All yeast strains used were obtained from the Department of Viticulture and Enology culture collection. Commercial isolates were purified individual colonies isolated from packets of commercial yeast that were then entered into the collection. Strains used were: UCD932 (Ba2 Lambrusco, Italian vineyard isolate) (Mortimer et al., 1994), UCD777 (isolate of commercial strain EC1118), laboratory strain W303 (MATa/MATα leu2-3,112 trp1-1, can1-100, ade2-1, ura3-1, his3-11,15), laboratory strain Y55 (MATa/MATα), DBVPG1106 (grape isolate, Australia), DBVPG 6040 (fermenting juice isolate, Netherlands). Yeast were maintained as frozen stocks in 50% glycerol solution and propagated on YPD [Yeast extract, peptone, glucose (2%)] medium. The bacterial strains used were obtained from the Department of Viticulture and Enology culture collection and are listed in Table 1. The source of the isolates is also listed. Isolates from natural sources were restreaked to purity and subjected to DNA sequence analysis for unequivocal identification prior to being entered into the collection. In one experiment direct commercial preparations of dried bacteria were used: Viniflora and VP41 as indicted in the text.

Bacteria from stuck fermentations were selected by centrifuging 1 mL of wine and resuspending the pellet in 100 μL of wine. This suspension was used to streak for isolation on ½ strength MRS (Difco) (1:1 with water) plates with 25 mg/L of Delvocid (DSM). The bacteria were re-streaked for isolated colonies and separate colonies were selected for identification from each sample. Bacteria were identified either using sequence comparisons of 16s rDNA or via MALDI-TOF using the Bruker Biotyper system according to the manufacturers specifications and matching to the custom database (Sogawa et al., 2011). In cases of unclear identification via MALDI, strains were subjected to 16s rDNA sequencing. DNA from the bacteria was isolated using the Promega Wizard genomic DNA kit (Fischer Scientific). Isolates were identified by PCR of the 16s ribosomal DNA using the 27F and 1492R primers (Frank et al., 2008). PCR was conducted in 50 μL reactions with 1 l of template DNA. Thermocycler conditions were an initial 1 min at 95°C followed by 35 cycles of 45 s at 95°C, 30 s at 56°C, 2 min at 72°C, and a final extension at 72°C for 8 min. PCR products were cleaned with a Promega Wizard or Qiagen PCR clean up kit. Samples were sequenced at the UC Davis DBS sequencing facility and results run on the NIH BLAST database to determine the closest match for taxonomic identification

The presence or absence of the [GAR⁺] phenotype was assessed on GGM medium as described (Brown and Lindquist, 2009). GGM agar contains (per liter): 10 g yeast extract (Becton, Dickinson and Company, United States), 20 g peptone (Becton, Dickinson and Company, United States), 25 mL of sterile 80% glycerol stock solution (Fisher Scientific, United State), 10 mL of filter sterilized 5% glucosamine stock solution (Acros Organics, United State), and 20 g granulated agar (Apex Bioresearch Products, United States). Briefly, yeast were pregrown to an absorbance of 0.4 A₅₈₀ in YPD and serially diluted five times at 1:5 for a total final dilution of 1:3125 and 3 μL of each dilution plated on GGM agar. Cell concentrations in the final dilution spot were approximately 15–20 cells per 3 μL. Plates were incubated at 30°C and evaluated daily for 7 days. Control platings on fully permissive YPD as well as YP glycerol plates were conducted for a qualitative assessment of growth on permissive and prion inducing media. All growth tests were run in biological triplicates.

