Yeast strains and hybrid generation
The yeast strains used in the study are listed in Table 1. The interspecific yeast hybrids H1 (P1 × P2) and H2 (P2 × P3) and intraspecific yeast hybrid H3 (P1 × P2) were generated using rare mating according to the method described in Krogerus et al. . Prior to mating, a uracil auxotroph (ura-) of P1 was selected on 5-fluoroorotic acid (5-FOA) agar , a respiratory-deficient mutant (rho-) of P2 was selected on YPDG agar containing 3% glycerol and 0.1% glucose, and a lysine auxotroph (lys-) of P3 was selected on α-aminoadipic (α-AA) agar  to allow for the selection of hybrids on minimal selection agar medium (0.67% Yeast Nitrogen Base without amino acids, 3% glycerol, 3% ethanol and 2% agar). The interspecific hybrid H1 was further mated with P2 (rho-) to form the interspecific triple hybrid T1 [(P1 × P3) × P2]. Prior to mating, a lysine auxotroph (lys-) of H1 was first selected as described above, after which ascospores of it were generated on 1% potassium acetate agar. Spore viability was calculated based on the amount of colonies formed from the dissection of up to 20 tetrads. Ascus walls were digested with Zymolyase 100T, after which spores of H1 (lys-) were mixed with P2 (rho-). Hybrid T1 was selected on minimal selection agar. Hybrid T2, a meiotic segregant of T1, was created by generating ascospores of Hybrid T1 on 1% potassium acetate agar and then dissecting the spores on YPD agar. All spore dissections were carried out using the Singer MSM 400 dissecting microscope (Singer Instruments, UK). The viable spore clones were then screened for the POF phenotype in a small-scale assay. 1 ml of YPM supplemented with 100 mg L−1 of trans-ferulic acid was inoculated with a colony of the spore clones, and they were allowed to incubate for 48 h at 25 °C. The POF phenotype was determined sensorially by examining for the presence (POF+) or absence (POF−) of the distinct clove-like aroma of 4-vinyl guaiacol. Hybrid T2 was a spore clone of T1 which did not produce 4-vinyl guaiacol in this assay. An overview of these brewing strains and their creation is presented in Fig. 1.
The hybrid status of isolates was confirmed by PCR as described in Krogerus et al. . Briefly, the rDNA ITS region was amplified using the primers ITS1 (5′-TCCGTAGGTGAACCTGCGG-3′) and ITS4 (5′-TCCTCCGCTTATTGATATGC-3′) and amplicons were digested using the HaeIII restriction enzyme (New England BioLabs, USA) as described previously . Amplification of the S. eubayanus-specific FSY1 gene (amplicon size 228 bp) and the S. cerevisiae-specific MEX67 gene (amplicon size 150 bp) was also performed using the primers SeubF3 (5′-GTCCCTGTACCAATTTAATATTGCGC-3′), SeubR2 (5′-TTTCACATCTCTTAGTCTTTTCC-AGACG-3′), ScerF2 (5′-GCGCTTTACATTCAGATCCCGAG-3′), and ScerR2 (5′-TAAGTTGGTTGTCAGCAAGATTG-3′) as described by Muir et al.  and Pengelly and Wheals . Additionally, the hybrid statuses of the intraspecific hybrid H3 and the triple hybrids T1 and T2 were confirmed by amplifying interdelta sequences using the delta12 (5′-TCAACAATGGAATCCCAAC-3′) and delta21 (5′-CATCTTAACACCGTATATGA-3′) primers as described by Legras and Karst .
