The Gcn4 transcription factor reduces protein synthesis capacity and extends yeast lifespan

In Saccharomyces cerevisiae, deletion of large ribosomal subunit protein-encoding genes increases the replicative lifespan in a Gcn4-dependent manner. However, how Gcn4, a key transcriptional activator of amino acid biosynthesis genes, increases lifespan, is unknown. Here we show that Gcn4 acts as a repressor of protein synthesis. By analyzing the messenger RNA and protein abundance, ribosome occupancy and protein synthesis rate in various yeast strains, we demonstrate that Gcn4 is sufficient to reduce protein synthesis and increase yeast lifespan. Chromatin immunoprecipitation reveals Gcn4 binding not only at genes that are activated, but also at genes, some encoding ribosomal proteins, that are repressed upon Gcn4 overexpression. The promoters of repressed genes contain Rap1 binding motifs. Our data suggest that Gcn4 is a central regulator of protein synthesis under multiple perturbations, including ribosomal protein gene deletions, calorie restriction, and rapamycin treatment, and provide an explanation for its role in longevity and stress response.

Shown are also the P-values in the two-sided Mann Whitney U test comparing mRNA fold-changes of genes belonging to a given ribosomal subunit (small or large) and those of all other genes. Boxes extend from the 25th to 75th percentiles (interquartile range (IQR)), horizontal lines represent the median, whiskers indicate the lowest and highest datum within 1.5*IQR from the lower and upper quartiles, respectively.

Supplementary Fig. 2. Polysome profiles reveal features of translation in individual strains
(a to f) Representative polysome profiles for (a) WT, (b and c) long-lived RPKOs and (d, e and f) short-lived RPKOs. The profiles of long-lived strains show the presence of half-mers (b and c), lower 60S-to-40S ratio, and reduced number of distinguishable polysomes peaks compared to the wild type strain (a). Short-lived strains resulting from deletion of components of the large ribosomal subunit show higher monosome and polysome peaks (d and e) compared to the wild type strain (a). A notable increase in the 60S-to-40S ratio was registered in the short-lived strain resulting from deletion of an RP from the small subunit (f).

Supplementary Fig. 3. Generalized uORFs skipping in long-lived RPKO strains (a to e) Ration of uORFs-to-CDS
and (e) Δrpl9a strains compared to the wild type strain. Each dot corresponds to a gene containing at least one uORF. The GCN4 gene is highlighted in orange. (f) Cumulative distribution functions for the different strains studied indicate that only long-lived strains show a significant decrease in the uORFs-to-CDS ratio compared to the wild type strain. The Mann Whitney U test was used to compare the distributions of ratios between RPKO and the wild type, and the Pvalues of the two-tailed test are indicated.
Supplementary Fig. 4. The effect of generalized uORF skipping on translation efficiency (a to d) Scatter plot of translation efficiency (TE) fold-changes of (a) Δrpl6a, (b) Δrpl15b, (c) Δrpl7a and (d) Δrpl9a with respect to the wild type strain, versus the difference in 5'UTR-to-CDS ratios for all genes containing uORFs. GCN4 is highlighted in orange, and genes showing significant changes (|ΔTE| >= 1.5) in translation efficiency have been labeled. (e and f) Boxplots of TE fold-changes in the long-lived (e) Δrpl7a and (f) Δrpl9a relative to the wild type strain for all genes (dark gray) and genes containing one or more uORFs (light gray). There are no significant differences in the TE distribution between the two groups. Boxes extend from the 25th to 75th percentiles, horizontal lines represent the median, whiskers indicate the lowest and highest datum within 1.5*IQR from the lower and upper quartiles, respectively. Boxplot notches indicate the 95% confidence interval around the median. and P Gal1-10 -GST-GCN4 strains relative to corresponding wild type strains. Boxes extend from the 25th to 75th percentiles, horizontal lines represent the median, whiskers indicate the lowest and highest datum within 1.5*IQR from the lower and upper quartiles, respectively.
Supplementary Fig. 8. Analysis of putative Gcn4-Rap1 co-regulation (a) Venn diagram depicting the intersection of previously described Rap1-regulated genes and the upregulated targets of Gcn4 inferred here from ChIP. (b) Position of the Gcn4 binding site relative to Rap1 binding site (in nucleotides, negative values indicate that Gcn4 binding is predicted to be upstream of Rap1, in the direction of transcription) in the upstream regions of downregulated (dark gray) and upregulated (light gray) targets (left). Schematic representation of the architecture of downregulated and upregulated promoters co-targeted by Gcn4 and Rap1 (right). ** indicates p-value < 0.01 in the two-tailed the Mann Whitney U test. Boxes extend from the 25th to 75th percentiles, horizontal lines represent the median, whiskers indicate the lowest and highest datum within 1.5*IQR from the lower and upper quartiles, respectively.
Supplementary Fig. 9. The DNA-binding domain of Gcn4 is required for the repression of RP genes (a) Chromatogram showing the wildtype GCN4 genomic sequence and the sequence after mutating two nucleotides to convert Serine to Leucine at position 242 (S242L). The nucleotide that were mutated are shown in red, both in the WT and the mutated sequence. (b) Scatter plot of mRNA fold-changes for the two strains overexpressing either Gcn4 (x-axis) or Gcn4 S242L (y-axis) compared to the wildtype strain. Red dots represent Gcn4 targets from the yeast database. (c) Cumulative distribution functions of mRNA expression changes observed in the two GCN4 overexpression strains relative to the wild type strain. (d) Number of genes that were differentially expressed in the two strains relative to wildtype (i.e. genes for which the mRNA abundance changed more than 2 fold relative to wildtype at a False Discovery Rate lower than 0.01). (e) Boxplots show the mRNA fold-changes for genes encoding amino acid biosynthesis (dark gray) and ribosomal (light gray) protein genes in the WT and S242L Gcn4-overexpressing strains with respect to wildtype. P-values were calculated using the two-sided Mann Whitney U test. Boxes extend from the 25th to 75th percentiles (interquartile range (IQR)), horizontal lines represent the median, whiskers indicate the lowest and highest datum within 1.5*IQR from the lower and upper quartiles, respectively.
Supplementary Fig. 10. Quality control of ribosome occupancy sequencing data.
(a) RPF read length distribution across the different strains studied. The height of the bar and the error bar show the mean and standard error of the mean, respectively, of read length density across replicates. (b) Metagene analysis of ribosome profiling reads. For each read length, the relative location of the P site with respect to the read start was inferred as the value for which the correct position of the start codon and the 3-nt periodicity was most apparent. The number of reads mapped around start and stop codons is shown for two different read lengths (29nt and 30nt) for the WT strain. The red dots correspond to the first positions of each codon in the CDS. (c) Density of RPF reads starting at each of the three reading frames (1 represents the first position of the codon). The height of the bar and the error bar show the mean and standard error of the mean, respectively, of the density for all read lengths showing 3-nt periodicity across all the ORFs (depending on the data set, read lengths of 28-34 nucleotides were typically used). (d) Metagene analysis of RPF reads across the different strains studied. RPF reads per codon (A-site) in a given ORF were individually normalized by the mean number of reads within the respective ORF, and then averaged across all the ORFs.