A Cell-Based Screen Reveals that the Albendazole Metabolite, Albendazole Sulfone, Targets Wolbachia

Wolbachia endosymbionts carried by filarial nematodes give rise to the neglected diseases African river blindness and lymphatic filariasis afflicting millions worldwide. Here we identify new Wolbachia-disrupting compounds by conducting high-throughput cell-based chemical screens using a Wolbachia-infected, fluorescently labeled Drosophila cell line. This screen yielded several Wolbachia-disrupting compounds including three that resembled Albendazole, a widely used anthelmintic drug that targets nematode microtubules. Follow-up studies demonstrate that a common Albendazole metabolite, Albendazole sulfone, reduces intracellular Wolbachia titer both in Drosophila melanogaster and Brugia malayi, the nematode responsible for lymphatic filariasis. Significantly, Albendazole sulfone does not disrupt Drosophila microtubule organization, suggesting that this compound reduces titer through direct targeting of Wolbachia. Accordingly, both DNA staining and FtsZ immunofluorescence demonstrates that Albendazole sulfone treatment induces Wolbachia elongation, a phenotype indicative of binary fission defects. This suggests that the efficacy of Albendazole in treating filarial nematode-based diseases is attributable to dual targeting of nematode microtubules and their Wolbachia endosymbionts.


Introduction
Wolbachia are intracellular maternally transmitted bacteria present in the majority of all insect species as well as some mites, crustaceans and filarial nematodes [1,2]. Wolbachia were initially studied in insects because they induce unconventional reproductive phenotypes including sperm-egg cytoplasmic incompatibility, feminization of males, male-killing, and parthenogenesis [3,4]. Wolbachia are essential endosymbionts of some filarial nematodes and recent studies demonstrated that they are the causative agent of African river blindness and also contribute to lymphatic filariasis [5,6]. One sixth of the world population is at risk of infection by Wuchereria bancrofti, Brugia timori and Brugia malayi, the filarial nematode species that cause lymphatic filariasis [7]. Wolbachia released from filarial nematodes into the human body trigger an inflammatory reaction that underlies the lymphedema and corneal occlusion associated with these neglected diseases [8,9,10,11,12,13,14,15,16,17,18].
Lymphatic filariasis and African river blindness have traditionally been treated through the administration of three drugs, singly or in combination: diethylcarbamazine (DEC), ivermectin (IVM) and albendazole (ALB). These drugs target the filarial nematodes associated with these diseases, namely Onchocerca volvulus, B. timori, B. malayi, and W. bancrofti [5,6,19]. DEC disrupts the nematodes by targeting the arachidonic acid metabolic pathway in the host [20]. IVM disrupts glutamate-gated chloride channels in the nematode that control release of excretory/secretory vesicles that would normally suppress the immune response [21,22]. ALB is a benzimidazole used to disrupt the nematode microtubule cytoskeleton [23]. Orally administered ALB is rapidly metabolized by in the intestinal mucosa and liver into albendazole sulfoxide (ALB-SO) and albendazole sulfone (ALB-SO2) [24,25]. ALB-SO is normally considered to be the ''active,'' form of Albendazole against systemic parasites, while ALB-SO2 is considered to be an inactive form of the drug [26]. All three drugs exhibit microfilaricidal effects [19]. The macrofilaricidal effects of ALB are not clear, though specific formulations induce worm sterility in animal models [27]. In addition, a number of clinical trials demonstrate that ALB when used in combination with DEC or IVM is macrofilaricidal [28,29,30,31,32].
Wolbachia are obligate symbionts of filarial nematodes required for normal embryogenesis, larval development and perhaps most significantly adult survival [33,34,35,36,37,38,39,40,41]. A recent study demonstrated that loss of Wolbachia in the adult results in high levels of apoptosis throughout the nematode [37]. Studies have also found that much of the pathology associated with filarial nematode diseases is due to induction of innate and adaptive host immune responses upon release of Wolbachia from their nematode hosts [8,9,10,11,12,13,14,15,16,17,18]. These discoveries suggest that compounds directly targeting Wolbachia may be a powerful alternative to the more traditional approaches for treating these diseases. The major advantage of this approach is that it targets adults as well as microfilaria and the Wolbachia will be eliminated prior to death of the nematode, reducing the destructive effects of the human immune response. In addition, loss of Wolbachia leads to a slow death of the adults, providing time for the infected individual to clear the dead nematodes without deleterious side effects [35,42,43,44,45,46,47]. Finally, antihelminthic drugs such as ivermectin, administered to patients co-infected with Loa Loa nematodes, can potentially trigger lethal encephalitis [48]. Loa loa does not require or maintain Wolbachia and thus will not be affected by anti-Wolbachia therapies, thereby avoiding these deleterious side effects [49]. The promise of the anti-Wolbachia based therapies in combating lymphatic filariasis has been demonstrated in clinical trials in which daily doses of doxycycline (DOX) for 4 weeks resulted in nematode sterility and death [6]. In addition, the pathologies associated with the infection, lymphedema and hydrocele, were dramatically reduced [8,9].
