OUP user menu

Global transcriptome analysis of the heat shock response of Bifidobacterium longum

Enea Rezzonico, Sofiane Lariani, Caroline Barretto, Gabriella Cuanoud, Gabriele Giliberti, Michèle Delley, Fabrizio Arigoni, Gabriella Pessi
DOI: http://dx.doi.org/10.1111/j.1574-6968.2007.00704.x 136-145 First published online: 1 June 2007


Bifidobacteria are natural inhabitants of the human gastrointestinal tract and have been widely used as functional foods in different products. During industrial processing, bacterial cells undergo several stresses that can limit large-scale production and stability of the final product. To better understand the stress-response mechanisms of bifidobacteria, microarrays were used to obtain a global transcriptome profile of Bifidobacterium longum NCC2705 exposed to a heat shock treatment at 50°C for 3, 7 and 12 min. Gene expression data highlighted a profound modification of gene expression, with 46% of the genes being altered. This analysis revealed a slow-down of Bi. longum general metabolic activity during stress with a simultaneous activation of the classical heat shock stimulon. Moreover, the expression of several genes with unknown function was highly induced under stress conditions. Three of these were conserved in other bacteria species where they were also previously shown to be induced by high temperature, suggesting their widespread role in the heat stress response. Finally, the implication of the trans-translation machinery in the response of Bi. longum cells to heat shock was suggested by the induction of the gene encoding the tmRNA-associated small protein B (SmpB) with concomitant high constitutive expression of the tmRNA gene.

  • Bifidobacterium
  • heat shock
  • microarray
  • transcriptome


Bifidobacteria are Gram-positive, heterofermentative and anaerobic bacteria that are found as natural inhabitants in the gut of humans and other warm-blooded animals (Ventura et al., 2004b). The interest in bifidobacteria has grown significantly mainly because their presence has been associated with a healthy microbiota, which led to their widespread use as functional food supplements (Picard et al., 2005; Chermesh & Eliakim, 2006; Szajewska et al., 2006).

To survive in the harsh gastrointestinal tract conditions, these microorganisms need to be able to adapt to a competitive and changing environment. Furthermore, for industrial applications, selected strains must respond to the adverse conditions encountered during the production process, for instance high temperatures during spray drying (Simpson et al., 2005) as well as during storage of the final product. The importance of inducible cell-protective systems for survival under stressful conditions has been shown in bifidobacteria (Schmidt & Zink, 2000; Maus & Ingham, 2003; Saarela et al., 2004) and in other microorganisms (Flahaut et al., 1996; Saarela et al., 2004). Moreover, the mechanisms underlying these stress responses have been extensively investigated in many bacterial species (Richmond et al., 1999; Yura et al., 2000; Helmann et al., 2001; Stewart et al., 2002; Weiner et al., 2003; De Angelis & Gobbetti, 2004; Gao et al., 2004; Barreiro et al., 2005). However, relatively little is known for bifidobacteria, partly due to the paucity of efficient molecular genetic tools for the analysis of gene functions. The analysis of the genome sequence of Bifidobacterium longum NCC2705 (Schell et al., 2002) revealed a number of physiological traits that partially explain its successful adaptation to the gut environment and the identification of conserved bacterial stress response genes allowed predicting some of its stress-protective mechanisms (Klijn et al., 2005). Using a proteomic approach, several stress-induced proteins have been identified in Bi. longum cells upon exposure to bile salts (Sanchez et al., 2005; Savijoki et al., 2005) or high temperatures (Savijoki et al., 2005). These included general stress-related chaperones, proteins involved in transcription and translation and few proteins from different metabolic pathways. A recent proteomic analysis performed on Bi. longum cells grown on different culture media showed that 21% of the predicted CDS were expressed and few of them were shown to display substrate-dependent expression (Yuan et al., 2006). Finally, a series of studies have been recently published describing the genetic organization and transcriptional regulation of few major stress-related chaperones of Bifidobacterium breve (Ventura et al., 2004a, 2005ae; Klijn et al., 2005). However, till date there have been no reports on genome-wide gene expression changes induced in a Bifidobacterium strain upon exposure to a stress stimulus.

