Parental strains

Three S. cerevisiae strains were selected for hybridization from a collection of 15 strains isolated from different wine-producing areas in Central Chile and previously described by Martinez et al. [30]. Similarly, three S. eubayanus parental strains were selected from a collection of strains isolated from different locations in Chilean Patagonia, exhibiting high fermentative capacity and representative of the different Patagonia-B lineages (PB-1, PB-2, and PB-3) [15]. The S. pastorianus Saflager W34/70 (Fermentis, France) strain was used as a commercial lager fermentation control. All strains were maintained in YPD agar (1% yeast extract, 2% peptone, 2% glucose and 2% agar) and stored at -80°C in 20% glycerol stocks. Strains are listed in Tab A in S1 Table.

Interspecific hybrids strains and mitochondria genotyping

Parental strains were sporulated on 2% potassium acetate agar plates (2% agar) for at least seven days at 20°C. Interspecific F1 hybrids were generated through spore-spore mating between S. eubayanus strains and S. cerevisiae strains (S1 Fig). For this, tetrads were treated with 10 μL Zymolyase 100 T (50 mg/mL) and spores of opposite species were dissected and placed next to each other on a YPD agar plates using a SporePlay micromanipulator (Singer Instruments, UK). Plates were incubated at two different temperatures, 12 and 20°C, for 2–5 days to preserve the cold- and heat-tolerant mitochondria, respectively, as previously described [31,32], resulting in nine different F1 hybrids (ranging from H1 until H9, Tab A in S1 Table). This procedure was repeated on 25 tetrads of each species, for each type of cross (H1 to H9) and temperature (12 and 20°C), resulting in 18 different cross x temperature combinations. Finally, colonies were isolated, re-streaked on fresh YPD agar plates, and continued to be incubated at 12 and 20°C. The hybrid status of isolated colonies was confirmed by amplification of rDNA-PCR (ITS1, 5.8S, and ITS2) using universal fungal primers ITS1 and ITS4 [33], followed by digestion of the amplicon using the HaeIII restriction enzyme (Promega, USA) as previously described [34] on one colony for each cross attempt (S1 Fig). Confirmed F1 hybrids were designated as H1 to H9 based on parental strains, followed by the hybridization temperature (12 or 20) and the colony number (i.e. H1.20–1 depicts cross 1 at 20°C (Tab A in S1 Table)). We identified the mitochondrial genotype by Sanger sequencing the mitochondrial COX3 gene as previously described [32].

Beer wort fermentation and metabolite screening

Fermentations were carried out in three biological replicates using previously oxygenated (15 mg/L) 12°P wort, supplemented with 0.3 ppm ZnCl₂ as previously described [17]. Briefly, pre-cultures were grown in 5 mL 6°P wort for 24 h at 20°C with constant agitation at 150 rpm. Cells were then transferred to 50 mL 12°P wort and incubated for 24 h at 20°C with constant agitation at 150 rpm. Cells were collected by centrifugation and used to calculate the final cell concentration to inoculate the subsequent fermentation according to the formula described by White and Zainasheff [35]. Cells were inoculated into 50 mL 12°P wort in 250 mL bottles covered by airlocks containing 30% glycerol. The fermentations were incubated at 12 or 20°C, with no agitation for 15 days and monitored by weighing the bottles daily to determine weight loss over time.

Sugar (glucose, fructose, maltose and maltotriose) consumption and ethanol production were determined by High-Performance Liquid Chromatography (HPLC) after 14 days of fermentation. Filtered samples (20 μL) were injected in a Shimadzu Prominence HPLC (Shimadzu, USA) with a BioRad HPX-87H column using 5 mM sulfuric acid and 4 mL acetonitrile per liter of sulfuric acid as the mobile phase at a 0.5 mL/min flow rate. Volatile compound production was determined by using HeadSpace Solid-Phase MicroExtraction followed by Gas Chromatography-Mass Spectrometry (HS-SPME-GC/MS) after 14 days of fermentation as previously described [18].

Phenotypic characterization

Hybrids and parental strains were phenotypically characterized under microculture conditions as previously described [36]. Briefly, we estimated mitotic growth in 96-well plates containing Yeast Nitrogen Base (YNB) supplemented with 2% glucose, 2% maltose, 2% maltotriose, 2% glucose and 9% ethanol, 2% glucose and 10% sorbitol, and under carbon source switching (diauxic shift) from glucose to maltose as previously described [37]. All conditions were evaluated at 25°C. Lag phase, growth efficiency, and the maximum specific growth rate (μ_max) were determined as previously described [38,39]. For the diauxic shift between glucose and maltose, lag time and μ_max were determined during growth in maltose. The parameters were calculated following curve fitting (OD values were transformed to ln) using the Gompertz function [40] in R (version 4.03).

