DNA micro-array-based identification of bile-responsive genes in Lactobacillus plantarum


  • Present address
    P.A. Bron, Alimentary Pharmabiotic Centre, Western Road, Cork, Ireland.

Michiel Kleerebezem, Wageningen Centre for Food Sciences/NIZO food research, P.O. Box 20, 6710 BA Ede, the Netherlands.
E-mail: michiel.kleerebezem@nizo.nl


Aims:  The purpose of this study was to determine the global transcriptional response in a food-associated lactic acid bacterium during bile stress.

Methods and Results:  Clone-based DNA micro-arrays were employed to describe the global transcriptional response of Lactobacillus plantarum WCFS1 towards 0·1% porcine bile. Comparison of differential transcript profiles obtained during growth of Lact. plantarum on plates with and without bile revealed 28 and 62 putative genes, of which the expression was at least 2·5-fold up- or down-regulated by bile, respectively. Approximately, 50% of these genes appeared genetically linked, and 12 bile-responsive gene clusters were identified. Seven of the identified bile-responsive genes and gene clusters encode typical stress-related functions, including glutathione reductase and glutamate decarboxylase, involved in oxidative and acid stress, respectively. Moreover, 14 bile-responsive genes and gene clusters were identified that encode proteins that are located in the cell envelope, including the dlt operon and the F1F0 ATPase.

Conclusions:  The identification of a relatively high number of genes encoding cell envelope functions indicates a major impact of bile acids on the integrity and/or functionality of the cytoplasmic membrane and cell wall.

Significance and Impact of the Study:  The data presented here provide valuable clues towards the defence mechanisms that play a role during bile stress in Lact. plantarum.


The human gastrointestinal (GI) tract represents an important ecological niche for food-associated micro-organisms. Following their consumption, food-grade bacteria, as well as food-borne pathogens, meet their host. Several biological barriers are encountered by bacteria during travel through and residence in the different parts of the host's GI tract, such as the gastric acidity in the stomach, the presence of bile salts in the duodenum and stress conditions associated with oxygen gradients that are steep at the mucosal surface, while the colonic lumen is virtually anoxic (Cummings and Macfarlane 1991). Moreover, considerable bacterial competition is encountered throughout the entire GI tract and notably in the colon where the bacterial population reaches densities of up to 1012 cells per gram of luminal content (Savage 1977). Because of the complex nature of host-specific and chemical conditions that are met by bacteria in the GI tract, many studies describe the in vitro response of bacteria to simplified models that mimic these harsh conditions, including several physiological and biochemical studies (Deuerling et al. 1995; Kullen and Klaenhammer 1999; Begley et al. 2002; Breton et al. 2002; Li et al. 2003; Bron et al. 2004).

Lactobacillus plantarum is a flexible and versatile lactic acid bacterium that is encountered in many fermented food products, including dairy, meat and many vegetable fermentations (Ruiz-Barba et al. 1991; Enan et al. 1996; Duran Quintana et al. 1999). Moreover, Lact. plantarum is currently marketed as a probiotic food ingredient in several products (Johansson et al. 1998; Molin 2001). In addition to the occurrence of Lact. plantarum in our diets, this microbe is frequently encountered as a natural inhabitant of the GI tract (Ahrne et al. 1998). Lactobacillus plantarum strain NCIMB 8826 has been shown to display high activity and survival rates in the human intestine in pharmacokinetic studies (Vesa et al. 2000). In addition, the complete 3·3-Mbp genome sequence of Lact. plantarum WCFS1, a single colony isolate of strain NCIMB 8826, has been determined (Kleerebezem et al. 2003). Recently, the observed GI tract persistence of Lact. plantarum WCFS1 has been studied by a genome-wide genetic screen, resulting in the identification of 31 putative genes, of which the expression appeared to be induced on plates containing porcine bile (Bron et al. 2004). In analogy with random gene disruption strategies applied in Listeria monocytogenes (Begley et al. 2002) and Enterococcus faecalis (Breton et al. 2002), this genetic screen in Lact. plantarum revealed that efflux pumps and changes in the architecture of the cell envelope are important for bile resistance of these bacteria. In addition, it was observed that exposure to bile greatly affected the morphology of Lact. plantarum WCFS1 (Bron et al. 2004). These findings are in good agreement with several physiological studies in other bacteria such as Propionibacterium freudenreichii and Lactobacillus reuteri which demonstrated that bile acids induce severe changes in the morphology of their cell envelopes (Leverrier et al. 2003; Taranto et al. 2003).

