(a) R² values (goodness-of-fit) for a linear-fit to the log₂ mRNA levels at the four time points (see inset for example of a single gene). As control, we performed the same analysis to 10,000 shuffled profiles in which each time-point is taken from a different gene (randomly selected), thus retaining the overall degradation of mRNA levels but shuffling the gene-specific degradation rates. 78% of the real profiles (compared with 18% of the shuffled profiles) obtained an R² value above 0.94 and were included in all further analyses. (b) Correlation of S. cerevisiae (blue) and S. paradoxus (red) mRNA degradation rates: (i) between different probes for the same genes (note that different probes typically have different hybridization intensities, yet the mRNA degradation rates are highly reproducible, see Materials and Methods), (ii) between biological repeat experiments, and (iii) between this work and a previous work that used a temperature-sensitive mutation in RNA polymerase II to block transcription. Note that although this previous work analyzed only S. cerevisiae, it has high correlations with our data for the two species. (c) Scatter-plot of mRNA degradation rates in S. cerevisiae and S. paradoxus, which have a genome-wide correlation of 0.78. (d) Patterns of mRNA degradation for the 12 oxidative phosphorylation genes included in the analysis in S. cerevisiae (blue) and S. paradoxus (red).

Classification of inter-species differences in mRNA degradation rates into cis and trans based on the extent of differences observed between the two hybrid alleles (see Materials and Methods). (a) Heatmap of the differences in mRNA degradation rates, log₂(S. cer/S. par), between the two species (left column), between the corresponding hybrid alleles (middle columns, reflecting only the cis component), and the subtraction of the species and hybrid differences (right columns, reflecting only the trans component). (b) mRNA degradation profiles of the two species (left) and the corresponding hybrid alleles (right) are shown for two examples of cis-differences (top) and one example of trans-difference (bottom).

(a) Target-sets of various TFs [31], RNA-binding proteins [14], and two complexes implicated in both transcription and degradation (Rbp4/7 [32] and Ccr4-Not [33]) were examined for enrichment with trans-coupled genes. 15 and 10 datasets had significant enrichment below a p value of 0.05 (full line) and 0.01 (dashed line), respectively, and these are shown in order of statistical significance. The total numbers of analyzed datasets and those with significant enrichments are shown in parentheses. (b) Same as in (a) after excluding targets of Rpb4, Ccr4, and Not5. (c) Same as in (a) for enrichment with cis-coupled genes. (d) Diverged TF binding [31] (red) or mRNA stability [15] (blue) motifs, which are intact only in S. cerevisiae (S. cer sites) or only in S. paradoxus (S. par sites), were identified by sequence analysis. The enrichment of diverged motifs (for all TFs combined or all stability motifs combined) was examined among all cis-coupled genes (All), cis-coupled genes predicted to be targets (Rpb4) or non-targets (All-Rpb4) of Rpb4, and for cis-coupled S. cer sites or cis-coupled S. par sites. (e) Same as (d) for enrichment with trans-coupled genes.

(a) Sliding window analysis (windows of 200 genes) for the percentage of inter-species differentially expressed genes (above 1.5-fold), using either mRNA levels (black) or estimated transcription rates (gray), as a function of the fold-change of inter-species differences in mRNA degradation rates. Dashed lines indicate the genome-wide percentage of differential mRNA levels (black) or transcription rates (gray). The green section includes small differences in mRNA degradation which may reflect technical variations, while the red section includes larger and biologically meaningful differences in mRNA degradation. (b) Scatter-plot of differential mRNA degradation rates versus differential mRNA levels for genes with (full circles) or without (empty circles) significant difference in mRNA degradation, and for oxidative phosphorylation genes (blue). The number of genes with significant difference in mRNA degradation is shown for each quarter, demonstrating an enrichment of genes with opposite effects of mRNA degradation and mRNA levels (genes with higher degradation in S. cerevisiae also tend to have higher mRNA levels, as the upper-right quarter has more genes than the lower-right quarter). (c) Sliding window analysis (windows of 200 genes) for the percentage of genes with opposite effects of transcription and degradation among those with differential mRNA degradation rate and either differential mRNA levels (black) or differential transcription rate (gray), as a function of the fold-change of inter-species differences in mRNA degradation rates. Dashed line indicates 50% opposite effects, as would be expected by chance if differential expression and differential degradation are independent. Green and red sections are as in (a). (d) Pie charts for the different combinations of differences in transcription and mRNA degradation, among the genes with differential mRNA degradation (right) and the genes with differential transcription (left). The analysis was performed with conservative estimates of coupling (only those with coupling as defined both by analyses of mRNA levels and by analysis of estimated transcription rates), while the percentages in parentheses show the results of a more relaxed analysis, in which either mRNA levels or transcription rates were sufficient to define coupling.

Inter-species differences in mRNA levels (or estimated transcription rates) and mRNA degradation were divided into the contribution of cis- and trans-mutations based on the hybrid data. The enrichment of opposite transcription and degradation effects was examined for each of the four combinations of cis and trans, by a sliding window analysis of the percentage of opposite effects as a function of the fold-changes in mRNA degradation.

The first model (Parallel coupling, left) assumes mutations in a single trans-factor that influences both processes and is consistent with the enrichment of trans-coupled genes with targets of Rpb4, Ccr4-Not, Pab1, and Npl3. The second model (Sequential coupling, right) assumes mutations that exert transcriptional effects (either in cis or in trans) and that these transcriptional effects then induce changes in mRNA degradation, for example, through a shuttling mechanism whereby Rpb4/7 (or other transcription-related molecules) binds to the mRNA co-transcriptionally and transports with it to the cytoplasm. This model is consistent with the enrichment of diverged TF motifs among cis-coupled genes. An opposite sequential coupling is also possible (dashed arrows), whereby mutations affect mRNA degradation and this effect then induces transcriptional changes, yet we do not find evidence to support it.