In this study, whole-genome DNA microarrays covering 97% of all Bi. longum NCC2705 CDS were used to analyse the time-dependent gene-expression changes induced in response to a severe sublethal heat shock. This allowed us to identify the Bi. longum heat stress response mechanisms and to compare them with other bacterial species.

Materials and methods

Bacterial growth conditions, heat stress and cell counting

Bifidobacterium longum NCC2705 was inoculated at 1% from an overnight culture into a 7 L bioreactor (new MBR Bioreactor AG, Zurich, Switzerland) filled with 4 L of MRS medium (BD, Basel, Switzerland). Cultures were grown at 37°C under anaerobic conditions. Bacterial cells suspensions (400 mL) were harvested in mid-exponential phase (OD600 nm of 0.7–0.8) and split into two aliquots that were immediately centrifuged at room temperature (RT) for 5 min at 3600 g. The supernatants were rapidly removed and the pellets suspended in prewarmed MRS medium at either 37 or 50°C and incubated in a water bath at the corresponding temperature. Ten milliliter culture aliquots were collected simultaneously at 3, 7, 12 and 19 min from the two cell suspensions and immediately centrifuged at RT for 5 min at 3600 g. Bacterial pellets were then immediately frozen in liquid nitrogen. For estimation of cell survival, culture aliquots were collected from the different time points and plated at three dilutions on MRS agar and incubated for 2 days in anaerobic jars before cell counting.

Production of microarrays, RNA extraction and hybridization

Glass microarrays spotted with 200–800 bp-long PCR-based amplicons targeting c. 97% of all Bi. longum NCC2705 CDS were described elsewhere (Parche et al., 2006). RNA extraction, RNA quality checks, labeling and hybridization were performed as previously described (Parche et al., 2006). Hybridizations were performed using RNAs from three independent experimental biological replicates for each time point and technical hybridization replicates were performed for some of the RNAs. Scanarray 4000 (Packard Biochip Technologies, Billerica, MA) was used for scanning of microarrays.

Microarray data analysis

Imagene 5.6 (Biodiscovery, El Segundo, CA) was used to extract data from the scanned images. The local average background was subtracted from the average spot signal values. Spots with low signal intensities for both Cy3 and Cy5 (i.e. less than threefold the average local background) were excluded, whereas spots displaying low signal intensities in only one channel were retained if the signal in the other channel was at least fivefold higher than the corresponding average local background signal. A within-array normalization was performed on the log2-transformed gene expression ratios (M-values) to bring the median of combined M-values within each array to zero. M-values from spot duplicates on each array were averaged first and then M-values from independent technical hybridization replicates were combined. As a result of these data-processing steps, M-values from the three biological replicates were obtained for the majority of the genes for each time point. Statistical analysis was based on an empirical Bayes method (Lönnstedt & Speed, 2002; Hatfield et al., 2003; Smyth, 2004) implemented in the software package limma for the R-computing environment (http://www.r-project.org). The threshold of significance was set to 1/no (α=~0.0007), where no is the number of genes with an expression value in at least two out of three biological repeats. Finally, genes were considered as differentially expressed if their absolute log2-transformed signal ratios were greater than 1 and their P-values smaller than α.

