The response regulator/histidine kinase pair LiaRS of Bacillus subtilis, together with its membrane-bound inhibitor protein LiaF, constitutes an envelope stress-sensing module that is conserved in Firmicutes bacteria. LiaR positively autoregulates the expression of the liaIH-liaGFSR operon from a strictly LiaR-dependent promoter (PliaI). A comprehensive perturbation analysis revealed that the functionality of the LiaFSR system is very susceptible to alterations of its protein composition and amounts. A genetic analysis indicates a LiaF:LiaS:LiaR ratio of 18:4:1. An excess of LiaS over LiaR was subsequently verified by quantitative Western analysis. This stoichiometry, which is crucial to maintain a functional Lia system, differs from any other two-component system studied to date, in which the response regulator is present in excess over the histidine kinase. Moreover, we demonstrate that LiaS is a bifunctional histidine kinase that acts as a phosphatase on LiaR in the absence of a suitable stimulus. An increased amount of LiaR – both in the presence and in the absence of LiaS – leads to a strong induction of PliaI activity due to phosphorylation of the response regulator by acetyl phosphate. Our data demonstrate that LiaRS, in contrast to other two-component systems, is non-robust with regard to perturbations of its stoichiometry.
Two-component systems (2CSs) are a ubiquitously distributed principle of signal transduction that allows a bacterial cell to respond to changes in environmental and cellular parameters (Stock et al., 2000; Mascher et al., 2006; Gao and Stock, 2009). Typically, these systems consist of a sensor histidine kinase (HK) and a cognate response regulator (RR). In the presence of a suitable stimulus (= input), the HK undergoes a conformational change, which leads to the activation of the catalytic domain and ultimately to the autophosphorylation of an invariant histidine residue. Subsequently, the phospho-HK serves as phospho-donor for the corresponding RR, leading to the phosphorylation of a conserved aspartate residue in the receiver domain of this RR. This phosphorylation leads to the activation of the RR, which can then mediate the cellular response, usually through differential expression of its target genes (= output). The system can be set back to the pre-stimulus state by dephosphorylation of the response regulator (Parkinson, 1993; Stock et al., 2000). This activity can be provided by specific phosphatases, but mostly it is the intrinsic property of the cognate HKs themselves, which are often bifunctional kinases/phosphatases. In the absence of a stimulus, such HKs act as phosphatases, thereby reducing or even preventing undesired phosphorylation of the cognate RR both from cross-talk and through the cellular pool of acetyl phosphate (Laub and Goulian, 2007).
The LiaRS system is one of over 30 2CSs encoded in the genome of the Gram-positive model organism Bacillus subtilis (Fabret et al., 1999). This 2CS was originally identified as one of the signalling devices orchestrating the cell envelope stress response in this organism (Mascher et al., 2003). LiaRS strongly responds to the presence of a number of cell wall antibiotics, such as bacitracin, daptomycin, nisin, ramoplanin and vancomycin (Mascher et al., 2004; Pietiäinen et al., 2005; Hachmann et al., 2009; Wecke et al., 2009) and is also weakly induced by other more unspecific stresses that interfere with envelope integrity, such as alkaline shock and secretion stress (Wiegert et al., 2001; Hyyryläinen et al., 2005). Upon activation, phosphorylated LiaR strongly induces expression from the strictly LiaR-dependent promoter PliaI upstream of the liaIH-liaGFSR operon, thereby also autoregulating expression of the LiaRS 2CS. This leads to the generation of two distinct transcripts: the major liaIH transcript and a transcript of the entire liaIH-liaGFSR locus. In the absence of a suitable stimulus, the expression of the LiaRS 2CS is ensured by a weak constitutive promoter, PliaG, located upstream of liaG (Jordan et al., 2006). Phylogenomic studies revealed that LiaRS is a conserved cell envelope stress-sensing 2CS present in most Firmicutes (low G+C Gram-positive) bacteria (Jordan et al., 2006, 2008). It is linked by genomic context and function to liaF, encoding a membrane protein that acts as a specific inhibitor of LiaRS. In liaF deletion mutants, the LiaRS 2CS is constitutively active even in the absence of envelope stress (Jordan et al., 2006). This observation was later confirmed by a study of the homologous system in Streptococcus mutans (Suntharalingam et al., 2009).
Previous studies have shown that the LiaR-dependent target promoter upstream of the liaIH operon (PliaI) is tightly regulated by at least three different transcriptional regulators, LiaR, AbrB and Spo0A (Jordan et al., 2007). The close interplay between the corresponding regulatory cascades ensures that this promoter is almost shut off during exponential growth. Under inducing conditions, the activity of PliaI can increase almost three orders of magnitude in a LiaR-dependent fashion (Mascher et al., 2004; Jordan et al., 2006), with its primary target, LiaH, becoming the most abundant cytosolic protein in B. subtilis (Wolf et al., 2010). LiaH is a homologue of the Escherichia coli phage shock protein (Psp) A. The latter maintains the proton motive force under extracytoplasmic stress conditions that affect the membrane integrity (Model et al., 1997; Darwin, 2005). Because of the similar structure of PspA and LiaH and an overlapping inducer spectrum, we hypothesize that LiaFSR mounts a PspA-like response (Wolf et al., 2010). The small membrane protein LiaI seems to act as membrane anchor for LiaH. The cellular function of LiaG is not known so far.
We have previously shown that the PliaI activity can be locked in constitutive ‘ON’ or ‘OFF’ states in a liaF or liaR mutant respectively (Wolf et al., 2010). Both of these extreme situations are exclusively Lia-dependent. In aiming to unravel the mechanism behind the tight regulation and impressive dynamic range of PliaI-dependent transcription initiation, we focused our attention on the mechanism of signal transduction mediated by the LiaFSR system.
This article describes the results of comprehensive genetic and biochemical studies on the stoichiometry of the LiaFSR system. A systematic deletion/complementation screen revealed that the LiaFSR system is very susceptible to alterations of its protein compositions and amounts. Under native conditions, the amount of LiaF exceeds that of LiaS. Likewise, an excess of the HK LiaS over LiaR is also crucial for the functionality of the LiaFSR system. Our data indicate that the stoichiometry of the Lia system is essential for the dynamic range and output observed at the level of its target promoter, PliaI. Increasing the abundance of either LiaS or LiaR leads to ‘locked-ON’ phenotypes, even in the absence of a stimulus. Hence, the Lia system behaves non-robustly with regard to its protein stoichiometry, in contrast to other 2CSs studied in this respect.
Hyperactivity of PliaI is the result of positive polar effects from inserted resistance cassettes
We previously noticed by β-galactosidase assay using a PliaI–lacZ reporter strain that the constitutive liaI promoter activity of a liaF mutant, in which the gene has been replaced by a kanamycin resistance cassette, was 10 times higher than the maximum promoter activity in the induced wild type (Jordan et al., 2006) (see Fig. 1A). Surprisingly, a liaS mutant constructed in a similar way also shows a ‘locked-ON’ behaviour with regard to the activity of the strictly LiaR-dependent PliaI, despite the absence of the HK responsible for activating LiaR (Fig. 1A). But here, the activity was lower and comparable to the induced wild type. Subsequent time-course experiments revealed a PliaI-dependent accumulation of β-galactosidase over time in the liaS mutant (data not shown), indicating a significantly increased LiaR-dependent basal expression. Both, the liaS::kan and the liaF::kan mutants no longer responded to the extracellular addition of sublethal bacitracin concentration (Fig. 1A), and in each case, the observed PliaI activity was still LiaR-dependent (data not shown). These observations could be explained either by postulating an important role of LiaS as a phosphatase under these conditions, and/or by assuming polar effects of the inserted kanamycin resistance cassette on the expression of downstream gene(s). We first investigated the second hypothesis.
