Yeast strains: Fermentation assays were carried out with two industrial yeast strains selected by Fermentec, Piracicaba, Brazil, and named PE2 and CAT1 (Basso et al. 2008). These strains were multiplied in low sugar must of sugar cane molasses diluted with water until to reach a cellular biomass to start the fermentation recycles. These strains are available under request.
Fermentation recycles: The fed-batch fermentations with yeast cell recycle were carried out in bench scale reproducing all steps that occur in a typical industrial process of alcoholic fermentation in Brazil as demonstrated in Figure 1A. Fermentations were carried out in triplicate for each strain. Multiplication of yeast cells is carried out with must of sugar cane molasses diluted with water (6% of sugars) while the fermentation recycles are carried out with with sugar cane must containing 18% sugars, being 50% from molasses and 50% from sugar cane juice, as described in the paper. After centrifugation, the yeast cells were resuspended in water and fed with sugarcane must containing 18 % sugars, being 50 % from molasses and 50% from sugarcane juice. Fermentations were conducted at 33 oC for 9 hours containing 10 % of yeast cells (w/v; about 109 cells/ml). At the end of the first cycle fermentative yeasts were centrifuged again and the cream of yeast treated with sulfuric acid to reach pH 2.5 for 1.5 hours at room temperature. The objective of this acid treatment is do reduce the bacterial contamination before another fermentation cycle (Basso et al. 2008). After the acid treatment, yeast cells were fed with sugar cane must to start a new fermentation. In the second cycle fermentation samples were collected for analysis of transcriptome. The sampling of fermented must containing yeast cells was made in three steps: after acid treatment (before feeding), termination of must feeding (three hours) and at the end of fermentation (nine hours). In each step, 1 ml was collected from each fermentation flask. The samples were immediately frozen in liquid nitrogen and stored at ultra low temperature (- 70 oC) for further analysis of transcriptome.
Analysis of fermentation parameters: The main parameters related to yeast cells and fermentation processes analyzed were: (i) yeast cell viability, by light microscopy (Nikkon E200) based on the colorimetric method of living/dead cells by erythrosine; (ii) fermentation rate, by loss of weight (CO2 released at each 1.5 hours); (iii) content of residual sugars (glucose, fructose and sucrose) and glycerol concentration in the fermented must at the end of fermentation [by High-Performance Anion Exchange Chromatography with Pulse Amperometric Detection (HPAEC-PAD, Dionex, model BioLC equipped with a PA-1 column)], and (iv) alcohol content in the fermented must, measured by vapor extraction followed by the measurement of the density of the distilled with a digital densimeter (Anton Paar, model DMA 4500) and estimation of the ethanol yield in relation to total sugars in the fermented must.
Sporulation and tetrad dissection: The diploid cells were patched in potassium acetate medium (potassium acetate, 10 g; yeast
extract, 1 g; glucose, 0.5 g; agar, 20 g) plates for 2 days at 30 °C. The cells were examined under microscope to look at tetrads. The patches were scrapped with the loop into 20 ïl of beta-glucoronidase and centrifuged at 30 oC for 15 minutes. Two hundred microliters of milli-Q water were added and 10-20 ïl were poured onto the edge of a dry and flat YPD plate. The solution was allowed to dry into the plate. Using a SINGER micromanipulator, the four spores were placed far enough away from each other in a way that the colonies would not grow into each other, using the grid 10 x 8 to dissect the
tetrads. The plate was incubated at 30 oC for 2 days.
RNA isolation and real-time PCR: For total RNA isolation, the yeast cells were disrupted by vortexing with glass beads and total RNA was extracted with Trizol reagent (Invitrogen, USA). Ten micrograms of RNA from each treatment were then fractionated in 2.2 M formaldehyde, 1.2 % w/v agarose gel, stained with ethidium bromide, and then visualized with UV-light. The presence of intact 25S and 18S ribosomal RNA bands was used as a criterion to assess the integrity of the RNA. RNAse free DNAse treatment was carried out as previously described (Semighini et al. 2006). After this, the total RNA was purification by RNeasy® Mini Kit (Qiagen) and the purified samples were measured in the NanoDrop® 2000 - Thermo Scientific (Uniscience).
RT-PCR detection of spliced HAC1 mRNA: RNA was isolated from wild-type (S288c), CAT-1, and PE-2, treated with DNAse RNAse-free, and RT-PCR was performed (Semighini et al. 2006). Equal concentrations of cDNA were subjected to 25 cycles of PCR amplification using PCR primers that recognize HAC1 cDNA 343 bp upstream of the intron (5'-TCAAGGGATTTCCAGAGC-3') and 56 bp downstream of the intron (5'-TCATGAAGTGATGAAGAAATCATT-3'). Because the HAC1 intron is 252 bp in length, unspliced HAC1 PCR product (HAC1u) is 651 bp, and spliced HAC1 PCR product (HAC1i) is 399 bp (Bicknell et al. 2007). The images generated were subjected to densitometric analysis of pixel intensity using the ImageJ software (http://rsbweb.nih.gov/ij/index.html).
