Novel LanT Associated Lantibiotic Clusters Identified by Genome Database Mining

Background Frequent use of antibiotics has led to the emergence of antibiotic resistance in bacteria. Lantibiotic compounds are ribosomally synthesized antimicrobial peptides against which bacteria are not able to produce resistance, hence making them a good alternative to antibiotics. Nisin is the oldest and the most widely used lantibiotic, in food preservation, without having developed any significant resistance against it. Having their antimicrobial potential and a limited number, there is a need to identify novel lantibiotics. Methodology/Findings Identification of novel lantibiotic biosynthetic clusters from an ever increasing database of bacterial genomes, can provide a major lead in this direction. In order to achieve this, a strategy was adopted to identify novel lantibiotic biosynthetic clusters by screening the sequenced genomes for LanT homolog, which is a conserved lantibiotic transporter specific to type IB clusters. This strategy resulted in identification of 54 bacterial strains containing the LanT homologs, which are not the known lantibiotic producers. Of these, 24 strains were subjected to a detailed bioinformatic analysis to identify genes encoding for precursor peptides, modification enzyme, immunity and quorum sensing proteins. Eight clusters having two LanM determinants, similar to haloduracin and lichenicidin were identified, along with 13 clusters having a single LanM determinant as in mersacidin biosynthetic cluster. Besides these, orphan LanT homologs were also identified which might be associated with novel bacteriocins, encoded somewhere else in the genome. Three identified gene clusters had a C39 domain containing LanT transporter, associated with the LanBC proteins and double glycine type precursor peptides, the only known example of such a cluster is that of salivaricin. Conclusion This study led to the identification of 8 novel putative two-component lantibiotic clusters along with 13 having a single LanM and 3 with LanBC genes. Putative lantibiotic clusters identified here hold the potential for the discovery of novel lantibiotic(s).


Introduction
Antibiotic compounds are in use for many decades in treating various bacterial infections and their continuous use has led to the development of drug resistant bacterial strains [1]. MRSA (methicillin resistant Staphylococcus aureus), VRE (vancomycin resistant Enterococci), MDR-TB (multi drug resistant Mycobacterium tuberculosis) and MDR-SP (Streptococcus pneumoniae) are some of the antibiotic resistant bacterial strains that are difficult to treat, therefore posing a need for new antimicrobial compounds. Lantibiotics are emerging as more promising drugs against these bacteria. Nisin was the first ''lantibiotic'' discovered from Lactococcus lactis [2] and is being used in food preservation since then due to its broad activity against many food spoiling and pathogenic bacteria such as Bacillus cereus, Listeria monocytogenes, Enterococci, Staphylococci and Streptococci. More than 40 years of its use in food industry has not led to any significant resistance [3], justifying the need and importance of study on lantibiotics.
Lantibiotics are Lanthionine containing antibiotics having a characteristic lanthionine ring responsible for their stability. The lanthionine ring is formed by dehydration and cyclization in the core peptide region of a 50-100 amino acid residues long precursor peptide from which the N-terminal leader peptide is cleaved off. The serine and threonine residues are dehydrated to form didehydroalanine (Dha) and didehydrobutyrine (Dhb) residues, respectively. Following dehydration, cyclization occurs by linking cysteine to Dha or Dhb residues thus forming a thioether linkage called as lanthionine or b-methyllanthionine. Lantibiotics exhibit their antibacterial activity by capturing lipid II precursor of peptidoglycan, thus, inhibiting cell wall formation [4], or by pore formation which leads to leakage and disruption of the membrane potential [5]. Nisin exhibits both of these activities. The leader peptide of lantibiotics helps in the recognition and interaction of lanthipeptide with different enzymes of its biosynthetic machinery [6,7]. It is the leader peptide which keeps the lantibiotic inactive, when present inside the cell [7,8] and is also required for its extracellular transport [9,10]. Lantibiotic biosyn-thetic machinery is organized into gene clusters, in single or multiple operons encoding the precursor peptide(s), modification enzymes, immunity proteins, a protease and a transporter. Following post-translational modifications, the lanthipeptide is exported outside the cell by an ABC (ATP binding cassette) transporter and the leader peptide is cleaved off by a protease. The cleavage is either concomitant with the export [11] or after the transport, by an extracellular protease that may (e.g. nisin) or may not (e.g. subtilin) be encoded in the cluster [12,13]. Due to a diversity in the lantibiotic compounds, several classification schemes have been proposed [14][15][16][17]. The widely used bacteriocin mining tool, BAGEL2 [18] is based upon the classification given by Cotter et al., who have classified the lantibiotics into two major classes. Class I lantibiotics includes lanthionine containing lantibiotics and class II includes small non-lanthionine bacteriocins. Class I lantibiotics are further classified into three subgroups [19]. Type IA lantibiotics like nisin [20] and microbisporicin [21] are modified by two separate enzymes LanB and LanC. LanB is responsible for the dehydration of selective serine and threonine residues in the core peptide region and LanC for cyclization leading to lanthionine formation. Type IA precursor peptides have a highly conserved proline at -2 position, which is a characteristic of LanBC modified precursor peptides. In Type IB clusters, lanthionine formation is carried out by a single LanM protein, which carries out both the dehydration and cyclization. Type IC lantibiotics are also modified by a single protein, LanL [22]. In most of the class I clusters, besides the precursor peptide and lanthionine synthetases, a two-component response regulator system (LanRK) is present, same as in quorum sensing, for concentration dependent biosynthesis of the lantibiotic [23]. To ensure that the producer organism is not affected by its own lantibiotic, immunity proteins LanI, LanF, LanE and LanG are also present which provide self-resistance. LanI is a lipoprotein and LanFEG are ABC transporters, which pump the lantibiotic outside, thus, preventing from self-damage [24,25].
Type IB lantibiotic cluster can further be either a single or twocomponent lantibiotic encoding cluster. Single component lantibiotics like mersacidin [26] and lacticin 481 [8] have a single antimicrobial peptide. The two-component lantibiotics like haloduracin [27,28] and lichenicidin [29,30] consist of two peptides designated as a and b peptides, which show antimicrobial activity through a synergistic effect. The two peptides are highly antimicrobial together but show little or no activity separately. In case of two-component lantibiotic, where the two precursor peptides show high sequence identity, like in case of cytolysin (i.e. 43%) from Enterococcus faecalis [31], there is a single LanM to process both the precursor peptides ( Table 1). In examples where the precursor peptides show low sequence identity, like 17% in case of lichenicidin [29] and 12% in haloduracin [27], two separate LanM enzymes process the respective precursor peptides. Some unique clusters have also been identified which contain more than two precursor peptides for a single LanM. For, example, upto 29 precursor peptides (with 37-84% sequence identity) have been identified in Prochlorococcus marinus MIT9313, of which 17 were experimentally found to be processed by a single LanM only [32], showing the highly promiscuous nature of the LanM.
Here, we adopted a strategy of using HalT transporter as a query sequence to identify novel type IB lantibiotic clusters by microbial genome database mining. The strategy resulted in identification of 54 strains, from the top 72 hits that were encoding novel HalT homologs in the genome. Out of these 54 strains, we could identify 24 novel lantibiotic clusters that were encoding genes for precursor peptides, modification enzyme, quorum sensing, transport and immunity.