Mid-parent and best-parent heterosis were determined as previously described [41,42], using Eq 1 and 2, where mid-parent heterosis denotes the hybrid deviation from the mid-parent performance and best-parent heterosis denotes the hybrid deviation from the better parent phenotypic value [43].

(1)

(2)

Where:

Experimental evolution

Experimental evolution was carried out at 20°C under two different media conditions (M and T): 1) YNB + 2% maltose supplemented with 9% ethanol (M) and 2) YNB + 1% maltose + 1% maltotriose supplemented with 9% ethanol (T). Experimental evolution assays in maltose were performed in a final volume of 1 mL in 2 mL tubes, while those in maltose and maltotriose were performed in a 96-well plate under a final volume of 200 μL. Each hybrid strain was first grown in 0.67% YNB medium with 2% maltose at 25°C for 24 h with constant agitation at 150 rpm. Each pre-inoculum was then used to inoculate each evolution line at an initial OD_600nm of 0.1, with three replicate lines per strain in medium M and four replicate lines in medium T. Lines in medium M were incubated at 20°C for 72 h. Lines in medium T were incubated for 144 h at 20°C. After this, cultures were then serially transferred into fresh medium for an initial OD_600nm of 0.1. Serial transfers were repeated for 250 generations in total (approximately seven months). The number of generations was determined using the formula log(final cells–initial cells)/log₂ [44]. Population samples were stored at -80°C in 20% glycerol stocks after 50, 100, 150, 200 and 250 generations. After 250 generations, three colonies were isolated for each replicate line on YPM solid medium (1% yeast extract, 2% peptone, 2% maltose and 2% agar) supplemented with 6% ethanol. The fastest growing colonies were stored at -80°C in 20% glycerol stocks. The fitness increase of each the 28 evolved line was determined as the ratio between the phenotypic value of a given line and the equivalent of its respective ancestral hybrid.

Genomic characterization

Genomic DNA was obtained for whole-genome sequencing using the YeaStar Genomic DNA Kit (Zymo Research, USA) and sequenced in an Illumina NextSeq500 following the manufacturer’s instructions. Variant calling and filtering were done with GATK version 4.3.0.0 [45]. Briefly, cleaned reads were mapped to a concatenated reference genome consisting of S. cerevisiae strain DBVPG6765 [46] and S. eubayanus strain CL216.1 [17] using BWA mem 0.7.17 [47], after which output bam files were sorted and indexed using Samtools 1.13 [48]. Variants were called per sample using HaplotypeCaller (default settings) generating g.vcf files. Variant databases were built using GenomicsDBImport and genotypes were called using GenotypeGVCFs (-G StandardAnnotation). SNPs and INDELs were extracted and filtered out separately using SelectVariants. We then applied recommended filters with the following options: QD < 2.0, FS > 60.0, MQ < 40.0, SOR > 4.0, MQRankSum < -12.5, ReadPosRankSum < -8.0. This vcf file was further filtered by removing missing data using the option–max-missing 1, filtering out sites with a coverage below 5^th or above the 95^th coverage sample percentile using the options–min-meanDP and–max-meanDP, and minimum site quality of 30 (—minQ 30) in vcftools 0.1.16 [49]. Sites with a mappability less than 1 calculated by GenMap 1.3.0 [50] were filtered using bedtools 2.18 [51]. As an additional filtering step, the ancestral and evolved vcf files were intersected using BCFtools 1.3.1 [52] and variants with shared positions were extracted from the vcf files of the evolved hybrids. Annotation and effect prediction of the variants were performed with SnpEff [53].

We used sppIDer [54] to assess the proportional genomic contribution of each species to the nuclear and mitochondrial genomes in each sequenced hybrid. In addition, we used the tool to identify potential aneuploidies within these genomes. CNVs were called using CNVkit (—method wgs,—-target-avg-size 1000) [55]. As the analysis was performed on a haploid reference (both parental genomes were present), a CNV of log2 = 1 corresponds to a duplication.

Since S. cerevisiae and S. eubayanus show divergence greater than 20% we could not detect loss of heterozygosity (LOH) by mapping to a single parental reference. As both hybrids appear to have one copy of each parental genome with no aneuploidies, LOH should appear as a loss of coverage segments that can be detected as copy number losses. LOH regions were detected by mapping reads to the concatenated genome and we used CNVkit [55] to detect genomic segments (1000 bp windows) showing CNVs with log2FC less than -2. Comparisons were made between parental strains vs ancestral hybrids and between ancestral hybrids with evolved hybrids.

RNA-seq analysis

Gene expression analysis was performed on ancestral and evolved hybrid strains H3-A and H3-E. RNA was obtained and processed after 24 h under beer wort fermentation in triplicates, using the E.Z.N.A Total RNA kit I (OMEGA) as previously described [37,56]. Total RNA was recovered using the RNA Clean and Concentrator Kit (Zymo Research). RNA integrity was confirmed using a Fragment Analyzer (Agilent). Illumina sequencing was performed in NextSeq500 platform.