Flow cytometry was also performed on the brewing yeast strains to estimate ploidy essentially as described by Haase and Reed . Cells were grown overnight in YPD medium (1% yeast extract, 2% peptone, 2% glucose), and approximately 1 × 107 cells were washed with 1 mL of 50 mM citrate buffer. Cells were then fixed with cold 70% ethanol, and incubated at room temperature for 1 h. Cells were then washed with 50 mM citrate buffer (pH 7.2), resuspended in 50 mM citrate buffer containing 0.25 mg mL−1 RNAse A and incubated overnight at 37 °C. 1 mg mL−1 of Proteinase K was then added, and cells were incubated for 1 h at 50 °C. Cells were then stained with SYTOX Green (2 μM; Life Technologies, USA), and their DNA content was determined using a FACSAria IIu cytometer (Becton–Dickinson, USA). DNA contents were estimated by comparing fluorescence intensities with those of S. cerevisiae haploid (CEN.PK113-1A) and diploid (CEN.PK) reference strains. Measurements were performed on duplicate independent yeast cultures, and 100,000 events were collected per sample during flow cytometry.
Whole genome sequences of the brewing strains P1 and P3 have been published previously [3, 50]. The other 6 brewing strains were sequenced by Biomedicum Genomics (Helsinki, Finland). In brief, an Illumina NexteraXT pair-end 150 bp library was prepared for each hybrid and sequencing was carried out with a NextSeq500 instrument. Pair-end reads from the NextSeq500 sequencing were quality-analysed with FastQC  and trimmed and filtered with Skewer . Alignment, re-alignment and variant analysis was carried out using SpeedSeq  and its FreeBayes SNP prediction and CNVnator copy number variation prediction modules [54, 55]. Reads of S. cerevisiae P2 were aligned to that of S. cerevisiae S288c , while reads of hybrid strains were aligned to concatenated reference sequences of strains P1 and P3 [3, 50] as described previously . SNPs predicted by FreeBayes with less than five left and right aligning reads were discarded. Prior to SpeedSeq variant analysis, alignments were filtered to a minimum MAPQ of 50 with SAMtools . Quality of alignments was assessed with QualiMap . The median coverage over 10,000 bp windows was calculated with BEDTools  and visualized with R (http://www.r-project.org/). The coverage of S. cerevisiae P2 and the 5 hybrid strains (H1-H3 and T1-T2) across the S. cerevisiae and S. eubayanus reference genomes are displayed in Additional file 1: Figure S3.
The fermentation kinetics and lipid compositions of the 8 brewing strains at two different temperatures (10 and 20 °C) and two different initial ethanol concentrations [0 and 8% (v/v)] were assayed in 100 mL shake-flask fermentations. The strains were grown in media containing 1% yeast extract, 2% peptone, 8% maltose, and up to 8% ethanol. Prior to the assay, the strains were pre-cultivated in media containing 1% yeast extract, 2% peptone and 4% maltose for 24 h at 20 °C. The OD600 of the pre-cultivations was measured, and the growth assays were started by inoculating 100 mL of media to a starting OD600 of 0.01 in triplicate flasks. Flasks were capped and then incubated at either 10 or 20 °C with light agitation (80 RPM) for up to 33 days. Fermentation was monitored (up to twice daily) by drawing 100 μL samples and measuring the refractive index (°brix) with a Quick-Brix 90 digital refractometer (Mettler-Toledo AG, Switzerland). Samples were drawn for lipid analysis at the end of the exponential fermentation phase. 15 mL samples of fermentation media were centrifuged for 5 min at 9000×g, after which the yeast pellet was washed twice in 15 mL of ice-cold deionized water. The washed yeast pellet was then transferred to a cryotube, and flash-frozen in liquid nitrogen. The samples were stored at −80 °C prior to lipid analysis. After fermentations were complete, the biomass concentration determined by drawing and centrifuging 30 mL samples of the fermentation media (10 min at 9000×g), washing the yeast pellets gained from centrifugation twice with 25 mL deionized H2O and then suspending the washed yeast in a total of 6 mL deionized H2O. The suspension was then transferred to a pre-weighed porcelain crucible, and was dried overnight at 105 °C and allowed to cool in a desiccator before the change of mass was measured. Fermentation curves for the fermentations were modelled based on the decrease in °brix over time using the ‘grofit’-package for R . The fermentation parameters were determined using the logistic model in ‘grofit’.