Here we describe the utilization of clone-based DNA micro-arrays to analyse the global transcriptional response of Lact. plantarum WCFS1 towards 0·1% porcine bile during growth on plates. This approach resulted in the identification of 28 and 62 potential genes, of which the expression appeared at least 2·5-fold up- or down-regulated in the presence of bile. Moreover, the expression levels of 12 gene clusters appeared to be regulated by bile. The observation that many of these genes and gene clusters encode cell-wall- and cell-membrane-associated functions strongly suggests that bile acids have a major impact on the integrity of the cell envelope.

Material and methods

Growth conditions and RNA isolations

Lactobacillus plantarum WCFS1 (Kleerebezem et al. 2003) was grown at 37°C in de Man-Rogosa-Sharpe medium (MRS; Difco, Surrey, UK) without aeration. Appropriate dilutions of a full-grown culture were plated on MRS containing 1% agarose, with or without 0·1% porcine bile salts (Sigma, Zwijndrecht, the Netherlands, B-8631). After 3 days of growth at 37°C, the plates contained approx. 100 colonies which were collectively resuspended in 3 ml MRS and immediately added to 12 ml quench buffer (60% methanol, 66·7 mmol l−1 HEPES, pH 6·5, −40°C) (Pieterse et al. 2006). Following quenching, the cells were pelleted by centrifugation at 3220 g for 10 min and cell pellets were resuspended in 0·4 ml ice-cold MRS. The cell suspensions were added to ice-cold tubes containing 1 g of zirconium glass beads, 400 μl of acidified phenol (Sigma), 100 μl of chloroform (Sigma), 30 μl of 10% SDS (Sigma) and 30 μl of 3 mol l−1 sodium acetate (pH 5·2) (Merck, Darmstadt, Germany). The cells were disrupted using two treatments of 40 s in a FastprepTM (Qbiogene Inc., Illkirch, France) interspaced by 1 min on ice. After centrifugation, 200 μl of the aqueous phase was used for RNA isolation using the High Pure system, which included a 1 h treatment with DNaseI (Roche Diagnostics, Mannheim, Germany).

Array design

The DNA micro-arrays were based on clones derived from the genomic library that was previously constructed for the determination of the complete genome sequence of Lact. plantarum WCFS1 (Kleerebezem et al. 2003). In total, 3692 genomic fragments were amplified by PCR from the genomic library using Supertaq (SphaeroQ, Leiden, the Netherlands) and vector-derived universal forward and reverse primers with 5′-C6 aminolinkers to facilitate cross-linking to the aldehyde-coated glass slides. The resulting amplicons were purified by ethanol precipitation and dissolved in 3× SSC (1× SSC contains 150 mmol l−1 NaCl and 17 mmol−1 sodium citrate, pH 7·2). Subsequently, the purified amplicons were arrayed in a controlled atmosphere on CSS-100 silylated aldehyde glass slides with quill pins (Telechem, Sunnyvale, CA, USA) in an SDDC 2 Eurogridder (ESI, Toronto, Canada). Afterwards the slides were dried and blocked with borohydride.

Fluorescent labelling and hybridization

Differential transcript levels were determined by two-colour (Cy5 and Cy3) fluorescent hybridizations of the corresponding cDNAs on the Lact. plantarum WCFS1 clone-based DNA micro-array. The RNA samples were labelled during reverse transcription using random hexamer primers and either Cy5- or Cy3-labelled dUTP (Amersham Biosciences, Freiburg, Germany). Unincorporated dyes were removed from the synthesized cDNA using autoseq G50 columns (Amersham Biosciences). The arrays were prehybridized for 45 min at 42°C in prehybridization solution (1% BSA, 5× SSC and 0·1% SDS). Cohybridizations of the labelled cDNA samples were performed overnight at 42°C in Easyhyb buffer (Roche Diagnostics) according to the manufacturer's protocol. The slides were subsequently washed twice in 1× SSC and 0·2% SDS, once in 0·5× SSC and twice in 0·2× SSC at 37°C.

