Regulation of Energy Balance by the Hypothalamic Lipoprotein Lipase Regulator Angptl3

Obesity is a growing global health concern that underlies the development of type 2 diabetes, hypertension, and cardiovascular diseases. Although many factors contribute to the obesity epidemic, chronic positive imbalance between energy intake and expenditure is thought to be the major cause of common forms of human obesity. For over a century, the hypothalamus has been considered to be a controlling center of energy balance (1). Specialized hypothalamic neurons sense the whole-body nutritional state, which is crucial for the hypothalamic regulation of energy balance (2).

Intracerebroventricular (ICV) administration of long-chain fatty acid (LCFA) suppresses food intake and glucose production (3), indicating that fatty acids can signal nutrient availability to the central nervous system (CNS) and thereby restrict the additional release of nutrients into the circulation. Accumulating data suggest that esterified LCFA (LCFA-CoA) is an important lipid metabolite in hypothalamic lipid-sensing pathways (4). Hypothalamic inhibition of carnitine palmitoyltransferase-1, which increases neuronal LCFA-CoA levels by inhibiting fatty acid oxidation, reduces food intake and glucose production (5). This effect is reversed by a blockade of LCFA-CoA formation, supporting a crucial role for LCFA-CoA in hypothalamic lipid sensing (5). By contrast, hypothalamic activation of AMPK, which decreases LCFA-CoA levels by increasing fatty acid oxidation, stimulates food intake (4,6).

Lipoprotein lipase (LPL) is a key enzyme in lipid metabolism that hydrolyzes triglycerides (TGs) in circulating TG-rich lipoproteins and promotes the cellular uptake of chylomicron remnants, cholesterol-rich lipoproteins, and free fatty acids by adipose tissue and skeletal muscle (7). LPL is highly expressed throughout the brain, including the hypothalamus and hippocampus (8). In the CNS, LPL modulates various aspects of neurobiology such as learning and memory by regulating neuronal lipid uptake (9). Notably, mice with a neuron-specific LPL deficiency develop obesity, indicating an important role of neuronal LPL in the control of energy balance (10). In these mice, hypothalamic LCFA levels are decreased, suggesting that LPL-dependent lipid uptake in the hypothalamus is essential for the maintenance of energy homeostasis.

The angiopoietin-like proteins (Angptls) are a family of proteins that share two structural domains with the angiopoietins: the N-terminal coiled-coil domain and the COOH-terminal fibrinogen-like domain (11). Among the eight Angptls, Angptls 3, 4, 6, and 8 are implicated in glucose, lipid, and energy metabolism (11,12). Angptl8/betatrophin controls TG trafficking from skeletal muscle and heart to adipose tissues upon food intake (12). Angptl6, also known as angiopoietin-related growth factor, controls peripheral fatty acid oxidation and energy metabolism (13). Angptl4 regulates plasma TG levels by directly inhibiting LPL activity and mediates the antiobesity effects of a germ-free condition (14,15). Moreover, we previously showed that Angptl4 regulates feeding behavior and energy metabolism via central mechanisms (16), extending the biological roles for Angptls to the CNS regulation of energy metabolism.

Angptl3 is known as an important regulator of circulating lipid metabolism (11,17). In the periphery, Angptl3 mRNA is exclusively found in the liver, where it is secreted into the circulation (18). Angptl3 potently inhibits the activity of LPL and endothelial lipase (19). Supporting these actions, Angptl3-deficient mice display lower circulating TG levels (20,21), whereas Angptl3 administration results in increased plasma lipid levels (17). In humans, loss-of-function mutations of Angptl3 are associated with familial combined hypolipoproteinemia (22). Based on our preliminary data showing hypothalamic Angptl3 expression, we investigated a potential regulatory role for Angptl3 in the central regulation of energy balance.

Full-length Angptl3 and Angptl4 were purchased from AdipoGen (Incheon, Republic of Korea), and apolipoprotein C3 (apoC3) was purchased from Sigma (St. Louis, MO). Triacsin-C (Tri-C) and AICAR were purchased from Sigma and Toronto Research Chemicals (Toronto, Ontario, Canada), respectively. Heparin was obtained from Halim Pharmaceuticals (Seoul, Republic of Korea).

