Delivery of circulating lipoproteins to specific neurons in the Drosophila brain regulates systemic insulin signaling

The Insulin signaling pathway couples growth, development and lifespan to nutritional conditions. Here, we demonstrate a function for the Drosophila lipoprotein LTP in conveying information about dietary lipid composition to the brain to regulate Insulin signaling. When yeast lipids are present in the diet, free calcium levels rise in Blood Brain Barrier glial cells. This induces transport of LTP across the Blood Brain Barrier by two LDL receptor-related proteins: LRP1 and Megalin. LTP accumulates on specific neurons that connect to cells that produce Insulin-like peptides, and induces their release into the circulation. This increases systemic Insulin signaling and the rate of larval development on yeast-containing food compared with a plant-based food of similar nutritional content. DOI: http://dx.doi.org/10.7554/eLife.02862.001


Introduction
Nutrient sensing by the central nervous system is emerging as an important regulator of systemic metabolism in both vertebrates and invertebrates (Lam et al., 2005;Levin et al., 2011;Rajan and Perrimon, 2012;Bjordal et al., 2014;Linneweber et al., 2014). Little is known about how nutritiondependent signals pass the blood brain barrier (BBB) to convey this information. Like the vertebrate BBB, the BBB of Drosophila forms a tight barrier to passive transport, and is formed by highly conserved molecular components (Bundgaard and Abbott, 2008;Stork et al., 2008;Abbott et al., 2010). Its simple structure and genetic accessibility make it an ideal model to study how nutritional signals are communicated to the CNS. Insulin and Insulin-like growth factors are conserved systemic signals that regulate growth and metabolism in response to nutrition. Although Drosophila do not have a single pancreas-like organ, they do produce eight distinct Drosophila Insulin/IGF-like peptides (Dilps) that are expressed in different tissues (Riedel et al., 2011;Colombani et al., 2012;Garelli et al., 2012). A set of three Dilps (Dilp2,3,5), released into circulation by Dilp-producing cells (IPCs) in the brain, have particularly important functions in regulating nutrition-dependent growth and sugar metabolism; ablation of IPCs in the CNS causes Diabetes-like phenotypes, slows growth and development, and produces small, long-lived adult flies (Rulifson et al., 2002;Broughton et al., 2005;Partridge et al., 2011). Systemic Insulin/IGF signaling (IIS) increases in response to dietary sugars, proteins and lipids. Sugars act on IPCs directly to promote Dilp release (Haselton and Fridell, 2010), but other nutrients are sensed indirectly through signals from the fat body-an organ analogous to vertebrate liver/adipose tissue (Colombani et al., 2003;Geminard et al., 2009;Rajan and Perrimon, 2012).
The Drosophila fat body produces two major types of lipoprotein particles: Lipophorin (LPP), the major hemolymph lipid carrier, and Lipid Transfer Particle (LTP). LTP transfers lipids from the intestine to LPP. These lipids include fatty acids from food, as well as from endogenous synthesis in the intestine . LTP also unloads LPP lipids to other cells (Van Heusden and Law, 1989; eLife digest How does an animal sense if it is well nourished or not, and then regulate its metabolism appropriately? This process largely relies on the animal's body deciphering signals that that are transmitted between different organs in the form of molecules and hormones. Many animals-ranging from insects to mammals (including humans)-also use their brains to sense and decipher these nutritional signals.
A signaling pathway involving the hormone insulin controls how various different animals grow and develop-and how long they will live-based on these animals' food intake. Insulin is produced in mammals by an organ called the pancreas. But in the fruit fly Drosophila, this hormone is produced by cells within different tissues, including the insect's brain.
The fruit fly is used to study many biological processes because it is easy to work with in a laboratory. Insulin-producing cells make and release insulin-like molecules into the insect's hemolymph (a blood-like fluid) in response to sugar and to other nutrients (which are detected via molecules generated in a fruit fly organ called the fat body). The fat body produces lipophorin, a protein which carries fat molecules in the hemolymph, and which is known to be able to move from the hemolymph to the brain and accumulate within the brain. The fat body also produces lipid transfer protein (or LTP), which transfers fats absorbed or made within the insect's gut onto lipophorin, and can also unload fat molecules to other insect cells. If LTP can also enter the brain, and what it might do there, was unclear.
