Role of the leader and structural regions of prelantibiotic peptides as assessed by expressing nisin-subtilin chimeras in Bacillus subtilis 168, and characterization of their physical, chemical, and antimicrobial properties.

Biosynthesis of lantibiotics such as nisin and subtilin involves post-translational modifications, including dehydration of serines and threonines, formation of thioether cross-linkages, translocation, cleavage of a leader sequence, and release into the medium. We have studied the cellular machinery that performs the modifications by constructing and expressing nisin-subtilin chimeric prepeptides in a strain of Bacillus subtilis 168 that possesses all of the cellular machinery for making subtilin except for the presubtilin gene. The chimeras consisted of a normal subtilin leader region (SL), fused to nisin-subtilin chimeric structural regions, one of which was SL-Nis11-Sub12-32, in which the N-terminal portion of the structural region was derived from nisin, and the C-terminal portion derived from subtilin. This chimera was accurately and efficiently converted to the corresponding mature lantibiotic, as established by reverse phase high performance liquid chromatography profiles, proton NMR spectroscopy, mass spectral analysis, and biological activity. A succinylated form of the chimera was also produced. Another chimera was in the reverse sense, with subtilin sequence at the N terminus and nisin sequence at the C terminus of the structural region (SL-Sub11-Nis12-34). It was processed into a heterogeneous mixture of products, none of which had the characteristics of a correctly processed polypeptide, but did contain a minor component that was active, with a specific activity that considerably exceeded nisin itself. These results, together with results published earlier, establish that processing requires specific recognition between the prelantibiotic peptide and the processing machinery, and in order for the processing to occur correctly, there must be an appropriate combination of the N-terminal part of the leader region and the C-terminal part of the structural region of the prepeptide.

Nisin (produced by Lactococcus lactis) and subtilin (produced by Bacillus subtilis) are the most thoroughly studied examples of lantibiotics, which are ribosomally synthesized antimicrobial peptides that are characterized by the presence of unusual lanthio and dehydro residues. Their structures are shown in Fig. 1. Their biosynthesis involves several post-translational modifications: dehydration of serines and threonines, formation of thioether cross-linkages between dehydro residues and cysteines, translocation, removal of a leader sequence, and release of the mature antimicrobial peptide into the extracellular medium (reviewed in Refs. [1][2][3]. Gene-encoded antimicrobial peptides constitute a family of natural products, whose known members are expanding rapidly in number and diversity and are produced by many kinds of organisms, ranging from bacteria to eukaryotes, including mammals (1, 4 -6). Their ubiquity among widely diverged organisms implies that they have had an opportunity to explore many strategies for achieving their antimicrobial properties, some of which are quite different from the mechanisms of classical antibiotics such as penicillin. They may therefore be able to supplement the arsenal of therapeutic antimicrobial agents that has been depleted as a result of the evolution of resistance among microbes. An advantage that is unique to gene-encoded antimicrobial peptides is that their structures can be readily manipulated by mutagenesis, which provides a facile means for constructing and producing the large numbers of structural analogs needed for structure-function studies and rational design. Whereas this advantage is shared by all gene-encoded antimicrobial peptides, the lantibiotics are unique in possessing the unusual dehydro and lanthio residues, which are absent from magainins (7)(8)(9), defensins (10 -13), or cecropins (14,15). This means that the lantibiotics offer chemical, physical, and hence biological properties that are not attainable by polypeptides that lack these residues. For example, the dehydro residues (Dha 1 and Dhb) are electrophiles, whereas none of the ordinary amino acids is electrophilic. The thioether crosslinkages are more resistant to breakage than the disulfide bridge and can better survive reducing conditions and extremes of pH and temperature (16).
