Pharyngeal Pumping and Tissue-Specific Transgenic P-Glycoprotein Expression Influence Macrocyclic Lactone Susceptibility in Caenorhabditis elegans

Macrocyclic lactones (MLs) are widely used drugs to treat and prevent parasitic nematode infections. In many nematode species including a major pathogen of foals, Parascaris univalens, resistance against MLs is widespread, but the underlying resistance mechanisms and ML penetration routes into nematodes remain unknown. Here, we examined how the P-glycoprotein efflux pumps, candidate genes for ML resistance, can modulate drug susceptibility and investigated the role of active drug ingestion for ML susceptibility in the model nematode Caenorhabditis elegans. Wildtype or transgenic worms, modified to overexpress P. univalens PGP-9 (Pun-PGP-9) at the intestine or epidermis, were incubated with ivermectin or moxidectin in the presence (bacteria or serotonin) or absence (no specific stimulus) of pharyngeal pumping (PP). Active drug ingestion by PP was identified as an important factor for ivermectin susceptibility, while moxidectin susceptibility was only moderately affected. Intestinal Pun-PGP-9 expression elicited a protective effect against ivermectin and moxidectin only in the presence of PP stimulation. Conversely, epidermal Pun-PGP-9 expression protected against moxidectin regardless of PP and against ivermectin only in the absence of active drug ingestion. Our results demonstrate the role of active drug ingestion by nematodes for susceptibility and provide functional evidence for the contribution of P-glycoproteins to ML resistance in a tissue-specific manner.


Introduction
The treatment of parasitic nematode infections in animals and humans relies on chemotherapy, and macrocyclic lactones (MLs) represent the most widely used drug class in veterinary medicine due to their high efficacy, low toxicity, and broad spectrum of target parasites [1]. However, drug resistance to MLs is now widespread in parasitic nematodes of ruminants [2], horses [3], companion animals [4,5], and humans [6]. With over 1.5 billion infected humans [7] and essentially all domestic animals exposed to nematode infections, the ongoing development and spread of ML drug resistance is an obstacle to maintaining health standards. In the pathogenic equine parasite Parascaris univalens, resistance against MLs is widespread, which poses a health threat to young animals worldwide [8].
In Caenorhabditis elegans and in parasitic nematodes, the primary pharmacological targets of MLs are the glutamate-gated chloride channels (GluCls) [9,10]. The irreversible binding of MLs to GluCls leads to a hyperpolarisation of the respective neurons that control locomotion, pharyngeal pumping, and egg laying, resulting in flaccid paralysis of the muscles and pharynx [9,10]. However, the underlying resistance mechanisms are insufficiently understood.
Transcription of Pun-pgp-9 was confirmed by RT-PCR for all Pun-PGP-9 transgenic strains by amplifying a 1043 bp PCR product, while no PCR product was amplified in the control strain ( Figure 1a). Protein localisation in the epidermis and the intestine were confirmed by the immunofluorescence detection of a FLAG-tag fused to the NH 2 -end of Pun-PGP-9 and visualised using a secondary DyLight405 antibody and confocal microscopy ( Figure 1b-e). Pun-PGP-9 protein expression in intestine-Pgp-9 line 1 appeared intestinal specific (Figure 1b). At a higher magnification, a strong apical expression was apparent ( Figure 1e). Expression in the epidermis-Pgp-9 strain was detectable in the epidermis ( Figure 1c). Specific blue fluorescence was not visible in the control strain following staining, while pharyngeal GFP expression was visible, indicating that the antibody staining was specific to the FLAG-tagged Pun-PGP-9 ( Figure 1d).
Pharmaceuticals 2021, 14, x FOR PEER REVIEW 3 of 18 derivatives are taken up by C. elegans via the intestine and cuticle-epidermis. In both tissues, Pgps have a protective function by reducing the worms' susceptibility to IVM and MOX.
Transcription of Pun-pgp-9 was confirmed by RT-PCR for all Pun-PGP-9 transgenic strains by amplifying a 1043 bp PCR product, while no PCR product was amplified in the control strain ( Figure 1a). Protein localisation in the epidermis and the intestine were confirmed by the immunofluorescence detection of a FLAG-tag fused to the NH2-end of Pun-PGP-9 and visualised using a secondary DyLight405 antibody and confocal microscopy ( Figure 1b-e). Pun-PGP-9 protein expression in intestine-Pgp-9 line 1 appeared intestinal specific (Figure 1b). At a higher magnification, a strong apical expression was apparent ( Figure 1e). Expression in the epidermis-Pgp-9 strain was detectable in the epidermis (Figure 1c). Specific blue fluorescence was not visible in the control strain following staining, while pharyngeal GFP expression was visible, indicating that the antibody staining was specific to the FLAG-tagged Pun-PGP-9 ( Figure 1d). Tissue-specific expression of Pun-PGP-9 in Cel-pgp-9 loss-of-function strain, strain tm830. Verification of transcription and tissue-specific expression. (a) Visible bands for all reverse-transcriptase (RT) PCRs on cDNA templates made from whole-worm total RNA, and no bands for no RT controls and the control strain. (b-e) Confocal microscope images of tissue-specific expression patterns of fixed, freeze cracked, and immunofluorescence-stained transgenic adult Caenorhabditis elegans (blue lookup table (LUT), 405 nM excitation), pharyngeal GFP expression (green LUT, 488 nm excitation) and merged images (DIC +405 nM + 488 nM). Primary antibodies target the FLAG-tag fused to Pun-PGP-9 and the secondary antibody is conjugated with DyLight405.