Because of the strong increase in LiaR-dependent PliaI activity, we suspected positive polar effects as the possible reason for the observed behaviour. Therefore, we used markerless deletion mutants of both genes, constructed with the pMAD vector system (Arnaud et al., 2004). The resulting PliaI reporter strains TMB216 (ΔliaS) and TMB331 (ΔliaF) now showed significantly reduced basal expression levels. While the latter confirmed the inhibitor function of LiaF, the overall promoter activity was now comparable to the induced wild type, irrespective of the presence or absence of bacitracin (Fig. 1A). Likewise, the ΔliaS strain now showed a locked-OFF behaviour comparable to the uninduced wild type, as expected (Fig. 1A; for values, see Table S2).
To rule out any artefacts derived from the PliaI–lacZ reporter system used to study LiaR activity, we verified the observed differences between the liaF::kan and the ΔliaF mutants by Western blot analysis, monitoring LiaH expression in the absence of bacitracin by specific polyclonal antibodies (Fig. 1B). Again, we observed significant differences in LiaH accumulation comparable to the results from the β-galactosidase assays described above. For the liaS::kan mutant, the strong positive polar effect of the kanamycin resistance cassette was directly visualized by Northern blots (Fig. 2). Figure 2A shows a schematic representation of the lia locus in the wild type and the liaS::kan mutant, as well as the expected and observed transcripts. The Northern blot revealed a significant overexpression of a liaR-specific transcript that correlates in size with a kan-liaR mRNA (Fig. 2B).
We next quantified the expression levels of liaF, liaS and liaR by real-time RT-PCR in the wild type (+/− bacitracin) and the allelic replacement/clean deletion mutants of liaF and liaS (Table 1). The results are in complete agreement with the data described above. Replacement of liaF by the kanamycin cassette results in a 10-fold increased level of liaS compared with the clean deletion. Likewise, liaR expression is increased 20-fold in the liaS::kan strain compared with the ΔliaS mutant. Taken together, our data demonstrate that the strong positive polar effect (10- to 20-fold induction of downstream genes) strongly affects the LiaR-dependent output at the level of target promoter (PliaI) activity.
bGenes quantified, using primer pairs TM0630/TM0631 (liaF), TM0628/TM0629 (liaS) and TM0093/TM0094 (liaR) (Ct, threshold cycle; n.a., not applicable; n.d., not determined).
WT (− Bac)
8.2 ± 0.1
49 ± 8
8.3 ± 0.2
47 ± 16
13.8 ± 0.3
WT (+ Bac)
2.7 ± 0.1
2165 ± 419
44 ± 1
3.0 ± 0.1
1791 ± 412
39 ± 4
9.3 ± 0.2
22 ± 2
0.8 ± 0.1
8568 ± 1983
189 ± 20
7.5 ± 0.2
77 ± 8
4.1 ± 0.2
910 ± 303
19 ± 0.1
8.9 ± 0.3
31 ± 2
8.5 ± 0.1
41 ± 8
13.3 ± 0.3
2 ± 1
LiaR is phosphorylated by the cellular pool of acetyl phosphate in the absence of LiaS
While positive polar effects could be identified as the reason for the observed PliaI hyperactivity in strains harbouring allelic replacement mutations, the strong promoter activity in case of the liaS::kan mutant was still puzzling. Activation of RR activity depends on a HK-dependent phosphorylation and subsequent dimerization. We therefore wondered if phosphorylation of LiaR played a role in the observed promoter activity in the liaS::kan mutant. To address this question, we constructed a mutant in which we introduced a single point mutation during the allelic replacement of liaS, which leads to an aspartate to alanine exchange in LiaR, thereby rendering the invariant site of RR phosphorylation dysfunctional. This strain (TMB247: liaS::kan, LiaR D54A) also showed a strong expression of a kan-liaR transcript, but no expression of the LiaR-dependent liaIH or liaIHGF-kan-liaR transcripts (Fig. 2B). Accordingly, the resulting PliaI–lacZ reporter strain (TMB232) only showed basal expression levels comparable to the uninduced wild type (Fig. 1A). While we cannot rule out that the introduced amino acid exchange somehow affected LiaR stability or folding, our data nevertheless strongly suggests that phosphorylation of LiaR is a prerequisite for the observed output in the absence of the cognate HK LiaS.
Acetyl phosphate has been described in the literature as a small molecule phospho-donor capable of phosphorylating response regulators in vivo, since it can reach intracellular concentration comparable to those of ATP (about 3–5 mM), at least in E. coli (McCleary and Stock, 1994; Wolfe, 2005; Klein et al., 2007). While the phosphatase activity of bifunctional HKs usually prevents undesired phosphorylation of their cognate RRs in the absence of suitable triggers, some RRs are readily phosphorylated in their absence, as has been demonstrated for the VanRS system of Streptomyces coelicolor (Hutchings et al., 2006). We therefore hypothesized a similar scenario for the Lia system.
Acetyl phosphate is produced from pyruvate via acetyl-CoA as part of the cellular overflow metabolism. Under normal conditions, acetyl-CoA is synthesized from pyruvate and metabolized by the tricarboxylic acid (TCA) cycle. Whenever too much acetyl-CoA is present in the cell to be metabolized via the TCA cycle, the excess acetyl-CoA is converted to acetyl phosphate. This reaction is catalysed by the enzyme phospho-transacetylase, which is encoded by the pta gene. Acetyl phosphate is then converted to acetate (a reaction catalysed by the acetate kinase AckA), which is released into the medium. If the cells grow on acetate, the order of biochemical reactions is inverted (Wolfe, 2005).
The cellular amount of acetyl phosphate can be influenced by the carbon sources supplied to the medium. For E. coli, the highest concentrations of acetyl phosphate were determined with pyruvate as carbon source (McCleary and Stock, 1994; Wolfe, 2005; Klein et al., 2007). We therefore decided to compare the PliaI activity of the liaS::kan mutant in the presence and absence of pta or ackA with varying carbon sources. The results are summarized in Fig. 3.
In the chemically defined CSE medium (Stülke et al., 1993), the PliaI activity strongly responds to the available carbon source, at least in the liaS::kan mutant. If succinate [0.56% (w/v) final concentration], an intermediate of the TCA cycle that does not feed into the overflow metabolism, is used as the sole carbon source, a significant reduction of the basal PliaI activity of the liaS::kan mutant is observed compared with LB medium. In contrast, addition of pyruvate [0.5% (w/v) final concentration] to CSE medium increased the promoter activity in this mutant more than 10-fold compared with LB medium and 100-fold compared with CSE alone (Figs 1A and 3). The wild type shows comparable PliaI activities in LB medium and CSE medium with or without pyruvate. Irrespective of the carbon source, the promoter activity drops to the same basal level in a pta mutant, thereby clearly demonstrating that most of the increased LiaR-dependent PliaI activity in the liaS::kan background can be attributed to stimulus-independent cross-phosphorylation from acetyl phosphate (Fig. 3 and data not shown), as has been demonstrated at least for some other 2CSs (Laub and Goulian, 2007). An ackA/liaS double mutant, which is not able to convert acetyl phosphate to acetate, shows PliaI activities comparable to the liaS::kan mutant (Fig. 3).
The discrepancy in PliaI activity between the ΔliaS and liaS::kan strains (Fig. 1A) seemed to indicate that phosphorylation of LiaR by acetyl phosphate only leads to a measurable output if LiaR is simultaneously overproduced, as is the case in the liaS::kan strain (Fig. 2B). To address this question, we introduced an additional copy of liaR under the control of a xylose-inducible promoter, into the ΔliaS reporter strain TMB216 (Table 2), resulting in strain TMB641, thereby simulating the situation of strain TMB019. Indeed, this strain showed a similar behaviour (Figs 3 and 1A). And again, the ‘locked-ON’ behaviour was completely dependent on acetyl phosphate, as demonstrated by the lack of promoter activity after introducing the pta::tet allele into this strain (Fig. 3).