Brankatschk et al.now discover that LTP can cross the 'blood brain barrier' in fruit fly larvae and can accumulate over time on their insulin-producing cells and the neurons in direct contact with these cells. This accumulation depends on the flies' diet: flies fed a diet made from yeast cells accumulated LTP on these neurons, while those fed only on sugar and proteins did not.
Furthermore Brankatschk et al. found that when they switched flies from a yeast-based to a plant-based diet, the larvae grew more slowly and the flies lived longer. Both of the diets contained abundant calories and nutrients, but contained slightly different kinds of fat molecules. The fly larvae on the plant-based diet also accumulated less LTP on their insulin-pathway neurons, and insulin signaling was reduced.
Branskatschk et al. also found that fat molecules from the yeast-based diet activated the cells of the blood brain barrier, and that this encouraged LTP to be transported the brain. Blocking LTP from crossing the blood brain barrier reduced insulin signaling, slowed the growth of the fly larvae, and extended the lifespan of the flies.
These findings of Brankatschk et al. thus reveal that fat-containing molecules carry information about specific nutrients to the brain. The extent to which these mechanisms operate in other animals-such as mammals-remains to be explored.  Magnified boxed region in (A) (B-B′) Total numbers of LTP-positive neurons/brain lobe (B) or Repo-positive glia/brain (B′) of wt YF larvae of different ages, quantified from 50-60 confocal sections per brain. Numbers indicate larval instar; superscripts indicate age in hours after egg collection. (C-E) Cartoons depict positions of all LTP-positive neurons (black dots) identified in confocal stack of three wt YF brains of different ages, stained for LTP and Elav. (F) Western blot showing equal volumes of hemolymph from control YF larvae, and larvae with FB-specific knock-down of ltp (ltpRNAi FB ). Blots are probed for Apo-LTPI, Apo-LTPII, Apo-LPPII and Cv-D. (G and G′) Total numbers of LTP-positive neurons/brain lobe (G) and Repo-positive glia/brain (G′) quantified from control YF larvae, and larvae where ltp has been knocked-down in neurons (ltpRNAi N ), glia (ltpRNAi G ), or FB (ltpRNAi FB ). Error bars indicate standard deviation. *** = p < 0.001, ** = p < 0.01, n.s. = not significant (Student's t test). Scale bars indicate 50 μm (A and C-E) or 5 μm (A′-A′″). DOI: 10.7554/eLife.02862.003 The following figure supplements are available for figure 1:   To understand the CNS functions of LTP, we sought to identify the neurons where it accumulated. We first investigated whether LTP co-localized with Dilp2-positive neurons in early third instar larval brains. Dilp2 is present not only in IPCs, which lie dorso-lateral to the big commissure, but also on three neurons per hemisphere that do not produce Dilps. These neurons recruit Dilp2 from IPCs using IMPL2, a Dilp2-binding protein (Bader et al., 2013). Although each hemisphere contains 7-8 neurons that express IMPL2 at this stage, only 3 of these detectably recruit Dilp2, and 2 of these 3 also accumulate LTP. These neurons can be unambiguously identified by the presence of both LTP and Dilp2, by their position, by lack of expression of dilp2-GAL4, and by expression of impL2-GAL4 ( Figure 2, Figure 2-figure supplement 1). Henceforth, we refer to these neurons as Dilp2recruiting neurons (DRNs). Starting in the third larval instar, LTP accumulates on two specific DRNs located dorsal to the big commissure ( Although LTP is found on other neurons at earlier stages, its accumulation on DRNs is developmentally regulated. We also sometimes (6/20 sampled CNS) detect LTP on a subset of IPC neurons ( Figure 2B). Thus, upon entering the brain, LTP accumulates on IPCs and specific neurons that recruit Dilp2.