Adult male C57BL/6 mice and Sprague-Dawley rats (8–10 weeks of age) were purchased from Orient Bio (Seoul, Republic of Korea). Animals were allowed free access to a standard chow diet (Cargill Agri Purina, Inc., Seoul, Republic of Korea) and water, unless otherwise indicated. Animals were housed under conditions of controlled temperature (22 ± 1°C) and a 12-h light-dark cycle, with the light on from 0800 to 2000 h. All procedures were conducted in accordance with the International Guiding Principles for Biomedical Research Involving Animals (National Institutes of Health, Bethesda, MD) and were approved by the Institutional Animal Care and Use Committee at the ASAN Institute for Life Sciences (Seoul, Republic of Korea). Agouti-related protein (AGRP) neuron–specific EGFP-expressing (AGRP-EGFP) mice were generated by crossing AGRP-IRES-Cre mice (obtained from Dr. Joel. K. Elmquist, University of Texas Southwestern Medical Center, Dallas, TX) with EGFP-floxed mice (The Jackson Laboratory, Bar Harbor, ME).

Stainless steel cannulae (26 gauge; Plastics One) were implanted into the third cerebral ventricle of the mice and rats, as described previously (23). After a 7-day recovery period, correct positioning of each cannula was confirmed by a positive dipsogenic response to angiotensin-2 (50 ng per mouse and 150 ng per rat). All peptides and compounds were dissolved in 0.9% saline solution just before use, and were administered via the ICV-implanted cannulae in a total volume of 2 μL. The majority of feeding studies were performed just before the end of the daily light cycle, unless otherwise indicated. Food intake and body weight were monitored for 24 h after the injections.

Small interfering RNAs (siRNAs) specific for murine Angptl3 (catalog #E-065291-00-0020) and LPL (catalog #STOP-130319-015) and a nontargeting scrambled siRNA (catalog #SCRO-101125-010) were purchased from Dharmacon (Chicago, IL) and injected into the bilateral mediobasal hypothalamus (MBH) as previously described (23). Successful knockdown was confirmed by determining hypothalamic Angptl3 and LPL mRNA levels at the end of the study. Gene knockdown was considered successful when hypothalamic Angptl3 and LPL mRNA levels fell to <30% of the average levels of the control group. Animals showing successful gene knockdown were included in the analysis.

Energy expenditure (EE) and respiratory quotient (RQ) were monitored using the Oxymax Comprehensive Laboratory Animal Monitoring System (CLAMS; Columbus Instruments, Columbus, OH) over 24–48 h after treatment. Animals were acclimatized in metabolic cages for 3 days before treatment.

Saline, Angptl3 (1 μg), Angptl4 (3.5 μg), apoC3 (0.1 μg), and heparin (5–10 units) were injected via ICV-implanted cannulae just before the beginning of the dark period, unless otherwise indicated. The MBH was collected at 2 h postinjection. Immediately after collection, MBH tissue was weighed and minced into small pieces. After rinsing with cold PBS to remove red blood cells, the minced tissue was homogenized in 60 μL of cold 20 mmol/L Tris buffer, pH 7.5, containing 150 mmol/L NaCl. After centrifugation at 10,000g for 10 min at 4°C, the supernatant was carefully collected and subjected to LPL activity assay following the manufacturer’s protocol (Cell Biolabs, San Diego, CA). The cerebrospinal fluid (CSF) of rats was collected 2 h after the ICV injection of heparin (20 units), as previously described (24). LPL activity was measured in 50-μL amounts of CSF.

The MBH was collected 2 h after ICV injection of saline, Angptl3 (1 μg), or Angptl4 (3.5 μg) into overnight-fasted mice during the early light phase. The tissue was then flash frozen in liquid nitrogen. Total AMPK activity was determined as previously described (25).