These studies also revealed that a three-week course of DOX was insufficient to produce significant mortality of the adult nematodes, highlighting the need to identify more potent anti-Wolbachia compounds [50]. To this end, we generated a Wolbachiainfected Drosophila cell line and conducted an automated, cellbased screen to identify lead compounds that reduced intracellular Wolbachia infection. This screen of two libraries totaling 4926 compounds yielded 40 anti-Wolbachia compounds, including several that structurally resembled ALB. Our follow-up testing indicated that ALB-SO2 directly targets Brugia by depolymerizing its microtubules. Here we demonstrate that ALB-SO2 also disrupts Brugia Wolbachia independently of its effects on the Brugia cytoskeleton. Furthermore, ALB-SO2 treatment of Brugia led to visibly elongated Wolbachia morphology indicative of a binary fission failure, consistent with a direct impact of ALB-SO2 upon B. malayi Wolbachia.

Development of a Drosophila cell line constitutively infected with Wolbachia
To identify compounds that affect intracellular Wolbachia titer, we generated new Drosophila tissue culture cells constitutively infected with Wolbachia [51] (see Methods). There is currently no stable nematode tissue culture line, nor any type of cell line constitutively infected with Wolbachia derived from filarial nematodes. Wolbachia-infected insect cells provide an opportunity to identify drugs that disrupt Wolbachia through conserved molecular pathways. The cell line used for this study, JW18, is particularly amenable to high throughput screening as Wolbachia are maintained in approximately 90% or more of the host cell population (n = 1053 cells scored). The Wolbachia load in 6.7% of infected cells ranges from 1-46 bacteria, while Wolbachia load in the other 93% of infected cells is obscured by crowding of the bacteria (n = 205 cells scored). The mitotic index of JW18 and tetracycline-cured JW18 cells, henceforth referred to as JW18TET, was 0.27% and 0.68% respectively, which are not significantly different according to Chi square tests (n = 1876 and 2339). Furthermore, no significant difference was observed in the frequency of binucleate cells between JW18 (9.1%, n = 873) and JW18TET cells (11%, n = 1081). Thus, Wolbachia do not exert an obvious influence on the regulatory or structural mechanisms governing progression of the cell cycle in the JW18 cell line. However this analysis does not preclude more subtle cycle cell effects.
To test whether Wolbachia exhibit normal interactions with the host cytoskeleton, we took advantage of the fact that this cell line constitutively expresses a GFP-Jupiter fusion protein that binds to and labels microtubules [52] (Figure 1). During interphase, Wolbachia are closely associated with Jupiter-GFP-labeled microtubules ( Figure 1A-B, Video S1). Live imaging demonstrates that Wolbachia move processively along those interphase microtubules (Video S2), consistent with earlier reports of Wolbachia-microtubule interactions [53,54,55,56]. During mitosis, Wolbachia were asymmetrically distributed throughout the cytoplasm in 82% of cells (n = 56, Figure 1C-F, Table S1), reminiscent of Wolbachia localization patterns observed in embryonic and larval neuroblasts [56]. These data indicate that Wolbachia distribution in the JW18 cell line is consistent with that of intact Drosophila tissues.

Automated cell based screening for anti-Wolbachia compounds
High-throughput cell-based screens using automated microscopy have proven effective in identifying new compounds targeting specific biological processes [57,58]. Here we used the JW18 cell line in a 384 well format to screen 2000 compounds from the Spectrum Collection and 2926 compounds from the National Cancer Institute for reduction of Wolbachia titer. These libraries include structurally and functionally diverse synthetic compounds, FDA-approved drugs, compounds with biological activity and a set of natural products. After a 5-day incubation period, the cells were fixed, stained, and imaged using automated robotics (Figure S1, Materials and methods). Customized software was used to analyze the images and score the percentage of Wolbachia-infected cells in each well. Treatment wells showing significant reduction of Wolbachia-infected cells in at least 2 of 3 replicates as compared to the JW18 control and the entire cell population in general were scored as preliminary hits. Compounds known to be generally hazardous were excluded, resulting in a finalized hit list.
18 compounds from the Spectrum library, and 22 compounds from the NCI library exhibited consistent, potent anti-Wolbachia activity in the screen ( Figure 2, Table S2). A number of these

Author Summary
Wolbachia-based neglected diseases currently threaten one sixth of the world population. Millions of people are infected with filarial nematodes that rely upon endosymbiotic Wolbachia bacteria for their survival. These Wolbachia ultimately induce an immune response that gives rise to African river blindness or lymphatic filariasis. Thus, targeting Wolbachia will prevent induction of disease symptoms while also eliminating the filarial nematode infection. To identify new, fast-acting anti-Wolbachia drugs, we tested candidate compounds in Wolbachiainfected insect cells using automated robotics. Some of the anti-Wolbachia compounds we discovered closely resembled Albendazole, an FDA-approved drug already used worldwide to combat filarial nematode infections. We found that a common Albendazole metabolite strongly suppresses Wolbachia density in fruit flies and filarial nematodes by disrupting Wolbachia growth. These findings suggest that Albendazole is effective in treating filarial nematode-based diseases because it independently targets both the nematode and its essential Wolbachia endosymbionts. This has immediate implications for treating lymphatic filariasis and African river blindness.
compounds are already known to exert antimicrobial activity, consistent with expectations from the assay. For example, one of the hits CID484401 (totarol acetate) is a known inhibitor of the essential bacterial division protein, FtsZ [59]. CID42640 is a DNA damaging agent that has been shown to inhibit Mycobacterium tuberculosis [60]). Significantly, the hit CID313612 is structurally similar to CID42640. CID16524 (pyronin B) is a quaternary ammonium compound, many of which serve as the antimicrobial agents in commercial disinfectants [61]. An additional hit compound, CID6364517 has also been shown in a prior screen to exert antibacterial activity against Streptococcus pyrogenes (Pubchem BioAssay AID1900 and AID1915, conducted by the Broad Institute). Furthermore, a number of chemotherapy drugs were identified by the screen: CID30323 (daunorubicin), CID31703 (doxorubicin), CID65348 (epirubicin) and CID4212 (mitoxantrone) were identified as hits, along with the derivitives CID5351490 (cinerubin B) and CID27590 (mitomycin B). Many of these drugs were originally isolated from bacteria in nature. It is presumed that these compounds are used to gain a competitive advantage over neighboring bacteria. As such, this class of drugs is referred to as to ''anticancer antibiotics'' [62,63,64,65].
