Six yeast strains belonging to our laboratory culture collection (Center for Genomic Biotechnology, National Polytechnic Institute, Reynosa, Mexico), obtained from fermenting mezcal must, were cultivated in yeast extract peptone dextrose (YPD) agar (containing in g/L: yeast extract 10, casein peptone 20, agar 20 and glucose 20; Sigma-Aldrich, Merck, Monterrey, Mexico) (4). Yeasts were labelled as: LCBG-Sc3Y4, LCBG-Sc3Y3, LCBG-Sc3Y2, LCBG-Sc3D5, LCBG-Sc3Y5, LCBG-Sc3Y8, and ScFerm (commercial wine strain S. cerevisiae Fermichamp, DSM Food Specialties B.V., Amsterdam, The Netherlands) that was used as control. Before each assay, a loop of each strain was pre-cultured in 250-mL Erlenmeyer flask with 50 mL of YPD broth (Sigma-Aldrich, Merck) for 18 h at 30 °C and 200 rpm in an incubator (Environ Shaker 3527; Lab-Line Instruments, Inc., Melrose Park, IL, USA).

Two culture media were used on yeast extract peptone (YP) plates (in g/L: yeast extract 10, casein peptone 20 and agar 20; Sigma-Aldrich, Merck) with 20 g/L of either glucose (YPD) or fructose (YPF; both Sigma-Aldrich, Merck), to evaluate the influence of the carbon source on the stress tolerance of these yeasts. Selection of thermotolerant yeasts was conducted using a microdrop assay (11). The plates were incubated at 10, 15, 20, 30 and 40 °C, and colony forming units were counted every 24 h until there was a visible growth, varying from 24 up to 216 h at 10 °C. The strains that were able to grow in the whole temperature range in both hexoses were considered to be thermotolerant.

Stress-tolerant yeasts were cultured in YP broth (in g/L: yeast extract 10 and casein peptone 20; Sigma-Aldrich, Merck) supplemented with 100 g/L of either glucose (YPD) or fructose (YPF; both Sigma-Aldrich, Merck). An inoculum concentration of 1.5‧10⁶ cell/mL was used for all the experiments, which were carried out in 50-mL Falcon tubes (Corning Inc., Corning, NY, USA) capped with cotton stoppers and maintained for 60 h without agitation at 30 or 40 °C. Cell population was determined by Neubauer chamber counting at different time intervals. Experiments were performed in triplicate.

Residual reducing sugars in YPD or YPF broth were detected colorimetrically by a modified dinitrosalicylic acid (DNS) method (27) measuring the absorbance at 580 nm in a DU 650 spectrophotometer (Beckman Coulter, Brea, CA, USA). Calibration curves were prepared using glucose or fructose as standard (Sigma-Aldrich, Merck). Production of glycerol, acetic acid and ethanol was determined by HPLC analysis at different time intervals using an HO1200 HPLC equipment (Agilent Technologies, La Jolla, CA, USA) with an Aminex HPX-87 column with a flow of 0.5 mL/min and a mobile phase of H₂SO₄ (5 mM) in Milli-Q water (Millipore, Burlington, MA, USA) for 30 min, and coupled to a refractive index (RI) detector. The samples were centrifuged at 14 400xg for 5 min at 10 °C in a Spectrafuge 24D (Labnet International Inc., Edison, NJ, USA), then diluted (1:10, V/V) in the mobile phase and filtered through nylon membranes (pore size 0.45 μm, 47 mm, GVS Magna™; Thermo Scientific, Mexico City, Mexico). Calibration curves within ranges of 0-20 (ethanol), 0-10 (glycerol) and 0-10 g/L (acetic acid) were set up for commercial standards (Sigma-Aldrich, Merck). Finally, experiments were performed in triplicate.

