| Edited by Hisaji Maki. Naotake Ogasawara: Corresponding author. E-mail: nogasawa@bs.naist.jp. Giyanto: Present address: Faculty of Agriculture, Bogor Agricultural University, Jalan Kamper Kampus, IPB Darmaga, Bogor, Indonesia. Note: Supplementary materials in this article are at http://wwwsoc.nii.ac.jp/gsj3/sup/84(4)Rukmana/ |
Many microorganisms living in soil produce antibiotics that inhibit the growth of competitors for limited nutrients in the environment. At the same time, microorganisms have developed defense systems to cope with external antibiotic challenge. The cell envelope is the target of many antibiotics, and, in Gram-positive bacteria, membrane alterations and dysfunction caused by antibiotics are sensed mainly by two classes of signal transduction systems: the ECF (extracytoplasmic function) sigma factors and the two-component signal transduction systems (TCSs). These systems induce appropriate counteractions to repair membrane damage and to ensure membrane functionality (Jordan et al., 2008). With both systems, membrane-embedded sensors (anti-sigma factors and sensor histidine kinases) activate cognate transcriptional regulators (ECF sigma factors and response regulators) upon signal perception. Recent transcriptomic approaches using gene arrays to comprehensively identify genes upregulated in Bacillus subtilis challenged with sub-inhibitory concentrations of antibiotics that interfere with cell envelope integrity, including vancomycin (Cao et al., 2002; Eiamphungporn and Helmann, 2008), bacitracin (Mascher et al., 2003), cationic antimicrobial peptides (CAMPs) (Pietiainen et al., 2005), daptomycin (Hachmann et al., 2009; Wecke et al., 2009), and friulimicin B (Wecke et al., 2009), have revealed that combinations of three ECF sigma factors (SigM, SigW, and SigX) and five TCSs (LiaRS, BceRS, YvcPQ, YxdJK, and YhcZY) are activated in response to cell envelope stress caused by antibiotic action.
The B. subtilis genome encodes seven ECF sigma factors, and the physiological roles of four of these (SigM, SigW, SigX, and SigY) have been investigated in some detail. It has been shown that the proteins act on partially overlapping regulons related to cell envelope homeostasis and antibiotic resistance (reviewed in Mascher et al., 2007). The roles of the other three factors (SigV, SigZ, and YlaC) remain elusive. The direct involvement of an ECF sigma factor in mediating resistance to antibiotics has been demonstrated for SigW, which regulates resistance genes for fosfomycin, sublancin, and the antimicrobial peptide SdpC (Butcher and Helmann, 2006). SigM induces a bacitracin-resistance determinant, bcrC (Cao and Helmann, 2002; Bernard et al., 2005). SigX regulates the dltA and pssA operons that affect the overall net charge of the cell envelope by incorporating positively charged groups into teichoic acids and the cytoplasmic membrane, respectively. Consistent with these observations, a sigX mutant strain displayed increased sensitivity to CAMPs (Cao and Helmann, 2004). However, the ECF sigma factors of B. subtilis recognize structurally similar promoter sequences, and regulatory overlap has been demonstrated. This often makes it difficult to evaluate the contribution of an individual system to genetic resistance to any particular antibiotic, because the effects of inactivation of one sigma factor may be compensated for by other factors. Indeed, triple inactivation of the sigM, sigW, and sigX genes in an undomesticated B. subtilis strain resulted in a greatly increased sensitivity to several cell wall-active antibiotics. In a number of instances, the antibiotic-sensitive phenotypes were significantly enhanced in the triple mutant strain relative to strains lacking only one or two factors (Mascher et al., 2007).
