Changes in the ratios of mRNA subpopulation in response to MMS. The fraction of poly(A)⁺ mRNA transcripts from genes related to ribosomes (blue), protein degradation (white), amino acid metabolism (green), cell rescue (yellow), and all others (red) was calculated for untreated and MMS-treated cells by using the GeneChip hybridization data. Hybridization intensities are directly related to absolute poly(A)⁺ mRNA level (16). Therefore, hybridization intensities of a given mRNA reflect the abundance of a given mRNA relative to the total amount of mRNA. Genes are grouped into these categories according to the S. cerevisiae Genome Database and the Munich Information Centre for Protein Sequences (30, 31).

Fluorescence image of S. cerevisiae cRNA hybridization to GeneChip oligonucleotide arrays corresponding to ORFs from YEL001c to YJL088w (chromosomes 5–10; ref. 30) probed with 10 μg of biotin-labeled cRNA prepared from S. cerevisiae DBY747 log-phase cells untreated (A) or treated (B) with 0.1% MMS for 1 h. At this dose and time, there was a >90% survival rate. Differential hybridization between A and B represents ORFs expressed at different levels before and after MMS exposure. Examples of repressed (left arrow), induced (middle arrow), and nonresponsive (right arrow) ORFs are indicated. These ORFs correspond to RPS26B, YFL061W, and ACT1 respectively, and the hybridization differences were 3.6-fold, 251-fold, and 1-fold, respectively.

Verification of GeneChip data by conventional Northern-blot analysis. Northern blots were prepared with total RNA isolated from untreated and 0.1% MMS-treated S. cerevisiae DBY747 log-phase cells grown at 30°C in rich media; RNA was from cells exposed to MMS for the indicated times. Blots were probed with 50 ORFs; 6 of them are shown, namely two induced [YFL061W (251-fold) and SNZ1 (≈65.8-fold)], two repressed [DBP2 (18.1-fold) and PHO3 (12.3-fold)] and two nonresponsive ORFs [CDC19 and ACT1], as determined by GeneChip analysis. The GeneChip data in this figure are taken from the experiment in Fig. 1. Each ORF is represented on the Ye6100 array by ≈20 oligonucleotide pairs. One member of each pair corresponds to a perfectly matched sequence from the ORF (top row); the other pair member contains a single-base mismatch (bottom row). The difference in intensity between the perfectly matched and the mismatched sequences is used to calculate an average intensity difference for each ORF.

Correlation of GeneChip and Northern-blot data. The correlation between ORF induction and repression values (in response to MMS) obtained from the GeneChip analysis and Northern-blot data are compared for 50 ORFs. Northern-blot hybridization levels were determined with a Bio-Rad Molecular Imager, and the GeneChip data are from the experiment in Fig. 1. (A) Data plots for genes showing less than a 20-fold change in transcript level (by GeneChip). (B) Data plots for all 50 ORFs. The ORFs were selected to represent the entire range of change (as indicated by the GeneChip data) with changes from a 19-fold decrease to a 251-fold increase. The names and Northern-blot values are YFL061W (>150), SNZ1 (>150), GTT2 (180.0), YKL071W (118.0), SNO1 (60.0), YNR065C (42.0), YIL165C (17.3), SHM1 (9.8), ARG11 (8.7), GIN3 (8.6), ASP3 (6.7), NPL4 (6.1), RNR3 (6.0), ECM29 (5.5), YLR080W (5.3), HIS5 (5.1), YOR227W (5.1), MAG1 (5.0), NTH1 (4.5), YPR1 (4.0), HOM3 (4.0), LAG2 (3.5), CTK3 (3.0), YGR130C (2.8), LHE1 (1.4), YJL131C (1.2), SSA1 (1.0), PHO5 (−15.2), URA7 (−11.4), DBP2 (−8.7), NSB1 (−8.2), PHO3 (−6.2), AAH1 (−5.3), GUA1 (5.1), SAH1 (−5.0), RKI1 (−4.0), YDL213C (−3.1), RPS16A (−3.0), NIP7 (−2.8), YIL158W (2.6), ANB1 (−2.5), TCM1 (−2.3), ENO1 (−2.2), INH1 (−2.0), RPL16A (−2.0), SPE2 (−1.7), CDC19 (−1.6), RPL17 (−1.6), YLR009W (−1.4), and ACT1 (−1.1). The GeneChip values are in Table 1. Dashed red lines represent a factor-of-two difference from a perfect match between Northern-blot and GeneChip analysis. The asterisk in A represents the PHO5 ORF (discussed in Results and Discussion) and ^† marks transcripts that were undetectable by Northern-blot analysis in untreated cells and whose fold-induction is likely to be less accurate. Note that these points were not included in the calculation of the correlation coefficient.