To unequivocally demonstrate that indeed acetyl phosphate is responsible for the observed stimulus-independent activation of PliaI, we also reversed the reaction by adding acetate to the CSE medium (60 mM final concentration), thereby driving the Pta-AckA pathway backwards. Under these conditions, we received similar results as described for CSE + pyruvate: the PliaI activity of the liaS::kan mutant again increases about 100-fold compared with CSE medium alone and drops to a basal level in an ackA::mls mutant, which is no longer able to produce acetyl phosphate from acetate anymore. The same basal PliaI activity is observed for the ΔliaS mutant, which does not overproduce LiaR (Fig. 3).
Taken together, the data obtained so far strongly suggests that LiaR can be phosphorylated by the cellular pool of acetyl phosphate, but only if LiaR is overexpressed in the absence of the cognate HK LiaS. This artificial susceptibility of LiaR for acetyl phosphate-dependent activation is therefore in good agreement with the results obtained for other bacterial RR (Laub and Goulian, 2007).
LiaS is a bifunctional HK that possesses a phosphatase activity
The data described in the previous section suggests that LiaS is a bifunctional kinase that functions as a phosphatase in the absence of inducing conditions. LiaS belongs to the HPK7 family of HKs, which harbours a HisKA_3 dimerization and histidine phosphotransfer domain. Within this domain, a conserved DxxxQ motif was recently identified (Huynh et al., 2010). For the HK NarX it was shown that the glutamine residue plays a critical role in phosphatase activity, and the exchange by an alanine, glutamate or histidine residue results in a kinase ON, phosphatase OFF protein (Huynh et al., 2010). Since this motif is also found in LiaS (data not shown), we introduced a copy of liaS, in which the conserved glutamine 164 residue was substituted by an alanine, into the ΔliaS mutant and analysed the PliaI activity by β-galactosidase assay (Fig. 4). In contrast to the wild type behaviour of the complementation mutant carrying a native liaS gene, the LiaS Q164A mutant shows a constitutive PliaI activity comparable to the induced wild type. These genetic findings, together with the physiological data described above, strongly suggest that LiaS possesses a phosphatase activity. This is in perfect agreement with biochemical in vitro evidence from the direct LiaS orthologues VraS of Staphylococcus aureus (Belcheva and Golemi-Kotra, 2008) and LiaSLm of Listeria monocytogenes (Fritsch et al., 2011), which were both demonstrated to be bifunctional HKs. In fact, all HKs belonging to the family HPK7 investigated so far, such as DesK or NarX/Q are bifunctional kinases (Schröder et al., 1994; Albanesi et al., 2004). In the case of LiaS, this phosphatase activity is very important to keep the output (PliaI activity) switched off in the absence of inducing conditions.
Overproduction of LiaS or LiaR – but not LiaF – affects the functionality of the LiaFSR system
The results obtained so far seemed to indicate that the LiaFSR system is very susceptible to changes in the relative stoichiometry of its three components. Specifically, the observed activation of LiaR in the absence of LiaS provoked the question: would an artificial increase of the RR be sufficient to result in a ‘locked-ON’ phenotype, even in the presence of LiaS? To address this question, we overproduced each of the three proteins in the wild type reporter strain, using the xylose-dependent pXT expression system (Derre et al., 2000).
Overproduction of the inhibitor protein LiaF had only mild effects on the functionality of the LiaRS 2CS. The maximum PliaI activity in the presence of bacitracin was reduced by a factor of three, while no change of the uninduced basal expression level was observed (Fig. 5A). The intact expression of a functional LiaF protein from this construct was verified by its ability to suppress the ‘locked-ON’ phenotype of a liaF mutant (data not shown). Therefore, we conclude that the LiaFSR system is relatively robust with regard to increasing LiaF concentrations. However, we cannot exclude that the mild effect on PliaI might also be the result of only a weak overproduction of LiaF. Since both the purification of LiaF (to raise antibodies against the protein) and also Western blots against an epitope-tagged functional LiaF, expressed from its native chromosomal position, failed (data not shown), this hypothesis can unfortunately not be verified experimentally at the moment.
In contrast, overproduction of LiaR resulted in a ‘locked-ON’ phenotype (Fig. 5A). As hypothesized, a xylose-dependent increase in the amount of LiaR therefore results in the same behaviour as observed in the liaS mutants, above. Since liaS is not overexpressed in this strain, the increase in PliaI activity without external stimuli might be caused by a phosphorylation via acetyl phosphate that can appear due to an inefficient dephosphorylation by LiaS. For LiaS, we observed an intermediate phenotype. Overproduction significantly increased the uninduced basal expression level without affecting the response to bacitracin (Fig. 5A).
Because of the weak Shine–Dalgarno (SD) sequence upstream of liaS (see below for details), we wondered if this intermediate behaviour was due to inefficient translation initiation, resulting in only a moderate increase in the cellular amount of LiaS. To investigate this hypothesis, we cloned two different FLAG3-tagged liaS alleles, one under the control of its native SD sequence into the vector pXT (Derre et al., 2000), the other with an optimized ribosome binding site provided by the vector pALFLAG3 (Schöbel et al., 2004) respectively. The resulting plasmids were then introduced into strain TMB216 (ΔliaS, PliaI–lacZ), resulting in strains TMB500 and TMB501 respectively (Table 2). The results of the β-galactosidase assay are shown in Fig. 5B. In the presence of xylose, the ΔliaS mutant did not show any promoter activity, as observed before (Fig. 1A). Expression of liaS under the control of its native SD sequence restored the phenotype to wild type behaviour, indicating that sufficient amounts of intact LiaS were produced. In contrast, expression of LiaS with the optimized SD sequence resulted in a ‘locked-ON’ behaviour (Fig. 5B). These differences could be directly correlated with the different amounts of LiaS in the cells. Western analysis with FLAG-tag-specific antibodies identified a strongly increased amount of LiaS protein in the membrane fraction when expressed with an optimized ribosome binding site, compared with the complementation with the native SD sequence (Fig. 5C). These results demonstrate that overproduction of both LiaS and LiaR severely perturbates the signal transduction mediated by the LiaFSR system, even in the presence of all other Lia proteins in their native amounts.
In contrast, the simultaneous overexpression of all three genes, liaF, liaS and liaR, which increases the cellular amount of these proteins simultaneously without changing their stoichiometry to each other, shows a comparable behaviour to the wild type (Fig. 5A). This result demonstrates that the absolute protein amounts of LiaFSR have no effect on the PliaI activity. As long as the ratio between LiaS and LiaR is maintained, the phosphatase activity of LiaS is sufficient to prevent stimulus-independent phosphorylation of LiaR by acetyl phosphate.
Taken together, the data obtained from our perturbation studies indicate that the LiaFSR system seems to behave non-robustly with regard to alterations in the cellular ratios of LiaS and LiaR relative to the other protein components of the Lia system. We hypothesized that the stoichiometry of LiaF:LiaS:LiaR is very important for the functionality of Lia-dependent signal transduction. To study this in more detail, we next performed an in-depth genetic analysis of the wild type expression levels of the three genes/proteins by determining transcription and translation initiation both independently and in conjunction. For LiaSR the results were subsequently also verified by quantitative Western blot analyses to estimate the number of proteins in the cell.
Transcription of liaR is almost 50-fold weaker compared with liaF or liaS
We first studied the expression of the three genes by quantitative real-time RT-PCR in the wild type strain W168, both in the absence and in the presence of bacitracin. The first condition monitors the intrinsic basal expression level from the constitutive promoter upstream of liaG, while the latter reflects the combined activity of PliaG and the LiaR-dependent PliaI(Fig. 2A). The results are given in Table 1, which summarizes the determined fold changes relative to the liaR transcription of the uninduced wild type. We detected almost equal amounts of liaF and liaS transcripts. Surprisingly, liaR expression was significantly lower under both conditions tested. The values indicate an overall ratio of transcription of 49:47:1 (liaF:liaS:liaR) for uninduced wild type cells.