Doxycycline (DOX) did not come up as a hit in this screen. This was unexpected as DOX has proven to be an effective anti-Wolbachia reagent in lab and clinical settings [6]. However, treating the same Wolbachia strain with DOX in the context of Drosophila oogenesis revealed marked Wolbachia susceptibility to DOX ( Figure 3A-C). This suggests that the mechanism accounting for this difference in DOX efficacy is dependent upon properties of the host cell rather than the Wolbachia strain. It may be the efflux pumps in the JW18 cell line are particularly effective at expelling DOX. Alternatively, the bacteriostatic effects of DOX may not readily be detected in our assay because of the relative growth rates of Wolbachia and the JW18 host cells. Regardless of the mechanism, it is not general, as this screen yielded a number of new potent anti-Wolbachia compounds.
To further investigate the hit compounds revealed by the screen, we first examined their structures. This revealed that the benzimidazole CID5382764 and the benzthiazoles CID5458770 and CID5351210 share structural similarity to ALB, the widely used, FDA-approved anthelmintic drug known for disrupting helminthic microtubules ( Figure 4). The hit compounds were also assessed for cytotoxicity by consulting prior screens run by the NCI against human cancer cell lines and measuring the impact of the compounds on cell proliferation in our own assay. This indicated that 11 of 22 hits from the NCI library and 7 of 18 hits from the Spectrum library exhibit unacceptable toxicity levels, while the remaining 22 hit compounds exhibited low or unknown cytotoxicity ( Figure 2, Table S2). By this measure, two of the ALBlike compounds, CID5458770 and CID5382764, were indicated to be non-toxic ( Figure 2).

Albendazole-like compounds reduce Wolbachia titer in Drosophila oogenesis
Drosophila has proven a particularly effective model system for in vivo compound testing of a wide array of biological processes [66].
Here we used the Drosophila oocyte as a secondary screen for retesting the ALB-like hits identified in the cell-based screen. This system provides the advantages that the Wolbachia strain is the same as in the JW18 cell line, and Wolbachia titer in the oocyte is known to greatly increase between stage 3 and 10A, approximately a 40-hour period [54,67,68]. Starved adult Drosophila were fed yeast paste containing compounds at a concentration of 100 uM for 24 hours. Stage 10A oocytes were fixed and labeled, followed by imaging and quantification of their Wolbachia as described in the Materials and Methods. Average titer measurements from single focal planes have been shown to be representative for comparing Wolbachia titer between different conditions [67]. DMSO control oocytes exhibited 268+/214.3 Wolbachia within a single oocyte focal plane ( Figure 3A-C). Oocytes treated with DOX exhibited 141+/212.4 Wolbachia (p,.001). Treatments with CID5458770 yielded oocytes displaying 174+/212.7 Wolbachia (p,.001). Furthermore, CID5382764-treated oocytes exhibited 161+/215.9 Wolbachia (p,.001, Figure 3). Thus, both of the nontoxic, ALB-like hits identified by the cell screen deplete Wolbachia titer in Drosophila oogenesis similarly to DOX.
These results motivated us to determine whether ALB and its common metabolites, ALB-SO and ALB-SO2 ( Figure 4), also affect Wolbachia titer in vivo. Wolbachia counts from single oocyte focal planes indicated that ALB and ALB-SO treatments did not significantly affect Wolbachia titer, with 216+/222.5 and 299+/ 230.0 Wolbachia detected per oocyte, respectively ( Figure 3C). However, ALB-SO2-treated oocytes exhibited markedly less Wolbachia than the control, with 167+/214.8 Wolbachia evident per oocyte (p,.001, Figure 3). This indicates that the ALB-SO2 metabolite exerts anti-Wolbachia effects in Drosophila.

Albendazole sulfone disrupts Wolbachia titer in B. malayi
To investigate whether the anti-Wolbachia effect of ALB-SO2 applies to a disease model, we treated adult B. malayi nematodes in vitro. After a 3-day incubation period, Wolbachia were imaged in the distal tip region of the Brugia ovary referred to as the ''mitotic proliferation zone'' due to enriched replication of host nuclei in this tissue [69]. Wolbachia densely populate this region of the distal ovary in DMSO controls ( Figure 3D). By contrast, Wolbachia titer is visibly depleted in worms treated with either ALB-SO2 or DOX ( Figure   Albendazole sulfone has no impact on microtubule organization in Drosophila A consistent Wolbachia-disrupting effect for ALB-SO2 raises the question of the mechanism of action of this metabolite. ALB and other benzimidazoles are known to bind to beta-tubulin and disrupt microtubule polymerization [70,71,72,73]. Previous studies from Drosophila have demonstrated that Wolbachia titer is affected by host microtubules [54,67]. This raises the possibility that the reduction of Wolbachia titer upon exposure to ALB-SO2 is due to an impact of this compound on microtubule organization. Prior mutant studies identified key amino acids within beta tubulin that are important for the microtubule-disrupting impact of benzimidazoles. Residues N165 and Y200 are thought to form a hydrogen bond that restricts accessibility to a benzimidazole binding site [74]. Interestingly, most of the beta tubulin homologs in Drosophila encode the residues that correspond to benzimidazole resistance ( Figure S2). This suggests that the Drosophila microtubule cytoskeleton should be unaffected by ALB-SO2 treatment.
