Transcriptional regulation of the operon encoding stress-responsive ECF sigma factor SigH and its anti-sigma factor RshA, and control of its regulatory network in Corynebacterium glutamicum

The sigH gene and the rshA gene encoding an anti-sigma factor of SigH form an operon

The genes encoding SigR (an ortholog of C. glutamicum SigH) in S. coelicolor and SigH in some mycobacteria (e.g. M. smegmatis and M. avium) are located in close proximity to the genes encoding their anti-sigma factors RsrA and RshA, respectively, which were found immediately downstream [12, 25]. The same arrangement of the sigH (cg0876) and rshA (cg0877) genes was described in the genomes of C. glutamicum ATCC 13032 [5] and C. jeikeium[26] (Figure 1a). Probably due to its small size of 267 nucleotides (89 amino acids), the rshA gene has not been annotated in two other C. glutamicum genome sequences, but it can also be found there by using a BLASTX search (data not shown). It is interesting to note that the absence of the rshA gene in the annotation of one of the C. glutamicum genome sequences [4] apparently misled the authors of a recent study [13], who picked the wrong ortholog from C. glutamicum in order to check for a functional complementation of rsrA in Streptomyces. It is not surprising that the above-mentioned study failed to show a functional complementation.

In all C. glutamicum genomes, the translational stop codon of sigH is only separated by two bp from the translation initiation codon of rshA, indicating an operon-like structure. The deduced RshA protein sequence from C. glutamicum is only moderately similar to that of RshA from M. tuberculosis (35%) and RsrA from S. coelicolor (28%). An amino acid sequence alignment between the three corynebacterial genes and their M. tuberculosis and S. coelicolor counterparts (Additional file 1) shows that RshA from C. glutamicum carries the conserved cysteine residues which mediate the interaction of SigH and RshA in the ZAS domain [13].

The sigH gene and the rshA gene form an operon-like structure in C. glutamicum. We therefore first analyzed their transcriptional organization by Northern hybridization. The blotting was performed with total RNA prepared from C. glutamicum RES167 (restriction-deficient variant derived from the ATCC 13032 type strain and its derived deletion mutant strains DN2 (carrying a deletion within sigH) and AS1 (carrying a complete deletion of sigHrshA). The blot was then hybridized with DIG-labelled RNA-probes derived from the sigH and the rshA genes, respectively. A single 1-kb transcript hybridized with the sigH riboprobe when total RNA isolated from the RES167 strain was used (Figure 1b). A transcript of the same length also hybridized with the rshA riboprobe. These results indicated that both genes are transcribed in a single mRNA from a promoter located upstream of the sigH gene. An additional transcript of approximately 370 bp was detected by using the rshA riboprobe. This transcript most likely only covered the rshA gene and suggested that another promoter (PrshA) is present within the sigH coding region.

To address the question of whether the promoters of the sigH and rshA genes are controlled by the sigma factor SigH, we used RNA isolated from the sigH deletion strain DN2 for Northern hybridization. We supposed that the SigH- dependent transcripts would not be found with DN2 RNA. Indeed, no signal was detected when the sigH probe was used, because the complementary region in the sigH gene was deleted in DN2. A transcript of around 550 bp was detected with the rshA-specific probe (Figure 1b). This transcript most probably initiated upstream of sigH (from the sigH promoter), since its length was that of the full-length transcript containing sigH-rshA minus the length of the deletion within sigH in DN2 (Figure 1a). These results suggested that the bicistronic sigH-rshA transcript is formed in a SigH-independent manner. In contrast, the rshA transcript was not detected with the rshA probe, although the deletion within sigH should not have removed the presumed rshA promoter. This result indicated that the rshA promoter is under the control of SigH.

Genes of the sigH-rshA operon are transcribed from multiple promoters of different types

To analyze the promoter regions of the sigH-rshA operon and of the rshA gene, DNA fragments (504 bp upstream of sigH and 301 bp upstream of rshA) were cloned in the promoter probe vector pET2, thus forming transcriptional fusions of the promoter-active fragments and the reporter gene cat coding for chloramphenicol acetyltransferase (CAT). The activity of the promoters was measured using the CAT enzyme activity in cell-free extracts of C. glutamicum (pET2sigH) and C. glutamicum (pET2rshA). The activity of PsigH during the exponential growth phase was 0.1±0.015 U (mg of protein)^-1 whereas the activity of PrshA was only 0.03 ±0.005 U (mg of protein)^-1. Negligible activity was detected with the empty vector pET2 (≤0.003 U (mg of protein)^-1). These measurements confirmed that rshA is also transcribed from the separate PrshA promoter.

