|Home | About | Journals | Submit | Contact Us | Français|
Triacylglycerol (TAG) homeostasis is an integral part of normal physiology and essential for proper energy metabolism. Here we show that the single Drosophila ortholog of the PXR and CAR nuclear receptors, DHR96, plays an essential role in TAG homeostasis. DHR96 mutants are sensitive to starvation, have reduced levels of TAG in the fat body and midgut, and are resistant to diet-induced obesity, while DHR96 overexpression leads to starvation resistance and increased TAG levels. We show that DHR96 function is required in the midgut for the breakdown of dietary fat, and that it exerts this effect through the CG5932 gastric lipase, which is essential for TAG homeostasis. This study provides insights into the regulation of dietary fat metabolism in Drosophila and demonstrates that the regulation of lipid metabolism is an ancestral function of the PXR/CAR/DHR96 nuclear receptor subfamily.
Fat metabolism is central to the process of energy homeostasis. When nutrients are abundant, dietary fat in the form of triacylglycerol (TAG) is broken down by gastric TAG lipases to release fatty acids. These fatty acids are absorbed by the intestine and used to resynthesize TAG, which is trafficked to the adipose tissue for storage. These TAG reserves can be accessed upon nutrient deprivation through the action of specific lipid-droplet associated lipases that release the fatty acids for energy production through mitochondrial fatty acid β-oxidation. Defects in these processes can lead to dramatic changes in TAG levels and a range of physiological disorders, including obesity, diabetes, and cardiovascular disease. The alarming rise in the prevalence of these disorders in human populations has focused attention on understanding the molecular mechanisms that coordinate dietary nutrient uptake with TAG homeostasis. As a result, many regulators of TAG metabolism have been identified, including SREBP, PPAR, and adiponectin. In spite of these advances, however, the molecular mechanisms that coordinate dietary fat uptake, synthesis, storage, and utilization, remain poorly understood.
Nuclear receptors (NRs) are ligand-regulated transcription factors that play a central role in metabolic control. They are defined by a conserved zinc-finger DNA binding domain (DBD) and a C-terminal ligand binding domain (LBD) that can impart multiple functions, including hormone binding, receptor dimerization, and transactivation. Many NRs are regulated by small lipophilic compounds that include dietary signals and metabolic intermediates, and exert their effects by directing global changes in gene expression that act to maintain metabolic homeostasis. This is exemplified by members of the mammalian PPAR, LXR, and FXR subfamilies, which play critical roles in adipogenesis, lipid metabolism, and cholesterol and bile acid homeostasis, respectively (Chawla et al., 2001; Sonoda et al., 2008).
In this study we use the fruit fly, Drosophila melanogaster, as a simple model system to characterize the NR subfamily represented by the Pregnane X Receptor (PXR, NR1I2), Constitutive Androstane Receptor (CAR, NR1I3), and Vitamin D Receptor (VDR, NR1I1) in mammals. Previous studies have defined central roles for these receptors in sensing xenobiotic compounds and directly regulating genes involved in detoxification (Timsit and Negishi, 2007; Willson and Kliewer, 2002). Initial studies showed that the single ancestral Drosophila ortholog of this NR subclass, DHR96, has similar functions (King-Jones et al., 2006). A DHR96 null mutant displays increased sensitivity to the sedative effects of phenobarbital and the pesticide DDT as well as defects in the expression of phenobarbital-regulated genes. These studies, however, revealed other potential roles for the receptor – in particular, unexpected effects on the expression of genes that are predicted to regulate lipid and carbohydrate metabolism (King-Jones et al., 2006). This observation is in line with recent studies that have implicated roles for the mammalian PXR and CAR NRs in metabolic control (Moreau et al., 2008). The molecular mechanism by which they exert this effect, however, remains undefined.
Here we show that DHR96 null mutants are sensitive to starvation and have reduced levels of TAG, while DHR96 overexpression leads to starvation resistance and elevated TAG levels. A series of studies using metabolic assays, diets, and the drug Orlistat, revealed that DHR96 mutants are defective in their ability to break down dietary lipid. This model was supported by microarray studies, which showed that many genes expressed in the midgut are misregulated in DHR96 mutants, including highly reduced expression of the gastric lipase gene CG5932. We show that CG5932 is required for proper whole animal TAG levels, and that selective expression of CG5932 in the midgut of DHR96 mutants is sufficient to rescue their lean phenotype. Taken together, these data support a role for the PXR/CAR/DHR96 NR subclass in lipid metabolism and define DHR96 as a key regulator of dietary TAG breakdown in the Drosophila midgut.
Previous studies have shown that DHR961 null mutants, grown under normal conditions, are viable and fertile with no apparent developmental defects (King-Jones et al., 2006). To assess potential metabolic roles for DHR96, we examined the ability of mutants to survive a period of complete starvation. All studies were performed using Canton-S wild-type flies (CanS) and DHR961 null mutants that had been crossed to CanS through nine generations of free recombination. Mature male flies were collected as 5–7 day old adults, transferred to starvation media, and the number of living animals was scored every four hours. DHR961 mutants die at a faster rate than the wild-type controls (Figure 1A). We also tested DHR962X animals, which carry two transgenic copies of DHR96 along with the endogenous wild-type locus. These flies express approximately two-fold more DHR96 protein than wild-type animals (M. Horner, unpublished results). The DHR962X animals are more resistant to starvation than the wild-type controls (Figure 1A). Taken together, these opposite effects of DHR96 loss-of-function and gain-of-function on the starvation response suggest that this factor plays a central role in maintaining energy homeostasis.
