Mutations in the Saccharomyces cerevisiae Succinate

Mutations in the Saccharomyces cerevisiae Succinate Dehydrogenase Result in Distinct Metabolic Phenotypes Revealed Through 1H NMR-Based Metabolic Footprinting Samuel S. W. Szeto,† Stacey N. Reinke,† Brian D. Sykes, and Bernard D. Lemire* Department of Biochemistry, School of Molecular & Systems Medicine, University of Alberta, Edmonton, Alberta, Canada T6G 2H7 Received August 28, 2010

Metabolomics is a powerful method of examining the intricate connections between mutations, metabolism, and disease. Metabolic footprinting examines the extracellular metabolome or exometabolome. We employed NMR-based metabolic footprinting and multivariate statistical analysis to examine a yeast model of mitochondrial dysfunction. Succinate dehydrogenase (SDH) is a component of both the tricarboxylic acid cycle and the mitochondrial respiratory chain. Mutations in the human SDH are linked to a variety of cancers or neurodegenerative disorders, highlighting the genotype/ phenotype complexity associated with SDH dysfunction. To gain insight into the underlying global metabolic consequences of SDH dysfunction, we examined the metabolic footprints of SDH3 and SDH4 mutants. We identified and quantified 36 metabolites in the exometabolome. Our results indicate that SDH mutations cause significant alterations to several areas of yeast metabolism. Multivariate statistical analysis allowed us to discriminate between the different metabotypes of individual mutants, including mutants that were phenotypically indistinguishable. Metabotypes were highly correlated to mutant growth yields, suggesting that the characterization of metabotypes offers a rapid means of investigating the phenotype of a new mutation. Our study provides novel insight into the metabolic effects of SDH dysfunction and highlights the effectiveness of metabolic footprinting for examining complex disorders, such as mitochondrial diseases. Keywords: mitochondria • yeast • metabolomics • phenotypic profiling • metabolic footprinting • succinate ubiquinone oxidoreductase

Introduction The emergence of metabolomics as an area of study holds great promise in understanding how metabolite levels influence phenotypes.1-3 Metabolomics provides a top-down, systems level readout of the biochemical status of the cell. The metabolome is thought to be the most sensitive and functional measure of the cellular state.4,5 Metabolomics has become particularly pertinent for the examination of disease conditions and their associated metabolic alterations and an important goal of metabolomics is the identification of useful biomarkers as prognostic or diagnostic tools.6-10 Metabolic footprinting characterizes the profile of the extracellular metabolites, consumed from or excreted into the media by cells; this is referred to as the exometabolome.5,11,12 This innovative strategy exploits the notion that changes in the exometabolome are a direct reflection of intracellular metabolic activity. Monitoring the exometabolome is efficient, noninvasive and not subject to the technical difficulties associated with isolating intracellular metabolites.5,11 Metabolic footprinting has been applied suc* To whom correspondence should be addressed. Department of Biochemistry, University of Alberta, Edmonton, Alberta, T6G 2H7, Canada. Phone: (780) 492-4853. Fax: (780) 492-0886. Email: [email protected]. † These authors contributed equally to this manuscript. 10.1021/pr100880y

 2010 American Chemical Society

cessfully in metabolomic studies of bacteria, yeast and human cell culture.11,13-15 Mitochondria and particularly the mitochondrial respiratory chain (MRC) play a fundamental role in aerobic metabolism. Although the primary role of the MRC is in energy metabolism and the maintenance of redox balance and cellular energy levels, MRC function also influences a multitude of metabolic pathways and cellular processes not directly connected with energy metabolism.16-19 MRC dysfunction thus represents the most common group of inborn errors of metabolism.20 The clinical presentations of mitochondrial diseases are numerous, affecting almost any tissue or organ system and being highly variable in the age of onset and severity.21 The factors that contribute to this variability are poorly understood, making diagnoses of mitochondrial disease and prediction of disease progression very challenging. A number of metabolites whose levels are linked to mitochondrial diseases has been identified but a consistent and reliable marker of MRC dysfunction is still lacking.22 To facilitate a better understanding of the molecular etiology of mitochondrial diseases and define the disease signature, a more systematic and comprehensive evaluation of the metabolic alterations resulting from MRC dysfunction is needed.15 Recently, Shaham et al. set out to address these issues by characterizing the metabolic profile of spent culture Journal of Proteome Research 2010, 9, 6729–6739 6729 Published on Web 10/21/2010

research articles medium, using a cultured cell model of mitochondrial disease. Their work demonstrated that examination of the metabolic footprint can provide insight into the biochemical derangements that occur due to mitochondrial disease; they also identified a new potential biomarker.15 To gain insight into the pathology of mitochondrial diseases, we investigated how the metabolome was affected in a S. cerevisiae model of succinate dehydrogenase (SDH) dysfunction. The yeast enzyme is functionally and structurally similar to its mammalian counterpart, making it a suitable model for study.23 SDH, also known as complex II or succinate:ubiquinone oxidoreductase, is a critical enzyme functioning in both the MRC and the tricarboxylic acid (TCA) cycle, forming a link between these two essential metabolic processes.24-26 SDH is a tetrameric, iron-sulfur flavoprotein that resides in the mitochondrial inner membrane and catalyzes the oxidization of succinate to fumarate coupled to the reduction of the lipid soluble electron carrier ubiquinone to ubiquinol. SDH is comprised of two domains; the catalytic domain, formed by Sdh1p and Sdh2p, where succinate oxidation occurs and the membrane anchor domain, consisting of Sdh3p and Sdh4p, the site of ubiquinone reduction.23,26 Since SDH participates in two fundamental metabolic pathways; its dysfunction can result in diverse array of disease presentations. In humans, mutations in the SDHA gene, encoding the flavoprotein subunit, can result in a collection of clinical phenotypes, including optic atrophy, ataxia, muscle weakness, myopathy and Leigh syndrome (a degenerative disorder of the central nervous system).27,28 Mutations in the four SDH genes have also been linked to a variety of cancers ranging from paragangliomas and pheochromocytomas to renal cell carcinomas and colorectal cancer.29-34 In our previous work, we characterized yeast models of point mutations in residues that are associated with tumor formation in humans.35-39 We focused our attention on residues lining the proximal ubiquinone-binding (QP) site of the membrane anchor domain. Residues from both membrane anchor subunits contribute to the formation of the QP site and are frequently mutated in paragangliomas and pheochromocytomas.40-42 We modeled tumorigenic mutations in the conserved yeast residues, SDH3 Arg-47 and SDH4 Asp-88.42 SDH3 Arg-47 was mutated to Cys, Glu and Lys and SDH4 Asp-88 to Asn, Glu and Lys.39 These mutants exhibited variable respiratory growth, which was indicative of the severity of the mutation.39 They also resulted in the accumulation of succinate. We hypothesized that mutation to a critical enzyme in energy metabolism, such as SDH, should result in global alterations to metabolism and that these changes would be reflected in the exometabolome. Profiling the metabolic changes resulting from SDH dysfunction may also reveal additional pathways that are dependent on SDH activity and facilitate a greater understanding into the molecular details underlying pathogenesis. In this study, we used a 1H NMR based metabolic footprinting approach to examine the metabolic profiles of SDH3 and SDH4 mutants. We were able to identify and quantify 36 metabolites, including nucleic acid constituents, organic acids, amino acids and sugars. Our findings show that mutations to SDH result in significant alterations in yeast metabolism as reflected by changes in the exometabolome. Using partial leastsquares discriminant analysis (PLS-DA), we were able to discriminate the various yeast strains from each other on the basis of their metabolic phenotypes, or metabotypes. These metabotypes were strongly correlated to mutant respiratory 6730