To measure LCFA, free fatty acids were extracted from 10–20 mg of hypothalamic tissue or from 80 μL of plasma using 400 μL methanol and isooctane. The extracted free fatty acids were derivatized using BCl₃-methanol (26). A standard mixture of fatty acid methyl esters obtained from Sigma-Aldrich was used to generate calibration curves. Fatty acid methyl ester was analyzed with a gas chromatography–mass spectrometry system (7890A/5975C detectors; Agilent) using a capillary column (HP-5MS; 30 m × 0.25 mm × 0.2 µm) and electron impact ionization. To measure LCFA-CoA, samples were prepared from 10−15 mg of hypothalamic tissues, as previously described (27). The dried samples were stored at −20°C and reconstituted with 30 μL H₂O/acetonitrile (50/50 volume for volume) prior to analysis. Liquid chromatography–tandem mass spectroscopy was performed using the 1290 Infinity Binary HPLC system (Agilent), the Qtrap 5500 system (ABSciex), and a reverse-phase column (2.1 × 150 mm; ZORBAX 300Extend-C18; Agilent). Analytical conditions were the same as those previously described (27). Hypothalamic TG contents were determined using a colorimetric assay (Abcam). MBH blocks were homogenized with 60 μL 5% NP-40. The homogenate was centrifuged, and 50 μL of the resulting supernatant was used in the assay.

Mice were perfused with 4% paraformaldehyde through the heart under anesthesia for 15 min after a 5-h fast. After postfixation and dehydration, coronal brain sections (14 μm thick) were cut using a cryostat (Leica, Wetzlar, Germany) and incubated with primary antibodies against the Angptl3 N-terminal fibrinogen-like domain (1:200, mouse; Abcam), microtubule-associated protein 2 (MAP2) (1:800, chicken; Millipore), glial fibrillary acid protein (GFAP) (1:300, rabbit; Abcam), apoC3 (1:100, rabbit; Santa Cruz Biotechnology), β-endorphin (1:10,000, rabbit; Phoenix Pharmaceuticals), c-fos (1:100, rabbit; Santa Cruz Biotechnology), LPL (1:100, rabbit; Santa Cruz Biotechnology), or EGFP (1:1,000, rabbit; Invitrogen) at 4°C for 24 h. After washing, slides were incubated with an Alexa Fluor 488–conjugated goat anti-mouse antibody, an Alexa Fluor 546–conjugated goat anti-rabbit antibody, or an Alexa Fluor 633–conjugated goat anti-chicken antibody (all from Invitrogen) at room temperature for 1 h. For nuclear staining, slides were treated with DAPI (1:10,000; Invitrogen) for 10 min before mounting. Triple immunofluorescence was examined by confocal microscopy (Leica).

Total RNA was extracted using TRIzol Reagent (Life Technologies). Hypothalamic mRNA expression levels of Angptls and LPL were determined by real-time PCR or semiquantitative RT-PCR (the primer sequences are shown in Supplementary Table 1) and normalized to levels of GAPDH mRNA.

Tissue proteins were extracted from hypothalamic tissue blocks, as previously described (16), and were subjected to immunoblotting using primary antibodies against Angptl3 (1:2,000, mouse; Abcam), LPL (1:1,000, rabbit; Santa Cruz Biotechnology), paired amino acid–converting enzyme 4 (PACE4) (1:1,000, rabbit; Santa Cruz Biotechnology), proprotein convertase subtilisin/kexin type 5 (PCSK5) (1:1,000, rabbit; Sigma), phosphorylated (Tyr705) and total signal-transduction-activated transcript-3 (Stat3), and phosphorylated (Thr308) and total Akt (all from Cell Signaling Technology). Hypothalamic tissue band density was measured with a densitometer (VersaDoc Multi Imaging Analyzer System; Bio-Rad, Hercules, CA).

Conditioned taste aversion (CTA) tests were conducted after sequential intraperitoneal injection of saline plus ICV saline, intraperitoneal lithium chloride (127 mg/kg, Sigma) plus ICV saline, and intraperitoneal saline plus ICV Angptl3 (1 μg), using a previously described protocol (28).