As a first step toward determining the basis for these effects on the starvation response, we measured the major forms of stored energy in the animal, glycogen and TAG levels, in control, DHR961, and DHR962X stocks, under both fed conditions and after 20 hours of starvation. Both the DHR961 mutant and DHR962X animals have wild-type levels of glycogen under normal feeding conditions (Figure 1B). The mutant, however, has significantly less glycogen upon starvation than the wild-type control. Interestingly, TAG levels are reduced in both fed and starved DHR961 mutants, and elevated in both fed and starved DHR962X animals relative to controls (Figure 1C). An ~2-fold reduction in TAG is also seen with a second DHR96 null allele, and is effectively rescued by a wild-type genomic DHR96 transgene (Figure S1). Taken together, these observations suggest that the starvation sensitivity of DHR96 mutants is caused by a deficit of stored energy in the form of TAG, and implicate a central role for DHR96 in TAG homeostasis.
To confirm and extend these results, we examined the distribution of neutral lipids in the major sites of fat storage in the animal, the fat body and midgut. Fat body cells were dissected from 1–2 day old control male flies, DHR961 mutants, and the DHR962X overexpression strain, and stained with Nile Red (Figure 1D–F). DHR96 mutant fat body cells are approximately half the size of those in controls, with an average ~3-fold reduction in lipid droplet size (Figure S2A,B). Conversely, fat body cells from the DHR962X overexpression strain are wild type in size, but have lipid droplets that are, on average, nine times larger than those in control fat cells (Figure S2A,B). Similar results were seen when the midguts of control, DHR961 mutant, and DHR962X animals were dissected and stained with Oil Red O (Figure 1G–I). Whereas low levels of neutral lipids can be detected in the gut epithelium of control flies (Figure 1G), no staining is evident in the gut epithelium of DHR961 mutants (Figure 1H), and large lipid droplets are present in the midguts of DHR962X animals (Figure 1I). These results support our measurements of whole animal TAG levels and indicate that DHR96 plays a central role in the process of lipid metabolism.
We next tested whether dietary conditions can alter the TAG levels in DHR96 mutant flies. Control CanS and DHR961 mutants were transferred to a previously characterized low calorie (0.5 SY) or high calorie (2.0 SY) medium and assayed for specific changes in TAG levels after one week (Mair et al., 2005). Control flies maintained on the high calorie diet display a significant increase in whole animal TAG levels relative to that seen in flies maintained on the low calorie diet (Figure 2A). Interestingly, DHR96 mutants grown on either the low calorie or high calorie media display a lean phenotype relative to control animals on the restrictive diet. These observations support the proposal that DHR96 mutants are genetically lean and indicate that they are resistant to diet-induced obesity.
It is possible that a reduced feeding rate could contribute to the lean phenotype in DHR96 mutants. To test this possibility, both control and DHR961 mutant flies were transferred to medium supplemented with a radioactive dCTP tracer and the amount of retained label was measured after a 12 hour period (Carvalho et al., 2005). DHR96 mutants display a mild (12–18%) decrease in the amount of ingested food relative to the controls (Figure 2B). No change in food content, however, was seen upon spectrophotometric measurement of extracts from flies grown on yeast paste supplemented with 0.05% Bromophenol blue (data not shown). Based on these assays, we conclude that a reduction in feeding rate is unlikely to be a major cause of the lean phenotype in DHR96 mutants. Similarly, there is no significant change in the amount of radiolabeled oleic acid retained in DHR96 mutants after a 12 hour feeding period, suggesting that they have no major defects in dietary fatty acid uptake (Figure 2C).
We next wanted to determine if dietary TAG breakdown contributes to the lean phenotype in DHR96 mutants. As a first step toward this goal, TAG levels were measured in control and DHR961 mutants that were maintained for five days on media supplemented with 2 µM Orlistat (Figure 3A). Orlistat (tetrahydrolipstatin) is a widely used over-the-counter weight loss drug (Heck et al., 2000). It acts inside the intestine as a competitive inhibitor of pancreatic and gastric lipases, preventing their interaction with dietary TAG and thus blocking fatty acid release and dietary lipid uptake. Control flies exposed to Orlistat display a significant reduction in fat levels (Figure 3A). A similar effect was seen in mutants for the fly perilipin homolog Lsd-2, which are lean due to reduced levels of fat body TAG, consistent with the conclusion that Orlistat specifically inhibits Drosophila gastric lipases (Grönke et al., 2003) (Figure S3A). In contrast, no effect was seen upon exposing DHR96 mutants to the drug. These observations were confirmed by staining midguts dissected from either untreated or Orlistat-treated control and DHR961 mutants with Oil Red O (Figure 3B–E). Control flies maintained on the high nutrient diet used for this study display clearly detectable levels of neutral lipids in their gut epithelium, and this level is significantly reduced in the midguts of animals exposed to Orlistat (Figure 3B,C). In contrast, DHR96 mutants display no detectable neutral lipids in their midguts, either in the presence or absence of drug (Figure 3D,E). These results raise the possibility that the lean phenotype in DHR96 mutants may arise, at least in part, from decreased TAG lipase activity in the midgut.