Scanning and data analysis

After the washing steps, slides were dried and scanned with a ScanArray Express 4000 scanner for both dyes (PerkinElmer, Boston, USA). Images were analysed by using ImaGene 4·2 software (BioDiscovery, El Segundo, USA). Criteria for flagging spots were as follows: (i) empty spots threshold 2·0 (ii) poor spots threshold 0·4 (iii) negative spots on. The resulting ImaGene output files were further processed by discarding spots flagged as empty, poor or negative by the ImaGene software, and the Cy5 and Cy3 signals being less than two times above the local background. Routinely over 80% of all spots passed all these quality criteria and the remaining high-quality spots were normalized. For each array, the raw spot intensities in both channels (I1 and I2) were converted to MA co-ordinates, where M = 2log(I1/I2) and A = 2log(I1 + I2)/2. Subsequently a fit of M as a function of A was calculated using a robust local linear regression algorithm (LOWESS), thereby generating a smooth curve through the datapoints in the MA plot (Cleveland 1981). The value of the smooth curve at the corresponding A co-ordinate was subtracted from each M value of the raw data to obtain the corrected M values. These corrected M values equal the corrected 2log(fold-regulation) values. To calculate a regulatory ratio for each gene, as far as they were represented by clones on the micro-array, a weighted average of the M values of all clones that overlapped with the gene of interest was calculated. The weight used for each clone was equal to the square of the overlap between gene and clone divided by the total length of the gene. Hence, this method weighs small overlapping fragments less than proportional compared with larger overlapping fragments.


Clone-based transcriptome analysis

Clone-based DNA micro-arrays were used to investigate the genetic bile response of Lact. plantarum WCFS1. The development of these clone-based arrays was initiated following the random sequencing of the Lact. plantarum genomic library and its primary sequence assembly during the genome sequence project (Kleerebezem et al. 2003). Subsequently, 3692 inserts from this genomic library were amplified. The resulting amplicons had an average size of approx. 1·2 kb, cover 80·8% of the Lact. plantarum WCFS1 genome and represent 2683 of the 3052 annotated genes (88%). The overlap of clones in the part of the genome that is covered resulted in an average 1·6-fold redundancy.

For the construction of ORF-based arrays a partial, pre-annotated genome sequence is the minimum requirement. Therefore, one advantage of the clone-based arrays employed here lies within the fact that its construction is independent of such sequence-based analyses. In contrast, an intrinsic disadvantage of clone-based arrays is that certain genes will be incompletely represented or completely lacking on such an array. Moreover, certain clones will represent two or more ORFs, which potentially result in masking of regulation of a specific gene by neighbouring genes that are regulated in the opposite direction. Consequently, interpretation of the regulatory factors for individual genes requires relatively complex data processing and in some situations is simply impossible. On the contrary, multiple clones can represent individual genes or gene clusters on a clone-based DNA micro-array. Although multiple clones complicate the assessment of the exact magnitude of regulation factors, they provide independent replicate measurements that intrinsically validate the conditional response of specific genes or gene clusters. Therefore, our analyses exploited the latter advantages of clone-based DNA micro-arrays and focused mainly on the bile response of gene clusters and individual genes that are represented by multiple clones.