To determine the transcriptional start points (TSPs) of the sigH-rshA and rshA transcripts, a primer extension analysis was performed (PEX) using the primer CM4 and total RNA isolated from C. glutamicum (pET2sigH) and C. glutamicum (pET2rshA), respectively. Three TSPs were located within the upstream region of the sigH gene. TSP1, TSP2 and TSP3 were mapped at nucleotide A in all cases, 22 nt, 89 nt and 93 nt upstream of the sigH start codon, respectively (Figure 2a). An identical result was achieved with the primer CM5 (data not shown). The putative −10 hexamers of the respective promoters, TAGAAT (P1), TAAAGT (P2) and TAGAGT (P3) are similar to each other and fit well to the consensus −10 hexamer TANANT of SigA-dependent promoters driving the expression of housekeeping genes in C. glutamicum[8]. The putative −35 sequences of P1, P2 and P3 are less similar to the consensus, which is a common feature of C. glutamicum housekeeping promoters. In conclusion, all three promoters seem to be SigA-dependent. Since yet another TSP signal could be recognized further upstream of TSP3 in some primer extension analyses using C. glutamicum (pET2sigH), the 348-bp upstream fragment (462 to 115 nt upstream of the sigH initiation codon, outside of P1, P2 and P3) was separately cloned in pET2 (resulting in pET2sigH 4). Using this transcriptional fusion, a CAT activity of 0.009±0.002 U (mg of protein)^-1 was determined. This result indicated that there is a promoter within this upstream fragment. With RNA isolated from C. glutamicum (pET2sigH 4) and primers CM4 or CM5, transcription start point at nt A, (TSP4) 131 nt upstream of the sigH initiation codon was determined by PEX (Figure 2b). The position of TSP4 was further confirmed by RACE analysis (data not shown). The hexamer TACATA located the appropriate distance from TSP4 and the hexamer TTGTTT (with a spacer of 19 nt) could function as the −10 and −35 sequences of another SigA-dependent promoter (P4), respectively (Figure 2c). A TGGTACATATGTTCTA sequence conforming to the consensus sequence of the SOS box, which was described as a LexA binding site in C. glutamicum[27], was found to overlap with the −10 region of P4.

Using total RNA from C. glutamicum (pET2rshA) and the CM4 primer, two TSPs were detected at nt G and A, 62 nt and 66 nt upstream of the rshA initiation codon (Figure 2d). TSP1 at the same G was detected by a weaker PEX result with the CM5 primer (not shown). The motifs TGGAAGA in the −35 region and TGTTAAA in the −10 region relative to TSP1 fit well to the consensus sequence of the −35 and −10 regions of the proposed SigH-dependent promoters of the M. tuberculosis (^G/TGGAA^C/TA −16 nt –^C/GGTT) [28] and SigR-dependent promoters of S. coelicolor (GGGAA T^G/C - 16 nt - ^C/GGTT G) [29] and also to the proposed C. glutamicum consensus of SigH-dependent promoters gGGAA ta - 16–19 nt - ^C/TGTT gaa [14] or ^G/TGGAA TA - 16–19 nt - ^C/TGTT GAA [8]. This result suggests that the PrshA promoter is under the control of SigH, which is in agreement with the results from the Northern hybridization experiments.

Global transcriptional profiling of the rshA deletion mutant revealed the majority of known SigH-dependent genes and novel ones

To discover genes that are under the control of SigH, we utilised the constructed C. glutamicum ΔrshA strain. We expected that SigH would be released from inhibition by the anti-sigma factor in this deletion strain and SigH-dependent genes might be expressed without applying any stress. A comparative microarray hybridization analysis was performed using total RNA isolated from C. glutamicum RES167 and its rshA deletion derivative growing under standard cultivation conditions (30°C) in shaking flasks. The signal intensity ratio (m) / signal intensity (a) plots deduced from hybridizations are shown in Figure 3 and the differentially transcribed genes are listed in Table 1. Altogether, 83 genes in 61 putative transcriptional units were found to be upregulated in the ΔrshA mutant compared to its parent strain. The highest ratios were observed for the genes previously described as members of the SigH regulon [14]. These data strongly confirmed the assumption that the SigH sigma factor would be highly active in the ΔrshA strain in which the functional rshA gene product is absent and are in line with the notion that RshA plays the role of an anti-sigma factor controlling SigH activity in vivo.

