Abstract
Paraherquamide is a novel natural anthelmintic product with a mode of action that is incompletely characterized. Nicotine and cholinergic-anthelmintic agonists of different chemical classes were used to produce contraction in Ascaris muscle strips. Paraherquamide and a semisynthetic derivative, 2-deoxy-paraherquamide, antagonized these responses. Analysis of the actions of the antagonists was made using the simple competitive model and nonlinear regression to estimate the pKB values of the antagonists. The analysis was tested using Clark plots. The pKB values for paraherquamide were: nicotine, 5.86 ± 0.14; levamisole, 6.61 ± 0.19; pyrantel, 6.50 ± 0.11; and bephenium, 6.75 ± 0.15. The pKB of nicotine was significantly different from the pKB values for levamisole, pyrantel, and bephenium, showing that paraherquamide can distinguish a subtype of cholinergic receptors sensitive to nicotine and a subtype of cholinergic receptors sensitive to levamisole, pyrantel, and bephenium. The pKB values for 2-deoxy-paraherquamide were: levamisole, 5.31 ± 0.13; pyrantel, 5.63 ± 0.10; and bephenium, 6.07 ± 0.13. The Clark plots of the antagonism illustrated the degree of fit to the competitive model for 2-deoxy-paraherquamide. 2-Deoxy-paraherquamide selectively antagonized the effects of bephenium; the pKB values of levamisole and pyrantel were significantly different from the pKB of bephenium. Paraherquamide and 2-deoxy-paraherquamide are selective competitive cholinergic antagonists that distinguish subtypes of cholinergic receptor inAscaris muscle corresponding to nicotine-, levamisole-, and bephenium-sensitive receptors.
Nematode parasite infections of humans and animals cause disease with loss of productivity, debility, and occasionally death. Ascariasis and hookworm infections are carried by 1.6 billion people throughout the world and in 2% of cases cause loss of life. The use of therapeutic compounds forms a major component of control, and the development of novel therapeutic agents is required to deal with the increasing levels of resistance to existing drugs.
Paraherquamide (Fig. 1) is a novel anthelmintic (Yamazaki et al., 1981) that is an alkaloid fermentation product originally isolated from Penicillium paraherquii. The anthelmintic property of paraherquamide was first identified using jirds infected with Trichostrongylus colubriformis (Ostlind et al., 1990). Paraherquamide produces paralysis of parasitic nematodes in culture, without an effect on ATP, suggesting that it does not act as a metabolic poison (Thompson et al., 1996). Interestingly, one of the toxic effects of paraherquamide in the dog (Shoop et al., 1990) is a prolapsed nictitating membrane, an effect that suggests antagonism of neuronal nicotinic receptors (nAChRs). Recently, it has been reported (E. W. Zinser, M. L. Wolfe, S. J. Bowman, E. M. Thomas, V. E. Groppi, J. P. Davis, D. P. Thompson, and T. G. Geary, manuscript submitted for publication) that paraherquamide and 2-deoxy-paraherquamide act as antagonists of nematode nAChRs and mammalian neuronal nAChRs.
Structure of anthelmintics paraherquamide, bephenium, levamisole and pyrantel. In paraherquamide, note the presence of the groups connecting N* in position 19 and O* in position 2 with a similarities to acetylcholine. The O* is missing in 2-deoxy-paraherquamide.
Nematode nAChRs have some pharmacological similarities to vertebrate neuronal receptors, but there are important differences. This is fortunate because it allows drugs like levamisole and pyrantel to be used for therapeutic purposes. Levamisole and pyrantel are potent anthelmintics that act as selective agonists on nematode nAChRs (Martin et al., 1996). Nematode nAChRs, like vertebrate neuronal nAChRs, are taken to be composed of a pentamer of subunits, and a number of different subtypes are present on muscle (Robertson et al., 1999; Richmond and Jorgensen, 1999). The distinctive pharmacology of some nematode nAChRs seems to relate to the distinctive molecular structure of the agonist binding site formed by part of the α-subunit (Fleming et al., 1997).