Bacterial strains were pre-grown in ½strength MRS broth (Difco) at 30°C to stationary phase. UCD932, UCD777, and W303 yeast strains were pre-grown in YP media (1% Yeast extract and 2% Bacto peptone) with 2% glucose to log phase on a roller drum at room temperature. Yeast cells were harvested at absorbance of 0.4 at A₅₈₀, washed twice with sterile water and resuspended at approximately 3–4 × 10⁶ cells/mL to be used for dilutions and plating. Bacterial impact on yeast growth on YP media with 2% glycerol and 0.05% glucosamine was measured by plating 3 μL of 1:5 serial dilutions of yeast such that the final dilution contains ~20 cells. Bacteria were grown aerobically in test tubes on a roller drum and diluted to an OD₅₈₀ of 2.0 and 3 μL plated across all spots at a consistent distance from the diluted yeast cultures. Growth and induction of [GAR⁺] was assessed after 5 days of incubation at 30°C. All tests were run in triplicate. To assess crossfeeding, control plates of YP no carbon source and YP glycerol, no glucosamine were also run for each sample assessing the level of growth nearest the bacteria as compared to that in the center of the plates. Colonies appearing near the bacteria were isolated, passaged on non-selective media, and then replated on GGM to assess growth phenotype and retention of the ability to grow on GGM in the absence of the bacteria. There were cases where non-wine ecosystem bacteria did lead to either cross-feeding or to apparent degradation of the glucosamine present and “induced” yeast did not retain the prion phenotype following passage on non-selective media, but those organisms were not included in this study as none originated from the wine ecosystem, but served to assure that these controls would identify cases of growth without prion induction.

Bacterial strains were grown in YP broth (1% Yeast extract and 2% bactopeptone) with 2% glycerol and 0.05% glucosamine (GGM media) at 30°C. The media with bacteria were clarified by centrifuging and filtered through a 0.22 μM Millex-GV syringe filters (Millipore SLGV033RS) to obtain spent media. UCD932 was pre-grown in YP broth with 2% glucose and cells were harvested at an absorbance of 0.4 at A₅₈₀ and washed with sterile water and inoculated into the spent media at approximately 6.67 × 10⁵ cells. Samples were taken after 4 h and 1:5 serial dilutions were plated on YP plates with 2% glycerol and 0.05% glucosamine and on YP plates with 2% glucose for dilution controls. Growth and [GAR⁺] induction were assessed after incubation at 30°C for 5 days. Samples were run in triplicate.

For the liquid induction assays, a culture of wild type [gar⁻] strain of UCD932 cells were grown to an OD₅₈₀ of 0.6–0.8 in YPD, washed, and diluted to an OD₅₈₀ of 0.4 in GGM with acetic acid at the concentrations indicated and incubated at room temperature on roller drum for 4 h. OD₅₈₀ was measured again after incubation and cultures were diluted back to OD₅₈₀ 0.4 to make five-fold serial dilution series. Serial dilutions were then plated on GGM agar as described above. Samples were run in triplicate.

Yeast cultures in log phase (about 10⁴ cells/mL) were spread plated on GGM media. Assay discs were saturated with 10 μl of different concentrations of acetic acid (0, 0.5, 1, 1.5, 2, 3, 4%) or lactic acid (0, 1, 2, 3%) or pyruvic acid (0, 0.5, 1, 2%) which corresponds to 0, 50, 100, 150, 200, 300, or 400 μg of acetic acid, 0, 100, 200, or 300 μg of lactic acid, and 0, 50, 100, or 200 μg of pyruvic acid respectively per spot. These were placed on each quadrant of the petri dish. Plates were incubated at 30°C and growth pattern was monitored for 2–5 days. Samples were run a minimum of four times.

UCD 26 (Lactobacillus kunkeii), UCD 175 (Acetobacter pasturianus), UCD215 (Acetobacter pasteurianus) were grown in ½ strength MRS media, 0.45 μM filtered Chardonnay juice, YP (1% Yeast extract and 2% bactopeptone) media with different carbon sources including 2% glucose, 2% fructose, 1% each of glucose and fructose and 2% glycerol, to stationary phase, then centrifuged and filtered through a 0.22 μm Millex-GV syringe filters (Millipore SLGV033RS). Acetic acid and lactic acid levels were measured by Anion analysis kit using Beckman Coulter ProteomeLab PA800 capillary electrophoresis system. Biological replicates were run for each strain per condition and technical replicates were done for each analysis.