To test the possibility that DHR96 mutants have a decreased ability to break down dietary fat, lysates were prepared from dissected control CanS and DHR961 mutant midguts and assayed for TAG lipase activity by examining their ability to cleave a glycerol tributyrate substrate. This experiment showed that lysates from DHR96 mutants have significantly reduced lipase activity relative to lysates from control animals, suggesting that decreased gastric lipase activity is a primary cause of the lean phenotype in DHR96 mutants (Figure 4A). To further test this model we attempted to bypass the need for midgut lipolysis by rescuing the lean phenotype through dietary supplementation with free fatty acids. Control CanS and DHR961 mutant flies were transferred to a lipid depleted medium that was either unsupplemented with lipid, or supplemented with a mixture of stearic acid and oleic acid (free fatty acids) or a mixture of glycerol tristearate and glycerol trioleate (TAG). Extracts were prepared from these animals after seven days and assayed for TAG levels (Figure 4B). As expected, DHR96 mutants maintained on the lipid-depleted medium continue to display the lean phenotype. Interestingly, the same phenotype is seen when DHR96 mutants are maintained in the presence of supplemented TAG, while the lean phenotype is effectively rescued by dietary supplementation with free fatty acids (Figure 4B). This observation indicates that an inability to break down dietary TAG is a major contributing factor to the lean phenotype seen in DHR96 mutants. Phospholipid digestion, however, appears to be normal in DHR96 mutants, as revealed by cleavage of the quenched fluorescent phospholipid PED6, which provides an accurate indicator of gastric phospholipase activity (data not shown) (Farber et al., 2001).
Expression of wild-type DHR96 effectively rescues the lean phenotype of DHR96 mutants, indicating that this phenotype can be attributed to a specific defect in DHR96 function (Figure S1). If, however, DHR96 regulates the breakdown of dietary fat, then we should be able to rescue the mutant lean phenotype by specifically expressing the wild-type receptor in the midgut of DHR96 mutants. Consistent with this proposal, specific expression of wild-type DHR96 in the fat body of DHR96 mutants does not have a significant effect on the low levels of TAG seen in these animals, while specific expression of wild-type DHR96 in the midgut of DHR96 mutants effectively rescues the lean phenotype (Figure 4C). This observation indicates that the reduced levels of TAG seen in DHR96 mutants arise from defects in midgut function. Taken together with our dietary rescue experiments, we conclude that DHR96 plays a central role controlling the breakdown of dietary fat.
Microarray studies were conducted to determine the molecular mechanisms by which DHR96 regulates lipid homeostasis. This study revealed that 136 genes are significantly affected by the DHR96 mutation in fed adult flies, with 94 genes displaying increased levels of expression and 42 genes showing decreased levels of expression (Table S1). A significant number of genes involved in cuticular structure and the peritrophic matrix, which acts as protective layer inside the lumen of the midgut, are up-regulated in the mutant, while many genes that encode predicted α-mannosidases and endopeptidases are expressed at lower levels. Genes more directly involved in metabolism are also over-represented in the list of DHR96-regulated genes. These include two genes that encode larval serum proteins: Lsp1γ (+2.2-fold in the DHR96 mutant) and Lsp2 (+3.1-fold). These proteins are synthesized by the fat body and are thought to provide a source of amino acids to support adult development during metamorphosis (Telfer and Kunkel, 1991). Interestingly, the gene that encodes the larval serum protein receptor, Fbp1 (+3.1-fold), is also regulated by DHR96, suggesting a central role for DHR96 in LSP function (Burmester et al., 1999). Mdr50, which is reduced in its expression in DHR96 mutants (−1.5-fold), is the fly ortholog of mammalian ATP-binding cassette subfamily B, member 4 (ABCB4), a protein that is primarily expressed in the liver and involved in phospholipid transport. Similarly, the fifth most down-regulated gene in DHR96 mutants, CG13325, encodes a protein with a predicted acyltransferase domain that could function in lipid transport. Finally, it is remarkable that the most significantly affected genes in DHR96 mutants include most of the Drosophila orthologs of the Niemann-Pick (NPC) disease genes that play central roles in cholesterol metabolism, including npc1b (+2.6-fold up) and five NPC2 family members: npc2c (+1.7-fold), npc2d (−13-fold), npc2e (+21-fold), and two genes that are below our 1.5-fold cutoff in expression level: npc2g and npc2h (both −1.4-fold).