Thus far, we have described LTP localization in larvae reared on a rich medium: 'yeast food' (YF) . Systemic IIS is high in these animals, as reflected by predominantly cytoplasmic localization of FOXO in fat body cells ( Figure 3E,E′) (van der Horst and Burgering, 2007). To ask whether neuronal LTP accumulation was regulated by nutrition, we asked how it responded to starvation. Late second instar larvae transferred from YF to food containing only glucose (GF) arrest growth and development. Systemic IIS is strongly reduced, as indicated by nuclear accumulation of FOXO in fat body cells ( Figure 3A,A′). Although LTP still circulates ( Figure 3G), it accumulates on many fewer neurons in glucose-fed larvae, and is never detected on DRNs ( Figure 3F-F″). Thus, the presence of glucose is insufficient to allow high systemic IIS or accumulation of LTP on most neurons. We next examined the role of lipids and proteins in promoting neuronal LTP accumulation and systemic IIS. We compared larvae fed on YF, with those transferred at late second instar to a lipid-depleted food (LDF), which contains a yeast autolysate that has been chloroform-extracted to remove lipid. LDF contains proteins and sugars, and has a similar caloric content to YF. However, larvae arrest on LDF because of the absence of sterols (Carvalho et al., 2010), and systemic IIS is low ( Figure 3B,B′). Brains of LDF-fed larvae contain more LTP-positive neurons than those of glucose-fed larvae, but fewer than YF-reared larvae. In particular, LTP is never detected on DRNs ( Figure 3F-F″).
To address the sterol-dependence of neuronal LTP accumulation, we supplemented LDF with 10 µM cholesterol (LDSF). Larvae transferred from YF to LDSF grow slowly and give rise to viable adults of reduced size (Carvalho et al., 2010). Consistent with this, nuclear FOXO staining in the fat body indicates that systemic IIS is low ( Figure 3C,C′; Carvalho et al., 2010). Sterol addition further Video 1. Confocal stack from wild type first instar larval brain probed for LTP (green), Dilp2 (red) and Repo (grey). Sections are spaced 1.5 µm apart, scale bars indicate 50 µm. DOI: 10.7554/eLife.02862.007 Video 2. Confocal stack from wild type second instar larval brain probed for LTP (green), Dilp2 (red) and Repo (grey). Sections are spaced 1.5 µm apart, scale bars indicate 50 µm. DOI: 10.7554/eLife.02862.008 increases the number of LTP-positive neurons, but still does not allow LTP accumulation on DRNs ( Figure 3F-F″, and Figure 3-figure supplement 1). Thus, a component present in yeast food, but not the chloroform-extracted yeast autolysate, is required for accumulation of LTP on DRNs, and for high-level systemic IIS.
We wondered whether lipids in general were required for LTP accumulation on DRNs. Interestingly however, experiments with plant food (PF) suggest that bulk lipid is not sufficient. PF contains no yeast and is based entirely on plant materials. It has the same caloric content as YF, and slightly more lipid with a different fatty acid composition (Carvalho et al., 2012, see 'Materials and methods'). Surprisingly, LTP is found only occasionally on DRNs in larvae transferred from YF to PF. Depletion of LTP on DRNs occurs within 16 hr of transfer and is reversible in the same time frame ( Figure 3F-F″ and Figure 4A,A′,C and Figure 3-figure supplement 2). Strikingly, despite the abundant calories derived from carbohydrates, proteins and lipids, feeding with PF specifically slows the larval growth phase without lengthening embryonic or pupal development. PF also dramatically extends adult lifespan compared to YF ( Figure 4E-G). This suggests that systemic IIS is reduced when larvae are fed with PF, compared to YF. Indeed, FOXO is predominantly nuclear in PF-reared larvae ( Figure 3D,D′).
