Combining two or more longevity-altering interventions and determining the resulting effect on lifespan is a common method for examining the relationship between such interventions. An important subset of this type of analysis occurs when one of the factors under study promotes longevity, such as daf-16 in Caenorhabditis elegans or SIR2 in Saccharomyces cerevisiae. For both of these genes, several studies have combined a lifespan shortening null allele with an intervention that extends lifespan. A resulting lifespan similar to that of the short-lived single mutant has generally been interpreted as suggesting that the factors act in the same pathway. In contrast, an intervention extending the lifespan of the short-lived mutant has been interpreted as suggesting that the factors act in genetically distinct pathways. Specific examples of this type of comparison are studies in which dietary restriction (DR) fails to extend lifespan in yeast (Lin et al., 2000), invertebrates (Rogina & Helfand, 2004; Wang & Tissenbaum, 2006) and mice (Li et al., 2008) when Sir2-orthologs are mutated. These data have been, and continue to be, interpreted by some to support a model in which DR promotes longevity and healthspan through the activation of sirtuins (Baur et al., 2010).
It has been previously reported that deletion of SIR2 blocks replicative lifespan (RLS) extension from DR by reduction in glucose and in strains lacking GPA2 or HXK2, two genetic mimics of DR, but not in a strain lacking the rDNA replication fork block protein, FOB1 (Kaeberlein et al., 2004). To examine the influence of deleting SIR2 on RLS extension more generally, we generated 30 additional double mutant strains in which a RLS-extending deletion was combined with deletion of SIR2. We also tested three additional methods of DR involving growth on alternative carbon sources (ethanol, glycerol or raffinose). Strikingly, none of these interventions resulted in a significant RLS extension relative to sir2Δ cells (Figs 1 and S2; Table S1).
One possible interpretation of these data is that each of the RLS-extending interventions acts upstream of Sir2, perhaps by promoting Sir2 activity. Two observations are inconsistent with this model. First, at least eight single-gene deletions that increase wild-type RLS, and all four forms of DR, significantly extend the RLS of sir2Δ fob1Δ cells (Fig. S2; Table S1), demonstrating that SIR2 is not absolutely required for RLS extension in these cases. Second, at least five long-lived deletion mutants show no indication of enhanced Sir2 activity in vivo, as measured by rDNA recombination or rDNA silencing (Fig. S3). A similar lack of increased Sir2 activity has been previously reported in cells subjected to DR (Kaeberlein et al., 2005; Riesen & Morgan, 2009; Smith et al., 2009). Interestingly, deletion of TOR1 caused a significant decrease in rDNA recombination, but this effect was independent of SIR2 (Fig. S3A).
An alternative explanation for these data is that loss of SIR2 alters ageing such that molecular processes that do not limit RLS in wild-type cells become limiting in sir2Δ cells. Sir2 has multiple functions, including repression of extrachromosomal rDNA circle formation (Kaeberlein et al., 1999), enhancing global rDNA stability and silencing (Gottlieb & Esposito, 1989; Smith & Boeke, 1997), promoting asymmetric inheritance of damaged proteins (Aguilaniu et al., 2003) and maintaining telomeric chromatin during ageing (Dang et al., 2009). Our observation that only deletion of FOB1 is sufficient to suppress the short RLS of sir2Δ cells suggests that (i) the primary RLS-limiting defect in sir2Δ cells is likely related to rDNA instability and (ii) none of the 32 deletions tested that slow ageing in wild-type cells is able to overcome this defect. One prior study reported that overexpression of Hsp104 could also suppress the short RLS of sir2Δ cells (Erjavec et al., 2007), raising the possibility that accumulation of damaged proteins in sir2Δ mother cells may also contribute to the reduced longevity.
While it is likely that many of the genes examined in this study do not require Sir2 for their effect on RLS, we do not believe that all of the 32 long-lived single-gene deletion mutants examined here necessarily act via Sir2-independent mechanisms. For example, deletion of SAS2, a histone acetyltransferase known to antagonize Sir2 effects on chromatin (Dang et al., 2009), extends wild-type RLS but fails to extend the RLS of sir2Δ fob1Δ cells (Fig. S2B). Thus, both functional and genetic evidence suggest that Sas1 likely acts in the same longevity pathway as Sir2.
This study provides a clear demonstration of the challenges associated with interpreting longevity epistasis data. In particular, the failure of a longevity intervention to extend lifespan in a short-lived background may not be informative regarding the mechanism of lifespan extension in the wild-type context. In the absence of strong evidence indicating that the lifespan shortening is caused by acceleration of the wild-type ageing process, caution is warranted when interpreting these types of data.