Identification and Characterization of the Structural and Transporter Genes for, and the Chemical and Biological Properties of, Sublancin 168, a Novel Lantibiotic Produced by Bacillus subtilis 168*

An antimicrobial peptide produced byBacillus subtilis 168 was isolated and characterized. It was named sublancin 168, and its behavior during Edman sequence analysis and its NMR spectrum suggested that sublancin is a dehydroalanine-containing lantibiotic. A hybridization probe based on the peptide sequence was used to clone the presublancin gene, which encoded a 56-residue polypeptide consisting of a 19-residue leader segment and a 37-residue mature segment. The mature segment contained one serine, one threonine, and five cysteine residues. Alkylation of mature sublancin showed no free sulfhydryl groups, suggesting that one sulfydryl had formed a β-methyllanthionine bridge with a dehydrobutyrine derived by posttranslational modification of threonine; with the other four cysteines forming two disulfide bridges. It is unprecedented for a lantibiotic to contain a disulfide bridge. The sublancin leader was similar to known type AII lantibiotics, containing a double-glycine motif that is typically recognized by dual-function transporters. A protein encoded immediately downstream from the sublancin gene possessed features of a dual-function ABC transporter with a proteolytic domain and an ATP-binding domain. The antimicrobial activity spectrum of sublancin was like other lantibiotics, inhibiting Gram-positive bacteria but not Gram-negative bacteria; and like the lantibiotics nisin and subtilin in its ability to inhibit both bacterial spore outgrowth and vegetative growth. Sublancin is an extraordinarily stable lantibiotic, showing no degradation or inactivation after being stored in aqueous solution at room temperature for 2 years. The fact that sublancin is a natural product of B. subtilis 168, for which a great deal of genetic information is available, including the entire sequence of its genome, suggests that sublancin will be an especially good model for studying the potential of lantibiotics as sources of novel biomaterials.

lis 168 was isolated and characterized. It was named sublancin 168, and its behavior during Edman sequence analysis and its NMR spectrum suggested that sublancin is a dehydroalanine-containing lantibiotic. A hybridization probe based on the peptide sequence was used to clone the presublancin gene, which encoded a 56-residue polypeptide consisting of a 19-residue leader segment and a 37-residue mature segment. The mature segment contained one serine, one threonine, and five cysteine residues. Alkylation of mature sublancin showed no free sulfhydryl groups, suggesting that one sulfydryl had formed a ␤-methyllanthionine bridge with a dehydrobutyrine derived by posttranslational modification of threonine; with the other four cysteines forming two disulfide bridges. It is unprecedented for a lantibiotic to contain a disulfide bridge. The sublancin leader was similar to known type AII lantibiotics, containing a double-glycine motif that is typically recognized by dual-function transporters. A protein encoded immediately downstream from the sublancin gene possessed features of a dual-function ABC transporter with a proteolytic domain and an ATP-binding domain. The antimicrobial activity spectrum of sublancin was like other lantibiotics, inhibiting Gram-positive bacteria but not Gram-negative bacteria; and like the lantibiotics nisin and subtilin in its ability to inhibit both bacterial spore outgrowth and vegetative growth. Sublancin is an extraordinarily stable lantibiotic, showing no degradation or inactivation after being stored in aqueous solution at room temperature for 2 years. The fact that sublancin is a natural product of B. subtilis 168, for which a great deal of genetic information is available, including the entire sequence of its genome, suggests that sublancin will be an especially good model for studying the potential of lantibiotics as sources of novel biomaterials.
Lantibiotics are bacterially produced antimicrobial peptides that possess unique chemical and biological properties owing to their containing a variety of unusual amino acid residues. Lantibiotics are defined as such by the presence of lanthionine or ␤-methyllanthionine, which are introduced by a posttranslational process in which serine or threonine is dehydrated to the corresponding dehydro residue, which then reacts in a Michael-type addition of a cysteine sulfhydryl group to the double bond of the dehydro residue to form a thioether link (reviewed in Refs. [1][2][3][4][5][6]. Mature lantibiotics typically contain one or more dehydro residues that do not participate in lanthionine bridges. The unique properties that are conferred by these unusual residues may result in their being useful components in the design of novel biomolecules (1,2,7,8).
One of the attractive features of lantibiotics is that they are comprised of gene-encoded polypeptide sequences, so their structures can be manipulated by protein engineering. Whereas this is simple in concept, putting it into practice requires the utilization of many different genetic and recombinant DNA techniques, including the removal and replacement of chromosomal segments with their genetically engineered counterparts. Ideally, these manipulations need to be done in such a way that the engineered lantibiotic analog be efficiently produced so that useful amounts of the analog are available for experimentation, which implies a need to engineer regulatory elements. Only a few bacterial strains have been sufficiently characterized to permit these manipulations to be performed in a convenient and facile manner. One such well characterized bacterial strain is Bacillus subtilis 168, which is second only to Escherichia coli in the extent to which tools of genetic and protein engineering have been developed, which has contributed to the extensive use of B. subtilis 168 for the industrial production of bio-engineered materials. The advantage of B. subtilis 168 over other bacterial strains has recently been increased even more by the availability of the complete sequence of the B. subtilis 168 genome (9).
It is in this context that we report the discovery of a new lantibiotic, which we have named sublancin 168, that is a natural product produced by B. subtilis 168. Although approximately 20 lantibiotics are already known, the fact that this new lantibiotic is endogenous to B. subtilis 168, and thus can be studied and manipulated using the powerful methods that are available in this strain, suggests that progress in our understanding of lantibiotics will be accelerated by our ability to study and manipulate sublancin and the genes associated with its production in its natural B. subtilis 168 host. In addition to this practical aspect of the discovery, sublancin 168 has structural features and physical properties, such as the presence of disulfide bridges and extraordinary stability, that are unprecedented among the known lantibiotics.