To investigate mechanisms underlying the differences in systemic IIS in larvae fed with YF and PF, we examined the pathway at different levels. We first focused on the neuronal activity of the IPCs. Neuronal activity correlates with higher levels of intracellular free calcium, which can be detected using the GCaMP reporter (Reiff et al., 2005). This GFP reporter increases its fluorescence in response to free calcium. We expressed the GCaMP construct either in IPC neurons (under the control of dilp2-GAL4) or in all neurons (under the control of rab3-GAL4) and compared GFP fluorescence in brains from YF and PF-reared larvae. While the activity of most neurons is not affected by feeding with these different foods ( Figure 5A,B), the activity of the IPCs was dramatically higher on YF than on PF ( Figure 5C,D; Videos 7 and 8). We quantified the number of IPCs exhibiting detectable fluorescence of the GCaMP reporter in 5 brains each of larvae fed with PF or YF. On YF, we detected activity of the GCaMP reporter in 4.8 ± 0.6 neurons per brain lobe, while on PF only 1 ± 1.3 neurons per brain lobe exhibited detectable reporter fluorescence. To ask whether higher neuronal activity of IPCs resulted in elevated release of Dilp2 into the hemolymph, we probed Western blots of hemolymph from YF and PF-reared larvae with antibodies to Dilp2. YF-reared animals have higher levels of circulating Dilp2 than PF-reared animals ( Figure 5E). Thus, IPCs release more Dilp2 when animals are fed with YF. We next examined different molecular readouts of IIS in different tissues of PF and YF-reared larvae. YF increases PI3K activity in salivary glands ( Figure 5F,G), as revealed by membrane localization Video 3. Confocal stack from wild type third instar larval brain probed for LTP (green), Dilp2 (red) and Repo (grey). Sections are spaced 1.5 µm apart, scale bars indicate 50 µm. DOI: 10.7554/eLife.02862.009 Video 4. Confocal stack from wild type third instar larval brain probed for Elav (red), Repo (grey), LTP (green) and DAPI (blue). Sections are spaced 1.5 µm apart, scale bars indicates 50 µm. DOI: 10.7554/eLife.02862.010 of the PH GFP reporter construct. Furthermore, phosphoAKT levels are higher in fat bodies of YF-reared larvae ( Figure 5H). Thus, feeding with YF activates IPCs, causing them to secrete more Dilp2 into circulation, thereby increasing systemic IIS.
To examine whether the differences between plant and yeast lipids were responsible for changes in IPC activity, we prepared chloroform extracts of PF and YF and used them to supplement LDF. Yeast lipids supported faster larval growth compared to plant lipids ( Figure 5I). Furthermore, the GCaMP reporter reveals that IPC neurons are only active when larvae are fed with yeast lipids ( Figure 5J,K; Videos 7 and 8). Thus, IPC neurons respond differently to the types of lipids present in YF and PF.
How does YF promote IPC activity and Dilp release? We wondered whether the YF-dependent localization of LTP to the DRNs might be responsible. To investigate this, we first asked whether IMPL2-expressing neurons actually contact the IPCs. We labeled the projections of IMPL2 expressing neurons by driving a transmembrane Cherry (Cherry) under the control of impL2-Gal4, and stained larval brains for both Dilp2 and IMPL2. A subset of Cherry-labeled projections, also containing IMPL2, extends from the DRNs towards the IPCs ( Figure 6B′,B″,B′″). Dilp2 staining reveals a subset of projections from the IPCs extending towards the DRNs ( Figure 6A-class 2 projections and Figure 6B,B′″). They meet in a region where IMPL2 begins to colocalize with Dilp2 ( Figure 6B′″). Thus, it is likely that DRNs and IPCs communicate with each other directly.
