A new family of glutamate-gated chloride channels in parasitic sea louse Caligus rogercresseyi: A subunit refractory to activation by ivermectin is dominant in heteromeric assemblies

Sea louse ectoparasitosis is a major threat to fish aquaculture. Avermectins such as ivermectin and emamectin have been effectively used against sea louse infestation, but the emergence of resistance has limited their use. A better understanding of the molecular targets of avermectins is essential to the development of novel treatment strategies or new, more effective drugs. Avermectins are known to act by inhibiting neurotransmission through allosteric activation of glutamate-gated chloride channels (GluCls). We have investigated the GluCl subunit present in Caligus rogercresseyi, a sea louse affecting aquaculture in the Southern hemisphere. We identify four new subunits, CrGluCl-B to CrGluCl-E, and characterise them functionally. CrGluCl-A (previously reported as CrGluClα), CrGluCl-B and CrGluCl-C all function as glutamate channel receptors with different sensitivities to the agonist, but in contrast to subunit -A and -C, CrGluCl-B is not activated by ivermectin but is rather antagonised by the drug. CrGluCl-D channel appears active in the absence of any stimulation by glutamate or ivermectin and CrGluCl-E does not exhibit any activity. Notably, the expression of CrGluCl-B with either -A or -C subunits gives rise to receptors unresponsive to ivermectin and showing altered response to glutamate, suggesting that coexpression has led to the preferential formation of heteromers to which the presence of CrGluCl-B confers the property of ivermectin-activation refractoriness. Furthermore, there was evidence for heteromer formation with novel properties only when coexpressing pairs E/C and D/B CrGluCl subtypes. Site-directed mutagenesis shows that three transmembrane domain residues contribute to the lack of activation by ivermectin, most crucially Gln 15’ in M2, with mutation Q15’T (the residue present in ivermectin-activated subunits A and C) conferring ivermectin activation to CrGluCl-B. The differential response to avermectin of these Caligus rogercresseyi GluClsubunits, which are highly conserved in the Northern hemisphere sea louse Lepeophtheirus salmonis, could have an influence on the response of these parasites to treatment with macrocyclic lactones. They could serve as molecular markers to assess susceptibility to existing treatments and might be useful molecular targets in the search for novel antiparasitic drugs.

rogercresseyi GluClsubunits, which are highly conserved in the Northern hemisphere sea louse Lepeophtheirus salmonis, could have an influence on the response of these parasites to treatment with macrocyclic lactones. They could serve as molecular markers to assess susceptibility to existing treatments and might be useful molecular targets in the search for novel antiparasitic drugs.

Introduction
Sea lice are ectoparasite copepods of the Caligidae family (order Siphonostomatoida). These small crustaceans attach to fish and feed on their epidermal tissue and blood and in a fish farming setting lead to morbidity and mortality with extremely high economic impact in the industry. Caligus rogercresseyi [1] is the most important parasite in salmonid farming in Chile causing increased costs and decreased productivity with high economic and social impact in the nationally important aquaculture industry [2][3][4]. Closely related Lepeophtheirus salmonis is responsible for high morbidity and mortality in Northern hemisphere salmon and trout aquaculture [5].
Various chemical treatments have been used to counter sea lice infestation of fish farms [2], among them are macrocyclic lactone avermectins emamectin and ivermectin widely used as antiparasitic drugs in human and veterinary medicine [6][7][8]. The importance of ivermectin has been widespread for more than 35 years, with its discoverers being distinguished with the Nobel Prize in Physiology or Medicine in 2015 [9]. In Chile, emamectin benzoate was administered orally with fish feed to provide long-lasting protection against all forms of attached sea lice, but the success of avermectins as antiparasitic treatment has been threatened by the emergence of resistance that has become a problem worldwide [10][11][12][13].
Inhibitory glutamate-gated chloride channels (GluCls) are Cys-loop ligand-gated ion channels (LGICs) made of five subunits whose association can form homomeric or heteromeric receptors [14,15]. Amino acids in transmembrane domain M2 and in the M1-M2 linker sequences. Three of these (Group A, S1 Fig) correspond to partial fragments of the already identified and characterised subunit [34] that we now rename CrGluCl-A, and one belonged to a glycine receptor and was therefore discarded from further analysis. Four transcripts (Group B, S1 Fig) exhibited overlapping with each other and their assemblage suggested their coding for a full GluCl peptide. The putative protein thus encoded was termed CrGluCl-B. The remaining 11 transcripts were assigned by BLAST analysis against transcripts in the data base for L. salmonis (https://licebase.org/) and genomic information for C. rogercresseyi at the https://ww.ncbi.nlm.nih.gov/. With this approach they could be classified in three putative separate subunits that we termed CrGluCl-C, CrGluCl-D and CrGluCl-E (Groups C-E, S1 Fig).
Full cDNAs were generated by RT-PCR from adult C. rogercresseyi specimens from which full-length putative peptide sequences of CrGluCl-B, CrGluCl-C, CrGluCl-D and CrGluCl-E were obtained. These sequences are shown together with that for CrGluCl-A in the multiple alignment of Fig 1. The predicted amino acid sequences for CrGluCl-A, -B, -C, -D and -E (Fig 1) all showed the general features characteristic of previously identified glutamate-gated chloride channel subunits. They all have putative signal peptides consistent with a plasma membrane protein expression. A large conserved N-terminal domain exhibited cysteine residues necessary to the formation of pLGIC signature Cys-loops. These were followed by four predicted transmembrane domains (M1-M4) and a shorter C-terminus. Amino acids identified to participate in the binding sites for glutamate and ivermectin [17] are largely conserved in the CrGluCls. Table 1 shows that identities and similarities between the CrGluCls are homogeneously around 50% and 63% respectively. These subunits most likely arise from separate genes.
Sea louse L. salmonis sequences present in the licebase (https://licebase.org/) and NCBI (https://www.ncbi.nlm. nih.gov/) databases were obtained using the CrGluCls identified here. Indeed, and as seen in S2 Fig, it is possible to identify virtually complete contigs for L. salmonis GluCls, LsGluCl, homologous to CrGluCl-B and -D, while contigs for putative LsGluCls -A, -C and -E can be generated on the basis of overlapping sequences.
A phylogenetic analysis of these putative C. rogercresseyi GluCl subunits was made including presently obtained GluCls from L. salmonis, and GluCls from other arthropods and chosen nematodes for which functional expression has been documented (Fig 2). Receptor subunits from nematodes and arthropods share separate ancestors. Within the arthropod subunits, those belonging to crustaceans C. rogergresseyi and L. salmonis made up a clade separate from those belonging to insects and the arachnid T. urticae. Each of the C. rogercresseyi subunits is orthologous to the L. salmonis gene products reported in S2 Fig This conclusion is supported by branch reliability estimates using bootstrapping.