Global transcriptome analysis and selection of bile-regulated genes and clusters

The genome-wide transcription profiles of Lact. plantarum WCFS1 grown on plates with or without 0·1% porcine bile were compared. The experimental procedures that were used during these transcriptome analyses were highly similar to those used during a previous genetic screen in our laboratory (Bron et al. 2004). An independent biological duplicate was performed using two DNA micro-array slides and RNA derived from Lact. plantarum colonies grown on different batches of plates. Notably, the Cy5 and Cy3 dyes used during cDNA synthesis on the RNA originating from cells grown with or without bile were swapped in the two DNA micro-array experiments. Moreover, technical variations introduced by cDNA synthesis and/or dye-swap effects were assessed by a third array experiment in which two cDNA samples were synthesized using the same RNA template derived from Lact. plantarum cells grown without bile (Fig. 1). The differences in gene expression in the technical duplicate appeared to be relatively small as compared with those found in the arrays that assessed the differential transcript profile of cells grown with or without bile (biological variation). These global analyses indicate that the specific response of Lact. plantarum towards bile is drastically larger than the experimental variation (Fig. 1). Nevertheless, the variation between biological duplicates appeared relatively large (data not shown). Therefore, this first study focused on those responses that were consistently observed in both experiments, and the primary selection of regulated clones was performed using stringent criteria: (i) consistent direction of regulation in both arrays (ii) at least 2·5-fold up- or down-regulation in both arrays and (iii) the same order of magnitude of regulatory factor (maximum tenfold difference between the two slides). Using these relatively stringent criteria, 62 and 28 ORFs were found to be down- and up-regulated during growth of Lact. plantarum on plates containing bile as compared with control plates lacking bile, respectively (Tables 1 and 2). Approximately 25% of these individual genes were represented by a single clone and, consequently, significant assessment of their bile induction or repression levels requires expansion of the number of arrays. The rest of the genes were represented by two or more clones, allowing more trustworthy assessment of their induction levels.

Figure 1.

Scatter plot of the M values of two array experiments. The M value is defined as the 2log of the ratio of the Cy5 and Cy3 signal intensities per spot determined for one array. The X-axis represents the M values from an array hybridized with two cDNAs derived from a single RNA sample, originating from Lact. plantarum cells grown without bile (M−/−). The Y-axis represents the M values from two RNAs originating from cells grown with and without bile (M−/+). The square indicates 2·5-fold regulation that was used as a cut-off value in the analyses, and the circles enclose bile-responsive clones.

