Intracellular metabolic changes in Saccharomyces cerevisiae and promotion of ethanol tolerance during the bioethanol fermentation process

Abstract

During the batch bioethanol fermentation process, although Saccharomyces cerevisiae cells are challenged by accumulated ethanol, our previous work showed that the ethanol tolerance of S. cerevisiae increased as fermentation time increased. However, the exact molecular mechanisms underlying the increased ethanol tolerance of S. cerevisiae are still poorly understood. In this study, a gas chromatography-mass spectrometry-based metabolomics strategy was used to determine the fermentation process-associated intracellular metabolic changes in S. cerevisiae cells. With the aid of partial least squares-discriminant analysis between two of the three fermentation stages (i.e., the lag, exponential, and stationary phases), 40 differential metabolites with variable importance and a projection value greater than 1 were identified. During the bioethanol fermentation process, S. cerevisiae cells could continuously remodel their membrane composition and structure to obtain higher ethanol tolerance. During the lag–exponential phase transition, in spite of a down-regulated TCA cycle, the increased ergosterol content combined with decreased saturated fatty acid content might be the most significant factor in making yeast cells more robust and ethanol-tolerant. During the exponential–stationary phase transition, a re-activated TCA cycle could provide plenty of energy, and the increased energy production together with the increased energy requirements might be partly responsible for the increased ethanol tolerance in the stationary phase. Moreover, the increased content of glycerol, trehalose, ergosterol and some amino acids also might jointly confer the yeast cells with higher ethanol tolerance. These results highlighted our knowledge about the relationship between the bioethanol fermentation process and ethanol tolerance, and could contribute to the construction of feasible ethanologenic strains with higher ethanol tolerance.

---

Introduction

The energy crisis and environmental pollution issues caused by the over-use of traditional fossil fuels (i.e., global warming and regional disasters)¹ promote the demand for alternative renewable energy sources.² Due to its renewability and low pollution, bioethanol is considered to be a good alternative to fossil fuels to ease the energy crisis and contribute to alleviating environmental pollution.³ Saccharomyces cerevisiae is the most used microbe for the industrial production of bioethanol.^3,4 However, while ethanol is the end product of fermentation, the accumulated ethanol itself can also function as a major inhibitor for S. cerevisiae cell growth.⁵ During the bioethanol fermentation process, the accumulated ethanol affects yeast cell viability, thereby contributing to the inhibition of complete bioethanol fermentation^6,7 and decreasing bioethanol production efficiency. However, interestingly, our previous results indicated that although ethanol gradually accumulated and could continuously decrease S. cerevisiae cell viability, the ethanol tolerance of S. cerevisiae cells was continuously increased as bioethanol fermentation time increased.⁸

During the batch fermentation process, S. cerevisiae cells undergo different stages (i.e., lag, exponential, and stationary phases),⁶ and yeast cells at different phases encounter different fermentative environments, including a changed ethanol content. For example, during alcoholic fermentation, S. cerevisiae mainly face hyperosmotic stress (180–260 g l⁻¹ sugar concentration) at early stages, and subsequently accumulated ethanol and the depletion of some essential nutritional substances, such as nitrogen sources, vitamins, and lipids.⁹ On the other hand, to maintain cell survival, S. cerevisiae cells try their best to respond or adapt to the changed fermentative environment. In other words, S. cerevisiae cells might proactively pursue changes to adapt to the changed fermentative environment. To some extent, S. cerevisiae cells might undergo a transient domestication through gradual adaption to the stepwise increased ethanol concentration during a batch fermentation process.⁸ Through such transient domestication, S. cerevisiae cells might change their physiological status, metabolisms and other cellular behaviors through self-regulation; and to some extent, these cell behavioral changes might confer higher ethanol tolerance to S. cerevisiae during the fermentation process. Indeed, yeast cells did regulate their cellular behavior to adapt to the changed fermentative environment.^6,10 For example, during very-high-gravity (VHG) bioethanol fermentation, significant gene expression changes secondary to the variation between carbohydrate metabolism and stress response were detected.¹¹ In addition, various fermentation stages also led to significant changes in the metabolite profiling of S. cerevisiae Y98-5 during fermentation of the Korean traditional spirit makgeolli.¹² Similarly, yeast cell physiology at respective growth stages can affect the following fermentation performance, such as the CO₂ holding capacity.¹³ Therefore, the changed cellular behavior could play a role in changing the fermentative environment of the subsequent fermentation stage. The changed cellular behavior combined with the changed fermentative environment would further influence cell performance during the next fermentation stage (Fig. 1). Specifically, the changed cellular behavior might also confer a higher ethanol tolerance to S. cerevisiae cells during the batch bioethanol fermentation process.

Fig. 1 Overview of the relationship between Saccharomyces cerevisiae cellular behavior and fermentative environments.

