Block of muscle nicotinic receptors by choline suggests that the activation and desensitization gates act as distinct molecular entities

Abstract

Ion channel block in muscle acetylcholine nicotinic receptors (AChRs) is an extensively reported phenomenon. Yet, the mechanisms underlying the interruption of ion flow or the interaction of the blocker with the channel's gates remain incompletely characterized. In this paper, we studied fast channel block by choline, a quaternary-ammonium cation that is also an endogenous weak agonist of this receptor, and a valuable tool in structure-function studies. Analysis of the single-channel current amplitude as a function of both choline concentration and voltage revealed that extracellular choline binds to the open-channel pore with millimolar apparent affinity (K(B) congruent with 12 mM in the presence of approximately 155 mM monovalent and 3.5 mM divalent, inorganic cations), and that it permeates the channel faster than acetylcholine. This, together with its relatively small size ( approximately 5.5 A along its longest axis), suggests that the pore-blocking choline binding site is the selectivity filter itself, and that current blockages simply reflect the longer-lived sojourns of choline at this site. Kinetic analysis of single-channel traces indicated that increasing occupancy of the pore-blocking site by choline (as judged from the reduction of the single-channel current amplitude) is accompanied by the lengthening of (apparent) open interval durations. Consideration of a number of possible mechanisms firmly suggests that this prolongation results from the local effect of choline interfering with the operation of the activation gate (closure of blocked receptors is slower than that of unblocked receptors by a factor of approximately 13), whereas closure of the desensitization gate remains unaffected. Thus, we suggest that these two gates act as distinct molecular entities. Also, the detailed understanding gained here on how choline distorts the observed open-time durations can be used to compensate for this artifact during activation assays. This correction is necessary if we are to understand how choline binds to and gates the AChR.

Figure 1.

Wild-type AChR single-channel inward currents elicited by various concentrations of choline. Membrane potential ≅ −100 mV. Display f _c ≅ 4 kHz. Openings are downward deflections. Fast open-channel block by choline is manifest as a concentration-dependent decrease in the single-channel current amplitude and as a prolongation of the (apparent) open times.

Figure 2.

Kinetic scheme used to interpret the effect of choline block on the kinetics of AChR gating and desensitization. For simplicity, this scheme only displays the diliganded receptor. The closed, open, and desensitized states are denoted as C, O, and D, respectively. Choline acting as an agonist is denoted as A. Choline acting as a blocker is denoted as B. The opening, closing, and entry-into-desensitization rate constants of the unblocked (β_U, α_U, D_U) and blocked (β_B, α_B, D_B) forms of the receptor are indicated. The ratio between the blocker dissociation rate constant and the blocker association rate constant gives the blocker dissociation equilibrium constant. In this paper, the dissociation equilibrium constant from the open-channel pore is denoted as K_B, whereas that from the closed-channel pore is denoted as G_B. The reaction steps that dictate the mean duration of choline-diliganded (apparent) openings (Eq. 2) are indicated as bold arrows.

Figure 3.

Voltage and concentration dependence of wild-type AChR block by choline. All data points were globally fitted with Eq. 1 assuming that choline can only dissociate back to the extracellular solution, that the voltage dependence of K_B can be expressed as K_B = K_B0e^0.04δV, and that the unblocked current (i ₀) is a linear function of the transmembrane potential (i ₀ = γV). The fitted values were: γ = 75.8 ± 0.4 pS, K_B0 = 12.5 ± 1.1 mM, and δ = 0. The estimates of K_B0 and δ can be regarded largely as phenomenological descriptors of choline block, but they are probably devoid of mechanistic meaning (see Results for a longer discussion). (A) Data displayed as a 3-D plot. (B) Data displayed as separate I-V curves at the indicated concentrations of blocker. Note that the solid lines were calculated using the parameters obtained from the global fit. The limited ability of Eq. 1 to fit the entire dataset, which is particularly evident at 2 and 5 mM choline, most likely reflects the inaccuracies of the permeation model used.

Figure 4.

Kinetic properties of wild-type OA₂ OA₂B bursts at −100 mV. (A) The mean durations of choline-diliganded open-blocked bursts, at different choline concentrations, were estimated as indicated in Materials and Methods. Fractional-current values were estimated as the ratio between the observed single-channel current amplitude and −7.58 pA (γ = 75.8 pS, from Fig. 3). Because of the inevitable variation in their estimates, the average single-channel current amplitude at some of the nonblocking concentrations of choline turned out to be somewhat greater than −7.58 pA. This explains why the fractional-current values of some of the points in the plot are larger than unity. Each experimental point corresponds to 1 of 15 different choline concentrations, between 200 nM and 50 mM. The solid line is the fit of the data with Eq. 3. The estimated unblocked channel shutting rate was (α_U + D_U) = 3,941 ± 240 s⁻¹. The estimated blocked channel shutting rate was (α_B + D_B) = 329 ± 72 s⁻¹. The dashed line plots show three hypothetical different scenarios. In all of them, the unblocked channel shutting rate is assumed to be the same (α_U + D_U = 3,941 s⁻¹), whereas the shutting rate of the blocked channel is assumed to be either infinitely slow, equal to that of the unblocked channel, or even faster. The predictions of these three plots are very different from one another and from the experimental observations. To facilitate the interpretation of this figure, the top x axis contains the choline concentration scale. Vertical error bars are standard errors. (B) The probability of an OA₂ OA₂B burst shutting (i.e., closing or entering a desensitized state) from OA₂ is shown as a solid line plot (Eq. 5), whereas the probability of shutting from OA₂B is shown as a dashed line (Eq. 4). The shutting rates estimated in A were used for computing these probabilities.

