Divergence of allosteric effects of rapacuronium on binding and function of muscarinic receptors

Background Many neuromuscular blockers act as negative allosteric modulators of muscarinic acetylcholine receptors by decreasing affinity and potency of acetylcholine. The neuromuscular blocker rapacuronium has been shown to have facilitatory effects at muscarinic receptors leading to bronchospasm. We examined the influence of rapacuronium on acetylcholine (ACh) binding to and activation of individual subtypes of muscarinic receptors expressed in Chinese hamster ovary cells to determine its receptor selectivity. Results At equilibrium rapacuronium bound to all subtypes of muscarinic receptors with micromolar affinity (2.7-17 μM) and displayed negative cooperativity with both high- and low-affinity ACh binding states. Rapacuronium accelerated [3H]ACh association with and dissociation from odd-numbered receptor subtypes. With respect to [35S]GTPγS binding rapacuronium alone behaved as an inverse agonist at all subtypes. Rapacuronium concentration-dependently decreased the potency of ACh-induced [35S]GTPγS binding at M2 and M4 receptors. In contrast, 0.1 μM rapacuronium significantly increased ACh potency at M1, M3, and M5 receptors. Kinetic measurements at M3 receptors showed acceleration of the rate of ACh-induced [35S]GTPγS binding by rapacuronium. Conclusions Our data demonstrate a novel dichotomy in rapacuronium effects at odd-numbered muscarinic receptors. Rapacuronium accelerates the rate of ACh binding but decreases its affinity under equilibrium conditions. This results in potentiation of receptor activation at low concentrations of rapacuronium (1 μM) but not at high concentrations (10 μM). These observations highlight the relevance and necessity of performing physiological tests under non-equilibrium conditions in evaluating the functional effects of allosteric modulators at muscarinic receptors. They also provide molecular basis for potentiating M3 receptor-mediated bronchoconstriction.


Background
Five subtypes of muscarinic acetylcholine receptors that belong to class A of G-protein coupled receptors have been identified [1]. The primary response of stimulation of the M 2 and M 4 subtypes of muscarinic receptors is activation of the G i/o class of G-proteins resulting in inhibition of adenylyl cyclase, whereas stimulation of M 1 , M 3 , and M 5 receptors leads to activation of the G q/11 class of Gproteins and stimulation of phospholipase C [2]. Muscarinic receptors mediate many diverse physiological functions that are selectively mediated by different receptor subtypes [3]. This is why discovery of selective ligands is of prime importance for clinical practice. However, due to the very conserved nature of the orthosteric binding site of muscarinic acetylcholine receptors the selectivity of orthosteric agonists is very poor [4]. Orthosteric antagonists that bind to less conserved amino acids located close to the orthosteric binding site display better selectivity than orthosteric agonists. Muscarinic allosteric ligands exhibit remarkable selectivity among receptor subtypes [5]. They interact mainly with the second and the third extracellular loops that are much less conserved than transmembrane segments creating the orthosteric binding site [6][7][8][9][10].
The extraordinary selectivity of allosteric modulators that is due to differences in both affinity and cooperativity [11] has attracted attention of pharmacologists in the past decade. Somewhat paradoxically, most of originally discovered and probably best studied allosteric compounds of muscarinic receptors are neuromuscular blockers [12][13][14]. By definition, these are competitive nicotinic acetylcholine receptor antagonists but many of them have high affinities and strong allosteric interactions, particularly at the M 2 subtype of muscarinic receptors.
In clinical practice, different competitive (nondepolarizing) neuromuscular blockers are employed to induce muscle relaxation to facilitate intubation during surgery. The neuromuscular blocker rapacuronium was withdrawn from clinical use due to high incidence of bronchospasm resulting in death [15]. Parasympathetic innervation of airways transmits signal via postsynaptic M 3 receptors that mediate acetylcholine-induced contraction and M 2 receptors that inhibit with high potency smooth muscle relaxation mediated by increase in cytoplasmic cAMP [16]. M 2 receptors are also located at parasympathetic cholinergic nerve terminals innervating smooth muscle and their stimulation inhibits acetylcholine (ACh) release [17]. In functional experiments on the guinea pig trachea preparation it was demonstrated that rapacuronium preferentially antagonizes M 2 over M 3 muscarinic receptors [18]. In addition, involvement of allosteric potentiation of ACh binding to muscarinic M 3 receptors in bronchospasm induced by rapacuronium was suggested, but not proven [19]. A very recent paper confirmed a unique behavior of rapacuronium compared to other skeletal muscle relaxants in vivo and demonstrated that rapacuronium potentiates bronchoconstriction evoked by both naturally released and exogenous acetylcholine, indicating an important role of postsynaptic M 3 receptors [20].
Because we have been interested in investigations of positive cooperativity of allosteric ligands with ACh binding [11,21] and allosteric agonists [22] these findings led us to analyze in detail the interactions of rapacuronium with acetylcholine binding and receptor activation of all subtypes of muscarinic receptors heterologously expressed in membranes of Chinese hamster ovary (CHO) cells. We demonstrate that rapacuronium binds to and exhibits negative cooperativity with ACh binding at all subtypes of muscarinic receptors. Surprisingly, low concentrations of rapacuronium potentiate ACh-induced signaling at the M 1 , M 3 , and M 5 receptor subtypes and accelerate ACh binding. This striking behavior is unparallel at other neuromuscular blockers.

