|Home | About | Journals | Submit | Contact Us | Français|
The molecular mechanism involved in tolerance and adaptation of ethanologenic Saccharomyces cerevisiae to inhibitors (such as furfural, acetic acid, and phenol) represented in lignocellulosic hydrolysate is still unclear. Here, 18O-labeling-aided shotgun comparative proteome analysis was applied to study the global protein expression profiles of S. cerevisiae under conditions of treatment of furfural compared with furfural-free fermentation profiles. Proteins involved in glucose fermentation and/or the tricarboxylic acid cycle were upregulated in cells treated with furfural compared with the control cells, while proteins involved in glycerol biosynthesis were downregulated. Differential levels of expression of alcohol dehydrogenases were observed. On the other hand, the levels of NADH, NAD+, and NADH/NAD+ were reduced whereas the levels of ATP and ADP were increased. These observations indicate that central carbon metabolism, levels of alcohol dehydrogenases, and the redox balance may be related to tolerance of ethanologenic yeast for and adaptation to furfural. Furthermore, proteins involved in stress response, including the unfolded protein response, oxidative stress, osmotic and salt stress, DNA damage and nutrient starvation, were differentially expressed, a finding that was validated by quantitative real-time reverse transcription-PCR to further confirm that the general stress responses are essential for cellular defense against furfural. These insights into the response of yeast to the presence of furfural will benefit the design and development of inhibitor-tolerant ethanologenic yeast by metabolic engineering or synthetic biology.
Bioethanol produced from renewable resources such as lignocelluloses is considered to be an attractive alternative to fossil fuels, for it is renewable, can make use of fast-rotation plants, produces fewer emissions, and generates no net carbon dioxide. Nevertheless, there are some barriers in the lignocellulosic-to-ethanol conversion process, including inhibitor tolerance, ethanol tolerance, and utilization of xylose (62). Inhibitors formed by acid-catalyzed hydrolysis of lignocelluloses, which include furan derivatives, weak acids, and phenolic compounds, reduce both the growth rate and fermentation of ethanologenic Saccharomyces cerevisiae (2). The mechanisms of inhibition acting upon yeast during fermentation of lignocellulosic hydrolysate have been studied intensively, but mainly with traditional methods such as metabolite analysis, enzyme activity analysis, metabolic flux analysis, and kinetic analysis (50). Furfural is one of the major inhibitors for lignocellulosic hydrolysates. Previous studies have shown that in most cases, furfural can be converted by yeast to furfural alcohol (12, 30). Sometimes furoic acid (64), furoin and furil (47), and acyloin products (28, 61) can also be detected in the medium under different sets of cultivation conditions. The genetic mechanisms involved in furfural tolerance have been investigated by screening an S. cerevisiae disruption library to find potential relative genes (18). Through gene cloning and enzyme activity study, Liu et al. found that the conversion of furfural is catalyzed by multiple aldehyde reductases (40).
The traditional methods described above can analyze only one or a few metabolites, proteins, or genes and are unable to globally assess the inhibition issue, which is complex and systematic. Moreover, previous work mainly focused on extracellular metabolites and the activity of some key enzymes, whereas what happens inside yeast cells in response to inhibitors remains a “black box” to us. Integration of different “omics” tools, including those of transcriptomics, proteomics, and metabolomics, into the study of systems biology is a potentially powerful approach to address these challenges (61, 68). Many proteomic, transcriptomic, and/or metabolomic studies of S. cerevisiae have provided us with an increasingly rich understanding of the response of this organism to various environmental perturbations. Investigation of genomic expression profiles of the ethanologenic yeast S. cerevisiae to HMF (5-hydroxymethylfurfural) stress conditions showed that up to several hundred genes were differentially expressed significantly in response to HMF treatment (41, 42, 58). Comparative lipidomics analysis has been applied to study the ethanologenic yeast response to different inhibitors, such as furfural, acetic acid, and phenol (69). The results of comparative proteome analysis (8, 23) and small-molecule metabolite profiling of ethanologenic yeast during industrial fermentation (13) have been previously reported, enhancing the molecular understanding of physiological adaptation of industrial strains for optimizing the performance of industrial bioethanol fermentation. However, the specifics of global protein expression in response to the presence of biomass conversion inhibitors have not yet been quantitatively measured for ethanologenic yeast.
Quantitative proteomics, i.e., quantifying protein expression levels in different sets of complex biological samples on a large scale, is critical for our understanding of biological systems and pathways as a whole. It is considered likely to be a potential cornerstone of systems biology in the near future (56). Current quantitative proteomic methods fall into three categories: the traditional two-dimensional (2D) electrophoresis, stable isotope labeling, and nonlabeling methods (45). Each of the different labeling methods undergoing development has its advantages and disadvantages (see reference 45 for reviews). Among quantitative proteomic methods, 18O stable isotope labeling is convenient to use, low in cost, highly specific in terms of specific 18O C-terminal modifications, and capable in theory of labeling proteins globally. The 18O-labeling method has demonstrated its applicability in differential comparative proteomics with biological applications performed using Porphyromonas gingivalis strain W50 (4), the human plasma proteome (55), breast cancer cells (5), and the low-molecular-weight serum proteome (26). 18O labeling is becoming a powerful labeling strategy for quantitative proteomics application.
