ClaR—a novel key regulator of cellobiose and lactose metabolism in Lactococcus lactis IL1403

In a number of previous studies, our group has discovered an alternative pathway for lactose utilization in Lactococcus lactis that, in addition to a sugar-hydrolyzing enzyme with both P-β-glucosidase and P-β-galactosidase activity (BglS), engages chromosomally encoded components of cellobiose-specific PTS (PTSCel-Lac), including PtcA, PtcB, and CelB. In this report, we show that this system undergoes regulation via ClaR, a novel activator protein from the RpiR family of transcriptional regulators. Although RpiR proteins are widely distributed among lactic acid bacteria, their roles have yet to be confirmed by functional assays. Here, we show that ClaR activity depends on intracellular cellobiose-6-phosphate availability, while other sugars such as glucose or galactose have no influence on it. We also show that ClaR is crucial for activation of the bglS and celB expression in the presence of cellobiose, with some limited effects on ptcA and ptcB activation. Among 190 of carbon sources tested, the deletion of claR reduces L. lactis growth only in lactose- and/or cellobiose-containing media, suggesting a narrow specificity of this regulator within the context of sugar metabolism.


Introduction
Lactic acid bacteria (LAB), including Lactococcus lactis, are the focus of intensive research within the field of carbohydrate catabolism and its regulation, important during industrial fermentation processes (Mayo et al. 2010). The most prevalent pathway for carbohydrate uptake in LAB is the phosphoenolpyruvate-phosphotransferase system (PEP-PTS), which links sugar transport with phosphorylation of incoming sugars. Phosphoenolpyruvate serves as the first phosphoryl donor to Enzyme I, which utilizes a complex phosphoryltransfer cascade to phosphorylate sugars entering bacteria via the transmembrane protein-EIIC (Lorca et al. 2010). Among a wide array of carbohydrates that can be internalized via PTS, in LAB these systems are best known for efficient transport of milk (lactose) and plant (β-glucosides) sugars, which are highly important for industrial fermentation processes involving LAB species.
Due to the economic importance of lactose fermentation, the metabolism of this sugar is studied extensively in L. lactis. These studies mostly focus on two pathways: (i) plasmidlocalized lactose-specific PTS (PTS Lac ) and (ii) the chromosomally encoded lactose permease-β-galactosidase system (Aleksandrzak-Piekarczyk 2013). The latter system transports lactose in an unphosphorylated form that enables its subsequent cleavage by β-galactosidase. Recently, we reported on the discovery of a third, alternative lactose uptake system that we found to be operative in L. lactis IL1403 (Aleksandrzak-Piekarczyk et al. 2011). This novel lactose utilization pathway engages chromosomally encoded components of a cellobiosespecific PTS (PTS Cel-Lac ). The proteins of PTS Cel-Lac are encoded in two distinct regions that encompass ptcA, ptcB (one region), and celB (second region), which code for EIIA, EIIB, and EIIC, respectively (Aleksandrzak-Piekarczyk et al. 2011). The latter region also contains the bglS gene, which encodes a P-sugar-hydrolyzing enzyme with both P-β-glucosidase and P-β-galactosidase activity. This BglS enzyme has been shown to promote the cleavage of cellobiose and lactose internalized via PTS Cel-Lac in L. lactis IL1403 (Aleksandrzak-Piekarczyk et al. 2005). Therefore, PtcAB, CelB, and BglS form a complete system specific for cellobiose and lactose uptake and hydrolysis in L. lactis IL1403. We surmise that the existence of PTS Cel-Lac is not limited to L. lactis IL1403 and may occur commonly among other lactococcal strains (Aleksandrzak-Piekarczyk et al. 2011). This was confirmed by the later study of Solopova et al. (2012) in L. lactis MG1363. In this bacterium, the alternative lactose metabolism pathway, despite being differently induced (by lactose or cellobiose), relies on the same PTS Cel-Lac components used by L. lactis IL1403.
