Specificity of Subtilin-Mediated Activation of Histidine Kinase SpaK

ABSTRACT

Autoinduction via two-component systems is a widespread regulatory mechanism that senses environmental and metabolic changes. Although the lantibiotics nisin and subtilin are closely related and share the same lanthionine ring structure, they autoinduce their biosynthesis in a highly specific manner. Subtilin activates only the two-component system SpaRK of Bacillus subtilis, whereas nisin activates solely the two-component system NisRK of Lactococcus lactis. To identify components that determine the specificity of subtilin autoinduction, several variants of the respective lantibiotics were analyzed for their autoinductive capacities. Here, we show that amino acid position 20 is crucial for SpaK activation, as an engineered nisin molecule with phenylalanine at position 20 (nisin N20F) was able to activate SpaK in a specific manner. In combination with the N-terminal tryptophan of subtilin (nisin I1W/N20F), SpaK autoinduction reached almost the level of subtilin-mediated autoinduction. Furthermore, the overall structure of subtilin is also important for its association with the histidine kinase. The destruction of the second lanthionine ring (subtilin C11A, ring B), as well as mutations that interfere with the flexibility of the hinge region located between lanthionine rings C and D (subtilin L21P/Q22P), abolished SpaK autoinduction. Although the C-terminal part of subtilin is needed for efficient SpaK autoinduction, the destruction of lanthionine rings D and E had no measurable impact. Based on these findings, a model for the interaction of subtilin with histidine kinase SpaK was established.

IMPORTANCE Although two-component systems are important regulatory systems that sense environmental changes, very little information on the molecular mechanism of sensing or the interaction of the sensor with its respective kinase is available. The strong specificity of linear lantibiotics such as subtilin and nisin for their respective kinases provides an excellent model system to unravel the structural needs of these lantibiotics for activating histidine kinases in a specific manner. More than that, the biosyntheses of lantibiotics are autoinduced via two-component systems. Therefore, an understanding of their interactions with histidine kinases is needed for the biosynthesis of newly engineered peptide antibiotics. Using a Bacillus subtilis-based reporter system, we were able to identify the molecular constraints that are necessary for specific SpaK activation and to provide SpaK specificity to nisin with just two point mutations.

INTRODUCTION

Lantibiotics are ribosomally synthesized and posttranslationally modified peptides that contain the name-giving unusual lanthionine ring structure (1). Besides their strong antimicrobial activity against a variety of Gram-positive organisms, lantibiotics are subdivided into four classes (2–4). Linear lantibiotics, such as nisin from Lactococcus lactis, subtilin from Bacillus subtilis ATCC 6633, entianin from B. subtilis DSM 15029, ericin S from B. subtilis A1/3, epidermin from Staphylococcus epidermidis, and gallidermin from Staphylococcus gallinarum, form a distinct group within the lantibiotics and are combined as class I (1, 5–9).

Nisin and subtilin exhibit identical lanthionine ring structures and differ by 14 amino acids (Fig. 1). Both peptides act as biosynthesis autoinducers via a two-component system (TCS), termed LanRK (10–13). The histidine kinases (LanK) are activated after the respective lantibiotic is sensed, so that the histidine kinase is autophosphorylated. Afterwards, the phosphate group is transferred to the respective response regulator, LanR, which in turn induces the transcription of the lantibiotic gene clusters. In the current context, the SpaRK TCS induces operons spaBTC (biosynthetic genes), spaIFEG (immunity genes), and the subtilin structural gene spaS (subtilin), whereas operons nisABTCIP and nisFEG are induced by the TCS NisRK (nisin). One major difference between the two systems is that NisRK is constitutively expressed and further increases with high extracellular nisin concentrations (12–14), whereas the expression of SpaRK has a growth phase-dependent expression pattern (11) and is regulated via the major transition state regulator AbrB as well as the alternative sigma factor H (10, 11, 15, 16). As described previously, high sugar concentrations reduce the dependency on the growth phase (17). Noteworthily, subtilin/entianin (see Fig. 1) activates specifically histidine kinase SpaK, while histidine kinase NisK is activated solely by nisin. No cross-activation occurs between these two systems (10, 11, 13, 18).

FIG 1.

Structures and amino acid sequences of subtilin-like lantibiotics. Differences between subtilin, nisin, and entianin are highlighted in yellow. A-S-A, meso-lanthionine; Abu-S-A, 3-methyl-lanthionine; Abu, α-aminobutyric acid; Dha, 2,3-didehydroalanine; Dhb, 2,3-didehydrobutyrine.

There is an urgent need for new antimicrobial drugs due to the considerable increase in multiresistant pathogenic bacteria, and genetically encoded peptide antibiotics are a valuable source for the development of new lantibiotics, as these peptides are easily accessible for genetic modification. Particularly for autoinducible lantibiotics, a detailed molecular understanding of the structural features that are important for autoinduction is needed.

To follow subtilin and nisin autoinduction, the promoter of the subtilin structural gene was fused with the β-galactosidase reading frame (P_spaS-lacZ) and integrated into B. subtilis 6633 at the amyE locus. Upon activation of histidine kinase SpaK, this construct allowed the quantification of peptide-mediated autoinduction by means of a β-galactosidase activity assay (18).

Although they are structurally very similar (Fig. 1), we previously showed that subtilin-like lantibiotics were able to activate solely SpaK but not NisK, whereas nisin specifically activated NisK but not SpaK. To understand the molecular features of subtilin that provide the specificity for SpaK activation, we gradually adapted subtilin step by step to nisin and tested the autoinduction capacities of the engineered subtilin variants. As previously described, the replacement of the N-terminal tryptophan (W1) of subtilin with aliphatic amino acids resulted in peptides with reduced SpaK autoinduction capacity, whereas exchanges of amino acids concerning positions 2 to 17 within the subtilin molecule had no significant impact on autoinduction (19). Surprisingly, nisin was able to quench subtilin-mediated SpaK activation to approximately 50% (20).