To give insights into the tolerance and adaptation of ethanologenic yeast to biomass conversion inhibitors at the protein level, comparative shotgun proteomic investigations combining 18O labeling with 2D liquid chromatography-tandem mass spectrometry (2D-LC-MS/MS) were performed here to systematically identify proteins by the use of an industrial strain of S. cerevisiae and to quantify cells treated with furfural compared with control cells under aerobic batch culture conditions. Quantitative real-time reverse transcription-PCR (RT-PCR) and metabolite analysis were utilized to provide orthogonal evidence for the comparative proteome results.
An industrial strain of S. cerevisiae, purchased from Angel Yeast Co., Ltd. (Hubei, People's Republic of China), in the form of alcohol instant active dry yeast, was utilized in this study. This industrial strain has the advantages of thermal resistance (38 to 42°C), low-acid tolerance (pH 2.5), and high glucose tolerance (60%) and can tolerate 13% (vol/vol) ethanol.
After recovery from a lyophilized form, S. cerevisiae was maintained on agar slants containing YEPD medium (2% glucose, 2% yeast extract, 1% peptone, and 2% agar). S. cerevisiae was initially grown in 250-ml conical flasks containing 50 ml of YEPD medium (2% glucose, 2% yeast extract, and 1% peptone) on a rotary shaker at 30°C and 160 rpm for 12 h. Subsequently, the 50-ml seed cultures were transferred into 2-liter conical flasks containing 450 ml of YEPD medium on a rotary shaker at 30°C and 90 rpm for approximately 12 h and grown to an optical density (OD) of about 3. Cells for the control experiment and the furfural treatment experiment were harvested from the same inoculation culture. An initial OD of 0.35 was used for aerobic bath cultures performed at 30°C in 2-liter conical flasks containing 450 ml of medium, with a stirrer speed of 90 rpm. The aerobic bath culture medium was composed of 10% glucose, 2% yeast extract, and 1% peptone. During exponential growth in the respiratory-fermentative phase, when the OD was approximately 3 to 4, 50-ml volumes of aerobic bath culture media containing 0 and 7.33 ml of furfural were introduced into the medium for the control experiment and the furfural treatment experiment, respectively. Samples for subsequent protein extraction were collected by centrifugation at 5,000 × g for 10 min at 4°C from furfural-treated and control cultures at 20 min and 2 h after the addition of furfural, respectively.
The concentration of furfural in lignocellulosic hydrolysates can range from 0.5 to 11 g/liter, and there are a broad range of other compounds that have inhibitory effects on microbial fermentation (2). Usually, the inhibitor concentrations used to test their effects on fermentation were 10 to 100 times larger than the concentration found in the hydrolysates (31). Therefore, based on our primary fermentation experiments and the pertinent literature, the final furfural concentration used was set at 17 g/liter for the study of the response of yeast to the presence of furfural under an extreme set of conditions.
Cell growth was determined by measuring the absorbance of the culture at 600 nm with a spectrometer (model 722 grating spectrometer; Shanghai No. 3 Analysis Equipment Factory, Shanghai, China). The concentrations of glucose, ethanol, glycerol, and furfural were measured by high-performance LC. Samples from an aerobic bath culture were first filtered through 0.22-μm-pore-size sterile filters and loaded onto an Aminex HPX-87H ion-exchange column (Bio-Rad, Hercules, CA) operated at 65°C and were then eluted with 5 mM H2SO4 at 0.6 ml/min. A refractive index detector was used. Cell viability was assessed by methylene-blue staining and a fluorescence microscope (Eclipse E800; Nikon, Japan).
The extraction of whole-yeast-cell proteins was conducted as described by Wang and Yuan (66), with minor modifications. After reduction and alkylation, the proteins were precipitated again by using cold organic solvent (ethanol/acetone/acetic acid, 50:50:0.1) overnight at −25°C, followed by centrifugation and lyophilization. The pellets were stored at −25°C until use.
Protein samples from the control experiment and the furfural treatment experiment were dissolved in digestion buffer (1 M urea, 100 mM NH4HCO3) made using 16O and 18O water (Isotec, Miamisburg, OH) (95%), with each solution maintained at a concentration of 1 μg/μl, and were digested with trypsin (Promega, Madison, WI) at a ratio of 50:1 at 37°C for 24 h. Then, additional trypsin was added to achieve a final ratio of 20:1, and the incubation was maintained at 37°C for another 18 h. Digestions were terminated by adding formic acid to the final volume concentration of 5%. The corresponding 16O- and 18O-labeled samples from the same time point were combined immediately before 2D-LC-MS/MS analysis.
Nanoflow LC-MS/MS analysis was performed by the use of a LCQ DecaXP Maxa mass spectrometer (Thermo Finnigan, Palo Alto, CA) under the control of the Xcalibur data system (Thermo Finnigan, Palo Alto, CA) as described by Wang and Yuan (66). There were some modifications in terms of salt steps and the elution gradient. A 14-step separation from the strong cation exchanger followed by a gradient elution from the reverse-phase chromatography results was utilized to separate the peptides. The 14 salt steps used 0, 25, 40, 50, 75, 100, 150, 200, 250, 300, 400, 500, 700, and 1,000 mM ammonium chloride, respectively. The elution gradient for the reverse-phase chromatography consisted of 1 min of 100% buffer A (5% [vol/vol] acetonitrile-0.1% [vol/vol] formic acid in water); a 70-min gradient to 30% buffer B (0.1% [vol/vol] formic acid in acetonitrile); a 20-min gradient to 50% buffer B; a 10-min gradient to 95% buffer B; 5 min of 95% buffer B; an 8-min gradient to 100% buffer A; and 7 min of re-equilibration at 100% buffer A.
