Abstract
The efficiency and efficacy of alcoholic fermentation by yeast is crucial for the winemaking process. Sluggish or arrested fermentations can negatively affect winery operations and wine quality. Here, we present a novel mechanism by which problem fermentations can arise. Yeast can induce a prion known as [GAR+] that allows the cell to circumvent glucose repression of alternative carbon substrates. We have confirmed that Saccharomyces cerevisiae strain UCD932 can spontaneously generate the [GAR+] phenotype and that this phenotype exhibits the genetic traits of a prion. Differences were observed in the fermentative behavior of UCD932 wild-type [gar−] versus [GAR+] yeasts in laboratory-scale model juice fermentations. To further understand these differences, fermentations were performed in Chardonnay juice to study the interaction of the [GAR+] prion and presence of sulfur dioxide (SO2) on fermentation kinetics, bacterial community composition, and volatile compound production. Cells harboring the [GAR+] prion displayed reduced fermentation capacity, which was especially evident in the absence of SO2. Presence of SO2 and fermentation time had the most significant effects on the types of bacteria present in the fermentation. However, [GAR+] yeasts without added SO2 were especially sensitive to bacterial competition. This difference was also reflected in the bacterial and volatile profiles of the finished wine. We hypothesize that the bacterial induction of the [GAR+] prion by yeast during fermentation is another possible mechanism by which stuck or sluggish fermentations may become established.
- [GAR+] prion
- glycerol glucosamine media
- LAB
- lactic acid bacteria
- minimal must media
- Saccharomyces
- SO2
- sulfur dioxide
The robustness of Saccharomyces cerevisiae fermentations is indispensable for the winemaking process. Yeasts are able to quickly and efficiently convert the glucose and fructose present in grape juice into ethanol through alcoholic fermentation. Problem fermentations arise when yeast are unable to efficiently convert sugar into alcohol, causing the fermentation to become sluggish or arrested, leaving behind residual sugar, and, in turn, raising the potential for microbial spoil age. The causes of problem fermentations are well documented, often stemming from several factors, such as insufficient amounts of nitrogen or other nutrients, inappropriate strain selection, and issues with temperature management (Ingledew and Kunkee 1985, Alexandre and Charpentier 1998, Bisson 1999). Here, we present evidence for another possible mechanism by which problem fermentations may arise.
S. cerevisiae displays a biologically hardwired response to glucose. When present at sufficiently high concentrations, glucose (as well as fructose) is preferentially metabolized over any other available carbon source. This phenomenon, known as “glucose-associated repression”, prioritizes the fermentation of glucose over the utilization of alternative fermentable and nonfermentable carbon sources (Kayikci and Nielsen 2015). Strong upregulation of genes associated with fermentation gives yeast a significant advantage over microbial competition, enabling rapid production of ATP, oxygen consumption, and ethanol accumulation. Therefore, loss of glucose repression may negatively affect fermentation performance (Dashko et al. 2014). Depletion of oxygen and nutrients, coupled with the production of ethanol, enables yeast to become the dominant organism when glucose is present, especially at the high sugar concentrations typically found in the natural environments of yeast, such as grape juice (Boulton et al. 1996). However, this innate yeast metabolic circuit can be circumvented by the induction of a novel yeast prion.
Yeast prions are intracellular proteins capable of existing in at least two stable states (Garcia and Jarosz 2014, Wickner et al. 2015). In yeast, these prions act as heritable epigenetic “switches” that allow a portion of a cell population to adapt dynamically to environmental stress (Shorter and Lindquist 2005, Newby and Lindquist 2013). The induction of these prion states depends on intracellular conditions, metabolite concentrations, and the family of heat shock proteins (HSPs) (Verghese et al. 2012). Since the specific protein conformational state is automatically established in progeny cells, the prion phenotype is “inherited” from the mother cell (Halfmann and Lindquist 2010). HSPs are activated in response to stress, can function as molecular chaperones regulating protein refolding and protein–protein interactions, and are also required for faithful propagation of the prion state (Kiktev et al. 2012, Liebman and Chernoff 2012).
A population of yeast undergoing stress will spontaneously induce these prion states at predictable frequencies, allowing a subset of the population to sample new phenotypes and/or modes of metabolism (Halfmann et al. 2010). Although their exact role is controversial, it is now clear that yeast prions do occur naturally in wild strains and can provide a benefit under the appropriate conditions (Halfmann et al. 2012). A hallmark of yeast prion biology is that these prion states often mimic mutant phenotypes, serving a parallel purpose, but in a dynamic manner without the threat of permanently affecting the population genome or long-term fitness (Lancaster et al. 2010, Garcia and Jarosz 2014). Thus, prions promote phenotypic diversity in the absence of genetic mutations, thereby protecting the original genotype under variable and stressful environmental conditions.
The [GAR+] (resistant to glucose-associated repression) prion allows yeast to bypass glucose repression (Brown and Lindquist 2009). [GAR+] establishment depends on Hsp70 and on the association of the plasma membrane ATPase Pma1 with Std1, a co-transcription regulator of hexose transporters (HXTs) (Lakshmanan et al. 2003, Horak 2013). [GAR+] displays all of the defining genetic characteristics of a yeast prion: it is reversibly curable, transferable by cytoduction, dominant, and exhibits 4:0 segregation exclusively during meiosis. Another consequence of [GAR+] induction is an ~40-fold drop in the expression of the ubiquitous, high-capacity sugar transporter HXT3, which is important for the efficacy of fermentation (Karpel et al. 2008). We hypothesized that this drop in expression of HXT3 in [GAR+] cells would lead to overall reduced fermentative rates and potentially to an inability to complete fermentation and to effectively dominate the microbial community. In addition, the discovery that bacteria, including species present during wine production, induce the prion state in wine strains of S. cerevisiae suggests that the prion-based modification to yeast metabolism may confer some benefits to bacterial populations present during fermentations (Jarosz et al. 2014a).
