Hybridization of Saccharomyces cerevisiae Sourdough Strains with Cryotolerant Saccharomyces bayanus NBRC1948 as a Strategy to Increase Diversity of Strains Available for Lager Beer Fermentation

The search for novel brewing strains from non-brewing environments represents an emerging trend to increase genetic and phenotypic diversities in brewing yeast culture collections. Another valuable tool is hybridization, where beneficial traits of individual strains are combined in a single organism. This has been used successfully to create de novo hybrids from parental brewing strains by mimicking natural Saccharomyces cerevisiae ale × Saccharomyces eubayanus lager yeast hybrids. Here, we integrated both these approaches to create synthetic hybrids for lager fermentation using parental strains from niches other than beer. Using a phenotype-centered strategy, S. cerevisiae sourdough strains and the S. eubayanus × Saccharomyces uvarum strain NBRC1948 (also referred to as Saccharomyces bayanus) were chosen for their brewing aptitudes. We demonstrated that, in contrast to S. cerevisiae × S. uvarum crosses, hybridization yield was positively affected by time of exposure to starvation, but not by staggered mating. In laboratory-scale fermentation trials at 20 °C, one triple S. cerevisiae × S. eubayanus × S. uvarum hybrid showed a heterotic phenotype compared with the parents. In 2 L wort fermentation trials at 12 °C, this hybrid inherited the ability to consume efficiently maltotriose from NBRC1948 and, like the sourdough S. cerevisiae parent, produced appreciable levels of the positive aroma compounds 3-methylbutyl acetate (banana/pear), ethyl acetate (general fruit aroma) and ethyl hexanoate (green apple, aniseed, and cherry aroma). Based on these evidences, the phenotype-centered approach appears promising for designing de novo lager beer hybrids and may help to diversify aroma profiles in lager beer.


Introduction
Interspecific hybridization is an evolutionary force that shapes novel phenotypic and genomic profiles and may lead, via adaptive introgression or polyploidization, to the formation of new species [1,2]. In allodiploid hybrids, two divergent sub-genomes integrate and evolve within the same nucleus, with different possible phenotypic outcomes [3]. Genes that have never been tested together by natural selection may generate negative epistatic incompatibility, resulting in phenotypically intermediate hybrids that are suboptimally adapted compared to the parents [4,5]. An alternative outcome of hybridization, known as Another solution to expand the portfolio of available brewing strains is to exploit strains from alternative food-related niches, such as sourdough, Brazilian spirits and sake [40][41][42][43][44][45]. However, these strains typically lack the ability to ferment wort at low temperature and have been proposed for ale brewing and other specialty beer styles [44]. In a previous study, for example, sourdough strains were proven to have good brewing aptitude and to produce "sahti" beer enriched in flavor compounds [45]. Here, we combined bioprospecting for novel yeasts from niches alternative to beer and hybridization and constructed interspecific hybrids between these S. cerevisiae sourdough strains and the cold-tolerant S. bayanus NBRC1948. This strain was recently demonstrated to be an S. eubayanus × S. uvarum hybrid with some introgression from S. cerevisiae [8,30,46,47]. We evaluated the best conditions to increase hybridization yield using a marker-less spore-tospore method. We also compared the technological behavior of the obtained triple hybrids at laboratory scale in comparison to S. cerevisiae × S. uvarum hybrids. Finally, the potential application of the best S. cerevisiae × NBRC1948 hybrid candidate was demonstrated in two 2 L-scale fermentation trials.

Strains, Culture Conditions and Chemicals
Strains used in the present work are listed in Table 1. For growth, yeasts were cultured on YPDA medium (1% w/v yeast extract, 1% w/v peptone, 2% w/v dextrose, 2% w/v agar) for 48 h at 28 • C and then stored at 4 • C for the duration of work. For long-term storage, the strains were maintained at -80 • C in YPD medium (1% w/v yeast extract, 1% w/v peptone, 2% w/v dextrose) supplemented with 25% (v/v) glycerol as cryopreservation agent. All media and chemicals used in this study were purchased from Sigma Aldrich (St. Louis, MO, USA), except where differently indicated. Strains used in the present work are listed in Table 1. For growth, yeasts were cultured on YPDA medium (1% w/v yeast extract, 1% w/v peptone, 2% w/v dextrose, 2% w/v agar) for 48 h at 28 • C and then stored at 4 • C for the duration of work. For long-term storage, the strains were maintained at -80 • C in YPD medium (1% w/v yeast extract, 1% w/v peptone, 2% w/v dextrose) supplemented with 25% (v/v) glycerol as cryopreservation agent. All media and chemicals used in this study were purchased from Sigma Aldrich (St. Louis, MO, USA), except where differently indicated.

