Changes in global gene expression of Vibrio parahaemolyticus induced by cold- and heat-stress

Vibrio (V.) parahaemolyticus causes seafood-borne gastro-intestinal bacterial infections in humans worldwide. It is widely found in marine environments and is isolated frequently from seawater, estuarine waters, sediments and raw or insufficiently cooked seafood. Throughout the food chain, V. parahaemolyticus encounters different temperature conditions that might alter metabolism and pathogenicity of the bacterium. In this study, we performed gene expression profiling of V. parahaemolyticus RIMD 2210633 after exposure to 4, 15, 20, 37 and 42 °C to describe the cold and heat shock response. Gene expression profiles of V. parahaemolyticus RIMD 2210633 after exposure to 4, 15, 20, 37 and 42 °C were investigated via microarray. Gene expression values and RT-qPCR experiments were compared by plotting the log2 values. Moreover, volcano plots of microarray data were calculated to visualize the distribution of differentially expressed genes at individual temperatures and to assess hybridization qualities and comparability of data. Finally, enriched terms were searched in annotations as well as functional-related gene categories using the Database for Annotation, Visualization and Integrated Discovery. Analysis of 37 °C normalised transcriptomics data resulted in differential expression of 19 genes at 20 °C, 193 genes at 4 °C, 625 genes at 42 °C and 638 genes at 15 °C. Thus, the largest number of significantly expressed genes was observed at 15 and 42 °C with 13.3 and 13 %, respectively. Genes of many functional categories were highly regulated even at lower temperatures. Virulence associated genes (tdh1, tdh2, toxR, toxS, vopC, T6SS-1, T6SS-2) remained mostly unaffected by heat or cold stress. Along with folding and temperature shock depending systems, an overall temperature-dependent regulation of expression could be shown. Particularly the energy metabolism was affected by changed temperatures. Whole-genome gene expression studies of food related pathogens such as V. parahaemolyticus reveal how these pathogens react to stress impacts to predict its behaviour under conditions like storage and transport.


Background
Vibrio (V.) parahaemolyticus is one of the causes of seafood-borne gastro-intestinal infections in humans worldwide [1]. It is widely found in marine environments and is isolated frequently from seawater, estuarine waters, sediments and raw or insufficiently cooked seafood (e.g. shrimp or bivalve molluscs) [2][3][4]. Consumption of or contact to raw or undercooked seafood containing V. parahaemolyticus in relevant numbers, might lead to human infections, mostly associated with gastroenteritis [5,6].
Different studies investigated the behaviour of V. parahaemolyticus under environmental stresses on the phenotypic level (e.g. cold shock, heat shock, high salt concentrations or bile supplementation) [7][8][9]. Nonetheless, the general mechanism of adaptation and survival under these conditions are not elucidated yet.
Within its ecological habitat and food chain, V. parahaemolyticus encounters changing temperature conditions. These temperature shifts will result in metabolic changes.
A cold shock resulting from a rapid downshift of the temperature, e.g. changing water temperatures or storage on ice, alters bacterial gene expression [10][11][12][13]. However the expression of V. parahaemolyticus resulting from cold shock is still poorly understood. The coldinduced gene expression profile of a clinical V. parahaemolyticus strain at 10°C has been examined by Yang et al. [13] in a time course analysis. Significant differential expression of almost 13 % of genes (n = 619) investigated, was found.
Temperatures in the marine habitat of V. parahaemolyticus usually do not exceed 25°C. In V. parahaemolyticus several stress proteins, e.g. heat shock protein (hsp) families such as Hsp60 (GroEL and GroES) and as Hsp70 (DnaJ, DnaK, GrpE) are produced in response to elevated temperatures [14,15]. In general those proteins are made in substantial amounts acting as chaperones, protecting cells from heat dependent denaturation [16][17][18]. In V. parahaemolyticus especially Hsp60 family proteins serve as general stress proteins and are found in several cell compartments and in substantial amounts [19].
In addition, changing temperature conditions can affect the pathogenicity of V. parahaemolyticus [19]. Chiang and Chou [20] demonstrated increased pathogenicity after heat shock response in V. parahaemolyticus as elevated toxin expression. Clinical strains alter expression of systems regulating virulence as well as systems indirectly related to host-pathogen attachment such as biofilm production and motility at 37°C [21]. However, environmental strains did not show this behaviour or exhibit decreased expression of biofilm production or motility related genes at higher temperatures. Sublethal heat shock of V. parahaemolyticus resulted in elevated expression levels of the gene encoding the thermostable direct hemolysin (TDH), one of the two prominent toxins enhancing its pathogenicity [19].
The aim of this study was to investigate gene expression profiles of V. parahaemolyticus after exposure to 4, 15, 20, 37 and 42°C. Moreover high regulation clusters e.g. toxins produced in response to temperature changes were to be identified.