In winemaking, sulfur dioxide (SO2) is used to both block oxidation reactions and moderate microbial growth, enhancing fermentation efficiency by S. cerevisiae (Constanti et al. 1998, Divol et al. 2012). Since the [GAR+] prion impacts the ability of the strains to ferment and dominate a fermentation, the presence of SO2 was hypothesized to mitigate the effects of prion induction in wild microbial communities if that induction indeed depends on the biological activity of the bacteria present. To test the hypothesis that bacterial induction of [GAR+] could lead to problem fermentations, we set out to evaluate the impact of this prion on fermentation kinetics, bacterial population composition, and volatile compounds produced in the fermentation of Chardonnay in the presence and absence of SO2.
Materials and Methods
Yeast strains and [GAR+] frequency test
The yeast strains used in this study are presented in Table 1. Single yeast colonies were inoculated with a sterile loop into 10 mL of yeast peptone dextrose (YPD) (1% Bacto Yeast Extract and 2% Bacto Peptone [Becton, Dickinson and Company], and 2% dextrose [Amresco]) and grown on a rotary shaker at 25°C overnight. The cultures were diluted 1:10 and absorption at 660 nm was measured, and 5-mL YPD cultures were inoculated at 0.05 A660 and grown to midlog phase (0.4 to 0.6 A660). Depending on optical density, the appropriate aliquots of the cultures were transferred to 1.5-mL Eppendorf microcentrifuge tubes such that the cells could be resuspended at 0.68 A660. The tubes were centrifuged at 2500 g for 2 min, the supernatant was removed, and pellets were resuspended in sterile water at 0.68 A660; 200 μL of suspension was transferred into a sterile 96-well plate and serially diluted 1:5 for five times. The serial dilutions were then either stamped or spotted (3.5 μL) onto glycerol glucosamine medium (GGM) (1% Bacto Yeast Extract, 2% Bacto Peptone, 2% glycerol [Fischer Chemical], and 0.05% glucosamine [Acros Organics]). Plates were incubated at 30°C, and the extent of growth in each dilution was assessed and photographed on days 3, 5, and 7.
Yeast strains used in this study.a
Prion curing and sporulation
The pAG42 plasmid containing a dominant-negative ssa1∆ K69M allele (Jarosz et al. 2014b) was obtained from the Jarosz laboratory (Stanford University). UCD932 [gar−] and [GAR+] strains were transformed with pAG42 according to the method of Gietz and Woods (2002) and plated on selective YPD plates containing 150 μg/mL hygromycin B (Invitrogen). Single colonies were selected and restreaked five times on selective YPD plates containing 150 μg/mL hygromycin B to cure the prion, and restreaked four times on 2% YPD to induce loss of the pAG42 plasmid. Cured colonies were subjected to the [GAR+] frequency test as described above.
Sporulation and dissection of UCD932 [gar−] and [GAR+] strains were performed according to Treco and Winston (2001). Sets of tetrads were selected, restreaked onto 2% YPD, and then used for a [GAR+] frequency test.
Laboratory-scale fermentations
Single UCD932 [gar−] and [GAR+] colonies were inoculated into 10 mL of minimal must medium (MMM) prepared at 208 mg/L yeast nitrogen equivalents as described by Giudici et al. (1993) and Spiropoulos et al. (2000). Cultures were grown on a rotary drum for two days, measured for absorbance, and inoculated in triplicate at 0.05 A660 into 150 mL of MMM in 250-mL baffled flasks. Fermentations took place at 25°C, with shaking at 120 rpm, and were weighed daily to track CO2 evolution and fermentation progression. Postfermentation, yeast viability was assessed with methylene blue staining (Sami et al. 1994). Average cell size was measured with a handheld Coulter counter, Scepter 2.0 (EMD Millipore). Amino acid analysis was performed at the UC Davis Proteomics Core, according to Cooper (2000). Residual sugar was measured with an Anton Paar Portable Density Meter (DMA 35N; Anton Paar). Prism 6 (GraphPad) was used to determine statistical significance by multiple t tests with a false discovery rate of 5%; rows were analyzed individually, assuming a constant standard deviation (SD).
Winery-scale fermentations
Unfermented Chardonnay juice had a sugar content of 24.6 Brix, a pH of 3.6, a titratable acidity (TA) of 5.0 g/L, a nitrogen by O-phthaldinaldehyde assay (NOPA), and ammonia concentrations of 147 mg/L and 78 mg/L, respectively. Single UCD932 [gar−] and [GAR+] colonies were inoculated into 10 mL of 0.45-μM sterile filtered Chardonnay and grown at 25°C on a rotary drum for two days. These cultures were used to inoculate 300 mL of sterile filtered Chardonnay at 0.05 A660 in 500-mL round-bottom flasks. The cultures were then grown at 25°C with shaking at 120 rpm for two days and used to inoculate at 0.05 A660 in triplicate 18.9 L of unfiltered Chardonnay that had been racked into static 22.7 L stainless steel drums.
One set of [gar−] and [GAR+] triplicates received no SO2, and the other set received 50 mg/L of SO2; ambient temperature was held steady at 17°C. Fermentations were sampled daily, and the Anton Paar DMA 35N meter was used to measure Brix and temperature. Ten mL samples were taken over the course of the fermentations to conduct standard wine chemistry tests via the Gallery Automated Photometric Analyzer (Thermo Scientific) with kits from Sigma-Aldrich.
