Boron-Pleuromutilins as Anti-Wolbachia Agents with Potential for Treatment of Onchocerciasis and Lymphatic Filariasis

A series of pleuromutilins modified by introduction of a boron-containing heterocycle on C(14) of the polycyclic core are described. These analogs were found to be potent anti-Wolbachia antibiotics and, as such, may be useful in the treatment of filarial infections caused by Onchocerca volvulus, resulting in Onchocerciasis or river blindness, or Wuchereria bancrofti and Brugia malayi and related parasitic nematodes resulting in lymphatic filariasis. These two important neglected tropical diseases disproportionately impact patients in the developing world. The lead preclinical candidate compound containing 7-fluoro-6-oxybenzoxaborole (15, AN11251) was shown to have good in vitro anti-Wolbachia activity and physicochemical and pharmacokinetic properties providing high exposure in plasma. The lead was effective in reducing the Wolbachia load in filarial worms following oral administration to mice.


■ INTRODUCTION
Parasitic nematodes of the family Filarioidea including Onchocerca volvulus, Wuchereria bancrofti, and Brugia malayi are responsible for significant disease burden in developing countries around the world. 1 Specifically, O. volvulus is the parasite responsible for River Blindness, and W. bancrofti, B. malayi, and Brugia timori cause lymphatic filariasis (elephantiasis). 2 Lymphatic filariasis is one of the leading causes of global disability and accounts for at least 2.8 million disabilityadjusted life years. Treatment/control and elimination programs for these infections have been in place for many years, but fall short of full effectiveness due to the demanding treatment paradigm required to break disease transmission. 3−5 More specifically, the long life span of adult worms (up to 15 years) requires annual to bi-annual mass drug administration of therapeutics such as ivermectin which only kills the juvenile microfilariae released by adult worms and sterilizes adult worms but is not macrofilaricidal. 5 Very recently, the results of a triple-drug treatment (ivermectin, albendazole, and diethylcarbamazine) clinical trial demonstrated that this combina-tion may require less frequent administration (perhaps once every 3 years), but only a marginal improvement of macrofilaricidal effects over a dual-drug therapy was noted. 6 Consequently, new approaches to kill the adult worms (macrofilaricides) are required in order for elimination timeframes of both diseases to be radically reduced. A short course of treatment (7 days or fewer) would likely be required for ease of implementation in the field.
A unique feature of these parasitic worms is the presence of obligate symbiotic bacteria of the Wolbachia genus. 7 It has been known for some time that classical antibacterial agents such as doxycycline can kill the bacteria present in the worms, which results in a reduced life span of the adult worm itself. 8−10 Unfortunately, doxycycline presents challenges as a drug for mass administration, including the requirement for long treatment periods (4−6 weeks) and contraindications in pregnancy and in children. 11 An anti-Wolbachia approach to the treatment of filarial infections has a number of patient benefits especially as the co-endemic eyeworm Loa loa does not harbor the endosymbiont and is therefore unaffected by treatment. It has been observed that concurrent killing of L. loa microfilariae by directly acting drugs in patients with >30 ;000 microfilariae per mL can have serious side effects including neurologic effects, coma, and death. 12 In addition, recent work has suggested that depleting Wolbachia in worms can also

Journal of Medicinal Chemistry
Article diminish the number of microfilariae that are able to develop in the insect vector, thus providing transmission blocking activity. 13 The underlying parasitology suggests that the discovery and development of new anti-Wolbachia drugs remains an attractive option for improving our ability to reduce the global burden of river blindness and elephantiasis and accelerate disease elimination goals.
