The Statin Target Hmgcr Regulates Energy Metabolism and Food Intake through Central Mechanisms

The statin drug target, 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR), is strongly linked to body mass index (BMI), yet how HMGCR influences BMI is not understood. In mammals, studies of peripheral HMGCR have not clearly identified a role in BMI maintenance and, despite considerable central nervous system expression, a function for central HMGCR has not been determined. Similar to mammals, Hmgcr is highly expressed in the Drosophila melanogaster brain. Therefore, genetic and pharmacological studies were performed to identify how central Hmgcr regulates Drosophila energy metabolism and feeding behavior. We found that inhibiting Hmgcr, in insulin-producing cells of the Drosophila pars intercerebralis (PI), the fly hypothalamic equivalent, significantly reduces the expression of insulin-like peptides, severely decreasing insulin signaling. In fact, reducing Hmgcr expression throughout development causes decreased body size, increased lipid storage, hyperglycemia, and hyperphagia. Furthermore, the Hmgcr induced hyperphagia phenotype requires a conserved insulin-regulated α-glucosidase, target of brain insulin (tobi). In rats and mice, acute inhibition of hypothalamic Hmgcr activity stimulates food intake. This study presents evidence of how central Hmgcr regulation of metabolism and food intake could influence BMI.


Introduction
Obesity has become an international problem, leading the World Health Organization (WHO) to declare it a global pandemic [1]. With its associated health problems, including type 2 diabetes, cardiovascular diseases, and some forms of cancer, obesity is correlated with decreased life expectancy. Even though obesity is associated with serious diseases, as a noncommunicable disease (NCD), it has not received the attention of other rapidly spreading infectious diseases. Although lifestyle influences, such as increased food availability and intake, along with decreased physical activity, have significantly increased the frequency of obesity, genome-wide association (GWA) studies suggest a substantial genetic contribution. Recent estimates indicate that heritable contributions may account for as much as 60% of rat procedures were approved by the institutional ethics committee at the University of Waikato. Procedures  Fly stocks were maintained at 25 • C with 60% relative humidity in a 12:12 h light:dark cycle. Unless otherwise stated, Drosophila stocks were maintained with Jazz-Mix Drosophila food (Thermo-Fisher Scientific, Göteborg, Sweden) supplemented with yeast extract (58 g/dL sugar:12 g/dL protein, referred to as high-sugar food). The following strains were used to knockdown Hmgcr: w 1118 , UAS-Hmgcr RNAi1 (P{KK101807}VIE-260B) from the Vienna Drosophila RNAi Centre (VDRC, Vienna, Austria) and w 1118 , UAS-Hmgcr RNAi2 (P{UAS-RNAi-HMGCR}10367-R3) from the National Institute of Genetics stock center (NIG, Mishima, Japan). Likewise, flies were used with the following GAL4 drivers: y[1] v [1];P{GawB}elavC155 w* (referred to as elav-GAL4, containing a neuron-specific promoter); w*; P{GawB}Aug21/CyO (referred to as Aug21-GAL4, containing a corpus allatum-specific promoter), P{w[+mC] = Ilp2-GAL4.R}EQU2 (referred to as Dilp2-GAL4, containing a promoter specific for insulin-producing cells (IPCs)), all from the Bloomington Stock Center (Bloomington, IN [21]; P{y[+t7.7] v[+t1.8] = TRiP.HMC05747}attP40 (UAS-FppsRNAi); and the following strain was used to knockdown Tobi: P{w[+mW.hs] = GawB}48Y and y1 v1; P{y[+t7.7] v[+t1.8] = TRiP.HMJ02101}attP40 (UAS-tobi RNAi ), both from the Bloomington Stock Center (Bloomington, IN, USA). All transgenic strains were initially crossed into the same w 1118 genetic background. Genetic crosses: Virgin female elav-GAL4, Dilp2-GAL4, Aug21-GAL4 or 48Y-GAL4 were crossed with male UAS-RNAi or overexpression lines. Two different controls were used: virgin female w 1118 crossed with male UAS lines, and virgin female GAL4 lines crossed with male w 1118 . The UAS-tobi overexpression line (referred to as tobi OE ) was a gift from Dr. Michael Pankratz [19]. Adult flies were collected immediately after they eclosed and placed at 29 • C for 5-7 days, to ensure maximum expression of the UAS line, before any experiments were performed. Unless otherwise stated, experiments were performed at 25 • C.

RNA Extraction
To obtain equivalent amounts of RNA, RNA was extracted by homogenizing either 25 male fly heads (for quantitative RT-PCR of Hmgcr, Ilp2, Ilp3, and Ilp5) or 10 male fly bodies (for quantitative RT-PCR of Adipokinetic Hormone, Akh) in PBS, all equally aged to between 5-7 days post-eclosion. An equal volume of phenol:choloroform:isoamyl alcohol solution (25:24:1) (Sigma-Aldrich, Malmö, Sweden) was added to the homogenized flies and mixed. The solution was centrifuged for 5 min at 12,000× g at room temperature. The aqueous phase was transferred to a new tube and an equal volume of chloroform was added, followed by centrifugation at the previously mentioned speed and time. An ethanol and silica suspension (1 g/mL) (Sigma, Malmö, Sweden) was added to the aqueous phase and incubated for 1 min before additional centrifugation. The pellet was washed with 70% ethanol and let dry. To remove any DNA, the sample was treated with DNAse (Thermo-Fisher Scientific, Göteborg, Sweden) for 30 min at 37 • C and subsequently, at 65 • C for 15 min. The pellet was resolved in 50 µL DEPC-H2O and incubated for 5 min. To remove the silica suspension, the sample was centrifuged, and the RNA solution was transferred to a new tube. A spectrophotometer (model ND-1000, Nanodrop) was used to measure total RNA concentration.

cDNA Synthesis
The High-capacity RNA-to-cDNA Kit (Applied Biosystems, Stockholm, Sweden) was used for cDNA synthesis and performed according to the manufacturer's instructions.

