Introduction

Ethanol tolerance has been correlated with membrane fluidity in yeasts, although such correlations can differ from one strain to another, and depend upon intrinsic properties of the particular yeast strain investigated. In many studies, membrane fluidity has been inferred indirectly from measurement of the unsaturation index (UI) of the yeast membrane fatty acids; higher UI has been taken to represent higher membrane fluidity (Fajardo et al. 2011; Henderson and Block 2014). Such studies have indicated that increased UI relates to improved ethanol tolerance (Henderson and Block 2014). Introduction into S. cerevisiae of a gene for unsaturated fatty acid synthesis (FAD2) combined with overexpression of the mono-desaturase OLE1 reportedly increased the UI and improved tolerance to exposure to 15% ethanol (Kajiwara et al. 2000; You et al. 2003).

However, membrane fluidity is not only influenced by fatty acid composition, but also other component(s) of the yeast plasma membrane, such as proteins and sterols (Alexandre et al. 1994b; Learmonth 2012). Therefore, the reports inferring fluidity from lipid composition need to be confirmed via more direct determination of membrane fluidity. Several studies have shown exposure of S. cerevisiae to high concentrations of ethanol results in increased membrane fluidity (Alexandre et al. 1994a; Learmonth 2012; Learmonth and Gratton 2002; Lloyd et al. 1993). Lloyd et al. (1993) reported that in S. cerevisiae grown in the presence of up to 9% (v/v) ethanol (as measured by EPR) mitochondrial membrane fluidity was similar, while microsomal membrane fluidity was markedly increased at 9% ethanol. Alexandre et al. (1994b) measured membrane fluidity by DPH anisotropy in addition to membrane lipid composition and concluded that higher membrane fluidity (with underlying higher oleic acid proportion and UI and lower sterol/lipid and sterol/protein ratios) was related to higher ethanol tolerance, as determined by growth in media containing ethanol. This conclusion was drawn from measurement of two different yeast species, S. cerevisiae (grown at 10% v/v ethanol) and Kloeckera apiculata (grown at 4% v/v ethanol). Learmonth and coworkers (summarized in Learmonth 2012) measured membrane fluidity either by DPH anisotropy or laurdan generalized polarization, and found that upon exposing S. cerevisiae strains to 20% (v/v) ethanol, membrane fluidity increased dramatically, followed by variable recovery. In addition, they found that the respiratory culture phase, characterized by lower membrane fluidity, correlated with reduced fluidization and higher ethanol tolerance. Further assessment of strains from the same yeast species could reveal different behaviors. Therefore, in the present study, we compared membrane fluidity of several strains of S. cerevisiae.

Methods

Microorganisms and fermentation conditions

Three S. cerevisiae strains were used in the present study, selected on the basis their high ethanol productivity and range of ethanol tolerances. The strains included the baker’s yeast A12 (Lewis et al. 1997), the wine yeast PDM (Mauri Yeast), and the sake yeast K7 (ATCC6422). The yeasts were grown in the defined medium Yeast Nitrogen Base (YNB) containing 1.5% glucose and 0.67% YNB (Difco). Starter cultures were inoculated from slopes and grown overnight (~ 16 h) at 30 °C and 180 opm in an orbital shaker (Paton). The ratio of flask size to culture volume was 4:1 to ensure adequate oxygen mixing. Starter cultures were used to inoculate experimental cultures to give an initial viable cell number of 106 cells/mL. The yeasts were grown at 30 °C in an orbital shaker (180 opm) for 30 h, and samples were aseptically removed by drawing off with a sterile pipette at inoculation and at 6 h intervals.

Growth parameter assessment

Yeast growth was monitored by measuring optical density of the culture at 600 nm (OD600 nm) using a Beckman DU 650 spectrophotometer, making dilutions where necessary. Viable cell numbers were assessed by light microscopy (×400 magnification) with the aid of a Neubauer-type haemocytometer, using the methylene violet staining method. Samples were prepared by mixing equal volumes of sample and methylene violet solution (0.01% w/v in 2% sodium citrate solution) (Smart et al. 1999).