Prior to lipid extraction, the frozen cell samples were freeze-dried at −55 °C overnight (Martin Christ Alpha 1-4 LDplus, Germany). For fatty acid (free and bound) and sterol analysis, 10 mg of freeze-dried sample was rehydrated into 200 µL of 0.9% sodium chloride solution and spiked with heptadecanoic acid (FFA 17:0; 5.36 µg) and glyceryl triheptadecanoate [TG(17:0/17:0/17:0); 21.11 µg]. Lipids were extracted with chloroform:methanol (2:1 v/v; 800 µL) by homogenizing the samples with grinding balls in a Retsch mixer mill MM400 homogenizer (Retsch GmbH, Haan, Germany) at 20 Hz for 2 min. After 30 min standing at room temperature, the samples were centrifuged at 10,620×g for 5 min and the lower layer was separated into glass tubes and evaporated to dryness under nitrogen flow.
The evaporation residues from lipid extractions were dissolved into petroleum ether (b.p. 40–60 °C; 700 µL). Fatty acids were transesterified with sodium methoxide (NaOMe; 0.5 M; 250 µL) in dry methanol by boiling at 45 °C for 5 min. The mixture was acidified with 15% sodium hydrogen sulphate (NaHSO4; 500 µL). The petroleum ether phase containing the fatty acid methyl esters (FAME) as well as free fatty acids (FFA) was collected into an Eppendorf tube and centrifuged (10,620×g; 5 min). Half of the petroleum ether layer was separated into a GC vial and evaporated, after which the residue was dissolved into hexane (100 µL) and taken into a vial insert. FAMEs were analysed on an Agilent 7890A GC combined with an Agilent 5975C mass selective detector controlled by MSD ChemStation software (Agilent Technologies Inc., Santa Clara, CA, USA). The column was an Agilent FFAP silica capillary column (25 m × 0.2 mm × 0.3 µm). Helium was used as carrier gas and the samples were injected in splitless mode. The oven temperature programme was from 70 °C (2 min) to 240 °C at a rate of 15 °C min−1, total run time was 39 min. The temperatures of the injector and MS source were 260 and 230 °C, respectively. The samples (1 µL) were injected by a Gerstel MPS injection system (Gerstel GmbH & Co. KG, Mülheim an der Ruhr, Germany) and the data were collected in EI mode (70 eV) at a mass range of m/z 40–600.
The other half of the petroleum ether layer was transferred to a GC vial and evaporated into dryness for the determination of free fatty acid and sterols. The residue was dissolved into 50 µL of DCM and derivatized with N-Methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA; 40 µL) and trimethylchlorosilane (TMCS; 10 µL) by incubating at 80 °C for 20 min. The samples (1 µL aliquots) were injected in splitless mode at 300 °C, and analysed on an Agilent DB-5MS column (30 m × 0.2 mm × 0.25 µm). The oven temperature programme was from 50 °C (1.5 min) to 325 °C at a rate of 10 °C min−1, total run time was 49 min.
For lipidomics analyses, approx. 5 mg of freeze-dried cell samples were rehydrated in 50 µL of 0.9% sodium chloride in Eppendorf tubes, mixed with 400 µL chloroform:methanol (2:1) and spiked with 10 µL of an internal standard mixture 1 [IS1; containing LysoPC(17:0), MG(17:0), DG(17:0/17:0), TG(17:0/17:0/17:0), PG(17:0/17:0), Cer(d18:1/17:0), PC(17:0/17:0), PE(17:0/17:0), CE(19:0), CA(14:0), C12(β)-d-GluCer and C8-l-threo-LacCer] (Avanti Polar Lipids, Alabaster, AL, USA; Larodan Fine Chemicals AB, Malmö, Sweden; Nu-Chek Prep, Inc., Elysian, MN, USA) at concentration levels of 0.4–3.2 µg/sample. The samples were homogenized with grinding balls in a Retsch mixer mill MM400 homogenizer at 25 Hz for 2 min and after 30 min standing were centrifuged at 10,620×g for 3 min. Before UPLC-MS analysis, a 20 µL aliquot of a labelled lipid standard mixture [IS2; containing LysoPC(16:0-D3), PC(16:0/16:0-D6) and TG(16:0/16:0/16:0-13C3)] (Avanti Polar Lipids, Alabaster, AL, USA) was added into the separated lipid extracts.
