Identification of Propofol Binding Sites in a Nicotinic Acetylcholine Receptor with a Photoreactive Propofol Analog*

Background: Propofol, a general anesthetic, potentiates GABAA receptors and inhibits nAChRs by binding to unknown sites. Results: Photoaffinity labeling with a photoreactive propofol analog is used to identify propofol binding sites in a muscle-type nAChR. Conclusion: Propofol binds to an intrasubunit site in the nAChR transmembrane domain. Significance: This study helps define the diversity of allosteric modulator binding sites in Cys-loop neurotransmitter receptors. Propofol, a widely used intravenous general anesthetic, acts at anesthetic concentrations as a positive allosteric modulator of γ-aminobutyric acid type A receptors and at higher concentration as an inhibitor of nicotinic acetylcholine receptors (nAChRs). Here, we characterize propofol binding sites in a muscle-type nAChR by use of a photoreactive analog of propofol, 2-isopropyl-5-[3-(trifluoromethyl)-3H-diazirin-3-yl]phenol (AziPm). Based upon radioligand binding assays, AziPm stabilized the Torpedo nAChR in the resting state, whereas propofol stabilized the desensitized state. nAChR-rich membranes were photolabeled with [3H]AziPm, and labeled amino acids were identified by Edman degradation. [3H]AziPm binds at three sites within the nAChR transmembrane domain: (i) an intrasubunit site in the δ subunit helix bundle, photolabeling in the nAChR desensitized state (+agonist) δM2-18′ and two residues in δM1 (δPhe-232 and δCys-236); (ii) in the ion channel, photolabeling in the nAChR resting, closed channel state (−agonist) amino acids in the M2 helices (αM2-6′, βM2-6′ and -13′, and δM2-13′) that line the channel lumen (with photolabeling reduced by >90% in the desensitized state); and (iii) at the γ-α interface, photolabeling αM2-10′. Propofol enhanced [3H]AziPm photolabeling at αM2-10′. Propofol inhibited [3H]AziPm photolabeling within the δ subunit helix bundle at lower concentrations (IC50 = 40 μm) than it inhibited ion channel photolabeling (IC50 = 125 μm). These results identify for the first time a single intrasubunit propofol binding site in the nAChR transmembrane domain and suggest that this is the functionally relevant inhibitory binding site.

Propofol, a potent intravenous general anesthetic, acts as a positive allosteric modulator of inhibitory ␥-aminobutyric acid receptors (GABA A Rs), 2 and this interaction is a major determinant of the anesthetic potency of propofol (1)(2)(3). In contrast, propofol inhibits excitatory nicotinic acetylcholine receptors (nAChRs) (4,5), which are also members of the Cys-loop superfamily of pentameric ligand-gated ion channels, as well as GLIC, a prokaryotic homolog that is a proton-gated cation channel (6). Identification of propofol binding sites in GABA A Rs and nAChRs is necessary to determine whether propofol binds to equivalent or distinct sites in Cys-loop receptors when it produces opposing effects as a positive or negative allosteric modulator.
Cys-loop ligand-gated ion channels each consist of five identical or homologous subunits that associate around a central axis forming the ion channel (7,8). Models of three-dimensional structures of these receptors have been derived from a cryoelectron microscopy structure of the Torpedo (muscletype, ␣ 2 ␤␥␦) nAChR (9) and from recently determined high resolution crystal structures of the homopentameric receptors GluCl, an invertebrate glutamate-gated chloride channel (10), and GLIC (11,12). Each subunit contains an N-terminal extracellular domain, a transmembrane domain made up of a loose bundle of four transmembrane helices (M1-M4), and an intracellular domain formed by the primary structure between the M3 and M4 helices. The M2 helices from each of the five subunits form the ion channel and the M1, M3, and M4 helices form an outer ring partly exposed to lipid.
Although there has been no direct identification of propofol binding sites in a GABA A R or nAChR, within GLIC crystals, propofol binds within an intrasubunit pocket formed by the four transmembrane helices (6). Within GABA A R, photoreactive derivatives of etomidate, another intravenous anesthetic, identified an intersubunit etomidate binding site in the transmembrane domain at the interface between ␤ and ␣ subunits, the same interface that contains the GABA binding site in the extracellular domain (13,14). Within the Torpedo nAChR, etomidate analogs have been shown to bind to three distinct sites in the transmembrane domain: the ion channel, the ␥-␣ subunit interface, and the ␦ subunit helix bundle (15,16). Each of these sites is a potential binding pocket for propofol, a potent inhibitor of the Torpedo nAChR expressed in oocytes (IC 50 ϭ 7 M (17)).