Although most of the differentially transcribed genes match those described by Ehira et al.[14], this study also found the genes mshC (cg1709; mycothiol synthesis) and mca (cg1127, mycothiol conjugate amidase) to be strongly deregulated, and qor2 (cg1553, quinone oxidoreductase) as weakly influenced in the ΔrshA mutant. All of these genes are apparently involved in redox homeostasis and were also found to be more strongly transcribed under disulphide stress conditions induced by diamide treatment (our unpublished results).

Interestingly, some heat-stress related genes previously reported to be SigH-dependent (dnaJ2, clpB, clpP1 and clpP2;[14, 16]) showed up only weakly in our analyses and some other previously identified members of the SigH regulon failed to exhibit the minimum threshold (m-value of 0.6 corresponding to 1.5-fold change) used. Genes that displayed differential expression values below this threshold were the dnaK-grpE operon, clpC, the non-essential sigma factor gene sigB and most genes of the suf cluster [14]. A differential transcription of clgR, a heat stress-responsive regulator, which was expressed from a SigH-dependent promoter accoding to Engels et al.[16], was also not detected in our experiments. This finding is similar to observations by Ehira et al.[14]). These discrepancies might be explained by additional regulatory systems negatively controlling the transcription of these genes in the absence of (heat) stress.

Genes identified for the first time as being triggered by the SigH-RshA regulatory network included uvrA (cg1560) and uvrC (cg1790), both coding for subunits of the Exinuclease ABC (nucleotide-excision repair), as well as uvrD3 (cg1555), one of three genes encoding DNA helicases similar to UvrD proteins in C. glutamicum, and a gene cluster (cg0184-cg0186) possibly involved in alkylated DNA repair. Together with the observation of a putative LexA-regulated promoter upstream of the sigH-rshA operon, this links the SigH network with DNA damage and repair.

Other newly identified genes code for components of the proteasome machinery, pup (cg1689; encoding a prokaryotic ubiquitin-like protein) and cg0998 (a trypsin-like serine protease). All these genes were found to be transcriptionally induced in the ΔrshA strain (Table 1).

Among the downregulated genes, only 7 exceeded the standard threshold m < −1 (fold change 0.5). These genes encode putative membrane proteins, hypothetical proteins and transporters (Additional file 2). Interestingly, the sigH transcript itself appeared to be less abundant in the rshA deletion mutant. Since this result was unexpected, we checked PsigH for mutations in this strain by PCR amplification and sequencing of the sigH 5’-upstream region. No mutations were found within 315 bp upstream of the sigH translational start codon (data not shown). It can be speculated that the sigH transcript is less stable in the ΔrshA mutant due to a change in its structure or due to the lack of stablisation effects by ribosomes translating rshA.

To validate the newly found potential SigH-dependent genes, we focussed our subsequent analyses on those from which new insights into the SigH regulon were expected. Therefore the genes potentially involved in response to disulphide stress, in protein degradation and in SOS response to DNA damage were included in the following q-RT-PCR experiments.

Differential transcription of selected SigH-dependent genes was validated by quantitative real-time RT-PCR

The microarray analyses found a number of novel candidate genes for the SigH regulon. To validate these results, we performed a q-RT-PCR with mshC, mca and mtr (involved in mycothiol synthesis and recycling [30, 31]), pup (encoding an ortholog of the recently identified prokaryotic ubiquitin-like protein in M. tuberculosis[32]), as well as uvrA and uvrD3 (SOS-response). Additionally, we chose the two genes with strongly enhanced expression in the ΔrshA strain, cg2838 (putative dithiol-disulfide isomerase) and cg3405 (NADPH:quinone reductase), which might be involved in defense against disulphide stress. The recently described small antisense RNA arnA that has been shown to be transcribed from a SigH-dependent, heat-shock-induced promoter [33] was also included in the q-RT-PCR analysis. The arnA transcript was not addressed in the microarray analysis, since only probes for protein-coding genes were used in the design of the microarray [34].

