A Saccharomyces eubayanus haploid resource for research studies

Since its identification, Saccharomyces eubayanus has been recognized as the missing parent of the lager hybrid, S. pastorianus. This wild yeast has never been isolated from fermentation environments, thus representing an interesting candidate for evolutionary, ecological and genetic studies. However, it is imperative to develop additional molecular genetics tools to ease manipulation and thus facilitate future studies. With this in mind, we generated a collection of stable haploid strains representative of three main lineages described in S. eubayanus (PB-1, PB-2 and PB-3), by deleting the HO gene using CRISPR-Cas9 and tetrad micromanipulation. Phenotypic characterization under different conditions demonstrated that the haploid derivates were extremely similar to their parental strains. Genomic analysis in three strains highlighted a likely low frequency of off-targets, and sequencing of a single tetrad evidenced no structural variants in any of the haploid spores. Finally, we demonstrate the utilization of the haploid set by challenging the strains under mass-mating conditions. In this way, we found that S. eubayanus under liquid conditions has a preference to remain in a haploid state, unlike S. cerevisiae that mates rapidly. This haploid resource is a novel set of strains for future yeast molecular genetics studies.


Results and discussion
Generation of a large collection of haploid derivates in S. eubayanus. S. eubayanus is a diploid yeast, homothallic, and has been isolated mainly from wild environments 36,49 . To generate heterothallic strains, we decided to knock out the HO gene in wild diploid strains from different lineages (Fig. 1a). The HO gene encodes a site-specific endonuclease that promotes mating-type switching from a to α, or vice versa, in haploid strains, and its deletion has been widely used for the generation of stable haploid cells in yeast 8,11,53 . To delete the HO gene, we used a plasmid expressing the Cas9 enzyme and a gRNA cloning site developed by Fleiss et al. 54 ; however, this plasmid had only been used in S. cerevisiae, without evaluating its efficiency in other Saccharomyces species.
We first evaluated the efficiency of the CRISPR-Cas9 system and the plasmid pAEF5 to obtain null mutants for the HO gene in the CL216.1 strain of S. eubayanus using four different gRNAs (gRNA1, gRNA2, gRNA3 and gRNA4, see methods), together with a 100-bp double stranded repair DNA fragment. HO gene deletion was observed in 11%, 0%, 42% and 0% of the investigated colonies for each gRNA (from 1 to 4), respectively. The deletion or mutation efficiencies obtained for S. eubayanus using CRISPR-Cas9 varies between studies, depending on the targeted region. In some cases, 100% efficiency was achieved in generating the targeted mutations (at the MALTX genes 55 ), whilst in others, the efficiencies varied between 3 and 40%, depending on the gene and the genetic background 56 . Also, diploid yeast cells exhibit a barrier in utilizing DSB-mediated transformation schemes because diploid cells prefer homologous chromosomes for DSB repair, rather than exogenous linear DNA, resulting in a minority of recovered strains acquiring the targeted modification 10 . However, the efficiencies obtained in our study are in agreement with the efficiencies reported elsewhere in different S. cerevisiae strains utilizing the CRISPR-based gene deletion approach (1.3-100%) 21   www.nature.com/scientificreports/ without the need to insert a selection cassette into the genome 20 . CRISPR-Cas9 has been used successfully in a few studies in S. eubayanus, generating null mutants 55 , integrating and overexpressing genes 58 , and introducing loss-of-function SNP mutations 56 . Therefore, based on our initial findings, gRNA3 was selected to repeat this procedure in a greater number of strains. We subsequently used a collection of S. eubayanus strains spanning three of the six main lineages in the species, together with a subset of admixed strains, comprising about 55% of the genetic diversity found in this yeast species 36,37 . In this way, the initial collection of strains for HO knockout considered 89 strains (Fig. 1b), of which 83 strains were previously isolated from 10 different locations in Chile belonging to the PB-1, PB-2 and PB-3 populations; admixed clusters (ADM), five North American strains that grouped in the PB cluster, and the reference strain from Argentina were also used (Supplementary Table S1). Overall, 59 of the 89 S. eubayanus isolates were successfully transformed with the plasmid pAEF5 (Fig. 1b), including 20 strains from PB-1, 11 from PB-2, 19 from PB-3 and nine ADM strains (Supplementary Table S1). S. eubayanus has been genetically modified using electroporation and the lithium acetate procedure, however, historically only three different genetic backgrounds have been mainly used 55,56,58 . In this context, this study achieved genetic editing on a broader panel of genetically-distinct S. eubayanus strains.