Functional expression of Caligus rogercresseyi putative glutamate-gated chloride channels
Evaluation of the functional properties of C. rogercresseyi GluCl subunits was carried out after expression in Xenopus laevis oocytes by microinjection of cRNA derived from cDNA of variants CrGluCl-A, -B, -C, -D and -E [34]. Server) are shown in red italics with the cut portions shaded in gray. Cysteine residues necessary to the formation of Cloops are highlighted in yellow; in grey and in red are residues involved in glutamate and ivermectin binding respectively. The position of the transmembrane domains and binding sites, as well as the residues involved in glutamate and ivermectin binding, are based on the crystal structure of the C. elegans GluClα [17]. Highlighted in red and below the sequences are amino acids that participate in ivermectin binding in C. elegans GluClα but are not conserved in the CrGluCl subunits. The sequences for CrGluCl-A, B, C, D, and E are deposited in the GenBank under identifiers KX599189, OP737386, OP737387, OP737388 and OP737389 respectively. Alignments generated using Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/). https://doi.org/10.1371/journal.ppat.1011188.g001

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi subunit behaves as a glutamate-gated chloride channel (Fig 3). The graph shows current traces obtained using two-electrode voltage-clamp of an oocyte previously injected with CrGluCl-B cRNA. Outward and inward currents were recorded respectively at 60 and -80 mV, which correspond to chloride flowing into and out of the oocyte respectively ( Fig 3A). Superfusion with increasing concentrations of L-glutamate led to a graded increase in current. Much as reported for CrGluCl-A [34], there is no desensitization of the response of CrGluCl-B to glutamate. In contrast with the glutamate response, ivermectin was without effect on the currents revealing a complete refractoriness to activation of CrGluCl-B by the antiparasitic drug. Fig 3B shows the dose dependence of the glutamate effect on CrGluCl-B measured at 60 mV. The average relationship, reported as circles in the figure, can be described by a Hill equation. Separate Hill equation fits to these experiments gave an EC 50 of 388 ± 203 μM and n H of 1.7 ± 0.1 (Mean ± SD, n = 16). The result of similar experiments using CrGluCl-A are also shown as downward pointing triangles. These summarise individual Hill equation fits giving an EC 50 of 7.2 ± 2.7 μM and n H of 1.65 ± 0.63 (Mean ± SD, n = 5). Previously published values for glutamate CrGluCl-A EC 50 and n H were 6.9 μM and 1.33 respectively [34].
Although all CrGluCl channels have roughly similar selectivity filters, some divergences occur at M2 that might alter ion selection. Divergences are present in CrGluCl-A at positions 2´, 17´and 18´; CrGluCl-B at positions 5´, 10´, 14´and 15´; CrGluCl-C at position 19´; CrGluCl-D at positions 16´and 17´; CrGluCl-E at positions 4´, 10´, 14´and 16´). In addition, changes of the structure elsewhere from M2 might also affect the conformation of the selectivity filter and its interaction with the permeant ion. Picrotoxin, on the other hand, is a selectivity filter blocker and these divergences in M2 might also affect its potency. We have therefore checked both selectivity and picrotoxin effect for the different CrGluCls. Fig 3C shows that the glutamate-dependent current was carried by chloride, as partial extracellular replacement Clwith a more impermeant anion sharply reduced outward current (Clinflux). This is corroborated by data from the current-voltage (IV) relation taken during glutamate application under normal and low extracellular chloride (Fig 3D). Data are taken from voltage-ramps applied during the experiment in panel C. The graph shows a current-voltage relation before addition of glutamate (No Glut). This small current reversed at a potential, E rev , of -50 mV. In the presence of glutamate (Glut), an increased current is seen to occur at all voltages tested with the current vs. voltage relationship nearly linear. In low Clextracellular solution E rev became depolarized and the IV relation appeared slightly inwardly rectified, as expected for a current carried by chloride. On average E rev for the glutamate-activated current

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi rogercresseyi GluCl subunits (underlined in red) compared to other arthropods and nematodes CrGluCl counterparts. The percentage of trees in which the associated taxa clustered together is shown next to the branches (100 bootstrap replicates). The tree is drawn to scale, with branch lengths measured in the number of substitutions per site (scale bar). The analysis involved 39 amino acid sequences. The GenBank accession numbers of the amino acid sequences are given in the Methods section. Evolutionary analyses were conducted in MEGA11 [55]. The GluCl subunits of the platyhelminth Schistosoma mansoni (SmaGluCl-1, SmaGluCl-2, SmaGluCl-3 and SmaGluCl-4) were used as control since they belong to an independent GluCl clade evolutionarily distinct from arthropod and nematode counterparts [43]. https://doi.org/10.1371/journal.ppat.1011188.g002

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi shifted from -25 ± 6.2, close to the chloride equilibrium potential of Xenopus oocytes [30], to 8 ± 8.2 mV in low chloride, giving a calculated P Gluconate /P Cl value of 0.2 ± 0.06 (means ± SD, n = 6). Fig 3E shows that 100 μM PTX inhibits the currents elicited by 1 mM glutamate. Average inhibition was 76 ± 9% (mean ± SD, n = 4). Ivermectin and other compounds of the avermectin family of macrocyclic lactones are potent and irreversible activators of CrGluCl-A and some other ionotropic invertebrate receptors. However, as shown above ivermectin fails to increase the activity of CrGluCl-B. Although avermectins do not produce the expected activation of CrGluCl-B, they act as antagonists of this subunit as illustrated in Fig 4A and 4B. In A a first application of the agonist glutamate elicits a robust and reversible activation of CrGluCl-B while subsequently added ivermectin is without effect. A second, post-ivermectin stimulation with glutamate however fails to elicit a response. There is no attenuation of the glutamate response upon a second addition of the agonist without intervening treatment with ivermectin, but the drug is able to abolish activity in the continued presence of glutamate ( Fig 4B). This antagonistic effect is also observed when using emamectin, a different avermectin (S3A and S3B Fig). Increasing concentrations of ivermectin were used to gauge the potency of its antagonism of the glutamate-elicited current of CrGluCl-B. Fig 4C shows CrGluCl-B currents recorded in the continuous presence of 3 mM glutamate as affected by ivermectin that elicits a graded inhibition. This is quantified in the dose response relation in Fig 4D. The data can be described by a Hill equation. Separate Hill equation fits to individual measurements gave an IC 50 of 342 ± 114 nM and n H of -2.2 ± 0.5 (mean ± SD, n = 7).