Journal of Proteome Research • Vol. 9, No. 12, 2010

Szeto et al. growth yields, identifying a relationship between these two parameters. Our study demonstrates the usefulness of metabolic footprinting in elucidating the biochemical details underlying complex diseases, such as mitochondrial disorders and provides insight into the metabolic perturbations resulting from SDH dysfunction.

Experimental Procedures Strains, Media and Culture Conditions. The S. cerevisiae strainssdh3W1(MH125,sdh3::TRP1),sdh4W2(MH125,sdh4::TRP1) and the various site-directed mutants used in this study have been described previously.39,43,44 Plasmids were introduced into the yeast strains by lithium acetate-mediated transformation.45 Strains were plated on solid SD medium containing casamino acids (0.5% w/v) without tryptophan for plasmid retention. Cells were grown to stationary stage (36 h at 30 °C) in 2 mL of YP medium containing 0.25% glucose to maximize the differences in metabotypes due to respiratory deficiency. This liquid medium is a modified version of solid medium previously used to examine acid secretion by the SDH3 and SDH4 mutant strains.39,46 The yeast media YPD and SD have been described.43 All experiments were performed under normoxic conditions. Preparation of Exometabolome Samples. After completion of the incubation period, 100 µL of culture were removed and used to measure the optical density at 600 nm. The remaining culture was centrifuged at 14 000× g for 2 min to pellet the cells. The clarified media were transferred to new microcentrifuge tubes. The samples were subjected to trichloroacetic acid precipitation (5% final concentration) and incubation on ice for 30 min. The samples were centrifuged at 14 000× g for 15 min. The supernatants were recovered, adjusted to pH 7.0 with 5 M NaOH, flash frozen in liquid N2 and stored at -80 °C until they could be lyophilized for 2 days and stored dry at 4 °C. The lyophilized samples were dissolved in 570 µL D2O (99.9%; Isotec Inc., Miamisburg, Ohio) along with 30 µL of 5 mM 2,2-dimethyl-2-sila 3,3,4,4,5,5-hexadeutero-pentane sulfonic acid (DSS-d6, Chenomx Inc., Edmonton, Alberta) as a chemical shift indicator and concentration standard for NMR analysis. The pH was recorded for calibration purposes and samples were centrifuged at 14 000× g for 3 min to remove particulate matter. Five-hundred ten microliters of supernatant were transferred to 5 mm diameter NMR tubes for data collection. 1 H NMR Spectroscopy and NMR Data Processing. Onedimensional 1H NMR spectra were acquired on a 600 MHz Varian Inova spectrometer (Varian Inc., Palo Alto, California) at 30 °C using a tnnoesy pulse sequence (circa Vnmr 6.1B software, Varian Inc.). A pulse width of 7.95 µs (90°) was used. All spectra had an acquisition time of 4 s, a preacquisition delay of 1 s, a mixing time of 0.1 s, a sweep width of 7200 Hz and 256 transients collected with 33,500 data points.39,47 All spectra were Fourier transformed without line broadening applied, referenced to the DSS-d6 singlet at 0 ppm, manually phased and baseline-corrected. Chenomx NMR Suite Professional software v5.1 (Chenomx Inc., Edmonton, Alberta) was used for identification and quantification of metabolites by computerassisted manual fitting of selected peaks. This software uses pattern recognition and line shape deconvolution to fit spectra.48 The spectral patterns of many metabolites contain more than one peak throughout the spectrum, often forming complex and unique sets of peaks. The resonance linewidths are input from the DSS standard. The baseline will vary between spectral

Metabolic Footprinting of Yeast SDH Mutants

research articles

Table 1. SDH Mutants Used in This Study and Their Phenotypic Propertiesa

The first approach assesses growth on solid glycerol-containing medium; growth on glycerol requires respiration because glycerol is a nonfermentable carbon source. Respiratory capacity was also determined using a growth yield assay in a galactose-containing medium. S. cerevisiae exhibits a diauxic shift when cultured on galactose, a fermentable carbon source; an initial fermentative growth stage is followed by a subsequent respiratory phase when the galactose is exhausted.44,51,52 We determined the spectral profiles of the acid-soluble extracellular metabolites in the spent culture media after growth using solution state 1H NMR (Figure 1). A metabolite profile of fresh culture medium was also established. The spectra possessed relatively flat baselines, which is indicative of samples containing minimal lipid or protein. The high spectral resolution and low level of contaminating material enabled accurate quantification of the identified metabolites. A total of 36 metabolites were identified, including amino acids, organic acids, nucleic acid constituents and sugars, demonstrating the wealth of biochemical information that can be extracted using this approach. We compared the metabolite concentrations of the medium before and after cell growth to determine which metabolites were consumed or excreted by the yeast strains during the culturing time. As shown in Figure 2, the concentrations of many of the metabolites quantified changed significantly (P < 0.05), reflecting their net consumption or excretion by the yeast cells. The overall trends in the changes in metabolite concentration were consistent between the SDH3 and SDH4 sets of strains. Two metabolites, glucose and niacinamide were completely consumed during growth as they were present in the original culture medium but absent after growth of the yeast strains. Of the nutrients consumed, the majority was amino acids (alanine, arginine, asparagine, aspartate, leucine, methionine, phenylalanine, proline, serine, threonine, tryptophan and tyrosine), but betaine, lactate and uracil were also consumed. We detected a variety of metabolites that were excreted into the medium, including butyrate, fumarate, succinate, nicotinate and significant quantities of acetate. Some metabolite concentrations were altered in all strains, while others were only altered in a certain subset of the strains; these latter changes correlated with respiratory capacity. For example, we only detected significant decreases in glycerol concentration for the wild-type strains and the respiration-competent mutants, consistent with previous results that the respiration-deficient mutants are unable to grow on a glycerol-containing medium. In contrast, acetate was found exclusively in the respiration-deficient strains.