Table 1. Lact. plantarum genes that are at least 2·5-fold down-regulated by bile. The gene clusters are indicated in bold and are presented in more detail in Fig. 2. The two ratios represent the ratio of signal (Cy5 and Cy3) of the individual arrays
GeneProductNo. of clonesRatio 1Ratio 2
lp_0005DNA repair and genetic recombination protein RecF3 −3·72·6
lp_0006DNA gyrase, B subunit3 −3·52·5
lp_0007DNA gyrase, A subunit4 −9·83·5
lp_0064Conserved hypothetical protein1 −2·8−3·0
lp_0175Maltose/maltodextrin ABC transporter, substrate-binding protein3 −4·0 -4·0
lp_0176Maltose/maltodextrin ABC transporter, permease protein3 −3·23·0
lp_0177Maltose/maltodextrin ABC transporter, permease protein2 −3·02·5
lp_0178Maltose/maltodextrin ABC transporter subunit (putative)2 −4·93·7
lp_0179Alpha-amylase2 −4·65·3
lp_0180Multiple sugar ABC transporter, ATP-binding protein111·36·1
lp_0239Hypothetical protein2 −5·3−3·0
lp_0240Conserved hypothetical protein2 −6·1−3·2
lp_0242Nucleoside diphosphate kinase2 −5·3−3·0
lp_0266Hypothetical protein2 −4·9−5·7
lp_0286Cellobiose PTS, EIIC4 −6·5−2·8
lp_0287Hypothetical protein2 −3·2−2·6
lp_0289Hypothetical protein1 −3·0−2·6
lp_0852Pyruvate oxidase3 −7·5−3·0
lp_1069NADH dehydrogenase2 −5·3−3·2
lp_1168Hypothetical protein1−10·6−4·9
lp_1169Glutamate dehydrogenase (NAD(P)+)4 −4·0−3·7
lp_1171Serine-type d-Ala-d-Ala carboxypeptidase3 −8·6−2·8
lp_1362Hypothetical protein2 −3·7−2·8
lp_1420Nisin resistance protein (putative)1 −3·0−4·0
lp_1544Response regulator2 −7·0−3·2
lp_1545Histidine protein kinase; sensor protein3 −9·2−3·0
lp_1558Phenylalanine-tRNA ligase, alpha chain3 −3·5−3·7
lp_1643Cell surface protein precursor11 −9·8−3·2
lp_1956ABC transporter, permease protein1 −3·73·2
lp_1957ABC transporter, permease protein3 −3·23·0
lp_1958ABC transporter, ATP-binding protein2 −3·03·2
lp_1959Transcription regulator2 −5·34·9
lp_1960Hypothetical protein2 −2·8−5·7
lp_1961Hypothetical protein1 −2·8−5·7
lp_1962DNA-directed RNA polymerase, sigma factor 422 −2·5−4·3
lp_2143Integral membrane protein1 −4·0−3·0
lp_2173Extracellular protein3 −5·33·2
lp_2174Cell surface protein precursor3 −4·63·2
lp_2647N-Acetylglucosamine/galactosamine PTS, EIIA336·84·3
lp_2648N-Acetylgalactosamine PTS, EIID348·56·1
lp_2649N-Acetylgalactosamine PTS, EIIC2 −9·84·6
lp_2776D-Serine dehydratase2 −4·0−3·5
lp_2936Thiamin biosynthesis lipoprotein ApbE1 −4·0−4·6
lp_2937Transcription regulator2 −8·6−3·7
lp_2975Extracellular protein3 −4·93·2
lp_2976Cell surface protein precursor (putative)2 −6·53·2
lp_2977Cell surface protein precursor2 −2·82·5
lp_2978Extracellular protein1 −4·32·6
lp_3091Conserved hypothetical protein1 −3·7−3·7
lp_3092Succinate-semialdehyde dehydrogenase (NAD(P)+)3 −5·3−4·6
lp_3172Xylose operon regulator4 −4·9−2·8
lp_3219Sucrose PTS, EIIBCA3 −3·52·6
lp_3220Alpha-glucosidase6 −5·32·5
lp_3221Transcription regulator4 −3·52·8
lp_3222Hypothetical protein3 −2·63·0
lp_3360Integral membrane protein1 −4·9 −2·8
lp_3362Choloylglycine hydrolase4 −9·8 −3·0
lp_3397Conserved hypothetical protein1 −2·8 −3·7
lp_3553MaltoseO-acetyltransferase3 −8·0 −4·3
lp_3554L-Arabinose isomerase518·4 −7·5
lp_3555L-Ribulose 5-phosphate 4-epimerase297·016·0
lp_3556L-Ribulokinase (putative)613·9 −8·6
Table 2. Lact. plantarum genes that are at least 2·5-fold up-regulated by bile. The gene clusters are indicated in bold and are presented in more detail in Fig. 2.
GeneProductNo. of clonesRatio 1Ratio 2
Lp_0255Cystathionine beta-lyase39·29·2
Lp_0256Cysteine synthase38·68·6
Lp_0512Ribosomal protein L3119·25·3
Lp_0547Cell division protein FtsH, ATP-dependent zinc metallopeptidase44·03·2
Lp_0609Glutamate-tRNA ligase44·33·5
Lp_0779Nucleotide kinase (putative)36·14·0
Lp_0780Conserved hypothetical protein44·94·6
Lp_0781Conserved hypothetical protein25·35·3
Lp_1026Ribosomal protein S713·23·7
Lp_1253Glutathione reductase14·34·9
Lp_1541Phosphogluconate dehydrogenase (decarboxylating)35·73·0
Lp_2002Hypothetical protein13·24·3
Lp_2003Transcription regulator (putative)13·24·0
Lp_2019D-Alanine activating enzyme DltA34·03·5
Lp_2020D-Ala-teichoic acid biosynthesis protein (putative)14·66·1
Lp_2364H(+)-transporting two-sector ATPase, beta subunit33·711·3
Lp_2365H(+)-transporting two-sector ATPase, gamma subunit33·012·1
Lp_2366H(+)-transporting two-sector ATPase, alpha subunit24·011·3
Lp_2367H(+)-transporting two-sector ATPase, delta subunit33·09·2
Lp_2368H(+)-transporting two-sector ATPase, B subunit22·68·6
Lp_2369H(+)-transporting two-sector ATPase, C subunit22·68·6
Lp_3014Extracellular protein22·83·5
Lp_3420Glutamate decarboxylase26·54·9
Lp_3421Gamma-d-glutamate-meso-diaminopimelate muropeptidase (putative)26·53·0
Lp_3536Choloylglycine hydrolase24·94·6
Lp_3537Hydrolase, HAD superfamily, Cof family14·93·7
Lp_3687Cell division protein15·74·6