All data are expressed as the mean ± SEM. Statistical analysis was performed using SPSS version 17.0 software (SPSS Inc., Chicago, IL). Comparisons between two groups were analyzed using the Student t test. Comparisons among three or more groups were conducted using repeated or one-way ANOVA followed by a post hoc least significance difference test. A P value <0.05 was considered statistically significant.

Real-time PCR analysis was first used to demonstrate the mRNA expression of Angptl3 and other Angptls in the MBH of normal mice (Fig. 1A). Immunoblotting analysis also revealed the presence of full-length Angptl3 (62 kDa) in protein extracts of the hypothalamic arcuate nucleus (ARC), the lateral hypothalamus (LH) area, and the paraventricular nucleus (PVN) collected from mice fasted for 5 h using the punch biopsy technique (Fig. 1B). A higher expression of Angptl3 was found in the ARC and LH compared with the PVN. To verify whether MBH Angptl3 protein expression is altered by nutrient availability (Fig. 1C), mice were either kept under fasted conditions for 24 h before being killed or allowed to resume food intake during the beginning of the dark period after a 24-h fast. Hypothalamic Angptl3 protein levels were higher in refed mice than in fasting mice (Fig. 1C), indicating that hypothalamic Angptl3 expression increases during the postprandial period.

To determine whether Angptl3 is expressed in hypothalamic neurons or glial cells in the mouse MBH, double immunofluorescence staining was performed using antibodies against Angptl3 and the neuronal marker MAP2 or the glial marker GFAP. Angptl3 immunoreactivity almost completely overlapped with MAP2 but not with GFAP (Fig. 1D), suggesting that Angptl3 is mostly expressed in hypothalamic neurons. High levels of Angptl3 expression were observed in the soma of MBH neurons, while lower expression was seen in neuronal processes (Fig. 1D).

The key appetite-regulating neuropeptides proopiomelanocortin (POMC) and AGRP are produced by a discrete population of MBH neurons that play a primary role in hypothalamic nutrient sensing (1,29). Double immunostaining for Angptl3 and the POMC product β-endorphin revealed that >90% of POMC neurons expressed Angptl3. Dual staining for Angptl3 and EGFP in AGRP-EGFP mice revealed Angptl3 expression in ∼40–50% of AGRP neurons (Fig. 1E). Notably, a considerable proportion of Angptl3-expressing neurons did not produce either POMC or AGRP.

Given the abundant expression of Angptl3 in the MBH, a brain area critical for the control of energy balance, we hypothesized that Angptl3 may be involved in the hypothalamic regulation of energy balance. To test this hypothesis, MBH Angptl3 expression was depleted by microinjecting Angptl3 siRNA into the bilateral MBH. Successful Angptl3 knockdown in the MBH was confirmed by examining hypothalamic Angptl3 mRNA and protein expression levels at the end of the study (Fig. 2A and B). Compared with controls injected with a nontargeting siRNA, mice with reduced hypothalamic Angptl3 expression consumed more food and recovered more quickly from postsurgical weight loss (Fig. 2C). Moreover, CLAMS revealed that these mice displayed a decreased EE and an increased RQ during the dark period when access to food was not restricted (Fig. 2D and E). Considering that hypothalamic Angptl3 expression increases upon food intake (Fig. 1C), hypothalamic Angptl3 may coordinate multiple metabolic responses to food intake, including satiety generation, stimulation of EE, and transition of nutrient oxidation from fat to carbohydrate.

The metabolic effects of exogenous administration of Angptl3 were then examined. Full-length mouse Angptl3 (0.3, 1, and 3 μg) was administered via an ICV-implanted cannula in the beginning of the light phase in mice fasted overnight. ICV injection of Angptl3 significantly suppressed fasting-induced feeding in a dose-dependent manner (Fig. 3A). The anorexigenic effect of a single ICV injection of Angptl3 was sustained for at least 24 h. Moreover, ICV injection of Angptl3 inhibited weight gain for 24 h postinjection (Fig. 3B). To test the possibility that ICV injection of Angptl3 may increase aversion to food, we performed a CTA test after ICV injection of Angptl3. In contrast to intraperitoneal injection of the established CTA inducer lithium chloride, which decreased saccharine consumption, ICV Angptl3 injection did not induce CTA (Fig. 3C), suggesting that Angptl3-induced anorexia was not due to a conditioned aversion to food.