Reads quality was evaluated using the fastqc tool (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and processed using fastp (-3 l 40) [57]. Reads were mapped to a concatenated fasta file of the DBVPG6765 and CL216.1 genome sequences. To account for mapping bias due to the different genetic distances of the parental strains to their reference strains, the L3 and CL710.1 parental strains were re-sequenced using WGS, after which genomic reads were mapped with BWA [47] to the DBVPG6765 and CL216 references and SNPs were called using freebayes [58]. These SNPs were used to correct the hybrid genome sequence using the GATK FastaAlternateReferenceMaker tool. RNAseq reads were mapped to this hybrid reference using STAR (-outSAMmultNmax 1, -outMultimapperOrder random) [59]. Counts were obtained with featureCounts using a concatenated annotation file [60]. Counts were further analyzed in R using de DESeq package [61]. A PCA analysis to evaluate the reproducibility of replicates was performed, after which two outlier replicates (H3-A replicate 3 and H3-E replicate 2) were removed. To analyze differences in allele expression, a list of 1-to-1 orthologous genes between both parental strains was identified using OMA [62]. Orthologous genes that differ more than 5% on their gene lengths were excluded. The differential allelic expression of these orthologous genes was determined using design = ~parental, with parental being “L3” or “CL710”. Furthermore, orthologous genes that showed differential allele expression depending on the ancestral or evolved strain background were assessed using an interaction term (~ parental:condition), with condition being “ancestral” or “evolved”. Finally, to evaluate differences between ancestral and evolved hybrid strains, all 11,047 hybrid genes (5,508 S. eubayanus and 5,539 S. cerevisiae) were individually tested for differential expression using DESeq2. Overall gene expression differences were evaluated using the design ~condition. For all analyzes an FDR < 0.05 was used to consider statistical differences. GO term enrichment analyzes on differentially expressed genes were calculated using the package TOPGO [63].

IRA2 gene validation

The S. cerevisiae IRA2 polymorphism was validated by Sanger sequencing. PCR products were purified and sequenced by KIGene, Karolinska Institutet (Sweden). The presence of the SNP in the evolved hybrid strains was checked by visual inspection of the electropherograms. Null mutants for the IRA2 gene in the S. cerevisiae subgenome were generated using CRISPR-Cas9 [64] as previously described [36]. Briefly, the gRNAs were designed using the Benchling online tool (https://www.benchling.com/) and cloned into the pAEF5 plasmid [65], using standard “Golden Gate Assembly” [66]. Ancestral and evolved hybrids were co-transformed with the pAEF5 plasmid carrying the gRNA and the Cas9 gene, together with a double-stranded DNA fragment (donor DNA). The donor DNA contained nourseothricin (NAT) resistance cassette, obtained from the pAG25 plasmid (Addgene plasmid #35121), flanked with sequences of the target allele, corresponding to 50-pb upstream of start codon and 50-pb downstream of the stop codon. Correct gene deletion was confirmed by standard colony PCR. All primers, gRNAs, and donor DNA are listed in Tab B in S1 Table.

FACS analysis

DNA content was assessed through the propidium iodide (PI) staining assay, as previously described [67]. Initially, cells were recovered from glycerol stocks on YPD solid media and allowed to grow overnight at 25°C. Subsequently, a portion of each patch was transferred into liquid YPD media and incubated overnight at 25°C. Then, 1 ml of each culture was harvested and suspended in 2.3 ml of cold 70% ethanol for fixation during 48h h at 4°C. Following fixation, cells were washed with sodium citrate (50 mM, pH 7) and 100 μl of cells resuspended in the same solution, and 1 μL of RNAse A (100 mg/mL) were incubated for 2 h at 37°C. Then, cells were stained with a solution containing PI (final concentration of 50 μg/mL) and sodium citrate (50 mM, pH 7), and incubated for 40 min at room temperature in darkness. Analysis was conducted on a BD FACSCanto II flow cytometer with excitation at 488 nm and fluorescence collection using an FL2-A filter, analyzing ten thousand cells per sample. Three strains with known ploidy (two S. cerevisiae -n and 2n- and one S. pastorianus -4n-) were employed as controls.

Statistical analysis

Data visualization and statistical analyses were performed with R software version 4.03. Maximum specific growth rates and total CO₂ loss were compared using an analysis of variance (ANOVA) and differences between the mean values of three replicates were tested using Student’s t-test and corrected for multiple comparisons using the Benjamini-Hochberg method. A p-value less than 0.05 (p<0.05) was considered statistically significant. Heatmaps were generated using the ComplexHeatmap package version 2.6.2. A principal component analysis (PCA) was performed on phenotypic data using the FactoMineR package version 2.4 and the factoextra package version 1.07 for extracting, visualizing and interpreting the results.