Motility Assays in Transgenic and Wildtype Caenorhabditis elegans
All mentioned half-maximal effective concentration (EC 50 ) values, fold changes, and other parameters calculated from the non-linear regression models as well as p-values from statistical significance tests are given in Supplementary Table S1 for IVM and Supplementary Table S2 for MOX and are not specifically referenced in the text.
Motility assays were conducted as described elsewhere [41] with modifications for the stimulation of PP as schematically summarised in Figure 2a. Prior to drug incubations, induction of PP by feeding OP50 or 5-HT treatment or the lack of PP in the absence of a stimulus was confirmed visually under a stereo microscope. By PP stimulation, MLs dissolved in DMSO within the incubation medium were actively ingested. After an 18-24 h incubation, the motility of worms incubated without drugs in 1% DMSO (negative, vehicle control) was tested; neither transgene expression nor induction of pharyngeal pumping by 5-HT or OP50 significantly influenced motility (Kruskal-Wallis test with Dunn's post hoc test) compared to the wildtype (WT, untransformed N2 Bristol strain worms)/OP50 − condition (no PP stimulation) (Figure 2b).

The Effect of Active Drug Ingestion on Ivermectin and Moxidectin Susceptibility
Strikingly, the EC 50 for the WT in the OP50 + condition was 11.1-fold lower than in the absence of OP50 bacteria (Figure 2c). In order to evaluate whether this effect is the result of potentially confounding factors such as IVM metabolisation by the OP50 bacteria or starvation, OP50 were substituted by 5-HT in the incubation medium, thereby eliminating the bacteria and food source but maintaining a stimulus for pharyngeal pumping. This resulted in an 8.0-fold decrease in the EC 50 compared to OP50 − to a level just above that achieved by incubation with OP50 bacteria (Figure 2c).
For MOX, stimulation of PP by OP50 also resulted in overall lower EC 50 values than in the absence of PP (OP50 − ), but the effect of PP was considerably smaller than that observed for IVM (Figure 2c,d).
Comparing PP induction by 5-HT + and OP50 + , IVM EC 50 values were higher when stimulated by 5-HT across all strains except intestine-Pgp-9 (all p < 0.05, extra-sum-ofsquares F test, Bonferroni corrected) while MOX EC 50 values varied slightly in the control and WT and were not significantly different in both Pun-PGP-9 expressing strains (both p > 0.05, extra-sum-of-squares F test, Bonferroni corrected) (Figure 2e).  Tables S1 and S2. Prior to the calculation, the no-drug negative control was set to 0.1 nM and all concentrations were log 10 transformed.