Maltose and Glucose Consumpion Tests
Maltose and glucose fermentation tests were carried out utilizing the procedure described by Kurtzman et al. [50]. Briefly, basal medium (0.45% w/v yeast extract, 0.75% w/v peptone and 0.0048% w/v bromothymol blue as pH indicator) was previously autoclaved, then sugar stock solution was added to a final concentration of 2% (w/v), and dispensed in test tubes with screw caps and containing inverted Durham tubes. The tubes were inoculated with 10 µL of microbial suspension, kept at 28 • C and monitored for 21 days for the production of gas. Scores were attributed according to Kurtzman et al. [50] as follows: +, strongly positive, insert filled within 7 days; s, slowly positive, insert slowly filled after more than 7 days; −, no gas production.
Growth ability was also evaluated by inoculating each strain at the final concentration of 10 5 cell/mL in YNB medium (0.67% w/v Yeast Nitrogen Base with amino acids and ammonium sulfate, BD Difco, Sparks, MD) supplemented with 2% (w/v) either glucose or maltose as carbon sources. Optical absorbance was measured at 600 nm after 48 h of incubation at 25 • C under shaking conditions (150 rpm) in three independent replicates [40].

Sporulation Efficiency, Spore Viability and Generation of Spore Clones
For sporulation tests, yeasts were sub-cultured in YPDA medium at 28 • C for 24 h, transferred to sporulation medium (ACM; 0.5% w/v sodium acetate, 2% w/v agar; pH 6.5) and incubated at 28 • C for a period of 14 days. Asci formation was microscopically checked after 3, 7, and 14 days and scored according to Kurtzman et al. [50].
Sporulation efficiency was determined by resuspending cells/asci mixtures from sporulation medium into physiological water (approximately 1 × 10 µL-loop into 1 mL physiological water). The total number of cells and the number of asci were counted in a Bürker chamber under an optical microscope (Carl Zeiss, Oberkochen, Germany). Sporulation efficiency was calculated as follows: For each strain, at least six tetrad asci were gently digested with 0.2 U of Zymolyase 20 T (AMSBIO, Abingdon, UK) for 20 min at RT and dissected on YPDA plates using a Singer MSM Manual micromanipulator device (Singer Instruments, Somerset, UK), according to [49]. Single spores were incubated at 28 • C for 48 h. Spore viability was calculated as follows: Spore Viability % = (n • of vital spores/n • of dissected spores) × 100. ( The obtained spore clones were streaked on YPDA plates and stored at 4 • C for the duration of work.

Mating Competence Assay
Spore clones were tested for mating competence in mixture cultures with either BY4741 (MATa) or BY4742 (MATα) mating testers on YPDA plates according to Kurtzman et al. [50]. Plates were incubated at 28 • C and conjugative bridges were checked at 8, 24 and 48 h after mixing the cultures.

Construction of Inter-Specific Hybrids
Hybrids were generated by the spore-to-spore mating method [49], with the exception of the S. cerevisiae × Saccharomyces cariocanus cross, where S. cerevisiae vegetative haploid monospore clones were used instead of spores. A scheme of hybridization strategy is reported in Figure 1. S. cerevisiae × S. uvarum RC2-10 and S. cerevisiae × S. bayanus NBRC1948 crosses were carried out in staggered mating (SM) and non-staggered mating (NSM), while S. cerevisiae × S. cariocanus cross in SM only. Mating involved spores from wild-type strains and/or spores from monospore clones in S. cerevisiae × Scheme 2. and S. cerevisiae × S. bayanus NBRC1948 crosses. Mating involved either spores from wild-type strains or mating-competent cells in S. cerevisiae × S. cariocanus CBS8841 cross. Abbreviations: SM, staggered mating; NSM, non-staggered mating; Sc, S. cerevisiae; Su, S. uvarum; Sbay, S. bayanus; Scar, S. cariocanus; P1, parental strain 1; P2, parental strain 2.
In detail, for each mating trial, at least three independent crosses were attempted. S. cerevisiae has been reported to have a germination time lower than other Saccharomyces non-cerevisiae species and this difference negatively impacts interspecies mating [51]. In non-staggered mating (NSM) trials both parental asci were dissected on YPDA mating plates and spores were immediately placed in contact with one another. In staggered mating (SM) trials, S. cerevisiae asci were dissected 4 h later than the Saccharomyces noncerevisiae asci in order to discern whether hybridization yield increases in SM compared to NSM assays. Mating plates were incubated at 30 • C for 3-4 h and hybrid candidates were streaked at least in duplicate in YPDA medium and then cultured in YPD broth for approximatively 20 generations in order to assure genetic stability of the new hybrids. Finally, hybrids were cryo-preserved at −80 • C in YPD medium supplemented with 25% (w/v) glycerol.
To test the effect of residence time in starvation conditions on mating propensity, cells were incubated on ACM medium and the plates containing spores were stored at 4 • C for 105 days. Asci age was calculated in days starting from the plating on sporulation medium. Mating propensity was calculated by spore-to-spore mating after 35 and 105 days of residence on ACM, respectively.

