An Antiparasitic Compound from the Medicines for Malaria Venture Pathogen Box Promotes Leishmania Tubulin Polymerization

Abstract

The few frontline antileishmanial drugs are poorly effective and toxic. To search for new drugs for this neglected tropical disease, we tested the activity of compounds in the Medicines for Malaria Venture (MMV) “Pathogen Box” against Leishmania amazonensis axenic amastigotes. Screening yielded six discovery antileishmanial compounds with EC₅₀ values from 50 to 480 nM. Concentration–response assays demonstrated that the best hit, MMV676477, had mid-nanomolar cytocidal potency against intracellular Leishmania amastigotes, Trypanosoma brucei, and Plasmodium falciparum, suggesting broad antiparasitic activity. We explored structure–activity relationships (SAR) within a small group of MMV676477 analogs and observed a wide potency range (20–5000 nM) against axenic Leishmania amastigotes. Compared to MMV676477, our most potent analog, SW41, had ~5-fold improved antileishmanial potency. Multiple lines of evidence suggest that MMV676477 selectively disrupts Leishmania tubulin dynamics. Morphological studies indicated that MMV676477 and analogs affected L. amazonensis during cell division. Differential centrifugation showed that MMV676477 promoted partitioning of cellular tubulin toward the polymeric form in parasites. Turbidity assays with purified Leishmania and porcine tubulin demonstrated that MMV676477 promoted leishmanial tubulin polymerization in a concentration-dependent manner. Analogs’ antiparasitic activity correlated with their ability to facilitate purified Leishmania tubulin polymerization. Chemical cross-linking demonstrated binding of the MMV676477 scaffold to purified Leishmania tubulin, and competition studies established a correlation between binding and antileishmanial activity. Our studies demonstrate that MMV676477 is a potent antiparasitic compound that preferentially promotes Leishmania microtubule polymerization. Due to its selectivity for and broad-spectrum activity against multiple parasites, this scaffold shows promise for antiparasitic drug development.

Graphical Abstract

Human leishmaniasis is endemic in nearly 100 countries, and 350 million people worldwide are at risk for this disfiguring (cutaneous or mucocutaneous) or lethal (visceral) disease.¹ Leishmaniasis is caused by obligate intracellular single-celled parasites of the Leishmania genus, which have two life cycle stages. The fast-growing promastigote, which lives in sandflies, transforms into the slow-growing amastigote inside human phagocytic cells, causing clinical disease.² The primary species of Leishmania used in this manuscript, Leishmania amazonensis, which belongs to the L. mexicana complex, can be grown in the tissue culture setting in vitro as promastigotes, axenic amastigotes (amastigotes in tissue culture media alone, without host cells present), or intracellular amastigotes (amastigotes grown inside macrophages (Mϕ)).

Current treatments are few and have significant side effects. Since resistance has emerged to these antileishmanial drugs^3-5 and effective Leishmania vaccines are decades away, there is an urgent need for novel chemotypes to find replacements for the drugs that are currently available. To accelerate drug discovery, MMV has coordinated screens of over 5 million compounds against Plasmodium, generating approximately 20 000 starting points for drug discovery and development.^6-9 Previously, a representative set of 400 compounds, called the Malaria Box, made a significant impact beyond the malaria field and stimulated medicinal chemistry efforts against many diseases, including leishmaniasis.^6,10,11 Due to the success of the Malaria Box, the MMV distributed another collection of 400 drug-like compounds, the Pathogen Box (www.PathogenBox.org), which are likely to show acceptable oral absorption and target an expanded set of pathogens.^12,13 Its name derives from the fact that each compound in the box has known activity against one or more pathogenic bacterial, fungal, or parasitic organisms. Known antiparasitic drugs and current antiparasitic lead compounds are also included. Similar to the Malaria Box, these compounds also reflect a cross-section of the chemical diversity available in MMV’s 20 000 hits, providing 374 starting points for oral drug discovery.