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi CrGluCl-C Fig 5 shows the results of the functional assay of CrGluCl-C. Contrasting with the results with variants A and B, CrGluCl-C activation by glutamate results in a transient response ( Fig 5A) at all concentrations tested. The concentration dependence of the peak response to glutamate is given in Fig 5B where it is compared with those of the A and B subunits. The average relationship (mean ± SD, n = 6), reported by upright triangles in the graph, can be described by a Hill equation. Separate fits to these experiments gave an EC 50 of 67 ± 25 μM and n H of 2.5 ± 0.56 (Mean ± SD, n = 6).
CrGluCl-C was only partially inhibited by PTX at 100 μM (23 ± 10% inhibition, mean ± SD, n = 9, Fig 6E). The same concentration of PTX elicits a 78 ± 2% inhibition (mean ± SD, n = 4) of CrGluCl-A (see Fig 6E for a dose-response curve taken from reference 34) and a 76 ± 9% inhibition of CrGluCl-B (mean ± SD, n = 4, see Fig 3E). Anion replacement experiments (Fig 5D and 5E) show that the current through CrGluCl-C is carried by chloride ions. Glutamate-activated outward current shown in Fig 5D was markedly decreased in a low chloride solution. Current voltage relations taken from voltage-ramps applied during this experiment are shown in Fig 5E. There was little rectification for the glutamate-induced current that has an E rev at~-32 mV. Reduction in extracellular chloride shifts E rev in a depolarized

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi direction and the IV relation acquires some inward rectification, consistent with anion permeation. In separate experiments the observed E rev shift was from -25 ± 6.1 to 8 ± 8.2 mV giving a calculated P Gluconate /P Cl value of 0.1 ± 0.09. Spontaneous membrane potential in the absence of agonist was -51 ± 11 (all data means ± SD, n = 5).
CrGluCl-C is sensitive to ivermectin as seen in Fig 6A that shows a recording of CrGluCl-C current at 60 mV and where addition of 0.3 mM glutamate produced the expected transient increase in outward current. After glutamate removal, ivermectin at 100 nM elicited a stable increase in outward current that could be partially inhibited by PTX at 100 μM. That the current is carried by chloride is shown by its decrease by partial removal of the anion from the extracellular solution. Current voltage relations taken from voltage-ramps applied during this experiment are shown in Fig 6B. These gave similar results as for glutamate-activated currents, with little rectification for the ivermectin-induced current which is shifted in the depolarized direction from IV taken before addition. A further shift is seen after low chloride solution perfusion. Respective E rev values were -55, -41 and 10 mV. In the presence of ivermectin the membrane potential shifted from -25 ± 8.9 mV in normal chloride concentration to 18 ± 11 mV in the low chloride solution, giving a calculated P Gluconate /P Cl value of 0.13 ± 0.08 (means ± SD, n = 5).
Ivermectin causes graded activation in anion current with increasing concentration and the effect of the drug is irreversible (Fig 6C). The circles in the plot in Fig 6D show the average

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi response of CrGluCl-C to ivermectin and a Hill equation fit to the data. Separate fits to data yielded an EC 50 value of 31.1 ± 14.4 nM and n H of 2.1 ± 0.67 (means ± SD, n = 7). These compare with values of 181 nM for EC 50 and n H of 2.1 previously reported for the CrGluCl-A subunit [34]. Maximal current reached at saturating ivermectin concentration was 3.7 ± 1.3 μA (n = 10), which compares with 3.5 ± 1.7 μA for the maximal current elicited by glutamate.
As seen above the sensitivity of CrGluCl-C to PTX appears lower than those of subunits A and B. This was explored further in the PTX inhibitory dose-response in Fig 6E. PTX inhibited ivermectin-induced CrGluCl-C current with an IC 50 of 275±96 μM (n = 7) and n H -1.2±0.22 (n = 7). PTX potency for CrGluCl inhibition is therefore two orders of magnitude lower than A. Current trace obtained at 60 mV during bath application of 300 μM glutamate and then ivermectin at 100 nM during the times shown in the boxes. The effect of PTX addition at 100 μM and the reduction in extracellular chloride concentration are also shown. B. Current voltage relations taken from voltage-ramps applied during the experiment in A before and after ivermectin addition as well as after superfusion with the low chloride solution. C. Effect of increasing concentrations of ivermectin on CrGluCl-C and the effect of partial external chloride removal. D. Effect of increasing concentrations of ivermectin on CrGluCl-C reported as average dose dependence of the response. Data (mean ± SD, n = 7) are normalized to the maximal effect of ivermectin. The line is a fit of the Hill equation to the values and it is compared to that measured for CrGluCl-A previously [34]. E. Effect of increasing concentrations of picrotoxin on CrGluCl-C current elicited by 1 μM ivermectin. Data (means ± SD, n = 9) are normalized to ivermectin-dependent current value before PTX addition. The line is a fit of the Hill equation to the values and it is compared to that measured for CrGluCl-A previously [34]. https://doi.org/10.1371/journal.ppat.1011188.g006

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi for the A subunit measured previously at IC 50 of 3.2 μM ( [34], fit also shown in the graph for comparison).

CrGluCl-D
Expression of subunit CrGluCl-D gives rise to large spontaneous currents ( Fig 7A) that amounted to 6.4 ± 0.76 μA (mean ± SD, n = 12) at 60 mV and could be blocked by the ligandgated anion channel inhibitor PTX. E rev of the spontaneous current was -25 mV (control in 7B), close to E Cl . Fig 7C shows that spontaneous outward current is markedly decreased upon lowering extracellular chloride and that although glutamate or ivermectin lack a direct activating effect they modulate PTX-sensitivity of the spontaneous current ( Fig 7C). Fig 7D shows IV curves taken from trace in C, and represent initial control spontaneous current with no addition and in the presence of PTX or low extracellular Cl -. Partial removal of extracellular chloride displaced E rev from -21 ± 10.2 to 14.5 ± 14.8 mV, giving a P Gluconate /P Cl permeability ratio of 0.21 ± 0.12 (means ± SD, n = 4). The sensitivity to PTX of CrGluCl-D was explored both in naïve oocytes expressing CrGluCl-D or in those exposed previously to 1 μM ivermectin. Fig 7E shows the respective dose response data. The IC 50 for PTX inhibition before ivermectin treatment was 16±8.8 μM while that after ivermectin increased to 221±100 μM (means ± SD, n = 11 and 12 respectively). The spontaneous membrane potential near E Cl of CrGluCl-Dexpressing oocytes, the inhibition of their spontaneous currents by PTX, and the changed

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi sensitivity to PTX induced by the GluCl agonist ivermectin treatment, all support the hypothesis that the observed activity corresponds to a permanently active GluCl receptor.

