Anesthetic-like Interaction of the Sleep-inducing Lipid Oleamide with Voltage-gated Sodium Channels in Mammalian Brain

[benzoyl-2,5-³H]Batrachotoxinin A 20-α-benzoate ([³H]BTX; specific activity 34 Ci/mmol) was purchased from NEN Life Science Products, Inc. (Boston, MA). The sodium channel activators veratridine and scorpion (Leiurus quinquestriatus ) venom were obtained from Sigma-Aldrich Canada Ltd. (Oakville, Ontario, Canada). The cis -9,10 and trans -9,10 isomers of oleamide were synthesized according to published procedures. 14,17All other chemicals and saline constituents for electrophysiologic studies were from Sigma (Poole, Dorset, United Kingdom) or Merck Ltd. (Poole).

Murine NIE115 neuroblastoma cells were supplied by Dr. D. E. Ray (MRC Toxicology Unit, University of Leicester, United Kingdom). Confluent cells were subcultured twice weekly and grown on small glass coverslips in Dulbecco’s modified Eagle medium (containing Glutamax-1; Gibco, Paisley, Renfrewshire, United Kingdom) supplemented with 10% fetal calf serum and 50 μg/ml gentamicin. Cells were incubated at 37°C in 5% CO²in triple-vented 35-mm culture dishes. In some experiments, 2% dimethyl sulfoxide was added to the growth medium, which increased Na⁺current density in the clamped somata but did not qualitatively alter oleamide sensitivity. Cells were selected for electrophysiologic experiments 24–36 h after plating.

Synaptoneurosomal fractions were isolated from the brains of male CD1 mice (20–25 g; Charles River Laboratories, St. Constance, Quebec, Canada) in sodium-free isolation saline (130 mm choline chloride, 5.4 mm KCl, 5.5 mm glucose, 0.8 mm MgSO⁴, and 50 mm HEPES, adjusted to pH 7.4 with Tris base) as previously described. 18Batrachotoxinin is a full agonist at the alkaloid neurotoxin recognition site on voltage-gated Na⁺channels, which functionally facilitates hyperpolarizing shifts in channel activation and markedly impairs channel inactivation. These properties are exploited by neurochemists to chemically elicit ion flux through the normally voltage-responsive ion channel. [³H]BTX binding was determined according to published methods 18,19with minor modifications. For equilibrium assays, binding medium (60 μl) consisting of isolation saline containing [³H]BTX (final concentration, 2.2–175 nm as required), scorpion venom (30 μg), together with bovine serum albumin and tetrodotoxin (final concentration, 1 mg/ml and 1 μm, respectively), was thoroughly mixed with the appropriate oleamide isomer (in 0.8 μl dimethyl sulfoxide) or dimethyl sulfoxide (solvent control). The scorpion venom promotes levels of specific binding to site II on the Na⁺channel macromolecule. The binding reaction was then initiated by adding synaptoneurosomes (100 μl; approximately 150 μg protein), and incubations were continued at 37°C for 50 min. To terminate the reaction, 3 ml of ice-cold wash saline (163 mm choline chloride, 1.8 mm CaCl², 0.8 mm MgSO⁴, and 5 mm HEPES, adjusted to pH 7.4 with Tris base) was added, and after brief vortexing, bound [³H]BTX was separated from free radioligand by rapid vacuum filtration through Whatman GF/C filters (Mandel Scientific, Guelph, Ontario, Canada). After two 5-ml washes with ice-cold saline, the radioactivity present on filters was quantitated using liquid scintillation counting. Nonspecific binding, determined in the presence of 0.25 mm veratridine, amounted to 17.9 ± 1.4% of total binding with [³H]BTX at 10 nm.

Potential effects of the sleep modulator on the time course of association of [³H]BTX with site 2 on the sodium channel complex were investigated by preincubating synaptoneurosomes with cOA (100 μm) or the solvent control in binding medium devoid of [³H]BTX for 10 min at 37°C . [³H]BTX (10 nm) was then added, and the increase in specific binding was monitored up to 30 min.

The effect of cOA on the kinetics of [³H]BTX:sodium channel complex dissociation was examined by equilibrating synaptoneurosomes with radioligand (10 nm) in binding medium for 50 min at 37°C. At this point, either veratridine (plus solvent control) or veratridine plus cOA (at 100 μm or 200 μm) was added and the incubations continued from 1–30 min before determination of specifically bound [³H]BTX-A.

