Vitellogenin family gene expression does not increase Drosophila lifespan or fecundity

One of the most striking patterns in comparative biology is the negative correlation between lifespan and fecundity observed in comparisons among species. This pattern is consistent with the idea that organisms need to allocate a fixed energy budget among competing demands of growth, development, reproduction and somatic maintenance. However, exceptions to this pattern have been observed in many social insects, including ants, bees, and termites. In honey bees ( Apis mellifera), Vitellogenin ( Vg), a yolk protein precursor, has been implicated in mediating the long lifespan and high fecundity of queen bees. To determine if Vg-like proteins can regulate lifespan in insects generally, we examined the effects of expression of Apis Vg and Drosophila CG31150 (a Vg-like gene recently identified as cv-d) on Drosophila melanogaster lifespan and fecundity using the RU486-inducible GeneSwitch system. For all genotypes tested, overexpression of Vg and CG31150 decreased Drosophila lifespan and did not affect total or age-specific fecundity. We also detected an apparent effect of the GeneSwitch system itself, wherein RU486 exposure (or the GAL4 expression it induces) led to a significant increase in longevity and decrease in fecundity in our fly strains. This result is consistent with the pattern reported in a recent meta-analysis of Drosophila aging studies, where transgenic constructs of the UAS/GAL4 expression system that should have no effect (e.g. an uninduced GeneSwitch) significantly extended lifespan in some genetic backgrounds. Our results suggest that Vg-family genes are not major regulators of Drosophila life history traits, and highlight the importance of using appropriate controls in aging studies.