Here, we assessed the Pathogen Box for activity against L. amazonensis axenic amastigotes, which yielded six discovery antileishmanial compounds with EC₅₀ values ranging from 50 to 480 nM. The top hit, MMV676477, also killed intracellular Leishmania spp., T. brucei, and P. falciparum at nanomolar concentrations. We describe this screen and our initial SAR studies that provide proof-of-concept for future optimization of the scaffold.

We next engaged in target identification studies for MMV676477 and derivatives. Using assays in intact parasites, we demonstrated that our scaffold affects Leishmania cell division, as well as selectively promotes microtubule polymerization within Leishmania parasites, resulting in several additional morphological changes in the parasite. Since Leishmania requires microtubule dynamics for multiple critical functions to survive throughout its life cycle, including chromosome segregation, flagellar motility, cell division, and structure maintenance,¹⁴ parasite microtubule polymerization is a plausible antiparasitic drug target. Furthermore, our assays on purified tubulin demonstrate that MMV676477 and derivatives directly bind to tubulin and facilitate its polymerization, with selectivity for leishmanial over porcine tubulin, providing one plausible mechanism for this scaffold’s antiparasitic activity.

In summary, our studies demonstrate that MMV676477 and its analogs are potent antiparasitic compounds that promote Leishmania microtubule polymerization. Although depolymerizing antileishmanial agents have been studied in the past, to our knowledge, no drugs currently on the market or in development selectively facilitate tubulin polymerization in parasites. Since MMV676477 has activity against multiple parasites, this scaffold shows promise for future antiparasitic drug development.

RESULTS

Screening the Pathogen Box Identifies Multiple Hit Antileishmanial Compounds.

To identify antileishmanial compounds, we used a microplate-based alamarBlue assay to quickly triage the 400 drug-like compounds available in the MMV Pathogen Box collection. We included the clinically approved antileishmanial agents amphotericin B and miltefosine as additional controls. We used a three-step process to screen the Pathogen Box: (1) identify hits against axenic amastigotes, (2) confirm the inhibitory concentrations of these hits against axenic amastigotes, and (3) evaluate these hits for their potency against intracellular parasites using fluorescence- and bioluminescence-based intracellular assays. The MMV Pathogen Box resource was tested at two concentrations (5 μM and 1 μM, 72 h end point) on axenic amastigotes. Full data are shown in Table S1.

Hits were defined as compounds that, at a concentration of 1 μM, decreased the relative fluorescence intensity signal to ≤70% of that produced by parasites incubated in vehicle (0.06% DMSO) only (Table S1). A total of 10 hits were identified, including four reference compounds: buparvaquone, delamanid, auranofin, and nitazoxanide (Table S1). All 10 hits were analyzed to estimate their EC₅₀ at 72 h $({\text{EC}}_{50}^{72\mathrm{h}})$. The two best hits, MMV676477 and MMV676412, both from the MMV set annotated as antimycobacterial, displayed similar potency to that of buparvaquone, delamanid, and amphotericin B.^15-18 Exemplar log-concentration–response curves for these two best hits (MMV676412 and MMV676477), compared to amphotericin B, are shown in Figure 1A, and the ${\text{EC}}_{50}^{72\mathrm{h}}$ values for all hits are reported in Table 1. MMV676412 and MMV676477 have an ${\text{EC}}_{50}^{72\mathrm{h}}$ ± standard error (SE) of 51 ± 3.7 nM and 79 ± 8.4 nM, respectively, and they both are nearly as potent as amphotericin B (EC₅₀^72h = 53 ± 4.4 nM) in vitro (Figure 1A). Since ${\text{EC}}_{50}^{72\mathrm{h}}$ data for both MMV676412 and MMV676477 were acquired using the alamarBlue assay, the reliability of these data and alamarBlue itself were independently confirmed by a bioluminescence assay with transgenic luciferase-expressing parasites (L. amazonesis^luc; see Methods for how this clone was generated) (Figure 1B). The ${\text{EC}}_{50}^{72\mathrm{h}}$ values estimated using either assay format were almost identical.

Figure 1.

