Report

Regulation of Yeast Replicative Life Span by TOR and Sch9 in Response to Nutrients

See allHide authors and affiliations

Science  18 Nov 2005:
Vol. 310, Issue 5751, pp. 1193-1196
DOI: 10.1126/science.1115535

Abstract

Calorie restriction increases life span in many organisms, including the budding yeast Saccharomyces cerevisiae. From a large-scale analysis of 564 single-gene–deletion strains of yeast, we identified 10 gene deletions that increase replicative life span. Six of these correspond to genes encoding components of the nutrient-responsive TOR and Sch9 pathways. Calorie restriction of tor1D or sch9D cells failed to further increase life span and, like calorie restriction, deletion of either SCH9 or TOR1 increased life span independent of the Sir2 histone deacetylase. We propose that the TOR and Sch9 kinases define a primary conduit through which excess nutrient intake limits longevity in yeast.

Calorie restriction (CR) is the only intervention known to increase life span in yeast, worms, flies, and mammals, but the molecular mechanism for this phenomenon has not been clear. In yeast, CR due to reduced glucose concentration of the culture medium increases replicative life span (the number of daughter cells produced by a given mother cell before senescence) by 20 to 40% (13). This increased life span has been attributed to activation of Sir2 (1), a histone deacetylase that is dependent on NAD (the oxidized form of nicotinamide adenine dinucleotide) (4) and that promotes longevity by inhibiting the formation of extrachromosomal ribosomal DNA (rDNA) circles (ERCs) in the nucleolus (5). Recently, however, the link between Sir2 and CR has been called into question with the discovery that Sir2 is not required for life-span extension by CR (3).

To identify genes that regulate longevity in the budding yeast, a large-scale analysis of replicative life span was conducted with the MATa haploid open reading frame (ORF) deletion collection, a set of ∼4800 single-gene–deletion strains (6). Because replicative life-span analysis requires labor-intensive micromanipulation of daughter cells from mother cells, fewer than 80 different genes have been previously examined for their effect on replicative life span (7). Here we examined the replicative aging properties of 564 single-gene–deletion strains (Fig. 1A; table S1).

Fig. 1.

TOR activity is an important modifier of yeast longevity. (A) The distribution of observed strain mean life spans for 564 single-gene–deletion mutants (broken line) shows an overrepresentation of short-lived (dark arrow) and long-lived (light arrow) mutants relative to expected mean life-span distribution (solid lines) for wild-type cells of the same sample size (n = 5). (B) Deletion of TOR1 increases life span. (C) Deletion of either RPL31A or RPL6B, ribosomal proteins transcriptionally regulated by TOR, increases life span. Mean life spans are shown in parentheses.

An iterative method was designed to identify ∼95% of strains with mean replicative life span at least 30% longer than wild type (8). For each single-gene–deletion strain, replicative life span was initially determined for five individual mother cells. If the mean life span was less than 26 generations, the strain was classified as not-long-lived (NLL). This lower cutoff value is predicted to result in misclassification of a long-lived strain less than 5% of the time (fig. S1). If the mean life span was less than 20, the strain was classified as short-lived (SL). If the mean life span was greater than 36, the strain was putatively classified as long-lived (LL), and an additional 10 cells were examined. This upper cutoff value is predicted to result in misclassification of a strain with wild-type life span less than 2% of the time. For the remaining strains with a five-cell mean life span between 26 and 36 generations, an additional five cells were analyzed (one iteration), and the same classification scheme was applied. This process was repeated until every strain was either classified as SL, NLL, or LL or until replicative life span had been determined for a minimum of 15 cells for each unclassified strain. The replicative life-span data for strains from which at least 15 mother cells had been assayed were compared with cell life-span data from wild-type mothers, matched by experiment, by using a Wilcoxon rank-sum test to generate a P value. Strains with P ≤ 0.1 were classified as LL, and strains with P > 0.1 were classified as having a life span not significantly extended (NSE). Of the 564 strains analyzed, 114 were classified as SL, 254 as NLL, 152 as NSE, and 44 as LL. Although nearly 20% of the gene deletions resulted in a significantly shortened life span, relatively few of these are likely to represent a true premature aging phenotype, because dysregulation of many different cellular processes will decrease fitness and longevity (9). For this reason, we focused on genes that, when deleted, resulted in increased replicative life span, reasoning that the proteins encoded by these genes must impede the normal aging process.