Approximately 50% of the regulated genes appeared to be genetically linked (Tables 1 and 2), suggesting clustered organization of these genes in operon-like structures, which could explain their concerted regulation. Therefore, these entire clusters, as based on gene annotation in the Lact. plantarum genome (Kleerebezem et al. 2003), were analysed in more detail (Fig. 2). To increase confidence, only clusters represented by at least three clones were used in the analyses, which in some cases included flanking genes that were not necessarily identified by our initial, stringent selection. Notably, some clones represent multiple genes (Fig. 2c, lp_1956), some genes are only partly represented (Fig. 2h, lp_3554 and lp_3557), and absolute induction levels obtained vary for different clones representing the same gene. The latter is especially true for clones that are located at the termini of the gene clusters identified and partially overlap with genes up- or downstream of the gene cluster (Fig. 2h, three clones representing lp_3553). On the contrary, the direction and magnitude of bile-mediated regulation is very similar for all clones representing the 12 complete gene clusters, thereby allowing multiple, independent measurements of the bile response of these gene clusters.

Figure 2.

Figure 2.

Schematic overview of the identified bile-responsive gene clusters. (a)–(h) represent the bile-repressed clusters lp_0004–lp_0007 (DNA gyrase), lp_0175–lp_0181 (maltose ABC transporter), lp_1956–lp_1959 (ABC transporter), lp_2173–lp_2175 (extracellular proteins), lp_2647–lp_2651 (N-acetylgalactosamine PTS system), lp_2975–lp_2978 (extracellular proteins), lp_3219–lp_3222 (sucrose PTS system) and lp_3553–lp_3557 (arabinose transporter), respectively. (i)–(l) represent the bile-induced gene clusters lp_0254–lp_0256 (metC-cysK operon), lp_0779–lp_0781 (nucelotide kinase), lp_2016–lp_2021 (dlt operon) and lp_2363–lp_2370 (F1F0 ATPase), respectively. Notably, representative clones are truncated at the first start or stop codon of the genes encompassed by the gene clusters and the presented fold-induction levels are the average of the values obtained in the two arrays performed. Genes presented as bold italic were initially identified using stringent selection criterion (Tables 1 and 2).

Figure 2.

Figure 2.

Schematic overview of the identified bile-responsive gene clusters. (a)–(h) represent the bile-repressed clusters lp_0004–lp_0007 (DNA gyrase), lp_0175–lp_0181 (maltose ABC transporter), lp_1956–lp_1959 (ABC transporter), lp_2173–lp_2175 (extracellular proteins), lp_2647–lp_2651 (N-acetylgalactosamine PTS system), lp_2975–lp_2978 (extracellular proteins), lp_3219–lp_3222 (sucrose PTS system) and lp_3553–lp_3557 (arabinose transporter), respectively. (i)–(l) represent the bile-induced gene clusters lp_0254–lp_0256 (metC-cysK operon), lp_0779–lp_0781 (nucelotide kinase), lp_2016–lp_2021 (dlt operon) and lp_2363–lp_2370 (F1F0 ATPase), respectively. Notably, representative clones are truncated at the first start or stop codon of the genes encompassed by the gene clusters and the presented fold-induction levels are the average of the values obtained in the two arrays performed. Genes presented as bold italic were initially identified using stringent selection criterion (Tables 1 and 2).