MATERIALS AND METHODS
Bacterial Strains, Cloning Vectors, and Culture Conditions-Sublancin was isolated from B. subtilis BR151, which is B. subtilis 168 (lys-3 metB10 trpC2), obtained from the Bacillus Genetics Stock Center, Ohio State University, Columbus, OH. Stocks were maintained on agar with Penassay broth (17.5 g of Bacto antibiotic medium 3/liter). Sublancin was produced by inoculating 1 liter of medium A with 10 ml of BR151 cultured for 16 h at 37°C with vigorous aeration. Medium A is as described previously (10,11), except it contained 2% sucrose instead of 10% sucrose. The culture was agitated vigorously by shaking 500-ml volumes at 200 rpm in 2-liter baffled flasks at 37°C for 28 h, whereupon the culture usually acquired a pinkish-brown color, a fruity odor, and a pH that had dropped to about 6. Good sublancin production was consistently obtained when these events were observed. For reasons that are not understood, the color, odor, and pH changes did not always occur, whereupon sublancin production was usually poor. Similar variability has been reported for subtilin production in B. subtilis ATCC 6633 (10).
Isolation of Sublancin 168 -The culture was acidified to pH 2.5 with concentrated phosphoric acid, and centrifuged to remove cells. The supernatant was made 1 M in NaCl and then applied, by a peristaltic pump, to a hydrophobic interaction column constructed with 25 ml of Toyopearl ® Butyl-650 resin (TosoHaas, Montgomeryville, PA) that had been equilibrated with 1 M NaCl, 50 mM NaAc, pH 4. Unbound proteins were eluted with several volumes of the loading buffer, and the sublancin was eluted with 50 mM NaAc, pH 4.0, or alternatively, with 30% acetonitrile. After being lyophilized, the residue was dissolved in a minimum amount of water that contained 0.1% trifluoroacetic acid, centrifuged to remove particulates, and applied to an analytical reverse-phase C-18 HPLC 1 column (Rainin/Varian, Walnut Creek, CA) in a Hewlett-Packard 1050 HPLC machine with a diode-array detector. Sublancin was eluted using a two-step gradient (solvent A was 0.1% trifluoroacetic acid in water, solvent B was 0.1% trifluoroacetic acid in acetonitrile), the first step going from 0 to 25% solvent B over 30 min, and the second step going from 25 to 35% solvent B over 30 min, using a 1.2 ml/min flow rate throughout. Fractions in the second step were assayed for antimicrobial activity; active fractions were pooled, lyophilized, and then subjected to a second round of HPLC purification using the same conditions as the first round. The elution profile was monitored at wavelengths of 214, 254, and 280 nm to detect the presence of peptide, dehydro residues, and aromatic residues, respectively. During the second round of HPLC purification, the activity was associated with a single absorbance peak, which was lyophilized and stored at Ϫ20°C.
Assay of Sublancin 168 Activity-Two methods were employed: a halo assay on plates, and a liquid assay in culture tubes. Both methods used Bacillus cereus T spores as the test organism. 250 mg of spores, prepared as described previously (12), were suspended in 30 ml of distilled water with a glass homogenizer, heat-shocked at 65°C for 2 h, centrifuged, and resupended in 50% ethanol. This suspension was sprayed onto the surface of medium A-containing agar plates using a Sigma spray unit. Prior to spraying, 10 -20-l volumes of serial dilutions of purified sublancin were spotted onto the plate. The plates were incubated at 37°C for 5-12 h to allow germination and growth of the spores, the diameters of the halos caused by sublancin inhibition were measured, and the minimum amount of peptide that was required to give an observable halo was noted. For the liquid assay, heat-shocked spores were suspended in sterile water to a final concentration of 2 mg/ml, and then added to culture tubes containing 1% Bacto-tryptone, 0.1 M Tris-P i , pH 6.8. Prior to adding the spores, serial dilutions of sublancin were added to the tubes. The final concentration of spores was 0.1 mg/ml. The tubes were incubated at 37°C for 3 h, using sufficient shaking to keep the spores well suspended. The cultures were then examined using phase-contrast microscopy to observe the extent to which the spores had undergone germination, outgrowth, and vegetative growth. The amount of sublancin required to prevent the spores from proceeding through outgrowth was noted.
Spectrum of Sublancin Antibiotic Activity-The ability of sublancin to inhibit growth of a variety of Gram-positive and Gram-negative bacterial strains was assessed using an agar-diffusion method, and for those strains that showed sensitivity to sublancin, a minimum inhibitory concentration (MIC) was determined. For the agar diffusion test, agar plates contained Difco brain heart infusion (Listeria monocytogenes, Lactococcus lactis, Enterococcus faecalis, Streptococcus pyrogenes) or Difco nutrient broth (B. cereus T, Bacillus megaterium, B. subtilis, Staphylococcus aureus, Staphylococcus epidermidis, Bordetella bronchiseptica, E. coli, Yersinia enterocolitica). A 1000-fold dilution of exponential cultures of the respective strains was made into molten top agar containing the appropriate medium, which was poured onto the agar plates. After solidification of the top agar, wells were made with an Ouchterlony punch and filled with 20 l (25 g) of sublancin solution. The plates were incubated for 24 h at 37°C, and the diameters of any halos of inhibition around the wells were measured. For those strains that showed a halo of inhibition, a MIC was determined by making a 100-fold dilution of an exponential culture of cells in tubes containing growth medium together with different concentrations (5, 10, 25, 50, or 100 g/ml) of sublancin, which were incubated with shaking at 37°C for 18 -30 h, until the respective control cultures without sublancin reached saturation. The MIC was that concentration of sublancin that completely suppressed growth of the cells.