To examine the effects of Angptl3 on whole-body energy metabolism, Angptl3 (1 μg) was injected ICV into freely fed mice just before the end of the light cycle, and energy metabolism was monitored using the CLAMS. Angptl3-treated mice displayed a higher EE and a lower RQ in the dark period compared with saline-injected mice (Fig. 3D), indicating that ICV Angptl3 stimulates EE and promotes fat consumption in peripheral tissues. Taken together with data from loss-of-function and gain-of-function studies, an increase in hypothalamic Angptl3 leads to a negative energy balance.

We further investigated whether hypothalamic neurons are responsive to Angptl3 by using c-fos immunostaining after ICV injection of Angptl3 (1 μg). The c-fos immunoreactive neurons were mostly found in the hypothalamic ARC of the Angptl3-treated mice (Fig. 3E), suggesting that the neurons primarily responsive to Angptl3 reside in this area. Since the ARC neurons produce important metabolic regulators such as POMC, AGRP, and neuropeptide Y (NPY), we tested the possibility that Angptl3 may exert its actions through regulation of these neuropeptides. Indeed, the mRNA expression of NPY and AGRP was profoundly decreased by ICV injection of Angptl3 (1 μg), while POMC mRNA expression was not altered (Fig. 3F). Thus, Angptl3 may regulate energy balance via the downregulation of hypothalamic NPY and AGRP.

We further sought to identify the signaling pathways by which Angptl3 regulates feeding behavior and body weight. Hypothalamic AMPK activity was determined after ICV injection of Angptl3 (1 μg), as AMPK is an important downstream mediator of Angptl4 in the hypothalamus (16). Consistent with a previous report, ICV injection of Angptl4 (3.5 μg) potently suppressed AMPK activity in the MBH at 2 h postinjection. By contrast, ICV Angptl3 injection had no effect on hypothalamic AMPK activity (Fig. 4A). The effects of ICV injection of Angptl3 on hypothalamic Stat3 and on phosphoinositide 3-kinase-Akt signaling were also examined, as many appetite regulators act through these signaling pathways (28,30). Hypothalamic Stat3 and Akt phosphorylation were not significantly altered by ICV-injected Angptl3 (Supplementary Fig. 1), indicating that these signaling pathways are not involved in the Angptl3-mediated regulation of energy balance.

In the periphery, Angptl3 strongly inhibit LPL activity (19). Hypothalamic LPL activity was therefore evaluated 2 h after ICV administration of Angptl3 in mice fasted for 5 h. Animals were kept under fasting condition until they were killed to avoid any possible effects of food intake on hypothalamic LPL activity. Contrary to its inhibitory effect on peripheral LPL, ICV injection of Angptl3 (1 μg) increased LPL activity in the MBH by ∼1.8-fold (Fig. 4B). Consistently, hypothalamic LPL activity was found to be lower in mice injected with Angptl3 siRNA than in mice injected with nontargeting siRNA after a 5-h fast (Fig. 4C). ICV injection of Angptl4 (3.5 μg) also caused a modest but significant increase in hypothalamic LPL activity (Fig. 4B).

We further investigated the mechanisms of Angptl3-mediated regulation of hypothalamic LPL activity. By using immunoblotting and immunofluorescence staining, we demonstrated that Angptl3 (1 μg) increased MBH LPL protein levels when ICV injected in mice fasted for 5 h just before lights went off, whereas it did not increase hypothalamic LPL mRNA levels (Fig. 4D–F). In the periphery, Angptl3 enhances LPL cleavage induced by proprotein convertases (furin, PACE4, and PCSK5) (31). Hypothalamic PACE4 and PCSK5 expression was unchanged after ICV injection of Angptl3 (Fig. 4E). Therefore, Angptl3 upregulates and activates hypothalamic LPL via as yet unidentified post-transcriptional regulation. On the other hand, the LPL assay we used in this study can also measure other lipase activities. Hypothalamic expression of hepatic lipase and endothelial lipase was not altered by Angptl3 administration (Fig. 4F), suggesting that LPL is a major hypothalamic lipase regulated by Angptl3.