strain

SDH3 SDH3 SDH3 SDH3 SDH3 SDH4 SDH4 SDH4 SDH4 SDH4

WT Arg47Cys Arg47Glu Arg47Lys KO WT Asp88Asn Asp88Glu Asp88Lys KO

growth on solid minimal glycerol media

growth yield in liquid galactose mediab

succinate/ fumaratec

+ + + + + -

100% 35 ( 4% 12 ( 1% 70 ( 6% 12 ( 1% 100% 72 ( 9% 80 ( 7% 10 ( 1% 10 ( 1%

60 ( 10 190 ( 30 170 ( 20 210 ( 30 190 ( 20 80 ( 10 180 ( 20 150 ( 40 170 ( 30 150 ( 20

a Reference is Szeto et al. (2007). b Values represent the percentage growths of the yeast strains compared to their respective wild-type controls. c From this study.

regions but not all of a metabolite’s peaks need to be used in the fitting. Fitting can be performed utilizing only the top portion of any peak but the overall fit must agree in all regions. In this way, baseline issues are minimized. Furthermore, we consistently utilize the same set of peaks for quantification. Metabolite Data Analysis. Metabolic data were subjected to multivariate statistical analysis using Simca P+ v12.0.1 software (Umetrics, Umeå, Sweden).49 Data were not transformed but were scaled to unit variance, dividing each variable by its standard deviation, and mean centered so as to treat all variables with equal importance. Data were initially visualized using the unsupervised principal component analysis (PCA) to note any outliers. Data were then treated with the supervised partial least-squares discriminant analysis (PLS-DA) to gain optimal separation between groups. R2 and Q2 values were used as indications of model fit and predictability, respectively. Validations of individual groups within each model were performed by 999 random permutations of y-variables (metabolites) while keeping x-variables (strain) intact. The measures of fit (R2 and Q2) from the permutations were compared with those of the original. Additionally, CV-ANOVA (analysis of variance testing of cross-validated predictive residuals) tests were performed to determine significant differences between groups in each model. The Pearson product-moment correlation coefficient and level of significance were determined as described.50

Results SDH Dysfunction Results in Severity-Dependent Alterations to the Exometabolome. Our previous work characterizing tumorigenic mutations modeled in the yeast SDH revealed elevated levels of succinate secreted into the medium, a hallmark of SDH dysfunction associated with oncogenesis.39 We hypothesized that mutations to SDH would cause additional perturbations to global metabolism and that these would also be reflected in changes to the exometabolome. To investigate this, we employed a 1H NMR metabolic footprinting approach to examine the exometabolome of yeast strains containing SDH3 and SDH4 mutations. These mutations, in order of decreasing respiratory capacity, are SDH3 Arg47Lys, Arg47Cys and Arg47Glu and for SDH4, Asp88Glu, Asp88Asn and Asp88Lys (Table 1). The respective knockout strains transformed with either the corresponding wild-type gene or with empty vector were also examined. The respiratory capacity of each strain was determined previously using two approaches.

To facilitate a comparison of the metabotypes between strains, we normalized the metabolite concentrations to the final optical densities achieved by the culture. These normalized metabolite concentrations were compared to those of the corresponding wild-type (Figure 3). All of the normalized metabolite concentrations, except for fumarate and isobutyrate, followed a consistent pattern, with their concentrations correlating with the severity of the mutation. The respirationcompetent mutants SDH3 Arg47Lys, SDH4 Asp88Asn and Asp88Glu had metabolite levels that were comparable or slightly higher than wild-types. Significantly higher concentrations were measured for the respiration-deficient strains. These results show that SDH dysfunction causes reproducible changes in metabolism that are reflected in the exometabolome. Journal of Proteome Research • Vol. 9, No. 12, 2010 6731

research articles

Szeto et al.

Figure 1. High resolution 600 MHz 1H NMR spectra of postgrowth medium. The aliphatic regions of representative spectra are shown. Key metabolites are labeled. BCAA, branched chain amino acids. (A) SDH3 wild-type, (B) ∆SDH3.

Furthermore, these changes are clearly dependent on the severity of the mutation. SDH Mutant Metabotypes are Differentiated using Multivariate Analysis. We employed multivariate analysis to further analyze the metabotypes. The data sets were not transformed but were scaled to unit variance and mean-centered. Principal component analysis (PCA) was used to inspect data variation and identify potential outliers. The metabotypes were subjected to a supervised extension of PCA, partial least-squares discriminant analysis (PLS-DA), to obtain the maximal separation between profiles and for statistical analysis.49 The resulting scatter plots reveal that strains can be discriminated on the basis of their metabolic footprints; the replicate data sets for each strain cluster into discernible groups (Figure 4). Each 6732


color-coded dot on the PLS-DA plots represents a replicate metabotype. The groups were separated mainly along component 1 within each model. Model performance can be assessed by the R2 and Q2 values. The SDH3 model possessed goodness of fit values, R2Xcum and R2Ycum, of 0.96 and 0.759, while the SDH4 model had values of 0.907 and 0.692, respectively. The Q2cum values are measures of predictability and were 0.663 and 0.593 for the SDH3 and SDH4 models, respectively. These values indicate that the models possess an excellent fit and are predictive of the data. Model validations using 999 permutations and five principle components for the SDH3 model (Supporting Information Figure 1) and four principle components for the SDH4 model were performed (Supporting Information Figure 2). There are two aspects of the validation plots


research articles

Figure 2. Ratio of metabolites in postgrowth to pregrowth medium. Postgrowth metabolite concentrations were normalized to the mass of dry protein-free lysates and compared to mass-normalized pregrowth media concentrations. Metabolite concentrations were not normalized to cell culture densities. (A) Amino acids from SDH3 strains. (B) Nonamino acid metabolites from SDH3 strains. Yellow, WT; red, Arg47Lys; green, Arg47Cys; black, Arg47Glu; blue, KO. (C) Amino acids from SDH4 strains. (D) Nonamino acids from SDH4 strains. Yellow, WT; red, Asp88Glu; green, Asp88Asn; black, Asp88Lys; blue, KO. Journal of Proteome Research • Vol. 9, No. 12, 2010 6733

research articles

Szeto et al.