Considerable SpaK autoinduction was also observed with an N-terminal fragment of the subtilin-like lantibiotic entianin comprising amino acids 1 to 20 (entianin_1–20) (20). However, the autoinduction of the N-terminal fragment entianin_1–19 was 10 times lower, though both fragments contain the three N-terminal lanthionine rings (20). This indicated a key role of amino acid position 20.

RESULTS

Amino acid position 20 is important for subtilin-mediated SpaK activation.

To investigate the possible importance of phenylalanine at position 20 (F20) of subtilin, we replaced F20 of subtilin with asparagine, which is the corresponding residue of nisin at this position. Indeed, this single mutation (F20N) caused a dramatic decrease in SpaK activation (Fig. 2). Interestingly, a saturating mutation analysis of amino acid 20 showed that the replacement of F20 by any aromatic, aliphatic, or small aliphatic amino acid had no negative effect on SpaK activation. By contrast, the exchange of F20 with charged amino acids or even amino acids with a free amino group (Q and N) prevented considerable SpaK autoinduction (Fig. 2), suggesting a hydrophobic interaction site of SpaK with subtilin residue 20.

FIG 2.

Autoinduction activity of the subtilin reporter strain B6633.MB1 upon addition of subtilin variants with changes at amino acid position 20. The β-galactosidase activities were measured 60 min after induction. Error bars represent the standard deviations for samples from two separate cultures. Each measurement was carried out at least in duplicate (n = 4). CTR, control (no lantibiotic was added).

It is noteworthy that all variants at position 20 with strongly reduced autoinduction capacity could be purified after expression of the respective genes in strain B15029.TSp01 (ΔetnS), which shows that they still have some residual capacity for SpaK autoinduction. The corresponding high-performance liquid chromatography (HPLC) chromatograms are presented in the supplemental material (see Fig. S3).

Subtilin F20N competes with subtilin for SpaK activation.

The significantly reduced SpaK activation by subtilin F20N could either result from a reduced capability to bind to the kinase itself or indicate that this amino acid is important for SpaK autophosphorylation or its dephosphorylation. To gain further insights into whether F20N influences binding in a negative way, we tested the peptide for quenching. The simultaneous addition of 6 nM F20N and wild-type subtilin reduced the subtilin-mediated SpaK activation by half (2,000 Miller units) (Fig. 3). A 2-fold excess of subtilin F20N relative to the wild-type subtilin further slightly reduced the subtilin-mediated SpaK activation (Fig. 3). The observed quenching of subtilin autoinduction by subtilin F20N is very similar to the previously reported quenching of subtilin autoinduction by nisin (20).

FIG 3.

Influence of subtilin F20N on subtilin wild-type-mediated induction of P_spaS-lacZ. β-Galactosidase activities in subtilin reporter strain B6633.MB1 were measured 60 min after induction. Error bars represent the standard deviations for samples from two separate cultures. Each measurement was carried out at least in duplicate (n = 4).

Interestingly, higher concentrations of subtilin F20N (48 nM) resulted in measurable SpaK activation, up to 20% of that observed with native subtilin. However, this activation seems less specific, as this was also observed previously for the nisin fragment nisin_1–20 and gallidermin (19, 20). This explains why subtilin F20N does not quench subtilin-mediated SpaK activation 100%. The observed quenching effect indicates that subtilin F20N can still bind to SpaK but cannot activate the histidine kinase at these concentrations. Interestingly, the F20N exchange had no measurable impact on antibiotic activity, as the MIC of the F20N peptide was comparable to that of subtilin (Table 1).

TABLE 1.

Determination of MICs^a

^aDeterminations were carried out in triplicates using indicator strain Kocuria rhizophila ATCC 9341.

Nisin N20F variant provides SpaK activation.

To further elucidate the importance of amino acid 20 for SpaK activation, a nisin variant was generated in which asparagine was replaced by phenylalanine (nisin N20F). Since nisin originates from L. lactis, the expression of nisin variants was achieved in an L. lactis-based expression system (L. lactis NZ9800) using replicative plasmids carrying the modified nisA structural gene (21).

Remarkably, nisin N20F was able to activate SpaK to a large extent, reaching approximately 70% of the autoinduction capacity compared with the autoinduction level of wild-type subtilin at equimolar concentrations. In a 2- to 3-fold molar excess, nisin N20F achieved a level of SpaK activation that was comparable to that of the wild type (Fig. 4). As nisin N20F showed efficient SpaK autoinduction, we tested if nisin could also quench nisin N20F autoinduction in a manner similar to that previously described for subtilin autoinduction. This was indeed the case (Fig. 4).

FIG 4.

Autoinduction activity of the subtilin reporter strain B6633.MB1 upon addition of nisin N20F. The β-galactosidase activities were measured 60 min after induction. Error bars represent the standard deviations for samples from two separate cultures. Each measurement was carried out at least in duplicate (n = 4). CTR, control (no lantibiotic was added).

Similar to the corresponding subtilin variants (subtilin F20E and F20K), the exchange of N20 in nisin with charged amino acids did not result in measurable SpaK activation. However, aliphatic or aromatic amino acids at nisin position 20 were also able to autoinduce SpaK significantly (data not shown). These results show that the inability of nisin to activate SpaK is mainly due to the asparagine residue at position 20 that prevents SpaK activation.

As previously shown, tryptophan at the N terminus of subtilin also had some impact on SpaK autoinduction, as the replacement of tryptophan by isoleucine in subtilin (subtilin W1I) reduced SpaK activation by almost 50%. To investigate the impact of the N terminus, a single amino acid exchange (nisin I1W) and a double exchange (nisin I1W/N20F) were constructed. Interestingly, the nisin I1W variant failed to activate SpaK (Fig. 5) at the examined concentrations. This showed that the N-terminal tryptophan residue was not sufficient for SpaK activation. However, the I1W exchange in combination with N20F (nisin I1W/N20F) further improved SpaK activation to a level 80% of that of the wild-type subtilin using equimolar concentrations (Fig. 5).