Bacterial sampling and sequencing
Samples (50 mL) were acquired from a consistent height in the fermentation vessels; this midpoint represented the composition of the suspended biomass at each sampling point. Samples were drawn preinoculation and then at days 3, 10, 18, and 32 postinoculation, transferred into 50-mL Falcon conical centrifuge tubes (Fisher Scientific), and immediately stored at −20°C until DNA extraction and library construction.
Sequencing library construction
Amplification and sequencing was performed as described previously for bacterial communities (Bokulich et al. 2012). Briefly, the V4 domain of bacterial 16S rRNA genes was amplified with primers F515 (5′-NNNNNNNNGTGTGCCAGCMGCCGCGGTAA-3′) and R806 (5′-GGACTACHVGGGTWTCTAAT-3′) (Caporaso et al. 2011), with the forward primer modified to contain a unique eight-nucleotide barcode (the italicized poly-N section of the primer above) and a two-nucleotide linker sequence (bold, underlined portion of the primer) at the 5′ terminus. Polymerase chain reactions (PCR) contained 5 to 100 ng DNA template, 1× GoTaq Green master mix (Promega), 1 mM MgCl2, and 2 pmol of each primer. Reaction conditions consisted of an initial 94°C for 3 min; followed by 35 cycles of 94°C for 45 sec, 50°C for 60 sec, and 72°C for 90 sec; and a final extension at 72°C for 10 min. Amplicons were combined into two separate, pooled samples at roughly equal amplification intensity ratios, purified with the QIAquick spin kit (Qiagen), and submitted to the UC Davis Genome Center DNA Technologies Core for Illumina paired-end library preparation, cluster generation, and 250-bp paired-end sequencing on an Illumina MiSeq sequencer.
Bacterial profile data analysis
Raw fastq files were demultiplexed, quality-filtered, and analyzed with QIIME v.1.8.0 (Caporaso et al. 2010b). The 250-bp reads were truncated at any site of more than three sequential bases receiving a quality score <Q10, and any read containing ambiguous base calls or barcode/primer errors was discarded, as were reads with <75% (of total read length) consecutive high-quality base calls (Bokulich et al. 2013). Reverse primer sequences were trimmed from the ends of ITS sequences after the demultiplexing.
Operational taxonomic units (OTUs) were clustered at 97% identity with the QIIME subsampled reference OTU-picking pipeline and with UCLUST (Edgar 2010) against the Greengenes 16S rRNA gene database (May 2013 release) (McDonald et al. 2012), modified as described previously (Bokulich and Mills 2013). OTUs were classified taxonomically against these same databases according to Ribosomal Database Project classifiers (Wang et al. 2007). Any OTUs comprising <0.01% of total sequences for each run were removed prior to further analysis (Bokulich et al. 2013, 2014). Bacterial 16S rRNA gene sequences were aligned in PyNAST (Caporaso et al. 2010a) against a reference alignment of the Greengenes core set (McDonald et al. 2012). From this alignment, chimeric sequences were identified and removed with ChimeraSlayer (Haas et al. 2011), and a phylogenetic tree was constructed from the filtered alignment with FastTree (Price et al. 2010). Sequences failing the alignment or identified as chimeras were removed prior to downstream analysis. Principal coordinates were computed from the resulting distance matrices to compress dimensionality into three-dimensional principal coordinate analysis plots, enabling visualization of sample relationships.
To determine whether sample classifications (i.e., sulfite concentration, [GAR+] status, and bacterial contamination) showed differences in phylogenetic or OTU diversity, ADONIS permutational multivariate analysis of variance (MANOVA) (Anderson 2001) and analysis of similarities (ANOSIM) (Clarke 1993) with 999 permutations were used to test the null hypothesis that sample groups were not statistically significant. For all classifications for which this null hypothesis was rejected, a Kruskal-Wallis test was used to determine which taxa differed significantly (with Bonferroni error correction) in sample groups receiving different sulfite doses and different inoculation strains.
Sample preparation for volatile analysis
10 mL of undiluted wine samples were pipetted into 20-mL round-bottom amber glass vials (Agilent Technologies) with 3 ± 0.02 g NaCl (Fisher Scientific). 2-undecanone was added to each vial as an internal standard to a concentration of 50 μg/L. Vials were sealed with BiMetal 20-mm crimp caps lined with PTFE/silicone septa (Sigma-Aldrich). All treatments were analyzed in triplicate, with samples being run in randomized order within 18 hr of loading onto the instrument. A 2-cm divinylbenzene/carboxen/polydimethylsiloxane (Supelco), 23-gauge solid-phase microextraction (SPME) fiber was used for sampling. Samples were warmed to 40°C and agitated at 500 rpm for 5 min before exposing the fiber for 30 min at 40°C with agitation at 250 rpm.
HS-SPME-GC-MS analysis
Samples were analyzed in a 6890 gas chromatograph coupled to a 5975 mass selective detector (MSD) (Agilent Technologies) equipped with a MPS2 autosampler (Gerstel). A DB-Wax capillary column (30 m, 0.25 mm i.d., 0.25 μm film thickness) (J&W Scientific) and SPME inlet liner (0.77 mm i.d.; Supelco) were used. The instrument was operated with Maestro software (ver. 1.2.3.1; Gerstel), and the data were analyzed with ChemStation software (ver. E.01.01.335; Agilent Technologies). Helium was used at the carrier gas at a constant flow of 1 mL/min. The retention time was locked to the internal standard, 2-undecanone, at constant flow to prevent retention-time drifting.