Our previous study mainly focused on the relationship between the variation in cell membrane structure and composition and ethanol tolerance during the bioethanol fermentation process.⁸ However, the mechanisms underlying the ethanol tolerance of S. cerevisiae are very complex and include many genes, proteins and metabolites, and a number of bio-processes,^14–16 such as valine, inositol, the trehalose synthetic pathway, the activation of the glutamate metabolism, and so on.^17,18 As a complex living system, S. cerevisiae adapting to or tolerating ethanol stress would be related not only to the membrane, but also even to the whole system. So, the next worthwhile issue was to learn about how S. cerevisiae cells acquired higher ethanol tolerance as fermentation time increased, at the systemic level. There are few studies about the influence of fermentation stages on yeast cell tolerance abilities at the systematic level. Metabolomics analysis can assess the last step in a series of changes that occur following an external stimulus¹⁹ or a pathological insult,²⁰ and the metabolite composition and content can comprehensively reveal the phenotype transformations that always happen in living systems. A gas chromatography-mass spectrometry (GC-MS)-based metabolomics strategy has been shown to be a sensitive approach to reveal stress responses in S. cerevisiae with high resolution.^21,22 These previous results showed that a GC-MS-based metabolomics strategy can provide us with a powerful platform for determining bioethanol fermentation process-associated biochemical changes in S. cerevisiae.

In this study, a GC-MS-based metabolomics strategy was used to identify S. cerevisiae metabolic changes secondary to the bioethanol fermentation process transition, herein contributing to an understanding of the mechanism underlying the increased ethanol tolerance of S. cerevisiae as fermentation time increased.

Materials and methods

Strains, media and culture conditions

The S. cerevisiae S288c strain (ATCC number 204508) used in this study was precultured in YPD broth (20% glucose, 1% yeast extract, 2% peptone) at 30 °C overnight. The precultured yeast cells were transferred to a fermentative YPD broth and the optical density of the initial cells, as determined at an absorbance of 600 nm, was adjusted to 0.1. The fermentative culture was performed at 30 °C, with shaking at 150 rpm in 250 ml cotton-plugged flasks containing 100 ml of YPD broth.

Preparation of metabolome samples

Metabolome samples of S. cerevisiae were prepared according to the procedures of Li et al.²² with some modifications. Two milliliters of the cultures were quickly harvested from the primary cultures after 2, 8 and 16 hours of incubation (i.e., the lag, exponential and stationary phases, respectively) and immediately transferred to 15 ml-tubes, containing 8 ml of pre-chilled 60% methanol kept at −40 °C, to quench the cultures. After quenching, the yeast cells were collected using centrifugation (8000g, −4 °C, 10 min). The supernatant was discarded, and a pellet spiked with internal standard (50 μl of ribitol in water, 0.5 mg ml⁻¹) was prepared for the extraction of intracellular metabolites. The samples were suspended in 0.75 ml of pre-chilled pure methanol kept at −40 °C, and then frozen in liquid nitrogen. The frozen suspension was thawed in an ice bath, and this freeze–thaw process was repeated three times before centrifugation (8000g, −4 °C, 10 min). The supernatant was collected and an additional 0.75 ml of pre-chilled pure methanol was added to the pellet. The mixture was vortexed for 30 s prior to centrifugation (8000g, −4 °C, 10 min). Both supernatants were pooled together and stored at −20 °C until use.

Derivatization

Samples were dried in a vacuum centrifuge dryer. For derivatization, 100 μl of methoxylamine hydrochloride in pyridine (20 mg ml⁻¹), the first agent, was added to the dried samples, prior to incubation at 30 °C for 2 h. One hundred microliters of N-methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA), the second derivatizing agent, was added to the samples, before incubation at 30 °C for 3 h to trimethylsilylate the polar functional groups. The derivate samples were rapidly filtered through 0.22 μm pore size filters before GC-MS analysis.

GC-MS analysis

Chromatography analysis was performed on a GC-MS system 7890B/5977A GC/MSD (Agilent Technologies Inc. America) equipped with an Agilent J&W Ultra Inert GC-MS column (30 m × 250 μm i.d., 0.25 μm film thickness; Agilent J&W Scientific, Folsom, CA). Samples (1 μl) were injected into the DB-5 capillary column in split injection mode, with a split ratio of 30 : 1 after dilution. Helium was used as the carrier gas at a constant flow rate of 1 ml min⁻¹. The injection, ion source, and ion source surface temperatures were set to 300 °C, 200 °C, and 280 °C, respectively. The GC oven was heated to 80 °C for 1 min, raised to 100 °C at a rate of 2 °C min⁻¹, raised to 300 °C at a rate of 15 °C min⁻¹ and then maintained at 300 °C for 6 min. Electron impact ionization (70 eV) in full scan mode (m/z 80–500) at a rate of 20 scans per s was used. Ribitol served as an internal standard to monitor batch reproducibility and to correct for minor variations that occurred during sample preparation and analysis. 5977MSD MassHunter Data Analysis Software was used to acquire mass spectrometric data and peak areas. The General Integrator (MSD Chemstation RTE Integrator) and NIST11.L database were used to search for and identify metabolites restricted to peaks detected with a total ion current (TIC). The compounds were also identified by comparing their mass spectra and retention times with those of commercially available reference compounds.

