Bacteriocins

The bacteriocin database has been updated and now contains almost 500 RiPPs (class 1, see Supplementary Figure S1), 230 unmodified bacteriocins (class 2) and 90 large (>10 kD) bacteriocins (class 3). Most records contain a link to NCBI or UniProt. Next to literature in general specific resources have been used to update our records such as RippMiner (11 (link)) and the MIBiG data repository (14 (link)). The database is available on http://bagel4.molgenrug.nl.

In addition to the secondary metabolite cluster types supported in the original release of antiSMASH (type I, II and III polyketides, non-ribosomal peptides, terpenes, lantipeptides, bacteriocins, aminoglycosides/aminocyclitols, β-lactams, aminocoumarins, indoles, butyrolactones, ectoines, siderophores, phosphoglycolipids, melanins and a generic class of clusters encoding unusual secondary metabolite biosynthesis genes), version 2.0 adds support for oligosaccharide antibiotics, phenazines, thiopeptides, homoserine lactones, phosphonates and furans. The cluster detection uses the same pHMM rule-based approach as the initial release (17 (link)): in short, the pHMMs are used to detect signature proteins or protein domains that are characteristic for the respective secondary metabolite biosynthetic pathway. Some pHMMs were obtained from PFAM or TIGRFAM. If no suitable pHMMs were available from these databases, custom pHMMs were constructed based on manually curated seed alignments (Supplementary Table S1). These are composed of protein sequences of experimentally characterized biosynthetic enzymes described in literature, as well as their close homologs found in gene clusters from the same type. The models were curated by manually inspecting the output of searches against the non-redundant (nr) database of protein sequences. The seed alignments are available online at http://antismash.secondarymetabolites.org/download.html#extras. After scanning the genome with the pHMM library, antiSMASH evaluates all hits using a set of rules (Supplementary Table S2) that describe the different cluster types. Unlike the hard-coded rules in the initial release of antiSMASH, the detection rules and profile lists are now located in editable TXT files, making it easy for users to add and modify cluster rules in the stand-alone version, e.g. to accommodate newly discovered or proprietary compound classes without code changes. The results of gene cluster predictions by antiSMASH are continuously checked on new data arising from research performed throughout the natural products community, and pHMMs and their cut-offs are regularly updated when either false positives or false negatives become apparent.
The profile-based detection of secondary metabolite clusters has now been augmented by a tighter integration of the generalized PFAM (22 (link)) domain-based ClusterFinder algorithm (Cimermancic et al., in preparation) already included in version 1.0 of antiSMASH. This algorithm performs probabilistic inference of gene clusters by identifying genomic regions with unusually high frequencies of secondary metabolism-associated PFAM domains, and it was designed to detect ‘classical’ as well as less typical and even novel classes of secondary metabolite gene clusters. While antiSMASH 1.0 only generated the output of this algorithm in a static image, version 2.0 displays these additional putative gene clusters along with the other gene clusters in the HTML output. A key advantage of this is that these putative gene clusters will now also be included in the subsequent (Sub)ClusterBlast analyses.

Using the HMMer3 tool (http://hmmer.janelia.org/), the amino acid sequence translations of all protein-encoding genes are searched with profile Hidden Markov Models (pHMMs) based on multiple sequence alignments of experimentally characterized signature proteins or protein domains (proteins, protein subtypes or protein domains which are each exclusively present in a certain type of biosynthetic gene clusters). Using both existing pHMMs (5 (link),11–13 ) and new pHMMs from seed alignments, we constructed a library of models specific for type I, II and III PK, NRP, terpene, lantibiotic, bacteriocin, aminoglycoside/aminocyclitol, beta-lactam, aminocoumarin, indole, butyrolactone, ectoine, siderophore, phosphoglycolipid, melanin and aminoglycoside biosynthesis signature genes. Additionally, we constructed a number of pHMMs specific for false positives, such as the different types of fatty acid synthases which show homology to PKSs. The final detection stage operates a filtering logic of negative and positive pHMMs and their cut-offs. The logic is based on knowledge of the minimal core components of each gene cluster type taken from the scientific literature. The cut-offs were determined by manual studies of the pHMM results when run against the NCBI non-redundant (nr) protein sequence database (ftp://ftp.ncbi.nlm.nih.gov/blast/db). All technical details on the pHMM library and the detection rules are available in Supplementary Tables S1 and S2, respectively.
Gene clusters are defined by locating clusters of signature gene pHMM hits spaced within <10 kb mutual distance. To include flanking accessory genes, gene clusters are extended by 5, 10 or 20 kb on each side of the last signature gene pHMM hit, depending on the gene cluster type detected. As a consequence of this greedy methodology, gene clusters that are spaced very closely together may be merged into ‘superclusters’. These gene clusters are indicated in the output as ‘hybrid clusters’; they may either represent a single gene cluster which produces a hybrid compound that combines two or more chemical scaffold types, or they may represent two separate gene clusters which just happen to be spaced very closely together.

The topical cream tested in this study contained bacteriocins from B. subtilis (1% weight/weight; Biodue S.p.A., Tavarnelle Val Di Pesa, Italy) in a moisturizer base.

The microbiological study was aimed at assessing whether topical application of the cream containing bacteriocins from B. subtilis was able to promote S. aureus decolonization in acne areas. A total of 12 patients (6 males and 6 females; age range: 22–35 years) were recruited from Italian private practices. The sample size for this pilot study was chosen based on feasibility and costs. Skin swab specimens from facial acne areas before and after 60 days of topical treatment (paired samples) were obtained by trained personnel. Quantitative real-time PCR for absolute S. aureus quantitation was carried out as previously described [17 (link)]. In brief, total nucleic acid from acneic skin swab specimens was amplified and the DNA concentration was quantified using the CFX96 Real-Time System (Bio-Rad, Hercules, CA, USA). Standard curves for the absolute abundance values of S. aureus in collected specimens were constructed using seven 10-fold dilutions (from 4 ng/µl to 4 fg/µl) of S. aureus USA300 [18 (link)].