On the other hand, the physiological roles of TCSs in mediating resistance to antibiotics (Fig. 1) are relatively unclear. BceRS induces the expression of downstream bceAB genes encoding an ABC transporter that facilitates bacitracin removal (Ohki et al., 2003). YvcPQ and YxdJK are structurally homologous to BceRS and induce the expression of downstream yvcRS and yxdLM, respectively, which are homologous to bceAB (Joseph et al., 2004; Giyanto, unpublished results). Expression of yxdLM and yvcRS is induced by CAMP LL-37 (Pietiainen et al., 2005) and nisin (Hansen et al., 2007), respectively. However, no phenotypic change has been detected in yxdLM or yvcRS mutants. The LiaRS system activates expression of the liaIHGFSR operon (Mascher et al., 2004). Notably, the liaH gene product is homologous to the Escherichia coli phage-shock protein PspA, which is induced under membrane stress conditions and is proposed to bind to membrane phospholipids to suppress proton leakage (Kobayashi et al., 2007). The expression of the lia operon is strongly induced by antibiotics that interfere with the lipid II cycle of peptidoglycan synthesis; these antibiotics include bacitracin, nisin, ramoplanin, and vancomycin (Mascher et al., 2004). However, liaRS inactivation did not affect the sensitivity of B. subtilis cells to these antibiotics. CAMP LL-37 (Pietiainen et al., 2005) and daptomycin (Hachmann et al., 2009; Wecke et al., 2009) are supposed to disrupt cell membrane structure and also induce lia operon expression, and the deletion of the liaH gene was very recently found to lead to a three-fold increase in susceptibility to daptomycin (Hachmann et al., 2009). In addition, LiaRS is proposed to induce the expression of the yhcZY-yhdA operon, the biological role of which is unknown (Jordan et al., 2006).
 View Details | Fig. 1 Operon structure of TCS genes specifically induced by cell envelope stress caused by antibiotics. Gene arrangements of operons containing TCS genes known to be specifically induced by cell envelope stress caused by antibiotics are indicated by bold arrows. Grey and black arrows show sensor kinase and response regulator genes, respectively. Transcriptional units and target sites of response regulators are indicated by arrows below and above gene maps, respectively. Dashed lines indicate putative regulatory circuits. The circle on a stem represents a transcriptional termination signal. |
It should be noted that in addition to the ECF sigma factor and TCS regulons discussed above, many genes, the induction mechanisms and biological roles of which are difficult to evaluate, were “upregulated” in transcriptomic analyses of responses to antibiotics. Evaluation of the biological roles of such genes requires further work, although some supposed elevations may simply represent background noise caused by the accuracy limitations of gene array experiments and/or difficulties in regulating the growth conditions of target and control cells.
During the course of preliminary experiments to survey antibiotics that activate YvcPQ and YxdJK TCSs, we found that the YvcPQ system was activated by enduracidin, the target of which is considered to be the transglycosylation step of peptidoglycan biosynthesis (Fang et al., 2006). This finding prompted us to assess B. subtilis responses to enduracidin using a high-density tiling chip previously described by our group (Ishikawa et al., 2007; Morimoto et al., 2008). Responses to bacitracin, which are well characterized in B. subtilis, were also explored in order to evaluate our analysis method. We used the quantitative advantage of the tiling chip to introduce a new criterion, an increase in transcriptional level, in addition to the conventional induction ratio, to distinguish genes of biological significance from those with lower induction ratios. Our results indicate that introduction of this criterion led to unambiguous identification of core transcription responses to antibiotics, with a reduction in the number of possible background genes, compared to previous results obtained using gene arrays. We identified 129 genes that were significantly upregulated by enduracidin and/or bacitracin. Notably, we found that inactivation of LiaRS TCS, which was very strongly induced by the two antibiotics, resulted in increased sensitivity to enduracidin, probably through a failure to induce LiaIH proteins. We noted that 33 genes belonging to the SigM regulon were induced by both antibiotics. Consistent with stronger induction of the SigM regulon in enduracidin-treated cells, inactivation of sigM resulted in increased sensitivity to enduracidin. In addition, and for the first time, we found that the Spx regulon was induced in cells challenged by enduracidin and bacitracin, suggesting that thiol-oxidative stress would be induced in cells challenged by antibiotics. These findings contribute to further our understanding of the molecular actions of genetic systems involved in antibiotic resistance.
The B. subtilis strains used in this study are listed in Table 1, and the primers used appear in Supplementary Table 1. To avoid any polar effect of selection marker fragments on transcription of downstream genes, clean deletion mutants of each gene in the lia operon (BAB1-6), were obtained by the method of Zhang et al. (2006), using the mazF gene as a counterselection marker, with a technical modification described in the accompanying paper (Supplementary Fig. 1, Morimoto et al., 2009). The pMUTinHis plasmid was employed to fuse a 12x-histidine tag to the 3’-terminus of liaR, as described previously (Ishikawa et al., 2006), to obtain BAB7. Disruption mutants of ECF sigma factor genes (Asai et al., 2008) were kindly provided by Dr. Kei Asai, Saitama University. Disruption mutants of bcrC and spx, constructed by an international consortium conducting B. subtilis genome functional analysis (Kobayashi et al., 2003), were obtained from the National BioResource Project on Bacillus subtilis (National Institute of Genetics, Japan).
