Oleic Acid Directly Regulates POMC Neuron Excitability in the Hypothalamus

Abstract

The mammalian CNS relies on a constant supply of external glucose for its undisturbed operation. However, neurons can readily switch to using fatty acids and ketones as alternative fuels. Here, we show that oleic acid (OA) excites pro-opiomelanocortin (POMC) neurons by inhibition of ATP-activated potassium (K_ATP) channels. The involvement of K_ATP channels is further supported by experiments in SUR1 KO animals. Inhibition of β-oxidation using carnitine palmitoyltransferase-1 inhibitors blocks OA-induced depolarization. The depolarizing effect of OA is specific because it is not mimicked by octanoic acid. Furthermore, OA does not regulate the excitability of agouti-related peptide neurons. High-fat feeding alters POMC neuron excitability, but not its response to OA. Thus β-oxidation in POMC neurons may mediate the appetite-suppressing (anorexigenic) effects of OA.

INTRODUCTION

The hypothalamus is the primary locus for integration of signals that influence energy balance. Among the hypothalamic nuclei, the arcuate nucleus (ARC) of the hypothalamus is considered to play a key integrative role between the initial afferent signals from the periphery and CNS responses (Flier 2004; Schwartz and Porte Jr 2005; Schwartz et al. 2000). Anorexigenic pro-opiomelanocortin (POMC) neurons of the ARC respond to circulating signals—such as glucose, leptin, insulin, ghrelin, and peptide YY (Acuna-Goycolea and van den Pol 2005; Cowley et al. 2001; Schwartz and Porte Jr 2005; Schwartz et al. 2000; Spanswick et al. 1997, 2000; Wang et al. 2004)—and contribute to the regulation of energy expenditure by releasing the anorexigenic melanocyte-stimulating hormones (α-MSH and γ-MSH), products of POMC that activate centrally expressed melanocortin-3 (MC3R) and melanocortin-4 receptors (MC4Rs) (Mountjoy and Wong 1997).

Fatty acid metabolism within the hypothalamus functions as a sensor of nutrient availability (for review, see Lam et al. 2005b). Long chain fatty acids (LCFAs), including oleic acid (OA, C18:1), act as nutrient abundance signals in the hypothalamus (Lam et al. 2005b; Lopez et al. 2005). Administration of LCFAs into the hypothalamus decreases agouti-related peptide (AgRP) and neuropeptide Y (NPY) mRNA expression and inhibits food intake as well as endogenous glucose production in the rat (Morgan et al. 2004; Obici et al. 2002). Similarly, central inhibition of carnitine palmitoyltransferase-1 (CPT1a), required for entry of LCFAs into mitochondria for fatty acid β-oxidation, decreases energy intake and glucose production (Obici et al. 2003; Pocai et al. 2006). Of particular interest in this regard is that 3 days of overfeeding abolishes the metabolic and anorexigenic effects of the central administration of LCFAs in rodents (Morgan et al. 2004). Thus such a defect in nutrient sensing may contribute to the susceptibility to obesity and insulin resistance in response to voluntary overfeeding.

In addition, malonyl-coenzyme A (malonyl-CoA) in the CNS plays an essential role in the control of food intake and energy expenditure (for review, see Wolfgang and Lane 2006). Inhibitors of fatty acid synthase (FAS) have been reported to reduce food intake by inducing a buildup of malonyl-CoA (Loftus et al. 2000). Lowering hypothalamic malonyl-CoA concentration by virally mediated overexpression of malonyl-CoA decarboxylase within the hypothalamus leads to increased food intake and weight gain (He et al. 2006; Hu et al. 2005). Furthermore, the brain-specific CPT1c that is a target of malonyl-CoA is expressed in the arcuate nucleus (Dai et al. 2007) and CPT1c KO mice have lower body weight and food intake (Wolfgang et al. 2006). In contrast, overexpression of CPT1c in the hypothalamus prevents excessive body weight gain when fed a high-fat diet (HFD) (Dai et al. 2007).

In this study, we delineated the cellular mechanisms by which oleic acid regulates the excitability of POMC neurons and the control of endogenous glucose production. The regulation of K_ATP channels in melanocortinergic neurons by acute and long-term treatment with nutrients may contribute to the control of ingestive behavior and substrate utilization.

METHODS

Slice preparation

All mouse care and experimental procedures were approved by the Institutional Animal Care Research Advisory Committee of the Albert Einstein College of medicine. We used POMC-CRE Z/eGFP and AgRP-CRE Z/eGFP mice, wherein the POMC-CRE and AgRP-CRE transgenes cause cell-specific recombination to induce expression of eGFP from the β-actin promoter (Figs. 1A and 5C; Balthasar et al. 2004; van de Wall et al. 2008). Transverse brain slices were prepared from FVB male and female mice at postnatal age 21–28 days. Animals were anesthetized with a mixture of ketamine and xylazine. After decapitation, the brain was transferred into a sucrose-based solution, bubbled with 95% O₂-5% CO₂, and maintained at about 3°C. This solution contained (in mM): sucrose 248, KCl 2, MgCl₂ 1, KH₂PO₄ 1.25, NaHCO₃ 26, sodium pyruvate 1, and glucose 10. Transverse coronal brain slices (200 μm) were prepared using a vibratome. Slices were equilibrated with an oxygenated artificial cerebrospinal fluid (aCSF) for >1 h at 32°C prior to transfer to the recording chamber. The slices were continuously superfused with aCSF at a rate of 2 ml/min containing (in mM): NaCl 113, KCl 3, NaH₂PO₄ 1, NaHCO₃ 26, CaCl₂ 2.5, MgCl₂ 1, and glucose 5 or 2.5 in 95% O₂-5% CO₂ at 30 ± 2°C. Osmolarity was adjusted with sucrose.

FIG. 1.

