Regulation of Drosophila Life Span by Olfaction and Food-Derived Odors

See allHide authors and affiliations

Science  23 Feb 2007:
Vol. 315, Issue 5815, pp. 1133-1137
DOI: 10.1126/science.1136610


Smell is an ancient sensory system present in organisms from bacteria to humans. In the nematode Caeonorhabditis elegans, gustatory and olfactory neurons regulate aging and longevity. Using the fruit fly, Drosophila melanogaster, we showed that exposure to nutrient-derived odorants can modulate life span and partially reverse the longevity-extending effects of dietary restriction. Furthermore, mutation of odorant receptor Or83b resulted in severe olfactory defects, altered adult metabolism, enhanced stress resistance, and extended life span. Our findings indicate that olfaction affects adult physiology and aging in Drosophila, possibly through the perceived availability of nutritional resources, and that olfactory regulation of life span is evolutionarily conserved.

As in many species, reduced nutrient availability (dietary restriction) increases life span in the fruit fly, Drosophila melanogaster, and leads to alterations in age-dependent patterns of gene expression, physiology, and behavior (14). Acute nutrient manipulation causes sudden and rapid changes in age-specific mortality (5, 6). Whole-genome expression data, containing age-dependent patterns of gene expression in diet-restricted long-lived flies and fully fed control flies (1), revealed that expression of genes encoding odorant-binding proteins was strongly affected by both age and nutrient availability (fig. S1).

To determine whether detection of food-related odors is sufficient to affect fly life span, we measured the life spans of flies in the presence and absence of odorants from live yeast. Yeast odorants were used because demographic and gene-expression data suggested that yeast availability is a major component of the longevity response to diet in Drosophila (79). Exposure to yeast odorants reduced life span in long-lived flies from two laboratory fly strains (Canton-S and yw) that had been subjected to dietary restriction (Fig. 1, A and C). Life span was further reduced when flies were allowed to consume yeast paste. The magnitude of the odorant effect was variable and usually small, relative to that caused by the consumption of yeast paste; odorant-mediated life-span reductions ranged from 6 to 18% in Canton-S flies and from 7 to 8% in yw flies (Fig. 1C). Such variability is reminiscent of the dietary-restriction response in flies, which depends on genetic background (8). Odorants are therefore sufficient to modulate life span, and currently unidentified odors may alter longevity with greater potency.

Fig. 1.

Exposure to yeast odorants alters life span in Drosophila. (A) Canton-S female flies were maintained under dietary restriction (5% SY medium) (triangles), with the addition of odorants from live yeast (circles) or yeast for consumption (squares). Sample sizes: control, n = 206; + odorant, n = 247; + live yeast, n = 193. (B) Canton-S female flies were unaffected by yeast odorants when fully fed (15% SY medium). Sample sizes: control, n = 269; + odorant, n = 267; + live yeast, n = 266. Error bars in (A) and (B) represent 95% confidence intervals on observed survivorship. (C) The effect of yeast odorants on longevity is repeatable and depends on diet. Replicate experiments showing mean life span (and SEM) for female Drosophila maintained under dietary restriction (5% SY, yellow symbols) and fully fed conditions (15% SY, blue symbols). Yeast odorants had no effect when flies were maintained in fully fed conditions (P > 0.05 in all cases). Each point represents life spans of 5 to 10 cohorts and a minimum of 206 (maximum of 403) animals. (D) Food consumption is unaffected by yeast odorants. Mean food consumption (and SEM) as measured by dye uptake (see the supporting online material) is shown. Multivariate profile analysis reveals no significant treatment effects (P > 0.05). The decline in dye uptake at later time points is a consistent finding that may reflect diurnal patterns of feeding and the dynamics of dye uptake and excretion. For survival data, block effects were adjusted for by analysis of variance (based on yeast paste treatments) to allow for comparisons across multiple experiments. P values were determined by means ofalog-rank test.