Membrane fluidity assessment

Membrane fluidity was assessed using steady state fluorescence spectroscopy, measuring generalized polarization (GP) of 6-dodecanoyl-2-dimethylaminonaphthalene (laurdan) following incorporation of the probe into yeast plasma membranes, as outlined by Learmonth (2012). For labelling, an aliquot of washed cells was standardized by diluting with centrifuged (8800×g) supernatant to an OD600 nm of 0.4. Laurdan (2.5 mM in ethanol) was added to the standardized washed cell sample to give final concentration of 5 µM, the samples were incubated for 60 min at 30 °C, in the dark with stirring.

Fluorescence emission intensities were measured on an ISS PC1 spectrofluorometer (ISS Inc. Champaign, USA) at 440 and 490 nm with excitation at 340 nm, with 8 nm slits for emission and excitation. Unlabeled cell suspension was used to measure background fluorescence which was subtracted from the labeled fluorescence readings.

Results were expressed as generalized polarization (GP) determined using Eq. 1.

$${\text{GP}}=\frac{{{{\text{I}}_{440\,{\text{nm}}}} - {{\text{I}}_{490\,{\text{nm}}}}}}{{{{\text{I}}_{440\,{\text{nm}}}}+{{\text{I}}_{490\,{\text{nm}}}}}}$$
(1)

Ethanol tolerance assessment

The concentration and time exposure to ethanol was based on the work of Chi and Arneborg (2000) and Lewis (1993) with slight modification. Samples of 1 mL of culture were centrifuged at 1500×g for 2–3 min, the supernatant growth medium decanted, and the pellet resuspended in the original volume of yeast nitrogen base without glucose (YNBNG). A 0.41 mL sample was added to a tube containing 0.09 mL of absolute ethanol and mixed immediately, exposing the cells to 18% v/v ethanol. The tube was incubated at 30 °C for 60 min. Survival was determined using two methods; counting of methylene violet stained suspensions and total plate count.

Methylene violet staining method

For the methylene violet staining method, the stress was relieved by making a fivefold (for 6 h culture) or tenfold dilution (for 24 h culture) in MilliQ water. The percentage of viable cells was calculated by dividing the number of viable cell by total cell number to determine the proportion viable. The result of this viability calculation was then treated as follows, referred as “viability reduction by methylene violet staining” (VR MVS), calculated using the Eq. 2.

$${\text{VR MVS}}\,(\% ) = \frac{{{\text{Control viable proportion}} - {\text{test viable proportion}}}}{{{\text{Control viable proportion}}}} \times 100\%$$
(2)

This method of calculation provides much more precise information on the effects of conditions on viability, as compared to simply determining and comparing % viabilities.

Total plate count method

After the stress period the samples were diluted using YNBNG to give serial decimal dilutions. Then, 100 µL of the diluted samples was then spread on YEP agar plates, and incubated for about 72 h at room temperature before counting the resultant colonies. The results were also expressed as “viability reduction by total plate count” (VR TPC), calculated using Eq. 3.

$${\text{VR }}{\text{TPC}}\,(\%) = \frac{{{\text{Total viable cell control}} - {\text{total viable cell test}}}}{{{\text{Total viable cell control}}}} \times 100\%$$
(3)

Measurement of glucose and ethanol

The amount of glucose and ethanol was determined by measuring the compounds in the growth media compared against the initial concentration at the beginning of fermentation (0 h) using HPLC. The HPLC system (Shimadzu) consisted of SIL-20A auto sampler, DGU 20A5 in-line degasser, LC-20AD solvent delivery module, CTO 20A column oven, RID 10A refractive index detector, and Class-VP software. The apparatus was connected to a PC with CBM 20A communication bus module. A Waters Sugar-Pak I HPLC column (part no. 85118) with dimension of 6.5 × 300 mm was used for separation of analytes. The mobile phase was deionized MiliQ water (resistivity ~ 18 Mohm) containing 5 mg/L CaNa2–EDTA (Sigma Aldrich) filtered through 0.45 µm filter. The mobile phase was passed through an in-line degasser to ensure that the mobile phase was gas free. The flow rate was maintained at 0.6 mL/min and the column temperature was maintained at 80 ºC. Prior to the initial use, and after running about 150 samples, the column was reconditioned by passing through a 500 mg/L CaNa2–EDTA solution at 0.5 mL/min flow rate and 80 °C for at least 2 h in the reverse direction.

