Mosquitocidal Activity and Mode of Action of the Isoxazoline Fluralaner

Mosquitoes, such as Aedes aegypti and Anopheles gambiae, are important vectors of human diseases. Fluralaner, a recently introduced parasiticide, was evaluated as a mosquitocide in this study. On Ae. aegypti and An. gambiae fourth-instar larvae, fluralaner had 24-h LC50 (lethal concentration for 50% mortality) values of 1.8 ppb and 0.4 ppb, respectively. Following topical application to adult Ae. aegypti, fluralaner toxicity reached a plateau in about 3 days, with 1- and 3-day LD50 (lethal dose for 50% mortality) values of 1.3 ng/mg and 0.26 ng/mg, suggesting a slowly developing toxicity. Fipronil outperformed fluralaner by up to 100-fold in adult topical, glass contact, and feeding assays on Ae. aegypti. These data show that fluralaner does not have exceptional toxicity to mosquitoes in typical exposure paradigms. In electrophysiological recordings on Drosophila melanogaster larval central nervous system, the effectiveness of fluralaner for restoring nerve firing after gamma-aminobutyric acid (GABA) treatment, a measure of GABA antagonism, was similar in susceptible Oregon-R and cyclodiene-resistant rdl-1675 strains, with EC50 (half maximal effective concentration) values of 0.34 µM and 0.29 µM. Although this finding suggests low cross resistance in the presence of rdl, the moderate potency, low contact activity, and slow action of fluralaner argue against its use as an adult mosquitocide for vector control.


Introduction
Gamma-aminobutyric acid (GABA) is present in both mammals and invertebrates and is an important inhibitory neurotransmitter [1]. Among the different types of GABA receptors, the GABA receptor-chloride channel complex is an important target site for insecticides, such as lindane and fipronil [2]. However, target site mutations have significantly reduced the utility of conventional GABAergic insecticides. Mutation A301S [3,4], which was originally described as A302S [5] and studied in cultured neurons from Drosophila melanogaster, confers ca. 100-fold and ca. 1000-fold resistance to picrotoxinin and lindane, respectively [6]. Therefore, development of insecticides that work on novel binding sites of this receptor could help avoid cross resistance problems and contribute to effective control of pest insects.
Isoxazolines and meta-diamides have emerged as second-generation GABAergic compounds in the search for novel insecticides [7]. Two representative isoxazolines, fluralaner ( Figure 1) and afoxolaner, were developed by scientists from Nissan Chemical Industries in Japan and DuPont in the U.S., respectively [8,9]. Both compounds have been approved by the U.S. Food and Drug Administration (FDA) to be commercialized as parasiticides. Fluralaner shows no cross-resistance in both in vivo and in vitro studies against various insects species compared to other classic GABAergic

Chemicals
Samples of dieldrin and L-aspartic acid were purchased from Sigma-Aldrich Chemical Co. . Rapeseed-oil methyl ester (RME) was purchased from UCY Energy (Alfter, Germany), and silicon oil was acquired from Dow Corning Co. (Auburn, MI, USA). Ethanol, acetone and dimethyl sulfoxide (DMSO) used as solvents were obtained from Sigma-Aldrich Chemical Co.
Fluralaner was extracted and purified from a commercial canine formulation of BRAVECTO™ Chews (for large dogs, 44-88 lbs, each containing 1 g of fluralaner), manufactured by Merck Animal Health. Isolation of the compound was performed with a Teledyne Isco flash chromatography system (Lincoln, NE, USA) using hexanes/ethyl acetate as eluent system. Solvents, hexanes and ethyl acetate were obtained from Acros Organics (Morris Plains, NJ, USA). Melting point was determined on a hot-stage apparatus and is uncorrected. Nuclear magnetic resonance (NMR) analyses were performed at the Nucleic Magnetic Resonance Facility of the University of Florida. NMR spectra were recorded in CDCl3 with tetramethylsilane as the internal standard for 1 H (500 MHz) and CDCl3 as the internal standard for 13 C (125 MHz).

Chemicals
Samples of dieldrin and L-aspartic acid were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO, USA). Fipronil (Figure 1) was donated by Rhône-Poulenc Ag Co. (now Bayer CropScience, Research Triangle Park, NC, USA), and Triton X-100 was acquired from Thermo Fisher Scientific Inc. (Waltham, MA, USA). Diethyl maleate (DEM) was purchased from Sigma-Aldrich Chemical Co., and piperonyl butoxide (PBO) and S,S,S-tributyl phosphorotrithioate (DEF) were obtained from Chem Service Inc. (West Chester, PA, USA). Rapeseed-oil methyl ester (RME) was purchased from UCY Energy (Alfter, Germany), and silicon oil was acquired from Dow Corning Co. (Auburn, MI, USA). Ethanol, acetone and dimethyl sulfoxide (DMSO) used as solvents were obtained from Sigma-Aldrich Chemical Co.
Fluralaner was extracted and purified from a commercial canine formulation of BRAVECTO™ Chews (for large dogs, 44-88 lbs, each containing 1 g of fluralaner), manufactured by Merck Animal Health. Isolation of the compound was performed with a Teledyne Isco flash chromatography system (Lincoln, NE, USA) using hexanes/ethyl acetate as eluent system. Solvents, hexanes and ethyl acetate were obtained from Acros Organics (Morris Plains, NJ, USA). Melting point was determined on a hot-stage apparatus and is uncorrected. Nuclear magnetic resonance (NMR) analyses were performed at the Nucleic Magnetic Resonance Facility of the University of Florida. NMR spectra were recorded in CDCl 3 with tetramethylsilane as the internal standard for 1 H (500 MHz) and CDCl 3 as the internal standard for 13 C (125 MHz).

