Transcriptomic analysis of Saccharomyces cerevisiae x Saccharomyces kudriavzevii hybrids during low temperature winemaking

Yeast strains and media

The yeasts strains included in this study were the strain Lalvin T73 (S. cerevisiae) and the strain IFO1802 (S. kudriavzevii), used as the parental species, and two S. cerevisiae x S. kudriavzevii hybrids, VIN7 and Lalvin W27, isolated from wine in South Africa and Switzerland, respectively. The yeast was grown and maintained in GPY medium (2% glucose, 0.5% peptone, 0.5% yeast extract) and plates (with 2% agar). Wine fermentations were performed in grape juice from the Tempranillo variety at 28°C or 12°C in vessels with 0.45 L of Tempranillo grape must.

Total RNA extraction and cDNA labelling

Each yeast strain had two independent fermentations and three samples were taken from each independent fermentation. Cells from each yeast strain fermentation were pelleted by centrifugation (4000 rpm/min, 5 min), at 12°C and 28°C. Cells were collected at the beginning of the exponential phase by taking samples two generations after inoculation. The RNA extraction protocol was based on subsequent treatments with phenol-tris, phenol-chloroform (5:1) and chloroform-isoamyl alcohol (24:1), and an ethanol precipitation with sodium acetate¹³. RNA concentrations and purity were determined using a Nanodrop spectrophotometer ND-1000 (Nanodrop Technologies™, Wilmington, DE). RNA integrity was checked by electrophoresis in agarose gel (1%). 2–4 μg of total RNA from each sample was linearly amplified using the Low RNA Input Fluorescent Linear Amplification kit (Agilent Technologies™, Ca, USA). 2–3 µg of amplified cRNA was used as template for cDNA synthesis. cDNA was marked indirectly using the SuperScript™ Indirect cDNA Labeling System (Invitrogen™, San Diego, CA). Cy3 and Cy5 mono-reactive Dye (Amersham GE Healthcare™, Amersham UK) were used as the fluorophores and dye incorporation was monitored using a Nanodrop spectrophotometer.

cDNA hybridization

A 200 to 300 pmol mixture of the labelled cDNA samples was concentrated (Concentrator Plus, Eppendorf™, Hamburg, Germany). Competitive hybridization was performed on a Yeast 6.4K Array with PCR-amplified ORFs of yeast S288c strain (Microarray Centre, UHN, Toronto, Ontario, Canada) in AHC hybridization chambers (ArrayIt Corporation, CA, USA) at 42°C overnight. Heterologous conditions as stated by Gamero et al.¹⁴ were employed to assure the hybridization of the S. kudriavzevii genome. The pre-hybridization solution contained 3X SSC, 0.1% SDS and 0.1 mg/ml BSA; the hybridization solution contained 0.1% SDS, 0.1 mg/ml of salmon DNA and 5X SSC. Microarrays were manually washed with different solutions containing different SSC 20X and SDS 10% concentrations (Sol.1: 0.1% SDS-2X SSC; Sol.2: 0.1% SDS-0.1X SSC; Sol.3: 0.1 SSC; Sol4: 0.01X SSC). The signal intensities of Cy3 and Cy5 were acquired with an Axon GenePix 4100A scanner (Molecular Devices, CA, USA) using GenePix Pro v.6.1 software, at a resolution of 10 µm.

Microarray data analysis

Microarray data were derived from three independent cDNA hybridization experiments. Background correction was performed with GenePix pro 6.0. Subsequent analyses were performed with the Acuity 4.0 software (Molecular Devices, CA, USA). Individual datasets were normalized to log₂ ratio value of 1 and data were filtered to remove spots that were not flagged and manually processed for print tip effect corrections. Only spots data with a minimum of two replicates were considered. Replicates were combined and medians were calculated. The first cut-off was the selection of the genes presenting at least 2-fold log₂ ratio values, according to the literature^15–17. These genes underwent a “GO term” enrichment analysis using the GO Term Finder tool in the Saccharomyces Genome Database (http://www.yeastgenome.org/). Regarding the statistics, False Discovery Rate (FDR) analysis and a significance level of 99% (p value < 0.01) were applied.