A sequence analysis of the liaFSR region identified two stem-loop structures at the very end of the liaS gene and close to each other (Fig. S1). While they lack both the strength and the poly-U run-off typical for classical rho-independent terminators, they could nevertheless be responsible for a significant amount of premature transcription termination that would account for the observed differences in transcript levels. Another possibility is that the 3′ end of the liaFSR-specific transcript is subject to RNase degradation.
Translation initiation of LiaS is severely impaired by its weak Shine–Dalgarno sequence
We next analysed the contribution of translation initiation to the overall expression of the three proteins. Towards that end, we used joining-PCR (see Experimental procedures) to fuse the strong constitutive promoter Pveg (Moran et al., 1982) with short chromosomal regions of 24 nt length, directly upstream and including the start codon of each of the three genes (Fig. 6A). The resulting PCR products were then cloned into the pAC7 vector (Weinrauch et al., 1991) to generate translational fusions with the lacZ gene. Therefore, any detectable β-galactosidase activity is the result of the promoter and SD sequence provided on the cloned fragments. Likewise, any difference between the activities of the three constructs is a direct consequence of the translation initiation sequences of the short DNA fragments shown in Fig. 6A. As a negative control, Pveg was cloned into pAC7 in a similar manner, but lacking a SD sequence. The B. subtilis wild type strain W168 was transformed with the four plasmids and also the empty vector, resulting in strains TMB466, TMB469 and TMB478–480 (Table 2). β-Galactosidase assays were performed with lysates from cells harvested during mid-exponential growth phase without induction. The results are shown in Fig. 6B.
Both the empty vector and the plasmid that contains Pveg without a SD sequence did not show any activity. The level of translation initiation of the three complete fragments differed significantly. While translation of the β-galactosidase from the SD sequences of liaF and liaR only differed two- to threefold, the β-galactosidase activity was 10- to 30-fold lower for liaS (Fig. 6B). This result correlates very well with the weakly conserved SD sequence of liaS (Fig. 6A), and the data shown in Fig. 5C.
Transcription and translation initiation combined indicate a LiaF:LiaS:LiaR ratio of 18:4:1
To study the combined effects of transcription and translation initiation on protein expression of the LiaFSR system in its natural genetic context, we next translationally fused the three fragments shown in Fig. 7A with a promoter-/SD-less lacZ gene using the vector pAC5 (Martin-Verstraete et al., 1992). Initially, we used the weak native promoter upstream of liaG (PliaG) (Jordan et al., 2006). But the resulting constructs did not give rise to any detectable β-galactosidase activity (data not shown). Cloning the fragments under control of the strong Pveg used above failed for the two longer fragments, most probably due to toxic effects of liaF expression in E. coli that we had already observed previously. We therefore used an engineered version of PliaG, harbouring an optimized −35 and extended −10 promoter region, and termed PliaG-opt. (Fig. 7B) to introduce as few alterations as possible and also avoid a strong expression of liaF during the cloning procedure. Thus, the β-galactosidase activity of all three reporter strains derived from these plasmids is under the transcriptional control of PliaG-opt. and under the translational control of the specific SD sequences of liaF, liaS or liaR respectively. All strains showed a weak, but clearly detectable β-galactosidase activity (Fig. 7C). Even the expression level for the longest and weakest construct (SDliaR) was significantly (more than twofold) above the background level of the empty vector. Based on the results shown in Fig. 7C, the overall native stoichiometry of LiaF:LiaS:LiaR is 18:4:1.
Quantitative Western blot analysis verifies an excess of LiaS over LiaR
So far, the findings on the native stoichiometry of LiaFSR are based on indirect genetic approaches. To support and substantiate these data, we next performed quantitative Western blot analysis to determine the cellular amounts of the proteins involved. For this, the corresponding genes were either N-terminally (liaF and liaR) or C-terminally (liaS) fused to a FLAG3 sequence that was integrated directly into the native lia locus via pMAD (Arnaud et al., 2004; see Experimental procedures for details). The PliaI activity of the generated strains was checked to ensure wild type behaviour and the functionality of the system in the presence of FLAG3-tagged proteins (Fig. S2). Furthermore, protein standards were expressed and purified from E. coli BL21(DE3) using the overexpression vector pProEx1, which generates N-terminal His6-tag fusions to the cloned genes. To be able to detect the proteins via Western blotting, an additional FLAG3-tag was fused C-terminal to these proteins. Quantitative Western blots were carried out as described in Experimental procedures. The standards were analysed by ImageJ software and band intensities were plotted against known amounts of proteins. Finally, the cellular amounts of proteins were calculated from the standard curve.
For LiaF, all attempts to (over)express and purify the protein failed, presumably because of the toxicity of LiaF production in E. coli, as mentioned above. Furthermore, the detection of a functional FLAG3-tagged LiaF protein was also not successful. This could be due to either an inefficient protein transfer during the Western blot procedure or the loss of the epitope tag in the course of LiaF processing. But for LiaS and LiaR the quantification succeeded. An example of a quantitative Western blot for both proteins is shown in Fig. 8A and B. The corresponding standard curves are represented in Fig. 8C and D. In LB medium grown until mid-exponential phase in the absence of an inducer, we determined about 10 molecules of LiaS per cell, but could not detect any LiaR. We therefore also quantified the protein amounts in cells induced with bacitracin. Here, B. subtilis W168 cells contained an average of 150 molecules of LiaS-FLAG3 per cell. Under the same conditions, we determined ∼ 20 molecules per cell for FLAG3-LiaR (Fig. 8E). Since the ratio of LiaS to LiaR should not be affected by bacitracin induction, this result should also be a reliable measure for the relative amounts of both proteins in uninduced cells. While the absolute numbers for both proteins calculated for individual experiments varied significantly (Fig. 8E) and should not be over-interpreted, especially since we only used one method to calibrate the protein standard curves, these direct measurements are nevertheless in very good agreement with the genetic data described above and support a clear excess (at least three- to fourfold) of LiaS over LiaR. This solid correlation between indirect genetic and direct biochemical data also demonstrates that the genetic approach described in Fig. 7 provides a reliable and easy-to-perform measure to estimate relative protein amounts in cells. Hence, although we failed to determine the exact amount for LiaF, it stands to reason to assume that its amount indeed exceeds that of LiaS, based on its genetic location, strong Shine–Dalgarno sequence and the corresponding results shown in Figs 6 and 7 and Table 1.
In this report, we have comprehensively investigated the native stoichiometry of LiaFSR by genetic and biochemical approaches and the effects of its perturbation on the functionality of this cell envelope stress-sensing system. The data obtained in this study is summarized in Table 3. A graphical model derived from these results is provided in Fig. 9. Taken together, we demonstrate a crucial role of maintaining conditions, in which the amounts of LiaF > LiaS > LiaR.
Table 3. Summary of the deletion/complementation studies
b0, protein amount equal to wild type; +, increased protein amount relative to wild type; (+), increased protein amount relative to wild type due to positive feedback regulation; −, no protein present due to deletion.
cThe activities – derived from the behaviour of the target promoter PliaI – is based on the assumption that LiaS is a bifunctional sensor kinase. Note that this has so far only been demonstrated experimentally by in vitro assays for the LiaS orthologues VraS from Staphylococcus aureus and LiaSLm from Listeria monocytogenes.
All values refer to uninduced conditions, except were labelled otherwise.