Insects
Fourth-instar Ae. aegypti larvae were kindly provided by the Center for Medical, Agricultural & Veterinary Entomology (CMAVE), U.S. Department of Agriculture-Agriculture Research Service, Gainesville, FL, USA. The larvae were fed a mixture of ground liver and yeast, maintained under 75% relative humidity and 28 • C, with a 12 h:12 h dark:light cycle, and reared to adulthood for bioassays. Eggs of An. gambiae were provided by BEI Resources under the CDC-MR4 program. The emerged larvae were fed with fish flakes (Tetra, Blacksburg, VA, USA), maintained under 75% relative humidity and 28 • C, with a 12 h:12 h dark:light cycle, and reared to adulthood for bioassays. Bioassays were performed on susceptible G3 (MRA-112) strain [13].
Susceptible (Oregon-R) and cyclodiene-resistant (rdl-1675) strains of D. melanogaster were used in bioassays and electrophysiology experiments. The Oregon-R strain was originally provided by Dr. Doug Knipple from Cornell University, Ithaca, NY, USA, and maintained in culture at the University of Florida since 2009. The rdl-1675 strain was purchased from the Bloomington Drosophila Stock Center at Indiana University, Bloomington, IN, USA. Both strains were reared at 21 • C and provided with artificial media purchased from Carolina Biological Supply, Burlington, NC, USA.

Larval Mosquito Bioassays
Compounds were dissolved in ethyl alcohol, followed by serial dilution to generate an appropriate number of concentrations. A 5 µL aliquot (or more depending on compound solubility) of this solution was added to 5 mL of tap water containing 10 fourth-instar larvae of each treatment group. Tap water treated with the corresponding amount of ethanol was used as the control group. Larvae were held at room temperature (21 • C) and observed periodically for behavioral effects, such as convulsions and paralysis, for the first 4 h after treatment. Larval mortality was recorded after 24 h, 48 h and 72 h, or until toxicity reached a stable plateau. In these assays, the plateau was defined as the treatment day where toxicity was not significantly different from the value observed 2 days later. Larvae were fed during the experiment with a few grains of ground fish flakes added to the petri dish every day. A Triton X-100 solution of 10 ppm was also tested to see whether it would increase the toxicity of fluralaner.
Paralytic activity of compounds to headless larvae was assessed as described by Islam and Bloomquist [14], with slight modification. Heads of fourth-instar larvae were detached with forceps and treated as described above, except in physiological saline instead of tap water. The saline was composed of (mM): NaCl (154), KCl (2.7), CaCl 2 (1.4), HEPES (4) at pH 7.2. Mosquito saline treated with 0.5% ethanol and 100 ppm L-aspartic acid were used as negative and positive controls, respectively. An insect pin was used to probe larvae every hour to observe swimming behavioral responses. Compared to the control group, larvae showing irregular behaviors such as consistent convulsion, twitching or only slight movement were counted as paralyzed. Each concentration was repeated on at least three different batches of mosquitoes. LC 50 (lethal concentration for 50% mortality) and PC 50 (concentration for 50% paralysis) values were calculated as described in the statistics section.

Adult Mosquito Bioassays
For topical toxicity assays [15], 5-7 days old adult female mosquitoes (non-blood fed, n = 10) were chilled on ice for 2 min, during which time 0.2 µL of insecticide solution (in ethanol) was applied to the thorax by a hand-held microdispenser (Hamilton, Reno, NV, USA). A solvent-only treatment was included in each experiment as a negative control. Following treatment, mosquitoes were transferred to paper cups covered with netting. A 10% sugar solution in tap water was supplied via a cotton ball placed on the netting and changed every day. Paper cups were held at room temperature (21 • C), and the mosquitoes were observed for behavioral effects such as convulsion or paralysis and the onset of toxicity for the first 4 h. To test for synergistic effects, synergists were applied topically to the abdomen of mosquitoes 4 h before treatments (PBO, 500 ng per mosquito; DEF, 200 ng per mosquito; or DEM, 1 µg per mosquito) at amounts that generally produced little or no mortality. Mortality was recorded after 24 h, 48 h, and 72 h, or until toxicity reached a stable plateau. Each dose was repeated on at least three different batches of mosquitoes. The LD 50 (lethal dose for 50% mortality) without synergist and LD 50 with synergist were calculated, as described in the statistics section. The synergist ratio was determined by the equation: LD 50 of insecticide alone/LD 50 of insecticide + synergist.
Mosquito injection bioassays [16] were performed under a microscope with a manual microsyringe pump (World Precision Instruments, Sarasota, FL, USA) and a fine glass pipette (TW100-4, World Precision Instruments) fabricated by a pipette puller (Sutter Instrument, Novato, CA, USA), and broken at the tip (ca. 20 µm). Compound solutions were prepared in mosquito saline (described in headless larvae assays) with 5% ethanol as a vehicle. A 5% ethanol solution in saline was used as control. Non-blood fed female mosquitoes (5-7 days post-emergence, n = 10) were chilled on ice and injected with 0.2 µL compound solution into the side of their thorax. Treated mosquitoes were then placed into paper cups and maintained as described above for topical assays. Each dose series was repeated on at least three different batches of mosquitoes, with LD 50 values calculated as described in the statistics section.
For feeding assays [17], female mosquitoes (5-7 days of age and non-blood fed) were starved for 6 h and chilled on ice for 2 min, then transferred to glass test tubes (n = 10). Compound solutions were prepared in 10% sugar water with 0.5% ethanol used as vehicle. The control group was assigned 10% sugar water with 0.5% ethanol alone. Cotton balls were treated with 1 mL of prepared compound solution and inserted at the top of the test tubes. Test tubes were maintained at room temperature (21 • C), with the cotton ball changed and solution reapplied every day. To assess synergist effects on feeding assays, synergists (compounds and doses as given above) were topically applied 4 h before treatment. Mortality data was recorded after 24 h, 48 h, and 72 h, or until toxicity reached a stable plateau, with each experiment repeated at least three times. LC 50 values were calculated as described in the statistics section.
A contact filter paper toxicity assay was conducted according to WHO protocol [18]. Adult female mosquitoes (n = 10) were 5-7 days of age and non-blood fed at the time of experimentation. Serial dilutions of each insecticide dissolved in ethanol were prepared prior to treatment and 2 mL of each concentration was applied to a 180 cm 2 (12 cm × 15 cm) filter paper (Chromatography Paper, Fisher Scientific Pittsburgh, PA, USA). Papers were left to dry for 24 h prior to use. Mosquitoes were chilled for 2 min on ice, after which they were transferred to a WHO cylindrical plastic holding chamber and held for 1 h to acclimatize. The mosquitoes were then gently transferred into the treatment chamber, which contained the treated paper, and exposed for 1 h in the vertical tube. After the 1-h exposure time, the mosquitoes were transferred to the holding chamber as described above, with 10% sugar solution supplied on the netting and changed every day. Holding chambers were left at room temperature (21 • C) after the treatment, and mortality was recorded after 24 h, 48 h, and 72 h, or until toxicity reached a stable plateau. Each experiment was repeated on at least three different batches of mosquitoes. LC 50 values with and without synergist, as well as synergist ratios were calculated, as described in the statistics section.
Additional surface contact assays were conducted in borosilicate glass test tubes (20 mm × 150 mm) with an inner surface area of ca. 85 cm 2 (Fisher Brand 14-961-33). Compounds were dissolved in acetone and 250 µL was added to each tube, with 250 µL of acetone as the control group. The test tube was carefully rotated to allow the acetone to evaporate and to distribute the compound evenly inside the tube. Adult female mosquitoes (n = 10) 5-7 days from emergence (non-blood fed) were chilled on ice for 2 min and transferred into compound-coated tubes. To ensure parallel comparison to WHO paper assay, in another set of experiments mosquitoes were transferred to clean tubes after 1-h exposure to insecticide-coated glass tubes. A cotton ball with 10% sugar water solution was used to seal test tubes and changed every day to ensure food supply for mosquitoes. Test tubes were maintained at room temperature (21 • C) and mortality data was collected after 24 h, 48 h, and 72 h, or until toxicity reached a stable plateau. Each experiment was repeated on at least three different batches of mosquitoes. LC 50 values were calculated as described in the statistics section.

