The molecular targets of ivermectin and lotilaner in the human louse Pediculus humanus humanus: New prospects for the treatment of pediculosis

Control of infestation by cosmopolitan lice (Pediculus humanus) is increasingly difficult due to the transmission of parasites resistant to pediculicides. However, since the targets for pediculicides have no been identified in human lice so far, their mechanisms of action remain largely unknown. The macrocyclic lactone ivermectin is active against a broad range of insects including human lice. Isoxazolines are a new chemical class exhibiting a strong insecticidal potential. They preferentially act on the γ-aminobutyric acid (GABA) receptor made of the resistant to dieldrin (RDL) subunit and, to a lesser extent on glutamate-gated chloride channels (GluCls) in some species. Here, we addressed the pediculicidal potential of isoxazolines and deciphered the molecular targets of ivermectin and the ectoparasiticide lotilaner in the human body louse species Pediculus humanus humanus. Using toxicity bioassays, we showed that fipronil, ivermectin and lotilaner are efficient pediculicides on adult lice. The RDL (Phh-RDL) and GluCl (Phh-GluCl) subunits were cloned and characterized by two-electrode voltage clamp electrophysiology in Xenopus laevis oocytes. Phh-RDL and Phh-GluCl formed functional homomeric receptors respectively gated by GABA and L-glutamate with EC50 values of 16.0 μM and 9.3 μM. Importantly, ivermectin displayed a super agonist action on Phh-GluCl, whereas Phh-RDL receptors were weakly affected. Reversally, lotilaner strongly inhibited the GABA-evoked currents in Phh-RDL with an IC50 value of 40.7 nM, whereas it had no effect on Phh-GluCl. We report here for the first time the insecticidal activity of isoxazolines on human ectoparasites and reveal the mode of action of ivermectin and lotilaner on GluCl and RDL channels from human lice. These results emphasize an expected extension of the use of the isoxazoline drug class as new pediculicidal agents to tackle resistant-louse infestations in humans.


Introduction
Lice are spread throughout the world in both low-and high-income countries, with heterogenous prevalence depending on the geographical area [1,2]. Pediculosis prevailing among school age children and homeless persons represents major economic, social but also public health concerns [3][4][5][6][7][8]. Human lice include the body louse Pediculus humanus humanus and the head louse Pediculus humanus capitis (order: Phthiraptera: Pediculidae). Both species are obligate blood feeding ectoparasites, living in clothes and in the scalp area, respectively [9]. Despite morphological, physiological and ecological differences, genetic studies showed that head and body lice have almost the same genomic and transcriptomic contents [9][10][11]. Furthermore, body lice are vectors of the pathogenic bacteria Bartonella quintana, Borrelia recurrentis and Rickettsia prowazekii, responsible for trench fever, louse-borne relapsing fever and epidemic typhus, respectively [4]. The control of lice infestation remains increasingly difficult due to the mode of transmission and the relative unreliable efficicacy of the available treatments [12]. Even though silicon-based suffocating products are widely used nowadays, their efficacy largely depends on the respect of their application and is limited against the nits [13]. Chemical insecticides such as organochloride (lindane), organophosphates (malathion), carbamates (carbaryl) and pyrethroids (pyrethrins and synthetic permethrin), which are the first line treatments recommended by the Centers for Disease Control and Prevention and by the American Academy of Pediatrics, have been widely used to control lice [14,15]. They have been proposed to inhibit γ-aminobutyric acid (GABA)-gated chloride channels, acetylcholinesterase and voltage-gated sodium channels, respectively, leading to the paralysis of the parasite. However, louse resistance to these products has been widely reported [16][17][18][19] as well as neurotoxic effects in children and toxicity on the environment [20][21][22][23][24][25]. Because macrocyclic lactones (MLs) are broad-spectrum neuroactive insecticides, acaricides and nematicides used to treat and prevent endo-and ectoparasites of humans and animals [26], ivermectin (a derivate of the ML avermectins) has emerged as a promising pediculicide and quickly became the recommended molecule for the control of body lice [27][28][29]. In invertebrates, ivermectin binds to glutamate-gated chloride channels (GluCls) leading to a reduction in motor activity, paralysis and death. However, the molecular basis for the action of ivermectin in louse is unknown. Recently, treatment failures of ivermectin against lice were observed [30,31]. In that respect, there is an urgent need to identify new druggable targets and develop new pediculicidal compounds.
Cys-loop ligand-gated ion channels (LGICs) are the major pharmacological targets of insecticides [32]. Among LGICs, the GluCls are found at inhibitory synapses of the nervous system exclusively in invertebrates [33], while GABA is the main inhibitory neurotransmitter in nerves and muscles in both vertebrates and invertebrates. GluCls and GABA receptors are constituted by the oligomerization of five identical or different subunits around a central pore [34]. In arthropods, the finding that the insecticidal activity of avermectins is mainly mediated by the GluCls was first achieved in the model insect Drosophila melanogaster [35]. For the GABA channels (GABACls), the screening for mutant flies that survive exposure to the insecticides dieldrin and fipronil, allowed the identification of the resistant to dieldrin gene or rdl [36]. When heterologously expressed in Xenopus laevis oocytes, the GluCl and RDL subunits formed functional homomeric glutamate-gated and GABA-gated chloride channels, respectively [35,37]. GluCl and RDL subunits were subsequently described in many other insect species although few of them were functionally assayed in vitro. Mutations in the GluCl subunit confers resistance to avermectins, whereas resistance to dieldrin, fipronil and picrotoxinin was shown to be associated to mutations in the RDL subunit using functional in vitro and in vivo experiments [32].
Interestingly, new synthetic molecules from the isoxazoline chemical class were recently proven to be effective ectoparasiticides against fleas and ticks [38]. The isoxazoline insecticides include fluralaner, afoxolaner, sarolaner and lotilaner [39][40][41][42]. They were shown to act as potent non-competitive antagonists of GABACls from D. melanogaster as well as the cat flea Ctenocephalides felis and the livestock tick Rhipicephalus microplus by molecular experiments and voltage-clamp electrophysiology. To a lesser extent, isoxazolines were also shown to act as antagonists of arthropod GluCls [39,[43][44][45]. In addition, these compounds were efficient at blocking GABA-elicited currents from dieldrin-and fipronil-resistant channels. Recent studies on the actions of these compounds on cloned GABACls have accounted for their higher selectivity for insect over vertebrate receptors [42,43]. Compounds from the isoxazoline group are now considered as the next generation of broad-spectrum insecticides effective against veterinary, agricultural, sanitary and stored product pests from eleven arthropod orders [46]. However, the pediculicidal activity of isoxazolines, the identification of the target ion channels in lice and their potential use for human health have not been investigated so far.
The objectives of our study were to decipher the molecular targets of ivermectin and lotilaner in the human body louse species Pediculus humanus humanus (Phh). In the present paper, we describe in the human body louse: the in vivo pediculicidal activity, the target identification and in vitro mechanism of action of the phenylpyrazole insecticide, fipronil, the ML insecticide, ivermectin, and the last marketed isoxazoline, lotilaner. For this, we identified the human body louse orthologues of the GluCl and RDL subunit genes and subsequently achieved their respective functional characterizations in Xenopus oocytes using two-electrode voltage-clamp. We found that the Phh-GluCl and Phh-RDL subunits give rise to glutamate-gated and GABAgated homomeric channels having distinct pharmacological properties. Hence, we provide mechanistic insights for a distinct mode of action of ivermectin and lotilaner. In the context of drug-resistance, this study highlights the pediculicidal potential of lotilaner and a possible extension of its use in new strategies to control human louse infestations.