Screening of genome database
Lantibiotic clusters were identified using PSI-BLAST, by taking the C39 peptidase domain containing transporter of haloduracin, HalT (Protein ID: NP_241317.1) as the driver sequence against genomic database on NCBI. PSI-BLAST was carried out using the default parameters, and the top 72 hits were taken for analysis.

Bioinformatic analysis of LanT homolog containing gene clusters
In cases where novel LanT-like genes were identified, the arrangement of adjacent genes was visualized using the genome viewer on NCBI. The individual ORFs obtained were subjected to both, BLAST with known lantibiotic clusters and analysis with the Conserved Domain Database (www.ncbi.nlm.nih.gov/structure/ cdd/cdd.shtml), to identify potential genes involved in lantibiotic production, immunity and quorum sensing. LanRK system was also located using Microbial Signal Transduction database (MiST, http://mistdb.com). Immunity and transporter proteins were analysed with TMHMM (http://www.cbs.dtu.dk/services/ TMHMM) and CELLO [33] to ensure their membrane association. The selected genomes were also analyzed using BAGEL2 [18], a web based bacteriocin mining tool. In clusters, where most of the genes were found to be present but LanA encoding genes were not annotated, intergenic regions were inspected using ORF finder (www.ncbi.nlm.nih.gov). Properties of the identified precursor peptides (Table S1) were analyzed with APD2 [34].

Phylogenetic analysis
Alpha and beta precursor peptides in a putative two-component lantibiotic cluster were identified by multiple sequence alignment with ClustalW2 and their phylogenetic analysis with known alpha and beta precursor peptides, of well-known lantibiotics. The evolutionary history was inferred using the neighbour-joining method in MEGA5 [35].

Results
Using LanT as a mining tool, an in silico approach was followed to identify novel type IB lantibiotic clusters in bacterial genome. LanT is a conserved C39 peptidase domain containing lantibiotic transporter and is an integral part of the Type IB clusters i.e. clusters encoding LanM and double glycine motif containing precursor peptide(s). HalT protein of haloduracin biosynthetic cluster [27,28] was selected as a driver sequence for PSI-BLAST search of the NCBI database. The top 72 hits obtained had a significantly low e-value (,6e-127) including those involved in the production of the lantibiotics like lichenicidin, g_sp_G11MC16, geobacillin, salivaricin ( Table 2), and some putative lantibiotic clusters [18,30,36], thus, proving the effectiveness of our mining approach. Besides the proven ones, 59 novel LanT homologs were identified in 54 bacterial strains (having 34-50% identity with HalT), among the representatives of actinobacteria, firmicutes, cyanobacteria, proteobacteria and chloroflexi, which have not been documented earlier for having any association with the lantibiotic production ( Table 2). Out of the 54 strains, we could identify 24 novel lantibiotic clusters. In rest of the strains, the identified HalT homolog was either orphan or a complete cluster could not be established. All these 24 clusters are discussed here and the identified lantibiotic precursor peptide sequences are given in detail (Table S1) along with their ClustalW alignment ( Fig. 1). They include 13 clusters having a single LanM and 8 encoding two LanM determinants. Three of the LanT homologs have been found to be associated with LanBC proteins, instead of LanM.

