Organisms and cultivation
S. cerevisiae ATCC 32167 was purchased from ATCC. S. cerevisiae deletion mutants with the parental phenotype BY4742 Matα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 were obtained from Open Biosystems (Heidelberg, Germany). The derived yeast deletion mutants Δlsm1 Δpat1 and Δdhh1 are resistant to the antibiotic geneticin and exhibit auxothrophies for histidine, leucine, lysine and uracil. Yeast cells were grown on YPD-agar plates (10 g/l yeast extract, 20 g/l peptone, 20 g/l glucose) containing 200 mg/l Geneticin. Cultivation was carried out first on complex YPD-medium at 30°C followed by a second pre-culture on defined synthetic medium containing (NH)4HPO4 1 g/l, (NH)4SO4 8.75 g/l, MgSO4·7 H2O 1 g/l, citric acid 1.1 g/l, CaCl2·2 H2O 0.15 g/l, glucose 20 g/l, 0.5 M Na-phosphate buffer (pH 6.0) 100 ml/l, 100× trace element solution 10 ml/l and 100× vitamin solution 10 ml/l. Lysine, leucine (each 120 mg/l), histidine and uracil (each 80 mg/l) were added from 100× stock solutions to allow growth of the deletion mutants. 100× vitamin solution consists of myo-inositol 301.5 mg/l, Ca-panthotenate 150 mg/l, thiamin/HCl 30 mg/l, pyridoxine/HCl 7,5 mg/l and biotin 0,15 mg/l. 100× trace element solution consists of FeCl3·6 H2O 75 mg/l, MnSO4·H2O 53 mg/l, ZnSO4·7 H2O 45 mg/l and CuSO4·5 H2O 12 mg/l . The main culture was carried out in defined synthetic medium using 1000 ml baffled shake flasks. The same medium without additional amino acids and uracil was also used for the cultivation of S. cerevisiae ATCC 32167 that is the wild type of the used reference strain BY4742.
For inter-strain comparison of intracellular metabolite concentrations, two parallel cultures were grown, and three samples were taken during the exponential growth phase. Cells were harvested at optical densities between 2 and 3. Exponential growth was observed for at least 30 min after taking samples for metabolite extraction to ensure that no changes in cell metabolism occurred, due to substrate limitation effects.
Cell dry weight (CDW) was determined gravimetrically after washing cells twice with double distilled water. After centrifugation (10 min, 8,000 × g, 4°C, Biofuge stratos, Heraeus, Hanau, Germany) the drying was performed at 80°C until constant weight was observed. Optical density (OD660 nm) was determined at 660 nm (Novaspec II, Pharmacia Biotech, Freiburg, Germany). The linear correlation between CDW and OD660 is CDW [g/l] = 0.51 [g/l]·OD660 nm. OD-CDW correlations of the four strains were identical.
Concentrations of extracellular glycerol, glucose and ethanol were determined by HPLC (Kroma System, Kontron Instruments, Neufahrn, Germany) with an Aminex HPX-87H column (300 × 7,8 mm; Bio-Rad, Hercules, USA) and an isocratic flow of 0.5 ml/min 22 mM H2SO4 at 35°C. Glucose and ethanol concentrations in the supernatant were determined using enzymatic kits (Boehringer Mannheim, R-Biopharm GmbH, Darmstadt, Germany) in addition to HPLC analysis. Intracellular amino acid concentrations were determined by HPLC using an OPA method .
Quenching and metabolite extraction
For metabolome analysis three independent samples of 10 ml were drawn from each of the two cultures during the exponential growth phase and transferred immediately under vigorous shaking to a 50 ml falcon tube, filled with pre-cooled quenching solution (-40°C, 60% methanol, 10 mM HEPES, pH 7.5). Separation of cells and medium was performed by centrifugation (5 min, 8,000 × g, -19°C, Biofuge stratos, Heraeus, Hanau, Germany) . The temperature was always below -25°C during the whole quenching process. After an additional washing step the metabolite extraction was achieved by incubating the pellet with 2 ml of boiling water for 15 min. Cell-debris was separated by centrifugation (10 min, 1,500 × g, 4°C, Labofuge 400R, Heraeus, Hanau, Germany) and extracted again with 750 μl chloroform (37°C, 10 min). The two phases were combined and separated again by centrifugation. The polar phase was further analyzed. The sampling process until quenching was not oxygen limited. This was ensured by measuring dissolved oxygen (DO) concentration using an optical method . DO concentrations were always above 30% of air saturation.
