Activation of heteroliganded mouse muscle nicotinic receptors

Abstract

The activation of the mouse muscle-type nicotinic acetylcholine receptor was studied in the presence of carbachol, and in the simultaneous presence of carbachol and choline. The channel currents were recorded under steady-state conditions using cell-attached single-channel patch clamp, and during transient exposures to the agonists using a piezo-driven fast application system. The presence of choline resulted in inhibition of currents elicited by carbachol. The inhibitory effect of choline manifested as a reduction in the effective opening rate (increase in the mean intracluster closed time duration) in single-channel recordings. In the fast application experiments, the peak current amplitude was reduced and the current rise time increased when choline was co-applied with carbachol. The data were analysed according to a model in which receptor interactions with carbachol and choline resulted in three types of ligation: receptors occupied by two carbachol molecules, receptors occupied by two choline molecules, and receptors in which one agonist binding site was occupied by carbachol and the other by choline, i.e. heteroliganded receptors. All three agonist-bound receptor populations could open albeit with different efficacies. The affinity of the resting receptor to choline was estimated to be 1–2 mm, and heteroliganded receptors opened with an opening rate constant of ∼ 3000 s⁻¹. The results of the analysis suggest that the presence of choline in the neuromuscular junction in vivo has little effect on the time course of synaptic currents. Nevertheless, the contribution of heteroliganded receptors should be taken into consideration when the receptor is exposed simultaneously to two or more agonists.

The nicotinic acetylcholine receptor (AChR) is located in the neuromuscular junction and the central nervous system where its function is to initiate or facilitate synaptic transmission (for review, see Lindstrom, 2003). The receptor is normally activated upon the binding of two ACh molecules to two distinct ligand binding sites. The receptor binds ACh with high affinity, and the conformational change of channel opening is rapid in the presence of ACh.

The activation of the AChR is not limited to ACh and the requirements for the structure of an agonist appear to be quite liberal. A compound as simple as tetramethylammonium is an efficacious agonist of the muscle-type AChR (Akk & Auerbach, 1996). At the other end of the spectrum of the agonist structure, it has been demonstrated that monoclonal antibodies against the ACh binding site can activate the receptor (Bufler et al. 1996). Also, oxotremorine M and muscarine, drugs traditionally classified as muscarinic ligands, can elicit the opening of the nicotinic receptor channel (Reitstetter et al. 1994; Akk & Auerbach, 1999).

While ACh is the primary nicotinic agonist in the neuromuscular junction and the central nervous system, nicotinic receptors in vivo can be exposed to a number of secondary activating compounds. These include choline, a by-product of ACh metabolism, and nicotine, both of which are weak agonists of the muscle-type AChR (Zhou et al. 1999; Akk & Auerbach, 1999). Other potential ligands of the AChR are various anti-nociceptive drugs and inhibitors of the acetylcholinesterase used in the treatment of Alzheimer's disease, such as epibatidine and its analogues or physostigmine, galantamine and others (Bradley et al. 1986; Pereira et al. 1993; Prince & Sine, 1998; Avalos et al. 2002). However, it should be mentioned that drugs belonging to the latter class probably interact with the receptor via an allosteric site (Maelicke et al. 2000; Pereira et al. 2002; Akk & Steinbach, 2005).

Depending on the concentrations of ACh and that of the secondary agonist, the presence of the latter may lead to potentiation or inhibition of the receptor function. For example, for neuronal α4β4 receptors, the responses to low concentrations of ACh are potentiated by low, but inhibited by high, doses of choline (Zwart & Vijverberg, 2000). In the presence of high concentrations of ACh, all doses of choline inhibit the response. Similar results have been obtained for (+)-tubocurarine, physostigmine and tacrine (Steinbach & Chen, 1995; Zwart et al. 2000).

Many of these findings are likely to be explained by the emergence of heteroliganded receptors in which one agonist binding site is occupied by ACh and the other by choline or another secondary agonist. Heteroliganded receptors may have a higher opening probability than receptors occupied by a single ACh molecule. In this case, choline acts as a co-agonist. In contrast, at high ACh concentrations, exposure to a low-efficacy agonist (e.g. choline) may competitively inhibit the response.

In the present manuscript, we have used single-channel patch clamp and fast drug applications to investigate the activation of the mouse muscle-type AChR by carbachol and choline. The receptors were expressed transiently in a mammalian cell line. By studying the current responses to carbachol and choline, separately and in combination, we have elucidated the affinity of the muscle-type nicotinic receptor to choline, and the gating properties of the heteroliganded receptor.

Methods

General methods

Adult mouse muscle-type nicotinic acetylcholine receptor subunits (α, β, δ and ε) were subcloned into a cytomegalovirus (CMV) promoter-based expression vector, pcDNAIII (Invitrogen, San Diego, CA, USA). The receptors were expressed transiently in HEK 293 or tsA 201 cells using a calcium phosphate-based transfection technique as previously described (Akk, 2002). In brief, 3.5 μg of DNA in the ratio of 2: 1: 1: 1 (α: β: δ: ε) was used per 35 mm culture dish. The cells were exposed to the precipitate overnight after which the medium in the dish was replaced and electrophysiological experiments could be performed.

The experiments were carried out using two principal patch-clamp techniques, the cell-attached and outside-out patch-clamp configurations (Hamill et al. 1981). The bath solution contained (mm): 140 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 10 glucose and 10 Hepes, pH 7.3. The pipette solution in the cell-attached experiments contained (mm): 142 KCl, 5.4 NaCl, 1.8 CaCl₂, 1.7 MgCl₂ and 10 Hepes, pH 7.4. In the outside-out configuration, the pipette solution contained (mm): 140 KCl, 5 MgCl₂, 5 EGTA, 10 glucose and 10 Hepes, pH 7.3. The agonists (carbachol and choline) were added to the pipette solution (cell-attached) or bath solution (outside-out). All chemicals were obtained from Sigma Chemical Co. Unless specified otherwise, the membrane potential was held at −50 mV (cell-attached configuration) or −40 mV (outside-out configuration). The membrane potential in the cell-attached experiments was determined from the reversal potential of ionic currents. The experiments were carried out at room temperature.

Ionic currents were amplified with an Axopatch 200B amplifier (Axon Instruments, Union City, CA, USA), and digitized at 500 kHz (cell-attached/single-channel) or 50 kHz (outside-out/macroscopic). The data were saved on a PC hard drive using a Digidata 1322 Series interface (Axon Instruments) or ISO2 hard-and software (MFK, Taunusstein, Germany).

Single-channel patch clamp and analysis

In most cases, the experiments were carried out at agonist concentrations which resulted in single-channel clusters (Sakmann et al. 1980). A cluster is defined as a series of channel openings and closings which is separated from other such series of events by relatively long closed periods during which all receptors in the patch are desensitized. When activated by high concentrations of efficacious agonists (e.g. carbachol), the channel closed durations within clusters are relatively short resulting in episodes of high open probability. Event detection (idealization) and kinetic analysis were carried out using the QuB suite (provided by Drs Qin, Auerbach and Sachs, SUNY at Buffalo, http://www.qub.buffalo.edu). The currents were idealized at 4–6 kHz. The dead time was set at 30–45 μs. Channel open and closed interval duration histograms were constructed using idealized intracluster channel events.

Basic characterization of receptor activation was performed with the help of two dose–response curves. Intracluster open probability (P_o) is defined as the fraction of time the receptor spends in the open state within a cluster. When plotted as a function of agonist concentration, P_o is the microscopic equivalent of the whole-cell dose–response curve.