Library Preparation and Sequencing
RNA-seq reads for the entire transcriptome were obtained using SOLiD 5500xl pairedend sequencing from Life Technologies. Initial quality analysis was performed using a proprietary 'XSQ Toolkit' provided by Life Technologies. Further analysis was performed using the 'Tuxedo suit' [21,22], mainly composed of three tools: TopHat, Cufflinks, and CummRbund. Reads were then aligned to the D. melanogaster reference genome (build dmel_r5.47_FB2012_05) obtained from flybase using TopHat with the pre-built bowtie index downloaded from the TopHat home page (http://ccb.jhu.edu/software/tophat/ index.shtml, accessed on 10 January 2022). Transcript assembly and abundance estimation was estimated using Cufflinks v2.0.2. Subsequently, differential expression tests were performed using cuffcompare and cuffdiff. The calculated p and q values (the FDRadjusted p value of the test statistic) from cuffdiff were used to determine the significance of differential expression.

Western Blot Assay
Five heads of male flies with 5 to 7 days of age were homogenized in 2X Laemmli sample buffer (Bio-Rad, Hercules, CA, USA) and equal amounts of homogenate were loaded onto SDS-PAGE gels and blotted according to standard protocol. Blots were probed with the following antibodies (all at 1:2000 dilution): (1) phospho-Drosophila Akt (Ser505) Antibody (Cell Signaling #4054), (2) Akt (pan) (C67E7) Rabbit mAb (Cell Signaling #4691), (3) E7 anti-beta-tubulin (Developmental Studies Hybridoma Bank). At least four experiments were performed, and multiple exposures were taken; Western blot quantification was done on lightly exposed blots using Licor Image analysis software (LI-COR Biosciences-GmbH). Blots were stripped and re-probed using the mild stripping protocol from Abcam (http://www.abcam.com/ps/pdf/protocols/stripping%20for%20reprobing.pdf, accessed on 10 August 2021). Primary antibodies were used in the following order-P-Akt, total Akt, β-tubulin. Secondary antibodies conjugated to HRP (Genescript) were used, and the signals were detected by chemiluminescence using the Enhanced ECL kit (Biorad).

Oxidative Stress Assay
Resistance to oxidative stress was performed according to [23]. In short, 5-7-dayold male flies previously maintained on either a high-sugar diet (58 g/dL sugar:12 g/dL yeast) or a low-sugar diet (10 g/dL sugar:10 g/dL yeast) were transferred into glass vials containing 1% agar, 5% sugar and 20 mM paraquat (Sigma, St. Louis, MO, USA) [23]. Dead flies were counted twice a day. For both assays, a total of 50 flies were used for each genotype and parameter (10 flies per vial). Survival differences were analyzed by proportional hazard analysis.

CAFE Assay
A vial, 9 cm by 2 cm (height by diameter), containing 1% agarose (5 cm high) (Invitrogen, Göteborg, Sweden) to provide moisture and humidity for the flies, was used for this assay [24]. A calibrated capillary glass tube (5 µL, VWR International) was filled with liquid food which contains 5% sucrose (Sigma, Sweden), 5% yeast extract (Sigma, Sweden), and 0.5% food-coloring dye. A layer of mineral oil was used to prevent the liquid food from evaporating. Five males, which were 5-7 days old, were put inside the chamber and the opening of the vial was covered with paraffin tape, with a capillary tube being inserted from the top through the tape. The experimental setup was kept at 25 • C, 50% humidity on a 12:12 h light:dark cycle. At least 10 replicates were performed for each genotype. All male flies were equally aged to 5-7 days post-eclosion.

Triglyceride Assay
The 5-7 day old flies (25 males) were homogenized with 100 µL of PBST buffer (1X PBS with 10% Tween 20), incubated at 70 • C for 5 min, and then centrifuged at maximum speed for 10 min. The supernatant was transferred into a new microcentrifuge tube and used as samples. The glycerol standard (Sigma, Stockholm, Sweden) was used to generate a standard curve with concentrations of 1.0, 0.8, 0.6, 0.4, and 0.2 mg/mL equivalent triolein concentration. One hundred µL of free glycerol reagent was added with 10 µL of blank (PBST), standards or samples, and initial absorbance at 540 nm was measured after incubation at 37 • C for 15 min. The concentration of free glycerol in the samples was calculated from a standard curve generated by these initial absorbance values. Then, 20 µL of triglyceride reagent was added to each standard and samples and incubated at 37 • C for 15 min. The final absorbance measurement was taken at 540 nm to calculate triglyceride concentrations from the generated standard curve. The protein concentration of each sample was measured with the Bio-Rad Protein Assay Kit (Bio-Rad, Hercules, CA, USA). The concentrations of free glycerides and triglycerides in samples (mg per mg of protein) were calculated from 10 replicates.

