Peptide Antibiotic Sensing and Detoxification Modules of Bacillus subtilis

Abstract

Peptide antibiotics are produced by a wide range of microorganisms. Most of them target the cell envelope, often by inhibiting cell wall synthesis. One of the resistance mechanisms against antimicrobial peptides is a detoxification module consisting of a two-component system and an ABC transporter. Upon the detection of such a compound, the two-component system induces the expression of the ABC transporter, which in turn removes the antibiotic from its site of action, mediating the resistance of the cell. Three such peptide antibiotic-sensing and detoxification modules are present in Bacillus subtilis. Here we show that each of these modules responds to a number of peptides and confers resistance against them. BceRS-BceAB (BceRS-AB) responds to bacitracin, plectasin, mersacidin, and actagardine. YxdJK-LM is induced by a cationic antimicrobial peptide, LL-37. The PsdRS-AB (formerly YvcPQ-RS) system responds primarily to lipid II-binding lantibiotics such as nisin and gallidermin. We characterized the psdRS-AB operon and defined the regulatory sequences within the P_psdA promoter. Mutation analysis demonstrated that P_psdA expression is fully PsdR dependent. The features of both the P_bceA and P_psdA promoters make them promising candidates as novel whole-cell biosensors that can easily be adjusted for high-throughput screening.

Peptide antibiotics are produced by a wide range of organisms and can be synthesized both ribosomally and nonribosomally (22). Nonribosomally synthesized antimicrobial compounds are produced mainly by bacteria and are often posttranslationally modified (17). They can form linear polypeptides, such as gramicidin (25), or cyclic molecules, such as bacitracin and polymyxins (29, 36). Glycopeptides (e.g., vancomycin and ramoplanin) consist of a peptide backbone, which is further modified by glycosylation and methylation (12). Ribosomally synthesized peptides, including lantibiotics and defensins, are more widespread and are produced by mammals, amphibians, insects, plants, and bacteria (17). They are often derived from small precursor peptides and are usually small (10 to 50 amino acids), with an overall positive charge and a significant number of hydrophobic residues (18).

Most peptide antibiotics target crucial steps in cell wall biosynthesis. The bacterial cell wall is a vitally important structure that gives the cell its shape, separates it from its environment, and acts as a molecular sieve (23). This makes it an important target for many antimicrobial compounds, which very often act by sequestering lipid II and by blocking transglycosylation and transpeptidation steps (47). Vancomycin, lantibiotics, ramoplanin, and many defensins bind different moieties of lipid II (23, 45, 46). Vancomycin binds to the C-terminal Lys-d-Ala-d-Ala of the pentapeptide chain of the cell wall precursor (7). Nisin and nisin-like lantibiotics bind the pyrophosphate of lipid II, whereas the binding site of mersacidin and related lantibiotics includes both the MurNAc-GlcNAc sugar moiety and the pyrophosphate (10). Ramoplanin requires the presence of MurNAc-Ala-Glu pyrophosphate in order to bind to lipid II. Bacitracin inhibits a different step of cell wall biosynthesis by binding undecaprenyl pyrophosphate and inhibiting its dephosphorylation, thereby blocking its recycling and, ultimately, cell wall biosynthesis (42).

Because the production of peptide antibiotics is widespread, the presence of an appropriate stress response system is necessary both for the producer strains as well as for those bacteria that are exposed to these compounds in their natural habitat. One type of detoxification system against peptide antibiotics found mainly in Gram-positive bacteria is a module consisting of an ABC transporter, which is genetically and functionally linked to a two-component system (TCS) (23, 24, 30). Upon sensing the signal (i.e., the presence of the antibiotic), the histidine kinase phosphorylates its cognate response regulator, which in turn induces the expression of the ABC transporter genes. The transporter facilitates the removal of the antibiotic compound from its active site (23).

While few of these systems have been experimentally characterized to date, all respond and mediate resistance to cell wall peptide antibiotics. In Staphylococcus aureus, the GraRS-VraFG system was previously found to respond to vancomycin, polymyxin B (34), gallidermin (19), and defensins (27). Homologous proteins mediate resistance to nisin in Lactococcus lactis (26) and to bacitracin in Streptococcus mutans (39, 51).

The genome of Bacillus subtilis contains three such peptide-sensing and detoxification (PSD) modules consisting of a TCS and an ABC transporter: BceRS-AB, YxdJK-LM, and YvcPQ-RS (Fig. 1 A and B). BceRS-AB (PSD1) was initially identified as a bacitracin-specific detoxification module (31, 38). Recently, it was also shown to respond to the defensin plectasin (46). The YxdJK-LM system (PSD2) responds to the human antimicrobial peptide LL-37 (40). The third system, YvcPQ-RS (PSD3), was initially described as a part of bacitracin stress response network (31).

FIG. 1.