Closer examination of the DHR96-regulated genes reveals that many are expressed exclusively, or most abundantly, in the midgut. Of the 132 genes that are misregulated in DHR96 mutants, 42 are included in the 2,695 midgut-specific genes identified by Li et al (2008)(32%; p-value=9×10−9). A similar overrepresentation of midgut-expressed genes can be seen by surveying the expression patterns on FlyAtlas (Chintapalli et al., 2007). For example, 12 of the 15 most down-regulated genes, and 9 of the 15 most up-regulated genes in DHR96 mutants, are highly expressed in the midgut (Figure S4).
Finally, we found that many of the DHR96-regulated genes are located next to one another in the genome. The gene clusters include three genes with predicted α-mannosidase activity, CG9463 (−3.3-fold), CG9466 (−5.9-fold), and CG9468 (−4.3-fold), all of which are among the most highly down-regulated genes in the mutant. A divergently transcribed pair of genes with predicted α-mannosidase activity are also misregulated in DHR96 mutants, CG5322 (−1.3-fold) and CG6206 (−1.8-fold), as well as two genes with predicted α-glucosidase activity, CG14934 (+1.5-fold) and CG14935 (−1.6-fold), and a cluster of two Jonah genes that encode predicted gastric peptidases, Jon65Ai and Jon65Aii (both −1.7-fold). Two genes with predicted sphingomyelin phosphodiesterase activity, CG15533 (+2.0-fold) and CG15534 (+1.3-fold), are regulated by DHR96. Moreover, many of the NPC genes that are misregulated in DHR96 mutants lie within gene clusters. These include npc2d (−13-fold), npc2e (+21-fold), and npc2c (+1.7-fold). An adjacent gene, fancl, which has no known function, is down-regulated 1.5-fold in DHR96 mutants. In addition, npc2g and npc2h, are located next to one another in the genome.
Two genes that encode predicted TAG lipases were identified in our microarray study of DHR96 mutants, CG5932 and CG31091 (Table S1). Both genes encode members of the α/β-hydrolase fold lipase family and are homologs of human gastric lipases, with 37% and 33% amino acid identity, respectively. In addition, both genes are expressed almost exclusively in the larval and adult midgut (Chintapalli et al., 2007), raising the possibility that their misexpression may contribute to the reduced ability of DHR96 mutants to break down dietary TAG. Validation of the effects of the DHR96 mutation on CG5932 and CG31091 expression by northern blot hybridization showed that CG31091 mRNA levels are very low, consistent with the CG31091 expression levels reported on FlyAtlas (Chintapalli et al., 2007) (data not shown). In contrast, CG5932 is abundantly expressed, is down-regulated by starvation in wild-type flies (Figure 5A, lanes 1,3), and this expression is significantly reduced in fed or starved DHR96 mutants (Figure5A, lanes 2,4). CG5932 is also the fourth most highly down-regulated gene identified in our microarray study. Chromatin immunoprecipitation of DHR96 protein from wild-type lysates revealed direct binding to sequences immediately upstream from CG5932 (Figure S5). This binding is not seen at a promoter that is independent of DHR96 regulation or in chromatin immunoprecipitation experiments using lysates made from DHR96 mutants, suggesting that it represents a specific DNA-protein interaction. Moreover, expression of wild-type DHR96 in the midgut of DHR96 mutants is sufficient to restore normal CG5932 expression (data not shown). Taken together, these observations suggest that DHR96 maintains TAG homeostasis through direct regulation of CG5932 expression.
If CG5932 is essential for TAG homeostasis then disruption of CG5932 function should lead to changes in TAG levels. To test this possibility, we used Act-GAL4 to drive the expression of a UAS-CG5932 RNAi construct. Whereas flies that carry either the Act-GAL4 driver alone or the UAS-CG5932 RNAi construct alone display normal levels of TAG and CG5932 expression, combining Act-GAL4 with the UAS-CG5932 RNAi construct resulted in significant reduction of both whole animal TAG levels and CG5932 mRNA accumulation (Figure 5B). Similar results were seen upon driving UAS-CG5932 RNAi with the midgut-specific Mex-GAL4 driver (Figure S6A). These animals are also sensitive to starvation, consistent with their lean phenotype (Figure S6B). In addition, purified CG5932 protein has lipolytic activity in vitro, as demonstrated by its ability to break down a glycerol tributyrate substrate (Figure S7). Taken together, these observations indicate that CG5932 plays a critical role in the midgut to regulate whole animal TAG homeostasis, most likely through its effect on the breakdown of dietary fat.
If the down-regulation of CG5932 expression in DHR96 mutants contributes to their inability to break down dietary TAG, then restoring CG5932 expression specifically in the midgut of these animals should rescue their lean phenotype. To test this possibility, we established transformant lines that carry two different insertions of a UAS-CG5932 transgene. As expected, the presence of either the midgut-specific Mex-Gal4 driver alone or each UAS-CG5932 transgene had little or no effect on the reduced TAG levels seen in the DHR96 mutant (Figure 6). Combining the Mex-GAL4 driver with either of the UAS-CG5932 transgenes in a DHR96 mutant, however, allows the mutant to recover normal levels of TAG (Figure 6). In addition, overexpression of CG5932 does not stimulate an increase in whole animal TAG levels, indicating that while CG5932 is necessary for TAG homeostasis, it is not sufficient to drive TAG accumulation (Figure S8). These results, combined with our observation that the lean phenotype in DHR96 mutants can be rescued by dietary supplementation with free fatty acids (Figure 4B), and the absence of an effect of Orlistat treatment in DHR96 mutants (Figure 3A) or Act>CG5932 RNAi animals (Figure S3B), suggests that DHR96 controls whole animal TAG levels through its regulation of the CG5932 gastric lipase.