The strong transcriptional induction observed in microarray analysis was validated for both cg2838 and cg3405 with 60-fold and 20-fold higher transcript levels, respectively (Figure 4). The genes mshC, mca, mtr, uvrD3, and arnA, were induced 3- to 4-fold and the weakest induction was observed for pup and uvrA with a 2-fold higher transcript level in the ΔrshA mutant than in the WT-strain. The reduction of the transcript level of sigH to around 50% of the WT level was also confirmed.

Experimental localization of SigH-dependent promoters and derivation of a consensus sequence

Several genes which exhibited higher transcript levels in the ΔrshA strain than in its parental WT strain in microarray analyses and/or in q-RT-PCR were chosen for promoter localization by TSP determination using primer extension analysis. Regions (300 to 400 bp) upstream of the initiation codons of the analyzed genes were used to construct transcriptional fusions with the cat gene in the vector pET2. TSPs within the mshC, mca, dnaJ2, uvrA and uvrD3 upstream fragments (carrying potential SigH-dependent promoters) were localized 141 nt, 207 nt, 100 nt, 46 nt and 56 nt upstream of the initiation codons, respectively. Examples of the results of primer extension analysis for dnaJ2 and uvrA are shown in Figure 5. The respective −10 and −35 regions which were compatible with the consensus sequence of the SigH-dependent promoters [8, 14] were found at the proper distances in all cases (Figure 6). In addition, transcriptional starts within mca and pup fragments and the respective sequence motifs resembling SigA-dependent promoters were localized upstream of these genes by primer extension (data not shown).

Further SigH-dependent promoters were searched for within the 5´-UTRs of the genes, which exhibited enhanced transcription in the ΔrshA strain in the microarray analyses, by motif searches using the program Bioprospector [35]. In addition to all previously defined promoters belonging to the genes of the SigH regulon [14], the promoter of arnA[33] and the promoters determined in this work (rshA, mshC, mca, dnaJ2, uvrA and uvrD3) were included in the training set. We searched for two 10-bp motifs with a gap of 15 to 23 bp. Using the Bioprospector program, 10 additional transcriptional units containing a conserved SigH-dependent promoter motif in their 5´-UTR were identified (Figure 6). The other 26 analyzed transcriptional units did not show up in these analyses. Their transcription initiation is possibly not directly SigH-dependent but rather upregulated by secondary effects under the conditions used. Six SigH-dependent promoters upstream of the identified genes were precisely localized by determination of the respective transcriptional start points. A refined consensus sequence based on the sequences of 45 SigH-dependent promoters was defined (Figure 7).

The sigH-rshA operon in C. glutamicum exhibits complex transcriptional organization including autoregulation

In this study we demonstrated the upregulation of the majority of the known SigH-dependent genes in the absence of an applied stress by removing its putative anti-sigma factor RshA. The gene encoding RshA was only annotated in the genome of C. glutamicum ATCC 13032, reported by Kalinowski et al.[5]. The rshA gene in two other sequenced C. glutamicum strains, in C. glutamicum ATCC 13032, reported by Ikeda and Nakagawa [4], and C. glutamicum strain R, reported by Yukawa et al.[6], is not annotated, probably because of its small size of 89 amino acids. However, the deduced RshA protein sequences are identical in the three genome sequences and similar to other anti-sigma factors from M. tuberculosis (RshA; [12]) or S. coelicolor (RsrA; [11]). RshA from C. glutamicum shares the conserved cysteine residues in the ZAS domain with its counterparts. These residues modulate the interaction with the SigH protein, a fact that has been experimentally determined for RshA and SigH in M. tuberculosis[12], RsrA and SigR in S. coelicolor[11], as well as for other members of the ZAS-domain containing protein family in actinobacteria [13]. The clear upregulation of all previously determined SigH-dependent genes in the constructed rshA mutant provides further proof that in C. glutamicum, RshA functions as an anti-sigma factor similar to M. tuberculosis RshA and S. coelicolor RsrA.

