Structure and function of the nicotinic acetylcholine receptor

Structure and function of the nicotinic acetylcholine receptor

The nicotinic acetylcholine receptor (NAR) is the most widely studied receptor ion channel. This is due to a very practical reason – availability. The receptor can be isolated in high yield from electric eel or electric ray, both of which use strong electric discharges to incapacitate their prey or for defence. In the electric organs of these fish, the receptor occurs in abundance in stacks of excitable cells (Figure 9.2). Importantly, however, the NAR and voltage-gated ion channels only occur on one side of the cell.

In the resting state, both sides of the cell will have the same membrane potential. As a consequence, the electric field vectors within the cell and within the entire stack will can-cel each other out (Figure 9.2a). Electric stimulation will depolarize one membrane in each cell and invert its electri-cal field. Now, all of a sudden, all field vectors in the entire stack will point into the same direction (Figure 9.2b), and thus will add up to one very strong electrical field. Each in-dividual cell will contribute about 150 mV (the potential of its two membranes connected in series). Since the overall voltage is about 500-1000 V, this obviously requires sever-al thousands of cells stacked on top of each other. More-over, since both voltage and current are needed to make an impact on the prey, each level within the stack will need to have a sizeable surface area, thus providing for a large num-bers of NAR molecules.

Overall structure

A variety of experimental approaches have been taken to study the structure of NAR. A very interesting one is `elec-tron crystallography', performed with NAR incorporated into artificial lipid membranes (liposomes) and there form-ing regularly packed arrays (Figure 9.3a, right). Electrons (not X rays, which are normally used in protein crystallog-raphy) scattered from these two-dimensional crystals will form a pattern that can be evaluated to yield a three-dimen sional map of the structure, provided that data are avail-able from various tilt angles of the crystals relative to the electron beam. The resolution of structural images thus ob-tained is not quite the same as that of X-ray crystallography. However, X-ray data are often very hard to obtain with inte-gral membrane proteins; in fact, in the structures available, the transmembrane parts are often missing².

Figure 9.3b shows `contour' maps of the NAR, in top and in side view. In the top view, the identities of the individu-al subunits are also indicated; these have been determined using samples labelled with antibody fragments specific for each subunits. Note that the subunit composition of the NAR varies between different tissues. The α₂βγδ pat-tern shown here holds for the fish electric organs ³, and for the muscle type NAR in humans. Neuronal NAR, includ-ing the one found in the autonomic ganglia, is α₂β₃ or α₃β₂. This structural difference accounts for the selective action of several agonistic and antagonic drugs on muscles or the ganglia, respectively (see below).

Figure 9.3. Electron microscopy and `electron crystallography' of the nicotinic acetylcholine receptor (NAR). a: NAR channels in liposome membranes. On the left, they are mostly random-ly oriented (but some form a more regular pattern). On the right, a regularly packed `two-dimensional crystal' has formed. Such samples can be used to obtain a three-dimensional structure at low resolution by electron crystallography. b: Electron crystal-lographic structure, represented as density contour maps. Left: Top view. Middle, right: Side view. The bilayer and the portions of the receptor protruding from it into both directions are visi-ble. The arrow in the right frame points to the acetylcholine bind-ing site.

In the side view, it can be seen that a significant part of the receptor protrudes from the membrane to the exterior. The `bottleneck' or gate is located at the level of the lipid bilayer. An additional portion of the receptor sits on the cytosolic surface of the receptor and may be important for its ion selectivity.

The bottleneck is lined by 5 α-helices (one each from the 5 subunits). In the closed state, they touch upon each oth-er, making direct hydrophobic contacts via some leucine residues (Figure 9.4). These helices move apart by a con-siderable distance during channel opening, creating a rather large free lumen.

Location of the acetylcholine binding site

The acetylcholine binding site was mapped onto the struc-ture by comparing the electron densities of ligand-bound and unbound samples. This site has also been extensively characterized with biochemical methods, which allow the assignment of amino acid residues involved in ligand bind-ing. One important technique consists in affinity labelling. Figure 9.5 summarizes one such study.

The compound used is a derivative of acetylcholine. It con-tains tritium (³H) and thus is radioactive; it also contains a photo-activatable reactive group (blue). If allowed to bind to the receptor and illuminated with UV light, this group will attach itself onto anything in the vicinity, including amino acid residues, even ones of very low intrinsic reac-tivity. This will lead to the incorporation of the radioactive label into the receptor. The gel shows the labelled α and γ subunits (as well as a third band that reportedly is a degra-dation product of the γ chain). Importantly, this labeling oc-curs selectively at the agonist binding site, since it is large-ly suppressed in the presence of excess carbachol, which is a synthetic agonist closely related to acetylcholine (see later).

After labelling, the chains were separated⁴ and individually cleaved with a site-specifically acting protease (V8) ⁵. Fig-ure 9.5b shows that, in the α chain, the bulk of the radioac-tivity is associated with one of the proteolytic fragments (named αV8-20). However, specific labelling (i.e., label-ing that can be suppressed by the specific competitor car-bamoylcholine) also occurs in other fragments.