In this study, we used 2-isopropyl-5-[3-(trifluoromethyl)-3H-diazirin-3-yl]phenol (AziPm), a recently developed photoreactive propofol analog (18), to identify propofol binding sites in the Torpedo nAChR. AziPm acts as a tadpole anesthetic similar in potency to propofol, and it potentiates GABA A R responses at anesthetic concentrations (18). Although AziPm is related structurally to propofol (see Fig. 1), it incorporates a photoreactive 3-trifluoromethyl-3-phenyldiazirine group, which can react with most amino acid side chains, including aliphatic side chains (19). AziPm also is related in structure to TID (Fig. 1), a potent Torpedo nAChR inhibitor (20) that has been shown by photoaffinity labeling to bind in a state-dependent manner to sites in the ion channel (21) and in the ␦ subunit helix bundle (22,23). Based upon competition radioligand binding assays, we found that propofol binds to the nAChR with higher affinity in the desensitized state than in the resting state, whereas AziPm binds with higher affinity in the resting state. Based upon the amino acids photolabeled by [ 3 H]AziPm, we show that it binds to three sites in the nAChR transmembrane domain: within the ␦ subunit helix bundle, in the ion channel, and at the ␥-␣ interface. Propofol binds to the site within the ␦ helix bundle as shown by the full inhibition of [ 3 H]AziPm photolabeling there. Propofol acts as an allosteric inhibitor of [ 3 H]AziPm photolabeling in the ion channel and potentiates [ 3 H]AziPm photolabeling at the ␥-␣ interface.

EXPERIMENTAL PROCEDURES
Materials-Torpedo nAChR-rich membranes, purified from Torpedo californica electric organs (Aquatic Research Consultants, San Pedro, CA) as described (24), contained from 1.5 to 1.7 nmol of [ 3 H]ACh binding sites/mg of protein. Non-radioactive AziPm was synthesized as described (18) where f(x) is the total radioligand binding (in cpm) at competitor concentration x, B 0 is the total radioligand bound in the absence of competitor, B ∞ is the residual radioligand binding at high concentrations of competitor, IC 50 is the drug concentration inhibiting radioligand binding by 50%, and n H is the Hill coefficient. For all fits with the exception of the propofol inhibition of [ 3 H]PCP binding, B ∞ was set equal to the nonspecific radioligand binding (B ns ) determined in the presence of 1 mM Carb, 0.1 mM tetracaine, or 0.1 mM PCP.
Photolabeling nAChR-rich Membranes-Experiments were performed using ϳ100 g or 10 mg of protein/condition for analytical and preparative scale photolabelings. For all experiments, Torpedo nAChR-rich membranes were resuspended at 2 mg of protein/ml in Torpedo physiological saline supplemented with 1 mM oxidized glutathione, an aqueous scavenger.  methanol and then resuspended for 30 min at a final concentration of ϳ1 M in the nAChR-rich membrane suspension before addition of other drugs. This was followed by an additional 30-min incubation on ice. Membrane suspensions were irradiated on ice for 30 min using a 365 nm UV lamp (Model EN-16, Spectronics Corp., Westbury, NJ) at a distance of 0.5 cm. Preparative scale photolabeling was performed under three paired experimental conditions to identify photolabeled amino acids: 1) in the resting state versus desensitized states (ϮCarb), 2) in the absence versus presence of 300 M propofol (without agonist), and 3) in the desensitized state (ϩCarb) in the absence versus presence of propofol.
SDS-Polyacrylamide Gel Electrophoresis and Proteolytic Digestions-For photolabeling experiments on an analytical scale, nAChR subunits were resolved after photolysis by Trisglycine SDS-PAGE on 8% acrylamide, 0.32% bisacrylamide gels that were stained with Coomassie Blue R-250. Samples were run in duplicate on two gels, one of which was prepared for fluorography using Amplify (Amersham Biosciences), and the other of which was used for measurement of 3 H incorporation into individual subunits quantified by liquid scintillation counting of excised gel bands.
rpHPLC and Sequence Analysis-nAChR subunit proteolytic fragments were fractionated on an Agilent 1100 binary reversed phase HPLC system using a C 4 Aquapore 7-m (100 ϫ 2.1mm) column with an upstream C 2 guard column at 40°C. Peptide elution was monitored by absorbance at 215 nm. Aqueous trifluoroacetic acid (0.08%) was used as solvent A, and acetonitrile (60%), 2-propanol (40%), trifluoroacetic acid (0.05%) was used as solvent B. A non-linear elution gradient increasing from 25 to 100% solvent B in 80 min was used at a flow rate of 200 l/min with fractions of 0.5 ml collected.