Data analysis

The three stages of yeast sample metabolite levels were compared with each other, after being normalized by weight and internal standard. The generated normalized peak areas (variables) were imported into the SIMCA package (Ver. 10.0) (Umetrics, Umea, Sweden) for multivariate statistical analysis. Principal component analysis (PCA) and partial least-squares-discriminant analysis (PLS-DA) were performed on the data after mean-centering and UV-scaling. These analyses employed a default 7-fold internal cross validation from which the R² and Q² (goodness of prediction) values, representing the explained variance and the predictive capability, respectively, were extracted. An unsupervised PCA was initially performed to obtain an overview of the GC-MS data from the three fermentation stages.

To identify the bioethanol fermentation process-associated metabolite changes, a supervised PLS-DA was subsequently carried out. On the basis of the loadings plots and variable importance in the projection (VIP) value threshold (VIP > 1) from the 7-fold cross-validated PLS-DA models, variables that were responsible for distinguishing between the different fermentation phases groups were selected.

The identified fermentation process-associated differential metabolites that were selected using the PLS-DA models were further analyzed using another multivariate statistical analysis method, hierarchical cluster analysis (HCA), to assess the predictive accuracy of the PLS-DA models. HCA was performed with the aid of Cluster 3.0, and visualized using TreeView 1.1.6 software.

Independent-sample T tests were performed on specific metabolites using SPSS13.0 for Windows to assess the statistical significance of the metabolic changes, and the standard error of the mean (SEM) was used as an error bar. Differences showing P-values less than 0.05 were considered statistically significant.

Results and discussion

During the batch bioethanol fermentation process, ethanol accumulates gradually⁸ and its inhibitory effect on yeast cells increases continuously. Our previous works also confirmed the inhibition from ethanol stress on S. cerevisiae cell growth.^8,22 However, ethanol tolerance of S. cerevisiae cells did not decrease as expected, but increased continuously as fermentation time increased.⁸ How the S. cerevisiae cells acquired higher ethanol tolerance during the fermentation process was a worthy issue. Our previous study indicated that during the fermentation process, S. cerevisiae cells could remodel their membrane to acquire higher ethanol tolerance.⁸ A similar conclusion was also proposed by Hiroshi and his colleagues.²³ However, the mechanisms underlying the higher ethanol tolerance of yeast cells would not be only limited to variation in cell membrane composition or structure.²² Besides changes in the cell membrane, many other changes including cellular metabolisms would also occur during the fermentation process, and might contribute to the acquired higher ethanol tolerance of yeast cells.^17,18 Yeast cells at different stages would encounter different fermentation environments (Fig. 1), and the cellular behavior, including metabolisms, would also be different. The fermentation process transition-associated metabolic changes might contribute to an understanding of how S. cerevisiae cells acquire higher ethanol tolerance as fermentation time increases.

Effect of different fermentation phases on the intracellular metabolite profiles of S. cerevisiae

PCA and PLS-DA score scatter plots with excellent fit and satisfactory predictive ability (Table 1) were obtained to represent the sample distribution in the new multivariate space (Fig. 2). Distinct clustering was observed among the different groups. These plots suggest that over 67% of variation in the data set was due to a change in fermentation stage, demonstrating that the fermentation stage was likely responsible for the metabolic perturbation observed in the data.

Table 1 Statistical data from partial least squares-discriminant analysis (PLS-DA) models for different phases of batch bioethanol fermentation

Sample	R^2X (cum)	R^2Y (cum)	Q^2 (cum)
All groups (PCA)	0.756	—	0.613
All groups (PLS)	0.753	0.928	0.921
2 h versus 8 h	0.742	0.980	0.903
2 h versus 16 h	0.756	0.990	0.948
8 h versus 16 h	0.679	0.998	0.989

---

Fig. 2 Score plots for groups from the lag phase (i.e., 2 h), the exponential phase (i.e., 8 h), and the stationary phase (i.e., 16 h) of the batch bioethanol fermentation process. (a) PCA-derived metabolite profiles. (b) PLS-DA-derived metabolite profiles. In the scores plot, the confidence interval is defined using the Hotelling's T2 ellipse (95% confidence interval), and observations outside the confidence ellipse are considered outliers.

Changes in intracellular metabolites during the batch bioethanol fermentation process

To confirm the fermentation process transition-associated metabolic variations, the class separation was further detected using a PLS-DA pairwise comparison. PLS-DA pairwise comparisons between two of the three fermentation stages (i.e., the lag, exponential, and stationary phases) suggested an obvious metabolic difference between the classes in each pairwise comparison from the first component (Fig. 3a, c and e). The PLS-DA models were well constructed with excellent fit and satisfactory predictive power (Table 1). The major metabolic perturbations that cause the above discriminations were identified from line plots of the X-loadings of the first component in the PLS-DA models (Fig. 3b, d, and f). The intensity of the peak presented the contribution to the class separation.

Fig. 3 PLS-DA model plots for pairwise comparisons between two of the three fermentation stages (i.e., the lag, exponential, and stationary phases). Cross-validated score plots of the pairwise comparison of (a) 2 h versus 8 h, (c) 2 h versus 16 h, and (e) 8 h versus 16 h. The two groups in each scores plot were separated along the first component. In the score plots, the confidence interval is defined using the Hotelling's T2 ellipse (95% confidence interval), and observations outside the confidence ellipse are considered outliers. Loading plots of pairwise comparisons of (b) 2 h versus 8 h, (d) 2 h versus 16 h, and (f) 8 h versus 16 h.