FIG 5.

Autoinduction activity of the subtilin reporter strain B6633.MB1 upon addition of nisin I1W, N20F and I1W/N20F peptides. The β-galactosidase activities were measured 60 min after induction. Error bars represent the standard deviations for samples from three separate cultures. Each measurement was carried out at least in triplicate (n = 6). CTR, control (no lantibiotic was added).

These results confirmed the major importance of amino acid position 20 for SpaK activation, whereas the N terminus additionally supports activation efficiency. Noteworthily, these mutations had no impact on the antimicrobial activity of the respective peptides, which shows that amino acid 20 and the N-terminal tryptophan are only of significance with respect to autoinduction (Table 1).

Flexibility of the hinge region influences SpaK activation.

In previous studies, we showed that a truncated entianin variant (entianin_1–20) had a 10 times lower SpaK activation than the full-length entianin (20), though this peptide comprises amino acids W1 and F20. This was indicative of an additional role of the C-terminal part (amino acids 21 to 32) for SpaK activation.

For linear lantibiotics such as subtilin and nisin, the hinge region located between the lanthionine rings C and D is an important structural feature. The mode of action of these class I lantibiotics depends on the flexibility of this hinge. After binding the two N-terminal rings to the cell wall precursor lipid II, the C-terminal part of the lipid II-associated lantibiotic integrates into the cell membrane. This integration leads to the formation of stable pores, which finally lead to the breakdown of the membrane potential (22–25). It has been shown for nisin that the integration of two proline residues within the hinge region leads to a dramatic loss of antibiotic activity. It was supposed that the increased rigidity of the molecule prevents pore formation (26).

To test the importance of the hinge region for SpaK autoinduction, two proline residues were placed at amino acid positions 21 and 22, resulting in subtilin L21P/Q22P. To verify the loss of flexibility of the hinge region, pore formation by subtilin L21P/Q22P was analyzed using a 3,3′-dipropylthiadicarbocyanine iodide [DiSC₃(5)] diffusion assay (Fig. 6). In this assay, the breakdown of the membrane potential, which is induced by pore-forming antibiotics, is visualized by the fluorescence of the emergent cyanine dye DiSC₃(5).

FIG 6.

Subtilin-mediated breakdown of the membrane potential using subtilin L21P/Q22P. DiSC₃(5) diffusion assay with the control strain B168.KO (without immunity proteins). Cells were incubated with DiSC₃(5) fluorescent dye and fluorescent leakage was monitored (emission, 670 nm; excitation, 544 nm). As a control, 1 μM valinomycin (orange) was added to the cells. After 150 s, either 1.25 μg · ml⁻¹ (green) or 2.5 μg · ml⁻¹ (blue) subtilin L21P/Q22P was added and the increase of relative fluorescence units (RFU) was monitored.

Whereas the positive controls subtilin and valinomycin produced a clearly measurable increase in fluorescence, no increase was visible for subtilin L21P/Q22P, even in 8-fold molar excess relative to subtilin. This confirms that the integration of two proline residues within the hinge region prevents pore formation (Fig. 6). This was in agreement with the drastically reduced MIC values of subtilin L21P/Q22P compared with that of wild-type subtilin (Table 1). Using the rigid subtilin L21P/Q22P variant for SpaK autoinduction, concentrations approximately 15 times higher were needed to achieve SpaK activation comparable to that of the wild type (Fig. 7). This shows that a proper hinge region is needed for maximal SpaK activation.

FIG 7.

Autoinduction activity of the subtilin reporter strain B6633.MB1 upon addition of subtilin L21P/Q22P peptide. The β-galactosidase activities were measured 60 min after induction. Error bars represent the standard deviations for samples from two separate cultures. Each measurement was carried out at least in duplicate (n = 4). CTR, control (no lantibiotic was added).

Due to expected cis-trans dynamics resulting in different conformations (27), subtilin L21P/Q22P reversed-phase (RP) HPLC purification revealed an asymmetric peak eluting over 10 min (see Fig. S3F). The rechromatography of samples taken within the asymmetric peak again revealed the broad elution profile, confirming the cis-trans dynamics. To ensure the homogeneity of the purified subtilin L21P/Q22P, we followed the masses within the entire peak. Mass analysis revealed the expected mass (m/z 3272.53), and no contaminating molecules were detected. This shows that the subtilin L21P/Q22P was purified to homogeneity.

Importance of lanthionine rings for subtilin-mediated autoinduction.

Although the subtilin F20N variant abolished SpaK activation at physiological concentrations, this variant quenched subtilin-mediated SpaK activation at equimolar concentrations to 50% (Fig. 3). As previously reported (20), such a quenching effect on SpaK was also found if nisin was applied together with the SpaK autoinducer subtilin (see also Fig. 4). These findings suggested that in addition to amino acids 1 and 20, the lanthionine ring structure might also be important for SpaK activation.

Furthermore, subtilin fragment subtilin_1–20 also showed SpaK activation but to a 10-fold-reduced extent, which argues for the importance of the C-terminal part of subtilin for SpaK activation. To examine the importance of the five lanthionine rings (rings A, B, C, D, and E; see also Fig. 1) on SpaK activation, subtilin variants with destructed lanthionine rings were analyzed. Lanthionine rings are posttranslationally built from the prepeptide in 2 steps (1). First, water is eliminated from a serine or a threonine residue, resulting in dehydroalanine or dehydrobutyrine, respectively. In a second step, cysteine is covalently linked to the respective double bonds, resulting in either meso-lanthionine or 3-methyl-lanthionine. To destruct lanthionine ring formation, either serine, threonine, or cysteine was replaced by alanine. Whereas rings B, D, and E could be successfully destructed, even via coexpression using a wild-type subtilin-like lantibiotic-producing strain, no peptide production was found for the ring A and C destructions, neither by respective replacements of serine (subtilin S3A) and threonine (subtilin T13A) nor by replacements of cysteine residues (subtilin C7A and subtilin C19A), suggesting that the biosynthetic machinery could not mature the resulting subtilin S3A and C7A variants. Subtilin C11A, which prevents formation of the second lanthionine ring (ring B), could be purified upon coexpression and tested for its ability to activate SpaK (Fig. 8). No SpaK activation was observed for subtilin C11A at the examined concentrations. Even with an 8-fold excess (48 nM) amount compared with that of subtilin, only marginal SpaK activation was detected. Furthermore, no quenching effect on subtilin SpaK activation was observed, even with an 8-fold-excess amount of subtilin C11A (Fig. 8). This clearly shows that lanthionine ring B is indispensable for SpaK activation. Furthermore, ring B mutants lost most of their antibiotic activity against the indicator strain Kocuria rhizophila ATCC 9341, with a 30-fold higher MIC value of 192 nM (Table 1).