MS/MS data were searched using the SEQUEST algorithm, and a database of S. cerevisiae open reading frames was downloaded from the Saccharomyces Genome Database on 9 March 2007. The parameters used for the analysis of MS/MS spectra were detailed in a previous study (66). In this study, carboxyl-terminal double 18Os were designated for use in variable modification. Then, SEQUEST output files were submitted to the PeptideProphet and ProteinProphet websites of the Seattle Proteomics Center (http://tools.proteomecenter.org) for statistical assessment of peptide and protein sequence matches, respectively (65). INTERACT software was utilized to organize and display the results. The accepted error rate for peptide and protein identification was controlled to remain below 10%.
The abundance ratios (18O/16O) for labeled peptide pairs were calculated using reconstructed ion chromatograms and the following equation, which is similar to one previously reported (70) but with slight modifications:
where I0, I2, and I4 represent the measured relative intensities for the first two isotopic variants of unlabeled peptides, monolabeled peptides, and dilabeled peptides, respectively, and M0, M2, and M4 represent the sums of the theoretical relative intensities for the first two isotopic variants of unlabeled peptides, monolabeled peptides, and dilabeled peptides, respectively. The natural isotopic distribution was calculated using the peptide sequence and the MS-isotope program (http://prospector.ucsf.edu/). The isotope distribution pattern for the 18O-labeled peptide was assumed to be the same as for the unlabeled peptide. The labeled and unlabeled peptide pair ratios were first sorted by protein locus and filtered using Dixon's test to omit the outliers. The remaining peptide ratios were used to calculate the protein means and standard deviations.
Total RNAs from three biological replicates were isolated using Trizol reagent (Invitrogen) for RT-PCR. Each biological replicate was analyzed thrice. Quantitative real-time PCR assays were performed using SYBR green PCR master mix (ABI) and an ABI 7300 real-time PCR system (ABI). The sequences of primers used for quantitative PCR are described in Table S1 in the supplemental material. The cycling program was as follows: an initial cycle of 2 min at 50°C and 10 min at 95°C, followed by 40 cycles of 10 s at 95°C and 30 s at 60°C. The disassociation analysis was carried out in a routine fashion by acquiring fluorescent readings for 1°C increases from 55 to 95°C. Data were analyzed using system 7300 SDS software to calculate threshold cycle (CT) values. Subsequently, the relative expression ratios for the genes were determined according to the following equations:
Rapid sampling, quenching, and metabolite extraction of biomass (in approximately 10 ml of culture broth) were performed according to the method of Luo et al. (44). All LC-MS experiments were carried out using an LCQ DecaXP Max mass spectrometer (Thermo Finnigan, Palo Alto, CA) under the control of the Xcalibur data system (Thermo Finnigan, Palo Alto, CA). A sample of 20 μl was loaded onto an Atlantis T3 4.6- by 250-mm column (Waters, Ireland) (5 μm pore size), which was equilibrated for 30 min before loading with 98% solution A (5 mM NH4HCO3 in water) and 2% solution B (80% methanol-5 mM NH4HCO3 in water). A 20-min gradient to 60% solution A and a 5-min gradient to 60% solution A were used. The column eluent was electrosprayed directly into a mass spectrometer. The mass spectrometer was operated in the negative-ion and selected-reaction-monitoring mode. The optimized parameters were as follows: ion-spray voltage, −4.5 kV; sheath gas and auxiliary gas, 33 and 5 (arbitrary units), respectively. The capillary temperature was 300°C, and the scan range was 140 to 850 m/z. The standard curve was obtained by analyzing standard solutions (NADH and NAD+; Sigma, St. Louis, MO) at five concentrations (50, 10, 1, 0.5, and 0.1 μg) selected for NADH and NAD+ concentration calculations. The levels of NADH/NAD+ and ATP/ADP were calculated according to the ratios of the integrated ion currents. Three biological replicate experiments were performed, with two injections for each sample.
The functional category and subcellular localization determinations for proteins were carried out with FunCatDB software (http://mips.gsf.de/projects/funcat), and biochemical pathways were classified and reconstructed by reference to the KEGG database (http://www.genome.jp/kegg/) and the Saccharomyces Genome Database (http://www.yeastgenome.org/). The Saccharomyces Genome Database was also utilized to obtain information on proteins. Proteins quantified with respect to two or more peptides were considered to be significantly changed when one of the following three criteria was satisfied: (i) expression changed by no less than ±1.5-fold and with relative standard deviations (RSDs) below 40%; (ii) determination of an 18O/16O ratio higher than 3 or lower than 0.33 and an RSD below 50%; or (3) all corresponding peptide abundances changed by no less than ±1.5-fold without regard to RSD values.
The physiological effect of furfural on the industrial S. cerevisiae strain was studied by comparison of cultivation with the addition of furfural (17 g/liter) into aerobic batch cultivations during exponential growth in the respiratory-fermentative phase to the control cultivations grown under the same conditions without the introduction of furfural. The results are presented in Fig. Fig.1.1. The addition of furfural almost completely suppressed cell growth, leading to very slow cell growth in the first 2 h after addition of furfural and a halt in biomass formation after that. Upon extended incubation for 9 h, no increase in growth was observed. Both ethanol formation and glucose consumption were inhibited, showing slow rates in the first 4 h and then stopping at 4 h after the addition of furfural. Glycerol formation was halted immediately, and there was no change in the glycerol concentration after the addition of furfural. In contrast, the furfural concentration in the medium decreased sharply during the first 4 h and slowly in the next 4 h and then showed no change with a residual concentration of about 6 g/liter, indicating that furfural had been converted by yeast cells to compounds of lesser toxicity. Using the methylene blue technique for cell viability determinations during the experiment, we observed that yeast cells were not able to recover from the inhibition caused by furfural and that all the cells died at 8 h after the addition of furfural (see Fig. S1 in the supplemental material).