Cells adhering to glass coverslips were transferred to a plexiglas recording chamber rigidly mounted on the stage of a Nikon diaphot inverted microscope (Nikon, Kingston, Surrey, UK). Cells were continually superfused with saline containing 140 mm NaCl, 5 mm KCl, 1.8 mm CaCl², 0.8 mm MgCl², 10 mm HEPES, pH 7.3, and supplemented with 0.1% dimethyl sulfoxide and 0.1% bovine serum albumin. Tetrodotoxin and cOA were applied quantitatively in this superfusing solution (flow rate, ∼5 ml/min; bath volume, < 0.5 ml). The whole-cell patch-clamp technique was used to voltage clamp the cells with a single microelectrode. A Burleigh (Harpenden, Herts, United Kingdom) PCS-5000 micromanipulator was used. Pipettes (3–4 MΩ) were fabricated from thin-walled borosilicate glass (1.6-mm OD; Hilgenberg-GMBH, Malsfeld, Germany) on a Mecanex electrode puller (Mecanex Ltd., Geneva, Switzerland). The pipette solution consisted of 10 mm NaCl, 20 mm TEA-Cl, 110 mm CsCl, 2 mm MgCl², 1 mm CaCl², 11 mm EGTA, 10 mm HEPES, pH 7.4. To reduce stray capacitance, the tips of the electrodes were silanized by immersion in Sigmacote (Sigma, Poole) after back-filling. Only spherical and unclumped cells were used in the study, and 75–90% series resistance compensation was applied directly from the Axopatch 200 preamplifier (Axon Instruments Ltd., San Diego, CA). Because I/V curves for the sodium channel shift to more negative potentials in the first few minutes after beginning cell dialysis, at least 6 min was allowed after membrane rupture before taking data. Analog data were filtered at 5 KHz and were digitized at 20–40 KHz using a CED1401 plus laboratory interface for analysis using patch and voltage-clamp software (CED, Cambridge, United Kingdom). Leak currents and residual capacitive artefacts were deducted offline in software. Details of, and rationale for, pulse protocols are given in Results or associated figure legends. All electrophysiologic experiments were conducted at room temperature (20–24°C). Conductance (g) through Na⁺channels was calculated by dividing the peak Na⁺current by the driving force (V-Er), where Er is the observed reversal potential in the clamped cell. Relative conductance was fitted to a Boltzmann function of the following form:

(g/g^max) = 1/(1 + exp[(V  ^1/2−V )/k]

where g^maxis the peak conductance, V^1/2is the voltage at which half-maximal conductance occurs, k is the slope factor, and V is the command voltage.

In binding studies, the inhibitor concentration which displaces 50% of specifically bound radioligand (IC⁵⁰) of cOA was calculated using the probit method. 20Effects of cOA on equilibrium binding parameters (K  ^dand binding maximum) of [³H]BTX were determined according to Bennett. 21Least squares regression analysis was used to fit lines to data as appropriate. Quantitative electrophysiologic data were analyzed using Microsoft Excel 97 (Microsoft Corp., Seattle, WA) and GraphPad Prism 3 software (San Diego, CA), and are cited as mean ± SEM (n). P  values were generated using the Student paired t  test.

cis -9,10-Octadecenoamide inhibited the specific binding of 10 nm [³H]BTX to mouse brain synaptoneurosomes in a concentration-dependent fashion and gave an IC⁵⁰of 39.5 μm (fig. 1). The trans  isomer of oleamide was considerably less effective, achieving 22% and 27% inhibition of [³H]BTX binding at 20 μm and 100 μm, respectively. The mechanism by which cOA interacts with site 2 of voltage-sensitive sodium channels was investigated using Scatchard analysis of [³H]BTX binding and by examining the effect of this sleep inducer on [³H]BTX binding kinetics. Figure 2shows the Scatchard analysis of the specific binding of [³H]BTX to synaptoneurosomes in the absence and presence of cOA. Regression analysis of these data demonstrated that the sleep modulator produces a fourfold increase in the K  ^dof [³H]BTX binding, without affecting the number of binding sites (binding maximum) available to the radioligand. Substances that reduce the association of [³H]BTX likely decrease binding by interacting with closed channels; however, no discernible affect of cOA on the time course of association of [³H]BTX with its binding site was detected (figs. 3A and B). The ability of cOA to interact with activated sodium channels in synaptoneurosomes was investigated by determining its effect on the dissociation of radioligand from the [³H]BTX:sodium channel complex. Application of either 100 or 200 μm cOA (combined with a saturating concentration of veratridine) to synaptoneurosomes previously equilibrated with [³H]BTX caused a concentration-related increase in the rate of dissociation of radioligand over and above that of veratridine alone (fig. 4).