N-terminal sequencing was performed on an Applied Biosystems Procise 492 protein sequencer with 1 ⁄ 6 of the material from each cycle used to quantify the amount of PTH-amino acid derivative obtained and 5 ⁄ 6 collected to measure the 3 H released. Pooled HPLC fractions containing 3 H-labeled peptides were drop-loaded at 45°C onto Applied Biosystems Micro TFA glass fiber filters except for fractions containing ␣M4 or ␦M1, which because of their low sequencing efficiency when loaded on glass fiber filters were loaded onto PVDF filters using Prosorb sample preparation cartridges. All filters were then treated with BiobrenePlus. To chemically isolate during sequence analysis the nAChR subunit fragments beginning at ␤Thr-273, ␥Thr-276, or ␦Thr-281 within the M2-M3 loops (28), sequencing filters were treated with o-phthalaldehyde prior to cycle 6 of Edman degradation at which point fragments contain a proline. o-Phthalaldehyde reacts with primary amines but not with proline, a secondary amine. This treatment prevents further sequencing of other fragments on the filter that do not contain a proline at that cycle (24,29).
The efficiency of photolabeling of amino acid residues within a sequenced fragment was calculated from the increase in 3 H release at that cycle of Edman degradation (cpm x Ϫ cpm (x Ϫ 1) ) and the increased mass of the PTH-amino acid released in cycle x. The masses of the released PTH-amino acids were quantified from their peak heights relative to the peak heights of PTHamino acid standards and then fit by non-linear least squares (SigmaPlot 11) to Equation 2, where f(x) is the mass of the peptide residue in cycle x (in pmol), I 0 is the initial amount of peptide (in pmol), and R is the average repetitive yield, a parameter dependent upon the amount of sample washed off and the efficiency of peptide bond cleavage at each cycle of Edman degradation. PTH derivatives whose amounts cannot be accurately measured (Cys, His, Ser, Trp, and Arg) were omitted from this fit. The efficiency of photolabeling of amino acid residues (cpm/pmol) was then calculated using Equation 3.
Molecular Modeling-A model of T. californica nAChR structure based on the cryoelectron microscopy structure of the Torpedo marmorata nAChR (Protein Data Bank code 2BG9 (9)) was used to dock, using the Discovery Studio (Accelrys, Inc.) software package, AziPm molecules within (i) the ion channel, (ii) the ␦ subunit helix bundle, and (iii) the ␥-␣ subunit interface. We docked AziPm using the CHARMm-based molecular dynamics simulated annealing program CDOCKER. A randomly oriented AziPm molecule was seeded at the center of binding site spheres positioned between M2-6Ј and M2-13Ј within the ion channel (r ϭ 10 Å); within the pocket formed by ␦M2-18Ј, ␦Phe-232, and ␦Cys-236 in the ␦ subunit helix bundle (r ϭ 8 Å); or at the level of ␣M2-10Ј in the ␥-␣ subunit interface (r ϭ 8 Å). The CDOCKER program was set up to first seed 50 randomly distributed replicas of AziPm in the center of the spheres, and high temperature molecular dynamics was used to generate 50 random conformations for each replica. For each conformation, 400 energy-minimized docking solutions were generated using simulated annealing and full potential minimization. Connolly surface representations defined by a 1.4-Ådiameter probe of the ensemble of the 15 docking solutions with the lowest CDOCKER interaction energies are shown for each of the binding sites.