Organization of the psdRS-AB and bceRS-AB loci and induction by peptide antibiotics. (A) Graphic representation of the psdRSAB and bceRSAB loci. Genes belonging to the psd and bce loci are shown in blue (two-component system) and green (ABC transporter); the genes flanking both operons are white. Promoters are marked with bent arrows, and putative terminators are represented by vertical bars and a circle. (B) Regulatory principle and genetic setup of the Psd and Bce biosensor strains. The response regulator (PsdR or BceR), activated by the sensor kinase (PsdS or BceS), binds to its target promoter and induces the expression of the ABC transporter encoding the psdAB or bceAB operon (detoxification) and lacZ (production of β-galactosidase). (C) Example of a qualitative β-galactosidase assay with nisin, actagardine, mersacidin, and gramicidin (disk diffusion assay). The reporter strain carrying a chromosomal P_psdA-lacZ fusion was used in soft-agar overlays on LB plates containing X-Gal. Bactericidal activity is visualized as the presence of a growth inhibition zone around the filter disk, and P_psdA-dependent induction is visualized as a blue ring around the inhibition zone. (D) Qualitative β-galactosidase assay with the subtilin producer strain B. subtilis ATCC 6633. P_psdA-lacZ reporter strain TMB299 was streaked out onto LB-X-Gal plates directly next to the B. subtilis ATCC 6633 cultures. The appearance of a blue zone in the reporter strain next to the subtilin-producing strain shows the induction of P_psdA.

In this study, we aimed to identify novel inducers for all three PSD modules. Using disk diffusion assays and promoter-lacZ fusions, we screened a wide variety of cell envelope-active compounds, including many peptide antibiotics. In addition, we performed a comprehensive meta-analysis of all previously published stress response microarray data sets in order to identify additional inducers of bceAB, yvcRS, and yxdLM expression.

We present evidence that the BceRS-AB system is not only a bacitracin-specific resistance determinant but rather a PSD module that responds to a broader spectrum of compounds, including the lantibiotics mersacidin and actagardine as well as the defensin plectasin. This module also mediates a certain level of resistance to these compounds.

For PSD3, it was recently shown that the weak induction of the yvcR promoter by bacitracin is the result of a cross-activation of the response regulator YvcP by the histidine kinase of the paralogous BceRS-AB system (42). In this study, we identified lipid II-binding peptides, mainly lantibiotics but also one lipopeptide, enduracidin, as inducers of yvcRS expression. We further demonstrate that the ABC transporter YvcRS confers resistance against its inducers. Based on the primary inducers and the resistance profile, we renamed the yvcPQRS locus to psdRSAB (for peptide antibiotic sensing and detoxification).

Our data demonstrate that the P_bceA- and P_psdA-based reporter strains are more sensitive and more specific biosensors for lipid II-binding peptide antibiotics than any of the established cell wall antibiotic biosensors currently available, such as the P_ypuA- and P_liaI-derived reporter strains (32, 53). We provide evidence indicating that both biosensors could easily be modified to accommodate high-throughput screens for novel antimicrobial compounds using pure compounds, culture supernatant, or even, directly, the producing strains.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

B. subtilis and Escherichia coli cells were routinely grown in Luria-Bertani (LB) medium at 37°C with agitation. For the induction of antibiotic production, B. subtilis ATCC 6633 (a subtilin-producing strain) and B. subtilis W168 (a sublancin-producing strain) cells were grown in medium A (3). All strains used in this study are listed in Table 1. Ampicillin (100 μg/ml) was used for the selection of plasmid pAC6 and its derivatives in E. coli. Kanamycin (10 μg/ml) and chloramphenicol (5 μg/ml) were used for the selection of the B. subtilis strains used in this study.

TABLE 1.

Strains, plasmids and oligonucleotides used in this study

Strain, plasmid, or nucleotide	Characteristic or descriptiona	Source or reference
Strains
E. coli DH5α	F′ endA1 hsdR17(rK− mK+) glnV44thi-1 recA1 gyrA (Nalr) relA1 Δ(lacIZYA-argF)U169 deoR[φ80 dlacΔ(lacZ)M15]	Laboratory stock
B. subtilis
W168	Wild type; trpC2	Laboratory stock
ATCC 6633	Wild type, subtilin producer	Laboratory stock
TMB035	W168 bceAB::Kan	42
TMB279	W168 amyE::[cat PbceA(−122-82)-lacZ]	42
TMB294	W168 psdAB::Kan	42
TMB299	W168 amyE::[cat PpsdA(−110-30)-lacZ]	42
TMB413	W168 amyE::[cat PpsdA(−105-30)-lacZ]	This work
TMB414	W168 amyE::[cat PpsdA(−94-30)-lacZ]	This work
TMB588	W168 amyE::[cat PyxdL(−194-57)-lacZ]	This work
TMB652	W168 amyE::[cat PpsdA(−110-30)-lacZ] psdR::Kan	This work
Plasmids
pAC6	lacZ fusion vector; integrates at amyE; chloramphenicol resistance	49
pDF602	pAC6 PpsdA(−105-30)-lacZ	This work
pDF603	pAC6 PpsdA(−94-30)-lacZ	This work
pER605	pAC6 PpsdA(−110-30)-lacZ	42
pPH601	pAC6 PyxdL(−194-57)-lacZ	This work
Oligonucleotidesb
PpsdA+30	AGTCGGATCCCGATAGGTTCGTTGTTTGCAACACG
PpsdA−105	AGTCGAATTCGTGAATGTGACAGCATTGTAAG
PpsdA−94	AGTCGAATTCACAGCATTGTAAGATTGGG
PyxdL+57	GCATGGATCCAGGACACTTGTCCTTTATAGG
PyxdL−194	GCATGAATTCTCTCCCGGTGAAGGGACATC
psdR-up-fwd	CAAAAGAAGAGCTATGGCG
psdR-up-rev	CCTATCACCTCAAATGGTTCGCTGCAAGCAAAATCCGATACACG
psdR-do-fwd	CGAGCGCCTACGAGGAATTTGTATCGCGGAAGGATGAAGCGGAATG
psdR-do-rev	GAAAACACGATGGTCATCAC
Kan-fwd	CAGCGAACCATTTGAGGTGATAGG
Kan-rev	CGATACAAATTCCTCGTAGGCGCTCGG
Kan-check-fwd	CATCCGCAACTGTCCATACTCTG
Kan-check-rev	CTGCCTCCTCATCCTCTTCATCC

^aThe positions of the cloned fragments are given relative to the “A” of the start codon of the corresponding gene.