The sigH-rshA gene organization is also conserved in all sequenced Corynebacterium strains available in NCBI database e.g. C. glutamicum, C. efficiens, C. jeikum[8], and in the more distantly related S. coelicolor. M. tuberculosis exhibits a similar organization, but a gene encoding a protein of unknown function is inserted between the sigH and rshA genes.

The transcriptional organization of the sigH-rshA operon in C. glutamicum is similar but not identical to that of M. tuberculosis and S. coelicolor. In C. glutamicum, four promoters upstream of sigH-rshA resemble house-keeping promoters which are recognized by SigA. The reason for having multiple promoters might ensure fine-tuning, either by the action of additional transcription factors or by the differing affinities of these promoters to SigA and SigB, the non-essential sigma factor of C. glutamicum that also targets house-keeping promoters [9]. Experimental observations are in line with this assumption: it was shown by a reporter fusion analysis (P-sigH::cat) that the activity of the sigH promoter rose in the stationary phase and after oxidative stress [15], whereas no significant changes in sigH transcript levels were detected after heat shock [18] or in the transition phase of growth [24]. In S. coelicolor, the sigR-rsrA operon is also transcribed from multiple promoters. There is one transcriptional start of sigR dependent on the housekeeping sigma factor SigA and another one dependent on SigR^Sc itself [25]. In M. tuberculosis, sigH is apparently only autoregulated by SigH [12].

A possible additional regulation of SigH in C. glutamicum might operate via the SigA-dependent promoter that was found in the 5´-UTR of the sigH gene, overlapping with a putative SOS-box [27] and therefore most likely blocked by the LexA repressor in the absence of a DNA-damaging agent.

The main difference from the related bacteria S. coelicolor and M. tuberculosis was the finding that in C. glutamicum, the rshA gene is transcribed by an additional promoter as a monocistronic transcript. We showed by Northern blotting and by PEX analysis that this transcription is SigH-dependent. It can be speculated that this transcriptional organization evolved to guarantee an excess of RshA protein over SigH at all times and therefore a fast shut-down of SigH-dependent transcriptional activation as soon as stress conditions end.

Expression analysis of the rshA mutant strain validated and extended the known SigH regulatory network

SigH is one of the major regulators, especially during heat stress, which also involves a number of different transcriptional regulators [8]. In contrast to studying the action of SigH in the presence of stress, we choose to uncouple SigH from RshA in order to assess its regulon without a possible stress-induced background. Using microarray analyses, we observed an induction of all SigH-dependent genes described by Ehira and coworkers in the rshA deletion mutant, with the exception of the dnaK-grpE operon, clpC, sigB and most genes of the suf cluster. Like Ehira and coworkers, working with overexpressing and deleting the sigH gene, we were unable to show a differential transcription of clgR. The rather weak transcriptional induction of some of the SigH-dependent heat-shock genes and the apparent absence of induction of the above-mentioned genes is explained by dominant effects exerted by known transcriptional regulators such as ClgR, HrcA, HspR, and/or SufR [14, 16, 18] (Figure 8). The additional action of these regulators might increase SigH activity under heat and/or oxidative stress. This might also hold for the sigB gene encoding the non-essential sigma factor of C. glutamicum. SigB is involved in gene expression in the transition phase of growth, and in our experiments sampling took place in the exponential phase of growth. Again, additional factors might be necessary to trigger the transcriptional activation of sigB by SigH.

Genes hitherto not described as being part of the SigH regulon included genes involved in mycothiol (MSH) synthesis and recycling. Besides thioredoxin (Trx), MSH is the major low-molecular mass thiol in corynebacteria, mycobacteria and streptomycetes [36]. The biosynthesis of MSH in C. glutamicum and two essential genes, mshC and mshD involved in the biosynthetic pathway have been described [37]. In our approach, we observed a SigH-dependent upregulation of mshC, coding for the second gene in mycothiol (MSH) synthesis, and mca as well as mtr, involved in mycothiol recycling (Figure 8).