Molecular Methods
Yeast DNA was extracted from single colonies with the lithium acetate-SDS method [52].
MAT genotyping was carried out on sourdough strains and their monospore clones using primers described by Huxley et al. [53]. Strains BY4742 (MATα) and BY4743 (MATa) were used as internal controls. Hybrids were validated by PCR-RFLP analysis of ITS1 spacer with HaeIII enzyme (Thermo Fisher Scientific, Waltham, MA, USA) [54] and PCR amplifications of FSY1 and MEX67 genes using species-specific primers [55]. Genotyping of hybrids after genome stabilization was done by (GTG) 5 fingerprinting assay according to Dakal et al. [56]. All PCR reactions were carried out in a T100 Thermal Cycler (Bio-Rad, Hercules, CA, USA). All PCR mixtures were performed in 20 µL of final volume containing 1 µL of colony DNA as template, 0.4 µM of each primer, 200 µM each dNTP, and 0.5 U DreamTaq DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions.

Wort Fermentations
Micro-scale trials were carried out according to Catallo et al. [45]. Fermentations were performed in duplicate with 100 mL of 15 • Plato ( • P) all-malt wort (96.36 g/L maltose and 40.18 g/L maltotriose) in 250 mL Erlenmeyer flasks, without agitation. Airlocks containing 2 mL of 85% glycerol were used to seal the flasks. Weight loss of the flasks due to CO 2 release was measured with an analytical balance and used to monitor fermentation progress. A "neutral" fermentation temperature of 20 • C was chosen to support growth of all strains. Final measurements and samples were taken after 14 days. Ethanol concentration and pH values were determined from the centrifuged and degassed fermentation samples using an Anton Paar Density Meter DMA 5000 M with Alcolyzer Beer ME and pH ME modules (Anton Paar GmbH, Graz, Austria). After washing with deionized H 2 O, each yeast pellet was transferred to a pre-weighed porcelain crucible, dried overnight at 105 • C and weighed to determine the dry mass content. Viability percentage was determined in a NucleoCounter ® YC-100™ as previously reported [57] and calculated as follows: Fermentation curves were modeled based on the weight loss trend over time using the "grofit"-package for R [58]. Maximum rate of fermentation (µ) and maximum fermentation efficiency (total amount of CO 2 released at the end of the fermentation; A) were determined using the spline-fitting method in "grofit".
For 2 L-scale fermentations, an inoculation loop was first used to transfer a small amount of yeast to 50 mL YPD medium and propagated under aerobic conditions on a shaker (100 rpm) at 20 • C. After propagation overnight, the cells were washed in sterile Milli-Q-filtered water and re-suspended to achieve a 20% slurry (200 g fresh yeast/L). This slurry was used to inoculate 1 L of all-malt 15 • P wort in a 2 L Erlenmeyer flask at a pitching rate of 1 g/L. After a two-day incubation period at 20 • C and shaking (100 rpm), the flask contents were centrifuged and yeast was again washed and re-suspended to achieve a 20% slurry. In this case the fermented wort, rather than water, was used for resuspension of cells. This slurry was used to inoculate 1.5 L of 15 • P wort within 2 L-scale sterile stainless steel cylindroconical vessels at a pitching rate of 1 g/L. Fermentations were conducted at 12 • C, as similarly low temperatures are typical for the production of lager-style beer, and low-temperature tolerance is an essential criterion for the selection of prospective lager brewing yeasts. Samples were taken aseptically throughout the fermentation period.
Samples for yeast-derived flavor compound determinations were drawn from the beer when fermentations were ended and immediately frozen until the analysis.

