Nucleosome loss leads to global transcriptional up-regulation and genomic instability during yeast aging

Abstract

All eukaryotic cells divide a finite number of times, although the mechanistic basis of this replicative aging remains unclear. Replicative aging is accompanied by a reduction in histone protein levels, and this is a cause of aging in budding yeast. Here we show that nucleosome occupancy decreased by 50% across the whole genome during replicative aging using spike-in controlled micrococcal nuclease digestion followed by sequencing. Furthermore, nucleosomes became less well positioned or moved to sequences predicted to better accommodate histone octamers. The loss of histones during aging led to transcriptional induction of all yeast genes. Genes that are normally repressed by promoter nucleosomes were most induced, accompanied by preferential nucleosome loss from their promoters. We also found elevated levels of DNA strand breaks, mitochondrial DNA transfer to the nuclear genome, large-scale chromosomal alterations, translocations, and retrotransposition during aging.

Keywords: DNA rearrangement; gene expression; histone occupancy; replicative aging.

Figure 1.

(A) Schematic diagram of the purification of old and young cells. (B) Old cells isolated using the MEP were analyzed for histone H3 protein levels in comparison with young cells. Equivalent numbers of old and young cells (strain ZHY2) were analyzed by Western blotting. The number of cell divisions was determined by counting bud scars after calcofluor staining. (C) MNase-seq data for a typical region of the genome. “Over” refers to an old sample with overexpressed H3/H4. (D) Distribution of nucleosome fuzziness scores, neighboring distances, and occupancy periodicities. Periodicity was calculated based on the power spectrum density analysis algorithm (Chen et al. 2008, 2010). (E) Accumulated nucleosome count plotted relative to sequence-predicted nucleosome occupancy. (Left) Plot for nucleosomes with positions shifted up to 100 bp and not <50 bp in old cells relative to young cells. (Right) Plot for nucleosomes whose positions are the same in old and young cells; therefore, the curves for the old and young cells completely overlap with each other. P-values are calculated based on KS test between the distributions of nucleosome occupancy values in old and young cells.

Figure 2.

(A) Nucleosome occupancy normalized based on spike-in control for all genes plotted relative to the TSS and TTS. All three replicates are plotted for each cell group. (B) Gene expression changes during aging are plotted from the RNA-seq analysis. Nucleosome occupancies for the same genes are shown on the right. Genes are ranked by the false discovery rate (FDR) of expression difference between old and young cells, with most induced genes on the top and least induced genes at the bottom. (C) Nucleosome occupancy plotted relative to the TSS; the top and bottom 500 genes ranked by expression up-regulation during aging are defined as most induced and least induced gene groups, respectively, and are plotted separately. (D) The fold change of nucleosome occupancy plotted relative to the TSS. Independent scales on the Y-axis are used to plot for old–young and over–young fold change. Horizontal dashed line indicates average fold change over all yeast genes. (E) Rank of gene expression change during aging plotted relative to the FDR value of gene expression change during aging (red) and during aging in the presence of H3/H4 overexpression (black). (F) RNA-seq data for three typical genes: MET5, YSC4, and FAB1. “Over” refers to an old sample with overexpressed H3/H4. Sky-blue arrows over a purple line indicate gene locus on genomic DNA, respectively.

Figure 3.

(A) Comparison of nucleosome occupancy changes during histone H3 transcriptional repression (left) and aging (middle) and comparing aging with histone H3 transcriptional repression (right). (r) Pearson correlation coefficient. (B) Comparison of gene expression changes during aging and histone H3 transcriptional repression.

Figure 4.