Trehalose and Glucose Assays
Four substrates were measured: circulating trehalose, stored trehalose, circulating glucose, and glycogen. All extracted substrates were converted to a glucose solution for final analysis via spectrophotometry. Male flies, aged 5-7 days post-eclosion, were collected and starved for either 0, 12, or 24 h in 1% agarose (Invitrogen, Göteborg, Sweden) vials and then frozen at −80 • C overnight. To collect hemolymph, 10 flies per replicate were weighed using a 1/10,000 scale (Denver instrument company, Goettingen, Germany), placed in PBS (pH 7.4) in a 1:5 ratio (mg of flies/µL of PBS) and decapitated via centrifugation (at 3000× g for 6 min at 4 • C). Hemolymph was used to measure circulating glucose and trehalose. To determine stored trehalose and glycogen, the remaining bodies of the 10 flies were homogenized in a 1:10 ratio of PBS (mg of flies/µL of PBS), homogenates were centrifuged at 12,000× g for 15 min at 4 • C, and supernatant was collected for analysis. Trehalose from the hemolymph or supernatant was converted to glucose using porcine kidney trehalase (Sigma, T8778) overnight at 37 • C. Glycogen from the supernatant was converted to glucose by amyloglycosidase from Aspergillus niger (Sigma, A7095) overnight at 25 • C. Lastly, glucose levels from all substrates were quantified using the Glucose Assay Kit (Liquick Cor-Glucose Diagnostic kit, Cormay, Siedlce, Poland) involving glucose oxidase and peroxidase according to the manufacturer's instructions. Briefly, glucose is oxidized to form gluconic acid and hydrogen peroxide by glucose oxidase. The hydrogen peroxide then reacts with 4-aminoantipyrine in the presence of peroxidase to form a colored solution where glucose concentration is proportional to the absorbance of light. Absorbance at 492 nm was measured for each replicate of each substrate on a multi-scan microplate spectrophotometer (model) and converted to a mM concentration of glucose using a linear regression obtained by a calibration curve made from a serial dilution of a sample with a known glucose concentration. Glucose measurements were then converted back to the units of their original substrates by fill-in-the-blank.

Subject and Housing Conditions
Adult male C57BL/6 mice (Taconic, Silkeborg, Denmark) weighing approximately 25-30 g at the beginning of the experiment were individually housed in type III standard Macrolon cages in a controlled environment in terms of temperature (20-22 • C) and air humidity (55%) with a 12:12 h light-dark cycle. Water and standard chow (R04, Safe, Augy, France) were available ad libitum at all times unless otherwise stated.

Hmgcr Expression in the Mouse Brain In Situ Hybridization
Design and synthesis of RNA probes: Antisense and sense probes were generated from commercial mouse cDNA clones (Source BioScience). The clones were sequenced (Eurofins MWG Operon, Ebersberg, Germany) and verified to be correct. Plasmids were linearized with appropriate restriction enzymes and the probes were synthesized using a 1 µg vector as a template with T7, Sp6, or T3 RNA polymerase in the presence of digoxigenin (DIG)labeled 11-UTP (Roche Diagnostics, Rotkreuz, Switzerland). Probes were controlled and quantified using the Nanodrop ND-1000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).
In situ hybridization on free-floating sections: Free-floating sections were washed in PBS followed by 6% hydrogen peroxide treatment at room temperature. After successive washing in PBS, the sections were treated with 20 µg/mL proteinase K (Invitrogen). The sections were post fixed in 4% formaldehyde (Thermo-Fisher Scientific, Sweden) before pre-incubation in hybridization buffer (50% formamide, 5X SCC pH 4.5, 1% SDS, 50 µg/mL tRNA) (Sigma Aldrich, Stockholm, Sweden), 50 µg/mL heparin (Sigma Aldrich, Stockholm Sweden) in PBS. Hybridization of sections in presence of 100 ng antisense probe/mL was performed overnight at 58 • C. As a control for in situ hybridization, the sense probe was used. To remove the unbound probe, the sections were washed with buffer 1 (50% formamide, 2X SSC pH 4.5, and 0.1% Tween-20 in PBS) followed by buffer 2 (50% formamide, 0.2X SSC pH 4.5, and 0.1% Tween-20 in PBS). The sections were incubated in blocking solution (1% blocking reagent) (Roche Diagnostics, Stockholm, Sweden) followed by overnight incubation combined with 1:5000 diluted anti-digoxigenin alkaline phosphates conjugated antibody (Roche Diagnostics Scandinavia, Stockholm, Sweden). Unbound antibody was washed away with sequential washes in 0.1% TBST. The sections were then developed with Fast Red (Roche diagnostics, Stockholm, Sweden).

Hmgcr Expression in Mouse Brain under Different Feeding Conditions
Feeding experiments: Two feeding experiments were performed in order to determine Hmgcr expression at baseline, during starvation, and in obese mice (n = 8 per group). Experiment 1: Hmgcr mRNA expression at baseline and after 24 h of starvation. Food was removed prior to the onset of the dark phase and the mice were decapitated on the next day. Control mice were sacrificed at the same time but had ad libitum access to food. Experiment 2: Hmgcr mRNA expression in obese mice. An additional cohort of male mice was maintained on either regular chow or a high-fat diet in addition to the regular chow for 8 weeks. Bodyweight measured at the start of the experiments did not differ between the two groups, but the mice on the high-fat diet had significantly increased in body weight compared to the controls after the 8 weeks.
RNA isolation and cDNA synthesis: After 8 weeks, the male mice were sacrificed by cervical dislocation within 2 hours of the early dark phase, and the brains were removed and dissected within 10 min. After dissection, the tissue was immersed in RNA-later solution (Ambion, Elk Grove, CA, USA) and kept at room temperature for 1 h. Individual tissue samples were homogenized by Bullet Blender (Next Advance, New York, NY, USA) in RNA-later solution. mRNA was extracted from tissue using Absolutely RNA Miniprep Kit (Agilent Technologies, Santa Clara, CA, USA) by following the included protocol. RNA concentration was determined using a NanoDrop ND-1000 Spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA). cDNA was synthesized using the First Strand cDNA Synthesis Kit (Fisher Scientific, Göteborg, Sweden) with random hexamers (Fisher Scientific, Sweden) as primers according to manufacturer's instructions.