Results
Saturation binding experiments ( Figure 1; Table 1) with 68 pM to 2 nM [ 3 H]NMS in cell membranes showed similar binding capacity (1 to 2 pmol of binding sites per mg of protein) and affinity (equilibrium dissociation constant (K D ) ranging from 205 pM at M 4 to 320 pM at M 2 receptors) for all receptor subtypes ( Figure. 1 Figure 3, Table 2). This is an established hallmark of allosteric receptor modulation. It had the strongest effect at M 2 receptors (7-fold decrease in rate of dissociation) and weakest effect at M 3 and M 5 receptors (40% decrease). While dissociation evoked by NMS was monophasic ( Figure 3 closed symbols) it became biphasic in the presence of 100 μM rapacuronium with the exception of the M 5 subtype. binding to all subtypes, as evidenced by a maximal limit to its effects on the affinity of the radioligand that differed as a function of radioligand concentration. These effects were strongest at the M 5 subtype (35-fold decrease in affinity) and weakest at the M 2 subtype (6.8-fold decrease in affinity; Figure 5, closed circles). While cooperativity of rapacuronium with high-(pα) and low-affinity (pβ) ACh binding was essentially the same at individual subtypes (  Negative logarithms of equilibrium dissociation constants (pK D ) and maximum binding capacities (B MAX in fmol/μg of protein) of radioligands were obtained from saturation experiments shown in Figure 1 by fitting Eq. 1 to the data. Values are means ± SE of fits to 3 independent experiments performed in quadruplicates. *P < 0.05; significantly different from control (radioligand alone) by ANOVA and Tukey-Kramer post-test.   Table 2.  (Table 4 and Figure 6, closed circles). The maximal effect of ACh (E MAX ) was about two-fold increase in basal binding at odd-numbered receptors and three-fold increase at evennumbered receptors with a rank order of efficacy of M 2 >M 4 >M 1 >M 5 >M 3 (range from 3.12 to 1.99-fold increase). In control conditions ACh EC 50 values were lower at even-numbered subtypes than at odd-numbered subtypes with a rank order of potency of M 2 >M 4 >M 3 >M 5 >M 1 (range from 0.25 to 6.31 μM) ( Table  5). While the EC 50 of ACh-stimulated [ 35 S]GTPγS binding was less than that of its low-affinity binding conformation by 178-times at M 2 and 23-times at M 4 receptors it was only 7.4-, 5.4-, and 4.7-times lower at M 1 , M 3 , and M 5 receptors, respectively. In comparison with its high-affinity binding, the EC 50 of ACh-stimulated [ 35 S]GTPγS binding was only 10-times higher at M 2 and 55-times at M 4 receptors but 130-260-and 300-times higher at M 1 , M 5 and M 3 receptors, respectively. E MAX was about two-fold increase in basal binding at odd-numbered receptors and three-fold increase at even-numbered receptors with a rank order of efficacy of M 2 >M 4 >M 1 >M 5 >M 3 (range from 3.12 to 1.99-fold increase) ( Table 5).
Measurements of ACh-stimulated [ 35 S]GTPγS binding in the presence of 0.1, 1 and 10 μM rapacuronium showed differential effects of rapacuronium on receptor activation by an orthosteric agonist at individual receptor subtypes (Figure 6 open symbols). At even-numbered subtypes 1 μM and 10 μM rapacuronium significantly increased ACh EC 50 , with lowering of E MAX at 10 μM rapacuronium. These results are in line with the effects of rapacuronium Negative logarithm of equilibrium dissociation constant of rapacuronium (pK A ) and factors of cooperativity (β) between rapacuronium and acetylcholine low affinity binding were obtained by fitting Eq. 4 to the data in Figure 5. Factors of cooperativity α and β are expressed as negative logarithms so that negative values represent negative cooperativity. Values are means ± SE of fits to 3 independent experiments performed in quadruplicates.  Table 3. Curves are fits of Eq. 3 (circles) and Eq. 4 (squares and triangles) to data prior to normalization. Binding parameters are summarized in Table 3.  Table 7).   Tables 4 and 5.  Table 7).