We tested whether diet-restricted flies might exhibit altered feeding behavior or altered investment in reproduction when exposed to nutrient-related odors, thereby accounting for the longevity effect. Increases in the intensity of either of these behaviors would reduce life span by compensating for our diet-restriction procedure or by augmenting the costs of reproduction, respectively. In our experiments, neither food consumption (as measured by the rate of dye ingestion, proboscis extension, and fecal output) nor fecundity was affected by yeast odorants (Fig. 1D and figs. S2 and S3). Behavioral alterations leading to increased nutrient intake or reproductive effort are therefore not responsible for reduced longevity upon exposure to yeast odorants.

The effects of yeast odorants on fly life span depend on nutrient availability. Longevity was not affected by yeast odorants when flies were fully fed (Fig. 1, B and C). Thus, the odor effect is a regulated biological response, and yeast odorants are not a generalized toxin that shortens life span. Our data support the hypotheses that diet- and odorant-mediated regulation of aging act at least partly through the same molecular pathway and that nutrient-related odors can rescue, albeit incompletely, the extension of longevity through dietary restriction. Consequently, the beneficial effects of dietary restriction may be due in part to the decreased perception of nutrient availability.

We next asked whether the loss of olfactory function is sufficient to increase life span. In vertebrates and insects, each olfactory neuron expresses a small number of odorant receptors that impart response characteristics of the neuron to specific odors (1013). Of the 62 putative odorant receptors in Drosophila, Or83b is atypical in that it is broadly expressed throughout olfactory tissues (14, 15). Or83b interacts with conventional odorant receptors and is required for their localization to the neuronal dendrite (1416). Loss-of-function mutations in Or83b limit spontaneous activity in many odorant-receptor neurons and severely reduce physiological and behavioral responses to a wide range of odorants (14, 16), including nutrient-related odors (fig. S4).

We measured the life span of flies carrying the Or83b2 allele, in which the first five of seven transmembrane domains were replaced with the w+ marker (14) (fig. S6). Or83b2 homozygous flies lack detectable levels of Or83b mRNA and protein, which suggests that it is a null allele (14). Fully fed female Or83b2 mutant flies exhibited a 56% increase in median life span when compared to appropriate wild-type animals (Fig. 2A). Males were also significantly long-lived, but the magnitude of the extension was generally smaller (Fig. 2B). Heterozygous flies exhibited intermediate longevity in both sexes, and heterozygous adult females showed a similar deficiency in attraction to live yeast paste (fig. S4, B and C). We found no evidence of such impairment in heterozygous mutant males (fig. S4D). It may be that odor-evoked behaviors are less affected or that different classes of odorants are critical for male longevity. Longevity was also extended when olfactory signaling was suppressed in Or83b-expressing neurons through the disruption of guanine nucleotide–binding protein (G protein) signaling (fig. S5). Because the Or83b mutant also disrupts G protein activity, it is possible that G protein function in olfactory neurons, rather than perception per se, influences life span.

Fig. 2.

Mutation of Or83b increases life span. (A and B) Both female and male flies carrying the Or83b2 mutation are long-lived. Sample sizes (females): +/+, n = 326; +/–, n =314; –/–, n = 320. Sample sizes (males): +/+, n = 317; +/–, n = 346; –/–, n = 305. Error bars represent 95% confidence intervals, and statistical comparisons by means of a log-rank test and Cox regression yielded P <1 × 10–6 for all paired comparisons except (B) (males) where P = 0.001 for +/+ versus +/–. (C) Or83b2 homozygous mutant flies expressing UAS-GFP:Or83b under control of the Or83b-Gal4 driver (bicolor triangles) have comparable life spans to those of background control animals (w1118, open circles). Controls for the transgenic constructs had no effect on life span (compare solid red and blue lines to green triangles). Data are presented for females and are qualitatively similar for males (fig. S7). Experiments were carried out with the use of 15% SY media and were repeated by means of independent transgenic insertions with identical results. Error bars represent 95% confidence intervals on observed survivorship.