A concern when making mutants of lantibiotics is the effect of the mutations on the post-translational modification process, because a mutation that disrupts processing makes the biosynthesis of the corresponding mature lantibiotic peptide impossible. All known lantibiotic prepeptides contain an N-terminal region that is cleaved during maturation, and for the Type A lantibiotics (e.g. nisin, subtilin, and epidermin), this leader region is highly conserved (17), and its participation in the orchestration of post-translational modification and secretion has been proposed (17,18). Certain mutations in the leader region of the nisin prepeptide have rendered the cell incapable of nisin production (19), whereas many mutations in the structural region of several lantibiotics that do not disrupt processing have been reported (e.g. Refs. 20 and 21). When the complete nisin prepeptide consisting of the nisin leader region and the nisin structural region (N L -Nis 1-34 ) was expressed in a subtilin-producing cell, no nisin-related peptide products were detectable (22,23). However, when a chimera consisting of the subtilin leader region and the nisin structural region (S L -Nis 1-34 ) was expressed in a subtilin-producing cell, an inactive nisin-like peptide was produced in which the leader region had been correctly cleaved and contained a full complement of unusual amino acids (22). The lack of activity was attributed to the formation of incorrect thioether cross-linkages (22). Similarly, when a prepeptide consisting of a subtilin leader region and a nisin structural region was expressed in a nisin-producing cell, the nisin structural region contained the unusual amino acids, but the leader was not cleaved (24). It has also been reported that expression of a prepeptide consisting of the nisin structural region fused to a subtilin-nisin chimeric leader region, S L 1-7 -N L 8 -23 -Nis 1-34 , forms active nisin when expressed in a subtilin-producing cell (23).
These results imply that the subtilin processing machinery in B. subtilis is not capable of recognizing the nisin prepeptide (which is ordinarily expressed in Lactococcus lactis) and converting it to nisin. However, the subtilin machinery will perform the modification reactions on the nisin structural peptide if it is attached to a subtilin leader region, although the modifications seem to be misdirected so that active nisin is not produced. Finally, the subtilin machinery will produce active nisin if the leader region is an appropriate combination of subtilin leader and nisin leader sequences. In this work, we explore the contribution of the structural region and its relationship to the leader region by the construction and expression of nisin-subtilin chimeras, which contain chimeric nisin-subtilin structural regions fused to the subtilin leader region. We have discovered that chimeras in which the C-terminal portion of the structural region correspond to subtilin are processed correctly and give active products, whereas those in which the C-terminal portion of the structural region corresponds to nisin produces a heterogeneous mixture of products, most of which, but not all, are inactive.

MATERIALS AND METHODS
Bacterial Strains, Cloning Vectors, and Mutagenesis-B. subtilis 168 strains and cloning vectors used were LH45erm⌬S (21), pTZ19U (Life Technologies, Inc.), pSMcat (21), and pACcat (this work). Structural mutants of subtilin were constructed and expressed using the cassette mutagenesis system as described previously (21). Synthetic oligonucleotides were annealed, filled in using Klenow fragment, restricted with EcoRI and HindIII, and cloned into the EcoRI-HindIII site of pTZ19U. The cloned fragment, which contained the mutation, was cut out with appropriate restriction enzymes and cloned into the corresponding site of the pSMcat vector or the pACcat vector. The mutated sequence was confirmed by sequence analysis of the cloned insert using the Sequenase version 2.0 sequencing kit from U. S. Biochemical Corp. This mutant gene was introduced into the chromosome of LHerm⌬S (from which the natural subtilin gene had been deleted) by transformation and Campbell-type integration using selection on chloramphenicol plates (21). Cloning vector pACcat was constructed by replacing the upstream BstEII site in the plasmid pSMcat with an SpeI site, thus making the downstream BstEII site a unique site.
Culture Conditions and Purification of Chimeric Peptides-Strains producing the mutant peptides were grown in Medium A (21,25) that was modified in that it contained 2% sucrose and 10 g/ml chloramphenicol. The culture was incubated with vigorous aeration for 25-35 h at 35°C, acidified to pH 2.5 with phosphoric acid, and heated to 121°C for 3 min to inactivate proteases. A 0.5 volume of n-butanol was added, stirred at 4°C for 2 h, allowed to stand at 4°C for 2 h, and centrifuged; 2.5 volumes of acetone were added to the supernatant, allowed to stand at Ϫ20°C for 16 h, and centrifuged. The pellet was lyophilized and resuspended in 20% acetonitrile with 0.05% trifluoroacetic acid. This was immediately purified on a C-18 reverse phase HPLC column using a trifluoracetic acid (0.05%)-water-acetonitrile gradient in which the acetonitrile varied from 0 to 100% over 30 min at a rate of 1.2 ml/min, unless indicated otherwise.
