If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* This work was supported by the Canadian Institutes of Health Research and a Canadian Graduate Scholarship (to C. J. B. D.). The on-line version of this article (available at http://www.jbc.org) contains supplemental Experimental Procedures, references, Figs. S1–S6, and Tables S1–S3.
Lipids influence the ability of Cys-loop receptors to gate open in response to neurotransmitter binding, but the underlying mechanisms are poorly understood. With the nicotinic acetylcholine receptor (nAChR) from Torpedo, current models suggest that lipids modulate the natural equilibrium between resting and desensitized conformations. We show that the lipid-inactivated nAChR is not desensitized, instead it adopts a novel conformation where the allosteric coupling between its neurotransmitter-binding sites and transmembrane pore is lost. The uncoupling is accompanied by an unmasking of previously buried residues, suggesting weakened association between structurally intact agonist-binding and transmembrane domains. These data combined with the extensive literature on Cys-loop receptor-lipid interactions suggest that the M4 transmembrane helix plays a key role as a lipid-sensor, translating bilayer properties into altered nAChR function.
Neurotransmission at chemical synapses is fundamental to the propagation of electrical signals within the nervous system. Central to this process is the ability of Cys-loop receptors to convert a chemical input into an electrical output by conducting ions across the synaptic membrane in response to neurotransmitter binding (
). At the molecular level, this not only requires the ability to bind agonist and conduct ions, but also the ability to effectively translate agonist binding into ion channel opening/gating. The agonist-binding sites, which are located on the extra-membranous surface of the receptor, are thus allosterically coupled to the distant transmembrane ion pore (Fig. 1). Factors that affect the ability of Cys-loop receptors to bind agonist, conduct ions, and/or couple agonist binding to ion channel gating have the potential to modulate the synaptic response, and thus influence the transmission of electrical signals.
One factor that is known to affect the activity of several Cys-loop receptors is the lipid composition of the membrane in which they are embedded (
). Initial studies showed that to preserve a fully functional nAChR, the receptor must be purified in the presence of exogenous lipid and then reconstituted into a membrane with a particular lipid composition (
). The consensus is that both cholesterol (Chol) and anionic lipids (such as phosphatidic acid; PA) are required in a reconstituted phosphatidylcholine (PC) membrane to provide an optimal environment. Chol and PA both increase the proportion of nAChRs stabilized in an agonist-activatable conformation (
). As the chemical labeling pattern of the non-activatable PC-nAChR is similar to that of the desensitized nAChR, it has been suggested that lipids modulate the natural equilibrium between resting and desensitized receptors (
). The lack of detailed characterization of the inactive PC-nAChR conformation has prevented the development of descriptive models of nAChR-lipid interactions. It is this lack of definitive insight into the lipid-dependent nAChR conformations that is addressed here.
As a first step toward understanding how lipids influence function, we characterize here the activatable and non-activatable conformations of the Torpedo nAChR stabilized in PC/PA/Chol (PC/PA/Chol-nAChR) and PC (PC-nAChR) membranes. We show that lipid-dependent inactivation is not related to agonist-induced desensitization. Instead, PC-nAChR adopts a novel conformation in which allosteric coupling between the agonist-binding sites and transmembrane pore is lost (Fig. 1D). Furthermore, uncoupling leads to a substantial increase in solvent accessibility, with minimal effects on nAChR secondary structure and thermal stability. Together, our data show that the lipid environment surrounding the nAChR transmembrane domain influences communication between the intact agonist-binding and transmembrane pore domains. In the context of recent structural and mutagenesis data, our findings suggest that the lipid-exposed transmembrane M4 helix acts as a lipid-sensor modulating interactions at a coupling interface between the two domains. The existence of this novel uncoupled conformation could explain how membrane-soluble allosteric modulators (including lipids) influence Cys-loop receptor function.
Preparation of nAChR Membranes
Native nAChR-enriched Torpedo membranes (i.e. native-nAChR) and affinity purified and reconstituted membranes of defined lipid composition were prepared as described previously (
) were performed essentially as described elsewhere. The fluorescence of ethidium bromide emission was measured at 590 nm with 500 nm excitation. [3H]Acetylcholine equilibrium binding was determined using an ultrafiltration assay described elsewhere (
). The same ultrafiltration protocol was used for measuring the total number of accessible [125I]α-bungarotoxin sites. See supplemental Experimental Procedures for a detailed description of all protocols.