To verify that the extended life span of Or83b2 flies was not due to heterosis or to comparison against a relatively weak control stock, we backcrossed the Or83b2 allele into two additional laboratory stocks (Canton-S and yw). In all cases, the longevity of mutant flies was considerably greater than that of their wild-type controls (Tables 1 and 2). The degree of life-span extension was independent of the longevity of the corresponding wild-type stock, establishing that loss of Or83b function extends life span in healthy animals (17).

Table 1.

Life-span data for female Or83b mutant Drosophila. Mean life spans of Or83b2 homozygous mutant and wild-type control females. The Or83b2 allele was backcrossed into each of three genetic backgrounds. For all comparisons, longevity extension in the mutant was calculated with respect to the wild-type background to which it had been backcrossed. All increases are statistically significant (P < 1 × 10–6), determined by means of a log-rank test and Cox regression. *YP represents live yeast paste added to the vials.

Genetic backgroundNutrient level (% SY)Control longevityOr83b longevityAbsolute change% Increase
n Mean(SE) n Mean(SE)
3% 319 50.1 (0.77) 311 61.6 (0.86) 11.5 23.0%
5% 319 50.1 (0.81) 304 62.5 (0.78) 12.3 24.6%
7.5% 324 43.9 (0.80) 315 58.9 (0.83) 15.0 34.1%
10% 313 44.8 (0.82) 318 58.6 (0.92) 13.8 30.7%
15% 314 41.6 (0.69) 329 55.6 (0.76) 14.0 33.7%
3% 249 58.0 (0.85) 234 76.1 (0.63) 18.1 31.2%
5% 244 60.0 (0.83) 239 76.5 (0.60) 16.5 27.6%
7.5% 254 62.0 (0.74) 236 76.0 (0.76) 14.0 22.6%
10% 241 61.9 (0.74) 238 75.6 (0.81) 13.7 22.1%
15% 242 50.5 (0.57) 246 69.4 (0.60) 18.9 37.4%
YP* 237 42.9 (0.84) 243 65.7 (0.73) 22.8 53.1%
3% 315 51.9 (0.89) 318 75.6 (0.86) 23.7 45.6%
5% 326 53.2 (0.90) 320 76.0 (0.81) 22.8 43.0%
7.5% 333 49.6 (0.83) 319 78.0 (0.77) 28.4 57.3%
10% 316 51.1 (0.79) 317 77.6 (0.84) 26.5 51.9%
15% 314 49.3 (0.79) 319 72.7 (0.78) 23.4 47.3%
Table 2.

Life-span data for male Or83b mutant Drosophila. Mean life spans of Or83b2 homozygous mutant and wild-type control males. The Or83b2 allele was backcrossed into each of three genetic backgrounds. For all comparisons, longevity extension in the mutant was calculated with respect to the wild-type background to which it had been backcrossed. All increases are statistically significant (P < 1 × 10–6), determined by means of a log-rank test and Cox regression.

Genetic backgroundNutrient level (% SY)Control longevityOr83b longevityAbsolute change% Increase
n Mean(SE) n Mean(SE)
3% 322 52.1 (0.83) 317 66.8 (0.75) 14.7 28.2%
5% 315 53.2 (0.67) 282 66.3 (0.80) 13.1 24.6%
7.5% 331 52.3 (0.71) 313 68.7 (0.77) 16.4 31.3%
10% 330 52.0 (0.76) 296 67.5 (0.82) 15.6 30.0%
15% 325 47.2 (0.74) 301 67.1 (0.85) 19.9 42.1%
3% 257 63.6 (0.81) 238 73.8 (0.57) 10.3 16.2%
5% 242 63.9 (0.89) 243 75.2 (0.57) 11.4 17.8%
7.5% 251 64.4 (0.75) 240 74.3 (0.72) 9.9 15.4%
10% 246 65.3 (0.92) 246 75.8 (0.67) 10.5 16.0%
15% 244 60.1 (0.73) 222 70.3 (0.69) 10.1 16.8%
3% 317 63.2 (0.80) 328 69.1 (0.72) 6.0 9.5%
5% 314 63.9 (0.77) 322 69.5 (0.77) 5.6 8.7%
7.5% 322 63.6 (0.80) 324 73.6 (0.74) 10.0 15.7%
10% 317 65.2 (0.77) 310 74.6 (0.77) 9.4 14.5%
15% 317 64.4 (0.80) 305 73.8 (0.71) 9.4 14.5%