To obtain stable haploid derivates, transformants were sporulated and dissected by micromanipulation (Fig. 1a). Only tetrads with four viable spores were considered for MAT locus genotyping. In this way, we obtained MATa and MATα haploid versions for 57 out of the 59 parental strains containing the plasmid pAEF5 (Fig. 1b, Supplementary Table S1). Strains CL1002.1 and CL611.1 did not sporulate and we were unable to obtain haploid derivates. Furthermore, in strains CL704.2 and CL836.1 we did not obtain four viable spores from the same ascus; however, as we identified MATa and MATα versions in both cases, both were included in the final haploid collection. Finally, we evaluated spore viability in 36 out of the 57 successfully transformed isolates, obtaining spore viabilities between 15 and 100% (Fig. 1c). In all, only one isolate exhibited spore viabilities lower than 50%, six isolates presented spore viabilities between 50 and 75%, and 30 isolates had spore viabilities over 75%. The high proportion of strains able to sporulate (~ 97%) agrees with previous data observed in S. eubayanus 41 Table S2). In general, we observed that for most conditions and genetic backgrounds, there were no statistically significant differences in maximum growth rates between haploid strains and their respective parental strain (Fig. 2a). In this sense, we observed that in S. eubayanus most haploids and diploids exhibited similar performance under the environmental conditions tested in this study, where no general fitness advantages were evident in either ploidy state. Previously reported observations in S. cerevisiae and S. paradoxus using a more extensive set of conditions and strains also demonstrated no differences between either haploidy or diploidy 61 . However, we did find certain cases where one or both haploid versions showed statistically significant differences under certain conditions (Supplementary Table S2). We found specific ploidy x environment interactions, particularly for ethanol tolerance, where Mata haploid strains exhibited significantly lower growth rates than diploid strains (p-value < 0.05, Fig. 2b). The trend for the opposite mating type version MATα was similar, however, we found marginally significant differences in this case (p-value = 0.06, Student t-test). Indeed, DNA damage agents were previously shown to favor diploid asexual proliferation in S. cerevisiae and S. paradoxus compared to haploid strains 61 .
To analyze the whole phenotypic landscape in the set of 57 diploid and haploids strains, we performed a PCA analysis (Fig. 2c). The PC1 and PC2 components explain 33.8% and 15% of the observed variance, respectively, and combined, the two components account for 48.8% of the overall variation. PCA shows that growth in glycerol correlates negatively with the rest of the phenotypes where the carbon source was glucose or maltose. Interestingly, the individual factor map indicates no significant separation pattern according to ploidy and/or haploid strain mating-type, suggesting comparable phenotypes among diploid and haploid strains.
In addition, we also performed hierarchical clustering of the phenotypic data ( Fig. 2d), observing three main clusters. Notably, 60% of the parental strains (34) clustered together with their two haploid versions, 18% (10) of the diploid parents clustered with one of the haploid versions, while 23% (13) distributed across different clusters compared to their two haploid versions. These results corroborate that most of our haploid strains are representative of their parental genetic backgrounds, as previously described 61 . For example, the CL715.1 strain and its haploid variants did not show statistically significant differences in seven of the eight media evaluated. However, the alpha version showed statistically significant differences when cultivated in 10% ethanol (Supplementary Table S2). Phenotypic differences between haploid and parental strains have also been observed in other yeast collections 9,11 . Therefore, this extensive collection of S. eubayanus represents a valuable resource for molecular genetics studies.
