Molecular evidence for dual pyrethroid-receptor sites on a mosquito sodium channel

Expression and Functional Characterization of a Mosquito Sodium Channel in Xenopus Oocytes.

A full-length sodium channel cDNA clone, AaNa_v1–1, was isolated from adults of Ae. aegypti. An initial attempt to express AaNa_v1–1 in Xenopus oocytes, however, produced only small sodium currents. It is known that robust expression of the Drosophila melanogaster sodium channel requires an auxiliary subunit protein temperature-induced paralytic E (TipE) (24, 25). We cloned a TipE-like cDNA, AaTipE, from Ae. aegypti and used it for coexpression with the AaNav1-1 cRNA (1:1 ratio) in Xenopus oocytes. Indeed, AaTipE significantly increased the amplitude of AaNa_v1–1 sodium currents (Fig. S1B). The voltage dependences of activation and steady-state inactivation (Fig. S1C) are similar to those of sodium channels from other insects (25⇓⇓–28).

One major feature of pyrethroid action on sodium channels is the induction of a slowly decaying tail current upon termination of depolarizing pulse (i.e., deactivation following repolarization of the membrane) in voltage-clamp experiments (29). Because pyrethroids preferably interact with the open state of insect sodium channels (19), to elicit large pyrethroid-induced tail currents, we applied a 100-pulse train of 5-ms depolarization from −120 to 0 mV, with a 5-ms interval. Pyrethroids are classified either as type I or type II based on their chemical structures, poisoning symptoms, and effects on the nervous system (29, 30). Compared with type I pyrethroids, such as permethrin (PMT), type II pyrethroids including deltamethrin (DMT) generally modify sodium channel gating more dramatically, causing a much more slowly decaying tail current during deactivation upon repolarization (19). This was also the case for the AaNa_v1–1 channel (Fig. S1 D and E). The resulting pyrethroid-induced tail currents serve as a measure of channel sensitivity to pyrethroids (31).

Five Mutations Confer the AaNa_v1–1 Channel Resistance to Pyrethroid Insecticides.

The establishment of a functional expression system for the AaNa_v1–1 channel enabled us to examine whether any of the seven pyrethroid-resistance–associated mutations found in Ae. aegypti (Fig. 1A) affect pyrethroid sensitivity of a native sodium channel. In addition, we also examined two mutations, leucine (L) to phenylalanine (F) or serine (S) mutation at amino acid position 1021 (L1021F/S) in domain II segment 6 (IIS6) because they were found in pyrethroid-resistant Anopheles and Culex mosquitoes (12). The L1021F mutation corresponds to L1014F in the house fly sodium channel and L993F in the cockroach sodium channel, both of which have previously been shown to reduce the sensitivity of respective sodium channels to pyrethroids (12, 14), whereas the L1021S mutation reduces the sensitivity of a Drosophila sodium channel to pyrethroids (32). Because S996P and D1794Y mutations are often found to coexist with V1023G in pyrethroid-resistant field populations of Ae. Aegypti (references in ref. 12), we also examined the effects of two double mutations. Accordingly, a total of nine single mutant and two double mutant AaNa_v1–1 channels were constructed. All mutants generated sufficient sodium currents in Xenopus oocytes for further functional analysis. None of mutations altered the voltage dependences of activation and inactivation (Table S1).