Concentration–response curves for two MMV Pathogen Box compounds for L. amazonensis axenic and intracellular amastigotes, compared to amphotericin. Log concentration–response curves using (A) alamarBlue and (B) luciferase for MMV676412 (■), MMV676477 (□), and amphotericin B (▲, gray) for axenic amastigotes. (C) Log concentration–response curves for the indicated compounds in L. amazonesis intracellular amastigotes (using luciferase and RAW 264.7 cells). Data shown represent the means of three biological replicates, with standard deviations (SD) indicated by error bars. ${\text{EC}}_{50}^{72\mathrm{h}}$ values for all hit compounds are detailed in Table 1 for axenic amastigotes and Table 2 for intracellular amastigotes.

Table 1.

EC₅₀ Values for the 10 Hit MMV Pathogen Box Compounds, Amphotericin, and Miltefosine against L. amazonensis Axenic Amastigotes^a

compound	EC50 ± SE (nM)	compound	EC50 ± SE (nM)
MMV676412	51 ± 3.7	amphotericin B	53 ± 4.4
MMV676477	79 ± 8.4	buparvaquone	12 ± 1.1
MMV688372	160 ± 1.8	delamanid	31 ± 3.5
MMV652003	220 ± 50	auranofin	130 ± 2.5
MMV595321	240 ± 70	nitazoxanide	590 ± 100
MMV090930	520 ± 120	miltefosine	1200 ± 280

^aReference drugs included in the MMV Pathogen Box are labeled by their names, rather than the MMV number. Mean ${\text{EC}}_{50}^{72\mathrm{h}}$ values calculated from three biological repeats (in nM) ± SE of these three ${\text{EC}}_{50}^{72\mathrm{h}}$ values are shown.

Compounds supplied in the MMV Pathogen Box have been tested for cytotoxicity against human cell lines, and this information is available online (https://www.pathogenbox.org/about-pathogen-box/supporting-information). Prior to testing the activity of our hits against intracellular parasites, we tested their cytotoxicity against the Mϕ cell line RAW 264.7, which we use as the host cell for assays of intracellular parasites and which continues to divide in tissue culture (Table S2).

MMV676477 Is a Potent and Selective Antiparasitic Hit Compound.

As outlined in Katsuno et al.,¹⁹ to be selected as a hit, at minimum, a compound should be cytocidal, with an intracellular EC₅₀ of less than 10 μM. We next analyzed the two most potent compounds, MMV676477 and MMV676412, against intracellular L. amazonensis amastigotes, along with amphotericin as a positive control. We used RAW 264.7 cells^33,34 to perform a bioluminescence assay with transgenic luciferase-expressing L. amazonensis (Figure 1C). We found that both amphotericin (${\text{EC}}_{50}^{72\mathrm{h}}=65\pm 26\text{nM}$) and MMV676477 (${\text{EC}}_{50}^{72\mathrm{h}}=500\pm 20\text{nM}$) were active against intracellular parasites, while MMV676412 was not (${\text{EC}}_{50}^{72\mathrm{h}}>2000\text{nM}$), and therefore we focused on MMV676477 for the remainder of these studies.

We note that all routine in vitro quantification of antileishmanial drug potency is quantification of “cytostatic” potency. Molecular studies of drug resistance have identified the need to rapidly define a hit’s cytocidal activity early in the drug development process and have emphasized prioritizing compounds that kill parasites rather than merely inhibiting their growth. Therefore, we also quantified the cytocidal action of MMV676477 by obtaining an LD₅₀ at 72 h (${\text{LD}}_{50}^{72\mathrm{h}}$, a lethal dose that kills 50% of the bulk population of the parasites relative to untreated control)^10,22 and compared it to the known cytocidal drug amphotericin. To do so, we employed an intracellular washout assay.^15,25 We found that MMV67477’s ${\text{LD}}_{50}^{72\mathrm{h}}$ was 650 ± 200 nM and amphotericin’s ${\text{LD}}_{50}^{72\mathrm{h}}$ was 82 ± 7.1 nM (Table 2). We next calculated a ratio of LD₅₀/EC₅₀ for MMV676477 and amphotericin. Low ratios indicate that the estimated EC₅₀ values represent killing of the parasites rather than simple growth inhibition, whereas higher ratios represent a cytostatic compound. For example, amphotericin B, a known cytocidal drug, has an estimated LD₅₀/EC₅₀ ratio of 1.3. MMV676477 also had an LD₅₀/EC₅₀ ratio of 1.3, suggesting cytocidal activity.