Pre-inocula of UCD932 without and with the [GAR⁺] prion were grown to stationary phase (48 h) in 10 mL sterile filtered Chardonnay juice on a roller drum at room temperature. Fermentations were carried out in triplicates in filtered Chardonnay juice with starting inoculum of approximately 3.3 × 10⁵ cells. Flasks were incubated with agitation at 120 rpm at 28°C. Fermentation progress was monitored by measuring weight loss due to carbon dioxide release. Samples were taken routinely through the fermentation for 10 days, clarified by centrifugation and filtered through 0.22 μM Millex-GV syringe filters (Millipore SLGV033RS) prior to storing at 4°C for amino acid analysis. Amino acid analysis was performed by the UC Davis Proteomics Core (University of California, Davis).

Samples of arrested wine fermentations were directly plated onto WLN medium (Difco) and S. cerevisiae colonies identified by colony morphology (Pallmann et al., 2001) and confirmed by cell morphology. Up to 8 isolated colonies were selected per wine sample and each colony inoculated into one well of a 96 well plate containing 100 μL of YPD broth and grown for 24 h with shaking at room temperature. A multipronged replica plater was used to transfer liquid cultures from the 96 well plates to GGM and YPD media plates and incubated at 30°C. Growth of the spots was assessed at 1 day for YPD and after 5 days for GGM. Growth on the GGM plates was compared to [GAR⁺] and [gar⁻] control strains: UCD932 [GAR⁺], UCD932 [gar⁻], UCD777 [GAR⁺], and UCD777 [gar⁻]. Isolates showing similar growth to the [GAR⁺] stains were selected as presumptive [GAR⁺] isolates.

Curing of the [GAR⁺] prion was adapted from Tapia and Koshland (2014). Isolated colonies were selected from the YPD plate of the screening process and inoculated into 200 μL of YPD broth using one isolate per well of a 96 well plate. Cells were grown for 24 h with shaking at room temperature. The absorbance of the wells were taken using a plate reader at A₅₈₀. Cell suspensions were washed by spinning the plate in a centrifuge and extracting excess water without disturbing cell pellet. Two hundred microliters of water was added, the cells resuspended and recentrifuged in the plate twice. Cell pellets were then resuspended in 100 μL of dilute phosphate buffered saline (1.25 mM PO43-, 17.125 mM NaCl, and 0.3375 mM KCl). Ten microliters of each isolate were added to 100 μL of sterile Milli Q water to be tip plated for viability count on both YPD and GGM. After day 1 for YPD and day 5 for GGM the number of isolated colonies bigger than 0.3 mM diameter were counted to obtain initial total viability and population penetrance of the [GAR⁺] prion levels. The remainder of the cells in the 96 well plate were placed in an area with approximately 60% relative humidity at 23°C for 1 month with the top cover slightly open by holding up the corners of the plate with stiff paper. After 1 month cells in the wells were rehydrated in 100 μL of dilute phosphate buffer saline and tip plated for viability after desiccation on both YPD and GGM.

Single colonies of UCD932 [gar⁻] and [GAR⁺] were picked from YPD plates, and inoculated into 5 mL of 2% YPD liquid and left to grown overnight on a rotary drum at room temperature (~25°C). Overnight cultures were inoculated into fresh 5 mL 2% YPD liquid cultures and allowed to reach mid-late exponential phase (~0.8 A₅₈₀). Cells were washed in sterile DI H₂O and concentrated to 1.2 A₅₈₀. A 96 well plate was filled with 250 μL of YP liquid base containing the corresponding percentage of added carbon stock solution. Washed cells were inoculated into wells at 0.05 A₅₈₀ using a Rannin LTS multichannel pipet (Mettler Toledo, Oakland, CA). A Breath Easy® sealing membrane was placed on the 96 well plate (Sigma-Aldrich, St. Louis, MO). The plate was then placed in a Synergy 2 BioTek plate reader, incubating at 30°C sampling every 30 min preceded by 40 s of variable shaking (BioTek, Winooski, VT). Samples were run in 8 biological replicates. Data represent the average of 8 independent wells.