Mca is the first gene in MSH recycling and was already shown to be transcribed in a SigH-dependent manner [14]. It encodes mycothiol S-conjugate amidase (Mca), which cleaves adducts (MSR) from the reaction of MSH with electrophiles to produce a mercapturic acid (AcCySR) and 1-O-(2-amino-2-deoxy-a-D-glucopyranosyl)-D-myo-inositol (GlcN-Ins) [30, 31]. GlcN-Ins is the substrate of MshC, and MSH is synthesized from the subsequent enzymatic reaction with MshD [37]. As was mentioned above, mshD was not observed to be transcribed in a SigH-dependent manner, but its transcription was induced by disulphide treatment in C. glutamicum (our unpublished results), indicating that mshD is transcriptionally regulated. In M. tuberculosis, all the genes of MSH synthesis seem to be transcribed constitutively [30]. There is a similar mechanism in S. coelicolor, with the difference that besides mca, mshA is transcriptionally induced as a direct target of SigR and the genes mshB, mshC and mshD are SigR-dependent, but apparently induced indirectly [38].

The SOS regulon of many bacteria, including E. coli, is involved in various cellular processes, e.g. nucleotide excision and recombination repair [39]. By deleting the gene encoding the SOS response regulator LexA in C. glutamicum, Jochmann and coworkers [27] defined the SOS response in C. glutamicum, with only one of the uvr genes, namely uvrC, showing up in the microarray as differentially transcribed.

In our approach we observed a SigH-dependent induction of three uvr genes (uvrA, uvrC, uvrD3). The induction of uvrC transcription was quite low in our experiments, most likely because of an additional repression by LexA. As mentioned in [27], the degree of induction of SOS gene expression depends on at least four parameters: (i) the affinity of LexA for the SOS box, (ii) the location of the SOS box relative to the promoter, (iii) the promoter strength, and (iv) the presence of any additional constitutive promoters [39–41]. In this context, it is apparent that SigH is involved in the SOS response in C. glutamicum, integrating it with the heat stress and thiol-oxidative stress defense systems into a general stress response network.

This is in accordance with a proposal made by Barreiro et al.[18]. The regulation of sigH in cases of severe stress (probably causing DNA damage) would release LexA from the SOS boxes and thereby activate an additional SigA-dependent sigH promoter.

The SigH regulatory network appears to also control other functions. An interesting novel finding was the enhanced transcription of components of the proteasome. The actinobacterial proteasome consists of functions for pupylation (a process similar to eukaryotic ubiquitinylation, which marks proteins that are to be degraded) and proteases. Our study connects the recently identified pupylation component Pup (prokaryotic ubiquitin-like protein) and PafA2 (proteasome acessory factor, responsible for Pup conjugation; [14]) with the SigH regulon and underlines that SigH also plays a significant role in protein quality control.

Based on the results obtained in this study and in previous studies, we propose an extended model of the SigH regulon in C. glutamicum (Figure 8) including the direct control of the stress reponse to disulphide and heat stress by RshA, involving the thioredoxin system and the mycothiol-recycling system to cope with thiol-depleting conditions. In an unstressed state, SigH is inhibited by the reduced form of RshA. The disruption of the SigH–RshA complex in C. glutamicum appears under severe heat shock or disulphide stress via a change in the conformation through the oxidation of RshA. The released SigH forms a functional RNAP holoenzyme with the core enzyme and induces the stress response by transcribing SigH-dependent genes, including those involved in disulphide and heat stress response. The feed-forward induction of the anti-sigma factor RshA enables the cell to quickly shut down the stress response, based on SigH-dependent transcription, after the stress ends. RshA, as the stress-sensing redox switch, is one of the targets of the biochemical pathways encoded by genes of the SigH network, namely those of the reducing compounds thioredoxin (Trx) and mycothiol (MSH). Direct induction of trxB1C generates the thiol Trx and the gene products of trxB, mtr, mca, and mshC reduce and/or recycle Trx and MSH, respectively, which are able to restore, together with other reductases and reducing compounds, the thiol redox balance and reverse the oxidation of cysteine residues in RsrA. In this closed loop, RshA is reduced to regain its functionality and binds SigH after redox homeostasis is reached. A similar model was developed for the thiol-depleting stress response in S. coelicolor by Newton and coworkers in 2008 [30, 31]. The transcriptional regulatory network controlled by SigH is highly connected to other regulators, modulating gene expression in response to other physical or chemical triggers. The heat-shock regulatory network that includes the regulators HspR and ClgR is an example of such a level of control.