Pro-opiomelanocortin (POMC) neurons express adenosine 5′-triphosphate (ATP)–activated potassium (K_ATP) channels. A: identification of POMC neurons in acute brain slice. Example of infrared–differential interference contrast (IR-DIC) image, green fluorescent protein (GFP) expression, and fluorescent dye (Alexa 568)–injected neuron. B: electrophysiological profile of POMC neuron. Responses of a POMC neuron to depolarizing/ hyperpolarizing current pulses recorded at the resting membrane potential (V_m) in current-clamp mode. The neuron shown, like most POMC neurons, has moderate spike accommodation in response to direct depolarization. V_m = −43 mV. C: whole cell current-clamp recording of POMC neurons in acute brain slice. On access in whole cell mode (1 mM [ATP]_internal), the resting potential was −50 mV. Within <1 min, V_m slowly hyperpolarized, reaching a maximum V_m of −71 mV after about 5 min. During this period, excitability dramatically declined. Bottom: POMC neuron activity on an expanded timescale. D: representative recording sample of the effect of the K_ATP channel blocker. Treatment with glibenclamide (20 nM) induced action potential (AP) discharge of POMC neuron (top). Bottom: neuronal activity on an expanded timescale. V_m = −47 mV. E: sample recording of V_m of POMC neurons in the presence of 5 mM [ATP]_internal. V_m was stable at −45 mV without any further membrane hyperpolarization and POMC neurons show spontaneous APs. Bottom: neuronal activity on an expanded timescale. F1 and F2: current–voltage (I–V) relationship of the tolbutamide-sensitive run-up current (I_run-up). F1 shows the whole cell membrane currents in response to ramping the V_m from −156 to −66 mV (ramp 90 mV/2 s) following I_run-up and after subsequent application of the K_ATP channel blocker tolbutamide (200 μM). Tolbutamide completely inhibited I_run-up. F2 represents the I–V relationship of the tolbutamide-sensitive current.

FIG. 5.

The depolarizing effect of OA is specific. A: sample recording of membrane potential of POMC neuron before, during, and after application of octanoic acid (C8; 50 μM). In contrast to the effect of OA (C18:1), octanoic acid (OC) did not depolarize POMC neurons in the same experimental conditions. V_m = −47 mV. B: population data for membrane potential with OC at 50 and 100 μM (OC: 50 μM: control: −61.6 ± 2 mV vs. plus OC: −61.5 ± 2 mV; n = 11; 100 μM: control: −61 ± 2.5 vs. −61.6 ± 2 mV; n = 14). C: identification of agouti-related peptide (AgRP) neurons in acute brain slice. Example of IR-DIC image and GFP expression. D: sample recording of membrane potential of AgRP neuron before, during, and after application of OA (50 μM). OA was without effect in the presence of CAPS. V_m = −48 mV. The vehicle control for OA perfusion was 0.1% EtOH.

Electrophysiological recordings

Brain slices were placed on the stage of an upright, infrared–differential interference contrast (IR-DIC) microscope (Olympus BX50WI) mounted on a Gibraltar X-Y table (Burleigh). Green fluorescent protein (GFP)–positive neurons were visualized with a ×40 water-immersion objective using epifluorescence imaging and patched under IR-DIC optics. Membrane currents and potentials were recorded with an Axopatch 200B patch-clamp amplifier or a Multiclamp 700B in whole cell configuration. Electrophysiological signals were low-pass filtered at 2–5 kHz, stored on a PC, and analyzed off-line with pCLAMP 10 software (Molecular Devices, Sunnyvale, CA). In whole cell voltage-clamp mode, series resistance was not compensated. The access resistance of POMC cells was typically 10–15 MΩ and the cell capacitance was typically 10–18 pF. Because current amplitude did not exceed 300 pA, an access resistance of 15 MΩ resulted in a voltage error of <4.5 mV. Access resistance was checked repeatedly during the experiments and recordings showing an increase of >20% were rejected. Whole cell patch-clamp recordings were made from >300 POMC-eGFP neurons in mouse brain slice preparations in 5 mM glucose, except as noted (Fig. 1, A and B).

The spike rate of POMC neurons was also recorded using cell-attached patch-clamp technique in voltage-clamp mode (Perkins 2006). The recording pipette was voltage-clamped at 0 mV. For amphotericin-based perforated-patch recording, the amphotericin B stock solution (30 mg/ml) was prepared in dimethylsulfoxide just before the recording session. The pipette was first filled at the tip with an internal solution containing (in mM): KCl 150 and HEPES 10 (pH 7.3) and subsequently back-filled with the same solution containing amphotericin B (150 μg/ml). Cell-attached and perforated recordings were also performed in the presence of glutamate, γ-aminobutyric acid type A (GABA_A), and glycine receptor blockers. Patch electrodes were pulled from thin-walled glass capillaries (WPI) using a temperature-controlled vertical puller (HEKA PIP5). Pipette resistance ranged from 2.5 to 4 MΩ.

The liquid junction potential between the electrode solution and the bath solution was compensated using the pipette offset control and the liquid junction potential between the cytosol and the pipette solution was calculated using the Henderson equation (Barry and Lynch 1991). The calculated liquid junction potential was about +16 mV. The reversal potential (E_rev) of the run-up current (I_run-up) was measured by ramping the membrane potential (V_m) from −156 to −66 mV for 2 s in voltage-clamp configuration. Membrane currents were also evoked between −146 and −66 mV (10-mV step for 500 ms) from a holding potential of −86 mV. The net I_run-up was obtained by subtracting the baseline current from the I_run-up.

6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX, 10 μM), d-amino-phosphonovaleric acid (d-AP-5, 50 μM), picrotoxin (100 μM), and strychnine (1 μM) (Sigma–Aldrich) were continuously present in aCSF except as noted. The internal solution contained (in mM): K-acetate, 115; KCl, 10; MgCl₂, 2; EGTA, 10; HEPES, 10; Na₂ATP, 1 or 5; and Na₂GTP, 0.5. For cell-attached recordings, the internal solution contained (in mM): NaCl, 140; KCl, 3.5; CaCl₂, 2.5; MgCl₂, 1; HEPES, 10; and glucose, 2.5. Throughout our electrophysiological experiments, we used a free form of OA without a carrier. Oleic acid and octanoic acid (Sigma–Aldrich) were prepared in 100% ethyl alcohol (EtOH) as 1,000× concentrated stock solution and dissolved at a final concentration in extracellular solution just before the recording session. After a stable baseline was established after several minutes of recordings, fatty acids were applied by bath perfusion for 5 min at a rate of 2 ml/min.