Propofol and AziPm Inhibition of [ 3 H]ACh, [ 3 H]Tetracaine, and [ 3 H]PCP Equilibrium Binding-In initial experiments, we compared the effects of propofol and AziPm on the equilibrium binding of [ 3 H]ACh to
Torpedo nAChR-membranes ( Fig. 2A). For nAChR-rich membranes in the presence of agonist at equilibrium, Ͼ99% of nAChRs are in the equilibrium desensitized state that binds ACh with highest affinity, whereas in the absence of agonist, ϳ85% of nAChRs are in the resting, closed channel state, and ϳ15% are in the desensitized state (30,31). Those conformations are interconvertible, and allosteric modulators such as PCP and proadifen increase the fraction of nAChRs in the desensitized state in the absence of agonist (stabilize the desensitized state) and enhance ACh equilibrium binding affinity (32,33). We measured [ 3 H]ACh binding at a concentration sufficient to occupy ϳ20% of sites, a condition chosen to detect either increases or decreases of ACh binding affinity. Under this condition, proadifen, a desensitizing aromatic amine non-competitive antagonist (33), maximally increased [ 3 H]ACh binding by 30%. Propofol at concentrations above 10 M also increased [ 3 H]ACh binding, producing a maximal increase of 20% at 300 M, with inhibition seen at 1 mM. In contrast, AziPm only inhibited [ 3 H]ACh binding with a steep concentration dependence (IC 50 ϭ 134 Ϯ 13 M; Hill coefficient, n H ϭ 2.5 Ϯ 0.5) ( Fig. 2A).
We also examined the effects of propofol and AziPm on the equilibrium binding of nAChR channel blockers and found that propofol bound with higher affinity to the nAChR in the presence of agonist, whereas AziPm bound preferentially to the nAChR in the absence of agonist.  [ 3 H]PCP binding by only 74 Ϯ 3% (two experiments) compared with the full inhibition produced by excess non-radioactive PCP. However, the concentration dependence was consistent with a single site model (IC 50 ϭ 47 Ϯ 5 M, n H ϭ 1.09 Ϯ 0.09). This partial inhibition of binding indicates an allosteric rather than competitive mechanism of inhibition such that when propofol is bound to the nAChR the [ 3 H]PCP dissociation constant increases ϳ4-fold.
In the absence of agonist, [ 3 H]AziPm was incorporated into each nAChR subunit with the ␦ subunit labeled most efficiently. Tetracaine reduced ␣ and ␤ subunit photolabeling by 40% and ␦ subunit labeling by 60%. Because tetracaine binds within the ion channel, this inhibition suggests that [ 3 H]AziPm photolabels residues within the ion channel. Propofol also inhibited labeling of the ␦ subunit by 60%, but it increased ␣ subunit incorporation by 20% (Fig. 3B). For nAChRs in the absence of agonist, the 1,900 cpm tetracaineor propofol-inhibitable 3 H incorporation in the nAChR ␦ subunit indicated photolabeling at 25 cpm/pmol of ␦ subunit (ϳ0.5% of subunits).
In the presence compared with the absence of Carb, nAChR subunit photolabeling was reduced by ϳ20%. Propofol inhibited ␦ subunit photolabeling by ϳ50%, whereas PCP reduced it by only 20%. As in the absence of agonist, propofol increased ␣ subunit labeling by ϳ20% in the presence of Carb.
To provide an initial localization of the photolabeled amino acids within the ␣ subunit, we characterized the distribution of [ 3 H]AziPm incorporation within the large, non-overlapping fragments of the ␣ subunit that can be generated by "in-gel" digestion with S. aureus V8 protease: fragments of 20 kDa (␣V8-20; beginning at ␣Ser-173 and containing ACh binding site Segment C and the M1-M3 transmembrane helices), 18 kDa (␣V8-18; beginning at ␣Thr-52 and containing ACh binding site Segments A and B), and 10 kDa (␣V8-10; beginning at ␣Asn-339 and containing the cytoplasmic MA and transmembrane M4 helices) (9,25). For the nAChR photolabeled in the resting state, 59, 3, and 38% of 3 H were recovered in ␣V8-20, ␣V8-18, and ␣V8-10, respectively (Fig. 3C). The pharmacologically sensitive photolabeling within the ␣ subunit was restricted to ␣V8-20 where tetracaine and Carb inhibited photolabeling by 60 and 40%, respectively, and propofol increased labeling by ϳ15%.  1-7). The mobilities of the ␣, ␤, ␥, and ␦ nAChR subunits, rapsyn (Rsn), and the Na ϩ /K ϩ -ATPase ␣ subunit (␣ NaK ) are indicated. B, gel bands containing nAChR ␣, ␤, ␥, and ␦ nAChR subunits and Na ϩ /K ϩ -ATPase ␣ subunit were excised from duplicate gels, and 3 H incorporation was determined by liquid scintillation counting. For each gel band, the average cpm Ϯ S.D. (error bars) are plotted. C, nAChR ␣ subunits were isolated from nAChRrich membranes (400 g of protein/ϳ650 pmol of ACh binding sites) photolabeled with [ 3 H]AziPm in the absence of other drugs (Control) or in the presence of tetracaine (100 M), propofol (300 M), or Carb (1 mM). The isolated ␣ subunit gel bands were loaded onto a second 15% polyacrylamide gel for in-gel digestion with V8 protease to produce four subunit fragments that were visualized by staining the gel with GelCode Blue (Pierce). The 3 H incorporation within the excised gel bands was determined by liquid scintillation counting. The locations of the four subunit fragments within the ␣ subunit primary structure are indicated above the graph.