Figure 3. Growth-normalized comparison of metabolites in postgrowth medium. Metabolites were quantified and normalized to the mass of dry protein-free lysates and to cell culture densities. (A) Nonamino acid metabolites in postgrowth media of SDH3 mutants. (B) Amino acids in postgrowth media of SDH3 mutants. Yellow, WT; red, Arg47Lys; green, Arg47Cys; black, Arg47Glu; blue, KO. (C) Nonamino acid metabolites in postgrowth media of SDH4 mutants. (D) Amino acids in postgrowth media of SDH4 mutants. Yellow, WT; red, Asp88Glu; green, Asp88Asn; black, Asp88Lys; blue, KO.

that address model validity. First, that none of the permutations outperform the original model (the original model points on the far right of the plot are higher than the blue and green data points to the left). The second aspect is that the Y-intercepts of the regression lines are below the acceptable cutoff values of 0.4 for the upper trend line and 0.05 for the lower trend line. 6734


The validation plots for all of data sets, with the exception of the SDH4 Asp88Glu mutant satisfied both criteria and indicate that these models are valid. The validation plot for the SDH4 Asp88Glu mutant had one permutation outperform the original data but satisfied all other criteria. To examine the statistical significance of the differences between the strains in the model,

research articles


Figure 4. PLS-DA model plots of exometabolome profiles. (A) SDH3 mutant strains. Yellow, WT; Arg47Lys; green, Arg47Cys; black, Arg47Glu; blue, KO. The model is characterized by the following parameters using 5 components: R2Xcum ) 0.96, R2Ycum ) 0.759, Q2cum ) 0.663. N ) 12 for all strains, CV-ANOVA p ) 4.16 × 10-23. (B) SDH4 mutant strains. Yellow, WT; red, Asp88Glu; green, Asp88Asn; black, Asp88Lys; blue, KO. The model is characterized by the following parameters using 4 components: R2Xcum ) 0.907, R2Ycum ) 0.692, Q2cum ) 0.593. N ) 12 for all strains, CV-ANOVA p ) 2.19 × 10-20.

CV-ANOVA tests were performed. These resulted in scores of p ) 4.16 × 10-23 for the SDH3 strains and p ) 2.19 × 10-20 for the SDH4 strains, indicating that the differences between the groups within each model are highly significant. The variable influence on projection (VIP) plots indicate that a large proportion of the metabolites identified contributes to the discrimination of the strains (Figure 5, A and B). A VIP score greater than 1 indicates that the metabolite has a significant influence on the separation of the groups within the model. Our model shows that at least two-thirds of the metabolites significantly contribute to the separation between groups; this pattern is unusual but demonstrates the importance of evaluating the entire metabolic profile rather than a few key metabolites. Interestingly, the most discriminating metabolites differed between the SDH3 and SDH4 models. For example, acetate ranked as the most important metabolite in the SDH4 model but its contribution to the SDH3 model was significantly lower. Consistent with the VIP plots, the loading plots reveal that the majority of the metabolites are responsible for the clustering of the sample groups (Figure 5, C and D). Squares on the loadings plot represent each strain group and triangles represent individual metabolites. Metabolites located furthest away from the center of the plot are responsible for the clustering of the groups whereas metabolites situated at the center of the

plot do not contribute. That is, the proximity of a particular metabolite in spatial location to a certain group reflects its increased relative concentration compared with that of other groups. The concentrations of those metabolites situated in the same spatial location of a sample group are increased relative to other sample groups and the loading plots show that the respiration-deficient strains are enriched in the majority of the metabolites identified. Metabotypes and Strain Growth Yields are Correlated. PLSDA of the data sets revealed a noticeable trend. The wild-type strains are on one side of scatter plot and the corresponding knockout strains appear on the other side. The three mutant strains are located between these two extremes and their positions appeared to correlate with their degree of respiratory capacity (Figure 4). For the SDH3 strains, the Arg47Lys data set clustered near the wild-type while the Arg47Cys and Arg47Glu sets clustered nearer the knockout data set. The Arg47Lys mutant is respiration-competent, while the Arg47Cys and Arg47Glu mutants are significantly impaired (Table 1). For the SDH4 strains, the Asp88Glu and Asp88Asn data sets cluster nearer the wild-type, while the Asp88Lys data set is on the same side as the knockout data set. For the SDH4 strains, Asp88Glu and Asp88Asn mutants are respiration-competent, while the Asp88Lys mutant is not (Table 1). We examined the relationship between the metabotypes and their growth phenotypes. The mean loading score for each of the data sets was plotted against the previously determined growth yields of these strains determined using a galactose growth assay.39 Visual inspection of the scatter plots suggests a good correlation between metabotype and growth yield (Figure 6). We determined the Pearson product-moment correlation coefficient for the SDH3 strains (P < 0.02) and the SDH4 strains (P < 0.05). The results indicate a strong correlation is present between the exometabolome and respiration-competence.

Discussion Metabolic footprinting is an innovative method for the functional analysis and characterization of a cell or organism’s metabolic status. Footprinting using the extracellular metabolome gives access to intracellular metabolic information in a noninvasive manner and may be particularly useful for clinical applications. To date, exometabolic studies have been successfully performed in a number of model systems, ranging from bacteria, to yeast and cultured human cells.5,11,14,15 In this work, we utilized a high resolution 1H NMR spectroscopic-based approach in conjunction with multivariate statistical analysis to examine a yeast model of mitochondrial dysfunction. We focused on determining the effects of single amino acid substitutions in the membrane domain of SDH, a key enzyme in the MRC and the TCA cycle. We were able to characterize the metabolic footprints from these mutants and identified a number of biochemically relevant molecules (Figure 1). Our results indicate that single SDH point mutations cause distinct alterations to cellular metabolism that are reflected in the exometabolome (Figure 2 and 3). The magnitudes of the perturbations correlate with the severity of the mutation, with larger changes in the exometabolome observed for mutants with decreasing respiration-competence (Figure 3). Respirationcompetent mutants show more modest alterations to their metabolic footprints. The metabolic footprints of the individual wild type and mutant strains were clustered and could be discriminated from each other by the multivariate statistical method PLS-DA (Figure 4). Additional analysis of the PLS-DA Journal of Proteome Research • Vol. 9, No. 12, 2010 6735

research articles

Szeto et al.