The efficiency of transport and subsequent metabolism of incoming sugars is tightly controlled by several regulatory proteins, which form the regulatory network necessary for sensing environmental conditions and setting catabolic capacities of the cell. Guédon et al. (2002) distinguish two groups of regulators-general and secondary regulators. The main example of such a general regulator is catabolite control protein A (CcpA) (Hueck and Hillen 1995). CcpA is a wellconserved protein operating in many low-GC Gram-positive bacteria and, together with its corepressor Ser-P-HPr, it binds to 14-nucleotide cis-acting DNA target sites known as catabolite responsive elements (cre), promoting carbon catabolite activation (CCA) or repression (CCR) (Weickert and Chambliss 1990). In L. lactis strains, CcpA has been shown to repress different genes associated with uptake of β-glucosides, fructose, galactose, or lactose, while activating the glycolytic las operon (Aleksandrzak-Piekarczyk et al. 2005Barrière et al. 2005;Luesink et al. 1998;Monedero et al. 2001).
In addition to global regulators, carbon catabolism might also be controlled by specific and local secondary regulators belonging to different protein families such as LacI, LysR, AraC, GntR, DeoR, RpiR, or BglG, which are widely distributed among LAB. Regulators from some of these families have been shown to control genes encoding utilization of αgalactosides, β-glucosides, fructose, lactose, maltose, sorbose, sucrose, and xylose in lactococci (Bardowski et al. 1994;Boucher et al. 2003;Mayo et al. 2010).
The YebF protein (now designated cellobiose-lactose regulatory protein-ClaR) from L. lactis IL1403, belongs to the RpiR family of regulators. In bacterial genomes, there are many genes that encode potential regulators belonging to the RpiR family (http://www.genome.jp/kegg/ssdb/). However, none of these genes have thus far been shown to play a role in carbohydrate catabolism among LAB. We previously suggested the probable involvement of ClaR in the activation of bglS, celB, ptcA, and ptcB genes, which encode components of the novel and the only pathway for lactose utilization in L. lactis IL1403 (Aleksandrzak-Piekarczyk et al. 2005. It has been demonstrated that the expression of these genes is tightly regulated by the general catabolite repression system, whereas bglS and celB also require the presence of cellobiose to be fully induced (Aleksandrzak-Piekarczyk et al. 2011).
In this report, we show that in L. lactis IL1403, genes encoding the PTS Cel-Lac components in the presence of cellobiose are regulated via a novel activator ClaR. Moreover, global phenotypic analysis via phenotype microarray suggests that ClaR is a specific regulator indispensable for cellobiose and lactose metabolism.
Construction of the claR deletion mutants and complementing plasmid The mutant was created by double crossover between pGhost9 harboring DNA fragments overlapping the claR gene and the chromosomal region containing these DNA fragments. The overlapping DNA fragments were amplified using the appropriate forward and reverse primer pairs ( Table 1). The claR upstream region was amplified with the primers claRPstI/claRERpr (Table 1) using ExTaq polymerase and was cloned as a PstI/EcoRI fragment into the corresponding sites in pGhost9. The claR downstream region was amplified using the primers claRERrv/claRSalI (Table 1) and was cloned as an EcoRI/SalI fragment into the vector carrying the upstream fragment, resulting in the pGhost9ΔclaR plasmid. The deletion plasmid was introduced into L. lactis IL1403 and IL1403ΔccpA. Homologous recombination was enforced by 10 −3 dilution in fresh G-M17-Em medium of the overnight culture of the lactococcal strains harboring pGhost9ΔclaR. Diluted cultures were incubated for 2.5 h in G-M17 at non-permissive temperature (38°C). Integrants containing pGhost9ΔclaR in the chromosome were selected at 38°C on G-M17 agar plates containing erythromycin.