Table 2.

Comparison of LD₅₀ and EC₅₀ Values for MMV676477^a

compound	EC50 ± SE (nM)	LD50 ± SE (nM)	LD50/EC50 ratios
MMV676477	500 ± 20	650 ± 200	1.3
amphotericin B	65 ± 26	82 ± 7.1	1.3

^a${\text{LD}}_{50}^{72\mathrm{h}}$ concentrations for MMV676477 and the known cytocidal antileishmanial drug amphotericin in L. amazonensis intracellular amastigotes were compared to the ${\text{EC}}_{50}^{72\mathrm{h}}$ concentrations for L. amazonensis intracellular amastigotes. Shown are mean ${\text{EC}}_{50}^{72\mathrm{h}}$ and ${\text{LD}}_{50}^{72\mathrm{h}}$ values (nM) ± SE calculated from three biological replicates. Ratios are calculated by dividing the LD₅₀ by the EC₅₀ value.

We next tested MMV676477 against several other protozoan parasites: axenic amastigotes of Leishmania donovani, Leishmania tarentolae, and blood-stage Plasmodium falciparum and Trypanosoma brucei. The compound’s activity was measured using alamarBlue (L. donovani and L. tarentolae), Malaria SYBR green I fluorescence (MSF; P. falciparum), and CellTiter-Glo luminescent assays (T. brucei), as described previously^10,23,24 (Table 3). MMV676477 demonstrated broad activity against all three parasite genera. In addition, MMV676477 has been recently reported as having potent activity against Neospora caninum, Cryptosporidium parvum, and Entamoeba histolytica (Table 3).^25-28 However, no target has been suggested for this compound.

Table 3.

Activity of MMV676477 against Additional Parasites^a

pathogen	MMV676477 EC50 ± SE (nM)
L. donovani axenic amastigotes	1300 ± 210
P. falciparum	420 ± 19
L. tarentolae axenic amastigotes	11 ± 1.3
T. brucei	190 ± 8.7
Neospora caninumb	13
Cryptosporidium parvumb	220
Entamoeba histolytica trophozoite proliferation at 10 μMb	56%

^a${\text{EC}}_{50}^{72\mathrm{h}}$ values for MMV676477 against L. donovani axenic amastigotes, blood-stage P. falciparum, and T. brucei. Shown are mean ${\text{EC}}_{50}^{72\mathrm{h}}$ values (nM) calculated from three biological replicates ± SE.

^bData obtained from literature.^26-28

Structure–Activity-Relationship (SAR) of MMV676477: Initial Characterization.

Resynthesis of MMV676477 confirmed its activity to be the same as in our initial screen (Supplemental Methods). To improve the potency and selectivity of MMV676477, as well as to identify regions of the compound that could be functionalized for mode-of-action studies, we synthesized 11 analogs (see Table 4 for structures and Supplemental Methods for synthetic strategy). The structure of MMV676477 is characterized by a central pyrazole ring linked through N1 to a pyrimidinone moiety (Table 4). An N-acylated amino group at the pyrazole C5 position provides an additional opportunity for diversification. Our preliminary SAR survey modified the pyrimidinone and N-acyl group. We generated compounds with a range of activities from 20 nM to >5000 nM against L. amazonensis axenic amastigotes (Table 4). Compared to the parental compound (MMV676477), SW41, the most potent compound in this series, showed about 4-fold improved potency against L. amazonensis (${\text{EC}}_{50}^{72\mathrm{h}}$ = 20 vs 79 nM; Figure 2, Table 4). Similarly, SW41 exhibited a 4-fold improved selectivity index (SI, calculated as the EC₅₀ for L. amazonensis axenic amastigotes divided by the EC₅₀ for RAW 264.7 cells) over MMV676477 (SI = 120 vs 28). SW41 was also tested against L. donovani axenic amastigotes and L. amazonensis intracellular amastigotes (Figure 2, Table 5). SW41 was 3–4-fold more potent than MMV676477 against L. donovani axenic amastigotes and L. amazonensis intracellular amastigotes (Figure 2, Table 5). Moreover, its therapeutic index was increased relative to the initial hit as a consequence of improved potency against the parasites, without a corresponding increase in toxicity toward the Mϕ cell line RAW 264.7. For the 12 compounds tested, their antileishmanial activity ranked as follows: SW41 > MMV676477 > SW73 > SW22 > SW74 > SW75> SW100/SW101 > SW102 > SW23/SW10 (Table 4). SW23 and SW10 were essentially inactive.

