Assessing Population Diversity of Brettanomyces Yeast Species and Identification of Strains for Brewing Applications

Abstract

Brettanomyces yeasts have gained popularity in many sectors of the biotechnological industry, specifically in the field of beer production, but also in wine and ethanol production. Their unique properties enable Brettanomyces to outcompete conventional brewer's yeast in industrially relevant traits such as production of ethanol and pleasant flavors. Recent advances in next-generation sequencing (NGS) and high-throughput screening techniques have facilitated large population studies allowing the selection of appropriate yeast strains with improved traits. In order to get a better understanding of Brettanomyces species and its potential for beer production, we sequenced the whole genome of 84 strains, which we make available to the scientific community and carried out several in vitro assays for brewing-relevant properties. The collection includes isolates from different substrates and geographical origin. Additionally, we have included two of the oldest Carlsberg Research Laboratory isolates. In this study, we reveal the phylogenetic pattern of Brettanomyces species by comparing the predicted proteomes of each strain. Furthermore, we show that the Brettanomyces collection is well described using similarity in genomic organization, and that there is a direct correlation between genomic background and phenotypic characteristics. Particularly, genomic patterns affecting flavor production, maltose assimilation, beta-glucosidase activity, and phenolic off-flavor (POF) production are reported. This knowledge yields new insights into Brettanomyces population survival strategies, artificial selection pressure, and loss of carbon assimilation traits. On a species-specific level, we have identified for the first time a POF negative Brettanomyces anomalus strain, without the main spoilage character of Brettanomyces species. This strain (CRL-90) has lost DaPAD1, making it incapable of converting ferulic acid to 4-ethylguaiacol (4-EG) and 4-ethylphenol (4-EP). This loss of function makes CRL-90 a good candidate for the production of characteristic Brettanomyces flavors in beverages, without the contaminant increase in POF. Overall, this study displays the potential of exploring Brettanomyces yeast species biodiversity to find strains with relevant properties applicable to the brewing industry.

Keywords: 4-ethylguaiacol; Dekkera bruxellensis; beta-glucosidase; brewing fermentation; genomics; high-throughput screening; maltose assimilation; phenolic off-flavor.

FIGURE 1

Geographical distribution and substrate of origin of the Brettanomyces strains included in the study.

FIGURE 2

Genomic phylogeny of Brettanomyces population. (A) Circular claudogram of 84 Brettanomyces species included in this study. Phylogenetic tree was produced by comparison of the predicted proteins in the whole genome of each strain. Strains used as reference genome are marked with (^∗) Old CRL isolates are marked with (#). Branches are colored according to species, tree nodes according to continent of origin, and metadata layer is colored according to substrate of isolation. Yeasts have been grouped in different genetic clusters according to its location in the tree and genetic cluster names are indicated in the external circle. (B) Same phylogenetic tree in radial form. Branch length and separation displays genomic divergence between Brettanomyces isolates.

FIGURE 3

Distribution of brewing-relevant genes across the B. bruxellensis UMY321 reference genome.

FIGURE 4

Genomic coverage of Brettanomyces strains included in the study. The reference genes spotted in Figure 3 were subtracted and a BLAST search was performed for each gene, respectively. To predict regions of loss, hits scoring > 95% of the total BLAST score were considered as valid for gene presence. Coloring in the genomic heatmap indicates absence/presence of the respective gene. Two strains showed a low read-mapping coverage and were removed from the figure (CRL-6, CRL-60).

FIGURE 5

(A) Growth curves of Brettanomyces strain collection in pilsner wort, shown by means (n = 3). Growth-related values (G-value, y-axis) were extracted with the image analysis software (Enzyscreen). Fermentation time is indicated in minutes (x-axis). (B,C) Acetic acid and ethanol content of beers produced. Error bars show standard deviation (n = 3). (D) Violin plot for the main volatiles measured in beers. Strains are grouped according to its genetic cluster. Sensory detection threshold is indicated by a continuous line. When the line is not present, measured values are below detection threshold. Plots were obtained using ggplot2 package in R. Quantified values and statistical test can be found in Supplementary Table S2 and Statistical Test 1, respectively.

FIGURE 6

Sugar utilization of a representative part of Brettanomyces collection. Strains grown in media with glucose (red) or maltose (blue) as sole carbon source. Shaded curves are representing standard deviations between triplicates. Y-axis represents yeast growth quantified by Bvalue in Biolog. X-axis represents time in hours. Graphs of the full-collection can be found at Supplementary Figure S1. Statistical tests of the current data on maltose assimilation can be found in Supplementary Material.

FIGURE 7

β-glucosidase activity in cultures grown on cellobiose as sole carbon source (OD600) and extracellular activity measured (units/L). Strains are grouped according to the identified ORF in the genome. Statistical tests of the current experimental data can be found in Supplementary Material.

FIGURE 8

(A) Screening of the 84 Brettanomyces strains for uptake of ferulic acid. Heatmap displays averaged values of OD325 (n = 3). Standard deviations can be found at Supplementary Table S2. 1 = B. naardenensis; 2 = B. custersianus; 3 = B. anomalus; 4 = B. bruxellensis; 5 = Blanc; ^∗ = CRL-90. ^∗∗ = Low growth. For ferulic concentration, see Supplementary Table S3. (B) Pilsner wort fermentation of CRL-2 and CRL-90. Average values (n = 2). (C) Phenolic acids (p-coumaric, ferulic) and volatile phenols (4-ethylphenol, 4-ethylguaiacol) in mg/L at the end of fermentation. (D) Genomic setup of CRL-90 compared to the reference CRL-49. The figure has been produced with BLAST tool, comparing the contig nr. 685 where DaPAD1 is present. In Supplementary Table S4, information about the presence of neighboring ORFs with its respective function is given.

FIGURE 9

Violin plot with phenotypical measurements of Brettanomyces population. Strains are grouped according to the genomic cluster assigned in Figure 2.

FIGURE 10

Bi-plot with principal component analysis variable loadings of Brettanomyces strains included in the study. Variable loadings are measured by cos2 function.