Cel-pgp-9 Loss-of-Function Strain Susceptibility Phenotype in Adult Stage
The control strain, which is the Cel-pgp-9 null mutant (CtrlEx1 [Cel-pgp-9(-); Cel-myo-2p::gfp:: Cel-unc-54_3 -UTR]), did not exhibit increased susceptibility to IVM or MOX (Supplementary Figure S1a,b). Serving as a control for extrachromosomal array transgene expression (pharyngeal GFP expression), the calculated EC 50 values in the control strain matched those of the WT for IVM. Likewise, the control strain's response to MOX was similar to that of the WT strain despite minor variations (Supplementary Figure S1c,d).
The control strain was used as a reference for both the Pun-PGP-9 overexpression strains, epidermis-Pgp-9 and intestine-Pgp-9, as it exhibits the same genetic background and transgenic extra-chromosomal expression.

The Effect of Epidermal Pun-PGP-9 Expression on Ivermectin and Moxidectin Susceptibility
Epidermal Pun-PGP-9 expression reduced susceptibility to IVM only in the absence of PP stimulation but to MOX regardless of PP.
Concerning IVM susceptibility, the epidermis-Pgp-9 line resembled the control strain in the presence of an OP50 PP stimulus ( Figure 3a), but in the absence of a PP stimulus, the EC 50 increased significantly by 2.9-fold compared to the control strain. Similarly, the MOX EC 50 in the absence of PP in the epidermis-Pgp-9 strain was also significantly increased ( Figure 3d). However, significantly increased MOX EC 50 values were observed regardless of PP in this strain ( Figure 3b). Notably, the differences to the control strain were only moderate, i.e., a 1.3-fold in the absence of PP (OP50 − ) and a 1.3-fold (OP50 + ) or a 1.5-fold (5-HT + ) increase in the presence of PP stimulation.

Cel-pgp-9 Loss-of-Function Strain Susceptibility Phenotype in Adult Stage
The control strain, which is the Cel-pgp-9 null mutant (CtrlEx1 [Cel-pgp-9(-); Cel-myo-2p::gfp:: Cel-unc-54_3'-UTR]), did not exhibit increased susceptibility to IVM or MOX (Supplementary Figure 1a,b). Serving as a control for extrachromosomal array transgene expression (pharyngeal GFP expression), the calculated EC50 values in the control strain matched those of the WT for IVM. Likewise, the control strain's response to MOX was similar to that of the WT strain despite minor variations (Supplementary Figure 1c,d. The control strain was used as a reference for both the Pun-PGP-9 overexpression strains, epidermis-Pgp-9 and intestine-Pgp-9, as it exhibits the same genetic background and transgenic extra-chromosomal expression.

The Effect of Epidermal Pun-PGP-9 Expression on Ivermectin and Moxidectin Susceptibility
Epidermal Pun-PGP-9 expression reduced susceptibility to IVM only in the absence of PP stimulation but to MOX regardless of PP.

The Effect of Intestinal Pun-PGP-9 Expression on Ivermectin and Moxidectin Susceptibility
Intestinal Pun-PGP-9 expression reduced susceptibility to both tested MLs, but this effect was always dependent on active drug ingestion.
Concerning IVM, intestinal Pun-PGP-9 expression reduced the susceptibility to IVM in the presence of a PP stimulus compared to the control strain ( Figure 3c) in two independent lines (Supplementary Figure S1c,d). In contrast, IVM EC 50 values for both lines were not significantly different from the control strain when PP was not stimulated (Figure 3c).
MOX EC 50 values in the two intestine-Pgp-9 lines were also significantly elevated when PP was stimulated by OP50 or 5-HT compared to the control strain ( Figure 3d). However, in contrast to IVM, the corresponding fold-changes were considerably smaller (<1.3-fold). In addition, all MOX EC 50 values for the two lines were overall lower than the MOX EC 50 values calculated for the epidermis-Pgp-9 line. In the absence of PP stimulation, EC 50 values did not increase compared to the control strain, which is in line with the observations for IVM.