Hmgcr Inactivation in the Mouse Brain
Hypothalamic cannula implantation: Male mice were anesthetized in an induction chamber using isoflurane (Baxter, Deerfield, IL, USA), first, at 4% to induce anesthesia, followed by 1.5% for maintenance, and then restrained onto a stereotaxic frame (Stoelting, Wood Dale, IL, USA). The animals subsequently received a subcutaneous injection of Carprofen (2 mg/kg body weight, Astra Zeneca, Södertälje, Sweden) to prevent postsurgical pain and inflammation. The hair was shaved off the heads of the mice, and the skin was sterilized. An incision was made to expose the skull and local anesthesia (Marcain, 5 mg/kg, Apoteket, Stockholm, Sweden) was added directly onto the surgical cut. The skull was wiped clean with sterile cotton swabs, followed by the application of hydrogen peroxide to enhance the visibility of bregma for correct coordinate determination. Target coordinates used were anterior-posterior −1.48 mm, laterally 0.5 mm, and dorsal-ventral −5.5 mm with respect to bregma. A steel guide cannula (26-gauge, 5.5 mm, World Precision Instruments, Friedberg, Germany) was lowered unilaterally by a micromanipulator (Stoelting, Wood Dale, IL, USA) through the cerebral cortex into the ventromedial hypothalamic nucleus. The guide cannula was secured to the skull using dental cement (GC, Tokyo, Japan). A dummy cannula (26-gauge, 6.0 mm, World Precision Instruments, Friedberg, Germany) was inserted into the guide cannula to prevent clogging and entrance of unfamiliar particles. After surgery, the animals were returned to their home cages and allowed to recover for at least one week prior to further experiments.
Hypothalamic injection of simvastatin: A 5 µL Hamilton syringe (World Precision Instruments, Friedberg, Germany) was attached to an internal cannula (26-gauge, 6 mm, World Precision Instruments, Friedberg, Germany) through polyethylene tubing. The internal cannula was inserted into the guide cannula extending 0.5 mm below the guide cannula tip. Infusions were made 1 h prior to the onset of the dark phase. The mice were mildly anesthetized through isoflurane inhalation and 0.5 µL of simvastatin (Sigma Aldrich, Malmö, Sweden) dissolved in 5% dimethyl sulfoxide (DMSO, Sigma Aldrich) or vehicle (5% DMSO) were infused manually at an approximate speed of 1 µL/min. After the injection, the needle was left in place for 30 s before removal of the internal cannula. Afterward, the mice were returned to their home cages.
Effects of simvastatin on nocturnal food intake in ad libitum-fed mice: One week before the experiment, the mice were habituated to eating food from Petri dishes. On the day of the experiment, the mice were injected with either simvastatin or vehicle 1 h prior to the onset of the dark phase. A Petri dish with an excess amount of food (8 g of standard chow) was then presented, and the mice's food intake was measured manually at 1, 2, 3, 6, 12, and 24 h after injection.
Verification of injection sites: After the completion of the experiments, the mice were sacrificed via transcardial perfusion through the left ventricle with PBS followed by 4% formaldehyde (Histolab, Gothenburg, Sweden) and post-fixed at 4 • C overnight. Methyl blue (Sigma Aldrich) was injected into the guide cannula to confirm correct placement ( Figure S1). The brains were cut into 200 µm sections using a Leica VT1000S microtome, and the injection site was checked by visual inspection based on Franklin and Paxinos 2007 [25] and Allen Mouse Brain Atlas 10 [26].

Rat Studies 2.4.1. Subject, Housing Conditions and Cannula Implantation
Male Sprague-Dawley rats (AgResearch, Hamilton, New Zealand) weighing 320-331 g at the beginning of the studies were single-housed in Plexiglas cages in a temperaturecontrolled facility (21 • C) with light-dark 12:12, turning on the light at 8 am. Water and standard laboratory chow (Sharpes Feed, Carterton, New Zealand) were available ad libitum unless stated otherwise. Animals were anesthetized with ketamine (100 mg/kg) and xylazine (20 mg/kg) and surgically equipped with a 26-gauge cannula (Plastics One) aimed at the lateral cerebral ventricle. Stereotaxic coordinates were as follows: 1.5 mm lateral to the midline, 1.0 mm posterior to bregma, and 3 mm below the surface of the skull. Dental acrylic was applied to secure the cannula to two screws inserted in the skull. The injector extended 1 mm beyond the tip of the guide cannula. The placement was verified after a 10-day recovery period and again, following the completion of the feeding experiments, via intracerebroventricular (ICV) injections of 100 ng angiotensin II (rats that drank less than 7 g of water post injection were excluded) [27]. Simvastatin was acquired from Sigma (Christchurch, New Zealand), and it was dissolved in DMSO just before each experimental trial Intracranial injections were performed using Hamilton syringes in a volume of 2 µL. Injection was delivered within 5 s and the injector remained in place in the guide cannula for an additional 15 s to permit diffusion of the drug from the cannula tip.