Figure 5. VIP and Loadings Plots. Variable importance of projection plots for exometabolites along component 1. (A) SDH3 PLS-DA model. (B) SDH4 PLS-DA model. Loadings plots using PLS-DA components 1 and 2. (C) SDH3 model. RfK, Arg47Lys; RfC, Arg47Cys; RfE, Arg47Glu. (D) SDH4 model. DfE, Asp88Glu; DfN, Asp88Asn; DfK, Asp88Lys. Squares, strain groups; triangles, individual metabolites.

models, using VIP and loadings plots, demonstrated that at least two-thirds of the metabolites significantly contributed to separation between groups (Figure 5). This pattern is highly unusual but demonstrates the importance of evaluating a more comprehensive metabolic profile, rather than a few key me6736


tabolites. We suggest that metabolic footprinting may be a robust method for discriminating between mutants that are otherwise phenotypically indistinguishable.11 The most severely impaired mutants, SDH3 Arg47Glu and SDH4 Asp88Lys were easily discriminated from their respective knockout strains by


research articles

Figure 7. Overview of carbohydrate and amino acid metabolism in yeast. Major energy-related metabolic pathways and amino acid biosynthetic and catabolic pathways are highlighted. Black text, metabolites with increased concentrations compared to wild-type when normalized to mass and cell culture density; gray text, metabolites that were not detected; boxed, metabolites completely depleted from the medium. BCAA, branched chain amino acids. The location of SDH in the TCA cycle is indicated. KYN ) kynurenine. Figure 6. Correlation between component 1 loading score and growth yield. Component 1 loading scores from each PLS-DA model are plotted against previously determined growth yields on galactose medium; the wild type is set to 100%.39 (A) SDH3 mutant strains. Pearson correlation coefficient R ) -0.97, degrees of freedom ) 3, p < 0.02. (B) SDH4 mutant strains. R ) 0.93, degrees of freedom ) 3, p < 0.05.

their metabotypes (Figure 3), even though the mutants and knockout strains exhibit similar growth phenotypes on fermentable and nonfermentable carbon sources (Table 1). Unlike the knockout strains, the SDH3 Arg47Glu and SDH4 Asp88Lys enzymes retain residual SDH activity when examined in vitro. These low levels of activity may be sufficient to account for the observable metabolic differences between these yeast strains. While previous studies using metabolic footprinting have focused mainly on characterizing gene deletion mutants, our results demonstrate footprinting can also discriminate between hypomorphic mutations. Our results also show a strong correlation between the metabotypes of the mutants and their respiratory competence (Figure 6). Metabotypes can therefore provide functional information on how phenotypes are governed or determined by metabolites.2 Our results also suggest that characterization of metabotypes may offer a rapid means of gaining insight into the phenotype of a new mutation (Figure 7). Metabolic analysis of these SDH mutants also provides insight into the effects of SDH dysfunction on cellular metabolism. One of the more interesting metabolic changes is the presence of large amounts of extracellular acetate found in the most dysfunctional mutant strains (Figure 2). Acetogenesis is

thought to be a direct consequence of TCA cycle impairment and has been observed in SDH, fumarase and malate dehydrogenase mutants.53 After glucose is depleted, cells undergo the diauxic shift to respiration utilizing ethanol.54 Acetyl-CoA is synthesized in the cytosol from ethanol via the pyruvate dehydrogenase bypass pathway. It is imported into the mitochondria via the carnitine acetyltransferase system.55-57 TCA cycle dysfunction will lead to a buildup of acetyl-CoA in the mitochondrial matrix. S. cerevisiae is unable to transport acetylCoA out of mitochondria and it is converted to acetate by the acetyl-CoA hydrolase.58 Acetate, in its protonated form, can leave the mitochondrial matrix by passive diffusion. Alternatively, an unknown mitochondrial carrier or exchanger may participate in acetate export.59-61 Excess acetate is likely excreted from the cell by various monocarboxylate transporters.62,63 It is interesting to note that extracellular acetate levels are a better predictor of SDH dysfunction in these mutants than is the succinate to fumarate ratio. In our previous work, all of the SDH mutants had elevated ratios of succinate to fumarate compared to the wild-type (Table 1).39 The succinate to fumarate ratios do not correlate with the severity of the mutation. That acetate excretion was only observed for the most severe mutations suggests that the respirationcompetent mutants have sufficient SDH activity to metabolize acetate via the TCA cycle; interestingly, these same mutants secrete succinate.39,64 Amino acid concentrations varied greatly between strains. Amino acids, particularly asparagine, serine and threonine were taken up from the medium by all strains (Figure 2). When normalized to cell growth, we noted that amino acids were not Journal of Proteome Research • Vol. 9, No. 12, 2010 6737

research articles utilized as readily in SDH-deficient strains as in the wild-type strains. Two reasons may explain this observation. First, the mutant strains likely have decreased protein synthesis. Second, amino acids may be used as an energy source upon ethanol depletion but the TCA cycle impairment of the SDH mutants may limit these pathways. Amino acids can be utilized as both carbon and nitrogen sources, with exception of the branched chain and aromatic amino acids, which can only serve as nitrogen sources in yeast.65,66 Metabolic profiling is emerging as a powerful tool for elucidating the molecular etiologies underlying complex human disorders. In this study, we used metabolic footprinting to examine the metabolic consequences of various levels of SDH dysfunction. Our findings suggest footprinting may offer unique advantages in understanding the severity of mutations. Other mitochondrial and metabolic disease models should also be amenable to a metabolomic approach, which may provide additional insight into the pathogenesis of complex diseases.