 View Details | Table 1 B. subtilis strains used in this study |
The custom Affymetrix tiling chip used in this study contains 55,430 25-mer probes for the coding strands of protein-encoding regions, at 25–30 bp intervals, and 72,218 probes for both strands of intergenic regions, at 2–3 bp intervals (Ishikawa et al., 2007). Wild-type B. subtilis cells were cultivated in LB medium at 37°C to an OD600 value of 0.2, and divided into three portions. Cells were further cultivated for 10 or 30 min with or without 0.025 μg/mL enduracidin (Sigma) or 50 μg/mL bacitracin (Sigma). Next, total RNA was extracted as described previously (Ishikawa et al., 2007). Synthesis of cDNA, terminal labeling, and tiling chip hybridization were performed essentially by following the Affymetrix instruction manual, as described earlier (Morimoto et al., 2008). Processing of hybridization signal data was performed using the Affy-package of the Bioconductor software suite (Gentleman et al., 2004). The background of raw hybridization signals was corrected using the RMA convolution model, and distributions of signal intensities from each experiment were normalized by the quantile method to compare signals under different test conditions (Irizarry et al., 2003). Next, based on the signal intensities of probes for each gene, individual gene transcription levels were calculated by the median polish method (Irizarry et al., 2003). The distribution of corrected transcription signals along the genome coordinate was visualized with the In Silico Molecular Cloning program, Array Edition (In Silico Biology).
Overnight cultures of wild-type and mutant cells, in LB medium at 37°C, were diluted to an OD600 value of 1.0, and further serially diluted 10-fold in LB medium. Next, 2 μL aliquots of cell suspensions were spotted onto LB agar plates with and without antibiotics, and incubated overnight at 37°C.
ChAP-chip analysis (chromatin affinity purification coupled with gene-chip analysis) was performed as previously described (Ishikawa et al., 2007), using BAB7 cells harboring the liaR gene fused with a 12x-histidine coding sequence at the 3’-terminus. Signal intensities of LiaR binding were plotted along the genome coordinate using the In Silico Molecular Cloning program. For quantitative PCR analysis, DNA fragments that co-purified with LiaR and those in the supernatant fraction used for affinity purification were extracted from the same amounts of cells, as in ChAP-chip analysis. Purified DNA fragments were dissolved in 20 μL H2O, followed by two-fold serial dilution, and used as templates for quantitative PCR. Promoter sequences of the lia and yhcZY-yhdA operons were amplified using KOD-plus-DNA polymerase (Toyobo) in the presence of 1.5 mM MgSO4, employing yhcY F-yhcY R and lia F-lia R primer pairs, respectively. PCR was performed for 30 cycles using 1 μL aliquots of diluted templates in 10 μL volumes of reaction mixture, and 6 μL aliquots of products were separated in 2% (w/v) agarose gels, followed by ethidium bromide staining.
Enduracidin is a lipodepsipeptide produced by the soil bacterium Streptomyces fungicidicus (Higashide et al., 1968). The antibiotic exhibits strong bactericidal activity against a broad spectrum of Gram-positive bacteria, including strains resistant to known antibiotics, and is thus an attractive target for further antibiotic development studies (Yin and Zabriskie, 2006). The primary mechanism of action is considered to be inhibition of the transglycosylation step of peptidoglycan biosynthesis, mediated by antibiotic-binding to the lipid II substrate (Fang et al., 2006). Bacitracin is a cyclic dodecylpeptide synthesized by B. licheniformis and some strains of B. subtilis. The antibiotic binds to undecaprenyl pyrophosphate (UPP) and prevents the dephosphorylation of UPP necessary for recycling of the lipid carrier (reviewed in Rietkotter et al., 2008).