Propofol Inhibition of [ 3 H]AziPm Photolabeling in the ␦
Subunit-To further characterize the effects of propofol on [ 3 H]AziPm photoincorporation, we examined the concentration dependence of propofol inhibition of ␦ subunit 3 H incorporation in nAChRs photolabeled in the absence and presence of Carb (Fig. 4). In the presence of agonist, propofol inhibition of [ 3 H]AziPm ␦ subunit labeling was consistent with a simple, single site model (n H ϭ 1) with an IC 50 of 37 Ϯ 8 M, which was similar to the IC 50 of 47 M for the inhibition of [ 3 H]PCP binding in the desensitized state (Fig. 2C). In the absence of agonist, propofol inhibited [ 3 H]AziPm ␦ subunit photolabeling with a concentration dependence (IC 50 ϭ 140 Ϯ 30 M, n H ϭ 2) similar to that seen for propofol inhibition of [ 3 H]tetracaine binding in the ion channel (IC 50 ϭ 125 Ϯ 14 M, n H ϭ 1.6 Ϯ 0.3) (Fig. 2B).
[ 3 H]AziPm and Propofol Bind in the nAChR ␦ Subunit Helix Bundle in the Desensitized State-In the presence of agonist, propofol inhibited ␦ subunit photolabeling to a greater extent than PCP, suggesting that propofol inhibits [ 3 H]AziPm photolabeling of a site other than the ion channel. The ␦ subunit helix bundle pocket was a likely candidate because [ 125 I]TID, a drug related in structure to AziPm (Fig. 1), binds within this pocket in the nAChR open and desensitized states, photolabeling amino acids in ␦M1 (␦Phe-232 and ␦Cys-236), ␦M2 (␦Thr-274 (␦M2-18Ј)), and ␦M2-␦M3 loop (␦Ile-288) (22,23). To determine whether there was propofol-inhibitable [ 3 H]AziPm photolabeling of this site, we isolated and sequenced fragments beginning near the N termini of ␦M2 and ␦M1 from nAChRs photolabeled in the presence of Carb Ϯ 300 M propofol. Sequence analysis of the ␦M2 fragment (Fig. 5A) revealed peaks of 3 H release in cycles 18 and 21, which indicated photolabeling of ␦Thr-274 and ␦Arg-277, residues positioned within the ␦ subunit helix bundle (9). Propofol inhibited labeling of these residues by 80 and 75%, respectively. N-terminal sequencing through ␦M1 (Fig. 5B)    of Carb, the fragment beginning at ␦Phe-206 was the primary sequence (I 0 ϭ 7 pmol, both conditions (Ⅺ)) with a secondary sequence beginning at ␦Met-257 (I 0 ϭ 0.5 pmol; not shown). For the ϩCarb sample, the peaks of 3 H release in cycles 27 and 31 indicate photolabeling of ␦Phe-232 (10 cpm/pmol) and ␦Cys-236 (3.2 cpm/pmol) that was reduced by Ͼ90% in the absence of Carb. For the ϪCarb sample, the peak of 3 H release in cycle 13 resulted from photolabeling of ␦Val-269 (␦M2-13Ј) in the secondary sequence that was reduced by Ͼ90% in the presence of agonist (see Fig.  7).
When photolabeling was performed in the absence of Carb, no 3 H incorporation was detected in ␦Phe-232, ␦Cys-236, or ␦Thr-274, indicating that if labeling occurred it was at Ͻ5% of the level seen in the presence of Carb (Fig. 5C). We also characterized [ 3 H]AziPm photolabeling in the ␦M2-M3 loop by isolating and sequencing a fragment beginning at ␦Thr-281. In the presence of Carb, photolabeling of ␦Ile-288, if it occurred, was at less than 3% of the level of ␦Phe-232 (data not shown).