Similar brewing fermentative profiles in haploid and diploid strains. Considering the relevance that S. eubayanus has gained in the last decade for lager beer fermentation, we evaluated the fermentative profiles of parental strains and their haploid versions under conditions of beer wort fermentation. For this, we selected four strains representative of different lineages (CL1106.1 (ADM), CL216.1 (PB-3), CL601.1 (PB-3) and CL715.1 (PB-2)) for a 15-day lager fermentation in a 12°Brix wort and monitored the fermentation outcome by estimating CO 2 loss. Overall, we did not observe statistically significant differences in the total CO 2 output between haploid strains and their respective parental strains, except for the CL601.1-a strain (Fig. 3). All strains completely consumed the sugars present in the beer wort, except for maltotriose, which was previously reported as not  www.nature.com/scientificreports/ being metabolized by S. eubayanus 62 . The haploid strains of CL216.1 and CL715.1-a showed significantly higher ethanol production than their respective parental strains, with differences of 0.74% (t-test, p < 0.001) and 0.64% (t-test, p < 0.001) for CL216.1-a and CL216.1-α strains, respectively, and 0.48% (t-test, p < 0.05) for the CL715.1a strain (Supplementary Table S3). In general, the results of the fermentation assay demonstrate that this set of haploid strains is representative of the phenotypic fermentative performance of their diploid parents. Obtaining phenotypically representative haploid variants of the wide diversity of S. eubayanus strains is of relevance for genetic and biotechnological studies. In particular, these strains can be used at the fermentative level, where genetic analyses explaining differences between strains at the phenotypic level have not been undertaken. For example, strains belonging to the PB-3 subpopulation have a higher fermentative capacity than strains belonging to PB-2 or PB-1 36 , reaffirming the existence of natural genetic variants in wild strains resulting in phenotypic diversity. In this way, this collection of haploid strains represents a valuable tool to be considered, for example, in the identification of QTLs.

Low levels of off-target mutations in sequenced haploid strains. Previous studies have described
off-target effects resulting from the use of a CRISPR-Cas9 system in genome editing 63 . In this sense, to evaluate off-target mutations, we sequenced the genome of three haploid strains, i.e., one spore of CL715.1, CL216.1, and CL601.1 (strains previously evaluated under fermentation conditions), using high-coverage whole-genome sequencing (WGS, Supplementary Table S4). After mapping the reads to the S. eubayanus reference genome (CBS12357) 55 , we performed variant calling and compared each haploid strain with their corresponding wildtype genotype that had been previously obtained using WGS 36 to discover novel off-target mutations. We initially found 6, 12, and 16 putative mutations in CL601.1, CL216.1, and CL715.1, respectively (Supplementary Table S5a). After manual inspection of the alignments, we confirmed two single-nucleotide mutations, both occurring in the CL216.1 haploid strain in chromosomes 4 and 8. Interestingly, the mutation located in chromosome 4 was upstream of the HO gene, within the donor DNA recombination region, where the new allele coincides with the reference sequence used to generate the donor DNA. Furthermore, the mutation at chromosome 8, located in the coding sequence of the PRO2 gene, was also observed in the CL216.1 diploid sequencing using long reads 35 , suggesting that this mutation arose before the CRISPR-Cas9 procedure, and was not detected in the original CL216.1 WGS 36 . Off-target mutations correspond to DNA modifications at unintended sites, such as SNPs, deletions, insertions or inversions 63 , and depend on gRNA and experimental conditions 64 . In Saccharomyces yeast, these off-target mutations could arise from NHEJ of endonuclease-mediated DSBs, by HR between the 100 bp DNA donor or by homologous sequences at other genomic locations similar to the target sites 57 . Studies have evaluated the presence of off-target mutations after genome editing in Saccharomyces yeasts, in which they have been shown to be rare in haploid or homozygous strains 56,57,65,66 . Furthermore, off-target effects are considered unlikely in a small genome such as that of yeast 21 , coinciding with the results obtained in our study. We also used haploid WGS to validate heterozygotes sites previously determined after WGS of diploid strains 36 . Strikingly, we found that out of the 807, 720 and 942 heterozygous sites called after WGS of the diploid strains CL216.1, CL601.1 and CL715.1, respectively, 98% of them were again identified as heterozygous sites in the haploid strains (Supplementary Tables S5b, S5b