Discussion
Our results clearly demonstrate that the neuromuscular blocker rapacuronium binds to all muscarinic receptor subtypes at physiologically relevant concentrations [18] and displays micromolar affinity and slight selectivity towards M 2 receptor. This selectivity is smaller than that of other neuromuscular blockers such as alcuronium, gallamine and pancuronium [23,24,Jakubík,unpublished data]. Like the majority of this class of compounds, rapacuronium acts as a negative allosteric modulator (alters dissociation kinetics and incompletely inhibits binding of orthosteric ligands) with respect to binding of both the natural agonist ACh (Figures 1, 4, 5, 8 and 9) and the classical antagonist NMS (Figures 1, 3, and 5). Rapacuronium exhibits complex effects on the kinetics of ACh binding (Figures 8 and 9) and subsequent receptor activation estimated from stimulation of [ 35 S]GTPγS binding (Figures 6 and 7). Functional effects differ from those of the prototypic negative allosteric modulators alcuronium and gallamine ( Figure 10, Table 7).  Figure 6. Values are means ± SE of fits to 3 independent experiments performed in quadruplicates. *P < 0.05; significantly different from control (Ach alone) by ANOVA and Tukey-Kramer post-test.   Table 6. Our observation of an allosteric mode of interaction between rapacuronium and muscarinic receptors is in agreement with reported slowing-down of NMS dissociation from M 2 and M 3 receptors by this drug [19]. The observed biphasic dissociation of NMS under non-equilibrium conditions in the presence of an allosteric modulator such as rapacuronium was described earlier [24].

Inverse receptor agonism by rapacuronium
Rapacuronium alone decreases [ 35 S]GTPγS binding. This effect is mediated by muscarinic receptors because it is not observed in membranes prepared from a native CHO cell line that does not express muscarinic receptors and thus cannot be explained by nonspecific effects on cell membranes. Instead, this effect can be related to an inverse agonistic effect of rapacuronium itself on constitutive receptor activity. This view is supported by previous demonstration of constitutive activity of muscarinic receptors [25,26] and by finding that the orthosteric antagonists NMS and atropine also decrease [ 35 S]GTPγS binding when applied alone [22,27]. In addition, both agonistic and inverse agonistic effects of allosteric modulators have already been observed [27,28].

Allosteric modulation of receptor activation by rapacuronium Both [ 3 H]ACh saturation binding experiments (Figure 1) and ACh vs. [ 3 H]NMS competition experiments (Figure 2)
show ACh high affinity binding in the nanomolar range without selectivity towards any of muscarinic receptor subtypes. The affinities of ACh at M 2 and M 4 receptors reported in this study are within the range of published values, being lower than those published by Lazareno et al. [11] but higher than the values reported by Haga et al. [29] or Gurwitz et al. [30]. This divergence is likely due to the dependence of the affinity of acetylcholine at its highaffinity site on many factors (e.g. receptor source, preparation, concentration of ions (mainly Mg 2+ , Na + ), residual concentration of GDP, temperature, etc.). Similarly, we found no subtype differences in ACh low affinity binding, which is in accordance with our previous studies [21]. Despite lack of binding selectivity, the potency and efficacy of ACh in stimulating [ 35 S]GTPγS binding are significantly higher at even-numbered than at odd-numbered subtypes. In other words, the M 2 and M 4 subtypes that preferentially couple with G i/o G-proteins display better coupling and larger receptor reserve than the M 1 , M 3 , and M 5 subtypes that preferentially couple with G q/11 G-proteins. Despite accumulating evidence for the existence of agonist-specific conformations of muscarinic and other G-protein-coupled receptors [31][32][33] it is generally accepted that the change in agonist potency in receptor activation follows a change in the affinity of its binding induced by an allosteric modulator. Thus, negative cooperativity between the allosteric modulator and the binding of an orthosteric agonist would lead to lower potency of agonist (e.g. pioneering experiments with gallamine of Clark and Mitchelson [12]) and positive cooperativity would result in higher potency of agonist [11,34]. Rapacuronium behaves in accordance with this view in case of the M 2 and M 4 subtypes. However, at the M 1 , M 3 and M 5 receptor subtypes, rapacuronium up to a concentration of 10 μM either increases or does not alter ACh potency or efficacy in inducing [ 35 S]GTPγS binding ( Figure 6), despite clear negative cooperativity with ACh binding (Figures 4 and 5). Although this observation may appear surprising at first glance it is perfectly in agreement with the hypothesis of multiple receptor conformations induced by orthosteric and allosteric ligands, and with the existence of conformations that exhibit low affinity for agonist binding but nevertheless activate second messenger pathways [26,31,35,36].

