HST2 Mediates SIR2-Independent Life-Span Extension by Calorie Restriction

See allHide authors and affiliations

Science  16 Sep 2005:
Vol. 309, Issue 5742, pp. 1861-1864
DOI: 10.1126/science.1113611


Calorie restriction (CR) extends the life span of numerous species, from yeast to rodents. Yeast Sir2 is a nicotinamide adenine dinucleotide (NAD+-dependent histone deacetylase that has been proposed to mediate the effects of CR. However, this hypothesis has been challenged by the observation that CR can extend yeast life span in the absence of Sir2. Here, we show that Sir2-independent life-span extension is mediated by Hst2, a Sir2 homolog that promotes the stability of repetitive ribosomal DNA, the same mechanism by which Sir2 extends life span. These findings demonstrate that the maintenance of DNA stability is critical for yeast life-span extension by CR and suggest that, in higher organisms, multiple members of the Sir2 family may regulate life span in response to diet.

A major cause of aging in the yeast Saccharomyces cerevisiae stems from homologous recombination between ribosomal DNA (rDNA) repeat sequences and the formation of extrachromosomal circles of rDNA known as ERCs, which accumulate exponentially and eventually lead to the death of the mother cell (1, 2). The Fob1 protein promotes ERC formation by stalling rDNA replication forks, which are inherently unstable structures (3). Accordingly, deletion of FOB1 reduces ERC formation and extends yeast life span (4).

In contrast, ERC formation is repressed by the activity of Sir2, an NAD+-dependent histone deacetylase (HDAC) that catalyzes the formation of heterochromatin at the rDNA locus (5-8) and is the founding member of the sirtuin deacetylase family. Additional copies of the SIR2 gene extend yeast life span by decreasing ERC formation, whereas deletion of SIR2 increases ERC formation and shortens life span (9), a phenotype that can be suppressed by deleting FOB1 (4).

The life span of most species including S. cerevisiae is extended by CR (10). Yeast and Drosophila lacking the SIR2 gene do not live longer when subjected to CR, suggesting that Sir2 underlies this life-span extension (11-15). However, there is increasing evidence that the situation is not so simple (16-18). Although CR is blocked by a sir2Δ mutation, CR can extend the life span of yeast cells lacking both SIR2 and FOB1 (18, 19). CR can also extend the life span of Caenorhabditis elegans worms lacking the worm SIR2 gene, sir-2.1 (20). Not surprisingly, there is much confusion and debate in the aging field regarding these findings, with some researchers postulating that Sir2 may not regulate life-span extension by CR and that yeast life-span extension is independent of suppression of ERC formation (19, 21).

To help resolve this issue, we screened for genes that could increase yeast life span via the same mechanism as that by which SIR2 increases life span by taking advantage of the positive correlation between rDNA silencing and longevity (2, 22). The marker genes MET15 and ADE2 were integrated into the rDNA locus such that increased rDNA silencing led to the accumulation of a brown pigment on Pb2+-containing media, or decreased growth on plates lacking adenine, respectively. A high-copy 2μ yeast genomic library was introduced into the reporter strain and, from ∼20,000 colonies, we isolated three genomic fragments that resulted in a robust increase in rDNA silencing (fragments A, B, and C) (Fig. 1A). The three candidates were then tested for rDNA recombination, which negatively correlates with life span. Candidates A and B, but not C, showed a marked decrease in rDNA recombination rate (fig. S1).

Fig. 1.

A screen for genes that stabilize the rDNA locus identifies HST2 (A) rDNA locus (RDN1) silencing assays. W303AR cells carrying the gene candidates on a 2μ plasmid were grown to log phase in YPD (2% glucose) and then spotted in 10-fold dilutions on SC media with and without adenine. Silencing of the RDN1::ADE2 reporter results in growth retardation on plates lacking adenine. (B) Overexpression of HST2 can suppress rDNA recombination even in the absence of Sir2. A plasmid carrying HST2 (pSP400-Hst2) was integrated into the URA3 locus of W303AR5 sir2::TRP1 and W303AR sir2::TRP1 fob1::hphr. W303AR5 cells containing the gene candidates were grown to log phase in 2% YPD and then plated on YPD plates with low adenine. rDNA recombination rates were calculated by determining the frequency of loss of ADE2 in the rDNA of strain W303AR5 at the first cell division after plating, as scored by the number of half red-sectored colonies per total number of colonies (30). At least 6000 colonies were examined for each data point. Results show average values and SD for at least three experiments. (C) Treatment with 5 mM nicotinamide (NAM) blocks life-span extension by CR (0.5% glucose or an hxk2Δ deletion). Life-span analyses were performed as described (11). Average life spans: W303 sir2Δ fob1Δ 2% glucose 24.7, 0.5% glucose 30.4, 2% glucose + 5mM NAM 14.8, 0.5% glucose + 5 mM NAM 15.9. (D) NAM blocks life-span extension by hxk2Δ in BY4742 sir2Δ fob1Δ. Average life spans: BY4742 sir2Δ fob1Δ 25.7; sir2Δ fob1Δ hxk2Δ 37.9; sir2Δ fob1Δ + 5mM NAM 25.9; sir2Δ fob1Δ hxk2Δ + 5 mM NAM 26.5.