Recent studies have implicated roles for mammalian PXR and CAR in controlling lipid metabolism, although little is known regarding the molecular mechanisms by which they exert these effects. Here we show that the single Drosophila ortholog of PXR and CAR, DHR96, plays an essential role in maintaining whole animal TAG levels through the proper breakdown of dietary fat. Our results indicate that DHR96 acts through a previously uncharacterized gastric lipase encoded by CG5932 to promote dietary lipid uptake and maintain TAG homeostasis.
Although DHR96 null mutants are viable and fertile, with no morphological defects, they die significantly more rapidly than genetically-matched control flies under starvation conditions, while DHR96 overexpression leads to starvation resistance (Figure 1A). The effects of these genotypes on the major forms of stored energy in the animal, glycogen and TAG, are consistent with their effects on the starvation response. DHR96 mutants have reduced levels of TAG under both fed and starved conditions, while DHR96 overexpression leads to increased TAG levels (Figure 1C). Although no effects are seen on whole animal glycogen levels in fed animals that either lack or overexpress DHR96, the mutants consume significantly more glycogen upon starvation than do controls (Figure 1B). This rapid utilization of glycogen stores is most likely due to the decreased energy contribution from TAG. Taken together, these observations suggest that the starvation sensitivity of DHR96 mutants can be attributed to their lean phenotype, while the starvation resistance of the DHR962X strain is due to their excess energy in the form of TAG. This proposal is supported by the observation that genetically elevating the levels of TAG in DHR96 mutants by introducing mutations in bmm or AKHR, which control distinct aspects of fat body TAG lipolysis (Grönke et al., 2007), effectively rescues their starvation sensitivity (Figure S9). In addition, the opposite effects of DHR96 loss-of-function and gain-of-function on both the starvation response and TAG levels argues that this receptor plays a central role in maintaining whole animal TAG homeostasis.
Several lines of evidence support the conclusion that DHR96 exerts its primary metabolic functions through the midgut. These include our initial observation that DHR96 mutants are resistant to treatment with the gastric lipase inhibitor Orlistat (Figure 3) and display reduced levels of midgut lipolytic activity (Figure 4A). In addition, dietary supplementation with free fatty acids, but not TAG, is sufficient to rescue the lean phenotype of DHR96 mutants, as is midgut-specific expression of wild-type DHR96 in a DHR96 mutant background (Figure 4C). We also see a dramatic effect on lipid levels in the midgut, where almost no neutral lipids are detectable in DHR96 mutants and enlarged lipid droplets are evident in DHR962X flies maintained on a normal diet (Figure 1H,I). Interestingly, while the lumen of the midgut is not evident in control animals, we clearly see material in the lumen of DHR96 mutant midguts or in control flies that are treated with Orlistat (Figure 1H, Figure 3C–E). In some cases, this material is stained by Oil Red O, suggesting that it may represent an increased level of undigested fat in these animals. This phenotype would be similar to that seen in humans who have defects in intestinal lipase activity (Ligumsky et al., 1990). Likewise, an increase in the passage of undigested dietary fat is a complication associated with Orlistat treatment in patients (Heck et al., 2000). Taken together, these observations support the conclusion that DHR96 acts in the midgut to regulate the breakdown of dietary fat.
An essential role for DHR96 in the midgut is further supported by our microarray study, which revealed that many DHR96-regulated genes are primarily expressed in this tissue. Interestingly, many of these genes have predicted roles related to the breakdown of dietary nutrients (Table S1). These include down-regulation of multiple genes with predicted α-mannosidase activity, which is involved in the breakdown of the complex sugars found on glycoproteins. Many genes that encode trypsins and endopeptidases are also expressed at reduced levels in DHR96 mutants as well as a few genes that encode predicted α-glucosidases, which are involved in the breakdown of dietary carbohydrates. In addition, a number of genes involved in the formation of the peritrophic matrix are more abundantly expressed in DHR96 mutants. This matrix is comprised of chitin and peritrophic proteins, and acts as a protective layer for the epithelial surface of the midgut (Hegedus et al., 2009). The peritrophic matrix also has critical roles in facilitating digestion. Only smaller molecules that arise from the initial digestion of complex nutrients, including peptides, sugars, and lipids, can move through the peritrophic matrix for final digestion and absorption by the midgut epithelium. These events are controlled by the selective partitioning of digestive enzymes to different sides of the peritrophic matrix as well as within the matrix itself. Thus, while midgut morphology appears normal in DHR96 mutants, the effect of the mutation on the peritrophic matrix could impact nutrient digestion and absorption.
