We used fifty regional S. cerevisiae isolates from 12 different viticulture places in Chile, which are part of the Microorganism Culture Collection of the Kayta company in collaboration with LAMAP-University of Santiago of Chile (Supplementary Table S1; Supplementary material A1). In addition, strains isolated from Argentine Patagonia (part of the LAMAP-culture collection) were included as a reference, due to their brewing capacities. Cryogenically preserved (−80°C) strains were cultured and maintained on YPM medium (3 g/L malt extract, 3 g/L yeast extract, 5 g/L peptone, 10 g/L glucose, 16 g/L agar) plates and stored at 4°C.

Samples were taken at the end of the fermentations carried out with grapes from these different viticulture places in Chile. They were diluted in 1 moL/L sorbitol and plated on YM agar (yeast extract 3 g/ l, malt extract 3 g/L, peptone 5 g/L and glucose 10 g/L) supplemented with ampicillin (100 mg/L). The taxonomic identity of these yeast isolates was pre-identified by growth on lysine and WL medium (Oxoid, Basingstoke, United Kingdom). Furthermore, yeast identification at the genus, and/or species, level was carried out by a 5.8S-ITS RFLP analysis as described Esteve-Zarzoso et al. (1999). To determine the autochthonous character of the selected yeasts, we used the Random Amplified Polimorphic DNA (RAPD) molecular technique, considering eight random primers (Echeverrigaray et al., 2000). We analyzed gel electrophoresis imaging using the Gel Analyser™ program, which allowed us to identify electrophoretic bands in each strain and constructing a dendrogram (Supplementary Figure S1; Supplementary material A1). At this stage, we analyzed L261, L170, 8jj, L169 and L249, and the commercial strains Fermicru XL™ (wine), Lalvin™ EC1118 (wine), Lallemand Nottingham™ (beer), WLP001™ (beer) and WLP004™ (beer). The figure shows that the Saccharomyces strains analyzed had a higher similarity (short distance) with the Fermicru XL strain. This strain has been confirmed as an indigenous Chilean wine strain. On the other hand, a higher distance with the commercial strains was observed (Supplementary Figure S1; Supplementary material A1).

Each strain was cultured in a medium where the primary carbon sources were maltose and maltotriose. We generated a pre-inoculum of each strain, cultivated in YPM broth, which was inoculated at 1 × 10⁶ cells /ml in 50 mL of YPMM broth (0.5% peptone, 0.5% yeast extract, 2% maltose and 1% maltotriose), using 100 mL Erlenmeyer flasks. The cultures were incubated at 28°C for 24 h with shaking (120 rpm). Maltose and maltotriose consumption was evaluated as a selection phenotype; the determination was carried out by HPLC, as described below.

At this stage, synthetic wort was prepared using malt extract with a concentration of 150 g/L, generating 1,060 g/mL of final density at pH 5.4. Wort was sterilized at 121°C for 15 min. We prepared the cultures in 250 mL Erlenmeyer flasks with gas outlets. We inoculated the culture volume of 100 mL considering 4 ×10⁶ cells/ml, this cell concentration allows the culture to develop. Cell viability was verified with methylene blue. Before inoculation, we aerated the wort by shaking for 5 min. All cultures were fermented for 7 days at 18°C without stirring.