Agonist-induced Conformational Change
Affinity purified nAChR, reconstituted into proteoliposomes with different lipid compositions (supplemental Fig. S1), exhibits varying abilities to respond to agonist. Infrared difference spectroscopy (Fig. 2) is an effective tool for probing conformational change in these membrane environments because the technique provides a comprehensive map of the vibrational, and thus structural changes induced in the entire nAChR upon agonist binding (
). Carbamylcholine (Carb) difference spectra recorded from native-nAChR, PC/PA/Chol-nAChR, and PC-nAChR all exhibit a number of positive vibrational bands due to nAChR-bound Carb (asterisk in Fig. 2A). The difference spectra also include vibrations near 1620 and 1514 cm−1, which reflect quaternary amine-aromatic interactions, and near 1690 cm−1, which reflects an interaction between the Carb ester carbonyl and an unidentified binding site residue (
). The presence of these (and other) vibrations indicates that Carb binds to the nAChR with a similar overall pattern of recognition in all three membrane environments. Membrane lipid composition has no effect on agonist-binding site integrity.
Carb difference spectra recorded from native-nAChR and PC/PA/Chol-nAChR also exhibit relatively intense positive amide vibrations near 1655 and 1547 cm−1 (shaded regions in Fig. 2A), which reflect Carb-induced structural changes in the polypeptide backbone. Because the positive intensity at these frequencies is lost when Carb difference spectra are recorded in the presence of the desensitizing local anesthetic, dibucaine (i.e. the nAChR is desensitized prior to Carb binding; Fig. 2B), we conclude that these positive vibrations result from the Carb-induced resting-to-desensitized conformational transition (
). Positive intensity is also absent at these frequencies (i.e. 1655 and 1547 cm−1) in difference spectra recorded from PC-nAChR, whether or not dibucaine is present. The difference spectra show that whereas native-nAChR and PC/PA/Chol-nAChR both undergo a Carb-induced resting-to-desensitized conformational transition, PC-nAChR does not. Why is PC-nAChR unresponsive to agonist? PC-nAChR may be unresponsive because it is already stabilized in the desensitized state (
To test whether PC-nAChR is desensitized, we first examined the fluorescence of ethidium binding to the nAChR in each of the three membrane environments (Figs. 2, C, and D, and supplemental S2). Ethidium binds with high affinity to a hydrophobic site within the ion channel pore of the desensitized (Kd = ∼0.3 μm), but not the resting (Kd = ∼1 mm) nAChR (
). Relative to ethidium in aqueous solution, the bound ethidium exhibits a greater fluorescence emission intensity, and its emission maximum shifts from 605 (aqueous solution) to 590 nm (nAChR-bound).
Addition of either native- or PC/PA/Chol-nAChR to an aqueous solution of ethidium leads to an apparent increase in fluorescence intensity (Fig. 2C), but this increase is not dibucaine displaceable and can be attributed to the scattering of incident light by the membrane vesicles (supplemental Fig. S2). The low ethidium fluorescence observed in the absence of agonist shows that the ion pore in both membrane environments is stabilized in a conformation with low affinity for ethidium, an affinity suggestive of the resting state. The addition of 500 μm Carb, however, leads to a substantial increase in dibucaine-displaceable ethidium fluorescence, accompanied by a shift in the emission maximum to 590 nm (Figs. 2C and supplemental S2). Both the increase in fluorescence and the shift in emission maximum indicate that Carb binding to the nAChR in native and PC/PA/Chol membranes shifts the nAChR pore into a conformation with a high affinity for ethidium. In both membranes, Carb binding stabilizes the nAChR in a desensitized conformation.
If the non-activatable PC-nAChR is desensitized, as has been suggested previously, then ethidium should bind with high affinity to PC-nAChR regardless of whether or not Carb is bound. Surprisingly there are essentially no changes in either the intensity or maximum of ethidium emission upon the addition of PC-nAChR to aqueous ethidium (Figs. 2C and supplemental S2C). This shows that ethidium does not bind with high affinity to the PC-nAChR pore in the absence of Carb, and that the channel does not adopt a high affinity ethidium-binding conformation. The ion channel in PC membranes is not stabilized in the desensitized state. Furthermore, the addition of 500 μm Carb does not lead to a change in the emission properties of ethidium. Given that this concentration of Carb is 10-fold greater than that used in the infrared difference experiments (i.e. 50 μm), which showed essentially equivalent levels of Carb binding in all three membranes, it can be concluded that Carb binds to PC-nAChR, but is unable to convert the nAChR pore into the desensitized conformation.