For the second dose–response curve, the slowest agonist-dependentcomponent of the intracluster closed time histograms was measured at different agonist concentrations. The inverse of this duration is defined as the effective opening rate (β′). As the agonist concentration is increased, β′ also increases, approaching the intrinsic channel opening rate constant (β, see Model 1) at saturating agonist concentrations. At high agonist concentrations, the macroscopic equivalent of β is the current rise time as measured from fast drug application experiments.

Curve fitting was carried out using two principal equations. The relationship between the P_o or β′ and carbachol concentration was estimated from fitting to the Hill equation:

(1)

where the response (P_o or β′) was measured at a specific carbachol concentration, EC₅₀ is the concentration of half-maximal effect and n is the Hill coefficient.

In cases where the inhibitory effect of choline was examined on cluster P_o or β′ at 1 mm carbachol, the curve fitting was carried out using the following equation:

(2)

where the response was measured at 1 mm carbachol and a specific concentration of choline, maximal response corresponds to the P_o or β′ at 1 mm carbachol in the absence of choline.

Receptor activation by carbachol or choline alone was studied using the standard activation scheme (Magleby & Stevens, 1972; Colquhoun & Hawkes, 1977):

Model 1

In Model 1, the closed, unliganded receptor (C) sequentially binds two agonist molecules, after which the channel opens (A₂O). The concentration of agonist is given by [A], k₊ and k₋ correspond to the agonist association and dissociation rate constants, β is the channel opening rate constant, and α is the channel closing rate constant. The A₂D corresponds to a short-lived (1–5 ms) intracluster desensitized state (Colquhoun & Sakmann, 1985; Salamone et al. 1999), not to be confused with the long-lived (seconds or minutes) desensitized state that terminates individual clusters (Franke et al. 1993; Auerbach & Akk, 1998). Rate constants k_+d and k_−d govern the transitions between the open and desensitized states.

Receptor activation in the presence of two species of agonists (carbachol and choline) was studied according to the following kinetic scheme:

Model 2

In Model 2, the vacant receptor (C) can bind either ligand A (e.g. carbachol) or ligand B (e.g. choline). If the approximate affinities of the receptor to the particular ligands are known a priori, the concentrations of the individual ligands can be adjusted so as to assure equivalent or preferential binding of one of the ligands. The monoliganded receptor AC (or BC) can bind either another molecule of ligand A (or ligand B), or a ligand of another type, and become a diliganded receptor A₂C, B₂C or ABC. The opening probability of a diliganded receptor depends on the β/k₋₂ ratio which may differ for different combinations of agonists.

Fast application measurements and analysis

A piezo-driven liquid filament application system was used to measure macroscopic responses from outside-out patches as described in Heckmann & Pawlu (2002). In brief, a P 810.30 piezo element (Physik Instrumente, Waldbronn, Germany) held in a brass cylinder was used with a glass tube (outer diameter 0.5 mm, wall thickness 0.175 mm; Hilgenberg, Malsfeld, Germany) connected to it. The tube was perfused with extracellular solution, to which agonist (carbachol or the combination of carbachol and choline) was added at the desired concentration. A miniature manual distribution six port (HVDX 6-5, Hamilton, Darmstadt, Germany) allowed the change between solution inflow from six reservoirs each filled with agonist-containing solution. The performance of the application system was tested by perfusing the glass tube with diluted extracellular solution and switching the ionic concentration at the tip of the patch pipettes.

The system had a solution exchange time of less than 50 μs when measured with an open patch electrode from changes in the liquid junction potential, or less than 100 μs when measured with an electrode containing a membrane patch. Short (10 ms) current traces were recorded at intervals of 1 s using ISO2 (MFK, Taunusstein, Germany) software. For the analysis of amplitudes and rise times, 7–25 traces were averaged.

The macroscopic patch clamp data were preprocessed and analysed with the QuB software (http://www.qub.buffalo.edu), using a maximum likelihood method (Milescu et al. 2005). The algorithm simultaneously estimates the rate constants and the number of channels in the patch, for a given state model, from the macroscopic current response to an arbitrary ligand pulse protocol. The pulse protocol we used allowed us to identify uniquely the kinetics and the channel count. The data were preprocessed in several steps. First, we eliminated traces with prominent noise artifacts. The remaining traces were aligned vertically with respect to the 0.5 ms immediately preceding the ligand pulse. We assigned a value of −2.4 pA to the amplitude of the open conductance class (an input parameter), as estimated from single channel experiments. The shape of the ligand pulse was approximated with a step, with rise and decay times, equal to one sampling interval. The change points of the ligand concentration were determined as those that gave the best fit, and they generally coincided with our visual pick. The estimated rate constants are in reasonable agreement with the values determined from single channel experiments. Aside from stochastic variations, the differences may be due to errors in approximating the ligand concentration and time profile, the single-channel amplitude, and the number of states and connections in the model.

Results

Receptor activation by carbachol – single-channel approach

Receptor activation was examined in the presence of 100–10 000 μm carbachol. Carbachol is an efficacious agonist for the muscle-type AChR, and single-channel clusters were observed throughout this concentration range. Sample clusters obtained at 200 and 2000 μm are shown in Fig. 1A. Cluster open probability (P_o) was determined at each agonist concentration and is plotted in Fig. 1B. Equation (1) was fitted to the data yielding a maximal open probability of 0.87 ± 0.03 with a half-maximal concentration at 646 ± 48 μm. The agonist-dependent component in the intracluster closed time histograms was plotted as the effective opening rate (β′, see Methods) (Fig. 1C). The curve saturated at 7778 ± 1581 s⁻¹ with a midpoint at 1382 ± 467 μm.

Figure 1. Activation of the AChR by carbachol.

A, single-channel clusters from the wild-type AChR elicited by 200 or 2000 μm carbachol. Openings are shown as downward deflections. Under each cluster, a segment at a higher resolution is given. B, cluster open probability (P_o) versus carbachol concentration. The P_o was calculated from the mean open and closed durations within the clusters where the mean closed duration did not include the short-lived desensitized state (A₂D, Model 1). The line was fitted using eqn (1) (see Methods). The fitting results are: P_o,max = 0.87 ± 0.03, EC₅₀ = 646 ± 48 μm; n = 1.3 ± 0.2. C, effective opening rate versus carbachol concentration. The effective opening rate is the inverse of the longest agonist-dependent component in the closed time histograms. The line was fitted using eqn (1). The fitting results are: β= 7778 ± 1581 s⁻¹, EC₅₀ = 1382 ± 467 μm, n = 2.0 ± 0.2. For B and C, each data point represents the mean P_o or β′ value from all clusters of one patch at each carbachol concentration. The number of clusters per patch ranged from 11 to 48.

The activation rate constant analysis was carried out using Model 1 which was fitted to single-channel currents obtained at 1 mm carbachol (n = 5 patches, total of 68 302 events). Two constraints were used to reduce the number of free parameters in the analysis. First, the channel opening rate constant, β, was held at the value obtained from the saturation of the β′ curve (7778 s⁻¹, Fig. 1C). It should be mentioned that such constraint is, in principle, not required. The analysis software allows estimation of the β value directly from the single-channel currents (e.g.Wang et al. 1997). Second, it was assumed that k₊₁ = 2 k₊₂ and k₋₂ = 2 k₋₁, and that the microscopic dissociation equilibrium constant of each site (K_D) is k₋₁/k₊₂, because it appears that in the mouse, adult-type AChR, the two transmitter binding sites have essentially equivalent K_D values for a number of agonists (Akk & Auerbach, 1996; Wang et al. 1997; but see Salamone et al. 1999).