Acknowledgment. We thank David Broadhurst for his insightful comments during the preparation of this manuscript. S.N.R. was supported by an AHFMR Studentship. This work is supported by Canadian Institutes of Health Research Grants MOP-37769 to B.D.S. and MT-15336 and MT-15290 to B.D.L. Supporting Information Available: Validation plots for the SDH3 PLS-DA models (SI Figure 1) and SDH4 PLS-DA models (SI Figure 2) are provided. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Oliver, S. G.; Winson, M. K.; Kell, D. B.; Baganz, F. Systematic functional analysis of the yeast genome. Trends Biotechnol. 1998, 16 (9), 373–8. (2) Fernie, A. R.; Trethewey, R. N.; Krotzky, A. J.; Willmitzer, L. Metabolite profiling: from diagnostics to systems biology. Nat. Rev. Mol. Cell Biol. 2004, 5 (9), 763–9. (3) Raamsdonk, L. M.; Teusink, B.; Broadhurst, D.; Zhang, N.; Hayes, A.; Walsh, M. C.; Berden, J. A.; Brindle, K. M.; Kell, D. B.; Rowland, J. J.; Westerhoff, H. V.; van Dam, K.; Oliver, S. G. A functional genomics strategy that uses metabolome data to reveal the phenotype of silent mutations. Nat. Biotechnol. 2001, 19 (1), 45– 50. (4) Holmes, E.; Wilson, I. D.; Nicholson, J. K. Metabolic phenotyping in health and disease. Cell 2008, 134 (5), 714–7. (5) Kell, D. B.; Brown, M.; Davey, H. M.; Dunn, W. B.; Spasic, I.; Oliver, S. G. Metabolic footprinting and systems biology: the medium is the message. Nat. Rev. Microbiol. 2005, 3 (7), 557–65. (6) MacIntyre, D. A.; Jimeńez, B.; Lewintre, E. J.; Martin, C. R.; Scha¨fer, H.; Ballesteros, C. G.; Mayans, J. R.; Spraul, M.; Garcia-Conde, J.; Pineda-Lucena, A. Serum metabolome analysis by 1H-NMR reveals differences between chronic lymphocytic leukaemia molecular subgroups. Leukemia 2010, 24 (4), 788–97. (7) Brindle, J. T.; Antti, H.; Holmes, E.; Tranter, G.; Nicholson, J. K.; Bethell, H. W.; Clarke, S.; Schofield, P. M.; McKilligin, E.; Mosedale, D. E.; Grainger, D. J. Rapid and noninvasive diagnosis of the presence and severity of coronary heart disease using 1H-NMRbased metabonomics. Nat. Med. 2002, 8 (12), 1439–44. (8) Lanza, I. R.; Zhang, S.; Ward, L. E.; Karakelides, H.; Raftery, D.; Nair, K. S. Quantitative metabolomics by 1H-NMR and LC-MS/ MS confirms altered metabolic pathways in diabetes. PLoS ONE 2010, 5 (5), e10538. (9) Jansson, J.; Willing, B.; Lucio, M.; Fekete, A.; Dicksved, J.; Halfvarson, J.; Tysk, C.; Schmitt-Kopplin, P. Metabolomics reveals metabolic biomarkers of Crohn’s disease. PLoS ONE 2009, 4 (7), e6386. (10) Griffin, J. L.; Shockcor, J. P. Metabolic profiles of cancer cells. Nat. Rev. Cancer 2004, 4 (7), 551–561. (11) Allen, J.; Davey, H. M.; Broadhurst, D.; Heald, J. K.; Rowland, J. J.; Oliver, S. G.; Kell, D. B. High-throughput classification of yeast mutants for functional genomics using metabolic footprinting. Nat. Biotechnol. 2003, 21 (6), 692–6.

6738


Szeto et al. (12) Mapelli, V.; Olsson, L.; Nielsen, J. Metabolic footprinting in microbiology: methods and applications in functional genomics and biotechnology. Trends Biotechnol. 2008, 26 (9), 490–7. (13) Mas, S.; Villas-Boâs, S. G.; Hansen, M. E.; Åkesson, M.; Nielsen, J. A comparison of direct infusion MS and GC-MS for metabolic footprinting of yeast mutants. Biotechnol. Bioeng. 2007, 96 (5), 1014–22. (14) Kaderbhai, N. N.; Broadhurst, D. I.; Ellis, D. I.; Goodacre, R.; Kell, D. B. Functional genomics via metabolic footprinting: Monitoring metabolite secretion by Escherichia coli tryptophan metabolism mutants using FT-IR and direct injection electrospray mass spectrometry. Comp. Funct. Genomics 2003, 4 (4), 376–91. (15) Shaham, O.; Slate, N. G.; Goldberger, O.; Xu, Q.; Ramanathan, A.; Souza, A. L.; Clish, C. B.; Sims, K. B.; Mootha, V. K. A plasma signature of human mitochondrial disease revealed through metabolic profiling of spent media from cultured muscle cells. Proc. Natl. Acad. Sci. U.S.A. 2010, 107 (4), 1571–5. (16) Wallace, D. C. Mitochondrial diseases in man and mouse. Science 1999, 283 (5407), 1482–8. (17) Scheffler, I. E. Mitochondria; Wiley-Liss: New York, 1999; pp xiv, 367. (18) Scheffler, I. E. A century of mitochondrial research: achievements and perspectives. Mitochondrion 2000, 1, 3–31. (19) Wallace, D. C. A mitochondrial paradigm of metabolic and degenerative diseases, aging, and cancer: A dawn for evolutionary medicine. Annu. Rev. Genet. 2005, 39 (1), 359–407. (20) Thorburn, D. R. Mitochondrial disorders: prevalence, myths and advances. J. Inherit. Metab. Dis. 2004, 27 (3), 349–62. (21) Smeitink, J.; van den Heuvel, L.; DiMauro, S. The genetics and pathology of oxidative phosphorylation. Nat. Rev. Genet. 2001, 2 (5), 342–52. (22) Mancuso, M.; Orsucci, D.; Coppede, F.; Nesti, C.; Choub, A.; Siciliano, G. Diagnostic approach to mitochondrial disorders: the need for a reliable biomarker. Curr. Mol. Med. 2009, 9 (9), 1095– 107. (23) Lemire, B. D.; Oyedotun, K. S. The Saccharomyces cerevisiae mitochondrial succinate:ubiquinone oxidoreductase. Biochim. Biophys. Acta 2002, 1553 (1-2), 102–16. (24) Ackrell, B. A. C.; Johnson, M. K.; Gunsalus, R. P.; Cecchini, G. Structure and function of succinate dehydrogenase and fumarate reductase. In Chemistry and Biochemistry of Flavoenzymes; Mu ¨ller, F., Ed.; CRC Press, Inc.: Boca Raton, FL, 1992; Vol. III, pp 229-97. (25) Ha¨gerha¨ll, C. Succinate:quinone oxidoreductases: variations on a conserved theme. Biochim. Biophys. Acta 1997, 1320 (2), 107–41. (26) Hederstedt, L.; Ohnishi, T. Progress in succinate:quinone oxidoreductase research. In Molecular Mechanisms in Bioenergetics; Ernster, L., Ed; Elsevier Science Publishers: New York, 1992; pp 163-97. (27) Ackrell, B. A. Cytopathies involving mitochondrial complex II. Mol. Aspects Med. 2002, 23 (5), 369–84. (28) Bourgeron, T.; Rustin, P.; Chretien, D.; Birch-Machin, M.; Bourgeois, M.; Viegas-Pequignot, E.; Munnich, A.; Ro¨tig, A. Mutation of a nuclear succinate dehydrogenase gene results in mitochondrial respiratory chain deficiency. Nat. Genet. 1995, 11 (2), 144–9. (29) Rustin, P.; Ro¨tig, A. Inborn errors of complex II - Unusual human mitochondrial diseases. Biochim. Biophys. Acta 2002, 1553 (1-2), 117–22. (30) Burnichon, N.; Brie`re, J.-J.; Libe´, R.; Vescovo, L.; Rivie`re, J.; Tissier, F.; Jouanno, E.; Jeunemaitre, X.; Beńit, P.; Tzagoloff, A.; Rustin, P.; Bertherat, J.; Favier, J.; Gimenez-Roqueplo, A.-P. SDHA is a tumor suppressor gene causing paraganglioma. Hum. Mol. Genet. 2010, 19, 3011–20. (31) Sugimoto, J.; Shimohira, M.; Osawa, Y.; Matsubara, M.; Yamamoto, H.; Goto, Y.; Nonaka, I. A patient with mitochondrial myopathy associated with isolated succinate dehydrogenase deficiency. Brain Dev. 2000, 22 (3), 158–62. (32) Rustin, P.; Munnich, A.; Ro¨tig, A. Succinate dehydrogenase and human diseases: new insights into a well-known enzyme. Eur. J. Hum. Genet. 2002, 10 (5), 289–91. (33) Baysal, B. E.; Rubinstein, W. S.; Taschner, P. E. Phenotypic dichotomy in mitochondrial complex II genetic disorders. J. Mol. Med. 2001, 79 (9), 495–503. (34) Habano, W.; Sugai, T.; Nakamura, S.; Uesugi, N.; Higuchi, T.; Terashima, M.; Horiuchi, S. Reduced expression and loss of heterozygosity of the SDHD gene in colorectal and gastric cancer. Oncol. Rep. 2003, 10 (5), 1375–80. (35) Bayley, J. P.; van Minderhout, I.; Weiss, M. M.; Jansen, J. C.; Oomen, P. H.; Menko, F. H.; Pasini, B.; Ferrando, B.; Wong, N.; Alpert, L. C.; Williams, R.; Blair, E.; Devilee, P.; Taschner, P. E. Mutation analysis of SDHB and SDHC: novel germline mutations in sporadic

