Yeast strains and growth conditions

All the strains are from Euroscarf collection. They derived from the Saccharomyces cerevisiae strain BY4742. Cells were grown in YPD medium (1% (w/v) bacto-yeast extract, 2% (w/v) bacto-peptone, 2% (w/v) glucose). Drug resistance assays were performed by tests with serial dilutions on YPD liquid medium in order to establish the most appropriate concentration of benomyl.

• Time-course experiments and microarray experiments

Different strains were grown in YPD liquid medium to an OD600 0.5. Addition of benomyl to 20 µg/ml or only DMSO to the same concentration was performed for various times (30 sec, 2 min, 4 min, 10 min, 20 min, and 40 min) starting with the same initial culture. Cells were harvested instantaneously by centrifugation, flash-frozen in liquid nitrogen, and stored at -80°C. Total RNA was isolated, purified and used to synthesize and to label cDNA as described on web site. For each time of exposure to benomyl and for each analysed strain, we performed 2 independent microarray experiments and two technical repetitions. Microarrays containing oligonucleotides to probe most of the open reading frames of S. cerevisiae were produced in our laboratory with an Omnigrid II Biorobotics robot. They were based on the principle of 40 mer oligonucleotides from MWG (www.mwg-biotech.com) covalent deposited onto Corning glass slides coated with pure gamma amino propil silane ( www.corning.com/lifesciences). The microarray protocol used is described on our web site. A total of 20 mg total purified RNA was used for each experiment. In each experiment, the cDNA corresponding to cells treated with benomyl was labelled with CY5-dUTP and cDNA from control cells treated with DMSO was labelled with CY3-dUTP. Dye swap experiments were carried out realized for each analysis. The arrays were read using a Genepix 4000A scanner (Axon) and analyzed with Genepix 3.0 software. Artefactual, saturated or low-signal spots were eliminated from the analysis. Linear regression of Cy5 against Cy3 channels (Marc and Jacq 2002) was applied to data in order to normalize fluorochrome channels. Duplicate spots were averaged, and values with an important log2(Ratio) difference were not used for the rest of the analysis.

- Missing value handling :
Gene expression patterns whose the minimum percentage of existing values was less than 80 % were eliminated from the rest of the analysis. The remaining missing values were replacing using the KNN-imputation method (Troyanskaya, Cantor et al. 2001).

- Selection of up- and down-regulated genes :
Log2(Ratio) data sets of times courses were used for filtering genes for which a more than two fold change is observed, at least twice and for two successive times along the course of the experiments.

• Chromatin immunoprecipitation analyses (ChIP assays)

YAP1 gene with N-terminal Myc tag on pRS316 plasmid was introduced in wild type yeast strain producing Yap1 under the control of the its own promoter. Cells were grown on YPD medium to an O.D. of 0.5. Addition of benomyl or DMSO to 20 µg/ml was performed for 5 min, 15 min, 40 min, and 60 min, starting with the same initial culture. DNA-binding proteins were cross-linked to DNA with formaldehyde in vivo (treatment for 15 min at room temperature). The chromatin was isolated by grinding with glass beads followed by sonication to shear DNA along with bound proteins into small fragments (1-3 kb). A 20 ml aliquot of the lysate was saved as the input fraction. To isolate the DNA-Yap1 complex, the samples were incubated with mouse anti-HA monoclonal antibody (Babco), and with UltraLink Immobilized Protein G (Pierce). Immunoprecipitates were extensively washed and centrifuged to recover a pellet (bound) and supernatant (unbound). Protein was eluted from the sepharose beads by heating at 95°C for 20 min. Cross-links were reversed by heating at 65°C for 6 h. The DNA was purified with PCR QiaQuick columns prior to PCR analysis. Both input and bound DNA were dissolved in 40 µl TE. 1µl of DNA preparation from each reaction was used for PCR. The Qiagen QuantiTect SYBR Green PCR kit was used for quantitative real-time PCR. We used LightCycler3 and repeated experiments at least twice. The various primer sets used for amplification of the 14 promoters are as followed :



The enrichment factor was calculated as follows :
1) the values of immunoprecipitated amplified DNA were divided by those for the negative control (a non coding fragment of chromosome V) included in each analysis and their mean values retained as the crude promoter-occupancy value
2) the ratio of amplified unbound DNA to negative control constituted the background value. The enrichment factor was obtained by dividing the crude promoter-occupancy value by the background value.

• Bioinformatic analyses

- Principal Component Analysis :
The analysis was completed with the statistical computing and graphics environment R http://cran.r-project.org/.

- KO Gene Cluster Analysis (KOGCA) :
The differential gene expression ratio induced by benomyl between the wild type and yap1Delta strains were plotted as a X (log2(Ratio) value observed in the wild type strain) versus Y (log2(Ratio) value observed in the yap1Delta strain) for the different time points (Figure 2, A).
When all the times were plotted on the same graph, lines could be drawn between points corresponding to the same gene. The resulting "profiles" were therefore characteristics of the differential expression of the considering genes along all the time course analysis. Similarity measure between the differential gene expression profiles can be calculated and cluster analysis was conducted. The resulting tree is finally split into several groups of genes (Figure 2, B) joined together (mean value) in the two dimensional graph (Figure 2, C).
Graphical representation, distance computation as well as hierarchical clustering were performed with a script to utilize library function in R http://cran.r-project.org/.

- REDUCE algorithm :
REDUCE was used as indicated in (Roven and Bussemaker 2003). REDUCE is a motif-based regression method for microarray analysis. This algorithm uses unbiased statistics to identify oligonucleotide motifs whose occurrence in the regulatory region of a gene correlates with the level of mRNA expression. Regression analysis is used to infer the activity of the transcriptional module associated with each motif. REDUCE is available online at http://bussemaker.bio.columbia.edu/reduce/.