In this article, we report the quantitative investigation of the cholinergic antagonistic effects of paraherquamide and 2-deoxy-paraherquamide, using the simple competitive antagonist model on nematode nAChRs in Ascaris body wall muscle. We report that paraherquamide is a potent antagonist in nematodes that distinguishes levamisole- and nicotine-sensitive nAChRs of nematodes and that 2-deoxy-paraherquamide distinguishes levamisole- and bephenium-sensitive receptors. These findings are important because they demonstrate multiple subtypes of acetylcholine receptor with different anthelmintic sensitivities in a parasitic nematode and they further define the mode of action for these anthelmintics.
Materials and Methods
Ascaris Preparation.
Adult Ascaris suum were collected weekly from the IPB packing plant (Storm Lake, IA) and returned to the laboratory in Locke's solution at around 35°C in a metal vacuum flask. TheAscaris were used for the contraction studies within 72 h because the ability to contract vigorously to cholinergic agonists declined after this period.
Two 2-cm body-flap preparations, one dorsal and one ventral, were made from each Ascaris female from the region anterior to the genital pore. The gut was teased away from the muscle strip and the lateral lines were removed from the edge of the flaps to produce two preparations (one dorsal and one ventral). Each flap, composed of the muscle field and the cut nerve cord, was monitored isometrically by attaching a force transducer in an experimental bath, maintained at 37°C, containing 10 ml of Ascaris Ringer (23 mM NaCl, 110 mM Na-acetate, 24 mM KCl, 6 mM CaCl2, 5 mM MgCl2, 11 mM glucose, 5 mM HEPES, pH 7.6, with NaOH), and bubbled with nitrogen. Eight baths were used simultaneously. After dissection, the preparations were allowed to equilibrate for 15 min under an initial tension of 2.0 g. The antagonist was then added to the preparation 15 min before the application of the first concentration of the agonist. We ran two controls (no antagonist) and three replicate antagonist concentrations during each experiment (eight preparations). The dorsal and ventral flaps were assigned randomly in the experiments to reduce any potential error associated with differences in receptor populations. The agonists were added cumulatively with 2- to 3-min intervals between applications, and the responses were steady changes in tension. Figure2B is a representative trace of a control agonist concentration-response plot with bephenium. Results for individual flaps were rejected if the maximum change in tension, at the highest agonist concentration, did not exceed 0.5 g. The mean maximum tension of the preparations, produced at each concentration of antagonist, was not significantly different (p > 0.05, F-test). The responses for each concentration were expressed as the percentage of the maximum tension produced by each individual flap preparation. The addition of small volumes of agonist did not significantly reduce the antagonist concentration in the tissue bath.
A, A. suum in hand for scale. B, representative Ascaris muscle contraction cumulative-concentration-response trace. The trace shows the effect of cumulative addition of bephenium to the muscle strip.
Drugs.
Paraherquamide and 2-deoxy-paraherquamide were obtained from Pharmacia Animal Health (Kalamazoo, MI). Nicotine hemi-sulfate, levamisole hydrochloride, pyrantel citrate, bephenium hydroxynaphthoate, and dihydro-β-erythroidine were purchased from Sigma-Aldrich (St. Louis, MO). Pyrantel tartrate, bephenium hydroxynaphthoate, paraherquamide, and 2-deoxy-paraherquamide were dissolved in DMSO and added to the bath. Concentrations of DMSO did not exceed 0.1% in the bath. Control experiments demonstrated that this concentration of DMSO had no effect on concentration-response relationships (data not shown).
Recording and Analysis.
Changes in isometric muscle tension responses were monitored using a PowerLab System (ADInstruments Pty Ltd., Castle Hill, Australia) that consists of the PowerLab hardware unit and Chart for Windows software (Microsoft, Redmond, WA). The system allows for recording, displaying, and analysis of experimental data. Sigmoid concentration-response curves for each individual flap preparation at each concentration of antagonist were described by the Hill equation.
The slope factor, equivalent to the slope of the Schild plot, was estimated using nonlinear regression and the following equation.
To display the effect of the antagonists paraherquamide and 2-deoxy-paraherquamide on the agonist EC50values, Clark plots of log EC50 versus log ([XB] + pKB) were used. This display allows the distribution of the data to be compared with the ideal for simple competitive block.
Results
Dihydro-β-Erythroidine.
Richmond and Jorgensen (1999) have described the presence of nicotine- and levamisole-sensitive cholinergic receptors on muscle of Caenorhabditis elegansand have used single concentrations of the antagonist dihydro-β-erythroidine to block electrophysiological responses to nicotine but not levamisole. We tested, in Ascaris, the effects of different concentrations of dihydro-β-erythroidine on the contraction-responses to nicotine and levamisole.