Conformation of the PC-nAChR Agonist Sites
We next probed the conformation of the PC-nAChR agonist-binding sites by comparing [3H]acetylcholine ([3H]ACh) binding to native-nAChR, PC/PA/Chol-nAChR, and PC-nAChR (Figs. 3A and supplemental S3). Equilibrium [3H]ACh binding curves for the native- and PC/PA/Chol-nAChR both show a saturable increase in specific binding in the 0–2.0 μm [3H]ACh range (Fig. 3A). The equilibrium dissociation constants (Keq) for [3H]ACh binding to native-nAChR (∼30 nm) and PC/PA/Chol-nAChR (∼65 nm; Table 1), are between those expected for resting (Kd = 800 nm) and desensitized (Kd = 2 nm) nAChRs (
). The measured equilibrium binding constant for native membranes is closer to that of the desensitized state because the resting-to-desensitized equilibrium shifts toward the desensitized conformation with increasing [3H]ACh (
If the agonist-binding sites of PC-nAChR adopt a desensitized conformation, high affinity, saturable binding, with a measured Keq characteristic of the desensitized conformation should be observed (i.e. Kd ≈ 2 nm). Instead only minimal [3H]ACh binding to PC-nAChR was detected over the same 0–2.0 μm [3H]ACh range (Fig. 3A). As a similar number of accessible agonist sites were used for all binding experiments (see supplemental Experimental Procedures and Table S1), the reduced binding of [3H]ACh to PC-nAChR can be attributed to lower occupancy, rather than saturation of a lower number of accessible sites. An estimate of the [3H]ACh equilibrium dissociation constant, based on the total number of accessible sites, is in the 3–10 μm range (Table 1), which is much lower than the expected affinity for the desensitized state. The lower binding affinity of PC-nAChR for [3H]ACh shows that the agonist-binding sites do not adopt a desensitized conformation, even after prolonged exposure (>1 h) to [3H]ACh. The PC-nAChR agonist-binding sites are stabilized predominantly in a low affinity conformation, reminiscent of the resting state.
nAChR Allosteric Coupling
If both the agonist-binding sites and transmembrane pore of PC-nAChR adopt low affinity ligand-binding conformations suggestive of the resting state, why does Carb binding fail to stabilize PC-nAChR in a desensitized conformation? The ethidium binding experiments show that Carb binding to the agonist sites does not alter the conformation (and thus ethidium affinity) of the transmembrane pore. This implies that the transmembrane pore is allosterically uncoupled from the agonist-binding sites. If this is the case, then the reverse should also be true, the binding of allosteric modulators to the pore should not influence the binding properties of the agonist sites. To test this hypothesis, we probed whether a pore-binding allosteric modulator could influence the affinity of PC-nAChR for [3H]ACh (Fig. 3B).
At constant, subsaturating concentrations of [3H]ACh, occupancy of the agonist-binding sites can be modulated by shifts in [3H]ACh affinity. We measured the ability of tenocylclidine (TCP), a pore-binding allosteric modulator of the nAChR (
), to increase the affinity of the nAChR for [3H]ACh. In both PC/PA/Chol-nAChR and native-nAChR, increasing concentrations of TCP up to ∼10 μm increase the fraction of [3H]ACh bound at equilibrium, showing that TCP converts the nAChR into a conformation with high affinity for [3H]ACh. The binding of TCP to the transmembrane pore converts the agonist-binding sites into a desensitized conformation (Fig. 3B). TCP concentrations above ∼10 μm decrease the amount of [3H]ACh bound, likely due to competitive displacement of [3H]ACh from the agonist-binding sites, as is commonly observed with other “pore-binding” allosteric modulators (
In contrast, concentrations of TCP up to 1 mm do not increase [3H]ACh binding to PC-nAChR, indicating that TCP does not allosterically modulate the affinity of nAChR for agonist. This is consistent with the finding that another non-competitive inhibitor, tetracaine, also has no effect on the conformation of PC-nAChR (
). These data confirm that the agonist-binding sites and transmembrane pore in PC-nAChR are allosterically uncoupled. Inactivation of the nAChR in PC membranes can thus be attributed to an uncoupling of the agonist-binding sites from the transmembrane pore, as opposed to stabilization of the desensitized state.
