Visual Overview
Abstract
Nicotinic acetylcholine receptors (nAChRs) are ligand-gated ion channels involved in neuromuscular transmission. In nematodes, muscle nAChRs are targets of antiparasitic drugs. Bephenium is an anthelmintic compound whose molecular action in the free-living nematode Caenorhabditis elegans, which is a model for anthelmintic drug discovery, is poorly known. We explored the effect of bephenium on C. elegans locomotion and applied single-channel recordings to identify its molecular target, mechanism of action, and selectivity between mammalian and C. elegans nAChRs. As in parasites, bephenium paralyzes C. elegans. A mutant strain lacking the muscle levamisole-sensitive nAChR (L-AChR) shows full resistance to bephenium, indicating that this receptor is the target site. Bephenium activates L-AChR channels from larvae muscle cells in the micromolar range. Channel activity is similar to that elicited by levamisole, appearing mainly as isolated brief openings. Our analysis revealed that bephenium is an agonist of L-AChR and an open-channel blocker at higher concentrations. It also activates mammalian muscle nAChRs. Opening events are significantly briefer than those elicited by ACh and do not appear in activation episodes at a range of concentrations, indicating that it is a very weak agonist of mammalian nAChRs. Recordings in the presence of ACh showed that bephenium acts as a voltage-dependent channel blocker and a low-affinity agonist. Molecular docking into homology-modeled binding-site interfaces represent the binding mode of bephenium that explains its partial agonism. Given the great diversity of helminth nAChRs and the overlap of their pharmacological profiles, unraveling the basis of drug receptor-selectivity will be required for rational design of anthelmintic drugs.
Introduction
The prevalence of human and animal infections with helminths, commonly known as parasitic worms, remains staggeringly high, thus the urgent need for concerted efforts in this area of research. Parasitic nematodes (roundworms) are not ideal laboratory animals because of the difficulty in genetic manipulation and the need for infecting host animals. In this context, the free-living nematode Caenorhabditis elegans has emerged as a powerful model system for anthelmintic drug screening: It shares physiologic and pharmacological characteristics with parasitic nematodes, it is no more dissimilar to parasitic nematodes than each individual species of parasite is to one another, and it is sensitive to most anthelmintic drugs (Holden-Dye and Walker, 2007).
Nicotinic receptors (nAChRs), which have a key role in muscle contraction and locomotion in vertebrates and invertebrates, are targets of anthelmintic agents. They belong to the Cys-loop receptor family of pentameric ligand-gated ion channels, which contain a large extracellular domain that carries the agonist binding sites and a transmembrane domain that forms the ion channel (Unwin, 2005; Kalamida et al., 2007; Cecchini and Changeux, 2015; Bouzat and Mukhtasimova, 2018). The agonist binding sites are located at subunits interfaces, where one subunit contributes to the principal face and the adjacent one, to the complementary face. Only α-type subunits, which contain a conserved disulfide bridge in loop C of the binding site, form the principal face of the binding site, whereas the complementary face can be formed by either α or non-α subunits (Sine, 2012).
Vertebrates contain 17 nAChR subunits, which yield a variety of receptors. The mammalian muscle nAChR is formed by four different subunits, α1, β1, δ, and γ or ε depending on whether it is embryonic or adult muscle, respectively, with an (α1)2βδ ε/γ composition. The two agonist binding sites occur at α1/δ and α1/γ or ε subunit interfaces. C. elegans possesses an extensive gene family of nAChR subunits (Jones et al., 2007; Jones and Sattelle, 2008). At its neuromuscular junction, the two main nAChRs are the nicotine-sensitive (N-AChR) and the levamisole-sensitive (L-AChR) (Richmond and Jorgensen, 1999). The L-AChR is the target of nematocide agents, such as levamisole and pyrantel, which, by acting as potent agonists, produce body-wall muscle hypercontraction, paralysis, and ultimately death. It is composed of five different subunits, three of which are essential (UNC-63, UNC-38, and UNC-29) and the other two (LEV-8 and LEV-1) nonessential (Boulin et al., 2008; Hernando et al., 2012). The disposition of the five subunits in the pentameric arrangement remains unknown. Null mutants of essential subunits show uncoordinated phenotypes and resistance to anthelmintic agents, whereas only partial resistance is observed in null mutants of nonessential subunits (Lewis et al., 1980). UNC-63, UNC-38, and LEV-8 are α-type subunits and can contribute to the principal face of the binding site. ACR-8, a muscle α-type subunit homologous to LEV-8 (Mongan et al., 1998), acts as a spare subunit for C. elegans L-AChR (Hernando et al., 2012).
Parasitic nematodes contain nAChR subunit genes that show high similarity with their C. elegans counterparts. Whereas C. elegans muscle contains two main nAChR subtypes (L- and N-) of fixed subunit composition, parasitic nematode muscle has a greater diversity of nAChRs owing to subunit duplication and to different arrangements of the orthologous subunits (Qian et al., 2006; Boulin et al., 2008, 2011; Buxton et al., 2014; Duguet et al., 2016; Verma et al., 2017). In different species, L-, N-, P- (pyrantel), M- (morantel), and B- (bephenium) subtypes have been described (Martin et al., 1997; Robertson et al., 2002; Qian et al., 2006; Verma et al., 2017).
Bephenium is an anthelmintic compound used against human and dog hookworms and gastrointestinal parasitic nematodes in sheep (Burrows, 1958; Copp et al., 1958; Young et al., 1958). Unlike levamisole and pyrantel, the activity of bephenium on the parasite model C. elegans as well as its mechanism and target sites are poorly known. Here we reveal the effect of this drug on C. elegans locomotion and we identify its molecular target, mechanism of action, structural basis of action, and differences in activation between mammals and C. elegans.
Materials and Methods
Drugs.