Figure 3 shows the effect of different concentrations (10–100 μM) of dihydro-β-erythroidine on the concentration-response relationships of levamisole and nicotine. Dihydro-β-erythroidine had no detectable effect on the levamisole response (N = 7, 5, 5, and 5 for the 0, 10, 30, and 100 μM concentrations of dihydro-β-erythroidine) but produced a modest parallel shift to the right, in the concentration-response curves for nicotine (N = 8 for each concentration of dihydro-β-erythroidine). The dose-ratio produced by 100 μM dihydro-β-erythroidine was 2.16. If the antagonism is taken to be competitive, we can make an estimate for the antagonist pKB using eq. 4.
Dihydro-β-erythroidine (DHβE) as an antagonist of nicotine (A) and levamisole (B). Note that dihydro-β-erythroidine produces some dose-dependent antagonism of the effects of nicotine but has no effect on the levamisole response (n ≥ 5 for control and each antagonist concentration).
Paraherquamide.
We found that paraherquamide was a more potent antagonist than dihydro-β-erythroidine in Ascaris, and we therefore examined the antagonism in greater detail. We selected agonists that were cholinergic anthelmintics belonging to different chemical classes for further examination. We chose (Fig. 1) the tobacco alkaloid nicotine, the imidazothiazole levamisole, the tetrahydropyrimidine pyrantel, and the quaternary nitrogen compound bephenium.
Figure 4 shows the antagonistic effects of paraherquamide (0.03–30 μM) on concentration-response plots for nicotine (N = 8, 5, 5, and 5 for the 0, 0.3, 3, and 30 μM paraherquamide), levamisole (N = 8, 5, 6, and 5 for 0, 0.3, 3, and 30 μM paraherquamide), pyrantel (N= 6, 5, 6, and 5 for 0, 0.3, 3, and 30 μM paraherquamide) and bephenium (N = 7, 6, 8, 7, and 3 for 0, 0.3, 3, 10, and 30 μM paraherquamide). The antagonism produced clear shifts to the right in all the concentration-response curves. As a test for the parallel nature of the shifts, we fitted all the concentration-response plots of single flap preparations (n = 5–8 at each antagonist concentration) with eq. 1, to estimatenH and EC50. We then tested the effect of antagonist concentration onnH and found no significant effect (p > 0.05, F-tests) for nicotine, levamisole, pyrantel, or bephenium and concluded that the shifts could be taken as parallel.
Concentration-response plots for the agonists nicotine (A), levamisole (B), pyrantel (C), and bephenium (D) in the presence of different concentrations of paraherquamide (n ≥ 5 for control and each paraherquamide concentration).
Since the concentration-response curves were shifted to the right in a parallel manner, we used the simple competitive antagonist model (eq.2) to describe the data. We also estimated the parameters of eq. 3using nonlinear regression to make an estimate of the Schild slope factor, N.
Table 1 shows the values that were estimated for nicotine, levamisole, and pyrantel. The Schild slopes for these agonists were not significantly different from 1. The slope for bephenium was significantly less than 1, suggesting deviation from the simple competitive antagonist model. For reasons we discuss later, we described the antagonism of all the agonists with the simple competitive model using eq. 2.
The mean ± S.E. values for the parameters of eq. 2 are shown in Table 1 along with their confidence limits. Interestingly, the goodness of fit, R2, showed that the fit for bephenium was comparable to nicotine and levamisole. As a further test of the fit of the simple competitive antagonist model, we plotted log EC50 versus log ([B] +KB) as the Clark plot. EC50 values were obtained from eq. 1,B is the concentration of paraherquamide, andKB is the antagonist dissociation constant (the antilog of −pKB). Figure5 shows these plots for nicotine, levamisole, pyrantel, and bephenium. These plots show the relationship between the experimental data and the line predicted by the simple competitive antagonist model.
Clark plots for the agonists nicotine (A), levamisole (B), pyrantel (C), and bephenium (D). Ordinate: mean ± S.E. log EC50. Abscissa: log (paraherquamide concentration + agonist dissociation constant). Note the degree of fit between the observed values and the line predicted by the simple competitive block model.