Of the 44 single-gene–deletion strains initially classified as LL, 13 result in a significant increase in replicative life span (Table 1). Verification was accomplished by determining the replicative life span for the corresponding gene deletion strain from the haploid MATa deletion collection and, in select cases, by generating a new deletion allele in the parental BY4742 strain. Of the 13 genes, FOB1 served as a proof of principle that our method can identify a true-positive aging gene, because deletion of FOB1 is known to increase life span by reducing the formation of ERCs (10). In two cases, gene deletions conferring increased life span occurred in overlapping ORFs encoded on opposite strands (REI1 contains YBR266C; IDH2 overlaps YOR135C), and longevity was comparable for overlapping deletion pairs (table S2). The identification of two different overlapping gene pairs from this screen suggests that a high fraction of true-positive genes were successfully identified.

Table 1.

Long-lived deletion strains. From a screen of 564 single-gene–deletion strains, 13 genes were found to increase replicative life span when deleted. GDP, guanosine diphosphate; GTP, guanosine 5′-triphosphate; PI3-like kinase, a kinase like phosphatidylinositol 3-kinase.

Deletion strain Protein function
bre5Δ Ubiquitin protease
fob1Δ rDNA replication fork barrier protein
idh2Δ Isocitrate dehydrogenase
rei1Δ Protein of unknown function with similarity to human ZPR9
rom2Δ GDP-GTP exchange factor for Rho 1p
rpl31aΔ Ribosomal protein L31
rpl6bΔ Ribosomal protein L6
tor1Δ PI3-like kinase involved in regulation of cell growth
ure2Δ Regulator of nitrogen catabolite repression
ybr238cΔ Protein of unknown function
ybr255wΔ Protein of unknown function
ybr266cΔ Hypothetical ORF overlapping REI1
yor135cΔ Hypothetical ORF overlapping IDH2

The most striking feature of the 10 (excluding the overlapping dubious ORFs and FOB1) newly identified aging genes is that 6 are implicated in the TOR signaling pathway. TOR proteins are highly conserved from yeast to humans and regulate multiple cellular processes in response to nutrients, including cell size, autophagy, ribosome biogenesis and translation, carbohydrate and amino acid metabolism, stress response, and actin organization (11). Yeast has two TOR proteins, Tor1 and Tor2. Tor2 is essential and, therefore, not represented in the deletion collection. Deletion of TOR1 was identified from this screen and found to increase both mean and maximum life span by ∼20% (Fig. 1B). Two downstream targets of Tor1 and Tor2 were also identified: Ure2, which regulates activity of the nitrogen-responsive transcription factor Gln3, and Rom2, a proposed activator of protein kinase C (12, 13). Deletion of three genes that are transcriptionally up-regulated by TOR increased life span: YBR238C, a gene of unknown function (14), and RPL31A and RPL6B, encoding two components of the large ribosomal subunit (Fig. 1C). Not all TOR-regulated ribosomal protein gene deletions examined conferred increased life span. Unlike the case in most multicellular eukaryotes, many of the ribosomal protein genes are duplicated in yeast (e.g., RPL31A and RPL31B), which allows for viable deletion of either paralog (but not both simultaneously). The relative importance of each paralog for ribosomal function, perhaps reflecting differential expression levels, may determine the longevity phenotype on deletion, with the gene coding for the more abundant member of the pair more likely to influence life span. Consistent with this idea, rpl31aD mother cells are long-lived and slow growing, whereas rpl31bD mother cells are not (fig. S2).