Amino Acid Sequence and Composition Analysis-Purified sublancin was sequenced from its N-terminal end using Edman degradation, using an Applied Biosystems (Foster City, CA) model 477A peptide sequencer with an on-line HPLC analyzer in the University of Maryland Core Facility (Baltimore, MD). Amino acid composition analysis was performed on HCl hydrolysates by Commonwealth Biotechnologies, Inc. (Richmond, VA). Sublancin was treated with ethanethiol in order to sequence through any dehydro residues, which are otherwise blocked, using the method of Meyer et al. (13). The modification mixture consisted of 280 l of ethanol, 200 l of sterile deionized water, 65 l of 5 M NaOH, and 60 l of ethanethiol. 150 l of this modification mixture was added to 50 g of freeze-dried sublancin and incubated under nitrogen for 1 h at 50°C. The pH was lowered by addition of 5 l of glacial acetic acid, and the product purified by HPLC as described above for sublancin.
NMR and Mass Spectral Analysis-One-dimensional NMR spectroscopy was performed with a Bruker AMX-500 NMR spectrometer interfaced to an Aspect 3000 computer using UXNMR software. Lyophilized sublancin was dissolved in 99.96% atom% D 2 O to exchange protons and lyophilized (done twice) and dissolved in D 2 O to a final concentration of 10 mg/ml. The proton spectra were recorded at constant 295 K in D 2 O with and without the suppression of the water solvent resonance. Mass spectral analysis was performed by PeptidoGenic Research & Co (Livermore, CA) on a Sciex API I Electrospray mass spectrometer. The reported masses are those calculated as the most probable values based on the different m/z forms.
Cloning of the Sublancin Gene-A B. subtilis 168 genomic library was constructed in bacteriophage using total chromosomal DNA from strain BR151 grown in 50 ml of Penassay broth. Cells were lysed with a mixture of lysozyme, sodium dodecyl sulfate, and proteinase K; the DNA was recovered and deproteinized with phenol-chloroform as described previously (11). The genomic DNA was partially digested with Sau3AI to give random fragments in a 12-23-kb size range, which were cloned into LambdaGEM-12 partially filled-in XhoI half-site arms obtained from Promega (Madison, WI), which were then packaged into E. coli cells using the protocol provided by the manufacturer. The library was screened for the sublancin gene using synthetic DNA oligomers whose sequences were chosen using the strategy of Lathe (14), based on the 16-residue N-terminal sequence of sublancin. Three single-sequence 48-mer probes were designed, each one with randomly chosen degenerate bases, and the synthesis was performed by Ransom Hill Bioscience (Ramona, CA). For those amino acid residues that appeared as unidentifiable blanks in the sequence, inosines were placed in the corresponding codons in the probes. The three probes were: probe 1, GGGTTGGGTAAAGCCCAAIIIGCGGCCTTGTGGTTACAGIIIGCTT-CC; probe 2, GGGTTGGGCAAAGCACAGIIIGCGGCTTTTTGGTTAC-AGIIIGCGTGC; probe 3: GGACTTGGTAAAGCGCAAIIIGCAGCTCTG-TGGCTTCAAIIIGCATGC.
The probes were radiolabeled with 32 P at their 5Ј ends using T4 polynucleotide kinase and hybridized to Southern blots of restriction digests of BR151 genomic DNA under a variety of temperature and ionic strength conditions in order to optimize the signal strength and specificity. Probe 1 gave a good signal when hybridized at 45°C in 6ϫ SSC and washed at 37°C in 2ϫ SSC, whereas probe 3 gave a good signal when hybridized at 45°C in 6ϫ SSC and washed at 45°C in 2ϫ SSC. A good signal for probe 2 could not be obtained, so its use was abandoned. The bacteriophage library was plated and transferred to duplicate nitrocellulose filters using standard procedures (15). One of the duplicate filters was hybridized to probe 1, and the other to probe 3. The only plaques selected for further study were those that hybridized to both probes. Several such dual-hybridizing plaques were picked, and their inserts were subcloned into pTZ plasmids and screened again with probes 1 and 3. Positive inserts were cloned into M13 and subjected to dideoxy sequence analysis. The DNA sequences were conceptually translated into six reading frames, which were searched for the Nterminal amino acid sequence of sublancin. When the sublancin se-quence was found, the actual DNA sequence that encoded the sublancin gene could be identified, which provided the sequence information needed to synthesize probes that were exactly homologous to the sublancin gene. These were used to identify library clones that contained sequences that surrounded the subtilin gene, which were then also subcloned and sequenced.

RESULTS
Isolation of Sublancin 168 -Our first observation of sublancin was when it appeared as a contaminant in acetone-butanol extracts of culture supernatants of B. subtilis LH45, which is a strain of B. subtilis 168 that has been genetically engineered to produce subtilin (16). When the acetone precipitate was dissolved in water and analyzed by reversed-phase HPLC, a peak that contained antimicrobial activity that emerged earlier than subtilin was observed. Its early emergence indicated that it is more hydrophilic than subtilin, but its appearance in the acetone precipitate suggested that it had subtilin-like physical properties. Several characteristics of the wild-type B. subtilis 168 strain from which LH45 had been derived suggested that wild-type B. subtilis 168 harbors an endogenous lantibiotic. 2 The possibility that this contaminating activity might be a new lantibiotic prompted its isolation and characterization.
To obtain this putative lantibiotic, a strain of B. subtilis 168 was cultured as described under "Materials and Methods," and the active material was recovered from the supernatant using a hydrophobic interaction column and purified to near-homogeneity on a reversed-phase HPLC column, as shown in Fig. 1. The active peak showed absorbances at 214, 254, and 280 nm; when the active peak was treated with ninhydrin, it gave the purple color that is characteristic of proteins and peptides. The antimicrobial substance was named sublancin 168, to connote its being an antimicrobial peptide that is produced by B. subtilis 168.
