CoA protects against the deleterious effects of caloric overload in Drosophila1

We developed a Drosophila model of T2D in which high sugar (HS) feeding leads to insulin resistance. In this model, adipose TG storage is protective against fatty acid toxicity and diabetes. Initial biochemical and gene expression studies suggested that deficiency in CoA might underlie reduced TG synthesis in animals during chronic HS feeding. Focusing on the Drosophila fat body (FB), which is specialized for TG storage and lipolysis, we undertook a series of experiments to test the hypothesis that CoA could protect against the deleterious effects of caloric overload. Quantitative metabolomics revealed a reduction in substrate availability for CoA synthesis in the face of an HS diet. Further reducing CoA synthetic capacity by expressing FB-specific RNAi targeting pantothenate kinase (PK orfumble) or phosphopantothenoylcysteine synthase (PPCS) exacerbated HS-diet-induced accumulation of FFAs. Dietary supplementation with pantothenic acid (vitamin B5, a precursor of CoA) was able to ameliorate HS-diet-induced FFA accumulation and hyperglycemia while increasing TG synthesis. Taken together, our data support a model where free CoA is required to support fatty acid esterification and to protect against the toxicity of HS diets.

The prevalence of obesity, or excessive TG accumulation in adipose tissue, results from dietary excess and is increasing in many parts of the world. Obesity is a risk factor for several diseases, including cardiovascular disease, T2D, and cancer. A growing consensus exists, however, that obesity is protective against the adverse biochemical effects of caloric excess (1)(2)(3)(4). Several studies have shown that obesity and T2D are accompanied by increases in FFAs, ceramides, DAG, and other potentially toxic lipids Hemolymph was collected and shipped frozen. All metabolite extraction was done by Metabolon. GC-MS (glycerol, cysteine) and LC-MS (positive ion monitoring mode; pantothenate, carnitine, palmitoyl-carnitine, and oleoyl-carnitine) were used to isolate the peaks representing each analyte, with injection standards used at fi xed concentrations to quantify the relative amounts of each metabolite in six biological replicates per diet. Metabolites were identifi ed by automated comparison of the ion features in the extracted samples to a reference library as described previously ( 29 ). Metabolite levels were normalized to protein content for FB, and to total volume for hemolymph.

CoA-SH determination
Eighty wild-type ( w 1118 ) wandering third-instar larvae were homogenized in 10 mM DTT/water and frozen at Ϫ 80°C. Homogenates were defrosted, TCA precipitated, and washed six times with ether to remove lipids. Aqueous fractions were dried, resuspended, and run on a Waters HPLC, as described ( 17 ).

Metabolic assays
TG, FFA, and glycogen were assayed from groups of six frozen wandering L3 larvae as previously described ( 17 ). Hemolymph was isolated from groups of fi ve to twenty wandering larvae and assayed as previously described ( 17 ).

RESULTS
In previous gene expression profi ling experiments of HS-fed insulin-resistant larvae, we saw a dramatic (10-fold) increase in whole body expression of vanin-like (CG32754), which is predicted to encode a CoA metabolic enzyme, pantetheinase, that converts pantetheine to pantothenate (also known as vitamin B5) and cysteamine ( 12 ). We previously observed changes in fatty acid synthesis and esterifi cation, which require CoA, in HS-fed larvae ( 17 ). We hypothesized that the increase in pantetheinase might be evidence of a homeostatic response to defects in CoA availability. FlyAtlas ( 30 ) data shows the highest levels of vaninlike expression in the midgut, with expression only in midgut, hindgut, and Malpighian tubule (fl yatlas.org). Therefore, we quantifi ed vanin-like expression in isolated guts.
HS feeding increased expression of vanin-like in the midgut by 4-fold ( Fig. 1B ), suggesting that the gut might contribute to maintenance of systemic CoA levels by producing pantothenate to be used in other tissues when CoA levels are depleted. We also observed a 50% increase in expression of the putative pantothenate transporter, CG10444 , in gut RNA-seq, although edgeR did not identify a signifi cant difference (data not shown), consistent with a need for pantothenate or CoA during HS feeding. Once exported from the gut, circulating pantothenate could be converted back into CoA to support fatty acid synthesis and esterifi cation in the FB, which is essential for larval survival on a HS diet ( 17 ) ( Fig. 1A ).
We focused subsequent studies of expression and metabolism in the FB by using RNA-seq and metabolomics to characterize CoA metabolism there. Thioester hydrolase (encoded by CG1774), an enzyme that catalyzes the production of CoA-SH from acyl-CoA rather than pantothenate, was increased 3-fold at the mRNA level in whole one for each fatty acid to enter the mitochondrion and a second CoA for thiolysis after transport ( 26 ). CoA also regulates intracellular redox state via its free sulfhydryl group ( 25 ) and it regulates ketone biogenesis via acetyl-CoA ( 27 ).
In this study, we set out to identify metabolites that could contribute to lipotoxicity in the face of caloric overload. Our studies revealed that CoA levels were reduced, while FFA levels were increased, in HS-fed larvae. Reducing FB levels of pantothenate kinase (fumble, CG5725) or phosphopantothenyl cysteine synthase (PPCS, CG5629), both of which catalyze steps in CoA synthesis from pantothenic acid (PA), led to reduced TG storage and an increase in the severity of HS diet-induced fatty acid accumulation. By contrast, supplementing the HS diet with the CoA precursor, PA, increased TG synthesis and lowered glucose and FFA levels in the presence of caloric overload. Thus, CoA is limiting in the face of a HS diet, and we propose that increasing CoA levels represents a novel therapeutic target for individuals with obesity-associated metabolic disorders.

