymes requiring FAD. Another cofactor which seems to be important during growth on ribose is lipoate, essential of the E2 component of the PDC. An up-regulation of lplA (1.0 -1.6) encoding lipoate-protein ligase, which facilitates attachment of the lipoyl moiety to metabolic enzyme complexes, was seen in all the strains, allowing the bacterium to scavenge extracellular lipoate [55,56].

Nucleoside catabolism
The L. sakei genome contains a multiplicity of catabolic genes involved in exogenous nucleoside salvage pathways, and the bacterium has been shown to catabolize inosine and adenosine for energy [7]. Three iunH genes are present in the 23K genome, which encode inosineuridine preferring nucleoside hydrolases responsible for conversion of inosine to ribose and purine base. The iunH1 gene was up-regulated in all the strains when grown on ribose (1.8-2.6), as was also the iunH2 gene in 23K (1.2). The deoC gene encodes a deoxyribose-phosphate aldolase, and is located in an operon structure preceding the genes deoB, deoD, lsa0798, lsa0799, deoR and pdp which encode phosphopentomutase, purine nucleoside phosphorylase, pyrimidine-specific nucleoside symporter, a putative purine transport protein, the deoxyribonucleoside synthesis operon transcriptional regulator (DeoR), and a pyrimidine-nucleoside phosphorylase, respectively. The complete operon was induced in all the strains, except for pdp only induced in 23K ( Table  1). The phosphorylases catalyze cleavage of ribonucleosides and deoxyribonucleosides to the free base pluss ribose-1-phosphate or deoxyribose-1-phosphate. The bases are further utilized in nucleotide synthesis or as nitrogen sources. The pentomutase converts ribose-1phosphate or deoxyribose-1-phosphate to ribose-5-phosphate or deoxyribose-5-phosphate, respectively, which can be cleaved by the aldolase to glyceraldehyde-3-phosphate and acetaldehyde. Glyceraldehyde-3-phosphate enters the glycolysis, while a putative iron containing alcohol dehydrogenase, encoded by lsa0258 up-regulated in all the strains (0.5-1.6), could further reduce acetaldehyde to ethanol (Figure 2). The obvious induced nucleoside catabolism at the level of gene expression was not seen by proteomic analysis [19].

Genes involved in glycerol/glycerolipid/fatty acid metabolism
During growth on ribose, a strong induction of the glpKDF operon encoding glycerol kinase (GlpK), glycerol-3-phosphate dehydrogenase (GlpD), and glycerol uptake facilitator protein was observed (Table 1), which is in correlation with the over-expression of GlpD and GlpK seen by proteomic analysis [19]. GlpD is FADH 2 linked and converts glycerol-3-phosphate to dihydroxyacetone-phosphate. An over-expression of GlpD was also reported when L. sakei was exposed to low temperature [57]. A glpD mutant showed enhanced survival at low temperature, and it was suggested that this was a result of the glycerol metabolism being redirected into phosphatidic acid synthesis which leads to membrane phospholipid biosynthesis [57]. Nevertheless, a down-regulation was observed of the lsa1493 gene (0.6-0.9) encoding a putative diacylglycerol kinase involved in the synthesis of phosphatidic acid, and of cfa (1.3-1.4) encoding cyclopropane-fatty-acyl-phospholipid synthase directly linked to modifications in the bacterial membrane fatty acid composition that reduce membrane fluidity and helps cells adapt to their environment [58]. Interestingly, LS 25 upregulated several genes (LSA0812-0823), including accD and accA encoding the αand ß-subunits of the multi-subunit acetyl-CoA carboxylase (Table 1). This is a biotin-dependent enzyme that catalyzes the irreversible carboxylation of acetyl-CoA to produce malonyl-CoA, an essential intermediate in fatty acid biosynthesis. In B. subtilis, the malonyl-CoA relieves repression of the fab genes [59]. We observed that also acpP, fabZ1, fabH, fabD and fabI (Table 1) encoding enzymes involved in fatty acid biosynthesis were induced in LS 25. The altered flux to malonyl-CoA may be a result of the decreased glycolytic rate. MF1053, on the other hand, showed a down-regulation of several genes in the same gene cluster. A higher level of acetate is produced when the bacterium utilizes ribose, and acetate lowers the pH and has a higher antimicrobial effect than lactate. Changes in the phospholipid composition could be a response to changes in intracellular pH. Protons need to be expelled at a higher rate when the pH drops. The LS 25 strain which showed faster growth rates than the other strains [9], was the only strain to up-regulate the F 0 F 1 ATP synthase (Table 1), which at the expense of ATP expels protons during low pH.