Structure Analysis of Sublancin 168 -The sublancin peptide was subjected to N-terminal sequence analysis using Edman degradation. It gave a sequence of Gly-Leu-Gly-Lys-Ala-Gln-Xaa-Ala-Ala-Leu-Trp-Leu-Gln-Xaa-Ala-Xaa-Xaa-Xaa. The blank cycles (Xaa) were those that did not show an identifiable amino acid, some of which could be due to Cys residues, which were underivatized and therefore not detectable. Other potential sources of blank cycles are the unusual amino acid residues typically found in lantibiotics. For example, lanthionine residues do not produce peaks that are identifiable as normal amino acids, and the dehydro residues block the sequence analysis because they spontaneously lose their N-terminal amino group and are therefore unable to react with the Edman reagent (17), thus bringing the sequence analysis to a halt. This dehydro-residue block can be alleviated by reacting the peptide with ethanethiol, which adds across the double bond, thus preventing loss of the N-terminal amino group (13). Sublancin was accordingly derivatized with ethanethiol, whereupon it was possible to sequence past the apparent block at position 16, and to obtain Gly both at positions 17 and 18; however, a blank was then encountered at position 19. The fact that ethanethiol derivatization alleviated the block at position 16 is strong evidence that residue 16 is a dehydro residue.
Cloning and Sequence Analysis of the Sublancin Gene-Since lantibiotics are biosynthesized from gene-encoded precursors, one approach to determine if sublancin is a lantibiotic is to see if it is gene-encoded, and if it is, to examine the gene and the operon in which it is found to see if they possess features that are characteristic of lantibiotics. To determine whether sublancin is a gene-encoded peptide, the N-terminal sequence was used to design a hybridization probe, which was then used to screen a B. subtilis 168 genomic library that had been constructed in bacteriophage . Clones containing positive signals were subjected to DNA sequence analysis. The probe design, screening, and sequencing are described under "Materials and Methods." Nearly 5 kb of sequence was obtained, which we published in a public data base as soon as it was complete (GenBank accession no. AF014938 (1997)). This sequence was searched for open reading frames (ORFs), which were in turn searched for the N-terminal amino acid sequence of sublancin 168. A 56-residue ORF, shown in Fig. 2, that contained a perfect match to the N-terminal sequence of the sublancin peptide was found near the center of the 5-kb sequence. In addition, a 332-residue ORF was found upstream from the sublancin gene, and about 560 residues of a partially complete ORF was found downstream from the sublancin gene. The locations of these three ORFs within the 5-kb sequence are shown in Fig. 2. Several months after our sequence was published in GenBank, the Bacillus Genome Project published the complete B. subtilis 168 genome (9), which mapped these genes at a position of 193.8°on the B. subtilis 168 chromosome.
The putative functions of the upstream and downstream ORFs were explored by searching the GenBank/EMBL nucleotide data bases for homologies to proteins with known functions. The 332-residue upstream ORF (denoted uvrX) showed extensive homologies to proteins involved in repair of uv damage to DNA, so a role in the biosynthetic pathway of sublancin seems unlikely. The 560-residue segment of the downstream ORF showed homologies to known ABC transporter proteins including PepT, which is the transporter that is responsible for secretion of Pep5 during its biosynthesis (19). The gene for this downstream ORF (denoted sunT) is therefore a strong candidate as the corresponding transporter that participates in the secretion of sublancin. Fig. 3 shows the segment of the DNA sequence that contains the sublancin gene (sunA), and the 5-prime end of sunT, together with their conceptual translations (SunA and the N-terminal portion of SunT), the putative promoter region of the sun operon, and the ribosome binding site of the SunA mRNA. The complete sequences and their conceptual translations are available as accession number AF069294 in GenBank.
If sublancin 168 is a lantibiotic, then SunA is presublancin, and accordingly should contain structural features that are similar to known prelantibiotics. Fig. 4 1. HPLC purification of sublancin 168. Elution of sublancin from a reversed-phase C-18 column using a water-trifluoroacetic acidacetonitrile gradient as described under "Materials and Mehtods." Profile shown is the second round of purification, in which the activity eluted in the middle of the second step of the gradient, which went from 25-35% acetonitrile over 30 min at 1.2 ml/min. The molecular mass of the material eluting in the activity peak was 3877.78 kDa.
with the type A lantibiotics, which are divided into two subtypes, AI and AII. The type A lantibiotics include those that are the most thoroughly studied, such as nisin A, subtilin, epidermin, and Pep5. Type A lantibiotics are characterized by being elongated and cationic with molecular masses ranging from 2151 to 4635 Da (1). The mature region of the sublancin peptide is cationic, and its predicted molecular mass is approximately 3900 Da (depending on what posttranslational modifications have occurred), and thus possesses characteristics of a type A lantibiotic. Type AI and type AII lantibiotics differ in their leader segments, with the AII leaders containing a GA/GS/GG ("double-glycine") sequence motif immediately preceding the FIG. 2. Sequence of presublancin 168. The conceptual translation of the 56-residue ORF in the middle of the 5-kb sequence is shown, cleaved into a 19-residue leader segment, and a 37-residue mature segment. The N-terminal end of the mature segment is a perfect match to the N-terminal amino acid sequence of the HPLC-purified sublancin shown in Fig. 1. Serine, threonine, and cysteine residues which are candidates for posttranslational modification during maturation are shown in an enlarged font. The uvrX gene encodes an ORF with homologies to proteins involved in uv repair, and is therefore presumed to be unrelated to sublancin biosynthesis. The sunA gene encodes the presublancin polypeptide as indicated, followed by a non-coding reagion, and then sunT gene, which encodes SunT, which is the putative sublancin transporter as described in the text and in Figs. 3 and 4. The direction of transcription of these genes is indicated by the horizontal arrows. P indicates the location of a consensus prokaryotic promoter site as described in the legend of Fig. 3.  (3). Sublancin shows homologies that are characteristic of type AII lantibiotics, including the "diglycine motif" found in leaders that are normally cleaved by dual-function transporters that contain a leader peptidase function (20), as described under "Results." cleavage site, and conserved EL/EV and EL/EM sequences upstream from the cleavage site. Double-glycine type leader peptides are unrelated to the N-terminal sequences utilized by the sec pathway, and the corresponding ABC transporters typically possess a dual function that both removes the leader peptide and translocates it across the cytoplasmic membrane (20). These features are shared by several non-lantibiotic antimicrobial peptides, including pediocin and lactococcin A, which are produced by Gram-positive bacteria; and by colicin V, which is produced by Gram-negative E. coli (20), suggesting that the double-glycine leader peptide may represent an evolutionary branch-point between the lantibiotic and non-lantibiotic peptides.