Acetylcholine chloride, levamisole hydrochloride, and bephenium hydroxynaphthoate were from MilliporeSigma (Merck, KGaA, Darmstadt, Germany).
Caenorhabditis elegans Strains and Culture.
Nematode strains used were: N2 Bristol wild-type; PD4251: ccIs4251;dpy-20(e1282), which produces green fluorescence protein (GFP) in all body wall muscles and vulval muscles; ZZ15: lev-8(x15), a lev-8 null mutant strain with partial resistance to levamisole; CB904: unc-38(e264)I, which lacks UNC-38 essential subunit that integrates L-AChR. All nematodes strains were obtained from the Caenorhabditis Genetic Center, which is funded by the NIH National Center for Research Resources. Nematode strains were maintained at 18–25°C using freshly prepared nematode growth medium (NGM) Petri dishes that had been spread with Escherichia coli (OP50) as a source of food (Rayes et al., 2007; Hernando, et al., 2012).
Isolation and Culture of C. elegans Embryonic Muscle Cells.
Embryonic cells were isolated and cultured as previously described (Christensen et al., 2002; Rayes et al., 2007). Briefly, adult nematodes were exposed to an alkaline hypochlorite solution (0.5 M NaOH and 1% NaOCl) and the eggs released were treated with 1 IU/ml chitinase (MilliporeSigma Co., St. Louis, MO) for 30–40 minutes at room temperature. The embryo cells were isolated by gently pipetting and filtering through a sterile 5-μm Durapore syringe filter (EMD Millipore, Bedford, MA) to remove undissociated embryos and newly hatched larvae. Filtered cells were plated on glass coverslips coated with poly-O-ornithine. Cultures were maintained at 24°C in a humidified incubator in L-15 medium (Hyclone, Logan, UT) containing 10% fetal bovine serum. Complete differentiation to the various cell types that comprise the newly hatched L1 larva were observed within 24 hours. Electrophysiological recordings were performed 1–5 days after cell isolation. Muscle cells are easily identifiable owing to their spindle-shaped morphology, which resembles that of body wall muscle cells in vivo (Christensen et al., 2002; Rayes et al., 2007; Hernando et al., 2012).
Isolation and Culture of C. elegans L2 Muscle Cells.
Adult nematodes were exposed to an alkaline hypochlorite solution (0.5 M NaOH and 1% NaOCl) and the eggs released were grown in petri dishes spread with dead E. coli (OP50) to avoid cell culture contamination. Synchronized larva 2 (L2) stage were used for cells isolation in sterile conditions. The worms were exposed to a SDS-dithiothreitol solution to weaken the cuticle. Then, the pronase E enzyme was added to disrupt the cuticle and release live cells from larval worms. When combined with mechanical disruption by pipetting, pronase and SDS-dithiothreitol treatment dissociated tissues and released single cells very efficiently. Filtered cells were plated on glass coverslips and maintained in culture like the embryonic muscle cells (Zhang et al., 2011; Zhang and Kuhn, 2013; Spencer et al., 2014). Electrophysiological recordings were performed 1–5 days after cell isolation.
Paralysis Assays.
Drug sensitivity was determined by paralysis assays on agar plates with young adult hermaphrodite worms from synchronized plates. Worms (about 30–50 animals) were placed on agar plates containing the tested drug at room temperature as described before (Hernando et al., 2012). Body paralysis was defined as the lack of complete body movement in response to prodding. The paralysis was evaluated under magnifying glass at the indicated time (every 30–60 minutes).
Mammalian Cell Expression of Muscle nAChR.
BOSC cells [derived from human embryonic kidney 293 cells, Pear et al., 1993] were transfected with mouse adult muscle nAChR subunit cDNAs using calcium phosphate precipitation, as described previously (Bouzat et al., 1994, 2000). The subunit cDNA ratio was 2α1:1β:1ε:1δ, with a total cDNA amount of 5 μg/35-mm dish. A plasmid encoding green fluorescent protein was also included to allow identification of transfected cells under fluorescence optics. Cells were used for single-channel measurements 2 days after transfection.
Patch-Clamp Recordings.
Single-channel currents were recorded in the cell-attached patch configuration at 20°C (Rayes et al., 2007; Hernando et al., 2012). The bath and pipette solutions contained 140 mM KCl, 5.4 mM NaCl, 1.8 mM CaCl2, 1.7 mM MgCl2, and 10 mM HEPES (pH 7.4). The stock solution for bephenium was prepared in dimethyl sulfoxide and the final dimethyl sulfoxide concentration used in all assays was lower than 0.1%. Single-channel currents were recorded using an Axopatch 200 B patch-clamp amplifier (Molecular Devices, Inc., Sunnyvale, CA), digitized at 10-microsecond intervals with the PCI 6111E interface (National Instruments, Austin, TX), recorded using the program Acquire (HEKA Instruments Inc., Bellmore, NY), and detected by the half-amplitude threshold criterion using the TAC 4.0.10 program (Bruxton Corporation, Seattle, WA) at a final bandwidth of 9 kHz. Open- and closed-time histograms were plotted using a logarithmic abscissa and a square root ordinate and fitted to the sum of exponential functions by maximum likelihood using the program TACFit (Bruxton Corporation). Clusters of channel openings were identified as a series of closely separated openings preceded and followed by closings longer than a critical duration (τcrit). This duration was taken as the point of intersection of the predominant closed component and the succeeding one (longer duration) in the closed time histogram (Bouzat et al., 2000; Corradi et al., 2009). For the mammalian nAChR, τcrit was defined by the point between the main closed component, which is sensitive to ACh concentration, and the previous one (typically 10–16 milliseconds for 30 μM ACh at 70 mV pipette potential). Cluster durations were determined from the longest duration component of open-time histograms, with τcrit set to the corresponding value, and therefore openings separated by closing briefer than this time constituted a cluster.