Quantitative reverse transcriptase-PCR (qRT-PCR)

Oligonucleotide primers (Table S1) were designed with the primer express software (Applied Biosystems, Foster City, CA). Validation of microarray data was carried out by qRT-PCR measurements on 11 genes selected according to their microarray gene expression values including four constitutively expressed (BL0118, BL0274, BL0301, BL1800), five significantly induced (BL0517, BL0519, BL1180, BL1250, BL1672) and two significantly repressed genes (BL0279, BL1182). cDNA was synthetized according to the protocol recommended by the supplier using random hexamers and the TaqMan Reverse Transcription Reagents (Applied Biosystems). Real-time PCR was performed on an ABI PRISM 7000 machine (Applied Biosystems) using the SYBR Green PCR Master Mix (Applied Biosystems), following the recommended protocol. Quantitative values were obtained using the comparative threshold-cycle method (ΔCT). Gene expression was assayed in RNAs from control and heat shock-treated samples from all three experimental replicates. Quantification of relative expression of ssrA and smpB (BL1180) genes was performed using the ΔΔCT normalization method using three housekeeping genes (BL0118, BL0274, BL0301) that were previously selected based on microarray data showing their constitutive expression under different experimental conditions (data not shown).

Operon prediction and classification of Bi. longum genes into functional categories

The number of Bi. longum genes organized in operons was identified using the method recently published by Price (2005). Classification of the NCC2705 genes into different clusters of orthologous groups (COGs) was based on the most recent COG classification system (66 genomes; http://www.ncbi.nlm.nih.gov/COG). Percentages of differentially expressed genes in each COG functional category were calculated by dividing the number of significantly induced/repressed genes in each category by the total number of retained genes in the corresponding category (genes were retained if expression data from at least two experimental replicates were available). Statistical analysis of the differential expression observed in each COG category was based on hyper-geometric distribution. The P-values assigned to the different COG categories express the significance of the difference observed between the number of differentially expressed genes measured in a given COG category and the overall level of differential expression.

Results and discussion

Survival of Bi. longum following heat shock

To study the changes in gene expression occurring in Bi. longum in response to heat stress, optimal sublethal stress conditions were experimentally defined in order to induce extensive differential gene expression without compromising the viability of the cells. After preliminary tests at different temperatures, it was decided to apply a temperature of 50°C to bacterial suspensions harvested in mid-exponential phase. Samples were collected 3, 7, 12 and 19 min after beginning of the heat shock and cell survival was assessed by counting cfu. No significant difference in survival rates was observed between control and heat-treated cells during the first 12 min (Fig. S1). However, after 19 min a reduction of cell viability was observed in one of the three experimental replicates (Fig. S1; −0.75 log-units), and hence it was decided that the focus of transcriptome analysis would be on the first 12 min.

Bifidobacterium longum rapidly reacts to heat treatment with a pronounced change in gene expression

The complete list of the average gene expression values obtained at 3, 7 and 12 min is presented in Table S2. Average expression data were further considered only for the genes exhibiting significant expression in at least two biological replicates (hereinafter defined as the transcriptionally active genes). This condition was fulfilled by 83% of all Bi. longum NCC2705 genes. Microarray data were validated by qRT-PCR measurements on eleven genes revealing a high level of correlation (R2=0.93) between the two methods (Fig. S2). Figure 1a summarizes the number of genes differentially expressed at each time point. In total, 442 genes (206 induced and 236 repressed) at 3 min, 655 genes (314 induced and 341 repressed) at 7 min and 625 genes (290 induced and 335 repressed) at 12 min displayed an absolute differential expression ≥twofold in response to the applied heat stress. This corresponds to 56% of the transcriptionally active genes being differentially expressed at least at one time point, i.e. 46% of all Bi. longum genes. According to an automated operon prediction method (Price et al., 2005), 52% of the differentially expressed genes were organized in 215 different operonic structures. It is noteworthy that the lowest inter-experimental variability was observed at 7 min, a time point that is sufficiently advanced to enable establishment of a viable and uniformly responding cell population (see distribution of statistically significant differentially expressed genes, Fig. 1b). The impact of heat treatment on Bi. longum gene expression is more pronounced compared with data obtained in other species. For instance, when Ba. subtilis (Helmann et al., 2001), Escherichia coli (Richmond et al., 1999) and Campylobacter jejuni (Stintzi, 2003) were subjected to heat treatments of 48, 50 and 42°C, respectively, only up to 20% of the genes were found to be differentially expressed. Moreover, in contrast to previous studies showing a transient transcriptome reorganization upon exposure to high temperature (Helmann et al., 2001; Stintzi, 2003; Gao et al., 2004), in Bi. longum most of the genes which were differentially expressed, remained induced and repressed throughout the time course. The most likely explanation for these differences is that the experimental stress conditions of this study were closer to the lethal threshold. Furthermore, different mechanisms of transcriptional regulation between these phylogenetically distant bacterial species might also explain the observed major effect of a heat treatment on Bi. longum transcriptional profile. To obtain an overview of the impact of high temperature on bacterial physiology, NCC2705 differentially expressed genes were grouped into functional categories according to the COG classification system (Tatusov et al., 1997). A clear bias towards heat-dependent regulation of genes involved in few specific cellular processes was observed (Fig. 2) and the functional categories exhibiting significantly different percentages of regulated genes will be discussed individually in the following sections.