To evaluate channel sensitivity to PMT and DMT, the percentage of sodium channels modified by pyrethroids was determined by quantifying the amplitude of pyrethroid-induced tail current (Fig. S1 D and E) (31). As shown in Fig. 1 B and C, of the six mutations located in IIS6, V1023G, L1021F, and L1021S mutations reduced AaNa_v1–1 channel sensitivity to both pyrethroids, whereas I1018M reduced AaNa_v1–1 channel sensitivity to PMT, but not to DTM. In contrast, I1018V and V1023I did not alter channel sensitivity to either pyrethroid (Fig. 1 B and C). Of the remaining three mutations, F1565C in IIIS6 reduced the sensitivity of the AaNa_v1–1 channel to PMT, but not to DMT (Fig. 1 B and C). This is in agreement with our earlier findings that this mutation reduces the sensitivity of cockroach sodium channels to type I, but not type II, pyrethroids (33). In addition, neither V1023G/S996P nor V1023G/D1794Y was more resistant to pyrethroids than the single mutants (Fig. 1 B and C). Altogether, our functional analysis using the AaNa_v1–1 channel provides direct evidence for the involvement of three Ae. aegypti mutations (I1018M, V1023G in IIS6, and F1565C in IIIS6) and also two Anopheles/Culex mutations (L1021F/S in IIS6) in AaNa_v1-1 channel insensitivity to pyrethroids. Furthermore, it is worth mentioning that although S996P and D1794Y mutations were concurrent with V1023G in pyrethroid-resistant field populations of Ae. Aegypti, neither of them imposes additive or synergistic effects on the V1023G-mediated pyrethroid resistance.

Molecular Mapping of a Second Pyrethroid-Receptor Site on the AaNa_v1–1 Channel.

Two Ae. aegypti kdr mutations confirmed in this study, V1023G in IIS6 and F1565C in IIIS6, are located in the previously identified receptor site (site 1) (Fig. S2). Surprisingly, the other three kdr mutations, I1018M, L1021F, and L1021S, are not close to site 1 (Fig. S2). Importantly, kdr mutations at L1021 (corresponding to L1014 and L993 in the house fly and cockroach sodium channels, respectively) are widespread in diverse pyrethroid-resistant populations of many pest species (14, 20), and the L993F mutation in the cockroach sodium channel has been shown to reduce pyrethroid binding using Schild analysis (18). These findings raise the possibility that site 1 may not be the only pyrethroid-receptor site on insect sodium channels, as was previously proposed by Vais et al. based on Hill plot analysis (19). Their work suggested that the M918T mutation in IIL45 (within site 1) in the Drosophila sodium channel reduces the number of apparent pyrethroid-binding sites per channel from two to one (19). However, the molecular identity of this putative second site remained elusive.

To explore the molecular identity of a second pyrethroid-receptor site on the AaNa_v1–1 channel, we have built a homology model of the inner-pore region of the AaNa_v1–1 channel using the X-ray structure of the potassium channel K_v1.2 (34). From here on, we replaced the sequence-based amino acid numbering with a nomenclature that is universal for sodium channels and other P-loop ion channels (35). This nomenclature, which is described in Fig. 2 legend, permits recognition of symmetric positions of amino acid residues in the four homologous domains within the same channel and also comparison with sodium channels from other species (Fig. 2A). When necessary, we also show the sequence-based residue number in parentheses. According to our model, the three mosquito kdr mutations, I^2i13(1018)M and L^2i16(1021)F/S, are located in the I/II domain interface outside of site 1, whereas the kdr mutation F^3i13(1565)C and a previously identified pyrethroid-sensing residue, F^3i16(1518), in the cockroach sodium channel (16) are situated in the II/III domain interface within site 1 (Fig. 2A). Remarkably, the two pairs of residues [I^2i13(1018) and L^2i16(1021) vs. F^3i13(1565) and F^3i16(1518)] are located in symmetric positions in the model. We therefore hypothesized that I^2i13(1018) and L^2i16(1021) could contribute to a second pyrethroid-receptor site. This site is formed by residues in IL45, IS5, and IIS6 (the IL45–IS5–IIS6 triangle), which would be analogous to site 1, formed by residues in IIL45, IIS5, and IIIS6 (the IIL45–IIS5–IIIS6 triangle).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Identification of a second pyrethroid-receptor site on the AaNa_v1–1 channel. (A) Topology of the sodium channel protein indicating key pyrethroid-sensing residues in site 1 (triangles) and site 2 (circles). Here we designate residues using the nomenclature that is universal for P-loop ion channels. A residue is labeled by the domain number (1–4, as I–IV in the topology), segment type (k, L45 linker; i, inner helix; o, outer helix), and the relative number of the residue in the segment (50). (B and C) Percentage of channel modification by permethrin (1 μM) (B) and deltamethrin (1 μM) (C). The number of oocytes for each mutant construct was >5. Error bars indicate mean ± SEM. Asterisks indicate significant differences from the AaNa_v1–1 channel as determined by using one-way analysis of variance with Scheffé's post hoc analysis (P < 0.05).