The important point that we can make here is that the pKB of paraherquamide for nicotine is 5.86 ± 0.14, and this is significantly less (p < 0.001, t test, degrees of freedom 43–52) than for levamisole (6.61 ± 0.19), pyrantel (6.50 ± 0.11), or bephenium (6.75 ± 0.15). The observations demonstrate that the receptors activated by nicotine are not the same as those activated by levamisole, pyrantel, and bephenium.
2-Deoxy-Paraherquamide.
We also tested the paraherquamide analog 2-deoxy-paraherquamide (see Fig. 1). We found that like paraherquamide it was an antagonist of nicotine, levamisole, pyrantel, and bephenium. Figure 5 shows the effect of 2-deoxy-paraherquamide on the log-concentration response plots. Concentration-response plots for levamisole (N = 6, 6, 6, and 6 for 0, 0.3, 3, and 30 μM antagonist), pyrantel (N = 6, 8, 7, and 8 for 0, 0.3, 3, and 30 μM antagonist), and bephenium (N = 7, 6, 6, 6, and 6 for 0, 0.03, 0.3, 3, and 30 μM antagonist), but not nicotine (N = 9, 10, 9, and 30 for 0, 0.3, 3, and 30 μM antagonist) were shifted to the right in a parallel manner by 2-deoxy-paraherquamide. Again, as a test for the parallel nature of the shift, we fitted all the concentration-response plots of single preparations with eq. 2. We then tested the effect of 2-deoxy-paraherquamide concentrations onnH; we found no significant effect (p > 0.05, F-tests) for levamisole, pyrantel, or bephenium. The effect of 2-deoxy-paraherquamide onnH of the nicotine slopes was significant (p < 0.0004, F = 10.2, degrees of freedom 3,18), an observation inconsistent with a simple competitive antagonism model.
We estimated N using eq. 3. Table2 shows the mean values and 95% confidence limits for N. The slope, N, for nicotine was clearly less than 1, with a mean value of 0.39 (95% confidence limits 0.10–0.69), again emphasizing that a simple competitive action was not adhered to. As with paraherquamide, we again fitted the simple competitive model of antagonism to describe the action of the antagonist and to estimate the pKB values with eq. 2. Table 2 shows the mean values and error estimates for the pKB values for levamisole, pyrantel, and bephenium. We do not include the error estimates for nicotine because the simple competitive antagonism model did not describe the experimental data well, and R2 was 0.50. To illustrate the fit of the experimental data to the simple competitive antagonist model and the estimated dissociation constant, we again used the Clark plots (Fig.6). The plots suggest that the model is suitable and describes the data for levamisole, pyrantel, and bephenium.
Concentration-response plots for the agonists nicotine (A), levamisole (B), pyrantel (C), and bephenium (D) in the presence of different concentrations of 2-deoxy-paraherquamide [0 (control), 0.3, 3, and 30 μM] (n ≥ 6 for each agonist/antagonist combination).
Clark plots for the agonists levamisole (A), pyrantel (B), and bephenium (C). Ordinate: mean ± S.E. log EC50. Abscissa: log (2-deoxy-paraherquamide concentration + agonist dissociation constant). Note the relationship between the observed values and the line predicted by the simple competitive model of competitive antagonism.
The pKB values estimated for 2-deoxy-paraherquamide are less than the pKB values of paraherquamide for the same agonist showing that paraherquamide is more potent than 2-deoxy-paraherquamide as an antagonist in A. suum. Also, the pKB of 2-deoxy-paraherquamide was significantly greater with bephenium as agonist than with levamisole and pyrantel as agonist. This implies that bephenium does not act on the same population of nAChRs as levamisole and pyrantel.
Discussion
Use of the Simple Competitive Antagonism Model.
We used classical pharmacological techniques to analyze the antagonist action of paraherquamide and 2-deoxy-paraherquamide. The simple model of competitive action is based on the law of mass action, using the reaction scheme below.
This simple model may not be universally valid for ion-channels that have two or more binding sites for agonist and antagonist molecules. Under these conditions, the slope of the Schild plot is predicted to be reduced under the law of mass action and under limiting conditions is described by the modified Schild equation (Williams et al., 1988), which has a slope less than 1. For the paraherquamide antagonism of bephenium and the 2-deoxy-paraherquamide antagonism of nicotine, we saw that the Schild slope factor was less than 1. One explanation then for the reduced slope factor may lie along the lines prescribed by Williams et al. (1988).