Quantitative PCR

Expression of genes NUG1, LSM8, PDC5, NSR1, GPD1 and GUT2 was investigated by qPCR. Normalization of gene expression was carried out using ACT1 and RDN18-1 as controls since their expression remains constant along fermentation. Primers were designed using Primer-BLAST (NCBI) and S. kudriavzevii and S. cerevisiae gene sequences were deposited in databases (www.ncbi.nlm.nih.gov; GSE90793). Forward and reverse oligonucleotides were synthetized to hybridize to the selected alleles of S. kudriavzevii or S. cerevisiae genes (Table 1). PCR Mastercycler pro (Eppendorf, Germany) confirmed the allele primer specificity and their annealing temperature. cDNA synthesis and RNA extraction were carried out as previously explained. qPCR runs were done in triplicate in a LightCycler® 480 Real-Time PCR System (Roche, Switzerland) and analyzed using the software from the manufacturer (LightCycler® Software, version 4.0). The relative gene expression was quantified by comparison with ACT1 and RDN18-1 expression, after confirm comparable PCR efficiency.

Paromomycin assays

Yeast cells grown overnight in GPY were diluted to 2 × 10⁶ cells/ml the next morning. They were then grown until the mid-log phase (approximately 1 × 10⁷ cells/ml), and 175 μl were inoculated on each GPY plate. A filter (1 cm diameter) with paromomycin (2 μg) was placed on the surface and plates were incubated at 12°C or 28°C until the lawn was formed. The assays were repeated twice.

S. cerevisiae x S. kudriavzevii hybrid strain fermentation performance under low temperature conditions

We carried out micro-fermentations in vessels with 0.45 L of Tempranillo grape must, mimicking previous work conditions⁴. The time needed for both hybrid strains to finish the fermentation at 28°C was similar, 5 and 6 days for Lalvin W27 and VIN7, respectively (Table 2). In contrast, at low temperature (12°C) the fermentation performance was quite different. W27 behaved more similar to the S. kudriavzevii type strain (11 days), taking 14 days to finish fermentation. Hybrid strain VIN7 performance was more similar to the S. cerevisiae parental strain (21 days) and took 23 days. The differences in fermentation kinetics amongst these strains revealed how allelic differences can determine the behaviour of oenological traits of interest.

Differential gene expression in S. cerevisiae x S. kudriavzevii hybrids at low temperature

To evaluate changes in the global expression of genes during acclimation to low temperature in the wine fermentation of natural must, sampling was done at the beginning of the exponential phase when cells are growing at a speed close to µ_max, two generations after inoculation. RNA from the samples was hybridised against the S288c microarray to study transcriptomic changes. An average of 86% sequence similarity exists between species of S. cerevisiae and S. kudriavzevii⁴, so we used heterologous hybridisation conditions for the microarrays. We observed that 95% of the total S288c gene spots were hybridised by S. kudriavzevii DNA⁴. Gene expression of each strain at both temperatures was analysed. Genes that were differentially expressed at 12°C and 28°C at a level that was considered significant were analysed further using the SAM (Significance Analysis of Microarrays) test with an FDR below 5% (Supplementary Table 1).

In S. cerevisiae x S. kudriavzevii hybrid strain VIN7 at low temperature, 18 genes were up-regulated, while 22 genes were down-regulated. For W27, 20 genes were up-regulated while 3 genes were down-regulated. GO analysis using the GO Term Finder on the Saccharomyces Genome Database was performed to evaluate the GO categories arising from the up- and down-regulated genes in both hybrid strains (Supplementary Table 2). Analysing the up-regulated genes in both strains after applying the GO-module online tool (http://www.lussiergroup.org/GO-Module) for false positives resulted mainly in GO terms related to “magnesium ion binding”, “thiamine pyrophosphate binding”, “branched-chain-2-oxoacid decarboxylase activity” and “pyruvate decarboxylase activity”. The last category appeared, because amongst the up-regulated genes we found the three pyruvate decarboxylases PDC1, PDC5 and PDC6, together with IDP2. The electron transport category was also significantly up-regulated in W27, similar to what has been previously described in S. cerevisiae and S. kudriavzevii⁴. Amongst other up-regulated GO-terms in W27 is also “amino acid metabolism” with a number of sub-categories, and in “heavy metal binding” in VIN7. These GO-terms were also previously observed in S. kudriavzevii. For the hybrid strain VIN7 the genes down-regulated at 12°C were related to rRNA synthesis, while for W27 no GO term categories were found.