It is obvious that the introduction of furfural into aerobic batch yeast cultivations had an immediate and drastic effect on the physiological behavior of yeast and that the response of yeast to the presence of furfural was a continued dynamic process. Confronted with furfural, the industrial yeast cells showed a notable decrease in cell growth, glucose consumption, ethanol production, and glycerol production but still tried to detoxify furfural to alleviate its inhibitory effects. These results may reflect a metabolic rearrangement in yeast to cope with furfural detoxification and with diverse effects caused by the presence of furfural, such as the arrest of cell growth and/or the accumulation of acetaldehyde, underscoring the need for utilization of systems biology approaches for further research into the effect of furfural on the industrial S. cerevisiae strain.
To study the proteomic response of the industrial S. cerevisiae strain under conditions of treatment with furfural, a shotgun yeast comparative proteome investigation has been carried out. Two equivalent whole-yeast extracts acquired from the control experiment and the furfural treatment experiment were trypsinized into peptides in H216O and H218O, respectively. Samples from the same time point (the 20-min or 2-h time point) were combined in a 1:1 ratio and subjected to three rounds of analysis using 2D-LC-MS/MS. Proteins were identified through database searching, and relative protein expression levels were determined by calculating corrected 18O/16O peptide ratios and using peak areas. A total of 2,037 and 3,655 peptides (corresponding to 205 and 309 proteins, respectively) derived from three replicate experiments run for both 20 min and 2 h were quantified manually. Of those proteins, 175 were quantified for both of the datasets. According to the data-mining criteria described above (see Materials and Methods) as used for analysis of proteins that are differentially expressed, 70 proteins were upregulated and 6 proteins were downregulated at 20 min, whereas 31 proteins were upregulated and 35 proteins were downregulated at 2 h in response to the presence of furfural.
To get an overview of the differentially expressed proteins and guide subsequent data analysis, determinations of the subcellular localizations and functional categories of the differentially expressed proteins were carried out using FunCatDB software. The numbers of differentially expressed proteins for each cellular compartment or functional category for time points 20 min and 2 h are shown in the form of 100% stacked columns, as displayed in Fig. Fig.22.
Subcellular localization distributions of differentially expressed proteins at 20 min were notably different from those seen at 2 h. Proteins with an abundance change at 20 min were not localized to the bud, golgi, or peroxisome compartments, whereas proteins with an abundance change at 2 h did not originate from the cell wall. The other subcellular compartments were well represented and had different proportions of differentially expressed proteins in each data set. The diverse subcellular localization distribution profiling results for 20 min of cultivation versus 2 h demonstrate that furfural has different effects on cellular compartments over time. When introduced into the aerobic batch culture, furfural first entered into the yeast cell through the cell wall and plasma membrane and affected yeast gene expression in as little as 10 min (39). As a result, the numbers of differentially expressed proteins localized to the cell wall, plasma membrane, and nucleus were much greater at 20 min than at 2 h. On the other hand, more compartments, including the bud, golgi, and peroxisome compartments, were affected by furfural at 2 h than at 20 min.
The function distribution of proteins with altered expression in response to the presence of furfural at 20 min and 2 h is presented in Fig. Fig.2B.2B. As a whole, differentially expressed proteins at 20 min were overrepresented compared to those seen at 2 h for most function groups, with the exception of the regulation of metabolism and protein function, unclassified protein, and development function groups. At 20 min of cultivation, furfural may affect yeast more severely and result in more proteins with changes in abundance than would be seen at 2 h, due to the higher concentration of furfural at the earlier time point (15.9 g/liter and 10.5 g/liter at 20 min and 2 h, respectively; see Fig. Fig.1).1). Yeast cells can convert furfural to less-toxic compounds, resulting in a decreased concentration of furfural over time. A dose-dependent response of ethanologenic yeast to the presence of furfural and HMF at concentrations from 10 to 120 mM has been characterized by Liu et al. (43). Proteins with abundance changes caused by the presence of furfural were localized to most compartments and were found to be involved in almost all the functions and pathways in yeast cells, revealing that the response of yeast to furfural is global and systematic. Furthermore, our observations also suggest that the response of yeast to furfural is a continued dynamic and complex process.