Table 4.

Structures and EC₅₀ Values for MMV676477 Structural Analogs^a

Compound ID	Structure	Amastigote EC50 ± SE (nM)	Promastigote EC50 ± SE (nM)	RAW 264.7 cells CC50 ± SE (nM)
	R =
MMV676477	2-SCH3	79 ± 8.4	2400 ± 35	2200 ± 410
SW100	H	1100 ± 84	2000 ± 130	690 ± 140
SW22		160 ± 3.2	6700 ± 53	1300 ± 42
SW72	4-OCH3	180 ± 8.2	7500 ± 880	> 10,000
SW41	3-CI, 4-F	20 ± 0.9	780 ± 29	2400 ±100
SW73	2-CF3	68 ± 3.9	3200 ± 87	1500 ± 140
	R =
SW75	H	290 ± 22	11200 ± 260	9900 ± 2800
SW23	−C(O)CH3	> 3000	> 50,000	> 10,000
SW101	−C(O)NH-(4-CH3-C6H4)	1100 ± 91	8200 ± 1600	3000 ± 250
SW102	−C(O)NH-(4-OCH3-C6H4)	2400 ± 160	13000 ± 1400	5200 ± 380
SW74	−C(O)-(3-pyridyl)	250 ± 40	9200 ± 1300	4200 ± 520
SW10		> 5000	> 50,000	> 10,000

^a${\text{EC}}_{50}^{72\mathrm{h}}$ and ${\text{CC}}_{50}^{72\mathrm{h}}$ values for structural analogs of MMV676477 against L. amazonensis axenic amastigotes and promastigotes and the Mϕ cell line RAW 264.7 are shown. Shown are mean ${\text{EC}}_{50}^{72\mathrm{h}}$ or ${\text{CC}}_{50}^{72\mathrm{h}}$ values (nM) calculated from three biological replicates ± SE. Structure of each compound is provided. See Supplemental Methods for synthesis of these analogs.

Figure 2.

Comparison of potency of MMV676477 and its structural analog SW41 against L. amazonensis axenic and intracellular amastigotes and L. donovani axenic amastigotes. Log concentration–response curves for MMV676477 (□) and SW41 (■) against (A) axenic and (B) intracellular amastigotes of L. amazonensis and (C) axenic amastigotes of L. donovani. SW41 curves shift toward the left, indicating improved potency. Data represent the means of three biological replicates, with SD indicated by error bars. ${\text{EC}}_{50}^{72\mathrm{h}}$ values are detailed in Table 5.

Table 5.

MMV6767477 and SW41 EC₅₀ Values for Additional Leishmania Life Cycle Stages and Species^a

	L. amazonensis		L. donovani
compound	axenic amastigotes, EC50 ± SE (nM)	intracellular amastigotes, EC50 ± SE (nM)	axenic amastigotes, EC50 ± SE (nM)
MMV676477	79 ± 8.4	500 ± 20	1300 ± 210
SW41	20 ± 0.9	220 ± 13	310 ± 38

^a${\text{EC}}_{50}^{72\mathrm{h}}$ values for MMV676477 and its structural analog, SW41, against L. amazonensis axenic and intracellular amastigotes and L. donovani axenic amastigotes. Shown are mean ${\text{EC}}_{50}^{72\mathrm{h}}$ values (nM) calculated from three biological replicates ± SE. See Figure 2 for ${\text{EC}}_{50}^{72\mathrm{h}}$ curves for all compounds shown in this table.