To examine whether the responses to OA are regulated by dietary history, two groups of animals were fed a standard diet (with 60% calories provided by carbohydrates, 20% by protein, and 5% by fat) or HFD (with 20% calories by carbohydrate, 20% by protein, and 60% by fat; D12492; Research Diets, New Brunswick, NJ) for 2 wk starting from postnatal age 14 days. We kept mice pups together with their dams until week 3 after birth. The pups were weaned onto HFD and provided HFD until sacrifice (a minimum of 14 days).

Statistics

Statistical analyses were performed on all POMC neurons examined using the paired t-test and one-way ANOVA followed by a Bonferroni post hoc procedure for between-group comparison in some experiments in which there were more than two variables. The mean values were reported from the entire population tested (Origin 7.0; OriginLab, Northampton, MA). Current–voltage (I–V) plots were compared by Student's t-test. Data were considered significantly different when the P value was <0.05. All statistical results are given as means ± SE.

RESULTS

POMC neurons express K_ATP channels

To examine the potential electrophysiological impact of fatty acid signaling in POMC neurons per se, we first examined the electrical characteristics of GFP-tagged POMC neurons in current-clamp configuration. We used 1 mM [ATP]_internal throughout our experiments, except as noted, because prior study showed that the resting concentration of cytosolic ATP in hypothalamic neurons was about 1 mM when monitored using genetically targeted luciferases (Ainscow et al. 2002). Immediately after breaking through to the whole cell recording configuration under these conditions, POMC neurons showed spontaneous discharge of action potentials (APs) when no current was delivered through the recording pipette (Fig. 1C). The average resting membrane potential of POMC neurons was −47 ± 1 mV (n = 21 neurons). As cell dialysis proceeded, about 70% of the neurons examined slowly hyperpolarized over a period of 5–10 min, reaching a maximum resting membrane potential of −66 ± 2 mV (n = 21 neurons).

Because prior studies have demonstrated the expression of K_ATP channels in POMC neurons (Dunn-Meynell et al. 1998; Ibrahim et al. 2003; Plum et al. 2006), we investigated whether the membrane hyperpolarization was associated with activation of K_ATP channels in POMC neurons. Treatment with the K_ATP channel blocker glibenclamide (20 nM) reversed the membrane hyperpolarization and elicited APs (Fig. 1D; n = 6 neurons). Because the gating of K_ATP channels is mainly regulated by the ratio of ATP/ADP in the cell (Ashcroft and Gribble 1998), we thus raised [ATP]_internal from 1 to 5 mM, given that a previous study showed that relatively high levels of ATP (>5 mM) inhibited K_ATP channels composed of SUR1 and Kir 6.2 subunits (Allen and Brown 2004). We found that raising [ATP]_internal induced no initial membrane hyperpolarization (Fig. 1E) and shifted the mean resting membrane potential from −65 ± 1.5 to −47 ± 1 mV (1 mM ATP; n = 22 neurons; 5 mM ATP; n = 10 neurons).

To further characterize the nature of the membrane hyperpolarization, we measured the E_rev by changing the membrane potential from −156 to −66 mV for 2 s in voltage-clamp configuration. Figure 1F represents a typical example of the whole cell membrane current in response to ramping the membrane potential following full current run-up (I_run-up) and subsequent application of the K_ATP channel blocker tolbutamide (200 μM). I_run-up was observed in about 60% of the neurons examined (n = 22 of 38 neurons). The I–V relationship revealed that the tolbutamide-sensitive current between −156 and −66 mV exhibited no outward rectification and that the reversal potential of the tolbutamide-sensitive current in 3 mM [K⁺]_external was −92.3 ± 3.8 mV, a value close to the calculated K⁺ equilibrium potential (n = 4; Fig. 1F2). Our data are consistent with prior studies showing the expression of K_ATP channel, composed of the SUR1 (ABCC8) and Kir6.2 (KCNJ11) in POMC neurons (Dunn-Meynell et al. 1998; Ibrahim et al. 2003; Plum et al. 2006).

Oleic acid directly excites POMC neurons and regulates endogenous glucose production

We then examined the direct action of OA on POMC neurons by measuring membrane potential in current-clamp mode. Treatment with oleic acid (OA, 50 μM) depolarized and increased the firing rate of POMC neurons in about 60% of the neurons tested (control: −63 ± 3 mV; plus OA: −57 ± 4; n = 8 neurons; P = 0.016), consistent with a recent report showing OA-induced excitation in a subpopulation of the neurons in the arcuate nucleus (Wang et al. 2006). The onset of OA action varied from 2 to 10 min. The enhanced electrical activity was not reversible following washout (>30 min).

We next eliminated presynaptic influences by adding a cocktail of glutamate, GABA_A, and glycine receptor blockers and determined whether OA directly altered the intrinsic electrical properties of POMC neurons. Under these experimental conditions, OA (50 μM) depolarized and induced AP discharge in around 40% of the neurons tested (Fig. 2, A and B; resting potential [V_m]: control: −65 ± 1 mV, plus OA: −61 ± 1 mV; n = 32 neurons; P = 0.01). The depolarization of POMC neurons was OA concentration dependent (Fig. 2C).

FIG. 2.

Oleic acid (OA) induces an increase in POMC neuron excitability. A and B: OA-induced excitation of POMC neurons. A shows an example of the whole cell membrane potential before, during, and after treatment with OA after having blocked synaptic transmission. OA depolarized and induced AP discharge of POMC neurons in the presence of CNQX, d-AP-5, picrotoxin, and strychnine (CAPS). V_m = −52 mV. B: pooled data of membrane potential with or without OA (open bars: without CAPS; filled bars: with CAPS). C: OA causes a concentration-dependent depolarization of POMC cells. Depolarization by OA was determined at 1, 5, 20, and 50 μM in 5, 9, 11, and 32 cells, respectively. D and E, top: sample membrane currents in response to voltage steps (from −146 to −66 mV; 10-mV step for 500 ms), with and without OA, at a holding potential of −86 mV (top). Bottom: the I–V curve shows the OA-sensitive component of current over this range of membrane potentials. E: the I–V curve shows that OA significantly reduced the amplitude of the I_run-up over this range of membrane potentials (n = 5; from −146 to −66 mV: P = 0.027, 0.026, 0.024, 0.044, 0.026, 0.004, 0.17, 0.16, 0.015). The vehicle control for OA perfusion was 0.1% ethyl alcohol (EtOH). F: representative whole cell recording sample of the electrophysiological effects of OA in the presence of guanosine-5′-O-2-thiodiphosphate (GDP-β-S, 2 mM). Under these experimental conditions, POMC neurons responded to OA. V_m = −47 mV.