Results and discussion
Understanding temperature-dependent changes in bacterial gene expression patterns is crucial when studying tenacity, invasion, and environmental related viability of bacterial species. Temperature-dependent expression changes as cues for tenacity and persistence within matrixes such as food or hosts and environment has led to genetic approaches defining temperature-induced genes of pathogens [22][23][24]. However, temperature-dependent induction of genes is an arbitrary parameter because appropriate temperatures for comparison to any other temperature must be assumed. In this study, we investigated temperature-dependent gene expression of V. parahaemolyticus in comparable growth phases under different temperatures.

Validation of microarray results
To confirm the results of microarray data analysis, a quantitative RT-qPCR was used. Six house-keeping genes were chosen to compare the data of the two techniques, whereof four were applied in the multilocus sequence typing (MLST) scheme of V. parahaemolyticus [25]. The house-keeping genes were encoded on both chromosomes, with one exception (cspA): cspA, dtdS, groES, pvsA, pyrC and tnaA. Four additional MLST genes used for normalization: pvuA, dnaE, recA and one locus of the 16S-23S intergenic spacer region. Gene expression values of microarray and RT-qPCR experiments were compared by plotting the log 2 values of both experiments against each other. An overall positive correlation (R 2 = 0.7008) between the two techniques could be shown (Fig. 1). The similarity of replicate samples at different temperatures was studied using hierarchical clustering with correlation as the distance measure (Fig. 2a). The samples at 42°C form the clearest cluster. Samples of 20 and 37°C cluster according to the temperature. Moreover, volcano plots of microarray data were calculated to visualize the distribution of differentially expressed genes at individual temperatures and to assess quality and comparability of hybridizations (Fig. 2b). Additionally, volcano plots enable the quick identification of expression changes within the gene sets by combination of statistical tests (adjusted p-value) and magnitude of changes.

Gene expression at 4°C, 15°C, 20°C and 42°C
We compared gene expression patterns within a temperature range of 4 to 42°C. Additionally, Database for Annotation, Visualization and Integrated Discovery (DA-VID) analyses were performed, highlighting regulation of genes connected in metabolic pathways (Additional file 1). Among all conditions, the strongest expression changes (regarding the number of differentially expressed genes and intensity of expression changes) were observed at 15°C (13.3 % of all genes) and 42°C (13 % of all genes). Since the highest number of genes with stable expression was found at 37°C, this temperature was chosen as reference. Genes with an adjusted p-value ≤0.05 and an absolute logarithmic fold change ≤±1.5 were considered significantly stable expressed. To demonstrate the temperature-associated differences in gene expression changes, temperature experiments were clustered via K-means-clustering (Fig. 3). The  (Table 1).
Our analysis identified differentially expressed genes under different temperature conditions. Compared to 37 at 4°C 4 % (n = 193) the genes showed significant expression changes, whereas incubation at 20°C resulted in a rate of approx. 0.4 % (n = 19). At 42°C, 13 % (n = 625) of differentially expressed genes were detected. The highest number of genes regulated, however, was found at 15°C with 13.3 % (n = 638) differentially expressed genes. Incubation at 15 and 42°C resulted in almost balanced expression patterns regarding the amount of upand down-regulated genes. At 4°C, 78 % (n = 150) of the significantly differentially expressed genes showed down-regulation, whereas only 22 % (n = 43) showed up-regulation. Additional information can be found in Additional file 2.