How does the activity of IMPL2-positive neurons influence that of IPCs? To address this question, we inhibited synaptic release by knocking down Rabs 3 and 27-two Rabs that both contribute to synaptic vesicle release (Mahoney et al., 2006;Pavlos et al., 2010). To do this, we used homologous recombination to generate N-terminal YFP fusions of Rab3 and Rab27 at their endogenous chromosomal loci. This renders them sensitive to knock-down by anti-GFP RNAi ( Figure 6-figure supplement 1). Rabs 3 and 27 are expressed almost exclusively in neurons (manuscript in preparation). Thus, even for GAL4 drivers that are active in many tissues, this approach will affect only the neurons that express the GAL4 driver. When anti-GFP RNAi is driven in the IPCs in a background homozygous for both YFPRab3 and YFPRab27, larvae exhibit the slow growth phenotype characteristic of reduced IIS ( Figure 6C). Reducing levels of Rab3/Rab27 in the IMPL2-expressing neurons slows larval growth ( Figure 6C), and reduces phosphorylation of AKT in the fat body ( Figure 5H). This suggests that signals sent by IMPL2-expressing neurons promote systemic IIS.
Since LTP localizes to DRNs only on yeast food, we wondered whether it was required to promote systemic Insulin signaling. We therefore asked whether IIS in YF-reared larvae depended on LTP. RNAi-mediated ltp knock-down in the fat body causes larval arrest in the second or third instar Video 5. Confocal stack from UAS:ltpRNAi/+ third instar larval brain probed for Repo (grey), Dilp2 (red) and LTP (green). Sections are spaced 1.5 µm apart, scale bars indicate 50 µm. DOI: 10.7554/eLife.02862.011 Video 6. Confocal stack from lpp-Gal4>UAS:ltpRNAi third instar larval brain probed for Repo (grey), Dilp2 (red) and LTP (green). Sections are spaced 1.5 µm apart, scale bars indicate 50 µm. DOI: 10.7554/eLife.02862.012 . Fat body cells contain lower levels of phosphoAKT ( Figure 5H) and predominantly nuclear FOXO ( Figure 7K,K′). Thus LTP is required for high-level IIS in the fat body on YF.
Reduced IIS in ltp mutants might reflect a requirement for LTP on DRNs. Alternatively, loss of LTP in other tissues might affect Insulin signaling indirectly. Blocking transport of LTP across the BBB in YF-reared larvae could distinguish these possibilities. We therefore examined requirements for different LDL receptor family proteins in BBB transport of LTP. LTP localizes normally to DRNs and other neurons in larvae double mutant for LDL receptor homologues LpR1 and LpR2 (Figure 7, and Figure 7-figure supplement 1-3; Khaliullina et al., 2009). Thus, LpR1 and LpR2 are not dominant transporters of LTP across the BBB. However, we discovered redundant functions for LRP1 and LRP2 in BBB transport of LTP. Both proteins are predominantly expressed in glial cells ( Figure 7A-F). Double mutants die as third instar larvae (not shown). Although LTP circulates at normal levels (Figure 7figure supplement 1B), the number of LTP-positive neurons is reduced (Figure 7-figure supplement 1A,A′). A similar phenotype is produced by glial-specific double knockdown of LRP1/2, which halves the number of LTP-positive neurons, and completely blocks localization to DRNs. Knockdown of either LRP alone produces intermediate phenotypes, suggesting they function redundantly ( Figure 7G-G′″, Figure 7-figure supplement 3). Thus, both LRPs transport LTP across the BBB to DRNs. Transport of LTP to other neurons has a less strict requirement for LRP1/2.
If the absence of LTP on DRNs reduces systemic IIS, then losing LRP1/2 in BBB glia should reproduce this effect. Indeed, glial-specific LRP1/2 knock-down slows larval growth and delays pupariation, and emerging adults live longer-all phenotypes suggesting reduced IIS ( Figure 7H-I,M). Consistent with this, levels of circulating Dilp2 are lower in these animals ( Figure 5E), AKT is less phosphorylated   Figure 5H), and FOXO is predominantly nuclear in larval fat body cells ( Figure 7L,L′). These data support the idea that blocking transport of LTP to DRNs reduces the release of Dilp2 by IPCs, thereby lowering systemic IIS.