Excision from the chromosome and removing of the integration vector from L. lactis were performed by growth of  (Bolotin et al. 2001), which is available from NCBI (http://www. ncbi.nlm.nih.gov/genome) with accession no. AE005176. To certain primers restriction sites were added for digestion with EcoRI, KpnI, NcoI, PstI and SalI integrants in the absence of antibiotic for at least 100 generations at the permissive temperature of 28°C. The genetic structure of the resulting claR deletion strains, L. lactis IL1403ΔclaR, was confirmed by colony PCR, determination of the strain sensitivity to Em, and sequencing of the DNA region containing the deleted claR gene. In order to complement the claR deletion, the claR gene with its putative promoter region was amplified using claRSmaf and claRPstR primers (Table 1), cloned into pGEM-T, and transferred into E. coli TG1. The resultant plasmid DNA was isolated, digested with PstI and SmaI, ligated to pIL253, digested with the same restriction enzymes, and transferred into L. lactis IL1403ΔclaR giving rise to the IL1403ΔclaRpIL253claR strain.
Cloning and activity measurement of the putative intrinsic terminator following the yebE gene A DNA sequence resembling a rho-independent terminator downstream of yebE was amplified using yebEterf and yebEterr primers (Table 1). Further cloning procedures into pGBT58 bearing the xylE gene that encodes catechol 2,3-dioxygenase (Table 1) and measurements of its expression determined by enzymatic assay were performed as described previously (Aleksandrzak-Piekarczyk et al. 2011). Its activity was determined in three independent experiments. As a reference point, measurements of catechol 2,3-dioxygenase activity were also performed in the wild-type pGBT58 vector, which contains no terminator located between the xylE gene and its promoter.
Cloning of putative claR promoter region upstream of the luxAB genes Construct consisting of pGEM-T, pJIM2374, and a non-coding region (amplified with yebEterf and yebEterr), containing a putative promoter of claR controlling the luxAB genes, was prepared as described previously (Aleksandrzak-Piekarczyk et al. 2011). Since pGEM-T is non-replicative in Gram-positive bacteria and pJIM2374 does not contain the rep gene, the recombinant pJIM2374claRp plasmid was transferred into L. lactis LL302, which encodes the RepA protein in its chromosome. The existence of a functional promoter in the cloned region was concluded from luciferase activity measurements performed in C-CDM.
Relative quantification of the gene expression by real-time quantitative PCR For real-time quantitative PCR (RT-qPCR), total RNA was isolated with the use of TRI Reagent (Sigma) according to the instructions of the manufacturer, from 25 ml of the IL1403 wild-type strain and its mutants (IL1403ΔccpA and IL1403ΔclaR) cultures harvested in midexponential phase (OD 600 =0.6). Cells were grown in G-CDM, C-CDM, Gal-CDM, A-CDM, or GalC-CDM. RNA was extracted from at least three independent cultures.
First-strand cDNA was synthesized from DNAse I (Sigma)-treated 2 μg RNA samples. The synthesis of cDNA was performed by the use of the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems) according to manufacturer's instructions.
Real-time quantitative PCR assays were carried out on the 7,500 Real Time PCR System (Applied Biosystems). Each reaction was carried out in reaction mixture containing the following: 1× concentrated commercial buffer supplied with polymerase (Metabion), 0.625 U Taq polymerase (Metabion), 4 mM MgCl 2 , 30,000× diluted SYBR Green (Sigma), 5 % DMSO, 0.5 ng/μl acetylated BSA (Sigma), 0.8 % glycerol (ROTH), 400 μM dNTP mixture (Metabion), forward and reverse specific primers (400 μM each), cDNA template, and water up to 25 μl of final volume. Reactions were performed with an initial denaturation step (95°C for 3 min) followed by 45 cycles of denaturation (95°C for 15 s) and primer annealing-extension (60°C for 1 min). Fluorescence was read during the annealing-extension step of each cycle. After cycling, melting-point temperature analysis was performed in the range of 60 to 95°C with temperature increments of 0.33°C. In each experiment, background range was adjusted automatically and the threshold for C t evaluation was adjusted manually. For each cDNA sample, six reactions were carried out using three template amounts, each in duplicate. For each gene studied, the amounts of cDNA were chosen individually (if possible the same for all genes) to obtain C t values in range between 14 and 34 cycles. Quality of results was evaluated based on expected C t differences among three cDNA amounts as well as product melting curves. Rare outlying results were omitted in calculations. The results were normalized by the use of L. lactis, the reference tuf and purM genes, which code for elongation factor TU and phosphoribosylaminoimidazole synthetase, respectively. Specific primers for each gene (Table 1) were designed using Primer Express software (Applied Biosystems). Before use, the primers were verified for equal efficiency of PCR reaction. The resulting data were processed using files exported from the qPCR cycler program and imported into Excel sheets to facilitate calculation of expression ratios between target and reference genes. These ratios were calculated by ΔC t method using geometric mean of reference genes C t s in each experiment separately.