Overall, the most remarkable group of regulated genes (either up or down-regulated) in the hybrid strains fell into the translation machinery efficiency category, as previously shown for S. kudriavzevii compared to S. cerevisiae at 12°C⁴. When comparing the two strains that showed high fermentation performance at 12°C (S. kudriavzevii IFO1802 and the hybrid W27) to the two strains with low fermentation performance at 12°C (S. cerevisiae T73 and VIN7), the cold adapted strains overexpressed 13 genes, including NUG1, involved in rRNA export¹⁸, and chaperones DDR48¹⁹ and SRP1²⁰ that couple proteasomes to polypeptides emerging from the ribosome (Supplementary Table 1).

Relative contribution of S. cerevisiae and S. kudriavzevii alleles to the total expression of key genes in the hybrids

An open question regarding yeast hybrids is the relative contribution of the different parental alleles to the total expression of specific genes. To test the role of the different alleles (S. cerevisiae or S. kudriavzevii) in the better adaptation to cold temperatures, we selected three differentially overexpressed genes in W27 in this work, NUG1, PDC5 and LSM8, and also three S. kudriavzevii cold stress markers described in different studies, NSR1, GPD1 and GUT2^3,4,21. We observed overexpression of the overlapping dubious ORF YJR023C for LSM8 and assumed that this overexpression was effectively in LSM8. NSR1, LSM8 and NUG1 are related to translation machinery efficiency, whereas GPD1 and GUT2 are related to glycerol metabolism, involved in cold adaptation. GPD1 and GUT2 are also involved in NAD⁺/NADH balance, with PDC5. In Table 3 we included the genomic configuration, obtained from previous work^6,11,12, for the selected genes in each of the two hybrid strains, showing that most genes have at least one copy of each parental allele with the exception of S. kudriavzevii allele losses for NSR1 in VIN7 and for GPD1 and GUT2 in the W27 strain. The results do not suggest a general correlation between the relative contribution of S. kudriavzevii alleles to total gene expression of key genes and the cryotolerance shown by W27.

W27 has one copy of the NSR1 S. cerevisiae allele, whereas VIN7 has three copies, but the S. cerevisiae allele expression is higher in the W27 strain. Thus, despite the higher number of total S. cerevisiae allele copies in VIN7, relative gene expression of the S. cerevisiae NSR1 allele is higher in W27, and also total relative expression is higher in W27. One explanation is that the S. kudriavzevii allele in W27 may promote expression of both alleles. NUG1 is another example of gene that does not show correlation between the number of copies and the level of expression. This gene has two copies of the S. kudriavzevii allele in W27 and one in VIN7 but shows higher expression in VIN7 than in W27. No expression was found for GUT2, but GPD1 expression was remarkably higher in the VIN7 strain mainly due to the high S. kudriavzevii allele contribution. There is also a negative correlation between the copy number of the S. cerevisiae allele for the PDC5 gene and gene expression. Despite the similar expression of the S. cerevisiae allele for the PDC5 gene between both strains, we observe that the Sc/Sk ratio is higher in VIN7 than W27, which suggests this gene may have impact on efficiency of cold vinifications. The Sc/Sk ratio is the relative expression of S. cerevisiae alleles divided by the relative expression of S. kudriavzevii alleles.

Effect of the translation inhibitor paromomycin

The expression of genes related to translation efficiency in the hybrid strains prompted us to study this phenotype. The translation efficiency of the hybrid strains compared to the parental strains at low temperature was analysed by testing their sensitivity to paromomycin, a potent inhibitor of translation²².

As it can be seen in Figure 1, the S. cerevisiae strain shows a growth inhibition halo at 12°C, whereas no inhibition halo can be seen in the S. kudriavzevii strain at 12°C, confirming that the translation efficiency is not compromised for this yeast at low temperatures. S. kudriavzevii showed growth defects at 28°C, whereas S. cerevisiae remained unaffected. The hybrid strains, on the other hand, clearly show growth defects at both temperatures. This means that 28°C is not an optimal growth temperature for either of these hybrids. An important difference between the hybrid strains is that W27 shows a narrower halo at 12°C compared to 28°C while for the VIN7 strain the situation is the opposite, supporting the idea of W27 being more similar to S. kudriavzevii strains in low temperature conditions.

Figure 1. Evaluation of the translation efficiency of S. cerevisiae, S. kudriavzevii and S. cerevisiae x S. kudriavzevii hybrids.

The inhibitory effect of the translation inhibitor paromomycin was evaluated by observing the halo diameter generated in the lawns of S. cerevisiae (T73), S. kudriavzevii (CR85) or S. cerevisiae x S. kudriavzevii hybrids (VIN7 and W27) growing in GPY plates at 28°C or 12°C. Representative experiments are shown, and the end of the inhibition halos are indicated with arrows.