De novo S. cerevisiae x S. eubayanus F1 hybrids show similar phenotypes as their parental strains

The S. cerevisiae and S. eubayanus parental strains were selected from a previously described collection of Chilean isolates by Martinez et al. [30] and Nespolo, Villarroel et al. [15], respectively (Tab A in S1 Table). Initially, three S. cerevisiae strains from vineyards were selected because they showed: i) the highest maximum CO₂ loss in beer wort (S2A Fig and S2 Table), ii) the best growth performance under maltotriose conditions (S2B Fig), and iii) the most efficient maltotriose uptake during microculture conditions (S2C Fig). These strains were L3, L270, and L348. The selection of S. eubayanus parental strains was determined by two criteria: i) to represent distinct lineages found in the Chilean Patagonia to maximize genetic diversity (one strain per lineage, PB-1, PB-2, and PB-3), and ii) to display the highest CO₂ loss during fermentation when compared to strains within their respective lineages based on previous assays [15]. In this way, we selected CL450.1, CL710.1 and CL216.1, from PB-1, PB-2, and PB-3, respectively. All strains were able to sporulate (Tab C in S2 Table).

Nine interspecific F1 hybrid crosses were performed through spore-to-spore mating at 12°C and 20°C, to promote the preservation of the cold- and heat-tolerant mitochondria, respectively, as previously described (S1 Fig) [31,32]. We obtained 31 interspecific hybrids (Tab A in S1 Table), which we phenotyped individually under microculture conditions resembling those encountered during beer wort fermentation, such as glucose, maltose, maltotriose, and in the presence of ethanol and simulating osmotic stress with sorbitol (S3 Table). Hierarchical clustering of the phenotypic data denotes three main clusters, where there was no discernible clustering of hybrids based on their parental strains or hybridization temperature, highlighting the considerable phenotypic diversity resulting from hybridization (Fig 1A). To describe the phenotypic landscape of the 31 hybrids more comprehensively, we conducted a PCA analysis (Fig 1B). The individual factor map shows that hybrids made at 20°C fall into the right upper quarter of the phenotype space, and are associated with a higher growth rate in media with maltose and glucose compared to hybrids made at 12°C. This was particularly the case for four hybrid strains (H1, H3, H4 and H6), involving parental strains L3, L270, CL216.1 and CL710.1 (all p-values < 0.05, one-way ANOVA, Tab B in S3 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Phenotypic characterization of interspecific F1 hybrids.

A) Hierarchically clustered heatmap of phenotypic diversity of 31 interspecific hybrid strains under microculture conditions. Phenotypic values are calculated as normalized z-scores. For the diauxic shift between glucose and maltose, lag time and μ_max were determined during growth in maltose. (B) Principal component analysis (PCA) using the maximum specific growth rates under six microculture growth conditions, together with the distribution of hybrid strains. Arrows depict the different environmental conditions. (C) Best-parent heterosis in the 31 interspecific hybrids evaluated under microculture conditions in triplicates. (D) Fermentation capacity for the 31 interspecific hybrids and parental strains at 12°C. Plotted values correspond to mean values of three independent replicates for each hybrid. Asterisk indicates different levels of significance compared to the commercial strain W34/70 (Student t-test; *** p≤ 0.001 and **** p≤ 0.0001). (E) Best-parent heterosis in the 31 interspecific hybrids evaluated under fermentation conditions at 12°C.

https://doi.org/10.1371/journal.pgen.1011154.g001

To assess the impact of hybridization on yeast fitness, we calculated best-parent and mid-parent heterosis coefficients across the 31 hybrids (Fig 1C and Tabs C and D in S3 Table). While some hybrids exhibited positive mid-parent heterosis in 5 out 7 conditions (Tab C in S3 Table), we generally did not observe hybrids with positive best-parent heterosis (BPH, Tab D in S3 Table), except for rare cases involving maltose utilization and growth rate during diauxic shift, where 2 and 5 hybrids, respectively, displayed positive values (Fig 1C). Overall, inter-species hybridization did not result in a significant enhancement of fitness in F1 hybrids.

Next, we assessed the fermentation capacity of the 31 hybrids in wort at low temperature (Fig 1D and S4 Table). Hybrids generated at 12°C displayed similar levels of CO₂ production compared to those obtained at 20°C (Fig 1D and Tab A in S4 Table, p-value = 0.17, one-way ANOVA). We did not observe any hybrids exhibiting superior fermentative capacity when compared to their respective parental strains (S3 Fig), and there was no evidence for hybrid vigour according to best-parent and mid-parent heterosis coefficients (Fig 1E and Tabs C and D in S4 Table). Neither parents nor hybrids reached the fermentative capacity of the commercial strain W34/70 (p-value < 0.05, one-way ANOVA).