Extensive analysis of one yeast strain vineyard isolate, UCD932, demonstrated that this yeast can induce the [GAR⁺] phenotype which displays all of the hallmark traits of a prion (Walker et al., 2016). Therefore, the ability of a broader array of wine strains to induce the prion was assessed. To evaluate prion induciblity of wine strains, cells were plated on glycerol glucosamine medium and on control media lacking glucosamine. All strains evaluated grew well on the basal medium containing glucose (YPD) as sole carbon and energy source which also served to quantify the number of viable cells in the original culture based upon number of colonies formed at the two greatest dilutions (Figure 1A). The second control medium was YP glycerol to assess ability to grow on glycerol as a carbon and energy source, Glycerol is in general a poorer energy source than glucose at 2% (both substrates were present at this concentration). One of the strains tested, S288C, was not able to grow well on the medium with glycerol as sole carbon and energy source without addition of glucosamine (YPG, Figure 1B), but all other strains tested were able to grow on this medium. Strains induce the [GAR⁺] prion at different frequencies on glycerol glucosamine medium (GGM) as seen in Figure 1C. This medium contains the same level of glycerol as Figure 1B but in addition contains the glucose mimetic glucosamine a non-metabolizable analog of glucose that induces glucose repression. Juice isolate strain DBVPG6040 showed the best induction of the prion with growth to the terminal dilution plated. Two other wine strains, UCD932 (grape isolate) and EC1118 (UCD777, commercial isolate) also showed growth at the terminal dilution on GGM plates (inoculated with [gar⁻] starting cultures) in previously published work (Walker et al., 2016). The two laboratory strains, Y55 and W303 displayed moderate induction on the GGM medium. Strain DBVPG1106, also a vineyard isolate, showed weak induction of the [GAR⁺] phenotype.

A hallmark of prion induction is the heritability of the associated phenotype. Colonies of UCD932 and EC1118 obtained from the terminal dilutions on GGM were grown on YPD for several generations and then replated onto GGM and compared to the parental strain that had not previously been plated on GGM (Figure 2), That the [GAR⁺] derivatives show full growth upon replating as compared to [gar⁻] parental strains after passage on non-selective media indicates that the ability to grow on alternative carbon sources in the presence of inducers of glucose repression represents a heritable change within the population.

Induction of the [GAR⁺] prion has been associated with a decrease in expression of the major hexose transporter, Hxt3, present during the active phase of wine fermentations (Brown and Lindquist, 2009). This modification of glucose utilization slows fermentation progression in juices such as Chardonnay consistent with a decrease in sugar uptake capacity (Walker et al., 2016). This observation was repeatable using Chardonnay from a different vintage (Supplemental Figure 1). Under these pure culture conditions the [GAR⁺] yeast are fully capable of completing the fermentation, catabolizing all available sugar, but the onset of fermentation is delayed and the fermentation curve differs in shape from that of the wildtype strain consistent with the observed impact of the prion on induction of the Hxt3 transporter.

Previous work demonstrated that bacteria in co-cultivation with S. cerevisiae on GGM medium could accelerate the induction of the [GAR⁺] prion (Jarosz et al., 2014a). A wide array of bacteria were able to induce on this medium (Jarosz et al., 2014a). However, many of these bacteria are not found naturally in the same ecosystems as S. cerevisiae. We therefore screened our collection of bacterial isolates from grapes, musts and wines to determine if native ecosystem organisms could also induce the formation of the prion. For this analysis strains were categorized as having one of three effects: (1) induction as evidenced by enhanced growth on GGM plates in the yeast columns nearest to the bacteria, (2) inhibition as evidenced by the absence of growth of yeast columns nearest to the bacteria, and (3) no effect, characterized by yeast growth near the bacteria matching that of columns in the middle of the plate (Figure 3). Inhibition can be observed for the acetic acid bacterium Acetobacter malorum (Figure 3A) and induction for the acetic acid bacterium Gluconobacter cerinus (Figure 3A) and for two commercial lactic acid bacterium preparations of Oenoccus oeni (Figure 3B). The commercial preparations were pregrown directly from the packets and purity of the commercial preparations was not assessed. Both of the commercial preparations showed induction (Figure 3B) in contrast to pure cultures of O. oeni (Table 1).