Expression of temperature shock response genes
Some gene clusters showed up-regulation of expression under one temperature and down-regulation under another. Chaperone encoding hsp70 family genes, such as dnaK (VP0653), as well as the hsp60 family groEL, groES (VPA0286, VPA0287) showed significant down-regulation of expression at 4, 15 and 20°C. On the contrary, a strong up-regulation at 42°C was observed (Additional file 2). Cold shock responding genes, such as cpsA (VPA1289-1291 and VP1889) as well as a cluster encoding genes classified as ascorbate and phosphotransferase (VPA0229-231) showed significant up-regulation at 4, 15 and 20°C whereas down-regulation occurred at 42°C. Yang et al. [13] investigated time dependent behaviour of a clinical V. parahaemolyticus strain at cold temperatures. Almost 13 % of genes (619 genes) were differentially expressed at least at one of the three points in time investigated [13]. For metabolism related gene categories down-regulation was dominant over up-regulation due to the generally reduced cellular protein pool resulting from a sudden temperature downshift [11].
These findings are confirmed by our data. Moreover under cold temperatures, non-metabolic functions (cell envelope, transport and binding proteins, regulatory functions, cellular processes and mobile and extrachromosomal element functions) as well as genes with unknown or unassigned functions showed a more frequently up-regulation than genes related to cell structure and trans-membrane transporting functions (Additional file 1). The cold shock protein/regulator CspA (VPA1289) showed an over 30-fold enhanced transcription. Additionally, an antagonistic regulation of cold and heat shock genes was detected: heat shock genes encoding heat shock proteins (hsp), ATP-dependent proteases and chaperons were mainly down-regulated after exposure to 10°C [11][12][13]. Our results confirm the findings that metabolism related genes at low temperatures were mainly downregulated and genes without relation to metabolism or of unknown function were mainly up-regulated. Additionally, antagonistic expression of cold and heat shock genes as well as a strong induction of cold shock proteins at low temperatures was observed for V. parahaemolyticus RIMD 2210633 (Additional file 2).