Bacterial strains, plasmids, oligonucleotide primers, media and growth conditions

DNA isolation, manipulation and transfer

Isolation of plasmid DNA from E. coli cells by an alkaline lysis technique was performed using a QIAprep Spin Miniprep Kit (Qiagen, Hilden, Germany). Chromosomal C. glutamicum DNA was isolated as described previously [44]. DNA amplification by PCR was carried out with KOD DNA polymerase (Merck, Darmstadt, Germany) or Phusion DNA polymerase (Finnzymes,Vantaa, Finland) and chromosomal C. glutamicum RES167 DNA as the template. PCR products were purified with a QIAquick PCR Purification Kit (Qiagen). All oligonucleotides used in this study (Additional file 1) were purchased from Metabion (Martinsried, Germany). All PCR setups were done according to the manufacturers´ protocols. Modification of DNA, analysis by agarose gel electrophoresis and ligation were performed using standard procedures [42]. E. coli was transformed with plasmid DNA using the method of Hanahan [45], C. glutamicum cells were transformed by electroporation [46, 47].

Construction of defined deletions in the C. glutamicum chromosome

The defined chromosomal deletions (ΔrshA, ΔsigH and ΔsigHrshA) were constructed in C. glutamicum RES167 using the gene SOEing procedure [48], the E. coli vector pK18mobsacB[49] and the conditional lethal effect of the sacB gene for selecting double recombinants after the transformation of C. glutamicum[49]. The selection of the resulting marker-less C. glutamicum strains ΔrshA, DN2 and AS1 and PCR confimation of the respective rshA (220 bp) sigH (450 bp) and sigHrshA (1340 bp) deletions within their chromosomes (Figure 1A) were carried out as described previously [50] using the primers listed in the Additional file 3.

Construction of plasmids

Fragments carrying the promoter regions of the genes sigH, sigH(P4), rshA, mshC, mca, dnaJ2 uvrA and uvrD were amplified from the chromosomal DNA of C. glutamicum by PCR with the primer pairs PSIGHF+PSIGHR, PSIGHF+PSIGH4R, PRSHAF+PRSHAR, PMSHCF+PMSHCR, PMCAF+PMCAR, PDNAJ2F+PDNAJ2R, PUVRAF+PUVRAR and PUVRDF+PUVRDR, respectively (Additional file 3). The primers carried the PstI, BamHI or BglII restriction sites. The PCR products were digested by the respective enzymes and cloned in the plasmid pET2 digested by PstI and BamHI. The resulting plasmid constructs were introduced into C. glutamicum by electroporation.

RNA isolation and quantitative real-time RT-PCR

RNA was isolated from exponentially growing cultures of both C. glutamicum RES167 and the ΔrshA strain grown in triplicate. The cells were harvested by centrifugation and the cell pellets were immediately frozen in liquid nitrogen. The cells were then resuspended in the RLT buffer provided with the RNeasy Mini Kit (Qiagen, Hilden, Germany) and disrupted with a Precellys 24 homogeniser (Bertin Technologies, France) at a speed of 6.5 for 30 s once.

Total RNA was purified with an RNeasy Mini Kit along with an RNase-Free DNase Set (Qiagen) and a DNase I Kit (Sigma-Aldrich, Taufkirchen, Germany) according to a previously published protocol [34]. RNA was quantified with a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE). Purified total RNA from C. glutamicum cultures was used in real time RT-PCR analysis performed with a LightCycler instrument (Roche Diagnostics, Mannheim, Germany) and a 2× SensiMix One Step Kit (Bioline, Luckenwalde, Germany). The verification of the resulting RT-PCR products was performed by melting curve analysis. The differences in gene expression were determined by comparing the crossing points of three samples measured in duplicate. The crossing points were determined using the LightCycler software (Roche Diagnostics). The calculation of the average crossing point (CP) was performed by first calculating the averages for each set of technical replicates and then by calculating the average of the three biological replicates. For each set of three biological replicates, the standard deviation was calculated (assuming a normal distribution of the CPs) and the combined standard deviation for the DeltaCP was approximated using the standard calculation for the propagation of uncertainity (assuming non-correlated errors).