Phenotypic testing of carbon source utilization The study of phenotypic changes in mutant cells compared to parental strains was carried out through growth tests and analysis of fermentation patterns.
Growth tests were performed on a Microbiology Reader Analyser, Bioscreen C (Labsystems) in 0.2 ml of CDM supplemented with the required sugars. Changes in the OD 600 of the bacterial cultures were recorded every 30 min of growth up to 100 h.
The fermentation patterns of 49 sugars were determined using the API 50CH test as specified by the manufacturer (BioMérieux) and recorded after 12, 24, and 36 h of incubation under aerobic or anaerobic conditions at 30°C.
Further metabolic profiles were measured globally by the Phenotype MicroArrays system (Biolog, USA) according to manufacturer's instructions. L. lactis strains were streaked on plates containing G-M17 agar. Colonies were scraped from the plates and titrated into IF-0a inoculating fluid (Biolog) with growth supplements and Biolog redox tetrazolium dye until the solution reached desired transmittance, according to standard protocols recommended by Biolog for Streptococcus species. One hundred microliters of aliquots was added to each well of carbon source plates (PM1 and PM2). The plates were incubated at 30°C in an aerobic OmniLog incubator plate reader, and the metabolic activity was measured kinetically by determining the colorimetric reduction of a tetrazolium dye. Phenotype MicroArrays use Biolog's redox assays, engaging cell respiration or fermentation as a universal reporter and provide precise quantitation of phenotypes. If the phenotype is strongly "positive" in a well, the cells are metabolically active, reducing a tetrazolium dye and forming a strong color. If their metabolic activity is slowed or stopped, less color or no color is formed. Data were collected approximately every 10 min over a 72-h period. This was a sufficient time for color development in the positive control wells, while the negative control wells remained colorless. Data were analyzed with the Biolog Kinetic and Parametric software. The PM1 and PM2 Biolog assays assess the ability of a bacterium to utilize any of 190 carbon compounds used in the assay as the sole carbon source.

Results
Structural characterization of DNA region encompassing the claR gene Figure 1 depicts the chromosomal region of the claR gene (formerly yebF) from L. lactis IL1403. ClaR reveals high-sequence similarity with transcriptional regulators of the RpiR family: it is a two-domain protein, with a 67-residue Nterminal DNA binding helix-turn-helix (HTH) motif and a 123-residue C-terminal sugar isomerase (SIS) domain ( Fig. 1a; http://pfam.sanger.ac.uk/). Located upstream of claR is the yebE gene, which encodes a hypothetical protein. Within the intergenic region between these two genes, a potential rho-independent terminator with the free energy value (dG) of −15 kcal/mol was localized (Fig. 1b, c). In order to verify its functionality, the terminator was cloned into the pGBT58 vector between a promoter and the xylE gene encoding a catechol 2,3-dioxygenase. In an E. coli strain harboring this construct levels of catechol, 2,3dioxygenase activity obtained reached 0.50±0.04 U, lower than the observed control value (pGBT58 without yebE terminator) of 1.25±0.16 U. Thus, the presence of the yebE terminator caused a 60 % reduction in catechol oxygenase activity when compared to the strain harboring the wild-type pGBT58 vector.