Figure 5.

αS269I AChR single-channel inward currents elicited by various concentrations of choline. Membrane potential ≅ −100 mV. Display f _c ≅ 4 kHz. Openings are downwards.

Figure 6.

Affinity of the αS269I mutant open-channel pore for choline. All measurements were done at −100 mV. The data were fitted with the following: Fractional current = K_B/(K_B + B). The estimated value of K_B was 10.4 ± 1.3 mM, quite close to the wild-type estimate of ∼12.5 mM. Vertical error bars are standard errors.

Figure 7.

Kinetic properties of αS269I OA₂ OA₂B bursts at −100 mV. The mean durations of choline-diliganded open-blocked bursts were estimated as in Fig. 4 A, for the wild type. Each experimental point corresponds to 1 of 10 different choline concentrations, between 200 μM and 50 mM. The solid line is the fit of the data with Eq. 3. The estimated unblocked-channel shutting rate was (α_U + D_U) = 873 ± 53 s⁻¹. The estimated blocked-channel shutting rate was (α_B + D_B) = 82 ± 17 s⁻¹. Note that, even though these two shutting rates are quite different from their wild-type counterparts, the unblocked-to-blocked ratios are very similar (∼11 in the mutant, ∼12 in the wild type). The top x axis contains the choline concentration scale. Vertical error bars are standard errors.

Figure 8.

Effect of choline block on the wild-type rate constant of entry into desensitization. The mean total open time within clusters (Eq. 8) turned out to be rather insensitive to the fractional current. This suggests that the operation of the desensitized gate, at least in the entry-into-desensitization direction, is largely unaffected by choline block. The average of the mean total open time within clusters across choline concentrations (horizontal dashed line) is 31 ± 4 ms. Thus, in the wild type, D_U ≅ D_B ≅ 32 s⁻¹ (see Fig. 2). Vertical error bars are standard errors.

Figure 9.

Effect of choline block on the αS269I mutant rate constant of entry into desensitization. This rate constant was estimated as in Fig. 8, for the wild type. A wider range of choline concentrations could be tested on this mutant because lower concentrations of the ligand were needed to start detecting clear clusters of single-channel openings. As was the case for the wild type, the mean total open time within clusters (Eq. 8) proved to be insensitive to the fractional current. The average of the mean total open time within clusters across choline concentrations (horizontal dashed line) is 57 ± 8 ms. Thus, in the αS269I mutant, D_U ≅ D_B ≅ 18 s⁻¹ (see Fig. 2). Vertical error bars are standard errors.

Figure 10.

The opening rate constant of choline-diliganded wild-type AChRs. The mean duration of intracluster shut intervals was measured between 14 and 50 mM choline. The average value across choline concentrations (horizontal dashed line) is 8.0 ± 0.7 ms. Vertical error bars are standard errors.

Figure 11.

The opening rate constant of choline-diliganded αS269I AChRs. After the initial shortening of intracluster shut intervals, owing to the increasing occupation of the transmitter binding sites, the mean duration of CA₂ CA₂B bursts remains insensitive to the concentration of choline. The average closed-burst duration in the 14–50 mM choline range (horizontal dashed line) is 0.52 ± 0.02 ms. Vertical error bars are standard errors.

Figure 12.

Mean duration of sojourns in the CA₂/CA₂B set of diliganded closed states as a function of blocker concentration. Eq. 13 is plotted for different G_B values assuming that the opening rate constant of the choline-diliganded unblocked AChR (β_U) is 125 s⁻¹. As the concentration of blocker increases, the mean duration of closed-diliganded intervals is expected to approach, asymptotically, the reciprocal of the blocked-channel opening rate constant (i.e., β_B = β_UG_Bα_B/K_Bα_U, from detailed balance). If G_B/K_B = 13, then β_B = β_U, and the mean duration of closed-diliganded sojourns becomes independent of the concentration of blocker.

Figure 13.