A Comparison of Moxidectin and Ivermectin Susceptibility in Transgenic and Wildtype Strains
Compared to IVM, EC 50 values for MOX were generally higher across strains, when PP was stimulated but lower in the absence of a PP stimulus ( Figure 4) as detailed below. For example, MOX EC 50 values in the WT in the presence of PP stimulation were significantly higher, by 6.7-and 7.6-fold, compared to IVM (Figure 4) for co-incubation with OP50 or 5-HT, respectively. In contrast, when a PP stimulus was not provided, the MOX EC 50 was not significantly different compared to that of IVM ( Figure 4). Across strains, only small fold-changes below 1.2-fold for the MOX EC 50 values were detected in the presence or absence of PP stimulation, which is in a sharp contrast to those for IVM. As for IVM, a marked EC 50 increase (up to 11-fold in WT) from PP stimulation to no PP was observed in all strains ( Figure 4). Likewise, Pun-PGP-9 expression only resulted in low fold changes for MOX EC 50 values (<1.5-fold) in comparison to considerably higher fold changes (e.g., >3.5fold in intestine-Pgp-9) for IVM ( Figure 4). Despite the marked effect on IVM susceptibility from intestinal Pun-PGP-9 overexpression, all EC 50 values for both intestine-Pgp-9 lines in the presence of PP stimulation remained lower compared to EC 50 values of any other strain in the absence of a PP stimulus (Figure 4). In contrast, MOX EC 50 values in the intestine-Pgp-9 strains in the presence and absence of PP were similar (Figure 4).

The Role of Pun-PGP-9 in Ivermectin and Moxidectin Susceptibility
The objective in this study was to elucidate the general functional role of Pgp expression in specific tissues in the context of ML susceptibility. To this end, we chose to exemplarily characterise Pun-PGP-9, as it has already been shown to interact with IVM in a yeast model [23]. An important question addressed in this work was whether there are differences in the protective effect of Pgp overexpression between different ML derivatives, by testing IVM as an avermectin and MOX as a member of the milbemycins, which are summarised in Figure 4.
Currently, functional evidence regarding the interaction between specific parasite Pgps and different MLs has mostly been inferred from indirect assays using the above-mentioned yeast model [23,24] or the cell line LLC-PK1 [24,29,30,44,45]. In these model systems, MLs were used as inhibitors for a secondary Pgp substrate with a discernible effect in the model systems to characterise the interaction between a transgenic parasite Pgp and different MLs. These studies have consistently suggested that MOX has a lower affinity to the examined nematode Pgps than IVM. This conclusion is further supported by our findings in C. elegans that Pun-PGP-9 overexpression induces larger susceptibility shifts for IVM than for MOX, which were observed without the use of a secondary substrate. Whether the low fold-changes for MOX have any biological significance remains to be elucidated [46]. However, Pun-PGP-9 overexpression reduced both MOX and IVM susceptibility, which indicates that both MLs are probably Pun-PGP-9 substrates. This substrate range could facilitate cross-resistance, which indeed has been reported in Parascaris sp. [47].
In C. elegans, only one study had previously demonstrated the impact of a transgenic Pgp on IVM susceptibility, by overexpressing Pun-PGP-11.1 under the control of the native Cel-pgp-11 promotor [41], which has been shown to primarily drive intestinal expression [48]. The concordantly observed moderate fold changes in this and Janssen et al., despite the extrachromosomal array overexpression, might indicate that Pgps cannot cause resistance by themselves in parasitic nematodes but rather that Pgps are contributors within a multigenic context as proposed by NGS studies in other nematode species [17,18].