We analyzed changes in transcriptome profiles after addition of 0.025 μg/mL enduracidin or 50 μg/mL bacitracin, which are sub-inhibitory for B. subtilis growth in LB liquid medium, using a custom Affymetrix tiling chip developed earlier by our group (Ishikawa et al., 2007). Western blot analysis using BAB7 cells expressing LiaR fused with a 12x-histidine tag and antibody against the His-tag indicated that expression of the lia operon was maximally induced after 30 min of enduracidin addition (data not shown), whereas northern blot analysis indicated that the expression of the yvcRS genes peaked at 10 min (Giyanto, unpublished results). Therefore, we analyzed transcriptome profiles at 10 and 30 min after addition of antibiotics, in comparison with the profiles of cells incubated without any drug. Responses to bacitracin, which are well characterized in B. subtilis, were also determined to evaluate our new analysis method using the tiling chip. After background correction of raw transcriptional signal intensities from each experiment, and normalization of the intensity distribution of each experiment to permit comparison of signals obtained under different test conditions, we calculated transcription levels (as units) of each gene under all experimental conditions explored, yielding a maximum expression value of about 45,000 units. Graphical representations of transcriptome profiles and transcription levels of each gene thus obtained are available in Supplementary Fig. 2 and Supplementary Table 2, respectively.
 View Details | Fig. 2 Transcriptional profiles of liaRS, bceRS, yvcPQ, yxdJK, and yhcZY operons. Transcriptional profiles of liaRS (A), bceRS (B), yvcPQ (C), yxdJK (D), and yhcZY (E) operons are shown. Transcriptional signals for each probe are indicated by vertical bars at appropriate genomic coordinates. At the bottom, gene arrangements are schematically shown. |
Next, using the quantitative advantage offered by tiling chip data, we searched for genes that were significantly upregulated in cells challenged by antibiotics, using two criteria. First, the transcription level was required to be increased by more than five-fold compared to that of control cells without drug addition (the induction ratio test). This is a common but stringent criterion in transcriptome analysis. Second, the difference in transcription levels between antibiotic-challenged cells and control cells (the induction level test) had to be significant, to permit identification of genes of biological significance among genes with lower induction ratios, assuming that a high level of induction would correlate with biological activity. We set the minimum value of the second criterion as 10,000 units, to extract a number of genes similar to that identified by the first criterion. These procedures identified 21–53 genes under each experimental condition; these genes are listed in Table 2.
 View Details | Table 2 Genes significantly upregulated by enduracidin and/or bacitracina,b) |
Changes in the transcription profiles of five operons containing TCS genes are known to occur in response to cell envelope stress caused by antibiotics (see Fig. 1), and are shown in Fig. 2. The lia operon showed the highest degree of induction by both antibiotics, using both the ratio and level criteria defined above, of the genes listed in Table 1. Transcription of bceAB encoding a probable bacitracin exporter was also induced by bacitracin, to a similar level.
Bacitracin has been reported to induce yvcRS expression (Mascher et al., 2003). We observed relatively weak induction of yvcRS transcription by bacitracin at 10 min after drug addition, and the induction fell at 30 min. Similarly, enduracidin induced bceAB, yvcRS, and yxdLM transcription only at 10 min, although enduracidin also induced transcription of the bceS sensor and bceR regulator in our experiments, by an unknown mechanism. Involvement of the BceAB transporter in bacitracin resistance has been previously demonstrated (Ohki et al., 2003), and, in the present work, we discovered an important role for the lia operon in mediation of enduracidin resistance, as described below. However, any biological significance of transient transcriptional induction of other TCSs remains unclear because inactivation of these TCSs has resulted in no obvious phenotypic change, at least in the reports that have appeared to date. Sensor kinases in the TCSs are characterized by unique N-terminal input domains, consisting of only two transmembrane helices with a very short periplasmic linker sequence (Mascher et al., 2003), and BceRS has been proposed to respond to bacitracin transport by BceAB rather than to the extracellular presence of the drug per se (Bernard et al., 2007). Thus, both systems are classified as intramembrane-sensing sensor kinases (Mascher, 2006). Transient activation of YvcPQ and YedJK TCSs by both antibiotics, and the BceRS TCS by enduracidin, might result from transient distortions of membrane structure by the action of antibiotics.