MMV676477 Promotes Microtubule Polymerization in Parasites.

Since we obtained MMV676477 from a phenotypic screen, our next goal was to identify its likely target. Upon review of the literature, we noted that MMV676477-treated L. amazonensis microscopically resembled Leishmania treated with the anticancer drug paclitaxel (“Taxol”, ${\text{EC}}_{50}^{72\mathrm{h}}$ for La = 760 ± 8.3 nM).²⁹ Paclitaxel stabilizes microtubules and prevents mitosis in cancer cells.³⁰ We therefore used paclitaxel as a benchmark positive control for antimitotic activity and an unrelated antileishmanial drug, miltefosine, as a negative control. To test whether MMV676477 affected Leishmania cell division, we treated both promastigote and amastigote forms of L. amazonensis with MMV676477 and analogs at their 72 h EC₅₀ concentrations (${\text{EC}}_{50}^{72\mathrm{h}}$), but only for 48 h to ensure that parasites were assessed prior to their death, and examined them by confocal immunofluorescence microscopy. Overall, we observed an increase in parasites that appeared to be undergoing cell division that were treated with paclitaxel or MMV676477, with many treated promastigotes seeming to be stuck together in groups of two, relative to the more common single promastigotes in the DMSO or miltefosine controls (Figure 3A). We next characterized this phenotype further (Figure 3A,B). In general, while undergoing mitosis, Leishmania promastigotes replicate their flagella (F) first, then their kinetoplasts (K), and then their nuclei (N). Parasites are therefore designated as 1F or 2F, 1K or 2K, or 1N or 2N.³¹ We calculated the total percentages of promastigotes in either 2F1K1N, 2F2K1N, or 2F2K2N treated with MMV676477, controls, and analogs, as we considered all of these categories to represent parasites that were undergoing cell division (or mid-cell division). With one exception described below, we found a ~2-fold increase in parasites that were in mid-cell division in paclitaxel, MMV676477, and active analog-treated promastigotes (Figure 3C in black; p < 0.05 by ANOVA compared to DMSO control). There was a similar increase in paclitaxel, MMV676477, and active analog-treated amastigotes that were in mid-cell division (Figure 3C in gray; p < 0.05 by ANOVA compared to DMSO control). Inactive analogs and miltefosine did not affect L. amazonensis cell division. Hence, the ability of analogs to affect L. amazonensis during cell division generally correlated with their antiparasitic activity, with the lone exception of SW75, which had antiparasitic activity (${\text{EC}}_{50}^{72\mathrm{h}}$ = 290 ± 22) but no effect on parasite cell division. However, SW75 is structurally different from the remaining analogs (Table 4), lacking a benzamide group. Therefore, it appears likely that SW75 exerts its effect through an unrelated and potentially nonspecific mechanism.

Figure 3.

Effect of MMV676477 and analogs on cell division in L. amazonensis promastigotes and amastigotes. (A) Exemplar fluorescence microscopy images of a single promastigote (DMSO control or miltefosine, treated for 48 h at the ${\text{EC}}_{50}^{72\mathrm{h}}$ concentration) or promastigotes undergoing cell division (paclitaxel or MMV676477, treated for 48 h at the ${\text{EC}}_{50}^{72\mathrm{h}}$ concentration). Promastigotes were labeled with anti-α-tubulin antibody (green; monoclonal antibody YL1/2, Invitrogen, catalog no. MA1-80017), anti-gp46 (membrane protein, red), and DAPI (nuclei, blue). Labeling intensity was normalized across samples linearly in Photoshop. Diagram on right for each promastigote shows location of flagella (F), kinetoplast (K) and nuclei (N). Gray box indicates area shown in panel B. Scale bar = 5 μm. (B) Enlarged images of exemplar promastigotes shown in panel A. Scale bar = 5 μm. (C) Quantification of antimitotic effects by microscopic analysis for DMSO (control), miltefosine (MLTF), paclitaxel (PTXL), MMV676477, and analog-treated promastigotes (black) and amastigotes (gray). Promastigotes were labeled as in panel A and counted as being in mid-cell division if they were in 2F1K1N, 2F2K1N, or 2F2K2N. Amastigotes were also labeled as in panel A, except that anti-P8 was used as a membrane protein marker, and parasites that had 2K were counted as being in mid-cell division. At least 200 parasites were analyzed per condition for three independent biological replicates (mean ± SE). *p < 0.05 by ANOVA compared to control conditions.