WT + liaF
WT + liaS
WT + liaR
WT + liaFSR
ΔliaS, pXT-liaS (opt. SD)
Three very important conclusions can be drawn from the observations reported in this article on the functionality of the LiaFSR system. First, LiaF exerts its function through LiaS, since it does not affect the LiaS-independent phosphorylation by acetyl phosphate (Fig. 3 and data not shown) and its inhibitory effect can be overcome by LiaS overproduction (Fig. 5). Second, in the absence of a stimulus, LiaF maintains LiaS in its phosphatase state. To ensure this, the inhibitor protein LiaF needs to be in excess over LiaS to keep the system silent in the absence of a stimulus. If LiaF is absent or if LiaS is strongly overexpressed even in the presence of LiaF, the output of the Lia system, PliaI activity, is switched on, even in the absence of an inducer (Figs 1 and 5). Third, while LiaS – as its orthologues – is a bifunctional histidine kinase, at least its phosphatase activity seems to be rather inefficient. This hypothesis is derived from the observation that LiaS needs to be more abundant than LiaR in order to prevent stimulus-independent phosphorylation of this RR by acetyl phosphate. If either of the two last prerequisites for LiaFSR functionality is severely perturbated by overexpressing either LiaS or LiaR, the system enters a ‘locked-ON’ state, in which full PliaI activity is reached even in the absence of a signal (Fig. 5A and Table 3). Taken together, the LiaFSR system seems to behave non-robust with regard to relative alterations of its protein stoichiometry. The implications of these observations will be discussed in the following sections.
LiaFSR stoichiometry and robustness
So far, only few 2CSs have been studied with regard to the relative cellular amounts of the sensor kinase and response regulator and how this affects the functionality of the 2CS.
The most detailed quantitative analyses were performed for the paradigmatic 2CS EnvZ-OmpR of E. coli. It was demonstrated that the HK EnvZ is present in significantly lower amounts than its cognate RR OmpR, with ∼ 100 and ∼ 3500 monomeric molecules per cell respectively (Cai and Inouye, 2002). Since HKs usually function as stable dimers, the physiological relevant HK2:RR ratio is 1:70. Subsequently, it was shown that the EnvZ-OmpR system is robust with regard to alterations of the amount of both proteins, as long as OmpR remains in excess over EnvZ (Batchelor and Goulian, 2003). Most recently, the stoichiometry of the VicRK (WalRK) 2CS from Streptococcus pneumoniae (Wayne et al., 2010) was determined. The amount of the HK was the stoichiometrical bottleneck, with the HK2:RR ratios of 1:14.
The data obtained for LiaFSR in this study stands in contrast to these observations. Here, the amount of the RR LiaR is about four- to eightfold lower than that of the cognate HK LiaS (Figs 8E and 9), and maintaining this excess of HK over RR is crucial for the functionality of the LiaRS system. Because of the small differences between the amounts of LiaS2 and LiaR, this system seems to be particularly vulnerable to perturbations of its stoichiometry (Fig. 5 and Table 3). In contrast, the systems described above either have already been shown or are believed to be robust. The insight into the nature of robustness in bacterial signalling is only slowly evolving (Goulian, 2004) and the overall design of the regulatory systems described above are too different to allow drawing general conclusions. But to the best of our knowledge, the data on the LiaFSR system of B. subtilis is the first report on a particularly non-robust 2CS. This unusual stoichiometry could be interpreted as an indication for the formation of a regulatory complex between LiaS and LiaR, in order to control RR activity. As long as LiaS is in excess over LiaR, all RR molecules can be controlled, whereas free LiaR can easily be activated by phosphorylation through the cellular pool of acetyl phosphate.
LiaR and acetyl phosphate
The phosphorylation of LiaR by acetyl phosphate occurs either in the absence of LiaS or when the RR is present in increased amounts relative to LiaS (Figs 3 and 5A). A number of studies have been performed in recent years to establish that acetyl phosphate can act as a small molecule phospho-donor for response regulators both in vitro and in vivo (reviewed in Wolfe, 2010). While the cellular amount of acetyl phosphate is sufficient to phosphorylate RRs (Klein et al., 2007), this activation is usually regulated by the phosphatase activity of the cognate sensor kinase, at least in case of bifunctional HKs (Laub and Goulian, 2007). This mechanism of preventing stimulus-independent activation of a RR by the phosphatase activity of its cognate HK in the absence of inducing conditions has been documented for a number of bacterial 2CS with bifunctional HKs, including E. coli CpxAR (Wolfe et al., 2008), S. coelicolor VanRS (Hutchings et al., 2006) and EnvZ-OmpR from Xenorhabdus nematophilus (Park and Forst, 2006).
Overall, our results on the phosphorylation of LiaR by acetyl phosphate are in good agreement with the conclusions derived from the above studies. But in contrast to most of these examples, we could clearly demonstrate a full activation of LiaR by acetyl phosphate even in the presence of LiaS, presumably as soon as the LiaS2:LiaR ratio favours LiaR. But recently, the in vivo sensitivity of a RR for cellular acetyl phosphate even in the presence of the cognate HK has also been demonstrated for the CpxAR system of E. coli. In this report, the authors argue that the response of the RR CpxR to intracellular acetyl phosphate might play a role in fine-tuning the envelope stress response of this 2CS (Lima et al., 2012).
The question if the in vivo activation of LiaR by acetyl phosphate is physiological relevant or just an experimental artefact from overproducing LiaR is difficult to answer at the moment. But there is accumulating evidence that this small phospho-donor can indeed serve as an important input for 2CSs (Wolfe, 2010). Remarkably, most of the 2CSs that respond to acetyl phosphate in vivo, such as E. coli CpxAR, or VanRS from S. coelicolor are involved in some aspect of cell envelope stress response (Hutchings et al., 2006; Wolfe et al., 2008). Moreover, a recent study in S. pneumoniae identified three 2CSs that respond to the cellular pool of acetyl phosphate during normal growth, including LiaRS, VicRK and CiaRH, all of which are associated with maintaining cell envelope integrity (Ramos-Montanez et al., 2010). This observation strongly supports the findings reported in our and the above-cited study on CpxAR and might argue for a role of acetyl phosphate for the functionality of LiaRS-like or even envelope stress response 2CSs in general. Indeed, it is an appealing hypothesis to postulate that for these 2CSs the primary extracellular trigger, envelope stress, which is detected (directly or indirectly) by their HKs, is integrated with an intracellular modulator – acetyl phosphate as a measure of the energy state of the cell – that feeds into the 2CS at the level of the RRs to fine-tune the output.
LiaF and inhibition of LiaRS-dependent signal transduction
The Lia system is atypical in that it requires the activity of a third protein for its functionality. LiaF has previously been identified as a specific inhibitor of the LiaRS 2CS (Jordan et al., 2006). This function was recently verified for the orthologous system in S. mutans (Suntharalingam et al., 2009). The present study not only confirms the initial observation for a markerless liaF deletion, but also indicates that LiaF exerts its inhibitory function through LiaS. It does not affect the LiaS-independent phosphorylation of LiaR by acetyl phosphate (Fig. 3). Moreover, the inhibitory effect of LiaF can be overcome by increasing the amounts of LiaS, even in the presence of LiaF (Fig. 5). Based on these observations and the bifunctionality of LiaS, we propose that LiaF arrests LiaS in the phosphatase mode in the absence of a suitable trigger. Hence, the default setting of LiaS alone – even in the absence of a signal – is ‘kinase ON’, whereas in the presence of LiaF, it is switched to ‘phosphatase ON’ (Fig. 9). This hypothesis is supported by the observation that increasing the amount of LiaS relative to LiaF gradually turns on PliaI activity even in the absence of a signal (Fig. 5A and B). As long as LiaF is in excess, only the presence of envelope stress releases this inhibition, thereby switching LiaS into its kinase mode, resulting in the phosphorylation of LiaR and hence a strongly increased PliaI activity. A further increase of the inhibitor does not significantly affect the functionality of the LiaFSR system (Fig. 5A), demonstrating that a signal can overcome the inhibition of LiaS by LiaF irrespective of the amount of excess in which LiaF is present relative to LiaS. The importance of maintaining such a ratio could again be interpreted as an indication for a physical interaction – for example the formation of a sensory/regulatory complex – between LiaF and LiaS. While so far we were unable to demonstrate this, future studies will hopefully shed some light on the mechanism of interaction between the two proteins.