and S5b), suggesting that most heterozygote calls in diploids were false positives, and do not represent real heterozygotes sites, but rather polymorphic repeated regions or technical sequencing artifacts. These results suggest that the CRISPR-Cas9 technique used in our study likely  www.nature.com/scientificreports/ generated a low number of off-targets in the manipulated strains. These results should be confirmed by sequencing a more significant number of haploid derivates.. Although we found few heterozygous sites validated after WGS of the three haploid strains (16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29) (Supplementary Tables S5b, S5b and S5b), meiotic events could also lead to the generation of structural variants and/ or chromosome mis-segregation. Hence, we sequenced each of the four spores corresponding to one complete tetrad of the CL216.1 strain using nanopore long-read sequencing (Supplementary Table S6). Long-reads further confirmed the low number of heterozygous sites occurring in S. eubayanus, limited to only one site that segregated in a clear 2:2 proportion among spores ( Supplementary Fig. S1). The draft genomes of each spore showed the absence of structural variants among spore genomes and demonstrated a flawless meiotic event ( Supplementary Fig. S2). In general, meiosis provides a large source of genetic diversity by rearranging allelic combinations 67 , which is important to consider when haploid strain collections are generated. The main source of genetic diversity is meiotic crossovers (COs), that are reciprocal exchanges of chromosome arms between homologous chromosomes and are crucial for accurate homolog segregation at meiotic division I, and their absence leads to mis-segregation of homologs and aneuploid gametes 68 . Another source of genetic diversity during meiosis is the generation of punctual mutations. These normally occur in meiosis at higher frequencies than in vegetatively grown cells of S. cerevisiae, as a result of DSB repair in meiosis during the recombination process 69 . In this sense, we did not observe aneuploid or structural variants in any of the four spores derived from the CL216.1 strain, or de novo mutations, suggesting correct COs and homolog segregation during the meiotic process in CL216.1.
Intra-specific hybridization is common in S. cerevisiae, but not in S. eubayanus. Finally, we sought to evaluate the utility of our haploid collection by conducting, as proof of concept, two approaches to generate intra-specific hybrids (Fig. 4). First, we selected five strains that are representative of the different S. eubayanus lineages (CL449.1, CL715.1, CL216.1, CL601.1 and CL1106.1). The "Matα" and "Mata" versions of each strain were mixed in equal proportions and grown for 34 generations (four transfers) in a defined medium, and the proportion of haploid and diploid cells determined after each transfer (Fig. 4a). We evaluated the generation of diploid strains by colony isolation and identification of mating-type after each transfer. We observed a drastic decrease in MATα colonies in all three replicates. In replicates 2 and 3, we observed a predominance of MATa colonies, while a predominance of diploid colonies was observed in replicate 1 (Fig. 5a). Interestingly, MATa and MATα haploid strains did not exhibit growth rate differences under similar microculture conditions, and therefore the differences observed are specific under the dynamic competitive environment. A selective advantage for a certain trait is strongly dependent on environmental dynamics, where antagonism or competition can impact a strain's success 70 . Previous studies demonstrated that sterile haploid mutants could exhibit a growth advantage over wild-type haploid cells 71 suggesting that expression of mating pathway genes could differently impact the fitness of MATa and MATα cells.
The diploid colonies in replicate 1 could originate from crosses between different strains, or crosses between two opposite mating types of the same genetic background. In addition, diploid colonies obtained from the same replicate may be clones from the same single hybridization event. To pinpoint the source of this variation, we genotyped the polymorphic DCR1 gene in eight diploid colonies to identify their respective parental strains ( Fig. 5b and Supplementary Table S7). Five out of eight diploid colonies were identified as intraspecific hybrids: Three hybrids from replicate 1 (colony numbers 1, 3 and 4) and two from replicate 2 (colony numbers 6 and 7). Interestingly, in replicate 1 the CL449.1 strain predominated in both types of colonies, that is intraspecific hybrids and in self-crosses between strains of the same genetic background. On the other hand, CL715.1 strain predominated in replicates 2 and 3 in the two intraspecific hybrids (Supplementary Table S7).