NMR and Mass Spectral Analysis-Samples for 1 H NMR spectral analysis were dissolved in deuterated water (99.96 atom% D, Aldrich Chemical Co.) lyophilized (repeated twice) to exchange protons, and dissolved in D 2 O to a final concentration of 2-3 mg/ml 1 H NMR spectra were obtained using a Bruker AMX-500 spectrometer interfaced to an Aspect 3000 computer. Spectra were obtained at a constant temperature of 295 K, using selective solvent suppression. Data processing was performed using UXNMR software. Mass spectral analysis was performed by PeptidoGenic Research & Co. (Livermore, CA) on a Sciex API I Electrospray mass spectrometer, which has an analysis range of over 200,000 Da with Ϯ0.01% accuracy, on 5-l samples at a concentration of about 5 pmol/l. The reported masses are those calculated as the most probable values based on the different m/z forms.
Measurement of Biological Activity-Biological activity was measured using an inhibition zone (halo) assay (21) and a liquid culture assay (26). HPLC fractions were tested for activity by spotting 15 l onto an agar plate (modified Medium A), incubated at 37°C for 15 min, sprayed with Bacillus cereus T spores, and incubated at 37°C for 16 h. Inhibition resulted in a clear zone (containing spores that were inhibited during outgrowth) surrounded by an opaque lawn of cells derived from the spores that had become vegetative. In the liquid culture assay, various concentrations of peptide were added to a suspension of B. cereus T spores in modified medium A, incubated in a rotating drum shaker at 30°C for 90 min; the inhibitory effects were evaluated using phase contrast microscopy and a Klett-Summerson colorimeter. Relative amounts of peptide were also estimated by integration of peak areas (measured at 214 nm) of the HPLC profiles, using nisin as a standard. It was assumed that the extinction coefficients of the mutant peptides are the same as nisin at this wavelength.
SDS-Polyacrylamide Gel Electrophoresis Analysis-The sizes of the peptides were estimated using Tricine-sodium dodecyl sulfate-polyacrylamide gel electrophoresis, which is designed for proteins in the range of 1-100 kDa (27), and uses a 4% stacking gel, a 10% spacer gel, and a 16.5% separating gel. Gels were silver-stained using kit 161-0443 from Bio-Rad, according to the manufacturer's instructions. Fig. 1 reveals that the number and locations of the thioether rings and the Dha residues are conserved. Each has one Dhb residue, but its position is not conserved. The N-terminal region is relatively conserved, except for three non-conservative differences out of the first 11 residues. Nisin has isoleucine at position 1, whereas subtilin has a bulky aromatic tryptophan. Subtilin has a positively charged lysine at position 2, whereas nisin has an unusual Dhb residue. Finally, subtilin has a negatively charged glutamate at position 4, whereas nisin has a neutral aliphatic isoleucine. In previous work, we changed the Glu 4 of subtilin to the Ile 4 of nisin, and obtained a mutant with enhanced chemical stability and activity (21). We wondered what would happen if we changed the other two residues as well, to give a subtilin analog with a nisin-like N terminus. This analog would have only hydrophobic residues at the N terminus, as well as a fourth dehydro residue at a location that is unfamiliar to the subtilin processing machinery, and if it were unable to process it properly, the entire processing pathway could abort. Since the subtilin machinery cannot process the S L -Nis 1-34 prepeptide to an active product (22), it is difficult to predict how the machinery would handle the S L -Nis 1-11 -Sub 12-32 prepeptide. We therefore constructed and expressed this prepeptide and examined its products.

Inspection of the structures of nisin and subtilin in
Construction and Expression of the Nis 1-11 -Sub 12-32 Chimera-Using the mutagenesis strategy shown in Fig. 2, residues 1, 2, and 4 in the subtilin structural region were changed to those of nisin. This chimeric gene was integrated into the chromosome of B. subtilis LH45erm⌬S, and the polypeptide products were isolated from the extracellular fluid. Fig. 3 shows the HPLC elution profile of the peptides isolated from cells in early stationary phase. A single large peak emerged from the column, and it possessed antimicrobial activity. Electrophoresis on Tricine-SDS gels showed a single major band with a relative molecular mass between 3,000 and 3,200, which was consistent with the material being the Nis 1-11 -Sub 12-32 chimera.