Statistical analysis

Raw data was initially compiled into Minitab 15® for Windows®. This software package was then used to perform one way analysis of variance (one way ANOVA) which compared the variance of each parameter (i.e. GP, viability reduction) between the three strains with the variability within each replicate experiment of each strain. When samples were taken at successive time points, repeated measures analysis were applied for the one way ANOVA. When significant (p < 0.05) differences were detected in the one way ANOVA, the test was followed by Tukey post hoc test to determine which data differed significantly.

Results

Growth parameter comparisons

Optical density at 600 nm was used to measure cell density, allowing comparison of growth of the three strains used in this experiment (Fig. 1). Generally, during the lag (0–6 h) and respiro-fermentative (6–12 h) phases the OD600 nm values were similar, the only significant difference was found at the 6-h time point where A12 has significantly higher OD600 nm compared to PDM (p = 0.028). After 12 h, A12 and PDM started to enter respiratory phase, whereas K7 was still in the respiro-fermentative phase at this time point. At 18 h, K7 started to enter its respiratory phase, and interestingly, from this point onwards, OD600 nm value for K7 was significantly higher than the other strains (p = 0.003). At 18 h, K7 had the highest OD600 nm value followed by PDM and A12. The cell density of K7 remained higher at 24 h (p < 0.001) and 30 h (p < 0.001). This data indicates the different growth patterns of the three yeast strains used in this experiment.

Fig. 1
figure 1

Optical densities of three yeast strains during fermentation. Cultures of the strains indicated were grown in YNB medium with 2% (w/v) glucose as carbon source under aerobic conditions at 30 °C. Data are the means of four independent experiments. Error bars represent standard deviation

Full size image

In addition to optical densities, cell density was also assessed by light microscopy with counting. The cells were examined under a light microscope after staining by methylene violet solution. From this method total cell counts, viable cell counts, cell viability percentage, and budding rates can be determined. Figure 2 shows the cell counts of the three yeasts during culture. Unlike the OD600 nm data, there were no significant differences in total cells measured by light microscopy, likely due to the higher variability of this data. Likewise, there were no statistically significant differences in budding rate nor total viable cell counts. However, the total cell counts evidenced similar trends to the OD600 nm data; that the cells were in lag phase from 0 to 6 h, respiro-fermentative phase from 6 to 12 h and afterwards in respiratory growth phase.

Fig. 2
figure 2

Comparison of total cell number (a), viable cell number (b), budding rate (c) and cell viability (d) determined by light microscopy for three different yeast strains, A12 (), PDM () and K7 (). Cultures were grown in YNB medium with 2% (w/v) glucose as carbon source under aerobic conditions at 30 °C. Data are the means of four independent experiments. Error bars represent standard deviation

Full size image

Budding rate data indicated a tendency to decrease during the lag phase (0–6 h). At 12 h the budding rate was increased and stayed more or less the same throughout the fermentation. No significant differences were detected for these parameters in any of the strains used in the present experiment.

Interestingly, cell viability data showed some significant differences between the strains. At the beginning of fermentation, K7 started with significantly higher viability compared to the others (p = 0.022). At 6 h, all the strains reached their highest cell viability values, although no significant differences were observed at this time point. At the 12 h time point, cell viability of A12 and PDM were drastically reduced, while K7 maintained its viability significantly higher (p < 0.001); this higher viability was maintained throughout the remaining culture time.

Glucose uptake and ethanol production rates of all strains did not show any significant differences (Fig. 3). Generally, glucose was exhausted between 12 and 18 h. Even though not statistically significant, there was a trend in which ethanol was rapidly produced after 6 h, and reaching its highest concentration at 18 h. The highest ethanol concentration at 18 h was recorded for PDM with ethanol concentration of 0.75 ± 0.24% (v/v), however this concentration was not significantly different to the other strains. After glucose was exhausted by 18 h, the ethanol concentration started to decrease due to respiratory growth.

Fig. 3
figure 3

Glucose utilization (G) and ethanol production (E) by three different yeast strains, A12 (), PDM () and K7 (). Cultures of the yeast strain indicated were grown in YNB medium with 2% (w/v) glucose as carbon source under aerobic conditions at 30 °C. Data are the means of four independent experiments. Error bars represent standard deviation

Full size image

Membrane fluidity comparisons

Membrane fluidity was assessed by generalized polarization (GP) of laurdan fluorescence emission (Parasassi et al. 1998), applied to yeast cell plasma membrane by Learmonth and Gratton (2002) and Butcher (2008). This method is simpler and more robust compared to other methods for determining membrane fluidity such as polarization and anisotropy measurements utilized in previous publications.