When a Monod-Wyman-Changeux allosteric model for competitive antagonism is used for the operation of the receptor ion-channel, the dose-ratio turns out to be a linear function of the antagonist concentration, even in the general case (Colquhoun, 1973). So the explanation for when the slope factors are less than one may lie elsewhere and relate to the presence of more than one nAChR receptor subtype. If there are multiple subtypes of nAChR present in the preparation and the agonist and antagonists are not sufficiently selective, it is possible, at low agonist concentrations, that the KB value of the antagonist for receptors with high agonist affinity dominates the shift in the concentration-response curve. At high concentrations of antagonist, the KB of the antagonist for low-agonist-affinity receptors may dominate the shift in the concentration-response curve. This could reduce the Schild slope factor and/or give rise to nonparallel shifts in concentration-response plots, as we saw with the antagonism of nicotine by 2-deoxy-paraherquamide.
By using eq. 2, we have imposed a slope factor of 1 for all the agonists to estimate the pKB for reasons of parsimony and because the approach also reduces over estimates of pKB (Black and Shankley, 1985). Thus, despite some known uncertainties, we feel justified in using the competitive model. Waud and Parker (1971) have commented that “I do not blush when caught using a competitive kinetic analysis heuristically when the underlying model may in fact be more along the allosteric lines… if you tell me theKB of a new antagonist, I know a lot about the drug.” The conclusions we draw in this article are based on the assumption that agonists acting on identical receptors will be affected to the same degree by the antagonist. Classically, agonists that act on the same receptors are expected to produce the same pKB (Arunlakshana and Schild, 1959).
Competitive Action of Paraherquamide and 2-Deoxy-Paraherquamide and nAChR Subtypes.
In this study, we have been able to show that paraherquamide and 2-deoxy-paraherquamide can behave as competitive antagonists at nematode nAChRs. We know that paraherquamide is a potent anthelmintic (Shoop et al., 1990). The potency of the competitive antagonist action seen here, with dissociation constants in the micromolar range, suggests that competitive nAChR antagonism contributes to the anthelmintic properties of this compound.
2-Deoxy-paraherquamide lacks a carbonyl group at the 2-position (Fig. 1) and was around 10× less potent as an antagonist in our assays. In whole-worm preparations with a cuticle barrier to penetrate this order of potency is reversed (Thompson et al., 1996), perhaps for reasons of access. Our observations here suggest that the presence of a carbonyl group, which presumably acts in hydrogen bonding to the receptor, is important for binding. The classic model of the cholinergic receptor consists of a site which binds the quaternary nitrogen, N+, and an osteophilic site that binds to =O of acetylcholine; the separation between the N+ and =O is a chain length of four atoms. We can see a similar structure in paraherquamide, starting with the carbonyl group in position 2 and ending in the methylated nitrogen in position 19 (Fig. 1). This portion of paraherquamide might bind to the nAChR. We point out that the amino acid structure of the ligand-binding has now been resolved at the atomic level, and the quaternary nitrogen of ligands is thought to stack onto tryptophan (position 143) making cation-π interactions (Brejc et al., 2001).
Interestingly, our analysis revealed that paraherquamide and 2-deoxy-paraherquamide showed pKBvalues that varied significantly when different agonists were used. We saw that paraherquamide discriminated between receptors sensitive to nicotine and receptors sensitive to levamisole, pyrantel, and bephenium. This observation is consistent with the observations ofRichmond and Jorgensen (1999), who described the presence of nicotine- and levamisole-sensitive receptors on C. elegansmuscle. We found that 2 deoxy-paraherquamide had a pKB for bephenium that was significantly different to that of levamisole and pyrantel. Again the evidence suggests that the receptors acted on by levamisole and pyrantel are not the same as those of bephenium. It is interesting to point out that a comparison of levamisole-sensitive and levamisole-resistant isolates of Haemonchus contortus showed that muscle contraction became less sensitive to levamisole with resistance but did not change its sensitivity to bephenium (Sangster et al., 1991). Both these lines of evidence support the view that bephenium-sensitive receptors are different to levamisole-sensitive receptors.
Multiple Subtypes of nAChRs in Mammals and Nematodes and Selective Ligands.