Impact of the heat shock treatment on the global gene expression levels in Bifidobacterium longum after 3, 7 and 12 min at 50°C. Only genes with expression data from at least two experimental replicates are shown. (a) Venn diagram displaying the number of differentially expressed genes at each time point; numbers in parentheses represent repressed genes (log2-transformed ratio ≤−1), while the other numbers show induced ones (log2-transformed ratio ≥1). Genes that were found to be up or downregulated in two or three time points are indicated in the intersections. (b) ‘M vs. A’ plots displaying the global distribution of relative gene expression values [log2(ratio)] as a function of the log2-transformed average absolute gene expression levels in control and heat-treated cells. The horizontal lines indicate the threshold of twofold induction, respectively repression. Black dots indicate genes displaying statistically significant differential expression (P<0.0007).


Differentially expressed genes categorized by functional classification according to the COG annotation at 3 (white bars), 7 (grey bars) and 12 min (black bars) after beginning of the heat shock. Percentages of induced, respectively repressed, genes in each category was calculated by dividing the number of significantly induced/repressed genes in each category by the total number of retained genes in the corresponding category. *, statistical significance (*, P-value≤0.05; **, P-value≤0.005; ***, P-value≤0.0005). Functional categories: J, translation, ribosomal structure and biogenesis; A, RNA processing and modification; K, transcription; L, replication, recombination and repair; D, cell cycle control, cell division, chromosome partitioning; V, defense mechanisms; T, signal transduction mechanisms; M, cell wall/membrane/envelope biogenesis; N, cell motility; U, intracellular trafficking, secretion, and vesicular transport; O, posttranslational modification, protein turnover, chaperones; C, energy production and conversion; G, carbohydrate transport and metabolism; E, amino acid transport and metabolism; F, nucleotide transport and metabolism; H, coenzyme transport and metabolism; I, lipid transport and metabolism; P, inorganic ion transport and metabolism; Q, secondary metabolites biosynthesis, transport and catabolism; R, general function prediction only; S, function unknown.

Extensive downregulation of genes involved in cell growth

The majority of the genes belonging to COG categories J (translation, ribosomal structure and biogenesis) were repressed 7 min after exposure to high temperature conditions (Fig. 2). This category contains mainly genes encoding proteins involved in the translation machinery, such as ribosomal proteins and aminoacyl-tRNA synthetases and several of them are also found among the genes exhibiting the strongest level of transcriptional repression (Table S3). A similar observation was also reported for the food-borne pathogen C. jejuni (Stintzi, 2003). This down-regulation of ribosomal genes could represent a survival strategy of stressed bacteria that repress their most energy-demanding cellular processes for the benefit of other protective functions. In line with this hypothesis, genes involved in cell division and chromosome partitioning (COG category D) were also rapidly down-regulated. These observations probably reflect a heat-induced halt of Bi. longum cell division. In fact, when Bi. longum was inoculated on plates and incubated at different temperatures ranging from 37 to 50°C, no growth was observed at temperatures equal to or higher than 45°C (data not shown). Similarly, data from a previous study on another heat-stressed Bi. longum strain (Savijoki et al., 2005) showed that shifting the cells from 37 to 47°C resulted in a 50–80% reduction of the growth rate. Finally, genes encoding proteins involved in other essential processes linked to an active metabolic state like cell-envelope biogenesis (M), energy production and conversion (C), and nucleotide transport and metabolism (F) were also primarily repressed.