The process was started by inoculating each yeast strain into an YPG (3 g/L yeast extract, 5 g/L peptone, and 10 g/L galactose) culture broth. It was necessary to use this medium to activate the metabolic pathways related to the consumption of maltose and maltotriose early (Higgins et al., 2001). In this YPG broth, each yeast was cultured for 48 h with shaken (200 rpm) at 28 ° C. We performed an adaptive evolution process, which comprised three phases. For the first phase, we used synthetic wort with a concentration of malt extract of 150 g/L (maltose and maltotriose concentrations around 50 g/L and 30 g/L, respectively). This wort had a final density of 1.060 g/mL. At the beginning of the process, the volume of inoculum was 1 × 10⁵ cells /ml, which was grown at 20°C with 200 rpm of agitation. During the evolution process, we monitored all cultures to collect cells in the middle of the exponential growth phase to inoculate 1 × 10⁵ cells/ml into fresh wort of equal concentration again. This activity was sequentially replicated many times to reach 100 generations (depending on the doubling time observed when monitoring each growth). For the second phase, we carried out the same protocol using wort with a malt extract concentration of 300 g/L (maltose and maltotriose concentrations around 120 g/L and 40 g/L, respectively). This wort had a final density of 1.125 g/mL (~29°P). This phase was carried out over 100 more generations (Figure 1). The cells were transferred in each passage, isolating colonies at each stage. Subsequently, a third phase of evolution was performed, where the culture temperature was reduced to 18°C and agitation was limited. We repeated this activity until a total of 600 generations were reached (Figure 1). Culture samples were taken for each repetition, and a cell count was performed in a Neubauer chamber. In addition, to monitor the cell viability, a pour-plating was carried out on soft agar plates, using wort at the same concentration of the liquid wort used. At the end of each evolution phase, between 3 and 10 colonies were taken per plate and were evaluated in YPM (yeast extract, peptone and maltose), using a maltose concentration of 120 g/L. For this, multi-well plates were used, where growth was evaluated through optical density quantification. These quantifications were carried out using Tecan^® equipment (Männedorf, Switzerland). The microplates were inoculated using six replicates at a concentration of 1 × 10⁵ cells /ml; they were incubated at 20 ° C with or without shaking. In the adaptive evolution process, the initial yeast strains were called “parental” and the strains obtained at the end of the evolution process were called “offspring” or “evolved strain.”

As detailed earlier, the evolution program consisted of isolated colonies using synthetic wort agar plates at each stage. Then, different maltose concentrations were used (150 or 300 g/L) depending on the stage. The colonies obtained for each strain during this process were evaluated for fermentation kinetics phenotypes in YPM broth (yeast extract, peptone, and maltose), using 120 g/L maltose concentration. This evaluation was carried out in 96-well Cell Culture Plates (Thermofisher, United States), using a microplate reader (Tecan, Swiss).

The selected “evolved” strains obtained from the adaptive evolution process were evaluated using Pale Ale wort. The strains were inoculated in wort at a cell concentration of 5×10⁶ cells / ml in triplicate, using a 10-liter reactor (Biocl, Chile). The cultures were incubated at 18°C without agitation for 7 days. Then, they were decanted at 4°C for 5 days, and then 300 mL were packaged in 330 mL amber beer bottles. Cells of evolved strains used for inoculation were obtained directly from their propagation in YPM-120 broth (3 g/L yeast extract, 5 g/L peptone, 10 g/L glucose, 120 g/L malt extract). They were incubated at 18°C for 2 days. On the other hand, cells of parental strains used for inoculation were obtained from an adaptation process in which maltose concentration was increased from 30 to 120 g/L. After this process, incubation temperature was decreased from 25°C to 18°C.

We used Pale Ale wort and Kent Golding hops because their combination makes a neutral beer, which allows us to evaluate the contributions of flavor and aromas provided by the yeast. We mixed Pale Ale malt grain with water (1:4) previously heated to 70°C. Next, the mixture was heated at 65°C for 90 min with stirring every 15 min. After 30 min we included Kent Golding hops (0.8 g/L) was added to the mixture. Then, we filtered the mixture, separated the broth from the malt grain, and heated this new mixture at 100°C for 30 min. Finally, the wort was rapidly cooled, filtered and stored at 4°C. The density and pH of the wort were 1,050 g / ml and 5.1, respectively.

Yeast growth was followed spectrophotometrically by measuring the absorbance at 600 nm. Viable cell counts were determined by plating on WL agar, and plates were incubated at 28°C for 2 days. Moreover, the viability of each inoculum was confirmed by cell counting in a Neubauer chamber, using methylene blue. Flocculation was determined by Helm’s test (Bendiak, 1994). Ethanol, glucose, fructose, glycerol, maltose, and maltotriose were quantified by high-performance liquid chromatography (HPLC), using a Shodex sugar SH1011 column (Phenomenex, United States). A Shimadzu Prominence HPLC equipment (Shimadzu, United States) was used. In addition, 5 mM H₂SO₄ was used as the mobile phase, in an isocratic flow of 0.6 mL / min, with the oven temperature set at 5°C.