This paper presents the first DNA micro-array analyses performed in Lact. plantarum, focusing on the bile response of this lactic acid bacterium. Only two biological replicates of the DNA micro-array experiment were preformed in this study. To increase confidence levels, we focused mainly on the genes and gene clusters which are represented by more than one clone on the DNA micro-array. Seven genes or gene clusters encoding typical stress-related functions appeared regulated by bile. Remarkably, the operon encoding the RecF protein and DNA gyrase, involved in the repair of point mutations during oxidative stress in several bacteria (Rui and Tse-Dinh 2003) is down-regulated, while bile acids are considered to induce oxidative stress. This finding is unexpected, especially as several other bile-mediated responses appear to reflect the oxidative stress induced by bile. As an example, the increased expression of glutathione reductase and the metC-cysK operon upon bile treatment could be involved in the protection against the oxidative stress imposed on Lact. plantarum by bile (Leverrier et al. 2003; Li et al. 2003). Notably, one of the bile-induced proteins that was identified in P. freudenreichii in a differential proteome analysis is the cysteine synthase encoded by cysK (Leverrier et al. 2003), suggesting that at least part of the defence against bile appears to be conserved among different Gram-positive bacteria.

The increased ftsH transcript levels upon treatment with bile could also relate to stress adaptation in Lact. plantarum. The ftsH gene encodes an ATP-dependent zinc metallopeptidase involved in cell-cycle control and membrane function in Escherichia coli (Tomoyasu et al. 1993) and Bacillus subtilis (Tomoyasu et al. 1993; Deuerling et al. 1995; Deuerling et al. 1997). Moreover, in B. subtilis, the ftsH gene is induced upon osmotic upshift and has a general role in stress adaptation (Deuerling et al. 1995). Although their role in stress response remains unclear, it is very remarkable that the expression of one of the four choloylglycine hydrolase encoding genes of Lact. plantarum (lp_3536, bsh1) is approx. fivefold induced by bile, while the expression of another bsh gene is reduced (lp_3362, bsh3). Finally, the DNA micro-array experiments demonstrated that transcription of the glutamate decarboxylase encoding gadB gene was significantly induced by bile. In addition, two other genes encoding proteins involved in glutamate metabolism are regulated by bile, namely glutamate dehydrogenase and glutamate tRNA-ligase (lp_1169 and lp_0609, Tables 1 and 2, respectively). It has been demonstrated in several intestinal microbes, including E. coli (De Biase et al. 1999) and L. monocytogenes (Cotter et al. 2001), that intracellular glutamate accumulation enhances the survival of these microbes during osmotic and acid stress. This protective effect is directly correlated to the glutamate decarboxylase activity that is involved in the conversion of glutamate into γ-aminobutyrate. The γ-aminobutyrate formed is subsequently exchanged for another extracellular glutamate via the gadC-encoded antiporter (Cotter et al. 2001), thereby the combination of these reactions results in the consumption of an internal proton, which is thought to generate the observed stress tolerance. (Cotter et al. 2001). The gadB and gadC genes of E. coli and L. monocytogenes are genetically linked and tandemly transcribed, which is a feature that is not conserved in Lact. plantarum. Nevertheless, the Lact. plantarum genome appears to encode a GadC homologue (lp_2799). However, the expression of this gene did not appear to be affected by bile in our experiments. Nonetheless, a role of the Lact. plantarum GadB in maintenance of the relative intracellular pH and proton motive force, possibly involving a constitutively expressed glutamate-γ-aminobutyrate antiporter, can certainly not be excluded.

Six bile-repressed genes and gene clusters encode transporters located in the cell membrane, namely two ABC transporters, of which one is annotated to be specific for maltose/maltodextrin, three PTS-systems specific for N-acetylglucosamine, cellobiose and sucrose and an arabinose transport protein. The observation that all these transporters are down-regulated by bile could reflect the major impact of this lipid-active compound on membrane integrity and stability. Possibly, bile stress induced loss of membrane integrity can partially be compensated by the down-regulation of the genes encoding nonessential membrane proteins. A similar rationale might explain the reduced expression of the membrane-associated protein encoded by lp_2143. The only membrane-associated function of which the transcript levels are enhanced during bile stress is the F1F0 ATPase system, which is involved in maintenance of the proton motif force at the expense of ATP (Senior 1990). The expression of the atp operon in Lactobacillus acidophilus, encoding F1F0 ATPase, appeared to be induced at low pH ranges (Kullen and Klaenhammer 1999). Moreover, it was recently demonstrated that the membrane-bound H+-ATPase activity in a strain of Bifidobacterium was increased in response to bile (Sanchez et al. 2005). The increased transcript levels of the complete F1F0 ATPase encoding gene cluster in Lact. plantarum in the presence of bile strongly suggests proton motive force dissipation in the presence of bile. Apparently, maintenance of proton motif force is a crucial factor during bile stress, as two systems potentially contributing to this process appear to be induced, generating proton motif force at the expense of ATP (F1F0 ATPase system) or glutamate (glutamate decarboxylase system, see above).