To test the second binding site model, we conducted mutational analysis of the following additional residues in the predicted site 2: I^1k7 and V^1k11 in IL45 and L^1o6, I^1o10, and T^1o13 in IS5 (Fig. 2A). The positions of these residues are analogous to those at site 1 that have been shown to confer pyrethroid resistance (Fig. 2A) (12, 14). In addition, we also included another residue, L¹ⁱ¹⁸ in IS6, whose position is analogous to the kdr mutation V^2i18(1023)G in IIS6 at site 1 that we have confirmed in this study. Three substitutions, I^1k7A, V^1k11A, and T^1o13W, altered the voltage dependence of activation and/or inactivation, whereas other substitutions did not (Table S2). In agreement with our prediction, I^1k7A, V^1k11A, and L¹ⁱ¹⁸G reduced the sensitivity of the AaNa_v1–1 channel to PMT and DMT, whereas I^1o10C reduced the sensitivity to PMT (Fig. 2 B and C). Interestingly, none of the three mutations in IS5 reduced the sensitivity to DMT and only one of them reduced the sensitivity to PMT (Fig. 2B). These results provide experimental evidence to support the existence of a second pyrethroid-receptor site in a pocket formed by the IL45–IS5–IIS6 triangle, but they also suggest that the precise architecture of site 1 and site 2 are not identical because, unlike site-1 mutations in IIS5, some of the corresponding mutations in IS5 did not have a strong effect on DMT or PMT action.

Refined Modeling of the Pyrethroid-Receptor Site 2 on the AaNa_v1–1 Channel.

With the identification of residues I²ⁱ¹³, L²ⁱ¹⁶, I^1k7, V^1k11, and L¹ⁱ¹⁸ as critical determinants for pyrethroid action at site 2, we used the Monte Carlo (MC) energy minimization method (MCM) to dock pyrethroids between the linker helix IL45 and transmembrane helices IS5, IS6, and IIS6. Preliminary docking of DMT in the K_v1.2-based model yielded two ensembles of binding modes in which the terminal CBr₂ and phenyl groups of DMT approached, respectively, either I^1k7 and L¹ⁱ¹⁸ (ensemble 1) or L¹ⁱ¹⁸ and I^1k7 (ensemble 2). Due to limited precision of homology models, energy alone could not be used to favor one of the ensembles. To resolve the problem, we searched for possible binding sites of tetramethylcyclopropane, the pyrethroid fingerprint model. Many random starting positions and orientations were generated and MC minimized. Calculations predicted that tetramethylcyclopropane fits a site between IL45, IS5, IS6, and IIS6 (Fig. 3A). The dimethylcyclopropane ring of DMT can approach this site in the first, but not the second ensemble. We further generated many initial placements of DMT and 1R-cis-permethrin (R-PMT), an active isomer of PMT (Fig. S3), between I^1k7 and L¹ⁱ¹⁸ of the open-channel model and searched for possible binding modes using many-stage MCM protocol (SI Materials and Methods). The predicted low-energy binding modes of DMT and R-PMT in the I/II interface are rather similar (Fig. 3 B and C). Most of the experimentally detected pyrethroid-sensing residues in IL45, IS6, and IIS6 form favorable contacts with the ligands. The CCl₂ or CBr₂ group binds between IL45 and IIS6, whereas the phenyl group binds between IS6 and IIS6, suggesting that the pyrethroid agonists increase the current by clamping the domain interface in its open state. Furthermore, the cyano group of deltamethrin directs toward the cytoplasm and binds in the I/II interface, where hydrophilic residues N²ⁱ²⁰ and S¹ⁱ²⁹ are located. Further mutational and electrophysiological experiments involving different pyrethroids will be necessary to elaborate a more detailed model for the binding of type I and type II pyrethroids in sites 1 and 2.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Predicted complexes of the open AaNa_v1–1 channel (K_v1.2-based model) with tetramethylcyclopropane (A), deltamethrin (B), 1R-cis-permethrin (C), and 1S-cis-permethrin (D). Cyan, orange, pink, and yellow surfaces are side chains of pyrethroid-sensing residues in segments IL45, IS5, IS6, and IIS6, respectively. The halogen atoms of pyrethroids bind between IL45 and IIS6. The phenyl rings of deltamethrin and 1R-cis-permethrin (R-PMT) bind between IS6 and IIS6. The phenyl ring of the inactive 1S-cis-permethrin (S-PMT) forms weaker contacts with L¹ⁱ¹⁸ and I^2o13 compared with R-PMT.