Differences between measurements were determined using the Mood test (Gibbons and Chakraborti, 2011) using the statistical software Statgraphics Centurion (v5.0). We observed significant differences when p-values were less than 0.05.

Two sensory analyses were carried out during this study. The first analysis was performed during “fermentative behaviour analysis of the selected strains in synthetic wort,” where the wort fermented was evaluated by five judges with Beer Judge Certification Program (BJCP) certification. The evaluation considered aroma and taste as quality parameters. A score scale was used to qualify, a score of 1 indicated “I dislike” and score 5 indicated “I like a lot.”

The second analysis was performed at the end of the study, in which beer was produced on a pilot scale. We prepared 10-liter reactors (Biocl, Chile) with Pale Ale wort for each inoculated yeast in triplicates. The yeasts used were (1) Lallemand Nottingham™ (brewer’s yeast) and (2) L261col5max (yeast evolved). The L261 (parent) strain was not evaluated because it did not grow in wort. Additionally, fermentation blanks, corresponding to Pale Ale wort, were included and prepared under the same conditions as the samples but not inoculated with yeast. Therefore, 300 mL were packaged in 330 mL amber beer bottles to be later evaluated by 25 judges with and without training.

A hedonic test was performed with a score range of 1 to 5. The details of this scale are shown in Supplementary Table S2 and Supplementary material A1. Attributes evaluated were appearance, color intensity, beer clarity, foam consistency, aromas, esters, flavor complexity and beer body.

In addition, a preference test (Ranking) was carried out, where we asked the judges which beer they liked the most.

For each beer analyzed, 3.5 mL of sample were deposited in a vial with 1.4 g of NaCl. These vials have a magnetic lid for the SMPE-GCMS technique. The analyses were performed using the Thermo TSQ DUO equipment, under the following conditions: HEADSPACE-SPME: Samples were incubated for 20 min at 30°C with agitation, using DVB/CAR/PDMS fibers. Fiber cleaning was performed at 250°C for 2 min before incubation and after injection, for each sample. GC/MS conditions: The injector was set to 200°C and operated in split-less mode with helium as the carrier gas at a flow rate of 1 mL/min. Column temperatures program (RTX-5 ms, Restek) was: 7 min at 35°C, then ramped to 200°C at 8°C/min and hold 5 min, then ramped to 250°C at 20°C/min. The mass range evaluated was 20–350 m/z with a transfer line temperature of 280°C and ionization source temperature of 250°C.

Standards used: 1-Butanol, 3-methyl-acetate, Ethyl butyrate, Ethyl 2-methyl butyrate, Ethyl acetate, 2,4,6 Trichloroanisole, Ethyl 3-hydroxyhexanoate, γ-Decalactone, 2,3 Butadione, Butyric acid, 2-Methoxy-4-vinylphenol, 2-butanol and Phenol, 2,4-dichloro were purchased from Merck-Sigma Aldrich (United States).

NIST Mass Spectral Library used: The library’s search for compounds by the library was performed by comparing the mass spectra obtained using the “National Institute of Standards and Technology” software version 2.2.

Cells of L261 and L261 col5 max strains were grown from glycerol stocks on YM plates at 28°C. A single colony was cultivated in 50 mL YM at 28°C without agitation for 24 h; 100 mL Erlenmeyer were used. The cells were collected, washed in TE, and its DNA was extracted using a Wizard kit (Promega, United States). DNA samples were submitted to whole-genome sequencing at Beijing Genomics Institute (BGI) using the Illumina HiSeq2500 technology to a sequencing depth of about 100X.