All patch-clamped cells were found to express fast inward currents in response to step depolarization. From a holding potential of −80 or −100 mV, 0.1–1 μm tetrodotoxin produced complete block of evoked whole-cell inward currents (n = 6) as previously reported for this cell line in the ionic environment used. 22,23Preliminary experiments showed that oleamide exerted a reversible block of peak Na⁺currents (evoked at 0.1 Hz) but that the onset and recovery kinetics were relatively slow, confirming earlier observations in neural membranes (fig. 5). 16Subsequently, 12–15-min incubations were used after collection of control data to ensure equilibration of the lipid amide with cell membranes and its presumptive receptor site. Under these conditions, cOA caused tonic inhibition of Na⁺currents in response to a voltage step to 0 mV (0.5 Hz) in a voltage-dependent manner. cOA 10 μm resulted in a 40 ± 9% reduction in peak current at a holding potential of −60 mV, compared with 16 ± 5% at a holding potential of −100 mV (n = 4;P < 0.05;fig. 5A). The fractional block was concentration-dependent at the two holding potentials used (fig. 5C), and curves appeared to saturate at 64 μm. At negative holding potentials, the curve was displaced to the right (log of concentration required to block 20% of the peak current was increased significantly;P < 0.01; n = 4), increasing the EC²⁰from approximately 2 μm at −60 mV to approximately 10.5 μm at −100 mV. The maximal cOA fractional block was significantly enhanced at depolarized holding potentials (maximal block at −100 mV, 40.8 ± 0.02%; at −60 mV, 81.1 ± 0.02%; n = 7;P < 0.0001).

cis -9,10-Octadecenoamide 3.2–64 μm inhibited Na⁺currents in a manner that was largely independent of activation voltage (fig. 6). No significant shifts in reversal potential for the peak evoked currents were detected even at high cOA concentrations (P > 0.5; n = 4 at 10 μm and n = 6 at 32 μm). Conductance–voltage curves were virtually superimposable before and after application of cOA (order of administration was randomized in these experiments), as shown in figure 6D. Compounding data from replicated experiments confirmed that V⁵⁰values were not significantly altered by the lipid amide even at the relatively high concentration of 32 μm (P > 0.05; n = 11).

Steady state inactivation curves (prepulse of 90 ms between −120 mV and −20 mV before a test pulse to +10 mV) were fit to a single Boltzmann and showed a concentration-dependent hyperpolarizing shift in the presence of cOA. The maximum value of this shift in the V⁵⁰for inactivation was −15.4 ± 0.9 mV at 32 μm cOA (n = 6;fig. 7A). At concentrations greater than 3.2 μm, the negative shifts were significant (P < 0.05) and concentration-dependent. The magnitude of this shift was fitted to the Hill equation (fig. 7B), yielding an apparent EC⁵⁰of 11.6 μm (± log SEM of 0.057), a slope of 2.8 ± 1.0 (17 df ), and a tendency to saturate at 64 μm, which was the highest concentration we could formulate at room temperature (fig. 7B). Recovery from inactivation was studied by eliciting current with a 10-ms test pulse to 0 mV from a holding potential of −100 mV with increasing time intervals after a 100-ms prepulse between −100 and 0 mV (fig. 8A). Normalized current data were plotted against recovery interval and fit to a single exponential for each cell examined, confirming that 10 μm cOA slowed recovery from inactivation, with mean τ increasing from 3.7 ± 0.4 ms to 6.4 ± 0.5 ms (n = 6;P < 0.001). Compounded data for recovery from inactivation are shown in figure 8B. In control physiologic salines, the peak Na⁺current in these cells did not decrement significantly during a train of 50-ms stimulus pulses to evoke maximal currents at 0.5–10 Hz (not shown). Under these conditions, 10–32 μm cOA did not show detectable frequency-dependent facilitation of block (fig. 9) at frequencies up to 10 Hz (P > 0.1; n = 5 at 10 μm). Visual inspection of data at other concentrations revealed no frequency dependent fade, but these data were not subjected to statistical analysis.