Expression of a rescuing Or83b transgene (15) under the control of an Or83b-Gal driver (14) restored normal life span to the Or83b2 mutant flies (Fig. 2C). The effectiveness of this construct was verified by genomic polymerase chain reaction and visually (fig. S6), and it rescues most (but not all) olfactory phenotypes (14, 15). The Or83b-Gal4 driver and the upstream activating sequence–green fluorescent protein Or83b transgene (UAS-GFP:Or83b) were inserted into different genomic positions and none affected life span on their own (Fig. 2C and fig. S7). These rescue data and the persistence of the longevity phenotype through extensive backcrossing to three different genetic backgrounds provide compelling evidence that loss of function in Or83b is the cause of increased life span in these animals.

Olfactory signaling modulates life span primarily by altering the onset of demographic senescence (Fig. 3). Mortality analysis suggests that olfaction shifts the mortality curve to earlier (in the case of yeast odorants) or later (in the case of Or83b mutation) ages. In females, the rate of increase in mortality with age was largely unaffected (Fig. 3, A and B, and fig. S8), an effect that is similar to diet restriction (1, 5). In males, the impact of olfaction on mortality rate was reduced later in life; mortality trajectories converge at the oldest ages (Fig. 3C).

Fig. 3.

Olfaction modulates the onset of demographic senescence. (A) Age-specific mortality rates for flies exposed to yeast odorants expressed on the natural log scale. Corresponding survival data are presented in Fig. 1A. (B and C) Age-specific mortality rates for Or83b mutant (B) females and (C) males and their controls (see also fig. S8). Corresponding survival data are presented in Fig. 2, A and B. Points represent observed mortality rates. Lines are obtained by smoothing hazard rates (3-day window) by means of a kernel smoother prior to log transformation.

We next investigated whether olfactory function was required for longevity extension through diet manipulation. We measured male and female life span in different nutritional regimes ranging from severe nutrient restriction [3% sugar/yeast (SY) media] to nutrient-replete conditions (15% SY media). Flies homozygous for the Or83b2 mutation were consistently longer-lived than those of appropriate control stocks (Tables 1 and 2). Despite their exceptional longevity, life span was further increased in the Or83b2 mutants by dietary restriction. Or83b is therefore not required for diet-mediated longevity. Consistent with the yeast odorant results, however, we do find evidence for interaction between the olfactory and diet pathways. The relative increase in median and mean longevity in Or83b2 flies was significantly greater when flies were maintained in well-fed conditions (Tables 1 and 2), and mutant animals were partially resistant to changes in diet (fig. S9). Thus, the Or83b mutation extends longevity largely, but not exclusively, through a diet-independent pathway.

Reduced early reproductive output is not required for extended longevity in Or83b2 mutants. We observed the largest life-span extension in very high nutrient conditions where flies were provided access to live yeast paste (see Table 1), and, under these conditions, homozygous Or83b2 mutant females had equal or greater reproductive output than control females (Fig. 4A). We consistently observed that Or83b2 mutants showed moderately reduced egg production from 24 to 48 hours post-eclosion, but the reason for this is unclear. Feeding rates are not reduced in mature flies (fig. S10), but reduced reproduction may be due to a delay in the onset of adult feeding, because the chemotaxis ability of mutant flies is compromised.