Direct modulation of POMC neuron excitability was supported further by experiments in which we examined OA action on membrane currents evoked by voltage steps in voltage-clamp configuration (also see Supplemental Fig).1 Figure 2D shows membrane currents evoked between −146 and −66 mV (10-mV increase in voltage step) from a holding potential of −86 mV following current run-up and subsequent application of OA. Oleic acid significantly decreased the amplitude of the run-up current (I_run-up) over this range of membrane potentials (mean inhibition: 27 ± 5% at −130 mV; n = 5; Fig. 2E).

Depolarizing effect of OA occurs independently of G-protein signaling

An orphan G-protein–coupled receptor, GPR40, has been shown to be activated by both medium- and long-chain fatty acids and its activation regulates insulin secretion in pancreatic cells (Briscoe et al. 2003; Itoh and Hinuma 2005; Itoh et al. 2003). Because hypothalamic neurons express GPR40, we examined whether the depolarizing effect of OA is due to GPR40 activation (Briscoe et al. 2003). We intracellularly applied guanosine-5′-O-2-thiodiphosphate (GDP-β-S, 1 and 2 mM; n = 4 and 8 neurons, respectively) into the cell because inclusion of a nonhydrolyzable form of GDP through patch pipette prevents G-protein–mediated modulation of ion channels. Under such experimental conditions, POMC neurons still responded to OA with a depolarization (control: −60 ± 3 mV, plus OA: −55 ± 3; n = 12 neurons; P = 0.04; Fig. 2F), consistent with the fact that this depolarization occurs independently of G-protein signaling.

Oleic acid inhibits K_ATP channels

Based on the expression of K_ATP channels in POMC neurons and the inhibition of I_run-up following treatment with OA, we sought to determine whether OA inhibits K_ATP channels, thus increasing the excitability of POMC neurons. We tested a K_ATP channel opener (i.e., diazoxide) rather than K_ATP channel blockers on the response of POMC neurons to OA because both OA and glibenclamide depolarized POMC neurons in our preparations. As shown in Fig. 3, diazoxide (200 μM) either reversed or completely blocked OA-induced depolarization (n = 7 neurons), indicating that OA depolarizes POMC neurons by closure of K_ATP channels. To further determine the contribution of K_ATP channels, we investigated whether high levels of internal ATP restrain the excitatory effect of OA. Under the experimental conditions in which [ATP] is 5 mM, OA (50 μM) no longer depolarized POMC neurons (Fig. 3, C and D; control: −50 ± 2 mV; OA: −49 ± 2 mV; n = 8).

FIG. 3.

Oleic acid inhibits K_ATP channels in POMC neurons. A and B: an example of the membrane potential before, during, and after treatment with OA and OA + diazoxide. Application of diazoxide (200 μM) reversed the depolarizing effect of OA (A). V_m = −48 mV. B: pre-exposure to diazoxide induced no depolarization following application of OA. V_m = −45 mV. C and D: OA no longer induced an increase in the firing rate of POMC neurons in the presence of 5 mM [ATP]_internal (top). Bottom: neuronal activity on an expanded timescale. V_m = −47 mV. D: pooled data of the firing rate of POMC neurons before and after treatment with OA at 1 and 5 mM [ATP]_internal. OA did not significantly change the mean firing rate of POMC neurons at 5 mM [ATP]_internal (1 mM ATP: control: 0 Hz; OA: 2.1 ± 0.5 Hz; n = 9; 5 mM ATP: control: 3.1 ± 1 Hz; OA: 2.5 ± 0.7 Hz; n = 6 neurons: P = 0.34). E: AP discharge of POMC neuron in sulfonylurea receptor 1 subunit knockout (SUR1 KO) animals. POMC neurons did not show initial membrane hyperpolarization (V_m = −49 mV; left). OA did not excite POMC neurons in SUR1 KO animals. Right: sample recording on an expanded timescale. F: schematic diagram of proposed OA metabolism in POMC neurons (CPT1: carnitine palmitoyltransferase-1). Long chain fatty acids (LCFAs) require CPT1 for mitochondrial β-oxidation. G and H: β-oxidation of OA excites POMC neurons. G: example recording of membrane potential before, during, and after treatment with malonyl-CoA (MA; 20 μM) and MA + OA. MA alone did not change the membrane potential, but completely inhibited the depolarizing effect of OA. V_m = −46 mV. Likewise, prior exposure to ST1326 (ST; 100 nM) blocked OA action (H). V_m = −55 mV. I: pooled data of membrane potential with CPT1 inhibitor alone and CPT1 inhibitor + OA (MA: −63 ± 3 mV vs. MA + OA: −65 ± 3 mV; n = 8; P = 0.27; ST: −58 ± 3 mV vs. ST + OA: −60 ± 4 mV; n = 7; P = 0.06). The vehicle control for OA perfusion was 0.1% EtOH.

The contribution of K_ATP channels was supported further by experiments using sulfonylurea receptor 1 (SUR1) subunit knockout (KO) mice (Seghers et al. 2000). SUR1 is a regulatory subunit in K_ATP channels and endows the channel with sensitivity to drugs such as sulfonylureas and K_ATP channel openers. We combined the SUR1 KO allele with the POMC-eGFP transgenes to identify POMC neurons in SUR1 knockouts and then determined whether OA-induced depolarization is due to inhibition of K_ATP channel. Such experiments revealed that POMC neurons in SUR1 KO animals showed spontaneous AP discharge without membrane hyperpolarization over the recording period (Fig. 3E; V_m = 47 ± 2 mV; n = 7). Importantly, treatment with OA (50 μM) no longer depolarized nor increased the firing rate of POMC neurons (Fig. 3E; mean change in membrane potential: 0.6 ± 1 mV; n = 6; firing rate: control: 9.2 ± 3 Hz; OA: 10.8 ± 4 Hz; n = 6), consistent with the involvement of K_ATP channels in OA-induced depolarization of POMC neurons.