The gene responsible for the phenotype of fragment A was identified as HST2 (homolog of SIR2) (fig. S2). HST2 encodes a NAD+-dependent deacetylase whose closest human homolog is SIRT2 (16, 23, 24). Although Hst2 is known as a cytoplasmic protein (16), it has recently been shown to also be present in the nucleus, where it interacts with chromatin and down-regulates the expression of subtelomeric genes in concert with Hst1 (16, 23). Unexpectedly, we found that overexpression of HST2 suppressed rDNA recombination not only in wild-type cells, but also in the absence of SIR2 (Fig. 1B). This showed that HST2 does not necessarily act via SIR2 as previously hypothesized (16), and raised the possibility that Hst2 might be the factor responsible for mediating Sir2-independent life-span extension by CR.

To test this hypothesis, we first determined whether inhibiting Hst2 enzymatic activity affected the ability of CR to extend life span. We used nicotinamide (NAM), a vitamin B3 precursor that has been shown to inhibit yeast sirtuins in vitro (25, 26) and in vivo (25, 27). As reported for other yeast strains (19), the life span of W303AR5 sir2Δ fob1Δ was extended by mild CR (Fig. 1C). However, in the presence of NAM, there was no significant effect of CR on life span. The same effect was observed for strain BY4742 carrying an hxk2Δ (hexokinase 2) mutation, which is a genetic mimic of intense CR (11, 19) (Fig. 1D). Together, these experiments were consistent with the hypothesis that Hst2 (or another sirtuin) is required for Sir2-independent life-span extension by CR.

To determine whether the effect was due to Hst2 specifically, we tested whether deleting the HST2 gene affected the ability of CR to extend life span. Although CR could extend the life span of an hst2Δ strain, it was unable to extend the life span of either sir2Δ hst2Δ (Fig. 2A) or sir2Δ fob1Δ hst2Δ strains (Fig. 2B). Thus, HST2 is necessary for SIR2-independent life-span extension by CR.

Fig. 2.

Hst2 is required for Sir2-independent life-span extension by CR. (A) Average life spans: W303AR5 (wild type, wt) 2.0% glucose 26.7, 0.5% glucose 31.1; hst2Δ 2.0% glucose 18.9, 0.5% glucose 24.2; sir2Δ 2.0% glucose 13.7, 0.5% glucose 14.0; sir2Δ hst2Δ 2.0% glucose 13.9, 0.5% glucose 12.2. (B) Average lifespans: W303AR5 sir2Δ fob1Δ 2.0% glucose 21.9, 0.5% glucose 29.6; hst2Δ fob1Δ 2% glucose 22.3, 0.5% glucose 28.3; sir2Δ fob1Δ hst2Δ 2.0% glucose 13.0, 0.5% glucose 12.9. (C) Average life spans: BY4742 (wild type, wt) 2.0% glucose 20.6, 0.5% glucose 30.4; sir2Δ fob1Δ 2% glucose 23.0, 0.5% glucose 28.9; sir2Δ fob1Δ hst2Δ 2.0% glucose 20.6, 0.5% glucose 21.7. (D) Average life spans: BY4742 (wild type, wt) 23.2; hxk2Δ 38.4; sir2Δ fob1Δ 26.6; sir2Δ fob1Δ hxk2Δ 39.1; sir2Δ fob1Δ hst2Δ 25.9; sir2Δ fob1Δ hst2Δ hxk2Δ 30.1.

To ensure that our results were not strain-specific, we also performed our experiments in BY4742, in which the Sir2-independent CR pathway was originally identified (19). Again, deletion of HST2 in the sir2Δ fob1Δ strain completely prevented life-span extension by CR (Fig. 2C). In the hxk2Δ CR mimetic, deletion of HST2 strongly abrogated but did not completely prevent life-span extension in the sir2Δ fob1Δ background (Fig. 2D), suggesting the possible involvement of another factor in life-span extension.

It has been postulated that Sir2-independent life-span extension is unrelated to rDNA recombination (19). However, our data provide strong evidence to the contrary. (i) Consistent with the life-span data, CR suppressed rDNA recombination to about 50% of the initial rate in the sir2Δ fob1Δ strain, the same reduction as seen for wild-type cells on CR or overexpressing HST2 (Fig. 3A). (ii) The ability of CR to suppress recombination is completely blocked by NAM (Fig. 3B), again consistent with the life-span data above. (iii) Deletion of HST2 from a wild-type strain led to an approximate doubling in the rates of rDNA recombination, which could be suppressed by a fob1 deletion (Fig. 3C), demonstrating that HST2 is necessary for rDNA stability, even in a wild-type strain. (iv) Although single deletions of HST2 or SIR2 in a fob1Δ strain resulted in rDNA recombination rates similar to or lower than those of the wild type, deletion of both HST2 and SIR2 in this background resulted in a fivefold increase in recombination (Fig. 3C). This elevated recombination was only slightly reduced by CR. Finally, we observed a highly significant negative correlation between life span and rDNA recombination rate (fig. S3). Although these data do not exclude the possibility that CR may mediate yeast life span independently of its effects on the rDNA, these data provide strong evidence that CR extends life span by suppressing rDNA recombination irrespective of whether SIR2 is present or absent. They also demonstrate that in a sir2Δ fob1Δ strain, Hst2 is critical for maintaining rDNA stability.