Hmgcr Inactivation in the Rat Brain
Experiment 1: Effect of ICV simvastatin on standard chow intake in overnightdeprived rats. Animals deprived of food overnight were injected at 10:00 with vehicle, 10, 30, or 100 nmol simvastatin ICV (n = 6-7/group), and chow was returned to hoppers just after the drug administration. Food intake was measured 1, 2, and 4 h post-injection. Experiment 2: Effect of ICV simvastatin on overnight standard chow intake in ad libitum-fed rats. Ad libitum-fed animals were injected ICV at 19:45 (15 min before the lights off) with vehicle or 30 nmol simvastatin (30 nmol was established to be the lowest orexigenic dose in Experiment 1; n = 7/group). Food present in the hoppers was immediately exchanged for pre-weighed pellets and food intake was measured 1, 2, 4, 6, and 12 h post-injection. Experiment 3: Effect of ICV simvastatin on episodic intake of palatable solutions. Our protocol was based on previous studies testing the effects of injectants on the short-term intake of palatable solutions in a no-choice scenario [23][24][25]. Animals were accustomed to receiving one of the palatable solutions for 2 h/day on 2 days (10:00-12:00) prior to the injection experiments to avoid neophobia: 10% sucrose (Sigma), 0.1% saccharin (Sigma), or milk (DGC, Hamilton, New Zealand). On the day of the experiment, animals were injected ICV with vehicle or 30 nmol simvastatin (n = 6-7/group) just before the presentation of the solution. Intake was measured 2 h post injection. Chow and water were removed during the 2-h palatable tastant exposure.

Statistical Analysis
Shapiro-Wilk or Kolmogorov-Smirnov normality tests were performed for all data according to the number and type of samples. For data with a normal distribution, one-way ANOVA was performed, while for data with a non-parametric distribution, the Kruskal-Wallis test was used, and in both cases, post hoc tests were used as appropriate for each case (see figure legends for details on the individual experiments). Mean and standard error from all replicates of each experiment were calculated. All analyses were performed with GraphPad Prism 4. Data with a p < 0.05 were considered significant.

Central Hmgcr Regulates Insulin Expression in Drosophila
To begin, we verified that Hmgcr was expressed in the insulin-producing cells (IPCs) ( Figure 1A). Starving 5-7 day old male flies for 24 h reduced Hmgcr transcription to nearly undetectable levels within the IPCs ( Figure 1B). Using quantitative RT-PCR, we validated that starvation significantly reduced Hmgcr head expression in males ( Figure 1C), whereas maintaining the flies on diets containing various concentrations of macronutrients had no significant effect ( Figure 1D).
Next, we wanted to determine if insulin signaling requires Hmgcr activity within the insulin-producing cells (IPCs) of the PI [14]. Since insulin is known to control Drosophila body size during development [28], we knocked down Hmgcr expression specifically within the IPCs (Dilp2-GAL4 > UAS-Hmgcr RNAi ) throughout larval development to determine if there was an effect on body size. Notably, raising Dilp2-GAL4 > UAS-Hmgcr RNAi IPC knockdown males (from now on referred to as Hmgcr males, Figure S2A) at 25 • C was sufficient to produce flies that were significantly lighter than controls ( Figures 1E and S2B). Raising Hmgcr males at 18 • C, which significantly reduces GAL4 activity [29], was sufficient to rescue the phenotype ( Figure S2B). Consequently, we analyzed insulin-like peptide (Ilp2, Ilp3, Ilp5) and glucagon-like (Adipokinetic hormone, Akh) gene expression in adult males. Ilp2, Ilp3, and Akh transcript levels were significantly increased in Hmgcr males fed normal lab fly food, which constitutes a high-sugar diet (58 g/dL sugar:12 g/dL protein) ( Figure 1F). On the other hand, maintaining Hmgcr males on a low-sugar diet (10 g/dL sugar:10 g/dL protein) significantly decreased the transcript levels of Ilp2 and Ilp3 compared to controls, while Akh transcript levels were normal ( Figure 1G). Maintaining wild-type males for 24 h on a high-sugar diet containing the Hmgcr inhibitor fluvastatin was sufficient to increase Akh transcript levels ( Figure 1H), while 5 days of fluvastatin treatment significantly induced the transcript levels of Ilp2, Ilp3, and Akh ( Figure 1I). We also demonstrated that Thor (Drosophila 4E-BP) expression was induced in fluvastatin fed flies ( Figure 1J), representative of reduced insulin signaling [30].
Next, we examined the protein levels of ILP2 and ILP3 in the IPCs, comparing controls and Hmgcr males fed either a low-sugar or high-sugar diet. Not surprisingly, in controls, abundant ILP2 expression was observed within the IPCs when flies were fed a highsugar diet (Figure 2A,B). On the other hand, much less ILP2 protein was visible in the IPCs from control flies maintained on a low-sugar diet (Figure 2A,B). In Hmgcr males, fed either a high-sugar or low-sugar diet, DILP2 expression was only observed in a few IPCs (Figure 2A,B). A second Hmgcr RNAi line showed a similar, although less severe, reduction in DILP2 expression ( Figure S3A,C). Apparently, in high-sugar diets, HMGCR knockdown flies increase transcripts of insulin-like peptides ( Figure 1F), but for some unknown reason, their translation does not occur (Figure 2A). Interestingly, knocking down Farnesyl pyrophosphate synthase (Fpps), which is downstream of Hmgcr in the mevalonate pathway, also significantly reduced the expression of DILP2 in the IPCs ( Figure S3B,D). In controls, there was no difference in DILP3 expression when comparing flies fed either a high-sugar or low-sugar diet; yet in Hmgcr males, no DILP3 protein was observed under either condition ( Figure 2C,D). To detect the significant difference between groups a Shapiro-Wilk test was performed to determine normality, then a one-way ANOVA with Tukey's post hoc test for multiple comparisons was performed, *** p < 0.005). (E) Equally aged controls (w 1118 > Hmgcr RNAi , Dilp2-GAL4 > w 1118 ) or where Hgmcr is knocked down in the pars intercerebralis IPCs (Dilp2-GAL4 > Hmgcr RNAi ), showing that the Hmgcr males are smaller than equally aged control flies. This experiment was repeated five times, with each replicate consisting of 10 males (5-7 days old) per genotype. (F) Equally aged controls and Hmgcr males maintained on a high-sugar diet (58 g/dL sugar:12 g/dL protein) were collected and processed for qPCR. (G) Equally aged controls and Hmgcr males, 5-7 days old, maintained on a lowsugar diet (10 g/dL sugar:10 g/dL protein) were collected and processed for qPCR. (H) Wild-type male flies were fed 0.5 mM fluvastatin for 24 h before being processed for qPCR. (I) Wild-type male flies were fed 0.5 mM fluvastatin for 5 days before being processed for qPCR (F-I): n = 10 replicates per genotype, with 25 fly heads per replicate for Ilp2, Ilp3, and Ilp5, or 10 bodies per replicate for Akh. To detect the significant difference between groups, a Shapiro-Wilk test was performed to determine normality, then a one-way ANOVA with Tukey's post hoc test for multiple comparisons was performed, * p < 0.05, ** p < 0.01). (J) Wild-type male flies were fed 0.5 mM fluvastatin for 24 h before being processed for qPCR (n = 10 replicates per genotype, with 10 bodies per replicate for Thor). In all graphs, error bars = SEM. In all experiments n = 10 male fly heads per genotype per parameter, 5-7 days old. To detect the significant difference between groups a Shapiro-Wilk test was performed to determine normality, then a one-way ANOVA with Tukey's post hoc test for multiple comparisons was performed, *** p < 0.005. In both graphs, error bars = SEM.