Induction of the dnaK and groE chaperone systems

A statistically significant overrepresentation of induced genes was furthermore found amongst members of the COG functional category O (posttranslational modification, protein turn-over, chaperones), which includes most genes belonging to the classical and extensively studied heat shock stimulon (Fig. 2). The dnaK operon (BL0516–BL0520), which includes genes encoding the major molecular chaperones DnaK, GrpE, DnaJ as well as the transcriptional repressor HspR, were among the most highly and rapidly induced genes. Similarly, the genes encoding the second major molecular chaperone system, namely groEL (BL0002) and groES (BL1558) and the corresponding transcriptional regulator HrcA (BL0718), were also induced but at lower levels (Tables 1 and 2). In Bifidobacterium breve, the dnak operon was not induced at mild heat shock temperatures of 43°C (Ventura et al., 2005e) while members of the groE machinery were shown to display the highest expression levels at 43°C in a range from 37 to 50°C (Ventura et al., 2004a). The different regulatory patterns displayed by these two major chaperone systems probably reflect their distinct and complementary roles in different stress responses as already described in other bacterial species (Susin et al., 2006). The role of dnaK in survival to heat stress was recently confirmed using a Bi. longum NCC2705 mutant constitutively overexpressing the dnaK operon, which showed significantly improved resistance to high temperature (B. Berger, unpublished results).

View this table:

Overview of gene expression ratios of potential stress-related genes

GeneRatio 3 min7 min12 minFunction
BL00012.43.63.3Cold shock protein (CspA)
BL00025.84.54.3Chaperone protein (GroEL)
BL00071.01.51.5Cold shock protein (CspB)
BL00090.90.90.8Universal stress protein
BL00100.60.50.5ATP-dependent endopeptidase clp ATP-binding subunit (ClpC)
BL00941.20.81.0ATP-dependent endopeptidase
BL051614.213.516.2Heat-inducible transcriptional repressor (HspR)
BL051714.918.214.6Chaperone protein (DnaJ)
BL051910.915.812.7GrpE protein (HSP-70 cofactor)
BL052012.716.210.1Chaperone protein (DnaK)
BL05511.21.21.3Possible endopeptidase (HtpX)
BL05556.46.05.5DO serine protease (DegP or HtrA)
BL05769.815.413.9Probable HSP20-family heat shock chaperone
BL07183.83.64.2Heat-inducible transcription repressor (HrcA)
BL07191.51.10.8Chaperone protein (DnaJ)
BL09430.90.40.9ATP-dependent endopeptidase clp ATP-binding subunit (ClpX)
BL09440.80.50.5ATP-dependent endopeptidase clp proteolytic subunit (ClpP)
BL09450.80.60.6ATP-dependent endopeptidase clp proteolytic subunit (ClpP)
BL10433.73.73.3DNA repair protein (RecN)
BL11804.46.16.0SsrA-binding protein (SmpB)
BL125018.416.311.2Endopeptidase Clp ATP-binding chain B (ClpB)
BL13103.94.43.3SOS-response transcriptional repressor (LexA)
BL13571.61.62.0RNA polymerase ECF-type sigma factor
BL13583.54.23.1ECF-type sigma factor negative effector
BL14149.315.19.6Transcriptional activator (ClgR; probe 1)
BL14142.93.12.8Transcriptional activator (ClgR; probe 2)
BL14152.11.82.0RecA protein
BL14163.02.32.0RecX protein
BL14183.03.32.4Ribosome-associated factor Y
BL14281.21.21.2RNA polymerase principal sigma factor; sigma 70
BL14390.90.91.2GTP pyrophosphokinase (RelA)
BL15585.74.14.010 kDa chaperonin (GroES)
BL16641.61.31.1Widely conserved protein in universal stress protein family
  • Significantly induced genes (P<0.0007).