Our results demonstrate that cOA is capable of inhibiting the binding of [³H]BTX to synaptoneurosomes, providing the first evidence for an interaction of this natural sleep-inducing substance with site 2 of voltage-gated sodium channels in mammalian brain. 24Modulation of voltage-sensitive sodium channels was proposed in our previous report, 16which described antagonism of a number of veratridine-dependent responses in functional synaptic preparations and suppression of sustained repetitive firing in pyramidal cells by cOA at low micromolar concentrations. The potency of cOA as an inhibitor in the [³H]BTX binding assay (IC⁵⁰= 39.5 μm) compares favorably with these results, given the differences in the level of functional complexity and the different requirements of these assays. In addition, marked inhibitory effects on [³H]BTX binding were only observed with the strongly hypnotic cis  isomer. Thus, it appears that perturbation of the [³H]BTX binding domain on voltage-gated sodium channels is mechanistically relevant to the sodium channel blocking action of this sleep inducer.

cis -9,10-Octadecenoamide caused a decrease in the affinity of sites labeled by [³H]BTX without influencing the total number of sites accessed by this radioligand. Our Scatchard data therefore support the idea of competitive inhibition of [³H]BTX binding by this sleep modulator. In addition, because the kinetic analyses revealed that 100 μm cOA was unable to affect the association rate of radioligand but accelerated its dissociation, we infer that an indirect allosteric inhibitory mechanism is responsible for the reduction in affinity. In addition, the dissociation experiments suggest that cOA may inhibit [³H]BTX binding by interacting preferentially with sodium channels to which [³H]BTX and scorpion venom are already bound (i.e. , an allosterically modified conformation of the channel), whereas our failure to slow the association of this radioligand with cOA likely excludes its binding to BTX–free (resting or able to be activated) channels. 25Overall, electrophysiologic and binding data suggest a preferential association with inactivated or toxin-modified channel conformations.

The inhibitory profile of cOA on alkaloid neurotoxin binding to sodium channels shows an interesting parallel with those reported for the class I antiarrhythmics, 26anticonvulsants, 27and both local and general anesthetics. 19,28The mechanism of competitive inhibition of [³H]BTX binding by cOA we describe is evidently distinct from the unsaturated insecticidal N -alkylamides, which bear some structural similarity to fatty acid amides but reduce the affinity of voltage-gated sodium channels for [³H]BTX by decreasing its rate of association. 29 

The negative shifts in inactivation curves and slowing of recovery from inactivation are entirely consistent with those exerted by depressant drugs, including anticonvulsants or local and general anesthetic molecules. 30,31These effects are likely to underpin the highly isomer-selective effects of the lipid on burst firing 16we previously reported in current-clamped rat neurones. The suppression of sustained repetitive firing by cis - (but not trans -) oleamide was strikingly similar to the effects of the anticonvulsant lamotrigine in similar experiments on the cultured cortical cells. 32 

Experiments to confirm the isomer selectivity of oleamide interactions with voltage-clamped cloned Na⁺channels are now underway. We could see no distinctive effect of cOA on channel activation kinetics or voltage dependence of activation gating, nor did the compound influence transmembrane gradients for Na⁺secondary to indirect effects on membrane transport or ionic buffering. The neuroprotective agent riluzole has a similar mode of action, and, in vivo , such drugs are strongly depressant at high concentration and have been described as novel anesthetics. 33Furthermore, sedation has long been an undesirable side effect of anticonvulsant drugs in a clinical setting. 34 