Lipid extracts were analyzed on a Waters Q-TOF Premier mass spectrometer combined with an Acquity Ultra Performance LC™ (UPLC) under the control of MassLynx software (v 4.1; Waters Inc., Milford, MA, USA). The column (at 50 °C) was an Acquity UPLC™ BEH C18 (2.1 × 100 mm with 1.7 μm particles). The solvent system included (A) ultrapure water (1% 1 M NH4Ac, 0.1% HCOOH) and (B) LC/MS grade acetonitrile/isopropanol (1:1, 1% 1 M NH4Ac, 0.1% HCOOH). In ESI− mode, the same solvent system but without acid was used. The gradient started from 65% A/35% B, reached 80% B in 2 min, 100% B in 7 min and remained there for the next 7 min. There was a 5 min re-equilibration step before the next run. The flow rate was 0.400 mL min−1 and the injected amount 1.0 μL (Acquity Sample Organizer at 10 °C). The ESI source was at 120 °C and the capillary voltage 3.0 and 2.5 kV in positive and negative mode, respectively. N2 was used as desolvation gas (800 L h−1) at 270 °C.
The data were collected at a mass range of m/z 300–1200 with a scan duration of 0.2 s. Reserpine was used as the lock spray reference compound in ESI+ mode and leucine enkephaline in ESI− mode. Data processing was carried out with MZmine software (version 2.20)  enabling peak integration and alignment of the peaks. An internal spectral MS/MS library was used for identification of the compounds.
Quantification of lipid subspecies was based on peak heights of internal standards. All monoacyl lipids except cholesterol esters, such as monoacylglycerols and monoacylglycerophospholipids, were normalized with LysoPC(17:0), all diacyl lipids except ethanolamine phospholipids were normalized with PC(17:0/17:0), all ceramides with Cer(d18:1/17:0), all diacyl ethanolamine phospholipids with PE(17:0/17:0), and TGs and sterylesters were normalized with TG(17:0/17:0/17:0) and CE(19:0), respectively. Other (unidentified) molecular species were normalized with LysoPC(17:0) for retention time <300 s, PC(17:0/17:0) for retention time between 300 and 410 s, and TG(17:0/17:0/17:0) for higher retention times.
To assess the roles of oleic acid (C18:1) on cold and ethanol tolerance in yeast, microcultures were carried out in media containing various concentrations of supplemented oleic acid (0 or 0.8 mM) and ethanol (1, 5 or 10% v/v). This concentration of oleic acid was chosen based on values found previously in literature . The microcultures were carried out in 100-well honeycomb microtiter plates at 15 and 20 °C (with continuous shaking), and their growth dynamics were monitored with a Bioscreen C MBR incubator and plate reader (Oy Growth Curves Ab, Finland). The wells of the microtiter plates were filled with 300 µL of YPDt medium (1% yeast extract, 2% peptone, 2% glucose, and 1% Tergitol NP-40) supplemented with oleic acid (0 or 0.8 mM) and ethanol (1, 5 or 10% v/v). Oleic acid was added to the media from a 100 × stock solution (80 mM oleic acid) prepared in 50% ethanol and 35% Tergitol NP-40. Precultures of the laboratory strains WT, ole1∆, and erg4∆ (Table 1) were started in 10 mL YPD medium (1% yeast extract, 2% peptone, and 2% glucose) and incubated at 25 °C with shaking at 120 rpm. The cultures were centrifuged and the yeast pellets were washed once with sterile deionized water. The yeast was then resuspended in YPDt medium to an OD600 value of 10. The microcultures were started by inoculating the microtiter plates with 3 µL of cell suspension per well (for an initial OD600 value of 0.1) and placing the plates in the Bioscreen C MBR. The optical density of the microcultures at 600 nm was automatically read every 30 min. Four replicates were performed for each strain in each medium. Growth curves for the microcultures were modelled based on the OD600 values over time using the ‘grofit’-package for R .