Kinetics of functional response
Analysis of the kinetics of [ 35 S]GTPγS binding shows that the facilitatory effects of rapacuronium on ACh-induced responses are evident after brief incubations (lasting minutes, Figure 7). This suggests that the facilitating effects of rapacuronium on ACh-induced response are a consequence of altered receptor kinetics rather than a change in agonist affinity at equilibrium. Extended time of incubation during which binding of ligands equilibrates may thus obscure the initial transient potentiation. Analysis of kinetics of ACh binding (Figures 8 and 9) showed that rapacuronium affects ACh kinetics differently than those of NMS. While rapacuronium slows down NMS association and dissociation at all receptor subtypes (Figure 3) it accelerates ACh association and dissociation at odd-numbered subtypes (Figures 8 and 9). Thus, rapacuronium doubles the magnitude of ACh binding at 15 seconds at these receptors such that association after 15 seconds is twice as much in the presence of rapacuronium. This effect, however, is counterbalanced by accelerated dissociation, resulting in an overall decrease in ACh affinity (neg-  Figure 8 and Eq. 7c to data in Figure 9. Values are means ± SE of fits to 3 independent experiments performed in quadruplicates. *P < 0.05; significantly different from control ([ 3 H]ACh alone) by t-test.  Table 6. ative cooperativity). Although combination of negative binding cooperativity on the one hand and acceleration of binding on the other could in principle be interpreted within the frame of the ternary receptor model. However, data of association and dissociation of ACh in the absence of rapacuronium do not conform to a simple bi-molecular interaction. As a result, the interaction between ACh and rapacuronium at muscarinic receptors is more complex and may involve allosteric extension of the tandem two-site model [37,38]. Theoretically, this extension of the model allows for coexistence of positive cooperativity between rapacuronium and the initial step of ACh binding and overall negative binding cooperativity under equilibrium. An enigmatic feature of our data, however, is that low concentrations of rapacuronium (0.1 and 1 μM) that do not affect the the rate of binding of ACh or its affinity at equilibrium at odd numbered subtypes leads to an increase in both potency and efficacy of ACh in receptor activation. Theoreticaly, allosteric extension of tandem two-site model allows for positive cooperativity between rapacuronium and ACh initial binding step in overall negative binding cooperativity under equilibrium and transient binding of these sub-threshold concentrations. However, these concentrations of rapacuronium had no effect on ACh association in binding experiments ( Figure  8, open circles). One possible explanation is that ACh bound to a peripheral site (of tandem two-site binding) is lost during filtration but is well reflected and amplified in GTPγS binding that is pseudo-irreversible. A more speculative explanation assumes that rapacuronium at submicromolar concentrations binds to another site on the receptor and facilitates receptor activation by ACh without significant interference with radioligand binding. This facilitatory effect is overcome at high concentrations of rapacuronium by negative cooperativity in binding of ACh induced by binding of rapacuronium to an allosteric binding site. A latent further increase in ACh binding after 5 min in the presence of 10 μM rapacuronium (Figure 8, hatched circles) suggests an even more complex mechanism of interaction of rapacuronium wih the receptor.

Effect of rapacuronium on the time course of [ 3 H]ACh dissociation
Modeling of such complex kinetics would require a model even more sophisticated than ternary extension of the tandem-two site model [38]. Additionally, differential effects of low concentrations of rapacuronium (1 μM and lower) on receptor binding and function would require inclusion of receptor activation (probably with several ligand-specific activation states) in the model and therefore renders modeling unachievable.
Comparison of the effects of rapacuronium with those of the prototypic allosteric modulators alcuronium and gallamine (Figures eleven and twelve) on M 3 receptors shows that acceleration of ACh kinetics is unique to rapacuronium among negative allosteric modulators. To our knowledge this is the first report of acceleration of binding of an orthosteric ligand by a negative allosteric modulator. This highlights unpredictability of kinetics of allosteric modulation based on compounds with similar behavior observed under equilibrium.