We next directly compared MMV676477, active analog, and paclitaxel-treated promastigotes to characterize whether there were additional gross morphological defects. We found that MMV676477 and paclitaxel-treated promastigotes had elongated flagella (Figure S1A). Quantification demonstrated a ~1.5-fold increase in flagella length (Figure S1B, p < 0.05 by ANOVA compared to DMSO control) when promastigotes were treated with MMV676477, paclitaxel, or active analogs for 24 h at their respective ${\text{EC}}_{50}^{72\mathrm{h}}$. We also found that promastigotes treated with paclitaxel, MMV676477, or active analogs at their respective EC₅₀^72h for 48 h were more rounded or pear-shaped (defined as parasites with a cell body width that was ≥70% of the cell body length²⁹) than DMSO controls or promastigotes treated with the unrelated drug, miltefosine (Figure S1C). Quantification demonstrated a 2-fold increase in pear-shaped promastigotes in paclitaxel, MMV676477, and active analog-treated groups (Figure S1D, p < 0.05 by ANOVA compared to DMSO control). Inactive analogs, the structurally dissimilar compound SW75, and miltefosine did not affect either flagella length or promastigote shape.

One reason why the MMV676477 scaffold might affect parasite cell division and morphology in manners similar to paclitaxel is that like paclitaxel, it might also affect microtubule polymerization, either directly or indirectly. Therefore, we determined whether MMV676477 affected the degree to which microtubules were polymerized in cells. First, we prepared cell extracts from compound-treated and untreated L. amazonensis after treatment for 24 h at the ${\text{EC}}_{50}^{72\mathrm{h}}$. Prior to sedimentation/centrifugation, the protein concentration for each sample was measured and normalized between samples, to ensure equal amounts of cellular material in each sample. Centrifugation of these extracts allowed separation of insoluble polymeric (pellet) and soluble dimeric (supernatant) tubulin. Western blot analysis revealed that both paclitaxel and MMV676477 promoted the partitioning of cellular tubulin toward the polymeric form (Figure 4A). Densitometric analysis of these blots showed that both MMV676477 and paclitaxel significantly increased the proportion of cellular polymeric tubulin (Figure 4B).

Figure 4.

Effect of MMV676477 and analogs on microtubule polymerization in L. amazonensis axenic amastigotes. (A) L. amazonensis axenic amastigotes were treated with the indicated compounds for 24 h at their ${\text{EC}}_{50}^{72\mathrm{h}}$. Dimeric (unpolymerized, supernatant) and polymeric (polymerized, pellet) tubulin were separated by differential centrifugation and subjected to Western blotting using an α-tubulin antibody. Prior to sedimentation/centrifugation, the protein concentration (for each sample) was measured and normalized between samples, to ensure equal amounts of cellular material in each sample. GAPDH was used as a loading control for supernatants. (B) Densiometric analyses of Western blot band intensity from three independent biological replicates ± SE; *^,#p < 0.05 by ANOVA compared to control conditions.

MMV676477 Promotes Polymerization of Purified Tubulin.