Appearance of a Succinylated Form of the Nis 1-11 -Sub 12-32 Chimera during Late Growth Stages, as Determined by Proton NMR and Mass Spectral Analysis-In an attempt to attain a higher yield of material, the culture was allowed to incubate into late stationary phase, whereupon the HPLC column profile showed two peaks, one being the original peak, with a second peak trailing slightly behind (Fig. 4, inset). The two peaks (Fig.  4, Early Peak and Late Peak) were collected separately and subjected to NMR spectroscopy as shown in Fig. 5. The spectra showed that the late peak was contaminated by the early peak, so the column was eluted with a shallower gradient (center, Fig.  4), and further experiments were performed using the more purified material. The antimicrobial activity is associated mainly with the early peak. The halo assays in Fig. 4 did not detect any activity in the late peak, but when higher concentrations were tested, its activity was found to about 10-fold lower than the early peak (data not shown). This is reminiscent of the observation that B. subtilis 6633 (the natural producer of subtilin) and LH45 (a subtilin-producing derivative of B. subtilis 168) produce two forms of subtilin (28); when B. subtilis 6633 is incubated into late stationary phase, there is an accumulation of subtilin that has been succinylated at its N terminus (29). The succinylated subtilin was significantly less active than the normal unsuccinylated subtilin. We therefore suspected that the late peak is the succinylated form of the early peak. This was confirmed by mass spectral analysis (Fig. 5), showing that the early peak consists mainly of a species with an M r ϭ 3185.98 (panel A), which conforms exactly to the expected 3186-Da mass of the mature Nis 1-11 -Sub 12-32 chi-

FIG. 2. Strategy for construction of Nisin-Subtilin chimeras.
Mutagenesis was performed in the plasmid pSMcat, which is a cassettemutagenesis plasmid that contains a copy of the subtilin structural gene upstream from a cat gene (21). When this plasmid is transformed into the B. subtilis 168 host LH45erm⌬S and selected on chloramphenicol, the subtilin gene is integrated into the chromosomal subtilin (spa) operon (21) at the site from which the natural subtilin gene has been deleted. The sequence of the Nis 1-11 -Sub 12-32 chimera and the nucleotide sequence that encodes it is shown (top), in which the 32-residue mature Nis 1-11 -Sub 12-32 sequence is numbered. Immediately below are the mutagenic oligonucleotides used to construct this sequence. The sequence of the S 1-11 -Nis 12-34 chimera and the oligonucleotides used to produce it are at the bottom. The Nis 1-11 -Sub 12-32 chimera constructed as shown in Fig. 2 was expressed, isolated, and subjected to HPLC chromatography. Samples were collected at 1-min intervals and assayed for activity using the halo assay. The major peak contained the only activity. A portion of this peak sample was subjected to SDS-polyacrylamide gel electrophoresis and silver-stained. The stained gel is shown in a panel beside the peak  Fig. 2 was expressed and isolated as for Fig. 3, except that the cells were grown for a longer time into the stationary phase, resulting in the appearance of a new peak on the HPLC column (inset), with the original peak (Early Peak) and the new peak (Late Peak). These peaks were separated by only 1 min, so were chromatographed using a shallower gradient (35-60% acetonitrile over 45 min, center), whereupon the early peak and late peak were separated by 4 min. The results of halo assays are shown above the center HPLC profile, with arrows indicating the positions in the profile from which the samples used for halo assays had been derived. mera. The late peak gave a mass of 3286.78 Da (panel B), which is consistent with an expected mass of 3286 Da for the succinylated form of the mature Nis 1-11 -Sub 12-32 chimera. We note that in order for these expected masses to occur, it is necessary for the chimeric prepeptides to have undergone the full panoply of posttranslational modifications in which 8 serines and threonines are dehydrated, 5 thioether cross-linkages are formed, and the leader region cleaved at the proper residue. The NMR spectrum of the early peak (panel D),and of the mixture (panel C) of the early and late peaks shows resonances that correspond to the Dhb 2 and Dha 5 resonances contributed by the nisin part of the molecule and to the Dhb 18 and Dha 31 residues contributed by the subtilin portion of the molecule. Succinylation of subtilin has been shown to cause a shift in the resonance of the Dha 5 residue (29), which can be attributed to a change in the chemical environment of Dha 5 caused by the presence of the N-terminal succinyl group. Since the Dhb 2 residue in the succinylated chimera is even closer to the succinyl group, a shift in its resonance would be expected. The spectrum shown in panel C, which includes resonances of the succinylated chimera, confirms these expectations, and shows a shifted resonance for Dha 5 (labeled as Dha 5 *), and for Dhb 2 (labeled as Dhab 2 *). We note that succinylation of the Nis 1-11 -Sub 12-32 chimera in the same manner as subtilin means that the cell treats the chimera in a completely normal way, and that the succinylation system must be able to tolerate the differences in the N-terminal end of the chimera, so must not be critical for recognition. The biological role of the succinylation is unknown.
