Technical Comments

Response to Comment on "Long-Lived Drosophila with Overexpressed dFOXO in Adult Fat Body"

Science  04 Feb 2005:
Vol. 307, Issue 5710, pp. 675
DOI: 10.1126/science.1104733

Reduced insulin/insulin-like growth factor signaling (IIS) has been shown to extend life span in Caenorhabditis elegans, Drosophila, and mice (13), and adipose tissue has been implicated in determination of life span extension by reduced IIS in these model organisms (36). We recently showed that dFOXO overexpression in adult female Drosophila fat body increases their life span, reduces their fecundity, and increases their resistance to paraquat (5). In contrast, Hwangbo et al. (6) failed to find any such life-span extension, using either the wild-type or an IIS-insensitive dFOXO transgene. On the basis of analysis of mortality data, Tatar (7) suggests that our conclusions were incorrect and that induction of dFOXO in the adult fat body did not increase life span.

Neither we (5) nor Hwangbo et al. (6) presented mortality rate data. Our conclusion that life span was extended was based on analysis of cumulative survival data. The age specificity of the effect awaits investigation with the necessary large cohorts of flies required for accurate mortality analysis. With the sample sizes (∼200 flies) that we used, day-to-day and age-specific variation in mortality rate is inevitable. Tatar (7) suggests that sporadic high, early mortality in our control groups explains our results. Additionally, he has incorrectly described the data that we provided him, which has led to an overestimation of deaths in our control groups. He states that “for instance, 57 control females died on day 22,” but this number is incorrect. Fifty-seven is the total number of deaths shown in figure 1C in (5); 50 control and 7 experimental females were dead on day 22, rather than the 57 controls.

Table 1 shows increases in the maximum life span (time of death of the last 10% of the flies to die) and the overall survival of the experimental flies in all three experiments in (5), as well as similar results in other data that we had gathered before the time of publication. To avoid any effects of data selection, we published the data from the first three experiments that we performed (5). Results from the additional experiments (two for each dFOXO transgenic line) (Table 1 and Fig. 1) also show highly significant increases in overall survival and maximum life span.

Fig. 1.

Survival experiments on flies given either +RU486 (solid circles) or –RU486 (open circles) food (replicates from Table 1). (A and B) Results of overexpression of dFOXO/S1106 driver (on chromosome 2 or 3) in females. (A) dFOXO on 3 (4)/S1106. (B) dFOXO on 2 (3)/S1106. (C) Control females expressing the S1106 driver.

Table 1.

Median life spans of females overexpressing dFOXO in the adult fat body using the S1106 Geneswitch driver and control females expressing the S1106 driver only. Also shown are the percentage change between treatments and the P values for comparison of survival and maximum life span between groups, calculated by log-rank and Wilcoxon tests, respectively. Log-rank tests were used for comparison of survival between groups. Wilcoxon nonparametric tests were used for comparing maximum life span between groups (last 10% of survivors). No maximum life-span data are available for replicates 3 (3) and 2 (2) because the experiments were terminated before all flies were dead.

Replicate Median (days) % Change Log rank (survival) Wilcoxon (maximum life span)
dFOXO on 3 (1) 36 (-RU486) 51 (+RU486) +41.6% P < 0.0001 P < 0.0001
dFOXO on 3 (2) 32 (-RU486) 39 (+RU486) +21.8% P < 0.0001 P = 0.0105
dFOXO on 3 (3) 52 (-RU486) 59 (+RU486) +13.5% P < 0.0001 NA
dFOXO on 3 (4) 47 (-RU486) 57 (+RU486) +21.3% P < 0.0001 P < 0.0001
dFOXO on 2 (1) 25 (-RU486) 38 (+RU486) +52% P < 0.0001 P = 0.0067
dFOXO on 2 (2) 38 (-RU486) 58 (+RU486) +52% P < 0.0001 NA
dFOXO on 2 (3) 39 (-RU486) 53 (+RU486) +35.9% P < 0.0001 P < 0.0001
S1106 39 (+RU486) 37 (-RU486) P = 0.1572

In our published account (5), we described results from another control experiment designed to rule out any S1106 Geneswitch-dependent, but dFOXO-independent, effects. In this experiment, we examined the life span of S1106/+ females, in the presence or absence of RU486 (Fig. 1C and Table 1). The presence of activated Geneswitch in the S1106/+ females has no significant effect on life span. Hence, our results clearly demonstrated that female flies overexpressing dFOXO in the adult fat body have extended mean and maximum life span compared with controls.

To make his point about early mortality in the control groups, Tatar (7) censored deaths occurring primarily in control groups over a 10-day period. However, arbitrary censoring of deaths at particular times—in the absence of any external evidence of problems with an experiment—constitutes selection of data and is a biased, and hence inappropriate, procedure. One of the major benefits of experiments with environmental inducers of gene expression, such as RU486, is that experimental and control groups are of the same genotype, are derived from the same batch of flies randomly allocated to experimental treatment, and are subsequently housed and treated under identical conditions. The food regimen differs only by the addition of RU486 or the carrier in the experimental and control groups, respectively. There was no external justification for the censorship of any data from these experiments.

The differences between the two sets of results (5, 6) await explanation. The differences could be related to differences in the composition of the fly food medium or the way in which the RU486 was administered. In addition, Hwangbo et al. (6) used mixed male and female cohorts, and the close confinement of mixed sex groups of flies is detrimental to the survival of both sexes and can lead to widely varying sex ratios during experiments (8, 9).


Related Content

Navigate This Article