Gene expression at 4°C and 15°C
At 4°C, only 4 % of the genes were differentially expressed. Primarily transcription regulators as well as RNA metabolic process clusters were up-regulated, highlighting the impact of low temperatures (4°C) on the overall gene expression (Fig. 4). Phadtare et al. [12] described concordant findings in E. coli. In our study, at 4°C mainly genes encoding hypothetical proteins, e.g. VP1888, VP2889, VP3030 and VPA1291 were up-regulated.
Additionally, genes of the energy metabolism (VP1381, VP1533, VP2005, VP2666, VP2987, VPA0092) reacted to 4°C by up-regulation. Especially VP1533, encoding a putative ATPase, is of great importance for energy production using glucose [26]. In particular cold shock proteins were highly expressed (cspA VP1889, 4.05 log 2 fold change). These findings, originally described by Yang et al. [13], were confirmed by our data. However, cold temperatures bias gene expression results due to lower activities of e.g. enzymes [27].
The global regulator sigma factor 38, rpoS (VP2553) and the osmoregulator ompR (VP0154) were up-regulated (3.5 and 4.1×). Sigma factor 38 is one of the most crucial sigma factors under e.g. extreme temperatures [28]. No Fig. 2 Overview of microarray results. a The dendrogram represents the result of hierarchical clustering with euclidean distance measure. The first number in the sample label represents temperature, the second number is the replicate number at given temperature. b Volcano plot exemplarily shown for 15°C data. The x-axis represents the log 2 of the fold change plotted against the -log 10 of the adjusted p-value. Red points indicate the differentially expressed genes with at least 2.0 fold change and statistical significance adjusted p <0.05 other sigma factors were up-regulated. Additionally, the tRNA methyltransferase spoU (VP0158) described by Persson et al. [29] was up-regulated (4.1×) as well. Persson et al. [29] were not able to detect differences in growth rates of the E. coli wild-type and a spoU mutant. However, growth temperatures were between 37 and 42°C in that study. Maybe a particular part of tRNA activation can be triggered by low temperatures. Since none of the known genes related to DNA damage VP2034 (imuA), VP2035 (imuB), VP2036 (dnaE2), VP2550 (recA) and VP2945 (lexA) were up-regulated, cold induced DNA damaging, triggering the SOS response, appears to be absent. At 4°C 11 DAVID-gene categories were identified in which a statistically significant number of genes (n = 186, p-value <0.05) was differentially regulated. Nine of these categories were related to transcription, DNA-binding and regulation of RNA metabolism. The two other categories were related to ABC-transporters or transmembrane domains. The expression of genes organized into functional categories at 4°C is shown in Fig. 5. The top five up-and down-regulated genes at 4°C are shown in Table 2.
Especially energy metabolism was down-regulated; out of 75 differentially expressed genes of this category 65 The numbers of differentially expressed genes are given in total as well as in proportion to the number of genes present on the microarray in brackets. Chr1: genes encoded on chromosome 1 encoding 3080 genes of which 3073 genes were represented on the array. Chr2: genes encoded on chromosome 2 encoding 1752 genes of which 1747 genes were represented on the array. Down: down-regulated genes; up: up-regulated genes genes (87 %) were repressed (Fig. 4). Genes related to the pentose phosphate pathway, glycolysis and the citric acid cycle were down regulated. We suggest, that production of central energy molecules such as ATP, NADPH and NADH was decreased because of down-regulated expression of corresponding genes. The vast majority of genes showing highest up-regulation, however, were of unknown function (Additional file 2). The putative virulence-associated protein VacB showed highest upregulation (128×), which has been described to react to environmental signals in Haemophilus influenza [30]. Gene expression of functional categories is shown in Fig. 5. In contrast to 4°C incubation, genes of the amino acid category and de novo DNA synthesis were induced In total 32 DAVID-gene categories with 475 differentially expressed genes were identified at 15°C. Ten categories were associated with transport and transporters. Two major groups formed the most important clusters: integral and intrinsic components of the membrane with 69 (11.6 %) genes each. Additionally, nine metabolism related categories were identified.
At 15°C 32 DAVID-gene categories with 475 genes showed differential regulation. Many of them were connected with membrane maintenance or metabolism. In E. coli it could be shown that at 12°C the membrane composition remains unchanged but enzyme activations are effected [31]. The top five up-and down-regulated genes at 15°C are shown in Table 3.

Gene expression at 20°C
At 20°C, no differential expression of metabolic pathways was detectable. Thus, a range of temperature between 15 and 20°C seems to describe the (lower) physiological border of the normal condition for the strain investigated. A total of 19 genes was differentially expressed. Gene expression is shown in Fig. 5. At 20°C solely genes related to categories associated with the degradation of peptides and proteins were identified: 'peptidase' (15.8 %), 'protease' , 'peptidase activity' and 'proteolysis' (21.1 %), respectively. The top five up-and down-regulated genes at 20°C are shown in Table 4.

Gene expression at 42°C
At 42°C transport and metabolism of carbohydratesrelated genes were up-regulated (Additional file 2). Again, a range of temperature between 42 and 37°C seems to describe the (upper) physiological border for the clinical strain investigated. Interestingly, more genes located on the small chromosome were differentially expressed during incubation at all temperatures. Especially, at 42°C almost twice as many small chromosome genes were differentially expressed. The higher intensity of expression Table 3 Top 5

up-and down-regulated genes at 15°C
Coloured boxes highlight at least 1.5 fold differential expression in either direction: red up-regulated, green down-regulated, fc (fold change) is given log 2 transformed; PTS phosphotransferase system; transcr. reg. transcriptional regulator, put. putative Table 2 Top 5