FIG 8.

Influence of the subtilin C11A variant in subtilin-mediated induction of P_spaS-lacZ. β-Galactosidase activities of P_spaS-lacZ in reporter strain B6633.MB1 after 60 min of induction. Error bars represent the standard deviations for samples from two separate cultures. Each measurement was carried out at least in duplicate (n = 4).

Surprisingly, ring D and E destructions did not diminish subtilin-mediated SpaK activation (data not shown). These results showed that none of these rings are important for SpaK activation, suggesting that the amino acids in the C-terminal part of subtilin are decisive for the SpaK activation that is 10-fold higher than that with subtilin_1–20 fragments missing rings D and E. Subtilin and nisin have a remarkable conformity at their C termini with positively charged lysine residues. To analyze if the C-terminal lysine has an input in SpaK autoinduction, the C terminus of subtilin was discharged after its replacement with valine (subtilin K32V). However, this did not interfere with the autoinduction capacity (data not shown). Other exchanges, where amino acids of subtilin were replaced by the corresponding amino acids of nisin, resulting in subtilins L21M, L24A, N27H, and S31H, also did not influence SpaK autoinduction. Although the C-terminal amino acids support SpaK activation, this is obviously not due to specific amino acids as shown here by a number of point mutations within this region and, more strikingly, by nisin I1W/N20F, where the entire C-terminal part of subtilin was exchanged.

DISCUSSION

Although there are many detailed descriptions of histidine kinases, the exact sensing mechanisms after a specific stimulus and the following transduction processes are still poorly understood. The vast majority of two-component systems sense a change in environmental conditions due to either physiological changes or the availability of certain nutrition sources. In comparison to these systems, during lantibiotic autoinduction, relatively large peptides specifically interact with the histidine kinase, which makes this system unique. Furthermore, histidine kinases SpaK and NisK are highly specific for their respective lantibiotics (10, 28). The strong specificity of subtilin for SpaK becomes even more impressive, as approximately 30 different histidine kinases could be identified in B. subtilis (29, 30). Although the primary structures of subtilin and nisin differ by 14 amino acids, both peptides have the same overall structure with 5 lanthionine rings.

So far, the classification of histidine kinases is based on the arrangement of the domains, and two major classes are distinguished (31). Most of the sensor histidine kinases (83%) are membrane-passing proteins such as BceS from B. subtilis, which is involved in bacitracin resistance (32). A recent structural analysis of certain domains of histidine kinases and response regulators provided some insights into the molecular events during signal transduction (33–35). However, three-dimensional (3D) structures of the sensor domains have been solved for only a few histidine kinases, but ligand binding and signal transduction are not fully understood (36, 37).

Understanding the structural features for SpaK and NisK activation becomes even more complex, as under physiological conditions, both kinases are probably homodimers and the respective monomers are connected via a dimerization domain (38, 39). This further complicates the understanding of the kinetics of autoinduction. Possibly, in the case of subtilin allosteric effects, the influence of the peptide binding to one monomer on the binding of a second peptide cannot be excluded.

Currently, no structural information for SpaK, NisK, or any related histidine kinase is available, and all information is based on structural predictions. For SpaK and NisK, two transmembrane domains are predicted, a periplasmic sensor/acceptor domain and a kinase/transmitter domain located in the cytoplasm (10, 28, 38, 39).

For both kinases, the two hydrophobic transmembrane domains are predicted to be within the N-terminal part of the protein, with a hydrophilic extracellular domain in between (10, 28), which is most likely the periplasmic sensor/acceptor domain. Both kinases belong to the EnvZ group of sensor protein kinases and share a protein identity of approximately 26% (40). The primary structures of the supposed sensor domains of SpaK and NisK are even more different and share only 16% identity. By initial assessments, histidine kinases SpaK and NisK seem entirely different.

Independent of their differences in primary structures, SpaK and NisK have similar arrangements of α-helices and β-sheets (see Fig. S2 in the supplemental material). Such an observation was also made for the subtilin and nisin immunity proteins SpaI and NisI, which also seemed to be entirely different based on their primary sequences but turned out to have very similar 3D structures (41). It is possible that SpaK and NisK also have similar 3D structures, and within the sensor domains, certain amino acids are responsible for the specificity. This suggests that the two histidine kinases SpaK and NisK might share mechanisms of autoinduction. The signal transductions upon autoinduction to the respective response regulators, SpaR and NisR, also provide some evidence for a common activation mechanism, as both response regulators belong to the OmpR family and share a high similarity of 42% (10, 40).

Recently, the binding of nisin to NisK has been proposed, but surface plasmon resonance studies using a NisK sensor domain isolated from Escherichia coli only showed a very weak affinity to nisin, possibly due to the fact that the NisK sensor domain was not properly folded (42). Here, we analyzed the reverse by assessing the particular features of subtilin that provide its specificity for SpaK activation. However, we cannot speculate on the kinetics of autoinduction that include homodimerization and affinity constants for the dissociation rate of the peptide from the kinase. Nevertheless, with our current knowledge, we can provide a simplified model that describes the constraints for subtilin/SpaK interaction at a first approximation.