During the analysis, the oven temperature was kept at 40°C for 5 min, then increased at a rate of 3°C per min up to 180°C and then at a rate of 30°C per min up to 240°C, followed by a hold for 10 min. The MSD interface was held at 240°C. The inlet temperature was 240°C, and the SPME fiber was desorbed in split mode with a 20:1 split ratio. The solvent delay was 2.5 min, and the detector was turned off from 3.8 min to 4.3 min during ethanol elution. The fiber was held in the inlet for 10 min to prevent carryover effects. An electron ionization source was used, with a source temperature of 230°C and an electron energy of −70 eV.
The wines were measured with synchronous scan and selected ion monitoring (SIM) mode. The scan parameters ran from 40 m/z to 300 m/z, and both scan and SIM acquisitions were optimized such that there was a minimum of 15 scans over each peak (King et al. 2012, Hjelmeland et al. 2013).
Compound identification and data processing
In total, 16 compounds were identified with the headspace solid-phase microextraction gas chromatography-mass spectrometry (HS-SPME-GC-MS) method, representing a set of volatile compounds commonly present in finished wines. All 16 compounds were confirmed by analyzing the retention times of reference compounds purchased from Sigma-Aldrich. Standards were diluted with 100% ethanol (Gold Shield Chemical), and the retention times of the authentic standards were matched to the compounds measured. The compounds were also verified with the quantifier/qualifier ion ratios and published retention indices reported for the DB-Wax column.
The MetaboAnalyst online platform was used to determine significance by ANOVA, and to perform principal component analysis (PCA) (Xia et al. 2015). Compounds determined to be significantly different among treatments were normalized to the highest value of each specific compound. Significance across treatments was determined by Fisher’s least significant difference test.
Results
Evaluation of [GAR+] induction frequency in vineyard isolate UCD932
We performed an initial screen of wine-associated S. cerevisiae isolates to assess [GAR+] autoinduction across different strain backgrounds. From this screen, an Italian vineyard isolate, UCD932 (Mortimer et al. 1994), was selected for further study. UCD932 is genetically tractable and defined genetically as homozygous and displayed a typical fermentation profile (Figure 1A).
Fermentation progression and [GAR+] induction in UCD932. (A) Fermentation progression in synthetic juice of UCD932 compared to other native (DBVPG6040), commercial (EC1118), or laboratory (W303) strains. (B) [GAR+] autoinduction frequency on GGM agar. Strains were grown to 0.4 A660, then serially diluted 1:5 from right to left. The GGM frequency assay is shown on day 5 postplating.
Wine strains were evaluated on the basis of the “strength” of the innate ability to induce the [GAR+] prion. Strength of induction was defined as the population frequency of autoinduction on GGM (i.e., the highest dilution showing growth on the plates) and speed with which colonies grow on GGM (i.e., the number of days needed to reach maximal colony density on the plate). UCD932 exhibited a moderate to high ability to autoinduce [GAR+] on GGM (Figure 1B). UCD932 induced the [GAR+] prion at a frequency of ~1 in 300 cells, a frequency similar to those of other wine strains tested, and significantly higher than the reported frequency for laboratory strains such as W303 (1 in 105 cells) (Brown and Lindquist 2009).
[GAR+] in UCD932 displays the genetic characteristics of a prion
S. cerevisiae can grow on GGM either because of induction of the [GAR+] prion or because of mutations enabling growth on this selective medium. Therefore, the frequency and nature of the heritable change leading to growth of UCD932 on GGM was investigated to determine whether the growth of UCD932 on GGM was due to prion induction or to some type of mutation or other adaptation. Four independent tests were used: (1) frequency of appearance of the phenotype, (2) frequency of appearance in strains with haploid versus those with diploid genomes, (3) the pattern of phenotype segregation after meiosis, and (4) the ability to reversibly cure the phenotype.
The appearance of a mutation occurs on average at a frequency of 6.3 × 10−5 per haploid genome in a diploid cell (Joseph and Hall 2004) and would be expected to be on the order of 3 in 10−9 ([6.3 × 10−5]2) for spontaneous mutations in both copies of a single gene in diploid strains. However, the frequency of appearance of prions is orders of magnitude greater than this mutational frequency, depending on the type of stress, the prion being induced, and the strain background (Garcia and Jarosz 2014). The frequency of appearance of the [GAR+] phenotype in diploid strains of UCD932 is roughly 1/300 (3 × 10−3), which is inconsistent with a genomic mutation being responsible for the observed phenotype.
Maintaining a strain on GGM for several generations caused the “strength” of the [GAR+] phenotype to increase, presumably by enriching and selecting for the prion within the yeast population community (Figure 2A). The detection of a spontaneous recessive mutation of a gene within a yeast cell depends on the number of copies of that gene that must be altered for the phenotype to become apparent. A mutation in a gene is therefore more readily detected in a haploid cell, which contains only one copy of that gene, than in a diploid cell in which both gene copies must be altered. In contrast, prion-based phenotypes are induced at the same frequency in haploid and diploid cells because these phenotypes are independent of gene copy number. Accordingly, the frequency of appearance of the [GAR+] phenotype was compared between haploid and diploid UCD932 cells. Haploid MATa and MATα mating types of UCD932 had identical frequencies of induction as compared to the diploid strains in spite of containing only a single copy of all chromosomes (Figure 2B), suggesting that ploidy played no significant role in [GAR+] phenotype induction, consistent with the expected prion-based phenotype.