Macroscopic currents were recorded in the whole-cell configuration as described before (Gumilar et al., 2003). Briefly, the pipette solution contained 134 mM KCl, 5 mM EGTA, 1 mM MgCl2, and 10 mM HEPES (pH 7.3). The extracellular solution contained 150 mM NaCl, 0.5 mM CaCl2, and 10 mM HEPES (pH 7.4). Pulses of 1.5 seconds of extracellular solution containing ACh or bephenium were applied. The solution exchange time was estimated by the open pipette and varied between 0.1 and 1 millisecond.
Molecular Modeling and Docking.
Homology models of the extracellular domain of two adjacent subunits were created using the structure of the human α4β2 nicotinic receptor in complex with nicotine (Protein Data Bank code 5KXI, Morales-Perez et al., 2016) and the chimera α7/AChBP (acetylcholine binding protein) crystalized with the partial agonist lobeline, which contains loop C partially closed (PDB 5AFM, Spurny et al., 2015). The amino acid sequences were aligned using ClustalW (http://www.ebi.ac.uk/Tools/msa/clustalw2/) and modeling was performed using MODELLER 9.19 (https://salilab.org/modeller/) (Sali et al., 1995). We tested the ligand docking in mouse α1/ε and α1/δ interfaces and in 10 different subunit interfaces of the L-AChR using the two different models for each interface. Ten models were generated for each interface, and the one with the lowest energy and the smallest percentage of amino acids in the disallowed region of the Ramachandran plot was selected for docking studies. ACh and bephenium were obtained from the PubChem database and docked separately using AUTODOCK 4.2 (Morris et al., 2009). A hundred genetic algorithm runs were performed for each condition. Clustering of the results was done with AUTODOCK on the basis of a root-mean-square deviation cut-off of 2.0 Å. To evaluate the capability of each interface to bind agonist, we compared the best binding energy (BBE) and the frequency of conformations that bind the agonist in the correct orientation in the binding pocket. Docking results were corroborated in two different procedures, and the most representative docking result was plotted with Discovery Studio Visualizer 4.5 (Dassault Systèmes BIOVIA, San Diego, 2016).
Statistics.
Experimental data are shown as mean ± S.D. Statistical comparisons were done using the Student’s t test or one-way ANOVA with Bonferroni’s multiple comparison post test. A level of P < 0.05 was considered statistically significant.
Results
Behavioral Effects of Bephenium on C. elegans.
To test if the anthelmintic drug bephenium is active in C. elegans, we synchronized populations of wild-type worms (N2 strain), exposed them to bephenium on agar plates, and measured the fraction of moving animals. We found that exposure to bephenium produced spastic paralysis. The decrease in the percentage of moving animals as a function of drug concentration revealed an IC50 value of 2.2 mM when animals were pre-exposed for 1 hour to bephenium at each tested concentration (Fig. 1A). As expected, identical behavior was observed in the PD4251 strain, which expresses GFP in muscle, and it was therefore used as a control strain for many assays.
Paralyzing effects of bephenium on C. elegans. Chemical structures of bephenium and levamisole are shown at the top of the figure. Assays were carried out on agar plates with synchronized populations of C. elegans at 20–22°C. (A) Dose-response curve for the paralyzing effect of bephenium (1-hour exposure) on wild-type C. elegans (N2 strain). The percentage of paralyzed animals was related to the control condition in the absence of bephenium and in the presence of the corresponding amount of dimethyl sulfoxide. Body paralysis was defined as the lack of complete body movement in response to prodding. Values are mean percentages ± S.D. (B) Comparison of the percentage of paralyzed animals containing (wild-type) or lacking L-AChRs (unc38(e264)I) after exposure of 1 or 2 hours to 3 mM bephenium. Values are mean percentages ± S.D. derived from seven dishes per condition. *** Indicates statistically significant differences with respect to the wild-type condition (P < 0.001). (C) Percentage of paralyzed worms exposed for 1 or 2 hours to 0.03 mM levamisole, 1 mM bephenium, or the two drugs combined. Values are mean percentages ± S.D. derived from three simultaneous assays for all conditions and two complete experiments performed on separate days. Statistical comparisons between conditions with drug combination and individual drugs (***P < 0.001; **P < 0.01; *P < 0.05).
Since one of the main targets of antiparasitic drugs is the levamisole-sensitive nAChR (L-AChR), we tested the action of bephenium on the strain unc-38(e264)I, which lacks the UNC-38 subunit, an essential component of L-AChR (Fleming et al., 1997; Richmond and Jorgensen, 1999; Culetto et al., 2004). In this strain, 3 mM bephenium paralyzed 3.00% ± 2.70% and 9.71% ± 3.95% of animals after 1- and 2-hour exposures, respectively. In contrast, for the PD4251 strain, which contains wild-type L-AChRs, the percentage of paralyzed animals was 59.57% ± 6.08% and 74.86% ± 7.17%, respectively (n > 25 worms per dish, n = 7 dishes per condition) (Fig. 1B). The resistance to bephenium of worms lacking functional L-AChRs indicates that this receptor is a key target involved in its paralyzing activity.
It has been proposed that in Haemonchus contortus the nicotinic subunit ACR-8 is key for bephenium sensitivity (Charvet et al., 2012). ACR-8 is present in C. elegans muscle as a spare subunit and it may replace LEV-8 in its absence (Hernando et al., 2012). We therefore determined the action of bephenium in the lev-8(x15) strain that lacks the LEV-8 subunit and may contain ACR-8 in the pentameric L-AChR arrangement (Hernando et al., 2012). Synchronized PD4251 (wild-type L-AChR) and mutant lev-8(x15) worms (L-AChR lacking LEV-8) were exposed to 1 mM bephenium on agar plates. Whereas 6.27% ± 4.67% and 8.65% ± 7.09% of mutant worms were paralyzed after 1 and 2 hours, respectively, 16.49% ± 8.09% and 25.43% ± 15.98% of control worms containing wild-type L-AChRs were paralyzed at 1 and 2 hours, respectively (n > 30 worms in each assay, n = 8 dishes per condition). At both hours of exposure, the percentage of paralyzed animals was statistically significantly different between control and mutant animals (P = 0.008 and P = 0.017 for 1 and 2 hours, respectively). Thus, the mutant strain is partially resistant to bephenium, as shown for levamisole (Hernando et al., 2012).