Genetics
Wild-type w 1118 lines were from the Vienna Drosophila Resource Center. TRiP control, UAS-PK i and UAS-PPCS i (stocks GL00149 and JF03206, respectively) UAS-RNA i lines were from Harvard's DRSC TRiP collection. UAS-Dcr2; r4-GAL4 was used to express transgenes in the larval FB ( 28 ).

Diets
The control (0.15 M sucrose) and HS (0.7 M sucrose) diets were made using a modifi ed Bloomington semi-defi ned food as described previously ( 12 ). D-PA hemicalcium salt was from Sigma (P5155) and was added in a volume of 30 l to a fi nal concentration of 0.3 to 3 mM, with 3 mM producing optimal phenotypic rescue. Water (30 l) was added to HS food as a control in these experiments. Wild-type w 1118 were used in PA supplementation experiments.

Expression analyses
FBs were isolated from wild-type w 1118 ; UAS-Dcr2; r4-GAL4 or w 1118 ; UAS-Dcr2; cg-GAL4 wandering third-instar larvae and RNA isolated and quantifi ed as described using Illumina Hi-Seq ( 17 ). Data from both control lines were pooled to increase power to detect expression differences due to diet. A total of 13 biological FB replicates per diet were analyzed over several lanes. These data were deposited at GEO (GSE76214). For guts, six biological replicates were isolated and RNA extracted. RT was used to make cDNA, which was then analyzed by quantitative (q)PCR . Primers used to detect vanin-like mRNA were 5 ′ -TCCCGAGGATCAGATA-AACC-3 ′ and 5 ′ -ACAGGGTCACCAGAAACTCC-3 ′ . Vanin-like levels were normalized to RpL32 mRNA ( CG7939 ) using 5 ′ -CAGCATA-CAGGCCCAAGAT-3 ′ and 5 ′ -GCACTCTGTTGTCGATACCC-3 ′ . Similar results were observed using a primer pair targeting a different region of vanin-like .