What mechanisms promote the BBB transport of LTP to DRNs on yeast food? We first wondered whether levels of LRP1,2 might change. However, we observed no obvious differences in LRP1,2 staining in the brain on YF vs PF (data not shown). However we were surprised to observe dramatically higher levels of free intracellular calcium in BBB glia when larvae are fed YF compared to PF. Driving the free intracellular calcium sensor GCaMP using repo-GAL4 (which is active in all glia) reveals a marked reduction in GFP fluorescence in BBB glia relative to other glia. We confirmed this result by specifically driving the reporter in BBB glia using moody-GAL4 ( Figure 8A-D).
To investigate whether Ca++ signaling in BBB glia was sufficient to promote transport of LTP to DRNs, we asked whether ectopically inducing Ca++ influx would allow LTP to accumulate on DRNs even when larvae were fed with PF. To do this, we drove the expression of TRPA1 with moody-GAL4 in PF-reared larvae and stained their brains for LTP and Dilp2. Strikingly, while LTP is undetectable on DRNs of wild type larvae reared on PF, it accumulates to high levels on these neurons when Ca++ influx is induced by TRPA1 expression in BBB cells ( Figure 8E,E′). Accumulation is specific to DRNs and neuronal LTP localization in these larvae precisely mimics its localization on YF. Furthermore, Western blotting shows that phosphorylation of AKT in the fat body increases compared to wild type larvae fed with PF ( Figure 8G). Taken together, these data show that YF increases free cytoplasmic Ca++ in BBB glia. This is sufficient for the specific transport of LTP to DRNs and elevated systemic IIS. Interestingly, the accumulation of LTP on other neurons does not appear to require elevated Ca++ in BBB cells.

Discussion
In summary, this work demonstrates a key requirement for lipoproteins in conveying nutritional information across the BBB to specific neurons in the brain. As particles that carry both endogenously synthesized and diet-derived lipids, lipoproteins are well-positioned to perform this function. Our data suggest that transport of LTP across the BBB to DRNs influences communication between DRNs and the Dilp-producing IPCs, increasing the release of Dilp2 into circulation. Since the IPCs also deliver Dilp2 to the DRNs, this indicates that these two neuronal populations may communicate bidirectionally. How might LTP affect the function of DRNs? One possibility is that it acts to deliver a signaling lipid to the DRNs. It could do so either directly, or indirectly by promoting lipid transfer from LPP, which is present throughout the brain (Brankatschk and Eaton, 2010). LTP enrichment on specific neurons may increase lipid transfer to these cells.
This work highlights a key function for BBB cells in transmitting nutritional information to neurons within the brain. Feeding with yeast food increases free calcium in BBB glia, which then increases transport of LTP to DRNs. How might BBB cells detect the difference between yeast and plant food? Our data suggest differences in the lipid composition of yeast and plant-derived foods are responsible. Our previous work has shown that the lipids in these foods differ in their fatty acid composition. Yeast food has shorter and more saturated fatty acids than plant food . How could these nutritional lipids affect the activity of BBB glia? Interestingly, differences in food fatty acid composition are directly reflected in the fatty acids present in membrane lipids of all larval tissues including the brain . Thus, it is possible that the bulk membrane properties of BBB glia are different on these two diets. Membrane lipid composition is known to affect a variety of signaling events (Simons and Toomre, 2000;Lingwood and Simons, 2010). Alternatively, yeast food may influence the specific fatty acids present in signaling lipids that activate BBB glia.  We demonstrate an unexpected functional specialization of the BBB glial network, which permits specific and regulated LTP transport to particular neurons. How this specificity arises is an important question for the future. We note that a subset of glial cells within the brain also accumulates LTP derived from the fat body. Could these represent specific transport routes from the BBB? An alternative possibility is that transport depends on neuronal activity. Mammalian LRP1 promotes localized transfer of IGF in response to neuronal activity (Nishijima et al., 2010). Could LTP delivery by LRP1 and LRP2 in the Drosophila brain depend on similar mechanisms? The remarkable specificity of LTP trafficking in the Drosophila CNS provides a novel framework for understanding information flow between the circulation and the brain.