Raw reads were cleaned using the following steps: (1) Remove reads with at least 40% of low-quality (Q ≤ 20) bases, (2) Remove reads with at least 40% of Ns, (3) Remove adapter contamination, and (4). Reads were mapped against the S. cerevisiae strain DBVPG6765 genome (Yue et al., 2017) using BWA, after which duplicates were marked using Picard Tools. Variant calling was performed using freebayes v1.3.6 and were further filtered using vcffilter (“QUAL > 1 & QUAL / AO > 10 & SAF > 0 & SAR > 0 & RPR > 1 & RPL > 1”). The effect on the encoded protein sequence of each variant was predicted using SnpEff v4.3. Variants of interest were inspected manually by examining the alignments on a genome browser. Aneuploidies were inspected by examining genome wide sequencing coverage using tools following the pipeline by O’Donnell et al. (2023). Copy number variants were calculated using CNVkit (Talevich et al., 2016).

Ploidy were evaluated using flow cytometer BD FACS Melody (BD Biosciences, Canada) and Fx Cycle stain (Thermofisher, United States) (Supplementary Figure S2; Supplementary material A1). Yeast cells were inoculated into 35 mL liquid YPD medium and incubated overnight at 25°C with 150 rpm of agitation to obtain 5–8×10⁶ cell/ml. Then, 10 mL of culture was centrifuged at 4000 rpm for 3 min and at 10°C, washed twice with distilled water, and 15 mL of 70% (v/v) ethanol was added. It was incubated at room temperature for 30 min and then centrifuged. The cells were washed 3 times with distilled water. The cells were counted under a microscope and 1×10⁶ cells/ml was taken in a microfuge tube and then the fluorophore PI/ARNasa FxCycle™ (Thermofisher, United States). It was incubated in the dark at 25°C for 1 h. Finally, cells were filtered with a 40 uM cell strainer Falcon™. S. cerevisiae BY4741 (haploid) and BY4743 (diploid) were used as control.

To select a group of S. cerevisiae strains exhibiting promising brewing traits to subjected to an adaptive evolution process, 50 yeast strains were screened for their ability to consume maltose and maltotriose in malt extract wort. These yeasts were isolated from vineyards from different geographical places in Chile, and are part of Kayta’s company culture collection in collaboration with LAMAP-University of Santiago of Chile (Supplementary Table S3; Supplementary material A1). All strains evaluated were able to consume maltose, but only 22 of them consumed maltotriose (Supplementary Table S3; Supplementary material A1). We selected 13 strains due to their higher maltotriose (7 g/L) and maltose (16 g/L) consumption (Supplementary Table S3, marked strains; Supplemental material A1). Furthermore, the results indicated that strains L-3536, L-515, and L-166 consumed 20% of maltotriose, which is 50% less than that of the selected yeasts. However, their maltose consumption was only 10, 80% less than the selected yeasts. This finding is notable because maltotriose, a trisaccharide composed of three glucose units, is generally considered more difficult for yeasts to metabolize than maltose, a disaccharide consisting of two glucose units.

A further selection of yeast strains was performed on synthetic wort to evaluate an array of traits relevant for brewing. The fermentative fitness of 13 strains was evaluated using synthetic wort (density of 1,060 g/mL). In this phase, the 12jj strain was promptly discarded due to its slow growth. We observed higher maltose consumption (around 53 g/L) in cultures inoculated with L718, L168, L159, L167, L249, or L170 strains. Additionally, higher maltotriose consumption (around 2 g/L) was observed in cultures where L440, L512, L2890, L718, L169, or L170 strains were used. Moreover, we observed the highest attenuation by the L261 strain (72.5%), and this strain had one of the highest ethanol productions (Table 1). Glucose and fructose present in wort were depleted by all strains evaluated. In general, all strains produced around 2 g/L of glycerol, and around 0.2 g/L of acetic acid (% CV < 10), although the L2890 strain showed higher acetic acid production, reaching 0.3 g/L. Additionally, sensory evaluation tests were carried out on the cultures evaluated. We performed these tests by judges certified by the Beer Judge Certification Program (BJCP). The judges added to the evaluation test two parameters: aromas and tastes (Supplementary Figure S3; Supplementary material A1). The judges evaluated beers produced by L169, L170, L718, 8jj, L261 and L167 strains, highlighting their differentiating aromas and tastes. The judges agreed that these descriptors could generate beers different from currently available commercial beers. Finally, we selected five strains to subject them to an adaptive evolution process (L-170, L-261, L-718, L-169, and L-167); these strains produced beer showing the highest attenuation (around 64%), and ethanol production (around 3% v/v). Moreover, these strains showed the highest maltotriose (2 g/L) and maltose (53 g/L) consumption. Additionally, we selected the 8jj strain for its innovative flavor, as it was highlighted by the judges during the sensory analysis. Therefore, the strain 8jj strain was also subjected to adaptive laboratory evolution.