Adult D. melanogaster Bioassays
Adult females (n = 10) from both strains (CS-OR and rdl-1675) that were 5-7 days of age were tested in feeding assays and glass contact assays, and the resistance ratio (LC 50 rdl-1675/LC 50 CS-OR) was assessed via both routes of exposure. Procedures for both assays were the same as for the mosquito bioassays, except that D. melanogaster was anesthetized by a constant flow of CO 2 for several seconds.

Electrophysiological Recording on D. melanogaster Larval CNS
Electrophysiological recordings from D. melanogaster third-instar larval CNS were performed as described previously [19], with slight modification. The CNS was dissected in physiological saline containing (mM) NaCl (157), KCl (3), CaCl 2 (2), HEPES (4), at pH 7.2, and either left intact or transected posterior to the cerebral lobes to eliminate the blood-brain barrier (BBB) and facilitate penetration of chemicals into the central synapses. A recording suction electrode was pulled with a pipette puller (Sutter Instrument, Novato, CA, USA) and broken at the tip (ca. 40 µm) by fine forceps before being filled with saline. Several peripheral nerve trunks were drawn into the electrode to record nerve activity descending from the CNS.
Electrical signals were amplified, digitized, and transmitted to the analysis software, wherein spikes were converted to a rate (spikes/s or Hz) by a PowerLab analog to digital converter hardware and LabChart 7 software (ADInstruments, Colorado Springs, CO, USA). To eliminate background noise, spikes were only tallied if they exceeded a fixed threshold, which was set when no peripheral nerves were attached to the suction electrode. After baseline frequency was established, 1 mM GABA was added to the saline bath to inhibit nerve activity, and allowed to incubate for 5 min (Figure 2 to seal test tubes and changed every day to ensure food supply for mosquitoes. Test tubes were maintained at room temperature (21 °C) and mortality data was collected after 24 h, 48 h, and 72 h, or until toxicity reached a stable plateau. Each experiment was repeated on at least three different batches of mosquitoes. LC50 values were calculated as described in the statistics section.

Adult D. melanogaster Bioassays
Adult females (n = 10) from both strains (CS-OR and rdl-1675) that were 5-7 days of age were tested in feeding assays and glass contact assays, and the resistance ratio (LC50 rdl-1675/LC50 CS-OR) was assessed via both routes of exposure. Procedures for both assays were the same as for the mosquito bioassays, except that D. melanogaster was anesthetized by a constant flow of CO2 for several seconds.

Electrophysiological Recording on D. melanogaster Larval CNS
Electrophysiological recordings from D. melanogaster third-instar larval CNS were performed as described previously [19], with slight modification. The CNS was dissected in physiological saline containing (mM) NaCl (157), KCl (3), CaCl2 (2), HEPES (4), at pH 7.2, and either left intact or transected posterior to the cerebral lobes to eliminate the blood-brain barrier (BBB) and facilitate penetration of chemicals into the central synapses. A recording suction electrode was pulled with a pipette puller (Sutter Instrument, Novato, CA, USA) and broken at the tip (ca. 40 µm) by fine forceps before being filled with saline. Several peripheral nerve trunks were drawn into the electrode to record nerve activity descending from the CNS.
Electrical signals were amplified, digitized, and transmitted to the analysis software, wherein spikes were converted to a rate (spikes/s or Hz) by a PowerLab analog to digital converter hardware and LabChart 7 software (ADInstruments, Colorado Springs, CO, USA). To eliminate background noise, spikes were only tallied if they exceeded a fixed threshold, which was set when no peripheral nerves were attached to the suction electrode. After baseline frequency was established, 1 mM GABA was added to the saline bath to inhibit nerve activity, and allowed to incubate for 5 min ( Figure 2).  Experimental compounds (e.g., fluralaner) were then added to the saline bath (1 µL of DMSO solution in a 1 mL bath volume) and mixed by gentle pipetting to assess their ability to reverse the inhibitory effect of GABA. The baseline (pre-treatment) and GABA inhibited nerve firing rates were each averaged over a 3 min period. The drug-induced nerve firing rate was averaged over 3 min periods after treatment ( Figure 2). Fipronil and dieldrin at a concentration of 10 µM were used as positive controls. If, after GABA treatment, the nerve firing rate recovered to at least 25% of the original pretreatment level, the toxicant was considered to have induced a positive response.
Drug effects on larval CNS with intact BBB also were assessed, with experimental compounds added to the bath after establishing the baseline frequency, and each drug concentration was replicated 3-9 times. Fipronil at 10 µM was used as positive control. Recording data obtained within 36 min after addition of experimental compounds were included in the analysis.