Figure 4 shows GP values of the three yeast strains. Significant differences between GP values are marked by bars that do not share the same letters. Statistical analysis using one way ANOVA showed that at during respiro-fermentative growth, GP values differed significantly for the yeast strains tested. The GP value was significantly higher for A12, while no significant difference was detected between K7 and PDM. This indicated that A12 had the lowest membrane fluidity, followed by K7 and PDM. Unlike the 6 h results, the respiratory 24 h cultures did not show any significant differences of GP values between strains tested. While the membrane fluidity of all three strains were similar during respiratory growth, comparison with respiro-fermentative phase cultures showed that the GP of each strain was significantly higher, i.e. the membrane fluidity was lower in respiratory phase (p < 0.001). Similar results were observed in previous studies by Learmonth and Gratton (2002), Butcher (2008) and Learmonth (2012).

Fig. 4
figure 4

Generalized polarization of the three yeast strains during respiro-fermentative (6 h) and respiratory (24 h) growth. Cultures of the yeast strain indicated were grown under aerobic conditions in YNB medium with 2% glucose as carbon source. Bars that do not share same letter are significantly different statistically according to one-way ANOVA and Tukey post hoc test. Error bars represent standard deviation

Full size image

Ethanol tolerance comparisons

Figure 5 shows viability reduction upon ethanol stress tested using TPC. Comparisons between strains revealed no significant differences in viability reduction during either respiro-fermentative (6 h) or respiratory (24 h) cultures, while for each strain the respiratory phase cultures evidenced significantly less viability reduction. Interestingly, even though not statistically significant, the viability reduction of PDM was slightly increased compared to the other strains at 24 h, and according to the statistical testing, did not differ significantly with A12 at 6 h. This result was confirmed by methylene violet staining, and will be discussed further.

Fig. 5
figure 5

Viability reduction induced by ethanol stress as determined by the total plate count (TPC) method. Cells of the yeast strains indicated were grown in YNB medium with 2% glucose under aerobic conditions at 30 °C until the indicated time points. Then, they were exposed to 18% v/v ethanol for 60 min, and subsequently diluted and grown on agar plates. Bars that do not share same letter are significantly different statistically according to one-way ANOVA and Tukey post hoc test. Error bars represent standard deviation

Full size image

The TPC test showed that the viability reduction by ethanol stress was significantly decreased (p < 0.05) from 6 to 24 h culture for all strain tested. This result suggests that ethanol tolerance was increased in respiratory phase cells as compared to respiro-fermentative cells, for all strains. However, the TPC test failed to distinguish any differences in ethanol tolerance between the strains either at 6 or 24 h of culture.

Methylene violet staining was used as an alternative method for determining ethanol tolerance (Fig. 6). While PDM tended towards higher values, there was no significant difference between the strains in viability reduction in either respiro-fermentative (6 h) or respiratory (24 h) growth. The tendency of PDM to higher viability reduction indicates its lower ethanol tolerance compared to the other strains. During respiratory growth, the viability reduction of PDM was significantly higher than for K7, but not A12. However, unlike the TPC method, MVS failed to detect any differences of viability reduction between 6 and 24 h culture, likely due to high variability in the counts.

Fig. 6
figure 6

Viability reduction induced by ethanol stress as determined by the methylene violet staining method. Cells of the yeast strains indicated were grown in YNB medium with 2% glucose under anaerobic conditions at 30 °C until the indicated time points. Then they were exposed to 18% v/v ethanol. Cells were counted under microscope (×400 magnification) after exposure to 18% v/v ethanol and staining using methylene violet in sodium citrate solution. Bars that do not share same letter are significantly different statistically according to one-way ANOVA and Tukey post hoc test. Error bars represent standard deviation

Full size image

Correlation between test parameters for different yeast strains

Correlation between parameters tested was evaluated using the Pearson correlation coefficient (Table 1). Correlation analysis for the overall data showed that there is a very strong positive correlation between time and GP value (r = 0.899; p < 0.001). The same result was also observed for individual strains with r > 0.900 and p < 0.001. This result confirmed the findings above and showed that the GP value increases with increasing culture age, which indicates that membrane fluidity of all strains is lower in cells from respiratory growth phase.