As part of our ongoing effort to capitalize on the unique properties of benzoxaboroles in modifying pharmacologic, physicochemical, and pharmacokinetic properties of existing drug scaffolds, we prepared analogs of the antibiotic class known as pleuromutilins. 14,15 This class of ribosomal protein synthesis inhibitors predominantly targets Gram-positive bacteria and had been extensively explored since the 1950s, but had not previously been shown to have activity against Wolbachia. The pleuromutilins have been shown to inhibit protein synthesis through binding to the peptidyl-transfer center of the ribosome, with a crystal structure of tiamulin (4) with the 50S subunit of Deinococcus radiodurans, providing the most direct evidence of this mechanism of action. 16 Recently, Nabriva Therapeutics has been developing lefamulin (2), which is currently in clinical trials for community-acquired bacterial pneumonia, demonstrating the potential of this class of antibiotic. 17 Furthermore, the starting material for our explorations, pleuromutilin (1), is readily available and inexpensive with synthetic modification, particularly of the hydroxyacetate at C (14), being synthetically straightforward. In addition to boronated analogs of the pleuromutilin core, we also obtained or prepared several clinically relevant non-boron analogs as summarized in Figure 1. 18−21

■ RESULTS
The boronpleuromutilins modified at the C(14) position of the pleuromutilin core were easily prepared via an S N 2 displacement reaction of the pleuromutilin tosylate or iodide (Scheme 1). These intermediates were synthesized from the commercially available pleuromutilin. For compounds wherein the vinyl group at C (12) was modified, ozonolysis of the pleuromutilin tosylate with a reductive workup provided the C(12) aldehyde (Scheme 2). Displacement of the C(14) tosylate with 7-fluoro-6-hydroxybenzoxaborole afforded the boronpleuromutilin, which in turn was condensed with hydroxyl amines to provide the oximes, or with amines and sodium cyanoborohydride to provide the amines. Hydrogenation of the C(12) vinyl group of the boronpleuormutilin afforded the corresponding C(12) ethyl analog.
The boronpleuromutilins with a C(14)-carbamate linker were prepared in two steps from 4-epi-pleuromutilin (72) 22 in a manner analogous to that described in the literature (Scheme 3). 23 For example, following conversion of 72 to the carbamoyl chloride 73, treatment with 6-aminobenzoxaborole 78 followed by a zinc chloride-mediated 1,5-hydride shift gave the C(14)-carbamate 74. Compounds 75−77 were also prepared in a similar manner.
The boron-containing heterocycles were prepared by a variety of synthetic sequences that we have described in previous publications. 24 As an example, 7-fluoro-6-hydroxybenzoxaborole was prepared by the route shown in Scheme 4. Protection of 3,4-difluorophenol (79) as the methoxymethyl ether was followed by ortho-metallation and trapping of the intermediate aryllithium with ethyl formate to provide the benzaldehyde derivative 81. Nucleophilic displacement of the para fluorine with benzyl alcohol under basic conditions provided 82. Deprotection of the methoxymethyl ether followed by conversion to triflate 84 permitted introduction of the boron via palladium-mediated borylation. Reduction of the aldehyde 85 with sodium borohydride followed by treatment with aqueous hydrochloric acid resulted in formation of the benzoxaborole 86. Finally, hydrogenolysis of the benzyl ether gave the desired 7-fluoro-6-hydroxybenzoxaborole intermediate 87.
We screened the pleuromutilin drug candidates in two primary in vitro assays that differed in the strain of Wolbachia used, the insect cell line used to host the bacteria, and in the method of detection. 25,26 We used data from both assays to help ensure robust selection of compounds for progression to subsequent in vitro ADME and in vivo pharmacokinetics (PK) experiments. While the absolute potency of individual compounds varied between the two assays in some cases, the structure−-activity relationship (SAR) trends observed in these assays were generally similar. Initial screening revealed that several of the clinically relevant pleuromutilins, for example, retapamulin (3) and valnemulin (5), showed good activity in in vitro assays; significantly less activity was observed for lefamulin (2), tiamulin (4), or for the unadorned pleuromutilin core (1, Table 1). More interestingly, attachment of a benzoxaborole ring with either an oxygen (8), nitrogen (25), or sulfur (33) linker at the 6-position to the C(14) sidechain provided compounds (hereafter referred to as boronpleuromutilins) of similar in vitro potency to valnemulin and retapamulin. Introduction of a methylene spacer between the pleuromutilin core and the benzoxaborole ring as in 35 significantly reduced potency. Attachment of the benzoxaborole ring via the 5-position also provided compounds (37−39) with significantly reduced in vitro activity.
We next turned our attention to substitution of the benzoxaborole core, where we found that introduction of a fluoro substituent at the 4-(13), 5-(14) or 7-(15) positions was generally tolerated with the C(7) analog 15 exhibiting very potent activity. Other small substituents at C(7) such as methyl (12), chloro (16), or methoxy (17) were also tolerated, but did not improve activity beyond the fluoro analog. As the pleuromutilins are intrinsically very lipophilic, we also examined introduction of an aminomethyl substituent onto the benzoxaborole core (19−23, 26−29), but these analogs were generally less active than the corresponding neutral compounds. Finally, we found that addition of two methyl groups at C(3) of the benzoxaborole (24,30), a strategy that had improved pharmacokinetic properties in other series, 27 significantly reduced potency. Initial results with other boron-containing heterocycles linked via C(14) of the pleuromutilin core were encouraging, as a number of these analogs were very potent in the in vitro Wolbachia assay (Table 2). In particular, the N-methanesulfonyl (40), N-acetyl (42), and N-methylcarbamoyl (45) derivatives of the diazaborine scaffold exhibited low nanomolar to picomolar potency. Unlike the benzoxaboroles, however, inclusion of a fluorine atom adjacent to the B−OH did not improve potency; in fact, these derivatives (47−51) were significantly less potent. The benzoxaborine 53, wherein the five-membered ring was expanded by one carbon to a six membered ring, was completely inactive.
We also explored modification of the C(12) vinyl substituent (Table 3). Work in the classical pleuromutilin literature had shown that this double bond could be reduced to provide an ethyl group (54), converted to the epoxide (55) or oxidized to an aldehyde (58) that could be functionalized in a variety of ways. 28 In particular, we sought to introduce small amines at C (12) to modify physicochemical and pharmacokinetic properties. Given the interesting properties of 15 (vide supra), this work was performed with the 7-fluoro-6oxobenzoxaborole linked via to C(14) of the pleuromutilin core.
Disappointingly, only the ethyl (54) derivative exhibited potency close to the C(12) vinyl analog 15; oximes (60−62) and amines (63−70) were significantly less potent. Some potency was regained by acylation of the amine (71), but this was not perceived as being of sufficient advantage to pursue further.
As a final area of exploration, we prepared several boronpleuromutilins where the C(14)-hydroxyacetate linker was replaced by a shorter carbamate (Table 4). This strategy had previously been explored in the pleuromutilins as a means to improve oral bioavailability. 23 The carbamates prepared (74−77) exhibited moderate-to-good in vitro potency.
Concurrent with our evaluation of the in vitro potency of the boronpleuromutilins, we also selected a number of active analogs for characterization in both in vitro absorption,    50 (nM) Wolbachia infected LDW1 cells (wMel) EC 50 (nM) distribution, metabolism, excretion (ADME) and in vivo PK experiments. It has been suggested that one of the main limitations of the pleuromutilins is high metabolic instability leading to poor PK. 29,30 In fact, the majority of the clinically relevant pleuromutilins are of limited oral bioavailability, which has hampered their utility for human applications. A notable exception is the recently described lefamulin (2), which is currently in phase 3 clinical trials as an oral treatment for a variety of bacterial infections. 31,32 In order to calibrate our expectations for the boronpleuromutilins, we screened several of the clinically relevant analogs in our in vitro ADME assays as well ( Table 5). As anticipated, tiamulin and retapamulin were found to be rapidly metabolized by mouse liver S9 fraction, whereas lefamulin was slowly metabolized. These classical pleuromutilins exhibited a range of binding to mouse plasma proteins, with free fraction (f unbound ) from 0.05−0.26. Slightly surprisingly, we found that lefamulin was not particularly permeable in our MDR1-MDCK assay, though the high stringency of this assay may underestimate the permeability of this compound through the gut. We had chosen the MDR1-MDCK assay to assess permeability over the related Caco2 assay due to throughput and the opportunity to more directly assess the impact of P-glycoprotein-mediated efflux in this cell line. In addition, it has also been reported that the MDR1-MDCK is an acceptable assay for evaluation of the permeability of compounds through the intestinal mucosa of humans. 33 The neutral boronpleuromutilins evaluated (8−12, 14, 25, 33) were generally metabolized quickly by mouse S9 (Cl int > 50 μL/min/mg), were highly protein bound (f unbound < 0.03), and were predicted to have good permeability based on the MDR1-MDCK assay (P app > 5 × 10 −6 cm/s). In contrast, boronpleuromutilins bearing an aminomethyl substituent (19−21, 23, 26) were less rapidly metabolized (Cl int < 10 μL/min/mg), much less protein bound (f unbound > 0.1), but were predicted to be of low permeability (P app < 1 × 10 −6 cm/ s). Later compounds incorporating boron-containing heterocycles other than the benzoxaborole (40 −42, 44−46, 52, 53) or modification of C(12) (55, 59) did not significantly change the overall in vitro ADME profile. One compound that stood out as having an attractive balance of properties was the 7fluorobenzoxaborole derivative 15, which exhibited modest clearance (Cl int = 33 μL/min/mg) and protein binding (f unbound = 0.03), and good permeability (P app = 14.1 × 10 −6 cm/s).
Based on these in vitro profiles and due to resource constraints, we selected representative compounds from the various classes (e.g., neutral, aminomethyl), different linkers (e.g., O, N) and boron heterocycles (e.g., benzoxaboroles, diazaborines) for progression to in vivo PK experiments as summarized in Table 6. When the in vivo PK in mice of these representative compounds were measured, we observed that the PK parameters between them were variable following both IV and oral administration, consistent with the in vitro properties described above. Several observations were made from these studies. First, and as expected, clearance following intravenous dosing was generally quite high, with only compound 15 exhibiting low-to-modest clearance. Second, incorporation of an aminomethyl group on the benzoxaborole core (19) significantly reduced oral bioavailability, as did replacement of the benzoxaborole with a diazaborine (45,52).

Journal of Medicinal Chemistry
Article Finally, oral exposure and bioavailability was generally good for the 7-fluorobenzoxaborole derivatives tested (15, 32, and 77). Given the relatively low clearance and high exposure exhibited by 15, coupled with its good biological potency, additional PK experiments were performed with this compound. These confirmed the attractive PK profile following oral administration to mice of 15, as systemic exposures at doses from 10 to 50 mg/kg were good, with C max values increasing dose-dependently, while AUC values remained relatively constant, suggesting dissolution-limited absorption of the compound from the gut ( Table 7).
Based on this combination of in vitro potency and in vivo PK data, we next examined the ability of 15 to deplete the Wolbachia symbiont from filarial worms in vivo. In all in vivo efficacy assays, depletion of Wolbachia was determined through comparison to the Wolbachia load found in worms present in animals that were treated with only the dosing vehicle as described in the Experimental Section. For our initial in vivo screen, groups of three CB.17 SCID mice were inoculated with 50 L3-stage B. malayi larvae via the peritoneal cavity and treated from point of infection with 15 for 7 or 14 days with oral doses of 25 mg/kg, BID. At point of necropsy, on day 14, fourth stage (L4) B. malayi larvae were isolated from the peritoneal cavity. Larvae were pooled per group and intraworm Wolbachia titers were quantified from groups of 10 larvae using qPCR. Included in the same experiment were a vehicle control group and two groups of doxycycline-treated animals as positive controls. We were encouraged to find that 15 reduced the Wolbachia load in the B. malayi larvae by 75.4 and 98.8% in the 7-and 14-day treatment groups, respectively. In this experiment, doxycycline reduced Wolbachia load by 76.3 and 96.7% following 7-and 14-day treatment at 50 mg/kg, QD, respectively (Table 8).
Values represent percent (%) median reduction of Wolbachia versus the vehicle control, as measured by qPCR of Wolbachia surface protein single copy gene (wsp) per B. malayi L4 larva (n = 10 per group) obtained from necropsies undertaken 2 weeks after first dose.
Encouraged by the profile of 15, and in preparation for more extensive in vivo characterization of this molecule, we conducted a number of preliminary in vitro and in vivo studies to assess the safety of this molecule. In a non-GLP Ames assay (Bioreliance, Rockville, MD), 15 was determined to be nonmutagenic. Furthermore, 15 was found to be negative in its potential to induce micronuclei in human peripheral blood lymphocytes (Bioreliance, Rockville, MD). In a panel of over 50 mammalian receptors, enzymes and ion channels (Eurofins Cerep, France), 15 was found to be without any significant effect at 10 μM, and it was also shown to have no significant (<10%) effect on the hERG potassium channel expressed in HEK cells at a concentration of 30 μM (ChanTest, Cleveland, OH). Finally in a preliminary safety study (Anacor, Palo Alto, CA), no adverse effects were observed in BALB/c mice treated with 15 at doses up to 200 mg/kg/day for 7 days.
We next progressed 15 to chronic in vivo infection models of adult filarial parasitism (the target life cycle stage for an anti-Wolbachia indication); one using B. malayi and one using Litomosoides sigmodontis. We initially chose to use the B. malayi SCID mouse model 34 as this worm species is a causative agent of LF in humans. 2 In this B. malayi assay, 100 L3-stage larvae were inoculated via the peritoneal route and adult infections were allowed to develop. At 6-weeks post-infection, the mice were treated with 15 (25 mg/kg, BID) for 7, 14, or 28 days. At the end of the experiment (12-week post-infection), worms were recovered from the mice and Wolbachia load was determined by qPCR. In this experiment, vehicle and minocycline (25 mg/kg, BID × 28 days) groups were included as negative and positive controls, respectively. Minocycline was

Journal of Medicinal Chemistry
Article chosen as the positive control for this study based on concurrent results in an L. sigmondontis model (vide infra), which suggested that this tetracycline antibiotic exhibited superior efficacy to doxycycline. 35 Slightly disappointingly, efficacy of 15 was not nearly as good in the adult model, with a 45.5% reduction of Wolbachia load observed for the 28 day treatment group. By comparison, minocycline provided 98.2% reduction in this experiment (Table 9). We hypothesize that the failure of 15 to show efficacy at the dose chosen in the B. malayi adult worm model is likely due to the increased washout period between treatment start and necropsy in comparison to the B. malayi L3 larval model, leading to a rebound of the Wolbachia with a suboptimal treatment regimen of 15.
Values represent percent (%) median reduction of Wolbachia versus the vehicle control, as measured by qPCR of Wolbachia surface protein single copy gene (wsp) per adult female B. malayi (n = 10 per group) obtained from necropsies undertaken 6 weeks after first dose.
In a second efficacy model, 35,36 we used mice infected with the rodent filarial nematode L. sigmodontis, another helminth species known to host Wolbachia, and treated these animals with 15 at a dose of 50 mg/kg, BID for 14 or 28 days. The L. sigmodontis model was chosen as an additional, confirmatory filarial model as it uses immunocompetent wild-type mice as host and initial studies demonstrated the efficacy of tetracycline against Wolbachia-containing filariae first in this model. 37 In addition, the infection could be maintained in mice using a natural infection protocol. 35,38−40 Upon completion of dosing, animals were maintained for a total of 64−77 days (to maintain a consistent total time following the final dose), at which time worms were recovered from the peritoneum and the thoracic cavity. Wolbachia load was determined by qPCR. Doxycycline was included as a positive control in this experiment. We were encouraged to find that 15 was able to reduce Wolbachia burden in L. sigmodontis by >99% at this dose and duration (Table 10). The stronger reduction of the Wolbachia load in the L. sigmodontis mouse model in comparison to the adult B. malayi mouse model is probably mostly due to the increased dose used for treatment with 15, although L. sigmodontis were previously reported to have an increased susceptibility for anti-Wolbachia drugs compared to B. malayi filariae. 41 Values represent percent (%) reduction of Wolbachia versus the untreated control, as measured by qPCR FtsZ/actin per mouse, and were obtained from necropsies performed 1 month after first dose.

■ DISCUSSION AND CONCLUSIONS
Previous work from our laboratories has demonstrated that boron-containing heterocycles can be useful in the treatment of various diseases, including infectious diseases of the developing world such as malaria, 42,43 human African trypanosomiasis, 27,44 and tuberculosis. 45,46 We have extended our work to look at modification of known classes of antibacterial agents as an approach to treat river blindness via the bacterial symbiont (Wolbachia) present in the worms that cause this disease. Clear clinical "proof of concept" for an anti-Wolbachia approach has been demonstrated with the tetracycline antibiotic doxycycline. 9 The utility of doxycycline is limited, however, by the requirement that patients must be treated for 4−6 weeks in order to achieve significant reduction of Wolbachia in the O. volvulus parasite. 8−10 More recently, it has been demonstrated that combination of doxycycline (200 mg/day for 3 weeks) with albendazole (800 mg/day for 3 days) may provide a shorter treatment course. 47 While moving closer to the desired short course (7 day or fewer) treatment paradigm, this combination therapy is still not ideal and does not address tetracycline contraindicated groups (children of age ≤8 and pregnant women).
Our work started by evaluating representatives of various classes of antibiotics including several clinically relevant pleuromutilins. This initial screening revealed that some activity was present in this class of compounds, and we prepared several analogs containing our benzoxaborole core. The synthesis of these boronpleuromutilins was facilitated by the ready availability of the tosylate derivative of the hydroxyacetate ester at C(14) of the pleuromutilin core. Simple nucleophilic displacement reactions of this to afford C(14)-functionalized boronpleuromutilins with nitrogen, oxygen, or sulfur linker atoms proceeded in good yield. Early SARs revealed that direct attachment of the benzoxaborole ring to the pleuromutilin core provided compounds with good in vitro potency. In contrast, inclusion of a linker group such as present in the clinically relevant pleuromutilins between the core and benzoxaborole ring were of poor activity. Based on our previous work in the benzoxaboroles, we focused our attention on compounds linked through the 6-position, as these were both synthetically accessible and generally had been observed to have good ADME properties. Inclusion of a variety of substituents such as halogen, small alkyl, or aminoalkyl at positions 3, 4, 5, and 7 of the 6-O linked benzoxaboroles gave compounds of similar in vitro activity and allowed us to explore the effect of these substituents on ADME and PK. We were particularly pleased to find that the 7-fluoro analog (15) exhibited a good balance of potency and physicochemical and ADME properties. In particular, we were encouraged by the modest metabolic stability in mouse microsomes (Cl int = 33 μL/min/mg) and good permeability through an MDR1-MDCK monolayer (P app = 14.1 × 10 −6 cm/s). We were slightly concerned about the high protein binding of 15 (f unbound = 0.03), though this was predicted based on the calculated high lipophilicity (log D 4.23) of this compound. While other compounds, particularly those with an amino-

Journal of Medicinal Chemistry
Article methyl substituent on the benzoxaborole core (e.g. 20) exhibited better metabolic stability (Cl int = 6 μL/min/mg), they were compromised by very low permeability (P app < 0.1 × 10 −6 cm/s). When 15 was dosed to mice by the intravenous route (5 mg/kg), we were encouraged by the relatively low clearance (Cl int = 505 mL/hr/kg), good exposure (AUC 0−24h = 8.85 h/ μg/mL). Following oral administration (10 mg/kg), 15 exhibited good exposure (AUC 0−24h = 14.0 h/μg/mL) and bioavailability (F = 61%). Additional PK studies at higher oral doses demonstrated that exposure increased with dose, though not dose proportionally, to very high levels (AUC 0−24h = 58.7 h/μg/mL at 400 mg/kg). In addition to the parent compound, we also tracked the major (and active) metabolite of 15 (15a, IC 50 = 193 nM) where hydroxylation at C(2) of the pleuromutilin core had occurred, as this represented about 10−15% of the parent dose at most time points. Metabolism of pleuromutilins at this position of the core is well precedented. 48, 49 Based on these observations, 15 was selected for evaluation in several in vivo models that had been established for evaluation of anti-Wolbachia compounds in mice. In the first study, we found that 15, when dosed orally at 25 mg/kg, BID, to SCID mice infected with larval stage B. malayi, efficacy (defined as >99% reduction in Wolbachia load in worms) was achieved following 14 days of dosing; 7 days dosing did not achieve full efficacy. In a second B. malayi model employing worms that had been allowed to mature to the adult stage, we were disappointed to find that 15 did not demonstrate measureable efficacy when dosed at 25 mg/kg, BID for 28 days. More encouraging data were obtained in the L. sigmodontis BALB/c mouse model, where 15 was found to show excellent efficacy (>99% Wolbachia depletion) when dosed at 50 mg/kg, BID for 14 or 28 days. Additional in vivo studies to address this hypothesis are ongoing and will be reported in due course.
In conclusion, exploration of a series of pleuromutilin derivatives incorporating a novel boron-containing heterocycle linked to C(14) of pleuromutilin core as anti-Wolbachia agents has resulted in the identification of 15 as a potential preclinical candidate. SAR developed in this program has demonstrated that optimal activity was obtained by direct linkage of the benzoxaborole to the pleuromutilin core via C(6) of the benzoxaborole, and that small substituents on the benzoxaborole ring system had an impact on potency and pharmacokinetic properties. Good-to-excellent exposure following oral administration of 15 to mice was achieved, and this translated into high levels of Wolbachia depletion in several in vivo models. We were encouraged by the efficacy of 15 in these models, prompting more extensive evaluation of this compound, both alone and in combination with other known anti-Wolbachia drugs. The outcome of these studies will be reported in due course.

■ EXPERIMENTAL SECTION
General. All chemicals were purchased from commercial suppliers and used as received. 1 H NMR spectra were recorded on a Bruker AVANCE 400 spectrometer. Chemical shifts are expressed in δ ppm referenced to an internal tetramethylsilane (δ = 0 ppm) standard. Abbreviations used in describing peak signals are br = broad, s = singlet, d = doublet, dd = doublet of doublets, t = triplet, q = quartet, m = multiplet. All final compounds were purified to have purity higher than 95% by reverse phase high-performance liquid chromatography (HPLC), supercritical fluid chromatography, normal phase flash chromatography, or crystallization. Preparative HPLC was accomplished using a Luna C18 250 × 30 mm, 10 μm column. The purity was assessed by reverse phase HPLC with a gradient of 5−95% acetonitrile in water (with or without acid modifier) and monitored by a diode array ultraviolet detector at 220 and 254 nm. Lowresolution mass spectra were recorded on a liquid chromatography− mass spectrometer in electrospray positive (ESI+) or negative (ESI−) modes.
Following 7 days of sterile incubation at 26°C, staining mediacontaining SYTO 11 DNA dye was added to each well. After 15 min, all media was removed from each well and fresh media (no stain) was added. Imaging of each well was accomplished using a PerkinElmer Operetta high-content imaging system. Five fields per well were imaged using a confocal 60× objective with the fluorescein filter (excitation filter: 460−490 nm; emission filter: 500−550 nm). Images were analyzed using the PerkinElmer Harmony software to score each intact cell on the basis of texture complexity of the cytoplasm. Full details of the image analysis have been published. 25 Compound sample wells were analyzed and normalized (along with the positive controls) against the vehicle (untreated) control to give a percentage reduction of Wolbachia-infected cells. Using the cell number analysis, compounds with a host cell number amounting to less than 50% of the vehicle control were classified as toxic and retested at a reduced compound concentration. Dose−response curves were generated with percentage reduction of Wolbachiainfected cells versus compound concentration, using 5−10 compound serial dilutions. Data were analyzed and compound EC 50' s determined using a 4-parameter logistic nonlinear regression model. EC 50 is defined as the compound concentration producing a 50% reduction of Wolbachia in the C6/36 cell line.
Anti-wMel Wolbachia High-Content in Vitro Assay. The assay was developed based on previous work 25 and carried out as previously described. 50 Briefly, D. melanogaster LDW1 cells naturally infected with wMel strain of Wolbachia were maintained in Shields and Sang M3 (SSM3) Insect Medium (Sigma, S3652) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (qualified, One Shot Format, Gibco, 26140-111) at 25°C, in flasks with unvented caps. Greiner 1536-well assay plates (Part. no. 789091) were coated with 1 mg/mL solution of Concanavalin-A lectin (MP Biomedicals, 02195283) to facilitate cell adhesion. Compounds were acoustically transferred into plates using the Echo 555 Liquid Handler (Labcyte Inc.). Cells were seeded at 4000 cells/well in SSM3 medium supplemented with 2% FBS using the MultiFlo FX Multi-Mode Dispenser (Biotek). After incubating at 25°C for 6 days, the cells were fixed with 4% paraformaldehyde (PFA) for at least 10 min and washed with phosphate buffered saline, pH 7, 0.1% Tween 20. Fluorescence in situ hybridization (FISH) was used to stain Wolbachia and 3 μM DAPI was used to stain DNA. The "bottle valve" dispenser with an angled-head (Kalypsys Inc, San Diego, CA) was used for fixation and staining. Stained cells were imaged using the CX5 Insight Cellomics high-content imaging instrument with a 10×

Journal of Medicinal Chemistry
Article objective (Thermo). Each well was analyzed using Compartmental Analysis in HCS Studio (Thermo) for cell number and Wolbachia content. Analysis outputs included (1) the total object number per well and (2) the total fluorescence intensity of all spots within each cell (averaged per well). Data were analyzed in Genedata Screener, Version 13.0.1-Standard by normalizing to DMSO (neutral) and inhibitor control-treated wells (12.5 μM doxycycline and 125 nM rifampicin). To determine the EC 50 's, dose response curves (11′ point, 1:3 dilution) were fitted using the four parameter Hill equation. Each compound was tested in triplicate.
Ethics Statement on Animal Studies. All in vivo experiments involving B. malayi were conducted at the Liverpool School of Tropical Medicine (LSTM), were approved by the ethical committees of the University of Liverpool and LSTM, and were conducted according to Home Office (UK) requirements. All animal experiments with L. sigmodontis were conducted at the Institute for Medical Microbiology, Immunology and Parasitology (IMMIP), University Hospital of Bonn, in accordance to the European Union animal welfare guidelines. All protocols were approved by the local authorities (Landesamt fur Natur, Umwelt und Verbraucherschutz, Cologne, Germany (AZ 84-02.04.2015.A507; 84-02.04.2012.A140). In vivo pharmacokinetic studies in mice were conducted either at Anacor Pharmaceuticals or Shanghai ChemPartner under appropriate review by Institutional Animal Care and Use Committees (IACUC) at each site.
B. malayi SCID Mouse Infection Models. A B. malayi SCID mouse infection model for anti-Wolbachia drug screening was used as previously described 34,51−53 In brief, B. malayi L3 larvae were generated by membrane blood feeding Liverpool strain filarial susceptible Aedes aegypti mosquitoes with microfilariae isolated from patently infected Meriones unguiculatus gerbils. CB.17 SCID mice were purchased from Charles River UK. Groups of 3 or 5 SCID mice were infected with either 50 or 100 L3, ip, for larval or adult B. malayi efficacy testing, respectively. Drugs were dosed from point of infection for larval testing or after an incubation period of 6 weeks for adultstage testing with variable length washout. Test drugs were administered orally at indicated doses and as detailed below. Doxycycline or minocycline positive controls were administered in water at indicated doses. At 2-weeks (larval testing) or 12-weeks postinfection (adult testing) motile B. malayi L4 or female parasites were isolated by peritoneal lavage, enumerated by microscopy and individually snap frozen. Total genomic DNA was isolated per worm, and QPCR quantifications of single copy B. malayi Wolbachia were undertaken per group of 10 worms per treatment/control, as previously described. 34,54 Efficacy of Wolbachia depletion is reported as median Wolbachia load per drug group expressed as a percentage of median vehicle control levels from animals treated only with vehicle.
L. sigmodontis in Vivo Mouse Model. Female BALB/c wild type mice were obtained from Janvier (Saint-Berthevin, France) and were housed at the animal facility of the IMMIP in individually ventilated cages on a 12 h light/dark cycle with food and water ad libidum. Mice were infected at 6−8 weeks of age via exposure to the natural mite vector Ornithonyssus bacoti containing infectious L. sigmodontis L3 larvae. 36 To compare the infection, the same batch of mite-containing bedding was used to infect all animals of one experiment. Then, 35 days post infection, time point adult worms have developed within the thoracic cavity, and mice were treated per oral gavage with the test compounds at concentrations indicated in the result section. As a vehicle, 1% CMC + 0.1% Tween80 was used. Negative controls were treated with vehicle only, and as a positive control, mice were treated orally with doxycycline (Sigma) in 10% DMSO/1× PBS. Necropsies were performed at 64−77 days postinfection using an overdose of isoflurane (Baxter, Germany) and adult worms were isolated from the thoracic cavity and peritoneum and enumerated. Realtime PCR was performed from remaining female adult worms to quantify Wolbachia FtsZ and L. sigmodontis actin values by qRT-PCR as previously described. 35 General PK Study Protocol. Animals were allowed to acclimate a minimum of 72 h prior to selection for study. On the morning of dosing, animals were weighed and grouped randomly. Animals received the test material by tail-vein injection (IV), or oral gavage (PO). After dosing, all clinical signs and observations were recorded, and if adverse effects were noted, the exact time that the animals returned to normal was also recorded. At prescribed timepoints, animals were anesthetized by isoflurane inhalation and blood samples were collected via retro-orbital bleed or cardiac puncture. After the blood sample collection, animals were euthanized by CO 2 inhalation followed by cervical dislocation.
Blood samples were collected into microtainer tubes (K 2 EDTA) on ice and processed to provide plasma by centrifugation for 6 min at approximately 2000g or 6000 rpm. Processed samples were transferred to analysis tubes that were placed on dry ice. Samples were analyzed by LC/MS/MS.

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jmedchem.8b01854.
Molecular formula strings (SMILES) and biochemical and in vitro ADME data (CSV)