Introduction
Aging (senescence) is an almost universal process in multicellular organisms, in which organismal function and performance decline with age 1,2 . Decreasing fertility and increasing mortality are general hallmarks of aging that are typically accompanied by a declining activity level, altered metabolic rate, and a higher susceptibility to predation, parasites and disease [3][4][5] . Despite its ubiquity, patterns and rates of aging vary enormously among, and within species. A large body of theory and experiment explores the evolutionary causes of this diversity; however, the underlying molecular mechanisms are still poorly understood 6,7 .
Evolutionary life history (LH) theory generally assumes that allocation of energy among the competing demands of growth, development, reproduction and somatic maintenance lead to functional trade-offs among these processes 8,9 . Consistent with this idea of resource allocation trade-offs, lifespan and fecundity are generally negatively correlated in comparisons among species 8,10 , A prominent exception to this pattern occurs in many social insects. In many ants, termites, and bees 11 , reproductive females are both long-lived and highly fecund relative to other species. For example, queen black garden ants (Lasius niger) can live for at least 28 years, while laying hundreds of eggs per day 12 , and queen honey bees (Apis mellifera) have a maximum lifespan of 3-5 years while laying thousands of eggs per day 13 . In contrast, most non-social insects have adult longevity of less than one year, and have lower fecundity than social insect queens 14 .
Investigations in honey bees suggest that the Vitellogenin gene (Vg) produces a yolk protein precursor that is synthesized in the abdominal fat body and acts as an antioxidant and promotes longevity in queen bees 15 . Differential expression of Vg has also been associated with the differences in lifespan between different kinds of worker bees: higher expression is seen in "winter bees" which have a lifespan of 10 months to 1 year, and lower expression in "summer bees" with a lifespan of 30-50 days [16][17][18] . RNAi knockdown of Vg expression in workers resulted in lower oxidative stress resistance 19 and shorter lifespan 20 . In contrast, RNAi knockdown of Vitellogeninencoding genes in the nematode Caenorhabditis elegans increased survival 21 and the expression of genes with similar function in the fruit fly Drosophila melanogaster (yolk protein genes) is negatively correlated with lifespan 22 . The strong support for a role for Vg expression in regulating lifespan and fecundity in honey bees, combined with conflicting results from other species, led us to ask if honey bee Vg has different functional properties than its homologs in other invertebrates. Specifically, we asked (1) if transgenic expression of honey bee Vg in fruit flies can regulate lifespan or fecundity, and (2) if over-expression of a related gene that is endogenous in flies has effects similar that of expression of honey bee Vg.
An extensive genetic toolbox allows time and tissue-specific manipulation of gene expression in D. melanogaster in ways that are not available in other organisms, and these techniques have been used to characterize the effects of many genes on fly lifespan and fecundity 23 . For example, manipulations of genes involved in the insulin signaling and the target-of-rapamycin pathways have been causally linked to lifespan regulation in flies (reviewed in 23,24 ). To our knowledge, however, the effects of Vg-family genes on agingrelated traits have not been investigated in flies. Unfortunately, D. melanogaster lacks a direct homolog of honeybee Vg. Instead, gene CG31150, recently annotated as crossveinless d (cv-d), encodes a Vg-like protein that is expressed mainly in the fat body 25 . The coding sequence for this gene is the most similar among all Drosophila genes to Vg-encoding genes in honey bee (37% similarity), C. elegans (40% similarity), chicken (38% similarity) and zebra fish (38% similar) 26 . CG31150 resembles other Vg family genes in having an N-terminal Vitellogenin N domain, DUF1943 Pfam motifs, and a partial von Willebrand Factor D (VWD) domain near the C terminus 25 . The biological functions of CG31150 are largely unknown, but it was recently implicated in lipid transport and bone morphogenic protein (BMP) signaling.
We used the bipartite GAL4/UAS system to manipulate the expression of both honey bee Vg and the endogenous CG31150 gene in flies, and to assess the effects of these manipulations on fly lifespan and fecundity. In this system, expression of a transgene is under the control of a promoter region derived from yeast, the Upstream Activation Sequence (UAS). This promoter region activates transgene expression only when it is bound to the GAL4 protein. Tissuespecific promoters enable spatial control of GAL4 protein production 27 . Temporal control of transgene expression can be achieved in several ways 28 , but one of the most convenient methods is using constructs in which the GAL4 DNA binding domain is fused to a progesterone receptor transcriptional activation domain, and is therefore only activated by systemic application of a progesteronereceptor ligand 29-31 . We used RU486 (mifepristone)-induced GAL4 drivers (also called the GeneSwitch system) to manipulate our target genes in adult fat body 32 . A major advantage of this system is that control flies (not exposed to RU486) have exactly the same genotype as transgene-expressing flies, so genetic background differences cannot contribute to differences in phenotype 33 . We manipulated the expression of both the honey bee Vg and D. melanogaster CG31150 in order to determine whether either or both can regulate lifespan or fecundity in flies. We included two constructs of each gene to account for potential position effects. We also included a series of controls to take into account possible phenotypic effects of RU486 or of the expression of the GAL4 protein 34,35 .