Cell number was determined spectrophotometrically using a standard curve constructed with correlation of the absorbance measured at 600 nm (DU 650 spectrophotometer; Beckman Coulter) and the number of cells of each strain present in a Neubauer chamber. Each experiment was performed in triplicate. Linear regressions and their fitting were: y=2‧10⁷x, with R²=0.9961, and 2‧10⁷x with R²=0.954 for ScFerm and Sc3Y8, respectively.

Flocculation was determined using the microflocculation method (21). Total cell number was calculated by extrapolation of the absorbance values obtained in each standard cell curve, and the number of flocculent cells was calculated by subtracting the number of cells that remained in the suspension after stopping the agitation, according to Ogata (28). As a negative flocculent control, the assay was conducted using deionized water instead of EDTA solution (Sigma-Aldrich, Merck). Experiments were performed in triplicate.

Flocculation was analyzed by confocal microscopy. Briefly, selected flocculent and non-flocculent strains were grown at 30 and 40 °C as described above. ScFerm was selected as the non-flocculent yeast. Cells were resuspended in citrate buffer (50 mM, pH=4.0; containing 8 mM CaCl₂; Sigma-Aldrich, Merck) and then 0.4% (m/V) Solophenyl Flavine 7GFE 500 (AK Scientific, Inc., Union City, CA, USA) was added. Incubation was conducted at room temperature in darkness. The samples were washed with citrate buffer. Images were visualized and acquired with a confocal microscopy TCS SP5 X (Leica Microsystems, New York, NY, USA) at magnifications 20× and 40× with and without 4.0 digital zoom (confocal microscopy software LAS X v. 2.0.2.14392; Leica Microsystems) at 490 and 510-550 nm for excitation and emission, respectively. Finally, deionized water instead of the citrate buffer served as a negative flocculent control.

Genomic DNA isolation was conducted in S. cerevisiae (Sc3Y8 and ScFerm) cells grown in YPD broth using the DNAzol kit (Molecular Research Center, Cincinnati, OH, USA). DNA amount and quality were determined by using NANODROP 2000 equipment (Thermo Scientific). DNA samples were adjusted at final concentration of 100 ng/μL. For real-time polymerase chain reaction (PCR) amplification of FLO1 and FLO11 genes, the following primer sets were used: FLO1-F 5´-ATGCCTCATCGCTATATGTTTTTG-3´ and FLO1-R 5´-GCTCCTGAGGCCACACTAGTTAG-3, and FLO11-F 5´-CCTCCGAAGGAACTAGCTGTAATT-3´ and FLO11-R 5´- AGTCACATCCAAAGTATACTGCATGAT-3´, which produce fragments of 68 and 103 bp, respectively (23). To amplify the FLO5 gene, a primer set was preliminarily designed for this study (FLO5-F 5’-ATGACAATTGCACACCACTGC-3’ and FLO5-R 5’-ATATATGCTGCATTCGAATATGTGG-3’) from S. cerevisiae S288c (NM_001179342.1) sequence. Then, using this FLO5 sequence, a new primer set was designed (FLO5-F2 5’-GCATCAGGAAGTACGGAAGTCA-3’ and FLO5-R2 5’-TGCTGCATTCGAATATGTGGA-3’), which corresponds to a 110-bp fragment. MEGA v. 4.1 (29) program was used to align and design these primer sets. PCR products were amplified in reaction mix that contained 2 µL MgCl₂ (50 mM; Bioline, Taunton, MA, USA), 2 µL dNTP mix (100 mM; Bioline), 2.5 µL PCR buffer (Bioline), 1.5 µL each primer (50 ng), 0.2 U Taq polymerase (Bioline), and 3 µL DNA template (100 ng). Initial denaturalization time was 5 min at 94 °C, followed by 35 cycles of 1 min at 94 °C, 1 min at 61 °C and 1 min at 72 °C, with a final extension cycle of 10 min at 72 °C. PCR products were verified and purified from an agarose gel (1%) using Wizard SV gel and PCR clean-up kit (Promega, Madison, WI, USA). These PCR products were then bound and cloned in a pGEM-T vector (Promega) and transformed using Escherichia coli DH10B competent cells (Invitrogen, Carlsbad, CA, USA). FLO1, FLO5 and FLO11 gene identity was confirmed by sequencing with an ABI 377 DNA sequencer equipment (Applied Biosystems, Foster City, CA, USA) with universal M13 primers and the BigDye terminator v. 3.1 cycle sequencing kit (Applied Biosystems).