Lantibiotic gene clusters identified in actinobacteria
Identification of novel Streptomyces associated lantibiotic gene clusters. Streptomyces are Gram-positive bacteria with a unique capacity for the production of a multitude of varied and complex secondary metabolites. A number of lantibiotics associated with Streptomyces have been identified previously, like type IB lantibiotics ancovenin [37] and duramycin C [38]. We have identified novel clusters, both with single and two LanM determinants, in four representatives of Streptomyces (Fig. 2), which are discussed here.
Streptomyces roseosporus NRRL 11379. S. roseosporus produces the antibiotic daptomycin, which is effective against vancomycin and methicillin-resistant Staphylococcus aureus. The draft genome sequence of S. roseosporus, revealed a cluster ( Fig. 2) with two putative precursor peptides (SrosN1_010100020955 and 60), a single LanM determinant (SrosN1_010100020965, 24% identical to HalM of haloduracin) and putative immunity genes (SrosN1_010100020980, 85 and 90). The precursor peptides had 37% identity, which was intermediate between haloduracin (22%) and cytolysin precursor peptides (,43%), that are processed by two and single LanM, respectively ( Table 1). No other LanM encoding gene was present in the whole genome. In order to have an idea of the relation between the percent identity among the precursor peptides and the number of LanMs that can process them, the putative precursor peptides identified in this biosynthetic cluster were compared with those of the well-known lantibiotics, for their identity score ( Table 1). The analysis indicated that, probably the precursor peptides are processed by a single LanM. In case of prochlorococcins also, a single LanM has been shown to process 17 out of 29 proposed precursor peptides, having 37-84% identity [32]. Therefore, it can be hypothesized that with an intermediate identity of 37%, they would be processed by a single LanM, thus, marking a probable threshold value in the identity of precursor peptides, in light of the examples of experimentally proven lanthipeptides.
Streptomyces hygroscopicus ATCC 53653. S. hygroscopicus, also known as S. himastatinicus ATCC 53653 is a soil isolate from Himachal Pradesh, India and produces the antitumor agent himastatin. It is an aerobic mesophilic terrestrial bacterium. The genomic annotation of S. hygroscopicus showed the LanM determinant (SSOG_05509, 22% identical to HalM) as conserved hypothetical protein, LanT determinant (SSOG_05511) as the lantibiotic transporter, and two putative precursor peptides (SSOG_05508 and 10) as the hypothetical proteins. Since, the sequence identity among these putative precursor peptides was very low i.e. 7% (Table 1), therefore we anticipated that it should have a second LanM also. When the same region was analysed with ORF finder (NCBI), we could identify a second LanM   Fig. 2) with a rather shorter precursor peptide encoding ORF than SSOG_05510, adjoining the LanT homolog. Phylogenetic analysis of both the putative precursor peptides (Fig. 3) showed that these are homologs of the precursor peptides of two-component lantibiotics. Both of these features led us to conclude that the identified cluster might encode for a two-component lantibiotic. Streptomyces bingchenggensis BCW-1. S. bingchenggensis is a soil bacterium isolated from Harbin, China. It produces milbemycins (commercially important anthelmintic macrolide compounds), other insecticidal antibiotics, and a few cyclic pentapeptides and is thus interesting for its secondary metabolite producing capabilities [39]. Our genome analysis of S. bingchenggensis resulted in the identification of a cluster (Fig. 2) consisting of a LanM determinant (SBI_06988; 18% identical to HalM) and two putative precursor peptides (SBI_06989 and 90), that were originally annotated as hypothetical proteins, and a LanT homolog (SBI_06987) of ,1600 amino acid residues, annotated as cytolysin B transport protein. The size of LanT appeared large enough, for it to be just a transport protein ( Table 3). Also the precursor peptides were found to have an identity of 20% among themselves, which being very low ( Table 1), suggested that the cluster should have two LanMs [27]. Phylogenetic analysis (Fig. 3) also suggested that this cluster might encode for a two-component lantibiotic cluster. Using CDD, we could identify a second LanM, fused with the C39 protease and the ABC transporter domain, in this large protein sequence (Fig. 4). Though the cluster was also analyzed by BAGEL2, it predicted this single ORF as LanM and could not identify the hidden transporter. It will be interesting to know with wet lab experiments, if this tri-domain protein is operative in the formation of an active lantibiotic. When this long putative transporter sequence was subjected to BLAST, no other similar example could be identified on NCBI Entrez database. The second LanM could only be identified by using a consolidation of BLAST, CDD and BAGEL2, as it was fused with the LanT.
Streptomyces viridochromogenes DSM 40736. S. viridochromogenes is a producer of avilamycin A, an oligosaccharide antibiotic. Its genome analysis revealed a lantibiotic cluster (Fig. 2) with two putative precursor peptides (SSQG_04192 and 93) and two LanM determinants (SSQG_04194 and 96, showing 18% and 17% identity with HalM), besides the LanT homolog (SSQG_04195). The two LanM genes were found to be flanking the transporter. The cluster might encode for a two-component lantibiotic, as suggested by the phylogenetic analysis of the precursor peptides (Fig. 3). Downstream to the cluster, many transcriptional regulators (TR) were present next to the hypothetical protein (SSQG_04197) and the flavin utilizing monoxygenase gene (SSQG_04198).
Mycobacterium tusciae JS617. M. tusciae is a slow growing scotochromogenic mycobacterium that has been found in tapwater. This organism was isolated from the lymph node of an immunocompromised patient and from the respiratory sample of cystic fibrosis patient. Unexpectedly, the cluster identified in M. tusciae turned out to be a type IA cluster instead of a type IB cluster (Fig. 2). It had a LanT homolog (MyctuDRAFT_5975), a LanB determinant (MyctuDRAFT_5976) and a putative LanA precursor peptide (MyctuDRAFT_5977). The core-peptide region of the identified precursor peptide had a high amount of glycines (,43%) with serine and threonine residues, but no cysteine at all. Interestingly, the LanC cyclase, which links the cysteine to Dha and Dhb residues for cyclization, was neither found anywhere near the cluster and nor in the bacterial genome.