CrGluCl-E
Functional assay of CrGluCl-E gave low spontaneous currents and failed to show any response to glutamate up to 1mM or to 1μM ivermectin. We presume that CrGluCl-E is an inactive form of glutamate receptor or that the protein fails to reach the oocyte plasma membrane (but see below).

Functional properties of coexpressed CrGluCl subunits
Invertebrate glutamate receptors belong to the Cys loop ligand-gated ion channel (LGIC) group of proteins that can assemble as homopentamers or heteromers of different stoichiometries. Most of the GluCl subunits successfully expressed in vitro have been homomeric [18] but there are instances of heteromeric assembly in coexpression experiments in vitro that give rise to activities with novel properties [35][36][37][38]. The subunit composition of native GluCls is not known in species with genes coding for multiple subunits. We undertook to explore whether the different subunits of C. rogercresseyi GluCl identified here might form heteromers with properties distinct from those of the individual subunits when coinjected in Xenopus oocytes.

CrGluCl-A and CrGluCl-B coexpressions
Coinjection of cRNA for subunits A and B yields activities with characteristics at variance with those of either of the subunits expressed separately. Fig 8A shows a recording of the currents arising after microinjecting Xenopus oocytes with a 4:1 CrGluCl-A:CrGluCl-B mixture of cRNAs. These oocytes responded to glutamate with prompt development of robust currents similar in kinetics to those elicited after expression of the separate subunits. There was only a small response to ivermectin and further glutamate addition evoked a little increase in current, reminiscent to what is seen when expressing CrGluCl-B on its own. The IV relationship of the current induced by glutamate was ohmic and had an E rev of -22 mV, a shift from the -52 mV E rev of the preaddition, control IV ( Fig 8B). Separate analysis of IV relations of glutamate-activated currents in oocytes expressing a 4:1 CrGluCl-A:CrGluCl-B mixture after partial chloride replacement gave a P Gluconate /P Cl permeability ratio of 0.21 ± 0.12 (mean ± SD, n = 3).
Similar results were obtained after expression of 1:1 and 1:4 CrGluCl-A:CrGluCl-B mixtures of cRNAs. A summary of the responses to 3 mM glutamate and 1 μM ivermectin of the three A:B mixtures tested is given in Fig 8C together with the responses of the oocytes injected singly with either CrGluCl-A or CrGluCl-B. All three mixtures presented robust responses to glutamate but those to ivermectin were quite small (notice axis brake and change of scale in the ordinate of Fig 8C). Compared to the glutamate response of CrGluCl-B those of CrGluCl-A and the 4:1 mixed expression are significantly smaller (P<0.001 by ANOVA analysis). The response to ivermectin of all, 4:1, 1:1 and 1:4 mixed A:B-expresing oocytes, did not differ significantly from the (lack of) response to ivermectin of CrGluCl-B, but all three though small, were nevertheless different from zero (P = 0.003, 0.023 and 0.004 respectively as analysed in one sample t-tests).
The EC 50 values of CrGluCl-A and -B differ significantly. That for CrGluCl-A, at 7 μM, is amongst the lowest reported. Those for helminths and insects CeGluClβ, DmGluClα, HcGluClα and AgGluCl are 380, 23, 28 and 30 μM respectively [36,39,40]. The CrGluCl-B EC 50 on the other hand is much higher at around 400 μM. Mixed expression of CrGluCl-A and CrGluCl-B glutamate dose-response (shown in Fig 8D together

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi

CrGluCl-C in coexpression with CrGluCl-B or CrGluCl-A
Fig 9A shows a current recording from an oocyte injected with a CrGluCl-C:CrGluCl-B 1:1 mixture. Similarly to the CrGluCl-A:CrGluCl-B coexpression, the 1:1 mixture of CrGluCl-C and CrGluCl-B gave rise to a functional receptor with novel properties. Indeed, 3 mM glutamate elicited robust current that, remarkably, lacked the marked desensitization seen with subunit C alone (Fig 5). In addition, 1 μM ivermectin failed to elicit a significant response. As for CrGluCl-B alone or in coexpression with subunit A, the glutamate response was persistently antagonized by previous ivermectin treatment. The IV curve of the glutamate-dependent current was linear and reverted at -22 mV compared with a resting E rev of -52 mV (Fig 9B). The ivermectin-induced IV relation did not differ markedly from the control IV curve. Fig 9C  compares the concentration dependence of the glutamate response of coexpressed CrGluCl-C and CrGluCl-B together with the response of the individual subunits. A Hill fit to the response of the combined subunits yielded an EC 50 of 53 ± 18.8 μM and n H of 1.9 ± 0.6 (means ± SD, n = 13), an increase in sensitivity in comparison to that of CrGluCl-B but similar to that of CrGluCl-C assayed separately. Concerning ivermectin activation, CrGluCl-C and CrGluCl-B coexpressed receptor was poorly sensitive in contrast the high sensitivity of CrGluCl-C and more in line with the insensitivity of the CrGluCl-B subunit. The responses to 3 mM glutamate and 1 μM ivermectin are compared in Fig 9D. CrGluCl-C and CrGluCl-B on their own had glutamate-evoked currents that did not differ significantly from one another, whilst the current of coexpressed subunits was significantly smaller than that of CrGluCl-B (P<0.05 as tested by ANOVA). While CrGluCl-C is activated by ivermectin (P<0.05 by ANOVA compared with CrGluCl-B), there was no difference between the coexpressed C and B subunits and CrGluCl-B on its own. The small response of the coexpressed CrGluCl-C/CrGluCl-B was nevertheless significantly different from zero (P<0.001 in a one sample t-test).
Oocytes microinjected with CrGluCl-C and CrGluCl-A responded to glutamate developing currents exhibiting steady as well as transient components (Fig 9E), suggestive of a sum of the activities of the individual subunits acting separately. The experiment also shows the irreversible response to ivermectin expected for either GluCl individually expressed as well as the sensitivity to PTX. A summary of the magnitude of the responses to 3 mM glutamate and 1μM ivermectin is shown in Fig 9F. IV relations in Fig 9G show that after activation by glutamate the current at the plateau of the response is outwardly rectified as that for CrGluCl-A on its own, whilst the behaviour turns to ohmic when the peak response is plotted, reminiscent of that of CrGluCl-C expressed alone. Fig 9H shows an experiment where the dose-response of the effect of glutamate on currents at 60 and -80 mV of the coexpressed CrGluCl-A and -C was assayed. Glutamate produced a graded effect, inducing currents with increasing desensitisation as the concentration increased, an effect that was more marked at the negative potential at which the putative contribution of CrGluCl-C subunits should predominate. Fig 9I shows dose-response curves taken at 60 and -80 mV to which Hill equations were fitted. At 60 mV a single Hill equation sufficed to describe the data adequately with an EC 50 of 7.9 ± 0.9 μM (Mean ± SD, n = 4). The data at -80 mV were better described by double Hill equations that yielded EC 50 values of 7.5 ± 2.0 and 79 ± 16.7 μM (Mean ± SD, n = 4), approaching those of the -A and -C subunits expressed on their own (7 and 67 μM, see also Table 2). These results suggest that after microinjection of subunits A and C cRNAs the corresponding expressed receptors function independently.