Metabolites with a VIP value greater than 1 demonstrate a great contribution to the separation of groups in the PLS-DA models.²⁴ The VIP plots demonstrate that a lot of metabolites contributed to the class separation (ESI Fig. S1†). Based on the X-loadings line plots and VIP plots, a total of 40 metabolite paired retention time-mass to charge ratio (RT-M/Z) variables contributing to the pairwise separation were selected according to the cutoff VIP value (VIP > 1) and significant difference (P < 0.05) (Table 2). A HCA plot of the 40 identified differential metabolites reflected a clustering pattern that was similar to the PLS-DA analysis results (Fig. 4). The HCA result was consistent with the PLS-DA models, which also verified the predictive accuracy of the PLS-DA models. The metabolites that were probably responsible for the fermentation process transition-associated intracellular metabolic perturbations mainly included glycolysis and TCA cycle intermediates, saccharides, amino acids and fatty acids.

Table 2 The intracellular metabolites of Saccharomyces cerevisiae identified using GC-MS that differ between the lag phase (i.e., 2 h), the exponential phase (i.e., 8 h) and the stationary phase (i.e., 16 h) of the batch bioethanol fermentation process^a

RT	Metabolite	2 h	2 h vs. 8 h	8 h	8 h vs. 16 h	16 h	2 h vs. 16 h
a Note. The VIP scores of all listed metabolites are greater than 1. The data represent the relative peak intensities and are presented as mean ± SEM. * P < 0.05; ** P < 0.01; *** P < 0.001.
4.024	Ethylamine	0.671 ± 0.088	**	0.086 ± 0.010		0.051 ± 0.013	***
5.849	Butanediol	0.000 ± 0.000		0.000 ± 0.000	**	0.514 ± 0.122	**
6.293	Ethanol amine	0.573 ± 0.185		0.230 ± 0.019	*	0.101 ± 0.037	*
8.020	Alanine	0.655 ± 0.065	***	0.116 ± 0.007	***	0.771 ± 0.025
8.628	Glycine	0.649 ± 0.083	**	0.103 ± 0.005	**	0.307 ± 0.034	**
10.305	Malic acid	0.776 ± 0.076	***	0.148 ± 0.016	***	0.000 ± 0.000	***
12.361	Valine	0.937 ± 0.074	***	0.275 ± 0.018	***	0.520 ± 0.023	***
12.854	Urea	0.837 ± 0.047	***	0.142 ± 0.017		0.150 ± 0.020	***
13.742	Phosphoric acid	0.724 ± 0.085	***	0.121 ± 0.009		0.092 ± 0.013	***
13.791	Glycerol	0.207 ± 0.033		0.205 ± 0.021	*	0.643 ± 0.137	**
14.038	Proline	0		0	***	0.775 ± 0.076	***
14.334	Succinic acid	0.797 ± 0.082	***	0.135 ± 0.009	*	0.170 ± 0.012	***
15.123	Serine	0.814 ± 0.081	***	0.094 ± 0.005	**	0.067 ± 0.005	***
15.485	Threonine	0.843 ± 0.066	***	0.111 ± 0.007	***	0.207 ± 0.012	***
15.699	Unknown1	0.616 ± 0.112	**	0.071 ± 0.011	*	0.030 ± 0.004	**
15.863	Unknown2	0.423 ± 0.126	*	0.049 ± 0.009	**	0.000 ± 0.000	*
16.143	Unknown3	0.804 ± 0.066	***	0.125 ± 0.008	***	0.038 ± 0.013	***
16.422	L-Methionine	0.115 ± 0.021	*	0.055 ± 0.008	**	0.712 ± 0.110	**
17.409	α-Ketoglutaric acid	0.880 ± 0.103	***	0.191 ± 0.017	***	0.690 ± 0.041
17.803	Glutamic acid	0.746 ± 0.086	***	0.112 ± 0.015	*	0.219 ± 0.030	**
17.869	Phenylalanine	0.829 ± 0.070	***	0.162 ± 0.014	***	0.020 ± 0.001	***
17.968	Ribonic acid	0.646 ± 0.094	***	0.092 ± 0.016	*	0.052 ± 0.007	**
18.247	Aspartic acid	0.470 ± 0.136	*	0.056 ± 0.017	*	0.000 ± 0.000	*
19.004	Phosphoglycerol	0.839 ± 0.150	*	0.091 ± 0.023	**	0.398 ± 0.020	***
19.267	Methyl-glucofuranose	0.771 ± 0.095	***	0.113 ± 0.013	***	0.005 ± 0.001	***
19.596	Xylose	0.700 ± 0.098	**	0.096 ± 0.009	***	0.004 ± 0.001	***
19.628	Glucose	0.682 ± 0.096	**	0.114 ± 0.013	**	0.004 ± 0.000	***
19.990	Fructose	0.667 ± 0.102	**	0.089 ± 0.008	***	0.024 ± 0.005	**
20.056	β-D-Glucopyranose	0.677 ± 0.094	**	0.106 ± 0.007	***	0.004 ± 0.001	***
20.220	Tyrosine	0.595 ± 0.092	**	0.260 ± 0.167	*	0.006 ± 0.002	**
20.796	Palmitic acid	0.563 ± 0.092	**	0.093 ± 0.007	**	0.047 ± 0.010	***
21.996	Octadecanoic acid	0.760 ± 0.075	***	0.095 ± 0.004	*	0.045 ± 0.014	***
22.029	Glucose-1-phosphate	0.614 ± 0.115	**	0.131 ± 0.008	***	0.054 ± 0.009	***
22.342	Galactose-1-phosphate	0.547 ± 0.146	*	0.031 ± 0.004	***	0.119 ± 0.013	*
23.904	Monopalmitin	0.437 ± 0.131	*	0.012 ± 0.001	**	0.003 ± 0.000	*
24.249	Trehalose	0		0	**	0.595 ± 0.076	**
24.890	Glycerin monostearate	0.204 ± 0.038	**	0.016 ± 0.002	***	0.347 ± 0.034	*
28.672	Ergosterol	0	***	0.859 ± 0.087	***	0.02 ± 0.006	*