To confirm the effects of hypothalamic LPL activation on food intake and body weight, mice were treated with heparin, which activates LPL by promoting LPL release from cells (32). Indeed, MBH LPL activity increased after ICV injection of heparin (10 units) (Fig. 5A). LPL activity in the CSF was also increased after ICV injection of heparin (20 units) in Sprague-Dawley rats (Fig. 5B). Similar to Angptl3, ICV injection of heparin suppressed nighttime food intake and decreased body weight when injected before the beginning of the dark period (Fig. 5C and D). These findings suggest that hypothalamic LPL activation induced by Angptl3 and heparin could lead to decreased food intake and body weight.

To determine whether hypothalamic LPL activity is regulated by feeding conditions, hypothalamic LPL activity was measured under fasting and refeeding conditions. Hypothalamic LPL activity increased after food intake (Fig. 5E), consistent with the effects of Angptl3 on hypothalamic LPL activity.

ApoC3, a known LPL inhibitor (33), was found to be highly expressed in ependymal lining cells and MBH neurons (Fig. 6A). In addition, a certain population of hypothalamic neurons coexpressed Angptl3 and apoC3. Therefore, apoC3 was next used to determine whether the inhibition of hypothalamic LPL can block the effects of Angptl3 on energy metabolism. In contrast to Angptl3 and heparin, ICV injection of apoC3 strongly suppressed hypothalamic LPL activity (Fig. 6B). To investigate the effect of hypothalamic LPL inhibition on energy metabolism, apoC3 (30–1,000 ng) was administered ICV to freely fed mice before the end of the daily light cycle. ApoC3 treatment stimulated nighttime food intake in a dose-dependent manner (Fig. 6C), although these effects were not sustained for 24 h postinjection (data not shown). The ability of apoC3 to block the effects of exogenous Angptl3 on hypothalamic LPL and metabolic behavior was tested next. ApoC3 (0.1 μg) was administered to freely fed mice at the beginning of the dark cycle, 30 min prior to the injection of Angptl3 (1 μg). ApoC3 pretreatment markedly blunted ICV Angptl3–induced changes to hypothalamic LPL activity, food intake, body weight, EE, and RQ (Fig. 6D–F), implying that hypothalamic Angptl3 regulates energy metabolism by activating hypothalamic LPL.

To further confirm a role for hypothalamic LPL as a downstream mediator of Angptl3, hypothalamic LPL was depleted by injecting an LPL-specific siRNA into the bilateral MBH (Fig. 6G). Similar to mice with a neuron-specific depletion of LPL (10), mice with an siRNA-mediated reduction in hypothalamic LPL had a higher body weight than mice injected with a scrambled control siRNA (Fig. 6H). In these mice, ICV Angptl3–induced anorexia and weight loss were significantly blunted (Fig. 6H).

Angptl3-induced hypothalamic LPL activation would be expected to increase LCFA uptake by the hypothalamic neurons (10). After uptake, LCFAs are esterified to LCFA-CoA by acyl CoA synthetase (ACS) and accumulate in hypothalamic neurons (4). Increased hypothalamic LCFA-CoA signals to the hypothalamus that energy is in excess, leading to reduced food intake, as depicted in Fig. 7A (4). Direct measurement of LCFA and LCFA-CoA revealed that ICV-injected Angptl3 significantly increased the ratios of hypothalamic versus plasma levels of both saturated (C14:0 and C16:0) and unsaturated (C18:1n9, C20:4n6, and C22:6n3) LCFAs (Fig. 7B). Furthermore, hypothalamic stearoyl-CoA was significantly increased at 2 h after Angptl3 administration (Fig. 7C). Hypothalamic oleoyl-CoA levels were also marginally elevated, while palmitoyl-CoA levels were unaltered by Angptl3 treatment. The neuronal malonyl-CoA levels have been suggested to be critical for the hypothalamic regulation of energy homeostasis (34). Notably, ICV-administered Angptl3 significantly increased the hypothalamic malonyl-CoA levels but not the hypothalamic TG content (Fig. 7D and E). Thus, an increase in hypothalamic malonyl-CoA and LCFA-CoA levels may contribute to the Angptl3-induced suppression of food intake.