In vivo toxicity tests of ivermectin and lotilaner on lice and nits
In order to determine the efficacy of selected insecticides, we first performed insecticide exposure tests on laboratory-reared human adult body lice and nits (Fig 1A). Briefly, lice were incubated with ivermectin, fipronil, picrotoxin and lotilaner at different concentrations. The effects of the compounds on louse survival and immobilization (S1 Table for knockdown groups) were subsequently analyzed at four timepoints (1, 2, 3, and 24 hours). After 24h, there was 4.5% mortality in the groups of lice treated with 1% DMSO that was used to dissolve the different compounds (Fig 1B and S2 Table). Similarly, 100 μM picrotoxin was not efficient to kill the lice after 24h whereas 91% of the lice were killed or immobilized with exposure to 100 μM fipronil at the same time. In our experiments, fipronil was used as a positive control and this result corroborated previous bioassays in the literature [47]. As expected, ivermectin was more efficient than fipronil, since the death rate reached 30%, 3h after application of Photographs of the human body lice from the laboratory-reared colony used in this study. Different life stages are represented from the left to the right: larvae in stage 1 during hatching (L1), larvae in stage 2 (L2), larvae in stage 3 (L3), female adult, male adult. B. Adult lice (n = 44 per condition) were treated with different concentrations of lotilaner, ivermectin, fipronil or picrotoxin and death or immobilization (knockdown) rate was determined throughout 24 hours. Water and 1% DMSO were used as negative controls. C. Nits were treated with 100 μM of lotilaner or ivermectin and hatching was determined between 6 and 9 days. Water and 1% DMSO were used as negative controls. The results are expressed as percentage of hatching from three independent experiments with n = 50 to 190 nits. The lines represent medians. The results are representative of three independent experiments. https://doi.org/10.1371/journal.ppat.1008863.g001

PLOS PATHOGENS
Molecular targets of insecticides in the human louse 100 μM, and 100% after 24h application of both 10 and 100 μM (Fig 1B). These findings confirmed the insecticide susceptible status of the laboratory-reared colony used here. Strikingly, after application of 100 μM lotilaner, the number of living lice declined rapidly. Not only the application of 1 μM lotilaner killed 14% of lice at 24 h but 10 and 100 μM lotilaner led to the death of all lice 3h after exposition (Fig 1B). Interestingly, when compared with other drugs, 10 μM lotilaner was more efficient than 10 μM ivermectin and 100 μM fipronil. Next, we analyzed the effect of 100 μM ivermectin or lotilaner on the nit hatching between six and nine days after drug application. The rate of nits hatched did not differ between the ivermectin-and the lotilaner-treated nits and the negative controls (water and 1% DMSO) thus revealing the absence of efficacy of the two products on the nits (Fig 1C). Altogether, these results confirm that fipronil and ivermectin are efficient at killing human adult body lice and provide the first in vivo evidence of a pediculicidal activity for lotilaner. These data prompted us to investigate the targets of ivermectin and lotilaner at the molecular level.