PC-nAChR Structure, Stability, and Solvent Accessibility
We next examined the structural and biophysical properties of this uncoupled conformation in our highly purified proteoliposome preparations. The structure of both PC-nAChR and PC/PA/Chol-nAChR was assessed by recording infrared spectra in deuterated solvent. The amide I/I′ band is sensitive to hydrogen bonding and thus protein secondary structure. The amide I/I′ bands of PC/PA/Chol-nAChR versus PC-nAChR are similar, but do exhibit subtle differences in band shape that become evident upon spectral deconvolution (supplemental Fig. S4A). These spectral nuances, however, result from different downshifts in the frequencies of amide I component bands, as a consequence of different levels of peptide 1H/2H exchange (see below and supplemental Fig. S4B). This is confirmed by the fact that the amide I/I′ band shapes are virtually indistinguishable when compared at equivalent levels of peptide 1H/2H exchange (
). Furthermore, Carb has no effect on nAChR secondary structure in either membrane environment. The uncoupled conformation of PC-nAChR thus has a similar secondary structure to that of both the resting (−Carb) and desensitized (+Carb) conformations of PC/PA/Chol-nAChR. Uncoupling must result from rearrangements in nAChR tertiary or quaternary structure.
The thermal stability of the uncoupled PC-nAChR is also similar to that of both the resting (−Carb) and the desensitized (+Carb) PC/PA/Chol-nAChRs (Figs. 4A and supplemental S6). In the absence of agonist, PC-nAChR undergoes a cooperative thermal denaturation with a 50% denaturation temperature (Td) of 52.35 ± 0.13 °C (supplemental Table S2). PC/PA/Chol-nAChR undergoes a slightly more cooperative thermal denaturation, at a slightly higher temperature (56.43 ± 0.84 °C). The addition of Carb leads to an increase in thermal stability of the nAChR in both membranes, but in PC this shift is smaller (Fig. 4A and supplemental Table S2). The reduced stabilizing effect of Carb on PC-nAChR implies that the physical interactions between the nAChR and agonist are weaker, consistent with the lower [3H]ACh binding affinity.
Although the secondary structure and thermal stability of the nAChR are relatively unaffected by uncoupling, PC-nAChR undergoes greater peptide 1H/2H exchange than PC/PA/Chol-nAChR (supplemental Fig. S4B). We examined the differences in peptide 1H/2H exchange further by recording infrared spectra as a function of time after exposure to 2H2O (Figs. 4B and supplemental S5). The uncoupled PC-nAChR exhibits greater peptide 1H/2H exchange than the PC/PA/Chol-nAChR at all time points throughout the entire 12-h period, including within seconds of exposure to 2H2O. The addition of Carb had no effect on nAChR peptide 1H/2H exchange in either membrane.
Each hydrogen exchange curve was accurately fit with a three-phase exponential decay function. The fits show that the differences in hydrogen exchange between the uncoupled PC-nAChR and both the resting and desensitized PC/PA/Chol-nAChR conformations result from an increase in the proportion (∼20%) of peptide hydrogens that exist in the rapidly exchanging pool (supplemental Table S3). The large increase in the number of rapidly exchanging peptide hydrogens comes at the expense of peptide hydrogens that were previously resistant to exchange. This suggests that uncoupling is accompanied by a large increase in the solvent accessibility of previously buried residues.
The effects of agonists and allosteric modulators on the nAChR are usually interpreted in terms of a simplified conformational scheme involving three main inter-convertible conformations: the resting, open, and desensitized states (Fig. 1D, Scheme 2) (
). The data presented here, however, show unequivocally that the lipid-dependent non-activatable nAChR adopts a novel conformation that is neither resting, nor desensitized. Lipid action at the nAChR therefore does not result from a modulation of the resting-to-desensitized equilibrium, instead another non-activatable conformation is involved. The identification of this novel conformation has repercussions for nAChR-lipid interactions, but may also have broader implications for our general understanding of nAChR function.
We previously hypothesized that lipids play a role in coupling agonist binding to channel gating (
). In support of this hypothesis, the functional hallmark of the lipid-dependent non-activatable nAChR is a loss of allosteric coupling between the agonist binding and ion channel functions of the receptor. Specifically, agonist binding does not induce the expected conformational change of the transmembrane pore. Reciprocally, TCP, a non-competitive inhibitor that binds to the nAChR pore, fails to trigger a conformational change in the extracellular agonist-binding domain. Effectively, the nAChR is no longer an agonist-activated ion channel because its agonist-binding site is functionally uncoupled from its transmembrane pore. For this reason we refer to this conformation as the lipid-dependent “uncoupled state.”
The existence of this uncoupled conformation leads to two important, but related questions. First, how does membrane lipid composition influence coupling between the agonist-binding sites and the transmembrane pore? Second, does uncoupling play a role in nAChR (or Cys-loop receptor) function in vivo?