If the double-glycine leader peptide of sublancin is cleaved by a protease that is a component of a dual-function transporter, then the transporter should contain an identifiable protease domain. Examination of the ORF that is immediately downstream from the sunA gene shows that the putative SunT protein shows such a protease domain. Fig. 5 compares SunT with two other ABC-transporter proteins: PepT, which is the transporter for the type AI lantibiotic Pep5 (which does not have a diglycine-type leader peptide), and LcnDR3, which is the transporter for the non-lantibiotic lactococcin DR (which does have a diglycine-type leader peptide). The LcnDR3 protein contains an N-terminal protease domain that consistently appears in the dual-function transporters that cleave the leaders that contain the diglycine motif (20), and this protease domain also appears in the SunT protein, and contains the conserved cysteine and histidine residues that are part of the active site of the proteolytic domain.
The fact that the leader segment of sublancin contains that conserved features that are typical of type AII lantibiotics constitutes evidence that sublancin is not only a lantibiotic, but is a type AII lantibiotic. The SunT protein supports this conclusion by showing the presence of the protease domain that is expected for a transporter of a type AII lantibiotic. Moreover, there is strong homology to PepT, which is a transporter of the lantibiotic Pep5, and lantibiotic transporters (LanT proteins) are generally conserved (2). All of these considerations are consistent with sublancin's being a lantibiotic of the AII type.
Biochemical Properties of Sublancin-For sublancin to be a typical lantibiotic, it should contain at least one lanthionine residue, either Lan or MeLan; and at least one dehydro residue, either Dha formed from serine, or Dhb formed from threonine. The putative mature region of sublancin contains only one serine (residue 16), and one threonine (residue 19). For sublancin to contain at least one dehydro residue and one lanthionine residue would require that both the Ser 16 and Thr 19 be converted to Dha and Dhb, respectively, and for one of them to form a cross-linkage with a cysteine, and for the other to remain as a dehydro residue. These possibilities can be distinguished by NMR spectroscopy. Both Dha and Dhb contain vinyl protons, which typically give resonance peaks in the ␦ ϭ 5.2-6.9 ppm region of the NMR spectrum, with Dha appearing as a doublet, and Dhb appearing as a quartet (7,(21)(22)(23). The NMR spectrum of sublancin is shown in Fig. 6. A portion of the NMR spectrum shows a doublet appearing at ␦ 6.2 ppm, which is in the middle of the vinyl proton region, and therefore argues that a dehydro residue is present, and its being a doublet further argues that it is a Dha. The other peaks are in the aromatic proton region (␦ ϭ 6.5-8.0), and can be attributed to the aromatic residues in sublancin. It is to be noted that the Edman degradation of native sublancin was blocked from residue 16 on, and this block was alleviated by reacting with ethanethiol, which is also consistent with residue 16 being a dehydro residue. Since the gene sequence shows a Ser at position 16, one can conclude that the Dha shown in the NMR spectrum is derived by post-translational dehydration of Ser 16 to Dha.
When sublancin was subjected to SDS-PAGE, it showed a single band that migrated at a position that corresponded to a molecular mass of approximately 4 kDa (data not shown). Ion-spray mass spectroscopy provided a more precise molecular mass of 3877.78 kDa, as shown in Fig. 1. The sublancin molecular mass as predicted from the amino acid sequence encoded in the sublancin gene is 3713.3 Da, assuming one MeLan, one Dha, and four cysteines existing in two disulfide bridges. There is thus a discrepancy of 164.48 Da between this predicted molecular mass and the actual molecular mass, which may be due to one or more additional modifications of the amino acids, as considered under "Discussion." Analysis of Disulfide Cross-linkages in Sublancin-The sublancin prepeptide contains five cysteine residues, which is the same number of cysteines as are present in the prepeptides of nisin and subtilin. However, in nisin, subtilin, and all other  (20), and the N-terminal end of PepT, the ABC-transporter that exports the lantibiotic Pep5 during biosynthesis (19). Pep5 is a type AI lantibiotic (non-diglycine leader), and its transporter does not possess a N-terminal proteolytic domain. However, the C-terminal ATP-binding domain of PepT shows strong homology to SunT (not shown). LcnDR3 does possess an N-terminal proteolytic domain, and a homologous counterpart appears in SunT, including the conserved histidine and cysteine regions (identified by enclosing boxes and by stars) that are part of the active site of the proteolytic domain (20). The GenBank accession numbers of SunT, LcnDR3 (also called LctT), and PepT are AF069294, U91581, and Z49865, respectively. known lantibiotics, all of the cysteine residues are converted to unusual residues such as the five Lan and MeLan in nisin (24) and subtilin (11), or the aminovinylcysteine in epidermin (25). For a natural lantibiotic to contain unmodified cysteines or disulfide cross-linkages is unprecedented, so the cysteine residues in sublancin were examined to see if any possessed the characteristics of either free sulfhydryl groups or disulfide bridges. The amino acid analysis that was employed cannot detect free cysteine residues, but can detect them as carboxymethyl-cysteine if they are alkylated prior to acid hydrolysis. Alkylation of native sublancin followed by amino acid analysis gave no detectable carboxymethyl-cysteine, which rules out the presence of free sulfhydryl groups (data not shown). Reduction of sublancin with dithiothreitol followed by alkylation gave 3.3 (suggesting a real value of 4, since the 3.3 is likely a minimum value, and the nearest integer value larger than 3.3 is 4) carboxymethyl-cysteines/mol of sublancin. SDS-PAGE and ionspray mass spectroscopy results described above established that sublancin exists exclusively as a monomer, so there cannot be any intermolecular disulfide bridges. These observations are all consistent with four of the cysteines of sublancin participating in two disulfide bridges, with the fifth cysteine having been converted to a MeLan residue by reacting with a Dhb residue (derived from post-translational dehydration of Thr 19 ), leaving the unreacted Dha 16 that is revealed in the NMR spectrum.