up-and down-regulated genes at 4°C
Coloured boxes highlight at least 1.5 fold differential expression in either direction: red up-regulated, green down-regulated, fc (fold change) is given log 2 transformed; transcr. reg. transcriptional regulator, put. putative changes in genes located on the small chromosome compared to genes located on the large chromosome can be explained by the higher number of genes related to transcriptional regulation and transport of various substances being located on the small chromosome [32]. Thus, most genes related to environmental stress response are encoded on the small chromosome.
Primarily, genes classified as 'cell metabolism' along with the genes classified as 'unknown' , reacted to the temperature upshift. Altogether, the expression of 625 genes was differentially expressed at 42°C. Expression of categorized genes is shown in Fig. 5. Out of the 87 'cell metabolism' genes, 55 % (n = 48) were classified as 'energy metabolism' related genes.
At 42°C, 38 DAVID-gene categories with a total of 423 differentially expressed genes were identified (Additional file 1). Amongst others, nine categories were related to cell-motion (flagella), eight categories to metabolic processes and six categories to RNA, DNA and transcription. Additionally, three categories were associated with homeostasis (ion, cation, chemical) and two categories with iron-siderophores and transport of siderophores. A distinct cluster on the second chromosome encoding the genes VPA0915-1042 ('cellular processes': n = 23, 'energy metabolism': n = 18,'transport and binding': n = 14,'regulatory functions': n = 14 and 'unknown': n = 36) showed upregulation at 42°C. The top five up-and down-regulated genes at 42°C are shown in Table 5.
However, no prior studies about genome wide gene expression responses exist for the temperatures investigated in this study.

Temperature dependent expression of virulence genes
Virulence genes in total showed no significant expression changes under different temperatures (Additional file 2). The expression of tdh was not significantly influenced by temperature changes, even though slight activation (2.1 log 2 fold change) was observed at 15°C. A putative haemolysin encoding gene (VP3048), was up-regulated at 4 and 15°C. This effect was described by Yang et al. [13], reporting an induction of this putative haemolysin after cold shock. The most prominent haemolysin tdh, however, was not significantly up-regulated (Additional file 2). The associated regulator opaR which recently has been shown to repress expression of T6SS in V. parahaemolyticus is down-regulated at 42°C [33]. We found that, genes located within the virulence pathogenicity island 7 (VPa-7) encoded on the small chromosome, VPA1312-1396 showed no reaction to thermal stimulations (Additional file 2). However, since the energy metabolism was affected Table 4 Top 5

up-and down-regulated genes at 20°C
Coloured boxes highlight at least 1.5 fold differential expression in either direction: red up-regulated, green down-regulated, fc (fold change) is given log 2 transformed; transcr. reg. transcriptional regulator, put. putative especially and mostly at cold temperatures, reduced classical virulence or changed expression rates were to be expected [34].
Virulence associated genes in general (tdh1, tdh2, toxR, toxS, vopC, T6SS-1: VP1386-1420), remained unaffected by heat or cold stress (Additional file 2). The T3SS-1 was found down-regulated at 15°C for nosA (VP1697) and up-regulated for the putative chaperone VP1687 at 42°C. However, the T6SS-1 located on chromosome 1 showed up-regulation at 42°C. This was to be expected since the T6SS-1 system reacts to warm climate in V. parahaemolyticus as described by Salomon et al. [35]. The cold shock gene cspA was downregulated, whereas heat shock genes encoding chaperones and protection via sugar metabolites were induced [13].

Conclusions
Based on our data, the optimal temperature range of the clinical V. parahaemolyticus strain investigated is between 20 and 37°C, since most of the genes were transcribed at a rather constant level.
Finally, it could be shown that the classical pathogenicity markers, T3SSs as well as T6SSs were not upregulated in response to thermal changes. However, large proportions (~30 %) of the differentially expressed genes are of unknown function. Summarized, this study successfully demonstrated that genome-wide gene expression changes in V. parahaemolyticus occur at 4, 15, 20, and 42°C.