Chemical Analysis
Yeast-derived flavor compounds were determined by headspace gas chromatography with flame ionization detection (HS-GC-FID). Briefly, samples of 4 mL were filtered (0.45 µm), incubated at 60 • C for 30 min, and then 1 mL of gas phase was injected (split mode; 225 • C; split flow of 30 mL/min) into a gas chromatograph equipped with an FID detector and headspace autosampler (Agilent 7890 Series; Palo Alto, CA, USA). Analytes were separated on a HP-5 capillary column (50 m × 320 µm × 1.05 µm column, Agilent, Palo Alto, CA USA). The carrier gas was helium (constant flow of 1.4 mL/min). The temperature program was 50 • C for 3 min, 10 • C/min to 100 • C, 5 • C/min to 140 • C, 15 • C/min to 260 • C and then isothermal for 1 min. Compounds were identified by comparison with authentic standards and were quantified using standard curves. 1-Butanol was used as internal standard (246 mg/L).
For sugar analysis, the samples collected at the end of the fermentation were centrifuged and the supernatants were filtered (0.45 µm) prior to storing in a freezer (-23 • C). Sugar content of wort was analyzed by HPLC. A Waters 2695 Separation Module and Waters System Interphase Module liquid chromatograph coupled with a Waters 2414 differential refractometer (Waters Co., Milford, MA, USA) was used. An Aminex HPX-87H Organic Acid Analysis Column (300 × 7.8 mm; Bio-Rad, Hercules, CA, USA) was equilibrated with 5 mM H 2 SO 4 (Titrisol, Merck, Germany) in water at 55 • C, and samples were eluted with 5 mM H 2 SO 4 in water at a 0.3 mL/min flow rate.

Statistical Analysis
Data on maltose and glucose consumption and on 2 L wort fermentation were statistically analyzed with two-way ANOVA (p < 0.05) followed by a Tukey's multiple comparisons test. Statistical analysis was performed on kinetics parameters and aroma concentrations with a one-way ANOVA and Tukey's test. All analyses were performed using GraphPad Prism software (GraphPad Software, Inc, San Diego, CA, USA). p values < 0.05 were considered as statistically significant.

Maltose Fermentation Screening
Maltose is one of the main fermentable sugars in sourdough because it is constantly generated by the degrading activity of amylase on starch [59]. Maltose represents 60% of carbohydrates in wort, a complex substrate also containing 10% of glucose and 25% of maltotriose as fermentable sugars [60]. Sourdough strains were proven to exhibit a good brewing aptitude [45] and could be promising parental candidates for brewing hybrid construction with Saccharomyces non-cerevisiae cryotolerant species. Fermentation tests showed that all S. cerevisiae strains, S. uvarum RC2-10, S. bayanus NBRC1948 ferment maltose, but not S. cariocanus CBS8841 ( Table 2).
To assess inter-strain differences in growth on maltose, biomass was also tested as absorbance after 48 h of fermentation in maltose and glucose-containing media. S. cerevisiae Y19, Y21, Y23 and Y26 were the best maltose-fermenting strains, while Y17 exhibited a slow-growing phenotype both in glucose and maltose-supplemented media. Strains Y19 and Y23 did not show any significant diversity in maltose and glucose consumption (p < 0.05) ( Table 2).  [50] as follows: +, strongly positive; s, slowly positive; -, no gas production. Optical density at 600 nm (OD 600 nm ) was evaluated after 48 h of fermentation in medium containing either glucose or maltose. Different superscript letters indicate significant differences (p < 0.05) in the same column, while different symbols (* and §) indicate significant differences in the same raw, as determined by ANOVA followed by Tukey HSD test. Superscript letters were attributed as follows: the highest value was marked as "a," the next value that is significantly different can be "b," and so on. Abbreviation: nd, not determined.