Lantibiotic gene clusters identified in firmicutes
Identification of novel Bacillus cereus associated lantibiotic clusters. Bacillus species are rod-shaped, endospore-forming aerobic or facultatively anaerobic, Gram-positive bacteria that are ubiquitous in nature. The many species of the genus exhibit a wide range of physiologic abilities that allow them to live in every natural environment. Many lantibiotics have been discovered from the phyla including lichenicidin, haloduracin, amylolysin etc.
Bacillus cereus FRI-35 plasmid p03. Genome analysis of FRI-35 revealed the presence of a two-component lantibiotic cluster (Fig. 5) with two LanM determinants (BCK_27583 and BCK_27638, 33% and 25% identical to HalM, respectively) and as many as eight putative double glycine motif containing precursor peptides, with 11-68% identity among them ( Table 1). The whole biosynthetic cluster was present on the plasmid p03 (Sequence: NC_018499.1), of its four plasmids. Eight precursor peptide encoding genes were identified, which included 3 identical copies each of a (BCK_27608, 13 and 18) and b peptides (BCK_27623, 28 and 33) and two more peptides (BCK_27598 and 03), that fall within the b peptides in the phylogenetic analysis (Fig. 3). All the three immunity proteins, LanF (BCK_27648), LanE (BCK_27653) and LanG (BCK_27658) with the LanT homolog (BCK_27663) and an additional LanP, S8 peptidase were also present in the cluster (BCK_27643).
Bacillus cereus MSX-A1. Genome analysis of MSX-A1 led to the identification of a lantibiotic cluster (Fig. 5) having four almost identical precursor peptides (80-95% identity; II5_05392, 93, 94 and 95), a LanM determinant (II5_05396, having 27% Figure 1. Diverse structures of identified putative lantibiotic precursor peptides. Thirty nine selected lantibiotic precursor peptides shown in ClustalW alignment using MEGA5 [35]. The locus tag is given on the left of the sequence and the amino acid position is given on the right. The cleavage site of the leader peptides is indicated by an arrow. The precursor peptide D187_01267 from C. fuscus DSM 2262 has eight cysteine residues, the highest among all the identified precursor peptides. doi:10.1371/journal.pone.0091352.g001 identity with HalM), a LanT homolog (II5_05402) and a LanP (II5_05403). A LanR determinant (II5_05397) of two-component response regulatory system was also present. The cluster was intermediated by an unrelated gene annotated as hypothetical protein, the trans-isoprenyl diphosphate synthase (II5_05400). Two putative ABC transporter genes (II5_05404 and 05) were also found downstream to the cluster next to LanP, which might have a role in immunity.
Bacillus cereus SJ1. Lantibiotic biosynthetic cluster in SJ1 (Fig. 5) was found to contain an apparently frameshifted LanM determinant (BCSJ1_09308, 13 and 18, showing 19%, 28% and 22% identity with HalM), alongwith three putative precursor peptides (BCSJ1_09323, 28, and 33), out of which only BCSJ1_09328 contain a cysteine residue and the other two, although had a double glycine motif, but were devoid of any cysteine ( Table 1). All these three putative precursor peptides had an identical leader sequence, with variation in the core-peptide region only. Similar to MSX-A1, there was an additional LanP determinant (BCSJ1_09343) present in the cluster, near the LanT homolog (BCSJ1_09338). A cytolysin immunity protein (CylI; BCSJ1_09298) was also found to be present in the cluster.
Bacillus cereus VD045. Bioinformatic analysis of the draft sequence of VD045 revealed the presence of two putative lantibiotic biosynthetic clusters (Fig. 5), present in the vicinity, on the three contigs (conti1.34, 1.35 and 1.36), with their dedicated transporters and immunity genes. First cluster had a single putative precursor peptide (IIE_05053), a LanM (IIE_05054, 19% identical to HalM) and a LanT determinant (IIE_05055), with the putative immunity proteins LanF (IIE_05056), LanE (IIE_05057) and MutE homolog (IIE_05058). In the second cluster, there were six putative precursor peptides (IIE_05066, 67, 68, 69, 70 and 71) with almost identical leader  (Fig. 3) of the precursor peptides suggested that these clusters might encode a putative twocomponent lantibiotic. LanA1 and LanA2 represents alpha and beta precursors. In Mycobacterium tusciae, the double glycine motif containing precursor peptide was found to be encoded with the LanB determinant, enzyme for dehydration of type IA precursor peptides. HP -Hypothetical Protein; TR -Transcriptional Regulator; immunity genes which cannot be assigned as LanF/E/G are shown in red color. doi:10.1371/journal.pone.0091352.g002 sequences (74-97% identity), a LanM (IIE_05073, 24% identical to HalM), a LanT (IIE_05074) and a LanP (IIE_05075) determinant. An intermediating NADPH dependent FMN reductase protein was also present in the this cluster.
Bacillus cereus VD166. VD166 genome was found encoding a lantibiotic cluster (Fig. 5) comprising of three putative precursor peptides (IK9_05148, 49 and 50), a LanM determinant (IK9_05147, 24% identical to HalM) and a LanT homolog (IK9_05143). Using BAGEL2, only a single putative immunity protein, LanF (IK9_05146) could be identified near the cluster. Since, the immunity genes are normally clustered together, therefore, when the two nearby ORFs (IK9_05144 and 45) were analyzed by TMHMM, CELLO and CDD, they turned out to be transmembrane proteins. Also, the size of these proteins was in the range of the immunity proteins ( Table 3) which led us to speculate that these ORFs might be encoding for the two immunity proteins, LanG and LanE.