The Biological Activity of the Nis 1-11 -Sub 12-32 Chimera-Nisin and subtilin can inhibit spore-forming food spoilage bacteria from undergoing outgrowth from spores to the vegetative state, as well as inhibit cells that are in the vegetative state (30). The mechanism of inhibition of these types of cells is different, as it has been shown that the Dha 5 residue is critical for subtilin to inhibit spore outgrowth, but not for subtilin to inhibit vegetative cells (31). The activity of the two purified forms of Nis 1-11 -Sub 12-32 were therefore measured against outgrowing spores and vegetative cells, and compared to nisin. Since the activities of subtilin and E4I-subtilin have previously been compared to nisin (21), the relative activities among all these forms can be inferred in terms of relative nisin units. The activity of Nis 1-11 -Sub 12-32 against spore outgrowth was estimated by the halo assay and the liquid assays, and against vegetative cells by the liquid assay. The chimera was active against both spore outgrowth and vegetative growth, and the specific activities of the chimera and nisin were so similar that they could not be distinguished in either their ability to inhibit spore outgrowth or to inhibit vegetative cells (data not shown). Accordingly, one sees inhibition of spore outgrowth at about 0.2 g/ml, and against vegetative cells at about 2 g/ml, with both the chimera and nisin. Based on previous measurements (21,31,32), this means that the Nis 1-11 -Sub 12-32 chimera is about 2-fold more active than E4I-subtilin and about 6 -8 times more active than natural subtilin.

Stability of the Dehydro Residues in the Nis 1-11 -Sub 12-32 Chimera during Incubation in Aqueous Solution-
The chemical and biological instability of subtilin have been correlated with the tendency of residue Dha 5 to spontaneously undergo chemical modification, which results in disappearance of the Dha 5 peak in the NMR spectrum and loss of biological activity (21,28,33). This instability of residue Dha 5 has been attributed to the participation of the carboxyl group of Glu 4 of subtilin in the modification process. Accordingly, changing Glu 4 to Ile 4 (E4I-subtilin) dramatically enhanced the chemical stability of the Dha 5 residue (21), with the chemical half-life of the Dha 5 residue increasing nearly 60-fold, from less than a day to 48 days. Since the Nis 1-11 -Sub 12-32 chimera has additional changes in the vicinity of the Dha 5 residue, the chemical stability of the dehydro residues was examined by taking the NMR spectrum of a sample that was incubated in aqueous solution for an extended period of time. A 3-mg amount of the Nis 1-11 -Sub 12-32 chimera (consisting of a mixture of the early peak and late peak as defined in Figs. 4 and 5) was dissolved in D 2 O at a pH of 6.0, placed in a closed NMR tube, and incubated in the dark at room temperature for 2.5 months. The NMR spectrum of this sample was determined from time to time, with the results shown in Fig. 6. The resonances of the dehydro residues changed very little during the course of the 72-day incubation period. The slight differences that are seen are readily attributable to variations introduced during base-line correction during computations with the spectral data. In contrast to the 0.8-day half-life of the Dha 5 residue in natural subtilin and its 48-day half-life in E4I-subtilin, the half-life of the Dha 5 residue in the Nis 1-11 -Sub 12-32 chimera is so long that it cannot be estimated from the 72-day time point. Longer incubation times were not performed. We conclude that the dehydro residues in the Nis 1-11 -Sub 12-32 chimera are extremely stable.
We conclude from these results that the Dha 5 residue is subject to profound changes in its chemical reactivity, ranging FIG. 5. NMR and mass spectra of the early peak and late peak. The early peak and late peak are those as defined in the HPLC elution profiles in Fig. 4. The mass spectrum of the early peak is shown in panel A, with a calculated mass of 3185.98 Da that is identical to the expected value of 3186 Da for the mature Nis 1-11 -Sub 12-32 chimera. The mass spectrum of the the purified late peak is shown in panel B, with its calculated mass of 3286.78 Da exactly corresponding to the 100 Da increase expected from addition of a succinyl group to the mature Nis 1-11 -Sub 12-32 chimera. The NMR spectrum of the early peak is shown in panel D. Identification of the Dhb 2 , Dha 5 , Dhb 18 , and Dha 31 peaks was by correlation with NMR spectra obtained previously for nisin (33) and subtilin (21). The NMR spectrum of a mixture of the early peak and late peak is shown in panel C. Resonances shifted by the presence of the succinyl group are identified by asterisks (Dhb 2 *, Dha 5 *).