The tonic blocking effects noted were not noticeably augmented by high-frequency stimulation of the channel complex. Frequency-dependent fade or open-channel block is a feature of many clinically useful drugs, which may reflect, in part, access of charged molecules at physiologic pH to an intracellular blocking site via  the polar lumen of the open channel. Molecules such as quinidine are notable for their use dependence, which reflects cumulative enhanced binding of drugs to open channels and failure to unbind completely between pulses. 35The latter form of frequency dependence has been correlated with aromaticity in uncharged drug molecules. 35Oleamide has a calculated log P of approximately 6.5 and, despite its amphiphilic character, bears no charge or aromatic residue. These physicochemical properties may result in relatively rapid off kinetics (fast unbinding, compared with rates of oleamide association with channel proteins) and explain the lack of use dependence in our experiments. The slow onset and recovery times for functional effects are presumably a reflection of partitioning of the extremely hydrophobic molecule into the membrane plane adjacent to its site of action at the channel. In the neuroblastoma cells, the molecule appears to promote inactivation or stabilize the inactivated conformation of the channel. The binding data suggest an interaction of cOA with site 2 of the sodium channel complex. However, the kinetics of modulation of oleamide with the channels in both biochemical and electrophysiologic experiments is notably slow, and the binding interaction with site 2 is indirect. We cannot exclude the possibility that this involves liberation of cytoplasmic messengers and reflects an allosteric effect secondary to changes in phosphorylation of the voltage-gated channel complex. In independent studies, cOA is unable to modulate the binding of guanosine 5′-O-(3-thiotriphosphate) to neural membranes at the concentrations used here, suggesting that G-protein–linked receptors per se  are not targets for the hypnogenic lipid. 10 

Our experiments demonstrate that oleamide has the capacity to concurrently exert inhibitory effects on presynaptic Na⁺channels and postsynaptic GABA^Areceptors. Both channels are widespread in the central nervous system (CNS), and both are crucial for arousal. For these reasons, the effects reported here are consistent with CNS depression and may contribute to the hypnogenic effects of the molecule. However, both of the ion channel targets appear to respond to levels of oleamide above those measured in the cerebrospinal fluid of sleep-deprived or “sleepy” laboratory animals (0.4 μm). 2At first glance, this questions the physiologic relevance of the reciprocal modulation proposed here. However, this line of argument does not take into account the scope for physiologic synergy between Na⁺channels involved in fast axonic conduction in virtually all CNS neurones and the most prevalent inhibitory channel in the brain. 36,37More importantly, it ignores the physical properties of oleamide and the widespread distribution of fatty acid amide hydrolase enzymes in the brain. 38Oleamide has a log P of 6.5; because it is made in biologic membranes and avidly degraded there, it may not be accurate to assume that concentrations in bulk cerebrospinal fluid reflect those seen at the membrane plane at localized sites of action or in key sleep nuclei. Yost et al.  15disclosed that exogenous oleamide is not anesthetic, nor does it synergize anesthetics in laboratory animals. A more recent study demonstrated that, like anesthetics, intraperitoneal oleamide potentiated the sedative action of diazepam and sodium pentobarbital. 39However, the in vivo  actions of the synthetic molecule are likely to be limited by hepatic and serum hydrolases and uptake into non-neural lipid compartments, hence reducing effective concentrations in the CNS. The in vivo  effects of oleamide in rodents were very controversial until it was appreciated that its formulation was crucial (lipid-containing vehicles are essential for hypnogenic and psychomotor effects), suggesting that oleagenous excipents may quench or retard the substantial metabolism of oleamide in the periphery and CNS. 2 

It is an intriguing possibility that oleamide may represent an endogenous ligand for depressant drug binding sites in mammalian CNS. Our results suggest that the hypnogenic isomer has a similar molecular mode of action to therapeutically important and widely prescribed drugs used globally for the treatment of anxiety, sleep disorders, epilepsy, and even to induce anaesthesia in the operating room. 14,40The pioneering work of Hughes et al.  41unequivocally established the presence of endogenous ligands for opiate pathways in mammalian nervous systems. Several candidates for benzodiazepine sites have been proposed, but flumazenil does not have marked anxiogenic or stimulatory effects in animals or patients, suggesting that we may not have a constitutively active endogenous benzodiazepine ligand. 42Increasingly selective receptor agonists and antagonists are important tools in the quest for endogenous neurohumoral agents and have recently been used to identify anandamide as an endogenous cannabinoid. 43Our studies demonstrate that, like barbiturates, general anesthetics, and anticonvulsants, oleamide can concurrently exert depressant effects on voltage-gated Na⁺channels and the GABA^Areceptor family. In the absence of selective antagonists for such drugs (or for oleamide), careful pharmacologic analysis both in vitro  and in vivo  will be required to examine the exciting proposal that fatty acid amides can physiologically activate depressant sites on these crucial ion channels and directly regulate sleep and arousal.

The authors thank Helen Jackson (Postgraduate Research Technician, Institute of Pharmacy and Chemistry, University of Sunderland) for expertly maintaining cultured neuroblastoma, which were kindly supplied by Dr. D. E. Ray (University of Leicester, United Kingdom).