Drosophila melanogaster has been a leading model for the study of aging for over 80 years [1,2,3,4,5]. The intensive use of Drosophila as a model for developmental biology has produced a wealth of genetic and molecular biological tools that are readily adapted to the study of aging. Aging is associated with characteristic changes at the physiological and molecular level, however organismal life span is still the best measure of aging rate. The most successful studies of aging generally involve manipulations that increase life span. Experimental alterations of environment or genetic makeup that cause decreased life span might create novel diseases or pathologies that do not usually limit life span in a normal individual. In contrast, a manipulation that increases life span is thought to be more likely to identify mechanisms that normally limit the life span of the organism.

Laboratory selection of Drosophila populations for late-life reproduction results in extended life span [6,7,8]. The genetic selection experiments demonstrate that life span has a large genetic component and is highly plastic. Increased life span was generally correlated with increased stress resistance including increased oxidative stress resistance [9,10]. An important question in the study of aging in invertebrates is whether life span can be increased without a trade off with metabolic activity. Certain manipulations that decrease metabolic activity cause increased life span, such as lower culture temperature [11]. In the genetic selection experiments, life span is increased without a reduction in metabolic activity [12,13].

Single gene mutations have been identified that increase Drosophila life span. These mutations identify negative regulators of life span, as mutations expected to decrease activity of the gene lead to increased life span. mth encodes a protein related to G-protein coupled receptors, and appears to negatively regulate life span, stress resistance and body size [14]. Indy is related to a mammalian cotransporter involved in membrane transport of Krebs cycle intermediates, leading to the suggestion that Indy mutations might increase life span by decreasing the availability of nutrients [15].

Certain mutant alleles of either the Inr or the chico genes can increase Drosophila life span [16,17]. Inr is homologous to the mammalian insulin receptor and insulin-like growth factor receptor, while chico is homologous to the mammalian insulin receptor substrate. The data indicate that an insulin-like signaling pathway negatively regulates life span, as had previously been found for the nematode C. elegans [18]. This pathway for life span regulation may be even more highly conserved during evolution, as there are many similarities with life span regulation in mammals and in yeast. The increased life span regulated by the insulin-like signaling pathway in C. elegans is associated with increased resistance to stress, including oxidative stress [18,19,20,21]. This has led to suggestions that increased oxidative stress resistance and decreased amounts of oxidative damage may be the mechanism by which mutation of the insulin-like signaling pathway causes increased life span. Consistent with this idea, life-span extension in C. elegans is associated with increased expression of the gene encoding the mitochondrial antioxidant enzyme manganese-containing superoxide dismsutase (MnSOD) [22]. In Drosophila, mutations of chico that extend life span increase total SOD enzyme activity; however, resistance to oxidative stress was not detectably increased, at least not as assayed by paraquat resistance [16]. Therefore it remains to be determined if mutation of the insulin-like pathway in Drosophila increases life span by increasing SOD activity or by some other mechanism(s).

Increased life span caused by overexpression of a gene by definition identifies a positive regulator of life span. Several genes involved in oxidative stress defense have been found to increase life span when overexpressed in transgenic flies. These include genes for both forms of SOD, the Cu/ZnSOD found in the cytoplasm and outer mitochondrial space, and MnSOD found in the inner mitochondrial space [23,24,25]. These enzymes convert the most common oxygen radical produced by the mitochondria, superoxide, into hydrogen peroxide and water. Abundant catalase enzyme then converts the hydrogen peroxide into molecular oxygen and water. Overexpression of Cu/ZnSOD or MnSOD increases life span up to about 40%, and the amount of life span increase is found to be proportional to the amount of enzyme increase in each case. The life span increase caused by SOD overexpression is not associated with a detectable tradeoff in metabolic activity, as oxygen consumption was normal throughout the extended life span of the flies [24]. Overexpression of the antioxidant enzyme peptide methionine sulfoxide reductase is also associated with increased life span [26]. Finally, several genes implicated in repair and stress responses have been shown to be able to extend Drosophila life span when the flies are cultured at elevated temperatures [27,28,29,30].