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi CrGluCl-D in coinjection with CrGluCl-B, CrGluCl-C or CrGluCl-A As reported above (Fig 7), expression of CrGluCl-D in oocytes gives rise to large spontaneous currents lacking sensitivity to glutamate or ivermectin but inhibited by PTX. When coinjected with CrGluCl-B in a 1:1 ratio, however, there was only a residual current consistent with the background conductance of uninjected oocytes. Oocytes coinjected with CrGluCl-D and CrGluCl-B cRNAs responded to glutamate or ivermectin addition with prompt increases in PTX-sensitive current (Fig 10A). Fig 10B shows

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi currents were linear and reversed at potentials depolarized with respect to preaddition control IV ( Fig 10C). The concentration-dependence of the glutamate effect of the coexpressed subunits ( Fig 10D) has an EC 50 of 36.5 ± 0.08 μM and n H of 0.63 ± 0.11 (mean ± SD, n = 6). Concentration-dependence of the ivermectin effect on the coexpressed receptors ( Fig 10E) has an EC 50 of 278 ± 53 nM and n H value of 2.6 ± 1.1 (mean ± SD, n = 6).
In contrast with the evidence for the formation of mixed GluCl subunits when coinjecting CrGluCl-D and CrGluCl-B cRNAs, it appears this is not the case when CrGluCl-D cRNA is microinjected together with that of CrGluCl-C. Indeed Fig 11A shows that oocytes expressing CrGluCl-D and -C exhibit the large PTX-inhibitable spontaneous currents expected for the -D subunit expressed on its own. Addition of glutamate or ivermectin elicited additional current, with that responding to glutamate exhibiting a transient component possibly mediated by CrGluCl-C subunits functioning independently. Fig 11B shows the magnitude of both the spontaneous, chloride-dependent current and the increases due to glutamate or ivermectin addition. IV curves in Fig 11C approach ohmic behaviour in the absence of stimulation and after glutamate or ivermectin challenge. The depolarization upon partial extracellular chloride removal, from -32 to 12 mV, shows the anion as the charge carrier. The mean ± SD of P Gluconate /P Cl permeability ratio deduced from these shifts in E rev was 0.15 ± 0.04 (n = 4).

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi Oocytes coinjected with CrGluCl-D and CrGluCl-A exhibit sizable chloride-dependent spontaneous currents similar to those of CrGluCl-D alone (Fig 12A). Fig 12B (left-hand  panel) compares the magnitude of spontaneous currents of oocytes microinjected with cRNA for CrGluCl-D with those with D and A CrGluCl subunits. Unlike what is seen with singly expressed CrGluCl-D, glutamate or ivermectin increased the basal spontaneous current (Fig 12B, right-hand panel). IV curve for spontaneous current shown in Fig 12C approaches ohmic behaviour as do those seen after glutamate or ivermectin simulation. Depolarization in low chloride solution, from -27 to 16 mV, shows the anion as the charge carrier. The mean ± SD of P Gluconate /P Cl permeability ratio deduced from these shifts in E rev was 0.19 ± 0.03 (n = 3).

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi

CrGluCl-E in coinjection experiments with CrGluCl-A, CrGluCl-B or CrGluCl-C
CrGluCl-E appears not to elicit any glutamate receptor activity when expressed in Xenopus oocytes. We tested whether coinjection with CrGluCl-A, CrGluCl-B or CrGluCl-C might promote activity with properties different from those seen of the subunits expressed singly.
Additional expression of CrGluCl-E does not appear to affect CrGluCl-A activity. Oocytes microintected with cRNA for CrGluCl-E and CrGluCl-A responded to glutamate and ivermectin (Fig 13A and 13C) much as expected from singly expressed CrGluCl-A. The glutamate response (Fig 13B) had an EC 50 value of 5.0 ± 0.5 μM and n H of 2.5 ± 0.3 (means ± SDs, n = 4), a sensitivity that is close to that of CrGluCl-A expressed on its own [34]. Similar to CrGluCl-A expressed alone, the IV relations of these mixed subunit experiments were markedly outwardly rectified (Fig 13D).
Additional expression of CrGluCl-E did not affect CrGluCl-B activity markedly. Superfusion with increasing concentrations of glutamate of an oocyte expressing CrGluCl-E and CrGluCl-B in a 1:1 ratio led to a graded increase in current with no evidence for

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi desensitization ( Fig 14A). Reduction in extracellular chloride caused a marked decrease in outward current while addition of channel blocker PTX at 100 μM strongly inhibited both inward and outward currents. In contrast with the glutamate response, ivermectin was without any activating effect, nevertheless a subsequent challenge with 1 mM glutamate was markedly antagonised. Fig 14B shows the dose dependence of the glutamate effect on the 1:1 mixed subunits measured at 60 mV. The relationship can be described by a Hill equation and the average response had an EC 50 value of 534 ± 239 μM and n H of 1.51 ± 0.08 (means ± SDs, n = 6), which compares with respective values of 388 ± 203 μM and 1.7 ± 0.1 (means ± SD, n = 16) for subunit B on its own. The EC 50 values did not differ (P = 0.08 by t-test), but the difference between n H values did reach statistical significance (P<0.05) Assaying glutamate response after coinjection of cRNA for CrGluCl-E with that of the CrGluCl-C subunit gave rise to dose-dependent robust currents that unlike those of CrGluCl-C on its own showed little desensitization (Fig 14C). The concentration dependence of the response to glutamate (Fig 14D) followed a Hill equation with EC 50 of 36.5 ± 22.5 μM and n H of 1.4 ± 0.4 (means ± SDs, n = 10). Fig 14C also shows that after glutamate removal, addition

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi of 1 μM ivermectin elicited an irreversible increase in currents that, very much as is the case with CrGluCl-C, could only be partially inhibited by 100 μM PTX. That chloride is the charge carrier is witnessed by the decrease in outward current after extracellular Clremoval from the extracellular solution.