As depicted in Fig. 9, the overall shape of linear lantibiotics and the flexibility of the hinge region, which is between the three lanthionine rings in the N-terminal part and the two lanthionine rings in the C-terminal part, contribute to SpaK activation. The importance of the C-terminal part of subtilin became evident as (i) subtilin L21P/Q22P makes the hinge region more rigid (see Fig. S1K) and (ii) N-terminal subtilin_1–20 fragments strongly diminished SpaK activation capacity compared with that from the entire subtilin.

FIG 9.

Hypothetical model of the subtilin-mediated SpaK activation. Subtilin mediates the activation of the histidine kinase SpaK, resulting in a conformational change that, in turn, leads to autophosphorylation. Based on our current data, the important structural components are highlighted in the simplified model. Adjacent to the C-terminal part, amino acid positions 1 and 20 and lanthionine ring B assume a leadership role within SpaK autoinduction. As the expression and purification of lanthionine rings A and C was not possible, one can only speculate on their role in autoinduction; therefore, the model does not account for them.

The importance of the overall lanthionine structure was shown (i) by the partial quenching effect of subtilin-mediated activation by nisin and those subtilin variants which were unable to activate SpaK at physiological concentrations and (ii) by the fact that at higher—but nontoxic—concentrations, nisin and the respective subtilin variants also produced SpaK activation to a small extent. Due to the increasing toxicity of the respective peptides, they could not be applied at higher concentrations to test their maximal SpaK activation capacity.

The importance of the lanthionine rings within the N-terminal part of subtilin is indicated by the subtilin C11A variant, where the lanthionine ring B has been destroyed. Subtilin C11A can neither autoinduce nor quench subtilin-mediated SpaK activation (Fig. 8; see also Fig. S1L), which provides strong evidence that the lanthionine ring B is important for the first association of subtilin with SpaK (Fig. 9). It is possible that rings A and C may also support SpaK activation; however, this could not be analyzed as the respective subtilin destruction variants could not be expressed.

Although the overall lanthionine structure is needed for SpaK activation, amino acid residue 20 of subtilin is decisive for the specificity of subtilin autoinduction. Phenylalanine is most effective for SpaK activation. Amino acids with aromatic and neutral side chains are also able to confer autoinduction, whereas any additional charges and additional amino groups prevented SpaK activation. As previously shown, the efficiency of SpaK activation is further enhanced by the N-terminal amino acid tryptophan (19). These findings were unambiguously proven when these two amino acids were exchanged in nisin and the SpaK activation capacity of the resulting nisin I1W/N20F reached nearly the same efficiency as that of subtilin.

As shown by the 10-fold lower autoinduction capacity of the subtilin_1–20 fragment, the C-terminal part of subtilin also supports SpaK autoinduction. However, neither rings D and E nor a particular amino acid within the C-terminal part of subtilin is needed for autoinduction efficiency. Even the entire exchange of the C-terminal parts of subtilin and nisin had nearly no impact on SpaK autoinduction, as shown by the nisin I1W/N20F variant. Therefore, we assume that the C-terminal part contributes to autoinduction in a less specific way. However, this requires flexibility at the hinge region (amino acids 20 to 22), as a rigid hinge region does not autoinduce, as shown by the subtilin L21P/Q22P variant.

In summary, our current data suggest that two different constraints support SpaK activation as shown in the hypothetical model (Fig. 9; see also Fig. S1). We hypothesize that in a first stage, ring B, possibly also supported by rings A and C, is important for the association of the peptide with the histidine kinase. Most likely at this stage, other lantibiotics with a similar ring structure can compete with subtilin binding. In a further step, amino acid position 20 supported by the N-terminal amino acid plays a key role to finally mediate SpaK activation.

MATERIALS AND METHODS

Bacterial and yeast strains.

Table 2 lists all relevant strains and plasmids used in this study. Oligonucleotides used for plasmid constructions are listed in Table 3. A detailed protocol for the construction of all strains and plasmids is available on request.

TABLE 2.