If we take all observations of this study together, the following scenario can account for the observed behaviour (Fig. 9). Under normal non-inducing conditions, the excess of LiaF over LiaS keeps the HK quantitatively in its phosphatase mode. As long as LiaS is also in excess over LiaR, it can prevent the phosphorylation of LiaR by acetyl phosphate, thereby keeping the system completely switched off. If under these non-inducing conditions LiaR is overexpressed, it then stoichiometrically outcompetes its phosphatase, LiaS, and hence becomes phosphorylated by acetyl phosphate.
On the other hand, if LiaS is overexpressed, it is then in excess over LiaF, which therefore fails to completely keep the HK in its phosphatase state. If there is only a mild overexpression (i.e. in the wild type under its own weak SD sequence), it only increases the basal level of the LiaR-dependent gene expression (Fig. 5A). Nevertheless, most of LiaS is still kept in the phosphatase state by LiaF. Therefore, the system is still inducible by bacitracin (Fig. 5A). A strong overexpression of LiaS (i.e. with an optimized SD sequence) results in a higher amount of LiaS in the kinase state and hence a full activation of the LiaR response even in the absence of an inducer (Fig. 5B and C).
What is most puzzling about this model is that bifunctional HKs are very often present in the cells in much lower amounts than the cognate RRs. Nevertheless, they are usually very well capable of dephosphorylating their RR in the absence of a stimulus. This argues either for a very inefficient phosphatase activity of LiaS, and/or for a high affinity of LiaR for acetyl phosphate, again supporting the idea that the energy state of the cell – as reflected by the intracellular pool of acetyl phosphate – might indeed be an important secondary input into LiaR-dependent gene expression, as discussed above. Taken together, the combination of the unusual stoichiometry of the Lia system, the requirement for an additional component in order to keep LiaS in its phosphatase state, combined with a high affinity of LiaR for acetyl phosphate seem to collectively argue for a physiological necessity of this particular and unusual design of the LiaFSR system for its proper functionality. But these intriguing possibilities are purely speculative at the moment and will require subsequent investigations.
The exact mechanism by which LiaF affects LiaS activity is unknown. But one appealing hypothesis is that LiaF could function as a stimulus-perceiving anti-kinase that keeps LiaS inactive in the absence of a trigger, presumably through direct protein–protein interaction. Upon sensing a signal, LiaF releases the HK, which then acts as a LiaR-specific kinase. Alternatively, LiaF together with LiaS could form the stimulus perception complex of the LiaFSR system. In this complex, LiaF would again keep LiaS in its phosphatase state in the absence of a trigger. Upon addition of bacitracin, the LiaFS complex would perceive the resulting envelope stress and again LiaS would switch to its default kinase-ON mode, thereby activating LiaR. Both possibilities would be in good agreement with the data obtained.
While the important role of the stoichiometry could well be a specialty of LiaFSR-like systems, due to the presence of a third inhibitory protein, our observations could also have a more general significance for 2CSs with HisKA_3-containing HKs.
A comparison of the stoichiometry and genetic arrangement in operons of EnvZ/OmpR-like and NarXQ/NarL-like 2CSs indicates a possible connection between protein ratios and operon structure. EnvZ-like HKs are usually encoded downstream of their cognate RR genes, potentially accounting for the observed stoichiometry with RR exceeding the cognate HK. In contrast, for NarXQ/NarL-like 2CSs the genetic order is usually inverted, which can be viewed as an indication for a molar excess of HK over RR for 2CSs. This might point towards a fundamental difference between HisKA- (EnvZ-like) and HisKA_3- (NarXQ-like) containing HKs with respect to their enzymatic (at least phosphatase) activities: A functional role of an excess of the HK over the cognate RR suggests that in such cases the phosphatase activity of the HK is weak. But more work on additional HisKA- and HisKA_3-containing HKs will be necessary in order to verify or falsify such a hypothesis.
Media and growth conditions
Bacillus subtilis and E. coli were routinely grown in LB medium or chemical defined CSE [Chemical defined Succinate (0.56% (w/v) Na-succinate) and Glutamate [0.75% (w/v) K-glutamate)] medium (Stülke et al., 1993) at 37°C with aeration. Ampicillin (100 μg ml−1) was used for selection of all plasmids in E. coli. Kanamycin (10 μg ml−1), chloramphenicol (5 μg ml−1), spectinomycin (100 μg ml−1), erythromycin (1 μg ml−1) plus lincomycin (25 μg ml−1) for macrolide–lincosamide–streptogramin (MLS) resistance, and tetracycline (10 μg ml−1) were used for the selection of the B. subtilis mutants used in this study.
Bacterial strains and plasmids
The strains of E. coli and B. subtilis are listed in Table 2. All B. subtilis strains used in this study are derivatives of the laboratory wild type strains W168 and CU1065 (W168 attSPβ). The plasmids used in this study are listed in Table 4.
The preparation of chromosomal DNA and transformation were performed according to standard procedures (Cutting and Van der Horn, 1990). E. coli plasmid DNA and restriction enzyme fragments were isolated by using the QIAprep spin miniprep and PCR purification kits respectively (Qiagen). DNA ligase (Fermentas), HotStarTaq Plus DNA Polymerase (Qiagen), and Phusion High-Fidelity DNA Polymerase (Finnzymes) were used according to manufacturer's instructions. All primers used for PCR are listed in Table S1.
Site-directed mutagenesis of liaR and liaS
To generate an amino acid exchange of the conserved aspartate of LiaR to alanine, we introduced a point mutation in liaR via the Combined Chain Reaction (CCR). This method was performed as described previously (Bi and Stambrook, 1997). In brief, liaR was amplified from chromosomal DNA of the wild type, using primers #0047 and #0048 flank the overall sequence of liaR while one internal mutagenesis primer (#0508) are positioned at the mutation site of interest (Table S1). The mutagenesis primer carries the point mutation where it mismatched with template as well as a phosphorylated 5′ end. During the PCR process, the extended specific forward primer was ligated with the mutagenesis primer by a thermostable DNA ligase (Ampligase) to create a liaR fragment with the expected point mutation.
To investigate the phosphatase activity of LiaS, an amino acid exchange of the glutamine residue within the conserved DxxxQ motif was generated via in vitro site-directed mutagenesis. The plasmid pKS727 (pXT-liaS) was used as template in a PCR reaction together with mutagenesis primers #2374 and #2375 that carry the desired point mutation. These primers are complementary to opposite strands of the plasmid. The extension of the primers results in a mutated plasmid. The PCR product was digested with DpnI to remove the methylated parental DNA template. The mutated plasmid was then transformed into E. coli XL1 blue competent cells.
Allelic replacement mutagenesis of liaS, pta and ackA using LFH-PCR
The Long Flanking Homology PCR (LFH-PCR) technique is derived from a published procedure (Wach, 1996) and was performed as described previously (Mascher et al., 2003). The constructed strains are listed in Table 2, and the corresponding primers are listed in Table S1.
Construction of translational B. subtilis Pveg–lacZ and PliaG-opt.–lacZ fusions
To investigate the translation initiation of liaF, liaS and liaR, ectopic integrations of Pveg-SDliaF/liaS/liaR–lacZ fusions were constructed based on the vector pAC7 (Table 4). For this purpose, one forward primer, which contains the Pveg sequence (#0856), and three reverse primer (#0857, #0898, #0899) were designed, which carry the Shine–Dalgarno sequences of liaF, liaS or liaR up to each corresponding start codon respectively (Table S1). Each reverse primer harbours 25 nucleotides at the 3′ end that is inverse and complementary to the 3′ end of the Pveg-forward primer, so that they can be fused by joining PCR. The resulting fragments were cloned into pAC7 via SmaI and BamHI, generating pKS1001–pKS1003 (Table 4). After B. subtilis transformation, the plasmids integrated into the amyE locus by double crossing-over, resulting in a stable integration of Pveg–lacZ fusions.