Evolved lines carrying the S. eubayanus mitochondria exhibit a greater fitness under fermentation

All results so far indicated that the de novo interspecific hybrids did not show any hybrid vigour, in none of the phenotypes assessed. We thus decided to subject hybrids to experimental evolution to enhance their fermentative capacity. We specifically selected four hybrids (H3.12–3, H4.12–4, H6.20–2, and H8.20–5) because they demonstrated the highest phenotypic values across kinetic parameters. From here on we will refer to these strains as H3-A, H4-A, H6-A, and H8-A (A for ‘ancestral’ or unevolved hybrid). These four hybrids completely consumed the sugars present in the beer wort, except for maltotriose, which may explain the lower fermentative capacity of the hybrids compared to the commercial strain W34/70 (Tab E in S4 Table). Furthermore, these four hybrids represent crosses made at 12°C and 20°C and they encompass all six parental genetic backgrounds. To enhance the fermentative capacity of these selected hybrids, they were subjected to adaptive evolution at 20°C for 250 generations under two distinct conditions: i) YNB supplemented with 2% maltose and 9% ethanol (referred to as "M" medium), and ii) YNB supplemented with 1% maltose, 1% maltotriose, and 9% ethanol (referred to as “T” medium). We evolved three lines independently per cross in medium M, and four independent lines per cross in medium T. These conditions were chosen because maltose is the main sugar in beer wort (approximately 60%) [68]. Considering that yeast typically consume carbon sources in a specific order (glucose, fructose, maltose, and maltotriose), we employed a combination of maltose and maltotriose to facilitate the utilization of the latter carbon source.

After 250 generations, the evolved lines showed different levels of fitness improvements, depending on the environmental conditions and their genetic background (Figs 2A and S4), with distinct fitness trajectories over time (S5 Fig). All evolved lines significantly increased in fitness in at least one of the evolution media and/or kinetic parameters assessed compared to their respective ancestral hybrids (Fig 2A and Tabs A and B in S5 Table; p-value < 0.05, one-way ANOVA). Interestingly, evolved lines from hybrids made at 12°C mating temperature (H3-A and H4-A) showed a more pronounced fitness increase in the T medium compared to those generated at 20°C (p-value = 3.327e-08, one-way ANOVA, Figs 2B and S4B), suggesting that hybrids with S. eubayanus mitochondria have greater potential for improvement than hybrids with S. cerevisiae mitochondria. We verified that the two ancestral H3-A and H4-A hybrids carried only S. eubayanus mitochondria by sequencing the COX3 gene, while H6-A and H8-A inherited the mitochondria from S. cerevisiae (Tabs C and D in S5 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Fitness of evolved lines under microcultures and fermentation conditions.

(A) Mean relative fitness (maximum OD_600nm) of evolved lines after 250 generations to their respective ancestral hybrids under microculture conditions. Evolved lines were evaluated in the same medium where they were evolved (M or T medium). (B) Comparison of mean relative fitness (maximum OD_600nm) shown in (A) between evolved lines from hybrids with S. eubayanus (Se) and S. cerevisiae (Sc) mitochondria. (C) Mean relative fitness (maximum CO₂ loss) of evolved lines after 250 generations to their respective ancestral hybrids under fermentation conditions at 12°C. (D) Comparison of mean relative fitness (maximum CO₂ loss) shown in (C) between evolved lines from hybrids with S. eubayanus and S. cerevisiae mitochondria. (E) Maltotriose uptake of evolved hybrid lines in maltose (M) and maltose/maltotriose (T), relative to the commercial lager strain W34/70. Ancestral hybrids are shown in grey, and hybrid lines with S. eubayanus and S. cerevisiae mitochondria are shown in blue and red, respectively. (F) The fermentative capacity of evolved individuals relative to the commercial lager strain W34/70 grouped according to the environmental condition used during experimental evolution and inherited mitochondria. Plotted values correspond to the mean of three independent biological replicates of each evolved line or strain. Asterisk indicates significant statistical differences between evolved lines and their respective ancestral hybrids in (A) and (C), between evolved lines with different inherited mitochondria in (B) and (D), and between evolved lines and the commercial lager strain in (E) and (F). Purple depicts Parental strains, brown the ancestral hybrid, and red and blue the Sc and Se evolved lines carrying mitochondria, respectively. Asterisk represents different levels of significance (Students t-test, * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001, ns not significant).