Central Hmgcr Regulates Insulin Signaling in Drosophila
To clarify if the loss of IPC Hmgcr has a direct effect on insulin signaling, we performed Western blot analysis to examine phospho-AKT (pAKT), a key molecule in the insulin signaling pathway, in the heads of flies fed a high-sugar diet [31]. Compared to controls, pAKT levels were significantly reduced in Hmgcr males fed a high-sugar diet ( Figure 3A,B), indicating a reduction in insulin signaling. It was reported that reduced insulin signaling protects flies against the effects of reactive oxygen species (ROS) [23]. Therefore, we measured the resistance to paraquat, which increases cellular ROS levels, using flies maintained on the two different diets. Hmgcr males maintained on a high-sugar diet survived on paraquat-containing food significantly longer than controls ( Figure 3C), while those maintained on a low-sugar diet were not significantly different from controls ( Figure 3D). raised on either a (C) high-sugar (58 g/dL sugar:12 g/dL protein) or (D) low-sugar diet (10 g/dL sugar:10 g/dL protein), aged 5-7 days posteclosion, were maintained on food containing 20 mM paraquat to determine their survival rate (n = 10 flies per genotype per replicate, with five replicates per experiment, the experiment was repeated five times).

Central Hmgcr Regulates Triglyceride and Carbohydrate Levels in Drosophila
Considering that insulin signaling regulates triglyceride and carbohydrate levels in Drosophila [32], we measured their concentrations before and during starvation in adult flies maintained on either a high-sugar or low-sugar diet. Hmgcr males maintained on a highsugar diet or fed a high-sugar diet and then starved for 12 h, had significantly increased triglyceride concentrations compared to controls ( Figure 4A). On the other hand, Hmgcr males maintained on a low-sugar diet had significantly lower triglyceride concentrations than controls when fed ad libitum ( Figure 4A). Similarly, Hmgcr males fed a high-sugar diet had elevated circulating glucose concentrations, which remained high after 12 h of starvation ( Figure 4B). Again, feeding Hmgcr males a low-sugar diet rescued the phenotype ( Figure 4B). Moreover, circulating trehalose concentrations were elevated prior to starvation in Hmgcr males fed a high-sugar diet, but not in Hmgcr males maintained on a low-sugar diet ( Figure 4C). Glycogen concentrations were not significantly different from controls on either diet ( Figure 4D).