In addition to genes that regulate different aspects of lipid metabolism, our microarray study of DHR96 mutants identified widespread effects on the expression of Drosophila homologs of NPC disease genes. The npc1b gene, which encodes an essential cholesterol transporter and the ortholog of mammalian NPC1L1 (Voght et al., 2007), is the tenth most highly up-regulated gene in DHR96 mutants. In addition, five of the eight Drosophila NPC2 genes are misregulated in DHR96 mutants: npc2c, npc2d, npc2e, npc2g, and npc2h (Huang et al., 2007). These genes encode homologs of mammalian NPC2, which is involved in intracellular cholesterol trafficking (Huang et al., 2007). Remarkably, two of these genes are the most highly up- and down-regulated genes identified in the mutant (npc2e and npc2d, respectively; Table S1). Moreover, many of these npc genes are located in clusters, suggesting that they are co-regulated by the receptor. Although the function of these npc2 genes is unknown, their disproportionate representation within the list of DHR96-regulated genes implies a critical role for the receptor in regulating cholesterol trafficking. Indeed, dietary cholesterol triggers a widespread transcriptional response in Drosophila that is dependent on DHR96 function (Horner et al., 2009). Moreover, this study showed that DHR96 mutants display defects in their ability to maintain cholesterol homeostasis when grown on a high cholesterol diet. Taken together with the results presented here, our studies suggest that DHR96 plays an essential role in the midgut to coordinate the processes of TAG and cholesterol breakdown, absorption and trafficking.
Two genes with predicted TAG lipase activity are expressed at lower levels in DHR96 mutants (Table S1). One of these genes, CG5932, is abundantly expressed in the larval and adult midgut, encodes a protein that is highly related to human gastric lipase (LIPF, 37% identity over 358 amino acids), and is essential for maintaining whole animal TAG levels (Figure 5). In addition, restoring CG5932 expression in the midguts of DHR96 mutants is sufficient to rescue their lean phenotype, defining this gene as a critical functional target of the receptor (Figure 6). Interestingly, CG5932 expression is also regulated by starvation, with reduced expression in the absence of food and increased expression upon refeeding (Figure 5A) (Gershman et al., 2006). This regulation is consistent with an essential role for CG5932 in the breakdown of dietary fat, where its expression is up-regulated when food is present. This response, however, is unaffected in DHR96 mutants, indicating that it is under independent control, possibly by known regulators of the starvation response such as dFOXO (Gershman et al., 2006).
The identification of CG5932 as a key functional target of DHR96 raises the question of how that regulation is achieved. We have tested a range of dietary parameters and candidate ligands using the GAL4-DHR96 ligand sensor, but have been unable to identify conditions that activate the DHR96 LBD (Palanker et al., 2006). In addition, the DHR96 binding site remains undefined. DHR96 has a unique P-box sequence within its DBD, which determines its DNA-binding specificity. This sequence is only shared by three C. elegans NRs, DAF-12, NHR-48, and NHR-8. Consistent with this observation, we found that DHR96 protein fails to bind to most canonical NR binding sites, except for weak binding to a palindromic EcR response element (Fisk and Thummel, 1995). One study has shown that DAF-12 displays preferential binding to a direct repeat of two distinct hexanucleotide sequences (AGGACA and AGTGCA), separated by five nucleotides (DR5) (Shostak et al., 2004). Whether DAF-12 contacts these sequences as a homodimer or a heterodimer with another NR, however, remains to be determined. The observation that many DHR96-regulated genes are arranged in clusters and DHR96 binds directly to a region upstream from the CG5932 start site provides an ideal context for defining its DNA binding specificity as well as determining the molecular mechanisms by which DHR96 coordinates target gene transcription.
Our studies of DHR96 raise the interesting possibility that the defects in xenobiotic detoxification seen in DHR96 mutants may arise, at least in part, from its role in regulating midgut metabolic activity. As noted in our original study, most of the genes that are regulated by phenobarbital in wild-type flies do so independently of DHR96 (King-Jones et al., 2006). These genes include representatives of the major classes associated with xenobiotic detoxification: cytochrome P450 monooxygenases (P450s), glutathione S-transferases (GSTs), carboxylesterases, and UDP-glucuronosyl transferases (UGTs). Moreover, some Phenobarbital-regulated genes that are misregulated in DHR96 mutants still show a transcriptional response to the drug, although that response is muted. These observations indicate that one or more other factors contribute to the transcriptional response to xenobiotic challenge in Drosophila. Interestingly, similar results have been observed in vertebrates, where less then half of the genes that are regulated by xenobiotics are affected in PXR and CAR mutant mice (Maglich et al., 2002; Ueda et al., 2002). It remains unclear, however, whether this lack of regulation might be due to functional redundancy between these mammalian receptors.