As implied by the KEGG pathway analysis of the differentially expressed proteins, furfural may have a great impact on the glycolysis pathway. After the addition of furfural into the aerobic batch cultures of the industrial S. cerevisiae strain, 12 out of 16 proteins quantified in the glycolysis pathway had been dramatically upregulated at 20 min, while only 1 of 17 proteins had been upregulated and 3 proteins downregulated at 2 h (Fig. (Fig.3).3). Evidently, the glycolysis of S. cerevisiae had first been rapidly induced by the addition of furfural and had then reached control levels at 2 h compared to the control experiment results (verified by NADH/NAD+ and ATP/ADP assays). This gives further evidence that the response of yeast to the presence of furfural is a continuing dynamic process, as described above. It is reasonable to believe that the expression levels of proteins catalyzing the reactions of glycolysis decrease over time, for glucose consumption had almost halted and wash out of the cell cultures had occurred at 8 h under the conditions that included treatment with furfural. The idea of the activation of glycolysis in the presence of furfural is supported by other studies. Taherzadeh et al. (61) determined that glycolysis inducing changes in abundance was immediately affected by furfural and that the reduction of furfural relied on active glycolysis by investigating conversion of furfural in aerobic and anaerobic bath cultures of S. cerevisiae CBS 8066 growing on glucose. Analysis of the metabolic flux distributions for aerobic steady-state cultures with and without furfural in the medium showed that the presentation of furfural in the medium resulted in an increase of 30% in the specific rate of glycolysis compared to the rate seen with furfural-free medium (27). Through metabolite analysis, Palmqvist et al. (48) found that furfural decreased cell replication without inhibiting cell activity and had a twofold effect on the kinetics of glucose metabolism in S. cerevisiae (i.e., the glucose metabolism rate was inhibited but the final ethanol yield was slightly increased at a nonlethal concentration of furfural) (48).
In contrast to the results seen with activated glycolysis, the growth and fermentation performance of the industrial S. cerevisiae strain were significantly retarded. However, notable reduction of furfural was observed in the first 4 h, demonstrating that activated glycolysis may correlate with the conversion of furfural. Presumably, the reduction of furfural to furfural alcohol may be catalyzed by alcohol dehydrogenases (ADHs), with NADH as a cofactor (12, 46, 67), which was evidenced here by the upregulation of Adh5p and Adh1p as described below. It is possible that glycolysis is activated by furfural to provide as much NADH as is required for the reduction of furfural, whereas ethanol production is reduced, because the reduction of furfural competes with the reaction of acetaldehyde to ethanol for both NADH and ADHs.
Downregulated levels of Gpd1p (at 20 min, not quantified; at 2 h, 0.46), Rhr2p (0.58; 0), and Hor2p (not quantified; 0.43) catalyzing glycerol biosynthesis were observed in furfural-treated cells compared to control cells (Fig. (Fig.3).3). Furthermore, as shown in Fig. Fig.1,1, the glycerol concentration stayed constant after the introduction of furfural to the aerobic bath cultures of yeast, indicating that furfural severely inhibits glycerol formation. It has been demonstrated that the reduction of furfural to furfural alcohol is preferred to glycerol production as a redox sink, subsequently resulting in the replacement of glycerol formation by furfural alcohol production (48, 61). Palmqvist et al. (49) and Taherzadeh et al. (61) observed that glycerol production was decreased in the presence of furfural under anaerobic conditions. Glycerol biosynthesis acts as a redox sink, providing additional reoxidation of cytosolic NADH. At the same time, NADH is the major cofactor required for reduction of furfural to furfural alcohol. As a result, glycerol production and furfural reduction compete for a shared pool of NADH, as is also observed with the reaction of acetaldehyde to ethanol and furfural reduction. In the presence of a high concentration of furfural, yeast reduces the production of glycerol to meet the urgent requirement of NADH for the conversion of furfural to furfural alcohol (see Fig. Fig.88).
As a central metabolic pathway, the tricarboxylic acid (TCA) cycle provides precursors for many compounds, including some amino acids, and generates useful amounts of ATP and NADH under aerobic conditions. None of the enzymes in the TCA cycle was identified at 20 min, whereas Cit1p, Aco1p, Aco2p, Idh1p, and Mah1p were identified and quantified at 2 h and all displayed a trend toward increased production, with Aco2p and Mdh1p noticeably upregulated in the presence of furfural. Analysis of the metabolic flux distributions for the aerobic steady-state cultures with and without furfural in the medium showed that the presence of furfural in the medium resulted in a 50% increase in the specific rate of the TCA cycle compared to the rate seen with furfural-free medium (27). The upregulation of enzymes at 2 h reveals that the TCA cycle can be activated in the presence of furfural to produce more NADH for the reduction of furfural (see Fig. Fig.88).
The pentose phosphate pathway (PPP) is an important carbohydrate metabolism pathway, oxidizing glucose to generate NADPH for reductive biosynthesis reactions within cells and ribose-5-phosphate for the synthesis of the nucleotides and nucleic acid. However, the expression levels of Gnd1p (at 20 min, 1.46; at 2 h, 1.08), Tkl1p (0.91; 0.98), and Tal1p (1.23; 1.30) involved in the PPP were not affected by the presence of furfural at either time point (Fig. (Fig.3).3). This observation is consistent with the results of a previous study showing that although selective-deletion mutants coded by genes in the PPP showed growth deficiencies in the presence of furfural, these mutants were inefficient in reducing furfural to furfural alcohol (18). Therefore, it is suggested that the PPP may have no direct correlation with furfural conversion under the conditions studied here.