Test of GABAergic Compounds on Mammalian GABA A Receptors
Effects of fluralaner, fipronil, dieldrin, and picrotoxin were tested on GABA α 1 β 3 γ 2 ion channels expressed in HEK293 cells by ChanTest Corporation, Cleveland, OH, USA. Briefly, HEK293 cells were transfected with cDNA of GABA α 1 β 3 γ 2 ion channels and maintained in Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 with addition of 10% fetal bovine serum, 100 U/mL penicillin G sodium, 100 µg/mL streptomycin sulfate, and 500 µg/mL G418. Extracellular buffer containing (mM): NaCl (137), KCl (4), CaCl 2 (1.8), MgCl 2 (1), HEPES (10), Glucose (10), at pH 7.4 was loaded into planar patch clamp (PPC) plate wells (11 µL per well). Cell suspension was then pipetted into the intracellular compartment (9 µL per well) of the PPC planar electrode. With whole-cell recording, effects of compounds on transfected HEK293 cells were detected in agonist mode (application of test compounds only) and in antagonist mode (application of GABA with test compound), with solution added 10 µL/s for 2 s. GABA and picrotoxin were used as agonist positive control and antagonist positive control, respectively.
The agonist effect of the test compounds and GABA (positive control) was calculated as: % activation = (I TA /I Max ) × 100%, in which I TA was the compound-induced current at various concentrations, and I Max was the mean current induced with 300 µM GABA. Inhibitory effect of the test compounds and picrotoxin on the channel was calculated as: % Inhibition = (I TA /I EC80 ) × 100%, where I TA was the GABA EC 80 -induced current in the presence of various concentrations of the test compound and I EC80 was the mean current elicited with GABA EC 80 . GABA EC 80 values were selected based on ChanTest historical data: 60 µM for α 1 β 3 γ 2 GABA ion channels.
Inhibitory concentration-response data were fitted to an equation of the form: % Inhibition = % VC + {(% PC − % VC)/[1 + ([Test]/IC 50 ) N ]}, where [Test] was the concentration of the compound, IC 50 was the concentration of the compound producing 50% inhibition, N was the Hill coefficient, % VC was the mean current at the passive control (picrotoxin EC 50 ), % VC was the mean current at the vehicle control (GABA EC 50 ) and % inhibition was the percentage of ion channel current inhibited at each concentration of the test compound. Nonlinear least squares fits were solved with the XLfit add-in for Excel software (Microsoft, Redmond, WA, USA).

Statistical Analysis
Control mortality was corrected by Abbott's formula and LC 50 , LD 50 , and PC 50 values were calculated by SAS software (Proc Probit, SAS 9.4, SAS Institute Inc., Cary, NC, USA). EC 50 and IC 50 values, and concentration-response curves of drugs in Drosophila larval CNS recordings were obtained by nonlinear regression to a four-parameter logistic equation in GraphPad Prism 4.0 software (GraphPad Software, Inc., San Diego, CA, USA). The two-tailed Student's t-test, one-way ANOVA with Bonferroni or Tukey's post-test used to analyze results between treatment group means, with a p < 0.05 considered to be statistically significant.

Larval Mosquito Bioassays
The 24-h LC 50 values of fluralaner on fourth-instar Ae. aegypti and An. gambiae larvae were 1.8 ppb and 0.4 ppb, respectively; a statistically significant difference (p < 0.05) by a factor of 4.5 ( Table 1). The toxicity of fluralaner on Ae. aegypti larvae also increased significantly (p < 0.05) by a factor of 3.6 after 48 h of exposure (Table 1). Aedes larval assays performed in mosquito physiological saline and in 10 ppm Triton X-100 solution displayed a small, but statistically-significant increase (p < 0.05) of LC 50 values to 2.9 ppb and 2.7 ppb, respectively, compared to those performed in tap water ( Table 1). The 24 and 48 h LC 50 values of fipronil on fourth-instar Ae. aegypti larvae were 23 ppb and 5.1 ppb, respectively, and showed 10-to 13-fold less potency compared to the toxicity of fluralaner in the same assay. The LC 90 values at the 48-h time point of fluralaner and fipronil on fourth-instar Ae. aegypti larvae were 2.2 ppb (1.4-6.0 ppb) and 11 ppb (8-21 ppb), respectively. Paralysis studies with fluralaner on headless and intact Ae. aegypti larvae yielded a significant (p < 0.05) 10-fold difference in PC 50 values (Table 1). Intoxication signs of fluralaner in mosquito larvae were spontaneous asynchronous contractions and convulsions, similar to signs of fipronil exposure.