To determine how bephenium affects the paralyzing effect of levamisole, we performed paralysis assays in the presence of bephenium (1 mM) and levamisole (0.03 mM) after 1 and 2-hour exposures. These concentrations are close to the IC50 values for the paralyzing effect of each drug. We found that 79.69% ± 13.04% (1 hour) and 90.93% ± 10.08% (2 hours) of adult worms were paralyzed with the drug combination, whereas 38.01% ± 9.75% (1 hour) and 72.78% ± 10.98% (2 hours) were paralyzed in the presence of levamisole and 29.73% ± 6.62% (1 hour) and 54.33% ± 9.67% (2 hours) in the presence of bephenium (Fig. 1C) (n = 6 simultaneous assays for all conditions, with >30 worms per each dish). There were statistically significant differences in the percentage of paralyzed worms between the drug combination and each individual drug (P < 0.001). The fact that the paralyzing effects of both drugs are quite additive suggests that they act at the same target, the L-AChR.
Bephenium Activates L-AChRs from C. elegans L1 Muscle Cells.
To unequivocally determine the molecular target of bephenium, we performed single-channel recordings from cultured embryonic cells that differentiate in vitro to the larva 1 (L1) stage (Christensen et al., 2002). We first used the PD4251 strain, which produces GFP in all body wall muscles and allows their easy identification in culture.
As shown in Fig. 2A, 50 μM bephenium elicits single-channel currents from cell-attached patches from L1 muscle cells. Channel openings were not detected under similar conditions in the absence of bephenium (n = 6). Single-channel activity elicited by bephenium shows a homogenous population of opening events of 3.56 ± 0.62 pA (n = 20) (100-mV holding potential) that appear mainly as isolated events and less often in short activation episodes composed of two or more openings (bursts) and at a constant frequency throughout the recording (at least 10 minutes) (Fig. 2A).
Single-channel currents elicited by bephenium from cultured L1 muscle cells. (A) Traces of single-channel activity in the presence of 50 μM ACh or bephenium from wild-type L1 muscle cells are shown at two different temporal scales. At the right, representative open- and closed-duration histograms are shown for each condition. Channel openings are shown as upward deflections. Pipette potential: 100 mV. (B) Traces of single-channel activity in the presence of 30 μM bephenium at the indicated holding potential are shown filtered at 9 kHz with channel openings as upward deflections. Middle: Corresponding amplitude histograms. Right: Current-voltage relationship. The conductance was obtained by linear regression analysis. Each point corresponds to the mean ± S.D. of three different recordings at each condition. (C) Continuous recordings from muscle cells derived from the lev-8(x15) mutant strain at 50 μM ACh and 30 μM bephenium. Pipette potential: 100 mV. Filter: 9 kHz. Single-channel activity of L-AChRs lacking LEV-8 decreases markedly with time and occurs in clusters.
The pattern of channel activity and the amplitude of the opening events resemble those of L-AChR (Rayes et al., 2007; Hernando et al., 2012). From this receptor, single-channel openings are readily detected in the presence of ACh (0.1–300 μM) or levamisole (0.1–100 μM) but not in its absence (Rayes et al., 2007; Hernando et al., 2012). At 50 μM ACh, channel activity appears as high-frequency opening events of amplitude similar to those activated by bephenium (3.92 ± 0.39 pA at 100 mV pipette potential, n = 8, P = 0.132). As with bephenium, opening events occur mainly isolated or in short bursts (Fig. 2A) and activity remains constant at least during a 7-minute recording (Hernando et al., 2012).
We recorded single-channel currents activated by ACh or bephenium at a range of holding potentials and constructed voltage-current relationships. The estimated single-channel conductance for bephenium was 34.8 ± 0.2 picosiemens (pS) (Fig. 2B), similar to that determined for ACh (35.0 ± 1.2 pS; Hernando et al., 2012), thus suggesting, once more, that channel activity elicited by bephenium arises from the L-AChR.
To unequivocally confirm this statement, we made single-channel recordings from L1 muscle cells obtained from the unc-38(e264)I strain that lacks the essential UNC-38 subunit. No single-channel activity was detected from L1 cells derived from this strain, which lacks L-AChRs either in the presence of 50 μM ACh (Rayes et al., 2007) or 50 μM bephenium (n = 10). This result is in full agreement with the behavioral results indicating that the strain lacking L-AChR is resistant to bephenium. Thus, bephenium-elicited channel activity arises from L-AChR.
We also determined if L-AChRs lacking LEV-8, which may contain the ACR-8 subunit instead of LEV-8, are activatable by bephenium by performing single-channel recordings from the lev-8(x15) strain. Single L-AChR activity in the presence of 30 μM bephenium was readily detected (Fig. 2C). The mean open duration of bephenium-activated channels (0.36 ± 0.16 milliseconds, n = 3 patches) was not statistically significantly different from that recorded with ACh as an agonist (0.30 ± 0.07 milliseconds, n = 3, P = 0.558, Hernando et al., 2012). Moreover, examination of the recordings showed that bephenium-activated openings appear clustered in long activation episodes, separated by prolonged closed times (Fig. 2C). This pattern of channel activity is similar to that described for ACh (Fig. 2C) and was shown to arise from increased desensitization of the L-AChR lacking LEV-8 (Hernando et al., 2012). Thus, bephenium is also capable of inducing desensitization of the L-AChR.