Species
Strain Fermentation Assay Optical Density (OD 600 nm )

Sporulation Efficiency and Spore Viability
In order to generate artificial hybrids by spore-to-spore mating, parental strains should sporulate and produce viable spores. Table 3 shows that the majority of strains sporulated after 3 days of incubation with a sporulation efficiency ranging from 17.2 to 35.9%. S. cerevisiae Y17 was the best sporulating strain (35.9%), while three strains, including S. uvarum RC2-10, and S. cerevisiae Y18 and Y26, showed an efficiency lower than 25%. Despite Y17 having the highest sporulation efficiency, it showed the lowest spore viability (29.2%), while other yeasts had spore viability values ranging from 45.8% for strains Y18 and Y19 to 66.7% for strain Y23. Overall, S. cerevisiae Y15, Y19, Y21 and Y23 exhibited the best phenotype in terms of both sporulation efficiency and spore viability. Three of them (Y19, Y21 and Y23) were the most suitable to grow on maltose as carbon source. Table 3. Sporulation efficiency, spore viability and MAT genotype. Strains sporulating after 3 and 7 days were scored as + and w (weak), respectively. Sporulation efficiency was determined as the mean of two independent replicates, while viability was calculated considering the dissection of six tetrads for each strain. All values of sporulation efficiency were statistically different based on one-way ANOVA. Abbreviation: nd, not determined.
Homothallism/heterothallism are critical factors that affect mating efficiency and hybridization yield [62]. Homothallic HO/HO diploid strains would result in tetrads with only diploid MATa/MATα spores, while heterothallic ho/ho strains would be associated with MATa and MATα progenies in a maximal ratio of 2:2, excluding death events. Conversely, heterozygous HO/ho strains can generate spores with either diploid or haploid genomes in a typical ratio of 2:2. To unravel the ability of sourdough strains to switch mating type, MAT multiplex PCR was also carried out on their meiotic segregants. All strains yielded the MATa/MATα segregant progenies with a frequency ranging from 46% to 73% (Figure 2). We also found MATa segregant clones (Figure 2) unable to sporulate and able to mate with mating tester BY4742 (MATα) (data not shown). All these results support that the MATa meiotic segregants are mating-competent and therefore suitable for outcrossing. Interestingly, MATα clones were rarely dissected, with the exception of two tetrads from strains Y18 and Y19 ( Figure 2). As BY4742 (MATα) was used as a positive control in PCR amplification of the MATα locus, we supposed that the majority of MATα haploid cells might be non-viable. Analysis of progeny distributions within each tetrad showed that diploid MATa/MATα occurred no more than twice for tetrad ( Figure 2). The coexistence of MATa/MATα and MATa spores in the same tetrad is congruent with a HO/ho genotype for the tested strains, suggesting that they are suitable for spore-to-spore mating.

Hybrid Construction
Sourdough strains Y15, Y19, Y21 and Y23 were chosen for hybrid construction on the basis of maltose consumption, sporulation efficiency and spore viability. Strain NBRC1948 (referred to as Sbay for convenience) was chosen for its peculiar genetic content, as being an S. eubayanus × S. uvarum hybrid with introgressed segments from S. cerevisiae [30]. S. uvarum RC2-10 (referred to as Su for convenience) and S. cariocanus CBS8841 (referred to as Scar for convenience) were chosen as alternative cryotolerant parents for comparative purposes.
Out of 190 attempted crosses, a total of 46 hybrids were generated and validated by molecular ITS1 PCR-RFLP assay and species-specific PCR (Supplementary Figure S1). Hybrid yield was higher than 22% for most of the cases ( Table 4). Out of 46 novel hybrids, three were between S. cerevisiae and S. cariocanus (Sc × Scar), 11 between S. cerevisiae and S. uvarum (Sc × Su), and 32 between S. cerevisiae and S. bayanus (Sc × Sbay). All three Sc × Scar hybrids, as well as three Sc × Su and 11 Sc × Sbay randomly selected hybrids were tested for sporulation. As opposed to Sc × Su hybrids, all the Sc × Sbay and Sc × Scar hybrids were able to sporulate (Supplementary Table S1).
Differences in germination time are some of the main prezygotic barriers among Saccharomyces species [63], so we performed pilot mating experiments under NSM and SM conditions. In the case of the S. cerevisiae × S. uvarum cross, we first attempted to gain hybrids by SM, which was previously reported to enhance hybridization yield [64]. Under SM, hybridization frequency was 27% for Y23 × RC2-10 crosses, whereas no hybrids were obtained when spores of S. cerevisiae Y15 were mated with spores of S. uvarum RC2-10 (Table 4). When the same strain pairs were tested under NSM conditions, hybrid frequency ranged between 20 and 31%, suggesting that differences in germination time were not relevant for these parental strains. In mating trials between S. cerevisiae and S. bayanus, SM did not enhance hybridization frequency compared with the NSM condition (Table 4).  S. cariocanus represents a particular S. paradoxus strain harboring four chromosomal translocations and which, consequently, is not interfertile with the S. paradoxus tester strain [18]. In S. cerevisiae × S. cariocanus crosses, we considered two individual interspecific pairings under SM, which was previously demonstrated to increase hybridization rates in S. cerevisiae × S. paradoxus mating trials [51]. Either spores from strain Y19 (MATa/MATα) or vegetative cells of the haploid monospore clone Y19.5A (MATa) were mated with spores from S. cariocanus CBS8841. By using 4 h of staggering time in both crosses, we obtained hybrid yields of 22% and 9% for Y19 × CBS8841 (spore-to-spore mating) and Y19.5A × CBS8841 (cell-to-spore mating), respectively.
During these pilot experiments, we also observed that asci maintained on ACM plates for long times (approximately 105 days) resulted in spores particularly prone to mate. To corroborate this observation, we performed Sc × Su and Sc × Sbay mating trials with both spores on ACM for 35 days and spores on ACM for 105 days. As shown in Table 5, old spores resulted in hybridization frequency higher than young spores, suggesting that spore age affects hybridization yield more than SM conditions.