Materials and methods
Drosophila stocks and maintenance The GeneSwitch driver strain S106 (w 1118 ; P{Switch1}106) and UAS-GFP strain (P{UAS-GFP.VALIUM10}attP2) were obtained from Bloomington Drosophila Stock Center. To control for position effects, we used two different transgenic strains for both Vg and CG31150, with the transgene inserted onto different chromosomes: CG31150-2 (w 1118 , UAS-CG31150-2/FM6), CG31150-4 (w 1118 , UAS-CG31150-4/TM3), Vg-1 (w 1118 , UAS-Apis VGD13-1/FM6), Vg-2 (w 1118 , UAS-Apis VGD13-2/TM3). These constructs were created by Eric Spana at the Duke University Model Systems Genomics Core Facility as follows. For the Vg constructs, four cDNAs were identified as Apis Vg from database searches, and obtained from RIKEN. All four were sequenced on both strands, and one, BH10008D13 was chosen for subsequent cloning as it matched the published sequence best. Site directed mutagenesis was used to remove an ATG from the parental cloning vector (CATTATACGAAGTTAGGGATCAG-GCCAAATCGGCCG where the underlined G marks the nucleotide that was changed from a T) so that a NotI/KpnI double digest would excise the Vg cDNA from the vector and not contain an incorrect start codon. The NotI/KpnI fragment was then ligated into a NotI/KpnI linearized pUAST. The subsequent clone was verified by sequencing on both strands, and pUAST-Apis_Vg was then transformed into Drosophila as described below. For the CG31150 constructs, cDNA GH05619 was available in the "Gold Clone" collection and was obtained from Robin Wharton (Duke University Medical Center). The cDNA was excised from GH05619 using EcoRV & XhoI and ligated into pUAST that had been digested by EcoRI, filled in with Klenow to make a blunt end, and then digested with XhoI to make compatible ends with the insert. The pUAST-CG31150 plasmid insert was sequenced on both strands. pUAST-Apis Vg and pUAST-CG31150 plasmid inserts were transformed into w 1118 by Model System Genomics of Duke University by standard techniques. All transgenic flies were mapped and balanced. Transgene sequences are shown in Supplementary Table S1.
Overexpression of the target gene in each of the four genotypes was achieved by crossing each transgenic stock with the GeneSwitch strain S106. Strain S106 drives expression specifically in the fat body and also in the digestive system 32 . This driver has been widely used in aging studies; for example, over-expression of dfoxo and dilp6 using this driver have been shown to increase lifespan 36,37 .
To produce female flies that expressed the target transgene, virgin females from the S106 stock were crossed to males from each of the UAS-transgenic strains. Half of the female offspring from this cross were reared with RU486-supplemented media throughout adult life (see below), thus inducing transgene expression, but only during the adult stage. The remaining females from each cross were reared on media supplemented only with RU486 vehicle (ethanol), and should not have expressed the transgene. Therefore each group of flies that expressed one of the target transgenes had a genotypematched control that differed only with respect to whether or not it was fed RU486 or vehicle.
To determine if RU486 exposure (or the GAL4 expression it induced) caused any change in lifespan or fecundity in these flies, we used an additional set of controls. We crossed the S106 driver strain to UAS-GFP flies, and exposed half of the female offspring to RU486 and half only to the ethanol vehicle. GFP is widely used as a reporter in D. melanogaster, and is believed to be non-toxic and not to influence endogenous gene expression at any stage during fly development 38 . Flies were kept on a 12:12 light: dark cycle at 25°C for their entire lifespan.
Similar drug concentration and delivery methods have been widely used in aging studies 32,35,36 , and our pilot data indicated that this concentration produced robust expression levels in the range that we desired (5-15 fold increase over non-induced controls). For the control food, an equal amount of 100% ethanol was diluted with water and added to the surface of standard fly food. Each type of food was made fresh twice per week. The vials were allowed to air dry for 48 hours prior to each transfer.
Lifespan assay For each genotype tested, female offspring were collected within 24 hours of emergence and split equally into experimental and control groups. For each genotype, 96 females were reared on media supplemented with RU486, and 96 were reared on control media. Flies were housed in standard rearing vials (VWR) at a density of 6 females per vial. All rearing vials were placed randomly in 10 × 10 vial racks to control for possible position effects. The flies were transferred to new media twice per week, and were counted nearly every day. In addition, 3 replicates of 10 females each from each treatment group per genotype were collected separately to be assayed for target gene expression. For these replicates, flies were sampled at 7 days post eclosion; GFP/S106 flies were examined for GFP expression using Zeiss LSM 5 PASCA fluorescent microscopy (Zeiss, Germany), and Vg/S106 and CG31150/S106 flies were flash frozen on dry ice for quantitative PCR assays.

Fecundity assay
Females described above were placed in vials with males from a wild-type laboratory strain (described Remolina et al. 39 ) when the females were 1 day old. 6 virgin females and 4 males were placed into each vial. The males were removed after 3 days, and females were transferred to new vials. Flies were transferred twice per week. After each transfer, the old vials were kept at 25°C, and all offspring eclosing within 14 days after the adults were removed and counted.