GluCl and RDL subunits are conserved in P. humanus humanus
Macrocyclic lactones and isoxazolines were previously shown to act both on arthropod glutamate-and GABA-gated channels. Therefore, we decided to identify and clone the respective P. humanus humanus GluCl and RDL subunits through a candidate gene approach. Searches for homologues of A. mellifera and D. melanogaster GluCl and RDL receptor subunits in P. humanus genomic/transcriptomic databanks, allowed the identification of two independent coding sequences for Phh-glucl and for Phh-rdl for which the respective complete coding sequences were obtained by RACE PCR experiments. All louse GluCl and RDL displayed the characteristic features of cys-loop LGIC subunits, including a signal peptide, an extracellular N-terminal domain containing the ligand binding sites, a disulfide bond formed by two cysteines that are 13 amino acid residues apart (cys-loop domain), four transmembrane domains (TM1-TM4) and a large intracellular loop between TM3 and TM4 that is highly variable (Fig 2A). Of importance is the sequence analysis of the pore selectivity filter located in the TM2 supporting the prediction with the signature of a chloride channel. Indeed, Phh-GluCl and Phh-RDL contain the PAR motif centered at the 0' position in the TM2 and conserved in other species (universal LGIC TM2 numbering system). A proline at position -2', followed by a small amino acid at position -1' like an alanine, an arginine at the 0' residue and a threonine at position 13' are the hallmarks of chloride-permeable LGICs [48].
The full-length coding sequences for the GluCl subunits included ORFs of 1353 and 1350 bp encoding proteins of 450 and 449 amino acids that were named Phh-GluCl-1 (MT321070) and Phh-GluCl-2 (MT321071), respectively. The alignement of Phh-GluCl-1 and Phh-GluCl-2 deduced amino-acid sequences is provided in Fig 2A. The two sequences are highly conserved with 97% identity and 98% similarity and only differ by eleven amino acid substitutions and an alanine insertion in the N-terminal part at position 59 in Phh-GluCl-1. (Figs 2A and S1). Noteworthy, all the amino acids involved in the glutamate binding were conserved.
A distance tree analysis of P. humanus humanus GluCl and RDL sequences including D. melanogaster, the honey bee Apis mellifera, the house fly Musca domestica, R. microplus and the red flour beetle Tribolium castaneum confirmed the orthologous relationships of the louse

PLOS PATHOGENS
Molecular targets of insecticides in the human louse subunit sequences with their respective counterparts (Fig 2B). GluCl amino acid sequences revealed that Phh-GluCls were found to be highly conserved, sharing 78%, 84%, 81% and 70% identities with the respective orthologues from A. mellifera, T. castaneum, D. melanogaster and R. microplus. Phh-RDL-1 and 2 shared 84%, 81%, 67% and 54% identities with their orthologues from A. mellifera, T. castaneum, D. melanogaster and R. microplus, respectively. These observations are in accordance with the relative position of Phthiraptera and Hymenoptera insect subfamilies and Acarids [49] and were further confirmed by supplementary analysis including other members from the GABA and GluCl groups from additional arthropod species (S1 and S2 Figs). In the phylogeny of GABA subunits, Phh-RDL-1/-2 were found to clearly cluster distinctly from the LCCH3, GRD and 8916 subunits, showing that each species has inherited an ancestral gene. Altogether, the phylogenetic observations support that we cloned the P. humanus humanus GluCl and RDL subunit orthologues allowing their subsequent assessment at the functional level.

P. humanus humanus GluCl and RDL subunits form functional homomeric channels in Xenopus oocytes
In previous studies, arthropod GluCl and RDL subunits were shown to reconstitute functional channels when expressed in Xenopus oocytes [32]. Hence, we hypothesized that the ortholog subunits from P. humanus humanus could form functional receptors. Following the cloning of two full-length coding sequences for Phh-GluCl and Phh-RDL subunits, their respective cRNAs were microinjected in X. laevis oocytes. To investigate the receptor functionality, responses to agonists were assessed by two-electrode voltage-clamp (TEVC) electrophysiology. In sharp contrast with Phh-GluCl-1 and Phh-RDL-1 subunits (Fig 3), injections of Phh-GluCl-2 (n = 11) and Phh-RDL-2 (n = 12) cRNAs never led to recordings of glutamate-and GABAevoked currents, respectively, indicating that these subunits did not form functional receptors (S3 Fig). Thus, Phh-GluCl-1 and Phh-RDL-1 will be henceforth refered to as Phh-GluCl and Phh-RDL in the article. When glutamate was perfused in the recording chamber, we could record fast inward currents in the μA range in oocytes injected with Phh-GluCl ( Fig 3A). As expected, this receptor desensitized extremely rapidly upon prolonged glutamate exposure. Then, we obtained the concentration-response curve by challenging the oocytes with glutamate concentrations ranging from 1 to 1000 μM. With current amplitudes normalized to the maximal response to 1000 μM, we found that the glutamate EC 50 value of Phh-GluCl was 9.3 +/-1.3 μM with a Hill coefficient of 1.9 +/-0.5 (n = 5) suggesting the presence of more than one glutamate binding site (Fig 3A and 3B). When Phh-RDL cRNAs were injected in oocytes, the application of 100 μM GABA resulted in robust currents with maximum amplitudes in the μA range (Fig 3D). The GABA concentration-response curve was characterized by an EC 50 of 16.0 +/-0.4 μM (n = 6) with a Hill coefficient of 2.3 +/-0.1 (n = 6) suggesting the binding of more than two GABA molecules (Fig 3E). As previously shown in the literature, RDL and GluCls are permeable to chloride anions [35,37,[50][51][52][53]. Here, the TM2 amino acid sequences of Phh-RDL and Phh-GluCl are respectively 100% identical with RDL and GluCls from other insect species such as A. mellifera, D. melanogaster, T. castaneum (Fig 2A), suggesting they are anionic receptors. In order to investigate the ion selectivity, we applied voltageramps to oocytes expressing Phh-GluCl or Phh-RDL and measured the reversal potentials for the respective glutamate-and GABA-sensitive currents in recording solution with varying concentrations of either chloride or sodium. Then, the reversal potentials were plotted against the chloride or sodium concentration (Fig 3C and 3F). In the standard recording solution containing 100 mM NaCl, glutamate-induced currents reversed at -23.39 +/-1.26 (n = 12) (Fig 3C). When the extracellular chloride concentration decreased from 104.5 to 4 mM Glutamate concentration-response curve for Phh-GluCl (mean +/-SEM, n = 5 oocytes). Data were normalized to the maximal effect of glutamate. C. Reverse potentialion concentration relationships of oocytes expressing Phh-GluCl obtained from voltage-ramps in presence of glutamate and using increasing concentrations of either Clor Na + to analyze the selectivity of this channel. D. Current traces recorded on oocytes injected with Phh-RDL cRNA in response to the application of 1 to 100 μM of GABA. Oocytes were clamped at -60 mV. Application times are indicated by the bars. E. GABA concentration response curve for Phh-RDL (mean +/-SEM, n = 6 oocytes). Data were normalized to the maximal effect of GABA. F. Reverse potential-ion concentration relationships of oocytes expressing Phh-RDL obtained from voltage-ramps in presence of GABA and using increasing concentrations of either Clor Na + to analyze the selectivity of this channel. https://doi.org/10.1371/journal.ppat.1008863.g003