Bacterial strains
V. parahaemolyticus RIMD2210633 was isolated from a patient suffering from diarrhoea in Japan in 1996 [32]. This strain harbours the tdh gene, lacks the trh gene and belongs to serotype O3:K6 [36]. This serotype has been detected in clinical as well as in environmental marine samples [37]. The strain has been sequenced by Makino et al. [32].
Prior use, the strain was stored in cryovials at −80°C (Cryobank; Mast Diagnostica, Bootle, England). For initial growth, cells were grown using a rotary shaker (Unimax 1010 and Incubator 1000; Heidolph, Schwabach, Germany) in alkaline peptone water (APW; 0.3 % Yeast-Extract, 1 % Peptone, 2 % NaCl; pH 8.6) at 37°C overnight. A 2 ml aliquot of the resulting culture was diluted to a total volume of 25 ml using APW and grown to an A 600 nm of 0.6. Cultures were grown at 37°C for 3.5 h in order to generate exponential phase cultures. After appropriate dilutions the A 600 was analysed again and aliquots consisting of 10 8 to 10 9 V. parahaemolyticus cells were transferred to 15 ml Falcon tubes, placed in a thermal mixer (Thermomixer comfort; Eppendorf, Hamburg, Germany) and incubated at different temperatures (42, 37 and 20°C) for 30 min. For stressing the cells at 4 and 15°C the entire incubation unit was placed in a conditioning cabinet (Rubarth Apparate, Laatzen, Germany) and bacteria were incubated at these temperatures for 30 min.

RNA preparation and reverse transcription for qPCR investigation
The cultures were centrifuged (2 min, 8000 × g) and the supernatant was discarded. The pellet was immediately resuspended in 1.5 ml RNAprotect Bacteria Reagent (Qiagen, Hilden, Germany) to minimize RNA degradation. Total RNA was isolated using the peqGold Bacterial RNA Kit (Peqlab, Erlangen, Germany). The obtained RNA was eluted into 43 μl of DEPC-treated, DNaseand RNase-free water (Carl Roth, Karlsruhe, Germany). Samples were then treated with DNase I along with Ribolock, an RNase-A, −B and -C inhibitor (Fermentas, Vilnius, Lithuania). RNA quantity was measured by spectrophotometry. RNA quality of each sample was Coloured boxes highlight at least 1.5 fold differential expression in either direction: red up-regulated, green down-regulated, fc (fold change) is given log 2 transformed; transcr. reg. transcriptional regulator, put. putative monitored via gel electrophoresis. Additionally, the RNA quality was assessed using the Agilent RNA 6000 Nano Kit on a 2100 Bioanalyzer (Agilent, Santa Clara, US).
Fluorescence-labeled cRNA generation for the microarray Prior to labelling, the RNA was initially transcribed in cDNA. Briefly, 200 ng of RNA were linear amplified using the full spectrum MultiStart primer (Biocat, Heidelberg, Germany) and Moloney murine leukemia virus reverse transcriptase (Agilent). The amplification was performed at 40°C for 2 h followed by 65°C for 15 min and stored at 4°C. The amplified cDNA, the full spectrum MultiStart primer and T7 RNA polymerase were used along with Cyanine 3-CTP (Agilent) generating labelled cRNA. Labeling was performed using the Quick Amp Labeling Kit (Agilent). The labeled cRNA was purified using the Qiagen RNeasy Mini Kit (Qiagen). A 3 μlaliquot was used for quality control. Experiments were performed using Agilent custom 8 × 15 k arrays (Agilent). The microarray field covers 99.75 % of all V. parahaemolyticus genes. In total, 3073 out of 3080 genes encoded on chromosome 1 and 1747 out of 1752 genes located on chromosome 2, are included. Each gene is represented by 1 to 10 probes (mean 3.15 probes per gene). Each probe consists of a 60mer located preferentially at the 3' terminus of the corresponding gene. The probe design was performed with the eArray Software a web-based Agilent application basing on the genome sequence of V. parahaemolyticus RIMD 2210633 (http://www.ncbi.nlm.nih.gov/genome/691?genome_assembly_id=167995). The cRNA samples were then hybridized to an individual microarray field.

Microarray hybridization and post hybridisation washing
For hybridizations on the microarray, three replicates of independently grown bacterial cultures were prepared for each temperature condition, for 37°C four replicates were used. Accordingly, three individually labeled cRNA sets were prepared for each temperature other than 37°C. Finally, 600 ng of the labeled and linear amplified cRNA was fragmented, added to 25 μl hybridization buffer mix of which a 20 μl aliquot (480 ng) was loaded on a microarray in a hybridization chamber (Biometra, Goettingen, Germany). The onechannel hybridization was performed at 65°C for 17 h and 10 rpm.
Washing of the slides was performed using preheated washing buffer (Gene expression wash buffer kit, Agilent). First the chamber was rinsed with washing buffer. Then the slides were washed once followed by a second washing step using washing buffer containing 0.01 % Triton X-102 (Agilent). The slides were dried using acetonitrile.