Microarray hybridization: For gene expression analysis commercially-available Agilent whole genome S. cerevisiae microarray [Yeast (V2) Gene Expression Microarray, 8x15K] was used. The microarray slides contain 15,208 probes for S. cerevisiae (S288C strain). The RNA samples were extracted with Trizol (Invitrogen, USA) and purified using the RNAeasy kit (Qiagen, Germany). cRNA labeling was performed according to the standard protocol described by Agilient using Two-Color Microarray-Based Gene Expression Analysis (Agilent Technologies, USA). Briefly, for cRNA synthesis and labeling 4-5 µg of total RNA was used. After labeling, 300 ng of Cy3 and Cy5-labelled cRNAs (specific activity >8.0 pmol Cy3-Cy5/µg cRNA) was fragmented at 60°C for 30 minutes in a reaction volume of 25 µl containing 1x Agilent fragmentation buffer and 2x Agilent blocking agent following the manufacturer's instructions. On completion of the fragmentation reaction, 25 µl of 2x Agilent hybridization buffer was added to the fragmentation mixture and hybridized to the S. cerevisiae microarrays slides for 17 hours at 65°C in an Agilent G2545A Hybridization Oven and on Agilent Rotator Rack. After hybridization, microarrays were sequentially washed: 1 minute at room temperature with GE Wash Buffer 1 (Agilent) and 1 minute with 37°C GE Wash buffer 2 (Agilent), then a 10 seconds Acetonitrile Wash (Agilent) followed by a 30 seconds Stabilization and Drying Solution wash (Agilent). Slides were immediately subjected to fluorescent detection using fluorescent detection with a GenePix 4000B (Molecular Devices, USA) with simultaneously scanning the Cy3 and Cy5 channels at an resolution of 5 µm. Laser was set at 100% and PMT gain was adjusted automatically for each slide using the program GenePix Pro (Molecular Device) according to the signal intensity of each array. Merged Cy3 and Cy5 TIFF images generated by the GenePix Pro were used to analysis in the Agilent Feature Extraction software (version 9.5.3.1, Agilent) using Linear Lowess algorithm to obtain background subtracted and normalized intensity values. The dye-normalyzed values generated in the Feature Extraction data files were used to upload the software Express Converter (version 2.1, TM4 available at http://www.tm4.org/utilities.html) which conveniently converts the Agilent file format to mev (Multi Experiment Viewer), a file format compatible to the TM4 softwares for microarray analysis (available at http://www.tm4.org/). The mev files were then uploaded in the MIDAS software where the resulting data were averaged from replicated genes on each array, from dye-swap hybridizations for each experiment and from two biological replicates using the tools "flip dye consistency checking" and "in slides replicates analysis" implemented in MIDAS software. The mev files generated were them loaded in MEV software (Multi Experiment Viewer) where differentially expressed genes was identified using one-class t-test (p>0.01). Significantly different genes were those whose mean log2 expression ratio over all included samples was statistically different from zero which indicates the absence of gene modulation. The resulting data were visualized based on similar expression vectors using Euclidean distance and hierarchical clustering with average linkage clustering method and K-means to group the genes in clusters to view the whole data set.
Physical protein-protein interactions (PPPI) network design and global topological analysis: The transcriptomic data gathered from yeast CAT-1 and PE-2 strains was used to obtain information about how the under- and overexpressed genes and their products interact in the context of physical protein-protein interactions (PPPI networks) in S. cerevisiae. In this sense, the data mining screening and network design of repressed or induced genes-associated PPPI networks was performed using Cytoscape software, version 2.6.3 (Shannon et al. 2003). For this purpose, we used the PPPI data of S. cerevisiae available in the Saccharomyces Genome Database (http://www.yeastgenome.org). The induced and repressed PPPI networks obtained from this first screening were then combined in a unique PPPI network by employing the union function of the Cytoscape core plugin Merge Networks (Figure 4). The union PPPI network was then analyzed with Biological Network Manager (BiNoM), a Cytoscape plugin available at http:// http://bioinfo-out.curie.fr/projects/binom in order to detect the strongly connected cluster (SCC). The degree of data overlapping between induced- and repressed-associated PPPI networks was obtained from an area-proportional Venn diagram analysis, available at http://bioinforx.com/free/bxarrays/overlap.php.
Network centralities and local topological analyses: Two major network centralities (node degree and betweenness) were computed from the SCC network using the Cytoscape plugin CentiScaPe 1.0 (Scardoni et al. 2009). The local topology of the network, defined as bottlenecks, was obtained from the threshold generated by each centrality calculated by CentiScape 1.0. In this sense, bottlenecks were defined as nodes with a value above the threshold calculated for node degree and betweenness. A bottleneck subnetwork containing was drawn using Cytoscape 2.6.3.
Gene ontology analysis: Gene ontology (GO) clustering analysis was performed using Biological Network Gene Ontology (BiNGO) (Maere et al. 2005) software, a Cytoscape plugin available at http://chianti.ucsd.edu/cyto_web/plugins/index.php. The degree of functional enrichment for a given cluster and category was quantitatively assessed (p- value) by hypergeometric distribution (Rivals et al. 2007) and a multiple test correction was applied using the false discovery rate (FDR) (Benjamini and Hochberg 1995) algorithm, fully implemented in BiNGO software. Overrepresented biological process categories were generated after FDR correction, with a significance level of 0.05.