PLOS PATHOGENS
Molecular targets of insecticides in the human louse (replacement of chloride by sodium acetate), Phh-GluCl exhibited a 13.4 +/-2.3 mV shift upon a 10-fold change in the chloride concentration (n = 7). In contrast, decreasing the extracellular sodium concentration from 100 to 0 mM (replaced by tetraethylammonium chloride) did not significantly shift the reversal potential (1.0 +/-0.5 mV shift per decade, n = 47). Similarly, Phh-RDL was highly sensitive to the extracellular chloride concentration (36.4 +/-1.2 mV shift upon a 10-fold change in the chloride concentration, n = 4), whereas it was almost insensitive to variations of the extracellular sodium concentration (2.3 +/-0.8 mV shift per decade, n = 4) (Fig 3F). These results indicate that both Phh-GluCl and Phh-RDL form chloride permeable GABA-gated channels that do not permeate sodium ions, consistently with the pore selectivity filter sequence (Fig 2A) and channels from other species [53,54]. In summary, these findings demonstrate that two novel functional homomeric receptors from the human body louse, Phh-GluCl and Phh-RDL, with respectively high affinity to glutamate or GABA, were reconstituted in Xenopus oocytes.

Differential effects of ivermectin and lotilaner on P. humanus humanus GluCl and RDL channels
Arthropod GluCl and RDL channels are well-known targets for insecticides [32]. In order to investigate their respective insecticides sensitivities, Phh-GluCl and Phh-RDL were challenged with ivermectin, lotilaner, fipronil and picrotoxin. Recordings from oocytes expressing Phh-GluCl revealed that ivermectin is a potent agonist of this receptor. A slow activation of the receptor was observed with application of ivermectin as low as 10 nM and large μA currents were recorded in response to 1 μM ivermectin with an average of 86.5 +/-8.2% (n = 6) of the 1 mM glutamate-elicited current (Fig 4A). The dose-response curve (Fig 4B) revealed an ivermectin EC 50 of 78.52 +/-30.22 nM (n = 3-9). As expected, the effect of ivermectin was not reversible. When applied on Phh-GluCl expressing oocytes, 1 to 30 μM lotilaner did not elicit any current and mediated no significant inhibitory effect on glutamate-evoked currents (p > 0.057) (Fig 4C and 4D). We subsequently tested the activity of selected insecticides on the Phh-RDL channel. As anticipated from the literature, lotilaner, picrotoxin and fipronil had no agonist effect (Fig 5). Interestingly, applications of 0.3 to 30 μM ivermectin during 90 seconds on Phh-RDL elicited slow and sustained currents which were not reversible (S4A Fig).
The maximum effect corresponding to 7.9 +/-1.3% (n = 6) of the 1 mM glutamate response was obtained with 1 μM ivermectin (S4B Fig). Then, we investigated the potential antagonistic effects of the selected insecticides on GABA recording traces. Fig 5A-5D illustrates results from application of 100 μM GABA before, during and after the sequential addition of increasing concentrations of insecticide. GABA-elicited currents were completely inhibited with 3 μM fipronil, ivermectin and lotilaner. On the other hand, picrotoxin had a weaker effect on Phh-RDL since the full inhibition of the GABA elicited current was observed with 30 μM. In accordance with the litterature [55], ivermectin acted as an inhibitor of Phh-RDL. More importantly, GABA responses were totally and irreversibly blocked with lotilaner as low as 0.3 μM suggesting that Phh-RDL is more sensitive to lotilaner than to the other insecticides fipronil, picrotoxin and ivermectin. To characterize this effect, concentration-inhibition response relationships were performed for the four drugs (Fig 5E). Picrotoxin, ivermectin and fipronil antagonized Phh-RDL in a concentration-dependent manner with IC 50 values of 1.16 +/-0.23 μM (n = 6), 342.5 +/-14.0 nM (n = 6) and 137.5 +/-35.1 nM (n = 6), respectively. Lotilaner resulted in the most potent dose-dependent inhibition of the GABA-mediated currents with an IC 50 value of 40.7 +/-6.8 nM (n = 6) (Fig 5E). Rank order potency series for each of the Phh-RDL antagonists tested was as follows: lotilaner > fipronil > ivermectin > picrotoxin. In order to determine the type of inhibition caused by lotilaner and ivermectin on Phh-RDL,