Individual proteolytic fragments, in turn, were isolated by electrophoresis and further purified by HPLC, and the la-belled residues within them were identified by protein se-quencing (Figure 9.5c). With the fragment shown, most of the label is found attached to two adjacent resides (Leu109 and Val 110).

While with the reagent used here only the α and the γ chains were strongly labelled, previous experiments (using different affinity probes with reactive groups attached to different sites of the acetylcholine molecule) have provided evidence that the δ chain is involved in ligand binding as well. In fact, there are two binding sites for acetylcholine in muscle-type NAR, located at the interfaces of the two α chains with the γ and the δ chains, respectively (cf. Fig-ure 9.3b).

The nature of the receptor-ligand interaction

Virtually all agonists and antagonists at the NAR share the positive charge of acetylcholine, most commonly (as in acetylcholine) in the form of a quaternary amino group⁶. What role does this positive charge have? The agonist bind-ing site of NAR is not rich in negative charges (aspartate or glutamate residues) but instead has multiple aromatic side chains. These can bind to cations by a peculiar mech-anism, called the `cation-π' interaction, in which the fixed charge of the cation is accommodated by the mobile π elec-trons of the aromatic ring. Using a quite sophisticated set of methods, it was determined that this mechanism indeed is responsible for the binding of acetylcholine to the NAR. These experiments were carried out as follows (Figure 9.6):

1.  The codons of individual tryptophan residues in the cloned α chain were exchanged for an amber stop codon (TAA) and the mutant mRNA was obtained by in vitro transcription.

2.  A suppressor tRNA – a tRNA carrying a mutant anti-codon that recognizes a complementary stop codon – was likewise generated in vitro and synthetically acy-lated with various fluorinated derivatives of tryptophan. This tRNA will selectively incorporate its amino acyl cargo at the mutant stop codon.

3.  The mRNA and the tRNA were both injected into frog oocytes.

4.  The frog oocytes expressing the mutant channels were studied using the `patch-clamp' method, which allows the characterization of single channels on intact cells⁷.

The non-natural tryptophan derivatives incorporated at the mutant stop codon contained various numbers of fluorine as substituents at the benzene ring that is part of the in-dole in the tryptophan side chain (cf. Figure 9.6c). In Fig-ure 9.6b, it is shown that with increasing numbers of fluo-rine substituents the sensitivity of the receptor to activation by acetylcholine decreased continuously, as ever higher amounts of the agonist were required to achieve half-max-imal response.

Fluorine is very small and is considered not to cause major steric changes when substituted for hydrogen. However, it is very strongly electronegative and will therefore pull π electrons out of the ring; this will weaken the cation-π inter-action. Figure 9.6c plots the observed receptor sensitivities (as logarithms of the EC₅₀⁸) against a theoretical parameter that describes the intensity of the cation-π interaction for tryptophan and its fluorinated derivatives. The correlation suggests that this interaction indeed is very important for the binding of acetylcholine to the NAR. By comparing the effects of substituting different tryptophan residiues, it was also determined that the most significant single residue in this interaction is tryptophan 149 in the α chain.

Receptor desensitization

An aspect of NAR function that is very important in its pharmacological manipulation is receptor desensitization. This phenomenon can be experimentally demonstrated in the set-up depicted in Figure 9.7: Stimulating electrodes produce presynaptic action potentials, which induce release of acetylcholine into the synaptic cleft. Acetylcholine will trigger postsynaptic action potentials, which in turn are detected with a second pair of electrodes. These postsynap-tic action potentials attain a reproducible, fairly constant height, and they are of very short duration, which reflects swift removal of acetylcholine by cholinesterase. If extra-neous acetylcholine is applied to this system (in a contin-uous fashion, so as to keep up a constant level in spite of cholinesterase activity), there is a steep initial depolariza-tion of the membrane, which then peters out over a time interval of a few seconds. Additional presynaptic stimuli during this time period are ineffectual. If the extrinsic sup-ply of acetylcholine is discontinued, cholinesterase rapid-ly cleans up, the membrane completely repolarizes, and its sensitivity to presynaptic triggers is restored in another few seconds.

Receptor desensitization can be understood in terms of a model of channel function that is similar to what we have seen before with voltage-gated channels (Figure 9.8). The receptor can cycle between three functional states: Resting (no current), open (current), and inactivated (no current). In the absence of ligand, the resting state is the most stable one, so most receptors will be in this state. Ligand binding favours both the open and the inactivated states over the resting state. Therefore, at sufficiently high ligand concen-tration, essentially all receptor molecules will leave the rest-ing state. Importantly, the open state is kinetically favoured over the inactivated state, which is a fancy way of saying that opening is faster than inactivation. Thus, most chan-nels will initially enter the open state (although some will become directly inactivated). However, the inactivated state is thermodynamically favoured over the open state, which means it is more stable than the latter. In fact, at high con-centrations of ligand, it is the most stable of all three states, and therefore upon continuous exposure to acetylcholine all the receptors that initially opened eventually wind up in the inactivated state. This explains the above observation of membrane repolarization and insensitivity during perfusion of the synapse with acetylcholine.