Adult Mosquito Bioassays
In adult mosquito bioassays, intoxication signs included an inability to stand upright, and flying along the bottom of holding container while on their backs. In topical assays, the 24 h LD 50 values of fluralaner on Ae. aegypti and An. gambiae were 1.3 ng/mg and 0.29 ng/mg, respectively (Table 2), a species difference of 4-fold. In time course studies with Ae. aegypti, the toxicity of fluralaner on mosquitoes increased progressively (Figure 3), and the LD 50 value following a single topical application decreased significantly (p < 0.05; one-way ANOVA, Tukey's test) on a daily basis until day 3 ( Table 2). At day 3, the LD 50 value was 0.26 ng/mg (Table 2), five-fold lower than that observed on day 1. At observation days 4, 5 and 6, the toxicity did not increase further (p > 0.05) compared to the LD 50 on day 3 (Figure 3), and at the two lowest doses tested did not reach 100% mortality, even after 7 days of observation ( Figure 3). For An. gambiae, fluralaner toxicity reached its plateau at day 2, with an LD 50 value of 0.21 ng/mg (Table 2), a difference of only 1.4-fold compared to 24 h. For comparison, topical assays of fipronil on Ae. aegypti found an LD 50 value after 24 h of 0.062 ng/mg ( Table 2). Toxicity of fipronil in Ae. aegypti topical assay stopped increasing significantly at day 2, with a LD 50 value of 0.021 ng/mg (Table 2). Fipronil was more active than fluralaner on Ae. aegypti in topical assays by 21-fold at 1 day post-exposure and 17-fold at day 2.    For injection bioassays of fluralaner on Ae. aegypti adult females, the 24-h LD50 value was 0.6 ng/mg (Table 3), with no increase of toxicity at the 48-h and 72-h time point. This LD50 value was less than that in 24-h topical assays by a factor of 2. The 24-h LC50 values for glass contact and feeding assays were 13 ng/cm 2 and 49 ppm, respectively ( Table 3). Toxicity of fluralaner by feeding and glass contact methods increased by factors of 2-to 4-fold after 48 h, compared to 24-h exposures. For comparison, fipronil had a glass contact LC50 value of 10 ng/cm 2 after 24 h and 0.7 ng/cm 2 after 48 h, which is a 14-fold increase in toxicity (Table 3).
In feeding assays, fipronil had LC50 values of 0.47 ppm and 0.12 ppm for 24 h and 48 h, respectively, and the toxicity did not increase significantly after 48 h (Table 3). Compared to fluralaner, fipronil had ca. 8-fold greater toxicity in glass contact assays at the 48-h time point and ca. 100-fold higher toxicity in feeding assays at 24-h and 48-h time points. At the 24-h time point in the glass contact assay, the difference in toxicity of fluralaner and fipronil was not statistically significant.
In WHO paper assays, fluralaner was dissolved in ethanol, with 3.6 mg/cm² of either silicon oil or RME added to the solvent. These formulations at 2 mg/paper (ca. 11,000 ng/cm 2 ) of fluralaner killed at most 12% at the 48-h time point. In glass contact assays with exposure time of 1 h to match the WHO paper assay, fluralaner displayed a LC50 value of 504 (152-1338) ng/cm 2 , which is significantly more toxic than that in WHO paper assays by a factor of at least 22-fold. For injection bioassays of fluralaner on Ae. aegypti adult females, the 24-h LD 50 value was 0.6 ng/mg (Table 3), with no increase of toxicity at the 48-h and 72-h time point. This LD 50 value was less than that in 24-h topical assays by a factor of 2. The 24-h LC 50 values for glass contact and feeding assays were 13 ng/cm 2 and 49 ppm, respectively ( Table 3). Toxicity of fluralaner by feeding and glass contact methods increased by factors of 2-to 4-fold after 48 h, compared to 24-h exposures. For comparison, fipronil had a glass contact LC 50 value of 10 ng/cm 2 after 24 h and 0.7 ng/cm 2 after 48 h, which is a 14-fold increase in toxicity (Table 3).
In feeding assays, fipronil had LC 50 values of 0.47 ppm and 0.12 ppm for 24 h and 48 h, respectively, and the toxicity did not increase significantly after 48 h (Table 3). Compared to fluralaner, fipronil had ca. 8-fold greater toxicity in glass contact assays at the 48-h time point and ca. 100-fold higher toxicity in feeding assays at 24-h and 48-h time points. At the 24-h time point in the glass contact assay, the difference in toxicity of fluralaner and fipronil was not statistically significant.
In WHO paper assays, fluralaner was dissolved in ethanol, with 3.6 mg/cm 2 of either silicon oil or RME added to the solvent. These formulations at 2 mg/paper (ca. 11,000 ng/cm 2 ) of fluralaner killed at most 12% at the 48-h time point. In glass contact assays with exposure time of 1 h to match the WHO paper assay, fluralaner displayed a LC 50 value of 504 (152-1338) ng/cm 2 , which is significantly more toxic than that in WHO paper assays by a factor of at least 22-fold. Synergist assays, in which PBO, DEF, or DEM was topically applied to adult female Ae. aegypti 4 h before treatment, were performed in topical and feeding assay formats (Tables 4 and 5). PBO displayed little effect on fluralaner toxicity in either assay, with synergist ratios of at most 2.2-fold. The synergist DEM decreased toxicity to fluralaner in the feeding assay (Table 4), but increased toxicity up to about 2-fold in topical treatments (Table 5). In contrast, DEF displayed strong synergism of fluralaner in feeding assays (Table 4), but weak synergism in topical assays (Table 5).  Table 3)/LC 50 of fluralaner + synergist; * Significant difference at p < 0.05, compared to data for fluralaner alone on the corresponding day (Table 3).

Adult D. melanogaster Bioassays
To confirm resistance in the rdl-1675 strain, dieldrin feeding assays were performed. In these studies, the susceptible strain (CS-OR) had a LC 50 value of 2.5 ppm (95% CI = 2.1-3.0 ppm) to dieldrin, whereas the resistant strain (rdl-1675) had no mortality when fed 100 ppm dieldrin, indicating a resistance ratio greater than 40-fold. In both glass contact and feeding assays, fluralaner displayed equivalent toxicity to susceptible and resistant strains of D. melanogaster ( Table 6). The LC 50 values of both D. melanogaster strains were not significantly different from each other at the 24-h and 48-h time point (Table 6). In glass contact assays, LC 50 values of fluralaner on Ae. aegypti and D. melanogaster were not significantly different from each other at recorded time points (Tables 3 and 6). However, the toxicity of fluralaner in feeding assays against D. melanogaster was greater than that against Ae. aegypti by a factor of ca. seven (Tables 3 and 6). As was observed for mosquitoes, mortality in D. melanogaster increased significantly at 48 h compared to 24 h (Table 6).