There are several possible mechanisms by which the metabolic functions of DHR96 could impact xenobiotic responses. First, a recent study of the expression patterns of Drosophila P450 genes showed that 34 of 60 genes that could be detected in third instar larvae are expressed in the midgut or hindgut (Chung et al., 2009). A similar over-representation of P450 genes is evident in a microarray study of midgut expressed genes (Li et al., 2008). These observations indicate that, contrary to previous assumptions, the gut and not the fat body may be a critical site for xenobiotic detoxification (Chung et al., 2009). If this is true, then the sensitivity of DHR96 mutants to phenobarbital or DDT treatment may be affected by the effects of this mutation on midgut physiology. This could occur through defects in the peritrophic matrix or through changes in the ability of the midgut to absorb lipophilic xenobiotic compounds such as DDT. An alternative possibility is that the sensitivity of DHR96 mutants to xenobiotics might be due to their decreased energy stores. Detoxification requires energy expenditure. For example, P450s consume NADPH or NADH for their oxidation of xenobiotics, UGTs consume glucose, and GSTs consume glutathione. Thus, the reduced levels of stored energy in DHR96 mutants might compromise their ability to properly inactivate toxic compounds. In addition, the reduced lipid stores in DHR96 mutants might exert an indirect effect on xenobiotic responses by lowering the ability of the animal to sequester toxins in the fat reserves of the animal. Thus, there are multiple pathways by which the midgut-specific metabolic defects associated with the DHR96 mutation might indirectly affect xenobiotic responses in these animals. Further studies are required to test this possibility and clarify the functional overlaps between the roles of DHR96 in lipid metabolism and xenobiotic detoxification.
Several recent studies have demonstrated roles for both PXR and CAR in lipid metabolism (Moreau et al., 2008). CAR can repress the transcription of genes encoding carnitine palmitoyltransferase and enoyl-CoA isomerase, key steps in lipid β-oxidation (Ueda et al., 2002). CAR mutant mice are also sensitive to starvation, much as we observe for DHR96 mutants, and lose weight more rapidly than wild-type mice when maintained on a low calorie diet (Maglich et al., 2004). Transgenic expression of a constitutively active form of PXR in the mouse liver leads to hepatic steatosis along with reduced expression of lipid catabolic genes and increased expression of genes involved in lipid synthesis (Zhou et al., 2006). Importantly, similar effects were observed upon pharmacological activation of PXR using a specific agonist, indicating that the endogenous receptor can contribute to lipid homeostasis (Hoekstra et al., 2009; Nakamura et al., 2007). Most recently, a mutation in CAR has been shown to normalize the elevated serum TAG levels seen in leptin deficient mice or in wild-type mice maintained on a high fat diet (Maglich et al., 2009). Conversely, treatment of wild-type mice with a selective CAR agonist leads to increased serum TAG levels, and this response fails to occur in a CAR mutant background. Taken together, these studies indicate that activation of PXR/CAR receptors leads to lipid accumulation while a loss of PXR/CAR activity leads to reduced lipid levels, defining a central role for these NRs in lipid homeostasis. Their role in normal lipid metabolism, however, remains unknown, although it may masked by functional redundancy between the two receptors. Similarly, the molecular mechanisms by which PXR and CAR can modulate lipid levels remain to be defined.
Our genetic studies of DHR96 demonstrate that these metabolic activities of PXR and CAR have been conserved through evolution, and represent an essential ancestral function for this NR subfamily. Moreover, the observation that DHR96 overexpression leads to lipid accumulation and DHR96 mutants are lean suggests that the molecular mechanisms that underlie these effects are also conserved across species. This conclusion is supported by genetic studies of the C. elegans member of this subfamily, DAF-12, which indicate that this receptor is also required for proper levels of stored fat (Gerisch et al., 2001). Our study provides further evidence of a role for this NR subfamily in normal lipid homeostasis and defines the control of dietary fat breakdown as a key step at which this regulation is achieved. In addition, this work provides a foundation for understanding how dietary lipid uptake can impact normal lipid metabolism in Drosophila and provides a genetic model for characterizing how dietary factors can lead to lipid metabolic disorders such as obesity.
The following stocks were used in this study: DHR961 (King-Jones et al., 2006), Cg-Gal4 (Hennig et al., 2006), Mex-Gal4 (Phillips and Thomas, 2006), Act-Gal4/CyO (Bloomington # 25374), bmm1 (Grönke et al., 2005), AKHR2 (Grönke et al., 2007), Lsd-2KG00149 (Grönke et al., 2003), w1118, DHR962X (M. Horner, unpublished), UAS-DHR96 (M. Horner, unpublished), and UAS-CG5932 RNAi (National Institute of Genetics stock 5932R-3). Flies were maintained on standard Bloomington Stock Center medium with malt at 25°C. The UAS-CG5932 P-element construct was made using oligonucleotide primers: 5’-ATAGAATTCATGAATCCAATCTTCTGCGC-3’ and 5’-ATACTCGAGCTAGCGACCTTCGTAGGAGT-3’ to amplify the CG5932 cDNA (Genbank NM_140972) from the DGRC LP10120 cDNA clone by PCR. Following digestion with EcoRI and XhoI, the cDNA was inserted into the corresponding sites of the pUAST vector. This UAS-CG5932 construct was then integrated into the w1118 genome.