Prior studies have demonstrated that LCFA and LCFA-CoA open, rather than close, K_ATP channels (Gribble et al. 1998; Larsson et al. 1996). Thus we directly introduced oleoyl-CoA (20 nM to 1 μM) into POMC neurons via patch pipette, anticipating a membrane hyperpolarization in POMC neurons. Contrary to our expectation, none of the cells tested was hyperpolarized by injection of oleoyl-CoA (n = 12). Instead, such manipulations shifted the resting membrane potential from −47 ± 1 to −40 ± 2 mV (n = 12 neurons; P = 0.0006; data not shown). These results are consistent with the fact that intracellular LCFA-CoA per se does not directly activate K_ATP channels in POMC neurons.

β-Oxidation of OA regulates POMC neuron excitability

Mitochondrial metabolism of LCFAs leads to an increase in [ATP] in the cell (Fig. 3F). Elevated intracellular [ATP] following β-oxidation in the cell would result in closure of K_ATP channels and consequent membrane depolarization. We thus evaluated this possibility by inhibition of CPT1. We used both an endogenous inhibitor of CPT1 (i.e., malonyl-CoA, 20 μM; for review, see Dowell et al. 2005) and an exogenous synthetic inhibitor of CPT1a (i.e., ST1326, 100 nM; Obici et al. 2003; Pocai et al. 2006). Pre-treatment with malonyl-CoA (20 μM) alone did not change membrane potential of POMC neurons (Fig. 3G). Following treatment with malonyl-CoA for 5 min, OA no longer excited POMC neurons (Fig. 3, G and I; 20 μM: malonyl-CoA: −63 ± 3 mV; malonyl-CoA + OA: −65 ± 3 mV; n = 8). Likewise, pre-exposure to the CPT1a inhibitor ST1326 (100 nM) completely blocked OA-induced depolarization without any changes in membrane potential of POMC neurons (Fig. 3, H and I; ST1326: −58 ± 3 mV; ST1326 + OA: −60 ± 4 mV; n = 7 neurons).

Oleic acid excites intact POMC neurons

We further examined the effect of OA under more physiological conditions, in which the intracellular environment remains intact by performing cell-attached recordings. Figure 4 shows cell-firing activity in cell-attached voltage-clamp mode (Perkins 2006). Under these experimental conditions, OA significantly increased the spike rate of POMC neurons as observed in whole cell recordings (control: 1.4 ± 0.3 Hz, OA: 1.7 ± 0.4 Hz; n = 13 neurons; P = 0.01; Fig. 4A). Interestingly, we found that the percentage of the POMC neurons that respond to OA is similar to that observed in whole cell patch-clamp recordings (i.e., cell-attached vs. whole cell: 44 vs. 38%). We next investigated the effect of the CPT1 inhibitor on OA-induced depolarization. Treatment with malonyl-Co-A alone reduced, but not significantly, the spike rate of POMC neurons from 1.3 ± 0.3 to 0.9 ± 0.3 Hz (n = 13; Fig. 4B). Following treatment with malonyl-CoA, addition of OA no longer changed the firing rate of POMC neurons (MA alone: 0.9 ± 0.3 Hz vs. MA + OA: 0.8 ± 0.3 Hz; n = 13; P = 0.39; Fig. 4B).

FIG. 4.

Oleic acid excites intact POMC neurons. A1 and A2: cell-attached voltage-clamp recording from POMC neurons with and without OA. Recording pipette is voltage-clamped at 0 mV. OA (50 μM) increased the spike rate of intact POMC neurons (A1). A2: histogram recorded from the neuron shown in A1 shows the time course of spikes before, during, and after application of OA. B1 and B2: sample recording of spikes before, during, and after addition of malonyl-CoA (MA) and MA + OA. Malonyl-CoA (20 μM) alone decreased the spike rate of POMC neurons. In the presence of MA, OA no longer excited POMC neurons (B1). B2 represents the time course of spike frequency recorded from the neuron shown in B1 before, during, and after application of MA alone and MA + OA. C1 and C2: representative cell-attached recording sample of the electrophysiological effects of glucose and OA. Lowering [glucose] from 5 to 2.5 mM decreased the spike rate of POMC neurons in about 60% of the neurons examined (n = 8). Under these experimental conditions, OA increased the frequency of spikes. C2: histogram shows the time course of the frequency of spikes recorded from the neuron shown in C1. D1 and D2: perforated patch-clamp recording sample from POMC neurons with and without OA. Likewise, OA increased the firing rate of POMC neurons. V_m = −49 mV. E1 and E2: representative perforated patch recording sample of APs before, during, and after addition of malonyl-CoA (MA) and MA + OA. OA no longer increased the firing rate following treatment with MA. V_m = −57 mV. The vehicle control for OA perfusion was 0.1% EtOH.

We then lowered the glucose concentration from 5 to 2.5 mM and then tested the effect of OA. Under these conditions, we found that, as described in prior work (Claret et al. 2007; Ibrahim et al. 2003; Parton et al. 2007; van den Top et al. 2007), roughly 60% of the POMC neurons examined are glucose-excited neurons. Lowering the glucose concentration from 5 to 2.5 mM decreased the spike rate of POMC neurons (5 mM: 2.4 ± 0.4 Hz; 2.5 mM: 1.1 ± 0.3 Hz; n = 16 neurons; P = 0.04; Fig. 4C). In the presence of 2.5 mM glucose, OA (50 μM) significantly increased the firing rate (from 1.1 ± 0.3 to 1.53 ± 0.3 Hz; n = 16 neurons; P = 0.04; Fig. 4C). However, it seems that the depolarizing effect of OA is independent of the nature of glucose effect on POMC neurons because OA increased the spike rate of the POMC neurons that are insensitive to glucose as well. Interestingly, the percentage of the POMC neurons responding to OA is higher in 2.5 than that in 5 mM glucose (i.e., 44 vs. 64%).