Ionotropic acetylcholine receptors are taken to be pentameric structures (Corringer et al., 2000; Sharples and Wonnacott, 2001). In vertebrate muscle, the subunits α, β, δ, and ε form the ion-channels with a stoichiometry of α2, β, δ, ε at the endplate. In vertebrate neuronal receptors, the channels are usually formed by a heteropentameric combination of α and β subunits, α2 to 7, α9, α10, and β2 to 4 in mammals, with an additional subunit, α8, in avian species. Homomeric combinations of some α subunits may occur, e.g., by α7, α8, or α9, but most channels are composed of α and β subunits. The rules governing the ability to combine to form functional ion-channels vary with the structure of the subunit.
The ligand binding sites (Corringer et al., 2000: Brejc et al., 2001) of individual nAChR receptors are composed of six amino acid loops, three (loops A, B, and C) from the α subunit and three (D, E, and F) from the adjacent β subunit. Since nAChRs have two or more α subunits, there are two or more ligand binding sites. Different nAChRs formed by different α and β subunit combinations have different agonist binding sites because the loops, ABC and DEF, will vary with the particular α and β subunits that form the ion-channel receptor. It is also a consequence that nAChRs composed of two identical α subunits and three dissimilar β subunits will have two nonequivalent binding sites. Consequently, selective agonists and antagonists are expected to distinguish between receptor subtypes and between nonequivalent sites on an individual nAChR.
A distinct but related family of nAChR subunits has been found inC. elegans in which there are 42 candidate nAChR subunits (Fleming et al., 1996, 1997; Baylis et al., 1997; Bargmann, 1998). The α subunits include: UNC-39, UNC-63, ACR-16 or Ce21, and ACR-3. The β subunits include: ACR-1, ACR-2, LEV-1, and UNC-29. An nAChR, sensitive to levamisole, may be composed of UNC-38 as an α-subunit along with UNC-29 and LEV-1 on muscle. Expression of ACR-2 + UNC-38, UNC-38 + ACR-3, and UNC-38 + UNC-29 + LEV-1 inXenopus oocytes produces small currents in response to levamisole. In contrast, expression of ACR-16, which is 47% identical to chicken α8, in Xenopus oocytes, produces channels sensitive to acetylcholine but insensitive to levamisole or pyrantel (Ballivet et al., 1996). In the intact nematode, levamisole pyrantel, morantel, and oxantel activate nAChRs on Ascaris muscle (Martin et al., 1996), and single channel recordings of nAChRs inOesophogostomum dentatum reveal the presence of four subtypes on somatic muscle (Robertson et al., 1999). There is therefore evidence for the presence of multiple subtypes of nAChR in nematodes with ligands that can be selective for different subtypes of receptor.
Significance of nAChR Receptor Subtypes for Anthelmintic Resistance.
There is strong evidence for nAChR receptor subtypes in vertebrates and, now, nematodes and that the different subtypes have different sensitivities to ligands. Also, we have presented evidence that nicotine, levamisole, and bephenium are selective ligands for different subtypes. Sangster et al. (1991) have described isolates of levamisole resistant H. contortus that have a reduced sensitivity to levamisole but not to bephenium. We suggest that one form of levamisole resistance is associated with a switch from levamisole-sensitive nAChRs to other nAChR subtypes. This mode of anthelmintic resistance would not be revealed as a change in the amino acid sequence of particular nAChR subunits (Hoekstra et al., 1997) but as a change in proportion of expressed nAChR subtypes.
The effect of paraherquamide and 2-deoxy-paraherquamide on whole nematodes seems to be that of paralysis by competitive antagonist action at different nAChR subtypes. If the potency of these antagonists is sufficient against all of the nAChR subtypes present in nematode parasites, these compounds could be useful for the control of types of levamisole resistant nematodes associated with changes in proportion of nAChR subtypes.
Footnotes
-
We are pleased to acknowledge the support of National Institutes of Health Grant RO1 A147194-02 awarded to R.J.M.
-
DOI: 10.1124/jpet.102.034272
- Abbreviations:
- nAChRs
- neuronal acetylcholine receptors
- DMSO
- dimethyl sulfoxide
- Received February 5, 2002.
- Accepted April 8, 2002.
- The American Society for Pharmacology and Experimental Therapeutics