The Intestinal and Transcuticlular-Epidermal Uptake of Ivermectin and Moxidectin
Interestingly, the extent of the protective effect against IVM resulting from the Pun-PGP-9 expression at the intestine or the epidermis was markedly influenced by the selective intestinal exposure from active drug ingestion by PP. By itself, this newly described effect of active drug ingestion on ML susceptibility was considerable and, for IVM, even surpassed the extent of the effect of Pgp overexpression in either tissue.
The differences between the effect on MOX and IVM susceptibility of active drug ingestion and intestinal or epidermal Pun-PGP-9 overexpression might be interpreted as differing uptake capacities at specific C. elegans tissues. As a general mechanism, PP stimulation is required for food uptake, and the increased uptake of incubation fluid also facilitates the accumulation of its contents [36].
For IVM, the strong effect of active drug ingestion and the PP dependency of the protective effect of intestinal Pun-PGP-9 expression indicate that higher intestinal exposure increases the overall concentration and effect of IVM. Nonetheless, the measurable impact of epidermal Pun-PGP-9 expression on susceptibility also indicates that IVM is taken up, although less efficiently, via the cuticle-epidermis.
In contrast, the effect of epidermal Pun-PGP-9 expression on MOX susceptibility regardless of PP could suggest that the epidermis-cuticle is the predominant MOX uptake route. For IVM and a few other anthelmintics, transcuticular-epidermal permeability had already been described by biophysical studies on cuticle-epidermis preparations of Ascaris suum and experiments using live A. suum with a surgically ligated pharynx [49,50]. Likewise, experiments in C. elegans have demonstrated a considerably quicker onset of pharyngeal paralysis in a cuticle defective C. elegans bus-8 loss-of-function strain [51]. A study by O'Lone and Campbell indicated that PP inhibition by refrigeration, and thus inhibition of oral ingestion, reduced the susceptibility to IVM in C. elegans [52]. The authors suggested that oral ingestion increases IVM susceptibility but that transcuticular-epidermal uptake is sufficient to induce paralysis at comparatively low concentrations [52], which is in high concordance with our observations and conclusions.
In contrast to IVM, the uptake of MOX in nematodes has not been studied at all. The absence of a similarly strong effect of PP on MOX susceptibility as observed for IVM could indicate that the cuticle-epidermis represents the main MOX penetration route. This conclusion is also supported by the lowest MOX susceptibility regardless of PP stimulation in the epidermis-Pgp-9 strain. The main factor determining permeability through the nematode cuticle is compound size [49,50,53], and MOX is considerably smaller than IVM due to the lack of the C13 glycosyl side chain. Additionally, lipophilicity is an ancillary but important factor for transcuticular permeability [49,50,53], and indeed IVM has a lower lipophilicity than MOX [54]. These differences in size and lipophilicity seem to offer a reasonable explanation for a less efficient transcuticular IVM uptake but cannot explain why intestinal drug exposure from active drug ingestion increased only IVM but not MOX susceptibility. Therefore, more detailed studies are needed to understand this difference. Both ML derivatives are very lipophilic and accumulate well in high-fat-content tissues in the host [55]. In this regard, the mainly intestinal and epidermal fat storage in C. elegans [56] should facilitate passive diffusion and accumulation in these tissues.