---

Fig. 4 Hierarchical cluster analysis of the 40 identified differential metabolites.

During the bioethanol fermentation process, yeast cells absorb as much sugar as possible for growth at the beginning of the fermentation. In this study, a continuously decreasing intracellular sugar content (i.e., glucose, fructose, xylose, galactose and 1-phosphate-glucose) (Table 2, Fig. 5, ESI Fig. S2†) also indicated that carbon sources were being consumed gradually and the Embden–Meyerhof–Parnas (EMP) pathway might be continuously activated during bioethanol fermentation. The continuous activated glycolysis pathway can provide plenty of metabolic intermediates for biomass demand.²⁵ As the nutrients were gradually depleted, the growth rate also decreased gradually and the biomass of yeast cells did not increase any more until the stationary phase.⁸ At this time point (the stationary phase, i.e., after 16 h of incubation), the metabolic flux might redistribute from biomass increase to the production of ethanol and other compounds (Fig. 6). In fact, the content of many compounds (i.e., some amino acids) did increase in the stationary phase, in comparison to in the exponential phase (Table 2, Fig. 5, ESI Fig. S2†).

Fig. 5 The fermentation process-associated intracellular metabolic changes in Saccharomyces cerevisiae as indicated through changes in metabolite levels. Red symbols denote significant increases (P < 0.05) and green symbols denote significant decreases (P < 0.05), whereas dark blue symbols denote no significant changes in metabolite levels (P > 0.05). Ala, alanine; Asp, aspartic acid; BCAA, branched chain amino acid; Gln, glutamine; Glu, glutamic acid; Gly, glycine; Ile, isoleucine; α-KG, α-ketoglutarate; Leu, leucine; Lys, lysine; Met, methionine; OAA, oxaloacetate; PEP, phosphoenolpyruvate; Phe, phenylalanine; Pro, proline; Ser, serine; Thr, threonine; Trp, tryptophan; Tyr, tyrosine; Val, valine.

Fig. 6 An overview of the effects of the fermentation process on the ethanol tolerance of Saccharomyces cerevisiae. Black arrows indicate mass flow, red arrows indicate positive regulation, and green bars indicate inhibition.

During bioethanol fermentation, ethanol would undoubtedly accumulate as fermentation time increased and could reach 5.03% v/v until the stationary phase (i.e., after 16 h of incubation), in YPD broth containing 20% v/v glucose.⁸ A higher ethanol level would enhance the energy requirements for yeast cell growth.²⁶ At first glance, the TCA cycle should keep activated to provide more energy.¹⁸ However, interestingly, the content of three TCA cycle intermediates (i.e., α-ketoglutarate, fumarate and succinic acid) decreased in the exponential phase, in comparison to in the lag phase, while the content of α-ketoglutarate and succinic acid re-increased in the stationary phase in comparison to in the exponential phase (Table 2, Fig. 5, ESI Fig. S2†). Such results indicate that the TCA cycle was inhibited before the stationary phase, but re-activated in the stationary phase (Fig. 6). The content of three TCA intermediates (i.e. succinate, citrate and malate) also decreased in an ethanol-tolerant yeast mutant and promoted the accumulation of glutamate, which can be converted from α-ketoglutarate, thereby contributing to the higher ethanol tolerance of the mutant.¹⁷ In this study, a decreased α-ketoglutarate content before the stationary phase might also confer ethanol tolerance on the yeast cells. The carbon flux was redistributed under 5% v/v ethanol stress from ethanol production to the TCA cycle as a result of the increased energy needed for maintaining cell demand.²⁷ During the exponential–stationary phase transition, the yeast cells might also turn from ethanol production to the TCA cycle, thereby providing plenty of energy (Fig. 6). The increased energy production together with the increased energy requirement might be also partly responsible for the increased ethanol tolerance in the stationary phase.