Table 1 Pearson correlation coefficients between time, GP, viability reduction by TPC and viability reduction by methylene violet staining for all strains
Full size table

Strong negative correlations were observed between viability reduction by TPC and time (r = ~ 0.8; p < 0.05) or GP (r = 0.65–0.79; p < 0.05) either for the overall data or individual strains, which showed that viability reduction by TPC decreases with increasing time and GP value. Since viability reduction is inversely related to ethanol tolerance, these results suggested that ethanol tolerance of all yeast strains was higher in respiratory phase cells, which also have lower intrinsic membrane fluidity.

Viability reduction by methylene violet staining was poorly correlated either with culture time, GP or viability reduction by TPC. A weak correlation (r ~ 0.500) for viability reduction by methylene violet staining with other parameters was observed for the A12 strain but not for the other strains. Significant but weak correlation was also observed for viability reduction by methylene violet staining with viability reduction by TPC for overall strain analysis (r = 0.370; p = 0.044).

Discussion

Comparison of growth parameters

Both the OD and total cell counts demonstrated that during the first 6 h, all yeast strains were in their lag phase, followed by respiro-fermentative growth. Interestingly, the OD data showed that under our conditions K7 had a longer respiro-fermentative phase of up to 18 h, compared to 12 h for A12 and PDM. Starting at the 18 h time point, the OD for K7 was significantly higher than the other strains, whereas before that, the OD for K7 was relatively lower than the others. This indicates that initially K7 grows more slowly than the other strains, but in the end it has the highest cell number.

While OD data indicated significant differences between the strains, no significant differences were found between strains for total cell counts. The total cell count data showed very high variability. This might be because the total cell number was measured via counting under a microscope, with counting being associated with intrinsically higher error rates. Another possibility is uneven distribution of the cells during sampling.

The cell viability data were interesting. The data showed that viability of A12 and PDM strains increased during the lag phase (first 6 h) of fermentation. For these strains the highest cell viability was recorded at 6 h and after that time the viability declined rapidly to relatively constant levels at about 75% of the 6 h data. Interestingly, cell viability of K7 was very high throughout the fermentation, never < 91%. Thus, K7 had better cell survival than the other strains, which is likely to be an advantage over the other strains. If this property is also manifest when the yeast grows in higher sugar concentration, would prolong the availability of viable cells to ferment the sugars and therefore increase the efficiency of fermentation. Therefore, of the strains trialed K7 seemed to have the most potential for further exploration in production of ethanol for biofuels or other purposes.

Notwithstanding, there were no significant differences between the yeast strains tested in glucose uptake or ethanol production during the fermentation. The highest ethanol concentration was detected at 18 h of fermentation, and ethanol started to decrease after that time point. Previous studies (Lewis et al. 1993b) showed that the decrease in ethanol was due to metabolic activity and not simply due to evaporation. As S. cerevisiae belongs to the Crabtee-positive yeast group (Piškur et al. 2006), after glucose depletion to about 0.2% (close to 18 h in our experiments), the yeast can begin to use ethanol as a carbon source, in the presence of oxygen.

Membrane fluidity and ethanol tolerance

Previous published works showed that the stress tolerance of yeast cell is related to the fluidity of the yeast cell plasma membrane. Alexandre et al. (1994a, b) compared membrane fluidity by measuring DPH anisotropy of two different yeast species with different degrees of ethanol tolerance, and found that yeast species with higher membrane fluidity had higher ethanol tolerance. Modification of the cell plasma membrane by incorporating more unsaturated fatty acids, thereby changing the unsaturation index of the plasma membrane has also been found to increase yeast tolerance against high ethanol concentrations (Kajiwara et al. 2000; You et al. 2003), freezing and also salt stress (Rodríguez-Vargas et al. 2007). Dinh et al. (2008) also found that yeast cells that developed tolerance against high ethanol concentration have a higher composition of unsaturated fatty acids, particularly oleic acid (C18:1).

Recent work by Dong et al. (2015) also found that as aerobic fermentations progressed through respiro-fermentative and respiratory/stationary phase, yeast membrane remodeling led to increases in unsaturated fatty acids, especially oleic acid. This change was found to be related with better ethanol tolerance. Similar results were found by Wang et al. (2015), who describe that more tolerant yeasts had higher unsaturated fatty acid composition. However as noted elsewhere, while fatty acid composition may influence membrane stability and fluidity, other factors combine to regulate membrane fluidity, which is not necessarily higher in more unsaturated membranes.