Fermentation and analysis
The set of eight brewing strains were characterized in fermentations performed in a 15 °Plato high gravity wort at 15 °C. Yeast was propagated essentially as described previously , with the use of a ‘Generation 0’ fermentation prior to the actual experimental fermentations. The experimental fermentations were carried out in duplicate, in 2-L cylindroconical stainless steel fermenting vessels, containing 1.5 L of wort medium. The 15 °Plato wort (69 g maltose, 17.4 g maltotriose, 15.1 g glucose, and 5.0 g fructose per litre) was produced at the VTT Pilot Brewery from barley malt. Yeast was inoculated at a rate of 15 × 106 viable cells mL−1. The wort was oxygenated to 15 mg L−1 prior to pitching (Oxygen Indicator Model 26073 and Sensor 21158, Orbisphere Laboratories, Switzerland). The fermentations were carried out at 15 °C until an apparent attenuation of 80% (corresponding to approx 7% ABV) was reached, until no change in residual extract was observed for 24 h, or for a maximum of 14 days.
Wort samples were drawn regularly from the fermentation vessels aseptically, and placed directly on ice, after which the yeast was separated from the fermenting wort by centrifugation (9000×g, 10 min, 1 °C). Samples for yeast-derived flavour compounds, fermentable sugars and 4-vinyl guaiacol analysis were drawn from the beer when fermentations were ended. Yeast viability was measured from the yeast that was collected at the end of the fermentations using a Nucleocounter® YC-100™ (ChemoMetec, Denmark).
The alcohol level (% v/v) and pH of samples was 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, Austria). The yeast dry mass content of the samples (i.e. yeast in suspension) was determined by washing the yeast pellets gained from centrifugation twice with 25 mL deionized H2O and then suspending the washed yeast in a total of 6 mL deionized H2O. The suspension was then transferred to a pre-weighed porcelain crucible, and was dried overnight at 105 °C and allowed to cool in a desiccator before the change of mass was measured. The measured pH values and suspended dry mass are presented in Additional file 1: Figure S4.
Concentrations of fermentable sugars (glucose, fructose, maltose and maltotriose) were measured by HPLC using a Waters 2695 Separation Module and Waters System Interphase Module liquid chromatograph coupled with a Waters 2414 differential refractometer (Waters Co., Milford, MA, USA). An Aminex HPX-87H Organic Acid Analysis Column (300 × 7.8 mm, Bio-Rad, USA) was equilibrated with 5 mM H2SO4 (Titrisol, Merck, Germany) in water at 55 °C and samples were eluted with 5 mM H2SO4 in water at a 0.3 mL min−1 flow rate.
Yeast-derived flavour compounds were determined by headspace gas chromatography with flame ionization detector (HS-GC-FID) analysis. 4 mL samples 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−1) 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, USA). The carrier gas was helium (constant flow of 1.4 mL min−1). The temperature program was 50 °C for 3 min, 10 °C min−1 to 100 °C, 5 °C min−1 to 140 °C, 15 °C min−1 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.
4-Vinyl guaiacol was analyzed using HPLC-PAD based on methods described by Coghe et al.  and McMurrough et al. . The chromatography was carried out using a Waters Alliance HPLC system consisting of a Waters e2695 Separations Module equipped with a XTerra® MS C18 column (5 µm, 4.6 × 150 mm) and a Waters 2996 Photodiode Array Detector. The mobile phase consisted of H2O/CH3OH/H3PO4 (64:35:1, v/v) and flow rate was 0.5 mL min−1. The diode array detector was used at 190–400 nm. 4-Vinyl guaiacol was quantified at 260 nm using standard curves of the pure compound (0.3–30 mg L−1).