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi Ivermectin activation of the anion current in oocytes microinjected with CrGluCl-E and CrGluCl-C cRNA is graded and carried by chloride (Fig 14E). The average response to ivermectin ( Fig 14F) yielded EC 50 and n H values of 37.4 ± 13.8 nM and 2.1 ± 0.67 (means ± SD, n = 5), very close to those for CrGluCl-C assayed on its own (31.1 nM and 2.1 respectively).

Structural basis for the lack of ivermectin activation of CrGluCl-B
Analysis of sequence differences between CrGluCl-B, refractory to ivermectin activation, and the ivermectin-sensitive subunit CrGluCl-A was focused on transmembrane segments M1-M3 that harbour points of contact of the drug in the CeGluClα structure [15]. Fig 15A shows sequence alignments of CrGluCl-A and -B with respectively ivermectin-sensitive and -insensitive α and β GluCl subunits of C. elegans and H. contortus. Six residues, highlighted in magenta in Fig 15A, differ between the C. rogercresseyi subunits. Three amino acids located in transmembrane domains M1, 2 and 3 respectively have been implicated in ivermectin binding to CrGluCl-A: L263, T305 and T315 (identified in red in Fig 15A [34]). L263 is conserved in CrGluCl-B and two other residues, I294 and I304, experience conservative changes to L in subunit A, and were therefore not considered further. In contrast, T305 and T315 are the nonconserved Q327 and M340 in the CrGluCl-B subunit. Further non-conservative differences in CrGluCl-B are Q279 and Y361. These residues were mutagenized in CrGluCl-B to those present in subunit -A to ascertain their role in ivermectin sensitivity.
The response to glutamate of the mutants (Fig 15B) was not significantly different from that of the WT receptor, except for Y361F and double Q279E-Y361F mutants, (ANOVA P<0.001). Ivermectin, on the other hand, had no effect on WT or M340T mutant, but produced small but significant effects on Q279E, Y361F and a double mutant carrying both modifications (one-sample t-tests). The effect of ivermectin on the singly Q327T mutated subunit, as well as that on the double and triple mutant were all significantly greater than those of the WT GluCl (ANOVA P<0.05). Currents elicited by 3 mM glutamate or 1 μM ivermectin in an oocyte expressing CrGluCl-B-Q279E-Y361F-Q327T triple mutant taken at 60 and -80 mV are shown in Fig 15C. The response to ivermectin of this triple mutant is irreversible and the activated current is sensitive to receptor blocker PTX. Dose response experiments (Fig 15D) confirm the ivermectin sensitivity of the triple mutated CrGluCl-B. Hill equations fitted to the individual data sets gave parameters of EC 50 94 ± 54 nM and n H 1.6 ± 0.22 (means ± SD), intermediate between those of the -C and -A subunits. Glutamate response occurs with significantly higher sensitivity (Fig 15E). Hill equation analysis of the data yielded average parameters of EC 50 103 ± 133 μM and n H 1.6 ± 0.22 (means ± SD).

Discussion
Avermectins such as ivermectin have been highly successful antiparasitic agents used widely in human and veterinary medicine. Ivermectin has been of central importance in the strategy to eliminate onchocerciasis, lymphatic filariasis, and strongyloidiasis (reviewed in [41]). Ivermectin and other avermectins are also important in the context of parasitic diseases affecting livestock and agricultural productivity [9]. In fish aquaculture, sea lice infestation was initially successfully controlled using avermectins, but more recently resistance to the drugs has developed by mechanisms that remain unclear [42]. The target of avermectins are GluCl receptors that are irreversibly activated by the drug thus conducing to parasite paralysis and death. The possibility that mutation of these ligand-gated chloride channels is responsible for resistance has been considered. Alternatively, the expression of naturally existing avermectin activationresistant receptor subunits could also be responsible for parasite evasion of the pesticidal agents. Exploration of these possibilities requires a so far lacking detailed knowledge of the

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi GluCls present in sea lice. In this report we identify and functionally characterise four novel GluCls subunits in Caligus rogercresseyi that together with a previously reported subunit [34] constitute a family of five receptor subunits encoded by different genes. Functional assays in Xenopus oocytes show that of the five glutamate-gated chloride receptors discussed three, CrGluCl-A, CrGluCl-B and CrGluCl-C, present bona fide receptor behaviour when expressed on their own. Expression of CrGluCl-D leads to constitutive chloride current consistent with spontaneously active GluCl channels, whilst CrGluCl-E did not present any activity upon stimulation with glutamate or ivermectin. Details of these results are summarised in Table 2 to facilitate comparisons between subunits.
CrGluCl-A and CrGluCl-C are activated by ivermectin, but CrGluCl-B is completely insensitive to activation by this antiparasitic macrocyclic lactone. Despite the failure to activate CrGluCl-B, both ivermectin and emamectin antagonise the effect of glutamate such that the response to the agonist is decreased after treatment with the drug. This phenomenon was observed using CeGluClβ (also called CeGLC-2) subunits [50]. The inhibition of glutamate activated current in CeGLC-2 occurs with IC 50 and n H values 1.28 μM and -1.74 [51], which compares with figures of 0.34 μM and -2.2 measured here for CrGluCl-B. Notice that we assume complete reversibility of PTX action. This is verified in the case of the subsequent glutamate response (see e.g. Fig 4A and 4B), but it has not been formally demonstrated for the avermectin effect.
Concerning activation of GluCls by avermectins, whilst CeGluClβ does not activate in response to ivermectin and other macrocylclic lactones, CeGluClα is activated by the drug [36,51]. HcGluClα and HcGluClβ are respectively sensitive and insensitive to ivermectin [37]. Coexpression of the α and β subunits of both C. elegans and H. contortus GluCls leads to heteromeric channels activated by both glutamate and ivermectin, but while the inclusion of the β subunit does not change the potency of ivermectin in C. elegans coexpression, the potency and efficacy of ivermectin in HcGluCl-α/β coexpression are diminished [37]. Remarkably however, coexpression of CrGluCl-B with either CrGluCl-A or CrGluCl-C gives rise to channels that, in addition to being more sensitive to glutamate than CrGluCl-B expressed singly, are not activated by ivermectin (Table 2). These results show that these subunits can form heteromeric assemblies with emerging new properties such as intermediate glutamate sensitivities but, importantly, they show that CrGluCl-B can confer these heteromers its refractoriness to activation by ivermectin. individual data sets. Fitted parameters were EC 50 94 ± 54 nM and n H 1.6 ± 0.22 (means ± SD). Curves describing ivermectin effect on CrGluCl-A and -C are also shown. E. Normalised response to glutamate of the triple mutant after fitting a Hill equation to the individual data sets is given as mean ± SD (n = 21). Average parameters were EC 50 103 ± 133 μM and n H 1.6 ± 0.22 (means ± SD). The fitted curve for CrGluCl-B dose response is also shown. https://doi.org/10.1371/journal.ppat.1011188.g015