The results are as follows: the association rate constant for carbachol is 23.3 ± 1.9 μm⁻¹ s⁻¹, the carbachol dissociation rate constant is 10672 ± 982 s⁻¹ giving the K_D as 458 ±56 μm. The channel closing rate constant at 1 mm carbachol was 1965 ± 19 s⁻¹. Since channel block may reduce the apparent channel closing rate, we obtained a separate estimate for α at lower carbachol concentrations. The simultaneous fitting of a simple C ⇌ O model to the currents recorded at 100 μm (1 patch) and 200 μm (3 patches) carbachol (total of 24 411 events) yielded 1506 ± 14 s⁻¹ for the apparent closing rate constant. Hence, the results suggest that 1 mm carbachol does not lead to significant linear channel block.

The rate constant analysis predicts that most bursts contain a single opening, i.e. once a channel closes it is more likely that a carbachol molecule dissociates from the receptor than that the channel reopens. The predicted number of reopenings per burst (n =β/2 k₋) is 0.36, and the duration of the gap associated with channel reopening τ= 1/(β+ 2 k₋) is 34 μs. We examined the burst structure in the data from one patch at 20 μm carbachol at −100 and +100 mV. The number and duration of short-duration gaps is predicted to be independent of voltage if the gaps originate from reopening from the A₂C state (Model 1). In contrast, if the gaps are due to open channel block then more gaps are predicted to be present at depolarized membrane potentials. The analysis yields 0.13 gaps of 41 ± 12 μs in duration per burst at −100 mV and 0.24 gaps of 38 ± 11 μs in duration per burst at +100 mV membrane potential. Hence, the data demonstrate the presence of short-lived gaps within bursts of openings, and suggest that the gaps are due to channel reopening from the diliganded closed state rather than channel block. The sample traces and the closed time histograms are given in Fig. 2.

Figure 2. Activation of the AChR by low concentrations of carbachol.

Single-channel currents and closed time histograms from the wild-type AChR activated by 20 μm carbachol at +100 mV (A) or −100 mV (B). Idealized traces are shown above or below the respective data traces. Idealization was carried out using program SKM (Qin et al. 1996) at 10 kHz. Histogram fitting was carried out using the QuB suite. At +100 mV, the mean closed durations were 38 ± 11 μs (24% of total) and 1.6 ± 0.1 ms. At −100 mV, the mean closed durations were 41 ± 12 μs (13% of total) and 6.9 ± 0.4 ms. The shorter-duration closed times are probably due to receptor reopening from the A₂C state (see Model 1). The longer-duration closed times are due to dwells in the monoliganded and unliganded states. The value of the longer-duration closed time component cannot be interpreted at 20 μm carbachol due to the absence of single-channel clusters.

Our estimates for the activation kinetics of the adult mouse AChR are in good agreement with previous estimates (Akk & Auerbach, 1999). The slightly lower K_D for carbachol estimated by us is probably due to the use of a more complete activation model containing the short-lived desensitized state, A₂D (Model 1). While the presence of the A₂D state in the kinetic scheme increased the likelihood of the fit, no systematic analysis of transition parameters between the open and desensitized states (k_+d and k_−d) was carried out.

Receptor activation by choline

Choline is a weak agonist of the adult-type mouse muscle AChR (Zhou et al. 1999). When the receptors were exposed to high, millimolar concentrations of choline, low open probability single-channel clusters could sometimes be observed (Fig. 3A). From such clusters, a channel opening rate constant of 50 s⁻¹ was estimated for receptor activation by choline. Due to a low β/k₋₂ ratio, the channel closing rate constant for choline can be estimated as the inverse of the apparent open duration. At 100 μm choline the channel closing rate constant is 1429 ± 82 s⁻¹ (mean open duration is 0.70 ± 0.04 ms; n = 1 patch, 1099 events). Using these values for β and α, we estimate the channel equilibrium gating constant (Θ) of 0.035 for choline. With such low gating efficacy no direct estimates for the receptor affinity to choline could be obtained from single-channel kinetics.

Figure 3. Activation of the AChR by choline.

A, single-channel currents from the wild-type AChR elicited by 100 μm or 10 mm choline, and the respective open duration histograms. Openings are shown as downward deflections. Continuous lines in the histograms are single-exponential fits with the mean open durations of 0.70 ± 0.04 ms (100 μm choline; 1099 events) or 0.41 ± 0.01 ms (10 mm choline; 4741 events). Closed duration histograms are not shown as no clusters were detected in the presence of 100 μm choline and due to the uncertainty in the number of channels in the patch the value of the closed times cannot be interpreted. At 10 mm choline, low P_o clusters were observed on rare occasions from which a β of 50 s⁻¹ was determined. B, single-channel clusters from the αS269I mutant receptor elicited by 5 or 20 mm choline, and the respective closed and open duration histograms. At 5 mm, the continuous lines in the closed and open time histograms are calculated from the rate constants of the fit to Model 1 (6627 events). The rate constants were: k₊ = 5.6 ± 2.6 μm⁻¹ s⁻¹, k₋ = 11726 ± 5528 s⁻¹, α = 559 ± 13 s⁻¹ with β constrained at 1500 s⁻¹ (Zhou et al. 1999). At 20 mm choline, the continuous lines in the closed and open time histograms are double or single-exponential fits, respectively (12 647 events). The mean closed times were 0.80 ± 0.02 ms (97%) and 6.62 ± 0.96 ms (3%). The mean open duration was 1.71 ± 0.03 ms.

To evaluate the affinity of the adult-type AChR to choline, we investigated the activation of the αS269I mutant receptor by choline (Fig. 3B). This congenital myasthenic syndrome mutation is known to enhance the channel equilibrium gating constant by inverse effects on the channel opening and closing rate constants while having little effect on receptor affinity (Croxen et al. 1997; Zhou et al. 1999). The channel opening rate constant of the mutant receptor in the presence of choline is ∼ 1500 s⁻¹, estimated from the saturation of the effective opening rate curve (Zhou et al. 1999) or the channel closed times at saturating choline concentrations (Fig. 3B).

Fitting Model 1 to single-channel currents from the mutant receptor elicited by 5 mm choline (n = 1 patch, 6627 events) yielded: k₊ = 5.6 ± 2.6 μm⁻¹ s⁻¹, k₋ = 11726 ± 5528 s⁻¹, K_D = 2094 ± 1386 μm. These values served as a starting point in the studies on the affinity of the wild-type receptor to choline.

Receptor activation by carbachol in the presence of choline – single-channel approach

Next, we investigated the effect of choline on receptor activation by 1 mm carbachol. Single-channel clusters elicited by 1 mm carbachol in the presence of 500 μm or 10 mm choline are shown in Fig. 4A. The presence of choline leads to a reduction in the cluster P_o. The relationship between the cluster P_o and choline concentration is given in Fig. 4B. The curve was fitted using eqn (2). According to the fit, P_o,max (the maximal open probability) is 0.65 ± 0.02, EC₅₀ = 15.3 ± 5.5 mm, and n = 0.92 ± 0.31. It should be mentioned that eqn (2) assumes that the P_o approaches a zero value at high choline concentrations which is not strictly true as the P_o in the presence of saturating choline concentrations is predicted to be approximately 0.02. The equation used in the fit does not address the mechanism of receptor inhibition but rather serves an illustrative purpose.

Figure 4. Choline affects the kinetic properties of single-channel clusters elicited by 1 mm carbachol.