All studies used 1–2 day old adult male flies that were aged 5–7 days prior to the experiment. Starvation sensitivity assays were conducted by transferring 15 samples of 20 flies of each genotype into vials containing 1% agar. Mortality was assayed every four hours as determined by a lack of responsiveness to touch. For glycogen and TAG assays, 5–10 mature adult male flies were homogenized in 100–200 μl PBST (PBS, 0.1% Tween 20), heated at 70°C for 5 minutes to inactivate endogenous enzymes, and the homogenate was cleared by centrifugation for 3 min. The supernatant was diluted 1:4 with PBST and assayed for glycogen levels as described (Palanker et al., 2009). TAG assays were conducted as described (Palanker et al., 2009), using the Stanbio triglyceride liquicolor kit (2100-430). The glycogen and TAG levels in each sample were normalized for total protein as determined by a Bradford assay. Both glycogen and TAG data were compiled from six samples collected from each genotype under each condition. All data is presented as normalized to a wild-type level of 100%. All assays were repeated three times and a representative experiment is presented in each figure. Nile Red staining was performed as described by Grönke et al. (2005), using Nile Red mounting media (20% glycerol in PBS, with a 1:10,000 dilution of 10% Nile Red in DMSO) and imaged on a Leica TCS SP2 confocal microscope using an excitation wavelength of 543 nm and a 600–650 nm emission spectrum. For Oil Red O stains, midguts from mature adult flies were dissected and fixed in 4% paraformaldehyde/PBS for 20 min. The midguts were washed with distilled water and incubated in 100% propylene glycol for 5 min. Specimens were then incubated at 60°C in Oil Red O stain (0.5% Oil Red O in propylene glycol), washed twice with propylene glycol, washed three times with PBS, and mounted in 20% glycerol/PBS for imaging.
All studies used 1–2 day old adult male flies that were aged 5–7 days prior to the experiment. Diet-induced obesity was assayed by transferring flies to either a low calorie 0.5 SY diet or a high calorie 2.0 SY diet for 7 days, after which extracts were prepared and assayed for TAG and protein (Mair et al., 2005). Food intake was measured by transferring flies to Bloomington cornmeal, yeast, and molasses medium supplemented with 0.1 µC/ml 32P-dCTP (Perkin-Elmer) for 12 hours, after which they were transferred to normal unlabeled food for 4 hours to remove label that was nonspecifically bound to the outside of the animal (Carvalho et al., 2005). Ten to twelve groups of 20 flies were collected, washed once with 0.1% BSA, and assayed for label retention on a scintillation counter. Fatty acid uptake was determined in a similar manner using 3H-oleic acid (Moravek Biochemicals) at a final concentration of 3 µC/ml. Treatment with Orlistat was conducted by growing flies on a high nutrient molasses medium (Bloomington cornmeal, yeast, and molasses media) with or without 2 µM Orlistat for 5–7 days. TAG levels were then assayed as above. All data shown are from two parallel data sets of six samples/stock/condition compiled together and repeated at least three times. Dietary lipid supplementation was performed by transferring mature adult males to lipid-depleted 1.0 SY for 5–7 days. This medium was prepared by extracting the yeast extract and agar components of the SY medium overnight with chloroform, followed by a second 4 hour chloroform extraction. Both components were then allowed to dry for 2–3 days in a fume hood. The lipid-depleted medium was supplemented with either free fatty acids (5 mg/ml stearic acid and 5 mg/ml oleic acid, ChemService) or TAG (5 mg/ml glycerol tristearate and 5 mg/ml glycerol trioleate, ChemService). All data shown are from two collections of six samples collected from each genotype under each condition. Each experiment was repeated at least three times.
Adult male CanS and DHR961 flies were collected 1–2 days after eclosion and allowed to age for seven days on normal medium. RNA was extracted from these animals using Trizol (Gibco) and purified on RNAeasy columns (Qiagen). All samples were prepared in triplicate to facilitate subsequent statistical analysis. Probe labeling, hybridization to Affymetrix GeneChip® Drosophila Genome 2.0 Arrays, and scanning, were performed by the University of Maryland Microarray Core Facility. Raw data was subjected to quantile normalization using R statistical analysis software and gene expression changes were determined using SAM 2.0, with a <5% estimated false positive rate and a 1.5-fold cut-off in expression level (Tusher et al., 2001). Comparison between microarray datasets was performed using Microsoft Access. Microarray data from this study can be accessed at NCBI GEO (accession number: XXX).
Statistical significance was calculated using an unpaired two-tailed Student’s t-test with unequal variance. All quantitative data are reported as the mean ± SEM.
We thank M. Horner for providing the DHR962X stock, B. Milash for advice on microarray analysis, M. Van Gilst for suggesting the use of Orlistat and for technical advice with these experiments, J. Tennessen, A.-F. Ruaud, L. Palanker, M. Horner, and M. Metzstein for helpful discussions, and M. Horner, A.-F. Ruaud, and J. Tennessen for comments on the manuscript. M.S. was supported by an NIH Developmental Biology Predoctoral Training Grant (5T32 HD07491). This research was supported by NIH grant 1R01DK075607.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.