^bSequences are given in the 5′→3′ direction. Restriction sites are underlined. Sequences highlighted in boldface type are inverse and complementary to the 5′ (up-reverse) and 3′ (do-forward) ends of the kanamycin cassette, respectively.

Construction of transcriptional promoter-lacZ fusions.

All strains, plasmids, and oligonucleotides used in this study are listed in Table 1. Ectopic integrations of P_psdA-lacZ and P_yxdL-lacZ fusions were constructed based on vector pAC6 (49). Promoter fragments of increasing lengths were generated by PCR. Standard cloning techniques were applied (44). The inserts were verified by DNA sequencing. The resulting pAC6-derived plasmids (Table 1) were linearized with ScaI and used to transform B. subtilis with chloramphenicol selection.

Promoter induction assays.

Screening for the induction of P_psdA, P_bceA, and P_yxdL was done by disk diffusion assays essentially as described previously (9). Briefly, the assays were carried out using soft-agar overlays of the reporter strains on LB plates containing 40 μg/ml X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside). Filter paper disks carrying 5 μl of stock solution (antibiotics normally at a concentration of 100 mg/ml; Pep5 and mersacidin at 1 mg/ml, actagardine at 50 mg/ml, duramycin at 10 mg/ml, and subtilin and sublancin as 5 μl of culture supernatants of B. subtilis ATCC 6633 and B. subtilis W168, respectively, grown overnight) were placed on top of the agar. The plates were incubated at 37°C. After incubation for 24 h, the plates were scored for the appearance of blue rings at or near the edges of the zones of growth inhibition produced by the diffusion of the antibiotics from the filter disks.

For quantitative measurements of β-galactosidase activity, cells were grown in LB medium at 37°C with agitation until they reached an optical density at 600 nm (OD₆₀₀) of ∼0.45. The culture was split, and an inducing substance (at a sublethal concentration) (Table 2) was added to one half, leaving the other half untreated (uninduced control). Both cultures were incubated for 30 min at 37°C. Cell pellets were resuspended in 1 ml of working buffer (20 mM β-mercaptoethanol, 60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄ [pH 7.0]) and assayed for β-galactosidase activity as described previously, with normalization to cell density (35).

TABLE 2.

Inducers of P_psdA, P_bceA, P_yxdL, and P_liaI expression^a

Antibiotic	Class	Chargeb	Lipid II bindingb,c	Pore formingb,c	Concnd	Inductione (reference)
						PbceA	PpsdA	PyxdL	PliaI
Bacitracin	Cyclic peptide	+2	−	−	50 μg/ml	+++ (42)	+ (42)	−	+++ (31)
Enduracidin	Cyclic lipopeptide	+4	+	−	0.025 μg/ml	−	+++ (43)	−	+++ (43)
Ramoplanin	Cyclic lipoglycopeptide	+2	+	−	5 μg/ml	−	−	−	+++ (32)
Vancomycin	Glycopeptide	+2	+	−	2 μg/ml	−	−	−	+++ (9)
LL-37	Cathelicidin	+6	−	+	ND	−	−	++ (40)	−
Plectasin	Defensin	+2	+	−	ND	+++ (46)	−	−	−
Actagardine	Lantibiotic	−1	+	−	3−10 μg/ml	+++	+++	−	−
Duramycin	Lantibiotic	+1	−	−	ND	−	−	−	+++
Gallidermin	Lantibiotic	+2	+	−	100 μg/ml	−	+++	−	+++ (8)
Mersacidin	Lantibiotic	−1	+	−	10 μg/ml	+++	−	−	−
Nisin	Lantibiotic	+5	+	+	2 μg/ml	−	+++	−	+++ (32)
Subtilin	Lantibiotic	+2	+	+	0.25%f	−	+++	−	+++ (8)

^aAll inducers are peptide antibiotics. Based on the meta-analysis of transcriptome data sets and our disk diffusion screen, the following compounds do not induce P_psdA, P_bceA, and P_yxdL (9, 21, 31, 37, 40, 56): cell wall biosynthesis inhibitors (amoxicillin, cefalexin, cephalosporin, cefotaxime, cefoxitin, daptomycin, d-cycloserine, friulimicin, oxacillin, penicillin G, Pep5, PG-1, phosphomycin, ristocetin, sublancin 168, and tunicamycin), membrane-active compounds and ionophors (gramicidin A, monensin, nigericin, nitrofurantoin, polymyxin B, poly-l-lysine, and Triton X-114), compounds interfering with DNA topology (ciprofloxacin, coumermycin, moxifloxacin, nalidixic acid, norfloxacin, and novobiocin), fatty acid biosynthesis inhibitors (triclosan and cerulenin), folate biosynthesis inhibitors (dapsone, sulfacetamide, sulfamethizole, and trimethoprim), inhibitors of protein biosynthesis (chloramphenicol, clarithromycin, clindamycin, erythromycin, fusidic acid, neomycin, puromycin, spectinomycin, and tetracycline), metal ions [Ag(I), Cd(II), Cu(II), Ni(II), Zn(II), and As(V)], and miscellaneous compounds (actinonin, azaserine, doxorubicin, ethidium bromide, hexachlorophene, and rifampin).