To quantify the fitness of the hybrid-diploid colonies after the final serial transfer, we evaluated five, two and one diploid colonies from replicate 1, 2 and 3, respectively, under microculture conditions in 2% glucose, 5% ethanol and 1.25 mM NaCl, comparing their fitness against the five parental strains (Supplementary Table S8). We observed a significantly higher average growth rate (p-value < 0.05, ANOVA) in diploid colonies compared to parental strains in the three conditions evaluated, suggesting a greater fitness in these diploid individuals. Other diploid Saccharomyces intraspecific hybrids have shown superior phenotypes than their diploid parental strains, representing a successful strategy to survive under challenging conditions by creating phenotypic diversity [72][73][74][75][76] . Finally, we evaluated the heterosis coefficient in the five intraspecific hybrids relative to their respective parental strains in the same three conditions. In this case, we observed mild positive mid-parent and best-parent heterosis (heterosis coefficient > 1) (Fig. 5c). These results agree with previous studies, where intraspecific hybrids showed low levels of mid-parent heterosis for relative competitive growth, with no significant relationship between genetic divergence and heterosis 77 .
Considering the low intraspecific hybridization rate in S. eubayanus, we evaluated whether this situation depended on environmental conditions or the Saccharomyces species. For this, we carried out a second experimental approach for the generation of intraspecific hybrids by mass-mating under beer wort, as a representative alternative environmental condition. Furthermore, to determine whether the low hybridization rate was specific to S. eubayanus or reflected a more general trend in the Saccharomyces genus, we evaluated the hybridization rate in S. cerevisiae compared to S. eubayanus (Fig. 4b). We selected four strains representative of the PB-2, PB-3 and ADM lineages (CL715.1, CL216.1, CL601.1 and CL1106.1) in S. eubayanus, and four strains representative of the Wine/European, Sake, West African and North America lineages (DBVPG6765, Y12, DBVPG6044 and YPS128) in S. cerevisiae. The "MATα" and "MATa" versions of each strain were mixed and grown in malt extract for seven days and for 34 generations (four transfers). Subsequently, we determined the proportion of haploid and diploid cells after each transfer by colony PCR (Fig. 4b). As previously observed, in S. eubayanus, we identified www.nature.com/scientificreports/ a drastic decrease in MATα colonies, a predominance of MATa colonies, and the lack of hybrids after the fourth transfer in the three replicates (Fig. 6a).
On the other hand, S. cerevisiae quickly returns to a diploid state, which represent more than 50% of the colonies analyzed after the second transfer (Fig. 6b). These results suggest that the intraspecific hybridization rate depends on environmental conditions and the Saccharomyces species. The predominance of haploid colonies of S. eubayanus may be due to several factors. The presence of colonies of both mating-types after the first a b  www.nature.com/scientificreports/ transfer, without the detection of diploid colonies, suggests that indeed the different strains of S. eubayanus are not mating, unlike in S. cerevisiae where after the first transfer, at least 30% of the colonies were diploid (Fig. 6b).
Previous studies have demonstrated that cells of different mating types grown in a liquid medium can optimize their decision to mate or to proliferate by detecting the ratio of opposite-sex partners to same-sex competitors 78 . Furthermore, specific sexual aggregation via α/a-agglutinins is required for mating in a suspension of cells in liquid, which is determined by the probability of random mating encounters and the interaction strength of sexual α/a-agglutinins 78,79 . Indeed, the production of pheromones and the sexual aggregation capacity of S. eubayanus have not been studied; both factors can vary between strains, may be different from those reported for other Saccharomyces species, and may impact their mating capacity 78

Parental strains
Diploid colonies

Conclusions
We obtained an extensive collection of haploid strains of S. eubayanus from different genetic backgrounds complementing previous resources generated for the species and the Saccharomyces genus. These strains were comprehensively phenotyped and represent a valuable and promising resource for studies in different aspects of biology, particularly for molecular genetic studies. Genomic analysis demonstrated the extremely low frequency of off-targets and no structural variants derived from the meiotic event. Finally, mass-mating experiments demonstrate the pertinence of the haploid collection for genetic studies and the preference of S. eubayanus to remain in the haploid state compared to S. cerevisiae, which quickly returns to a diploid state.