According to an in silico analysis, the claR gene is preceded by at least two putative promoters (Fig. 1b, c) suggesting multiple potential transcription start sites. None of the identified potential −10 and −35 promoter regions upstream of claR had a full agreement with promoter consensus sequences, defined as TATAAT and TTGACA, respectively (Browning and Busby 2004). One of these potential promoters was localized upstream of the yebE rho-independent terminator (Fig. 1b, c). In order to verify whether in the DNA region upstream of claR a functional promoter (s) may exist, a DNA fragment encompassing the abovementioned putative promoters was inserted in front of the luxAB reporter genes in the promoter probe pJIM2374 vector and introduced into L. lactis LL302. Considerable luciferase activity was detected when the claR upstream region was tested indicating functionality of at least one promoter embedded there.
Phenotypes of claR deletion mutants In our previous study, the lactose-fermenting ccpA mutant was randomly mutagenized using pGhost9::ISS1. By the use of this procedure, on X-Gal supplemented plates, we have isolated several βgalactosidase-negative phenotype mutants in the ccpA mutant background. Among them, the inactivation of claR (yebF) in the ccpA mutant led to the loss of its lactose fermentation ability (Aleksandrzak-Piekarczyk et al. 2005). Thus, in this study, deletion of claR was performed both in the IL1403 wild-type strain and IL1403ΔccpA resulting in IL1403ΔclaR and IL1403ΔccpAΔclaR mutants, respectively. Subsequently, IL1403ΔclaR and IL1403ΔccpAΔclaR growth was tested in CDM supplemented with various sugars and compared with their respective parental strains. In C-CDM, the IL1403ΔclaR mutant initially showed no growth ability, and a minor increase in optical density of cells was observed after prolonged incubation (Fig. 2a). Deletion of ccpA in the IL1403ΔclaR strain led to partial growth restoration of the obtained IL1403ΔccpAΔclaR strain in C-CDM (Fig. 2a). Both IL1403ΔclaR and IL1403ΔccpAΔclaR displayed no growth in l-CDM in comparison to their parental IL1403 and IL1403ΔccpA strains (Fig. 2b). The supplementation of this medium with an inducing concentration of cellobiose (LC-CDM) did not restore mutants' growth (Fig. 2c). Deletion of the claR gene in the IL1403 and IL1403ΔccpA strains had no effect on mutants' growth in arbutin-, esculin-, galactose-, glucose-, or salicinsupplemented medium (data not shown).
After a 24 h incubation period, the API 50CH test confirmed the complete inability of IL1403ΔclaR to ferment lactose and cellobiose. While the inability to assimilate lactose by both IL1403ΔclaR and IL1403ΔccpAΔclaR mutants lasted for 36 h, efficient utilization of cellobiose was observed in IL1403ΔclaR after the initial 24 h (Fig. 2d). When assessed by API 50CH, no differences were found between the IL1403ΔccpAΔclaR and its parental IL1403ΔccpA within the context of cellobiose fermentation (Fig. 2d).
Global measurements of metabolic activity using the Phenotypic MicroArray approach indicated that deletion of claR led to complete abolishment of lactose utilization in IL1403ΔclaR. In contrast, when cellobiose was available as a sole carbon source, this caused only a slight decrease of IL1403ΔclaR metabolic activity compared to wild type (data not shown).
No further pronounced differential phenotypes could be identified between strains tested on all other 47 sugars available on API 50CH or the 188 carbon sources available on Biolog PM1 and PM2 arrays.
Transformation of pIL253claR into IL1403ΔclaR, which led to the creation of the IL1403ΔclaRpIL253claR strain, fully complemented the effects of claR deletion, restoring the mutant's growth in cellobiose-and/or lactose-containing medium (data not shown).
ClaR is expressed constitutively at low level and independently from CcpA-mediated regulation Apart from direct regulation of gene expression mediated by CcpA binding to cre sequences (Kowalczyk et al. 2007), indirect control has been observed previously (Ludwig et al. 2002). Although the claR upstream DNA region lacks a sequence resembling the cre box, its expression could be still regulated by CcpA in an indirect manner. To test whether this phenomenon also occurs in the case of regulation of claR and whether its expression is influenced by the presence of different sugars, levels of claR expression were estimated by real-time quantitative RT-PCR in response to arbutin, cellobiose, galactose, and glucose in IL1403 and IL1403ΔccpA. In both of these strains and in presence of all sugars tested, the claR gene was constitutively expressed at a low level reaching an approximate mean relative gene expression level value of 0.05 (data not shown). This suggests that claR expression is neither induced nor repressed by any of the sugars tested and does not undergo CcpAdependent regulation.