PA, carnitine, acyl-carnitine, cysteine, and glycerol determination
FBs were isolated from wild-type w 1118 ; UAS-Dcr2; r4-GAL4 wandering third-instar larvae and immediately placed in PBS on ice, then homogenized before shipping on dry ice to Metabolon. which encode the intermediate steps in the CoA biosynthesis pathway. We also observed increases in mRNA encoding two other enzymes that produce free CoA (Ppat-Dpck and HMG-CoA synthase, Fig. 1G, H ). Therefore, we decided to study this pathway biochemically.
The gene expression results shown in Fig. 1 suggested compensatory regulation of CoA production in HS-reared larvae. Therefore, we hypothesized that CoA might limit fatty acid esterifi cation into TG in the face of a HS diet ( Fig. 2A ). First, we characterized the levels of free CoA in whole larvae reared on control or HS diets. HS feeding led to a signifi cant decrease in CoA concentrations in whole animals ( 12 ). We observed a corresponding increase in thioester hydrolase expression in FBs upon HS feeding ( Fig.  1C ). The transcriptional upregulation of both vanin-like and thioester hydrolase in gut and FB, respectively, could occur in response to a defi cit in CoA. CoA is required to produce fatty acid substrates for lipogenesis in the FB and for over 100 other reactions ( 21,22 ). Examining the pathway more closely in FB gene expression datasets, we noted upregulation of catalytic steps in CoA synthesis including pantothenate kinase (CG5725, known as fumble in Drosophila ; Fig. 1D ), although there was no change in expression of genes encoding the enzymes PPCS and Ppcdc ( Fig. 1E, F ), metabolism accumulated in both the FB (glycerol and carnitine; Fig. 2D, E , respectively) and hemolymph (palmitoyl-carnitine and oleoyl-carnitine; Fig. 2F, G , respectively). Glycerol requires CoA in order to generate TGs from FFAs, whereas carnitine, palmitoyl-carnitine, and oleoylcarnitine all require CoA in order to promote fatty acid ␤ -oxidation ( Fig. 2A ). Our metabolomic analyses were not sensitive enough to detect free CoA or fatty acyl-CoAs in these samples, but support the model nonetheless.
third-instar larvae ( Fig. 2B ), consistent with a trend toward decreased CoA shown in our previous work ( 17 ). Metabolomic analyses of FBs and hemolymph supported a role for pantothenate in lipogenesis during challenge with HS diets. Pantothenate levels were decreased in the FB, consistent with an increase in shunting of pantothenate toward CoA synthesis via increased pantothenate kinase expression in FBs from larvae fed HS diets ( Fig. 2C ). At the same time, intermediates that require free CoA to promote lipid we concluded that CoA is likely limiting for several processes on HS diets, including TG storage and fatty acid ␤oxidation. Therefore, we tested to determine whether we could rescue some of the effects of HS diets by increasing CoA levels. To do this, we supplemented the HS diet with the CoA precursor, PA, to test whether it could increase the ability of larvae to store TG when fed HS diets. PA supplementation signifi cantly increased TG storage in HS-fed larvae ( Fig. 4A ) and reduced FFA concentrations ( Fig. 4B ). PA supplementation also increased weight and reduced hemolymph glucose in larvae reared on HS diets, consistent with a potential increase in insulin signaling ( Fig. 4C, D ). The decrease in hemolymph glucose was likely not due to incorporation into disaccharides or polysaccharides because PA did not affect levels of trehalose or glycogen ( Fig. 4E, F ). PA supplementation did not affect free CoA-SH levels ( Fig.  4G ). We also tested whether PA could improve insulin sensitivity in cultured FBs, as measured by exogenous insulin stimulation of Akt phosphorylation at serine 505. No significant improvement in insulin responsiveness was seen when PO 4 -Akt was measured in insulin-stimulated FBs from wandering larvae reared on HS compared with HS + PA (4H). As an additional dietary approach, we tested whether adding cysteine to pantothenate could improve any HS-induced Because steady-state levels of metabolites do not refl ect the metabolic fl ux through that pathway, we took a functional approach to test whether the conversion of pantothenate to CoA was important for tolerance of HS diets. We targeted two enzymes that catalyze CoA synthesis from pantothenate, the rate-limiting pantothenate kinase ( PK also known as fumble , CG5725 ) and phosphopantothenoylcysteine synthase ( PPCS , CG5629 ), which, like fumble , is required for growth of Drosophila S2 cells ( 31 ). Because the FB is the primary site of TG synthesis in the fl y, we used FB-specifi c RNAi to reduce CoA biosynthesis from pantothenate by targeting PK/fumble and PPCS in this tissue ( Fig. 1A ). The r4-GAL4 driver was used in combination with a UAS-Dicer2 transgene to maximize degradation of endogenous mRNA. Knockdown of either gene product led to reduced numbers of larvae on HS diets (not shown), consistent with an essential role for CoA in supporting TG synthesis. Mutant larvae were leaner, with increased FFA levels, when compared with wild-type controls reared on the same HS diet ( Fig. 3A, B ). These larvae also exhibited reduced size, compared with HS-fed wild-type larvae ( Fig.  3C ). Surprisingly, no increase in hemolymph glucose concentration was observed in either mutant ( Fig. 3D ).
Given that knockdown of the CoA synthetic enzymes, pantothenate kinase and PPCS, exacerbated phenotypes, not tolerate cysteine supplementation ( 32 ) (data not shown). A modest but nonsignifi cant increase in FB cysteine levels was observed by LC-MS (1.5-fold, P < 0.2), suggesting that cysteine depletion was not a contributing factor to the observed reduction in CoA.
phenotypes, as cysteine also contributes to CoA synthesis as a substrate for PPCS ( Fig. 1A ). No improvements were observed relative to pantothenate supplementation, and decreased weights were observed with cysteine alone, in agreement with another recent study showing that fl ies do It is interesting to note that some studies have shown a benefi t for cysteine supplementation in T2D ( 53,54 ). We presume that a number of metabolites have the potential to become rate-limiting under different physiological conditions. Nonetheless, our data support a substrate-limited model where increasing the production of CoA benefi ts animal health in the face of a HS diet.
PA is available over-the-counter as calcium pantothenate in vitamin B5 supplements. In a recent study, pantothenate supplementation promoted CoA-dependent keto genesis and improved liver function in an animal model of nonalcoholic fatty liver disease ( 23 ). We propose that vitamin B5 represents a potential therapy for insulin resistance resulting from overnutrition. Although pantothenate supplementation would be expected to increase adiposity, our work suggests a signifi cant benefi t in terms of metabolic health. PA's low cost and toxicity profi le make it an especially attractive target for future clinical studies.