^bbased on the BACTIBASE database (http://bactibase.pfba-lab-tun.org/) (16).

^c+, yes; −, no.

^dConcentration resulting in the highest level of induction in quantitative β-galactosidase assays. ND, not determined.

^e−, no induction; +, weak inducer; +++, strong inducer.

^fPercent B. subtilis ATCC 6633 supernatant.

Determination of growth inhibition and the MIC.

For concentration-dependent induction and killing experiments, cells were grown in LB medium to the mid-log growth phase (OD₆₀₀ of ∼0.45), and antibiotics were added to the cultures as indicated. An uninduced culture was used as a negative control. The cultures were incubated with agitation at 37°C. A sample was taken after 30 min for β-galactosidase assays (see above), and the turbidity of the remaining culture was measured for at least 4 h to monitor the concentration-dependent effects of the antibiotics on cell growth.

MIC assays were performed in Mueller-Hinton medium. Strains W168 (wild type), TMB035 (bceAB::Kan), and TMB294 (psdAB::Kan) were inoculated to an OD₆₀₀ of 0.05, and different concentrations of antibiotics were added to the medium. Cultures were incubated with agitation at 37°C for 6 h before the determination of cell density. The MIC was defined as the lowest concentration of antibiotic that fully inhibited growth.

Allelic replacement mutagenesis using LFH-PCR.

The long-flanking homology PCR (LFH-PCR) technique is derived from a previously reported procedure (55) and was performed as described previously (31). Briefly, a kanamycin resistance cassette was amplified by PCR using vector pDG780 (15) as a template. Two primer pairs were designed to amplify ∼1,000-bp DNA fragments flanking the region to be deleted at its 5′ and 3′ ends. The resulting fragments are here called the “up” and “do” fragments. The 3′ end of the up fragment as well as the 5′ end of the do fragment extended into the gene to be deleted in a way that all expression signals of genes up- and downstream of the targeted genes remained intact. Extensions of ∼25 nucleotides were added to the 5′ ends of the up-reverse and the do-forward primers that were complementary (opposite strand and inverted sequence) to the 5′ and 3′ ends of the amplified resistance cassette. A total of 100 to 150 ng of the up and do fragments and 250 to 300 ng of the kanamycin cassette were used together with the specific up-forward and do-reverse primers at standard concentrations in a second PCR. In this reaction the three fragments were joined by the 25-nucleotide overlapping complementary ends and simultaneously amplified by normal primer annealing. The PCR products were used directly to transform B. subtilis W168. Transformants were screened by colony PCR using the up-forward primer with a reverse check primer annealing inside the resistance cassette (Table 1). The integrity of the regions flanking the integrated resistance cassettes was verified by sequencing of PCR products of ∼1,000 bp amplified from the chromosomal DNA of the resulting mutants.

RESULTS AND DISCUSSION

Screen for inducers of bceAB, yxdLM, and psdAB expression.

In order to identify specific inducers of the three detoxification modules of B. subtilis, we used reporter strains (Table 1) carrying a chromosomal pAC6-based transcriptional lacZ fusion integrated at the amyE locus in the presence of an intact TCS-ABC system (Fig. 1B). Disk diffusion assays were used to screen a variety of peptide antibiotics and other cell-envelope-active compounds for their ability to induce β-galactosidase expression. In this assay, filter plates were placed onto soft-agar overlays on LB plates containing X-Gal. The bactericidal activity of a given antibiotic was visualized as the presence of a growth inhibition zone around the filter disk, and the promoter induction was visualized as a blue ring around the inhibition zone.

To gain a more comprehensive understanding of the inducer spectra, these in vivo studies were complemented with an in silico meta-analysis of a large panel of genome-wide expression profiles of B. subtilis after treatment with inhibitory compounds of different modes of action. This analysis included membrane-active compounds; antibiotics targeting fatty acid biosynthesis, folate biosynthesis, cell wall biosynthesis, translation, DNA topology, and glycosylation; cationic antimicrobial peptides; and metal ions (9, 21, 31, 37, 40, 56) (see footnotes to Table 2 for details). Of all the compounds tested, only the peptide antibiotics listed in Table 2 acted as inducers of the three PSD modules. For comparison, the results obtained for an already established peptide antibiotic biosensor, based on P_liaI (32), are also listed.

PSD1.

The BceRS-AB system was initially identified as a part of the bacitracin stress response network and is an important bacitracin resistance determinant in B. subtilis (31, 38). Its expression was recently reported to be upregulated after treatment with a fungal defensin, plectasin (46). Using disk diffusion assays, we found that the expression of P_bceA is also strongly induced by two lipid II-binding lantibiotics, actagardine (formerly gardimycin) and mersacidin. We did not observe any induction after treatment with other lantibiotics, including nisin, sublancin, and duramycin (Table 2). We also did not identify any other compound from the microarray meta-analysis that induced the expression of P_bceA.