With regards to the first question, our biochemical and biophysical data can now be interpreted in terms of the nAChR structural model. This model shows that the agonist-binding and ion channel functions reside in two distinct structural domains, demarked by an abrupt change in secondary structure (FIGURE 1, FIGURE 2, FIGURE 3, FIGURE 4, FIGURE 5A) (
). The two domains meet at an interface located ∼10 Å above the bilayer surface. This interface consists of several loops from the agonist-binding (β1-β2, β6-β7 or Cys-loop, and β8-β9) and transmembrane pore (M2-M3L) domains, as well as the covalent link between them (β10-M1). Complex interactions between these interfacial loops are critical for effective communication between the two domains (
). Lipid-dependent uncoupling must result from structural rearrangements that ultimately disrupt these inter-domain interactions. Weakened associations between the two domains with increased exposure of previously buried interfacial residues could account for the increased solvent accessibility observed with the uncoupled state.
How could lipids alter inter-domain interactions that occur ∼10 Å above the bilayer surface? One possibility is that lipids disrupt these inter-domain interactions by altering the packing of transmembrane helices. Transmembrane helix packing could ultimately dictate the conformation of the M2-M3 and β10-M1 linkers, thereby influencing crucial interactions at the interface between the transmembrane pore and agonist-binding domains (
Lipid-dependent uncoupling could also stem from more localized structural rearrangements at the lipid-nAChR interface. The five M4 helices (one from each subunit) are the most lipid-exposed transmembrane segments (Fig. 1C) (
). According to the current nAChR structural model, the C-terminal end of M4 extends beyond those of the other transmembrane helices and is located in close proximity to several highly conserved residues within the Cys-loop (Fig. 5, B and C). This is significant given that previous experiments have shown that the extended length of M4 is an absolute requirement for nAChR function (
), conspicuously aligns with Gln-435 in M4 of the Torpedo α-subunit, a residue that is poised to interact with the highly conserved αPhe-137 within the Cys-loop (Fig. 5C). Clearly, the C-terminal end of M4 is an important, but unappreciated part of the interface between the agonist-binding and transmembrane domains. By interacting directly with both membrane lipids and residues in the Cys-loop, M4 is perfectly situated to relay bilayer properties to this coupling interface. We propose that lipid-dependent bilayer properties are important for positioning the C-terminal end of M4 so that it effectively contacts the Cys-loop. M4/Cys-loop interactions in turn affect the ability of the Cys-loop to communicate with the remainder of the transmembrane domain, particularly the all important M2-M3 linker (Fig. 5D). M4 essentially acts as a lipid-sensor relaying bilayer properties to the coupling interface.
By binding to discrete transmembrane sites, lipids (and other allosteric modulators, including steroids) may stabilize the nAChR transmembrane domain in conformations that permit effective communication with the agonist-binding domain, perhaps by facilitating interactions between the C-terminal end of M4 and the Cys-loop. Indeed, the identification of a steroid-binding site between M4 and M1-M3 in the mouse GABAA receptor, as well as a similar modulatory site in the rat α7 nAChR, appear to support this (
Alternatively, a physical (rather than a chemical) property of the bilayer may dictate the favorability of helix-helix versus helix-lipid interactions. Tryptophan substitutions in M4, which in model peptides stabilize transmembrane helix associations (
). Presumably the M4 tryptophan substitutions drive increased association between M4 and M1-M3, which in turn facilitate interactions between the C-terminal extension of M4 and the Cys-loop.
Undeniably M4-lipid interactions play a key role in Cys-loop receptor function in native (or native-like) membrane environments. M4-lipid interactions even affect the open probability of human nAChRs (
). So whereas, M4-mediated uncoupling explains our in vitro data, the existence of this uncoupled conformation may have broader implications for Cys-loop receptor function in vivo. Lipids (and potentially other allosteric modulators) could influence the pool of agonist-activatable receptors by modulating an equilibrium between functional coupled and non-functional uncoupled states (Fig. 1D, Scheme 1).
From a pharmacological perspective, it is clear that membrane lipid composition can modulate the efficacy of nAChR agonists. Although we have concentrated on highly simplified and therefore non-physiological membranes, what we observe likely represents extremes at opposite ends of a spectrum. Subtle alterations in lipid composition may fine-tune the synaptic response in vivo. Indeed, the fact that endogenous lipids modulate agonist responses in other Cys-loop receptors directly supports this (
). Although the lipid sensitivity of these bacterial channels is presently unknown, their shortened M4 segments preclude a role for the C-terminal end of M4 at the coupling interface between their extra-membranous and transmembrane domains. It is thus tempting to speculate that the emergence of lipid sensitivity in the Cys-loop receptor superfamily coincided with the increasing demands of an evolving nervous system.
We thank Drs. Z. Yao and M. Pelchat for use of their equipment and N. Lavigne, N. Vuong, and S. Wang for technical assistance.