We also performed perforated-patch recordings to further support that intact POMC neurons are modulated by OA. Oleic acid (50 μM) increased the frequency of AP discharge from 1.4 ± 0.4 to 1.8 ± 0.5 Hz (n = 11 neurons; P = 0.02) and depolarized the membrane potential from −44 ± 3 to −39 ± 2 mV (n = 11 neurons; P = 0.001). The increased frequency was completely abolished by treatment with malonyl-CoA (20 μM; MA: 1.7 ± 0.4 Hz; MA + OA: 1.5 ± 0.4 Hz; n = 9 neurons). Additionally, we noted that malonyl-CoA alone decreased, but not significantly, the frequency of AP discharge (control: 2.6 ± 0.3 Hz; MA: 1.7 ± 0.4 Hz; n = 9 neurons). All the data we collected from cell-attached and perforated-patch recordings thus confirm our data obtained from whole cell recordings and support our hypothesis that elevated ATP concentration produced from OA through mitochondrial β-oxidation inactivates K_ATP channels, causing membrane depolarization.

Oleic acid action is both cell and substrate specific

To examine whether the depolarizing effect of OA was mimicked by any substrate of oxidation, we tested the effect of a short-chain fatty acid (i.e., octanoic acid: C8). We found that treatment with octanoic acid (50 μM) had no effect on the electrical activity of POMC neurons under the same experimental conditions (i.e., in the presence of a cocktail of glutamate, GABA_A, and glycine receptor blockers and 1 mM [ATP]_internal: Fig. 5, A and B; V_m: before: −62 ± 4 mV; after: −63 ± 4 mV; n = 11 neurons). The same concentration (i.e., 50 μM) of octanoic acid may provide very different numbers of calories to the POMC neurons. Given that the calories themselves may be an important part of the regulation of nutrient-related neurons, we also tested the effect of a high concentration (i.e., 100 μM) of octanoic acid. Raising octanoic acid concentrations from 50 to 100 μM induced no depolarization in the POMC neurons examined (Fig. 5B; before: −61 ± 2.5 mV; after: −61.6 ± 2 mV; n = 14), consistent with the fact that the depolarizing effect of OA is specific and that CPT1-mediated β-oxidation of OA, but not CPT1-independent β-oxidation of octanoic acid, regulates K_ATP channels in POMC neurons.

We also investigated whether the OA-induced depolarizing effect is cell specific using transgenic mice that express GFP only in agouti-related peptide (AgRP) neurons in the arcuate nucleus (Xu et al. 2005) (Fig. 5, C and D). We found that, in contrast to the direct effect of OA on POMC neurons, the excitability of AgRP neurons was not altered by treatment with OA (50 μM) under the same experimental conditions (i.e., in the presence of a cocktail of glutamate, GABA_A, and glycine receptor blockers and 1 mM [ATP]_internal; Fig. 5D; control: V_m = −59.3 ± 5, plus OA: V_m = −58.5 ± 5; n = 8 neurons; P = 0.32).

High-fat feeding alters excitability of POMC neurons

We examined whether the POMC neuronal response to OA is regulated by dietary history. Two groups of mice were fed a standard diet or HFD (with 20% calories provided by carbohydrate, 20% by protein, 60% by fat) for 2 wk. Following 2 wk on the assigned diet regimen, we tested acute electrophysiological responses to OA. Initial experiments revealed that OA induced membrane depolarization in both groups ([ATP]_internal: 1 mM; Fig. 6A). The responses to OA thus appear to be the same in both groups.

FIG. 6.

The excitability of POMC neurons is regulated by the levels of nutrients. A: whole cell current-clamp recording of POMC neurons from animals fed a regular chow or a high-fat diet (HFD) in 1 mM [ATP]_internal. POMC neurons from 2 groups showed membrane hyperpolarization on access in whole cell mode in 1 mM [ATP]_internal. The responses to OA were the same in both groups. V_m = −52 and −50 mV, respectively. B: whole cell current-clamp recording of POMC neurons from animals fed regular chow or HFD in 5 mM [ATP]_internal. POMC neurons in a regular diet group showed spontaneous APs (left) without any hyperpolarization. Bottom: POMC neuron activity on the expanded timescale. V_m = −46 mV. Right: under the same experimental conditions, POMC neurons from animals fed HFD showed membrane hyperpolarization. After roughly 5 min in whole cell configuration, no AP discharge was observed. Bottom: POMC neuron activity on the expanded timescale. V_m = −47 mV. C: representative recording sample of the electrophysiological effect of OA in 5 mM [ATP]_internal. Treatment with OA (50 μM) was without effect. V_m = −48 mV (left). In contrast, OA depolarized POMC neurons and induced AP discharge (right). Bottom: POMC neuron response to OA on the expanded timescale. V_m = −47 mV. D: population data for the resting membrane potential of POMC neurons from animals fed a regular diet (RD) or HFD in 5 mM [ATP]_internal. No difference was observed in the resting membrane potential over a period of 5 min in the RD group, whereas V_m significantly hyperpolarized in POMC neurons from animals fed HFD. E: population data for membrane potential of POMC neurons following treatment with OA in 5 mM [ATP]_internal. OA depolarized POMC neurons in the HFD group, but not in the RD group. F: pooled data of the firing rate of POMC neurons following treatment with OA in 5 mM [ATP]_internal (RD: control: 3.1 ± 1 Hz, OA: 2.5 ± 0.7 Hz; n = 6; P = 0.34; HFD: control: 0 Hz, HFD: 2.5 ± 1.6 Hz; n = 5 neurons). The vehicle control for OA perfusion was 0.1% EtOH.

However, we found altered excitability of POMC neurons from animals fed HFD under experimental conditions where [ATP]_internal was high (i.e., 5 mM). POMC neurons from animals fed a regular diet showed spontaneous discharge of APs over a period of whole cell recording in 5 mM [ATP]_internal (Fig. 6B; n = 10 neurons). In contrast, POMC neurons from the HFD group showed membrane hyperpolarization on access in whole cell configuration in 40% of the neurons examined in the presence of 5 mM [ATP]_internal (Fig. 6B; n = 19 neurons tested). The mean resting membrane potential was shifted from −48 ± 1 to −55 ± 2 mV (P = 0.003) over a period of 5–10 min (Fig. 6D). Such hyperpolarization at 5 mM [ATP]_internal was never observed in neurons from animals fed regular chow. Furthermore, OA (50 μM) depolarized and induced AP discharge of POMC neurons in hypothalamic slices from mice fed HFD (Fig. 6, C–F; control: −61 ± 2 mV, plus OA: −55 ± 2 mV; n = 9 neurons; P = 0.01), whereas POMC neurons from animals fed the regular diet did not increase their rate of AP discharge with OA exposure at the same concentration of ATP (Fig. 6, C–F).