Methods
Strains. The S. eubayanus strains used in this study were selected from a collection of strains that have been published previously 36,37 . The strains comprised 83 Chilean strains collected from ten different localities 36 , one Argentinian strain 41 and five North American strains 45 (Supplementary Table S1). All strains were maintained on YPD solid medium (1% yeast extract, 2% peptone, 2% glucose, 2% agar). Yeast haploid strains are available upon request to the corresponding author.

Generation of null mutants by CRISPR-Cas9.
HO null mutants were generated using a CRISPR-Cas9 method 22 to generate stable haploid strains. The guide RNAs (gRNAs) were designed using the HO coding sequences from the reference strain CBS12357 55 and the Benchling online tool (https:// www. bench ling. com/). We selected the four gRNAs with the highest "on-target" score, an optimized score for 20 bp gRNAs with NGG PAM, developed by Doench et al. 64 . The gRNAs were synthesized as single-stranded oligonucleotides (Macrogen, Korea), adding the SapI overhangs, 5'-ATC and 5'-AAC, and were phosphorylated (10 µM final concentration) using 10 units of T4 polynucleotide kinase (New England BioLabs, NEB) in a final volume of 10 µL at 37 °C for 1 h. Double-stranded oligonucleotides were generated by annealing equimolar amounts of phosphorylated single-stranded oligonucleotides under the following conditions: denaturation at 96 °C for 6 min and then cooling to 23 °C with a ramp of 0.1 °C/s. The four gRNAs were separately cloned in the plasmid pAEF5 54 (a gift from Gilles Fischer, Addgene plasmid #136,305), using standard "Golden Gate Assembly" 85  Finally, each diploid strain was co-transformed with the plasmid carrying the gRNA together with the Cas9 gene, and the donor DNA using the standard lithium acetate protocol 86 with temperature modifications. Briefly, the strains were grown overnight in 5 mL YPD under constant agitation at 20 °C. Then, 300 µL overnight culture were inoculated in 5 mL YPD and incubated under the same conditions for 4 h or until the culture reached 10 8 cells/mL. From this, 1 mL was washed three times with distilled water, then washed twice in 1 mL of lithium acetate 0.1 M (LiAc) and cells were recovered by centrifugation. The pellet of cells was mixed with 240 µL polyethylene glycol 3350 50% w/v (PEG 3350), 36 µL 1 M LiAc, 20 µL denatured salmon sperm DNA (10 mg/mL), 500 ng of a plasmid carrying the gRNA and the Cas9, and 24 µL of donor DNA for the repair of CRISPR-induced DSBs. The cells were resuspended by vortex mixing and incubated for 45 min at 20 °C. Then, a heat shock was performed at 37 °C for 20 min. After transformation, cells were washed with 1 mL distilled water and plated on YPD with 0.2 mg/mL hygromycin for selection. Plates were incubated between 3 to 5 days at 20 °C. Correct gene deletion was confirmed by colony PCR using GoTaq DNA polymerase (Promega). All primer, gRNAs and donor DNA sequences are listed in Supplementary Table S9.