Experimental Limitations and Relevance for Parasitic Nematodes
The novel insights from the present study are, to some extent, limited to the model nematode C. elegans that was utilised, since the required forward genetic tools are not yet available for Parascaris sp. or any related parasitic nematode species [57]. However, it may be anticipated that the general conclusions of the tissue-specific protective function in C. elegans may be extrapolated as a general functional mechanism of Pgps in nematode barrier tissues, such as the epidermis and the intestine, in light of a common nematode body structure. Nevertheless, it should be noted that parasitic nematodes exhibit differences in size, life cycle, stage-specific feeding behaviour, genetics, e.g., in their receptor repertoire [58] and, hence, in their drug susceptibility [59]. For example, IVM and MOX EC 50 values for adult C. elegans were several fold higher than those measured in the adult stage of examined parasitic nematodes in vitro, e.g., T. circumcincta [60]. The reasons for these differences remain elusive and have been a matter of speculation. In non-feeding third larval stages, exposure to MLs (or any other anthelmintic) is restricted to the cuticle and, indeed, non-feeding larval stages of parasitic nematodes exhibit considerably lower susceptibilities [60,61]. These differences might be explained by an increased susceptibility from active IVM ingestion. Furthermore, the sheath (i.e., the residual cuticle of the second larval stage) of the third-stage strongylid larva might also reduce susceptibility by limiting drug penetration into the worm. Regarding the uptake of MLs in adult stages of parasitic worms, an in vivo study using an ML-resistant population of the sheep parasite H. contortus found no difference of initial drug accumulation in parasites between MOX, IVM, and abamectin 0.5 days after treatment [62]. However, MOX concentrations in the host dropped significantly compared to IVM and abamectin despite similar plasma levels within 2 days post treatment, suggesting that drug elimination may vary between different ML derivatives [62]. In Ostertagia ostertagi, a cattle parasite dwelling on the surface of the abomasal mucosa and feeding by sucking on the mucosa, IVM levels in the worms were shown to correlate with IVM concentration in the abomasal mucosa and were maximised by subcutaneous drug application [63]. These findings would support that parasite nematode IVM susceptibility is also increased by feeding. In the same study, IVM levels in the cattle parasite, Cooperia oncophora, which dwells in the small intestine, correlated well with intestinal content, and parasite exposure was maximised by oral drug application [63]. Overall, the study by Leathwick et al. highlights how parasite exposure considerably differs between parasitic species with regard to their biology [63], a concept that can also be extended to the different developmental stages, e.g., migrating/histotropic or hypobiotic stages. Based on the feeding behaviour, the blood-sucking H. contortus would be expected to experience high intestinal ML exposure during the early peak in plasma concentrations and, thereafter, prolonged low intestinal exposure, while the parasite cuticle would be exposed from the drug in the abomasal fluid [62,64,65]. Measuring IVM concentrations in adult H. contortus 3 days after subcutaneous injection, Lloberas et al. reported only a weak correlation with plasma levels, possibly because of the timing of the measurement [64]. In contrast, intraruminal IVM application resulted in very high abomasal drug concentration, which also resulted in high concentration in the parasite tissues 3 days after IVM application, suggesting an efficient transcuticular uptake [64].
Regarding the observed tissue-specific expression patterns, some reservations remain whether expression was exclusive to the targeted tissues, the intestine and the epidermis. The representative images in Figure 1 show the principal expression pattern differences observed in the stained worms between the lines with transgenic expression induced by two well-characterised promotors, ges-1p and col-19p. However, extrachromosomal array overexpression might result in background expression in other tissues and in particular epidermal specificity was difficult to verify as fluorescence was visible throughout the worm and staining appeared variable along the length of all stained worms. The fluorescence visible in the entire worm might be explained by autofluorescence as it appeared in both the green and blue channels (Figure 1c). The variability in the strength of the staining was also observed in other worms and could be the result of a non-uniform antibody exposure and deformation resulting from the patchy freeze-cracking. 3D-images would be better suited to verify tissue specificity and to rule out background expression in unwanted tissues [66].

Conclusions
Our results demonstrate that Pgps can contribute to reducing ML susceptibility in a tissuespecific manner in nematodes and that active drug ingestion increases ML susceptibility. The observed dependency of the protective effect of intestinal or epidermal Pgp overexpression on active drug ingestion would suggest a role of Pgps in barrier fortification by reducing tissue drug permeability as a possible mechanism of a Pgp-mediated ML susceptibility as illustrated in Figure 5. As the tissue-specific expression patterns vary considerably between the diverse Pgp lineages in different nematode species [23,27], the protective capacity of a specific Pgp lineage will vary between species and developmental stages.