Swan and Watson (1999) found that yeasts enriched in with oleic acid (C18:1) were more tolerant to ethanol and heat than those enriched with linoleic (C18:2) or linolenic acid (C18:3). These experiments followed on from previous experiments where heat and oxidative stress tolerance were studied in aerobic and anaerobic lipid-supplemented cells (Steels et al. 1994). In the latter experiments, unsupplemented anaerobic cells with largely saturated fatty acid profiles were more resistant to heat stress than C18:1 or C18:3 supplemented cells. These experiments suggested that increasing fluidity by introducing more unsaturation to fatty acyl chains leads to more susceptible cells. Although direct measurement of membrane fluidity was not reported in these studies, they support a counter view that decreased membrane fluidity correlates with higher ethanol tolerance. Notwithstanding that lipid unsaturation does not necessarily correlate with increased membrane fluidity (Learmonth 2012), measurements of yeast membrane fluidity by DPH anisotropy or laurdan generalized polarization (Learmonth 2012) have also supported the intuitive proposition that ethanol tolerance is inversely related to membrane fluidity.

In this study ethanol tolerance of the yeast cell was monitored using two different methods, total plate count (TPC) and methylene violet staining (MVS) after exposing the cell to 18% ethanol. The total plate count is widely used for ethanol tolerance determination of yeasts (Chi et al. 1999; Krause et al. 2007). Methylene violet staining is proposed as a better method for monitoring yeast cell viability compared to the traditional methylene blue staining method (Smart et al. 1999), and offers the advantage that viability can be assessed much more rapidly. As these two methods can be used to distinguish live and dead cells, we tried to apply the methylene violet staining to calculate viability reduction of ethanol stressed cells, and investigate whether offers advantages in ethanol tolerance test.

Our experiments demonstrated that for all three yeast strains, the membrane fluidity was significantly lower in respiratory phase cells (24 h culture) compared to respiro-fermentative cells (6 h). Considering assessment of viability reduction by methylene violet, we found that at 6 h culture, A12 which had significantly higher GP value than PDM showed relatively lower viability reduction. The same trend was seen for K7, although the higher GP was not significant. This indicates that lower membrane fluidity correlates with higher tolerance against ethanol, which acts by fluidizing membranes. However, viability reduction by TPC failed to distinguish ethanol tolerance of the strains at 6 h culture. Similar phenomena were also observed for the 24 h cultures, in which PDM, which tended to have the lowest GP, tended to have higher viability reduction, either determined by total plate count or methylene violet staining. However, for the 24 h cultures the differences were not statistically significant.

The results of the present study contradict most of the previously published works which indicated that higher membrane fluidity leads to higher ethanol tolerance (Alexandre et al. 1994b; Dinh et al. 2008; You et al. 2003). The previous studies used different approaches to determine membrane fluidity of yeasts. They mainly measured increasing unsaturated fatty acid, expressed as the unsaturation index (Alexandre et al. 1994b; Dinh et al. 2008; Rodríguez-Vargas et al. 2007; You et al. 2003) or direct measurement of membrane fluidity by measuring DPH anisotropy (Alexandre et al. 1994a). Studies which simply infer fluidity from lipid unsaturation data do not account for the impacts of the many other factors that can modulate membrane fluidity (Learmonth 2012; Learmonth and Gratton 2002).

Alexandre et al. (1994b) also stressed that membrane fluidity deduced from unsaturation index does not always reflect the real condition of the membrane fluidity, therefore they used DPH anisotropy to directly measure membrane fluidity. They reported that ethanol tolerance is connected with higher membrane fluidity. However, their data showed that S. cerevisiae which had higher ethanol tolerance did not always show higher membrane fluidity compared to K. apiculata which had lower ethanol tolerance. For example, when the cells were grown in the presence of ethanol and further exposed to 12% (v/v) ethanol, K. apiculata showed a lower anisotropy value (0.133 ± 0.002) compared to S. cerevisiae (0.138 ± 0.003), which indicates that K. apiculata had higher membrane fluidity, while this yeast is more susceptible to high ethanol concentration. This finding is consistent with the results of this study and also with a hypothesis that lower intrinsic membrane fluidity helps to minimize the fluidizing impact of high ethanol concentrations.