Characterization of Bephenium as an Agonist and Open-Channel Blocker of L-AChR.
To characterize the action of bephenium as an L-AChR agonist, we recorded single-channel currents at a range of concentration and compared with those elicited by ACh. The minimum bephenium concentration that allowed detection of opening events was 0.5 μM, which is in the same order as that for ACh (0.1–0.5 μM) (Rayes et al., 2007; Hernando et al., 2012). At a wide range of bephenium concentration (0.5–250 μM), channel activity appeared mainly as isolated opening events (Fig. 3A). At 50 μM, open-time distributions showed a main open component of 0.21 ± 0.04 milliseconds (n = 6), which is similar to the 50 μM of ACh-elicited channels (0.25 ± 0.03 milliseconds, n = 3). As shown in Fig. 3A, the open duration decreases as a function of bephenium concentration, indicating additional open-channel block mediated by the drug. The reduction in open-channel lifetime is statistically significant at concentrations higher than 100 μM, as shown for ACh (Hernando et al., 2012).
Activation of L-AChR as a function of bephenium concentration. (A) L-AChR channels were recorded from cultured muscle cells of PD4251 worms at different bephenium concentrations and are shown at different temporal scales. Filter: 9 kHz. Holding potential: 100 mV. Representative open duration histograms are shown. Channel openings are shown as upward deflections. (B) Relationship between the reciprocal of the mean open duration (s−1) as a function of bephenium concentration. The slope of the curves fitted by linear regression estimates the value of the forward blocking constant. Each point corresponds to the mean ± S.D. of three different patches for each condition.
The decrease in open duration at high concentrations indicates that bephenium also acts as an open-channel blocker. To characterize this effect, we used the simplest linear scheme (Scheme 1),
(scheme 1)In scheme 1, C corresponds to closed, O to open, and OB to blocked states. β and α are the opening and closing rates, respectively, and k+b and k−b, are the forward blocking and unblocking rates, respectively. From linear regression analysis, we estimated the blocking constant from the slope of the relationship of the inverse of open-channel lifetime, which is an approximation for α, and bephenium concentration (B in scheme 1) (Fig. 3B). The estimated value, 2.1 × 107 s−1 M−1, is similar to those reported for other open-channel blockers (Neher and Steinbach, 1978; Rayes et al., 2001).
Closed time histograms were fitted by two or more components. The main closed component corresponds to closing between isolated openings. This duration was 170.77 ± 66.68 milliseconds (n = 3) at 50 μM bephenium, whereas it was 19.25 ± 8.53 milliseconds at 50 μM ACh (n = 3, and see Rayes et al. (2007)).
L-AChR Is the Target of Bephenium at L2 Developmental Stage.
We also evaluated whether bephenium recognizes a different target at a different developmental stage. To this end, we used another strategy that allows culture of L2 cells (Zhang et al., 2011). In the cultures, muscle cells are recognized by the typical morphology and have a greater size than those of L1 cultures (Zhang et al., 2011).
From these L2 cells, we detected for the first time the single-channel activity elicited by ACh and compared it with that elicited by bephenium. In general, the patterns of channel activity elicited by ACh and bephenium were similar to those recorded from L1 cells, indicating that L-AChR is the main target for bephenium also in the L2 stage. Channel openings of a homogenous amplitude class were detected for both agonists, whose mean amplitudes at 100 mV were 3.20 ± 0.12 pA (n = 4) and 3.36 ± 0.10 pA (n = 3) for ACh (100 μM) and bephenium (50 μM). The mean open times were similar between bephenium and ACh: 0.16 ± 0.01 milliseconds (n = 4) and 0.18 ± 0.05 milliseconds (n = 4) for 100 μM ACh and 50 μM bephenium, respectively (P = 0.63).
Bephenium Is a Weak Agonist and a Potent Channel Blocker of the Mammalian Muscle nAChR.
To determine if bephenium is selective for nematode nAChRs, we recorded single channels from cells expressing mammalian adult muscle nAChRs (Fig. 4). Bephenium elicited single nAChR currents at concentrations as low as 1 μM; under the same conditions channel openings were detected at 0.5 μM ACh. However, channel openings were markedly briefer than those activated by ACh. Open-time distributions of 10 μM bephenium-activated channels were well fitted by a main component of 0.11 ± 0.05 milliseconds (n = 3), whereas the duration of the main open component was 1.22 ± 0.32 milliseconds in the presence of 1–100 μM ACh (n = 6 cells). At ACh concentrations higher than 10 μM, mammalian muscle nAChR opens typically in clusters of well defined activation episodes (Bouzat et al., 2000; Bouzat and Sine, 2018) (Fig. 4A). Each activation episode begins with the transition of a single receptor from the desensitized to the activatable state and terminates by returning to the desensitized state. At 30 μM ACh, clusters of openings (mean cluster duration 144 ± 70 milliseconds) showed intracluster closings whose mean duration was 1.80 ± 0.29 milliseconds (n = 6). In contrast, when nAChRs were activated by bephenium, even at concentrations as high as 300 μM, clusters were not distinguished (Fig. 4B). A similar behavior has been reported for other nAChR weak agonists (Akk and Steinbach, 2003). These results demonstrate that bephenium opens mammalian nAChR channels with greater latency than ACh, and once opened, channels close faster than in the presence of ACh, thus indicating that it is a weak agonist of mammalian nAChRs. In addition, the decrease in open-channel duration as a function of bephenium concentration indicates that it also acts as an open-channel blocker of the mammalian muscle nAChR. The analysis with scheme 1 as a basis gave an estimate of k+b = 7.98 × 107 M−1s−1 for the blocking process, which is 4-fold higher than that at the L-AChR, indicating that it is a more potent channel blocker at the mammalian receptor.