Plasmids and Plasmid Construction
Plasmids for transgenesis were assembled using the NEB HIFI DNA Assembly Kit Furthermore, our findings emphasise that more attention should be placed on how target parasite species take up ML drugs, since this might lead to differences in susceptibility to individual ML derivatives. Whether changes in barrier permeability represent a relevant mechanism of anthelmintic resistance remains to be elucidated in parasitic nematodes.
In conclusion, this study significantly improves the understanding of a Pgp-mediated ML resistance mechanism by demonstrating how transgenic Pgp expression at specific barriers can impact the susceptibility to different ML derivatives. Furthermore, the differing relevance of active drug ingestion for IVM and MOX susceptibility suggests thus far unknown pharmacological differences and demonstrates the importance of drug barriers and uptake routes for susceptibility.

Generation of Caenorhabditis elegans Strains and Maintenance
Caenorhabditis elegans WT strain Bristol N2 was obtained from the Caenorhabditis Genetics Centre (CGC; University of Minnesota, Minneapolis, MN, USA) and the Cel-pgp-9 deletion strain tm830 [Cel-pgp-9(-)] was obtained from the National BioResource Project (NBRP; Tokyo, Japan). Strains were maintained at 20 • C and standard conditions on NGM plates [67].

Verification of Pun-PGP-9 Expression by RT-PCR and Immunofluorescence
Transcription of Pun-pgp-9 in GFP-positive offspring of injected worms was confirmed by reverse transcriptase (RT)-PCR (primers and cycle conditions in Supplementary Table  S3) using the S7 Fusion enzyme (Mobidiag) and a cDNA or a no-RT control template. cDNA was generated from DNAse-treated RNA. In addition, tissue-specific Pun-PGP-9 protein expression was examined with immunofluorescence. Freeze cracking and antibody staining were performed in a 1.5 mL tube as described elsewhere [69] with minor adaptations using 4% formaldehyde and 50% methanol for fixation followed by freeze-cracking worms three times in liquid nitrogen and thawing in water at room temperature. Following the freezethaw cycles, tubes were shaken at 37 • C for 1 h. To remove the fixative, worms were centrifuged at 11,000× g for 1 min and washed with PBS-T (phosphate-buffered saline +0.5% Triton-X100) four times, removing all liquid during the last aspiration of PBS-T. Prior to antibody staining, worms were incubated in 1 mL PBS-BSA (PBS + 1% bovine serum albumin) overnight at 4 • C under mild shaking. The next day, worms were centrifuged for 1 min and then incubated for 24 h at 4 • C and mild shaking in 500 µL PBS-BSA containing a monoclonal (FG4R), mouse-derived anti-FLAG IgG antibody diluted 1:200 (ab125243, A85282 antibodies.com). The following day, worms were washed again five times in PBS to remove unbound primary antibody. Once again, worms were incubated for 24 h at 4 • C in 500 µL of PBS-BSA containing DyLight 405 conjugated polyclonal donkey anti-mouse antibodies diluted 1:300 (DyLight™ 405 AffiniPure Donkey Anti-Mouse IgG (H+L), Jackson ImmunoResearch). Following another five washes with PBS-BSA, all liquid was completely removed, and worms were transferred in~25 µL to an untreated microscope slide. After adding a drop (~25 µL) of VECTASHIELD ® mounting medium and sealing with nail polish, specimens were examined on an Eclipse Ti-U inverted research confocal microscope (Nikon, Tokyo, Japan) with a 20× and a 40× objective at excitation wavelength 405 nm to visualise antibody-specific staining (DyLight405) and 488 nm excitation to visualise pharyngeal GFP expression. Differential interference contrast (DIC) pictures were taken to visualise worm anatomy. Images were taken using VisiView 4.3.0 at 16 bit. ImageJ was used to pseudocolour and merge channels [40]. Fluorescent images at 405 nM and 488 nm excitation wavelength were visualised in the blue or green lookup table (LUT) scale provided by ImageJ.