Identification of novel lantibiotic clusters in other firmicutes.
Ruminococcus flavefaciens FD-1. R. flavefaciens is a Gram-positive cocci in the order 'Clostridiales' of firmicutes, which is a predominant anaerobic cellulolytic rumen bacterium. It was isolated from bolus containing ruminal microorganisms, which improve rumen function in calves. The culture has the largest known number of fibre-degrading enzymes. In R. flavefaciens genome, a cluster ( Fig. 6) with as many as 12 putative a and b precursor peptides (Fig. 3)  Two more immunity proteins (Rflaf_010100011249 and 54) were identified in this cluster, as that were found in B. cereus VD166, discussed earlier.
Caldicellulosiruptor bescii DSM 6725. C. bescii, also known as Anaerocellum thermophilum DSM 6725 is a thermophilic, cellulolytic, anaerobic bacterium isolated from a hot spring on the Kamchatka peninsula in Russia. It is the most thermophilic cellulose-degrading organism known [40]. Genome analysis of C. bescii revealed the presence of a lantibiotic cluster (Fig. 6), with a single putative precursor peptide (Athe_1106), a LanM determinant (Athe_1107, having 26% identity with HalM), a LanT homolog (Athe_1112) and a LanF determinant (Athe_1109). Two more ORFs (Athe_1110 and 11) next to the LanF were identified as immunity proteins. Two N-acetyltransferases (Athe_1105 and 13) were also found at immediate upstream and downstream of the cluster. Since, the lantibiotic paenibacillin has been shown to be acetylated by an acetyltransferase [41], these N-acetyltransferases probably might be playing a similar role here.
Staphylococcus epidermidis plasmid SAP_106A. S. epidermidis is a Gram-positive bacterium, belonging to the genus Staphylococcus. It is part of the human skin flora (commensal), and is also found in the mucous membranes of animals. In S. epidermidis genome, a cluster was identified on plasmid SAP106A (Fig. 6), containing a single putative precursor peptide (SAP_106A002), a LanM determinant (SAP_106A003, having 22% identity with HalM) and an additional LanP protein, besides the C39 protease of LanT homolog (SAP_106A001). A LanF determinant (SAP_106A004) and a probable immunity protein encoding gene (SAP_106A005) was also identified in the cluster.
Clostridium perfringens B str. ATCC 3626. C. perfringens is a rod-shaped, motile mesophilic pathogen belonging to firmicutes, which causes dysenteria, enterocolitis, enterotoxemia, food poisoning and gas gangrene. It was isolated from the intestinal contents of a lamb. Its genome analysis revealed a lantibiotic biosynthetic cluster (Fig. 6) with two putative precursor peptides (AC1_A0448 and 49), a LanM determinant (AC1_A0450) besides the LanT homolog (AC1_A0446). An atypical NADPH-dependent FMN reductase gene (AC1_A0447), as found in some of the other clusters, discussed elsewhere, was also present.
Bacillus sp. 7_6_55CFAA_CT2. This culture has been isolated from the healthy gut biopsy tissue from a patient with Crohn's disease. Its genome was found to contain a cluster (Fig. 6) with three putative precursor peptides (HMPREF1014_04854, 55 and 56) and two LanM determinants (HMPREF1014_04853 and 57, 26% and 33% identical to HalM, respectively). Phylogenetic analysis suggested that the cluster might encode for a putative twocomponent lantibiotic (Fig. 3). An S8 peptidase, LanP (HMPREF1014_04862) was also found next to the three hypothetical proteins, besides the bifunctional LanT homolog (HMPREF1014_04858).