Genetic characteristics of the [GAR+] prion in UCD932. All strains were grown to 0.4 A660, and then serially diluted 1:5 from right to left. (A) Preadapted UCD932 [GAR+] shows enhanced growth on GGM compared with UCD932 [gar−]. (B) Comparison of [GAR+] autoinduction frequency in haploid UCD932 MATa and MATα strains (top two strains) with diploid UCD932 [gar−] and [GAR+] strains (bottom two strains). (C) Following meiosis of UCD932 [GAR+], all four spores exhibited the [GAR+] phenotype. (D) Curing of the [GAR+] prion via the expression of a dominant-negative copy of ssa1Δ contained on the pAG42 vector. A [GAR+] frequency test was performed after the cured UCD932 [GAR+] pAG42 had been passaged on nonselective media to induce loss of the plasmid (Newby and Lindquist 2013); growth of the UCD932 [gar−] and [GAR+] strains is shown for comparison.
However, an equivalent frequency of appearance of mutations in diploid and haploid UCD932 would also be consistent with the [GAR+] phenotype, due to a dominant genomic mutation rather than a prion or a recessive genomic mutation. In the case of a dominant mutation, only one copy of the gene needs to be altered, and so the impact of ploidy is negated. Dominant genomic mutations can be differentiated from prion-based phenotypes according to the pattern of phenotype segregation following sporulation. Upon sporulation of diploid cells carrying a prion, all four spores will show the prion-associated phenotype (4:0), as all cells inherit the same cytoplasm of the parent. If the phenotype is due to a mutation in a single dominant gene, however, the spores will show a 2:2 segregation of the phenotype as chromosomes will segregate during division. Therefore, UCD932 [GAR+] cells were sporulated and dissected, and a frequency test was performed on the resulting spores to define the mode of inheritance.
Non-Mendelian or 4:0 segregation of the [GAR+] phenotype was observed for all tetrads dissected from diploids isolated from growth on GGM, further confirming that [GAR+] in UCD932 meets the defining genetic characteristics of a prion rather than a dominant genomic mutation (Figure 2C). Interestingly, the spores from the wild-type [gar−] UCD932 yielded two spores that grew out to one more dilution than the other two, suggesting genetic factors in the strain backgrounds that can enhance or reduce the ability to induce [GAR+] (Figure 2C).
Another hallmark of prion biology is the reversibility of the prion state. [GAR+] depends on yeast Hsp70 Ssa1 for establishment and faithful propagation (Brown and Lindquist 2009). The [GAR+] prion may be lost, or “cured”, by expressing a defective copy of the HSP70 gene, SSA1 (Jarosz et al. 2014b). UCD932 was transformed with a plasmid carrying the dominant-negative ssa1∆K69M allele. Expression of this defective Hsp70 copy over multiple generations led to a loss of [GAR+] by titration, reflected in frequency tests of the transformed yeasts (Figure 2D). This loss of the [GAR+] prion in UCD932 required more generations than has been reported for laboratory strains, but the phenotype was curable in an Hsp70-dependent manner. In conclusion, all the tests confirmed the prion-based nature of the [GAR+] phenotype in UCD932 strains growing on GGM selective media.
Yeast harboring [GAR+] display a reduced fermentation capacity in synthetic and sterile juice
Induction of [GAR+] results in an ~40-fold drop in HXT3 transcript level (Brown and Lindquist 2009). Since HXT3 is the major catabolic transporter active during grape juice fermentation, the fermentative ability of [GAR+] versus [gar−] strains of UCD932 was compared in a variety of conditions. To mitigate the effects of using inherently variable grape juice, fermentations were also performed in a defined synthetic model juice, MMM (Giudici et al. 1993, Spiropoulos et al. 2000).
The fermentation performances of both UCD932 [gar−] and [GAR+] were assessed in MMM with 208 mg/L of yeast available nitrogen (calculated directly from amino acid and ammonium ion analyses) (Figure 3A). Despite the same level of inoculation, [GAR+] cells lagged behind their [gar−] counterparts early in the fermentation. The [GAR+] cells were able to achieve a similar fermentation rate after the initial lag phase. Postfermentation analysis showed that the [GAR+] fermentations had a statistically significant but only slightly higher residual Brix ([gar−]: −2.3 ± 0.1 Brix, [GAR+]: −2.1 ± 0.1 Brix), and, consequently, lower ethanol ([gar−]: 13.52% ± 0.34%, [GAR+]: 13.25% ± 0.13%). It was also observed that the [GAR+] cells were on average larger than [gar−] cells, and had increased viability postfermentation (Figure 3B).
Comparison of fermentation behaviors of UCD932 [GAR+] and [gar−] strains. (A) Synthetic juice fermentations. UCD932 [GAR+] cells exhibit a lag phase at the beginning of fermentation. (B) Postfermentation cell size and viability. UCD932 [GAR+] cells are on average larger, and are viable longer than their [gar−] counterparts. Significance as determined by t tests is indicated by an asterisk (*). (C) Chardonnay juice fermentation. UCD932 [GAR+] shows a longer lag phase and lower fermentation rates than [gar−]. (D) [GAR+] cells leave more nitrogen-containing compounds in solution postfermentation than do [gar−] cells; significance as determined by t tests is indicated by asterisks (*).
UCD932 [gar−] and [GAR+] were also assessed in sterile filtered Chardonnay (Figure 3C). [GAR+] cells in juice displayed a slight lag phase and reduced fermentation rate when compared with MMM. This observation could be explained by the differences in nutritional composition between the grape juice and the defined medium. Similar to fermentations in MMM, [GAR+] cells in sterile juice display residual Brix values ([gar−]: −2.4 ± 0.1 Brix, [GAR+]: −0.8 ± 0.04 Brix) and produced less ethanol ([gar−]: 15.58% ± 0.11%, [GAR+]: 15.07% ± 0.22%). Postfermentation analysis indicated that [GAR+] wines had significantly higher concentrations of certain amino acids left in solution (Figure 3D). All of these data strongly indicated that the fermentation efficacy of [GAR+] cells was reduced in both synthetic and natural juices, with a more severe effect seen in the juice used in this study.