All three classes of impacts on yeast growth were observed among the bacteria evaluated with, in some cases, all members of a species showing the identical effect and in others the effect varying by strain within a species (Table 1). This analysis was conducted using yeast strain UCD932. Two bacterial classes, the acetic acid bacteria and the lactic acid bacteria, are commonly found on the surfaces of grapes and may persist in the wine. Many of these organisms are associated with arrested or slow fermentations. In general the majority of the species assessed in the genus Acetobacter were inhibitory on GGM media. Of the 19 strains tested in this genus, 11 were inhibitory (58%), 6 showed no effect and 2 (10%) were inducing. The two strains showing induction were isolates of A. pasteurianus one of which was isolated from wine. Strains of multiple species of Gluconobacter, also classified as an acetic acid bacterium, were evaluated. Of the 14 strains from 4 species, 8 (57%) were clearly inducing and 5 demonstrated no effect. One strain showed a variable response meaning that some yeast strains were induced and others showed no effect (strain 514). Two species showed strain variability: two G. cerevisiae isolates were inducing and the third showed no effect. Similarly for G. oxydans 4 of the 6 strains evaluated were neutral and two were inducing with one isolate showing a strain effect in induction (514). All isolates from the other two species, G. cerinus and G. frateurii were inducing. None of the Gluconobacter strains isolated from wines was inhibitory of yeast growth on the GGM plates. Gluconacetobacter is rarely isolated from grapes or wine but the ability of type strains to induce the prion was evaluated (Table 1). One species, Gluconacetobacter liquefaciens induced and the other, Gluconacetobacter hansenii showed no effect.

Three genera of lactic acid bacteria, Oenococcus, Lactobacillus, and Pediococcus, are commonly found on grapes and in wines and can be associated with arrest of fermentation. All but one isolate of Pediococcus, P. parvulus 258, were capable of induction of the [GAR⁺] prion. The two Oenococcus species evaluated also were unable to induce the prion in contrast to the two commercial preparations of Oenococcus used in Figure 3. Nine species of Lactobacillus were analyzed: L. brevis, L. casei, L. curvatus, L. hilgardii, L. kunkeei, L. lactis, L. mali, L. plantarum, and L. sakei, all of which have been isolated from grapes and/or wine. The L. sakei strain, which came from an arrested wine fermentation, was inhibitory similar to the acetic acid bacteria. None of the isolates of L. brevis and L. hilgardii showed an effect. All isolates tested of L. casei, L. curvatus, L kunkeei, L. lactis, and L mali were able to induce the prion state. Three of the five isolates of L. plantarum showed induction of the prion and the other two isolates showed no effect. Interestingly the 3 strains showing an effect were all isolated from arrested wine fermentations. Thus, although the ability to induce prion formation is widespread it is not universal among isolates of the same species.

Other genera of bacteria can be isolated from wines although much less frequently. Isolates of Bacillus megaterium and B. gingsengijumi obtained from wines were both able to induce the prion as were isolates of Staphylococcus. One isolate of Paenibacillus from wine was also able to induce the prion. Thus, the ability to induce the prion spans the spectrum of bacterial isolates obtained from wines.

The spot plate assay and diffusible nature of the inducer leaves open the possibility that what is being observed is not prion induction but cross-feeding of substrate-limited colonies of yeast. Inducing bacteria make end products such as pyruvate, lactate, and acetic acid that could serve as growth substrates. To differentiate between cross-feeding and prion induction two types of experiments were conducted. Given the heritable nature of prion induction, simple restreaking of the growth adjacent to the bacterial isolates on GGM agar would lead either to growth in the absence of the bacteria if induction had occurred or to a lack of growth in the absence of the bacteria if cross-feeding were responsible for the observations of growth on the initial plate. In all cases growth on GGM occurred in the absence of bacteria. A colony of UCD932 induced by Acetobacter pastuerianus (UCD 175) was obtained from a GGM plate adjacent to spots of the inducing bacterium A. pasteurianus. The colony was restreaked onto the non-selective medium, YPD, and retested on GGM. The retested strain grew well at all dilutions on the GGM medium as compared to the original [gar⁻] strain (Figure 4).