To test whether ALB-SO2 affects microtubule organization in Drosophila oogenesis, a combination of approaches was used. Intracellular Wolbachia localization was examined, taking advantage of the prior finding that Wolbachia concentrate at the oocyte posterior cortex in a microtubule-dependent manner [55,75]. In this study, posterior Wolbachia localization was detected in 95% of DMSO controls and 93% of ALB-SO2-treated oocytes (n = 56 and 46, respectively, Figure 5G-H). This differed significantly from oocytes treated with the microtubule-disrupting drug, colchicine, where only 21% exhibited posterior Wolbachia localization (p,.001, n = 13, Figure 5I) [55]. The Drosophila oocyte cytoskeleton was also directly examined by immunolabeling microtubules [76]. Stage 10B oocytes are known to undergo large-scale, microtubule-dependent cytoplasmic streaming, coincident with formation of microtubule bundles. As cytoplasmic streaming is a highly dynamic process, this bundling varies somewhat between oocytes, and changes within individual oocytes over time [76,77]. DMSO controls and oocytes treated with ALB-SO2 exhibited microtubule bundling at stage 10B, while the cytoplasm of colchicine-treated stage 10B oocytes was devoid of filamentous structure ( Figure 5J-L) [77]. Thus, ALB-SO2 did not affect the overall orientation or structure of the oocyte microtubule cytoskeleton in Drosophila, though this compound dramatically decreases Wolbachia titer ( Figure 3). This indicates that ALB-SO2 reduces Wolbachia titer through a microtubule-independent mechanism in Drosophila.

Albendazole sulfone disrupts Wolbachia titer in Brugia via a microtubule-independent mechanism
The findings above raise the question of whether ALB-SO2 disrupts Wolbachia titer independently of microtubules in B. malayi, analogous to Drosophila. The beta-tubulin genes of B. malayi carry the amino acid changes of N165S and Y200F that are associated with susceptibility to benzimidazoles like ALB-SO2 ( Figure S2). To assess the impact of ALB-SO2 on B. malayi microtubules in vivo, whole-mount immunostaining was performed. Examination of Brugia hypodermal chords revealed a dense meshwork of microtubules in DMSO-treated controls ( Figure 5A-C). In contrast, treatment with ALB-SO2 disrupted the Brugia microtubule cytoskeleton, although linear remnants remained visible throughout the hypodermal chord ( Figure 5D-F). This demonstrates that ALB-SO2 disrupts much of the host microtubule cytoskeleton in B. malayi.
To next determine whether the titer of Brugia Wolbachia relies upon host microtubules, we treated B. malayi with colchicine. To verify that colchicine disrupts microtubules in B. malayi, immunostaining was performed in the hypodermal chords of colchicinetreated worms as above. This analysis revealed that microtubule structure and organization was largely eliminated by colchicine treatment. To then assess the impact of colchicine on Wolbachia titer, Wolbachia were imaged in the distal ovary. Interestingly, colchicine-treated worms exhibited an average of 5.1+/20.93 Wolbachia per host nucleus (n = 193 bacteria scored), a value that is not significantly different from its DMSO controls by Chi square test ( Figure 3G). This indicates that microtubule disruption has little impact on Wolbachia titer in B. malayi. This suggests that ALB-SO2 is also unlikely to suppress Wolbachia titer through its microtubule-disrupting effects, and alternatively implicates a microtubule-independent mechanism for ALB-SO2 suppression of Wolbachia titer in B. malayi.
Albendazole sulfone induces Wolbachia elongation in B. malayi ovary To further pursue the mechanism by which ALB-SO2 disrupts Wolbachia, we examined its impact on Wolbachia morphology in Brugia tissue. In the mitotic proliferation zone of the distal ovary, ALB-SO2 treatment corresponded to elongation of Wolbachia nucleoids as compared DMSO controls ( Figure 6C-F). Wolbachia were also immunostained with an antibody raised against Wolbachia FtsZ, which crisply defines the boundaries of the Wolbachia cytoplasm [78]. The FtsZ staining also revealed an elongated Wolbachia morphology in ALB-SO2-treated Brugia relative to the DMSO control ( Figure 6G-H). Furthermore, quantification of Wolbachia length indicated that only 2.5% of Wolbachia in the DMSO control exceeded 2.5 mm in length, whereas 17% of Wolbachia in ALB-SO2 treated worms exceeded this length (n = 201 and 126, respectively) (p,.001) ( Figure 6I). This demonstrates that ALB-SO2 induces Wolbachia elongation in the Brugia ovary.
Wolbachia were also examined in the hypodermal chords, which are syncytial somatic tissues that run along the entire length of the Brugia body wall. Comparing Wolbachia in DMSO and ALB-SO2treated hypodermal chords revealed little difference in bacterial morphology ( Figure 6A-B). This indicates that Wolbachia in the hypodermal chord do not respond to ALB-SO2, in contrast to Wolbachia in the distal ovary of B. malayi.