Bacitracin is a cyclic dodecylpeptide (Fig. 2) that binds undecaprenyl pyrophosphate and inhibits cell wall biosynthesis by preventing the recycling of this lipid carrier (48). The remaining three inducing compounds, plectasin, mersacidin, and actagardine, also inhibit cell wall biosynthesis but by binding lipid II (5, 46). Actagardine and the closely related mersacidin belong to the class of lantibiotics with compact globular structures (57) (Fig. 2). Both compounds share a conserved structure that is predicted to form the lipid II-binding pocket (50) and bind the MurNAc-GlcNAc pyrophosphate (10). Plectasin is also thought to bind the pyrophosphate moiety of lipid II (46). Surprisingly, ramoplanin, which has been predicted to possess a backbone fold similar to that of actagardine and mersacidin and, therefore, a similar mechanism of action (10), does not induce the expression of the bceAB operon (Table 2). Therefore, the exact nature of the signal sensed by the PSD1 module BceRS-AB remains to be elucidated.

FIG. 2.

Schematic structures of peptide antibiotics inducing the Psd and Bce systems. Amino acids are represented by labeled gray circles. Charged amino acids are highlighted black (positive charge) or white (negative charge). Abu, aminobutyric acid; Chp, 3-chloro-4-hydroxyphenylglycine; Cit, citrulline; Dha, didehydroalanine; Dhb, 2,3-didehydrobutyrine; Dpg, 3,5-dichloro-4-hydroxyphenylglycine; End, enduracididine; HAsn, β-hydroxyasparagine; Hpg, hydroxyphenylglycine; Man, mannose; Orn, ornithine; aThr, allo-threonine. The induction of the Bce/Psd system is indicated by B+/P+.

PSD2.

For the YxdJK-LM module, we did not identify any novel inducers of P_yxdL (Table 2). The yxdL promoter was previously described to respond to the cationic antimicrobial peptide LL-37 (40). This cathelicidin is produced by human neutrophiles and shows antimicrobial activity against both Gram-positive and Gram-negative bacteria (52). LL-37 was shown previously to induce the expression of a homologous system in S. aureus (27). As it is unlikely that a soil bacterium like B. subtilis responds specifically to a human neutrophil peptide, we suggest that the actual physiological inducer shares some chemical properties with LL-37 but remains to be identified.

PSD3.

The PsdRS-AB system was initially described to respond to bacitracin (31, 38). We have recently shown that the weak bacitracin induction of the system is a result of the cross-activation of PsdR by the paralogous BceRS-AB system (42). Using disk diffusion assays, we found that P_psdA is induced by a lipid II-binding lipopeptide, enduracidin, and the lipid II-binding lantibiotics nisin, subtilin, actagardine, and gallidermin. No induction was observed for other lantibiotics, including mersacidin, Pep5, sublancin, and duramycin, as well as for lipid II-binding antibiotics of other classes, including vancomycin and ramoplanin (Fig. 1C and Table 2).

The majority of inducers of P_psdA expression are cationic lipid II-binding lantibiotics, namely, nisin, subtilin, and gallidermin (Fig. 2). They all share an N-terminal lipid II-binding motif, a so-called pyrophosphate cage (57). Actagardine, another inducing lantibiotic, differs from the above-mentioned compounds, as it lacks a positive charge (Table 2). However, it was proposed previously that lantibiotics of this family require Ca²⁺ ions to obtain full activity and that these ions improve the interaction with the bacterial membrane by conferring a positive overall charge (6). Mersacidin, a lantibiotic closely related to actagardine, does not induce psdAB expression. These two lantibiotics are highly similar (Fig. 2), but the activity spectrum of actagardine is different from that of mersacidin: while actagardine is most active against streptococci and displays low-level activity against staphylococci, mersacidin is most active against staphylococcal species (50), including methicillin-resistant S. aureus (MRSA) (28). As actagardine induces P_psdA expression whereas mersacidin does not, the PsdRS-AB system has to be able to distinguish between those two closely related compounds.

Enduracidin, a cyclic lipopeptide with a high level of similarity to ramoplanin (33) (Fig. 2), is the only inducer of P_psdA expression found in this study that is not a lantibiotic. The induction of psdAB expression by enduracidin was recently confirmed in an independent microarray study (43). The significance of this finding remains unclear, but the differential behavior of P_psdA for two pairs of very closely related compounds (mersacidin-actagardine and ramoplanin-enduracidin) strongly suggests that the Psd system responds to a very specific antimicrobial quality of these related compounds that goes beyond their known structural and/or functional features.

Screen with lantibiotic-producing strains.

For the initial screen of inducing antibiotics, we used disk diffusion assays with pure compounds or supernatants of the lantibiotic-producing strains (Fig. 1C). Subsequently, we wanted to test if it is also possible to use the reporter strain for the direct screening of lantibiotic producers. Therefore, we streaked out P_psdA reporter strain TMB299 on LB-X-Gal plates directly next to B. subtilis ATCC 6633, which produces subtilin, a lantibiotic that in the disk diffusion assay induced P_psdA expression. The appearance of a blue color only on the side of TMB299 adjacent to the producer strain shows that it is possible to visualize the induction not only by pure substances but also directly by producer strains (Fig. 1D). Because of their specificity and sensitivity, both the P_psdA and P_bceA reporter strains are promising candidates for the development of a whole-cell-based biosensor for the identification of novel peptide antibiotics from compound libraries and culture supernatants or even directly from antibiotic-producing colonies (Fig. 1).

P_psdA and P_bceA are induced by cell wall antibiotics in a concentration-dependent manner.

