Tryptophan Fluorescence Reveals Conformational Changes in the Acetylcholine Binding Protein*

The recent characterization of an acetylcholine binding protein (AChBP) from the fresh water snail, Lymnaea stagnalis, shows it to be a structural homolog of the extracellular domain of the nicotinic acetylcholine receptor (nAChR). To ascertain whether the AChBP exhibits the recognition properties and functional states of the nAChR, we have expressed the protein in milligram quantities from a synthetic cDNA transfected into human embryonic kidney (HEK) cells. The protein secreted into the medium shows a pentameric rosette structure with ligand stoichiometry approximating five sites per pentamer. Surprisingly, binding of acetylcholine, selective agonists, and antagonists ranging from small alkaloids to larger peptides results in substantial quenching of the intrinsic tryptophan fluorescence. Using stopped-flow techniques, we demonstrate rapid rates of association and dissociation of agonists and slow rates for the α-neurotoxins. Since agonist binding occurs in millisecond time frames, and the α-neurotoxins may induce a distinct conformational state for the AChBP-toxin complex, the snail protein shows many of the properties expected for receptor recognition of interacting ligands. Thus, the marked tryptophan quenching not only documents the importance of aromatic residues in ligand recognition, but establishes that the AChBP will be a useful functional as well as structural surrogate of the nicotinic receptor.

Ligand-gated ion channels, of which the nicotinic acetylcholine receptor is a prototypic structure, are composed of five subunits whose ␣-carbon chains traverse the membrane four times (1,2); their hydrophobicity and size preclude conventional structural studies at atomic resolution by x-ray crystal-lography or nuclear magnetic resonance spectrometry. Recently, an acetylcholine binding protein (AChBP) 1 from the fresh water snail, Lymnaea stagnalis, has been characterized, crystallized, and its structure determined (3,4). The crystal structure shows virtually all of the features predicted from a host of affinity labeling, site-specific mutagenesis, and subunit assembly studies conducted on the nicotinic receptor for over 2 decades (1,2,5). Although the isolated protein shares ligand recognition characteristics with its closest mammalian homolog, the pentameric ␣7 receptor (4), details on its ligand specificity, binding kinetics, and conformational changes remain unknown. These questions are critical to ascertaining whether the snail protein has the recognition properties and conformational states to serve as a functional as well as a structural surrogate of the extracellular domain of the nicotinic receptor. To this end, we have expressed the binding protein in a mammalian system from a chemically synthesized cDNA of 637 bp. The cDNA contains restriction sites at various locations to allow for substitution of encoding receptor segments into the cDNA template of the binding protein. Upon ligand binding, AChBP shows major changes in fluorescence emitted from five tryptophans on each subunit, providing an intrinsic detection system to monitor the stoichiometry and kinetics of ligand binding.

EXPERIMENTAL PROCEDURES
Gene Synthesis and Protein Expression-We synthesized seven double-stranded oligonucleotides between 80 and 126 bp reflecting codon usage in mammalian cells and containing appropriate overhangs for ligation (6). These were assembled into three ligation products that were then inserted into construction vectors and their sequence confirmed by automated sequencing. After digestion with appropriate restriction enzymes followed by band isolation, the inserts were ligated into a p3ϫFLAG-CMV-9 expression vector (Sigma) containing a preprotrypsin leader peptide followed by a N-terminal 3ϫFLAG epitope. 2 The expression plasmid also contained neomycin acetyltransferase for clonal selection. Transfection of the plasmid into human embryonic kidney cells produced an epitope-attached glycoprotein secreted primarily into the medium. Selection with G418 yielded a stable cell line secreting AChBP. Ultraculture medium (Bio Whittaker) was collected at 3-day intervals from multitier flasks for several weeks. Adsorption onto a FLAG antibody column followed by elution with the 3ϫFLAG peptide yielded purified protein in quantities between 1 and 2.5 mg/liter.
Fluorescence Assays-Fluorescence measurements were performed on a Jobin Yvon-Spex Fluoromax 2 fluorometer (Instruments S. A., Inc., Edison, NJ). AChBP was excited at 280 nm and emission intensity monitored for 0.1-s intervals at unitary wavelengths between 337 and 343 nm.