Generation of stable haploid strains. Isogenic haploid strains were constructed by deleting the HO gene using a CRISPR-Cas9 method as described above. Initially, the four gRNAs were evaluated separately in the CL216.1 strain, determining the efficiency of each gRNA as the number of colonies with the proper deletion divided by the total number of colonies evaluated by colony PCR. The most efficient gRNA was used to transform the other strains in the collection. Diploid Δho cells were sporulated on 2% potassium acetate agar plates (2% agar) for at least seven days at 20 °C. Meiotic segregants were obtained by dissecting tetrad ascospores treated with 10 µL Zymolyase 100 T (50 mg/mL) on YPD agar plates with a SporePlay micromanipulator (Singer Instruments, UK). Spores from four viable spore tetrads were selected to determine the mating type by colony PCR of the MAT locus 8 using GoTaq DNA Polymerase (Promega). All the primers used in these experimental procedures are listed in Supplementary Table S9.
Spore viability. Strains were sporulated on 2% potassium acetate agar plates as described above. Tetrad ascospores were treated with 10 µL Zymolyase 100 T (50 mg/mL) and dissected with a SporePlay micromanipulator. At least 20 ascospores per strain were dissected on YPD plates and incubated at 20 °C for four days. Spore viability was calculated using the formula (n° of viable spores/n° of dissected spores) × 100.
Genotypic characterization. Genomic DNA from a single spore of CL216.1-a, CL715.1-a and CL601.1-a was prepared for whole-genome sequencing using the Qiagen Genomic-tip 20/G kit (Qiagen, Germany) as previously described 36 and sent for DNBseq sequencing (BGISEQ-G400 platform). Reads were filtered and trimmed using Fastp 0.20.1 (-3-l 50-cut_mean_quality 30) 87 . Cleaned reads were mapped to the S. eubayanus reference strain CBS12357 using BWA mem 0.7.17 88 , after which PCR duplicates were marked using Picard tools. Mapping files corresponding to the original diploid strains were obtained from 36 . Variants were jointly called on haploid and diploid alignments using freeBayes 1.3.1 89 . Variants were analyzed in R and visually inspected using the IGV genome browser 90 .
To perform nanopore sequencing of a full tetrad of the CL216.1 strain, DNA was extracted from each spore using the Quick-DNA HMW MagBead Kit (Zymo Research). High molecular weight genomic DNA was purified using magnetic beads following the kit manufacturer's protocol. DNA integrity was verified using an agarose gel and DNA quantity was assessed using Quantus (Promega, Madison, Wisconsin, US.). Libraries were prepared following the manufacturer's protocol, barcodes were added to each library, and these were sequenced on a single R9.4 flow cell using a Minion (Oxford Nanopore Technologies, UK). The raw fast5 files were transformed to fastq files and de-barcoded using Guppy v5.0.14 with the "super high accuracy" model. Long reads were mapped to the S. eubayanus reference CBS12357 with minimap2 (ax map-ont -secondary = no) 91 and variants were called using PEPPER (-ont_r9_guppy5_sup) 92 . Long-read assemblies of each spore were generated using Flye v2.9 (-nano-hq -scaffold) 93 . To evaluate synteny, each assembly was compared using nucmer 94  www.nature.com/scientificreports/ Fermentations were carried out in three biological replicates using previously oxygenated (15 mg/L) 12°Plato (°P) beer wort, supplemented with 0.3 ppm ZnCl 2 . The pre-cultures were grown in 5 mL 6°P wort for 24 h at 20 °C in constant agitation at 150 rpm. Inocula were then transferred to 50 mL 12°P wort and incubated for 24 h at 20 °C in constant agitation at 150 rpm. The cells were collected by centrifugation and used to calculate the final cell concentration for use in each fermentation according to the formula described by White and Zainasheff 97 . Cells were inoculated into 50 mL 12°P wort in 250 mL bottles and airlocks with 30% glycerol. The fermentations were incubated at 12 °C, with no agitation for 15 days and monitored by weighing the bottles daily to determine weight loss over time.
Metabolites quantification by HPLC. Glucose, fructose, maltose, maltotriose and ethanol concentrations were determined by High-Performance Liquid Chromatography (HPLC) after 15 days of fermentation as previously described 35,48 . Samples were obtained extracting 0.5 mL fermented beer wort and filtered using 0.22 μm filters. 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 98 . Glucose, fructose, maltose and maltotriose uptake were estimated as the difference between the initial and final concentration before and after fermentation, respectively.