A, single-channel clusters elicited by 1 mm carbachol in the absence and presence of 500 μm or 10 mm choline. Openings are shown as downward deflections. The presence of 10 mm choline reduced cluster open probability. B, cluster open probability in the absence and presence of 0.2–10 mm choline. The line was fitted to eqn (2). The best-fit parameters are: max P_o = 0.65 ± 0.02, K_D = 15.3 ± 5.5 mm, n = 0.92 ± 0.31. C, channel effective opening rate in the absence and presence of 0.2–10 mm choline. The line was fitted to eqn (2). The best-fit parameters are: max β′ = 3157 ± 179 s⁻¹, K_D = 5.9 ± 1.8 mm, n = 0.82 ± 0.25. Dotted line represents simulated data using Model 2 and the rate constants presented in the text assuming that heteroliganded receptors do not open, i.e. choline acts as a pure competitive inhibitor. D, inverse mean open duration in the absence and presence of 0.2–10 mm choline. No appreciable change in the apparent open durations was detected in the presence of choline at concentrations up to 10 mm.

The P_o is calculated from channel open and closed times. The channel closed times were affected by the addition of choline. At all agonist (carbachol or choline) concentrations the intracluster closed time histograms were adequately fitted with the sum of two exponentials – the faster and predominant activation-related component and a component corresponding to dwells in the short-lived desensitized state (A₂D). The duration of the activation-related closed time (1/β′) was affected by the presence of choline in the agonist mix. At higher choline concentrations the value for β′ dropped (closed time durations increased). The relationship between β′ and choline concentration is given in Fig. 4C. Fitting to eqn (2) demonstrates that the β′ curve saturates at 3157 ± 179 s⁻¹, the EC₅₀ of the curve is at 5.9 ± 1.8 mm and n = 0.82 ± 0.25.

In contrast, the channel apparent open times were not significantly affected by the addition of choline to the extracellular medium. The channel open durations at 1 mm carbachol and various concentrations of choline are shown in Fig. 4D. The mean open durations were 0.66 ± 0.01 ms at 1 mm carbachol (determined from simultaneous fitting of data from 5 patches, total of 40 284 events), and 0.65 ± 0.01 or 0.71 ± 0.01 ms in the presence of 1 mm carbachol and 200 μm (n = 2 patches, 11180 events) or 5000 μm choline (n = 3 patches, 5848 events), respectively.

The absence of the effect of choline on channel open times is not surprising. We could think of two reasons why the addition of choline may alter channel open times. First, the open times may be affected if the receptors simultaneously occupied by carbachol and choline had a different lifetime. However, the data presented above demonstrate that the open times are similar for receptors exposed to carbachol or choline (0.66 versus 0.70 ms, respectively). We may plausibly assume that receptors occupied by one carbachol and one choline molecule also have a comparable open duration. Hence, all agonist combinations result in similar open durations. Second, the apparent open durations may be affected by drug-induced unresolved channel block (Adams, 1976; Neher & Steinbach, 1978) when increasing doses of choline are coapplied with carbachol. However, we have shown previously that a reduction in the amplitude of a single-channel event is not accompanied by an increase in the apparent burst duration as might be expected in the linear blocking mechanism due to unresolved blockage events (Akk & Steinbach, 2003).

Kinetic analysis of the single-channel data obtained in the presence of carbachol and choline was carried out using Model 2 (see Methods) in which the receptor can bind two carbachol molecules, two choline molecules, or one carbachol and one choline molecule, and where all combinations of ligation lead to channel opening. To reduce the number of free parameters several constraints were applied. First, the carbachol binding and gating steps were constrained to the values obtained from receptor activation by carbachol alone (see above). Hence, it was assumed that the binding of choline to a ligand binding site does not affect the affinity of carbachol to the other, unoccupied site. Second, the channel opening rate constant in the presence of choline was fixed to a value obtained from receptor activation by choline alone (50 s⁻¹, Zhou et al. 1999). Third, it was assumed that the choline association and dissociation rates are the same for both binding sites. The equivalency of binding sites in the mouse adult-type receptor applies to ACh and a number of other ligands (Akk & Auerbach, 1996). Here, we have assumed the same for choline. Finally, the analysis was carried out using the choline dissociation rate constrained at 15 000 s⁻¹. Without this constraint the analysis did not converge to a well-defined set of rate constants. It should be pointed out that changes in the imposed k₋ value had relatively little effect on the estimated K_D value for choline or on the opening rate constant of heteroliganded receptors (Fig. 8). Further justification of constraints is given in the Discussion.

Figure 8. The effect of the imposed choline dissociation rate constant on the results of single-channel kinetic analysis.

The effect of k₋ on choline association rate constant (A), K_D for choline (B), the opening rate of heteroliganded receptors (C) and the log-likelihood of the fit (D) was studied. The analysis was carried out on the data from one patch (Patch no. 4 in Table 1) at 1 mm carbachol + 2 mm choline. Single-channel kinetic analysis of heteroliganded receptors was carried out with the choline dissociation rate constant constrained at 15 000 s⁻¹.

Data from nine patches recorded at 1 mm carbachol and 1–10 mm choline were analysed individually. Table 1 summarizes the results of the analysis. The averaged value for the opening rate constant of heteroliganded (carbachol–choline) receptors is 3208 ± 1209 s⁻¹, the channel closing rate constant is 2786 ± 1133 s⁻¹, and the average K_D of the adult-type AChR for choline is 1295 ± 867 μm. The equilibrium gating constant (Θ=β/α) calculated for the heteroliganded receptor is 1.15. Thus, the opening rate constant of the heteroliganded receptor is approximately half of that for the receptor occupied by two carbachol molecules. The K_D of the muscle-type AChR to choline is approximately 3-fold greater than the K_D for carbachol.

Table 1.

Single-channel kinetic analysis of heteroliganded receptors

Patch no.	[Carbachol] (mm)	[Choline] (mm)	βcarbachol-choline (s−1)	αcarbachol-choline (s−1)	k+choline(μm−1 s−1)	KD,choline (μm)	No. of events in patch
1	1	1	5552 ± 1357	3403 ± 840	18.6 ± 3.9	806	12035
2	1	2	4131 ± 1809	2864 ± 1398	15.1 ± 4.9	993	2952
3	1	2	3934 ± 813	3228 ± 916	18.8 ± 3.3	798	25006
4	1	2	3492 ± 951	1503 ± 500	13.0 ± 3.1	1154	29592
5	1	2	2805 ± 591	2491 ± 648	20.9 ± 3.6	718	15880
6	1	5	2103 ± 590	1714 ± 541	15.0 ± 3.0	1000	5703
7	1	5	2613 ± 1518	4960 ± 2032	9.1 ± 2.8	1648	4584
8	1	10	2691 ± 683	3420 ± 1245	4.3 ± 0.9	3488	10678
9	1	10	1553 ± 940	1489 ± 793	14.3 ± 5.5	1049	1109
Mean	1	NA	3208 ± 1209	2786 ± 1133	14.3 ± 5.1	1295 ± 867	NA

The results from the analysis of data from nine patches are shown. Data were analysed using Model 2. The opening and closing rate constants of the heteroliganded receptors are given by β_{carbachol-choline} and α_{carbachol-choline}, respectively. The k_+choline is the association rate constant for choline, and K_D,choline corresponds to the equilibrium dissociation constant of the receptor assuming a k₋ of 15 000 s⁻¹. NA, not applicable.