(A) The percentage of genes with TATA boxes in each of 500 genes ranked according to expression change during aging is shown. (R) Spearman's rank correlation coefficient. (B) The sequence-predicted nucleosome occupancy for the 500 most induced and the 500 least induced genes during aging, plotted relative to the TSS. (C) Accumulated count of promoters plotted relative to predicted minimal nucleosome occupancy. P-value is calculated based on KS test of the occupancy value distribution. (D) The 500 most and least induced genes during aging, ranked by chromatin remodeler occupancy, are defined as bound and not bound, respectively. (E, left) A list of chromatin-binding proteins that bind (Y) or do not bind (N) to the most induced or least induced genes during aging. We retrieved the top and bottom 500 genes bound by each protein and calculated their percentage of overlap with the most induced genes; if the percentage exceeded 15%, we then defined the protein as binding most induced genes. (Right, top) Genes were ranked by Taf9-binding intensity from highest to lowest and divided into groups, each containing 500 genes. For each group, the percentage of genes most induced or least induced during aging was calculated. The percentage for each group was then plotted relative to the Taf9-binding rank for each group. Similar plots were drawn for Tfb1 (middle) and Vps72 (bottom).

Figure 5.

(A) Real-time RT–PCR analysis of the fold increase in the expression of the indicated Ty elements during aging. Average and standard deviation of three independent experiments are plotted. (B) Real-time PCR analysis of the fold increase in the genomic DNA content of the indicated Ty elements during aging. Average and standard deviation of three independent experiments are plotted. (C). Replicative life span of isogenic yeast strains BY4741 (wild type [WT]; N = 31) and a strain deleted for YLR194C (ylr194cΔ; N = 32). (D) Replicative life span of isogenic yeast strains wild type (WT; BY4741) and pGAL-YLR194c grown on raffinose (No Gal; left panel) or 0.5% Gal medium (right panel; N = 40) to induce expression of Ylr194c.

Figure 6.

(A) Average MNase-seq read coverage on each chromosome or chromosome segment. (B) Average genome sequencing read coverage on each chromosome or chromosome segment in the case of chromosome XII. (C) Nucleosome occupancy across the entire chromosome XII. (D) Analysis of intact yeast chromosomes by PFGE at different time points during aging. (Left panel) SYBR safe staining pattern of the chromosomes. (Right panel) Southern hybridization using a chromosome XII right arm-specific (distal to the rDNA locus) probe. During aging, a longer chromosome appears (labeled with an asterisk), along with apparent fragmentation of the chromosomes. (E) Circos plot of chromosomal translocations that were significantly increased (P < 1 × ⁻³) in old cells relative to young cells. The rDNA locus is indicated by a black bar in the outer ligh-gray circus, and the LTR loci are marked on the inner dark-gray circus, as are the Ty elements. Links shown in blue indicate translocations connecting from either the rDNA region or the mtDNA to other chromosomes. Black links indicate other interchromosome translocation. Red links indicate intrachromosome translocation.

Figure 7.

(A) Quantitation of the percentage of young and old cells with detectable γH2A foci, as measured by immunofluorescence. Error bars represent standard deviation of three independently isolated sets of old and young cells. (B) ChIP-seq of γH2A occupancy in the rDNA and mtDNA of young and old cells. The data shown are not normalized to histone content or DNA content. (C) Quantitation of the percentage of young and old cells with detectable incorporation of biotin dUTP onto DNA ends by TdT, as detected by immunofluorescence. Error bars represent standard deviation of three independently isolated sets of old and young cells. (D) Fold change of unique reads or nonunique reads in each sequenced sample during aging. Error bars indicate standard deviation among replicates. (E) Fold change of rDNA reads, mtDNA reads, reads on Ty elements, or all the other reads in the genome sequencing, MNase-seq, and γH2A-ChIP-seq. Error bars indicate standard deviation among replicates. (F) Validation of increased mtDNA content in aged cells by quantitative PCR. Average and standard deviation of three independent experiments are shown, normalized to a spike in control added according to equivalent cell number. (G) Functional assay of transfer of mtDNA into the nucleus in a strain carrying a TRP1 gene in the mitochondrial genome (red) and a strain lacking the TRP1 gene (purple), as determined by frequency of colony formation on −TRP plates. Shown are the average and standard deviation of three independent experiments performed at the indicated times during aging using the MEP.