Lantibiotic gene clusters identified in cyanobacteria.
Chamaesiphon minutus PCC 6605. C. minutus is a free living mesophilic cyanobacteria belonging to order Chroococcales, which was isolated from freshwater at Berkeley, USA. Its genome was found to encode a putative two-component lantibiotic cluster (Fig. 7) with three putative precursor peptides (Cha6605_2219, 21 and 23) and two LanM determinants (Cha6605_2218 and 22, 29% and 24% identical to HalM), besides the identified LanT homolog (Cha6605_2216).

Lantibiotic gene clusters identified in proteobacteria.
Myxococcus xanthus DK 1622. M. xanthus is an aerobic mesophilic terrestrial proteobacteria. It has a large genome (9.14 Mb) containing more than 1500 gene duplications. These gene duplications more commonly involve genes used in cell to cell signaling, molecule sensing and transcription control. A lantibiotic cluster (Fig. 7) with two putative precursor peptides (MXAN_2855 and 56), a LanM determinant (MXAN_2857, having 27% identity with HalM) and the LanT (MXAN_2853) homolog was identified in its genome. Two more peptides, Table 3. Approximate size of the lantibiotic biosynthetic proteins.

LanA 50-100
LanB 600-1000 LanC ,400 LanT (ABC Transporter) ,600 LanT (ABC Transporter+ C39 Peptidase) ,700 LanT*(ABC transporter+ C39 Peptidase+LanM) ,1600 LanM 900-1000  Cystobacter fuscus DSM 2262. C. fuscus is a Gram-negative aerobic bacteria of the order Myxococcales. Similar to M. tusciae JS617, its genome analysis also identified a putative type IA lantibiotic cluster (Fig. 7) comprising of a putative precursor peptide (D187_01267), a LanB (D187_01271) and a LanC determinant (D187_01270) associated with the LanT homolog (D187_01268). Two more LanT homologs without any associated modification enzymes were also identified elsewhere in the genome ( Table 2), which might also be involved in processing of the double glycine motif containing precursor peptides. The precursor peptides processed by LanBC proteins generally have a proline at -2 position, but here instead, double glycine motif was present. Noteworthy here is the presence of as many as eight cysteine residues of the total sixteen modifiable residues in the core peptide region (Fig. 1), maximum in any lanthipeptide discovered till date. These eight cysteine residues may contribute to form eight lanthionine rings, thus, probably making it more stable.
Corallococcus coralloides DSM 2259. C. coralloides is an aerobic terrestrial mesophilic Gram-negative bacteria of phylum proteobacteria. Genome analysis of C. coralloides also revealed the presence of a novel type IA cluster associated with the C39 containing LanT protein (Fig. 7). Three modification enzymes, LanB (COCOR_05225), LanC (COCOR_05228) and LanO (COCOR_05227) determinants were present in the cluster, along with the double glycine motif containing precursor peptide (COCOR_05226), as was in M. tusciae and C. fuscus discussed above.
Lantibiotic gene cluster identified in chloroflexi.
Ktedonobacter racemifer DSM 44963. K. racemifer is an aerobic, filamentous, non-motile, spore-forming Gram-positive heterotroph, which was isolated from a soil sample in Italy and represents a new phylogenetic class Ktedonobacteria, though it shares some morphological features with the actinobacteria. This culture is of special interest because it was the first cultivated representative of a deep branching unclassified lineage of otherwise uncultivated environmental phylotypes, tentatively located within the phylum 'chloroflexi' [42]. The draft genome sequence of K. racemifer revealed the presence of a lantibiotic biosynthetic cluster ( Fig. 7) with as many as ten putative precursor peptides (Krac_11033, 37, 38, 39, 40, 41, 42, 46, 47 and 50), two LanM determinants (Krac_11032 and 36, showing 30% and 27% identity with HalM) and the LanT homolog (Krac_11043). A second LanT homolog (Krac_11909; not shown) not associated with any lantibiotic precursor peptide was also present in the genome. Since there are two LanMs present in the cluster, so it was expected that the phylogenetic analysis would distribute the putative precursor peptides among the two clades of a and b precursors. Instead, the precursor peptides formed a separate group, unrelated with a and b precursor peptides (Fig. 8), suggesting that the cluster is probably not a two-component lantibiotic cluster.