Testing specific genes by overexpression in transgenic flies is a candidate gene approach, meaning that the investigator must make an educated guess as to which genes are worth assaying by this relatively labor-intensive method. Recently genetic methods have been developed that allow for the efficient overexpression of random genes in Drosophila. P type transposable elements have been engineered to have transcriptional promoters directed out through the end of the element [31,32]. These elements will often insert upstream of gene coding regions, causing overexpression of the gene and mutant phenotypes. The PdL P element contains an outwardly directed tet-on promoter [33,34]. The transcription of this promoter is activated upon feeding the fly doxycycline, leading to conditional gene overexpression and conditional mutations. Conditional gene overexpression mutations have several potential advantages for study of aging and life span. By waiting until flies are young adults to activate gene overexpression, effects can be identified that are specific to adult aging. The system also provides powerful controls for genetic background effects, because control (no DOX) and over-expressing (+DOX) flies have identical genetic backgrounds and therefore any differences in life span must be due to DOX administration and gene overexpression. The PdL system allows the investigator to search for genes that can increase life span without having to guess ahead of time which genes these might be. Therefore the approach has the potential to identify genes and pathways not previously known or suspected to be involved in life span regulation.

In the rtTA transgenic construct the powerful and tissue-general actin5C promoter drives expression of the rtTA transcriptional transactivator [33]. When flies are fed doxycycline, rtTA binds to target sequences in the synthetic tet-on promoter in the PdL P element. This outwardly-directed promoter often causes overexpression of genes near the PdL insertion site and yields various mutant phenotypes [34]. PdL is 8.4 kilobases (kb) long and contains the P element inverted repeats, the mini-white+ transformation marker gene, and the 500 base-pair (bp) tet-on promoter directed out through the 3' end of the element. The tet-on promoter contains no open reading frame (ORF) or translation initiation sequences. Therefore PdL must insert upstream of an endogenous translation intiation site (ATG) to cause overexpression of a protein. The transcripts initiated within PdL will therefore contain 280 bp of 5' untranslated sequence derived from PdL itself.

To search for genes that might extend life span, new insertions of PdL were created on the second and third chromosomes by appropriate crosses to a strain expressing P element transposase (Figure 1). Approximately 10,000 males bearing a new PdL insert and an rtTA transcriptional transactivator insertion were generated. The males were generated in eight staggered groups, each of around 1,200 individuals. In an attempt to increase the variety of mutations identified, two different rtTA insertion strains were used. The first six cohorts of males contained an insertion of the rtTA construct called rtTA(3)E2 that is associated with moderate levels of target-gene expression and low levels of leaky expression in the absence of DOX [33]. The last two cohorts of males contained an rtTA insertion called rtTA(3)M1 which is associated with a high level of target-gene expression and greater leaky expression. To obtain adult flies of defined age, the second cross was cultured at 25°C in urine specimen bottles (Figure 1). Prior to eclosion of the majority of pupae, bottles were cleared of adults and flies were allowed to emerge over the next 48 hours. The majority of the males will have mated during this time. The males only were then removed and were designated 1 day old, and were maintained at 25°C at 40 per vial in culture vials with food supplemented with 250 μg/ml DOX, and passaged to new vials every 48 hours. The expectation was that some of these flies might live longer due to the overexpression of a beneficial gene. For the first four cohorts, at approximately 30% survival of the cohort, each of the males was individually mated with four third-chromosome balancer stock virgin females. This was so that a stable line could be recovered later if that fly was found to be one of the longest lived. Only about 21% of these males were found to be fertile, meaning that there was selection for males that maintained fertility during aging. After 4 days, the fertile males were removed from the crosses, placed in individual numbered vials, and passaged every two days until they were all dead. For the rest of the cohorts the males were mated at approximately 65% survival of the cohort, which increased the frequency of fertile males to about 32%. Stable lines were recovered for the approximately 1.4% longest-lived flies from each cohort. These 144 lines should therefore be enriched for ones in which PdL is causing overexpression of a gene with a benefit for life span. In addition, lines in which PdL causes overexpression of a gene with large negative effects on life span should have been eliminated by this step.

Each of the 144 new PdL insertion lines was crossed to the appropriate rtTA chromosome strain, and males containing both constructs were assayed in large cohorts (around 400 flies) for doxycycline-dependent effects on life span (Figure 2). The mean life spans of the different strains in the absence of DOX varied greatly, from 40 to 84 days. Both increases and decreases were observed in the presence of DOX. For the 110 lines tested with rtTA(3)E2 there was an average 1.8% increase in the presence of DOX (Figure 2a), and for the 34 lines tested with rtTA(3)M1 an average 3.0% increase in the presence of DOX (Figure 2b), consistent with previous results that DOX itself has little to no effect on life span. However, several lines were identified where there was a significantly greater than average increase in life span in the presence of DOX. To confirm these results, these lines were assayed one or more additional times (Table 1). As a control, Oregon-R wild-type flies were repeatedly crossed to the rtTA strains and assayed for life span in parallel. The Oregon-R controls exhibited changes in life span from -5.14% to +4.71%, with an average change of +0.45%. In contrast, six PdL lines were found to yield significant and reproducible increases in life span in the presence of DOX, ranging from 5 to 17%.