PLOS PATHOGENS
Molecular targets of insecticides in the human louse we performed a competitive experiment (Fig 6). Briefly, a fixed concentration of lotilaner (1 nM and 10 nM) or ivermectin (100 nM and 1 μM) was continuously applied on oocytes while performing a GABA dose-response. The EC 50 values and maximum current amplitudes were 16.0 +/-0.4 μM and 97.95 +/-0.69% (n = 6) for GABA alone, 14.3 +/-1.9 μM and 69.16 +/-2.54% (n = 5) in the presence of 1 nM lotilaner and 11.5 +/-4.0 μM and 48.56 +/-4.90 (n = 6) in the presence of 10 nM lotilaner (Fig 6A). The presence of lotilaner did not significantly change the GABA EC 50 (p > 0.49 and p > 0.32 for 1 nM and 10 nM, respectively) but reduced significantly the maximum current amplitudes obtained with a saturating GABA concentration (p < 0.0001), which is a characteristic feature of a non-competitive antagonism. The competition experiment with ivermectin showed the remarkable ivermectin mode of action on Phh-RDL. Indeed, coapplication of low ivermectin concentration (100 nM) and low GABA concentration (3 μM) increased the current amplitude compared to the application of GABA alone highlighting the potentiation effect of ivermectin (Fig 6B). A higher concentration of ivermectin (1 μM) had a dual mode of action with a weak current activation (S4 Fig) and an

PLOS PATHOGENS
Molecular targets of insecticides in the human louse inhibition of the GABA-elicited current indicating both an agonistic and antagonistic effect of ivermectin (Fig 6B). Altogether, these results demonstrate that P. humanus humanus GluCl and RDL characterized in Xenopus oocytes display original pharmacologies regarding insecticides. Ivermectin acted as a strong agonist of Phh-GluCl while showing a double action on Phh-RDL with agonist or antagonist effects depending on the concentration. Finally, the antiticks and fleas lotilaner had no effect on Phh-GluCl, while it was a potent and selective noncompetitive antagonist of RDL of human lice.

First functional characterization of LGICs from human lice
In the present study, we report the first functional expression and pharmacological characterization of any human louse LGICs. Indeed, despite the social and public health impacts of human lice and the growing issue of resistance to pediculicides and since the complete sequencing of the body louse genomes, there was to our knowledge no functional characterization of any LGICs from lice so far [10,29]. Because glutamate and GABA are important neurotransmitters in the insect central nervous system and their receptors are targets for insecticides, we focused our attention on glutamate-and GABA-sensitive receptors. GluCl and RDL channel subtypes of LGICs have been extensively characterized in many arthropod species starting with the model fruit fly D. melanogaster and ranging from agricultural and sanitary pests to animal ectoparasites and disease vectors, but also beneficial insects like the honey bee [32,33]. Here, we show that P. humanus humanus GluCl and RDL subunits gave rise to homomeric channels when functionally expressed in Xenopus oocytes. This experimental expression system provided a unique means to assess the pharmacological and biophysical properties of these receptors. Even though two putative genes encoding respectively a GluCl and a GABA receptor have been identified in the genomic database, the molecular cloning of complete cDNAs revealed at least two potential full-length transcripts for both candidates (i.e. Phh-GluCl-1, Phh-GluCl-2 and Phh-RDL-1, Phh-RDL-2). These transcripts could either be alternative spliced isoforms or result from RNA A-to-I editing, a mechanism of RNA edition described in lice [56]. In insects, alternative splicing and RNA A-to-I editing are known to increase diversity of LGICs subunits [57][58][59][60]. Undoubtedly, further molecular investigations at the DNA level would be required to confirm this hypothesis. Here, we report that both Phh-

New insights into the mode of action of ivermectin on lice
Despite the massive use of pediculicidal compounds during the past decades, there was to our knowledge no functional characterization of their molecular targets in lice. Macrocyclic lactones are widely used in both veterinary and human medicine, with ivermectin being typically the recommended molecule for the control of body lice [27][28][29]. This broad-spectrum drug is highly efficient against a wide range of pathogens including endo-and ectoparasites such as nematodes, acarians and insects [26]. In comparison with other pediculicides, ivermectin presents some major advantages. First, it has long been recognized that it is a safe pharmaceutical in human health due to low level of toxicity for mammalians, in sharp contrast with classical pediculicides [70]. Indeed, former drugs such as fipronil, lindane or malathion are now forbidden for human use in several countries because of their toxicity and environmental impact [20][21][22][23][24][25]. Second, ivermectin is currently the only molecule used as an oral treatment against lice. For head lice, oral route avoids suboptimal shampoo application, thus reducing the emergence and transmission of resistant lice [28,[71][72][73]. For body lice, oral dosing is of particular interest to treat people having a limited access to basic hygiene services. However, despite these remarkable advantages, ivermectin's efficacy is currently compromised by the emergence of resistant lice [30,31].
Here, our bioassay confirmed the pediculicidal activity of ivermectin on human body lice. Furthemore, we addressed for the first time the mode of action of ivermectin at the molecular level on louse glutamate-and GABA-gated receptors. We showed that ivermectin preferentially activated Phh-GluCl channel, and Phh-RDL in a much lower extent. In addition, ivermectin blocked the RDL receptor with an IC 50 4-fold higher than the ivermectin EC 50 for Phh-GluCl. Hence, these findings suggest that Phh-GluCl is the preferential target of ivermectin. Interestingly, functional silencing using RNAi has already been shown to be an efficient and promising approach to investigate the role of detoxification genes in ivermectin tolerance of human body lice [74]. In order to confirm that GluCl is the main target of ivermectin in lice, it would be interesting to explore in vivo the phenotype of Phh-GluCl siRNA silenced lice as previously demonstrated for abamectin in T. cinnabarinus and in the crop whitefly pest Bemisia tabaci [64,75]. Finally, following the recent observation of GluCl single-nucleotide-polymorphisms (SNPs) in phenotypically ivermectin-resistant head lice [30], our results provide a unique means to explore the molecular mechanisms underlying louse resistance to ivermectin. In a further step, such electrophysiology studies in Xenopus oocytes will enable to determine the functional relevance of GluCl SNPs detected in ivermectin-resistant lice [51,55,62,76,77].