Trashing Assay
To prepare stock solutions, IVM and MOX (Sigma-Aldrich, St. Louis, MI, USA) were dissolved in DMSO and frozen at −20 • C. A saturated 40 mM 5-HT stock solution was prepared by dissolving serotonin creatinine sulphate monohydrate (Sigma-Aldrich) in S-medium [67] through vigorous vortexing and was either immediately frozen at −20 • C or used directly.
Bleach-synchronised L1 [67] were grown to adult stage on NGM plates (72 h). Oneday-old adults were washed with S-medium by repeatedly allowing worms to sink to the bottom of a 15 mL centrifuge tube, discarding, and refilling to remove all bacteria. The arrest of PP caused by the absence of food was confirmed visually after one hour of acclimatisation with an inverted microscope. In each individual experiment, 12 worms (only GFP-positive worms in case of transgenic strains) were transferred into 6-well plates (Sarstedt) with 2 mL S-medium final volume containing either no OP50 bacteria (OP50 − ), OP50 bacteria at OD 600 = 0.5 (OP50 + ) or 4 mM 5-HT (OP50 − /5-HT + referred to as 5-HT + in the text for simplicity), the latter two stimulating PP, which was visually confirmed before adding the drug. Plates were sealed with parafilm to avoid evaporation and incubated in the dark at 20 • C and 150 rpm for 18-24 h. The next day, worms were transferred by pipetting to agar-coated Petri dishes filled with S-medium without bacteria and were allowed to acclimatise for 1 min. Then, thrashes of individual worms were counted for one minute on a stereo-fluorescence microscope.
Caenorhabditis elegans adult worms that were incubated in S-medium without bacteria did not readily move, and even soft touch stimulus and shaking did not induce any movements. However, transfer with a pipette induced strong thrashing, hence this step was performed across all experiments and all incubation conditions to detect ML-induced paralysis. Likewise, adult worms incubated with a sufficient supply of OP50 bacteria only thrashed occasionally, but the transfer of worms by pipetting induced vigorous thrashing.

Statistical Analysis
For each concentration, strain, and condition at least 12 worms per day were tested on three separate days (total n ≥ 36). Before log 10 transformation of drug concentrations, vehicle controls were set to concentrations of 0.1 nM. Four-parameter non-linear logistic regression models and statistical analyses were calculated and visualised using GraphPad Prism 8.3.0 (GraphPad Software, San Diego, CA, USA) constraining the bottom values to ≥0. In each graph, the negative control concentration was visualised as "0 M no drug" with a break in the x-axis. Forest plots visualising corresponding half-maximal effective concentration (EC 50 ) and 95% confidence intervals (CI) values were visualised using ggplot2 [42] in R v4.0.3 [43]. Statistical differences in EC 50 values were calculated with the extra-sum-of-squares F test applying the Bonferroni-Holm correction for multiple testing and considering a p-value smaller than 0.05 as significant. For comparison, each Pun-PGP-9-expressing transgenic strain was compared to the control strain at the respective condition. Moreover, the control strain was compared to the WT at the respective condition. The EC 50 values of the WT at the condition without food (OP50 − ) and without food plus 5-HT (5-HT + ) were compared to the condition with food (OP50 + ). In addition, statistical comparisons of EC 50 values between MOX and IVM were calculated for the WT. Fold changes were calculated based on EC 50 values. To examine the impact of the feeding stimuli and transgene expression on motility, the motility response of all worms incubated without drug (negative no drug controls) were pooled per strain and per condition (n = 72) and a Kruskal-Wallis test with a Dunn's post hoc was conducted comparing all conditions in each strain to the WT OP50 − condition in GraphPad.
Author Contributions: A.P.G.: Conceptualisation, methodology, investigation, formal analysis, data curation and visualisation, writing-original draft preparation, review and editing, funding acquisition. J.K.: Conceptualisation, methodology, data curation and visualisation, writing-review and editing, supervision, funding acquisition. C.N.: conceptualisation, writing-review and editing, supervision. C.L.C.: conceptualisation, writing-review and editing, supervision, funding acquisition. A.H.: conceptualisation, methodology, writing-review and editing, supervision. G.v.S.-H.: conceptualisation, data curation and visualisation, writing-review and editing, supervision, funding acquisition. All authors have read and agreed to the published version of the manuscript. Data Availability Statement: Plasmids are available upon request and will be deposited on Addgene.org. The data from this study are available on request from the corresponding author.