Insulin antagonizes the depolarizing effect of oleic acid

The observed membrane hyperpolarization in the presence of high levels of internal ATP following HFD feeding may be due, at least in part, to altered ATP sensitivity of K_ATP channel in POMC neurons. In fact, high-fat feeding increases levels of plasma insulin (Morgan et al. 2004; Wang et al. 2001). Interestingly, prior study has shown that insulin, but not leptin, regulates K_ATP channels in POMC neurons by decreasing their sensitivity to ATP via phosphatidylinositol-3,4,5-triphospate (PIP3) (Plum et al. 2006). Based on these previous studies, our initial experiments tested whether insulin prevents OA-induced depolarization by activation of K_ATP channels in POMC neurons. Because POMC neurons were silent at 1 mM [ATP]_internal (Fig. 1C), we monitored the firing rate of POMC neurons at relatively high levels of internal ATP (see Figs. 3C and 6B) and examined the effect of insulin (100 nM). Figure 7A shows a typical example of insulin action on the firing rate of POMC neurons at 5 mM [ATP]_internal. We found that insulin (100 nM) significantly decreased the firing rate of POMC neurons from 4.3 ± 0.5 to 3.1 ± 0.5 Hz (Fig. 7A; n = 9 neurons). We next examined whether insulin antagonizes OA action because insulin hyperpolarized rather than depolarized POMC neurons in our preparations. We found that pre-exposure to insulin (100 nM) completely blocked OA-induced depolarization (Fig. 7B). POMC neurons did not respond to OA following treatment with insulin (V_m: insulin alone: −67 ± 1 mV vs. insulin + OA: −67 ± 1 mV; n = 8).

FIG. 7.

Insulin blocks the depolarizing effect of OA. A: sample recording of membrane potential of POMC neuron before, during, and after application of insulin (100 nM) in the presence of 5 mM [ATP]_internal. Treatment with insulin decreased the firing rate of POMC neurons (top). Bottom: POMC neuronal activity on an expanded timescale. V_m = −45 mV. B: representative recording sample of the effect of insulin on OA-induced depolarization. Prior exposure to insulin (100 nM) completely abolished OA action. V_m = −47 mV. The vehicle control for OA perfusion was 0.1% EtOH.

DISCUSSION

Our present work provides direct electrophysiological evidence for cellular mechanisms underlying the anorexigenic effect of oleic acid in several sets of experimental results. First, we demonstrate that OA directly regulates the electrical activity of POMC neurons, enhancing the anorexigenic tone exerted by the melanocortinergic system. Second, we outline the cellular mechanisms underlying the depolarizing effect of OA. Elevated ATP concentration produced from OA through mitochondrial β-oxidation inactivates K_ATP channels, causing membrane depolarization (see Fig. 8). Studies of identified POMC neurons in SUR1 knockouts support the contribution of K_ATP channels. Third, we demonstrate that the effect of OA is not mimicked by a short-chain fatty acid. Thus although ATP production via β-oxidation of long-chain fatty acids is necessary for POMC neuronal modulation, other factors including CPT1 also appear to be necessary. Finally, we demonstrate that high-fat feeding alters POMC neuron excitability.

FIG. 8.

Schematic diagram of proposed OA action in POMC neurons. On entry into POMC neurons, OA is rapidly esterified to oleoyl-CoA. The transfer of oleoyl-CoA to the mitochondria requires CPT1 that is located on the outer mitochondrial membrane. This may be controlled by malonyl-CoA, which is largely derived from glycolysis (the formation of malonyl-CoA from acetyl-CoA is catalyzed by acetyl-CoA carboxylase [ACC]). Prior work has indeed demonstrated that glucose metabolism by the CNS increases the level of malonyl-CoA in the hypothalamus (Wolfgang et al. 2007). Elevated β-oxidation raises [ATP] in the cell and subsequently inhibits K_ATP channels. Inhibition of K_ATP channels increases POMC neuron excitability. High-fat feeding and associated hyperinsulinemia may lower ATP sensitivity of K_ATP channels in POMC neurons, causing altered excitability of POMC neurons. The regulation of K_ATP channels in POMC neurons by nutrient-related signaling molecules is essential for controlling food intake as well as endogenous glucose production.

Recently, Wolfgang et al. (2006) reported that the brain-specific CPT1c KO mice exhibited decreased food intake and body weight, consistent with the involvement of CPT1 in maintaining energy homeostasis. However, it has been shown that CPT1c did not appear to participate directly in fatty acid oxidation (Wolfgang et al. 2008), although CPT1c is an outer mitochondrial integral membrane protein (Dai et al. 2007) and binds malonyl-CoAs like CPT1a (liver isoform) and CPT1b (muscle isoform) (Wolfgang and Lane 2006). Based on these findings, it has been proposed that CPT1c may play a regulatory rather than a direct metabolic role in the CNS (Wolfgang et al. 2008). If this is the case in POMC neurons, the most likely explanation is that CPT1c may be implicated in the control of neuronal activity when oleic acid, but not octanoic acid, is present because the transfer of short-chain fatty acids to mitochondria is independent of CPT1. Interestingly, OA did not alter AgRP neuronal activity, although NPY/AgRP neurons also express K_ATP channels (Gyte et al. 2007). Although CPT1c is highly expressed in regions of the hypothalamus critical for the regulation of energy balance including the paraventricular nucleus, ventromedial hypothalamus, and the arcuate nucleus (Dai et al. 2007), only a subset of neurons in these nuclei expresses CPT1c and the phenotype of CPT1c-expressing neurons remains to be determined. It is possible that the AgRP neurons did not respond to OA because of the lack of CPT1c. Thus decreased AgRP and NPY mRNA expression by OA (Morgan et al. 2004) appears to be due, at least in part, to indirect effects of OA via OA-sensitive hypothalamic neuronal circuit or to neuronal activity–independent mechanisms.