Activation of mammalian nAChR by bephenium. Single-channel recordings from BOSC cells transfected with mammalian α1, β, ε, and δ subunit cDNAs. (A) Mammalian nAChR channels activated by 30 μM ACh or bephenium are shown at two different time scales. In the presence of ACh, clusters of channel activity can be clearly distinguished. A representative cluster is shown at the lower amplitude resolution. No clusters are distinguished in the presence of bephenium. At the right, representative open-time histograms are shown for each condition. Holding potential: 70 mV. Channels are shown as upward deflections. (B) Mammalian nAChR channels activated by bephenium at different concentrations. No clusters are distinguished at any concentration. At the right, representative open-time histograms are shown. Holding potential: 70 mV. Channels are shown as upward deflections. (C) Single channel currents in the presence of 30 μM ACh alone or together with 30 μM bephenium were recorded at a pipette potential of +70 mV (top) or −70 mV (bottom). Channel openings are shown as upward deflections at +70 mV and as downward deflections at −70 mV.
To further explore the action of bephenium at mammalian muscle nAChRs, we analyzed its effect on nAChR activation by ACh at positive and negative pipette potentials. At +70-mV holding potential, the presence of 30 μM bephenium produced important changes in the activity pattern elicited by 30 μM ACh (Fig. 4C). Openings were statistically significantly briefer: The duration of the main open component was reduced from ∼1 to 0.36 ± 0.01 milliseconds (n = 3). Clusters appeared remarkably different in the presence of bephenium (Fig. 4C). Whereas in the presence of 30 μM ACh the duration of the main closed component was between 1 and 2 milliseconds, an additional and novel major slower component of 17.9 ± 4.9 milliseconds was detected with the drug combination (n = 3). To determine if the effect of bephenium on ACh-channel activity was the result of binding site competition or to channel block, we analyzed the changes at a negative holding potential (−70 mV). The basis for this study is the fact that channel block could be different at the different membrane potentials because bephenium is positively charged, whereas competition for the extracellular agonist binding site is less voltage-dependent. At −70-mV pipette potential, nAChR activated by 30 μM ACh appeared in clear clusters, with a mean closed component of 2.60 ± 0.64 milliseconds and a mean open component of 0.14 ± 0.04 milliseconds (n = 3). This type of activity did not differ significantly from that recorded with the combination of ACh and 30 μM bephenium at the same potential (Fig. 4C). In these recordings, clusters were clearly distinguished, and the mean durations of the main open and closed components were 0.17 ± 0.02 and 2.49 ± 0.19 milliseconds, respectively. Increasing bephenium concentration to 100 μM in the presence of 30 μM ACh did not affect remarkably the mean duration of the main closed component at the negative pipette potential (3.0 ± 0.11 milliseconds) but significantly enhanced it at +70 mV (46 ± 11 milliseconds, n = 3) compared with ACh alone.
These results indicate that in the presence of ACh, bephenium acts mainly as a channel blocker at positive holding potentials. Moreover, the fact that it does not affect significantly the duration of closings of ACh-activated channels at negative holding potentials indicates that it cannot compete with ACh and it is therefore a low-affinity agonist.
To further confirm the weak activation, we recorded macroscopic currents elicited by bephenium (10–1000 μM) at 50- and −50-mV holding potentials. We found that at both potentials the maximal peak current for bephenium was lower than 2% of that elicited by a saturating ACh concentration (n = 6), and in many cases was indistinguishable from the background noise. Thus, bephenium activates mammalian nAChRs very weakly.
Bephenium Docks into the ACh Binding Site.
To gain insights into how bephenium may activate nAChR by binding to the orthosteric agonist binding site we performed molecular docking studies. We made homology models of the extracellular domain of two adjacent subunits to form functional agonist binding sites on the basis of two different crystal models that contain loop C of the principal face of the binding site in different conformations. Loop C has been shown to be essential for triggering channel activation and moves from an open to a closed conformation after agonist binding (Billen et al., 2012). We used the α4β2 structure crystalized with nicotine that contains loop C closed (PDB 5KXI) and the chimera α7/AChBP crystalized with the partial agonist lobeline that contains loop C partially closed (PDB 5AFM). The disposition of the five L-AChR subunits in the pentamer as well as the number of functional binding sites remain unknown. We therefore modeled 10 possible subunit interfaces of C. elegans L-AChR: UNC-63/UNC-29, UNC-38/UNC-29, LEV-8/LEV-1, LEV-8/UNC-63, LEV-8/UNC-38, LEV-8/UNC-29, UNC-63/UNC-38, UNC-38/UNC-63, UNC-63/LEV-1, and UNC-38/LEV-1 (the latter four only for the model with loop C partially closed), and the two muscle subunit interfaces that comprise the agonist binding sites (α1/δ and α1/ε). For ACh, the conformations considered as favorable were those showing the previously described cation-π interactions between the amino group of the ligand and aromatic residues of the binding pocket (W55, Y93, W149, Y190, Y198, numbering corresponds to α1 subunit) (Dougherty, 2007).