In our models, the terminal aromatic group of deltamethrin binds between helices IS6 and IIS6, whereas the dibromoethenyl moiety binds between helices IIS6 and IL45. This orientation of the pyrethroid molecule ensures that the bulky rigid dimethylcyclopropane moiety fits between helices IL45, IS5, IS6, and IIS6, into a seemingly void space “below” the gating hinge (position i14). O’Reilly et al. (15) proposed a model in which the dimethylcyclopropane moiety of acrinathrin and bifenthrin binds “above” the gating hinge, in a region between two inner helices and the pore helix. In the X-ray structures of open potassium channels, contacts between alpha helices above the gating hinge are tighter than those below the gating hinge. Fitting the bulky dimethylcyclopropane moiety in a region of loose intraprotein contacts seems more likely than in a region of tight contacts.

In the completely extended conformation of DMT, the most distant heavy atoms (the terminal-phenyl p-carbon and a bromine) are ∼14 Å apart. The most distant pyrethroid-sensing residues in both K_v1.2- and Na_vAb-based homology models of the AaNa_v1–1 channel are I^1k7 and L¹ⁱ¹⁸. C^β atoms of these residues are 18.8 and 21.4 Å apart in the respective models, suggesting that simultaneous interaction of DMT with both I^1k7 and L¹ⁱ¹⁸ is less likely in the closed channel than in the open channel. To test this, we MC minimized the Na_vAb-based model with DMT constrained between I^1k7 and L¹ⁱ¹⁸. In the low-energy binding modes of the completely extended conformations, DMT approached I^1k7, but not L¹ⁱ¹⁸ (Fig. S4A).

Another striking feature of pyrethroids is their extreme stereospecificity for insecticidal activity: only certain stereoisomers are active (1). For example, among a total of four permethrin isomers, 1R-cis and 1R-trans isomers are active, whereas 1S-cis and 1S-trans isomers are inactive (1). Interestingly, although the inactive 1S-cis PMT (S-PMT) (Fig. S3) does not induce any tail current, it competitively inhibits the tail current induced by R-PMT, suggesting that S-PMT competes for the same binding sites with R-PMT (18, 36, 37). Consistent with these data, our modeling correctly predicted that S-PMT can bind to receptor site 2, but forms weaker contacts with L¹ⁱ¹⁸ and I²ⁱ¹³ compared with R-PMT (Fig. 3D). The extreme lipophilicity of pyrethroids prevents evaluation of pyrethroid binding using conventional radioligand binding assays (12). As an alternative method, we have been using Schild analysis to determine the binding affinity of pyrethroids to sodium channels (18, 36, 37). In this study, we conducted Schild analysis of two additional site 2 mutant channels: I^1k7A and V^1k11A. The competitive replacement of R-PMT by S-PMT was significantly less efficient for I^1k7A and V^1k11A channels, compared with the wild-type channel (Fig. S5 and SI Results). Our results indicate that both I^1k7 and V^1k11 are involved in the binding of both S-PMT and R-PMT. Superposition of X-ray structures of the pH-gated bacterial potassium channel KcsA and the voltage-gated bacterial sodium channel Na_vAb (38) indicates similar folding of the transmembrane region in the pore module, which supports our model. However, further refining of our model will be necessary when an X-ray structure of a eukaryotic sodium channel in the open state becomes available.

Contribution of Site 2 to Pyrethroid Selectivity on Insect vs. Mammalian Sodium Channels.