Stopped-flow Kinetics-Stopped-flow measurements were obtained using an Applied Photophysics SX.18MV (Leatherhead, UK) stoppedflow spectrofluorometer. Excitation was at 280 nm, and a cut-off filter at 305 nm was used to collect the fluorescence signal. Measurements of binding of the dansyl choline analogues employed 280 nm excitation and measured the enhanced fluorescence using a 420-nm cut-off filter. Rates of association and dissociation were estimated from the slope and ordinate intercept of plots of overall rate of fluorescence change versus ligand concentration. Dissociation rates were also estimated in several cases by reacting the preformed complex with a large excess of gal-* This work was supported by NIGMS Grants R37-GM 18360 and T32-GM 07752 and The National Center for Research Resources, National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

RESULTS AND DISCUSSION
Purification of AChBP secreted into the medium produced a single band on SDS gels migrating at ϳ35 kDa (Fig. 1a). Treatment with PNGase to remove N-linked oligosaccharides enhances the migration rate considerably. N-terminal sequencing and matrix-assisted laser desorption ionization mass spectrometry after deglycosylation yielded a sequence and mass consistent with cleavage of the leader peptide. Negative staining electron microscopy showed typical rosette structures expected of a pentameric subunit assembly ( Fig. 1, b and c). Hydrodynamic analysis revealed a Stokes radius of 57 Å from gel filtration and a sedimentation coefficient of 4.9 S from sucrose density gradients; values were also consistent with pentamer formation.
Stoichiometry of ligand binding was estimated from AChBP tryptophan quenching through titration by high affinity ligands ( Fig. 2). In separate preparations, this yielded values of 4.7-5.6 mol/mol of pentamer or 0.94 to 1.1 mol/mol of 26,551dalton subunit based on quantitative amino acid analysis. These data show with epibatidine as a high affinity ligand ϳ50% quenching of the intrinsic tryptophan fluorescence, a typical quenching value for most of the quaternary and tertiary amines used in this study. Since these ligands, with the exception of the dansyl choline derivatives, lack the spectral overlap with tryptophan emission necessary for fluorescence resonance energy transfer (FRET) (7), the precise mechanism of quenching is unknown. However, at least two (Trp-53 and Trp-143) and perhaps a third (Trp-82) of the five tryptophans are found in proximity to the bound ligand (3). The organic cation in fitting into this aromatic nest of tryptophans and tyrosines may disrupt aromatic connectivity established between the side chains (8). Upon binding of ligands lacking the capacity for FRET, tryptophan fluorescence quenching has also been observed for acetylcholinesterase (9); its binding site is at the base of a narrow gorge whose base and walls are lined with aromatic residues (10). In the muscle and ␣7 nAChR, residue substitution of tryptophans homologous to residues 53 and 143 reduces ligand affinity, but does not eliminate binding (11,12), and a charge-transfer complex involving one or more tryptophans has been proposed to stabilize various ligand complexes (13,14). Systematic substitution of other aromatic residues for the five tryptophans should enable one to delineate further their individual contributions to quenching of fluorescence.
Stopped-flow kinetic studies of ligand association are shown in Fig. 3; data for several ligands, monitored by AChBP fluorescence quenching, are tabulated in Table I. Rates of association and dissociation approach the time resolution of the stopped-flow technique (ϳ1 ms). A comparison of agonists shows that the bimolecular rates of agonist binding approach the diffusion limitation and the rate constants found in single channel or voltage-current relaxation analyses for the muscle or neuronal nicotinic receptor (15)(16)(17)(18)(19)(20). Rates of dissociation are slower than dissociation of ligand from the activatible receptor, but of nearly the same magnitude as dissociation from the open channel state, and more rapid than dissociation from the presumed desensitized state (15)(16)(17)(18)(19)(20). Hence, agonists bind and dissociate with rates expected from electrophysiologic studies for an open channel state of the receptor. In fact, certain residues in the transmembrane span of the receptor have been shown to enhance dissociation of ligands from the activatible, closed channel state of the receptor (21). Moreover, the AChBP structure may resemble more closely the open channel conformation of the Torpedo acetylcholine receptor (22).