Laboratory-Scale Wort Fermentation
Novel Sc × Sbay (Y15.2B × NBRC1948, Y19 × NBRC1948 and Y21 × NBRC 1948) and Sc × Su (Y23 × RC2-10) hybrids were selected for laboratory-scale wort fermentation together with their respective parental strains. One synthetic Sc × Su hybrid constructed for wine fermentation, namely LS3, was added for comparative purposes. In Figure 3 and Supplementary Figure S2, fermentation profiles show that Sc × Sbay hybrids retained the fermentation performance of the best Sc parental strains, while Sc × Su hybrids exhibited an intermediate kinetic curve compared with the parents.
Growth curve analysis revealed that Sc × Sbay hybrid Y15.2B × NBRC1948 exhibited higher maximum fermentation rate (µ) and maximum fermentation level (A) (p < 0.05) compared with the tested strains. This hybrid fermented faster than other hybrids and slightly faster compared with the Sc parental strain Y15.2B (Figure 4). Ethanol production ranged from 5.8% to 7.3% v/v after 12 days of fermentation at 20 • C (Supplementary Table S2). Sc × Sbay hybrids outperformed the Sc × Su hybrid in ethanol yield (p < 0.05). In contrast to previous studies involving ale strains crossed with S. eubayanus [28,29,65], both Sc × Su and Sc × Sbay hybrids were intermediate in alcohol production compared with the parental strains. Sourdough strains and their hybrids with NBRC1948 yielded ethanol at more than 82% of the theoretical value. Su strains were significantly lower in viability than Sbay NBRC1948 (p < 0.05).  S. cerevisiae, hybrids and S. bayanus/S. uvarum are represented as solid light gray, hatched gray and solid gray columns, respectively. Different letters indicate significant differences (p < 0.05) within each triad of hybrid and parents based on one-way ANOVA; different symbols ( §, # and *) indicate significant differences (p < 0.05) among hybrids Y23 × RC2-10, Y19 × NBRC1948 and Y15.2B × NBRC1948 based on oneway ANOVA. Superscript letters were attributed as follows: the highest value was marked as "a," the next value that is significantly different can be "b," and so on.

Two L-Scale Wort Fermentation
Based on the laboratory-scale fermentation results, we selected Sc × Sbay hybrid Y15.2B × NBRC1948 for 2 L-scale wort fermentation at 12 • C. Under these conditions, the hybrid exhibited a transgressive phenotype relative to its parents; it was able to ferment efficiently the wort sugars and also to tolerate low temperatures ( Figure 5), a trait that was not detectable from the previous small-scale fermentations that were carried out at the higher temperature of 20 • C. The S. cerevisiae parent performed relatively well in the first 10 days of fermentation, but the alcohol yield was limited in the latter stage of the fermentation, apparently due to an inability to utilize efficiently the available maltotriose, which was present in the final beer at approx. 20 g/L ( Figure 5). In contrast, the S. bayanus parent fermented more slowly, but was more efficient in terms of overall alcohol yield and sugar consumption, with residual sugar concentrations being similar to those of the hybrid strain.
The benefit of hybridization was seen not only in terms of fermentation performance, but also in beer quality. Four flavor-active aroma volatiles showed positive differences in concentration ( Figure 6). Acetaldehyde, which imparts an unpleasant chemical taste to beer and is particularly noticeable in lager beers, was at the flavor threshold in the S. bayanus beer. Low levels were observed in the S. cerevisiae beers, as well as those beers created with the hybrid strain. 3-methylbutyl acetate, which imparts a banana or pear flavor to beer and is one of the most desirable volatile compounds, was present at concentrations above the flavor threshold in the S. cerevisiae and hybrid strain beers. The S. bayanus beer concentrations were lower than the typical flavor threshold values and are not expected to contribute in any positive way to the overall flavor profile. A similar result was found for ethyl acetate (general fruit aroma) and ethyl hexanoate (green apple, aniseed, and cherry aroma) ( Figure 6).