from the most reactive state observed in natural subtilin, to the least reactive state observed in the Nis 1-11 -Sub 12-32 chimera; with E4I-subtilin being in between. Somewhat surprisingly, the biological activity displayed by these structural variants varies inversely with the reactivity, with the unstable subtilin having the lowest activity, and the highly stable Nis 1-11 -Sub 12-32 chimera displaying the greatest activity. The fact that the chemical reactivity of Dha 5 varies inversely with biological activity argues that role of the Dha 5 residue in the antimicrobial mechanism is not related to its chemical reactivity in a simple fashion, and that other factors, such as the specificity imposed by the peptide sequence surrounding the dehydro residue, are also important.
Properties of the Sub 1-11 -Nis 12-34 "Reverse Chimera"-An important feature of the S L -Nis 1-11 -Sub 12-32 chimeric prepeptide is that the subtilin processing machinery was able to correctly recognize and process it into its corresponding mature form. Since the same machinery cannot successfully process S L -Nis 1-34, it is clear that there is something in the Nis 12-34 region that disturbs the subtilin processing machinery. If this is the case, we reasoned that the machinery would not be able to correctly process a chimera that contained this Nis 12-34 region. We accordingly constructed a chimera that was a reverse (S L -Sub 1-11 -Nis 12-34 ) of the previous one, in that it contained subtilin sequence at the N terminus of the structural region and nisin sequence at the C terminus. This chimera was constructed using the strategy described in Fig. 2, which was integrated into the chromosome and expressed. The corresponding polypeptide was recovered from the culture supernatant using the butanol-acetone extraction method, and further purified by RP-HPLC as shown in Fig. 7. A major peak emerged somewhat earlier than expected for the Sub 1-11 -Nis 12-34 chimera, but it was devoid of activity. Moreover, mass spectral analysis showed it to have an M r ϭ 3544.47, which is 56.47 mass units greater than the expected 3488 Da. It is clear that something has gone wrong in the processing of the prepeptide. Following this large peak was a small peak that showed activity in the halo assay, also shown in Fig. 7. The amount of material in this peak is quite small, and the mass spectral analysis shows it to be very heterogeneous, consisting of at least a half-dozen species; none of which corresponded to the mass expected for the Sub 1-11 -Nis 12-34 chimera. Instead of an expected mass of 3488 Da, values of 3079 (expected Ϫ408, 13% of total), 3193 (Ϫ295, 27%), 3322 (Ϫ166, 12%), 3437 (Ϫ51, 27%), and 4174 (ϩ686, 21%) were obtained. None of these masses are readily explained in terms of simple processing defects, such as the dehydrations to give Dha 5 and Dha 33 (there is no Dhb expected) not having occurred, or the leader not cleaved. The small size of several species indicates proteolysis has occurred. Although we cannot tell which species is (or are) responsible for the activity, the amount of any of them, or even the total of them, is so small that the specific activity of whatever is responsible for the inhibitory activity is much higher than nisin itself. To show this, the total area of the active peak consists of no more than 10 g of peptide, of which 0.13 g was used for the halo assay shown in Fig. 7. This possesses an activity equivalent to 0.5 g of nisin (data not shown). If all of the components in the peak were equally active, they would be about 4-fold more active than nisin. The amount of the various components in the peak ranges from about 12% to 27% of the total. If all of the activity is due to just one of the components, then this component would be about 15-35 times as active as nisin, depending on whether it is a major or minor component. Determining the actual active species and its activity will require that the active component be purified to homogeneity and studied further. Although we do not know what contributes to this high activity, this observation may constitute a serendipitous path to the design of lantibiotic analogs with superior antimicrobial properties. DISCUSSION Our ability to incorporate the unusual dehydro and lanthio type amino acids into lantibiotic analogs and non-lantibiotic polypeptides depends on the ability of the lantibiotic processing machinery to cope with foreign precursor sequences. Our working hypothesis is that the leader region is primarily responsible for engaging the prepeptide with the processing machinery, and, once engaged, serines and threonines are dehydrated with little regard for the sequence in which they reside. Cysteines then react with particular dehydro partners in accordance with the forces of folding and conformation that exist within the polypeptide in a manner that is reminiscent of the specific selection of disulfide bond partners in polypeptides such as ribonuclease A and insulin (34). Although