The effect of SO2 on UCD932 [gar−] and [GAR+] fermentation kinetics
The aforementioned experiments used media and juices that had been filter sterilized prior to introduction with the yeast. However, under typical winemaking conditions, fermentations are not sterile, and fermentation progression depends on the ability of a yeast strain to successfully dominate the fermentation.
UCD932 [gar−] and [GAR+] cultures pregrown in sterile filtered UCD Chardonnay juice were inoculated in triplicate into 18.9-L treatments of UCD Chardonnay. One set of treatments had no added SO2 and another set had 50 mg/L of total SO2 added. As expected, the 50 mg/L SO2 treatments fermented considerably faster and more robustly than fermentations without added SO2 (Figure 4A). Fermentations without added SO2 were slower to start, indicating a higher microbial load and increased microbial competition (Bokulich et al. 2015).
UCD932 fermentation of Chardonnay (A) Fermentation profile of UCD932 [GAR+] and [gar−], with and without SO2 addition. (B) COX1 PCR analysis of colony-isolate identities. Treatments with 50 mg/L SO2 and [gar−] without SO2 displayed COX1 banding patterns consistent with UCD932, whereas the [GAR+] without SO2 displayed a mixture of COX1 banding patterns.
During fermentation, samples of yeast cells were plated and evaluated with COX1 PCR to confirm their identity as UCD932 [gar−] or [GAR+] (Figure 4B) (Lopez et al. 2003). Strains isolated from the [gar−] fermentations showed the COX1 banding patterns indicative of UCD932. However, strains from the [GAR+] treatment without SO2 showed a mixture of COX1 profiles, suggesting that [GAR+] cells do not dominate wild strains of Saccharomyces as well as their [gar−] progenitors (Figure 4B).
Addition of SO2 strongly affected fermentation performance of the different treatments (Figure 4A). Consistent with our observations in MMM and sterile juice, [GAR+] treatments lagged behind those of their [gar−] counterparts, especially at the start of the fermentation (Figure 4A). The [GAR+] cells without SO2 displayed the most problematic fermentation profile. This treatment took almost a week to begin fermentation and exhibited consistently sluggish kinetics. Both of the [GAR+] conditions were unable to efficiently finish fermentation, indicated by a higher residual sugar and lower alcohol content (Table 2). Wine chemistry was also affected as a function of SO2 and [GAR+]. In treatments with no added SO2, malate was relatively depleted, especially in the [GAR+] without SO2 treatment, indicating increased bacterial activity. Furthermore, the [GAR+] without SO2 treatment also exhibited an increase in both TA and volatile acidity (VA) (Table 2).
Final wine chemistry of UCD932 in unfiltered UCD Chardonnay.a
Inoculation, SO2, and [GAR+] affect the bacterial community structure in fermentation
Given the initial complexity of the microbial community in juice, we were interested in how the bacterial community would change as a function of SO2 addition, the presence of [GAR+] versus [gar−] yeast cells, or both. To determine the bacterial composition in these wines, samples were taken throughout the course of fermentation and analyzed by 16S rRNA marker gene sequencing (Bokulich et al. 2015). Component analysis helped visualize how the treatments differed in bacterial community composition at each sampling point (Figure 5A). Addition of SO2, inoculation, and day of fermentation exerted the strongest effects on bacterial diversity. Initially, the treatments had similar bacterial community compositions. As the fermentations progressed, the treatments differentiated into high versus low relative abundances of various bacterial taxa.
Bacterial population composition during fermentation. (A) The data points represent individual fermentations on each day sampled. Points are separated along Component 1 by phylogenetic compositional differences; thus, the further apart data points are, the more different the bacterial populations. (B) Relative-abundance bar graphs showing the ratios of different bacterial taxa in samples taken over the course of fermentation. The addition of 50 mg/L SO2 strongly inhibited the appearance of spoilage-associated lactic acid bacteria (LABs) regardless of [GAR+] status. Treatments without added SO2 displayed higher relative abundances of LABs, with the [GAR+] treatments showing significantly more of the spoilage-associated taxa.
For accuracy and simplicity, the bacterial taxa are presented at the family level. Statistical analysis identified significant taxa across the treatments according to the relative abundance of each clade at the different sampling points. Relative-abundance plots helped visualize how the significant taxa changed as a function of SO2 addition, presence of [GAR+] cells, and fermentation progression (Figure 5B). Addition of 50 mg/L of SO2 severely limited the appearance of spoilage-associated Lactobacillaceae. Compared with the SO2 treatments, all treatments without added SO2 displayed high abundances of Lactobacillaceae. The [GAR+] without SO2 treatment had a particularly high relative abundance of spoilage-associated lactic acid bacteria (LABs) at the later time points.
Impact of prion status and SO2 on the composition of wine volatile compounds
Informal sensory evaluations suggested that wines produced from [GAR+] and [gar−] cells displayed differences in aroma. To quantify these differences, we used HS-SPME combined with high-resolution GC-MS to analyze the volatile compounds produced in the winery-scale fermentations. Some volatiles produced in fermentations indicate yeast stress, whereas others are strongly associated with bacterial metabolism (Boulton et al. 1996, Fairbairn 2012).
Using the SPME technique, we were able to identify and quantify several volatile compounds across all the wines. PCA showed a clear segregation of the SO2 and prion conditions (Figure 6A), driven by a set of significant volatile compounds present in each treatment (Figure 6B). With ANOVA, we identified 10 volatile compounds that were significantly different among the treatments and gave insight into the underlying sensory differences among these treatments. The levels of significant compounds in the treatments are presented as relative values normalized to the highest peak area (Figure 6C).