Since it has been reported that the central carbon metabolism of S. cerevisiae is controlled to a large extent via posttranscriptional mechanisms in chemostat studies (11, 32, 33), quantitative RT-PCR was not carried out to substantiate our findings concerning the involvement of the central carbon metabolism in cellular response to the presence of furfural. Instead, the levels of intracellular NADH, NAD+, NADH/NAD+, and ATP/ADP were measured by LC/MS (Fig. (Fig.4).4). The intracellular concentrations of NAD+ and NADH were decreased at least twofold and fourfold, respectively, leading to a lower NADH/NAD+ ratio in furfural-treated cells compared with control cells at both time points. In contrast, the ATP/ADP ratio was increased significantly at 20 min but not at 2 h. The urgent requirement of furfural reduction caused a severe shortage of NADH, leading to the upregulation of some enzymes in the central carbon metabolism to provide as much NADH and ATP as possible in cells treated with furfural compared with control cells at 20 min and 2 h (although the effect was less significant at 2 h). The need for ATP may be smaller than that for NADH in furfural-treated cells. Thus, the levels of intracellular NAD+, NADH, and NADH/NAD+ were decreased whereas the ATP/ADP ratio was increased after the addition of furfural. These results corroborate the comparative proteome results described above. NAD+ and NADH are involved in various biological processes, including aging, apoptosis, cell death, energy metabolism, mitochondrial functions, calcium homeostasis, antioxidation-generation of oxidative stress, gene expression, and so on (71). It has been suggested that NAD+ depletion mediates poly(ADP-ribose) polymerase-1-induced cell death (1). Previous work has suggested that NAD+ and NADH may be involved in apoptosis: it was reported that selective inhibitors of NAD+ synthesis can induce apoptosis (24) and that NADH/NADPH depletion is an early event in apoptosis (52). Thus, the decrease in both NAD+ and NADH levels in yeast cells upon treatment with furfural, resulting from an inhibition of synthesis of these compounds or from accelerated degradation, may be ultimately responsible for the cell death observed. Furthermore, our data suggest that acetaldehyde likely accumulates in the culture during furfural reduction due to a decreased NADH concentration in the cell. In a previous study, the accumulation of acetaldehyde after the addition of furfural was observed and the effect of furfural on cell replication was shown to be related to acetaldehyde formation (48). Acetaldehyde has been found to exert inhibition effects on yeast growth (59), which is consistent with the arrest of growth observed in this study. Obviously, the presence of furfural caused several secondary effects, including the drop in both NAD+ and NADH levels, the accumulation of acetaldehyde, and a contribution to the general stress environment, which in turn may affect the cellular metabolism in yeast cells treated with furfural.
The altered expression levels of most proteins catalyzing the reactions of central carbon metabolism (with the exception of those involved in the PPP) revealed that the addition of furfural led to a central carbon metabolism rearrangement in yeast cells in order to induce toleration of furfural. Glycolysis and/or the TCA cycle were stimulated to provide sufficient NADH for efficient conversion of furfural. Also, NADH was diverted from glycerol synthesis and ethanol production into furfural alcohol formation. Thus, the formation of ethanol and glycerol was inhibited during the adaptation of yeast to furfural (see Fig. Fig.88).
Among the seven ADHs in yeast cells, three were quantified at 20 min and six at 2 h (Fig. (Fig.5).5). When 17 g of furfural/liter of medium was used, the expression of Adh5p was markedly upregulated (i.e., it was more than 4 times higher than that seen in furfural-free medium at both time points) whereas translational levels showed no significant changes. This discrepancy for Adh5p at the protein level and the transcript level has been reported by other groups (6, 7, 9). Adh5p can catalyze the conversion of acetaldehyde to ethanol, with activity apparent only in an adh1 and adh3 double-deletion strain (57), and this conversion can be induced by the presence of dimethyl sulfoxide (72). Adh6p was significantly upregulated at 20 min (with one peptide quantified) and showed no change at 2 h. Larroy et al. reported that Adh6p is an NADPH-dependent ADH of broad substrate specificity that is able to reduce aldehydes, including cinnamaldehyde, veratraldehyde, and furfural (34). In addition, the reduction of 5-hydroxymethyl furfural and furfural with NADPH as a cofactor was increased in cell-free crude extracts from Adh6p-overexpressing strains (51). Adh1p, the major enzyme catalyzing the reaction of acetaldehyde to ethanol, displayed mediate increases at both time points, which were verified by quantitative RT-PCR (see Fig. Fig.7).7). Adh1p has been suggested in previous studies to be an enzyme possibly catalyzing the reduction of furfural to furfural alcohol (21). The overexpression of Adh1p can increase the formaldehyde resistance of S. cerevisiae (20). Since ethanol formation was severely inhibited at both time points, the enhancement of ADHs was not related to the reaction of acetaldehyde to ethanol. Furthermore, it has been reported that the reduction of furfural to furfural alcohol is likely catalyzed by ADHs (12, 46, 67). Thus, in the context of the literature and our experiments, it is possible that Adh5p, Adh1p, and Adh6p may catalyze the reduction of furfural to furfural alcohol. However, further validation studies are necessary to confirm this hypothesis, since the induction of Adh5p, Adh1p, and Adh6p by furfural may be associated with other biological processes.
Adh2p, Adh4p, and Sfa1p were detected and quantified only at 2 h. Adh2p and Sfa1p had decreased expression levels, whereas the expression abundance of Adh4p was not affected by the presence of furfural. Adh2p, unlike other ADHs, catalyzes the reaction of ethanol to acetaldehyde and is repressed in the presence of glucose (9). Addition of furfural inhibited the glucose consumption and led to higher glucose concentrations in the furfural treatment experiment (Fig. (Fig.1),1), and this in turn repressed the expression of Adh2p. The downregulation of Sfa1p is consistent with the lower ethanol concentration in the medium with furfural compared to the furfural-free medium at 2 h, as it had been reported that Sfa1p is induced by the presence of ethanol (17). The decreased levels of expression of Adh2p and Sfa1p suggest that the ripple effect imposed by the presence of furfural exists and becomes more significant with the passage of time after the addition of furfural. Thus, ADHs may play a role in the tolerance and adaptation of ethanologenic yeast to furfural.