Quantitative PCR
For each genotype, we pooled 10 flies from each replicate and homogenized whole flies using cordless motors (VWR) and RNasefree pellet pestles (Kimbel Chase). Total RNA was extracted from whole bodies using a PicoPure RNA isolation kit (Arcturus), according to the manufacturer's recommended protocol. RNA purity and quantity were measured using a Nanodrop spectrophotometer (Thermo Scientific). RNA was DNase-treated with a DNAfree RNA kit (Zymo Research) and reverse transcribed using the Superscript III (Invitrogen). We conducted qRT-PCR using SYBR green master mix (Applied Biosystems), and ribosomal gene Rp49 as an endogenous control. The primer sequences were: For transgenic GFP flies (UAS-GFP/S106), expression of GFP was assessed by fluorescent microscopy.

Statistical analysis
For each transgenic line, we compared Kaplan-Meier (productlimit) survival estimates between RU486-and vehicle-fed flies by To determine if transgene expression had any effect on lifespan in the Vg and CG31150 lines, after accounting for the lifespanextending effects of RU486 exposure, we compared the effects of RU486 across all genotypes. Proportional hazards models indicated that there was significant heterogeneity among genotypes in their response to RU486 treatment overall (χ 2 = 11.2, df=4, p=0.02), with RU486 exposure associated with significantly higher mortality in the Vg and CG3115 transgenic flies than in the GFP transgenic flies in each pairwise comparison (χ 2 = 6.5, df=1, p=0.01, hazard ratio = 1.77 for CG31150-2 vs GFP; χ 2 = 7.8, df=1, p=0.005, hazard ratio = 1.85 for CG31150-4 vs GFP; χ 2 = 8.23, df=1, p<0.005, hazard ratio = 1.87 for Vg-1 vs GFP; χ 2 = 4.0, df=1, p<0.05, hazard ratio = 1.56 for Vg-2 vs GFP). These results suggest that Vg and CG31550 transgene over-expression decrease fly lifespan because (1) GFP over-expression is unlikely to increase lifespan, and in one study was shown to decrease lifespan 41 , and (2) if transgene expression had no effect in Vg and CG31550 lines, we should have observed a similar increase in lifespan in the RU486fed flies driven solely by lifespan-extending effects of RU486 and/ or GAL4 expression. using log-rank and generalized Wilcoxon chi-square tests of the homogeneity of survival functions between groups. These tests were conducted using JMP Pro 10 statistical software (SAS Institute Inc., Cary, North Carolina, United States). Log-rank tests are more sensitive to survival differences that occur late in life while Wilcoxon tests are more sensitive to differences earlier in life 40 . Results of both tests were consistent in all the analyses reported here, so we report only the log-rank test results. To determine if activation of transgene expression by RU486 had different effects in different genotypes, we use used Cox proportional hazard models with predictor variables that included genotype, RU486 treatment, and genotype-by-treatment interaction. A significant interaction in this model would indicate that genotypes responded differently to RU486 treatment. These tests were conducted using the phreg procedure of SAS v. 9.3. Flies that died accidentally or escaped during transfers were recorded as censored on the day they died or escaped. For fecundity assays, the mean age-specific fecundity for each vial was calculated by dividing the number of offspring by the number of female parents alive in the vial. We then compared fecundity among genotypes and treatment groups using repeatedmeasures ANOVA of mean fecundity, with age of the females as the repeated measure. These analyses were conducted in JMP Pro 10.