Discussion
Bioprospecting efforts in brewing seek to utilize yeasts from environments other than the brewery in order to augment and/or diversify flavor properties of the final product [44]. Sourdough is a bio-reservoir of particular interest due to the occurrence of maltose-positive S. cerevisiae strains with QPS/GRAS status [27,67] and potentially wellaccepted by consumers for their provenance from artisanal food systems [68]. The search for new S. cerevisiae sourdough strains as wild stocks could have great potential for wheat and other specialty beers [40,45]. Our sourdough strains are well-suited both for fermenting maltose and for producing flavorful molecules like ethyl-and acetate-esters [45]. However, they do not display the complete pattern of industrial adaptive signatures specific for lager brewing, making the exploitation of genetic improvement techniques mandatory. The aim of this work was to demonstrate the suitability of a phenotype-centered strategy based on outcrossing of selected strains derived from non-brewing environments for increasing genetic and phenotypic diversity of starter cultures for lager-type fermentations.
Crossbreeding between maltotriose-fermenting S. cerevisiae ale yeasts and cryotolerant S. non-cerevisiae species is effective for creating novel GM-free synthetic hybrids based on mating between cells with opposite mating type by using marker-assisted approaches of spore-to-spore mating, mass mating and rare mating [21,28,29,33,65]. As our sourdough strains are able to sporulate and produce viable spores, they were mated with the cryotolerant S. bayanus strain NBRC1948 to generate synthetic hybrids via a marker-less spore-to-spore mating approach. Generally, mass mating is fast procedure that is preferred over the spore-to-spore method and, recently, novel methods such as fluorescence-activated cell sorting (FACS) [69] were proposed for increasing and assisting marker-less hybrid recovery after rare mating. Here we demonstrated that spore-to-spore mating can produce hybrids without the requirement of any selectable phenotypes for the parental strains and at relatively high frequency if the conditions of mating have been optimized. For instance, differences in germination time may contribute to reproductive isolation between species [51,63]. In contrast to what was previously found in outcrossing between S. cerevisiae and the phylogenetically distant species S. uvarum, isolated spores of S. cerevisiae sourdough strains and S. bayanus NBRC1948 appeared to germinate almost synchronously and did not require any staggered mating. This may depend on the peculiar genetic makeup of the selected cryotolerant parent NBRC1948, which contains several introgressed S. cerevisiae genomic segments [46]. We also found that a prolonged exposure to starvation positively affects mating propensity of isolated spores, enhancing the recovery of Sc × Sbay hybrids. Furthermore, heterothallic parental strains or wild strains with at least one inactive copy of the HO gene should be preferred over homothallic yeasts for breeding as isolated spores harboring the inactive ho allele are forced to restore diploidy by outcrossing. Here, we found that all sourdough strains gave a pattern of meiotic segregants compatible with a HO/ho genotype. Despite homothallism representing the most common life cycle in S. cerevisiae [60,70], heterothallic strains are frequently found in nature [71,72], while heterozygous HO/ho strains were also isolated in industrial environments for Brazilian spirit production [73]. The extent of loss of function in HO genes from natural sourdough strains deserves further investigations in future. Like in other baker's yeasts [74], the MATa idiomorph prevails in meiotic segregants, suggesting two possible explanations. MATa monospore clones could be more viable than MATα ones or, alternatively, sourdough strains could have a MATa/MATa/MATa/MATα genotype. Overall, these results highlight the propensity of sourdough strains for outcrossing and provide some guidelines for optimizing successful outcrosses in spore-to-spore mating experiments without any GM technology utilized.