An MWC-type of kinetic scheme (Monod et al., 1965) that includes blocking and unblocking steps. This kinetic model was used to test the hypothesis that the prolongation of (apparent) open times with increasing choline concentrations is due to the increasing rebinding of choline to the transmitter binding sites in the open state (Eqs. 14 and 15), rather than to a more local effect of choline obstructing the closure of the activation gate. To this end, the expected time constant value of the slowest component of the open-time distribution, at each choline concentration, was computed numerically as the reciprocal of the smallest eigenvalue of the open-state submatrix of −Q (−Q_oo). Since choline block was assumed not to affect the kinetics of gating, desensitization or choline binding/unbinding to/from the transmitter binding sites, only one of the two stacked MWC schemes needed to be considered (say, the one with bold symbols). The model was further simplified by making the plausible assumption that desensitization of open-unliganded and open-monoliganded AChRs is negligible (bold gray symbols). Thus, the kinetic scheme used for the computation of eigenvalues ended up being the one shown with bold black symbols, which further assumes that both transmitter binding sites are functionally equivalent and independent. Since choline-elicited activations consist largely of single openings (unlike ACh-elicited activations, for example), the Q_oo partition only includes states OA₂, OA, and O. The values of the rate constants used for the calculations were as follows. Choline-diliganded closing rate constant = 297 s⁻¹ (a putative value derived from Fig. 4 A and Fig. 8); choline-diliganded opening rate constant = 125 s⁻¹ (from Figs. 10 and 12); unliganded closing rate constant = 12,000 s⁻¹ (Grosman, 2003); unliganded opening rate constant = 1.2 × 10⁻³ s⁻¹ (from the arbitrary, but reasonable, assumption that the unliganded gating equilibrium constant is 10⁻⁷); choline-monoliganded closing rate constant = 8,500 s⁻¹ (a reasonable value, intermediate between its diliganded and unliganded counterparts); choline-monoliganded opening rate constant = 1.75 s⁻¹ (from detailed balance); choline-association rate constant to each closed-state transmitter binding site = 100 μM⁻¹ s⁻¹ (reasonable assumption); choline-dissociation rate constant from each closed-state transmitter binding site = 4.10⁵ s⁻¹ (from the K_D value of 4.1 mM in Purohit and Grosman, 2006); choline dissociation rate constant from each open-state transmitter binding site = 1,806 s⁻¹ (from the putative value of 3,612 s⁻¹ for the dissociation rate of choline from the diliganded open state, as discussed in Discussion); choline association rate constant to each open-state transmitter binding site = 895 μM⁻¹s⁻¹ (from detailed balance); entry-into-desensitization rate constant = 32 s⁻¹ (from Fig. 8); recovery-from-desensitization rate constant = 0.01 s⁻¹ (reasonable assumption). The reciprocal of the computed eigenvalues are shown in Fig. 14.

Figure 14.

What causes the prolongation of (apparent) openings? The reciprocal of the smallest eigenvalues of the open-state −Q_oo submatrix (i.e., the values of the slowest open-time constants) were numerically computed for the kinetic scheme in Fig. 13 at choline concentrations between zero and 1 M (dashed line). The figure also replots the data points in Fig. 4 A (i.e., the experimentally estimated values of the slowest open-time constants), along with the fit with Eq. 3 (solid line). It is evident that the “prolongation effect” of choline is due to a local effect of choline block on the closure of the activation gate (solid line), rather than to the increasing binding of choline to the open-state transmitter binding sites (dashed line). The shape of the dashed line plot largely depends on the values of the rate constants of diliganded-channel closing, entry into desensitization, and agonist association/dissociation to/from the open-state transmitter binding sites. The plot is rather insensitive to all other parameters in the kinetic scheme of Fig. 13 and to the assumption of equivalence and independence of the transmitter binding sites. It is interesting to realize that, as a general phenomenon in ligand-gated ion channels, the prolongation of open intervals with increasing ligand concentrations is not necessarily due to the slower closing rate constant of the open-blocked channel. As shown by the dashed line, the “open-state binding” hypothesis makes a comparable, yet clearly distinguishable, prediction.

Figure 15.

Relative dimensions of choline and the AChR's closed-channel pore. The five M2 segments (black) and a molecule of choline (red) are shown in surface representation (probe radius = 1.4 Å). The M1, M3, and M4 segments from the five subunits (orange) are shown in ribbon representation. Along the pore's longest axis, the choline molecule was positioned at the intracellular entrance of the pore, at the level of the selectivity filter (between the −2' and 2' positions of M2; Corringer et. al., 1999). Views from both the extracellular and the intracellular compartments are displayed. The protein atomic coordinates were taken from the PDB file 1OED (Miyazawa et al., 2003), a model of the transmembrane portion of the closed conformation of the Torpedo AChR. The atomic coordinates of choline were taken from the crystal structure of choline tetraphenylborate. The molecular image was made with VMD (Humphrey et al., 1996).