While Alexandre et al. (1994b) used anisotropy of the fluorescent probe DPH, the present work used a different method of membrane fluidity determination. The two approaches have been demonstrated to provide similar conclusions (Learmonth 2012). We used generalized polarization of laurdan for determining membrane fluidity. Parasassi et al. (1990) explained that laurdan is very sensitive to the polarity of the environment, and further investigation by the same group also revealed that the spectrum is red shifted with decreasing membrane fluidity (Parasassi et al. 1998). The different methods used by Alexandre et al. (1994b) and the present experiment may lead to different results, although in general, the techniques report consistent trends.

Another thing to note is that the GP significantly increased in transition from respiro-fermentative to respiratory growth phase, for all strains. Thus membrane fluidity is relatively higher in respiro-fermentative cells which are also less tolerant to a range of stresses (Learmonth 2012; Learmonth and Gratton 2002; Lewis et al. 1993a, b). The decreased membrane fluidity related to increased ethanol tolerance, in agreement with previous observations (Learmonth 2012). Similar phenomena were also observed by Swan and Watson (1999) and Steels et al. (1994) which suggested that ethanol tolerance is related to lower membrane fluidity. In this study correlation analysis for all parameters tested provides further evidence that there are strong correlations between culture time, viability reduction by TPC and GP. The correlations with time are essentially reporting the difference between respiro-fermentative (6 h) and respiratory (24 h) cells.

Thus we believe there is substantial evidence to support the view that decreased membrane fluidity results in higher ethanol tolerance. Ethanol acts as a polar solvent, partitioning into the membrane and increasing their fluidity. Therefore cells with a lower intrinsic membrane fluidity can tolerate higher concentrations of ethanol before fluidization reaches lethal levels. Furthermore in previous studies of the response to ethanol stress (summarized in Learmonth 2012), adaptation to ethanol and partial recovery of membrane fluidity were shown. The modulation of membrane fluidity may be achieved by a combination of factors including changes in proportion of lipid classes (e.g. relative amounts of phosphatidyl choline and phosphatidyl ethanolamine) fatty acid chain length and unsaturation, and changing membrane protein profiles. Ethanol affects membrane fluidity, stability and permeability, leading to loss the proton gradients across the plasma membrane. Other effects of ethanol include inhibition of enzyme and transport protein activity, thus in particular, levels and function of the plasma membrane H+ ATPase and associated proteins such as Hsp30 are important in adaptation to ethanol stress.

Poor correlation was detected for viability reduction by methylene violet with other parameters. However, as described earlier, this technique succeeded in distinguishing ethanol tolerance of different yeast cells at a particular time point (6 or 24 h), which the TPC method failed to achieve. In contrast, viability reduction by TPC can distinguish ethanol tolerance of yeast cells from different time points, which methylene violet staining failed to distinguish. Therefore, each method has its own limitations and advantage, which should be considered when selecting a method for determination of ethanol tolerance. Generally, from our observations, we could use methylene violet staining when we intend to discriminate ethanol tolerance of different yeast strains at a particular time point, and the TPC method for better discriminating responses a particular yeast strain at different time points during culture.

The lack of correlation between viability assessment by methylene violet staining and total plate count is particularly troubling. It is likely that the high variability observed in the cell counts led to the lack of correlation, and sampling techniques could perhaps be further refined. In addition, the methods report different phenomena, in that methylene violet staining may report live cells that have lost the ability to replicate, while only cells able to divide will be counted after growth on plates. Additionally, the methylene violet approach was established to provide a more reliable assessment of viability than methylene blue staining, although it may suffer from a similar problem in assessment of viability of highly stressed cell populations.

Conclusions

This study provides a comparison of the membrane fluidity and ethanol tolerance of different yeast strains. The three yeast strains in this experiment evidenced different responses, highlighting the strain dependence of adaptive responses. Unlike most previously published studies, this study concludes that yeast strains with lower membrane fluidity have higher ethanol tolerance. Correlation analysis showed strong correlations between growth phase (respiro-fermentative, 6 h, vs. respiratory, 24 h), membrane fluidity and ethanol tolerance. Even though viability reduction by methylene violet staining did not show any strong correlation with the other parameters measured, it may be useful in some circumstances. Finally, among three strains used in the present experiment, A12 and K7 evidenced better ethanol tolerance, while growth parameter comparison revealed that K7 had better performance compared to the other strains. Therefore, of the strains tested, this strain has the most potential for further exploration.