Expression of celB, bglS, and ptcBA genes is positively regulated by ClaR in the presence of cellobiose Previously, we showed that genes encoding PTS Cel-Lac components (celB and ptcBA) and  Fig. 1 The organization of ClaR domains (a), the claR chromosomal region (b), and nucleotide sequence of the yebE-claR intergenic region (c). The ClaR domains (a) and the genes of the claR chromosomal region (b) are drawn to scale. Stem-loop structures denote rho-independent terminators. Gray-shaded sequences (c) highlight the potential −10/−35 promoter regions upstream of claR. The putative ribosome binding site (RBS) is underlined, and the start and stop codons of claR and yebE, respectively, are shown as black shading. Stars following yebE indicate a rho-independent terminator, the functionality of which was measured in this work was used to support the growth of IL1403ΔclaR, since this mutant is unable to utilize cellobiose as a carbon source. Lactose was not included in the array of sugars due to L. lactis IL1403 poor growth parameters (growth rate, lag phase, and optical density of the final culture) in comparison with other sugars used (such as cellobiose, galactose, or glucose).
The striking difference between the levels of bglS and celB expression in IL1403 wild-type and IL1403ΔclaR was evident when the strains were grown in GalC-CDM. In this case, levels of celB and bglS expression in the presence of ClaR (in IL1403) were significantly higher than the values obtained in the absence of ClaR (in IL1403ΔclaR) reaching ClaR activation ratio (calculated as a quotient of relative gene expression in IL1403 and IL1403ΔclaR) of 8 and 36, respectively (Table 2). In the case of ptcBA, ClaR activation ratios were close to 3, suggesting minor ClaR-dependent regulation ( Table 2).
The lowest transcript levels were detected in IL1403 wildtype and IL1403ΔclaR cells growing under repressive conditions (G-CDM) due to probable downregulation of bglS, celB, and ptcBA expression by CcpA. Levels of bglS and celB expression were only slightly varied, whereas the ptcBA transcription was increased in both strains in medium supplemented with galactose (Table 2). Comparable expression levels between IL1403 wild-type and IL1403ΔclaR strains in the presence of galactose or glucose (ClaR activation ratio ∼1; Table 2) imply that these sugars do not elicit ClaR-dependent activation of bglS, celB, and ptcBA.
Substantially higher transcription levels of all genes tested were detected when the wild-type strain was grown in a medium supplemented with cellobiose (C-CDM) as a sole carbon source ( Table 2). The addition of galactose (GalC-CDM) caused a 2-4-fold transcription decrease of bglS, celB, and ptcAB when compared to C-CDM. This may suggest that the presence of galactose leads to the repression of transcription in a similar manner to glucose.
Furthermore, to determine the influence of cellobiose on celB, bglS, ptcB, and ptcA transcription in the presence (in IL1403) or absence (in IL1403ΔclaR) of ClaR cellobiose, activation ratios were calculated as a quotient of relative gene expression in the wild-type strain IL1403 or IL1403ΔclaR x-axis = time (in hours), and the y-axis = optical density at 600 nm. Sugar fermentation patterns of the L. lactis IL1403 wild-type strain and its derivatives: IL1403ΔccpA, IL1403ΔclaR, and IL1403ΔccpAΔclaR determined by the API 50CH test (d). The results are shown for 12, 24, and 36 h of incubation; +, good fermentation; +/−, weak fermentation; −, no fermentation (d) grown in GalC-CDM and in Gal-CDM. These ratios obtained in IL1403 varied from 25 (bglS) and 7 (celB) to as low as 3-4fold (ptcBA) implying that cellobiose has different levels of influence on the activation of associated genes (Table 2). These cellobiose activation ratios were similar to ClaR activation values calculated for GalC-CDM suggesting that this cellobiose-dependent regulation may be due to ClaR action. This was confirmed by the observation that the cellobiose activation ratios obtained in the absence of ClaR were generally around a factor of 1, indicating a lack of genetic expression regulation in IL1403ΔclaR even in the presence of cellobiose (Table 2).