CNS Recordings on D. melanogaster
Initial experiments compared the solvent responses of intact and severed (BBB disrupted) larval CNS preparations. Intact and severed susceptible (CS-OR) CNS treated with DMSO showed a consistent nerve firing pattern, which is cyclic oscillations ( Figure 4A-C), but overall spike rates declined over the 30-min observation period ( Figure 4D). Health 2017, 14, 154 10 of 17

Adult D. melanogaster Bioassays
To confirm resistance in the rdl-1675 strain, dieldrin feeding assays were performed. In these studies, the susceptible strain (CS-OR) had a LC50 value of 2.5 ppm (95% CI = 2.1-3.0 ppm) to dieldrin, whereas the resistant strain (rdl-1675) had no mortality when fed 100 ppm dieldrin, indicating a resistance ratio greater than 40-fold. In both glass contact and feeding assays, fluralaner displayed equivalent toxicity to susceptible and resistant strains of D. melanogaster ( Table 6). The LC50 values of both D. melanogaster strains were not significantly different from each other at the 24-h and 48-h time point (Table 6). In glass contact assays, LC50 values of fluralaner on Ae. aegypti and D. melanogaster were not significantly different from each other at recorded time points (Tables 3 and 6). However, the toxicity of fluralaner in feeding assays against D. melanogaster was greater than that against Ae. aegypti by a factor of ca. seven (Tables 3 and 6). As was observed for mosquitoes, mortality in D. melanogaster increased significantly at 48 h compared to 24 h (Table 6).

CNS Recordings on D. melanogaster
Initial experiments compared the solvent responses of intact and severed (BBB disrupted) larval CNS preparations. Intact and severed susceptible (CS-OR) CNS treated with DMSO showed a consistent nerve firing pattern, which is cyclic oscillations ( Figure 4A-C), but overall spike rates declined over the 30-min observation period ( Figure 4D).  On severed CS-OR larval CNS, discharge rates dropped to 72% ± 11% of baseline firing rate 30 min after the addition of DMSO. Firing frequencies of intact CS-OR and rdl-1675 (resistant) CNS at 30 min after DMSO addition were 85% ± 3% and 89% ± 18%, respectively, which were greater than that of severed CS-OR CNS ( Figure 4D). This difference, however, was not statistically significant.
On CS-OR larval CNS preparations, 10 µM fluralaner showed neuroexcitatory effects within five minutes of addition to the bath ( Figure 5A,B). After the treatment of 10 µM fluralaner, firing patterns on larval CNS displayed more rapid cyclic oscillations on top of a greater tonic discharge rate, consistent with a disruption of normal pattern generation. Blocking of nerve discharge was observed following the neuro-excitation, and showed some evidence of greater effect in the severed preparations; viz., there is more residual firing evident in the intact preparations. The effects of 10 µM fipronil on intact and severed CN-OR larval CNS were similar to that of 10 µM fluralaner; an excitatory action followed by inhibition of nerve firing ( Figure 5C,D). Excitation and nerve blocking effects of fluralaner were concentration-dependent, and lower concentrations required longer incubation times to cause either effect. This relationship is evident in electrophysiological traces (Figure 6), and explored in measurements of time to 50% block of nerve firing (Figure 7). In both susceptible and resistant CNS preparations, reduction in firing occurs more quickly and is more extensive at higher concentrations ( Figure 6). Similar results were obtained on intact rdl-1675 larval CNS, where 10 µM fluralaner and fipronil blocked nerve discharge significantly faster than 0.1 µM fluralaner (Figures 6 and 7). Time to block of nerve firing was used to quantify effects on the CNS. However, there was not a statistically significant difference in time to 50% nerve block for fluralaner between intact and severed CNS preparations from the susceptible Oregon-R strain (Figure 7). Given the variability inherent to spontaneous nerve discharge recordings, no other analyses were attempted.  Excitation and nerve blocking effects of fluralaner were concentration-dependent, and lower concentrations required longer incubation times to cause either effect. This relationship is evident in electrophysiological traces (Figure 6), and explored in measurements of time to 50% block of nerve firing (Figure 7). In both susceptible and resistant CNS preparations, reduction in firing occurs more quickly and is more extensive at higher concentrations ( Figure 6). Similar results were obtained on intact rdl-1675 larval CNS, where 10 µM fluralaner and fipronil blocked nerve discharge significantly faster than 0.1 µM fluralaner (Figures 6 and 7). Time to block of nerve firing was used to quantify effects on the CNS. However, there was not a statistically significant difference in time to 50% nerve block for fluralaner between intact and severed CNS preparations from the susceptible Oregon-R strain (Figure 7). Given the variability inherent to spontaneous nerve discharge recordings, no other analyses were attempted.

Intact Severed
For comparison, dieldrin, fipronil, and fluralaner were applied to severed susceptible and resistant CNS preparations pretreated with 1 mM GABA to inhibit nerve discharge. Dieldrin (10 µM) stimulated recovery of firing in CS-OR CNS ( Figure 8A), but on rdl-1675 CNS yielded virtually no recovery of nerve activity ( Figure 8B). Application of 0.1% DMSO had no effect on GABA-treated CNS preparations (data not shown). At 3 µM, all preparations from both strains showed a positive response (recovery to at least 25% of baseline firing frequency within a 30 min observation period) to fluralaner ( Figure 8C,D). Similarly, fipronil at 10 µM showed similar recovery of firing on severed larval CNS in both susceptible and resistant strains ( Figure 8E,F). There was no response to 30 nM fluralaner for either Drosophila strain, but higher concentrations were effective (Figure 9). For both susceptible and cyclodiene-resistant CNS preparations, 100 nM fluralaner was the lowest concentration that reversed, even partially, the GABA inhibitory effect (Figure 9). The EC 50 values for fluralaner on CS-OR and rdl-1675 severed CNS preparations were 0.34 µM (95% CI: 0.06-1.8 µM) and 0.29 µM (95% CI: 0.09-0.92 µM); not significantly different from each other.  For comparison, dieldrin, fipronil, and fluralaner were applied to severed susceptible and resistant CNS preparations pretreated with 1 mM GABA to inhibit nerve discharge. Dieldrin (10 µM) stimulated recovery of firing in CS-OR CNS ( Figure 8A), but on rdl-1675 CNS yielded virtually no