DISCUSSION
Previous studies have shown a reduced capacity for TG synthesis in obesity that is accompanied by increases in FFAs, ceramides, and DAG, all potential mediators of lipotoxicity. Still, it remains unknown what mechanisms limit the ability of animals to store excess carbons from dietary sugar as TG. We noticed a dramatic upregulation in the expression of CoA synthetic enzymes that prompted us to take a closer look at these steps of the pathway. The CoA pool is known to be limiting for several metabolic processes, including the TCA cycle, ketogenesis, lipogenesis, and mitochondrial fatty acid import and ␤ -oxidation ( 21,26,33 ). Although we did not probe all of these pathways, our data support a model where CoA is limiting in the face of caloric excess, reducing animal fi tness by contributing to metabolic lipotoxicity.
The Drosophila gut may be an important source of pantothenate. The fl y gut is known to harbor commensal bacteria that regulate nutritional status and might help to provide pantothenate, as has been demonstrated in mammals (34)(35)(36). We observed measurable quantities of this nutrient in isolated guts, although no change in pantetheine or pantothenate levels was observed upon HS feeding (data not shown). Increased gut expression of genes predicted to encode the pantetheine hydrolase vanin-like and the pantothenate transporter, CG10444, may represent an attempt of the gut to compensate for inadequate CoA levels and suggests a concerted systemic effort to provide this nutrient to the FB.
One open question is: what metabolites indicate an increased requirement for pantothenate in peripheral tissues? The carnitine-acyl carnitine system is one way in which free CoA pools are maintained in cells ( 37,38 ). Serum acyl-carnitine concentrations refl ect an excess of intracellular acyl groups, increasing when fatty acid oxidation is defective in the presence of increased FFAs ( 39,40 ). It follows that these acyl-carnitines might accumulate when metabolic fl ux is reduced during insulin resistance. Increased long-chain carnitine esters have been observed in the serum, liver, muscle, and urine of individuals with obesity and T2D (41)(42)(43)(44)(45)(46), although reduced levels of long-chain acyl-carnitines have also been associated with metabolic syndrome and T2D ( 47 ). Rodent models of obesity and T2D also accumulate acyl-carnitines (48)(49)(50). In Drosophila , acyl-carnitines decline with age, along with obesity ( 51,52 ). Perhaps circulating acyl-carnitines signal a demand for CoA to enable proper fatty acid esterifi cation into TG in the FB and adipose. Our data support a model where CoA bioavailability enables metabolic fl exibility and channeling of the endocrine fatty acid pool.
Another potential rate-limiting substrate for CoA synthesis in the face of caloric overload is cysteine ( Fig. 1A ), although our data suggest that cysteine is not limiting in the context of caloric overload. Cysteine supplementation alone slightly reduced fi tness on HS diets and did not rescue HS phenotypes (data not shown). Metabolite analysis showed that cysteine levels were slightly elevated in HS-fed FBs compared with controls. Further increasing cysteine levels could adversely affect redox status in the FB, impairing cellular processes and masking any benefi t to lipogenesis.