Expression of the FLO1, FLO5 and FLO11 genes was determined on selected (Sc3Y8 and ScFerm) strains grown on YPD broth at 30 and 40 °C. Cells were collected after 12, 24, 48, 60 and 72 h. RNA isolation was carried out using the TRI reagent kit (Molecular Research Center) according to manufacturer’s instructions. RNA (1 µg) was treated with 1 U RQ1 RNase-Free DNase I (Promega), adjusting the volume to 10 µL, and incubated for 30 min at 37 °C. Reaction was stopped by adding 1 µL RQ1 DNase Stop Solution (Promega) followed by an incubation for 10 min at 65 °C. Genomic DNA contamination was verified by using agarose gels at 1.3% and by PCR reaction as follows: RNA treated with DNase was used as negative amplification template for the UBC6 gene. This encodes a constitutive ligase EI protein involved in the ubiquitin-protein degradation (30). Genomic DNA was used as positive amplification control with the primers (UBC6-F 5’-GATACTTGGAATCCTGGCTGGTCTGTCTC-3’ and UBC6-R 5’-AAAGGGTCTTCTGTTTCATCACCTGTATTTGC-3’). PCR products were amplified in a 25-µL reaction containing 2 µL MgCl₂ (50 mM; Bioline), 2 µL dNTP mix (100 mM; Bioline), 2.5 µL PCR buffer (Bioline), 1.5 µL each primer (50 ng), 0.3 µL Taq polymerase (Bioline), and 3 µL genomic DNA template (100 ng) or RNA treated with DNase (50 ng). Initial denaturalization time was 5 min at 94 °C, followed by 35 cycles at 94 °C for 1 min, 1 min at 59 °C and 1 min at 72 °C, with a final extension cycle at 72 °C for 10 min. cDNA synthesis was carried out using 1 µg RNA and the GoScript^TM Reverse Transcription System (Promega). Briefly, a final reaction volume of 20 µL containing 1 μg of DNase I‐treated RNA was mixed with 0.5 μg Oligo(dT)15 and random primers. Then, the reaction was incubated at 70 °C for 5 min, chilled in ice, and an RT Master Mix containing GoScript™ 1× reaction buffer, 5 mM MgCl₂, 0.5 mM each dNTP, 1 U Recombinant RNasin Ribonuclease Inhibitor and 15 U/μg GoScript™ Reverse Transcriptase was added. Reaction was incubated at 25 °C for 5 min, then at 37 °C for 1 h, and finally at 70 °C for 15 min. A kanamycin-positive control of RNA provided in the kit was used to estimate the yield of cDNA synthesis. cDNA amount was verified in a NANODROP equipment (NANODROP-1000; Thermo Scientific) and kept at -60 °C until its use.

Expression of FLO genes was determined by the ΔΔCq method (31) using UBC6 gene as a reference (30). Amplification was standardized in a real-time thermal cycler (GeneAmp PCR-System 9700; Applied Biosystems). Briefly, 50 ng/µL cDNA were used as template in 20 μL PCR reaction mix containing 10 μL PCR Master Mix (2X; Bioline) with SYBR Green (Applied Biosystems), and 0.5 pmol of each forward and reverse primer set. Conditions of amplification were similar to the previous section. In addition, negative control for each gene was included. Fluorescent signal of SYBR Green was standardized using the ROX dye included in the SYBR Green PCR Master Mix (Applied Biosystems).

Gene relative expression (R) was calculated using the following equation:

where E_Ref and E_Target are the efficiencies of the reference and target gene, respectively, and C_tRef and C_tTarget are the threshold cycle (C_t) average values of the reference and target gene, respectively (31). Significant differences (p<0.05) among the mean values were determined by Tukey’s test using one-way ANOVA and StatSoft Statistica software, v. 8.0 (32).