  • Ratio of gene expression between heat-shock treated cells and control cells.

View this table:

Fifty genes displaying the strongest level of induction after 7 min at 50°C

GeneRatio 5 min7 min12 minFunction
BL051714.918.214.6Chaperone protein (DnaJ)
BL125018.416.311.2Endopeptidase Clp ATP-binding chain B (ClpB)
BL052012.716.210.1Chaperone protein (DnaK)
BL051910.915.812.7GrpE protein (HSP-70 cofactor)
BL05769.815.413.9Probable HSP20-family heat shock chaperone
BL14149.315.19.6Possible DNA binding protein (transcriptional regulator)
BL051614.213.516.2Heat shock transcriptional regulator (HspR)
BL02657.512.19.8Fragment of arabinose permease
BL057313.111.88.6Hypothetical protein
BL125111.311.89.1Glutamyl-tRNA synthetase
BL13115.78.810.2Hypothetical protein with LysM domain of membrane-bound lytic murein transglycosylases
BL057110.78.75.4Putative esterase
BL01749.38.17.3Narrowly conserved hypothetical protein
BL05727.37.75.8Hypothetical membrane protein with unknown function
BL05995.17.56.9Hypothetical protein
BL06626.17.57.2Hypothetical protein
BL00037.97.43.9Hypothetical protein
BL17274.87.35.9Hypothetical protein
BL05234.57.14.8Possible xylosidase or glucosidase
BL16733.96.74.9Possible lactaldehyde reductase
BL01754.86.47.0Probable AraC/XylS-type transcriptional regulator
BL13096.06.33.4Probable cation efflux protein
BL11646.16.12.8Probable solute binding protein of ABC transporter system for sugars
BL11804.46.16.0SsrA-binding protein (SmpB)
BL05556.46.05.5Possible DO serine protease
BL17002.85.64.3Polypeptide deformylase
BL05323.95.64.4Transmembrane transport protein possibly for shikimate
BL08413.85.44.9Hypothetical protein
BL14912.65.42.3Hypothetical protein
BL18044.85.32.9Hypothetical protein
BL15203.35.23.9Possible NagC/XylR-type transcriptional regulator
BL01074.05.16.1Probable LacI-type transcriptional regulator
BL03314.45.03.4Probable PfkB family carbohydrate (sugar) kinase
BL11982.75.03.7Probable serine/threonine-protein kinase
BL06613.34.96.4Hypothetical protein
BL01613.24.94.5Hypothetical protein
BL03684.44.93.8Hypothetical protein
BL02022.24.94.7Hypothetical protein
BL11892.54.95.3Hypothetical protein
BL04225.44.94.8Narrowly conserved hypothetical protein
BL04233.54.85.8Sugar permease of ABC transporter system
BL12932.54.86.2Similar to xylulose kinase
BL14373.44.85.1Hypothetical cytosolic proteins
BL18034.04.87.6Hypothetical protein
BL02612.04.84.5ABC-type sugar transport systems, permease components
BL15595.64.74.1Cystathionine gamma-lyase
BL01543.74.63.2ATP binding protein of ABC transporter
BL01893.14.62.5Sugar permease of ABC transporter system
BL00025.84.54.3Chaperone protein (GroEL)
BL12162.84.53.6Probable transcriptional regulator with cyclic nucleotide-binding domain
  • Significantly induced genes (P<0.0007).

  • Ratio of gene expression between heat-shock treated cells and control cells.