To what extent might this be relevant to vertebrate systems? While it is clear that the vertebrate brain (unlike that of Drosophila) does not depend on lipoproteins to supply it with bulk sterols  , 2), repo-GAL4>lrp1,2RNAi (lane 3), lpp-GAL4>ltpRNAi (lane4) and YFP-rab27;YFP-rab3;impl2-GAL4>gfpRNAi (lane 5). Larvae were reared on yeast food (YF) or plant food (PF) as indicated. (I) shows percentages of larvae that have pupariated at different days after egg laying (dAEL). Larvae were reared on yeast food (red, n = 49), on plant food (black, n = 34), or on lipid-depleted food supplemented with either yeast lipids (light red, n = 23) or plant lipids (grey, n = 24). (J and K) show GFP fluorescence from collapsed confocal stacks of whole brains expressing GCaMP under the control of dilp2-GAL4 from larvae fed on lipid-depleted food supplemented with yeast lipids (J) or lipid-depleted food supplemented with plant lipids (K). See also Videos 7 and 8. DOI: 10.7554/eLife.02862.020 (Orth and Bellosta, 2012), this does not rule out possible functions for these particles in nutrient sensing. The vertebrate cerebrospinal fluid is rich in many types of HDL particles, including those containing ApoA-1, which is not expressed in the brain-this suggests that at least some lipoprotein particles in the brain may derive from the circulation (Wang and Eckel, 2014). Consistent with this idea, ApoA-I can target albumin-containing nanoparticles across the BBB in rodents (Zensi et al., 2010). Recent work suggests that lipoproteins may be the source of specific Long Chain Fatty Acids that signal to the hypothalamus to regulate glucose homeostasis, since neuronal lipoprotein lipase is required for this process (Wang et al., 2011;Wang and Eckel, 2014). Thus, it would be interesting to investigate whether circulating mammalian lipoproteins might reach a subset of neurons in the hypothalamus.
It has been known for some time that increasing the amount of yeast in the diet of lab grown Drosophila melanogaster increases the rate of development and adult fertility, but reduces lifespan (Sang, 1949;Leroi et al., 1994;Mair et al., 2005). Here, we show that flies have evolved specific mechanisms to increase systemic IIS in response to yeast, independently of the number of calories in the diet or its proportions of sugars proteins and fats. What pressures could have driven the evolution of such mechanisms? In the wild, Drosophila melanogaster feed on rotting plant material and their diets comprise both fungal and plant components. Drosophila disperse yeasts and transfer them to breeding sites during oviposition improving the nutritional resources available to developing larvae (Markow and O'grady, 2008). Yeast that are able to induce more rapid development of the agents that disperse them may propagate more efficiently. On the other hand, it has been noted that Drosophila species that feed on ephemeral nutrient sources like yeasts or flowers have more rapid rates of development (Markow and O'grady, 2008) than other species. It may be that, even within a single species, the ability to adjust developmental rate to the presence of a short-lived resource is advantageous. Humans subsist on diets of both plant and animal materials that during most of evolution have differed in their availability. It would be interesting to investigate whether Insulin/IGF signaling in humans might respond to qualitative differences in the lipid composition of these nutritional components.

UAS:membrane tagged mcherry
An 840 bp fragment containing the predicted signal sequence of lipophorin was fused to a HA-epitope triplet and cloned 5′ to an mcherry cDNA. This was fused at its 3′ end to a CD8 transmembrane domain sequence that was flanked at the 5′ end with the MYC-epitope and the 3′ end with a V5 epitope. The construct was cloned into pCM43 vector containing a 5xUAS-hsp70 promoter element followed by a multiple cloning site, a SV40 element and mini-white gene.

Lipid extraction
40 g PF or YF were homogenized in 100 ml Chloroform. The resulting homogenates were centrifuged and the lower organic phase was removed and evaporated. The residues were re-suspended in 2 ml Chloroform and 30 µl of these solutions were applied to lipid-depleted food.