Compared with that in the lag phase, the intracellular content of both trehalose and glycerol increased in the stationary phase, while it did not change in the exponential phase (Table 2, Fig. 5, ESI Fig. S2†). Similarly to fermentation process transition-associated TCA cycle changes, such results for trehalose and glycerol also preliminarily indicate that the mechanisms underlying the ethanol tolerance of yeast cells in the stationary phase were different from those in the exponential phase. Although a recent viewpoint suggested that yeast cell tolerance to various stresses depends not on trehalose, but on the trehalose-6P synthase (Tps1) protein,²⁸ trehalose is traditionally regarded as a protectant, improving the ethanol tolerance of S. cerevisiae, and the trehalose content could reflect the yeast cell tolerance to ethanol stress.²⁹ For example, an engineered S. cerevisiae strain with higher levels of intercellular trehalose had significantly higher fermentation rates and bioethanol yields than wild strains during high gravity fermentation.³⁰ Moreover, trehalose accumulation could also enhance the acetic acid tolerance of S. cerevisiae.³¹ Our previous work also indicated that the environmental pH value continuously decreased between 2 hours and 16 hours during the bioethanol fermentation process.³² In this study, the accumulation of trehalose in the stationary phase could facilitate yeast cell survival, withstanding the decreased environmental pH value.

Glycerol, especially intracellular glycerol, is also considered as an important cryoprotectant³³ and could play a vital role in maintaining the redox balance^34,35 and the osmoadaptation of S. cerevisiae³⁶ through the high osmolarity glycerol (HOG) pathway.^37,38 In this study, the increased glycerol might also contribute to the higher ethanol tolerance of S. cerevisiae in the stationary phase. Moreover, an accumulation of glycerol could also occur at the expense of a biomass increase, and hyperosmotic stress would induce a redirection of glycolytic flux from biomass increase to glycerol accumulation.³⁶ In this study, increased glycerol might also indicate that glycolytic flux might reroute from biomass production towards the production of glycerol and other compounds in the stationary phase (Fig. 6).

The cytomembrane is the first assaulted target from ethanol stress,³⁹ and on the other hand, is also an important protective substance to mitigate ethanol stress. Fatty acids, especially unsaturated fatty acids (UFAs), are necessities in the cell membrane and play vital roles in the tolerance or resistance of yeast cells to ethanol stress.^6,10 Fatty acid content would change in response to ethanol stress. For example, the monounsaturated fatty acid (MUFA) content increased while the saturated fatty acid (SFA) content decreased in lyophilized S. cerevisiae, while being exposed to 15% v/v ethanol.⁴⁰ During bioethanol batch fermentation, S. cerevisiae cells might also remodel the membrane composition or structure to counteract the gradually accumulated ethanol stress. Our previous work suggested that during the bioethanol batch fermentation process, the percentage of UFAs increased continuously, while the percentage of saturated fatty acids (SFAs) decreased.⁸ In this study, the content of hexadecanoic acid (C_16 : 0) and octadecanoic acid (C_18 : 0) decreased continuously during the fermentation process (Table 2, Fig. 5, ESI Fig. S2†). However, probably due to the detection limit of GC-MS, a change of palmitelaidic acid (C_16 : 1) and oleic acid (C_18 : 1) content could not be detected as in our previous study.⁸ In spite of this, the continuous decreased hexadecanoic acid (C_16 : 0) and octadecanoic acid (C_18 : 0) content also suggested a reduction in membrane integrity; or in other words, decreased hexadecanoic acid (C_16 : 0) and octadecanoic acid (C_18 : 0) content enhanced membrane fluidity, thereby making the membrane more changeable. A more changeable membrane would facilitate yeast cells adapting to a depletion of nutrients and accumulated ethanol as fermentation time increased (Fig. 6). However, too drastic a change in the membrane structure would be unfavorable for cell viability. As ergosterol could strengthen the cell membrane rigidity,⁴¹ the increased ergosterol content in the exponential phase (Table 2, Fig. 5, ESI Fig. S2†) might be from self-protection by the S. cerevisiae cell, in order to avoid drastic membrane structural changes through counterbalancing the increased membrane fluidity and maintaining membrane structural integrity as much as possible as the fermentation time increased. Moreover, a higher ergosterol content can also promote yeast cell viability.⁴²

In addition to alanine, methionine and proline, levels of the other amino acids identified in the present study decreased in the exponential and stationary phases in comparison to in the lag phase (Table 2, Fig. 5, ESI Fig. S2†). More interestingly, in addition to aspartic acid, phenylalanine and tyrosine, the levels of all the other amino acids identified in the present study increased in the stationary phase in comparison to in the exponential phase (Table 2, Fig. 5, ESI Fig. S2†). Such results indicate that the amino acid metabolic model continuously changed during the bioethanol fermentation process; moreover, such changes might also confer a different ethanol tolerance on S. cerevisiae cells at various fermentation stages, especially a difference in ethanol tolerance between the exponential and stationary phases. The addition of some amino acids could enhance the ethanol tolerance of yeast cells through stabilizing the membrane structure.⁶ A high content of intercellular L-proline could promote the ethanol tolerance of S. cerevisiae.⁴³ Increased levels of the amino acids identified here might confer a higher ethanol tolerance to S. cerevisiae cells in the stationary phase than in the exponential phase (Fig. 6).

