Absence of effects of Sir2 overexpression on lifespan in C. elegans and Drosophila.

Overexpression of sirtuins (NAD(+)-dependent protein deacetylases) has been reported to increase lifespan in budding yeast (Saccharomyces cerevisiae), Caenorhabditis elegans and Drosophila melanogaster. Studies of the effects of genes on ageing are vulnerable to confounding effects of genetic background. Here we re-examined the reported effects of sirtuin overexpression on ageing and found that standardization of genetic background and the use of appropriate controls abolished the apparent effects in both C. elegans and Drosophila. In C. elegans, outcrossing of a line with high-level sir-2.1 overexpression abrogated the longevity increase, but did not abrogate sir-2.1 overexpression. Instead, longevity co-segregated with a second-site mutation affecting sensory neurons. Outcrossing of a line with low-copy-number sir-2.1 overexpression also abrogated longevity. A Drosophila strain with ubiquitous overexpression of dSir2 using the UAS-GAL4 system was long-lived relative to wild-type controls, as previously reported, but was not long-lived relative to the appropriate transgenic controls, and nor was a new line with stronger overexpression of dSir2. These findings underscore the importance of controlling for genetic background and for the mutagenic effects of transgene insertions in studies of genetic effects on lifespan. The life-extending effect of dietary restriction on ageing in Drosophila has also been reported to be dSir2 dependent. We found that dietary restriction increased fly lifespan independently of dSir2. Our findings do not rule out a role for sirtuins in determination of metazoan lifespan, but they do cast doubt on the robustness of the previously reported effects of sirtuins on lifespan in C. elegans and Drosophila.

The role of sirtuins in ageing was discovered in budding yeast (Saccharomyces cerevisiae), where over-expression of SIR2 increases replicative lifespan[5]. It was then reported that elevated sirtuin levels increase lifespan in the nematode C. elegans[1,2,6] and the fruitfly Drosophila[3], suggesting an evolutionarily ancient role of sirtuins in longevity assurance[7]. Dietary restriction (DR), reduced food intake short of starvation, extends lifespan in organisms ranging from yeast to mammals[8], and initial studies suggested that DR increases lifespan by activating sirtuins in yeast[9], C. elegans[10] and Drosophila[3]. Pharmacological activation of sirtuins has thus been widely promulgated as a potential means to mimic DR and slow ageing in humans[11]. However, several aspects of the role of sirtuins in ageing have proved controversial[12]. Subsequent studies have suggested that sirtuins do not mediate DR effects on ageing, at least in budding yeast and C. elegans[13,14]. The plant-derived polyphenol resveratrol and other compounds have been reported to activate sirtuins and extend lifespan[15,16]. More recent findings have challenged both effects[17-20]. We therefore re-examined the effects of sirtuin over-expression on lifespan in C. elegans and Drosophila. In particular, we wished to exclude the possibility that the increased longevity observed in strains with sirtuin gene over-expression are caused by differences in genetic background, or by the mutagenic effects of transgene insertion, which frequently confound studies of the genetics of ageing[4].

We first examined a high copy number sir-2.1 transgenic C. elegans strain (LG100) carrying the integrated transgene array geIn3 [sir-2.1 rol-6(su1006)] (Ref. 1). As expected, this strain was long lived (Fig. 1a; Table S1). However, outcrossing (x5) of geIn3 to wild type (N2) abrogated the increase in longevity (Fig. 1a, Table S1) without affecting SIR-2.1 protein levels (Fig. 1b). This loss of longevity upon outcrossing was verified by an independent research team (Table S2).