The number and location of disulfide bridges was further explored by analysis with proteolytic enzymes. The native form of sublancin, and the denatured form, and the denatured-reduced form of sublancin were all resistant to trypsin, despite the presence of a Lys at position 4 and an Arg at position 33. When the denatured and reduced sublancin was alkylated, trypsin cleavage gave detectable amounts of fragments of 3200 and 1581 Da, neither of which is an expected product. Sublancin was more sensitive to chymotrypsin, with even the native molecule being substantially degraded, to give products of 1392 and 1823 Da. The first is consistent with a polypeptide consisting of residues 1-11 being cross-linked by a disulfide bridge to a peptide consisting of residues 36 and 37 (G 1 -W 11 -S-S-C 36 -R 37 , with an expected value of 1392 Da), and the second is consistent with a polypeptide consisting of residues 1-11 crosslinked by a disulfide bridge to a peptide consisting of residues 33-37 (G 1 -W 11 -S-S-R 33 -R 37 , with an expected value of 1, 823 Da); with chymotrypsin having cleaved at typical major cleavage sites (Trp 11 , Tyr 32 , Phe 35 ). From this, we can conclude that native sublancin has a disulfide bridge between Cys 7 and Cys 36 . To decide upon the location of the second disulfide bridge, we compare sublancin to other type A lantibiotics, and note that formation of a thioether link between Cys 22 and Dhb 19 , to give a Aba 19 -Ala 22 MeLan-type cross-linkage would put a two-residue Gly 20 -Gly 21 sequence in the ring enclosed by the MeLan cross-link, which is similar to the two-residue Pro 9 -Gly 10 sequence enclosed by the Aba 8 -Ala 11 MeLan cross-link in both nisin and subtilin. Moreover, formation of this particular MeLan bond is consistent with the observation that the Cysdehydro partner selection in lantibiotics consistently involves a dehydro residue that is on the N-terminal side of the Cys residue. Assuming that the MeLan that actually forms conforms to these standard patterns, then Cys 22 will react with Dhb 19 , which would require the second disulfide bridge to form between Cys 14 and Cys 29 , as shown in Fig. 7.
Spectrum of Antimicrobial Activity of Sublancin 168 -The lantibiotic family of antimicrobial peptides shows broad spectrum activity against Gram-positive bacteria, and very little activity against Gram-negative bacteria (1, 2). To see if sublan- FIG. 6. Proton NMR specrum of sublancin 168. The part of the spectrum that encompasses portions of the aromatic and vinyl proton regions is shown. The doublet resonance centered at ␦ ϭ 6.2 is identified as Dha 16 for reasons described under "Results." There were no other peaks in the vinyl proton region. The peaks in the aromatic region are assumed to represent the aromatic residues in sublancin.

FIG. 7. The locations of thioether and disulfide bridges in sublancin.
The position of the lanthionine residue and the pattern of disulfide bridge formation were inferred from the NMR spectrum, Nterminal amino acid sequence analysis, amino acid composition analysis, reaction with sulfhydryl-directed agents, proteolytic digestions, and conserved features among lantibiotics as described under "Results." The representation of the sublancin structure as three open circles and a salt-bridge between the N-terminal amino group and the C-terminal carboxyl group is arbitrary, since no information about the secondary structure of sublancin is available. cin conforms to this pattern, we tested the ability of sublancin to inhibit growth of the battery of Gram-positive and Gramnegative bacterial species used by Cleeland and Squires (26) to evaluate the spectrum of activity of antimicrobial agents. As described under "Materials and Methods," the strains were first assayed for susceptibility to sublancin in an agar-diffusion test. Next, the MIC for susceptible strains was determined in liquid culture. The results in Table I show that the antibiotic spectrum of sublancin is consistent with its being a lantibiotic, in that inhibition was observed only among Gram-positive strains of bacteria. However, not all the tested strains of Grampositive bacteria were sensitive, and those that were sensitive varied considerably in their sensitivity to sublancin. Whereas Bacillus megaterium 14581 and Bacillus subtilis 6633 were inhibited by 5 g/ml sublancin, Bacillus cereus T and Staphylococcus aureus 12600 required more than 100 g/ml for complete inhibition to occur.