Data handling and microarray analysis
Scanning was carried out using the Agilent G2565CA scanner with a resolution of 5 μm. After scanning, tiff-files were analysed and raw data was extracted using Feature Extraction Software (Agilent). Data processing was performed using Bioconductor V 2.12 package of the software R. At first, background corrected spot intensities (signal gProcessedSignal in the Agilent protocol GE1_107_Sep09) were retrieved and bad quality spots were removed using the outlier detection flags of the Agilent protocol. Further, the signal values were normalized using quantile normalization and log 2 transformed [38]. Linear modelling and empirical Bayes methods, implemented in the R package Limma [39], were used to detect the differentially expressed genes between two groups, in this case, the control and treatment sample groups. Raw p-values were adjusted using the Benjamini and Hochberg multiple adjustment method [40]. Genes with an adjusted p-value ≤0.05 and an absolute logarithmic fold change ≥1.5 were considered significantly differentially induced, while genes with an absolute logarithmic fold change ≤ −1.5 were considered repressed. Annotation of genes was performed according to Yang et al. [13] and updated using two new gene entries at NCBI (http://www.ncbi.nlm.nih.gov/ gene), KEGG (http://www.genome.jp/kegg/) and Gene Ontology (http://www.geneontology.org/). Finally, enriched terms were searched in annotations as well as functional-related gene categories using the Database for Annotation, Visualization and Integrated Discovery (DAVID V 6.7, Fisher exact test) [41,42]. The gene lists generated via DAVID enable to highlight gene sets which show a higher proportion of differentially expressed genes compared to other categories. This eases identification of pertinent biological processes to the according temperature. The identified categories are presented in the Additional file 1. K-means clustering of genes with similar gene expression was performed using Genesis V 1.7.6 [43]. Heat maps were generated using BioNumerics V 6.01 (Applied Math, St. Martens-Latem, Belgium). Volcano plots were generated via GraphPad V 5.04, (Graph-Pad, San Diego, US). Integrated graphical views were generated using Circos plot [44]. The transcriptomics data were supplied as experiment GSE60815 at Gene Expression Omnibus according to MIAME regulations. All differentially expressed genes with a log 2 fold change >1.5 and adjusted p-value <0.05 of each condition are supplied in Additional file 2.

qRT-PCR
For generating cDNA, a 1 μg RNA aliquot was used and reversely transcribed by the RevertAid Premium First Strand cDNA Synthesis Kit and random hexamer primers according to the manufacturer's instructions (Fermentas). Additionally, 1 μg of total RNA was used as RT-negative control following the same protocol with additional reaction buffer instead of the enzyme mix. Resulting cDNAs as well as RT-negative controls were diluted 1:50 in DNase-and RNase-free water. 1 μl of each sample was used for qRT-PCR.
Specific oligonucleotide primer pairs were used for PCR (Table 6). New primers or new primer pairs were designed with Primer3 software (http://frodo.wi.mit.edu/) and synthesized (Metabion, Martinsried, Germany). The amounts of cDNA of all genes were determined by qRT-PCR assays in 12.5 μl reaction volume. Conditions for the reactions were: 6.25 μl of 2× SsoFast Eva Green Supermix (BioRad, Hercules, US), 0.5 μM of each primer, 1 μl of cDNA; 1 × 95°C for 3 min, 45 × 95°C for 10 s and 57°C for 15 s in a BioRad C1000 cycler with an CFX96 optical head. Validation of specific products was done via melting curve analysis, consisting of an initial heating at 95°C for 10 s, followed by a stepwise temperature increase from 68 to 88°C with an increment of 0.2°C for 5 s. Threshold cycle values were calculated via regression analysis using CFX manager V 2.0 (BioRad). Differentially expressed genes were identified and analysed with the option 'gene study' of CFX manager software. The genes pvuA, dnaE, recA and a locus of the 16S-23S intergenic spacer region (1623S) were used for normalization via ΔΔC(q)-method.