Wolbachia elongation occurs independently of host microtubules
A role for ALB-SO2 in inducing Wolbachia elongation could be due to an indirect effect from disrupting host microtubules, or through an alternative mechanism. To distinguish between these possibilities, B. malayi ovaries were treated with colchicine and assessed for Wolbachia length. Examination of DMSO vs. colchicine-treated Brugia reveals no obvious differences in Wolbachia morphology between conditions ( Figure 6J-K). Wolbachia length was also quantified, and neither DMSO nor colchicinetreated tissues harbored Wolbachia exceeding 2.5 mm in length ( Figure 6I). This differs significantly from ALB-SO2-treated Wolbachia, which frequently exceeded 2.5 mm in length (p,.001) ( Figure 6I). Thus, overall microtubule disruption in Brugia did not change Wolbachia morphology as was seen for ALB-SO2 treatment. This indicates that ALB-SO2 induces Wolbachia elongation independently of host microtubules in Brugia malayi.

Discussion
Here we report the first high-throughput cell-based screen using automated microscopy to identify anti-Wolbachia compounds. The cell line used in this screen was established from primary culture preparations from Wolbachia-infected Drosophila embryos bearing the GFP-tagged microtubule binding protein Jupiter (Figure 1). Wolbachia have been stably maintained and transmitted in this line for years. Live analysis of interphase JW18 cells reveals that Wolbachia closely associate with and move along microtubules (Video S1), consistent with previous studies showing that Wolbachia   transport and positioning are microtubule-dependent [54,55,56]. Although there is variability from generation to generation, approximately 90% of JW18 cells are infected with variable amounts of bacteria. This is consistent with previous studies of Wolbachia-infected Drosophila, Spodoptera frugiperda and Aedes albopictus cell lines showing widely variable bacteria loads per cell [79,80,81,82,83,84]. Perhaps this variation in Wolbachia load is due in part to asymmetric partitioning of Wolbachia in mitotic JW18 cells (Figure 1, Table S1). This same Wolbachia strain has previously been reported to segregate asymmetrically in mitotic embryonic Drosophila neuroblasts and Aa23 tissue culture cells [56,82]. In addition, variants have been identified in D. simulans in which asymmetric segregation of Wolbachia may occur in the germline stem cells of the oocyte [85]. This contrasts with other evidence that Wolbachia partition evenly during mitosis in early Drosophila embryogenesis and in A. albopictus C7-10R tissue culture cells [56,83]. Future work is needed to address the underlying mechanisms that govern segregation of Wolbachia in mitotic cells.
Our automated, image-based screens of 4926 compounds from the NCI and Spectrum libraries revealed 40 compounds that depleted Wolbachia titer. Of these, 22 were classified as either nontoxic or of unknown toxicity (Figure 2, Table S2). Among the more characterized hit compounds, we identified totarol acetate, an inhibitor of the key bacterial division protein, FtsZ. This finding is consistent with a previous report describing FtsZ as a potential target for anti-Wolbachia therapy [86]. We also identified pararosaniline pamoate and pyrvinium pamoate as Wolbachiadepleting hits. These drugs have been employed as antihelmintics to treat patients infected with roundworms like Enterobius vermicularis and the trematode Schistosoma japonicum [87,88]. Our result raises an additional possibility that these hosts rely upon endosymbiotic bacteria to support their viability, and an antibacterial effect of these drugs thus indirectly compromises the host. No other Secernentea outside of the Filariodea have been reported to carry Wolbachia to date, though the Rhabditida order is the only one to have been systematically tested [89]. However, the Trematodes Acanthatrium oregonense and Nanophyetus salmincola have been reported to harbor the mutualistic bacteria Neorickettsia risticii and N. helminthoseca, respectively [90]. Furthermore, estradiol was also identified by our screen as a Wolbachia-depleting compound. This finding may help to explain why women exhibit symptoms of lymphatic filariasis less often than men in endemic areas, where 10% or fewer of females exhibit symptoms as compared to 10-50% of men [91]. Since Wolbachia can induce a TLR-2 and TLR-6-mediated inflammatory response analogous to that observed in lymphatic filariasis patients, the literature invokes a contribution of Wolbachia to the pathology of lymphatic filariasis [8,9,10,15,16,17]. Perhaps higher estradiol levels in female patients help to suppress Wolbachia load, thereby preventing TLR-2/6 induction and development of lymphatic filariasis symptoms.