Regulation mechanisms
Little is known about the regulation of catabolic pathways in L. sakei. Starting from ribose uptake, the rbs operon may be both relieved from repression and ribose induced. Presumably, a dual regulation of this operon by two opposite mechanisms, substrate induction by ribose and CCR by glucose may occur in L. sakei. The ccpA gene was not regulated, consistent with this gene commonly showing constitutive expression in lactobacilli [42,60]. The local repressor RbsR is homologous with CcpA, both belonging to the same LacI/GalR family of transcriptional regulators. RbsR was proposed to bind a cre-like consensus sequence located close to a putative CcpA cre site, both preceding rbsU [28]. RbsR in the Gram-positive soil bacterium Corynebacterium glutamicum was shown to bind a cre-like sequence, and using microarrays, the transcription of no other genes but the rbs operon was affected positively in an rbsR deletion mutant. It was concluded that RbsR influences the expression of only the rbs operon [61]. Similarily, in the L. sakei sequence, no other candidate members of RbsR regulation could be found [28]. However, experiments are needed to confirm RbsR binding in L. sakei. In Bacillus subtilis, RbsR represent a novel interaction partner of P-Ser-HPr in a similar fashion to CcpA [62]. The P-Ser-HPr interaction is possible also in L. sakei as the bacterium exhibits HPr-kinase/phosphatase activity.
A putative cre site is present in the promoter of lsa0254 encoding the second ribokinase (Table 2), and this gene is preceeded by the opposite oriented gene lsa0253 encoding a transcriptional regulator with a sugar binding domain which belongs to the GntR family. This family of transcriptional regulators, as well as the LacI family which RbsR and CcpA belong to, are among the families to which regulators involved in carbohydrate uptake or metabolism usually belong [63]. The GntR-type regulator could possibly be involved in regulating the expression of the second ribokinase, or of the inosine-uridine preferring nucleoside hydrolase encoding iunH1 gene which is located further upstream of lsa0254. C. glutamicum possesses an operon encoding a ribokinase, a uridine transporter, and a uridine-preferring nucleoside hydrolase which is co-controlled by a local repressor together with the RbsR repressor of the rbs operon [60,61,64]. It is possible that such co-control could exist also in L. sakei. Ribose as well as nucleosides are products of the degradation of organic materials such as DNA, RNA and ATP. The simultaneous expression of the rbs and deo operons as well as the other genes involved in ribose and nucleoside catabolism (Figure 2) allows the bacterium to access the different substrates simultaneously and use both ribose as well as nucleosides as carbon and energy source. DeoR shows 51% identity to the B. subtilis DeoR repressor protein [65,66]. Genes encoding deoxyribosephosphate aldolase, nucleoside uptake protein and pyrimidine nucleoside phosphorylase in B. subtilis are organized in a dra-nupC-pdp operon followed by deoR, and ribose was shown to release DeoR from DNA binding and thus repression of the operon genes are alleviated [65][66][67]. The B. subtilis pentomutase and purine-nucleoside phosphorylase are encoded from a drm-pupG operon which is not negatively regulated by DeoR, though both operons are subject to CcpA mediated CCR [65,66,68]. As a cre site is found preceding the L. sakei deoC (Table 2), the operon could be regulated by CcpA as well. It is interesting that deoR is the only strongly induced transcriptional regulator gene in all three strains, and the encoded regulator has sigma (σ) factor activity. We can only speculate whether it could function as activator of transcription on some of the regulated genes in this study.
Expression of the Xpk encoding gene of Lactobacillus pentosus was reported to be induced by sugars fermented through the PKP and repressed by glucose mediated by CcpA [69]. Indeed, the cre site overlapping ATG start codon of L. sakei xpk ( Table 2) indicates relief of CcpA-mediated CCR during growth on ribose. Also for several genes involved in alternative fates of pyruvate, putative cre sites were present (Table 2).
Several genes and operons involved in transport and metabolism of various carbohydrates such as mannose, galactose, fructose, lactose, cellobiose, N-acetylglucosamine, including putative sugar kinases and PTSs, were induced during growth on ribose (Table 1), and as shown in Table 2, putative cre sites are located in the promoter region of many of these up-regulated genes and operons. 23K showed an up-regulation of genes involved in the arginine deiminase pathway, and 23K and LS 25 showed an up-regulated threonine deaminase ( Table 1). The arcA and tdcB both have putative cre sites in their promoter regions (Table 2). Thus ribose seems to induce a global regulation of carbon metabolism in L. sakei. The identification is based on the genome sequence of L. sakei strain 23K, and the consensus sequence TGWNANCG NTNWCA (W = A/T, N = A/T/G/C), confirmed in Gram-positive bacteria [39] was used in the search, allowing up to two mismatches (underlined) in the conserved positions except for the two center positions, highlighted in boldface. a mismatch to consensus sequence is underlined b position of cre in relation to the start codon c suggested co-transcribed genes or genes organized in an operon d cre in preceding gene encoding hypothetical protein e gene not regulated in this study A putative cre site precedes the glp operon (Table 2), suggesting regulation mediated by CcpA. However, regulation of the L. sakei GlpK may also occur by an inducer exclusion-based CcpA-independent CCR mechanism as described in enterococci and B. subtilis [70,71], and as previously suggested by Stentz et al. [15]. By this mechanism, glycerol metabolism is regulated by PEPdependent, EI-and HPr-catalyzed phosphorylation of GlpK in response to the presence or absence of a PTS substrate. In the absence of a PTS sugar, GlpK is phosphorylated by P-His-HPr at a conserved histidyl residue, forming the active P-GlpK form, whereas during growth on a PTS sugar, phosphoryl transfer flux through the PTS is high, concentration of P-His-HPr is low, and GlpK is present in a less active dephospho form [20,70,71]. This conserved histidyl residue (His232) is present in L. sakei GlpK [20], and Stentz et al. [15] reported that whereas L. sakei can grow poorly on glycerol, this growth was abolished in ptsI mutants.