Macroscopic responses to carbachol and choline

Macroscopic currents were recorded from outside-out patches in response to fast applications of 1 mm carbachol, or the combination of 1 mm carbachol and 500, 2000 or 5000 μm choline. For 1 mm carbachol, the peak amplitudes ranged from −133 to −23 pA at V_M = −40 mV. With a P_o of 0.52 and a single-channel conductance of 60 pS, this translates roughly into 18–106 channels in the patch. During each application, the receptors were exposed to the agonist for 4 ms. In most cases, no desensitization was observed during this period. Successive applications were separated with intervals of 1 s.

Figure 5A shows averaged current traces recorded in the presence of 1 mm carbachol, 1 mm carbachol + 0.5 mm choline, 1 mm carbachol + 2 mm choline and 1 mm carbachol + 5 mm choline. The averaging was done using several (7–25) successive data traces.

Figure 5. Choline increases the rise times and reduces peak amplitude of macroscopic currents elicited by 1 mm carbachol.

A and B, macroscopic responses to 1 mm carbachol, 1 mm carbachol + 0.5 mm choline, 1 mm carbachol + 2 mm choline and 1 mm carbachol + 5 mm choline at two time resolutions. The currents are averages of 7–25 traces from one patch. Increasing concentrations of choline reduced the peak amplitude and increased the rise time. C, relative amplitudes of responses elicited by carbachol alone or carbachol + choline. The data are means ± standard deviations (s.d.) from 9 (0.5 mm choline), 14 (2 mm choline) or 9 (5 mm choline) experiments. D, relative 20–80% current rise times in response to carbachol or carbachol + choline. The means ±s.d. are from data from 3 (0.5 mm choline), 4 (2 mm choline) or 3 (5 mm choline) experiments.

The data demonstrate that the addition of choline to 1 mm carbachol results in depression of peak current (Fig. 5C). The average response for coapplied 500 μm choline was 91%, for 2000 μm choline 83%, and for 5000 μm choline 64% (1 mm carbachol alone corresponds to 100%).

The presence of choline also affected the current rise times (Fig. 5D). The 20–80% rise time was 0.14 ms for 1 mm carbachol. With choline added to the extracellular solution, the rise times increased to 0.16 ms (+0.5 mm choline), 0.17 ms (+2 mm choline) and 0.21 ms (+5 mm choline).

The macroscopic data from one patch were analysed using a maximum likelihood-based method (Milescu et al. 2005), implemented in the QuB software (Fig. 6). We performed a global fitting of averaged traces of the data collected at 1 mm carbachol, carbachol + 2 mm choline, and carbachol + 5 mm choline using Model 2 (see Methods).

Figure 6. Maximum likelihood-based fitting of macroscopic current responses.

A–C, averages of macroscopic data traces in response to 1 mm carbachol (A), carbachol + 2 mm choline (B) and carbachol + 5 mm choline (C). The fits were obtained using the sum of squares method in the QuB software (http://www.qub.buffalo.edu, Milescu et al. 2005), while imposing different sets of kinetic constraints. The results of the fits are given in Table 2. D, the idealized application pulse of carbachol or carbachol + choline. The change points of agonist concentration (t_start and t_end) were determined as the change points corresponding to the best fit (minimum sum of squares).

The results of the analysis are summarized in Table 2. In the analysis, three combinations of constraints were used, based on the previous results from single-channel work with carbachol and choline. In the first approach (Macroscopic 1), we constrained only the channel opening rate constant for choline while optimizing the other parameters (seven free parameters for rate constants plus one for channel count). As in the analysis of single-channel currents, we assumed that both agonist binding sites have equivalent affinities to an agonist. In addition, we constrained the closing rate constants for all three ligand combinations to a same value. In the second combination of constraints (Macroscopic 2), the binding and unbinding rates for carbachol, and the channel opening rate constants for carbachol and choline were constrained to the values obtained from single-channel analysis. In this approach, there were four free parameters. Finally, we used an approach in which all rates except the channel closing rate constant were fixed to the values estimated from the single-channel work (Macroscopic 3).

Table 2.

The comparison of rate constant estimates obtained from steady-state single-channel recordings (single-channel) and fast drug application experiments using three sets of constraints (macroscopic 1–3)

Parameter	Single-channel	Macroscopic 1	Macroscopic 2	Macroscopic 3
k+carbachol (μm−1 s−1)	23.3 ± 1.9	52.4 ± 6.5	(23.3)	(23.3)
k−carbachol (s−1)	10672 ± 982	4666 ± 732	(10672)	(10672)
KD,carbachol (μm)	458 ± 56	90 ± 18	458	458
βcarbachol (s−1)	(7778)	6289 ± 1571	(7778)	(7778)
αcarbachol (s−1)	1965 ± 19	4358 ± 336	3340 ± 121	3320 ± 23
Θcarbachol	3.96	1.44 ± 0.38	2.33	2.34
k+choline (μm−1 s−1)	14.3 ± 5.1	15.9 ± 4.5	4.3 ± 0.4	(14.3)
k−choline (s−1)	(15000)	29788 ± 11734	6670 ± 262	(15000)
KD,choline (μm)	1049	1873 ± 908	1551 ± 157	1049
βcholine (s−1)	(50)	(50)	(50)	(50)
αcholine (s−1)	1429 ± 82	4358 ± 336	3340 ± 121	3320 ± 23
Θcholine	0.035	0.011	0.015	0.015
βheteroliganded (s−1)	3208 ± 1209	2273 ± 311	5043 ± 388	(3208)
αheteroliganded (s−1)	2786 ± 1133	4358 ± 336	3340 ± 121	3320 ± 23
Θheteroliganded	1.15 ± 0.64	0.52 ± 0.08	1.51 ± 0.13	0.97
SS	NA	4739	5220	5716

Fitting was performed using Model 2. Under all conditions, it was assumed that the two binding sites have equivalent affinities for an agonist. Abbreviations: k₊, agonist association rate constant; k₋, agonist dissociation rate constant, K_D, equilibrium dissociation constant (k₋/k₊); β, channel opening rate constant; α, channel closing rate constant; Θ, equilibrium gating constant (β/α); SS, sum of squares. In the analysis of single-channel currents, the values for k₊, k₋ and α of carbachol were obtained from the analysis of currents recorded in the presence of 1 mm carbachol. The β values for carbachol and choline were obtained from the saturation of the effective opening rate curve (Fig. 1C) or from Zhou et al. (1999). The α value for choline was obtained as an inverse of the apparent open duration at 100 μm choline. The values for k₊, β and α for heteroliganded receptors are from Table1. The macroscopic currents were analysed using three sets of constraints. In the first set of conditions (macroscopic 1), only the β value for choline was fixed. Macroscopic 2 assumed the carbachol binding and gating rates from the single-channel studies while optimizing the choline binding rates and the β for heteroliganded receptors and in macroscopic 3, only the channel closing rate constant varied. Constrained rate constants are shown in parentheses.

Of the three conditions, the best fit (sum of squares, SS = 4739) was obtained under the set of conditions in which only the channel opening rate constant for choline was fixed. The results of the fit give a K_D of 90 μm for carbachol, a K_D of 1.9 mm for choline and an opening rate constant of 2273 s⁻¹ for heteroliganded receptors. For comparison, the results from the single-channel experiments gave a K_D of 458 μm for carbachol, a K_D of 1.3 mm for choline and 3208 s⁻¹ for the opening rate constant of heteroliganded receptors. Hence, with the exception of the K_D value for carbachol, there is a good agreement between the estimates obtained from single-channel and macroscopic responses.

A somewhat worse fit (SS = 5220) was obtained when the binding rates for carbachol were constrained to the values obtained from single-channel analysis. With the k₊ and k₋ for carbachol fixed at 23.3 μm⁻¹ s⁻¹ and 10672 s⁻¹, respectively, the choline association rate constant was 4.3 μm⁻¹ s⁻¹ and the choline dissociation rate constant was 6670 s⁻¹ (K_D = 1.5 mm). The opening rate constant for heteroliganded receptors was 5043 s⁻¹ under these conditions.