Central Hmgcr Regulates Food Intake in Drosophila
Previously, it was determined that insulin signaling in Drosophila regulates feeding behavior [33,34]; therefore, we assessed food intake in Hmgcr males. Interestingly, Hmgcr males maintained on a high-sugar diet were hyperphagic ( Figure 5A), while Hmgcr males fed a low-sugar diet showed normal food consumption ( Figure 5A). In support of these results, systemic inhibition of Hmgcr, by feeding wild-type flies a high-sugar diet containing fluvastatin, also induced hyperphagia ( Figure 5B); while wild-type flies maintained on a low-sugar diet containing fluvastatin ate normally ( Figure 5B). Of note, loss of Hmgcr expression in the corpus allatum, a peripheral endocrine gland where the Hmgcr enzyme is highly active [35,36] and shown to be regulated by insulin [14], had no effect on food intake ( Figure 5C).
Hmgcr males fed a high-sugar diet have a reduced insulin response; to substantiate this, we examined the expression of an insulin regulated α-glucosidase gene, known as target of brain insulin (tobi), whose expression was shown to be inhibited by increasing sugar levels [19]. When Hmgcr males were maintained on a high-sugar diet tobi expression was reduced compared to similarly feed controls ( Figure 5D). On the other hand, feeding control flies a low-sugar diet increased tobi expression and in Hmgcr males fed a low-sugar diet tobi expression was similar to controls ( Figure 5D). When maintained on a high-sugar diet, flies where tobi was knocked down in the midgut (48Y-GAL4 > tobi RNAi ) were hyperphagic and inhibiting Hmgcr activity with fluvastatin did not increase total food intake ( Figure 5E). Yet, overexpressing tobi (48Y-GAL4 > tobi OE ) in the midgut significantly inhibited food intake, and feeding these flies fluvastatin increased intake to normal levels ( Figure 5F). Figure 5. IPC Hmgcr regulates feeding in adult males (A) An intake assay [19] was used to assess total food intake in equally aged 5-7 day old adult male flies, fed either a high-sugar or low-sugar diet over a 24 h period. Different letters indicate similar groups (i.e., 'a' is significantly different than 'b' or 'c' and so on. One-way ANOVA with Tukey's post hoc test for multiple comparisons, p < 0.05). (B) Adult male flies, 5-7 days old, maintained on either a high-sugar diet, were fed fluvastatincontaining food for 24 h and total food intake was measured every 12 h (A,B: n = 10 replicates with 5 males per replicate. To detect the significant difference between groups, a Shapiro-Wilk test was performed to determine normality, then a one-way ANOVA with Tukey's post hoc test for multiple comparisons was performed, * p < 0.05, ** p < 0.01). (C) A CAFE assay was used to assess total food intake in flies fed a high-sugar diet over a 24 h period in adult males where Hmgcr is specifically knocked down in the corpus allatum (Aug21-GAL4 > Hmgcr RNAi ) (As per standard protocols, n = 10 replicates with 5 males per replicate, per genotype. To detect the significant difference between groups a Shapiro-Wilk test was performed to determine normality, then a one-way ANOVA with Tukey's post hoc test for multiple comparisons was performed). (D) Controls and Hmgcr males were maintained on either a high-sugar or low-sugar diet before being collected and processed for qPCR to determine tobi expression levels (n = 10 replicates per genotype, 10 whole bodies per sample for tobi. To detect the significant difference between groups, a Shapiro-Wilk test was performed to determine normality, then a one-way ANOVA with Tukey's post hoc test for multiple comparisons was performed. *** p < 0.001). (E,F) Tobi was either € knocked down or (F) overexpressed in gut endoderm, equally aged 5-7 day old male flies were raised on a high-sugar diet with or without the Hmgcr antagonist fluvastatintin. (E) Knocking down tobi increased total food intake, this was not affected by fluvastatin, while (F) overexpressing tobi inhibited total food intake, this was rescued by feeding the files fluvastatin. (In (E,F), where present, different letters indicate a similar group. n = 10 replicates with 5 males per replicate. To detect the significant difference between groups a Shapiro-Wilk test was performed to determine normality, then a one-way ANOVA with Tukey's post hoc test for multiple comparisons was performed, * p < 0.05, ** p < 0.01). In all figures, error bars = SEM.
To learn more about how Hmgcr regulates food intake, RNA-seq analysis was performed to identify changes in the expression of genes that regulate food intake in Hmgcr males. Our analysis revealed that only three genes, CrzR, CG10477, and Akh, had a significant change in Hmgcr males, all of them decreasing their expression (Table 1).

Downregulation of CNS Hmgcr Leads to Increased Food Consumption in Rodents
To understand if the novel finding that IPC Hmgcr activity regulates food intake in flies is conserved in mammals, we carried out different experiments in rodents (mice and rats). In mammals, it is known that peripheral insulin signaling to the hypothalamus regulates feeding behavior [32,37]. Thus, we first performed RNA in situ hybridization on adult mouse brains to map Hmgcr expression. Hmgcr was prominently expressed in all major brain areas, including the cortex, amygdala, hippocampus, and hypothalamus ( Figure 6A-L, Table 2). Within the hypothalamus, Hmgcr was especially abundant in areas known to regulate food intake, such as the ventromedial nucleus of the hypothalamus (VMH) ( Figure 6G) and the arcuate nucleus (ARC) ( Figure 6K).
Next, we sought to downregulate central Hmgcr activity by injecting simvastatin into the mouse hypothalamus (see Supplemental Figure S1C for injection sites), then measuring its acute effect on feeding behavior. A single injection of 150 nmol simvastatin into the mouse hypothalamus had a moderate and transient effect on ad libitum food intake, which appeared 3 h after injection ( Figure 6M). Similar to flies, administration of simvastatin peripherally had no effect on feeding behavior in mice ( Figure 6N). In addition, consistent with the fly data, Hmgcr expression in the mouse hypothalamus was significantly decreased after 24 h of food deprivation ( Figure 6O).
Next, using rats, we looked further into the effect of downregulating hypothalamic Hmgcr activity on food intake. ICV administration of simvastatin at 30 and 100 nmol significantly increased standard chow intake following overnight food deprivation ( Figure 7A). Similar to mice, a moderate and transient increase in food consumption also occurred at 3 h post-injection in ad libitum-fed rats ( Figure 7B). On the other hand, ICV simvastatin had no effect on food palatability-driven intake. We gave non-deprived ani-mals episodic access to palatable 0.1% saccharin, 10% sucrose or milk solutions (i.e., a non-carbohydrate sweet tastant, a sweet carbohydrate and a nutritionally complex palatable tastant-all of them being either devoid of energy or energy-dilute) and saw no effect of simvastatin on consumption stimulated by pleasant taste ( Figure 7C).  [25] and Allen Mouse Brain Atlas [26]. Black scale bar, 1 mm. (M) Various concentrations of simvastatin were injected directly into the hypothalamus of mice, after which, the total amount of food consumed was measured (n = 10 mice per concentration, one-way ANOVA with Bonferroni post hoc test for multiple comparisons, * p < 0.05). (N) Various concentrations of simvastatin were injected peripherally into the peritoneal of mice, after which the total amount of food consumed was measured (n = 10 mice per concentration, one-way ANOVA with Bonferroni post hoc test for multiple comparisons, * p < 0.05). (O) Relative level of Hmgcr transcript in the hypothalamus from starved male mice (n = 10 qPCR runs; one-way ANOVA with Bonferroni post hoc test for multiple comparisons, * p < 0.05). In all graphs, error bars = SEM. Table 2. Graded levels of Hmgcr mRNA expression in the mouse brain. A sign − means that no expression of Hmgcr mRNA was detected by RNA in situ hybridization. The signs +, ++ and +++ mean significant Hmgcr mRNA detection with respect to the areas without detection with p < 0.05, p < 0.01, p < 0.001, respectively.