Figure 1

C. elegans: Longevity of LG100 and NL3909 is not attributable to sir-2.1 over-expression

a, b. Outcrossing of LG100 removes life extension without affecting SIR-2.1 protein levels. Data in b derived from Western blots (mean of three trials each using an independent protein preparation). A representative Western blot is shown in Fig S1a. qRT-PCR showed that sir-2.1 mRNA is also elevated in both strains (data not shown). c, LG100-derived Dyf, non-Rol segregant lines are long-lived while non-Dyf, Rol lines are not. d. Non-Dyf Rol segregant lines have elevated SIR-2.1 levels, while Dyf, non-Rol lines do not. e, f. sir-2.1 RNAi does not suppress LG100 longevity, but reduces SIR-2 protein levels. g, h. Outcrossing of NL3909 removes life extension without affecting SIR-2.1 protein levels. See Tables S1, S3, S4 and S5 for lifespan statistics for a, c, e and g, respectively. Error bars, S.E.M.. *0.01< P < 0.05; ** 0.001 < P < 0.01; ***P < 0.001, n.s., not significant; Student’s t test (two tailed). One remaining possibility is that the outcrossed sir-2.1 strains both contain second site mutations that suppress longevity effects. However, daf-2 RNAi strongly induced longevity in both (data not shown), arguing against the presence of a general suppressor of longevity in each case.

LG100 exhibited a neuronal dye-filling (Dyf) defect[22] that did not segregate with the transgene upon outcrossing (Fig. S2A). Dyf mutants often exhibit extended lifespan[23]. To determine whether the longevity of LG100 might be attributable to a dyf mutation, we derived from it three Dyf, non-Rol lines (lacking geIn3) and three non-Dyf, Rol lines (carrying geIn3). Dyf, non-Rol lines were long-lived and showed wild-type SIR-2.1 protein levels (Fig. 1c,d, Table S3). Non-Dyf, Rol lines showed elevated SIR-2.1 protein levels but had wild-type lifespans. Dyf mutant longevity appeared to be partially daf-16 dependent (Fig. S2B), as seen previously for other Dyf mutants[23]. The co-segregation of longevity with this dyf mutation but not geIn3 was previously noted by another research team (S.S. Lee, Cornell University, personal communication). Furthermore, knock-down of sir-2.1 expression in LG100 using RNA-mediated interference did not suppress longevity, despite lowering SIR-2.1 protein to wild-type levels (Fig. 1e,f; Table S4). Taken together, these results imply that the longevity of LG100 is attributable to an unidentified dyf mutation (or possibly another mutation closely-linked to the dyf locus), and that high level over-expression of sir-2.1 is not sufficient to increase lifespan in these strains.

A low copy number transgenic strain (NL3909) over-expressing sir-2.1 (Ref. 7) is also long-lived[2]. We confirmed the increased lifespan of NL3909 (pkIs1642 [sir-2.1 unc-119] unc-119(ed3)) relative to the control strain NL3908 (pkIs1641 [unc-119] unc-119(ed3)) (Fig. 1g, Table S5). We also observed an apparent elevation of SIR-2.1 protein in NL3909 relative to NL3908 (Fig. 1h). Outcrossing (x6) of NL3909 once again abrogated longevity (Fig. 1g, Table S5) without affecting SIR-2.1 protein levels (Fig. 1h, Fig. S1c). This effect of outcrossing was independently verified (Table S6). Thus, the longevity of NL3909 also appears to be attributable to genetic background effects rather than to pkIs1642.

The duplication mDp4 includes the sir-2.1 locus, and the mDp4–containing strain DR1786 is long lived[1]. We too found that DR1786 is long-lived, and also shows elevated sir-2.1 expression. However, longevity was not suppressed by sir-2.1 RNAi (Fig. S3, Table S7) suggesting causation by factors other than sir-2.1, either on mDp4 or elsewhere in the genome.

In Drosophila over-expression of dSir2 reportedly increases lifespan relative to wild-type controls[3]. Over-expression was achieved using the GAL4, UAS binary system[24], with the largest increases in lifespan produced by combination of EP-UAS-dSir2 (dSir2) with a ubiquitously-expressed tubulin-GAL4 driver. We outcrossed these two transgenes (x6) into the control white Dahomey (wDah) background. Assayed on a medium similar to that used in the original study, EP-UAS-dSir2/tubulin-GAL4 flies were longer lived than wild-type controls, as previously reported[3] (Fig. 2a). However, they were not longer lived than the tubulin-GAL4/+ control flies (Fig. 2a). This implies that life extension is due to transgene-linked genetic effects other than over-expression of dSir2. Lifespan was assayed on a range of food media (see Methods for details) to test for nutrient-dependence of any effect. However, in no case were EP-UAS-dSir2/tubulin-GAL4 flies longer lived than one or both transgenic controls (Fig. S4).