The second type of experiment performed was to conduct the induction during liquid growth in spent medium. A. pasteurianus UCD175 was grown in liquid GGM media to an absorbance of 2 at 580 nm, removed by filtration and the yeast incubated in the spent medium for a period of 4 h. Following incubation yeast cells were collected and washed twice then serial dilutions plated onto GGM. Incubation of UCD932 in the spent medium (top panel, Figure 5) resulted in growth on GGM to the terminal dilution. Incubation in liquid GGM alone or in synthetic must Triple M (Spiropoulos et al., 2000) did not result in better growth on GGM plates (Figure 5). Washing of the cells did not impact ability to grow on GGM indicating that a heritable change had been induced in the yeast during the incubation period in the spent medium. Thus, Saccharomyces ecosystem bacteria induce the prion state in this yeast via diffusible factors.

The bacteria listed in Table 1 as isolated from wine were isolated from arrested or protracted wine fermentations and several (20) were found to be capable of inducing the [GAR⁺] prion as were bacteria isolated from grape must (7 isolates). However, in order to understand the ecological impact of the phenomenon of bacterial induction it is important to determine if it can occur during typical batch fermentation of grape juice. Arrested wine samples were sought for the 2014 vintage and used to isolate both bacteria and yeast. These samples were solicited as simply being arrested fermentations and bacterial inhibition represents only one class of arrest (Bisson, 1999). Our goal was to assess these samples for the presence of inhibitory or inducing bacteria and to isolate viable yeast if possible to assess the presence of the [GAR⁺] phenotype in these yeasts. Samples of arrested fermentations were received from 40 wineries from which viable yeast were isolated. Not all of the wines had viable yeast and some had been reinoculated with commercial strains which could potentially interfere in the analysis for the prion but all were included in the survey for prion induction. Of the 127 wines received 59 yielded isolatable viable yeast cells on WLN medium and 1–8 colonies were selected from each plate for a total of 400 yeast isolates analyzed. Seventy-one of the 400 isolates from 28 wines showed viable yeast capable of growth on GGM media matching that of known [GAR⁺] controls (Figure 6). Thus, of the original 127 wines 22% contained yeast populations with evidence of induction of the [GAR⁺] prion phenotype. Only one of the commercial wine samples containing [GAR⁺] yeast also contained a viable inducing bacterial species, L. mali suggesting that if induction were solely mediated by bacteria in the natural environment those bacteria may become non-viable or non-culturable later in the fermentation.

The spot plate assay detects growth and it was possible that colonies showing growth on this medium in the replica plating assay may derive from original colonies that are mixtures of [GAR⁺] and [gar⁻] cells. Therefore, all the original colonies of the 71 isolates showing good growth on GGM were re-assessed using tip-plating directly comparing the numbers of cells growing on GGM to the permissive yeast medium YPD. Upon quantitative replating 20 of the original populations growing as [GAR⁺] displayed growth on GGM at a population density greater than 1% with the highest density being 52%. These 20 strains were then tested to confirm that the growth on GGM was due to the induction of [GAR⁺].

Desiccation of yeast cells has been shown to “cure” the [GAR⁺] prion state due to loss of the prion conformation and recreation of the wildtype conformational form upon regrowth of the desiccated cells (Tapia and Koshland, 2014). To determine if the 20 isolates showing good growth on GGM were indeed expressing the [GAR⁺] prion vs. a genomic mutation all 20 isolates were subjected to desiccation for 30 days then media added and cell growth assessed on GGM and control non-selective medium YPD. Seventeen of the 20 strains tested showed loss of the ability to grow on GGM consistent with curing of the prion state with curable isolates identified from each of the 13 wines (Table 2). The remaining three strains showed complete loss of viability following the desiccation process and therefore curing percentage could not be evaluated. However, this analysis showed that the [GAR⁺] prion can be induced in wines during fermentation demonstrating that prion induction can occur in a native yeast ecosystem. The curable isolates obtained represented 7 of the original 40 wineries.