Results
Quantitative PCR confirmed that Vg and CG31150 were overexpressed by the GeneSwitch system Flies carrying S106 GAL4 driver and each of the target genes, Vg-1, Vg-2, CG31150-2, CG31150-4, that were fed on RU486 had a >5 fold increase in target mRNA expression compared with their genetically matched controls (Table 1). In addition, all 30 GFP/ S106 flies that were fed on RU486 showed fluorescent green under the fluorescent microscope while none of their genetically matched controls did (data not shown).
Increasing Vg and CG31150 expression does not extend lifespan in Drosophila To test whether Vg or CG31150 overexpression affected lifespan in Drosophila, we examined survival of flies with overexpression of Apis Vg and Drosophila CG31150. Contradictory to the hypothesis that an increase in the target gene expression would increase lifespan, we observed no significant differences between control and overexpression flies for any of the four experimental genotypes tested (Table 2, Figure 1a-1d). A proportional-hazards model indicated that there were no significant differences among the four Vg and CG31150 genotypes in their response to transgene activation (χ 2 = 0.97, df=3, p=0.81 for genotype-by-RU486 treatment interaction). As an additional control, we assessed whether the RU486 treatment itself was associated with changes in lifespan by comparing RU486-fed and vehicle-fed UAS-GFP/S106 flies. Surprisingly, the RU486-fed flies had a significantly longer lifespan than the controls in this comparison, indicating that RU486 itself (or the GAL4 expression induced by it) had a positive effect on lifespan in this assay (Table 2, Figure 1e). Thus, the lack of lifespanextending effects of Vg and CG31150 transgene expression were not due to confounding effects of the method of induction because this method appeared to increase, rather than decrease lifespan in this experiment.  age-by-treatment effects were observed in every genotype, Table 3, Figure 2).
In two-way ANOVA models that included all genotypes, there was no significant genotype-by-treatment interaction (F=0.111, df=4,149, p=0.978) indicating that all transgenic lines, including GFP/S106, responded similarly to RU486 exposure. It therefore appears likely that the decline in fecundity in transgene-expressing flies is caused by RU486 itself or by the expression of GAL4, rather than target-gene overexpression.
Increasing Vg and CG31150 expression does not increase fecundity in Drosophila To investigate whether overexpression of Vg or CG31150 increased fecundity in Drosophila, we compared age-specific and lifetime fecundity between flies overexpressing these genes and genetically matched controls. We observed a significant overall reduction in fecundity of the RU486-fed flies in all UAS/S106 genotypes, including the control genotype GFP/S106 (Table 3, Figure 2). The difference between treatments was greatest during mid-life at ages of peak egg-laying, and less at early and late ages (significant   been characterized. These genes include the major hemolymph lipid carrier (apolpp) and regulators of interorgan lipid transport (mtp and apoLTP). Products of these genes might therefore play major roles in regulating energy allocation and life history patterns, and future studies of their effects on aging-related traits are warranted.
In addition to testing effects of Vg and CG31150, our results highlighted the importance of using multiple controls in aging studies. If we had not used the GFP/S106 controls, we would not have identified the effects of RU486 exposure in increasing female fly lifespan and decreasing fecundity, and we would have concluded that expression of our target genes had no effect on lifespan and that they decreased fecundity. Aging-related phenotypes are sensitive to many environmental variables, genetic background, and interaction between genotype and environment. A recent meta-analysis of Drosophila aging experiments reported that transgenic constructs of the GAL4/UAS expression system that should have had no phenotypic effects (GAL4 alone, UAS alone, or noninduced GeneSwitch constructs) significantly extended lifespan in the w 1118 genetic background 46 , consistent with our results. Other studies have reported significant reductions in lifespan caused by RU486 exposure in some genotypes 35 , suggesting that strain-specific effects of RU486 exposure are common. Experiments using inducible transgenic constructs therefore require multiple sets of controls to reliably assay genetic regulation of aging phenotypes.

Competing interests
No competing interests were disclosed.