Finally, with all rate constants except the channel closing rate constant fixed to the values obtained from single-channel analysis, the worst of the three fits was obtained (SS = 5716).

Channel block in the presence of high concentrations of choline

Exposure to high concentrations of nicotinic agonists as used in this study can lead to channel block affecting the rate constant estimates. At the single-channel level, classical, linear open channel block manifests as a reduction in the current amplitude and an increase in the apparent open duration (Sine & Steinbach, 1984; Ogden & Colquhoun, 1985). In macroscopic responses, channel block may result in an initial overshoot of the current upon agonist application before settling at a lower, steady-state level, and in rebound current upon the removal of agonist which is due to relief from the blocked state (Liu & Dilger, 1991; Maconochie & Steinbach, 1998).

We did not see evidence of channel block affecting the channel open durations in the single-channel experiments. The durations of channel openings elicited by 1 mm carbachol remained constant when choline, at concentrations up to 10 mm, was added to the pipette solution (Fig. 4D). It has been shown previously that channel block in the presence of choline does not lead to an increase in the burst duration (Akk & Steinbach, 2003). This deviation from the linear blocking scheme may be explained by blocked channels retaining the ability to close or by very short lifetimes of the blocked state (a high unbinding rate of the blocker). To test the affinity of choline to the blocking site we examined the reduction in the single-channel current amplitudes at various choline concentrations. The relationship between the concentration of choline and the single-channel amplitude is shown in Fig. 7. In these experiments, the receptors were activated by 1 mm carbachol at −50 mV. Increasing concentrations of choline coapplied with carbachol resulted in a reduction in the single-channel amplitude with a K_B at 19.4 ± 4.7 mm.

Figure 7. Choline reduces single-channel amplitude.

Mean amplitudes of single-channel currents elicited by 1 mm carbachol in the presence of various concentrations of choline at −50 mV. Each data point corresponds to the mean amplitude of channel openings from one patch. The line was fitted to: response = max. response/(1 +[choline]/K_B. The best-fit parameters are: max. amplitude = −3.39 ± 0.09 pA, K_B = 19.4 ± 4.7 mm.

Such a high value for K_B suggests that burst lengthening due to linear channel block would be relatively insignificant at most choline concentrations. In the linear blocking scheme, the apparent burst duration is given as (1/α) (1 + ([blocker]/K_B)). The equation predicts a ∼ 5% (1/19.4) decrease in the apparent closing rate constant when 1 mm choline is coapplied with carbachol. Using the apparent α value for carbachol (1506 s⁻¹) and K_B value for choline (19.4 mm), a burst duration would be increased from 0.66 to 0.70 ms.

We also examined the blocking effect of 0.5–5 mm choline on macroscopic currents elicited by 1 mm carbachol. As channel block is usually voltage dependent (e.g. Sine & Steinbach, 1984), we compared the macroscopic current amplitudes at −40 and +40 mV. For typical voltage-dependent channel block, one expects the ratio of the current at −40 and +40 mV to decrease with increasing choline concentration. The results of this analysis are given in Table 3. The ratio remained almost constant (if anything, it increased with increasing choline concentration) suggesting that block is not affecting the current amplitude at −40 mV. The observed difference in the current amplitudes at the two potentials is expected because the channel closing rate constant and, hence, the channel open probability depend on the membrane potential. We can estimate the predicted difference by calculating the channel open durations and open probability at the two voltages using the known voltage dependency of the channel closing rate for the mouse muscle adult-type receptor from previous studies (Akk & Steinbach, 2000). The channel open duration for carbachol at 0 mV in excised patches is 0.21 ms and the voltage sensitivity, f= 0.32. From these values, using a relationship τ=τ(0 mV) exp(Vf/25 mV), we calculate the open durations at −40 mV (0.35 ms) and +40 mV (0.13 ms). Data from Fig. 4C give 3158 ± 371 s⁻¹ for the channel effective opening rate, i.e. 0.32 ± 0.04 ms for the activation-related component in the closed time histograms. As the channel effective opening rate is unaffected by voltage (Auerbach et al. 1996; Akk & Steinbach, 2003) we can use this β′ value to calculate the cluster open probability at both −40 and +40 mV. So, the predicted cluster P_o is 0.52 at −40 mV and 0.28 at +40 mV (ratio of 1.86). The comparison of peak amplitudes in Table 3 gives similar ratios for currents obtained at the two membrane potentials consistent with the interpretation that channel block has little effect on the macroscopic current amplitudes at choline concentrations up to 5 mm.

Table 3.

The comparison of macroscopic current amplitudes at −40 and +40 mV

Agonists	Amplitude at −40 mV (pA)	Amplitude at +40 mV (pA)	Ratio I−40/I+40
1 mm carbachol	−41.3	24.2	1.7
+0.5 mm choline	−38.2	19.8	1.9
+2 mm choline	−36.4	17.2	2.1
+5 mm choline	−27.5	14.8	1.9

If choline has no blocking effect on the currents, the predicted ratio of peak amplitudes is 1.9. The amplitudes are different at −40 versus +40 mV due to voltage dependency of the channel closing rate constant. The data suggest that block is not affecting current amplitudes at −40 mV.

However, in our calculations, we are assuming that the cluster P_o is the dominant determinant of the peak amplitude of macroscopic currents. Other factors, such as rapid desensitization and rectification may affect the macroscopic peak current estimates.

Discussion

For efficient gating, the AChR must bind two agonist molecules to two distinct sites. Although ACh is considered the major nicotinic agonist in vivo, receptor activation can be elicited by the metabolite of ACh, choline, whose potency on some types of heterologously expressed neuronal-type AChRs is even greater than that of ACh (Zhao et al. 2003). Native AChRs are presumably exposed to a mix of ACh, choline and other potential activators, hence the emergence of heteroliganded nicotinic receptors can be envisioned.

Heteroliganded receptors are formed when one of the agonist binding sites is occupied by an agonist of one type while the other site binds another type of agonist. The primary agonist for most AChRs is likely to be ACh. The secondary agonist can be supplied by natural processes such as the activity of acetylcholinesterase, whose product, choline, can activate the nicotinic receptor. In addition, secondary agonists can become available during smoking (nicotine) or treatment of neurological disorders (physostigmine, galantamine). Usually the affinity and/or efficacy of the secondary agonist compound on the muscle-type AChR is low, either due to natural selection (choline) or chosen so intentionally to minimize the side-effects in the case of pharmacological drugs. Despite the low potency, it is reasonable to expect that the presence of a secondary agonist affects the activation of the AChR by ACh.

Three possible types of effects by secondary agonists can be envisioned: (i) coagonism, (ii) competitive inhibition, and (iii) cross-desensitization. The particular type or combination of interactions depends on the intrinsic properties of the secondary agonist, the concentration of ACh, and on the concentration profile of the secondary agonist. The coapplication of a secondary agonist with low doses of ACh increases the effective ligand concentration and enhances receptor activation as a result of coagonism (Zwart & Vijverberg, 2000). At high ACh concentrations, the addition of a low-efficacy secondary agonist may lead to competitive inhibition of channel opening. Finally, prolonged exposure to low concentrations of a secondary agonist may result in desensitization of the receptor and reduce the response to subsequent applications of ACh (Paradiso & Steinbach, 2003; Uteshev et al. 2003).