https://doi.org/10.1371/journal.pgen.1011154.g002

Next, we assessed the fermentative capacity of the evolved lines under conditions resembling beer wort fermentation (12°Brix and 12°C) (Figs 2C and S6 and Tabs A and B in S6 Table). We did not observe a significant increase in CO₂ production levels in the evolved lines of the H6-A and H8-A hybrids in either M or T media (Figs 2C and S6 and Tab B in S6 Table, p-value < 0.05, one-way ANOVA). However, we found a significant greater CO₂ production in the evolved lines of H4-A, evident in both evolution media, indicative of higher fermentation activity. The evolved lines of H3-A under T media also demonstrated a slightly higher CO₂ production (Fig 2C and Tab B in S6 Table, p-value < 0.05, one-way ANOVA, for H4 evolved lines and p-values of 0.0708 and 0.05149 for H3 evolved lines in M and T, respectively). Thus, both evolved hybrid lines generated at cold-temperature, carrying S. eubayanus mitochondria, showed a greater increase in CO₂ production than hybrids carrying the S. cerevisiae mitochondria (Fig 2C). Specifically, hybrids with S. eubayanus mitochondria increased their maximum CO₂ loss by 10.6% when evolving in M medium (p-value = 0.003698, one-way ANOVA) and by 13% in T medium (p-value = 1.328e-08, one-way ANOVA) (Fig 2D). This was predominantly due to an elevated maltotriose uptake (Fig 2E and Tab C in S6 Table). Notably, the fermentative capacity of these hybrids reached that of the commercial strain (Tab D in S6 Table, p-value > 0.05, one-way ANOVA). These findings strongly suggest that lines derived from hybrids generated at colder temperatures carrying S. eubayanus mitochondria and evolved in a complex maltose/maltotriose medium (T), significantly enhanced their lager fermentative capacity due to an increased maltotriose uptake during beer wort fermentation.

Isolation of evolved genotypes with improved fermentative capacity and maltotriose uptake

To isolate individual representatives from the evolved population lines, we obtained one single genotype from each of the four hybrid lines at 250 generations (28 genotypes in total), which were then subjected to phenotypic evaluation in beer wort. These individual genotypes showed similar fermentation profiles as the population-level analyses above (Fig 2F). Evolved hybrid genotypes carrying S. eubayanus mitochondria and evolved in T medium (maltose/maltotriose, H3-E and H4-E), showed higher CO₂ production compared to H6-E and H8-E (p-value < 0.05, ANOVA, S7 Fig). The genotypes with the largest significant fitness increase were derived from lines H3-3 and H3-4 evolved in T conditions (S7 Fig), which exceeded the commercial strain. Interestingly, two genotypes deriving from H6-A (carrying the S. cerevisiae mitochondria) evolved in T medium also showed a CO₂ loss similar to the commercial strain (p-value = 0.90372, one-way ANOVA).

To focus more in-depth on the evolved lines with the highest fermentative capacity and carrying the S. eubayanus mitochondria (H3-4 and H4-1 evolved in T medium, S7 Fig), we isolated three colonies from each of these two lines to evaluate their fermentative capabilities. Notably, the CO₂ loss kinetics among these genotypes were comparable (p-value > 0.05, one-way ANOVA), with genotype #1 from line H3-4 exhibiting the highest CO₂ loss (Fig 3A). All these genotypes’ fermentation profiles closely resembled that of the W34/70 commercial lager strain, underscoring the significantly high fermentative capacity of these novel hybrids (p-value > 0.05, one-way ANOVA, Fig 3A). All genotypes consumed the maltotriose in the T medium completely (Tab A in S7 Table), and ethanol production ranged from 3.50% to 3.78% v/v (Fig 3B), which is similar to the commercial strain (p-value > 0.05, one-way ANOVA). One genotype (H4-1-C3) showed a remarkable 7.1% increase in ethanol production compared to the commercial strain (Fig 3B, p-value = 0.001, one-way ANOVA).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Fermentation performance of evolved hybrid individuals.

(A) Maximum CO₂ loss (g/L) for three different isolated genotypes (C1-C3) from evolved lines H3-4 and H4-1, ancestral hybrids (H3-A and H4-A) and commercial lager strain (W34/70). (B) Ethanol production (% v/v) for strains evaluated in (A). (C) Hierarchically clustered heatmap of volatile compounds production for strains evaluated in (A). Phenotypic values are calculated as normalized z-scores. For (A) and (B), plotted values correspond to the mean of three independent replicates. The (*) represents different levels of significance between hybrids and commercial lager strains (Student t-test, ** p < 0.01, **** p < 0.0001).