In order to determine whether Angptl3 acts through the hypothalamic lipid-sensing pathway, the ACS inhibitor Tri-C was administered prior to Angptl3 to inhibit the formation of LCFA-CoA (Fig. 7A) (4,5). Consistent with our hypothesis, the blockade of ACS by Tri-C significantly attenuated the Angptl3-induced reduction in food intake and body weight (Fig. 7F).

By contrast, AMPK activation inhibits the hypothalamic lipid-sensing pathway by stimulating mitochondrial LCFA-CoA oxidation through sequential acetyl-CoA carboxylase inhibition and carnitine palmitoyltransferase-1 activation (Fig. 7A) (4). To inhibit the hypothalamic lipid-sensing pathway, a low dose (0.2 nmol) of AICAR was centrally administered prior to Angptl3 injection in mice that had been fasted overnight. AICAR treatment alone did not further increase fasting-induced hyperphagia and weight gain, as hypothalamic AMPK activity would already be elevated in these fasted mice (Fig. 7G). Nevertheless, pretreatment with AICAR negated the Angptl3-induced anorexia and weight reduction (Fig. 7G). Consistent with these feeding effects, prior administration of Tri-C and AICAR attenuated the Angptl3-induced increase in the hypothalamic stearoyl-CoA content (Fig. 7H). These data highlight the importance of hypothalamic lipid-sensing pathways in the Angptl3-mediated regulation of energy balance.

Accumulating data suggest that hypothalamic lipid sensing is important for the maintenance of energy homeostasis. Neuronal LPL appears to modulate an early and important step in hypothalamic lipid sensing by controlling lipid uptake into the neurons. However, the regulatory mechanisms of hypothalamic LPL activity are not well understood.

In the current study, we demonstrated that Angptl3, a well-known regulator of peripheral LPL, is highly expressed in hypothalamic neurons including POMC and AGRP neurons. Moreover, ICV administration of Angptl3 increased hypothalamic LPL activity, whereas the inhibition of hypothalamic Angptl3 suppressed it. This finding was unexpected and contrasts with the inhibitory effects of these Angptls on peripheral LPL (19). In the periphery, LPL activity is modulated by cleavage of LPL at the linker region between its N-terminal catalytic domain and COOH-terminal lipid-binding domain (31). LPL cleavage promotes its dissociation from the cell surface, thereby inactivating it. Angptl3 enhances LPL cleavage induced by proprotein convertases such as furin, PACE4, and PCSK5 (31). In our current study, Angptl3 administration significantly increased the hypothalamic LPL protein expression, but had no effect on hypothalamic LPL mRNA levels and PACE4 and PCSK5 protein levels. Therefore, Angptl3 activates hypothalamic LPL through as yet unidentified post-transcriptional regulation.

We have shown that altered hypothalamic Angptl3 expression affects multiple aspects of energy metabolism. The siRNA-mediated inhibition of hypothalamic Angptl3 caused increased food intake, decreased EE, and increased RQ, leading to a positive energy balance. Conversely, ICV injection of Angptl3 induced anorexia via a mechanism unrelated to taste aversion, and stimulated EE and fat use during the dark period. ICV injection of Angptl3 increased the c-fos expression level in hypothalamic ARC, indicating that Angptl3 may act primarily on ARC neurons. Furthermore, Angptl3 administration potently inhibited hypothalamic NPY and AGRP expression, which may account for Angptl3-induced anorexia and weight loss.