The analysis of yeast growth at different temperatures (10, 15, 20, 30 or 40 °C) showed that the carbon source affected the growth behaviour since all the Saccharomyces strains grew on fructose at all the temperatures (Table 1), but only Sc3Y8 strain was able to grow on glucose at 10 °C. De la Torre-González et al. (11) reported that this Sc3Y8 strain was able to grow at concentrations of fructose up to 850 g/L at 29 °C, and in an ethanol shock test, it was able to survive at 25% ethanol (4). Previously, it had been documented that lower temperatures affect the fructose consumption in Saccharomyces strains (12) and that this sugar may also decrease the osmotic pressure and ethanol stress tolerance level in Saccharomyces and non-Saccharomyces strains (11). The differences in stress tolerance that these strains show are in concordance with the theory of stress cross-response (15). Finally, the Sc3Y8 strain was selected for further studies because of its thermotolerance in the two sugar sources analyzed.

The flocculation capability of the most thermotolerant mezcal strain, Sc3Y8, was further analyzed and compared to an industrial wine strain (ScFem), which is known to ferment grape must sugars to dryness (33). Both selected strains were analyzed in cultures with glucose or fructose (100 g/L), with the aim of comparing the fermentation yields under optimal (30 °C) and suboptimal (40 °C) temperature conditions. In general, all analyzed strains at 30 °C showed an exponential growth up to 24 h and entered the stationary phase at 60 h (Fig. 1). Interestingly, Sc3Y8 strain, which exhibited a wide thermotolerance in solid culture, showed the lowest growth under these conditions. At 40 °C, both Saccharomyces strains had a lower growth than at 30 °C. Regarding sugar consumption, these strains show different profiles since only the ScFerm completely consumed both sugars at 30 °C (after 30 h), but left 50 and 23 g/L of glucose and fructose, respectively, at 40 °C. The Sc3Y8 strain had a high residual sugar concentration of 17 and 31 g/L of glucose and fructose, respectively, at 30 °C, and overall it was less efficient than the industrial strain at this temperature. At 40 °C the residual sugar concentration was similar to the one of ScFerm. Different fructose uptake at different temperatures has been previously reported by Dumont (12), who tested five industrial wine yeast. Fructose and glucose consumption capacity by these industrial strains at increased temperatures was analyzed, with and without nitrogen limitation, and it was observed that glucose was consumed at higher rate at the beginning of the fermentation and fructose at a slower steady rate. Similar results were reported in some mezcal yeasts (4), including the Sc3Y8 selected in this study. Additionally, Dumont et al. (12) observed that at the latter stage of fermentation, the industrial wine yeasts that are considered fructophilic are able to continue uptaking fructose while leaving a higher residual glucose concentration in the medium. The same authors documented that the higher the temperature, the better the fructose consumption by all except one of the tested strains (a low temperature specialist, as stated by the authors), and it is worth noticing that their higher temperature (28 °C) tested is similar to the optimal (30 °C) temperature used in our study. Hence, the results shown in this work concerning an enhanced fructose consumption capacity at high temperatures are indeed one of the most interesting phenotypical characteristics of this mezcal strain. It is clear that more effort must be made to clarify this different kinetic behaviour in relation to fructose in this strain.