Physiological implications
Our observations are consistent with functional ex vivo and in vivo physiological experiments demonstrating an increase of acetylcholine-evoked muscle contraction of guinea pig trachea rings by rapacuronium [18][19][20].
Although they confirm proposed allostetic interaction between rapacuronium and ACh [19] they do not conform to the proposed positive binding cooperativity at the M 3 receptor subtype. Although rapacuronium at concentrations below 10 μM binds to and decreases the affinity of acetylcholine at equilibrium at all subtypes of muscarinic receptors, it accelerates association of ACh and enhances its potency and efficacy in functional responses at the M 3 receptor as evident from [ 35 S]GTPγS binding. The initial acceleration of the rate of association of ACh Values of observed rates of association (k obs ) and equilibrium binding (B eq ) of 40 nM [ 3 H]ACh with M 3 receptors were obtained by fitting Eq. 6 to data in Figure 10 Table 7.  [16]. In contrast, rapacuronium at clinically relevant concentrations strongly reduces the affinity of binding of ACh and also its potency and efficacy in activating M 2 receptors. This pattern of effects should lead to an increase in ACh release by interrupting its M 2 receptor-mediated presynaptic autoinhibition [17] and to the inhibition of postsynaptic M 2 receptor-mediated muscle relaxation. In contrast, the decrease of ACh affinity at the M 2 and M 4 subtypes is accompanied by a decrease in both potency and efficacy of stimulating

Conclusions
Although rapacuronium exerts negative cooperativity with binding of ACh to all muscarinic receptor subtypes at equilibrium it accelerates the rate of ACh binding at odd numbered subtypes. At concentrations below 10 μM, it increases the potency and efficacy of ACh in increasing the rate of [ 35 S]GTPγS binding at odd-numbered subtypes.
The time between acetylcholine release and termination of its action by acetylcholinesterase is in the range of a fraction of a second. Therefore, the effects of allosteric modulators in the early non-equilibrium stage of receptor signaling are therapeutically more important than effects on acetylcholine equilibrium binding, as the latter conditions do not occur in vivo. Our study demonstrates a case of dichotomous effects of the allosteric modulator rapacuronium on ACh equilibrium binding on the one hand and on the kinetics of ACh binding on the other. Our observations emphasize the necessity to employ fast functional assays in screening for potential allosteric modulators of neurotransmission that much better simulate physiological conditions than long-lasting equilibrium binding experiments.

Radioligand binding
All radioligand binding experiments were carried out on membranes in 96-well plates at 30°C in the incubation medium described above supplemented with freshly prepared dithiothreitol at a final concentration of 1 mM, essentially as described by Jakubík et al. [23]. ]GTPγS was carried out for 20 min and free ligand was removed by filtration as described above. Filtration and washing with ice-cold water lasted for 9 s (wash-aspirate button time).
After filtration filters were dried in vacuum for 1 h while heated at 80°C and then solid scintillator Meltilex A was melted on filters (105°C, 90 s) using a hot plate. After cooling the filters were counted using a Wallac Microbeta scintillation counter.

Data analysis
In general binding data were analyzed as described in Jakubík et al. [21]. Data were preprocessed by Open Office 3.0 http://www.openoffice.org and subsequently analyzed by Grace 5.1.18 http://plazma-gate.weizman.ac.il/ and statistics package R http://www.r-project.org on Mandriva distribution of Linux.
The following equations were fitted to data: in the presence of rapacuronium at concentration x normalized to the absence of rapacuronium; [L] concentration of radioligand; K D , equilibrium dissociation constant of radioligand; K A ; equilibrium dissociation constant of rapacuronium; α, factor of cooperativity between radioligand and rapacuronium [40]. Cooperativity factor greater than 1 denotes negative cooperativity and less than 1 positive cooperativity. Due to its log-normal error distribution factors of cooperativity are expressed as negative logarithms (pα) through the manuscript so negative values denotes negative cooperativity and positive value denotes positive cooperativity.
Concentration-response y, radioactivity in the presence of agonist at concentration x normalized to radioactivity in the absence of agonist; E MAX , maximal increase by agonist; EC 50 , concentration of agonist producing 50% of maximal effect; nH, Hill coefficient.
Time course of association y, radioligand binding at time x; k obs , observed rate of association; equilibrium binding B eq = Bottom + Span.
Time course of dissociation or or y, radioligand binding at time x normalized to binding at time 0; k off1 and k off2 , dissociation rate constants; f 2 , fraction of binding site with dissociation rate constant k off2 . When both Eq. 7a and 7b were fitted to data the better fit was chosen based on sum of squares F-test and runs test.
For fitting parameter estimates close to one expected were entered manually, parameters were constrained to reasonable range, the tolerance value was set to 0.01 and iteration steps to 30. Initial values of slope factors were always 1 constrained to 0.8 to 1.2 range.