H]AziPm Photolabeling in the M2 Ion Channel Domain Is Inhibited by Agonist (nAChR Desensitization) or by Propofol-[ 3 H]AziPm photoincorporation into ␣M2-6Ј
provided evidence that AziPm binds within the ion channel and that this binding is sensitive to agonist and to propofol. To further characterize [ 3 H]AziPm photolabeling within the ion channel, fragments beginning at the N termini of ␤M2 (␤Met-249) and ␦M2 (␦Met-257) were also isolated and sequenced from nAChRs photolabeled with [ 3 H]AziPm (i) in the absence and presence of Carb to compare photolabeling in the resting and desensitized states and (ii) in the absence and presence of 300 M propofol. When the fragment beginning at ␤Met-249 was sequenced, the peaks of 3 H release in cycles 6 and 13 of Edman degradation established labeling of ␤Ser-254 (␤M2-6Ј) and ␤Val-261 (␤M2-13Ј) that was inhibited Ͼ90% by agonist (Fig. 7A) and ϳ90% by propofol (Fig. 7B). When the fragment beginning at ␦Met-257 was sequenced, the peak of 3 H release in cycle 13 established labeling of ␦Val-269 (␦M2-13Ј) that agonist inhibited by Ͼ90% (Fig. 7C) and propofol inhibited by 85% (Fig. 7D). Because positions M2-2Ј, -6Ј, -9Ј, -13Ј, -16Ј, and -20Ј line the lumen of the ion channel (in all nAChR subunits and Cys-loop receptors (8)), our results demonstrate that [ 3 H]AziPm binds in the ion channel in the resting state.
[ 3 H]AziPm Photolabeling within ␣M4-Within the nAChR transmembrane domain, the M4 helices have the greatest exposure to lipid (9), and photolabeling with [ 125 I]TID identified five amino acids forming a strip on the ␣M4 helix at the lipid interface (35). We characterized [ 3 H]AziPm labeling within ␣M4 by sequencing a fragment beginning at ␣Tyr-401 isolated from tryptic digests of ␣V8-10. Peaks of 3 H release in cycles 12 and 18 of Edman degradation (Fig. 8C) established agonist-insensitive labeling of ␣Cys-412 and ␣Cys-418, the amino acids photolabeled most efficiently by [ 125 I]TID. When we sequenced fragments isolated from nAChRs photolabeled in the presence of 300 M propofol, we found that [ 3 H]AziPm photolabeling of ␣Cys-412 and ␣Cys-418 was also insensitive to propofol (Table 1).

DISCUSSION
Propofol has been reported to be a potent inhibitor of ␣4␤2 neuronal (IC 50 ϭ 5 M) and Torpedo nAChRs (IC 50 ϭ 7 M) but not mouse muscle nAChR (IC 50 ϭ 45 M) (5,17). In this study, we identified the Torpedo nAChR binding sites for [ 3 H]AziPm, a novel photoreactive propofol analog, and we used inhibition of [ 3 H]AziPm photolabeling to identify propofol binding sites. We summarize in Table 1 the photoincorporation efficiency (cpm/pmol) and pharmacological specificity for the labeled amino acids. Based upon the locations of the photolabeled res-idues in the structure of the nAChR (Fig. 9), we found that [ 3 H]AziPm binds to three distinct sites within the nAChR transmembrane domain: 1) within the ␦ subunit helix bundle in the desensitized state (photolabeling ␦M2-18Ј and ␦Phe-232/ ␦Cys-236 in ␦M1), 2) in the ion channel in the resting state (photolabeling ␣M2-6Ј, ␤M2-6Ј, ␤M2-13Ј, and ␦M2-13Ј), and 3) at the ␥-␣ and/or ␤-␣ transmembrane interface (photolabeling of ␣M2-10Ј). Within the nAChR structural model, computational docking calculations predict that AziPm can bind to each of the three sites, and the predicted binding modes are shown in Fig. 9 in Connolly surface representation.    (Table 1). Surprisingly, propofol inhibition of [ 3 H]PCP binding in the ion channel was characterized by essentially the same concentration dependence (IC 50 ϭ 47 Ϯ 5 M, n H ϭ 1.09 Ϯ 0.09), mediated by an allosteric mechanism, consistent with an ϳ4-fold reduction of [ 3 H]PCP affinity. Although it is possible that propofol allosterically inhibits [ 3 H]PCP binding by occupying a site in the ion channel with the same affinity as for the site in the ␦ subunit helix bundle, a more plausible interpretation would suggest that propofol occupying the site in the ␦ helix bundle results in an allosteric inhibition of [ 3 H]PCP binding in the ion channel. Consistent with this interpretation, for the nAChR in the desensitized state, non-radioactive TID acts as a competitive inhibitor of [ 3 H]PCP binding (20), and PCP fully inhibits [ 125 I]TID photolabeling at positions M2-2Ј and -6Ј in the ion channel. At the same time, PCP acts as an allosteric inhibitor of [ 125 I]TID photolabeling in the ␦ subunit helix bundle, reducing photolabeling by ϳ40% (23).