Expression of yhcZY-yhdA was also initially induced by the two antibiotics, and the expression has been suggested to be under the control of LiaRS (Jordan et al., 2006). To examine this possibility, we determined the LiaR binding site(s) on the genome using a modified ChIP-chip method (ChAP-chip) (Ishikawa et al., 2007), employing a strain harboring the liaR gene fused with a 12x-histidine coding sequence at the 3’-terminus. The results shown in Fig. 3A (see also Supplementary Fig. 3) demonstrate LiaR binding only in the promoter region of the lia operon, at 30 min after addition of enduracidin, at which time lia operon transcription was maximally induced. Additionally, we sought enrichment of the promoter sequence of the yhcZY operon in the ChAP fraction by quantitative PCR, at both 10 and 30 min after addition of the drug, with negative results (Fig. 3B). Although LiaR-dependent expression of the yhcZY operon has been demonstrated in LiaF-inactivated cells, in which LiaRS is constitutively activated (Jordan et al., 2006), our results suggest that yhcZY-yhdA expression would also be transiently induced by initial distortion of the cell membrane, independent of LiaRS status in wild-type cells.
 View Details | Fig. 3 LiaR binding profile on the B. subtilis genome. (A) LiaR binding signals in lia and yhcYZ operons 30 min after the addition of enduracidin are shown by vertical bars at appropriate genome coordinates, together with transcriptional signals in enduracidin-treated and control cells. (B) DNA fragments that co-purified with LiaR and those in the supernatant fraction used for affinity purification were used as templates for PCR amplification. Quantitative PCR was performed with primers specific for the promoter sequences of the lia and yhcZY operons, using serial dilutions of DNA fragments, as indicated. |
Additionally, HtrA and HtrB proteases under the control of the CssRS TCS (which senses abnormal membrane proteins) (Darmon et al., 2002) were induced by antibiotics, as discussed below.
Expression of the lia operon is strongly induced by antibiotics that interfere with the lipid II cycle of peptidoglycan biosynthesis, although liaRS inactivation did not affect the sensitivity of B. subtilis cells to such antibiotics (Pietiainen et al., 2005). A recent study disclosed that daptomycin, which is supposed to disrupt cell membrane structure directly, also induced expression of the lia operon, and that deletion of the liaH gene led to a three-fold increase in susceptibility to daptomycin (Hachmann et al., 2009). We created clean deletion mutants of each gene in the lia operon to avoid any polar effect of selection marker fragments on transcription of downstream genes, and examined their sensitivities to enduracidin and bacitracin by spotting serial dilutions of wild-type and mutant cells onto LB agar plates containing 0.005 or 0.01 μg/mL enduracidin and 50 μg/mL bacitracin (Fig. 4). Growth inhibition by enduracidin was more severe on agar plates, compared to liquid medium, for an unknown reason. We included in analysis a bcrC mutant known to be sensitive to bacitracin, as a control. The data of Fig. 4 show that inactivation of bcrC increased the bacitracin sensitivity of B. subtilis cells, but inactivation of the lia operon did not affect sensitivity. Notably, we found that inactivation of liaR or liaS resulted in an enduracidin-hypersensitive phenotype. Furthermore, we demonstrated that not only the LiaH protein (homologous to the E. coli PspA protein), but also the membrane protein LiaI co-induced with LiaH, played an important role in mediation of enduracidin resistance. The LiaR binding specificity shown in Fig. 3 suggests that the increased enduracidin sensitivity of liaS and liaR mutants was attributable to a failure to induce liaIH expression. It was interesting to note that inactivation of liaG had no effect on enduracidin resistance, but liaF inactivation, that induced constitutive expression of the liaRS operon (Jordan et al., 2006), led to increased resistance to the antibiotic.
 View Details | Fig. 4 Effect of inactivation of genes of the lia operon on enduracidin and bacitracin resistance. Overnight cultures of wild-type and mutant B. subtilis cells in LB medium, grown at 37°C, were diluted to an OD600 value of 1.0, and further serially diluted as indicated. Next, 2 μL aliquots of cells were spotted onto LB agar plates with or without antibiotics, and incubated overnight at 37°C. |
The E. coli homolog of LiaH, PspA, is induced under membrane stress conditions and is proposed to bind to membrane phospholipids to suppress proton leakage (Kobayashi et al., 2007). Although the mechanisms of action of enduracidin and daptomycin are different, the antibiotics are structurally related lipodepsipeptides produced by Streptomyces species. The primary mode of action of daptomycin is considered to be direct disruption of the functional integrity of the cytoplasmic membrane to induce depolarization. However, inhibition of peptidoglycan biosynthesis, either directly or indirectly, is also thought to be a feature of daptomycin action (Muthaiyan et al., 2008). Conversely, although the primary target of enduracidin is the transglycosylation step of peptidoglycan biosynthesis, the drug might also induce a distortion of the membrane structure similar to that caused by daptomycin, which needs to be counteracted by LiaH to re-establish membrane functionality. The lia locus of Bacillus species consists of five core genes, homologs of liaIHFSR, whereas only liaFSR-homologous genes are conserved in more distantly related firmicutes bacteria (Jordan, et al., 2008). Although the sensitivity of the liaI mutant to daptomycin has not been examined, LiaI might play an important role in supporting LiaH function in Bacillus species.
