Abstract
The genome sequencing and mining of microorganisms from unexplored and extreme environments has become important in the process of identifying novel biosynthetic pathways. In the present study, the biosynthetic potential of Paenibacillus sp. strains 23TSA30-6 and 28ISP30-2 was investigated. Both strains were isolated from the deep oligotrophic Krubera-Voronja Cave and were found to be highly active against both Gram-positive and Gram-negative bacteria. Genome mining revealed a high number of biosynthetic gene clusters in the cave strains: 21 for strain 23TSA30-6 and 19 for strain 28ISP30-2. Single clusters encoding the biosynthesis of phosphonate, terpene, and siderophore, as well as a single trans-AT polyketide synthase/non-ribosomal peptide synthetase, were identified in both genomes. The most numerous clusters were assigned to the biosynthetic pathways of non-ribosomal peptides and ribosomally synthesized and post-translationally modified peptides. Although four non-ribosomal peptide synthetase gene clusters were predicted to be involved in the biosynthesis of known compounds (fusaricidin, polymyxin B, colistin A, and tridecaptin) of the genus Paenibacillus, discrepancies in the structural organization of the clusters, as well as in the substrate specificity of some adenylation domains, were detected between the reference pathways and the clusters in our study. Among the clusters involved in the biosynthesis of ribosomally synthesized peptides, only one was predicted to be involved in the biosynthesis of a known compound: paenicidin B. Most biosynthetic gene clusters in the genomes of the cave strains showed a low similarity with the reference pathways and were predicted to represent novel biosynthetic pathways. In addition, the cave strains differed in their potential to encode the biosynthesis of a few unique, previously unknown compounds (class II lanthipeptides and three non-ribosomal peptides). The phenotypic characterization of proteinaceous and volatile compounds produced by strains 23TSA30-6 and 28ISP30-2 was also performed, and the results were compared with those of genome mining.