Evaluation of the hybrids under lager brewing conditions clearly showed the potential of the S. cerevisiae × S. bayanus hybrid combination for industrial application. With one particular parental combination, the hybrid benefited from increased maltotriose utilization and higher concentration of esters in the resultant beers. While yeast hybrids occur in both natural and industrial environments [14,15,[75][76][77], they appear to be more common (or more persistent) in the latter. This suggests that the hybrid state confers a distinct advantage in such environments. Recent work involving artificially created hybrids has confirmed this suggestion, with newly created hybrids showing promise for application in a number of fermentation environments including beer, cider, and wine production [24,29,35,[78][79][80]. The de novo creation of lager yeast hybrids has been of particular interest due to the commercial importance of the S. pastorianus lager yeast and the low level of diversity within this group. Recreation of the S. cerevisiae × S. eubayanus hybrid has shown clearly that the success of S. pastorianus under low-temperature wort brewing conditions was due to two-parent transgression. The hybrid benefited from the superior ability of S. cerevisiae to utilize wort sugars (particularly maltotriose), and the psychrotolerance of S. eubayanus. Hybrids therefore had a competitive advantage over both ale yeasts and wild Saccharomyces yeasts when fermenting wort at low temperature [28]. The advantage is, however, not necessarily parent-species specific; any combination of species that introduces the required phenotypes should be effective. Nikulin et al. [33], for example, created yeasts with a lager yeast phenotype by combining S. cerevisiae with yeasts other than S. eubayanus. Saccharomyces mikatae, for example, was equally effective at transferring the cold-tolerant phenotype to the hybrid. There is therefore scope for creating yeasts with industry-relevant phenotypes without recapitulating the original species combination [81]. The example in the present study is the first to show that a yeast with a phenotype suitable for brewing can be created even when neither parent has been used as production strain for brewing. This phenotype-centered approach to parent selection increases the genetic diversity that may be tapped for hybrid design.
Previous attempts to create lager hybrids have utilized S. cerevisiae as the maltose-, and maltotriose-positive partner [28]. Of note here was the finding that the S. bayanus parent appeared to be more adept at wort sugar utilization than the S. cerevisiae parent. This, at first, may seem counterintuitive, given that maltotriose utilization is not typical for either of the species (S. eubayanus or S. uvarum) included in the species complex [33]. However, the S. bayanus strain utilized in this study (NBRC1948) is known to possess a considerable amount of S. cerevisiae-introgressed DNA and, in particular, a large (70 kb) region containing the maltose transporter gene MAL31, and the maltose and maltotriose transporter gene (MTY1/MTT1). These genes and a number of other S. cerevisiae-derived genes related to osmotic stress resistance, anaerobic growth, and sucrose utilization may have been key to the survival of this strain as a contaminant in a brewery environment (the original source of the strain) [40]. This same introgression is feasibly responsible for the improved brewing efficiency of the hybrid created in this study. It is perhaps significant here that the Mtt1 transporter is known to have a high affinity for maltotriose relative to other transporters [82], and this may explain the greater ability to consume the trisaccharide during fermentation. It may be hypothesized that the sourdough S. cerevisiae parent possesses a lower affinity transporter such as Malx1. It has also been found that the Mtt1 transporter performs well at low temperature relative to other yeast maltotriose transporters [83].
In conclusion, this study proves that synthetic triple hybrids involving phenotypebased selection of non-brewing parents could be useful for diversification of lager yeast stock cultures. Outcrossing experiments defined environmental conditions that optimize hybrid recovery without any utilization of auxotrophic marker and/or GM techniques. In being an S. eubayanus x S. uvarum hybrid, strain NBRC1948 exhibits cold-tolerance, but is also able to ferment maltotriose better than S. eubayanus due to the presence of MTY1/MTT1, a gene encoding the cold-tolerant maltotriose transporter described in S. pastorianus [82,84]. S. cerevisiae sourdough strains have good fermentative aptitude in wort and positively affect aroma profile, a phenotype that is positively inherited by hybrids. Phenotype-driven selection of parents followed by outcrossing is a promising biotechnological solution that combines different benefit phenotypes into a single brewing culture.

Conflicts of Interest:
The authors declare no conflict of interest. Kristoffer Krogerus and Frederico Magalhães are from VTT Technical Research Centre of Finland Ltd., the company had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.