To further quantify our data, we analyzed the induction of the psdA and bceA promoters as a function of the concentration of the inducing compound. To this end, we performed quantitative β-galactosidase assays for the inducing antibiotics (Fig. 3A and 4A). In the concentration-dependent induction experiments, cultures of the P_psdA and P_bceA reporter strains were grown to the mid-log growth phase, and antibiotics were added to the cultures. After an induction for 30 min, a sample was taken for β-galactosidase assays. These assays not only confirmed all the compounds identified in the disk diffusion assay as being strong inducers (increase in induction ranging from 100-fold for P_bceA after induction with actagardine to 800-fold for P_psdA after treatment with the same lantibiotic) but also demonstrate a concentration-dependent induction of both promoters (Fig. 3A and 4A).

FIG. 3.

Concentration-dependent induction of P_psdA, lysis curves, and MICs of B. subtilis cultures treated with actagardine, enduracidin, gallidermin, nisin, or subtilin (supernatant of B. subtilis strain ATCC 6633). (A) β-Galactosidase activities, expressed as Miller units, of P_psdA-lacZ reporter strain TMB299 induced with the above-mentioned compounds. A log scale is applied on the y axis for reasons of clarity, due to the high dynamic range of β-galactosidase activities. (B) Concentration-dependent killing of TMB299. The times of antibiotic addition are indicated by arrows. The concentrations of actagardine, gallidermin, nisin (in μg/ml), enduracidin (in ng/ml), and subtilin (percent B. subtilis ATCC 6633 supernatant) that affect the growth of B. subtilis are indicated. (C) MIC assays for B. subtilis cultures treated with actagardine, enduracidin, gallidermin, nisin, and subtilin. Wild-type (•) and psdAB deletion mutant (○) strains were inoculated to an OD₆₀₀ of 0.05 in Mueller-Hinton medium with different concentrations of antibiotics. Cultures were incubated with agitation at 37°C for 6 h before the determination of the cell density (OD₆₀₀). The MIC was defined as the lowest concentration of antibiotic that fully inhibited growth.

FIG. 4.

Concentration-dependent induction of P_bceA, lysis curves, and MICs of B. subtilis cultures treated with actagardine, bacitracin, and mersacidin. (A) β-Galactosidase activities, expressed as Miller units, of reporter strain TMB279 induced by actagardine, bacitracin, or mersacidin. A log scale is applied on the y axis for reasons of clarity, due to the high dynamic range of β-galactosidase activities. (B) Concentration-dependent killing of B. subtilis. The times of antibiotic addition are indicated by arrows. The concentrations of actagardine, bacitracin, or mersacidin (all in μg/ml) that affect the growth of B. subtilis are indicated. (C) MIC assay for the wild type (•) and isogenic bceAB deletion mutant (○). See the legend of Fig. 3C for experimental details.

PSD3.

A strong induction of the P_psdA promoter was already observed for 0.5 μg/ml nisin, and it reached its maximum at 4 μg/ml. A similar picture was obtained for subtilin, where 0.125% of the B. subtilis ATCC 6633 supernatant strongly induced the P_psdA promoter, with the highest induction being observed after the addition of 0.5% supernatant. Induction by gallidermin was visible at a concentration of 10 μg/ml, reaching its maximum at 100 μg/ml. Enduracidin induced P_psdA expression at a much lower concentration (5 ng/ml), with the highest induction observed at 160 ng/ml (Fig. 3A). Actagardine was the strongest inducer of P_psdA expression, with 800-fold induction observed with 10 μg/ml.

PSD1.

The bceA promoter was induced by actagardine at 0.03 μg/ml, with the highest induction observed with 3 μg/ml (Fig. 4A). However, the induction was weaker than that observed for the psdA promoter, reaching only ∼100-fold. Bacitracin and mersacidin induced P_bceA to comparable levels, with the induction reaching its maximum at 100 μg/ml and 10 μg/ml, respectively (Fig. 4A).

At higher concentrations, a strong decrease in the β-galactosidase activity was observed for all antibiotics tested, indicating cellular damage. Therefore, we also measured the turbidity of the remaining culture of the reporter strain used for the β-galactosidase assay for at least 4 h postinduction to monitor the concentration-dependent effects of the antibiotics on cell growth. A rapid lysis was observed for the concentrations of antibiotics that led to a decrease in the β-galactosidase activity (Fig. 3B and 4B). These results demonstrate that cellular lysis interfered with the synthesis of β-galactosidase, as was observed previously for other antibiotic reporter strains (32).

The ABC transporters BceAB and PsdAB confer resistance to compounds inducing their expression.

As shown above, the psdA and bceA promoters are strongly induced by peptide antibiotics. The genes under the control of these promoters, psdAB and bceAB, respectively, encode two subunits of ABC transporters, ATP-binding protein and permease. For the PSD1 module it was previously shown that the ABC transporter BceAB mediates resistance to bacitracin (4, 42). This prompted the question of whether the same is true for the other inducing compounds and if the paralogous ABC transporter PsdAB also confers resistance to its inducers. Therefore, we determined the corresponding MIC values with Mueller-Hinton medium for the wild type and the isogenic transporter deletion mutants. As indicated by the lysis curves (Fig. 3B and 4B), the cultures often lyse rapidly after the addition of peptide antibiotics but resume growth after a couple of hours, presumably due to turnover and degradation of the compounds. For MIC determinations, we therefore measured the OD₆₀₀ after 6 h of incubation (see Materials and Methods). This time was sufficient for all antibiotics to develop their full inhibitory effect while simultaneously being short enough to prevent that growth already resumed (data not shown).