Additionally, inactivation of bcrC seemed to reduce enduracidin resistance at the higher concentration. Although BcrC was first identified as a bacitracin-resistance determinant, subsequent studies demonstrated that BcrC has an undecaprenyl pyrophosphate (UPP) phosphatase activity (Bernard et al., 2005). The bcrC mutant has been reported to be sensitive to paraquat stress (Cao et al., 2005), in addition to bacitracin. Thus, inactivation of bcrC will influence resistance to broader stresses through unknown changes in cell envelope structure.
Among the 129 genes listed in Table 2, we found that 33 belonged to the SigM regulon reported by Eiamphungporn and Helmann (2008). It is known that activated ECF sigma factors auto-activate their own expression, and the self-activation of sigM transcription was evident in our analysis. Generally, transcription induction of the SigM regulon by enduracidin and bacitracin was similar at 10 min after drug addition, but further induction at 30 min was observed only in the presence of enduracidin. Transcriptional levels of seven ECF sigma factors encoded in the B. subtilis genome are summarized in Table 3. Although sigV is included in Table 2, the sigV transcriptional levels were much lower than those of sigM. The transcriptional levels of sigW were highest among the ECF sigma factor genes assessed under our growth conditions, but the levels were not affected by either antibiotic. Transcription of sigX was also relatively high, although the level fell after 30 min incubation with antibiotics. Accordingly, we could not identify genes that might be upregulated by SigW or SigX after the addition of either antibiotic (Supplementary Table 2).
 View Details | Table 3 Transcription levels of ECF sigma factor genesa) |
Next, we examined the effect of inactivation of each gene encoding an ECF sigma factor on sensitivity to enduracidin and bacitracin. The data of Fig. 5 indicate that single inactivations of ECF sigma genes did not affect bacitracin sensitivity, as reported previously (Mascher et al., 2003). The bcrC gene is known to be the target of SigM, but bcrC has been suggested to be a target of SigW, SigX, and SigV. These latter sigma factors would compensate for inactivation of SigM function. On the other hand, consistent with stronger induction of the SigM regulon in the presence of enduracidin, inactivation of sigM increased sensitivity to the antibiotic, to a level similar to that seen in the liaR mutant (Fig. 5), suggesting that some gene(s) specifically induced by SigM also plays an important role in enduracidin resistance. However, the decreased enduracidin resistance displayed by the sigX mutant at the higher concentration was unexpected. In contrast, change in bacitracin resistance was not observed on LB plates containing 100 or 150 μg/mL bacitracin (data not shown). Transcriptomic analysis of the sigM and sigX mutants, seeking to identify genes specifically regulated by these sigma factors in the presence of enduracidin, are required to clarify the molecular mechanisms of involvement of the factors in enduracidin resistance.