To test the hypothesis that prion-induced yeast would be less efficient at depletion of nitrogen from the environment, depletion of amino acid from grape juice was compared in [GAR⁺] vs. [gar⁻] derivatives of the same yeast strain, UCD932. The majority of the amino acids were depleted in fermentations with the [gar⁻] cells within the first 24 h (Table 3). At the 48 h time sample low residual levels of ammonia and gamma-amino butyric acid were present. Higher levels of proline persisted but proline utilization requires the presence of molecular oxygen and these fermentations although mixed were oxygen limited in this strain. In contrast, in the [GAR⁺] strain depletion lagged significantly behind [gar⁻] strains with the pattern of consumption at 72 h matching that of the non-induced strain at 48 h (Table 3). This is consistent with the slower start of fermentation in the prion-induced strain in this same Chardonnay juice (Supplemental Figure 1) as the initial lag in onset of fermentation is roughly 24 h longer for the [GAR⁺] strain. Thus, when grown in pure culture the induction of the prion impacts fermentation initiation but does not affect the ability of the cells to completely ferment available sugar. Interestingly tryptophan levels increased in the medium for both strains suggesting tryptophan synthesized within the cells is released during fermentation and persists in the medium.

These studies were conducted using pure cultures of the yeast. However, a decreased rate in nitrogen depletion by the yeast in the presence of bacterial competitors would enhance the ability of the bacteria to proliferate during the early stage of fermentation. We have seen a similar slowdown in early oxygen consumption in [GAR⁺] yeast strains (Walker, to be reported). Therefore, the benefit to the bacteria from induction of the prion is obvious: they are able to slow yeast metabolism thereby reducing the rate of nutrient and oxygen depletion as well as potentially the production of inhibitory metabolites enabling the bacteria selecting for those bacteria able to proliferate and obtain energy under these oxygen limiting conditions and adapt to the anaerobic conditions of the fermentation. These findings are consistent with analyses of the microbiome of [GAR⁺] and [gar⁻] yeast in the presence and absence of the antimicrobial agent sulfur dioxide (Walker et al., 2016).

The above studies indicate that bacterial induction is mediated by a diffusible factor produced during the growth of the bacterium in GGM. A clue to the nature of the induction process came from analysis of the cross-feeding hypothesis as the mechanism of growth on GGM. The major acetic acid and lactic acid bacteria metabolites are pyruvate, DL-lactic acid, and acetic acid. To test cross-feeding various concentrations of these acids were spotted onto a lawn of UCD932 yeast on GGM (Figure 7A) at varying concentrations. The spot plates with pyruvate and lactic acid showed no impact on surrounding yeast growth for any of the concentrations evaluated. However, some of the acetic acid spots did show a halo of growth on the plates. This was especially evident for moderate levels of acetic acid. At lower levels no stimulation of growth on GGM occurred. To determine if this was cross feeding or induction, colonies from the growth halo around dots with 200 and 400 μg acetic acid levels were selected, washed, and replated onto GGM (Figure 7B). Yeast around the 200 μg acetic acid dot show evidence of prion induction. Interestingly, the 400 μg sample shows better growth than the no acetic acid treatment control but less than the 200 μg halo.

We next asked if acetic acid alone was sufficient for induction on GGM. To address this question a series of experiments were performed adding acetic acid at varying concentrations to liquid GGM and incubating yeast for 4 h, then washing the cells and plating on GGM (Figure 8). It is evident that even with this treatment at moderate levels of acetic acid growth of the serial dilution culture on GGM extends further than for cells exposed only to GGM during the 4 h incubation period (compare the most dilute sample where roughly 10 colonies are visible for the terminal dilution for the 1 and 2 g/l incubations but not the no acetic acid control). The 3 g/L treatment shows fewer colonies in the final dilution as compared to the 1 and 2 g/L samples. This suggests that induction is reduced at higher acetic acid levels in the environment. There was no difference in growth on YPD, YP glycerol, or YP no carbon source following acetic acid treatment as compared to non-treated cells. A difference in growth was only observable in the presence of the glucose mimetic glucosamine. In order to observe growth on GGM the acetic acid exposure had to occur in the presence of the growth substrate glycerol. Thus, acetic acid appears to be able to induce the prion in the presence of an utilizable alternative growth substrate although less effectively than the induction seen in spent medium.