The MIC for the psdAB mutant was reduced ∼2-fold for nisin and subtilin and ∼8-fold for gallidermin compared to the wild type (Fig. 3C). These results demonstrate that the PsdAB transporter indeed confers resistance to all of the P_psdA-inducing compounds, presumably by acting as an ATP-driven peptide antibiotic-specific resistance pump.

Similar effects were observed for the bceAB mutant (Fig. 4C). The BceAB transporter confers a high level of bacitracin resistance (the MIC for the bceAB mutant is reduced ∼30 fold), as was reported previously (31, 38). It also confers a more moderate level of resistance to actagardine and mersacidin (∼2- to 4-fold). These results demonstrate that the BceAB transporter, like PsdAB, mediates resistance to a broader spectrum of peptide antibiotics. However, in both cases, the efficiency of removal varied significantly between the different compounds irrespective of the strength of induction (Fig. 3 and 4).

Identification of the minimal PsdR-dependent promoter region for the psdAB operon.

In contrast to the BceRS-AB system, the regulation of the paralogous PsdRS-AB module is poorly understood. The psdRSAB operon encodes a response regulator, a histidine kinase, and the two subunits of an ABC transporter (ATP-binding protein and a permease), respectively (Fig. 1A). Two σ^A-dependent promoters can be identified in this locus, one upstream of psdR (response regulator) and a second weak σ^A-dependent promoter upstream of psdA (ATP-binding protein). A potential transcriptional terminator downstream of psdS can be predicted albeit with a low ΔG° value (−6 kcal mol⁻¹) (24). The expression of the psdAB genes under inducing conditions was previously verified by Northern blotting (31), and the presence of a longer transcript, psdRSAB, was also detected (24, 41). These results demonstrated a constitutive basal level of expression of the whole psdRSAB operon, with a much higher level of expression of the psdAB genes under inducing conditions (Fig. 1A).

The data presented so far, together with the knowledge gained from the BceRS-AB system, suggest that the response regulator PsdR is activated by its cognate histidine kinase PsdS in the presence of lantibiotics and binds to its operator sequence in the psdA promoter region, resulting in a strong induction of psdAB expression and, therefore, lantibiotic resistance (Fig. 1B). To verify this hypothesis, we analyzed the regulatory elements upstream of the inducible psdAB operon in more detail.

A −10 consensus sequence for σ^A can be predicted upstream of the psdA gene. No clear −35 sequence can be found at the appropriate position from the −10 sequence (Fig. 5A). Such a situation is often found for promoters regulated by transcriptional activators. Response regulators that act as transcriptional activators usually bind DNA via short binding sites (inverted or direct repeats) a few nucleotides upstream of the −35 promoter element (54).

FIG. 5.

Functional analysis of the psdA promoter. (A) Intergenic sequence between psdS and psdA. All features are marked underneath the respective lines of the sequence. The end of psdS and the start codon of psdA are indicated below the sequence. The −10 P_psdA promoter fragment and the putative ribosome-binding site are denoted by −10 and RBS, respectively. The inverted repeat sequence of the putative PsdR-binding site is boxed. The 5′ ends of the fragments used for the promoter deletion analysis are labeled according to their positions relative to the psdA start codon. The minimal promoter fragment for nisin-dependent induction is underlined. (B) Graphical representation of the intergenic region and the fragments used for the promoter deletion analysis. The features of the region are represented by black boxes and labeled as described above. (C) β-Galactosidase assay for the promoter deletion analysis. Black bars indicate the uninduced control sample for each strain, and gray bars represent the sample induced with nisin.

To localize the binding site of the response regulator, we used a lacZ-based promoter deletion approach. Progressively shorter fragments of P_psdA (all ending at position +30 relative to the ATG start codon of psdA) (Fig. 5B) were used to generate pAC6-based transcriptional lacZ reporter fusions integrated at the amyE locus. Strains were grown until mid-log phase and treated with nisin (2 μg/ml). All constructs had very low lacZ expression levels in the uninduced state (Fig. 5C). In cells containing reporter fusions that included at least 105 bp of the upstream psdA promoter region, β-galactosidase activity was strongly induced by the addition of nisin to the medium, while a fragment extending to position −94 showed no induction (Fig. 5C).

Next, we demonstrated the PsdR dependence of P_psdA induction by lantibiotics in a psdR deletion mutant by repeating the β-galactosidase assays with strains carrying the P_psdA reporter fusion in the presence of an intact PsdRS-AB system (TMB299) and in the psdR deletion background (TMB652): nisin-dependent induction was completely abolished in the psdR deletion mutant (133 ± 11 and 0.19 ± 0.05 Miller units for the wild type and the psdR mutant, respectively). Therefore, the induction of the psdAB genes in the presence of lantibiotics is completely PsdR dependent.

Based on the results from the promoter deletion experiments, we identified an 8-nucleotide imperfect (two mismatches) inverted repeat with a 4-nucleotide spacing region (ATGTGACAgcatTGTAAGAT) at positions −99 to −70 (Fig. 5A) as a possible site for DNA binding by PsdR. This binding sequence bears similarity to the operator upstream of P_bceA in the paralogous BceRS-AB module and has also been predicted by a comprehensive bioinformatic analysis of response regulator-specific binding sites in low-GC Gram-positive bacteria (11). Moreover, the specific binding of PsdR to this DNA region was demonstrated previously by DNase footprint experiments (14). Taken together, these results demonstrate that PsdR binds to the inverted repeat upstream of the P_psdA promoter and activates the expression of the psdAB operon in the presence of suitable inducers.