Fig. 4.

Or83b2 flies exhibited altered physiology and enhanced stress resistance. (A) Total reproductive output through day 14 is not different between Or83b mutant and control females. Data represent average daily egg production per female (n = 60 females per genotype). Error bars represent 95% confidence intervals on the mean. Or83b2 mutants had significantly lower egg production over the first observation period (two-tailed Student's t test, P = 0.0005) and higher levels of egg production at age 8 and 12 days (two-tailed Student's t test, P = 5.7 × 10–5 and P = 0.009, respectively). (B and C) Or83b2 homozygous mutant flies are starvation resistant. Sample sizes (females): +/+, n = 79; –/–, n = 79. Sample sizes (males): +/+, n = 100; +/– n = 94. Statistical comparisons and P values were determined by means of a log-rank test. Error bars represent 95% confidence intervals on observed survivorship. (D) Or83b2 homozygous mutant flies have increased triglyceride levels. Triglyceride levels are presented after normalization to total weight. Error bars represent 95% confidence intervals. TAG, triacylglycerol. (E and F) Weight and CO2 production for Or83b2 homozygous mutant and wild-type flies. Each point represents one fly (n = 15 for each sex and genotype). Box plots represent the median (horizontal line), 95% confidence intervals (box), and range (whiskers) of the data.

Or83b2 flies exhibited a range of phenotypes indicative of altered physiology and enhanced stress resistance. Females were more resistant to hyperoxia (mean longevity = 110 ± 2.04 hours and 123 ± 1.0 hours for control and Or83b2 mutant females, respectively; P <1 × 10–6, determined by means of a log-rank test). Mutants are resistant to starvation (Fig. 4, B and C), and females have significantly elevated levels of triglyceride, the primary lipid-storage molecule in Drosophila (Fig. 4D), despite their similar overall size and weight (Fig. 4E). Mutant males have higher but statistically indistinguishable levels of triglyceride, which suggests that life span and fat content are separable in this sex. The observed increases in longevity and stress resistance do not result from decreased metabolic rate (Fig. 4F).

Aging and longevity in Caenorhabditis elegans are regulated by sensory function through antagonistic effects of specific gustatory and olfactory neurons (18, 19). Although the specific environmental cues that regulate longevity in C. elegans are unknown, sensory regulation of aging largely involves insulin/IGF (insulin-like growth factor) signaling (18). Modulation of aging by gustatory neurons is entirely insulin signaling (i.e., daf-16)–dependent, whereas longevity extension by the ablation of olfactory neurons has a large daf-16–independent component (18).

Sensory systems and insulin-mediated longevity regulation are evolutionarily conserved (20, 21). Thus, as in C. elegans, olfaction may affect aging in Drosophila through altered insulin signaling and subsequent modulation of transcription factor dFOXO (the fly ortholog to daf-16). However, expression levels of Drosophila insulin-like peptides show no consistent differences in Or83b2 mutant flies (fig. S11). Consistent with normal levels of insulin signaling, we found that expression of Thor—the Drosophila homolog of mammalian 4E-BP and a primary target of dFOXO (22)—is not elevated in the body of mutant animals (fig. S11). Olfactory regulation of aging in Drosophila may therefore contain a substantial component that is independent of insulin signaling.

We have identified a nutrient-related olfactory cue (odorants from live yeast) and a gene involved in olfaction (Or83b) that limit fly life span. Olfactory-receptor function constrains the beneficial effects of dietary restriction, indicating that consumption is not the only way that nutrient availability modulates longevity (4). Genetic dissection of the roles of conventional odorant receptors in the life span of Drosophila may reveal additional candidate odors and neural circuits for longevity regulation.

Supporting Online Material

Materials and Methods

Figs. S1 to S11


References and Notes

View Abstract

Navigate This Article