Discussion
The claR gene product from L. lactis IL1403 belongs to the RpiR family of potential regulators, of which members are abundantly represented in many other related or distantly related bacteria from different genera (http://www.kegg.jp/ kegg/ssdb/). The strongest ClaR orthologs are found only in other Lactococcus species (e.g., L. lactis subspecies cremoris MG1363 and SK11, and L. lactis subspecies lactis KF147). There are eight paralogs (encoded by yidA, yecA, yljC, yfeA, yugA, gntR, and yleF) from the RpiR family, with ClaR being the prototypical example, encoded in the L. lactis IL1403 genome (Bolotin et al. 2001;Guédon et al. 2002). Thus far, no function has been assigned to any of the RpiR family genes in L. lactis. However, the conservation of ClaR orthologs found in many different bacterial species suggests an important role in host organisms. Indeed, few members of the RpiR family have been shown to function as positive or negative transcriptional regulators of genes involved in the metabolism of different carbon sources. For example, in Bacillus subtilis, maltose metabolism is positively regulated by GlvR (Yamamoto et al. 2001). Until now, GlvR was the only representative from the family of RpiR regulators characterized within a group of Gram-positive bacteria. All other functionally characterized examples originate from Gram-negative bacteria, where RpiR family regulators appear to downregulate gene expression. These RpiR-like repressors have been shown to regulate the metabolism of ribose and N-acetylmuramic acid in E. coli, glucose in Pseudomonas putida, and inositol in Caulobacter crescentus, Salmonella enterica, and Sinorhizobium meliloti (Boutte et al. 2008;Daddaoua et al. 2009;Jaeger and Mayer 2008;Kohler et al. 2011;Kröger and Fuchs 2009;Sørensen and Hove-Jensen 1996). Members of the RpiR family contain a HTH motif at their N-terminus and a SIS domain on their C-terminal end. The SIS domain is a phospho-sugar-binding domain found in many proteins that regulate the expression of genes involved in synthesis of phospho-sugars (Bateman 1999;Teplyakov et al. 1998).  Cellobiose activation ratio was calculated as a quotient of relative gene expression in the wild-type strain IL1403 or IL1403ΔclaR grown in GalC-CDM and in Gal-CDM It is tempting to speculate that this type of regulation is operative in L. lactis IL1403; theoretically, phosphorylated upon entry into the IL1403 cell via CelB cellobiose or lactose bind to the SIS domain of ClaR. Such association may result in a stimulation of the ClaR regulator and allows its binding to the respective DNA regions, resulting in ClaR-dependent regulation of gene expression. We postulate that through this mechanism the expression of bglS, celB, and ptcBA could be activated in the presence of cellobiose or lactose. Evidence confirming this assumption includes strong observed decreases in bglS, celB, and, to a lesser extent, ptcBA mRNA levels in response to ClaR absence upon the growth of a claR mutant in cellobiose-containing medium ( Table 2). This effect is not observable when galactose or glucose is present as a sole carbon source, indicating that these sugars have no activating role toward ClaR.