Discussion
Our findings allow us to present a possible model for how HMGCR links to BMI maintenance, as well as how statins could interfere with this maintenance. Using the model system Drosophila melanogaster, we demonstrate that central Hmgcr activity, via the mevalonate pathway, regulates insulin signaling, leading to increased lipid storage, hyperglycemia, and hyperphagia and that this regulation is dependent on carbohydrate consumption. The hyperphagia phenotype was recapitulated in rodents fed a normal diet, where statin inhibition of Hmgcr activity in the hypothalamus led to increased neuronal activity in regions known to regulate food intake. These results provide a strong argument for continued studies of the influence of central HMGCR activity on energy metabolism and food intake as a mechanism for its involvement in the BMI maintenance.
The insulin-glucagon system is highly conserved between flies and mammals [28,30,31,34]. In flies, the insulin system is responsible for regulating energy metabolism and feeding behavior through the insulin-producing cells (IPCs) located in the brain, central for maintaining energy homeostasis [38,39]. We demonstrate that knocking down Hmgcr expression in the central IPCs results in increased circulating glucose and increased triglyceride levels, as well as decreased insulin signaling, similar to an insulin resistance-like state in mammals. All the phenotypes resulting from suppression of Hmgcr expression in the IPCs (e.g., hyperphagia, hypoglycemia, and increased lipid levels) were induced by a high-sugar diet and returned to the normal state when flies were maintained on low-sugar food (equal parts carbohydrate and yeast). In flies, insulin is known to regulate feeding through its interactions with octopaminergic, dopaminergic, and Neuropeptide F (NPY in humans) signaling [40,41]. In fact, loss of the Insulin-like receptor (InR) in NPF neurons was sufficient to increase food intake when flies were fed a normal lab high-sugar diet [41]. In that study, the authors did not investigate the effect of feeding the flies different diets. We determined that, on a high-sugar diet, loss of Hmgcr expression inhibits insulin signaling, which leads to increased food intake. This could mean that when flies are fed a high-sugar diet, insulin is required to signal to NPF neurons to inhibit overeating. On the other hand, on a less energetic low-sugar diet, insulin signaling may be dispensable.
We also studied the effect of IPC Hmgcr on the expression of the conserved insulinregulated α-glucosidase, target of brain insulin (tobi). Tobi expression is regulated by both the insulin-like peptides (ILPs) and the fly glucagon analogue Adipokinetic hormone (AKH). In their screen to identify genes regulated by adult insulins, Buch et al. [19] discovered that loss of the ILPs leads to a significant reduction in tobi expression. In fact, tobi was the gene most affected by ILP loss [19]. Interestingly, even though the ILPs were required for tobi expression, high circulating sugar levels inhibited tobi, while high levels of protein increased tobi expression. They went on to show that this increase in tobi expression required AKH signaling. In our study, we confirm that higher levels of carbohydrate intake inhibit tobi expression. Interestingly, similar to ILP loss, inhibiting Hmgcr expression in the IPCs reduced tobi levels even further (see Figure 5D). In mammals, glycogen metabolism can regulate feeding behavior, where inhibition of glycolysis increases feeding [42,43]. Therefore, through its glycogen regulating activity, Tobi could control feeding in Drosophila. In support of this, flies, where tobi was knocked down in the midgut, were hyperphagic, and inhibiting Hmgcr activity via fluvastatin did not increase total intake (see Figure 3C). On the other hand, overexpressing tobi significantly inhibited food intake, and feeding these flies fluvastatin increased intake back to normal levels (see Figure 3D). Moreover, Hmgcr males maintained on a low-sugar diet, where tobi levels are normal, were not hyperphagic. Thus, it seems Tobi could control feeding behavior downstream of central Hmgcr activity.
Many mechanisms controlling feeding behavior and energy balance are conserved between flies and mammals [44,45]. Importantly, in our study, we found that inhibition of Hmgcr activity in the fly hypothalamus-like structure (pars intercerebralis) or rodent (mouse and rat) hypothalamus led to a hyperphagic phenotype. In flies, this was most likely due to inhibition of insulin signaling, which has previously been shown to regulate food intake [32,33]. In mice, we demonstrate that Hmgcr is highly expressed in the arcuate nucleus (ARC). Moreover, inhibiting Hmgcr in the rat hypothalamus, via statin injection, led to increased neuronal activity in the ARC and paraventricular nucleus (PVN). Interestingly, the ARC contains two classes of insulin-regulated neurons, the orexigenic neuropeptide Y/agouti-related protein (NPY/AgRP) neurons and anorexigenic pro-opiomelanocortin/cocaine and amphetamine-regulated transcript (POMC/CART) neurons, that signal to the PVN (reviewed by [46,47]).
Overall, manipulating Hmgcr gene expression in the brain by both, genetic tools, or statin drugs, might explain in large part the link between HMGCR and the susceptibility to obesity by increased food intake (high energy intake) and fat storage, possibly by causing insulin resistance peripherally and within the brain. Further studies are needed to tease out how central Hmgcr activity, especially within the hypothalamus, regulates feeding behavior.

Conclusions
This study presents evidence of how the central regulation of Hmgcr can modify metabolism and food intake and, therefore, could explain, to a certain extent, how it influences BMI.