Line PdL(3)3E36 was associated with the largest increase in life span (average 12%) (Table 1). The PdL insert in this line was located in the large first intron of the encore gene [35], with the promoter oriented in the sense direction (Figure 3a). PdL caused the expression of a pair of approximately 1.4 kb transcripts that contain a 44 amino acid ORF identified by the Drosophila genome project as CG14975, as determined by northern blot with a CG14975 probe. These transcripts were not detected in the absence of DOX in the adult males and this putative gene was named Red herring (Rdh). Northern analysis using a probe from encore exon 3 indicated that levels of the normal sized encore transcript were not detectably altered by DOX, as indicated by the arrowhead (Figure 3a). However, DOX appeared to induce expression of both larger and smaller RNAs containing encore coding-region sequences, resulting in a smear of hybridization in the upper part of the lane, as indicated by the asterisk. Representative survival curves are presented (Figure 3a).

Line PdL(2)4G14 contains an insert at the 5' end of the VhaSFD gene, encoding the vacuolar proton pump ATPase (H⁺-ATPase) SFD subunit [36]. DOX caused an approximately fourfold overexpression of the VhaSFD transcript and an average 8% increase in life span (Figure 3b). Representative survival curves for the PdL(2)4G14 line are presented (Figure 3b). Line PdL(3)2C33 contains an insert at the 5' end of a gene with homology to a maltose permease from Bacillus stearothermophilus [37], which was named Sugar baby (Figure 3c). DOX caused an approximately 8.5-fold overexpression of the Sugar baby transcript, and was associated with an average 6% increase in life span.

Three genes with previously characterized mutant phenotypes were also identified by PdL insertions. Line PdL(3)8S64 contained an insert at the 5' end of the filamin gene 'A' transcript (Figure 3d). Filamin is an actin-binding protein involved in cytokinesis and ring canal formation [38,39]. filamin A was overexpressed approximately fivefold in the presence of DOX and associated with an average 8.5% increase in life span. Expression of the filamin 'B' transcript was not detectably altered. Line PdL(3)8S25 contained an insert at the 5' end of the 'B' transcript of the four wheel drive (fwd) gene (Figure 3e). fwd encodes a 1-phosphatidylinositol 4-kinase (PI 4-kinase) that regulates actin organization and ring canal formation during germline cytokinesis [40]. fwd B was overexpressed approximately 10-fold and associated with an average 8% increase in life span. Finally, line PdL(3)8R128 had an insert at the 5' end of the CTP:phosphocholine cytidylyltransferase 1 (Cct1) gene (Figure 3f). Cct1 encodes a homolog of the rate-limiting enzyme in phosphatidylcholine biosynthesis [41]. Cct1 was overexpressed approximately 1.7-fold and was associated with an average 7% increase in life span.

The genetic screen described here had several unusual aspects, and the potential to identify a novel set of genes involved in life-span regulation. First, the screen was designed to identify positive regulators of life span, that is, genes that will increase life span when overexpressed. Most genetic screens for life-span regulators in Drosophila and other organisms, such as C. elegans, have involved loss-of-function mutations and therefore have identified negative regulators of life span [14,15,42,43]. Second, mutants were identified on the basis of the phenotype of increased life span under normal conditions, as opposed to some surrogate phenotype expected to correlate with life span, such as increased stress resistance. Because the PdL mutations generated here were screened directly for extended life span under normal conditions, the screen had the potential to identify genes and pathways not previously suspected to have a role in aging and life-span regulation.

Results were reported recently for a screen for positive regulators of life span using a different P-element-based gene overexpression strategy [29]. In that study a heat-stress inducible system was used, and lines were screened for life span at the moderate heat stress condition of 30°C. Several genes were identified that increased survival when overexpressed, including genes involved in heat stress and oxidative stress responses. It is interesting to note that there is no overlap with the genes identified here with PdL at 25°C. This may be due to the small fraction of the genome that has been tested so far, or it may indicate that different mechanisms limit life span under normal and moderate heat-stress conditions.