To investigate the expression levels of liaF, liaS and liaR, ectopic integrations of PliaG-opt.-SDliaF/liaS/liaR–lacZ fusions were constructed in a comparable fashion, based on the vector pAC5 (Table 4). Three fragments including liaG up to the start codon of liaF, liaS or liaR, respectively, were amplified from wild type chromosomal DNA, using the forward primer #0579, which introduces the optimal liaG promoter sequence, as well as the reverse primer #0580, #0581 or #0582 (Table S1). The resulting fragments were cloned into pAC5 via SmaI and BamHI, generating pER503–pER505 (Table 4). After B. subtilis transformation, the plasmids integrated into the amyE locus by double crossing-over, resulting in a stable integration of PliaG-opt.–lacZ fusions.
Measurement of promoter activity by β-galactosidase assay
Cells were inoculated from fresh overnight cultures and grown in LB medium or CSE medium at 37°C with aeration until they reached an optical density at 600 nm (OD600) of ≈ 0.4. The culture was split, adding bacitracin (50 μg ml−1 final concentration) to one-half (induced sample) and leaving the other half untreated (uninduced control). After incubation for an additional 30 min at 37°C with aeration, 2 ml of each culture was harvested and the cell pellets were frozen and kept at −20°C. The pellets were resuspended in 1 ml of working buffer and assayed for β-galactosidase activity as described elsewhere, with normalization to cell density (Miller, 1972).
Preparation of total RNA for quantitative real-time RT-PCR and Northern blotting
Total RNA was extracted from 4 ml of culture with and without bacitracin (50 μg ml−1 final concentration). Bacitracin was added to the culture at an OD600 of 0.5 (mid-exponential phase), and the cultures were incubated for 30 min at 37°C with aeration before the cells were harvested and rapidly frozen at −70°C. RNA was prepared using the RNeasy kit (Qiagen) according to the manufacturer's protocol. The RNA was treated with Baseline-ZERO DNase (EPICENTRE) to remove remaining traces of chromosomal DNA that would interfere with the subsequent reaction. The success of this treatment was verified by a lack of product in a standard PCR, using the same primers as for the real-time reverse transcription-PCR (RT-PCR).
Quantitative real-time RT-PCR
Measurement of transcript abundance was performed by quantitative real-time RT-PCR, using the iScript One-Step RT-PCR Kit (Bio-Rad) according to the manufacturer's procedure, with minor modifications. In brief, 100 ng of DNA-free total RNA was used in a total reaction volume of 20 μl with 0.3 μM of each primer (Table S1). The amplification reaction was carried out in an iCycler (Bio-Rad). Expression of rpsE and rpsJ, encoding ribosomal proteins, was monitored as a constitutive reference. Expression of liaF, liaS or liaR of the uninduced wild type was calculated as the fold change based on the CT values for each gene, as described previously (Talaat et al., 2002).
Probe preparation and Northern blot analysis
Internal fragments of liaIH and liaR (∼ 500-nucleotide length) were amplified by PCR using the primer pairs listed in Table S1. The PCR fragments were purified by using the PCR purification kit (Qiagen), and 1 μg of each fragment was labelled with digoxygenin (DIG) by in vitro transcription using the DIG RNA labelling mix (Roche) and the T7-RNA polymerase (Roche) according to manufacturer's protocol.
For Northern blot analysis, 5 μg or 10 μg of total RNA were denatured and loaded on a formaldehyde agarose gel. After electrophoresis, the RNA was transferred to a nylon membrane (Roche) in a downward transfer using 20× SSC (3 M NaCl, 0.3 M sodium citrate, pH 7.0) as transfer buffer. The RNA was cross-linked by exposing the damp membrane to UV light. The blot was pre-hybridized at 68°C for 1 h with pre-hybridization solution [0.2% (w/v) SDS, 0.1% (w/v) N-lauroylsarcosinate, 5× SSC, 50% (v/v) formamide, 2% (w/v) blocking reagent] and labelled probe was added to the hybridization tube. Hybridization was performed overnight at 68°C. The next day, the membrane was washed twice with low-stringency buffer [2 × SSC, 0.1% (w/v) SDS] at room temperature for 5 min, followed by two high-stringency washes [0.1× SSC, 0.1% (w/v) SDS] at 68°C for 15 min. For the detection of labelled probe, the DIG Nucleic Acid Detection Kit (Roche) was used. Therefore, the blot was removed from the hybridization tube and placed in a box with 1× buffer 1 [10× buffer 1 is 1 M maleinic acid, 1.5 M NaCl, 0.3% (v/v) tween20, pH 7.5] for 5 min at room temperature. The membrane was pre-incubated with buffer 2 [10% (v/v) 10× buffer 1, 1% (w/v) blocking reagent] for 30 min, treated with the antibody against DIG conjugated with alkaline phosphatase (AP) (Roche) for 30 min, and washed three times with 1× buffer 1 for 10 min. The blot was wrapped in plastic wrap, treated with the AP substrate CDP-Star (Roche) at a dilution of 1:200, and analysed using a LumiImager (PeqLab).
Preparation of B. subtilis cell fractions for Western blotting
The methodology was based on a published procedure (Heinrich et al., 2008) with the following modifications: B. subtilis strains were grown in LB medium and 50 ml of cells with an OD600 of 0.5–0.8 were harvested by centrifugation. The cells were washed and resuspended in 1 ml of cold disruption buffer (50 mM Tris-HCl, 100 mM NaCl, pH 7.5). Samples were sonicated (Cell disrupter UP 200 s, Dr Hielscher, Stuttgart) on ice and an aliquot of 100 μl was removed (whole cell fraction, W). Cell debris of the remaining 900 μl were removed by centrifugation at 5.000 g for 15 min at 4°C. The supernatant (800 μl) was ultracentrifuged at 70.000 g for 1 h at 4°C. The supernatant was removed (soluble protein fraction, S) and the membrane pellet (membrane protein fraction, M) was washed in cold disruption buffer, ultracentrifuged again (70.000 g, 30 min, 4°C), dissolved in 100 μl of Laemmli buffer and heated for 5 min at 95°C. The protein content of the W and S fractions was established according to Bradford. For SDS-PAGE and Western blotting 20 μg of samples of the W fractions, 20 μl of the S fractions, and 20 μl of the M fractions were loaded to each lane.
Western blot analysis
Western blot analysis was performed by a wet-blotting procedure, using a Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad) according to manufacturer's protocol. After protein transfer, the polyvinylidene diflouride (PVDF) membrane (Bio-Rad) was incubated with blotto [1× TBS (50 mM Tris, 150 mM NaCl, pH 7.6), 2.5% (w/v) skim milk] overnight at 4°C to prevent unspecific binding. On the next day the membrane was treated with the primary antibody anti-FLAG (Sigma) or anti-LiaH [polyclonal rabbit antisera that were raised against purified His10-LiaH (Jordan et al., 2007) at SEQLAB, Göttingen] at a dilution of 1:5000 for 3 h at room temperature. Then, the membrane was washed three times with blotto following by the addition of the secondary antibody (anti-rabbit IgG, conjugated with AP, Promega) at a dilution of 1:100.000 for 30 min. After further three washes with blotto the membrane was incubated with buffer 3 (100 mM Tris, 100 mM NaCl, pH 9.5) for 5 min. The blot was wrapped in plastic wrap, treated with the AP substrate CDP-Star (Roche) at a dilution of 1:100, and analysed using a LumiImager (PeqLab).