Additionally, the hypothesis that ClaR plays a role in lactose and cellobiose metabolism is strongly confirmed by the growth deficiencies of claR mutants in cellobiose-and/or lactose-supplemented media (Fig. 2a-c). Deletion of ccpA led to the partial growth restoration of the IL1403ΔccpAΔclaR mutant in cellobiose-supplemented media, but not in lactosesupplemented media (Fig. 2a-c). Thus, it seems reasonable to conclude that both mutants are barely or not able to utilize these sugars due to a lack of observed bglS and celB transcriptional activation by ClaR. Indeed, in the absence of ClaR, transcription of bglS and celB decreases 36-and 8-fold, respectively (Table 2). This may result in a severe decrease in BglS and CelB, which are indispensable for L. lactis IL1403 growth on cellobiose and lactose as shown previously (Aleksandrzak-Piekarczyk et al. 2005. When assayed globally (by Phenotype MicroArrays and API 50CH), no additional pronounced differences in utilization of tens of different carbon sources were detected between the claR mutants and their parental strains. This implies a narrow but significant specificity of ClaR. The contradictory phenomenon that IL1403ΔclaR is barely able to grow on cellobiose (Fig. 2a) while efficiently respiring on this sugar (when tested by phenotype MicroArrays) could be explained by the fate of cellobiose metabolism in IL1403. Previously, we showed that the EIIC CelB permease is the only protein dedicated to cellobiose and lactose uptake in L. lactis IL1403 (Aleksandrzak-Piekarczyk et al. 2011), while BglS is the only glycosidase crucial for lactose hydrolysis (Aleksandrzak-Piekarczyk et al. 2005). Although cellobiose is also a substrate for BglS, it was postulated that one or more additional enzyme(s) capable of hydrolysis of this sugar may exist in IL1403 and may be subject to CcpA-mediated negative regulation (Aleksandrzak-Piekarczyk et al. 2005). Moreover, results of the current study clearly indicate that the transcription of bglS tightly depends on its activation by ClaR, whereas transcription of celB is to a lesser extent dependent on the activation by ClaR (Table 2). Thus, we may assume that in IL1403ΔclaR, small amounts of CelB are still present and may transport traces of lactose or cellobiose. Internalized by IL1403ΔclaR, lactose does not serve as an energy source due to a lack of BglS resulting from the absence of ClaR. On the other hand, in IL1403ΔclaR, small amounts of cellobiose-derived metabolites, which are formed after cellobiose uptake via CelB and activity of different than BglS hydrolases, are insufficient for effective mutant cell division, resulting in the limited mutant's growth.
The results determined during this study by RT-qPCR indicate that claR is weakly expressed under all conditions tested (in presence of different sugars [ Table 2] and in presence or absence of the CcpA protein [data not shown]), in which it reaches the relative gene expression levels only slightly above zero. This is in agreement with the fact that many transcriptional regulators exist endogenously in low amounts, and, in the case of ClaR, it may be due to several reasons. First, both of the putative promoters found upstream of the claR gene show incomplete correlation with the consensus sites for both −10 and −35 sequences, and thus may drive a low level transcription. Second, one these putative promoters yield transcript that starts upstream of the rhoindependent terminator sequence (Fig. 1b, c) that may prematurely terminate claR transcription. We proved that it is a moderately strong terminator, as it arrested the transcription of claR with about 60 % efficiency. Thus, it might be assumed that the generation of full-length transcript covering the claR gene is likely achieved by a low amount of read-through transcription. However, we cannot exclude the possibility that the efficiency of read-through termination could be strengthened by the influence of unknown antitermination factors.
Despite the narrow specificity of ClaR within the context of sugar metabolism, it seems that this regulator has an invaluable role in lactococcal cells. We demonstrated that its activity is crucial for the assimilation of two sugars that are commonly found in ecosystems inhabited by L. lactis strains-plant material (cellobiose) and milk (lactose). Having such a regulator allows for a settlement of more diverse environments, and thus gives L. lactis an advantage over other microorganisms. The unprecedented role of the ClaR protein is made even more significant by the fact that the metabolism of lactose and β-glucosides is of great economic importance for biotechnological processes involving L. lactis, in regard to the production of both fermented milk and plant products. Obtaining detailed knowledge related to sugar metabolism and the regulation of associated gene expression in L. lactis may contribute to the improvement of mechanisms controlling significant cellular processes in these bacteria. In the case of industrial microorganisms such as L. lactis, modification of defined regulatory networks may drastically affect the properties of the bacteria and have implications on bioprocesses. Finally, we can assume that through our research, we can introduce changes in the metabolic potential of Lactococcus strains, which by themselves are not able to assimilate lactose. We can initiate this process through the inactivation of ccpA or activation of other genes by addition of cellobiose. In contrast to plasmid-located lac-operons, genes encoding PTS Cel-Lac components are located on the chromosome, which ensures their stability, a potentially important feature for industrial applications.