 View Details | Fig. 5 Effect of inactivation of ECF sigma factors on enduracidin and bacitracin resistance. The sensitivities of wild-type and mutant cells to enduracidin and bacitracin were examined as described in the legend to Fig. 4. |
The Spx protein is a unique regulator interacting with the α subunit of RNA polymerase and reorganizing gene expression in response to thiol-oxidative stress (Nakano et al., 2003; Reyes and Zuber, 2008). The Spx protein represses expression of target genes by blocking interaction of gene activators with the α subunit of RNA polymerase. The Spx protein also activates expression of a number of genes by recruiting RNA polymerase through recognition of a motif located upstream of promoters (Reyes and Zuber, 2008). An oxidized form of Spx, in which the N-terminal CXXC motif is in the disulfide state, is essential for formation of the active Spx-RNA polymerase complex, and spx expression is under complex regulation, including activation by SigM. Indeed, Spx expression was induced by enduracidin and bacitracin in our analysis. Furthermore, we found that 21 of the genes listed in Table 2 were activated by Spx (Nakano et al., 2003), strongly suggesting that thiol-oxidative stress is induced by enduracidin and bacitracin challenge. In addition to induction of the CssRS regulon (Darmon et al., 2002) as described above, we observed the induction of proteases belonging to the CtsR regulon that is involved in protein quality control in the cytoplasm (Miethke et al., 2006). Activation of these genes might result from damage to proteins caused by thiol-oxidative stress. Supporting this idea, co-induction of CtsR and Spx regulons has been observed after treatment of B. subtilis cells with 6-brom-2-vinyl-chroman-4-on (chromanon), 2-methylhydroquinone (2-MHQ), and salicylic acid, all of which would cause protein damage (Duy et al., 2007; Nguyen et al., 2007). The B. subtilis sigI gene encodes an alternative sigma factor of the sigma70 family, and is involved in heat-shock resistance (Tseng and Shaw, 2008). The mreBH gene listed in Table 2 was recently found to be the target of SigI, and the transcriptional level of sigI was increased by the two antibiotics (Supplementary Table 2), although the levels did not satisfy our criteria for extraction of genes for inclusion in Table 2. The phenomena discussed above may also be related to damage to proteins.
Although induction of thiol-oxidative stress by antibiotics targeting the cell envelope has not been previously reported, obvious TrxA induction has been observed in daptomycin- and friulimicin B-treated cells (Wecke et al., 2009), suggesting that thiol-oxidative stress induction might be general under cell wall stress conditions induced by antibiotics. However, inactivation of spx did not affect the sensitivity of B. subtilis cells to enduracidin and bacitracin (data not shown). Further studies are necessary to clarify the relationships between thiol-oxidative stress induction and actions of antibiotics interfering with cell envelope integrity.
The list of genes upregulated by enduracidin and/or bacitracin (Table 2) contains about 40 additional genes whose induction mechanisms and biological significance in relation to drug resistance remain unclear. Induction profiles of a number of these genes were similar to those of genes of the SigM or Spx regulons, and some of the “other” genes may be under the control of SigM or Spx. Several genes were transiently induced at 10 min after drug addition. Induction of the gerA operon by bacitracin was observed by Mascher et al. (2003), and gerA was also induced by daptomycin (Hachmann et al., 2009). Mascher et al. (2003) observed induction of the ytrABCDEF operon by bacitracin, whereas we found induction only in the presence of enduracidin. Induction of these genes may be very sensitive to subtle changes in cell envelope structure, as is the case with some of the TCSs discussed above. Further experiments are necessary to evaluate the biological significance of these “other” genes in relation to antibiotic resistance.
Mascher et al. (2003) additionally proposed that the SigB-dependent general stress regulon is activated by bacitracin. However, expression of genes listed as belonging to the sigB regulon by Hoper et al. (2005) was not greatly affected in our experiments (Supplementary Table 2), except for yvgN, suggesting that the growth conditions used by Mascher et al. (2003) more severely affected B. subtilis cells than did our experimental conditions.
Although previous transcriptomic studies of transcriptional responses to antibiotics focused on upregulated genes to identify genes that would involve in antibiotics resistance, we additionally examined significantly downregulated genes by antibiotics treatment. We identified 85 genes whose transcription level was reduced more than five hold in ratio and/or more than 10,000 units in level by enduracidin and/or bacitracin compared to control cells (Supplementary Table 3). Interestingly, 39 genes among them encode transporters, mainly for sugars, nucleotides, and ions, or belong to oprerons encoding them. In addition, products of other 14 genes are also associated with cell membrane, including ATP sythetase and cytochrome aa3 quinol oxidase complexes. Downregulation of these genes was observed essentially both in enduracidin- and bacitracin-treated cells, suggesting that these changes would be consequences of changes in metabolite flow due to perturbation of membrane structure by antibiotics. In addition, downregulation of genes encoding translational machinery (8 genes) and metabolic enzymes (10 genes) occurred mainly in cells treated with enduracidin for 30 min. In sum, in agreement with general view, we could not clearly detect candidate genes that might be involved in antibiotics resistance, among significantly downregulated genes.