It is difficult to estimate the number of genes that were tested in the PdL screen, but it is certain to have been only a tiny fraction of the genome. In previous experiments approximately 7% of PdL insertions were found to cause lethal and visible phenotypes when the PdL promoter was activated during larval and pupal development [34]. Seven percent is certainly a large underestimate of the frequency at which genes are overexpressed, as embryogenesis was not tested, and not all genes will be lethal or have an obvious visible phenotype when overexpressed. The original 10,000 PdL insertions generated here went through an initial enrichment step of unknown efficiency, to yield the 144 lines that were tested in large cohorts. Lines in which a gene was overexpressed that had a positive effect on life span should have been enriched in the final 144 lines, whereas ones with a significant negative effect should have been eliminated. The enrichment step appears to have been at least partially successful, as the largest negative effect on life span observed among the 144 lines tested was -7.5%, and PdL is known to be able to create mutations with negative effects of up to -30% [34]. The molecular characterization of the mutations suggests a low frequency of false positives. If the six PdL insertion strains were a random set, then PdL would be expected to sometimes be in the wrong orientation and/or downstream of the ATG translation start. However, all six lines were found to have PdL insertions at the 5' end of a gene and oriented in the sense direction, and in all six lines a transcript was overexpressed that was expected to contain a complete ORF.

The magnitude of life-span increases caused by the PdL mutations was generally small, ranging from 5 to 17%. This is in contrast to overexpression of SOD genes or certain single gene mutations such as mth, where life-span increases range from 20 to 50% [14,23,24,25]. There are at least two possible reasons for the relatively small life span increases observed here. First, as only a fraction of the genome has been tested, it may be that genes with larger effects exist but have not yet been identified. Second, the tet-on system is sometimes leaky, meaning that at certain genomic locations there can be significant expression of the tet-on promoter in the absence of DOX [33]. It is possible that in some of the lines the increase in life span between -DOX and +DOX is reduced because there is already some gene overexpression and life-span extension in the -DOX flies. Such a line might be expected to have an unusually high -DOX life span relative to the other lines, and it is notable that this appears to be the case for the lines bearing mutations in the VhaSFD and Sugar baby genes. When this difference is included in the calculations, the life span increases for VhaSFD and Sugar baby are approximately 20% and 17%, respectively.

The largest average increase in life span caused by DOX (12%) was obtained for the PdL(2)3E36 insertion in the large first intron of the encore gene. A pair of approximately 1.4 kb transcripts was expressed containing a 44 amino acid ORF with no detectable homologies. While there are a number of Drosophila genes encoding functional ORFs of this size, it is also possible that these transcripts do not encode a functional protein but rather function at the RNA level to affect life span. There are no cDNAs in the databases containing these intronic sequences, and we were unable to detect these transcripts in wild-type adult males using northern blots. The data suggest that these transcripts are not expressed under normal conditions, or are expressed at very low levels. Further experiments will be required to determine the mechanism of life-span extension in this line, and the elusive nature of the intronic gene suggested the name Red herring. The northern analyses also indicated induction of a small amount of a large transcript containing the encore coding region. Therefore it is also possible that the life span extension in this line is caused by an increase in the expression of encore, encore is a member of a novel family of proteins with multiple functions during Drosophila oogenesis [44].

Two of the genes identified appear to be involved in membrane transport. The VhaSFD gene encodes the Drosophila homolog of the vacuolar H⁺-ATPase SFD subunit. The vacuolar ATPase is an ATP-driven proton pump found in the endomembranes of eukaryotic cells [36,45,46,47]. It is composed of a cytoplasmic ATPase domain called V₁ that contains VhaSFD and other subunits, and a multisubunit membrane-channel component called V₀. The vacuolar ATPase is involved in the acidification and energization of organelles such as Golgi-derived vesicles, clathrin-coated vesicles, synaptic vesicles and lysosomes. The proton pumping is required for processes such as protein trafficking, receptor-mediated endocytosis, neurotransmitter release and intracellular pH regulation and waste management. The free membrane V₀ domain has also been implicated in membrane fusion events. In higher organisms the vacuolar H⁺-ATPase is also found in the plasma membrane of specialized epithelial cells where it functions in bulk transport. In Drosophila this is the Malpighian (or renal) tubule. The VhaSFD subunit is required for the ATPase activity and assembly of the vacuolar H⁺-ATPase and may function as a dissociable regulatory subunit. Therefore, overexpression of VhaSFD may affect life span by increasing activity of the vacuolar H⁺-ATPase.

The Sugar baby gene is related to a maltose permease from Bacillus stearothermophilus [37], suggesting that it may function in transport of maltose or other sugars across a membrane. Recently, loss-of-function mutations in another putative membrane transporter, Indy, have been found to increase life span [15]. It is possible that overexpression of the putative transporter encoded by Sugar baby could increase life span by increasing uptake of a particular sugar and altering energy metabolism pathways.