Molecular targets of insecticides in the human louse
Although the putative role of those SNPs are to be determined, we can reasonably assume that mutated subunits can lead to GluCl altered functionality, probably resulting in ivermectin resistance. Additionaly, screening on ivermectin-resistant GluCls recombinantly expressed will provide a powerful tool to optimize the design of new pediculicidal compounds. Further investigations are required to understand the mechanisms of ivermectin resistance and to define new control strategies to avoid emergence of resistance as well as to preserve this molecule efficacy.

On the mode of action of isoxazolines in arthropods
With the recent introduction of new generation of compounds in the isoxazoline class (e.g. lotilaner), the development of detailed understanding of the actions of isoxazolines is essential. Arthropod RDL channels are major targets for several insecticidal classes including dieldrin itself and fipronil [32] and they were recently shown to be the primary target for isoxazolines [39][40][41][42][43]. Therefore, we hypothesized that they could act on RDL from Pediculus humanus humanus. In addition to the pharmacological characterization of Phh-RDL and Phh-GluCl, our results show that the inhibitory effects of lotilaner depend on the receptor subtype. Indeed, lotilaner efficiently and selectively blocked the GABA response on Phh-RDL, whereas Phh-GluCl was not affected, even with the use of a concentration of 30 μM representing 737-fold the lotilaner IC 50 for Phh-RDL. In order to confirm these findings, it will be interesting to silence the rdl gene using RNAi in human body lice [64]. Our results clearly demonstrated the selective antagonistic action of lotilaner on RDL over GluCl channels contrary to fluralaner, which was reported to inhibit both RDL and GluCls from M. domestica and R. microplus [39,43,78,79]. Whether these differences might be due to the compounds themselves or the pharmacological properties of receptors from different species remain to be investigated. Although some studies recently indicated that four amino acids in the transmembrane subunit interface were closely related to the antagonism of M. domestica RDL by fluralaner [79], the mechanism of action of lotilaner on RDL receptors at the molecular level remains to be deciphered. Likewise, further efforts are needed to better understand the molecular basis of the insect selectivity of isoxazolines over vertebrate channels. Contrary to homomeric arthropod RDL channels, the vertebrate GABA are heteromeric channels made of different subunits. Recent studies indicated that fluralaner and lotilaner were highly selective of arthropod RDL channels, while poorly efficient on rat brain membranes (IC 50 > 10 μM), rat GABA channels (IC 50 > 30 μM) and canine GABA channels (IC 50 > 10 μM), thus evidencing for low mammalian toxicity [39,42,43]. Resistance to isoxazolines has not been reported so far and dieldrin resistance is likely not to confer cross-resistance to fluralaner [43]. Furthermore, the antagonistic effect of fluralaner and lotilaner on RDL receptors from M. domestica and D. melanogaster was not impaired by known mutations confering fipronil and dieldrin resistance [42,43]. Likewise, the distinct mechanisms of action for ivermectin and lotilaner seem to indicate that resistance to ivermectin will not extend to the isoxazolines. Identification of homologous genes encoding RDL subunits and investigation of new expressed recombinant RDL from many other arthropod species including important insect pest species, acari and crustaceans will enhance our understanding of the action of isoxazolines. The potency and selectivity of isoxazolines indicate an important future for this class of compounds.

From dog to human: Isoxazoline insecticides as promising pediculicidal candidates
The isoxazoline class is a new generation of insecticidal compounds recently introduced for insect chemical control including ectoparasites such as acari and fleas [44,45]. Since the