In this study, we have examined the activation of the muscle-type AChR by carbachol and choline. The agonists were selected to satisfy two conditions: (i) the agonists must have widely differing efficacies, and (ii) the efficacies of both agonists must be within the detection limit of the equipment to assure our ability to reliably measure receptor activation. A combination of choline and carbachol satisfies these requirements. The efficacies of the two agonists differ by > 100-fold, and the predicted macroscopic current rise times are within the detection limit. The goal of the experiments was to evaluate the gating properties of heteroliganded receptors occupied by carbachol and choline, and to get an estimate for the affinity of the muscle-type receptor for choline. The results of the work can be used to better understand and predict the behaviour of the nicotinic receptor in the presence of exogenously supplied agonists or drugs which have modulatory effects on the nicotinic receptor.

Previous studies on heteroligation

Previously, the phenomenon of heteroligation has been studied on muscle-type AChRs using ACh and (+)-tubocurarine (Steinbach & Chen, 1995), ACh and metocurine or atracurium (Fletcher & Steinbach, 1996), and ACh and decamethonium (Liu & Dilger, 1993). Similar studies on neuronal nicotinic receptors have been carried out using ACh and choline (Zwart & Vijverberg, 2000), and ACh and tacrine or physostigmine (Zwart et al. 2000).

Steinbach & Chen (1995) showed that when exposed to a mix of ACh and (+)-tubocurarine ((+)TC), the fetal AChR could be activated by the binding of two ACh molecules or two (+)TC molecules, or with one agonist binding site occupied by ACh and the other by (+)TC. The respective efficacies for each ligation combination differed. Receptors which had bound two ACh molecules had the highest probability of opening, followed by receptors with one each of ACh and (+)TC. Finally, receptors occupied by two (+)TC molecules had the lowest opening probability. No activation by (+)TC was observed for adult-type receptors nor was there any indication in the data suggesting that adult-type receptors can open with one (+)TC and one ACh molecule bound.

Zwart & Vijverberg (2000) studied the activation of the neuronal α4β4 nicotinic receptor in the presence of ACh and choline, and concluded that the ability of choline to act as a coagonist is greater than its ability to activate the receptor alone, i.e. activation in the presence of ACh and choline is more efficacious than in the presence of choline alone.

It should be mentioned that for a number of drugs for which potentiation of ACh-elicited responses is observed, no direct activation of the AChR by the drug is detected. These include atracurium, a non-depolarizing neuromuscular blocking agent (muscle-type AChR, Fletcher & Steinbach, 1996), and acetylcholinesterase inhibitors such as physostigmine and tacrine (α4β2 and α4β4 neuronal, Zwart et al. 2000; muscle-type, Prince et al. 2002). It is possible that a low gating efficacy contributes to the absence of discernible macroscopic currents.

Even though the apparent potentiation mechanisms of the compounds described above are similar, the site and type of interaction may differ. Choline and other typical nicotinic receptor partial agonists most probably act by interacting with a site that coincides with that for ACh. In contrast, there is now strong evidence suggesting that physostigmine and other acetylcholinesterase inhibitors interact with a unique site on the nicotinic receptor (Storch et al. 1995; Schrattenholz et al. 1996; Akk & Steinbach, 2005).

Our findings demonstrate that receptors which have bound one carbachol and one choline molecule have a higher gating efficiency than those which have bound two choline molecules.

Gating efficacy of heteroliganded receptors

The contributions that ligand binding to a single binding site makes towards the channel gating event can be estimated by comparing the equilibrium gating constant for the heteroliganded receptor (Θ_HET = β_HET/α_HET) with the Θ values for receptors that have bound two carbachol (Θ_carbachol) or two choline molecules (Θ_choline). If the contributions from the binding sites are equal then the ratio of Θ_carbachol/Θ_HET should equal the ratio of Θ_HET/Θ_choline. Using the Θ values available from this study (Θ_carbachol = 3.96; Θ_choline = 0.035; Θ_HET = 1.15), the two ratios equal 3.4 and 32.9. Thus, the data suggest that a receptor with the combination of choline–carbachol bound gates more efficiently than expected.

The independence of binding site contributions to channel gating can be estimated from the coupling coefficient (Ω) and coupling energy (ΔG) (Horovitz & Fersht, 1990; Hidalgo & MacKinnon, 1995). Based on previous data (see above; Liu & Dilger, 1993; Steinbach & Chen, 1995), we have calculated the coupling coefficients and coupling energies for AChR activation by decamethonium or curariform agonists in the presence of ACh. The results are presented in Table 4. For all agonist combinations, the coupling coefficient was less than unity suggesting that heteroliganded receptors open more efficiently than expected based on the gating properties of receptors occupied by one type of agonist.

Table 4.

Additivity in the AChR gating

Agonists	Carbachol + choline	ACh + decamethonium	ACh + (+)-tubocurarine	ACh + metocurine
Θprimary	3.96	20	100	100
ΘHET	1.15	2.5	0.42	0.72 (0.01)
Θsecondary	0.035	0.016	0.0003	0.00008
Ω	0.105	0.0512	0.17	0.0154 (80)
ΔG (kcal mol−1)	−1.33	−1.75	−1.05	−2.46 (2.59)
Receptor type	Adult muscle (αβδε)	Fetal muscle (αβγδ)	Fetal muscle (αβγδ)	Fetal muscle (αβγδ)
Reference	Present work	Liu & Dilger (1993)	Steinbach & Chen (1995)	Fletcher & Steinbach (1996)

The contributions of agonist binding to a single site towards the gating event were estimated by comparing the gating equilibrium constants for heteroliganded receptors (Θ_HET), receptors occupied by ACh or carbachol (Θ_primary), and receptors occupied by a secondary agonist, i.e. decamethonium, (+)-tubocurarine, metocurine or choline (Θ_secondary). The coupling coefficient (Ω) was calculated as (Θ_primary×Θ_secondary)/(Θ_HET)². The coupling energy was calculated as RTln(Ω). For ACh + metocurine, the data were calculated under two conditions: with both ligands binding to the same site with high affinity or that the site with high affinity for ACh had low affinity for metocurine (shown in parentheses). The Ω values of less than 1 indicate that heteroliganded receptors open more efficiently than expected based on the gating efficacies of the individual agonists assuming equal contribution by the binding sites to channel gating.

Comparison of rate constant estimates from the single-channel and macroscopic experiments

There was generally a good agreement between the rate constant estimates obtained from the steady-state single-channel recordings and the fast application macroscopic current measurements. The biggest discrepancy between the estimates obtained from the two kinds of analysis was for the receptor affinity to carbachol. The K_D for carbachol was 458 μm or 90 μm when estimated from the single-channel experiments or the macroscopic currents, respectively.

One possible explanation for this is the variance in the channel activation properties when the receptor is exposed to different monovalent inorganic cations. It has been shown previously that the K_D of the nicotinic receptor to ACh is > 50% greater when the receptor is bathed in KCl compared with NaCl (Akk & Auerbach, 1996). For receptors activated by tetramethylammonium (TMA), a switch from NaCl to KCl leads to an even bigger shift in the estimated K_D value (G. Akk, unpublished data). In the present project, the single-channel experiments were carried out using a KCl-based pipette solution. In contrast, the outside-out patches in the fast-application experiments were exposed to a NaCl-based bath solution. It is conceivable that a similar mechanism underlies the differences in the K_D values for carbachol in the cell-attached and outside-out patch configurations. However, it should be pointed out that our estimates for the K_D for choline did not show a similar variance: the K_D for choline was relatively uniform in the single-channel experiments and macroscopic current measurements.