Protein kinase A (PKA) and Sch9 are nutrient-responsive protein kinases that modulate replicative aging in yeast (1, 15). Mutations that decrease PKA activity increase replicative life span and have been suggested as genetic models of CR (1, 3). TOR is thought to act both upstream and parallel to PKA, whereas Sch9 is thought to act in a pathway parallel to PKA and TOR (16, 17). TOR, PKA, and Sch9 regulate the expression of common downstream targets, including ribosomal proteins, such as Rpl31a and Rpl6b (18, 19). CR of tor1D or sch9D cells failed to significantly increase the life span of these long-lived mutants (Fig. 2, A and B), which indicates that, similar to PKA, Sch9 and TOR are targets of CR in yeast. CR by growth on low glucose, or mutations resulting in decreased PKA activity, increase life span additively with deletion of FOB1 (3). Deletion of TOR1 or deletion of SCH9 also resulted in an additive increase in life span when combined with deletion of FOB1 (Fig. 2, C and D). The already long life span of the sch9D fob1D or tor1D fob1D mother cells was not further increased by CR (Fig. 2E). Life-span extension by CR also occurs independently of Sir2, as long as ERC formation is kept low through deletion of FOB1 (3). Deletion of either TOR1 or SCH9 also increased the life span of sir2D fob1D cells (Fig. 2F). These epistasis experiments suggest that decreased activity of the nutrient-responsive kinases Sch9 and TOR in response to CR results in increased replicative life span in yeast.

Fig. 2.

TOR1 or SCH9 deletion mutants are genetic mimics of CR. (A) CR fails to further increase the life span of cells lacking TOR1. (B) CR fails to further increase the life span of cells lacking SCH9. (C) Deletion of TOR1 increases life span additively with deletion of FOB1. (D) Deletion of SCH9 increases life span additively with deletion of FOB1.(E) Deletion of either TOR1 or SCH9 fails to increase the life span of calorie-restricted fob1D cells. (F) Deletion of either TOR1 or SCH9 increases the life span of sir2D fob1D double-mutant cells. Mean life spans are shown in parentheses.

Life-span extension by CR in yeast was initially characterized in the short-lived strain background PSY316 (1). PSY316 is unique among yeast strains used for longevity studies in that, although CR increases lifespan by 30 to 40%, deletion of FOB1 or overexpression of SIR2 fails to result in increased life span (20). To determine whether TOR activity is a general or strain-specific determinant of replicative life span, we examined the effect of TOR1 deletion on life span and Sir2 activity in the PSY316 background. Deletion of TOR1 significantly increased life span in PSY316, but had no effect on Sir2-dependent silencing at telomeres, similar to the effect of CR by growth on low glucose (Fig. 3, A and B). Thus, like CR, decreased TOR activity is a strain-independent mechanism to achieve enhanced longevity in yeast.

Fig. 3.

Decreased TOR activity, like CR, is a strain-independent modifier of replicative life span. (A) Deletion of TOR1 increases life span in the PSY316 background. Mean life spans are shown in parentheses. (B) Deletion of TOR1 and CR have no effect on Sir2-dependent silencing of a telomeric URA3 marker gene, as measured by survival in the presence of 5-FOA, in the PSY316 background. An extra copy of Sir2 (SIR2-ox) increases silencing of telomeric URA3. WT is wild-type.

TOR activity is a primary determinant of replicative aging in yeast, and genetic analysis indicates that Sir2-independent life-span extension by CR is mediated by reduced signaling through TOR, Sch9, and PKA, resulting in down-regulation of ribosome biogenesis. Recently, an alternative model has suggested that Sir2-independent CR is caused by decreased ERC formation, resulting from nuclear relocalization and activation of the Sir2 homolog Hst2 (21). However, as long as ERC formation is maintained at a low level, CR increases life span to a greater extent in cells lacking Sir2 than in cells where Sir2 is present, seemingly inconsistent with Hst's simply playing a role redundant to Sir2's. CR increases life span additively with deletion of FOB1, which suggests a mechanism for CR that is independent of ERCs. ERCs also affect aging only in yeast, whereas the longevity-promoting role of CR has been evolutionarily conserved.

Decreased activity of TOR and Sch9 orthologs increases life span in Caenorhabditis elegans (22, 23) and Drosophila melanogaster (24), as does mutation of the TOR-regulated S6 kinase (24), which promotes ribosomal protein maturation in multicellular eukaryotes. Therefore, the data presented here are consistent with a model whereby CR increases life span through a highly conserved, Sir2-independent signaling network from nutrients to ribosomes.

Supporting Online Material

www.sciencemag.org/cgi/content/full/310/5751/1193/DC1

Materials and Methods

Figs. S1 to S3

Tables S1 and S2

References and Notes

References and Notes

View Abstract

Navigate This Article