Regulation of proteolytic enzymes

None of the genes encoding the subunits of the Clp protease complex were induced upon heat shock, namely the substrate-recognition subunits ClpC (BL0010) and ClpX (BL0943) and the genes encoding the proteolytic subunits ClpP1 (BL0944) and ClpP2 (BL0945) (Table 1). Similarly, Bifidobacterium breve clpC, clpP1 and clpP2 genes were recently shown to be induced in response to moderate heat shock regimens, but not at 50°C (Ventura et al., 2005b, d). The constitutive expression of these genes was unexpectedly accompanied by strong induction of the Bi. longum transcriptional activator ClgR (BL1414; Tables 1 and 2) that was proposed to be involved in the induction of the ClpP locus in Bifidobacterium breve under moderate temperature-shock conditions (Ventura et al., 2005b, d). HtrA (BL0555), a widely distributed serine protease exhibiting both chaperone and protease activities in many bacteria (Pallen & Wren, 1997), was strongly induced at all time points in Bi. longum (Tables 1 and 2), suggesting its likely implication in the degradation of heat-induced misfolded proteins.

Induction of transcriptional regulators

As shown in Fig. 2, a significant overrepresentation of induced genes was found within the COG category K (transcription) including 30 induced transcriptional regulators, one probable serine/threonine-protein kinase (BL1198) and a transcriptional regulator belonging to the ROK family (BL0074) (Table S4). The list of induced transcriptional regulators included 2 two-component system (TCS)-response regulators (BL0316, BL0903), which displayed similar induction levels as the corresponding membrane-bound sensor histidine kinases (BL0317 and BL0902; see Table S2). TCS are well-known signal transduction systems employed by bacteria to detect environmental changes and mediate differential gene expression (Mascher et al., 2006). Surprisingly, 10 LacI-type sugar-responsive-repressors were also found among the induced regulators suggesting an effect of the heat treatment on carbohydrate metabolism. Unfortunately, a precise functional role could be assigned only to few regulators including the above-mentioned HspR and HrcA (Narberhaus, 1999) and the transcriptional repressor LexA (BL1310) (Table 1). In E. coli and in several other bacterial species, LexA controls the SOS response system, including genes involved in inhibition of cell division, error-prone replication or excision repair (Fernandez De Henestrosa et al., 2000; Campoy et al., 2005). LexA was shown to interact with the RecA protein, which acts as a sensor of environmental stresses inducing DNA damage and promotes the autocatalytic cleavage of LexA, hence preventing its binding to target recognition sequences. In this study, recA (BL1415), recX (BL1416), another member of the RecA-mediated signaling pathway, and recN (BL1043), which encodes a protein involved in DNA repair were also slightly induced upon heat stress. Taken together, these results indicate that stressed Bi. longum respond to changes/defects/errors in their chromosomal DNA, by induction of their SOS response system.

Induction of trans-translation protein salvage pathway

Ribosome ‘stalling’ can occur in bacteria for different reasons, e.g. as consequence of low tRNA levels, lack of stop codons in truncated mRNAs or inefficient translation termination. In a process known as trans-translation, a tmRNA molecule (encoded by the ssrA gene) acts as tRNA and mRNA to append a peptide tag to the nascent polypeptide and ‘rescue’ the ribosome (Withey & Friedman, 2003; Moore & Sauer, 2005). The tagged proteins are then rapidly degraded by specific proteases. In Ba. subtilis, tmRNA is induced under severe stress conditions and its presence is required for efficient growth (Muto et al., 2000). SmpB is a dedicated tmRNA-binding protein that is essential for the trans-translation process (Karzai & Sauer, 2001). Microarray data obtained in this study revealed a significant and strong induction of smpB (BL1180) at all time points (Tables 1 and 2) that was further confirmed by qRT-PCR. Because the original Bi. longum microarray design did not include ssrA gene-specific probes, qRT-PCR was used to assess the relative ssrA gene expression levels in heat shock treated and control cells. Surprisingly and in contrast to what was observed in Ba. subtilis, ssrA was strongly and constitutively expressed in Bi. longum. To the best of the authors' knowledge, this is the first report that describes the regulation of the tmRNA/SmpB system in response to heat stress to occur by high constitutive expression of the ssrA gene with a concomitant induction of the smpB gene. These results suggest that in Bi. longum, induction of SmpB is required under stressful conditions in order to effectively protect tmRNA molecules against degradation by endogenous RNAses. A model supporting this hypothesis was recently proposed in Caulobacter crescentus, where SmpB was shown to protect tmRNA from RNAse R degradation in vitro, and the levels of SmpB protein were shown to correlate with tmRNA stability in vivo (Hong et al., 2005).