Concerning the production of metabolites, at the end of cultivation (60 h), glycerol, ethanol and acetic acid concentrations increased in all selected strains when grown on fructose at 40 °C (Table 2). Sc3Y8 strain showed higher metabolic accumulation at this temperature. These different fermentation profiles as a consequence of the increase in the temperature and the type of used sugar could be due to their different ethanol tolerance and their differences in the hexose transporter efficiencies (15, 33). Interestingly, glycerol was the metabolite with the higher increase (>7 times) at 40 than at 30 °C (Table 2). It has been reported that temperature affects the glycerol accumulation in S. cerevisiae strains (8) since it acts as an osmoregulator between the cytosol and the environment (34). Similar to the accumulation of glycerol, the alcohol and acetic acid levels also increased (2 to 3 times) when these selected strains were grown at 40 °C using fructose as a carbon source, particularly Sc3Y8 strain. In general, the use of fructose at 40 °C increased the metabolite production by the yeasts, similar to what had been previously reported for some industrial S. cerevisiae strains (12). It is worth noticing that under stressful (temperature, osmotic pressure, etc.) conditions, Saccharomyces strains tend to overproduce glycerol and acetic acid, and to produce less biomass, as corroborated with the colony count in Table 1, i.e. the number of viable cells that grew in glucose or fructose at the two highest temperatures tested for all the strains. This general trend can be observed for both hexoses when comparing metabolite profiles (Table 2) at 30 and 40 °C. These results suggest that the two analyzed strains have a different metabolic response at 40 °C, with the mezcal yeast (Sc3Y8) producing more ethanol at this temperature, probably due to its original niche of isolation, which is a fermentation that is carried out at high fructose concentration and uncontrolled temperature.

Flocculation has been determined in some yeast species, mainly in S. cerevisiae, as a mechanism to overcome stress conditions (i.e. alcoholic or nutrient depletion) (22, 35). It was observed only in Sc3Y8 strain (Fig. 2). This phenomenon was CaCl₂-dependent since it was not observed in the presence of water or buffer without calcium (data not shown). Confocal microscopy analysis of Sc3Y8 strain using the fluorophore Solophenyl Flavine 7GFE 500 (AK Scientific, Inc.) allowed to assess some changes in the cell wall conformation at 40 °C, similarly to the observations made by Soares (19), using the fluorophore Calcofluor White M2R in cells that were subject to osmotic stress (1 mol/L NaCl) or a brief heat-shock (52 °C for 5 min). In this study, the temperature stress was maintained during fermentation. The fluorophore Solophenyl Flavine 7GFE 500 has not been previously used in flocculation analysis so it was first tested to determine its incubation time, concentration and viable cell number. The fluorophore did not show adverse effects on the cell viability during or after flocculation assay, so it was suitable for cell observations. It was determined that the best mass per volume ratio of Solophenyl Flavine was 0.4% at an incubation time of 15 min. Observations at 30 °C in the absence of CaCl₂ evidenced oval structures with a higher fluorescent intensity, likely the scars generated during budding (Fig. 3). In contrast, at 40 °C, the fluorophore recognized the whole cell wall and also allowed to observe cell lysis (Fig 3a). To confirm that the fluorescence detected at 40 °C is a consequence of the cell wall recognition by the Solophenyl Flavine and not their inclusion in the lysed cells, a viability assay was conducted in parallel. Fig. 3b and Fig. 3c compare the fluorescent analysis and the viability determination (plate count). In general, a reduction in the viability at 40 °C was observed although Sc3Y8 strain was slightly more viable than ScFerm (non-flocculent strain used as a control). Regarding fluorescent analysis, similar behaviour was observed between both S. cerevisiae strains at 30 °C, but these values increased significantly in the ScFerm strain at 40 °C. Considering the results at 40 °C, the increase in the fluorescence might be a consequence of the reduction in the viability and of the cellular lysis detected in the ScFerm strain, in contrast to the more temperature-tolerant Sc3Y8 strain, which exhibited a lower reduction of their fluorescence even at 40 °C.