Figure 2

Drosophila: Absence of effects of dSir2 on lifespan

All lines were outcrossed into wDah (+/+) a, Lifespan in flies over-expressing dSir2 driven via tubulin-GAL4 (tub-GAL4) is longer than wild type, but not than the tubulin-GAL4 /+ genetic control. Median lifespans: +/+ = 39 days, dSir2/tubulin-GAL4 = 59 days, dSir2/+ = 53 days, tubulin-GAL4 /+ = 60 days. dSir2/tubulin-GAL4 vs. dSir2/+, P = 0.0006; dSir2/tubulin-GAL4 vs. tubulin-GAL4 /+, P = 0.9295; dSir2/tubulin-GAL4 vs. +/+, P <0.0001 . b, Lifespan in flies over-expressing dSir2-Myc9 is longer than in wild type, but not than in the tubulin-GAL4 control. Median lifespans: +/+ = 39 days, dSir2-Myc9/tubulin-GAL4 = 67, dSir2-Myc9/+ = 41 days, tubulin-GAL4/+ = 60 days. dSir2-Myc9/tubulin-GAL4 vs. dSir2-Myc9/+, P = 0.0001; dSir2-Myc9/tubulin-GAL4 vs. tubulin-GAL4/+, P = 0.1354; dSir2-Myc9/tubulin-GAL4 vs. +/+, P <0.0001. Compared using log rank test, n=200. c, Effect of dietary restriction on Drosophila lifespan is not dSir2 dependent. Flies were assayed over five concentrations of SYA media and data are presented as the median lifespan on each food concentration. All lines were outcrossed into Canton S (+/+). P values confirm that all flies respond normally to DR when median lifespans are compared for DR vs. fully-fed (FF) conditions[30].

Lack of an observable effect on lifespan could reflect the relatively modest increase in dSir2 expression in EP-UAS-dSir2/tubulin-GAL4 flies, both in terms of levels of mRNA (Fig. S5) and protein (+35% relative to wild type, Fig. S6). We therefore created lines with a higher level of over-expression of dSir2 (UAS-dSir2-Myc9/tubulin-GAL4). Here dSir2 mRNA and protein levels were robustly increased relative to wild type (+318% relative to wild-type protein levels; Fig. S5, S6). We examined recombinant protein raised in E. coli to check that the presence of the Myc tag did not interfere with dSir2 histone deacetylase (HDAC) activity, as measured by deacetylation of the fluorophore-containing p53 (Fluor de Lys) or native acetylated histone H4 substrates, and it did not (Fig. S7). We also found that dSir2 HDAC activity was unaffected by addition of resveratrol in either assay (Fig. S7). We saw no increase in lifespan in UAS-dSir2-Myc/tubulin-GAL4 flies relative to tubulin-GAL4/+ controls, on food medium similar to that used in the original study (Fig. 2b) or relative to either control on a range of other media (Fig. S4b,c,f). An independent research team also saw no increase in lifespan in tubulin-GAL4/UAS-dSir2-Myc9 flies (Fig. S8). These results suggest that the previously observed longevity of EP-UAS-dSir2/tubulin-GAL4 flies was not attributable to elevated expression of dSir2, and that stronger, ubiquitous over-expression of dSir2 also does not extend fly lifespan.

The role of sirtuins in the extension of lifespan by DR in yeast and C. elegans is controversial, with multiple groups reporting that sirtuins are not required for life span extension from DR in both organisms[8]. In Drosophila, it was reported that DR does not increase lifespan in dSir2 deletion mutant flies[3]. We tested this too, using the deletion alleles dSir2 (tested previously[3]) and dSir2. We first outcrossed these alleles (Fig. S9a) into the Canton S wild type (see Methods), used in the previous DR study[3]. We then checked the effect of each allele on dSir2 gene expression. dSir2 abrogated dSir2 mRNA, implying that this is a null allele. By contrast, dSir2, which contains a relatively small deletion at the 5′ end of the gene, did not reduce dSir2 mRNA levels (Fig. S9b,c).

To reassess the role of dSir2 in DR in Drosophila, we compared lifespans of wild-type (Canton S), dSir2 and dSir2 homozygotes. All genotypes responded similarly and normally to DR in trials conducted by two independent research teams (Fig. 2c, Fig. S10), hence the effect of DR on lifespan did not require dSir2.