https://doi.org/10.1371/journal.pgen.1011154.g003

To compare the aroma profile of the H3-4-C1evolved hybrid to the lager strain, we identified volatile compounds (VCs) by HS-SPME-GC-MS in the fermented wort. This assay allowed us to identify 15 and 14 compounds in the evolved and commercial lager strains, respectively. We observed significant differences for 11 different compounds (Fig 3C, p-value < 0.05, one-way ANOVA, Tab B in S7 Table), including ethyl esters and higher alcohols. For example, the evolved strain produced significantly more fatty acid ethyl esters, such as dodecanoic and tetradecanoic acid ethyl esters (p-value = 0.013 and 0.002, respectively, one-way ANOVA), which could provide sweetish notes. The commercial lager strain on the other hand produced higher amounts of other fatty acid ethyl esters, such as octanoic acid and nonanoic acid (Fig 3D, p-value = 0.001 and 0.0001, respectively, one-way ANOVA), and citronellol (p-value < 0.0001, one-way ANOVA), which is associated with a citrus aroma. The H3-4-C1evolved hybrid produced detectable levels of 4-vinphenol, associated with clove-like, spicy, and phenolic aromas, commonly found in Belgian ale and wheat beers, which was completely absent in the lager strain. These results demonstrate that the aroma profiles of the evolved hybrid differ from the commercial lager strain. Therefore, the beers produced with the evolved hybrid would have a different profile towards a more herbal, spicy, and phenolic character. In contrast, those produced with the commercial strain would have a more citrusy and refreshing profile.

Mutation in IRA2 affects fermentation capacity in the evolved hybrid

To identify mutations in evolved hybrids associated with their improved fermentative capacity, we sequenced the genomes of the two genotypes exhibiting the highest CO₂ production levels, specifically H3-4-C1 and H4-1-C1 (from here on referred to as H3-E and H4-E; with ‘E’ for evolved hybrid) that were evolved in the maltose/maltotriose T medium (Tab A in S8 Table). Genome sequencing revealed that these two strain backgrounds had equal contributions from both parental genomes and that they had euploid, diploid genomes with no detectable aneuploidies (Tab B in S8 Table), containing the S. eubayanus mitochondria. We only detected small LOH regions that were associated with the hybridization process, mainly losses of S. eubayanus DNA at subtelomeric regions (Tab C in S8 Table) but also a small LOH region partly spanning the HKR1 gene located in chromosome II, between the H4-A hybrid compared with the CL216.1 parent (Tab C in S8 Table).

We then identified de novo single nucleotide polymorphisms (SNPs) in the evolved hybrid genomes that were absent in the ancestral hybrids. We found 3 and 4 SNPs in the H3-E and H4-E backgrounds, respectively (Tab D in S8 Table). The evolved hybrids presented a similar number of SNPs per genome. In H3-E, we found 1 and 2 SNPs in the S. cerevisiae and S. eubayanus parental genomes, respectively, while H4-E had 1 and 3 SNPs in the corresponding parental genomes. We identified an anticipated stop-codon in the IRA2 allele (encoding for a GTPase-activating protein, Fig 4A) in the S. cerevisiae subgenome, and a missense mutation in CKB2 (encoding for a subunit of casein kinase 2) and CMC1 (encoding for a mitochondrial protein necessary for full assembly of Cytochrome c oxidase) in the S. eubayanus subgenome (Tab D in S8 Table) in H3-E. While in H4-E, we identified a missense mutation in MDS3 (a putative component of the TOR regulatory pathway), FSF1 (predicted to be an alpha-isopropyl malate carrier), and ASC1 (core component of the small ribosomal subunit) in the S. eubayanus subgenome, and an upstream gene mutation in SSK2 (MAP kinase kinase kinase of HOG signalling pathway) in the S. cerevisiae subgenome (Tab D in S8 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Genomic analysis of evolved hybrids.

(A) SNP present in the IRA2 gene in the S. cerevisiae subgenome in the H3-E hybrid. (B) Maximum OD_600nm of ira2Δ^Sc mutant strains under microculture conditions. Mutant and wild-type strains were evaluated in the T medium. (C) CO₂ loss kinetics for ira2Δ^Sc mutant and wild-type strains. (D) Maltotriose uptake (%) for strains evaluated in (C). For (B), (C), and (D), plotted values correspond to the mean of four independent replicates. The (*) represents different levels of significance between mutant and wild-type strains (Student t-test, * p < 0.05, *** p < 0.001, **** p < 0.0001).

https://doi.org/10.1371/journal.pgen.1011154.g004

To detect additional genetic changes not identified in the individual clones and to track the relative frequencies of the de novo mutations in the evolution lines, we sequenced whole population samples at increasing time points of experimental evolution (at 50, 100, 150, 200, and 250 generations; S8 Fig and Tabs D and E in S8 Table). In this way, we identified 1 and 2 additional variants in the H3 and H4 evolved lines, respectively. For example, in the H4-evolved lines, we identified mutations within the SCC2 and NSE1 coding regions of two genes related to DNA replication and repair processes (Tab D in S8 Table). Subsequently, we estimated the IRA2 mutation frequencies on the H3-evolved line. The IRA2-L2418* polymorphism arose before generation 50 and was completely fixed by 150 generations. We also determined the ploidy of this population at the end of the evolution process (250 generations), confirming that ploidy levels did not change during the experimental evolution process in the H3 population, maintaining a diploid state (S9 Fig).