Filamin is an actin-binding protein with a homolog in humans called actin-binding protein 280 (ABP280) [38,39]. Filamin binds to several cell-membrane proteins and intercellular ligands involved in signal transduction, and appears to act as a downstream effector in remodeling of the actin cytoskeleton. During Drosophila oogenesis, filamin localizes to the ring canals, which are membrane-associated actin cytoskeletal structures that create intercellular bridges between germline cells. Loss-of-function mutations of filamin disrupt the normal function of these actin cytoskeletal structures. Filamin has several additional interactions with potential relevance to aging studies. Presenilins are membrane proteins found in both humans and Drosophila, and mutations in human presenilins are associated with early onset familial Alzheimer's disease (FAD) [48]. Drosophila filamin binds to both Drosophila and human presenilins, and filamin interacts genetically with presenilin in Drosophila [49]. Dominant adult phenotypes produced in Drosophila by overexpression of presenilin can be suppressed by overexpression of filamin. Filamin also interacts specifically with Drosophila Toll and Tube proteins [50]. Toll and Tube are members of a signal transduction pathway that activates the Rel transcription factor Dorsal and is involved in developmental patterning and in the adult immune response. Several of the immune-response genes regulated by Rel transcription factors are dramatically induced during Drosophila aging (G.L., J. Carrick and J.T., unpublished data).

fwd encodes a phosphatidylinositol 4-kinase (PI 4-kinase) that converts phosphatidylinositol (PI) to phosphatidylinositol 4-phosphate (PIP) [40]. This is the first step in the synthesis of the key regulatory membrane phospholipid PIP2, which is generated from PIP by a PI 4,5-kinase. PIP2 interacts directly with a number of proteins, regulating both their sub-cellular localization and their activity. Data from yeast and mammalian systems implicates homologous PI 4-kinases and PIP2 in regulation of secretion and membrane vesicle trafficking [51,52]. In Drosophila, fwd gene function is required for the formation of ring canals and cytoplasmic bridges during cytokinesis in male meiosis. PIP2 is also the precursor to the important second messengers PIP3 and diacylglycerol. PIP3 is involved in the insulin-like signaling pathway that affects life span in Drosophila and C. elegans [53], and it will be of interest to determine if fwd overexpression is somehow affecting this pathway.

CTP:phosphocholine cytidylyltransferase (Cct) catalyzes the conversion of phosphocholine to CDP-choline and is the rate-limiting enzyme in the phosphatidylcholine biosynthetic pathway [41]. Phosphatidylcholine is a major membrane phospholipid and is a precursor to second messengers involved in signal transduction at the membrane, including the PIP, PIP2 and PIP3 discussed above. There are two genes encoding Cct in Drosophila (Cct1 and Cct2), and mutations in the Cct1 gene disrupt patterning during oogenesis (T. Gupta and T. Schupbach, personal communication). Overexpression of Cct1 might affect life span by increasing synthesis of phosphatidylcholine and altering membrane structure and/or phospholipid signaling pathways.

It is striking that most of the genes identified in this screen as positive regulators of Drosophila life span are implicated in functions at the membrane, including regulation of transport, phospholipid metabolism, signal transduction and actin cytoskeleton organization. It is tempting to speculate that some of these genes may be acting in the same pathway(s) for life-span regulation. For example, both filamin and fwd are involved in the function of ring canals, which are membrane-associated actin cytoskeletal structures that create intercellular bridges between germline cells during early mitotic divisions [38,39,40]. encore is also required for normal early mitotic division of the germline cells [44]. In C. elegans, germline cells have recently been shown to send signals that regulate life span of the adult organism [54,55,56], and it will be of interest to determine if any of the genes identified here increase life span through effects on the germline. Both fwd and Cct1 are involved in phospholipid metabolism, and phospholipid signaling pathways are implicated in life-span regulation in Drosophila and other organisms [53]. The conserved insulin-like life-span regulatory pathway includes a PI 3-kinase that converts PIP2 into PIP3. The enzymes encoded by fwd and Cct1 are both in the biosynthetic pathway for PIP2 and therefore might affect the insulin-like pathway by altering the availability of this substrate. Further experiments will be required to confirm the role of these genes in life-span regulation, and to determine their interactions with each other and in known or novel life-span regulatory pathways.

A table showing the calculation of 95% confidence intervals for all the datasets is available.