K_ATP channels sense the fluctuation of [ATP]_internal and can be used to measure the submembrane ATP concentration (Gribble et al. 2000). The gating of K_ATP channels is mainly regulated by the ratio of ATP/ADP in the cell (Ashcroft and Gribble 1998). The observed inhibition of K_ATP channel can be interpreted by an increase in [ATP]_internal following treatment with OA. We should note that CPT1c does not catalyze acyl transfer from fatty acyl-CoA to carnitine. If POMC neurons express only CPT1c, but not other isoforms of CPT1 (i.e., CPT1a and CPT1b), we do not expect such inhibition of K_ATP channels. However, the anorexigenic and metabolic effects of OA were abolished by genetic and biochemical inhibition of CPT1a (Obici et al. 2003; Pocai et al. 2006). In addition, the endogenous CTP1 inhibitor malonyl-CoA alone decreased the excitability of POMC neurons in cell-attached configuration, consistent with the idea that there is de novo synthesis of LCFAs in the hypothalamus and that hypothalamic neurons may use fatty acids as an alternative fuel. It should be noted that CPT1c KO mice fed HFD are more susceptible to obesity, suggesting that CPT1c is protective against the effects of HFD on weight gain (Wolfgang et al. 2006). In line with these data, OA was still effective following high-fat feeding in our preparations. In contrast, short-term overfeeding abolishes hypothalamic lipid sensing (Morgan et al. 2004; Pocai et al. 2006). Thus we speculate that, depending on the availability of nutrients, CPT1a and/or CPT1c may control feeding-related neuronal activity. A defect in hypothalamic nutrient sensing by POMC neurons may contribute to the susceptibility to obesity and insulin resistance in response to voluntary overfeeding and excess adipose tissue expression.

Low levels of OA either depolarized or hyperpolarized the arcuate neurons, but only in about 10% of neurons examined (Wang et al. 2006). However, we observed OA action only at concentrations >5 μM and around 40% of neurons examined responded to OA with a membrane depolarization. This discrepancy may be explained, at least in part, by different experimental conditions including external glucose and OA concentrations used. Indeed, Wang et al. (2006) reported that OA regulates the arcuate neurons in a glucose-dependent manner. Importantly, glucose metabolism by the CNS increases the level of malonyl-CoA in the hypothalamus (Wolfgang et al. 2007). In this scenario, the balance between carbohydrates and LCFA concentrations may determine POMC neuronal activity. This may be supported by our findings that the percentage of the POMC neurons responding to OA is higher in 2.5 than that in 5 mM glucose (i.e., 64 vs. 44%). It thus appears that the effect of OA may depend, in part, on the availability of external glucose and that CPT1 in POMC neurons is essential for integrating carbohydrate and lipid signaling. Another difference is the absence of the inhibitory effect of OA. Because OA decreased food intake (Obici et al. 2002), we expected that OA had an inhibitory action on orexigenic AgRP neurons, but this was not the case in our preparations. It is well known that the arcuate neurons are heterogeneous and express diverse neuropeptides and neurotransmitters (Crown et al. 2007). Future studies are needed to determine which types of neurons are modulated and the cellular mechanisms underlying inhibitory versus excitatory actions of OA.

The inhibition of K_ATP channels by OA in POMC neurons may result in decreased food intake as well as reduced glucose production. Recent work has reported that in POMC cell-restricted PTEN (phosphatidylinositol-3,4,5-triphospate [PIP3] phosphatase) KO mice, POMC neurons showed a marked hyperpolarization associated with increased K_ATP channel activity, leading to hyperphagia (Plum et al. 2006). Moreover, a recent work of Parton et al. (2007) reported that the POMC neurons expressing mutant Kir6.2 that prevent ATP-mediated closure of KATP channels did not respond to raising the glucose concentration and that this genetic manipulation impaired the whole body response to a systemic glucose load. These prior data thus suggest that K_ATP channels are critical regulators of POMC neuronal activity and the regulation of food intake.

In our present work, the properties of K_ATP channels in POMC neurons appear to be altered following high-fat feeding because high levels of internal ATP were not able to inactivate K_ATP channels, causing membrane hyperpolarization. It has been shown that phosphatidylinositol phosphates control ATP inhibition of K_ATP channels by attenuating ATP sensitivity (Baukrowitz et al. 1998; Shyng and Nichols 1998). One of the major downstream pathways following activation of the insulin receptor involves the phosphorylation of PIP2 to PIP3 through the activation of PI3K (Backer et al. 1992). Importantly, enhanced PIP3 signaling in POMC neurons activates K_ATP channels (Plum et al. 2006). Thus one of the most likely explanations is that high-fat feeding and associated hyperinsulinemia reduce ATP sensitivity of K_ATP channels in POMC neurons. Alternatively, cytoskeletal remodeling may be an important contributor to the cellular signaling mechanisms of high-fat feeding. In fact, insulin's effect on K_ATP activity in isolated arcuate neurons was inhibited by the PI3K inhibitor as well as by the marine sponge toxin jasplakinolide, which binds to F-actin, resulting in its stabilization and prevention of depolymerization to its monomer G-actin (Mirshamsi et al. 2004). In this scenario, we expect that chronic hyperinsulinemia during high-fat feeding would remodel the cytoskeletal structure, rendering K_ATP channels less sensitive to ATP. Interestingly, a short-term high-fat diet increased the in vivo activity of CPT1 by a decrease in the levels of malonyl-CoA and the turnover of LCFA-CoAs in the mediobasal hypothalamus (Pocai et al. 2006). As a result, tonically increased levels of ATP via β-oxidation of LCFAs in POMC neurons may also alter the property of K_ATP channels.

In summary, our present data support the idea that the mitochondrial β-oxidation of oleic acid is a critical step for the regulation of the excitability of POMC neurons. The regulation of K_ATP channels in POMC neurons by both acute and long-term treatment with nutrients and nutrient-related hormones may contribute to the control of food intake and to maintenance of weight balance.

GRANTS

This work was supported by Skirball Institute for Nutrient Sensing and National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Grant DK-057621 to S. Chua and NIDDK/Albert Einstein College of Medicine/Diabetes Research and Training Center Grant P60 DK-020541, an award from the New York Obesity Research Center, and a junior faculty award from the American Diabetes Association to Y.-H. Jo.