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi The failure of the ivermectin activating effect is also present in experiments using a 1:1 expression ratio of CrGluCl-B and CrGluCl-A. If the assembly of these heteromers occurred randomly, a binomial expansion-based calculation would suggest only a 3% of homopentameric receptors implying that a single CrGluCl-B subunit would be necessary to render the pentamers non-activated by ivermectin. Surprisingly, a similar result is obtained when oocytes are injected with the subunit ratios of 1:4 and 4:1. This would imply a preferential assembly as complexes containing at least one CrGluCl-B subunit. This would be in line with results obtained using heteromeric C. elegans GluClα and β subunits which are shown to assemble with a fixed stoichiometry [14].
Two other subunits described here do not elicit regular CrGluCl receptor behaviour upon functional assay. CrGluCl-D generates what might be considered spontaneously active GluCl behaviour while no currents are observed when expressing CrGluCl-E, perhaps suggesting it does not reach the membrane surface. That they can both function as subunits of distinct heteromeric assemblies is suggested by coinjection experiments. Indeed, coexpressing CrGluCl-D together with CrGluCl-B leads to channels that are closed in the absence of either glutamate or ivermectin but that readily activate in their presence. Interestingly receptors arising from CrGluCl-D and CrGluCl-B coexpression, in contrast to subunit B on its own or in combination with other subunits, are sensitive to ivermectin and have high sensitivity to glutamate (Table 1). We have no explanation for this. Examination of predicted ivermectin binding sites of the subunits reveals high conservation along all five Caligus receptors. The only exception is present in CrGluCl-D M3 where Gly323 replaces a threonine in subunits B, C and E, and isoleucine in A, but we have not tested the functional consequence of this replacement.
CrGluCl-E shows no channel activity on its own and appears not to affect the behaviour of CrGluCl-B in coinjection experiments. Expressing together subunits E and C gives rise to channels responding to glutamate and ivermectin similarly to CrGluCl-C on its own. These heteromeric channels, however, differ in that the characteristic CrGluCl-C desensitising response to glutamate is absent when CrGluCl-E is expressed together with CrGluCl-C, in which case the resulting channel possesses sustained currents in the presence of the agonist. In contrast to the association of CrGluCl-A and -B, CrGluCl-A does not appear to give rise to new heteromeric assemblies with subunits C, D or E. We cannot discard, however, that heteromers between these subunits might form with very similar rectification properties, EC 50 and n H values to those of CrGluCl-A on its own. Future studies with ivermectin and other agonists might help to elucidate this point further.
Interestingly, a recent paper [38] describes a variety of GluCl subunits in the parasitic nematodes Brugia malayi and Parascaris univalens which are orthologues to C. elegans receptors AVR-14B, GLC-2, GLC-3 and GLC-4. Functional expression of these reveals no or very low activity as GluCls when expressed on their own. Robust glutamate-activated responses can be seen, however, in coexpression experiments. These data, together with our observations of conventional receptor behaviour of heteromers including subunits CrGluCl-D or CrGluCl-E, suggest that subunit integration into mixed pentameric assemblies might be of high relevance to the expression of otherwise inactive subunits as functional glutamate receptors.
An important challenge for future research on the GluCl receptors of C. rogercresseyi (and L. salmonis) is that of whether subunit assemblies formed in vitro by the recombinant proteins represent receptors that are present in the copepod. This uncertainty stems from near complete lack of knowledge of many aspects of the GluCl biology, including in which cells the GluCls are expressed, at what life stage, possible sexual dimorfism in expression of the channels, among others. Another potentially important point to consider is whether, as shown for the acetylcholine receptors of C. elegans [52], there are auxiliary subunits for the GluCls that might affect their heteromerisation, localisation or activity. Work based upon imunolocalisation has been exploited to demonstrate the coexpression of certain GluCl subunits both in C. elegans and the parasitic nematodes C. oncophora and H contortus, but these interesting new data fall short of demonstrating actual formation of heteromeric receptors (reviewed in [53]). Future research in these aspects of the CrGluCl family should help in our knowledge of the expression patterns of the subunits in the sea louse and might help in our understanding of the mechanisms of avermectin resistance.
Exploration of the possible molecular determinants of the absence of activation of CrGluCl-B by ivermectin identified amino acids located in transmembrane domains M1-M3 that diverged from ivermectin-activated subunits such as CrGluCl-A. Mutating CrGluCl-B residues Q279, Y361 and Q327 to their homologues present in CrGluCl-A conferred ivermectin-activation to this normally refractory subunit. By far the largest contribution to this changed ivermectin sensitivity resided in Q327 of the -B subunit, which is T in the CrGluCl-A subunit. This threonine residue at position 15' of M2 [54], has been identified in ivermectin docking simulations in CrGluCl-A as making H-bond contact with the cyclohexene ring of the drug [34]. Interestingly, recently Kaji at al. [51] have identified the homologous to Q15' of C. elegans GluClβ (S15' in the α subunit) as essential for the resistance to activation of this receptor to ivermectin. This residue is also conserved in ivermectin non-activable by H. contortus GluClβ [37], suggesting that perhaps a common mechanism underlies lack of receptor activation by ivermectin across species.
In the present paper we have explored the ivermectin effect on the CrGluCls, taking it as a "representative" molecule of the macrocyclic lactone family. We have shown that emamectin acts in a similar way to ivermectin on CrGluCl-A, where it appears to share identical binding site, and on CrGluCl-B. It will be interesting to address the issue of the sensitivity the CrGluCls to an extended number of macrocyclic lactones that are known to have a diversity of effects on this type of subunit.
In summary, we have discovered a family of five GluCl subunits in the salmon fish parasite copepod Caligus rogercresseyi. Of the five subunits identified, and dubbed CrGluCl-A to -E, CrGluCl-B is refractory to activation by ivermectin, an important antiparasitic macrocyclic lactone. CrGluCl-B is also able to confer this refractoriness to functionally distinct heteromeric glutamate assemblies generated by its coexpression with CrGluCl-A or CrGluCl-C. This might contribute to the resistance of this parasite to conventional treatments based on macrocyclic lactones. We suggest that these Caligus rogercresseyi GluCl subunits, which appear highly conserved in Lepeophtheirus salmonis, another parasite of importance in fish aquaculture, could serve as molecular markers to assess susceptibility to existing treatments and could become useful molecular targets in the search for novel antiparasitic drugs.

Ethics statement
All animal procedures were approved by the Centro de Estudios Científicos (CECs) Institutional Animal Care and Use Committee and were performed following "Guidance on the housing and care of the African clawed frog, Xenopus laevis" recommendations from Research Animals Department-The Royal Society for the Prevention of Cruelty to Animals (http:// www.rspca.org.uk/xenopus).