research articles


(36)

(37)

(38)

(39)

(40)

(41) (42)

(43)

(44)

(45) (46)

(47)

(48) (49) (50)

head and neck paraganglioma and familial paraganglioma and/ or pheochromocytoma. BMC Med. Genet. 2006, 7 (1), 1. Schiavi, F.; Boedeker, C. C.; Bausch, B.; Peczkowska, M.; Gomez, C. F.; Strassburg, T.; Pawlu, C.; Buchta, M.; Salzmann, M.; Hoffmann, M. M.; Berlis, A.; Brink, I.; Cybulla, M.; Muresan, M.; Walter, M. A.; Forrer, F.; Valimaki, M.; Kawecki, A.; Szutkowski, Z.; Schipper, J.; Walz, M. K.; Pigny, P.; Bauters, C.; Willet-Brozick, J. E.; Baysal, B. E.; Januszewicz, A.; Eng, C.; Opocher, G.; Neumann, H. P. Predictors and prevalence of paraganglioma syndrome associated with mutations of the SDHC gene. J. Am. Med. Assoc. 2005, 294 (16), 2057–63. Neumann, H. P.; Pawlu, C.; Peczkowska, M.; Bausch, B.; McWhinney, S. R.; Muresan, M.; Buchta, M.; Franke, G.; Klisch, J.; Bley, T. A.; Hoegerle, S.; Boedeker, C. C.; Opocher, G.; Schipper, J.; Januszewicz, A.; Eng, C. Distinct clinical features of paraganglioma syndromes associated with SDHB and SDHD gene mutations. J. Am. Med. Assoc. 2004, 292 (8), 943–51. Milunsky, J. M.; Maher, T. A.; Michels, V. V.; Milunsky, A. Novel mutations and the emergence of a common mutation in the SDHD gene causing familial paraganglioma. Am. J. Med. Genet. 2001, 100 (4), 311–4. Szeto, S. S.; Reinke, S. N.; Sykes, B. D.; Lemire, B. D. Ubiquinonebinding site mutations in the Saccharomyces cerevisiae succinate dehydrogenase generate superoxide and lead to the accumulation of succinate. J. Biol. Chem. 2007, 282 (37), 27518–26. Bayley, J. P.; Devilee, P.; Taschner, P. E. The SDH mutation database: an online resource for succinate dehydrogenase sequence variants involved in pheochromocytoma, paraganglioma and mitochondrial complex II deficiency. BMC Med. Genet. 2005, 6 (1), 39. Sun, F.; Huo, X.; Zhai, Y.; Wang, A.; Xu, J.; Su, D.; Bartlam, M.; Rao, Z. Crystal structure of mitochondrial respiratory membrane protein complex II. Cell 2005, 121 (7), 1043–57. Oyedotun, K. S.; Lemire, B. D. The quaternary structure of the Saccharomyces cerevisiae succinate dehydrogenase: Homology modeling, cofactor docking, and molecular dynamics simulation studies. J. Biol. Chem. 2004, 279 (10), 9424–31. Oyedotun, K. S.; Lemire, B. D. The carboxyl terminus of the Saccharomyces cerevisiae succinate dehydrogenase membrane subunit, Sdh4p, is necessary for ubiquinone reduction and enzyme stability. J. Biol. Chem. 1997, 272 (50), 31382–8. Oyedotun, K. S.; Lemire, B. D. The Saccharomyces cerevisiae succinate-ubiquinone oxidoreductase. Identification of Sdh3p amino acid residues involved in ubiquinone binding. J. Biol. Chem. 1999, 274 (34), 23956–62. Gietz, D.; St Jean, A.; Woods, R. A.; Schiestl, R. H. Improved method for high efficiency transformation of intact yeast cells. Nucleic Acids Res. 1992, 20 (6), 1425. de Kok, J.; Muller, J. L. M.; Slater, E. C. EPR studies on the respiratory chain of wild-type Saccharomyces cerevisiae and mutants with a deficiency in succinate dehydrogenase. Biochim. Biophys. Acta 1975, 387, 441–50. Reinke, S. N.; Hu, X.; Sykes, B. D.; Lemire, B. D. Caenorhabditis elegans diet significantly affects metabolic profile, mitochondrial DNA levels, lifespan and brood size. Mol. Genet. Metab. 2010, 100 (3), 274–82. Weljie, A. M.; Newton, J.; Mercier, P.; Carlson, E.; Slupsky, C. M. Targeted profiling: quantitative analysis of 1H NMR metabolomics data. Anal. Chem. 2006, 78 (13), 4430–42. Eriksson, L. J. E.; Kettaneh-Wold, N.; Trygg, J.; Wikstro¨m, C.; Wold, S. Multi- and megavariate data analysis. Part 1: Principles and applications, 2nd ed.; Umetrics Academy: Umeå, Sweden, 2001. Cann, A. J. Maths from scratch for biologists; John Wiley & Sons: Chichester, England, 2003.