An adaptive evolution process on synthetic wort was performed to improve brewing traits of the selected five strains. To create a strain capable of consuming maltose as a carbon source, maltose concentration was increased at three points during the evolution process. At the end of each of the three phases of the evolution process (described in Materials and Methods), we isolated 3 to 10 colonies per plate. In each test round, the strains showing the shortest lag phase were selected to continue the adaptive evolution process [growth kinetics shown in (Supplementary Figure S4; Supplementary material A1)]. Finally, strains that showed significantly higher growth than the parental strain were selected. Therefore, evolved colonies originating from the parental strains L261 (designated L261 col5 max), L169 (designated L169 max), and 8jj (8jj designated max) were selected. The name scheme was based on the highest growth of the group under evaluation (max). In the case of L261, where several colonies showed similar high growth, we randomly selected colony number 5 (col5). These strains are called “evolved strains” hereafter.

We evaluated the three evolved strains in natural Pale Ale wort. In addition, we used the commercial brewing strain “Nottingham” as a fermentation control. We carried out fermentation experiments and determined the following parameters: sugar consumption, ethanol production and attenuation. Pale Ale wort initially contained 65.6 g/L of maltose, 25.1 g/L of maltotriose, 12 g/L of glucose, and 3.8 g/L of fructose. All the strains depleted the glucose and fructose present in the wort. Interestingly, the L261 col5 max strain consumed higher levels of maltose and maltotriose than the other evolved strains. Likewise, higher ethanol production and attenuation (74%) were observed for the L261 col5 max strain (Table 2).

Additionally, we determined the flocculation percentage by applying the Helm’s test (Bendiak, 1994), obtaining 78% for L261 col5 max and control strains (commercial strain Nottingham™). Then, the L261 col5 max strain was selected for further evaluation.

A hedonic test was performed using a 1-to-5 scale. The details of this scale are indicated in Supplementary Table S2 and Supplementary material A1. This analysis compared the beers produced by the evolved strain L261 col5 max and the control. The parental strain L261 was not evaluated because it was not able to grow under brewing conditions. According to the judge’s opinion, beers produced by L261 col5 max strain showed higher scores in esters complexity compared to the control strain, the commercial beer brewing strain Lallemand Nottingham™ (Figure 2); these differences were statistically significant (Mood test, p < 0.05). Interestingly, the other sensory attributes evaluated did not show statistically significant differences. In addition, the preference test placed at 1st place beers produced by L261 col5 max strain (Figure 2).

Volatiles produced in beers done by the evolved and control commercial strains were analyzed by GC–MS, using 12 standard compounds and compounds present in the library. Only four of the twelve standard compounds were detected, which were also quantified. The beers produced by the L261 col5 max strain had a greater presence of “green” odors such as green apple (Ethyl 2-methyl butyrate), green grass (Ethyl acetate), cidre (2-Methylbutyl octanoate, 2-methylbutyl ester), green fruity (Ethyl trans-4-decenoate) and citrus (Linalool (terpene)) compared to beers produced using the Nottingham™ strain. Additionally, the evolved strain produced more fatty alcohol than the commercial Nottingham strain, and showed particular odors such as sarsaparilla (2-Undecanol (fatty alcohol)) and floral oily (1-Hexadecanol, 2-methyl-(fatty alcohol)), which demonstrates its aromatic differentiation (Table 3). In contrast, the beers produced by the Nottingham strain mainly showed the presence of compounds related to apricot pear (Octanoic acid, ethyl ester), floral rose, (Acetic acid, 2-phenylethyl ester), and green banana (Hexanoic acid, ethyl ester) (Table 3).