Strains and plasmids used in this study

Strain or plasmid	Genotype or descriptiona	Source or referenceb
Strain
B. subtilis
B15029	Wild type (Ent+)	DSM 15029
B6633	Wild type (Sub+)	ATCC 6633
B6633.MB1	ΔspaS amyE::PspaS-lacZ (Specr Cmr Sub−)	18
B15029.AK28	amyE::PspaS-spaS(C11A) (Specr Neor)	This work
B15029.SK1	ΔetnS amyE::PspaS-spaS(C26A) (Specr Neor)	This work
B15029.SK3	ΔetnS amyE::PspaS-spaS(C28A) (Specr Neor)	This work
B15029.TSp01	ΔetnS (Specr Ent−)	2
B15029.TSp03	ΔetnS amyE::PspaS-spaS (Specr Neor)	2
B15029.TSp04	ΔetnS amyE::PspaS-spaS(F20N) (Specr Neor)	This work
B15029.TSp61	ΔetnS amyE::PspaS-spaS(F20E) (Specr Neor)	This work
B15029.TSp62	ΔetnS amyE::PspaS-spaS(F20D) (Specr Neor)	This work
B15029.TSp63	ΔetnS amyE::PspaS-spaS(F20Y) (Specr Neor)	This work
B15029.TSp64	ΔetnS amyE::PspaS-spaS(F20W) (Specr Neor)	This work
B15029.TSp65	ΔetnS amyE::PspaS-spaS(F20R) (Specr Neor)	This work
B15029.TSp70	ΔetnS amyE::PspaS-spaS(L21P/Q22P) (Specr Neor)	This work
B15029.TSp73	ΔetnS amyE::PspaS-spaS(F20K) (Specr Neor)	This work
B15029.TSp74	ΔetnS amyE::PspaS-spaS(F20A) (Specr Neor)	This work
B15029.TSp79	ΔetnS amyE::PspaS-spaS(F20I) (Specr Neor)	This work
B15029.TSp80	ΔetnS amyE::PspaS-spaS(F20V) (Specr Neor)	This work
L. lactis
NZ9800	NZ9700 derivate ΔnisA	54
CG13	NZ9800 ΔnisA PnisA-nisA(N20F) (Cmr)	This work
CG34	NZ9800 ΔnisA PnisA-nisA(I1W/N20F) (Cmr)	This work
CG40	NZ9800 ΔnisA PnisA-nisA(I1W) (Cmr)	This work
E. coli
DH5α	recA1 endA1 gyrA96 thi hsdR17(rK− mK+) relA1 supE44 ϕ80ΔlacZΔM15 Δ(lacZYA-argF)U169	Laboratory stock
K. rhizophila ATCC 9341	Test strain for MIC determinations	ATCC 9341
S. cerevisiae
CEN.PK2	MATa/α ura3-52/ura3-52 trp1-289/trp1-289 leu2-3,112/leu2-3,112 his3Δ1/his3Δ1 MAL2-8c/MAL2-8c SUC2/SUC2	43
Plasmid
pAK15	bla amyE′ PspaS-spaS(C11A) Neor ′amyE	This work
pCG2	bla amyE′ PspaS BamHI Neor ′amyE	19
pCG8	E. coli, B. subtilis, S. cerevisiae shuttle vector carrying PnisA HIS3 (Ampr Cmr)	This work
pCG13	PnisA-nisA(N20F) (Cmr)	This work
pCG34	PnisA-nisA(I1W/N20F) (Cmr)	This work
pCG40	PnisA-nisA(I1W) (Cmr)	This work
pSK1	bla amyE′ PspaS-spaS(C26A) Neor ′amyE	This work
pSK3	bla amyE′ PspaS-spaS(C28A) Neor ′amyE	This work
pTSp103	bla amyE′ PspaS-spaS(F20D) Neor ′amyE	This work
pTSp104	bla amyE′ PspaS-spaS(F20E) Neor ′amyE	This work
pTSp105	bla amyE′ PspaS-spaS(F20Y) Neor ′amyE	This work
pTSp106	bla amyE′ PspaS-spaS(F20W) Neor ′amyE	This work
pTSp107	bla amyE′ PspaS-spaS(F20R) Neor ′amyE	This work
pTSp111	bla amyE′ PspaS-spaS(L21P/Q22P) Neor ′amyE	This work
pTSp113	bla amyE′ PspaS-spaS(F20A) Neor ′amyE	This work
pTSp114	bla amyE′ PspaS-spaS(F20K) Neor ′amyE	This work
pTSp122	bla amyE′ PspaS-spaS(F20I) Neor ′amyE	This work
pTSp123	bla amyE′ PspaS-spaS(F20V) Neor ′amyE	This work
pCI372	High-copy cloning vector, Cmr	5
pNZ8148	Vector carrying nisA promoter, Cmr	6
pAUT3	pNZ8148 PnisA-lacZ	19

^aEnt⁺, entianin producer; Sub⁺, subtilin producer; Cm^r, chloramphenicol resistant; Spec^r, spectinomycin resistant; Neo^r, neomycin resistant; MAL2-8^c, dominant mutant allele causing partially constitutive but glucose-repressible MAL2 gene expression. The amino acid exchanges and the numbers corresponding to their positions in subtilin/nisin variants are presented in parentheses.

^bDSM, German Resource Centre for Biological Material; ATCC, American Type Culture Collection.

TABLE 3.