Two of the strongest inducing bacteria, Lactobacillus kunkeei and A. pasteurianus (strain 175) were grown under a variety of growth conditions and the levels of acetic and lactic acid measured in comparison to a strain of A. pasteurianus that is inhibitory (215) (Table 4). Under all conditions L. kunkeei made higher levels of both acids than either A. pasteurianus strain. L. kunkeei displayed high production of both acids in Chardonnay grape juice. The low acetic acid production in the Acetobacter strains indicates as expected that other energy substrates are being used by these strains under these growth conditions yielding end products other than acetic acid. Oxidation of ethanol is the primary source of acetic acid in the acetic acid bacteria so the observations of low acetic acid production under these inducing conditions is expected. Inoculation with both L. kunkeei and A. pasteurianus 175 yielded similar to slightly higher levels of production for both acids suggesting production is slightly higher than simply additive.

The induction of the [GAR⁺] prion slows yeast metabolism but also enables use of alternative carbon sources in the presence of glucose. If there were no benefit to the yeast of induction of the prion it would seem logical that wine strains would evolve mechanisms to prevent its formation thereby assuring dominance of mixed culture fermentative environments. The retention of the ability to induce the [GAR⁺] prion in the presence of bacteria suggests the yeast gain some benefit from doing so. Use of an alternative carbon source when continued use of glucose would be toxic is a sound metabolic strategy. Since acetic acid can enter the cell if protonated by simple diffusion its presence could place the cells under cytoplasmic acidification stress and the ability to utilize glucose and fructose fermentatively which also places cells under acidification stress, may be compromised by the presence of acetic acid. In this scenario the reduction in fermentation is to balance the forces leading to cytoplasmic acidification. Spent media and acetic acid induction experiments conducted in control media lacking an alternative substrate did not lead to induction of the prion suggesting that the presence of a usable alternative substrate is important for the establishment of the prion.

Grape juice is an equimolar mixture of glucose and fructose with total sugar levels on the order of ~25% w/v (Boulton et al., 1996). Low concentrations of sucrose, another fermentable sugar, are present but other alternative carbon sources readily utilized by yeast are in short supply. However, both the lactic acid bacteria and acetic acid bacteria can make reduced or oxidized versions of sugars in juice that may represent alternative carbon sources for the yeast. L. kunkeei is a fructophilic lactic acid bacterium able to balance cofactors during metabolism by reducing fructose to mannitol at quite significant levels (Bisson et al., 2016b). Thus, when proliferating in grape juice both acetic acid and mannitol will be present. Similarly the acetic acid bacteria obtain energy from partial oxidation reactions and can oxidize both glucose and fructose again at significant levels providing the yeast with gluconic acid and its further oxidized derivatives as well as oxy-fructose and derivatives (Drysdale and Fleet, 1989; Barbe et al., 2001). An important question was to determine if the induction of the [GAR⁺] prion enabled utilization of these substrates.

Growth on various substrates of isogenic [GAR⁺] and [gar⁻] strains of UCD932 were compared (Figure 9). As expected growth in the presence of glucose (YPD) or in the absence of an added carbon source was identical in the two strains (Figure 9A). Galactose is a substrate showing glucose repression (Carlson, 1987) but use of this substrate is quickly induced on low glucose and the two strain show similar usage of this substrate with the [GAR⁺] isolate displaying a shorter lag for transit to use of this compound (Figure 9A). The growth of strains was compared in substrates subjected to glucose repression, melibiose, and raffinose (a trisaccharide yielding melibiose) (Neigeborn and Carlson, 1984; Carlson, 1987) and growth on both melibiose and raffinose was better for the [GAR⁺] isolate. Growth on glycerol, also a non-preferred substrate, was likewise better for the prion-induced strain (Figure 9B). Growth on oxidized (gluconic acid) and reduced (mannitol and sorbitol) substrates of glucose and fructose were also analyzed. In all three cases the [GAR⁺] strain showed better growth on these substrates (Figure 9C). Induction of the [GAR⁺] prion therefore enables ready growth on substrates normally subjected to glucose repression including the oxidized and reduced forms of glucose and fructose that can be produced by acetic acid and lactic acid bacteria during growth in grape must.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.