(51) Oyedotun, K. S.; Lemire, B. D. The quinone-binding sites of the Saccharomyces cerevisiae succinate-ubiquinone oxidoreductase. J. Biol. Chem. 2001, 276, 16936–43. (52) Oyedotun, K. S.; Yau, P. F.; Lemire, B. D. Identification of the heme axial ligands in the cytochrome b562 of the Saccharomyces cerevisiae succinate dehydrogenase. J. Biol. Chem. 2004, 279 (10), 9432–9. (53) Romano, J. D.; Kolter, R. Pseudomonas-Saccharomyces interactions: influence of fungal metabolism on bacterial physiology and survival. J. Bacteriol. 2005, 187 (3), 940–8. (54) Kruckeberg, A. L.; Dickinson, J. R. Carbon metabolism. In The metabolism and molecular physiology of Saccharomyces cerevisiae, 2nd ed.; Dickinson, J. R., Schweizer, M., Eds.; CRC Press: Philadelphia, PA, 2004; pp 42-103. (55) Kispal, G.; Cseko, J.; Alkonyi, I.; Sandor, A. Isolation and characterization of carnitine acetyltransferase from S. cerevisiae. Biochim. Biophys. Acta 1991, 1085 (2), 217–22. (56) Kispal, G.; Sumegi, B.; Dietmeier, K.; Bock, I.; Gajdos, G.; Tomcsanyi, T.; Sandor, A. Cloning and sequencing of a cDNA encoding Saccharomyces cerevisiae carnitine acetyltransferase. Use of the cDNA in gene disruption studies. J. Biol. Chem. 1993, 268 (3), 1824–9. (57) Pronk, J. T.; Wenzel, T. J.; Luttik, M. A.; Klaassen, C. C.; Scheffers, W. A.; Steensma, H. Y.; van Dijken, J. P. Energetic aspects of glucose metabolism in a pyruvate-dehydrogenase-negative mutant of Saccharomyces cerevisiae. Microbiology 1994, 140 (Pt 3), 601–10. (58) Buu, L. M.; Chen, Y. C.; Lee, F. J. Functional characterization and localization of acetyl-CoA hydrolase, Ach1p, in Saccharomyces cerevisiae. J. Biol. Chem. 2003, 278 (19), 17203–9. (59) Boubekeur, S.; Bunoust, O.; Camougrand, N.; Castroviejo, M.; Rigoulet, M.; Gue´rin, B. A mitochondrial pyruvate dehydrogenase bypass in the yeast Saccharomyces cerevisiae. J. Biol. Chem. 1999, 274 (30), 21044–8. (60) Wolfe, A. J. The acetate switch. Microbiol. Mol. Biol. Rev. 2005, 69 (1), 12–50. (61) Rivie`re, L.; Moreau, P.; Allmann, S.; Hahn, M.; Biran, M.; Plazolles, N.; Franconi, J. M.; Boshart, M.; Bringaud, F. Acetate produced in the mitochondrion is the essential precursor for lipid biosynthesis in procyclic trypanosomes. Proc. Natl. Acad. Sci. U.S.A. 2009, 106 (31), 12694–9. (62) Margarida, C.; Sandra, P.; Odı´lia, Q.; Isabel, S.-S. Transport of carboxylic acids in yeasts. FEMS Microbiol. Rev. 2008, 32 (6), 974– 94. (63) Sandra, P.; Frederic, D.; Sonia, B.; Claude, J.; Margarida, C. Ady2p is essential for the acetate permease activity in the yeast Saccharomyces cerevisiae. Yeast 2004, 21 (3), 201–10. (64) Pollard, P. J.; Brie`re, J. J.; Alam, N. A.; Barwell, J.; Barclay, E.; Wortham, N. C.; Hunt, T.; Mitchell, M.; Olpin, S.; Moat, S. J.; Hargreaves, I. P.; Heales, S. J.; Chung, Y. L.; Griffiths, J. R.; Dalgleish, A.; McGrath, J. A.; Gleeson, M. J.; Hodgson, S. V.; Poulsom, R.; Rustin, P.; Tomlinson, I. P. Accumulation of Krebs cycle intermediates and over-expression of HIF1R in tumours which result from germline FH and SDH mutations. Hum. Mol. Genet. 2005, 14 (15), 2231–9. (65) Dickinson, J. R., Nitrogen metabolism. In The metabolism and molecular physiology of Saccharomyces cerevisiae, 2nd ed.; Dickinson, J. R., Schweizer, M., Eds.; CRC Press: Philadelphia, PA, 2004; pp 104-16. (66) Hazelwood, L. A.; Daran, J. M.; van Maris, A. J.; Pronk, J. T.; Dickinson, J. R. The Ehrlich pathway for fusel alcohol production: a century of research on Saccharomyces cerevisiae metabolism. Appl. Environ. Microbiol. 2008, 74 (8), 2259–66.

PR100880Y

Journal of Proteome Research • Vol. 9, No. 12, 2010 6739

Mutations in the Saccharomyces cerevisiae Succinate

Recommend Documents