With the PDB 5KXI model, in all mammalian and nematode interfaces, ACh docking resulted in a main energetically favorable model with ACh oriented with its quaternary amine toward the membrane side or lower part of the cleft, just like carbamylcholine in AChBP (Celie et al., 2004). The positively charged group showed the potential to form the typical cation-π interaction with the indol group of W149 in loop B. There is also the potential for ACh to form additional H-bonds with different residues of the complementary face. In all tested interfaces, the frequency of active conformations was higher than 40% (from 40% to 100%) and the BBE of ACh was similar among all tested interfaces (−4.3 to −5.1 kcal/mol; Celie et al., 2004). Bephenium also docked into all interfaces, suggesting that it can activate nAChR through the orthosteric agonist binding site, but there were important differences with ACh docking. Although bephenium has a quaternary ammonium group, it was unable to make the typical cation-π interaction with W149 or with other aromatic groups at the cavity of all tested interfaces (Fig. 5, A and C). Another difference with ACh was a higher heterogeneity of conformations, suggesting that it can adopt different conformations at the binding site as reported for some partial agonists (Hibbs et al., 2009). Instead of cation-π, bephenium showed π-π interactions between one of its aromatic group and aromatic groups of residues Y93, W149, Y190, Y198, and W55 (Fig. 5, A and C). It also showed the potential to form H-bonds, and other types of bonds such as π-alkyl, with residues of the complementary face. Other interesting interactions were π-sulfur with the conserved cysteines C192 and C193. The BBE was more negative for bephenium than ACh at all interfaces, ranging from −6.7 in LEV-8/UNC-63 to −8.7 kcal/mol in α1/δ. The comparison between mammalian and worm interfaces did not show important differences in the orientations of bephenium at the binding site (Fig. 5, A and C).
Molecular docking of bephenium into homology-modeled binding-site interfaces. Homology models of the extracellular domain of two adjacent subunits were created using the structure of PDB 5KXI [(A) and (C), loop C closed] and PDB 5AFM [(B) and (D), loop C partially closed]. (A) and (B) correspond to the LEV-8/UNC-38 interface and (C) and (D) to α1/ε interface. Loops from the principal face are shown in green and from the complementary face in blue. Bephenium is shown in yellow. Cation-π interactions are shown in orange, π-π in pink, H-bonds in light blue, π-sulfur in red, and π-alkyl in purple.
Since partial agonists produce a partial contraction of loop C around the binding site (Billen et al., 2012), we explored homology models in this conformation (Fig. 5, B and D). Bephenium docked into the binding sites of all interfaces. All conformations showed similar BBE to those of 5KXI models (−6 to −8 kcal/mol). Importantly, in these models bephenium showed the potential to make the typical cation-π interactions with W149 and/or Y198 (Fig. 5, B and D). The frequency of conformations with cation-π interactions in all worm interfaces was lower than 10%, except for LEV-8/UNC-38 interface. Interestingly, in the LEV-8/UNC-38 interface, 80% of the conformations showed cation-π interactions, indicating that it may be an activatable site for bephenium. For the muscle nAChR, the typical cation-π interaction was observed in the α1/ε interface (about 18% frequency) but not in the α1/δ interface, suggesting that only one of the two sites could be activated by bephenium, which may explain the partial agonism.
Discussion
Parasitic nematodes negatively impact human and animal health. It has been estimated that as many as one-third of the world’s population harbors a helminth infection. Owing to the limited number of drugs available, there is an urgent need for concerted efforts on this area of research. Muscle nAChRs mediate body muscle contraction during locomotion in nematodes and are important targets of anthelmintic drugs (Jones et al., 2007). Despite the differences in subunit composition and pharmacology of muscle nAChRs among different parasites, as well as between parasites and C. elegans, C. elegans provides a means to decipher L-AChR subunit composition in parasites and has become an important model organism in the study of antiparasitic drugs (Brown et al., 2006; Neveu et al., 2010). Thus, the molecular characterization of C. elegans nAChRs is useful for highlighting similarities with parasite nAChRs and for providing information for recapitulating parasite nAChRs in C. elegans (Blanchard et al., 2018).
Whereas C. elegans nAChRs have been pharmacologically classified into nicotine- and levamisole-sensitive receptors, a wider spectrum of pharmacological classes has been reported for parasite nAChRs. Bephenium-sensitive receptors are included as an additional nAChR class. Nevertheless, the molecular actions of bephenium on C. elegans have not been explored to date.
The analysis of single L-AChR channel properties from C. elegans mutants has indicated that five different subunits (UNC-63, UNC-38, UNC-29, LEV-1, and LEV-8) assemble into a main functional form of L-AChR (Rayes et al., 2007; Hernando et al., 2012). The lack of the nonessential subunits LEV-1 or LEV-8 leads to functional L-AChRs with different channel properties (Hernando et al., 2012). Thus, in wild-type muscle the assembly of the five subunits appears to be constrained, but in the absence of nonessential subunits, different pentameric arrangements occur. In parasites, the assembly appears to be not so strict, since the orthologous subunits, such as UNC-29 and UNC-38, can combine into pentamers with different stoichiometries, thus resulting in a broader spectrum of pharmacological classes (Williamson et al., 2009). Sensitivity to bephenium has been attributed to Asu-UNC-63 subunit in Ascaris suum (Williamson et al., 2009) or to ACR-8 subunit in H. contortus (Boulin et al., 2011; Charvet et al., 2012).
We demonstrated here that UNC-38 null mutants are resistant to the bephenium-paralyzing effect and lack single-channel activity elicited by this drug, thus indicating that L-AChR is the target for bephenium. We also demonstrated that in the L2 stage single-channel activity of L-AChR elicited by ACh is similar to that in L1 and that L-AChR is also the target of bephenium at this developmental stage.
We showed that bephenium and levamisole have additive paralyzing effects and the combination of the two drugs has significantly higher efficacy, which is of importance since drug combinatorial therapies are convenient strategies for overcoming the increasing resistance of parasites.
L-AChR activation is elicited by low concentrations of bephenium. The channel activity pattern as well as the mean duration of openings at concentrations below the blocking ones are identical to those determined in the presence of ACh, in contrast to the marked differences between ACh and bephenium actions in the mammalian muscle nAChR. The only difference between ACh- and bephenium-elicited L-AChR channel activity is that the mean duration of the main closed component is systematically more prolonged with bephenium. Closed times from patches in which either the number of channels is not known or episodes of a single receptor molecule (clusters) cannot be distinguished do not provide relevant kinetic information (Bouzat and Mukhtasimova, 2018; Bouzat and Sine, 2018). However, the duration of the main closed component of recordings of ACh-activated L-AChR channels is relatively constant among patches and is dependent on agonist concentration (Rayes et al., 2007). Thus, we can speculate that the prolonged closed durations in the presence of bephenium with respect to equal concentrations of ACh may arise from a lower efficacy and/or slow channel block.