Several changes in the expression levels of genes encoding cell-wall-associated proteins were observed. The expression of lp_1643 and lp_3421, encoding two individual cell surface proteins, appear to be down- and up-regulated by bile, respectively. Moreover, the transcript levels of two clusters encoding multiple extracellular proteins were reduced (lp_2173–lp_2175 and lp_2975–lp_2978). The exact role of these extracellular proteins in Lact. plantarum during bile stress remains to be established. Nevertheless, the remarkable changes in transcription level of several genes encoding cell-wall-associated functions clearly indicate a changed cell wall composition during bile stress. A striking observation in this respect is the increased expression of the dlt operon, involved in d-alanylation of teichoic acids (Delcour et al. 1999). This observation suggests an increased molecular decoration of lipoteichoic acid in the presence of bile, which would influence the charge of the cell wall and might provide a defence mechanism against bile. Besides the increase in expression of the serine-type d-Ala-d-Ala carboxypeptidase encoded in the dlt operon, expression of a second gene encoding this function appeared to be reduced during bile stress. In conclusion, the fact that 14 genes and gene clusters encoding cell-envelope-associated functions display altered expression levels indicate prominent changes in cell envelope architecture during bile stress.

The global transcriptional response towards bile presented here was investigated during growth on plates, closely resembling the conditions used during a genetic screen performed previously in our lab (Bron et al. 2004). Several genes and gene clusters identified here as bile-regulated encode functions that have previously been associated with stress response and/or adaptation in different bacteria (Deuerling et al. 1995; Cotter et al. 2001; Li et al. 2003). Moreover, the major impact of bile acids on cell envelope integrity is clearly exemplified by the change in expression of 14 genes and gene clusters encoding cytoplasmic membrane and cell-wall-associated functions. In analogy, the genetic screen already elucidated the importance of efflux pumps and cell envelope architecture in the response towards bile (Bron et al. 2004). These findings are in good agreement with the dramatic changes in morphology of Lact. plantarum cells upon bile stress (Bron et al. 2004), which has also been observed for other Gram-positive bacteria such as P. freudenreichii (Leverrier et al. 2003) and Lact. reuteri (Taranto et al. 2003).

The genetic screen (Bron et al. 2004) and the DNA micro-arrays presented here, both aim at analysis of the bile response in Lact. plantarum during growth on solid media. However, the genes identified by the genetic screen did not appear to be consistently up-regulated by bile according to the array analyses presented here. This apparent discrepancy might be explained by the fact that the genetic screen primarily identifies differentially active promoter elements in a multicopy plasmid system, while the DNA micro-array studies aim at differential transcript profiling. Moreover, as the genetic screen employed was based on complementation of a mutation in the essential alanine racemase encoding gene (alr), it is likely that this screen primarily identifies genes that are differentially expressed consistently during growth on plates. In contrast, the transcript levels assessed with DNA micro-arrays represent the abundance of mRNAs at one specific time point and thus reflect the response to the physicochemical conditions present at that time point. Consequently, these analyses provide a snapshot of the bacterial expression profile, namely at the time of RNA isolation. Hence, within the experimental set-up used here, it could very well be that the alr complementation screen has identified genes involved in the initial stress response of Lact. plantarum upon bile treatment, while DNA micro-array analyses have unravelled the later, bile adaptive response. Taken together, the complementary approaches used (genetic screen and DNA micro-arrays) have provided valuables clues towards the defence mechanisms that play a role during bile stress in Lact. plantarum.