Grant information
This research was supported by National Science Foundation grant DEB 0848337 and National Institutes of Health grant AG022824 to KAH.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Dataset 1. Drosophila lifespan and fecundity data set.
Lifespan raw data: For each genotype, the death date for each individual was recorded. When a fly died of unnatural causes, the date was recorded under the columns titled "censored". "Emerged" refers to the date that the individual enclosed. "Cell #" indicates the position of the vial in the 10 × 10 vial racks.
Fecundity raw data: For each genotype, the number of offspring that enclosed before the day of transfer was recorded, along with the number of mothers that were alive at the time. "Cell #" indicates the position of the vial in the 10 × 10 vial racks.
qPCR raw data: Target gene mRNA expression level is derived by relative quantification (RQ) after normalizing to Rp49. CT, CT mean and CT standard deviation were calculated.
Click here to access the data.

Discussion
In this experiment, Vg and CG31150 overexpressing flies and their genetically matched controls did not differ significantly in lifespan. However, our negative control GFP/S106 showed significant increase in lifespan when fed RU486. The extended lifespan in the GFP/S106 flies is likely due to the supplement of RU486 in the diet or expression of the GAL4 protein, rather than GFP overexpression. If this lifespan-extending effect of RU486 and/or GAL4 activation also occurred in Vg-and CG31150-overexpresing lines we investigated, then over-expression of these genes likely had a corresponding negative effect on lifespan.
Vg and CG31150 overexpressing flies showed reduced fecundity comparing with their genetically matched controls. GFP/S106 flies fed with RU486 showed a quantitatively similar reduction in fecundity. These results suggest that over-expression of Vg and CG31150 do not affect fecundity in Drosophila. Instead the effect is attributable to RU486, the expression of GAL4, or both. Tarone et al. (2012) 42 reported that expression levels of Drosophila yp family genes were negatively correlated with longevity among genotypes derived from a natural D. melanogaster population. These genes, while unrelated to Vitellogenins, are thought to carry out some functions normally associated with Vitellogenins (e.g., they are believed to comprise the major storage protein in fly embryos) 26, 43 . Similarly, our results suggest that Vg-family gene expression affects fly lifespan. This result contrasts with effects of Vg expression in hymenoptera 15,16,19,44,45 . The specific pathway through which these genes affect fly lifespan is not known, but our results suggest that it was unlikely due to trade-offs with fecundity, since no increase in fecundity was detected in transgene-overexpressing flies.
In addition to CG31150, three other Vg-family genes have been described in D. melanogaster 26 , but their effects on aging have not

Supplementary material
Supplementary Table S1. cDNA sequences of Apis Vg and Drosophila CG31150 in the transgenic constructs.
The data are clearly presented and support the conclusions above, however, the generality of the conclusions must be tempered somewhat given the limited scope of the study.
In previous studies life span extension was observed when transgenes were expressed Drosophila with certain drivers, but not others, including some cases where the drivers tested had similar tissue specificities. Those results indicate that the precise tissue-specificity and/or level of transgene expression can be important. In the present study the transgenes were expressed Vg using a Gene-Switch driver strain that yields expression in only a subset of the fat-body tissues, so it remains possible that over-expression of the genes using a driver with more extensive Vg expression pattern in the fat body and/or other tissues might create a life span increase.
The conclusions regarding the possible cause of the life span increase in the control cross are based on the assumption that over-expressed GFP has little or no phenotypic effect. However the one study cited reports a significant (negative) effect of GFP expression on aging Drosophila phenotypes (using different drivers). If GFP can have negative effects in some tissues it seems possible it might have positive effects when expressed in other tissues (for example, the "negative" effect of killing the IPCs or the germ cells can increase fly life span). For these reasons it is not clear whether the life span increase observed in the present study results from the drug, the activated Gene-Switch, or the GFP. Finally, it should be noted that this result was not replicated. These considerations limit the support for a possible negative life span effect of the Vg transgenes as suggested in the Abstract and Discussion.
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed. Competing Interests: speculative, it still might be worth going into a bit more detail about what is known and thought about the molecular mechanisms underpinning how effects aging in honey bees and more Vg clearly define the predictions. a priori Summary: The title is appropriate Abstract is adequate Everything is well explained and presented The conclusions are sound and sufficiently conservative, I do make some suggestions above.
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed. Competing Interests: