Discovery and Characterization of a Potent and Selective Inhibitor of Aedes aegypti Inward Rectifier Potassium Channels

Vector-borne diseases such as dengue fever and malaria, which are transmitted by infected female mosquitoes, affect nearly half of the world's population. The emergence of insecticide-resistant mosquito populations is reducing the effectiveness of conventional insecticides and threatening current vector control strategies, which has created an urgent need to identify new molecular targets against which novel classes of insecticides can be developed. We previously demonstrated that small molecule inhibitors of mammalian Kir channels represent promising chemicals for new mosquitocide development. In this study, high-throughput screening of approximately 30,000 chemically diverse small-molecules was employed to discover potent and selective inhibitors of Aedes aegypti Kir1 (AeKir1) channels heterologously expressed in HEK293 cells. Of 283 confirmed screening ‘hits’, the small-molecule inhibitor VU625 was selected for lead optimization and in vivo studies based on its potency and selectivity toward AeKir1, and tractability for medicinal chemistry. In patch clamp electrophysiology experiments of HEK293 cells, VU625 inhibits AeKir1 with an IC50 value of 96.8 nM, making VU625 the most potent inhibitor of AeKir1 described to date. Furthermore, electrophysiology experiments in Xenopus oocytes revealed that VU625 is a weak inhibitor of AeKir2B. Surprisingly, injection of VU625 failed to elicit significant effects on mosquito behavior, urine excretion, or survival. However, when co-injected with probenecid, VU625 inhibited the excretory capacity of mosquitoes and was toxic, suggesting that the compound is a substrate of organic anion and/or ATP-binding cassette (ABC) transporters. The dose-toxicity relationship of VU625 (when co-injected with probenecid) is biphasic, which is consistent with the molecule inhibiting both AeKir1 and AeKir2B with different potencies. This study demonstrates proof-of-concept that potent and highly selective inhibitors of mosquito Kir channels can be developed using conventional drug discovery approaches. Furthermore, it reinforces the notion that the physical and chemical properties that determine a compound's bioavailability in vivo will be critical in determining the efficacy of Kir channel inhibitors as insecticides.


Introduction
Mosquitoes are vectors of protozoan, filarial nematode, and viral pathogens that cause numerous human diseases, including malaria, lymphatic filariasis, and dengue fever. These diseases impose an enormous burden on global health and profoundly impair socioeconomic advancement in developing countries [1]. The overuse of a limited number of insecticides has led to the emergence of insecticide-resistant populations of mosquitoes, which is hampering the effectiveness of vector control efforts [2,3,4]. Consequently, there is a need to identify new molecular targets against which insecticides can be developed and deployed.
An emerging body of evidence from our group supports the idea that inward rectifier potassium (Kir) channels represent viable targets for insecticide development [5,6,7]. Kir channels are tetrameric proteins that conduct K + ions across the cell membrane and thereby generate an ionic current that underlies various cellular functions. Dipteran insects possess three major Kir channel subtypes, denoted Kir1, Kir2 and Kir3. In Drosophila melanogaster, there are three genes that encode Kir channels (DrKir1, DrKir2, DrKir3), which play important roles in osmoregulation, immunity, and development [8,9,10,11]. In Aedes aegypti, there are five Kir channel genes (AeKir1, AeKir2A, AeKir2B, AeKir2B' and AeKir3), which are expressed in various body segments and tissues such as the carcass (thorax and abdomen), head, Malpighian tubules, midgut, and hindgut [6,12]. We showed previously in vitro that the A. aegypti Kir1 (AeKir1) channel mediates strong inward rectifying K + currents that are blocked by barium and the small molecule inhibitors, VU573 and VU590 [7,12,13]. Moreover, a hemolymph injection of either VU573 or VU590 inhibits the excretion of urine by adult female mosquitoes, leads to abdominal bloating, and incapacitates mosquitoes within 24 h [5].
Taken together, the above studies indicate that Kir channels represent promising molecular targets for insecticides that have a novel mechanism of action by disrupting the renal-dependent regulation of extracellular fluid homeostasis (i.e., renal failure). However, in mammals, Kir channels regulate the electrical excitability of neurons and cardiac cells, hormone secretion, and transport of K + ions across epithelial tissues of the kidney and gut [14]. Missense mutations that perturb the activity of Kir channels cause human diseases of the heart, nervous system, pancreas, and kidney [15,16,17]. Thus, efforts aimed at developing insecticides to target Kir channels must verify that lead compounds do not perturb the functions of mammalian Kir channels.
As such, the above 'tool' compounds VU573 and VU590 allowed us to establish proof-of-concept, but are not suitable for insecticide development, in part, because they inhibit mammalian Kir channels with greater potency than AeKir1 [18,19]. Here, we aim to discover new chemical probes of AeKir1 channels that exhibit improved potency and selectivity compared to the tool compounds by optimizing and validating an existing fluorescent thallium (Tl + ) flux-based assay of AeKir1 function [5] for highthroughput screening (HTS) of small molecule libraries. Screening approximately 30,000 small molecules from the chemical library of the Vanderbilt Institute of Chemical Biology (VICB) resulted in the identification of 283 compounds with activity against AeKir1 channels. We focus on the in vitro and in vivo activity of one of these compounds, N-(3-methoxyphenyl)-2-methyl-1-propionylindoline-5-sulfonamide (VU625), which exhibits nanomolar affinity and is highly selective for AeKir1 over mammalian Kir channels.

Materials and Methods
Tl + flux assays Tl + flux assays were performed essentially as described previously [13,18,19]. Briefly, stably transfected T-Rex-HEK-293 cells expressing AeKir1 channels were cultured overnight in 384-well plates (20,000 cells/20 mL/well black-walled, clearbottomed BD PureCoat amine-coated plates (BD, Bedford, MA) with a plating media containing DMEM, 10% dialyzed FBS and 1 mg/mL tetracycline. The next day, the cell culture medium was replaced with a dye-loading solution containing assay buffer (Hanks Balanced Salt Solution with 20 mM HEPES, pH 7.3), 0.01% (w/v) Pluronic F-127 (Life Technologies, Carlsbad, CA), and 1.2 mM of the thallium-sensitive dye Thallos-AM (TEFlabs, Austin, TX). Following 1 hr incubation at room temperature, the dye-loading solution was washed from the plates and replaced with 20 mL/well of assay buffer.
The plates were transferred to a Hamamatsu Functional Drug Screening System 6000 (FDSS6000; Hamamatsu, Hamamatsu (or Bridgewater, NJ), Japan) where 20 mL/well of test compounds in assay buffer (as prepared below) were added and allowed to incubate with the cells for 20 min. After the incubation period, a baseline recording was collected at 1 Hz for 10 s (excitation 470620 nm, emission 540630 nm) followed by a Tl + stimulus buffer addition (10 mL/well) and data collection for an additional 4 min. The Tl + stimulus buffer contains in (mM) 125 NaHCO 3 [19].
Tl + flux data were analyzed as previously described [19,21,22] using a combination of Excel (Microsoft Corp, Redmond, WA) with XLfit add-in (IDBS, Guildford, Surrey, UK) and OriginPro (OriginLab, Northampton, MA) software. Raw data were opened in Excel and each data point in a given trace was divided by the first data point from that trace (static ratio) followed by subtraction of data points from control traces generated in the presence of vehicle controls. The slope of the fluorescence increase beginning 5 s after Tl + addition and ending 15 s after Tl + addition was calculated.
2-methyl-1-(2,2,2-trifluoroacetyl)indoline-5-sulfonyl chloride: Chlorosulfonic acid (22 mL, 330 mmol, 9 equiv.) was added to a 100 mL round bottom flask equipped with a reflux condensor, and cooled to 0uC. To this, 2,2,2-trifluoro-1-(2methylindolin-1-yl)ethan-1-one, (8.5 g, 37 mmol, 1 eq.) was added dropwise. The reaction mixture was removed from the ice bath. The vial was heated to 40uC for 1 hour. The reaction was subsequently cooled to room temperature and PCl 5 (7.7 g, 37 mmol, 1 equiv.) was added slowly. After gas evolution ceased, the reaction mixture was heated to 80uC for 1 hour. The reaction mixture was cooled to room temperature and then placed in an ice bath. Water was added very slowly to the reaction mixture. Subsequently, DCM was added and the reaction was filtered through a phase separator. The organic layer was concentrated under reduced pressure and used without subsequent purification (6.46 g, 53%).

Patch clamp electrophysiology
T-REx-HEK293-AeKir1 cells were voltage clamped in the whole-cell configuration of the patch clamp technique after overnight induction with tetracycline (1 mg/ml) essentially as described earlier [5]. Briefly, patch electrodes were pulled from silanized 1.5 mm outer diameter borosilicate microhematocrit tubes using a Narishige PC-10 two-stage puller. Electrode resistance ranged from 3.5 to 5.5 MV when filled with the following intracellular solution (in mM): 135 KCl, 2 MgCl 2 , 1 EGTA, 10 HEPES free acid, 2 Na 2 ATP (Roche, Indianapolis, IN), pH 7.3, 275 mOsm. The standard bath solution contained (in mM): 135 NaCl, 5 KCl, 2 CaCl 2 , 1 MgCl 2 , 5 glucose, 10 HEPES free acid, pH 7.4, 290 mOsm. Whole-cell currents were recorded under voltage-clamp conditions using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA). Electrical connections to the amplifier were made using Ag/AgCl wires and 3 M KCl/agar bridges. Electrophysiological data were collected at 5 kHz and filtered at 1 kHz. Data acquisition and analysis were performed using pClamp 9.2 software (Axon Instruments). All recordings were made at room temperature (20-23uC).

Heterologous expression of AeKir1 and AeKir2B in Xenopus oocytes
AeKir1 and AeKir2B channels were expressed heterologously in Xenopus laevis oocytes as described previously [6]. In brief, defolliculated Xenopus oocytes (purchased from Ecocyte Bioscience, Austin, TX) were injected with 10 ng (0.35 ng/nL) of either AeKir1 or AeKir2B cRNA and cultured for 3-7 days in OR3 media at 18uC. Oocytes injected with 28 nl of nuclease-free H 2 O served as controls.

Electrophysiology of Xenopus oocytes
All electrophysiological experiments on Xenopus oocytes were performed at room temperature. The compositions of the solutions used in these experiments are shown in Table 1. When present, VU625 was dissolved in solution III or solution V to a final concentration of 0.1, 1, 5, 15, or 50 mM (0.05% DMSO). All solutions were delivered by gravity to a RC-3Z oocyte chamber (Warner Instruments, Hamden, CT) via polyethylene tubing at a flow rate of ,2 ml/min. Solution changes were made with a Rheodyne Teflon 8-way Rotary valve (Model 5012, Rheodyne, Rohnert Park, CA).
Electrophysiological recordings from oocytes were conducted as described previously [6] In brief, each oocyte was transferred to the holding chamber under superfusion with solution I and impaled with two conventional-glass microelectrodes backfilled with 3 M KCl (resistances of 0.5-1.5 MV) to measure membrane potential (V m ) and whole-cell membrane current (I m ), respectively. Current-voltage (I-V) relationships of oocytes were acquired as described previously [6]. In brief, the oocytes were subjected to a voltage-stepping protocol consisting of 20 mV steps from 2140 mV to +40 mV (100 ms each). After the conclusion of the voltage-stepping protocol, the clamp was turned off and a new solution was superfused through the chamber for ,90 s before acquiring another I-V relationship. All V m and I m values were recorded by a Digidata 1440A Data Acquisition System (Molecular Devices) and the Clampex module of pCLAMP. The I-V plots were generated using the Clampfit module of pCLAMP.
To evaluate the inhibition of AeKir1 and AeKir2B activity by VU625, we focused on the maximal inward currents elicited by the voltage-stepping protocol, which occur at a voltage of 2140 mV. For AeKir1 oocytes, the background, inward currents in solution II (i.e., 0.5 mM K + ) were subtracted from those in 1) solution III (i.e., 10 mM K + ) to calculate the total inward current for an oocyte before exposure to VU625 (I A ), and 2) solution III with VU625 to calculate the inward current after exposure to the small molecule (I B ). The percent inhibition of the inward current was calculated by subtracting I B from I A and then dividing by I A . For AeKir2B oocytes, a similar protocol was followed and similar calculations were made, except solution IV replaced solution II and solution V replaced solution III.

Mosquito colony
The Aedes aegypti mosquito colony used in the present study is identical to that described previously [6]. As before, only adult female mosquitoes 3-10 days post emergence were utilized for experiments.

Mosquito toxicology experiments
Adult female mosquitoes for injection were anesthetized on ice and impaled through the metapleuron using a pulled-glass capillary attached to a nanoliter injector (Nanoject II, Drummond Scientific Company, Broomall, PA). Each mosquito received a single hemolymph injection of 69 nL of solution. The injection solution consisted of a potassium-rich phosphate buffered saline (K + -PBS), 15% DMSO, 1% b-cyclodextrin, 0.1% Solutol, and a concentration of VU625 to deliver the doses indicated. In experiments where probenecid was used, water-soluble probenecid (Biotium, Hayward CA) was included in the injection solution at 50 mM, thereby providing a dose of 3.4 nmol per mosquito.
The K + -PBS solution consisted of the following in mM: 92.2 NaCl, 47.5 KCl, 10 Na 2 HPO 4 , and 2 KH 2 PO 4 (pH 7.5). A total of 10 mosquitoes were injected for a given treatment or dose, and then were placed into small cages within a rearing chamber (28uC, 80% relative humidity, 12:12 light:dark) and allowed free access to a solution of 10% sucrose. The mosquitoes were observed at 24 hr after injection. For each treatment, 3-7 replicates of 10 mosquitoes each were performed.

Mosquito excretion experiments
The excretory capacity of mosquitoes was measured as described [6]. In brief, after anesthetizing mosquitoes on ice, their hemolymph was injected as described above with 900 nL of a K + -PBS vehicle containing 1.15% DMSO, 0.077% b-cyclodextrin, and 0.008% Solutol, or the vehicle containing VU625 (0.77 mM) to deliver a dose of 690 pmol of VU625 per mosquito. In experiments where probenecid was used, the vehicle was supplemented with water-soluble probenecid (3.08 mM) to deliver a dose of 3.4 nmol of probenecid per mosquito. After injection, the mosquitoes were placed immediately in a graduated, packed-cell volume tube (MidSci, St. Louis, MO; 5 mosquitoes per tube) and held at 28uC. The volume of urine excreted at 60 min post injection was measured as described previously [6], and all mosquitoes were confirmed to be alive at the end of 60 min period. For each treatment, 6-18 independent trials of 5 mosquitoes per treatment were performed.

Statistical analyses
Tl+ flux assay. The Z9 value was calculated as described earlier [21], using the following formula: where SD is standard deviation, p and n are vehicle control and compound inhibited flux values respectively.
To compare the effect of DMSO on AeKir1-mediated Tl + flux, a one-way ANOVA was performed with a Tukey's multiple comparison test. Prism software (GraphPad Software) was used to generate CRC from Tl + flux. Half-inhibition concentration (IC 50 ) values were calculated from fits using a four parameter logistic equation.
Mosquito toxicology and urine excretion. Prism (Graph-Pad Software) was used to generate a dose-response curve for the toxicity of VU625; the doses (x-axis) were first log transformed and then the mortality data was normalized using Prism, where the smallest value and largest values in a data set equal '0%' and '100%', respectively. The data were then fitted using a 'biphasic' algorithm (,100 constraint) to calculate potencies (ED 25 and ED 75 values). To compare 1) the toxic effects among the vehicle, probenecid, VU625, and VU625 + probenecid treatments, and 2) the excretory capacity among the vehicle, probenecid, VU625, and VU625 + probenecid treatments, one-way ANOVAs were performed with Newman-Keuls posttests.

Discovery of novel AeKir1 inhibitors via HTS
In an effort to discover mosquito-specific inhibitors of AeKir1, we optimized a Tl + flux assay for HTS of large libraries of chemically diverse small molecules. The assay utilizes a monoclonal T-REx-HEK293 cell line that expresses AeKir1 from a tetracycline-inducible promoter [5]. The fluorescent dye, Thallos, is used to report the inward flux of Tl + through the AeKir1 channel pore in a population of cells plated in individual wells of a 384-well plate. As shown in Figure 1A, overnight induction of AeKir1 expression with tetracycline leads to a robust Tl + flux compared to control cells that were not treated with tetracycline. This assay enables more than 300 compounds to be tested simultaneously in a single plate, and thousands of compounds to be tested daily, for effects on AeKir1 activity.
The assay was validated for HTS by meeting a series of performance benchmarks. First, the assay was tested for its tolerance to the small-molecule vehicle DMSO at concentrations up to 10% v/v. As shown in Fig. 1B, the Tl + -flux mediated by AeKir1 is unaffected by DMSO concentrations up to 1.3% v/v as compared to the 0% DMSO control (one-way ANOVA, P , 0.0001). Next, the assay was tested for uniformity and reproducibility of HTS performance. As shown in Fig. 1C, the average Z9 statistic for these experiments was 0.6960.05 (Z9$0.5 is suitable for HTS), indicating that the assay is robust and will enable modulators of AeKir1 to be identified in HTS with a low falsepositive rate.
Approximately 30,000 compounds from the VICB library were screened at a nominal concentration of 10 mM for inhibition of AeKir1. From this primary screen and following confirmation testing in tetracycline-induced and uninduced T-REx-HEK293-AeKir1 cells (see Methods), 283 authentic channel-dependent modulators were selected for further study. Because our ultimate goal is to develop Kir channel inhibitors that are active against mosquitoes and not humans, these 'hits' were subsequently tested for dose-dependent activity against a panel of mammalian Kir channels, which included Kir1.  [18,19,21]. Fourpoint concentration response curves (CRCs) were generated for the 283 compounds, resulting in 17 inhibitors with 11 unique chemical scaffolds that exhibited dose-dependent inhibition of AeKir1 with IC 50 values below 5 mM and little to no activity (IC 50 $30 mM) against mammalian Kir channels (data not shown). These compounds were subsequently purchased from commercial vendors, freshly dissolved in DMSO, and assayed in 11-point CRCs against AeKir1 via the Tl + -flux assay. From the aforementioned Tl + flux assays, one compound-N-(3-methoxyphenyl)-2-methyl-1-propionylindoline-5-sulfonamide, termed VU625-was found to inhibit AeKir1 in 11-point CRCs with an IC 50 of 0.32 mM (95% CI: 0.25-0.39 mM) and a Hill coefficient value of 0.98 (95% CI: 0.8-1.2) ( Fig. 2A-C). VU625 also had no significant effects on the mammalian Kir channels assayed via Tl + flux with the exception of G-protein coupled Kir channels comprised of Kir3.1/3.2 subunits (IC 50 = 8.6 mM; Table S1). Furthermore, in radioligand displacement assays against 68 mammalian GPCR's, ion channels, and transporters, 10 mM VU625 was active (defined as.50% ligand displacement) against only three targets: adenosine A1 receptor (76% displacement), melatonin MT1 receptor (56% displacement) and 5-HT 2B receptor (69% displacement) (Table S2).
To further confirm the activity of VU625 obtained from Tl +flux assays, we used patch-clamp electrophysiology to assay the inhibition of AeKir1 expressed in T-REx-HEK293 cells. In wholecell patch clamp recordings, VU625 inhibited AeKir1 channel activity with an IC 50 of 96.8 nM (95% CI: 75.4-124.2 nM) and a Hill coefficient value of 1.02 (95% CI: 0.8-1.3) (Fig. 3A, B).
In a previous paper, we demonstrated that other small molecule inhibitors of AeKir1 (i.e., VU573 and VU590) can have different pharmacological effects on AeKir2B [6]. Thus, we sought to determine the effects of VU625 on AeKir2B channel activity, utilizing Xenopus oocytes heterologously expressing AeKir2B. AeKir1 expressing oocytes served as positive controls. Figure 3C shows that VU625 inhibits AeKir1and AeKir2B-mediated K + currents with IC 50 values of 3.8 mM (95% CI: 2.3-6.3 mM) and 45.1 mM (95% CI: 31.7-64.2 mM), respectively. Thus, VU625 inhibits both AeKir1 and AeKir2B channels, albeit with greater affinity for AeKir1. It should be noted that the reduction in VU625 potency observed in Xenopus oocytes compared to HEK cells is typical for a small-molecule inhibitor of Kir channels and has been observed for structurally diverse compounds and Kir channels [5,6,19,23].  Chemical lead optimization and structure-activity relationships Because of its potency, clean ancillary pharmacology and chemical tractability (Figures 2-3, Tables S1-S2), VU625 was selected for lead optimization (3a, Table 2). We partitioned the compound into three areas for structure-activity relationship (SAR) exploration denoted as the sulfonamide, central core, and southern amide portions (Fig. 4A). The first generation libraries held the sulfonamide and the central core sections constant and diversified the southern amide portion ( Table 2). The synthetic scheme (Fig. 4B) for this portion was straightforward and started with protection of the amine with trifluoracetamide (TFAA, pyridine) followed by sulfonyl chloride formation (ClSO 3 , PCl 5 ). Next, the sulfonamide was formed, the protecting group was removed, and either the amide or sulfonamide was formed (see Methods for details). Little tolerance for steric bulk was seen in this portion of the molecule. That is, the trifluoroacetamide (VU0477197, 3b, Table 2) retained potency (0.58 mM), however, larger aromatic amides were much less active (3c-g, Table 2). The same trend was observed for the sulfonamide compounds, with smaller sulfonamides retaining nanomolar activity (VU0477691, 3k, 0.76 mM; VU0477692, 3l, 0.82 mM) and the larger aromatic group leading to less activity (3h, Table 2).
Next, we evaluated the left-hand sulfonamide portion of the molecule, however, all efforts to change the 3-methoxyaryl moiety led to significant reductions in potency (see Table S3). Finally, we explored the central core with the intent of establishing the minimal pharmacophore needed for activity against AeKir1. To this end, the indoline core was replaced with simple aryl, heteroaryl or biaryl groups which all led to compounds with much reduced activity (.10-fold loss of potency). However, an interesting SAR was seen with very closely related 6,6-or 6,5indole or dihydroquinolinone-like structures (4a-f, Table 3). The simple N-methyl indole (VU0481807, 4a, 0.55 mM, Table 3) retained most of the activity as VU625 and addition of a 2-methyl (VU0486620, 4b, 0.97 mM, Table 3) led to a further minor reduction in activity. Expanding the ring system and addition of a lactam (4c-e, Table 3) was not productive. Lastly, removal of the methyl group in the indoline system of VU625, led to a ,3-fold loss of potency (VU0483404, 4f, 1.15 mM, Table 3).  The left-hand sulfonamide portion offered the least amount of SAR traction as only the 3-methoxyphenyl (and weaker 2methoxyphenyl) sulfonamide provided activity. Replacement with an amide, or substituting the 3-methoxyphenyl for other aryl groups all led to less potent compounds. Replacement of the proponamide in the southern fraction was tolerated as long as the substituent was small and aliphatic. Sulfonamides could be exchanged, although there was an observed ,2-3-fold loss of activity. Lastly, the central core was also important for potency. Only very similar compounds such as indole and des-methyl indoline were tolerated.

VU625-induced toxicity is increased by probenecid
Injection of the AeKir1 channel inhibitors VU573 or VU590 into the hemolymph of adult female A. aegypti mosquitoes leads to their incapacitation and/or death within 24 h [5,6]. Surprisingly, injection of a high dose (i.e. 690 pmol per mosquito) of VU625, which is a more potent inhibitor of AeKir1 than VU573 and VU590, into the hemolymph of mosquitoes had no significant effects on mosquito behavior or survival within 24 h (Fig. 6). Thus, we hypothesized that the lack of in vivo effects could be due to poor bioavailability of VU625 as a result of metabolic detoxification and/or excretion by the mosquito.
We therefore tested whether probenecid, which is a broadspectrum inhibitor of organic anion transporters (OATs) and ATP-binding cassette (ABC) transporters [24,25,26], would improve the efficacy of VU625. Interestingly, both probenecid and VU625 have a sulfonamide moiety in their chemical structure (Fig. S2). As shown in Fig. 6, the injection of probenecid (3.4 nmol per mosquito) along with VU625 (690 pmol per mosquito) significantly increases the toxicity of VU625 within 24 h compared to injection of VU625 or probenecid alone. The abdomens of these mosquitoes were not severely bloated and obvious sub-lethal effects (e.g., loss of flight) were not apparent.
We next sought to characterize the dose-response relationship of VU625 in mosquitoes when co-injected with a constant dose of probenecid (3.4 nmol per mosquito). As shown in Fig. 7, coinjection of VU625 with probenecid induces mortality in mosquitoes within 24 h in a biphasic manner with 25% and 75% efficacious doses (ED 25 and ED 75 ) of 9.96 pmol and 502 pmol, respectively. This biphasic dose-response relationship suggests the inhibition of at least two distinct molecular targets for which VU625 has different affinities, which is consistent with the inhibition of both AeKir1 and AeKir2B channels by VU625 (Fig. 3C).

VU625-induced reduction of urine excretion is enhanced by probenecid
We showed previously that pharmacological inhibition of AeKir1 with the small molecule inhibitors VU573 and VU590 leads to a decrease in the excretory capacity of A. aegypti mosquitoes after loading their hemolymph with 900 nl of a PBS vehicle [5] [6]. Therefore, we sought to similarly determine the effects of VU625 on mosquito excretory capacity. As shown in Fig. 8, mosquitoes injected with the PBS vehicle excreted 644624.18 nL of urine within the next hour. Consistent with the toxicity studies, we found that adding VU625 (0.77 mM) to the vehicle, which delivers 690 pmol of VU625 per mosquito, did not significantly decrease the excretory capacity (583.3629.52 nL/ female) compared to the vehicle controls (Fig. 8). Interestingly, adding probenecid (50 mM) to the vehicle, which delivers 3.4 nmol of probenecid per mosquito, causes a small but significant reduction in excretory capacity to 467.8633.53 nL/ female, suggesting a potential role of probenecid-sensitive transporters in urine excretion. However, adding both VU625 (0.77 mM) and probenecid (50 mM) to the vehicle significantly decreases the excretory capacity the furthest to 236.7624.53 nL/ female.

Discussion
Here, we report the discovery of VU625, the first submicromolar inhibitor of a mosquito Kir channel. VU625 is one of 283 confirmed AeKir1 inhibitors identified in a HTS of approximately 30,000 compounds from the VICB library. It was chosen for lead optimization based on its potency (IC 50 = 96.8 nM), greater than 80-fold selectivity for the AeKir1 channel over 8 mammalian Kir channels, and clean ancillary pharmacology among a panel of 68 critical mammalian off-targets comprised of voltage-gated ion channels, ion transporters, and receptors (i.e., neurotransmitter, peptide, and G-protein coupled). VU625 is the most potent and selective mosquito Kir channel inhibitor reported to date.
This study provides proof-of-concept that conventional drug discovery approaches can be employed successfully to identify small-molecule tools for probing the physiology of insect Kir channels and potential lead compounds for insecticide development. A similar approach has been used recently in insecticide discovery efforts targeting mosquito G-protein coupled receptors [27].
VU625 exhibits inhibitory activity against both AeKir1 and AeKir2B, albeit with greater affinity for AeKir1. To date, we have reported the activity of two other small-molecule inhibitors of mosquito Kir channels that exhibit differential pharmacology. VU590 is a selective inhibitor of AeKir1 over AeKir2B, whereas VU573 inhibits AeKir1 and activates AeKir2B [6]. Thus, VU625 potentially represents a broad-spectrum, small-molecule blocker of mosquito Kir channels, pending the characterization of its effects on the other mosquito Kir channels (AeKir2A, AeKir2B' and AeKir3 channels), which to date have not yet been expressed functionally in a heterologous system ( [12]; Denton and Piermarini, personal observations). Once the distinguishing pharmacological properties of each of these Kir channel inhibitors are fully characterized, they can potentially be employed to determine the relative contributions of Kir channel subtypes in the physiology of various mosquito tissues. This would provide an important chemical tool set to validate and complement studies of mosquito Kir channels that employ functional genetic approaches (e.g., RNA interference). Given the superior in vitro potency of VU625 compared to the AeKir1 inhibitors VU573 [5] and VU590 [6], we expected VU625 to elicit superior in vivo toxicity. Thus, we were surprised when high doses of VU625 elicited no observable effects on mosquito survival or excretory capacity when injected directly into the hemolymph. Since mosquitoes have evolved robust protective mechanisms for detoxifying and excreting xenobiotics that would harm them otherwise [28,29], we investigated whether the molecule may be detoxified and/or excreted.
Preliminary experiments with PBO did not improve the efficacy of VU625, suggesting that detoxification of the compound by cytochrome P450s is unlikely to contribute to its poor in vivo efficacy. The co-injection of VU625 with probenecid rescued not only its toxicity, but also its effects on excretory capacity, which suggests that VU625 is likely a substrate of OATs and/or ABC transporters in the mosquitoes and may be rendered ineffective in vivo through excretion. The potent toxicity of VU625 when coinjected with probenecid may be due to the ability of VU625 to inhibit at least two Kir channels, some of which are expressed in the central and peripheral nervous systems, such as Kir1 and Kir2B' [8,12,30,31], and/or a synergistic effect of probenecid that maintains high circulating concentrations of VU625 by preventing its renal excretion. Indeed, it is conceivable that the sulfonamide moiety in the structures of VU625 and probenecid causes them to be substrates for OATs and/or ABC transporters. Overall, these findings highlight efficient xenobiotic transport mechanisms in mosquitoes that render a nanomolar inhibitor of AeKir1 (VU625) ineffective in vivo, even when introduced directly to the hemolymph. The tissues that contribute to the excretion of VU625 remain to be determined, but presumably involve the Malpighian tubules and/or gut [32,33].
Lastly, the medicinal chemistry efforts put forth in the present study may be a valuable first step in determining which structural moieties are important for the excretion of VU625 by xenobiotic transporters and/or its in vivo activity in mosquitoes. Future Table 3. Structure-activity relationships and lead optimization summary for the central core portion of VU0077625 scaffold. studies should assess the in vivo efficacy and probenecid-mediated clearance of the VU625 analog series we generated to determine if any of these compounds exhibit potent toxicity in mosquitoes without probenecid.

Perspectives
Here, we show a direct relationship between in vitro pharmacology and in vivo toxicity of VU625, which is consistent with our previous studies [5,6] suggesting that Kir channel inhibitors are promising chemicals for insecticide development. To date, none of the Kir channel inhibitors we have reported (i.e. VU573, VU590, VU625) exhibit toxicity when applied to the cuticle (Piermarini, unpublished observations), which is a waxy, lipophilic structure that creates a physical barrier to insecticide permeation into the hemocoel of mosquitoes. This lack of topical activity severely limits the potential use of the present Kir channel inhibitors as active compounds for incorporation into insecticide-treated bed nets and indoor-residual sprays. The efficacy of common insecticides, such as permethrin, is dependent in part on their lipophilic nature [34,35]. Thus, future chemistry efforts will focus on lipophilic inhibitors of Kir channels. Furthermore, prioritizing initial HTS 'hits' according to their hydrophobicity may facilitate the discovery of more suitable small-molecules compounds for insecticide development.