Oligonucleotides used in this study

Oligonucleotide	Sequence (5′ to 3′)	Descriptiona
Amplification
CG-Dok-82	TGTCTAGATTATTTGCTTACGTGAATA	nisA universal oligonucleotide
CG-Dok-83	CGGAATTCTAGTCTTATAACTATAGTGA	nisA universal oligonucleotide
In vitro mutagenesis
TSp52	GCATTGCAAACTTGCAATCTTCAAACAC	Forward F20N
TSp53	GTGTTTGAAGATTGCAAGTTTGCAATGC	Reverse F20N
Gap repair
AK41	TGTGAAAGTCTCTAAACAAGACTCAAAAATCACTCCGCAATGGAAAAGTGAATCACTTTGTACACCAGGAGCAGTAACTGGTGC	Subtilin C11A
AK42	GAGATTTTGCAGTTACAAGTTAGTGTTTGAAGGAAGCAAGTTTGCAATGCACCAGTTACTGCTCCTGGTGTACAAAGTGATTCACTTTTC	Subtilin C11A
CG11	ACTTGCTTCCTTCAAACACTAACTTGTAACTGCAAAATCTCTAAATAAGTAAAACCATTAGCATCACCTTGCTCTGACTCCTTGCACT	Universal oligonucleotide
CG31	TGTGAAAGTCTCTAAACAAGACTCAAAAATCACTCCGCAATGGAAAAGTGAATCACTTTGTACACCAGGATGTGTAACTGGTGCATTGCAA	Universal oligonucleotide
CG40-Dok	GCTGTTTTCATGAAACAACCCATCAGAGCTCCTGTTTTACAACCGGGTGTACATAGCGAAATACTTGTAATGCGTGGTGATGCACCTGAATCTTTCTTCG	Nisin N20F
CG41-Dok	CAGGAGCTCTGATGGGTTGTTTCATGAAAACAGCAACTTGTCATTGTAGTATTCACGTAAGCAAATAAGTACTGCAGGCATGCGGTACCACTAG	Nisin N20F
CG-Dok-93-RV	GTTTTCATGAAACAACCCATCAGAGCTCCTGTTTTACAACCGGGTGTACATAGCGAAATACTTGTCCAGCGTGGTGATGCACCTGAATCTTTCTTCGAAACA	Nisin I1W/N20F
CG-Dok-94-FW	ACAGGAGCTCTGATGGGTTGTTTCATGAAAACAGCAACTTGTCATTGTAGTATTCACGTAAGCAAATAAGTACTGCAGGCATGCGGTACCACTAGTTCTAGAG	Nisin I1W/N20F
CG-Dok-108	TTTCATGTTACAACCCATCAGAGCTCCTGTTTTACAACCGGGTGTACATAGCGAAATACTTGTCCAGCGTGGTGATGCACCTGAATCTTTCTTCGAAACAG	Nisin I1W
CG-Dok-109	ACAGGAGCTCTGATGGGTTGTAACATGAAAACAGCAACTTGTCATTGTAGTATTCACGTAAGCAAATAAGTACTGCAGGCATGCGGTACCACTAGTTCTAGAGA	Nisin I1W
SK1	TTATTTAGAGATTTTGCAGTTTGCAGTTAGTGTTTGAAGGAAGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin C26A
SK2	ACTTGCTTCCTTCAAACACTAACTGCAAACTGCAAAATCTCTAAATAAGTAAAACCATTAGCATCACCTTGCTCTGACTCCTTGCACT	Subtilin C26A
SK5	TTATTTAGAGATTTTTGCGTTACAAGTTAGTGTTTGAAGGAAGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin C28A
SK6	ACTTGCTTCCTTCAAACACTAACTTGTAACGCAAAAATCTCTAAATAAGTAAAACCATTAGCATCACCTTGCTCTGACTCCTTGCAC	Subtilin C28A
TSp199	CTTCAAACACTAACTTGTAACTGCAAAATCTCTAAATAAGTAAAACCATTAGCATCACCTTGCTCTGACTCCTTGCACTTCTGAG	Universal oligonucleotide
TSp200	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTTGAAGCCAGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin F20W
TSp201	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTTGAAGTGCGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin F20A
TSp202	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTTGAAGATAGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin F20Y
TSp203	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTTGAAGATCGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin F20D
TSp204	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTTGAAGCTCGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin F20E
TSp206	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTTGAAGTTTGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin F20K
TSp207	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTTGAAGTCTGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin F20R
TSp210	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTGGTGGGAAGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTT	Subtilin L21P/Q22P
TSp211	CCACCAACACTAACTTGTAACTGCAAAATCTCTAAATAAGTAAAACCATTAGCATCACCTTGCTCTGACTCCTTGCACTTCTGAG	Subtilin L21P/Q22P
TSp220	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTTGAAGTACGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTTCCA	Subtilin F20V
TSp221	TTATTTAGAGATTTTGCAGTTACAAGTTAGTGTTTGAAGGATGCAAGTTTGCAATGCACCAGTTACACATCCTGGTGTACAAAGTGATTCACTTTTCCA	Subtilin F20I

^aLetters and numbers indicate the amino acid residue exchange and corresponding positions in subtilin/nisin variants.

For in vivo cloning by gap repair, the Saccharomyces cerevisiae strain CEN.PK2 (43) was transformed according to the method described by Schiestl and Gietz (44). The recombinant plasmids were isolated from CEN.PK2 according to Robzyk and Kassir (45) and, for further amplification, were transformed into E. coli DH5α. E. coli plasmid isolation was carried out using an alkaline extraction procedure. The transformation of B. subtilis was performed as described previously (20), whereas L. lactis was transformed by electroporation (46).

Bacterial and yeast strain growth conditions.

For optimized production of subtilin and its variants, the corresponding B. subtilis strains were grown at 37°C in medium A (7, 47). L. lactis strain NZ9700/NZ9800 derivates were grown overnight at 30°C in M17 broth (Fluka) containing 0.5% glucose without aeration, whereas the cultivation for optimized production of nisin and nisin variants was performed in TY medium (containing 3.125 g · liter⁻¹ tryptone, 6.25 g · liter⁻¹ yeast extract, 62.5 mg · liter⁻¹ MnSO₄·H₂O, and 156.25 mg · liter⁻¹ MgSO₄). The reporter strain B6633.MB1 (18) for autoinduction tests was grown in TY medium containing 0.3 M NaCl.

For selection, the following antibiotic and concentrations were used: neomycin, 15 μg · ml⁻¹; spectinomycin, 100 μg · ml⁻¹; and chloramphenicol, 5 μg · ml⁻¹ (or 10 μg ml⁻¹ for L. lactis). When two or more antibiotics were used simultaneously, the concentrations were reduced by half. E. coli DH5α strains with recombinant plasmids were grown in terrific broth (TB; 12 g tryptone, 24 g yeast extract, and 4 ml glycerol with the addition of 900 ml H₂O and 100 ml KPP buffer [0.17 M KH₂PO₄, 0.72 M K₂HPO₄]) containing 100 μg · ml⁻¹ ampicillin at 37°C. S. cerevisiae was grown at 30°C in YPD medium (1% yeast extract, 2% peptone, 2% glucose) or if necessary, in synthetic dropout medium (0.5% ammonium sulfate, 0.17% yeast nitrogen base, 2% glucose) without uracil/histidine for plasmid marker selection.

Plasmid construction.

All plasmids and oligonucleotides used in this study are listed in Tables 2 and 3. The construction of the various mutated spaS genes was performed as described before (19) using pCG2 as the backbone. The resulting plasmids with mutated subtilin structural genes were transformed into B15029.TSp01 and B15029 wild type as formerly published (20). For the generation of a template applicable for mutagenesis, the modified nisA gene was amplified with Phusion polymerase (New England BioLabs) using oligonucleotides CG-Dok-82 and CG-Dok-83 (Table 3). The resulting product was subsequently digested with EcoRI and XbaI and cloned into pCI372.

Purification of lantibiotics.