GC/MS analysis of intracellular metabolites
Prior to GC/MS analysis 100 μl of the cell extract was lyophilized (Lyovac GT2, GEA Lyophil GmbH, Huerth, Germany). The remaining powder was derivatised using 25 μl 20 g/l methoxylamine/HCl in pyridine for oximation (30 min, 80°C) followed by 50 μl N-methyl-N-trimethylsilyl-trifluoroacetamide (MSTFA) for silylation. Sample analysis was carried out on a HP6890 GC System using a Mass Selective Detector 5973 (Agilent Technologies, Waldbronn, Germany). Full scan mass spectra were acquired ranging from 30 to 600 amu using a scan rate of 9 scans/s. Separation was performed using a 60 m × 0.2 mm I.D. fused silica HP-5 ms column (Agilent Technologies, Waldbronn, Germany) and a helium carrier gas flow rate of 0.7 ml/min. After 1 min at 70°C the temperature was increased to 75°C (1°C/min) followed by a second increase to 315°C (5°C/min) and finally to 340°C (25°C/min). 1μl of derivatised sample was injected splitless for 2 min using a PTV with a temperature gradient from 75°C to 340°C with a rate of 360°C/min. injector. Temperatures of ion source and transfer liner were 200°C and 240°C. For electron impact ion generation a 70-eV electron beam was used. Intracellular amino acid concentrations were determined by HPLC using the OPA method .
Metabolite identification and deconvolution
Metabolite identification and signal deconvolution was performed using the AMDIS software (version 2.64, National Institute of Standards and Technology, Gaithersburg, USA) as well as the Enhanced ChemStation software (Agilent Technologies, Waldbronn, Germany). Identification of metabolites is based on the comparison of measured mass spectra and retention indices (RI) with in-house library entries  of pure component mass spectra and retention indices or the National Institute of Standards and Technology (NIST) library. Retention indices were scaled using a standard of 33 hydrocarbons ranging from C8 to C40. To determine the differences among the reference and the deletion strains a relative quantification is sufficient. Relative quantification of metabolites was carried out using peak areas of characteristic m/z ratio intensities of these metabolites that were normalized to the integral of the total intensity over the whole chromatogram.
PLS modeling and PCA
To find statistically significant differences between the metabolite profiles of the reference strain and the three deletion strains, the partial least squares to latent structures (PLS) regression method was applied . This is commonly used in the analysis of data from multivariate studies to find significant relationships between different classes, here the reference strain and deletion strains. The PLS estimation resulted in three sets of model parameters each one comparing one of the deletion strains with the reference strain. Calculation of the model parameters was carried out using SIMCA-P 11.5+ (Umetrics, Malmö, Sweden). To reduce the impact of artifacts and noise during the modeling process, pareto scaling was used for all variables. For data visualization and selection of interesting metabolites loading plots with jack-knifed confidence intervals were chosen as described by Wiklund et al. .
Euclidian distances between different strains were calculated using Matlab (version 22.214.171.124, The Math Works, Inc., Boston, USA). The Euclidian distance d(x, y) between two points x = (x1
) and y = (y1
) in a n-dimensional space is defined as
For each knock-out strain the Euclidian distance to the reference strain was calculated and visualized in the form of a dendrogram. Each vector includes the normalized areas of the corresponding metabolites which are statistically significant based on the PLS-DA (PLS-discriminant analysis).
Peptone was obtained form Bacto (Sparks, USA). Yeast Extract and nutrient agar were obtained from Difco (Sparks, USA). MSTFA was obtained form Macherey-Nagel (Düren, Germany). All other chemicals used were of analytical grade and obtained either from Sigma-Aldrich (St. Louis, USA) or Fluka (Buchs, Switzerland).