A well-known property of pyrethroids is their differential action on insect vs. mammalian sodium channels (30, 39). Several mammalian sodium channel isoforms, such as rat Na_v1.2 (rNa_v1.2) and rat Na_v1.4 (rNa_v1.4), are virtually insensitive to pyrethroids in in vitro expression systems (27, 40, 41), which contributes to the favorable selective toxicity of pyrethroids (30, 39). Previously, the I^2k11M substitution, which corresponds to the super kdr mutation M^2k11T in IIL45 (site 1), was found to significantly enhance the sensitivity of rNa_v1.2 and rNa_v1.4 channels to pyrethroids (40, 41). Our discovery of site 2 prompted us to investigate the role of this site in mediating pyrethroid selectivity. We first explored a sequence alignment between insect and mammalian sodium channels (Fig. S6) to determine species-specific residues at site 2 that would explain the selectivity of pyrethroids between insect and mammalian sodium channels. Site 2 residues are highly conserved in insect sodium channels, but some of them are substituted in mammalian sodium channels (Fig. S6). Specifically, all insect sodium channels have V^1k11 and I^1o10, whereas nine rat (also human) sodium channels have L^1k11 and M^1o10 at respective positions, except for T^1o10 in rNa_v1.8 (Fig. S6). Furthermore, mutational analysis of site 2 residues in the cockroach sodium channel validates the importance of site 2 in pyrethroid sensitivity in another insect sodium channel (Figs. S7 and S8). Next, we performed amino acid substitution experiments by introducing insect site-2 residues into rat sodium channels. We found that the M^1o10I substitution alone in rNa_v1.2 or rNa_v1.4 channels did not significantly increase the sensitivity of these channels to DTM (Fig. 4 B and C). The L^1k11V substitution significantly increased the DMT sensitivity of rNa_v1.4, but not rNa_v1.2 (Fig. 4 B and C). Interestingly, the M^1o10I/L^1k11V double mutations significantly enhanced the DMT sensitivity of both rNa_v1.2 and rNa_v1.4 channels (Fig. 4 B and C). Most remarkably, the L^1k11V/M^1o10I/I^2k11M triple mutations produced channels that are much more sensitive to DTM than single- or double-mutant channels (Fig. 4 B and C). Taken together, these results demonstrate that site 2 is critically important for the selective action of pyrethroids on insect vs. mammalian sodium channels.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Transferring AaNa_v1–1 site 2 residues into mammalian sodium channels enhances mammalian sodium channel sensitivity to deltamethrin. (A) Topology of the mammalian sodium channel indicating site 1 and site 2 residues substituted by respective AaNa_v1–1 channel residues. (B and C) Percentage of channel modification by deltamethrin of rNa_v1.2 (B) and rNa_v1.4 (C) and mutant channels. The number of oocytes for each mutant channel was >5. Error bars indicate mean ± SEM. Asterisks indicate significant differences as determined by using one-way analysis of variance with Scheffé's post hoc analysis (P < 0.05).

In the absence of X-ray structure of a eukaryotic sodium channel with bound pyrethroids, it is impossible to rule out the possibility that mutations of the residues in the proposed site 2 allosterically modify the binding of pyrethroids to the previously predicted site 1. However, several observations argue against such a possibility. First, the Hill analysis suggests that there is more than one pyrethroid binding site (19). Second, many residues in site 2 are rather far from the nearest residues in site 1. For example, beta carbons of L¹ⁱ¹⁸ (site 2) and V²ⁱ¹⁸ (site 1) are as far as 13.3 Å apart. Third, the single site-1 model could not explain several kdr mutations, which can now be explained in view of the dual receptor-site model. Fourth, our computational model predicted several previously unknown pyrethroid-sensing residues in site 2. Indeed, subsequent experiments confirmed most of these predictions. Finally, site 2 contains pyrethroid-sensing residues that differ substantially between mammalian and insect channels (Fig. S6). A model with dual pyrethroid binding sites is therefore consistent with the well-known and fundamentally important selective toxicity of pyrethroids toward insects.