Discussion
Genome mining is being extensively and successfully done these days using bioinformatic tools, to identify novel genes and gene clusters. Recently, many lantibiotic biosynthetic clusters have been identified using microbial genome database mining approach, by taking advantage of the conserved nature of the genes encoded in the lantibiotic clusters. For example, LanC has been used as a query for the identification of type IA lantibiotic clusters [43], LanM for type IB clusters [30,44], C39 protease domain for double glycine motif containing precursor peptides [45] and recently HylD and LanT was used for the identification of bacteriocins in cyanobacteria [44]. In the present study, a similar approach was followed to identify novel lantibiotic biosynthetic clusters. HalT transporter of haloduracin biosynthetic cluster was taken as a query to screen the bacterial genomes on NCBI Entrez. HalT has a N-terminal C39 protease and a C-terminal ABCtransporter, for haloduracin precursor peptides. This strategy resulted in the identification of 59 novel LanT homologs in 54 bacterial genomes (more than one in a single genome; Table 2). The detailed characterization of all these LanT homologs and the nearby genome sequence, led to the identification of 24 novel lantibiotic clusters, described here. The bacteria encoding these clusters were the representatives of actinobacteria, firmicutes, cyanobacteria, proteobacteria and chloroflexi.
Many examples of actinobacteria have been reported earlier for producing lantibiotics [46]. This study led to the identification of five more actinobacteria (Fig. 2) encoding lantibiotic clusters namely S. hygroscopicus ATCC 53653, S. bingchenggensis BCW-1, S. viridochromogenes DSM 40736, S. roseosporus NRRL 11379 and Mycobacterium tusciae JS617. The three actinobacteria, ATCC 53653, BCW-1 and DSM 40736 were found encoding a putative two-component lantibiotic cluster, as suggested by the phylogenetic analysis of the precursor peptides (Fig. 3) and the presence of two LanMs. In ATCC 53653, the ORF encoding a second LanM was missed out in the genomic annotation while in BCW-1, it was hidden and fused with the LanT homolog (Fig. 4). It was only the sequence identity among the precursor peptides (7% and 20%), that gave us a clue for the possible presence of a second LanM. To draw a relation between the sequence identity among the precursor peptides and number of LanMs that are required for the processing, examples of already known and putative lantibiotic clusters were taken and analysed ( Table 1). We could conclude therefrom that for upto 20% sequence identity, two LanMs are required for processing of the precursor peptides and above 37% identity, a single LanM is sufficient.
A majority of known lantibiotics are produced by firmicutes [46]. In this study, we have identified seven strains of Bacillus cereus (Fig. 5) namely FRI-35, VD107, MSX-A1, SJ1, VD045, VD166 and Rock1-3, encoding novel lantibiotic clusters with a diversity in the sequence of the precursor peptides. Earlier, Xiong et al had identified a cluster containing six identical precursor peptides in another strain of B. cereus Q1, which were named as cereicidins [36]. Other firmicutes that were identified included the anaerobic bacteria: R. flavifaciens FD-1 and C. bescii DSM 6725; facultative anaerobe: S. epidermidis and aerobic bacteria: C. perfringens, Bacillus Str. ATCC 3626, Bacillis sp. 7_6_55CFAA_CT2. The strains FRI-35, FD-1 and 7_6_55CFAA_CT2 were found encoding a putative two-component lantibiotic cluster. FRI-35 was found to be encoding 8 and FD-1 as 12 precursor peptides, this being maximum in any firmicute identified till date, although multiple precursor peptides encoding clusters have been identified earlier in cyanobacteria [32,44] and chloroflexi [30]. The lantibiotic clusters Figure 5. Cluster organization of the putative lantibiotic biosynthetic genes identified in the strains of Bacillus cereus. Cluster organization representing a diversity among subspecies of B. cereus, which belongs to Firmicutes. Only the strain, B.cereus FRI-35 was found to encode a putative two-component lantibiotic cluster (Fig. 3). Annotation with connected lines represent identical precursor peptides. doi:10.1371/journal.pone.0091352.g005 Figure 6. Cluster organization of the putative lantibiotic biosynthetic genes identified in firmicutes other than B. cereus. Lantibiotic clusters identified in other Firmicutes besides the strains of Bacillus cereus. The clusters identified in R. flavefaciens FD-1 and Bacillus sp. 7_6_55CFAA_CT2 encode two LanM determinants and phylogenetic analysis of the identified precursor peptides suggested that these clusters might encode for a two-component lantibiotic (Fig. 3). doi:10.1371/journal.pone.0091352.g006 identified in the firmicutes, FRI-35 and S. epidermidis were found on plasmids, like the previously characterized lantibiotics cytolysin, nisin, salivaricin etc.