Stability of Hybrid T2
The karyotype stability of Hybrid T2 was evaluated after 10 successive batch cultures (corresponding to approximately 65 cell generations) in two different media. Media 1 consisted of 1% yeast extract, 2% peptone, 1% maltose and 1% maltotriose, while Media 2 consisted of 1% yeast extract, 2% peptone, 2% maltose and 14% sorbitol. Media 2 was used to mimic the osmotic stress occurring during high-gravity wort fermentations. Hybrid T2 was first grown overnight in YPM at 25 °C. This culture was used to inoculate 1 mL of Media 1 or Media 2 in duplicate to a starting OD600 of 0.1. The cultures were allowed to grow for 7 days at 18 °C, after which they were used to inoculate a fresh 1 mL aliquot of Media 1 or Media 2 to a starting OD600 of 0.1. This was repeated for a total of 10 successive cultures (10 weeks). After this, 20 μL aliquots of the cultures were spread out on YPM agar, and a total of 8 colonies were randomly selected and isolated for further analysis (2 isolates per duplicate per media).
The karyotype stability of the 10-week isolates was assessed by determining their DNA content by flow cytometry as described above, and their karyotypes by pulsed-field gel electrophoresis (PFGE). PFGE was carried out essentially as described previously . Sample plugs were prepared with the CHEF Genomic DNA Plug Kit for Yeast (Bio-Rad) according to the manufacturer’s instructions with minor modifications. Instead of lyticase treatment, the plugs were treated with 0.1 mg mL−1 Zymolyase 100T in buffer containing 1 mM dithiothreitol. The sample plugs were loaded into the wells of a 1.0% pulse field certified agarose (Bio-Rad) gel. PFGE was performed at 14 °C in 0.5 × TBE buffer [89 mM Tris, 89 mM boric acid, 2 mM EDTA (pH 8)]. A CHEF Mapper XA pulsed field electrophoresis system (Bio-Rad) was used with the following settings: 6 V cm−1 in a 120° angle, pulse length increasing linearly from 26 to 228 s, and total running time of 40 h. A commercial chromosome marker preparation from S. cerevisiae strain YNN295 (Bio-Rad) was used for molecular mass calibration. After electrophoresis, the gels were stained with ethidium bromide and scanned with Gel Doc XR+ imaging system (Bio-Rad).
Data and statistical analyses were performed with R (http://www.r-project.org/). The distributions of the lipidomic data were estimated by histograms and the Shapiro–Wilk test, and the lipidomic data was consequently log10-transformed to correct for skewed distributions. The change in lipid composition compared to the control cultivation at 20 °C and 0% supplemented EtOH was tested by Student’s t test (two-tailed, unpaired, and unequal variances). To control for multiple testing, the p values were further adjusted for Benjamini–Hochberg false discovery rate (FDR). Strain-specific differences in fatty acid, squalene and ergosterol concentrations were tested with one-way ANOVA with Tukey’s post hoc test. Multivariate analysis was performed with Partial Least Squares (PLS) and PLS-Discriminant Analysis (PLS-DA) using the ‘ropls’ package in R . PLS-DA was initially performed on the lipid data of all samples in order to determine whether the lipid compositions of the yeast in low temperatures or high alcohol levels could be distinguished from those at control conditions (20 °C, 0% supplemented EtOH). PLS models were constructed from the fermentation and lipid data obtained at the different temperatures and supplemented ethanol levels in order to elucidate which lipid species contributed positively and negatively to fermentation performance at those conditions. The Y response variable of the models was the maximum fermentation level divided by the lag time (A and λ from Table 2, respectively), while the X predictor variables were the combined dataset of the compositions of fatty acids, squalene and ergosterol obtained from GC/MS analysis and the compositions of the 60 lipid species obtained from UPLC/MS lipidomics analysis. PLS(-DA) models were cross-validated (Q
2 > 0.5 was considered significant ), and the significance of the Q
2 value was tested with 200 random permutations of the X dataset.