Changes in wine volatile composition as a function of SO2 and [GAR+]. (A) PCA of the individual treatments. (B) A PCA biplot shows which compounds are responsible for differentiating the treatments in the PCA plot to the left. (C) Relative normalized plots visualize the significance of a treatment on a particular compound. For each compound, the y-axis represents the compound peak area for each sample normalized to the compound peak area of the sample with the highest relative amount of that compound. Different letters above the bars denote significant differences among the treatments as determined by Fisher’s least significant difference test.
The largest difference was observed in ethyl lactate, an ethyl ester of lactic acid indicating malolactic fermentation (MLF) and bacterial metabolism (Ribéreau-Gayon et al. 2006) (Figure 6C). Samples from the UCD932 [GAR+] treatment without SO2 displayed a strong peak of ethyl lactate, but this compound was undetectable in samples from all the other treatments. This observation was consistent with the bacterial communities present in these treatments. Differences in phenylethyl acetate were driven by SO2 addition, higher for juices without SO2, and lower for juices with added SO2. Also indicative of bacterial metabolism, the samples of the UCD932 [GAR+] treatment without added SO2 displayed a larger peak of acetic acid than the other treatments. Ethyl octanoate, a fatty acid–derived ester (Vianna and Ebeler 2001), was high across all treatments except for UCD932 [GAR+] without SO2. Interestingly, the presence of [GAR+] cells affected two volatiles differently, depending on SO2 addition. Isobutanol, a higher alcohol derived from valine metabolism (Dickinson et al. 1998), increased in samples fermented without SO2 compared with those with added SO2. Conversely, ethyl butanoate, a fruity fatty acid ester (Vianna and Ebeler 2001), was decreased in samples from treatments without SO2 compared with the SO2 treatments. The differences in the volatile compound data were consistent with our observations of the presence or absence of specific bacterial profiles.
Discussion
UCD932, isolated from a Lambrusco grape must in Emilia Romagna, Italy (Mortimer et al. 1994), was selected as a model strain for studying the effects of [GAR+] on fermentation kinetics. UCD932 is genetically tractable and induces the [GAR+] prion at a moderate to high frequency. This work has demonstrated that the phenotypic change enabling UCD932 to bypass glucose repression and, in turn, to grow on the selective GGM medium is prion-based.
In laboratory-scale fermentations with synthetic juice, the differences in fermentation initiation and progression, although statistically significant, were relatively modest between UCD932 [gar−] and [GAR+] cells. This lack of a strong difference was likely due to the presence of ample macro- and micronutrients and to the lack of stress and microbial competition inherent to fermentations of a sterile, defined medium. The reduction in HXT3 transporter expression does not in and of itself appear to lead to sluggish or arrested fermentations, suggesting that other transporters unaffected by [GAR+] are able to provide hexose transport and facilitate eventual completion of fermentation under these permissive growth conditions. Interestingly, previous research has shown that fermentation will progress in strains entirely lacking the HXT3 gene (Karpel et al. 2008), suggesting that the low levels of HXT3 that remain in [GAR+] cells may play a regulatory role in central glucose metabolism or that the induction of this prion impacts cell physiology beyond simple reduction of HXT3 transcript levels and loss of sugar-uptake capacity.
Analysis of the mechanism of [GAR+] prion induction in both laboratory and wine strains of S. cerevisiae has demonstrated that this prion state can be induced by bacteria present in the environment (Jarosz et al. 2014a). Interestingly, we have previously shown that bacterial species isolated from natural grape juices and wines, particularly those isolated from arrested fermentations, efficiently induce prion formation in yeast strains (Jarosz et al. 2014a, unpublished observations). In a typical fermentation, S. cerevisiae dominates these other microorganisms by quickly depleting oxygen and nutrients in juice and by producing ethanol as a by-product of hexose fermentation (Boulton et al. 1996). These combined strategies strongly inhibit the growth of other organisms that need oxygen and nutrients for growth. Accordingly, bacterial strains that can alter yeast metabolism could benefit nutritionally from this reduction in yeast biochemical activity. Consistent with this hypothesis, UCD932 [GAR+] without SO2 treatments displayed distinctly different ratios of organisms and aroma composition when compared to the [gar−] strain under the same conditions.
Given the widespread presence of bacteria capable of inducing the [GAR+] prion in many commercial and vineyard strains of S. cerevisiae, a much higher incidence of sluggish fermentations would be predicted than is observed commercially (Bokulich et al. 2015). We hypothesized that the use of SO2 to control bacterial communities would mitigate their effects on induction of the [GAR+] prion. The addition of 50 mg/L SO2 to juice exerted a strong effect on fermentation kinetics and bacterial community composition in [GAR+] yeast strains. Through plating and microscopy examinations, and as previously reported (Bokulich et al. 2015), lower initial microbial loads were observed in treatments with added SO2 in fermentations subsequently inoculated with UCD932 [GAR+]. This observation suggests that the greater negative impact seen in fermentation progression of the [GAR+] fermentations was in part due to the simultaneous proliferation of the bacteria.
