Fig. 1. Construction and properties of the chimerical transcription factor Pdr1*GAD. (A) The DNA-binding domain of Pdr1 was fused to the C-terminal domain of Gal4. The chimerical protein was designed to contain three N-terminal HA epitopes and the SV40 nuclear targeting signal (NLS). This chimerical gene was conditionally expressed under the control of the GAL1-10 promoter. (B) Time course expression of the protein Pdr1*GAD after galactose induction. The western blot shows two bands reduced to one single band, at the expected size, after phosphatase treatment, suggesting that the protein can be phosphorylated (data not shown). (C) Upon galactose induction, the cycloheximide resistance phenotype is similar for yeast strains deleted for both PDR1 and its functional homologue PDR3 (Δ1Δ3) expressing either the Pdr1-3 gain-of-function or the Pdr1*GAD chimera.

Fig. 2. Clustered display of upregulated genes from a time course production of Pdr1*GAD. Cluster analyses have been conducted and represented as indicated (Eisen et al., 1998). The first column represents the results obtained when the Pdr1-3 gain-of-function mutant was constitutively expressed in a strain deleted for both PDR1 and PDR3. The six subsequent columns present the upregulated genes at different times after galactose induction of the Pdr1*GAD construct (2, 4, 7, 10, 14, 18 h). All measurements are relative to the GAD control taken at the same time of induction. Genes were selected if their expression profile showed a continuous activation that goes beyond a 2-fold induction factor for at least one time-point. The colour scale ranges are indicated in the lower part. The early upregulated genes are presented in the upper part; they all contain at least one PDRE in their promoter. The two last lines correspond to ACT1 and PDA1 as internal controls, which are not sensitive to the production of Pdr1*GAD. The complete set of data is available at www.biologie.ens.fr/yeast-publi.html.

Fig. 3. Comparison of the upregulated genes in PDR1-3 and PDR1*GAD experiments. (A) Venn diagram showing the overlap between the PDR1-3 and PDR1*GAD upregulated genes in microarray experiments, and the set of yeast genes containing at least one Pdr1-binding site (PDRE) in their promoter. The names of the corresponding genes are available on the web site version of this figure (www.biologie.ens.fr/yeast-publi.html). Twenty-four genes are upregulated by both proteins. 191 other genes which have a PDRE element are neither regulated by Pdr1-3 nor by Pdr1*GAD. Respectively 2 and 3 PDRE-less genes are specific for PDR1-3 or PDR1*GAD expression. Four of them (HXK1, INO1, URA1, YER067w) could account for differences in culture and growth conditions between PDR1-3 and PDR1*GAD experiments and might be considered as indirect effects. The last one (YOR152c) shares its promoter with PDR5, the major target of Pdr1. (B) Promoter sequence analysis of the 24 genes upregulated by either Pdr1-3 or Pdr1*GAD reveals a core consensus sequence, which is TCCG(C/T)GGA. A systematic study of the promoter of the 23 genes represented in Figure 2 led to a more precise definition of the PDRE consensus. Note that the flanking regions are rather pyrimidine rich on one side (left) and purine rich on the other side (right).

Fig. 4. The microarray results were analysed by PCA (Dysik and Jonassen, 2001) for determining the two key variables represented on the two axes (PCA1 versus PCA2). This analysis clearly distinguishes the group of repressed genes (left, green) and the group of activated genes (right, red) from the bulk of the genes (central, blue). More interestingly, the upregulated genes are distributed along a gradient, which could reflect the existence of distinct subgroups of genes (see text). The expression profiles of repressed (A) and upregulated (B) genes are represented.