2X SD minimal medium (13.4% yeast nitrogen base without amino acids, 100% glucose). The SD minimal medium was supplemented with 1.25 mg/l ergosterol, 0.164 g/l Tween 80, and 0.35 mg/l oleic acid, to satisfy the lipid requirements of the yeast cells during anaerobic growth. Acetoin was added at various concentrations (from 0 to 300 mM) at the start of fermentation, as previously described by Celton et al. [1].

At an OD600 of 3.0, corresponding to mid-exponential growth phase, the concentration of extracellular metabolites (glucose, glycerol, organic acids) was measured in the supernatant (these data will be used to constrain the stoichiometric model, as described below). Cells wer ... e harvested and hydrolyzed overnight with 6 M HCl for the determination of amino acids labeling patterns. One glucose derivative (glucose pentacetate) and two amino-acid derivatives (ethyl chloroformate (ECF) and dimethyl formamide dimethyl acetal (DMFDMA)) were analyzed by GC-MS [2], as previously described by Gombert et al. [3]. From the raw GC-MS data, the summed fractional labeling (SFL) of each fragment was calculated as follow: SFL = 100 × [(1.m 1 + 2.m 2 + …. + n.m n) × (m 0 + m 1 + m 2 + … + m n)-1, with m o the fractional abundance of the lowest corrected mass and mi > 0 the abundance of molecules with higher corrected masses. The labelling data are available in request.
The data set used for 13 C-flux analysis included the 25 SFLs calculated from labeling experiments and 22 measured fluxes, including the drain of metabolic intermediates to biomass and the formation of seven metabolites, giving a total of 47 items of experimental data. The distribution of carbon in central carbon metabolism (CCM) was estimated with the metabolic model described by Gombert et al.[3], modified to take into account features specific to fermentation [4]. The network used is described in additional file 2. Flux calculations (60 reactions) were carried out with Matlab 7, as previously described [5,3]. Differences between experimental and simulated SFLs and between experimental and simulated fluxes were minimized by an iterative procedure.
-----------------------------------References---------------------------------
[1] Celton M, Goelzer A, Camarasa C, Fromion V, Dequin S. Metab Eng. 2012, 14: 366-79. http://doi.org/f32tf6
[2] Christensen B, Nielsen J. Met Eng. 1999, 1: 282-290. http://doi.org/bjrp5z
[3] Gombert AK, dos Moreira Santos M, Christensen B, Nielsen J. J Bacteriol. 2001, 183: 1441-10. http://doi.org/crqbjx
[4] Heux S: Ingénierie métabolique et analyse13C-flux du métabolisme central des levuresSaccharomyces cerevisiaeœnologiques.PhD Thesis. 2006, University of Burgundy, Sciences de la vie et de la santé.
[5] Christensen B, Nielsen J. Met Eng. 1999, 1: 282-290. http://doi.org/bjrp5z