Culture supernatants of subtilin- and subtilin-variant-producing strains were harvested by centrifugation and purified according to a previously published protocol (20, 47–49) with modifications. To the supernatants, 0.5 volume of precooled n-butanol was added and the mixture was stirred for 2 to 3 h at 4°C. The emerged emulsion was transferred to precooled centrifuge tubes and subsequently centrifuged for 30 min at 4°C at 20,400 × g. The butanol phase was removed and immediately mixed (briefly) with 2 volumes of −20°C acetone. After an overnight incubation at −20°C, the solution was centrifuged (30 min at 4°C at 20,400 × g). After removing the supernatants, the pellets were dried under vacuum and stored at −20°C. For further purification, the frozen pellets were resuspended in 1 ml HPLC eluent A (20% acetonitrile [HPLC grade], 0.1% 2,2,2-trifluoroacetic acid [TFA]) per 50 ml culture supernatant. Afterwards, portions of the solution were loaded onto semipreparative reversed-phase (RP) HPLC columns (Gemini, 5 μm, NX-C₁₈, 110 Å, 250 mm by 10 mm; Phenomenex, Torrance, CA) using eluent A (20% acetonitrile [HPLC grade], 0.1% TFA) and eluent B (99.9% acetonitrile [HPLC grade], 0.1% TFA). The subtilin variants were separated using a linear gradient of eluent B over 40 min. The absorbance was monitored at 214 nm and 280 nm. The collected fractions were dried under vacuum and resuspended in 5% acetonitrile for in vivo analysis or in 30% acetonitrile plus 0.1% TFA for mass spectrometric analysis.

Nisin A producer L. lactis NZ9700 or strains expressing nisin variants were first precultured twice (1:100) in 4.2% M17 broth (Fluka) with 0.5% glucose at 30°C without aeration. Two liters of TY were inoculated with 10 ml preculture and incubated for 20 h at 30°C. Cells were harvested by centrifugation (7,000 × g for 20 min), and the cell pellets were resuspended in 300 ml of 70% 2-propanol (HPLC grade) and 0.1% TFA and stirred at room temperature for approximately 3 to 4 h. The cell debris was removed by centrifugation (7,000 × g for 20 min) and the supernatants were retained. The 2-propanol was evaporated via a rotary evaporator and afterwards, the pH was adjusted to 4 using NaOH. A solid-phase extraction (SPE) C₁₈ column (10 g, 60 ml; Phenomenex) was preequilibrated in a first step with 100% methanol and subsequently with water. The column was washed with 120 ml of 30% ethanol, and the elution was performed using 60 ml of 70% 2-propanol with 0.1% TFA. Aliquots were applied to a preparative RP-HPLC column as previously described (50). The quantification of subtilin, nisin, and the corresponding variant peptides was performed as described previously (17).

Measurement of promoter activities by β-galactosidase assay.

The samples for β-galactosidase assays were prepared as described previously (20, 51). In short, a fresh overnight culture of subtilin reporter strain B6633.MB1 was inoculated in TY medium containing 0.3 M NaCl to an optical density at 600 nm (OD₆₀₀) of 0.1 (18). After an optical density of approximately 1 was reached, 2-ml aliquots of the culture were transferred into small test tubes containing different concentrations of the lantibiotic. After 1 h of incubation at 37°C, samples were taken. Cells were harvested by centrifugation and stored at −20°C for further analysis via the β-galactosidase assay. The β-galactosidase activity was measured as described previously (51).

Antibiotic activities of lantibiotic variants and hybrids using the MIC.

For determination of the MICs, the lantibiotic-sensitive indicator strain K. rhizophila ATCC 9341 was used and, in this regard, the lowest concentration that completely prevented growth of the cells was determined the MIC value. The cultivation of K. rhizophila was carried out at 37°C in LB on a shaking device. Using a fresh overnight culture, the medium was inoculated to an OD₆₀₀ of 0.001. Twofold serial dilutions of the lantibiotics were made and added to 2 ml cultures. After 12 h of incubation at 37°C, the growth was evaluated.

DiSC₃(5) diffusion assay.

Pore-forming lantibiotics have a severe impact on the membrane potential of Gram-positive organisms. To visualize the transmembrane potential (ΔΨ) of B. subtilis cells, the positively charged carbocyanine 3,3′-dipropylthiadicarbocyanine iodide [DiSC₃(5)] (52) was used as summarized by Breeuwer and Abee (53) with major modifications. As a control, the dodecadepsipeptide valinomycin was used for all DiSC₃(5) diffusion assays. This ionophore is highly specific for potassium ions and transports them across the membrane, leading to the breakdown of the membrane potential. For the assay, cultures were grown until the mid-exponential phase (OD₆₀₀ of 0.9 to 1.5) and harvested at 16,000 × g for 5 min. The cells were resuspended in DiSC₃(5) buffer (50 mM KPP [pH 7], 300 mM KCl, 0.1% [wt/vol] glucose) to a titer of 5 × 10⁸ cells · ml⁻¹ (OD₆₀₀ of 1 for B. subtilis was estimated to correspond to 2.5 × 10⁸ cells · ml⁻¹). From this cell suspension, 3 ml was transferred to fluorescence cuvettes and immediately used for the DiSC₃(5) diffusion assay (storage for more than 10 min was not possible). The fluorescence cuvettes with 3-ml cell suspensions (5 × 10⁸ cells · ml⁻¹) were placed into a Fluorolog-3 spectrofluorometer (Horiba Jobin Yvon, Inc.) with stirring at 750 rpm, and the emission maximum at 670 nm (slit, 2 nm) with excitation at 544 nm (slit, 10 nm) was recorded at an interval of 0.1 s for a time period of 1,000 s. Precisely 50 s after beginning the test, 2.5 μM DiSC₃(5) fluorescent dye was added (2.5 μl of a 3 mM stock solution). After 150 s, the maximal fluorescence quenching was reached (0.1 × 10⁶ to 0.2 × 10⁶ relative fluorescence units [RFU]). At this minimal fluorescence level, the test component was added (either valinomycin [1 μM] or lantibiotics at different concentrations). The fluorescence release was monitored until the maximal fluorescence (in RFU) was reached and no further increase was detectable.