An examination of the largely uncharacterized hits from the JW18 screen revealed that CID5458770, CID5382764 and CID5351210 structurally resemble ALB (Figure 4). Follow-up studies revealed that the non-toxic ALB-like compounds CID5458770 and CID5382764 exhibited Wolbachia-depleting activity in Drosophila oogenesis as well. This raised the possibility that ALB or its metabolites may have a similar activity in vivo. These studies surprisingly revealed that the ALB-SO2 metabolite depleted intracellular Wolbachia titer in Drosophila oogenesis, while ALB and ALB-SO did not (Figure 3). Furthermore, ALB-SO2 disrupts Wolbachia titer and morphology and alters host microtubules in B. malayi (Figure 3, 5, 6). These findings contrast with routine descriptions in the literature of ALB-SO2 as an inactive metabolite. It appears that ALB-SO2 inactivity has been interpreted from the lower abundance of the ALB-SO2 metabolite in human serum and urine relative to ALB-SO [24,25,26] as well as the relatively weaker ability of ALB-SO2 to compete against unlabeled benzimidazoles for binding to Haemonchus contortus tubulin in vitro [72]. However, the sequence of Brugia beta tubulin predicts susceptibility to benzimidazoles like ALB-SO2 (Figure 4, Figure S2), and evidence from other non-nematode systems supports a role for ALB-SO2 as an active metabolite. An in vitro competitive binding assay using colchicine showed that ALB-SO2 is slightly better at disrupting polymerization of Ascaris suum tubulin than either ALB or ALB-SO [92]. Studies from the tapeworm Echinococcus multilocularis showed ALB-SO2 to be similarly effective to ALB-SO in inducing structural defects and worm lethality. Other studies of the microsporidian parasites Encephalitozoon cuniculi, E. hellem and E. intestinalis indicated that ALB-SO2 is at least 5 times more effective at inhibiting growth than either ALB or ALB-SO [93]. Building upon those prior findings, this study is the first we are aware of to definitively show an active role for the ALB-SO2 metabolite in filarial nematodes.
A surprising outcome from this study was that ALB-SO2 disrupts Wolbachia independently of its effects on host microtubules. Previous studies have shown that host microtubules and microtubule-based motor proteins facilitate replication of Salmonella and Chlamydia [94,95,96,97,98,99,100,101]. Prior work in Drosophila indicates that normal levels of Wolbachia rely in part upon intact host microtubules as well [54,67]. This study confirms that Brugia microtubules are vulnerable to ALB-SO2 in vivo, raising the possibility that ALB-SO2 acts indirectly upon Wolbachia through the host cytoskeleton. However, colchicine treatments that eliminate Brugia microtubules had no significant impact on Wolbachia titer or elongation. Furthermore, ALB-SO2 suppressed Wolbachia titer in Drosophila even though there was no detectable impact on host microtubules in this system. Thus, ALB-SO2 does not disrupt Wolbachia through an influence on the host microtubule cytoskeleton.
The elongation of Wolbachia induced by ALB-SO2 suggests a possible mechanism of action for this compound. It has been widely documented that extensive bacterial elongation, referred to as filamentation, ensues when binary fission is disrupted, due to continued growth of the bacteria despite the failure of abscission [73,102]. Thus, Wolbachia elongation in ALB-SO2-treated ovaries suggests that ALB-SO2 is preventing abscission of growing and replicating Wolbachia. By contrast, ALB-SO2 treatment had no visible impact on Wolbachia length in Brugia hypodermal chords. This suggests that the Wolbachia bacteria residing within the hypodermal chord are in a non-growing steady state as compared to the proliferating Wolbachia population in the distal region of the B. malayi ovary. If confirmed, this implies that bacteriostatic drugs may not be effective at clearing Wolbachia from the hypodermal chords.
ALB-SO2 could disrupt Wolbachia binary fission by targeting a number of different factors. Proper binary fission relies upon assembly of FtsZ filaments at the future fission site, which then leads to recruitment of numerous other factors that help stabilize and constrict the division septum [103]. ALB has been shown to suppress titer and induce filamentation of Mycobacterium tuberculosis [104], and numerous studies demonstrate that ALB-like compounds target and disrupt FtsZ in Staphylococci, Escherichia coli and M. tuberculosis [73,104,105,106,107,108,109,110,111,112,113,114,115]. A role for ALB-SO2 in disrupting Wolbachia FtsZ function would be consistent with this body of work. However, given staining variability and resolution limits of the Brugia system, it is not currently possible to distinguish whether FtsZ concentration or distribution in vivo is significantly different between DMSO and ALB-SO2-treated conditions. An alternative possibility is that ALB-SO2 disrupts binary fission without directly targeting FtsZ. For example, some ALB-like compounds can intercalate into DNA and thus serve as potential DNA damaging agents [73]. DNA damage is known to induce the FtsZ-inhibitor SulA, leading to bacterial filamentation [116,117]. It is further possible that ALB-SO2 targets other as-yet unrecognized factors that are key to initiation and execution of Wolbachia binary fission.
This study has redefined the mechanism by which ALB acts against filarial nematodes. Building upon work by others, the data of this study suggest that ALB administered to humans disrupts B. malayi by a two-fold impact. First, ALB and its metabolites disrupt the filarial microtubule cytoskeleton, leading to rapid death of microfilariae. Second, our study indicates that the ALB-SO2 metabolite is directly targeting Wolbachia. Certain formulations of ALB were previously shown to lead to Brugia sterility in animal models, but the mechanism underlying this effect was unclear [27]. It is possible that microtubule disruption by ALB and its metabolites directly compromises the structure of the adult germline in Brugia. However, DOX-based elimination of Wolbachia from the female germline has also been shown to induce apoptosis in germline cells [37]. Perhaps ALB-SO2-mediated disruption of Wolbachia contributes further to destruction of the Brugia germline.
ALB-SO2 may serve as a valuable asset in the campaign against Wolbachia-based disease. Used in tandem with conventional antibiotics, ALB-SO2 may shorten the effective treatment time of anti-Wolbachia treatments from the current 4 to 6 weeks [5,6]. It is likely that more potent derivatives of ALB-SO2 may be discovered with enhanced the specificity for Wolbachia. Identification of the ALB-SO2 target will enable optimization of the compound to provide a possible alternative route for future treatment of African river blindness and lymphatic filariasis.