Confocal microscopy analysis using the Solophenyl Flavine in the presence of CaCl₂ showed cell aggregates only in the Sc3Y8 strain. These cell aggregates showed apparent different sizes depending on the temperature conditions (data not shown). Finally, at 40 °C, less lysed cells were observed in the Sc3Y8 than in ScFerm strain, supporting the previous viability assay. These results suggested that the flocculation phenotype in the Sc3Y8 strain may contribute to its thermotolerance due to the changes in the cell wall conformation and integrity, allowing a higher viability. Also, it has been reported that temperature affects the flocculation phenomenon at the level of cellular interactions, probably as a consequence of the flocculin denaturalization (19). Flocculation observed in Sc3Y8 cells slightly increased when the temperature was raised from 30 to 40 °C, although it was not statistically significant. Additionally, the floccule size in the Sc3Y8 strain is apparently bigger at 40 than at 30 °C. It has been reported that small changes in the cell wall hydrophobicity can also affect the floccule size (22). The differences found in the staining pattern caused by the fluorophore Solophenyl Flavine are not directly correlated with the cell viability (36). Nevertheless, it seems to be related to changes in the cell envelope caused by the increase of temperature. Recently, a transcriptomic relationship between the stress response and the flocculation was identified in a wine S. cerevisiae (F6789A) strain after deletion of the FLO5 gene. These genes regulate stress tolerance and are related with cell wall reorganization and adhesion (26).

With the aim to better understand the flocculation phenomenon in the Sc3Y8 strain, a FLO gene expression analysis was conducted. Genetic regulation, specifically involving lectins, has been related to physiological and epigenetic changes that in turn might be involved in cell wall synthesis (15). Gene fragments of the selected FLO1, FLO5 and FLO11 genes were successfully amplified generating sequence fragments of 68, 193 and 103 bp, respectively (Table S1). FLO1, FLO5 and FLO11 transcript accumulation was analyzed in these Saccharomyces strains on YPD broth at 30 and 40 °C. Results showed that FLO1 gene transcripts were constant at 30 °C in both Saccharomyces strains. However, at 40 °C a slight but significant (Tukey's test; p<0.05) induction (6%) was observed after 12 h, particularly in the Sc3Y8 strain. ScFerm also exhibited a reduced induction (3.8%) at 60 h of culture (Fig. 4). A stronger FLO5 expression in Sc3Y8 strain was induced at 40 (up to 20-fold) than at 30 °C. This transcript induction was statistically different among strains after 12 h, and had a maximal value of 175% at 48 h in the Sc3Y8 strain, which coincides with the beginning of the stationary phase (Fig. 1). The non-flocculent strain (ScFerm) kept constant its FLO5 gene expression at both temperatures.

Our results for the flocculent Sc3Y8 strain suggest a transcriptional induction, at least of FLO1 and FLO5 genes, as a consequence of the temperature increase. Expression of both genes was constant in non-flocculent strain ScFerm, with only a slight increase at 40 °C. FLO5 gene transcription in Sc3Y8 strain strongly increased at both temperatures, but it was the highest at 40 °C. Di Gianvito et al. (26) have recently demonstrated the importance of the FLO5 gene during flocculation and related it to stressful (ethanol at 20%) conditions in different Saccharomyces wine strains (10). Recently, this gene has been deleted in a S. cerevisiae wine strain (F6789A-Δ flo5) causing an up-regulation of genes related to assembly, shape and cell adhesion, and genes related to the organization of the cell wall, among others (26). The transcription of FLO1 gene has also been related to flocculation ability in Saccharomyces strains, although with a lower transcriptional activation of FLO5 gene (10), in line with our observations. The predictive protein products of FLO1 and FLO5 are 96% similar and both are subtelomeric genes but they are not alleles (37). The regulated expression of these FLO1 and FLO5 genes leads to specific phenotype intensities, resulting in cell aggregation, cell surface properties, and flocculation (10, 23). Finally, the expression profile of the FLO11 gene showed that at 30 ° C, there was a slightly higher statistically significant difference in strain Sc3Y8 at 24 and 48 h (increase in expression of 2 and 3.2%, respectively). At 40 °C, there was a constantly low expression of FLO11 in both strains. This gene has been related to a wide phenotype variety, such as flocculation, the invasive growth and flor formation (38), probably depending on the culture conditions. Further studies using other cell aggregation or cell-surface adherence genes as AMN1 and FLO8 must be conducted to clarify the relation between the flocculation and the stress response, specifically when these cell properties are involved (39).