Fig. 3.

CR extends life span via HST2 by suppressing rDNA recombination. (A) CR and overexpression of HST2 can suppress recombination by an equivalent amount. The rDNA recombination assay was performed as described in the legend to Fig. 1B, but CR treated strains were grown to log phase in 0.5% YPD before plating. (B) Treatment with NAM (5 mM) prevents CR from suppressing recombination. (C) Deletion of either HST2 or SIR2 results in increased rDNA recombination. A fob1Δ deletion suppresses hyper-recombination in a sir2Δ strain but only if HST2 is present. Recombination assays were performed as described in the legend to Fig. 1B. HST2 was disrupted in W303AR5 using the kanr marker, and FOB1 was disrupted using the hphr marker (31, 32).

Although the deletion of HST2 blocked the ability of CR to extend life span in the sir2Δ fob1Δ strain, it was formally possible that this was caused by toxic levels of ERCs in the strain, precluding alternative CR pathways from taking effect. Therefore, we determined whether HST2 could increase life span when overexpressed in order to test whether HST2 is a bona fide longevity gene (9). Consistent with the ability of HST2 to increase rDNA silencing and decrease rDNA recombination (Fig. 1 and fig. S1), overexpression of HST2 in W303AR5 sir2Δ fob1Δ extended life span to the same extent as CR in this strain background (Fig. 4A), as well as in a wild-type strain (fig. S4). No additive effect of HST2 overexpression and CR was observed, indicating that HST2 and CR extend life span of sir2Δ fob1Δ mutants through the same pathway (28).

Fig. 4.

Additional HST2 promotes life span in the absence of Sir2 and Hst1 plays a supportive role. (A) The average and maximum life span of W303AR5 sir2Δ fob1Δ are increased by overexpression of HST2 similar to growth on 0.5% glucose. Life spans were analyzed as described (11). Average life spans: W303AR5 sir2Δ fob1Δ 2.0% glucose 25.1, 0.5% glucose 29.9; W303AR5 sir2Δ fob1Δ HST2 overexpression (o/e) 30.5. (B) Deletion of HST1, HST2, or HST4 increases rDNA recombination. (C) Deletion of HST1 blocks the residual life-span extension by hxk2Δ in a sir2Δ fob1Δ hst2Δ strain. Average life spans: BY4742 sir2Δ fob1Δ 24.8; sir2Δ fob1Δ hxk2Δ 41.8; sir2Δ fob1Δ hst2Δ hst1Δ 22.7; sir2Δ fob1Δ hst2Δ hst1Δ hxk2Δ 24.8. (D) Deletion of HST1 does not prevent CR from suppressing recombination in a wild-type or a sir2Δ fob1Δ strain.

Next, we investigated whether the residual life-span extension seen for the hxk2Δ mutant (a mimic of intense CR) lacking SIR2 and HST2 (Fig. 2C) was due to the activity of another sirtuin. As previously reported (16), deletion of HST1 markedly increased rDNA recombination in a wild-type strain (Fig. 4B). Although deleting HST3 and HST4 together has been shown to decrease chromosomal stability and increase mitotic recombination (29), we did not observe increased rDNA recombination in a W303AR5 hst3Δ hst4Δ strain, although recombination in an hst4Δ single mutant is about twice as high as that in the wild type. Because deletion of HST1 had the greatest effect on rDNA recombination, we suspected that Hst1 might be the factor responsible for the residual life-span extension. This hypothesis was consistent with our finding that the general sirtuin inhibitor NAM completely blocked the life-span extension of a sir2Δ fob1Δ strain by hxk2Δ (Fig. 1D) and a recent report that Hst1 functions in the nucleus with Hst2 in gene silencing (23). Whereas deletion of either HST3 or HST4 in this strain did not affect the ability of hxk2Δ to extend life span (fig. S5), deletion of HST1 completely eliminated the residual life-span extension provided by hxk2Δ in the BY4742 sir2Δ fob1Δ hst2Δ strain (Fig. 4C).

In a previous study, the life span of a sir2Δ fob1Δ hst1Δ strain was extended by CR (19), leading the authors to conclude that HST1 plays no role in CR. Indeed, in agreement with this finding, we find that CR is effective in suppressing recombination of such a mutant (Fig. 4D). However, this study implies that HST2 underlies the CR-mediated life-span extension of this strain and that HST1 plays a minor role that is observed only in the absence of SIR2 and HST2.

Our results show that HST2 is responsible for Sir2-independent life-span extension by CR and that it does so by suppressing rDNA recombination, the same mechanism by which SIR2 extends life span. These findings highlight the importance of genomic stability as a determinant of yeast life span and raise the likelihood that multiple members of the sirtuin family in higher organisms also play critical roles in maintaining genomic stability and possibly in extending life span during times of adversity.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article