To explore the genetic mechanisms underlying the differences observed in beers produced by the parental L261 and the evolved L261 col5 max strains, we performed whole genome sequencing of both strains using Illumina technology. Initial inspection of novel variants appearing in the evolved strain showed a high number of sites that changed from a heterozygous state in the parental strain (9,163 sites) to either homozygous reference (4,053 sites) or homozygous alternative (4,305 sites) in the evolved strain. This strongly suggested the complete loss of a genome copy (haplotype) during the adaptive evolution process. We further examined alleles with loss of function mutations in a heterozygous state in the parental strain to determine which allele the evolved strain inherited (Supplementary Table S1; Supplementary material A2). Thirteen genes were found to be heterozygous for a non-functional allele in the parental strain but homozygous for the functional allele in the evolved strain. These included genes involved in nutrient transport, utilization and biosynthesis in starving condition such as HXT13, AGP3 and VBA2 genes, respectively (Kocaefe-Özşen et al., 2022). Moreover, we identified genes related to lipid biosynthesis (SPT23), and genes encoding Chromatin Remodeling Factors related to DNA repair such as RSF1 (Lans et al., 2012; Min et al., 2013; Price and D’Andrea, 2013); and genes associated with adaptation to stressful or changing environmental conditions (SSQ1, DOG1, PKR1).

Interestingly, genes involved in Pheromone-Regulated Membrane protein (PRM7 and PRM9), related to Gcn4 under starving condition (Ayers et al., 2020), were inherited as functional allele in the evolved strain.

Instead, the evolved strain inherited 11 alleles that had loss of function mutations. These included the glycogen debranching enzyme codified by GDB1 (Walkey et al., 2011), a protein with homology Bul1p, involved in down-regulation of the general amino acid transporter Gap1p codified by BSC5 (Novoselova et al., 2012); and the gene encoding the zinc-finger protein involved in transcriptional regulation of genes required for glycerol consumption codified by RSF2 (Böhm et al., 1997; Lu et al., 2005; Tkach et al., 2012). Moreover, OAF3, SHH4, CFR1 and HXT11 genes were inherited loss of function alleles. Specifically, OAF3 gene codes for a putative transcriptional repressor with Zn(2)-Cys(6) finger; which negatively regulates transcription of metabolic pathways involved in fatty acid degradation (MacPherson et al., 2006; Smith et al., 2007; Tkach et al., 2012); SHH4 gene codes for a putative alternate subunit of succinate dehydrogenase (SDH) (Chang et al., 2015); CRF1 gene codes for a transcriptional repressor of ribosomal protein (RP) gene transcription, via the TOR signaling pathway-codified (Kumar et al., 2021; Martin et al., 2004), and HXT11 gene codes a hexose transporter to a broad range of substrates (Ozcan and Johnston, 1999).

Next, we explored novel mutations in the evolved strain. We found only four novel mutations; interestingly, three were located in the YNL034W gene, which encodes a protein required for sporulation. However, only one of these three mutations had a predicted change in the protein sequence (Supplementary Table S1; Supplementary material A2).

We inspected whether there were aneuploidies in the evolved strain by checking relative coverage differences across chromosomes. The parental strain did not show any aneuploidy. However, we found 19 aneuploidy events in the evolved strain, with at least one per chromosome, varying from +4 copies at the ends of chromosomes 1 and 6, to +1 copies at the start and ends of chromosomes 4, 11,12 and 13 (Supplementary Table S1; Supplementary material A2). These results were corroborated by analyzing copy number variations between the evolved and parental strains, where we found 226 genes showing higher copy numbers in L261 col5 max, including mostly duplicated copies (Supplementary Table S1; Supplementary material A2). Among these genes, we found duplications in IMA1, MAL13, and MAL11, related with maltose metabolism. We also found 160 genes that lost a copy in the evolved strain, three of which were found at a complete loss: FLO5, PAU8, and SEO1. Moreover, we observed non-synonymous changes in sequence of the FLO1 gene.