Conclusions

In summary, this work illustrated the utility of a GC-MS-based metabolomics strategy to evaluate the batch bioethanol fermentation process-associated intracellular biochemical changes in S. cerevisiae cells, thereby contributing to a better understanding of the molecular mechanisms underlying the continuous increased ethanol tolerance of S. cerevisiae cells during the batch bioethanol fermentation process. The metabolomics results were comprehensive and clearly shed light on an intriguing study and the different factors that confer increasing ethanol tolerance on yeast cells in the different bioethanol fermentation stages. During the bioethanol fermentation process, S. cerevisiae cells could continuously remodel their membrane composition and structure to make the membrane more and more changeable, and the more changeable membrane further conferred a higher ethanol tolerance on the S. cerevisiae cells. However, other mechanisms underlying the high ethanol tolerance in the exponential and stationary phases, in comparison to the lag phase, might be different. During the lag–exponential phase transition, in spite of the down-regulated TCA cycle, the increased ergosterol content combined with the decreased SFA content might be the most significant factor in making yeast cells more robust and ethanol-tolerant. While during the exponential–stationary phase transition, the re-activated TCA cycle could provide plenty of energy, and the increased energy production together with the increased energy requirements might be partly responsible for the increased ethanol tolerance in the stationary phase. Moreover, the increased content of glycerol, trehalose, ergosterol and some amino acids, and the decreased SFA content, might jointly confer the yeast cells with higher ethanol tolerance.

Conflict of interest

Financial disclosures: the authors have no financial conflict of interest.

Abbreviations

EMP	Embden–Meyerhof–Parnas pathway
GC-MS	Gas chromatography-mass spectrometry
HCA	Hierarchical cluster analysis
HOG	The high osmolarity glycerol
MSTFA	N-Methyl-N-(trimethylsilyl)trifluoroacetamide
MUFAs	Monounsaturated fatty acids
PCA	Principal components analysis
PLS-DA	Partial least squares-discriminant analysis
RT-M/Z	Retention time-mass to charge ratio
SEM	Standard error of the mean
SFAs	Saturated fatty acids
TCA	Tricarboxylic acid
TIC	Total ion current
Tps1	Trehalose-6P synthase
UFAs	Unsaturated fatty acids
VHG	Very-high-gravity
VIP	Variable importance in the projection

Acknowledgements

This work was supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (No. XDB14010301), the Young Marine Science Foundation, SOA (No. 2013106) and the Fundamental Research Funds of the First Institute of Oceanography, SOA (No. 2013G32).