Generation of cultured cells
The JW18 cell line bearing the Jupiter-GFP transgene was generated according to the method described in Karpova et al, 2006. 1 to 15 hour old embryos derived from Wolbachia-infected flies carrying a Jupiter-GFP transgene were homogenized and plated in flasks. During the next six months of maintenance, five of the initial twenty seed flasks converted into immortal tissue culture lines. The JW18 cell line was selected for further pursuit due to its planar growth pattern and stable abundant Wolbachia infection. Cells were maintained at 25-26uC in Sang and Shields media containing 10% fetal bovine serum, split weekly at a 1:3 dilution. A cured version of the JW18 line, referred to as JW18TET, was generated by treating the cells with tetracycline [34,118,119]. A Chi-square test was used to test for differences in the frequency of mitosis between JW18 and JW18TET cells.

Screening approach
Several small chemical libraries were used in this study. The Spectrum Collection (MicroSource Discovery Systems, Inc) of 2000 compounds contains 1000 drugs with known pharmacological properties, 600 natural products, and 400 other bioreactive compounds. The NCI Diversity Set I contains 1990 compounds selected for structural uniqueness and fewer than 5 rotatable bonds. The NCI Mechanistic Set contains 879 compounds, selected for their diversity of impact in human tumor cell lines screened by the NCI. The NCI Challenge Set contains 57 compounds that stood out in the NCI tumor cell screens because of the unusual patterns of cell lethality and resistance induced by these compounds. The stock concentration of the compounds in these libraries was 10 mM. Furthermore, these libraries were reformatted into 384-well plates by the UCSC screening center. Specifically, compounds were placed into columns 3-22 of the plates, leaving columns 1-2 and 23-24 vacant to allow for untreated control wells.
Cells were plated in 384-well, clear bottom plates (Griener Bioone) pre-coated with 0.5 mg/mL Concanavalin A. JW18 cells were added to 22 columns, and JW18TET cells were added to the remaining 2 columns at a concentration of 6500 cells per well. After the cells adhered to the plates for 4-6 hours, compounds were transferred into 20 columns of JW18 cells in the center of the plate using a Janus MDT pin tool. The final concentration of compound was 100 uM per well. All treatments were distributed into 3 plate replicates.
After a 5-day incubation with the compounds at 25uC, the cells were prepared for imaging. Cells were fixed for 20 minutes in 4% formaldehyde and rinsed with PBS using an automated BioTek liquid handler. All staining solutions were administered using a Multidrop robot, with extensive rinsing between treatments. Mouse anti-histone (MAB052, Millipore) and goat anti-mouse Alexa 594 (Invitrogen) was diluted to 1:1250 in PBS/0.1% Triton. A stock solution saturated with DAPI was used at a final concentration of 1:40. After staining, PBS+sodium azide was added to all wells of the plates.

Screen data analysis
Stained plates were imaged using an ImageXpress Micro system (Molecular Devices, Sunnyvale CA). 10 images were acquired per well at 406 magnification. These images were analyzed using customized analysis software provided by Molecular Devices. The software routine masks any areas where clumps of cells are detected, based upon intensity of the Jupiter-GFP. The boundaries of the remaining cells and their nuclei are recognized based upon the Jupiter-GFP and anti-histone stains. A mask is applied to the nuclei, thus obscuring the histone and DAPI signal from those areas. A threshold for DAPI fluorescence detection is set to detect as much Wolbachia as possible in JW18 control cells while minimizing detection of background DAPI signal in JW18TET control cells. The remaining cytoplasmic DAPI, specifically labeling Wolbachia, is scored in individual cells to determine whether each cell is infected with Wolbachia. The cutoff value distinguishing ''infected'' from ''uninfected'' cells is 4000-5000 cytoplasmic DAPI fluorescence units per cell. This is a stringent limit, as values from 10 randomly selected 384-well plates indicate that the untreated JW18TET cells exhibit an average of 2383 +/ 278.30 cytoplasmic DAPI fluorescence units per cell as compared to untreated JW18 cells, which exhibit an average of 26146 +/ 2880.4 DAPI fluorescence units per cell.
A spreadsheet from the data analysis software indicates the quantity of Wolbachia-infected cells versus total cells measured in each well. An average and standard deviation were calculated to assess the frequency of Wolbachia infection in JW18 and JW18TET control cells, using the descriptive statistics function in SPSS (IBM). These values were used to calculate a Z9 factor for each plate [120]. The Z9 factor represents 1 -(the sum of the standard deviations for each control divided by the absolute value of the difference between mean values for each control.) Z9 factors regarded as acceptable by the field range from 0-1. Our Z9 factors range from 0.2 to 0.65 per plate, verifying that the infected and uninfected controls were clearly distinguishable by the assay. To identify preliminary ''hit'' compounds that substantially reduce intracellular Wolbachia titer, an initial hit range was calculated to lie between the JW18 average infection frequency 2 3 standard deviations, and the average JW18TET infection frequency + 3 standard deviations [120]. To enable comparison of hits identified on different treatment plates, the scaling of this initial hit range was next reset to span from 0-1, thus applying a uniform, normalized hit range to all plates.
To further increase the stringency for identifying hits, we also calculated an average infection frequency for all JW18 cells on the plate (treated or not), as most treatment wells are expected to be indistinguishable from untreated controls. Wells that lay within 3 standard deviations of the mean were removed from the hit list. Hit wells identified in only 1 of 3 replicates were also removed from the hit list. From the remaining hit candidates, we further identified compounds designated as hazardous by the National Cancer Institute, which includes alkylating agents, corrosives, carcinogens, explosives, flammables, oxidizers, poisons and known toxins. Any hits falling into these hazard classes were removed from further consideration. The hit compounds not excluded by these stringent criteria were designated as finalized hits.