Effect of Sublancin on Bacterial Spore Outgrowth-The ability of sublancin to inhibit bacterial spore outgrowth was also determined. This is important, because it has been demonstrated that an intact Dha 5 residue in both subtilin (27) and nisin (28) is required in order for inhibition of bacterial spore outgrowth to occur. The fact that an intact Dha 5 residue is unnecessary to inhibit exponentially growing cells established that the mechanism by which subtilin and nisin inhibit spores is different than the mechanism by which they inhibit growing cells (27). The ability of sublancin to inhibit bacterial spore outgrowth was tested using the same methods as for subtilin (27), which included a halo assay in which an agar plate was sprayed with a suspension of B. cereus T spores, and dilutions of sublancin were spotted onto the plate, which was incubated to permit the spores to germinate, outgrow, and grow exponentially to make a confluent lawn. Clear zones occur where the sublancin has been able to inhibit the development of spores into vegetative cells. The other method was to incubate dilutions of sublancin with spores suspended in growth medium, and use phase-contrast microscopy to observe the stage of inhibition. The latter method established that sublancin permits the spores to germinate, to change from the phase-bright dormant state to the germinated phase-dark stage; whereupon further development (swelling, elongation, emergence, divi-sion) are inhibited (data not shown). In this liquid assay, the concentration of sublancin required to inhibit spore outgrowth was about 0.1 g/ml, (27 nM), which is significantly less than the concentration of nisin (40 nM) or subtilin (80 nM) that is required to inhibit outgrowth of these same spores (27,29). It is notable that sublancin is about 1000-fold more effective in inhibiting spore outgrowth than in inhibiting the same cells in exponential growth. The corresponding ratio for subtilin is 30-fold, which means that, although sublancin is slightly better at inhibiting spore outgrowth than is subtilin, subtilin is substantially better at killing the corresponding exponentially growing cells than is sublancin. The requirement for an intact Dha 5 residue in nisin and subtilin in order to exhibit sporostatic activity suggests that the Dha 16 residue of sublancin may also play an important role in the sporostatic processes, but this cannot be certain without additional experiments.
While examining the halos caused by sublancin on the lawns of cells produced by spraying the plate with spores, we noted a discrepancy between the appearance of the halos produced by sublancin and the halos produced by either nisin or subtilin (data not shown). With nisin and subtilin, extended incubation of the plates for several days did not result in any change in the size or appearance of the halos, which remained completely clear. In contrast, incubation of the plates that contained sublancin halos resulted in occasional colonies growing up in within the halos, and a tendency for the surrounding cells to encroach across the perimeter of the clear zone, to cause the size of the halo to diminish slightly with time. The fact that the sublancin halos diminished in size whereas the nisin and subtilin halos did not, is explained by the relatively poor activity of sublancin against vegetative B. cereus T cells, so once the spores had developed into vegetative cells, they were able to encroach into the halo. However, the appearance of colonies within the clear zone suggested something else, which is that a small fraction of the spores that had been inhibited by sublancin at the post-germination stage were able to overcome this inhibition and proceed through outgrowth to the vegetative stage. If so, this is in contrast to nisin or subtilin, both of which have been shown to bind and inhibit spores irreversibly (30). To determine whether sublancin binding and inhibition to germinated spores is reversible, the B. cereus T spores were germinated for 3 h in the presence of various concentrations of sublancin ranging from 0.1 to 100 g/ml, centrifuging the inhibited spores out of the culture, and resuspending the spores in fresh medium without sublancin. The washed spores were then incubated for an additional 2-6 h and examined by phasecontrast microscopy. Whereas most of the spores remained unchanged, a small percentage (about 1%) were clearly proceeding through outgrowth, and eventually reached the vegetative stage and proliferated. This result is consistent with the appearance of colonies within the halos on the plates, where the colonies represent those inhibited spores that recovered after the sublancin had diffused away. The tendency of sublancin to dissociate from the spores may be a consequence of its containing only a single dehydro residue instead of the three that are present in nisin and subtilin, as is considered under "Discussion." The concentration of sublancin used to treat the spores prior to washing had no effect on the outcome of the recovery experiment, with the 0.1 g/ml treatment showing the same effect as the 100 g/ml treatment. This shows that the spore sites to which the sublancin become associated are saturated at very low levels of sublancin.
Stability of Sublancin 168 -Antimicrobial peptides that are chemically stable are better suited for practical applications than are unstable ones. The chemical stability of sublancin was therefore assessed when it was an unpurified component of the culture supernatant, and after it had been purified by HPLC chromatography. Activity was assessed using the agar-plate halo assay against bacterial spores. Culture supernatant stored at room temperature showed little change in halo size during the first 4 days, but showed significant loss after 1 week. Culture supernatants stored at either 4°C or Ϫ20°C showed no change in halo size after 6 months. HPLC-purified sublancin was remarkably stable, and one sample was stored as a 10 mg/ml solution of sublancin in sterile D 2 O, pH 6.5, in an NMR tube for 2 years (protected from light), after which its activity remained undiminished and its NMR profile unchanged (data not shown). Sublancin was stable to a wide range of pH values when either phosphoric acid or ammonium acetate buffers were used to adjust the pH of culture supernatants over a range of 1.5-9.5. The samples were assayed after incubating them for 2 h at 4°C. The pH 9.5 halo was diminished slightly, but the halos produced by the lower pH samples were unchanged. Finally, a sample of the culture supernatant that was autoclaved for 3 min at 121°C showed undiminished activity. These stability characteristics resemble those of nisin, which is very stable at low pH and can survive autoclaving at pH 2.5 without damage, but is fairly unstable above pH 7 (31). However, the ability of sublancin to survive in aqueous solution, at a pH that is nearly neutral, for 2 years without any apparent chemical or biological degradation shows that it is a peptide whose intrinsic stability is extremely high. This extraordinary stability may prove to be a useful characteristic, perhaps enhancing the utility of sublancin in practical applications, or as a model compound whose study may inspire strategies for enhancing the stabilities of non-sublancin antimicrobial peptides.