Cloning, expression and purification of recombinant N-terminal His6- and C-terminal FLAG3-tagged LiaS and LiaR
The liaR and liaS genes were amplified from B. subtilis W168 genomic DNA using primer pairs #1530/#1164 or #0958/#0962, respectively, and subsequently fused to a FLAG3 epitope tag (amplified from pALFLAG3rsiW with primers #0960/#1161) by a second joining PCR for detection via Western blot analysis. PCR products were cloned into the pProEx1 expression vector (Life Technologies) via NdeI and HindIII or BamHI and HindIII, respectively, generating plasmids pKSEx102 (His6-LiaR-FLAG3) and pKSEx103 (His6-LiaS-FLAG3) (Table 4). For overexpression, E. coli BL21(DE3)/pLysS was transformed with pKSEx102 or pKSEx103 and grown in LB medium. In mid-exponential phase (OD600 of 0.4–0.6), protein expression was induced by the addition of 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). Cultures were harvested 3 h (His6-LiaR-FLAG3) and 16 h (His6-LiaS-FLAG3) after induction. Cell pellets were stored at −80°C until further purification.
Purification of His6-LiaR-FLAG3
The cell pellet was resuspended in 15 ml of loading buffer [20 mM Tris-HCl (pH 7.5), 300 mM NaCl, 5 mM MgCl2, 10% (v/v) glycerol, 5 mM imidazole (pH 8.0)] and cells were disrupted by sonication. The lysate was centrifuged at 20.000 g and 4°C for 1 h. The supernatant was loaded on a gravity flow column containing 1 ml of Ni2+-nitrilotriacetic acid (NTA) Superflow resin (Qiagen). After washing steps with loading buffer and loading buffer containing 50 mM imidazole, His6-LiaR-FLAG3 was eluted from the column using loading buffer with imidazole concentrations of 100 mM, 200 mM and 500 mM. All fractions were analysed by SDS-PAGE and fractions containing the most pure His6-LiaR-FLAG3 protein were collected, quantified by Bradford assay using the Roti-Nanoquant kit (Roth), and used as standard for quantitative Western blot analyses.
Purification of His6-LiaS-FLAG3
The cell pellet was resuspended in 12 ml of disruption buffer (50 mM Tris-HCl, 100 mM NaCl, pH 7.5) and cells were disrupted by sonication. The cell debris were removed by centrifugation (5.000 g, 4°C, 15 min) and the supernatant was used to prepare the membrane protein fraction by ultracentrifugation as described above. The membrane pellet was resuspended in 1 ml of disruption buffer and the protein concentration was measured via BCA assay. The solution was diluted with loading buffer to receive a final protein concentration of 5 mg ml−1. To solubilize the membrane proteins, 0.5% (w/v) n-Dodecyl-β-d-maltoside (DDM) was added and gently mixed at 4°C for 1 h. After solubilization, the protein solution was ultracentrifuged (70.000 g, 4°C, 1 h). The supernatant was loaded on a Ni2+-NTA column and His6-LiaS-FLAG3 was purified as described for His6-LiaR-FLAG3 using buffers that contain 0.02% (w/v) DDM. Purified His6-LiaS-FLAG3 protein was quantified by Bradford assay using the Roti-Nanoquant kit (Roth) and used as standard for quantitative Western blot analyses.
Chromosomal FLAG-tagging of LiaFSR
To quantify the cellular amounts of LiaFSR, we integrated the FLAG3-tag sequence directly into the W168 chromosome C-terminal of LiaS or N-terminal of LiaF and LiaR. This was done by using the pMAD shuttle vector (Arnaud et al., 2004). The regions about 600 bp upstream and downstream of the position of FLAG integration were amplified using primers listed in Table S1, thereby introducing a 66 bp extension containing the whole FLAG3 sequence to the 3′ end of the up-fragment and a 25 bp extension to the 5′ end of the down-fragment which is complementary to the 3′ end of the FLAG3-tag sequence. The two fragments were fused in a second joining PCR, and the resulting fragment was cloned into pMAD via BamHI and NcoI, generating pKS101 (FLAG3-liaF), pKS104 (liaS-FLAG3) and pKS105 (FLAG3-liaR) (Table 4). The generation of the mutants basically followed the established procedure (Arnaud et al., 2004). In brief, B. subtilis W168 was transformed with pKS101, pKS104 or pKS105, respectively, and incubated at 30°C with MLS selection on LB agar plates supplemented with X-Gal (100 μg ml−1). Blue colonies were selected and incubated 6 h at 42°C in LB medium with MLS selection, resulting in the integration of the plasmids into the chromosome. Again, blue colonies were picked from LB (X-Gal) plates and incubated for 6 h in LB medium without MLS selection. Subsequently, the liquid culture was shifted to 42°C for 3 h, and the cells were then plated on LB (X-Gal) plates, this time without selective pressure. White colonies that had lost the plasmids were picked and checked for MLS sensitivity. The resulting strains, TMB1141 (liaS-FLAG3), TMB1155 (FLAG3-liaF) and TMB1201 (FLAG3-liaR) were analysed by PCR and sequencing for the integrity of the desired genetic modifications.
Determination of cellular amounts of LiaSR by quantitative Western blot analysis
Cellular amounts of LiaS-FLAG3 or FLAG3-LiaR were determined in strain TMB1141 or TMB1201 respectively. Cells were grown in LB medium until mid-exponential phase (OD600 at 0.4–0.6). The cultures were split, adding bacitracin (20 μg ml−1 final concentration) to one-half (induced sample) and leaving the other half untreated (uninduced sample). After incubation for an additional 30 min at 37°C with aeration, 10 ml of each culture was harvested and the cell pellets were frozen and kept at −80°C. Additionally, the amount of harvested cells was analysed on agar plates. The cells were resuspended in 1.1 ml of disruption buffer (50 mM Tris-HCl, 100 mM NaCl, pH 7.5) and cells were disrupted by sonication. The cell debris were removed by centrifugation (5.000 g, 4°C, 15 min) and the supernatant was used to separate the soluble and membrane protein fractions by ultracentrifugation as described above. The soluble protein fractions were concentrated up to 50 μl using Vivaspin 500 concentrator tubes (Sartorius) and the membrane pellets were resuspended in 50 μl of Laemmli buffer. Ten microlitres of each sample was loaded onto a 12.5% SDS gel together with the purified standards using 10–200 fmol of His6-LiaS-FLAG3 and 10–100 fmol of His6-LiaR-FLAG3. The Western blot was performed by a wet-blotting procedure, using a Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad) according to manufacturer's protocol. After protein transfer, the polyvinylidene diflouride (PVDF) membrane (Macherey-Nagel) was incubated with blotto [1× TBS (50 mM Tris, 150 mM NaCl, pH 7.6), 2.5% (w/v) skim milk] overnight at 4°C to prevent unspecific binding. On the next day the membrane was treated with the primary antibody anti-FLAG (Sigma) at a dilution of 1:2000 for 1 h at room temperature. Then, the membrane was washed four times with blotto following by the addition of the secondary antibody (anti-rabbit IgG, conjugated with HRP, Promega) at a dilution of 1:2.000 for 1 h. After further four washes with blotto the membrane was incubated with 1× TBS for 5 min. The blot was wrapped in plastic wrap, treated with the HRP substrate Ace Glow (Peqlab) according to manufacturer's protocol, and analysed using a LumiImager (PeqLab). The blot was analysed by ImageJ software. The band intensities of the standard proteins were plotted against the known protein amounts and these curves are referred to as standard curves. The protein amounts of LiaS-FLAG3 and FLAG3-LiaR were calculated from the standard curves.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (DFG; grant MA2837/1-3 and MA2837/3-1) and the Fonds der Chemischen Industrie (FCI). The authors would like to thank Eva Rietkötter and Maria Braun for plasmid constructions.