PLOS PATHOGENS
Molecular targets of insecticides in the human louse discovery of isoxazolines, their insecticidal spectrum has become broader to agricultural, sanitary and animal insect pests belonging to eleven orders missing out the order of Phthiraptera [39][40][41][42][43]45,46]. In the present study, we hypothesized that the isoxazolines could represent attractive candidate drugs to extend the pharmacopeia for louse control. Among the different compounds from the isoxazoline family, we selected the insecticide lotilaner as a proof-of-concept model for the isoxazoline effectiveness as pediculicidal for human use. Isoxazolines are licensed for protection of companion animals against fleas and ticks and therefore present a reduced toxicity to mammals. In addition, isoxazolines have a long-lasting effect after a single oral administration [42,80]. In dogs, lotiIaner remains active against ticks and fleas during one month with an efficacy ranging from 98 to 100% [81,82]. Here, we assessed the pediculicidal potential of lotilaner against different stages of body lice. Strikingly, our bioassays on human body lice demonstrated that the adulticide effect of lotilaner was higher than that of ivermectin and fipronil. Whereas 10 μM ivermectin treatment killed 100% of adult lice after 24h, we obtained the same result with 10 μM lotilaner in only 3 hours following exposure. This fast activity is in accordance with the 4h-12h needed to kill ticks and fleas in lotilaner treated dogs [81,82]. Based on dog data, we can speculate that a single oral dose could provide a pediculicidal activity above 17 days corresponding to the complete life cycle of the human lice [9,19]. Therefore, the long-lasting activity of isoxazolines could overcome the absence of activity on embryonated eggs and also prevent re-infestations with a single oral dose. Considering its relatively benign toxicological profile for mammals along with rapid knockdown of lice, isoxazolines seems to be an excellent candidate drug to tackle louse infestations in humans. Nevertheless, the innocuity of isoxazolines towards human remains to be proven as well as its pharmacokinetics. Interestingly, Miglianico et al. recently reported the high efficacy of two other veterinarian drugs from the isoxazoline family (i.e afoxolaner and fluralaner) against a panel of insect species representing major vectors for human tropical diseases [83]. Computational modelling highlighted both the feasibility and the relevancy of using isoxazoline for human medication. In accordance, our results strongly support isoxazoline compounds as major candidates for drug repurposing alone or in co-administration with ivermectin. Indeed, new combinations of active compounds have become frequent in veterinary medicine (e.g. anthelmintics). Alike ivermectin, a useful feature of lotilaner is its mode of administration through oral route, which is more tractable to human than topical formulations. Since ivermectin and lotilaner are efficient at killing human lice, it is tempting to suggest that a combination of both drugs may have a considerable potential, resulting in an additive or a synergistic effect allowing the control of emerging ivermectin resistant lice and delay or overcome resistance. Such a therapeutic option was recently achieved with the release of fluralaner combined with moxidectin [84,85], afoxolaner plus milbemycin oxime [86] and sarolaner combined either with selamectin [87,88] or moxidectin and pyrantel on the veterinary market for small animals [89]. Hence, it is reasonable to conclude that further investigations on the efficacy of lotilaner alone or in combination with ivermectin for general use as pediculicides would be particularly pertinent and justified. We hope that the results presented here contribute to an increased understanding on the mechanisms behind the physiology of neurotransmission and of the mode of action of the pediculicidal drugs in body lice at the functional level and will provide new therapeutic strategies to ensure the sustainable and effective control of louse infestations.

Ethics statement
All animal care and experimental procedures were conducted in strict accordance with the European guidelines for the care and use of laboratory animals and were approved by French

Molecular targets of insecticides in the human louse
Ministry for Research and the regional Val de Loire ethics committee (no 19) as a protocol registered under the permit number APAFIS#8455-2017010616224913 v3.

Chemical assays on body lice
The colony of human body lice (Pediculus humanus humanus) were provided by Kosta Y. Mumcuoglu from the Kuvin Center for the Study of Infectious and Tropical Diseases, Hebrew University-Hadassah Medical School, Jerusalem, Israel. This laboratory-reared colony adapted to rabbit blood were fed four times per week for 20-30 min on rabbits and maintained at 30 ± 1˚C and 60-70% relative humidity without exposure to any drugs [90]. Survival and immobilization were monitored by microscopical observations at 1, 2, 3 and 24 h after contact of 44 lice at different stages of the life cycle with insecticides at concentrations ranging from 1 to 100 μM or 1% DMSO on a Whatman grade 1 filter paper in a 55 mm Petri dish following the knockdown classification from KD + to KD ++++ (S1 Table), as previously described [12]. Nits (50 to 190 per condition) were incubated with lotilaner or ivermectin and hatching was monitored between six and nine days. Water and solution of 1% DMSO were used as negative controls. Pictures were recorded with a Motic Digital Microscope 143 Series and the Motic Image Plus 3.0 software.

Cloning of full-length cDNA sequences of GluCl and rdl from Pediculus humanus humanus
Total RNA was extracted from 30 mg of lice (15 lice) using the Total RNA isolation Nucleospin RNA kit (Macherey-Nagel) according to the manufacturer's recommendations. Complementary DNA (cDNA) was synthesized using the GeneRacer kit with the SuperScript III reverse transcriptase (Invitrogen) following the manufacturer's recommendations. Using the A. mellifera and D. melanogaster GluCl and RDL subunit sequences as querries, tBLASTn searches in NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi) allowed the identification of partial genomic contigs (AAZO01006897 and AAZO01005501 for Phh-glucl and AAZO01005738, AAZO01000267 and AAZO01000266 for Phh-rdl) and partial mRNA sequences (XM_002429761 for Phh-GluCl and XM_002422861.1 for Phh-rdl) containing the 3' cDNA ends. Primers designed on the sequences retrieved from the BLAST searches are reported in S3 Table. Then, the corresponding 5' cDNA ends were obtained by RLM-RACE experiments using the Generacer kit with two rounds of PCR as described elsewhere [91]. After the identification of the cDNA 5' ends, new primers were designed. The full-length complete coding sequences of Phh-GluCl and Phh-rdl were subsequently amplified by nested PCRs with the proofreading Phusion High-Fidelity DNA Polymerase (Thermo Scientific) using two pairs of primers. Then, PCR products were cloned into the transcription vector pTB-207 [92] using the In-Fusion HD Cloning kit (Clontech) as described previously [93]. Recombinant plasmid DNA was purified using EZNA Plasmid DNA Mini kit (Omega Bio-Tek) and the sequences were checked (Eurofins Genomics). The novel complete coding sequences of the two Phh-GluCl and Phh-RDL subunits were depositied to Genbank under the accession numbers MT321070 to MT321073. The constructions were linearized with the MlsI restriction enzyme (Thermofisher) and cRNAs were synthetized in vitro using the mMessage mMachine T7 transcription kit following the manufacturer's recommendations (Ambion). Lithium chloride-precipitated cRNAs were resuspended in RNAse-free water and stored at -20˚C.