To determine the phenotypic impact of the stop-codon detected in the IRA2 gene, we performed a CRISPR-Cas9 gene editing targeting the S. cerevisiae IRA2, generating null mutants (ira2^Sc) in the evolved and non-evolved hybrids. We evaluated growth under microculture conditions in the same evolutionary medium (T) and under beer wort fermentation (Fig 4B). This assay revealed that ira2^Sc mutants in the H3-A hybrid background had a 12.5% lower OD_max under maltose/maltotriose conditions compared to H3-A (Fig 4B, p-value = 0.01213, one-way ANOVA), but still a similar fermentative capacity (Fig 4C, p-value = 0.79685, one-way ANOVA). In the H3-E hybrid, the null ira2^Sc mutant showed a 16.6% higher OD_max under microculture conditions (Fig 4B, p-value = 0.00042, one-way ANOVA) and a significantly lower fermentative capacity under beer wort, with a 13.6% decrease in CO₂ production (Fig 4C, p-value = 0.02315 one-way ANOVA, S9 Table) and a 10.8% decrease in the maximum CO₂ loss rate (S9 Table, p-value = 0.02268 one-way ANOVA). This decrease in the fermentative capacity in the H3-E null mutant correlates with a lower maltotriose uptake (68.6%, Fig 4D). These results suggest that the stop-codon in IRA2 in the evolved hybrids does not necessarily lead to a loss of protein function, but instead to a complex genetic interaction in the H3-E background promoting a trade-off between biomass and fermentative capacity, which is likely partly responsible for the phenotypic differences during the evolutionary process.

Copy number variants of genes related to maltose metabolism are associated with improved fermentative capacity in evolved hybrids

Since ira2 null mutants did not restore the full increase in fermentative capacity of the evolved hybrids, we examined genes exhibiting copy number variation (CNVs) in H3-E and H4-E hybrids (Fig 5A and Tab F in S8 Table). Both H3-E and H4-E hybrids contained changes in copy number, particularly in the MAL gene family (Fig 5A and Tab F in S8 Table). For example, we identified 2 and 4 extra copies of the MAL13 and MAL11 genes in H4-E and H3-E, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Copy number variation and differential gene expression analysis.

(A) Copy number variations (CNVs) between H3-E and H4-E hybrids relative to their ancestral hybrids found in S. cerevisiae chromosome 7. Coding genes located within bins showing CNV calls higher than 1 copy (yellow rectangles) are shown. (B) Volcano plot showing differential expressed genes (DEGs) between H3-E and H3-A hybrids. The red and blue dots represent up-regulated and down-regulated genes in the H3-E hybrids, respectively. (C) Orthologous genes showing an interaction between allelic expression and experimental evolution. (D) Model depicting genes exhibiting mutations after the experimental evolution assay (highlighted in orange) and involved in pathways related to the detection, regulation, uptake, and catabolism of maltotriose. Phosphorylation is indicated in red. In blue highlight transporters involved in sugar consumption.

https://doi.org/10.1371/journal.pgen.1011154.g005

To determine the impact of these mutations and the CNVs in the transcriptome of the H3-E hybrid, we estimated transcript abundance under beer fermentative conditions in the evolved and non-evolved hybrid. We identified 40 Differentially Expressed Genes (DEGs, FDR < 5%, Tab G in S8 Table), where 21 and 19 genes were up- and down-regulated in the evolved hybrid relative to its hybrid ancestor, respectively. Interestingly, we found that S. cerevisiae alleles for IMA1, MAL11, and MAL13 were up-regulated in H3-E, which correlates with the increased gene copy number (Fig 5B). A GO term analysis showed that genes involved in maltose metabolic processes were up-regulated and genes in cell wall organization were down-regulated in the evolved hybrid (Tab H in S8 Table).

To measure the impact of cis-variants on allelic expression within each parental subgenome, we estimated allele specific expression (ASE) in the evolved and non-evolved hybrids (Fig 5C and Tab I in S8 Table). Seven genes showed ASE differences between the evolved and ancestral hybrid, likely originating from mutations in regulatory regions acquired during experimental evolution (Fig 5C and Tab I in S8 Table). Of these, one and six ASE differences in the H3-E hybrid represented up-regulated alleles in the S. cerevisiae and S. eubayanus subgenomes, respectively (Tab I in S8 Table). Interestingly, we detected the up-regulation of the REG2 allele related to sugar consumption with a 2.1 higher fold change in the S. cerevisiae subgenome, which is involved in the regulation of glucose-repressible genes (Fig 5C), correlating with the higher maltose and maltotriose consumption levels in the evolved hybrid.