In this study, we were unable to verify the effect of sirtuin over-expression on lifespan in either C. elegans or Drosophila. Increased lifespan was seen in two C. elegans lines with elevated sir-2.1 expression derived from independent studies, as previously reported, but in each case this was abrogated by outcrossing. sir-2.1 over-expression does exert effects on traits other than lifespan. For example, geIn3 is neuroprotective in a worm model of neuron dysfunction in Huntington’s disease[25] and, importantly, this effect is not attributable to the dyf mutation (Fig. S11). Moreover, both NL3909 and its outcrossed derivative are thermotolerant (M. Somogyvári, C. Sőti, unpublished data). In Drosophila, lines over-expressing dSir2 were longer-lived than wild type controls, as previously reported, but they were not longer lived than lines containing the appropriate transgenic controls. That all transgenic lines were longer lived than the Dahomey wild type into which they had been outcrossed could reflect heterosis in the vicinity of the transgene inserts, or a mutagenic effect of the GAL4 insert.

Lifespan was not increased by over-expression either of sir-2.1 from its own promoter in C. elegans, or dSir2 ubiquitously from a heterologous promoter in Drosophila. Our findings call into question the robustness of earlier reports of a role of sirtuins in longevity-assurance based upon over-expression in C. elegans and Drosophila, and also on the role of dSir2 in the response to DR in Drosophila. However, sirtuins can affect lifespan in animals under certain conditions: C. elegans daf-2(e1370) mutants are hypersensitive to genetic effects on lifespan[26], and here deletion of sir-2.1 reproducibly increases lifespan[6] (Fig. S12).

Our finding that resveratrol does not activate HDAC activity of dSir2 using a native histone H4 peptide is consistent with earlier findings with yeast SIR2 and mammalian SirT1 (Ref. 17,18). Resveratrol increased Drosophila lifespan in one study[27] but not another[21]. In principle, this could reflect sensitivity of resveratrol effects to subtle differences in culture conditions. If this were the case, our findings would imply that such effects are not attributable to direct activation of dSir2 by resveratrol.

METHODS SUMMARY

Nematode strains and maintenance

Nematodes were maintained on nematode growth medium (NGM) agar at 20°C, with Escherichia coli OP50 bacteria as a food source. Nematode strains used included: wild type (N2), GA707 wuEx166 [rol-6(su1006)] (rol-6 control), LG100 geIn3 [sir-2.1 rol-6(su1006)] dyf-?(wu250), NL3909 pkIs1642 [sir-2.1 unc-119] unc-119(ed3), and the control strain NL3908 pkIs1641 [unc-119] unc-119(ed3).

Nematode lifespan measurements

These were performed as described[28], at 20°C. To prevent progeny production, 5-fluoro-2′-deoxyuridine (FUdR) was added to seeded plates, to a final concentration of 10, 40 or 50 μM. Before testing effects of RNA-mediated interference (RNAi) on lifespan, worms were kept for 2 generations on the RNAi bacteria. Statistical significance of effects on lifespan was estimated using the log rank test, performed using JMP, Version 7 (SAS Institute).

Fly stocks and maintenance

tubulin-GAL4 and dSir2 were obtained from the Bloomington Stock Center. dSir2-Myc2 and dSir2-Myc9 lines were generated by germ-line transformation into strain w. dSir24.5/SM6B, dSir217/Cyo and Canton S were gifts from S. Pletcher, J. Rine and S. Helfand. All lines were outcrossed at least 6 times into the relevant controls. Experiments were performed at 25°C on a 12h : 12 h light-dark cycle at constant humidity.

Fly lifespan assays

Flies were bred at standard density, allowed to mate for 48 hours after emerging then sorted into 10 females per vial. Vials were changed every 48 hours, and deaths per vial scored until all flies were dead. Over-expression studies n=200. dSir2 mutant studies n=100. For statistical methodology, see above.

dSir2 deacetylation assays

We used both the SirT1 Fluorimetric Drug Discovery Kit (Enzo Life Sciences) and an HPLC-based acetyl-histone H4 deacetylation assay[29]. dSir2 and dSir2-Myc were cloned into pET SUMO (Invitrogen) and purified on HisPur cobalt spin columns (Thermo Scientific).