Although it has been mentioned before that furfural may result in the accumulation of reactive oxygen species, vacuole and mitochondrial membrane damage, and chromatin and actin damage in S. cerevisiae (39), there has been no previous study focusing on the stress effects caused in S. cerevisiae by the presence of furfural. The levels of abundance of 23 proteins related to stress response displayed significant changes either at one time point or at both time points (Fig. (Fig.6A).6A). Analysis with respect to functional category showed that these proteins are involved in the response of yeast to unfolded proteins, oxidative stress, osmotic and salt stress, DNA damage, and nutrient starvation.
There are eight proteins related to unfolded protein response (UPR), including five HSP70 proteins: Ssb1p, Ssb2p, Ssc1p, Ssz1p, and Kar2p. At 20 min after the addition of furfural, proteins related to the folding, sorting, and translocation of newly synthesized polypeptide chains such as Egd2p (16), Hsp10p (25), and Ssb1p were upregulated. At 2 h, no change in the expression of Egd2p and Ssb1p was seen, whereas Hsp10p was slightly downregulated, as shown by the results of quantification of one peptide. Instead, another group of proteins (Ssb2p, Ssc1p, and Ssz1p) related to the folding, sorting, and translocation of newly synthesized polypeptide chains displayed increased expression levels at 2 h after the addition of furfural. It is worth noting that Kar2p was upregulated at 2 h, since Kar2p not only is induced by UPR but is also involved in the regulation of UPR through interaction with Ire1p. The induction of these proteins related to the UPR implies that the addition of furfural may lead to the accumulation of unfolded proteins, subsequently resulting in triggering of the UPR. Further evidence was observed when the results of stimulation of ribosome proteins were recorded (Fig. (Fig.6B).6B). At 20 min, all of the 32 ribosomal proteins quantified in this study exhibited a trend to increased regulation changes as a group; 17 of those proteins were significantly upregulated in response to furfural. At 2 h, 16 of 34 ribosomal proteins quantified showed distinctly increased expression levels, while 1 protein was slightly downregulated. At both time points, 27 proteins were quantified, with 11 proteins upregulated. The main function of ribosome is to organize protein synthesis. The upregulation of ribosomal proteins suggests that protein synthesis in the aerobic batch culture containing furfural may have been accelerated compared to that seen with furfural-free cultures under the same conditions. The acceleration of protein synthesis may lead to the accumulation of unfolded proteins in the endoplasmic reticulum (ER) lumen, in turn activating the UPR to restore protein-folding capacity and adapt to new conditions caused by the presence of furfural. The UPR can protect cells against ER stress, but when this objective cannot be achieved within a certain time period or when ER stress is prolonged, the UPR can initiate cell death or apoptosis. The concentration of furfural applied in this study was so high that high-intensity and long-term ER stress existed, and washout of cultures occurred at 8 h.
Proteins that respond to oxidative stress represent the second-largest group among the stress-response-related proteins that showed abundance changes in the presence of furfural. The thioredoxins Trx1p and Trx2p, heat shock protein Hsp12p, and superoxide dismutase Sod1p were significantly upregulated at 20 min but showed no abundance change at 2 h in response to the presence of furfural. Expression of Tsa1p, a housekeeping thioredoxin peroxidase, was first notably upregulated at 20 min and then slightly downregulated at 2 h. The expression level of Ahp1 was downregulated at both time points after the addition of furfural into aerobic batch cultures of S. cerevisiae. Ahp1p is a thiol-specific peroxiredoxin protecting cells from oxidative damage by reducing hydroperoxides (35). Yhb1p, a nitric oxide oxidoreductase detoxifying nitric oxide (38), was quantified only at 2 h and showed an increased expression level. The oxidative stress may be related to the decrease of the NAD+, NADH, and NADH/NAD+ levels, since it has been suggested that NAD+ and NADH influence antioxidation and the generation of oxidative stress (71).
At both time points, furfural induced the expression levels of Rps3p (at 20 min, 1.70; at 2 h, 1.42) and Stm1p (3.25; 1.87), which respond to DNA damage. Genetic analyses have suggested that Stm1p, a G4 quadruplex and purine motif triplex nucleic acid-binding protein, participates in several biological processes, including interaction with ribosome and subtelomeric Y′ DNA (63), telomere maintenance by interaction with Cdc13p, and apoptosis (25, 36). Furthermore, its accumulation induces cell death (36). Like the UPR, the upregulation of Stm1p at both time points may be involved in the washout of cultures at 8 h. In contrast, Rps3p is essential for viability (15) and is involved in DNA damage processing and with apurinic-apyrimidinic endonuclease activity (29). Thus, the upregulation of expression of these two proteins reveals that furfural causes DNA damage in S. cerevisiae.