Because ACR-8 was proposed to be required for bephenium action in H. contortus (Charvet et al., 2012) and for levamisole sensitivity in parasite nAChRs (Blanchard et al., 2018), we explored its action in a C. elegans mutant strain lacking LEV-8, which may be replaced by ACR-8 in L-AChR (Hernando et al., 2012; Blanchard et al., 2018). We found that the mutant L-AChR is also activatable by bephenium, suggesting that both LEV-8 and ACR-8 can mediate bephenium responses in C. elegans. Moreover, bephenium is capable of inducing desensitization of the mutant L-AChR. This capability may act as a protection against sustained activation leading to paralysis, thus explaining the partial resistance of the strain lacking LEV-8 to bephenium. In parasitic nematode genomes, acr-8 orthologs are widely distributed, in contrast to lev-8 orthologs that are only present in a subset of parasitic species (Blanchard et al., 2018). Thus, C. elegans strains carrying parasite ACR-8 subunits could be useful models for exploring the kinetics of bephenium-activated channels (Blanchard et al., 2018).
Our studies showed that bephenium activates mammalian muscle nAChRs by acting as a weak agonist and a potent channel blocker, thus performing mainly as an inhibitor of ACh-activated channels. The main features, revealing its weak action at mammalian nAChRs, are the lack of clusters of openings over a range of concentrations, which differs from the activation by ACh or full agonists (Akk and Steinbach, 2003; Bouzat and Mukhtasimova, 2018; Bouzat and Sine, 2018), and 10-fold reduced open duration with respect to ACh-elicited channels. Moreover, if its activity had been tested only at the macroscopic level, its action as a partial agonist would not have been detected. Combined application of ACh and bephenium revealed that the main action of the antiparasitic drug is inhibition of ACh-activated channel activity by voltage-dependent channel block and that it is a low-affinity agonist since it does not affect ACh-activation at negative holding potentials (at which channel block is reduced).
Our molecular docking studies have provided insights into nAChR activation by bephenium binding to the orthosteric site. The ligand-binding cavities are located between subunit interfaces. The α subunits provide the principal face of each agonist site. The face is formed by loops A–C, which contribute to a nest of aromatic residues (Y190, W149, Y198, Y93) that stabilizes the ammonium of the agonist through cation-π interactions and/or hydrogen bonding (Xiu et al., 2009). Loop C closes to cap the agonist, an event associated with the triggering of channel opening (Mukhtasimova et al., 2009; Unwin and Fujiyoshi, 2012). Loop C adopts an uncapped conformation in the absence of agonist or in the presence of antagonists (Miller and Smart, 2010; Billen et al., 2012). X-ray structures of AChBP in complex with partial agonists showed an incomplete contraction of loop C (Hibbs et al., 2009; Billen et al., 2012; Spurny et al., 2015). We therefore generated binding site interfaces on the basis of two structural models, one with loop C closed and one with a partially closed conformation (as with partial agonists). In both models, bephenium docked into the binding site of all interfaces, indicating that it can act as an orthosteric agonist. For the model with loop C closed the main features of bephenium binding were: 1) the lack of cation-π interaction with W149 in loop B that has been reported for most nAChR agonists (Billen et al., 2012; Post et al., 2017), 2) the potential to make π-π interactions between one of its aromatic groups and the aromatic groups of key tyrosine and tryptophan residues, and 3) the potential of novel π-sulfur interactions with the conserved cysteines at loop C of the principal face. The results suggest new binding modes and explain the low efficacy resulting from the lack of cation-π interactions. Alternatively, bephenium may cause an incomplete contraction of loop C, which was evaluated in the second model. In these models, bephenium showed the potential to form the typical cation-π interactions with W149 and/or Y198. The requirement of a partially closed loop C to allow these key binding site interactions may explain the partial agonism. Although almost all worm interfaces also showed the capability of forming cation-π interactions, the frequency was remarkably higher in the LEV-8/UNC-38 interface. This higher frequency and the possibility of multiple binding site interfaces may explain the higher efficacy of bephenium activation of the L-AChR compared with mammalian nAChRs. Unfortunately, the disposition of the subunits in the pentameric arrangement of L-AChR subunits remains unknown and, therefore, we cannot be assured that this interface occurs in the native receptor. Remarkably, the α1/δ interface did not show cation-π interactions. Thus, the very weak action of bephenium for the mammalian nAChR may be explained by receptor activation from a single orthosteric binding site (α1/ε interface).
Our study provides novel insights into the molecular bases of anthelmintic drug action, which paves the way for the development of novel drugs.
Authorship Contributions
Participated in research design: Turani, Hernando, Corradi, Bouzat.
Conducted experiments: Turani, Hernando, Corradi.
Performed data analysis: Turani, Hernando, Corradi, Bouzat.
Wrote or contributed to the writing of the manuscript: Turani, Corradi, Bouzat.
Footnotes
- Received June 12, 2018.
- Accepted August 29, 2018.
This work was supported by grants from Universidad Nacional del Sur (UNS); Agencia Nacional de Promoción Científica y Tecnológica (ANPCYT); Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) Argentina; and Bill and Melinda Gates Foundation (OPP1098404) to C.B.
Abbreviations
- ACh
- acetylcholine
- BBE
- best binding energy
- GFP
- green fluorescence protein
- L-AChR
- levamisole-sensitive nAChR
- nAChR
- nicotinic acetylcholine receptor
- τcrit
- critical duration
- Copyright © 2018 by The American Society for Pharmacology and Experimental Therapeutics