CS-OR rdl-1675
Intact  For comparison, dieldrin, fipronil, and fluralaner were applied to severed susceptible and resistant CNS preparations pretreated with 1 mM GABA to inhibit nerve discharge. Dieldrin (10 µM) stimulated recovery of firing in CS-OR CNS ( Figure 8A), but on rdl-1675 CNS yielded virtually no recovery of nerve activity ( Figure 8B). Application of 0.1% DMSO had no effect on GABA-treated

Potency on Mammalian GABA Receptors
Fluralaner was screened on a mammalian GABA A receptor construct under contact by ChanTest TM . Only GABA displayed agonist activity among the compounds tested, and it had an EC 50 value of 11 µM on mammalian GABA A α 1 β 3 γ 2 receptors. Fluralaner and dieldrin showed low potency for block of these receptors, with IC 50 values above 30 µM. For comparison, fipronil and picrotoxin had IC 50 values of 4.9 µM and 3.9 µM, respectively.

Discussion
Gassel et al. [11] performed toxicity comparisons of fluralaner with other commercialized compounds, including fipronil. On Ctenocephalides felis (cat flea), Ae. aegypti, Lucilia cuprina Meigen (sheep blowfly), and Stomoxys calcitrans Linnaeus (stable fly), fluralaner outperformed dieldrin and imidacloprid, as well as deltamethrin, except on S. calcitrans. Ozoe et al. [10] also reported that fipronil outperformed fluralaner on C. felis by a factor of five in dry film contact assays. For mosquitoes, Gassel et al. [11] reported a single finding that fluralaner (48 h LC 90 value = 1.2 ppt) was ca. 16,000-fold more potent than fipronil on Ae. aegypti first-instar larvae (48 h LC 90 value = 20 ppb). In the present study, fluralaner gave a 48 h LC 90 value of 2.2 ppb on fourth-instar Ae. aegypti larvae, differing from the data on first-instar larvae by over 1800-fold. However, fipronil showed a 48-h LC 90 value of 11 ppb, similar to toxicity observed by Gassel et al. [11] for first instars (20 ppb for >90% mortality). In addition, on fourth-instar Ae. aegypti larvae, fipronil was 10-to 13-fold less potent than fluralaner. These data suggest that fluralaner might be an excellent larvicide, although the life stage of the mosquito larvae has a large impact on chemical sensitivity. On adult Ae. aegypti, fipronil had higher toxicity than fluralaner in topical (7-to 21-fold), feeding (ca. 100-fold), and glass contact assays (8-fold at 48 h). These findings, plus the greater speed of action, suggest fipronil would be a better overall adult mosquitocide, at least in the absence of resistance.
Bioassays and CNS recordings provide some insight into the slow toxicity of fluralaner. It took 3 days for fluralaner toxicity to plateau, and its toxicity was not enhanced much by injection (about two-fold compared to topical application). In contrast, the LD 50 value by injection can be reduced by more than 10-fold compared to topical treatments for carbamates such as propoxur [16], another compound that must reach central synapses to exert its effects. Cuticle thickness is known to have a significant and positive correlation with the time to knock down by permethrin, suggesting that thicker cuticle led to a slower rate of insecticide penetration [20]. For larval assays of Ae. aegypti, the concentration leading to 50% paralysis on intact larvae was 10-fold higher than that for headless larvae. Thus, the cuticle of fourth-instar Ae. aegypti proved to be a more important factor influencing fluralaner toxicity than in adults. The large size (molecular weight = 556) and high lipophilicity (log p = 5.0) could influence fluralaner penetration of barriers and contribute to its slowly developing toxicity. These barriers would include the blood-brain-barrier, and although there was no significant difference between speed of nerve discharge block in severed vs. intact D. melanogaster larval CNS, the blood brain barrier in adult mosquitoes may have a different permeability to fluralaner.
At present, there is no published information on the metabolism of fluralaner in insects. A decrease in toxicity was observed with DEM in adult sugar feeding assays (Table 5), but the overall effect was small, as was the potentiation of toxicity by PBO. Both findings argue against significant glutathione-S-transferase and P450 monooxygenase metabolism. DEF is a potent carboxylesterase inhibitor [21] and in this study had the most significant positive synergist ratios in feeding assays with Ae. aegypti adults. However, little synergism was noted in topical applications of fluralaner, suggesting that the factor responsible resides in the alimentary system. No ester linkage is present in the fluralaner molecule, but it does have two adjacent amide groups. Carboxylamidases have been identified in different insect species, including Lepidoptera, Orthoptera, and Dictyoptera, which can use p-nitroacetanilide as a model substrate, and were found most abundantly in the midgut [22,23]. The only purified carboxylamidase studied from insects was insensitive to DEF [22]; however, a carboxylamidase might be present in adult Ae. aegypti that metabolizes fluralaner in a DEF-sensitive fashion. Alternatively, DEF might have other mechanisms for potentiating the oral toxicity of fluralaner. All these possibilities await further investigation.
Fluralaner showed no target site cross resistance with other GABA receptor-directed compounds. Ozoe et al. [10] reported fluralaner to be equally effective in topical assays on dieldrin-resistant and susceptible strains of houseflies, with LD 50 values of 1.01 ng/mg and 0.85 ng/mg, respectively. These values were similar on a per mg insect weight basis to the topical toxicity found for fluralaner on Ae. aegypti adults (1.3 ng/mg at 24 h). Additionally, Asahi et al. [24] showed that fluralaner had similar potent toxicity on fipronil-resistant and susceptible strains of Laodelphax striatellus Fallén, whereas fipronil showed a resistance ratio of about 1700. These findings are in agreement with results of both feeding and glass contact assays on dieldrin-resistant and susceptible strains of D. melanogaster reported here, where fluralaner showed similar activity at all recorded time points.
On homo-oligomeric RDL (resistant to dieldrin) GABA receptors, fluralaner showed no cross-resistance to classical GABA non-competitive antagonists. Gassel et al. [11] reported fluralaner to be a potent blocker of dieldrin-resistant D. melanogaster and C. felis RDL recombinant, homomultimeric receptors in cell-based fluorescence dye assays, with IC 50 values as low as 2.8 nM and 1.7 nM, respectively. In two-electrode voltage clamp (TEVC) recordings of oocytes expressing M. domestica RDL receptors, fluralaner had similar IC 50 values of 2.8 nM on rdl-type receptors with A299S mutation and 5.3 nM on wild-type receptors [10]. Fluralaner also had an IC 50 value of 12 nM on fipronil-resistant RDL receptors from a plant-feeding mite, Tetranychus urticae Koch, in TEVC experiments, whereas 30 µM of fipronil only blocked 27% of GABA-induced current [24]. In the present study, fluralaner was tested on native D. melanogaster GABA receptors, in situ, instead of heterologously expressed RDL receptors. In line with data on expressed RDL receptors, fluralaner showed similar potent activity on native dieldrin-resistant and -susceptible GABA receptors of D. melanogaster larval CNS, with EC 50 values of 0.29 µM and 0.34 µM, respectively. The lower potency in CNS preparations can be attributed to the need for penetration into the neuropile, as well as any differences between native and recombinant homomultimeric receptors. These results further document that fluralaner has potent activity on insect strains that are resistant to classical GABAergic compounds.
Gassel et al. [11] also reported 18-fold resistance to fipronil in D. melanogaster homo-oligomeric RDL GABA receptors (Ser isoform). In the present study, no resistance to fipronil was observed in RDL larval CNS preparations, at least when tested at 10 µM (Figure 8). We would note that this concentration is over 10-fold greater than the IC 50 for blocking the Ala isoform in homo-oligomeric receptors, reported to be 663 nM [11]. So, the resistance may be largely circumvented at this concentration.
Good selectivity of fluralaner between mammals and invertebrates also has been demonstrated in both in vivo and in vitro studies. Currently commercialized as a parasiticide, fluralaner was reported to be safe to dogs at or above recommended treatments [25,26]. In radioligand binding studies on rat brain membranes, 10 µM of fluralaner showed around 40% inhibition of radiolabeled 4-ethynyl-4-N-propylbicycloorthobenzoate (EBOB) binding, which is more than 2000-fold less sensitive than its binding to housefly GABA receptors [10,12]. Additionally, fluralaner had low activities on either recombinant β 3 homopentamers or α 1 β 2 γ 2 heteropentamers [10,12]. In the present study, mammalian GABA A α 1 β 3 γ 2 receptors were tested against fluralaner and fipronil. In line with previous reports, the IC 50 value of fluralaner was higher than 30 µM. Moreover, the IC 50 value of fipronil was 4.9 µM, suggesting fluralaner has a lower toxicity to mammals than fipronil.
The mode of action and toxicology of isoxazolines such as fluralaner are similar to another new insecticide class, the meta-diamides, which also work on invertebrate GABA receptors. According to Nakao et al. [27], meta-diamide 7 has potent activity on three mutant GABA receptors that are resistant to GABA receptor-directed non-competitive antagonists. Additionally, mutations G336M in M3, I277F and L281C in M1 reduce the activity of fluralaner on RDL GABA receptors, while having only a small impact on the activity of fipronil. Further molecular modeling suggests that meta-diamide binds to T9' to S15' region in M2, close to the avermectin target site and different from the classical convulsant site [28]. Homology modeling also suggests that fluralaner might share the same binding site as meta-diamide 7 [28]. According to most recent Mode of Action Classification Scheme from the Insecticide Resistance Action Committee (IRAC) [2], GABAergic insecticides have been categorized into Groups 2A (cyclodiene and organochlorines, i.e., chlordane), 2B (phenylpyrazoles, i.e., fipronil), while glutamate-gated chloride channel activators, avermectins and milbemycins, are in Group 6. As of this writing, isoxazolines and meta-diamides have not received a category from IRAC, but these compounds will probably be assigned to a new category because of their novel actions on RDL GABA receptors.