DISCUSSION
The evidence that sublancin 168 is a lantibiotic is strong. The presublancin gene sequence encodes a serine residue at position 16 of the mature region, which can serve as the precursor to dehydroalanine. Sequential Edman degradation was blocked at position 16, which is characteristic of dehydro residues. As has been demonstrated for dehydro residues in other lantibiotics, the block was alleviated by derivatization with ethanethiol. The NMR spectrum of sublancin showed a doublet in the vinyl proton region of the spectrum, which is consistent with the presence of a dehydroalanine. Sequence analysis of the sublancin gene showed a leader segment with homologies to known type II lantibiotics, including the "double-glycine" sequence motif immediately preceding the cleavage site, indicating that it is probably translocated by a dual-function ABC transporter that both translocates the peptide and proteolytically cleaves the leader segment. The gene immediately downstream from the sublancin gene confirms this, in that it encodes a protein that is homologous to known dual-function transporters, with an identifiable proteolytic domain in addition to a transporter domain. Although the presublancin gene encodes five cysteines, reaction of sublancin with an alkylating agent failed to demonstrate the presence of a free sulfhydryl group, which is consistent with at least one of the cysteines having reacted with a dehydrobutyrine residue to form a ␤-methyllanthionine bridge. The spectrum of activity of sublancin is similar to other lantibiotics in that it is active against a variety of Gram-positive bacteria and inactive against Gramnegative bacteria. It also showed strong inhibition of bacterial spore outgrowth in addition to inhibition of exponentially growing cells, as is seen with both nisin and subtilin. However, unlike nisin and subtilin, washing sublancin-inhibited spores could cause a small percentage (about 1%) of them to proceed through outgrowth and then grow vegetatively, suggesting that the inhibitory effect of sublancin against spores is slightly reversible. For both nisin and subtilin, it has been demonstrated that the mechanism of inhibition of spore outgrowth is different from the inhibition of vegetative growth, in that an intact dehydroalanine is required for spore outgrowth inhibition, but not for vegetative growth inhibition. The fact that sublancin contains only one dehydro residue compared with the three dehydro residues in nisin and subtilin may account for sublancin showing reversibility of inhibition of spore outgrowth, whereas nisin and subtilin do not. It has been suggested that the dehydro residue can react with a nucleophilic target (2,27,29), in which case the larger number of possible attachment points of nisin and subtilin could reduce the likelihood of dissociation and reversal of inhibition, although this explanation is hypothetical.
With this report of the discovery and characterization of sublancin 168, the family of known lantibiotics increases in both size and scope, and there are now over 20 known lantibiotics (2). A striking feature of lantibiotics is their diversity in terms of structure, chemical properties, and biological properties (1,2). The defining characteristic of lantibiotics is that they contain the unusual amino acid lanthionine or ␤-methyllanthionine, which are formed by posttranslational dehydration of serine or threonine, respectively, followed by a Michael-type nucleophilic addition of a cysteine sulfhydryl across the double bond. Because of this mechanism, the presence of the lanthionine requires that the cell possess the machinery to dehydrate serines and/or threonines in addition to the ability to form the thioether linkage. Reflecting this, all the currently known lantibiotics possess at least one lanthionine and one dehydro residue in the mature peptide, although there is little reason to believe that exceptions to this are impossible. Especially notable is that, prior to our discovery of sublancin, all the cysteine residues in known lantibiotics had undergone posttranslational modifications, and never existed as disulfide bridges or free sulfhydryl groups. Sublancin breaks this trend in that only one of its five cysteines has been posttranscriptionally modified, and the other four cysteines instead participate in two disulfide bridges.
Lantibiotics can be considered as a subset of the prodigious number of ribosomally synthesized antimicrobial peptides that have been discovered recently, many of which are produced by eukaryotic organisms, such as the defensins and cecropins (1,32). Mammalian and insect defensins, tachyplesins, and plant thionins all tend to be disulfide-rich, typically containing two or three disulfide bridges within a peptide consisting of 30 -40 amino acid residues (33). The ubiquity and frequency of disulfide bridges argues an important role, perhaps by their ability to impose conformational constraints on the peptide and contribute to conformational and chemical stability. Because the thioether of the lanthionine bridge contains one sulfur atom instead of two, the lanthionine would be expected to be more conformationally constrained than the disulfide. Moreover, the lanthionine is insensitive to redox conditions, while the disulfide is easily broken under mild reducing conditions. In view of the apparent superiority of the lanthionine bridge in terms of conformational and chemical stability, it is somewhat surprising that sublancin contains one lanthionine and two disulfides, instead of the three lanthionines and no disulfides that are found in other lantibiotics such as subtilin, which is produced by B. subtilis ATCC 6633, and nisin. The fact that sublancin possesses both types of linkages suggests that having both types confers a selective advantage. It has been observed that antimicrobial peptides represent a remarkable example of convergent evolution, in which a wide variety of organism types have evolved antimicrobial peptides of common function from very different ancestral origins (33). Perhaps sublancin repre-sents a converging evolutionary branch-point between prokaryotic lantibiotics and eukaryotic defensins, in which sublancin has taken advantage of both types of linkages.
One aspect of the sublancin structure that this work does not resolve is the reason why the molecular mass of sublancin, as determined by ion-spray mass spectroscopy, is 164.48 Da greater than expected from the amino acid composition. This is very likely due to an unidentified posttranslational modification. A precedent for additional posttranslations is seen in subtilin, which is partially succinylated at the N-terminal end, and the extent to which this modification occurs increases as the culture ages (8,34). However, succinylation is not a possible explanation of the molecular size discrepancy in the case of sublancin for two reasons. One is that succinylation blocks the N terminus against Edman degradation (8,34), but sublancin is not blocked. A second reason is that the succinyl group should increase the molecular mass by only 100 Da instead of 164.48 Da. Our results that provided molecular masses of sublancin fragments after chymotrypsin degradation establish that the modification must occur at a residue that lies between Trp 11 and Arg 33 , because the disulfide cross-linked chymotryptic fragment consisting of (G 1 -W 11 -S-S-R 33 -R 37 ) has a molecular mass that is exactly that predicted from the amino acid composition. The molecular mass discrepancy therefore must come from the modification of one (or more) of the residues that resides outside this fragment.