the results presented here do not prove this hypothesis, they are consistent with it and therefore support it. There are now several known instances in which prelantibiotic peptides undergo processing reactions, but give rise to inactive products. These are summarized in Table I. Examples are the S L -Nis 1-34 chimera, which produces a processed (22) but inactive (22,23) product when expressed in a cell that possesses the subtilin machinery, and the S L -Sub 1-11 -Nis 12-34 chimera (this paper), which produces a heterogeneous mixture of products that are mainly inactive, although at least one active form is produced. This means that although N L -Nis 1-34 is an authentic lantibiotic precursor, the subtilin machinery seems incapable of coping with it, and its gene products have not been detected (22,23). However, if you place the subtilin leader in front of the nisin structural region to give S L -Nis 1-34 , a processed, but inactive, product is produced by the subtilin machinery (22). From this, we conclude that the subtilin leader is competent in engaging the processing machinery, but there is something about the conformational and folding interactions between the leader and structural region in the S L -Nis 1-34 construct that causes some of the processing reactions, perhaps the partner selection in thioether -Nis 1-34 construct is processed properly to give active nisin (23) argues that critical conformational interactions are restored when an appropriate N-terminal sequence element from the nisin leader region is combined with a C-terminal sequence element of the subtilin leader. However, this combination of leader sequence elements must be appropriately complemented by the structural region. Whereas the S L -Nis 1-11 -Sub 12-32 construct is processed correctly (this work), the S L -Sub 1-11 -Nis 12-34 construct is not. Moreover, the processing reactions for the latter construct really run amok, giving a complex mixture of mainly inactive products. Surprisingly, at least one component in this mixture was active. Since none of the components had the mass of a correctly processed product, this activity must be due to an incorrectly processed component, and its specific activity was at least 4-fold and as much as 35-fold higher than nisin itself. We would like to know more about what is responsible for such high activity, because it may provide insight about how to design lantibiotics that are dramatically more effective than the natural forms. We conclude from these results that correct processing of the prelantibiotic peptide requires specific conformational communication between the N-terminal portion of the leader region and the C-terminal portion of the structural region.
These results also provide new insight about the relationship between the structure of lantibiotics and their chemical properties and biological activity. Subtilin and nisin are highly disparate in their chemical stability and specific activity, with nisin being superior to subtilin in both categories. The Nis 1-11 -Sub 12-32 chimera has the superior properties of nisin, showing that the 3 residues that differ in the N-terminal regions of nisin and subtilin are primarily responsible for the disparity between nisin and subtilin. We note that Nis 1-11 -Sub 12-32 has a very hydrophobic N-terminal region, which may facilitate insertion of the lantibiotic into the membrane, which is its target of action (26,(35)(36)(37). However, another possible explanation for its elevated activity is the presence of a second dehydro residue (Dhb) at position 2 in the Nis 1-11 -Sub 12-32 chimera. One might expect that the Dhb 2 would have a dramatic effect on the antimicrobial properties of the chimera, since it is so close to the critical Dha 5 and might cooperate in reacting with its microbial target; however, if it does, there could be no more than a 2-fold effect. This illustrates a frustrating aspect of our knowledge about lantibiotics. The ubiquitous occurrence of the unusual residues among the many known lantibiotics argues that they are conserved because they have important functions. Except for the critical role of Dha 5 in inhibition of spore outgrowth, functions that clearly justify this ubiquitous occur-  7. HPLC profiles and mass spectra of the Sub 1-11 -Nis 12-34 chimera. The HPLC profile of the Sub 1-11 -Nis 12-34 chimera is shown in the center. The halo assays of samples from the profile are shown at the bottom, with the arrows showing were the samples were derived from the profile. The only activity in the profile appeared at the position of the small peak appearing at 16 min. The mass spectrum at the left was obtained for the material in the large peak indicated by the arrow. The mass spectrum at the right was obtained for the material in the small peak (which was active) indicated by the other arrow.
rence have yet to be identified. This suggests that there are some very important attributes of lantibiotics that we know nothing about.