Identification and cloning of glutamate activated chloride channels
To identify putative GluCl subunits, we queried the NCBI Transcriptome Shotgun Assembly Sequence (TSA) Database for "glutamate-gated chloride channel Caligus". The obtained sequences were translated and BLASTed (tBLASTn) against the C. rogercresseyi genome

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi (WGS, BioProject ID 280098, NCBI) and against transcripts of L. salmonis database (https:// licebase.org/). Primers for each of the putative new subunit found were designed to span the full length of the open reading frame (ORF) for each of them (B-E) (S1 Table, and arrows in Fig 1). Complete cDNAs were generated by RT-PCR from adult C. rogercresseyi samples from which putative full-length peptide sequences of CrGluCl-B, CrGluCl-C, CrGluCl-D and CrGluCl-E were obtained. Each PCR product was cloned in pGEM-T easy vector and sequenced. To express the clones in Xenopus laevis oocytes, ORF of CrGluCl-B, -C, -D and -E were cloned in pGH19 vector from pGEM-Teasy/CrGluCl using the following restriction enzymes: BamHI and XbaI (-B) and EcoRI (-D and -E). CrGluCl-C was cloned first in pCR3.1 vector with NotI and then into pGH19 with SpeI/XbaI. The CrGluCl/pGH19 plasmid were linearized with NheI (-A and -B) or XhoI (-C, -D, -E) and used as template for the synthesis of capped cRNA (complementary RNA) using the mMessage Machine T7 kit [34]. Mutations were made by site-directed mutagenesis using the QuickChange method and confirmed by sequencing.

Phylogenetic analysis
The evolutionary relationship between CrGluCl subunits and GluCls from L. salmonis, other arthropods and chosen nematodes, was studied. Phylogenetic and molecular evolutionary analyses were conducted using MEGA version 11 [55]. The sequences were aligned using the MUSCLE algorithm and the phylogenetic tree was reconstructed using the maximum likelihood method with default parameters and 100 bootstrap replicates. Reliability for internal branch was assessed using the bootstrapping method. Graphical representation and edition of the phylogenetic tree were performed using

Xenopus oocyte preparation and cRNA injection
Adult female Xenopus laevis frogs were anaesthetized by immersion in a1g/L solution tricaine (Ethyl 3-aminobenzoate methanesulfonate, Sigma) buffered to pH 7 with NaHCO 3 . Surgery was performed 30-40 min later, once the frog was completely anaesthetized judging by loss of reflexes. Ovarian lobes were removed through 1 cm abdominal incision and then kept at 16˚C in Barth's modified solution containing (in mM): 88 NaCl, 1 KCl, 0.4 CaCl 2 , 0.3 Ca (NO 3 ) 2 , 0.8 MgSO 4 , 2.4 NaHCO 3 , 5 Hepes, 2.5 sodium pyruvate, 100 μg/ml gentamicin sulfate, 100 U/ ml penicillin, 100 μg/ml streptomycin, 250 μg/ml amphotericin B, 0.22μm filtered and adjusted to pH 7.5 with Tris-Base. Osmolarity was brought to 220 mOsm with saccharose. The oocytes were manually defolliculated on the same day of extraction and the following day 20 ng cRNA in nuclease-free water was injected into stage V-VI oocytes using a FemtoJet microinjector

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi (Eppendorf). Finally, the injected oocytes were stored in 96 well plates containing Barth's modified solution and kept at 16˚C until experiment day.

Voltage-clamp assay of receptor activity in Xenopus laevis oocytes
Current measurements at controlled membrane potentials (voltage-clamp experiments) were performed using two intracellular microelectrodes in defolliculated Xenopus oocytes previously microinjected with 20 ng of cRNA. Voltage-clamp recording was performed at room temperature using a TURBO TEC-10CX amplifier (npi electronic GmbH, Tamm, Germany) and PClamp 10.6 software (Axon Instruments, USA), 1 to 4 days after oocyte injection. The oocytes were recorded in a chamber (Model RC-1Z, Warner Instrumenst, USA) under continuous superfusion. The electrodes had a resistance of 0.5-2 MO when filled with 3 M KCl. Reference was an Ag-AgCl electrode connected to the bath via a 3 M KCl in 3% agar bridge. Currents were recorded continuously during the experiments without interruptions for solution changes. Oocytes were held at -30 mV and a voltage protocol consisting of a voltage drop to -100 mV for 25ms followed by a 360 ms ramp from -100 mV to 60mV applied with a period of 565 ms was continuously given. The signal filtered at 1 kHz was acquired with a Digidata 1440A Analog-to-Digital Converter and analyzed with Axon pClamp 10.6 software. The solution that bathed the oocytes during the electrophysiological recordings had the following composition (mM): 115 NaCl, 2 KCl, 1.8 CaCl 2 , 1 MgCl 2 , 10 HEPES pH 7.5 obtained with NaOH. For the low chloride solution, all the NaCl was replaced with the gluconate salt of Na. This decreases chloride concentration from 122.6 to 7.6 mM while increasing that of gluconate from 0 to 115 mM. The stock solutions of the drugs in dimethyl sulfoxide (DMSO) were: 10 mM emamectin, 10 mM ivermectin, 100 mM picrotoxin. These were diluted in bath solution to obtain final concentrations. The highest DMSO concentration used was found to have no effect on CrGluCl currents.

Data analysis
To obtain the Dose-Response curves, the currents measured at the 60mV potential for each added agonist concentration (glutamate or avermectin), were extracted and a 4-parameter Hill equation (Formula 1) fit using Sigmaplot 12.3 was performed. The currents were normalized according to the extrapolated maximum current delivered by the Hill adjustment. In this way, the concentration that generates 50% of the maximum effect (EC 50 ) and the Hill coefficient (n H ) for each situation were obtained.
The analyses were done using the Excel or Sigmaplot 12.3 programs. Statistical figures are given as averages with their standard deviation. Multiple comparisons were made by ANOVA with Bonferroni correction or by Kruskal-Wallis one way ANOVA on ranks. Comparisons between two samples were done by t-test. One sample t-test was used to ascertain whether effects were significantly greater than zero. Statistical analyses were performed using Sigmaplot 12.3 and P <0.05 was considered as the significant level. The number of experiments quoted (n) refer to the number of independent oocytes assayed. They originate from at least 2 separate oocyte isolation procedures from different frogs.

PLOS PATHOGENS
A GluCl isoform non-activable by ivermectin is dominant in heteromeric receptors of C rogercresseyi Supporting information S1 Table.