Nine of the previously genetically characterized type IB clusters [46] and four of the clusters identified here in actinobacteria, completely rely on a bifunctional LanT. However, in six of the identified firmicutes, an additional LanP protease was found encoded in the cluster, namely in FRI-35, MSX-A1, SJ1, VD045 (Fig. 5), 7_6_55CFAA_CT2 and S. epidermidis (Fig. 6). The presence of two proteases in a single cluster suggests that there might be more than one cleavage site on the same precursor peptide [44]. While LanT processes and transports a precursor peptide with double glycine motif, the LanP is a membrane bound extracellular protease [20], which acts on the N-terminus of the P(QRS) motif of the precursor peptides. The latter motifs were found present in the leader region of the precursor peptides (Text S1) before the double glycine motif in the strains FRI-35, MSX-A1 and SJ1 while in VD045, it was present in the core peptide region. In 7_6_55CFAA_CT2 and S. epidermidis, the motif was not identified. So, it will be interesting to know how the presence of both the proteases fit into the sequence of events in lantibiotic biosynthesis. In 2009, Begley et al reported the presence of lantibiotic clusters by in-silico analysis in the representatives of cyanobacteria, chloroflexi and proteobacteria, which had not been associated with lantibiotic production earlier [30]. Later, it was experimentally confirmed in cyanobacteria [32]. Thereafter, Wang et al reported the widespread occurrence of bacteriocin gene clusters in cyanobacteria, suggesting the phyla to be a prolific source of bacteriocins [44]. Here, we have added two more representatives of cyanobacteria; Synechocystis sp. PCC 7509 and a new genus Chamaesiphon minutus PCC 6605 (Fig. 7), both encoding a putative two-component lantibiotic cluster (Fig. 3) and one more genus to chloroflexi, the Ktedonobacter racemifer encoding a lantibiotic cluster.
To the best of our knowledge, salivaricin is the only known example where the double glycine type precursor peptide is modified by LanBC (instead of LanM), processed and transported by a bifunctional LanT [47]. In our study, similar biosynthetic clusters were identified in an actinomycete, M. tusciae JS617 (Fig. 2) and in proteobacteria, C. coralloides DSM 2259 and C. fuscus DSM 2262 (Fig. 7). LanBC enzymes usually process the precursor peptides, which have a conserved proline at -2 position and FNLD motif around the position -20 and -15 in the leader, thus playing a crucial role in the post-translational modification [48]. Experimental evidences are therefore required to conclude that without the presence of these crucial motifs, as has been noticed in the above three clusters, the identified precursor peptides could still undergo post-translational modification by LanBC.
Many different kinds of post-translational modifications, other than the lanthionine formation have been reported recently in bacteriocins [41,49]. In the present study, besides the normal genes present in the lantibiotic cluster, some additional atypical genes were found like NADPH dependent FMN reductase in SJ1, VD045, C. bescii DSM 6725 and C. perfringens B str. ATCC 3626 and N-acetyltransferases were found in close proximity to the clusters in DSM 6725 and ATCC 53653. These atypical genes might be involved in additional post-translational modifications of the lantibiotics. N-terminal acetylation of proteins is believed to protect the proteins from degradation [50] and, thus, might provide an additional stability to the encoded lantibiotic, besides that by the lanthionine group.
Many of the previous, putatively [30] and experimentally characterized lantibiotics including nisin have been identified on transmissible genetic elements. Here also, the clusters identified in B. cereus VD045, B. cereus Rock1-3 and K. racemifer DSM 44963 were present in close proximity with the transposable genetic elements, which might be involved in mobilizing the cluster to other species and also for their mobility within the genome.
From several studies done on lantibiotics, it has been felt that the search for smaller antimicrobial peptides is difficult because of their small size and low homology, and are therefore missed out during the conventional genomic annotation. However, the relatively larger transporter gene of the lantibiotic biosynthetic cluster i.e. LanT is rarely unannotated and can be identified by BLAST search. The in silico analysis of the nearby region for the lantibiotic modification genes can then lead to the identification of the precursor peptides. Thus, identifying lanthipeptides using conserved regions of the larger proteins of the lantibiotic biosynthetic cluster proves out to be a better approach. The putative lantibiotic clusters identified here using the above said strategy is a major expansion to the already reported putative lantibiotics, and if taken for focussed wet lab experiments, it might expand our library of antimicrobial peptides against the drug resistant pathogens. The identified bacteria can be taken up for in vivo production of the lantibiotic or an in vitro reconstitution approach can be adopted. Some findings like the presence of LanBC modification enzymes for processing peptides with double glycine motif, fusion of the modification enzyme with the transporter and presence of unrelated genes like the FMN reductases and N-acetyltransferases in the lantibiotic biosysnthetic cluster organization has further pointed towards the need of experimental characterization of these lantibiotic biosynthetic systems. The use of BAGEL2 poses limitations in identification of the whole biosynthetic cluster that can be overcome by a consolidation of well established analytical bioinformatic tools used here. In nutshell, our results provide a higher level of confidence in the novel identified clusters, to proceed for wet lab experiments, for the discovery of novel lantibiotic(s).

Supporting Information
Table S1 List of all the putative lantibiotic precursor peptides identified in this study.

(XLSX)
Text S1 Processing sites present in the precursor peptides, identified in the clusters encoding two proteases, i.e. the C39 protease of the LanT transporter and the S8 peptidase, LanP.