Identification of a novel two-peptide lantibiotic, lichenicidin, following rational genome mining for LanM proteins.

Lantibiotics are ribosomally synthesized peptide antimicrobials which contain considerable posttranslational modifications. Given their usually broad host range and their highly stable structures, there have been renewed attempts to identify and characterize novel members of the lantibiotic family in recent years. The increasing availability of bacterial genome sequences means that in addition to traditional microbiological approaches, in silico screening strategies may now be employed to the same end. Taking advantage of the highly conserved nature of lantibiotic biosynthetic enzymes, we screened publicly available microbial genome sequences for genes encoding LanM proteins, which are required for the posttranslational modification of type 2 lantibiotics. By using this approach, 89 LanM homologs, including 61 in strains not known to be lantibiotic producers, were identified. Of these strains, five (Streptococcus pneumoniae SP23-BS72, Bacillus licheniformis ATCC 14580, Anabaena variabilis ATCC 29413, Geobacillus thermodenitrificans NG80-2, and Herpetosiphon aurantiacus ATCC 23779) were subjected to a more detailed bioinformatic analysis. Four of the strains possessed genes potentially encoding a structural peptide in close proximity to the lanM determinants, while two, S. pneumoniae SP23-BS72 and B. licheniformis ATCC 14580, possess two potential structural genes. The B. licheniformis strain was selected for a proof-of-concept exercise, which established that a two-peptide lantibiotic, lichenicidin, which exhibits antimicrobial activity against all Listeria monocytogenes, methicillin-resistant Staphylococcus aureus, and vancomycin-resistant enterococcus strains tested, was indeed produced, thereby confirming the benefits of such a bioinformatic approach when screening for novel lantibiotic producers.