The Pkc1p-mitogen-activated protein kinase (Pkc1p-MAP kinase) pathway regulates cell wall maintenance and integrity, which are essential for the growth and the integrity of proliferating cells (60). The Pkc1p-MAP kinase pathway is reported to be negatively regulated by Zeo1p (at 20 min, 2.3; at 2 h, 1.3) (19) and Lsp1p (at 20 min, not quantified; at 2 h, 2.83) (73), so the upregulation of expression of Zeo1p at 20 min and of Lsp1p at 2 h revealed that this pathway is inactivated in the presence of furfural, leading to lack of cell wall integrity and defective cell growth. This hypothesis is further supported by the downregulation of Rho1p (at 20 min, not quantified; at 2 h, 0.31), which is required for the Pkc1p localization to sites of polarized growth throughout the cell cycle and also to regions of cell wall damage (3, 54). Rho1p also regulates cell wall synthesizing enzyme 1, also called 3-beta-glucan synthase (Fks1p and Gsc2p) (14, 54). Expression of Act1p, a structural constituent of the cytoskeleton that is involved in cell polarization, endocytosis, and many other cytoskeletal functions (53), was significantly upregulated at 20 min and unchanged at 2 h. The altered expression of proteins that are involved in cell integrity, growth, and survival strongly implies that the presence of furfural had caused a rearrangement of cell structure and damage to yeast cell integrity, growth, and survival. Moreover, Pho2p, which participates in nutrient starvation, was significantly upregulated at 20 min. PHO2 is known as a transcriptional activator of PHO5 and PHO81 (phosphate utilization), HIS4 (histidine biosynthesis), CYC 1, TRP4, and HO; it also activates expression of the ADE1, ADE2, ADE5, ADE7, and ADE8 genes, which are involved in the metabolic pathway of purine nucleotide biosynthesis (10, 37).
To confirm the protein expression data obtained by 18O labeling, the transcript levels of the 10 selected stress-related proteins described above were measured by quantitative RT-PCR at 20 min and 2 h (Fig. (Fig.7).7). The quantitative RT-PCR results for Act1p, Hsp10p, Hsp12p, Pho2p, Tsa1p, and Zeo1p at 20 min and for Lsp1p and Stm1p at 2 h are consistent with the relative quantitative protein expression results obtained using 18O labeling. Furthermore, the levels of expression of Zeo1p, Hsp10p, and Hsp12p changed in the same direction at both the protein level and the transcription level, although the changes at the transcript level were statistically significant whereas those at the protein level were not. Generally, there were discrepancies between quantitative RT-PCR results and the relative quantitative protein expression results such as those seen with Ahp1p at both time points. These discrepancies may have been due to the fact that, apart from the effect determined by the amount of mRNA present, the protein expression level is influenced by protein turnover and posttranslational modifications. What is more, the mRNA molecules may be relatively unstable compared to proteins in general, contributing to the difference in turnover rates between mRNA and protein, as reported in a previous study using S. cerevisiae (22). The quantitative RT-PCR results provide orthogonal evidence of the reliability of the relative quantitative protein expression results determined using 18O labeling.
Clearly, furfural not only influences yeast with respect to primary carbonate metabolic pathways and protein synthesis but also causes the formation of a complex stress environment in yeast cells. The expression levels of proteins involved in common stress responses, including the UPR, oxidative stress, osmotic and salt stress, DNA damage, and nutrient starvation, were altered due to the complex stress environment formed by the addition of furfural to the aerobic batch culture. The redirection of resources toward stress defense may lead to diminishing amounts of free available energy supplied by catabolism for cell growth and insufficient ATP for phosphorylation of glucose to form glucose-6-phosphate, which is critical for the utilization of glucose. Thus, cell growth and glucose consumption are inhibited by furfural. With the passage of time, yeast cells first display a lag phase and eventually adapt to furfural stress, but when the concentration of furfural is too high and requires too much NADH, the yeast cell is severely damaged and washout occurs (Fig. (Fig.88).
The addition of high concentrations of furfural to the aerobic batch cultures of an industrial strain severely inhibited biomass growth, glucose consumption, ethanol production, and glycerol production. Relative quantitative proteomic data here provide a deeper understanding of the molecular mechanism involved in the response of S. cerevisiae to furfural under aerobic batch conditions. Together with upregulation of Adh5p and Adh1p, activation of glycolysis and/or the TCA cycle and repression of glycerol biosynthesis were observed, suggesting that the reduction of furfural to furfural alcohol catalyzed by Adh5p and Adh1p with NADH as a cofactor may be a potential pathway for the conversion of furfural to compounds of lesser toxicity. What is more, proteins involved in stress response were also differentially expressed due to the complex stress environment whose formation was caused by the presence of furfural. The redirection of resources toward furfural conversion and stress defense may lead to the inhibition of yeast cell growth and fermentation. Quantitative RT-PCR and metabolite analysis (of the levels of NADH, NAD+, NADH/NAD+, and ATP/ADP) were utilized to provide orthogonal evidence supporting the comparative proteomics results. Secondary effects due to the presence of furfural were observed and may have been related to the inhibitory effects of furfural. These insights into the response of yeast to furfural will benefit the design and development of inhibitor-tolerant ethanologenic yeast strains for lignocellulose-bioethanol fermentation, which is one of the significant challenges for cost-competitive bioethanol production.
We are very grateful for the financial support from the National Science Fund of China for Distinguished Young Scholars (project 20425620), the Key Program (project 20736006), National Basic Research Program 973 of China (grant 2007CB714301), the International Collaboration Project of MOST (grant 2006DFA62400), and Key Projects in the National Science & Technology Pillar Program (grant 2007BAD42B02).
Published ahead of print on 10 April 2009.
†Supplemental material for this article may be found at http://aem.asm.org/.