Given the proposed role for AChBP in scavenging the neurotransmitter in synapses (4), an association rate for acetylcholine similar to that found for the receptor would be expected to achieve efficient function. Also, the kinetic constants for the antagonist, d-tubocurarine, are of comparable magnitude with those found electrophysiologically for the receptor (23). No evidence for appreciable cooperativity of binding was found in the kinetic profiles.
Epibatidine and ␣-bungarotoxin have equilibrium dissociation constants for AChBP that differ by only an order of magnitude, yet the rate of association for the peptide antagonist is ϳ200-fold slower than for the nicotinic agonist (Table I). If the association reactions for ␣-bungarotoxin are run at higher concentrations, two steps in the reaction are evident with the rapid step being bimolecular-and concentration-dependent. The unimolecular step (k 2 ϭ 0.34 s Ϫ1 ) shows a small enhancement in fluorescence that diminishes, but does not eliminate, the over- all fluorescence quenching. The slower ␣-neurotoxin kinetics of association, which also has been well documented in nicotinic receptors from several species (24 -26), and the linked unimolecular step seen here are suggestive of the ␣-bungarotoxin locking the AChBP into a distinctive conformational state. The single tryptophan in ␣-bungarotoxin has ϳ2% of the fluorescence intensity of the five tryptophans in each AChBP subunit. The fluorescence change in the slow step could emerge from enhanced fluorescence of the single tryptophan of the toxin in the complex or a slight enhancement of the AChBP tryptophans after formation of the intial complex. Irrespective of the tryptophans contributing to the signal differences, the slower unimolecular isomerization points to differing positions of the tryptophans in the initial complex and the final equilibrium state.
The tris-quaternary antagonist, gallamine, when associated with the receptor, results in an enhancement of the tryptophan fluorescence, suggesting that the stabilization of this ligand may differ from the other agonists and antagonists studied. The three triethylammonioethyl groups that emanate from the pyrogallol ring probably preclude full insertion of the ring into the aromatic pocket. Rather stabilization involving the quaternary ammonium ligands and anionic moieties at the subunit interface account for the different binding orientation of gallamine, resulting in fluorescence enhancement (Fig. 4). Since gallamine binding appears mutually exclusive with the other agonists and antagonists listed in Table I (3) showing the ␣ carbon chain and the space filling residues for two subunits. The exposed portion of the docked gallamine is shown in green, tryptophan side chains in yellow, and selected anionic side chains in red. B, expanded view of the ␣ carbon chain with the tryptophan side chains in yellow, tyrosine in blue, and anionic residues in red. Note the positioning of the triethylammonio moieties near the anionic side chains (Glu-110, Glu-149, Glu-163, Glu-190; Asp-108) at the binding site. The pyrogallol ring is sandwiched between the vicinal cystines at 187 and 188 and isoleucine 112 side chain on the neighboring subunit. The position of gallamine docked by computation may be contrasted with HEPES found in the crystal structure (3). of the various antagonists (Table I), as well as measuring the stoichiometry of ligand binding for ligands that quench to a lesser extent than epibatidine (Fig. 2).
Quite apart from establishing ligand specificity for the AChBP, intrinsic tryptophan fluorescence quenching affords a universal means of directly following ligand binding to the AChBP without the necessity of developing competition studies with radioactive or fluorescent ligands. Substitution of sequences unique to particular receptor subtypes may allow one to examine selectivity of various AChBP-receptor chimeric sequences fashioned after the neuronal or muscle subtypes of receptor. The ligand binding kinetics seen with the AChBP reveals similarities to kinetics anticipated for ligand binding to the many subtypes of nicotinic receptor. Physical measurements in solution should enable one to correlate conformation with kinetic parameters of ligand recognition, and add another dimension to investigating specificity of this unique binding protein in relation to the larger family of receptor-related offspring.