Microarray hybridization

The hybridization of whole-genome oligonucleotide microarrays was performed as described previously [51], using 8 μg of total RNA from C. glutamicum cultures for cDNA synthesis. The normalization and evaluation of the hybridization data was done with the software package EMMA 2 [52] using a signal intensity (A-value) cut-off of ≥7.0 and a signal intensity ratio (M-value) cut-off of ±0.6, which corresponds to relative expression changes equal to or greater than 1.5-fold.

Northern blot analysis

The DIG-labeled RNA probes for the sigH and rshA genes for transcript analysis were obtained by in vitro-transcription with T7 RNA polymerase, NTP-DIG-label mix (Roche Diagnostics) and gene-specific primers with a T7 promoter-sequence attached to the reverse primer (Additional file 3). Prior to hybridization, the probes were denatured by incubation at 95°C for 10 min.

Northern blot analysis was performed as described by Homuth et al.[53] with the following modifications. Total RNA samples (5 μg), purified by using the RNeasy Mini Kit along with the RNase-Free DNase Set (Qiagen) and the DNase I Kit (Sigma-Aldrich) according to a previously published protocol [34], were separated under denaturing conditions in 1% agarose-formaldehyde gels in 1xMOPS (morpholinepropanesulfonic acid) running buffer and stained with ethidium bromide. Separated RNA was transferred to a Hybond-N membrane (GE Healthcare, Freiburg, Germany) by vacuum blotting. Hybridization and detection were carried out as follows. After being baked at 120°C for 0.5 h, the membrane was prehybridized under stringent conditions at 68°C for 1 h in 50% formamide and 5x SSC (1x SSC is 0.15 M NaCl plus 0.015 M sodium citrate) without the probe to block reactive membrane binding sites, and in the second step hybridized with digoxigenin (DIG)-labelled RNA probes (50 ng/ml) at 68°C overnight. The hybridized membrane was washed to remove the hybridization solution, first twice for 10 min in 2× SSC-0.1% (wt/vol) sodium dodecyl sulfate (SDS) at room temperature and then three times for 15 min in 0.1× SSC-0.1% (wt/vol) SDS at 68°C, and hybridization signals were detected according to the manufacturer’s instructions (Roche Anti-Digoxigenin-AP, Fab fragments 2 μl and CDP-Star) with a Luminescent Image Analyzer LAS-3000 (Fujifilm Europe, Düsseldorf, Germany). The sizes of the detected signals were determined by comparing with the prior ethidium-bromide-stained High Range Marker (Fermentas, St. Leon-Roth, Germany), marked on the membrane.

Primer extension analysis

C. glutamicum cells were cultivated in 2xTY medium at 30°C, harvested at OD₆₀₀ = 3.5, and frozen at −70°C. The pellet was resuspended in distilled water and approximately 0.2 × 10⁸ cells were disintegrated with a FastPrep FP120 (BIO101) (6x20 s, speed 6.0) using glass beads. The suspension was cooled for 5 min on ice between runs. The cell debris was removed by centrifugation and total RNA was isolated from the extract using a High Pure RNA Isolation Kit (Roche Diagnostics). The primer extension analysis was essentially done as described previously [54]. Reverse transcription was performed with SuperScript III transcriptase (Invitrogen, Carlsbad, CA) using 30 μg RNA and 5 pmol Cy-5-labeled primer CM4 or CM5 (Additional file 3) complementary to the vector pET2. Specific Cy5-labeled primers XMSHC, XMCA and XUVRD (Additional file 3) were used to determine the transcriptional start points of the genes mshC, mca and uvrD, respectively. PAA gel electrophoresis was run with the synthesized cDNA simultaneously with the DNA sequencing products generated with the same labeled primer in an A.L.F. Sequencer (GE Healthcare, Munich, Germany).

Chloramphenicol acetyltransferase (CAT) assay

The CAT activity was essentially measured as described previously [55]. C. glutamicum strains harboring the vector pET2 with promoter-carrying fragments were cultivated in complete 2xTY medium to OD₆₀₀ = 3 to 3.5. The cells were rapidly chilled on ice and disrupted with a FastPrep FP120 homogenizer (BIO101) (Thermo Scientific). The specific CAT activity in the cell-free extracts was determined photometrically at 412 nm as described by Shaw [56]. One unit (U) of enzyme activity was defined as 1 μmol of chloramphenicol acetylated per minute.