Other possible explanations for the differences in the K_D estimates for choline include statistical and biological variability, and experimental artifacts. Finally, the process of patch excision itself may affect some parameters of channel activation (Akk & Steinbach, 2000).

Constraints used in the analysis

A number of constraints were used in the kinetic analysis of heteroliganded receptors. Some of these constraints were based on the results of analysis of the receptor function when it was exposed to a single type of agonist. It is reasonable to assume that the binding rate constants of the first ligand molecule to an unoccupied receptor are unaffected by the general presence of other types of ligands in the extracellular solution. Hence, the binding rate constants for carbachol to the first site can be constrained to be equal to those observed in the absence of choline. Similarly, once both binding sites are occupied by carbachol, the receptor should open with the same rate regardless of the presence of choline in the extracellular medium. The same reasoning applies to receptors fully occupied by choline.

There is no direct evidence suggesting that the agonist binding sites have equivalent affinities to choline, but we made this assumption for the following reasons. First, for a wide range of nicotinic agonists, fitting without the constraint of equivalent binding sites does not lead to significant increase in likelihood (Akk & Auerbach, 1996; Wang et al. 1997). Second, we studied the activation of the αS269I mutant receptor by choline. The effects of this mutation can be accounted for by an increase in the gating equilibrium constant, i.e. an increase in β and a decrease in α (Zhou et al. 1999), allowing recording of single-channel clusters at different agonist concentrations. Our analysis using Model 1 suggests that there is no significant increase in the likelihood of the fit when the analysis is carried out using a model in which the binding sites have unequal affinities compared with one with constrained equivalent binding sites. Finally, fitting with fewer free parameters has distinct practical advantages, such as computation speed and convergence.

In the studies of heteroliganded receptors, the choline dissociation rate constant was constrained at 15 000 s⁻¹. Without this constraint the analysis did not converge. The value for the choline dissociation rate constant was chosen based on the binding properties of other nicotinic agonists (Akk & Auerbach, 1996). The effect of the imposed k₋ on the resulting k₊, K_D and β_HET values and the log-likelihood of the fit is shown in Fig. 8. In this analysis, one patch (Patch no. 4 in Table 1) was analysed using Model 2 while varying the constrained value for the choline dissociation rate constant. The increase in the k₋ value resulted in an increase in the k₊ value without affecting strongly the K_D or β_HET values.

One additional, implicit constraint was used, namely the assumption that the binding of ligand A does not affect the affinity of the other binding site to ligand B, i.e. the binding of choline to one binding site does not influence the affinity of the other binding site for carbachol. We have no experimental evidence supporting this. Further studies will have to be conducted to study how the binding of various ligands to one binding site affects the properties of the other binding site.

Does choline affect the functioning of the neuromuscular junction?

Our results demonstrate that the affinity of the muscle-type nicotinic receptor to choline is 1–2 mm. It has previously been found that the plasma concentration of choline does not exceed 20 μm (Savendahl et al. 1997). Coupled with the low gating efficacy of the muscle-type AChR in the presence of choline, it is safe to conclude that only a negligible fraction of receptor openings under physiological conditions is due to ambient choline.

On the other hand, the high turnover number of acetylcholinesterase in the neuromuscular junction ensures a rapid rise in local choline concentration which peaks during the decaying phase of the ACh concentration pulse. This may lead to the binding of choline to receptors which have lost one of the two ACh molecules and a prolongation of the current decay phase.

To test this possibility, we simulated synaptic responses to pulses of ACh and choline using Model 2. The activation rate constants for ACh are from Akk (2002). The activation rate constants for choline are from the present manuscript (Table 2, single-channel). In calculating the gating rate constants for heteroliganded receptors we have assumed that the coupling coefficient for acetylcholine–choline is the same as for the agonist pair carbachol–choline (0.1, Table 4).

Macroscopic responses to a pulse of ACh in the absence and presence of a pulse of choline are shown in Fig. 9. The ACh pulse had a rapid rise time (10 μs), high amplitude (1 mm) and a decay with a time constant of 300 μs. For the choline pulse, we varied three parameters. First, the decay time constant for the choline pulse was held at 0.1, 1 or 10 ms (Fig. 9A). Second, we varied the delay of the choline pulse with respect to the ACh pulse from 0 to 2 ms (Fig. 9B) and finally, the amplitude of the choline pulse was kept at 0.1, 0.5 or 1 mm (Fig. 9C). The variance in the parameters of the choline pulse reflects our ignorance of the true time course of choline in the neuromuscular junction.

Figure 9. Modelling of the effect of choline on macrocopic responses from the neuromuscular junction.

The current responses were produced using Model 2. The rate constants for ACh-mediated activation were: k₊ = 242 μm⁻¹ s⁻¹, k₋ = 30632 s⁻¹, β = 41090 s⁻¹, α = 2043 s⁻¹. The rate constants for receptors activated by choline were: k₊ = 14.3 μm⁻¹ s⁻¹, k₋ = 15000 s⁻¹, β = 50 s⁻¹, α = 1429 s⁻¹. The opening rate constant for heteroliganded receptors was calculated assuming a Ω of 0.1 yielding 5414 s⁻¹. A, time courses of the ACh and choline pulses and the resulting current traces. The ACh pulse had rise time constant of 10 μs, a plateau at 1 mm for 100 μs and a decay phase with a time constant of 300 μs. The choline pulse started 0.4 ms after the ACh pulse. It had a rise time constant of 300 μs, peak at 0.5 mm and a decay phase with time constants of 0.1, 1 or 10 ms. B, time courses of the ACh and choline pulses and the resulting current traces. The parameters of the ACh pulse are as in A. The choline pulse had a decay time constant of 1 ms and a delay of 0, 0.4 or 2 ms following the onset of the ACh pulse. C, time courses of the ACh and choline pulses and the resulting current traces. The parameters of the ACh pulse are as in A. The choline pulse started at 0.4 ms following the onset of the ACh pulse and had a decay time constant of 1 ms. The maximal choline concentration was 0.1, 0.5 or 1 mm. The boxed region of the current decay phase is shown also at a higher resolution. Changes in the choline decay phase, the separation of ACh and choline pulses, and the choline peak concentration had minor effects on the current decay phase. The simulations were carried out with the ion channel simulator in the QuB suite, using the scripting language developed by L. Milescu.

The comparison of modelled current time courses suggests that the presence of choline has no significant effect on the decay of endplate currents. Intuitively, this can be inferred from the paths available for the monoliganded receptor AC (Model 2) and the rates governing the transitions. Even at the peak of the choline pulse, the dissociation of the remaining ACh molecule is more than four times more likely than the binding of a choline molecule. Moreover, of the receptors which do bind one ACh and one choline, only 10% open.

Summary

Ligand-gated ion channels in vivo are exposed to a mix of ligands that can alter the response of the receptor to the endogenously supplied neurotransmitter. Studies of channel activation in the presence of different agonists can be used to determine the properties of the receptor when the binding sites are occupied by ligands of different type. The results of these studies may be useful in predicting the effects of drugs that can activate the nicotinic receptor.

The results of our experiments suggest that the presence of choline in the neuromuscular junction has little effect on the time course of synaptic currents, mainly due to the relative affinities of the muscle-type AChR to ACh and choline, and the relative concentrations of the two agonists. However, the contribution of heteroliganded receptors may be much greater in other situations, e.g. for mutant receptors (Zhou et al. 1999) or for nicotinic receptors in the central nervous system where the receptors are usually exposed to much lower concentrations of ACh, and so the general concept of heteroligation should not be discounted when describing nicotinic receptor activation in vivo.