The cytotoxicity of each hit from the Spectrum and NCI library screens was determined using a stepwise classification process. The Pubchem Bioassay Database was mined to assess the impact of our hits in prior mammalian tumor cell cytotoxicity screens conducted by the NCI. The majority of our hit compounds have already been tested for cytotoxicity in 45-115 screens conducted by the NCI. For those hits, if over 50% of the NCI screens indicated the drug to be cytotoxic, we designated that hit as ''toxic'' in our listing. If 10%-50% of the NCI screens indicated cytotoxic properties for that compound, we classified it as ''moderately toxic.'' If less than 10% of NCI screens indicated toxicity, we classified it ''non-toxic.'' A subset of our hit compounds have been run in 2 or fewer prior NCI screens. For those, hits, a preliminary cytotoxicity designation was assigned based upon the cell density reported by our screen. The average cell density per well was first calculated for each set of treatments, and treatments that failed to shift the cell density of a single well more than +/233% from the plate average were designated as ''non-toxic.'' Treatments that increased the average cell density to more that 33% over the average density were classified as ''toxicity unclear'', while treatments that reduced average cell density to 33%-66% below the average cell density were classified as ''moderately toxic.'' Drug treatments in Drosophila and Brugia ALB (Sigma), ALB-SO and ALB-SO2 (Santa Cruz Biotechnology) were dissolved into DMSO to create 10 mM stock solutions.
The flies used for this study, w; Sp/Cyo; Sb/TM6B, were reared on fly food consisting of 0.5% agar, 7% molasses, 6% cornmeal, and 0.8% killed yeast. Newly eclosed flies were collected, reared for 3 days, starved one day, and then fed compounds of interest for one day. For titer assessment experiments, compounds were diluted to a final concentration of 100 uM in yeast paste. Equivalent amounts of carrier DMSO diluted into these nutrient sources were used as a control.
Brugia microfilariae were provided by TRS (Athens, Georgia). These were resuspended in 2 mL tissue culture media containing 50 uM of each compound of interest, except for colchicine, which was administered at 20 uM. Control worms were provided an equivalent dilution of carrier DMSO alone. Microfilariae were incubated with the compounds for 1 day at 37 C with 7.5% CO2. Adult Brugia were incubated for 3 days.
Live imaging of tissue culture cells Tissue culture chamber slides were coated with 0.5 mg/mL Concanavalin A, followed by addition of JW18 cells. After a 24hour incubation at 25uC, cells were exposed to Syto-11 (Molecular Probes) for one minute at a dilution of 1:50,000. Cells were imaged on a Leica SP2 confocal microscope at 1006 magnification and 2.756optical zoom using FITC filters. Images were acquired at 5second intervals for up to 10 minutes.

Tissue staining and analysis
A combination of fixation and staining methods was used. Propidium iodide and microtubule staining of Drosophila ovarian tissue was also done using established methods [54,55,76,77]. Brugia staining was performed as previously described [121]. Briefly, worms were sectioned and fixed in PFA 3.2% for 10 minutes, rinsed in PBS+0.1% Triton-X100, and incubated overnight in RNAseA (10 mg/mL) in PBST, prior a 30 second incubation in a propidium iodide solution (1 mg/mL diluted100X in PBST). Worm fragments were washed for 1 minute in PBST and mounted in DAPI Vectashield mounting medium (Vector Labs). For FtsZ staining, worm fragments were incubated with rabbit anti-FtsZ [78] in PBST after a 1:500 dilution, after RNAse treatment, washed three times for 10 minutes, before adding a CY5-conjugated anti-rabbit secondary antibody, followed by 3 washes of 10 minutes in PBST and mounting in DAPI Vectashield mounting medium. Microtubules were stained using the monoclonal antibody DM1a (Sigma) at 1:250 and an Alexa488conjugated goat anti-mouse secondary antibody (Molecular Probes) was used at 1:250. Mouse anti-histone H1 (Millipore MAB052) was used at 1:500, and rabbit anti-phospho-histone H3 (Millipore) was used at 1:1000.
Data collection was conducted as previously. All tissues were imaged using Leica SP2 and Leica SP5 confocal microscopes. Wolbachia were quantified in single focal planes of stage 10A Drosophila oocytes using established methods [67]. To measure Wolbachia length in the Brugia ovaries, 11 images representing a 2 micrometer-thick Z-stack were merged to make a single image, followed by assessment of bacterial length using the Leica SP2 line quantification function. Average Wolbachia titer values associated with drug treatments were normalized against their respective DMSO controls as previously to ensure comparability between experiments [67].
Statistical analysis was conducted using established methods. The ANOVA function in SPSS was used to evaluate Wolbachia titer differences in Drosophila oogenesis and differences in Wolbachia length in B. malayi. Chi-square tests were used to compare the frequency of Wolbachia posterior localization in Drosophila oogenesis and to evaluate ratios of Wolbachia per host nucleus in the Brugia ovary. Figure S1 Overview of chemical screen strategy. (TIF) Figure S2 Comparing beta-tubulin residues 165 to 200. Yellow highlighting showss benzimidazole-resistant residues. Green indicates amino acid changes associated with benzimidazole susceptibility. (TIF)