The responses of bacterial cells to challenge by external antibiotics are very complicated, consisting of relatively specific responses to the dysfunctional activities of antibiotics and rather general responses to ensure cell functionality. The robustness of the latter responses often makes it difficult to evaluate the contribution of any particular genetic system, because the inactivation of one system might be compensated for by induction of other systems. However, an understanding of bacterial cell responses to antibiotics, at the molecular level, is important to understand the mechanisms of action of antibiotics and the emergency responses of antibiotic resistance systems. Toward this goal, transcriptional responses of B. subtilis cells to antibiotics have been extensively studied as a model system.
In the present work, we used a high-density tiling chip to assess transcriptional responses in B. subtilis cells challenged by antibiotics. Using the quantitative advantage of the tiling chip, we introduced a new criterion, an increase in transcriptional level, in addition to the conventional induction ratio, to identify genes of biological significance among genes with lower induction ratios. Our results indicate that introduction of this criterion led to unambiguous identification of core transcription responses to antibiotics, with a reduction in the number of possible background genes, compared to previous results employing gene arrays. Indeed, induction of about half of the genes belonging to the SigM and Spx regulons was identified, using induction levels to distinguish these genes from those of numerous genes with lower induction ratios. Although the minimum induction level to identify significantly upregulated genes was arbitrarily set at 10,000 units, it is interesting to note that among 35 genes added to the upregulated gene list by decreasing the criterion to 8,000 units, 14 belonged to the SigM or Spx regulons (data not shown), indicating that resetting of the minimum value would not greatly affect the main conclusions of our work.
In addition, timecourse analysis showed potential to clarify the complex and dynamic changes in the transcriptome caused by the need to adapt to antibiotic stress. Timecourse analysis suggested that expression of several genes might be transiently induced by initial perturbations in cell envelope structure. However, induction of transcription of genes that are more directly involved in antibiotic resistance persists for longer periods.
The present study, which represents the first transcriptomic analysis of responses of a bacterial cell to challenge by enduracidin, showed that multiple factors contribute to enduracidin resistance. Notably, we determined that inactivation of LiaRS TCS resulted in increased sensitivity to enduracidin, probably caused by a failure to induce LiaHI proteins. Enduracidin may have the ability to cause a distortion of membrane structure similar to that created by the related antibiotic daptomycin, which needs to be counteracted by LiaH to ensure cell functionality. We also demonstrated that LiaI, conserved in Bacillus species, plays an important role in supporting LiaH function. These findings contribute to a further understanding of the molecular mechanisms of action of enduracidin and daptomycin, and the role of LiaHI in maintaining the integrity of membrane structure. Consistent with induction of the SigM regulon, inactivation of sigM increased sensitivity to enduracidin, to a level similar to that of the liaR mutant. In addition, and unexpectedly, SigX also seemed to contribute to enduracidin resistance. Further studies are necessary to clarify the molecular mechanisms of involvement of these sigma factors in enduracidin resistance. In addition, and for the first time, we found that the Spx regulon was induced in cells challenged by antibiotics, suggesting that thiol-oxidative stress would develop in cells treated with enduracidin or bacitracin. Upregulation of spx expression by SigM may contribute to Spx regulon induction. However, the biological significance of this induction in the context of antibiotic resistance awaits further examination, as spx inactivation did not affect drug sensitivity.
We used a custom tiling chip to quantitatively monitor the genome-wide transcriptional profile. It is now possible to analyze transcriptional profiles quantitatively by massive sequencing of cDNAs using new high-throughput sequencers (Wang et al., 2009), instead of employing a high-density tiling array. Thus, quantitative transcriptome analysis is now available for any bacterium the genome sequence of which is known, and may be employed to more clearly disclose dynamic transcriptional responses of bacteria to environmental stimuli, including challenge by an antibiotic.
This work was supported by KAKENHI (via a Grant-in-Aid for Scientific Research) of the Priority Areas Systems Genomics, from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. We are grateful to Shu Ishikawa for critical reading of manuscript. We thank the project ‘Development of a Technology for the Creation of a Host Cell’, supported by the New Energy and Industrial Technology Development Organization (NEDO), for supplying B. subtilis tiling chips, and Sachiko Iida for financial support to AR.
|
Index