Conclusions and outlook.

The aim of the present study was to identify inducers of the three PSD modules, BceRS-AB (PSD1), YxdJK-LM (PSD2), and PsdRS-AB (PSD3), of B. subtilis. By combining an in silico meta-analysis of available microarray data sets with disk diffusion assays in vivo, we screened a wide range of antimicrobial compounds, including a number of peptide antibiotics. Using a promoter-lacZ fusion strain, we identified lipid II-binding lantibiotics as the main group of inducers of P_psdA expression. We also identified two closely related lantibiotics, actagardine and mersacidin, as novel inducers of bceAB expression, which was previously thought to be a bacitracin-specific resistance pump (31, 38). In contrast, we were not able to identify novel inducers of P_yxdL.

The induction of psdAB expression is completely dependent on the response regulator PsdR. We characterized the minimal P_psdA promoter and identified an inverted repeat that is necessary for PsdR-dependent P_psdA induction and presumably represents the PsdR-binding site. Moreover, we demonstrated that the psdAB genes, which encode an ABC transporter, confer resistance to this group of lantibiotics. Therefore, both the PsdRS-AB and the BceRS-AB systems constitute stand-alone PSD modules.

(i) Inducer specificity.

The most puzzling result, and an important open question that remains to be answered, concerns the nature of the true stimuli sensed by both PSD modules. Both systems have a more specific inducer range of lipid II-interfering peptide antibiotics than does the P_liaI-based biosensor (Table 2). Strikingly, on the one hand, they can discriminate between highly similar compounds, such as the mersacidin-actagardine or enduracidin-ramoplanin pair (PSD1 and PSD3) (Fig. 2), while on the other hand, they are also able to respond to compounds even as different as bacitracin and plectasin (PSD1). Nevertheless, all compounds belong to the group of peptide antibiotics that seem to share a cellular target: they all require the pyrophosphate moiety of the lipid carrier of cell wall biosynthesis, undecaprenyl pyrophosphate, as a docking interface to exhibit their antimicrobial activity (7, 47). While most of the compounds identified in our screen as being inducers of PSD1 and PSD3 activity bind to lipid II, bacitracin directly binds undecaprenyl pyrophosphate, the subsequent intermediate of the lipid II cycle (48).

The second surprising result is the apparent lack of a correlation between the strength of induction and the rate of resistance conferred by the induced ABC transporters. Actagardine is the strongest inducer of the PsdRS-AB system. However, the corresponding ABC transporter does not confer any detectable resistance. On the other hand, PsdAB confers significant resistance against gallidermin, despite a 10-fold-lower induction level (Fig. 3). Similarly, mersacidin is as potent an inducer of bceAB expression as bacitracin. However, the degree of resistance is almost an order of magnitude lower for mersacidin than for bacitracin (Fig. 4).

(ii) PSD modules as novel biosensors.

Bacterial reporter strains based on antibiotic-inducible promoters are an efficient tool for detecting novel bioactive compounds. The well-defined regulatory responses of Bacillus subtilis to different types of (antibiotic) stresses and the ease of genetic manipulations make this bacterium a preferred model organism for studying the mode of action of antibiotics by transcriptomics, proteomics, and whole-cell-based biosensors (1, 2, 20, 21, 53, 56).

Two comprehensive studies identified sets of B. subtilis promoters responding to antibiotics interfering with major biosynthetic pathways (i.e., the biosynthesis of DNA, RNA, proteins, the cell wall, and fatty acids) or specific classes of antibiotics (20, 53). Each of these promoters responds to a wide range of compounds that affect the respective pathways. But some of the identified promoters either have a relatively high basal expression level or are only moderately induced (3- to 10-fold). Therefore, the noise-to-signal ratio (and, hence, robustness) of these biosensors is not always ideal. A biosensor based on the P_liaI promoter is both more specific and much more robust (32). It has a very low level of background activity and is induced 50- to 500-fold in the presence of compounds interfering with the lipid II cycle of cell wall biosynthesis, such as bacitracin, ramoplanin, and vancomycin (Table 2). This biosensor strain has recently been adapted for high-throughput screens in microtiter plate bioassays (8), again demonstrating the potential of whole-cell biosensors for large-scale screens of novel antimicrobial compounds.

Whole-cell biosensors suitable for high-throughput screening need to be compound or pathway specific, robust, and sensitive (13). Based on the data presented in this study, the P_psdA and P_bceA reporter strains fulfill these criteria. They are based on an established organism, have very low levels of intrinsic activities and are strongly (more-than-100-fold) induced by a small set of peptide (l)antibiotics that bind lipid II (Fig. 2). We have demonstrated that they can be applied to analyze purified compounds, culture supernatants, or the producing strains directly (Fig. 1, 3, and 4 and Table 2). Despite our current lack of an exact definition of the nature of the stimuli sensed by the PSD modules, our data indicate that the combined use of the reporter strains derived from PSD1 and PSD3 represents a useful addition to the pool of B. subtilis biosensors currently available. They allow the identification of different but related subsets of peptide antibiotics that bind the pyrophosphate moiety of the lipid carrier of cell wall biosynthesis. Such biosensors will be very beneficial for the screening of strain collections and compound libraries, given the great potential of peptide antibiotics as an addition/alternative to the established antibiotics currently in clinical use.