Sequencing of the bacterial taxa revealed detailed information about the bacterial community structure during fermentation and as a function of SO2 addition and presence of the [GAR+] prion. The sequencing technology used measured only relative ratios of organisms, and did not allow us to distinguish between live and dead organisms. Nor could it accurately quantify the populations of bacterial species present in the fermentations. Enterobacteriaceae represents a large family of gram-negative bacteria that most likely enter the winery on the surface of grapes (Nisiotou et al. 2011, Bokulich et al. 2012, 2014). In our samples, this family represented genera like Citrobacter, Erwinia, Pantoea, and others, which are all commonly associated with plant surfaces (Leff and Fierer 2013). The Enterobacteriaceae likely make relatively few stylistic contributions to the final wines. The Lactobacillales order encompasses all types of LABs. Lactobacillaceae was the most significant taxa observed and is represented by the spoilage-associated Lactobacillus and Pediococcus genera (Bartowsky 2009). Treatments without SO2 showed a dramatic increase in the relative abundance in Lactobacillaceae on days 10, 18, and 32. The low relative abundance of these bacteria in the treatments with SO2 suggested that the spoilage-associated LABs grew opportunistically, exploiting the lack of inhibitory SO2.
Because the sequencing technology used was very sensitive, a large number of very-low-abundance taxa were also identified. Steptophyta represents chloroplasts from plant material and is not significant (David et al. 2014). Within the low-abundance taxa, significant differences were identified, but only a few of these differences were of enological significance. Oenococcus was observed in the treatments without SO2, which may help explain why these treatments had undergone partial MLF (Solieri et al. 2010). Gluconobacter was also observed at slightly higher relative abundances in the treatments without SO2, helping to explain the increase in VA in these cultures (Bartowsky et al. 2003). Many of the other low-abundance taxa identified are not known to contribute to the winemaking process.
The results of the analysis of volatile compound composition were consistent with observations of the fermentation kinetics and bacterial community composition. The fermentations with SO2 overall were associated with higher amounts of the fatty acid–derived esters ethyl decanoate and ethyl butanoate. The fermentations without SO2 were associated with an increase in phenylethyl acetate, hexyl acetate, and ethyl hexanoate. The positive association of fatty acid esters in the SO2 fermentations is indicative of the yeast dominance and their highly active metabolism (Saerens et al. 2010). The unsulfited fermentations were strongly correlated with compounds associated with yeast stress and non-Saccharomyces microbial contributions, indicating increased complexity enabled by the lack of SO2.
The [GAR+] fermentations with SO2 were positively correlated with ethyl butanoate and negatively associated with isobutanol. The relative reduction in isobutanol might have been due to overall less flux of amino acids in [GAR+] cells, since isobutanol is produced in small quantities from the valine biosynthetic pathway (Dickinson et al. 1998). Both [GAR+] fermentations displayed relatively elevated amounts of hexanol, indicating an altered metabolic state in [GAR+] yeast. The [GAR+] without SO2 treatment showed high levels of ethyl lactate, indicative of advanced MLF and bacterial metabolism (Ribéreau-Gayon et al. 2006).
Previous work has shown that certain bacterial strains can produce a small molecule that, in a titratable manner, can induce [GAR+] in a yeast population (Jarosz et al. 2014a). The benefits of this induction to the bacterial community are obvious, such as reduced yeast activity resulting in less competition for nutrients and reduced yeast oxygen consumption and ethanol production. However, if induction of this prion conferred no advantage to the yeast, one would expect that yeast strains would evolve to lose the ability to induce the prion under mixed-community growth conditions. We and our collaborators have observed relatively high frequencies of [GAR+] in yeast associated with wine, insects, and fruits (Halfmann et al. 2012, Jarosz et al. 2014b). The much higher frequencies of prion formation in commercial wine strains and vineyard isolates than in laboratory strains suggests that the yeast do benefit from induction of the prion in natural environments. It is unclear which benefits yeast accrue from prion induction, but the ability of yeast to utilize growth substrates more broadly, together with a reduction in ethanol production and glucose consumption, may benefit the long-term survival of the populations. Indeed, we have observed statistically significantly higher viabilities for [GAR+] strains postfermentation than for their [gar−] counterparts. Other physiological benefits are also likely and are the subject of continuing investigations.
Conclusion
Depending on the growth conditions, establishment of the [GAR+] prion by cells of S. cerevisiae negatively affected fermentation progression, encouraged the persistence and diversity of bacterial taxa, and altered the aromatic composition of the wines. The addition of SO2 to nonsterile juice fermentations conducted by [GAR+] strains greatly reduced, but did not fully eliminate, these effects of the prion. The emergence and persistence of wine bacterial communities in fermentations inoculated with [GAR+] strains demonstrated that bacteria can benefit from the presence of this prion in yeast populations, thus explaining the rationale for evolution of inducibility of this yeast metabolic state by the bacteria normally found in the same wine environment. The benefit to S. cerevisiae of [GAR+] induction in wine and grape environments remains to be elucidated, but such elucidation will provide key insights into the microbial community dynamics of grapes and wine fermentations.
Acknowledgments
The authors thank Charles Brenneman, Tyler Burke, Chad Masarweh, Karen Kalanetra, and Morgan Lee for logistical and technical support during this study. The authors also thank Daniel Jarosz, Vidhya Ramakrishnan, and Lucy Joseph for helpful discussions. GAW was supported by the American Wine Society Educational Foundation Scholarship, the American Society of Enology and Viticulture Scholarship, the Henry Jastro Shields Scholarship, the Wine Spectator Scholarship, the Paul Monk Scholarship, and the David E. Gallo Educational Enhancement Fund. Partial funding for AKH was provided by the Mario P. Tribuno Memorial Research Fellowship and the Wine Spectator Scholarship. This work was supported in part by funding from the American Vineyard Foundation and by a grant from the Sloan Foundation.
- Received September 2015.
- Revision received January 2016.
- Accepted February 2016.
- ©2016 by the American Society for Enology and Viticulture
Literature Cited
Vol 67 Issue 3
Jump to section
Related Articles
Cited By...
More from this TOC section
Similar Articles
Advertisement