Induction of genes with unknown function

Many of the genes whose expression was altered by a temperature increase encode proteins without homology to any members of a COG group (Fig. 2). In an effort to identify new general stress response genes, the 141 hypothetical genes significantly induced at 7 min were compared using blastp against a nonredundant bacterial protein database (NCBI). Potential gene orthologs were further considered only if the query sequence displayed an amino acid sequence identity greater than 30% over at least half of its length. For the 103 Bi. longum NCC2705 sequences fulfilling these criteria, the literature was searched for published stress-related global gene expression studies on the species carrying corresponding gene orthologs. This analysis resulted in the identification of three NCC2705 genes with a hit in another organism, in which the corresponding orthologous gene was previously shown to be induced under similar heat shock conditions: BL0408 protein displayed 52% identity over 87% of its length with the Ba. subtilis yvgZ-encoded unknown protein (Helmann et al., 2001); BL0503 displayed 39% identity over 62% of its length with the Mycobacterium tuberculosis Rv3654c-encoded hypothetical protein (Stewart et al., 2002); finally BL0850 was 33% identical over 75% of its length with the Shewanella oneidensis SO3342-encoded hypothetical protein (Gao et al., 2004). The presence of potential orthologs of these genes induced under similar heat-stress conditions in relatively distant bacterial species suggests that they might play a common role in the response to this stress stimulus.


This study describes the first genome-wide gene expression profile analysis of a Bifidobacterium species in response to heat stress, a condition that is highly relevant to the industrial production of probiotic strains. Results of this study show that 46% of the Bi. longum genes were differentially expressed. These include common bacterial stress-responses such as the induction of chaperones and the down-regulation of the translation, cell division and chromosome-partitioning machineries. In parallel, several transcriptional regulators were rapidly induced suggesting their involvement in the observed reshuffle of the gene-expression profile. Some less-characterized responses were also identified, such as the induction of the trans-translation machinery, which could play a major role for survival of Bi. longum cells under high temperature conditions. Overall, these data significantly contribute to the understanding of Bi. longum stress-response mechanisms and build the basis for future more in-depth functional studies.

Supplementary material

The following supplementary material is available for this article:

Tab. S1. Oligonucleotide primers used for qRT-PCR.

Tab. S2. Bifidobacterium longum NCC2705 gene expression values at 3, 7 and 12 minutes after beginning of the HS.

Tab. S3. 50 genes displaying to the strongest level of repression after 7 minutes at 50°C.

Tab. S4. List of genes belonging to the COG functional category K induced after 7 minutes at 50°C.

Fig. S1. Comparison of the survival rate between control and heat-treated cultures during the analyzed time period, measured by plate counting method. Values for each group at each time point represent the average of three experimental replicates.

Fig. S2. Validation of microarray results by real-time qRT-PCR. Sports represent single log2-transformed relative gene expression ratios between heat-stressed and control bacteria, measured by microarray (X-axis) or by real-time RT-PCR (Y-axis) at 3 (○) 7 (*) and 12 minutes (◆) after beginning of the heat shock.

This material is available as part of the online article from: http://www.blackwellsynergy.com/doi/abs/10.1111/j.1574-6968.2007.00704.x (This link will take you to the article abstract). Please note: Blackwell Publishing is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.


A. Klijn is thanked for fruitful discussions, and R.D. Pridmore, B. Berger and C.H. Ahrens for critical reading of the manuscript.


  • Editor: Ezio Ricca


View Abstract