Conclusions
In this study, the toxicity of fluralaner against Ae. aegypti, An. gambiae, and D. melanogaster was assessed in various exposure routes. Compared to fipronil, fluralaner had a more slowly developing toxicity, and generally a 7-to 100-fold lower potency in adult bioassays. These findings suggest that fipronil would be a better overall mosquitocide than fluralaner in the absence of resistance. The data also imply that the moderate potency, low contact toxicity, and slow action of fluralaner might preclude its use as a mosquitocide for vector control, despite its favorable mammalian selectivity and lack of cross resistance in rdl carrying insects.
For larval assays on Ae. aegypti, the concentration leading to 50% paralysis of intact larvae was 10-fold higher than that on headless larvae, so the cuticle seems to be an important factor influencing fluralaner toxicity. Compared to fluralaner, fipronil was 10-to 13-fold less potent in larval assays, which differed from the results seen in adult mosquito bioassays.
In synergism assays, DEF was found to increase fluralaner toxicity by 3.8-8 fold in feeding assays on each day of exposure, with other synergists less active. This finding suggests that carboxylamidases might be involved in the metabolism of fluralaner in mosquitoes.
This study provides additional evidence for selectivity and lack of cross resistance of fluralaner. It was tested on mammalian GABA A α 1 β 3 γ 2 receptors, which gave an IC 50 value larger than 30 µM. Additionally, feeding and glass contact assays of fluralaner against susceptible and rdl strains of D. melanogaster showed similar activity, consistent with the equal sensitivity of larval CNS recordings towards fluralaner in both strains.