Mannose-PTS
As mentioned in the introduction, the PTS plays a central role, in both the uptake of a number of carbohydrates and regulatory mechanisms [20][21][22]. Encoding the general components, ptsH showed an up-regulation in MF1053 and LS 25 (1.2 and 0.9, respectively), while all the strains up-regulated ptsI (0.8-1.7). The manLMN operon encoding the EII man complex was surprisingly strongly up-regulated during growth on ribose in all the strains (Table 1). By proteomic analysis, no regulation of the PTS enzymes was seen [19]. The expression of HPr and EI in L. sakei during growth on glucose or ribose was previously suggested to be constitutive [14], and in other lactobacilli, the EII man complex was reported to be consistently highly expressed, regardless of carbohydrate source [72][73][74]. Notably, PEP-dependent phosphorylation of PTS sugars has been detected in ribose-grown cells, indicating that the EII man complex is active, and since no transport and phosphorylation via EII man occurs, the complex is phosphorylated, while it is unphosphorylated in the presence of the substrates of the EII man complex [8,73]. The stimulating effect exerted by small amounts of glucose on ribose uptake in L. sakei, which has also been reported in other lactobacilli [74,75], was suggested to be caused by dephosphorylation of the PTS proteins in the presence of glucose, as a ptsI mutant lacking EI, as well as P-His-HPr, was shown to enhance ribose uptake [15,16,76]. Stentz et al. [15] observed that a L. sakei mutant (strain RV52) resistant to 2 deoxy-D-glucose, a glucose toxic analog transported by EII man , and thus assumed to be affected in the EII man , did not show the same enhanced uptake [15]. It was concluded that EII man is not involved in the PTS-mediated regulation of ribose metabolism in L. sakei. The mutation was though not reported verified by sequencing [15], and other mutations could be responsible for the observed phenotype. The L. sakei EIIAB man , EIIC man and EIID man show 72, 81, and 82% identity, respectively, with the same enzymes in L. casei, in which mutations rendering the EII man complex inactive were shown to derepress rbs genes, resulting in a loss of the preferential use of glucose over ribose [75]. Furthermore, in L. pentosus, EII man was shown to provide a strong signal to the CcpA-dependent repression pathway [73]. The hprK gene encoding HPrK/P which controls the phosphorylation state of HPr was strongly upregulated (1.2-2.0) in all three strains. HPrK/P dephosphorylates P-Ser-HPr when the concentration of glycolytic intermediates drop, which is likely the situation during growth on ribose [20,22,24].
Numerous genes encoding hypothetical proteins with unknown function were also found to be differentially expressed (Table 1), as well as several other genes belonging to various functional categories. For most of these, their direct connection with ribose metabolism is unknown, and is likely an indirect effect.

Conclusions
The ability to ferment meat and fish is related to the capacity of the bacterium to rapidly take up the available carbohydrates and other components for growth. The importance of this process, especially to the meat industry, stimulates research aimed at understanding the mechanisms for transport and metabolism of these compounds, with the ultimate goal to be able to select improved strains. Genome-wide transcriptome analyses with DNA microarrays efficiently allowed the identification of genes differentially expressed between growth on the two carbohydrates which L. sakei can utilize from these substrates. Moreover, microarrays were a powerful tool to increase the understanding of the bacterium's primary metabolism and revealed a global regulatory mechanism. In summary, the ribose uptake and catabolic machinery is highly regulated at the transcription level, and it is closely linked with catabolism of nucleosides. A global regulation mechanism seems to permit a fine tuning of the expression of enzymes that control efficient exploitation of available carbon sources.

Additional material
Additional file 1: Table S3. Primer and probe sets used for qRT-PCR. Presents the primer and probe sets used for validation of microarray data by qRT-PCR analysis. Table S4. Comparison of microarray data with qRT-PCR results of L. sakei strain LS 25 grown on ribose compared with glucose. Presents gene regulation values (log 2 ) from the qRT-PCR analysis in comparison with microarray data.