Interestingly, the central metabolic effects of Angptl3 were similar to those of Angptl4 (16), indicating a functional similarity between these closely related Angptls (11). Despite similarities in structure and function, Angptl3 and Angptl4 act through different signaling pathways in the hypothalamus. Indeed, Angptl4 potently suppressed hypothalamic AMPK activity, while Angptl3 did not.

Recently, the involvement of neuronal LPL in body weight metabolism was demonstrated in mice with a neuron-specific deficiency in LPL (10). These mice developed obesity, which may be due to transient hyperphagia and reduced EE. In line with these findings, inhibition of hypothalamic LPL by ICV apoC3 and intra-MBH injection of LPL siRNA increased food intake and attenuated Angptl3-induced anorexia. By contrast, hypothalamic LPL activation induced by ICV administration of Angptl3 and heparin caused anorexia and weight loss. These data confirm a key role for hypothalamic LPL in controlling energy balance. However, some discrepancy exists between our findings and the phenotypes in neuronal LPL knockout mice (10). The metabolic effects of acute hypothalamic LPL inactivation may differ from those of chronic neuronal LPL depletion. Otherwise, Angptl3 might act through mechanisms other than hypothalamic LPL.

Hypothalamic n-3 fatty acid levels were reduced in mice lacking neuronal LPL (10). Consistent with these levels, we found that hypothalamic LCFA and LCFA-CoA levels increased after ICV injection of Angptl3. These findings support an important role for neuronal LPL in neuronal uptake of TG-rich lipoprotein-derived fatty acids. Notably, LPL activity in the hypothalamus increased in the postprandial state. Given that hypothalamic Angptl3 expression increased under the same condition, it can be reasonably speculated that Angptl3 mediates postprandial LPL activation in the hypothalamus.

Interestingly, the LPL inhibitor apoC3 was expressed in ependymal lining cells and MBH neurons. Hypothalamic expression of these LPL regulators strongly implies the presence of local mechanisms that regulate LPL activity depending on nutrient availability. Along with increased LCFA-CoA levels, Angptl3 treatment significantly elevated the MBH content of malonyl-CoA, an important lipid metabolite that affects systemic energy metabolism. Although the mechanism mediating this effect is unclear, the anorexigenic effect of Angptl3 may be partly attributable to increased hypothalamic malonyl-CoA levels.

Enhanced LPL activity may activate hypothalamic lipid-sensing pathways by increasing neuronal uptake of LCFA. Neuronal LCFA-CoA levels are critical for hypothalamic lipid sensing, and are affected by both LCFA esterification and mitochondrial oxidation of LCFA (Fig. 6A) (4). By blocking LCFA esterification with Tri-C and by activating LCFA oxidation with AICAR, we demonstrated that hypothalamic lipid-sensing pathways mediate the effects of Angptl3 on food intake, body weight, and hypothalamic LCFA-CoA levels. Since hypothalamic sensing of circulating lipid levels also controls hepatic glucose production (4), it will be interesting to investigate the effects of hypothalamic Angptl3 on peripheral glucose metabolism.

In summary, this study identified hypothalamic Angptl3 as a key regulator of energy balance, which acts through hypothalamic LPL and lipid-sensing pathways.

Acknowledgments. The authors thank Julie Roda and Stephen Cooke of Bioedit Corporation for their help in the preparation of the manuscript.

Funding. This study was supported by grants from the National Research Foundation of Korea (NRF-2013R1A1A3010137 and 2013M3C7A1056024), the ASAN Institute for Life Sciences (2013-326), and the Korean Diabetes Association (to M.-S.S., 2012).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. H.-K.K. and M.-S.S. performed the experiments and the data analysis and wrote the manuscript. B.-S.Y. and J.H.C. contributed to the discussion. G.M.K., S.Y.G., C.H.L., and H.S.L. performed the experiments. H.J.Y. performed the experiments and contributed to the discussion. M.-S.K. wrote the manuscript and directed the study. M.-S.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.