The Identification of Novel Diagnostic Marker Genes for the Detection of Beer Spoiling Pediococcus damnosus Strains Using the BlAst Diagnostic Gene findEr

Abstract

As the number of bacterial genomes increases dramatically, the demand for easy to use tools with transparent functionality and comprehensible output for applied comparative genomics grows as well. We present BlAst Diagnostic Gene findEr (BADGE), a tool for the rapid prediction of diagnostic marker genes (DMGs) for the differentiation of bacterial groups (e.g. pathogenic / nonpathogenic). DMG identification settings can be modified easily and installing and running BADGE does not require specific bioinformatics skills. During the BADGE run the user is informed step by step about the DMG finding process, thus making it easy to evaluate the impact of chosen settings and options. On the basis of an example with relevance for beer brewing, being one of the oldest biotechnological processes known, we show a straightforward procedure, from phenotyping, genome sequencing, assembly and annotation, up to a discriminant marker gene PCR assay, making comparative genomics a means to an end. The value and the functionality of BADGE were thoroughly examined, resulting in the successful identification and validation of an outstanding novel DMG (fabZ) for the discrimination of harmless and harmful contaminations of Pediococcus damnosus, which can be applied for spoilage risk determination in breweries. Concomitantly, we present and compare five complete P. damnosus genomes sequenced in this study, finding that the ability to produce the unwanted, spoilage associated off-flavor diacetyl is a plasmid encoded trait in this important beer spoiling species.

Fig 1. The use-oriented prediction of diagnostic marker genes using BADGE.

(a) Workflow illustrating the structure of a comparative genomics project with the aim to identify and validate diagnostic marker genes (DMGs). Each box corresponds to a single step within the straightforward procedure. (b) The prediction of group A specific DMGs is shown using a fictive example with two members of group A and two members of group B. A description of all steps can be found in the corresponding boxes. Genome = here: sequences of chromosome and all extrachromosomal elements (contigs); ORFs = open reading frames, here: sequences of all ORFs; DMG = diagnostic marker gene.

Fig 2. Shared gene content of P. damnosus genomes.

The number of genes shared between strains is shown for each combination, once for all chromosomal and once for all plasmid encoded genes. In case of the plasmid comparison, we included the plasmidome of the rapid beer spoiling L. brevis BSO 464 [35]. Brewery strains and the winery isolate TMW 2.1536-NB are color labeled for differentiation. Shared genes were predicted using BADGE with default settings. NB = no spoilage potential, SB = strong spoilage potential according to beer spoilage test.

Fig 3. Visual illustration of BADGE results of the P. damnosus beer spoilage example.

(a) Illustration of BADGE comparison of the group of beer spoiling strains with the group of non-spoiling strains. The number of genes, shared by all five strains is also shown. All calculations were done with BADGE using default settings. (b) Percentage of SB-DMGs regarding their function based on RAST annotation. (c) BLAST ring image for whole genomes of all five strains. All rings are described from the inside to the outside: ring 1 (black) shows the whole genome of TMW 2.1535-SB as reference with bp coordinates; ring 2 (black) shows the GC content of TMW 2.1535-SB; ring 3 consists of arcs with different lengths and alternating coloration (grey / blue) representing the contigs (chromosome and plasmids) of TMW 2.1535-SB; ring 4 (red) shows all blast hits of TMW 2.1536-SB ORFs versus its own genome illustrating the coding density; ring 5 (orange) shows all blast hits of TMW 2.1533-SB ORFs versus the reference; rings 6–8 (blue shades) show all blast hits of the NB strains ORFs versus the reference, ring 9 contains all DMGs predicted by BADGE mapped to the reference, located in the gaps of the rings 6–8. Note that a particular DMG may occur more than once in a genome. Therefore the number of DMG labels does not correspond to the number of DMGs calculated by BADGE. (d) BLAST ring image for plasmids only. Most DMGs are located within clusters, e.g. in a cluster containing genes for fatty acid synthesis (FAS). For descriptions of the rings see (c).

Fig 4. DMG evaluation and validation.

20 strains of P. damnosus were tested for the presence of the predicted DMGs and already published DMGs using PCR. Note that the published DMGs horA and horC were not predicted by BADGE, as both genes are present in two of three non-spoiling genomes. Still, all strains were tested for their presence and results are shown for comparison. A positive reaction is indicated with a dot. Significant correlation to spoilage potential is indicated using the colors red (very high) and orange (moderate to high). Genome sequenced strains are underlined.