Generation of intra-specific hybrids by mass-mating. Five different genetic backgrounds were selected to generate intra-specific hybrids by mass-mating (CL216.1, CL601.1, CL715.1, CL1106.1 and CL449.1) (Fig. 4a). First, one colony from each S. eubayanus strain and mating type was cultured in 0.67% YNB medium (Difco, France) with 2% glucose at 25 °C in constant agitation at 150 rpm. Each pre-inoculum was then utilized to prepare a co-culture at an initial OD 600nm of 0.1 of each strain in 150 mL 0.67% YNB medium with 2% glucose. This co-culture was divided in three 50 mL replicates and incubated at 25 °C in constant agitation at 150 rpm during 72 h. Subsequently, the cultures were used to inoculate fresh 50 mL cultures at an initial OD 600nm of 0.1, and this procedure was sequentially repeated four times. After each incubation, colonies were isolated for each replicate on YPD solid medium and the mating type determined by colony PCR as described above (at least 10 colonies per replicate). The number of generations was determined using the formula log 2 (OD 600 final /0.1) 99 .
Diploid colonies identified after the last incubation were characterized under microculture conditions as described above, together with the parental diploid and haploid strains. The conditions evaluated were: 2% glucose at 25 °C, 2% glucose with 1.25 mM NaCl, and 2% glucose with 5% ethanol. Mid-parent and best-parent heterosis were determined as previously described 77,100 , using Eqs. (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 101 . where: µmax h : growth rate of a hybrid.
Intra-specific hybrids by mass-mating were also generated under fermentation conditions (Fig. 4b). Four different genetic background were selected: CL216.1, CL601.1, CL715.1 and CL1106.1. The pre-culture conditions were the same as described above; each haploid strain was grown in 5 mL 6°P wort for 24 h at 20 °C in constant agitation at 150 rpm, then transferred to 50 mL 12°P wort and incubated for 24 h at 20 °C in constant agitation at 150 rpm. Each pre-inoculum was used to prepare a co-culture at a final concentration of 1 × 10 6 cells/mL and transferred to three replicates to obtain a final concentration of 1.5 × 10 6 cells/mL in 50 mL 12°P wort in 250 mL bottles and airlocks with 30% glycerol. The fermentations were incubated at 20 °C, with no agitation for 7 days and monitored by weighing the bottles daily to determine weight loss over time. At the end of fermentation, the cultures were used to inoculate fresh 50 mL 12°P wort at an inoculum density of 1.5 × 10 6 cells/mL, repeating the fermentation and re-inoculation four times. Also, we used four different S. cerevisiae strains that represent the main lineages described in the species: YPS128 (North American, NA), DBVPG6044 (West African, WA), Y12 (Sake, SA) and DBVPG6765 (Wine/European, WE) (Supplementary Table S10) 8 . The number of generations was determined using the formula log 2 (final cells/1.5 × 10 6 ) 102 . Finally, after each fermentation, the proportions of diploid and haploid cells were determined by colony PCR of colonies isolated for each replicate on solid YPD medium, as described above.
Hybrid genotyping. Diploid colonies identified after the last incubation in the mass-mating experiment were genotyped to determined their intraspecific hybrid status by sequencing the DCR1 gene 45 . Genomic DNA of eight diploid colonies was extracted using the Wizard® Genomic DNA Purification Kit (Promega) according to the manufacturer's instructions. PCR reactions were performed in a final volume of 25  Data and statistical analysis. Data visualization and statistical analyses were performed with R software version 4.0.3. The maximum specific growth rates and total CO 2 loss were compared using an analysis of variance (ANOVA) and the mean values of the three replicates were statistically analyzed 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. A principal component analysis (PCA) was performed using the Facto-MineR package version 2.4 to compute principal component methods and the factoextra package version 1.07 for extracting, visualizing and interpreting the results. Heatmaps was generated using the ComplexHeatmap package version 2.6.2.