Sequence analysis and phylogeny
Signal peptide and transmembrane domain were predicted using SignalP4.1 [94] and SMART [95], respectively. Deduced amino-acid sequences of Phh-GluCl and Phh-RDL were aligned

PLOS PATHOGENS
Molecular targets of insecticides in the human louse using the MUSCLE algorithm [96] and further processed with GeneDoc (IUBio). The phylogenetic distance trees were generated on amino-acid sequences by the SeaView software [97] using BioNJ Poisson parameters and bootstrap values were calculated with 1000 replicates as described previously [98].

Electrophysiology in Xenopus laevis oocytes and data analysis
Defolliculated Xenopus laevis oocytes were purchased from Ecocyte Bioscience and maintained in incubation solution (100 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 .2H 2 O, 1 mM MgCl 2 .6H 2 O, 5 mM HEPES, 2.5 mM C 3 H 3 NaO 3 , pH 7.5 supplemented with penicillin 100 U/ mL and streptomycin 100 μg/mL) at 19˚C. Each oocyte was microinjected with 57 ng of cRNA encoding Phh-GluCl or Phh-RDL using a Drummond nanoject II microinjector. Three to five days after cRNA injection, two-electrode voltage-clamp recordings were performed with an oocyte clamp OC-725C amplifier (Warner instrument) at a holding potential of -80mV or -60 mV to assess the expression of the GluCl or RDL channels. Currents were recorded and analyzed using the pCLAMP 10.4 package (Molecular Devices). Concentration-response relationships for agonists were carried out by challenging oocytes with 10 s applications of increasing concentrations of compounds. The peak current values were normalized to the response to 1000 μM glutamate or 100 μM GABA giving the maximum current amplitude for GluCl and RDL channels, respectively. For Phh-GluCl, the dose-response relationship for ivermectin was performed by applying each concentration of ivermectin for 20-30 s. Each oocyte received only one ivermectin concentration (� 30 nM) followed with a high ivermectin concentration (� 100 nM). Data were normalized to a first 100 μM glutamate apllication. For RDL, the effect of antagonists was evaluated, by a first preincubation of each antagonist alone for 90 s, followed by the co-application with 100 μM GABA for 10 s. Oocytes were then washed with recording buffer and sequentially preincubated with an increasing antagonist concentration followed by coapplication with 100 μM GABA. The observed responses were normalized to the response induced by 100 μM GABA alone performed prior to challenging with the antagonist. The concentration of agonist required to mediate 50% of the maximum response (EC 50 ), the concentration of antagonist required to inhibit 50% of the agonist response (IC 50 ) and the Hill coefficient (nH) were determined using non-linear regression on normalized data with GraphPad Prism software. Concentration-response curves for agonists were fitted with the following equation: Where Y is the normalized response, H is the Hill slope and X is the logarithm of concentration. The same equation was used for the concentration-response curves for antagonists by replacing logEC 50 by logIC 50 . Results are expressed as mean +/-SEM and statistical analysis

PLOS PATHOGENS
Molecular targets of insecticides in the human louse were performed using One-Way ANOVA with Tukey's Multiple Comparisons Test. The competitive experiments on Phh-RDL were performed as previously described [42] with a concentration-response relationship in the presence of GABA alone or with 1 nM or 10 nM lotilaner. A pre-incubation of 1 nM or 10 nM lotilaner for 90 s was followed by a coappllication with increasing concentrations of GABA (1-1000 μM) for 10 s with a 30 s wash between each concentration. Current amplitudes were normalized with a first 100 μM GABA application. The corresponding concentration response curves were fitted using the following four parameter Hill equation: Where Bottom is the response at the minimal concentration and Top is the maximal response. Statistical comparison were performed using unpaired Student's t test. Chloride and sodium permeabilities were conducted as described previously [53]. In brief, the reversal potential of the GABA-induced currents was measured with a 1500-ms-long ramp of voltage from -60 to +60 mV in standard saline recording solution (100 mM NaCl, 2.5 mM KCl, 1 mM CaCl 2 , 5 mM HEPES, pH 7.3) as well as in recording solutions containing a concentration range of either chloride (NaCl replaced by sodium acetate) or sodium (NaCl replaced by tetraethylammonium chloride).

Materials
Glutamate, GABA, fipronil, picrotoxin and ivermectin were purchased from Sigma-Aldrich. Glutamate and GABA were directly dissolved in recording solution. Fipronil, picrotoxin and ivermectin were first dissolved at 10-100 mM in DMSO and then diluted in recording solution to the final concentrations in which DMSO did not exceed 1%. Lotilaner (Credelio, Elanco, USA) was a commercial formulation purchased in a local pharmacy store. Tablets containing 450 mg lotilaner were dissolved in DMSO to obtain a 100 mM lotilaner stock solution and subsequently diluted in recording solution at a final DMSO concentration less than 0.003%.  Table. Knockdown groups. Lice from groups KD +++ and KD ++++ were considered as dead [12]. (TIF) S2 Table. Rates of mortality of P. humanus humanus exposed to lotilaner, ivermectin, fipronil and picrotoxin. (TIF) S3

PLOS PATHOGENS
Molecular targets of insecticides in the human louse