Notes and references

1. R. C. Saxena, D. K. Adhikari and H. B. Goyal, Renewable Sustainable Energy Rev., 2009, 13, 167–178.
2. J. Goldemberg, Science, 2007, 315, 808–810.
3. F. W. Bai, W. A. Anderson and M. Moo-Young, Biotechnol. Adv., 2008, 26, 89–105.
4. H. Widiastuti, J. Y. Kim, S. Selvarasu, I. A. Karimi, H. Kim, J. S. Seo and D. Y. Lee, Biotechnol. Bioeng., 2011, 108, 655–665.
5. F. H. Lam, A. Ghaderi, G. R. Fink and G. Stephanopoulos, Science, 2014, 346, 71–75.
6. J. M. Ding, X. W. Huang, L. M. Zhang, N. Zhao, D. M. Yang and K. Q. Zhang, Appl. Microbiol. Biotechnol., 2009, 85, 253–263.
7. S. Izawa, Biosci., Biotechnol., Biochem., 2010, 74, 7–12.
8. S. J. Dong, C. F. Yi and H. Li, Int. J. Biochem. Cell Biol., 2015, 69, 196–203.
9. M. Novo, A. Mangado, M. Quirós, P. Morales, Z. Salvadó and R. Gonzalez, PLoS One, 2013, 8, e74086.
10. M. Ma and Z. L. Liu, Appl. Microbiol. Biotechnol., 2010, 87, 829–845.
11. C. G. Liu, Y. H. Lin and F. W. Bai, Biotechnol. J., 2013, 8, 1332–1340.
12. H. R. Kim, J. H. Kim, B. H. Ahn and D. H. Bai, Mycobiology, 2014, 42, 353–360.
13. N. R. Mohammad, D. Emmie, J. Pieter, P. Anali, J. V. Kevin and M. C. Christophe, Food Microbiol., 2014, 39, 108–115.
14. S. Kubota, I. Takeo, K. Kume, M. Kanai, A. Shitamukai, M. Mizunuma, T. Miyakawa, H. Shimoi, H. Iefuji and D. Hirata, Biosci., Biotechnol., Biochem., 2004, 68, 968–972.
15. X. H. Hu, M. H. Wang, T. Tan, J. R. Li Jr, H. Yang, L. Leach, R. M. Zhang and Z. W. Luo, Genetics, 2007, 175, 1479–1487.
16. M. C. Teixeira, L. R. Raposo, N. P. Mira, A. B. Lourenco and I. Sa-Correia, Appl. Environ. Microbiol., 2009, 75, 5761–5772.
17. S. Kim, J. Kim, J. H. Song, Y. H. Jung, I. S. Choi, W. Choi, Y. C. Park, J. H. Seo and K. H. Kim, Biotechnol. J., 2016, 11, 1221–1229.
18. E. Ohta, Y. Nakayama, Y. Mukai, T. Bamba and E. Fukusaki, J. Biosci. Bioeng., 2016, 121, 399–405.
19. B. V. Ravenzwaay, G. C. Cunha, E. Leibold, R. Looser, W. Mellert, A. Prokoudine, T. Walkb and J. Wiemerb, Toxicol. Lett., 2007, 172, 21–28.
20. A. Saghatelian and B. F. Cravatt, Curr. Opin. Chem. Biol., 2005, 9, 62–68.
21. M. Z. Ding, B. Z. Li, J. S. Cheng and Y. J. Yuan, OMICS, 2010, 14, 553–561.
22. H. Li, M. L. Ma, S. Luo, R. M. Zhang, P. Han and W. Hu, Int. J. Biochem. Cell Biol., 2012, 44, 1087–1096.
23. T. Hiroshi, T. Miki, K. Akari and K. Yoshito, Appl. Environ. Microbiol., 2005, 71, 8656–8662.
24. S. S. Szeto, S. N. Reinke, B. D. Sykes and B. D. Lemire, J. Proteome Res., 2010, 9, 6729–6739.
25. A. Stincone, A. Prigione, T. Cramer, M. C. Wamelink, K. Campbell, E. Cheung, V. O. Sandoval, N. M. Grüning, A. Krüger, M. T. Alam, M. A. Keller, M. Breitenbach, K. M. Brindle, J. D. Rabinowitz and M. Ralser, Biol. Rev. Cambridge Philos. Soc., 2015, 90, 927–963.
26. J. Pagliardini, G. Hubmann, S. Alfenore, E. Nevoigt, C. Bideaux and S. E. Guillouet, Microb. Cell Fact., 2013, 12, 298–301.
27. K. M. Yang, N. R. Lee, J. M. Woo, W. Coi, M. Zimmermann, L. M. Blank and J. B. Park, FEMS Yeast Res., 2012, 12, 675–684.
28. M. Petitjean, M. A. Teste, J. M. François and J. L. Parrou, J. Biol. Chem., 2015, 290, 16177–16190.
29. H. B. Kwon, E. T. Yeo, S. E. Hahn, S. C. Bae, D. Y. Kim and M. O. Byun, FEMS Yeast Res., 2003, 3, 433–440.
30. P. M. Wang, D. Q. Zheng, X. Q. Chi, O. Li, C. D. Qian, T. Z. Liu, X. Y. Zhang, F. G. Du, P. Y. Sun, A. M. Qu and X. C. Wu, Bioresour. Technol., 2014, 152, 371–376.
31. Y. Yoshiyama, K. Tanaka, K. Yoshiyama, M. Hibi, J. Ogawa and J. Shima, J. Biosci. Bioeng., 2015, 119, 172–175.
32. C. F. Yi, F. L. Wang, S. J. Dong and H. Li, Can. J. Microbiol., 2016, 62, 827–835.
33. J. Tulha, A. Lima, C. Lucas and C. Ferreira, Microb. Cell Fact., 2010, 9, 82.
34. R. Ansell, K. Granath, S. Hohmann, J. M. Thevelein and L. Adler, EMBO J., 1997, 16, 2179–2187.
35. A. Valadi, K. Granath, L. Gustafsson and L. Adler, J. Biol. Chem., 2004, 279, 39677–39685.
36. E. P. Kurdziel, C. Kuehn, B. Nordlander, D. Klein, K. K. Hong, T. Jacobson, P. Dahl, J. Schaber, J. Nielsen, S. Hohmann and E. Klipp, PLoS Comput. Biol., 2013, 9, e1003084.
37. J. L. Brewster, D. T. Valoir, N. D. Dwyer, E. Winter and M. C. Gustin, Science, 1993, 259, 1760–1763.
38. S. Hohmann, Microbiol. Mol. Biol. Rev., 2002, 66, 300–372.
39. M. Ma, P. Han, R. Zhang and H. Li, Can. J. Microbiol., 2013, 59, 589–597.
40. K. M. Archana, R. Ravi and K. A. Anu-Appaiah, J. Food Sci. Technol., 2015, 52, 6770–6776.
41. X. L. Tao, D. Q. Zheng, T. Z. Liu, P. M. Wang, W. P. Zhao, M. Y. Zhu, X. H. Jiang, Y. H. Zhao and X. C. Wu, PLoS One, 2012, 7, e31235.
42. F. Aguilera, R. A. Peinado, C. Millán, J. M. Ortega and J. C. Mauricio, Int. J. Food Microbiol., 2006, 110, 34–42.
43. H. Takagi, Appl. Microbiol. Biotechnol., 2008, 81, 211–223.

---

Footnotes

† Electronic supplementary information (ESI) available: ESI Fig. S1 and S2. See DOI: 10.1039/c6ra19254h

‡ These authors contributed equally to this paper.

---