Does Propofol Bind in the nAChR Ion Channel?-AziPm binds in the ion channel, and propofol inhibits [ 3 H]AziPm channel photolabeling. However, this inhibition occurs by an allosteric, rather than a competitive, mechanism. Based upon equilibrium radioligand binding assays, propofol binds to the Torpedo nAChR preferentially in the desensitized state, whereas AziPm binds preferentially in the resting, closed channel state (Fig. 2)  H]TDBzl-etomidate photolabeling of ␣M2-10Ј was enhanced in the presence of the desensitizing channel blockers PCP and proadifen. This binding site for etomidate derivatives was located at the ␥-␣ interface based upon the photolabeling of ␥Met-295/␥Met-299. In the absence of photolabeling of similar amino acids in ␥M3 or ␤M3, we do not know whether [ 3 H]AziPm binds at the ␥-␣ or ␤-␣ interface (or both). In contrast to [ 3 H]AziPm, [ 125 I]TID did not photolabel ␣M2-10Ј, ␥Met-295, or ␥Met-299, although it photolabeled ␥Asn-300 and other amino acids in the M3 helices that are exposed to lipid (35).
Propofol Binding Sites in Cys-loop Receptors-Propofol acts as an inhibitor of muscle and neuronal nAChRs and of GLIC, and our results identify an intrasubunit propofol binding site in the Torpedo AChR transmembrane domain equivalent to the site identified in GLIC (6). It remains to be determined whether propofol binds to an equivalent intrasubunit binding site in the GABA A receptor where it acts as a positive allosteric modulator. For the Torpedo nAChR, the positive allosteric modulator TDBzl-etomidate does not bind within the pocket formed by the ␦ subunit helix bundle. Instead, it binds to an intersubunit site at the interface between ␥ and ␣ subunits (15). However, for the homopentameric ␣7 nAChR, mutational analyses provide evidence for an intrasubunit binding site for positive allosteric modulators within the helix bundle pocket (39,40).
General anesthetics of diverse chemical structure act as GABA A R positive allosteric modulators, and the results of early mutational analyses were interpreted in terms of structural models that predicted intrasubunit binding sites for alcohols and volatile anesthetics as well as intravenous anesthetics, including propofol and etomidate (41). However, improved GABA A R homology models indicate that many of the positions determining anesthetic sensitivity do not project within an intrasubunit pocket (42). In addition, photoreactive etomidate analogs identify an intersubunit etomidate binding site at the interface between ␤ and ␣ subunits that contains the GABA binding site in the extracellular domain (13,14), and in crystals of GluCl, a homopentameric invertebrate glutamate-gated chloride channel, the positive allosteric modulator ivermectin is bound at this intersubunit binding site (10). With the availability of [ 3 H]AziPm, which reacts broadly with aliphatic and nucleophilic side chains and identifies propofol binding sites in the Torpedo nAChR, and other recently developed propofol . The photolabeled amino acids are represented in stick format (i) within the ␦ subunit helix bundle (in green; ␦Thr-274, ␦Phe-232, and ␦Cys-236), (ii) within the ion channel (in cyan; ␣Ser-248, ␤Ser-254, ␤Val-261, and ␦Val-269), and (iii) within the ␥-␣ interface (in red; ␣Ser-252). C, enlarged side view of the nAChR transmembrane domain with the ␥ and ␣ ␦ subunits and the M4 helices of the ␣ ␥ , ␤, and ␦ subunits removed for better visibility of the photolabeled amino acids and the AziPm binding pockets in the ␦ subunit helix bundle and in the ion channel. analogs (15), it may soon be possible to identify propofol binding sites in GABA A Rs.