Since our studies in intact parasites suggested that the MMV676477 scaffold promotes microtubule polymerization in cells, we next determined whether these effects were the result of activity on tubulin itself. To determine whether MMV676477 directly regulates microtubule assembly, assembly competent tubulin from Leishmania tarentolae was purified as previously described^32,33 (Figure 5A). Leishmania amazonensis and L. donovani are BSL-2 organisms, posing technical challenges for large-scale protein purification. L. tarentolae causes disease in lizards but not humans; it is a BSL-1 model organism and grows to high densities in liquid culture (~1 × 10⁸ cells/mL). Tubulin is highly conserved between L. tarentolae and other Leishmania species, including L. amazonensis (>98% identity) and L. donovani. Based on this degree of conservation, we concluded that L. tarentolae tubulin would be a suitable surrogate for use in protein studies, as has been done previously.^32,33 Purified L. tarentolae tubulin (3 mg/mL) was treated with serial dilutions of MMV676477 while the absorbance was measured at 340 nm over time at 30 °C, as shown previously.^32,33 Representative turbidity curves are shown in Figure 5B. MMV676477 promoted L. tarentolae tubulin polymerization in a concentration-dependent manner (estimated MMV676477 EC₅₀ = 0.5 ± 0.1 μM; paclitaxel EC₅₀ = 1.3 ± 0.1 μM).

Figure 5.

Effect of MMV676477 on purified Leishmania and mammalian tubulin in vitro. (A) SDS-PAGE gel showing purification of assembly competent L. tarentolae tubulin by ion exchange chromatography. L. tarentolae lysate (1) was centrifuged (pellet, 2; supernatant, 3), filtered (4), loaded on a DEAE-sepharose column (flow-through, 5), and washed (6). Tubulin was eluted with high salt (7, 8). (B) Representative turbidity curves for MMV676477-treated purified Leishmania tubulin (3 mg/mL tubulin, 1% DMSO (= orange line)). DMSO (10%) was used to provide maximal tubulin polymerization and absorbance values, and EC₅₀ values were estimated (see Methods). For Leishmania tubulin, MMV676477 EC₅₀ = 0.5 ± 0.1 μM; maximum A340 using 10% DMSO was 0.4. (C) Correlation among purified L. tarentolae (3 mg/mL) tubulin polymerization activity of MMV676477 analogs at 10 μM and 1 μM. Absorbance signal (A340) changes were normalized to maximal A340 changes observed following treatment with 10% DMSO versus the negative control (no drug, 0% tubulin polymerization). (D) Correlation among MMV676477 analogs between purified L. tarentolae relative tubulin polymerization activity at 10 μM and antiparasitic ${\text{EC}}_{50}^{72\mathrm{h}}$ for L. amazonensis axenic amastigotes. (E) Correlation among MMV676477 analogs between purified L. tarentolae relative tubulin polymerization activity at 1 μM and antiparasitic ${\text{EC}}_{50}^{72\mathrm{h}}$ for L. amazonensis axenic amastigotes. (F) Representative turbidity curves (all concentrations in μM) for purified porcine tubulin (3 mg/mL tubulin, 1% DMSO (= dark blue line)) treated with MMV676477. For porcine tubulin, MMV676477 EC₅₀ = 11 ± 3.2 μM. Maximum absorbance (A340, 10% DMSO) = 0.4 for MMV676477.

We next tested the effects of MMV676477 analogs on L. tarentolae tubulin polymerization at 1 and 10 μM compound concentrations (Figure 5C). As the structurally dissimilar compound SW75 had no effect on relative tubulin polymerization, we did not include it in this analysis. This plot showed a strong and significant correlation (p < 0.0001, r² = 0.9) and allowed ranking of the compounds from most to least promotion of polymerization: SW41 > MMV676477 > SW73 > SW22 > SW74 > SW100/101 > SW102/SW23/SW10. This ranking was exactly the same as the ranking of compounds’ antiparasitic activity. Linear regression analysis also allowed us to compare the ability of compounds to promote in vitro polymerization (at 1 and 10 μM) with their antiparasitic activity on axenic amastigotes (Figure 5D,E). The 10 μM polymerization data showed a strong and significant correlation with the antiparasitic ${\text{EC}}_{50}^{72\mathrm{h}}$ data (p = 0.002, r² = 0.7) (Figure 5D). Using the more stringent cutoff criteria of polymerization activity at 1 μM, the correlation with the antiparasitic ${\text{EC}}_{50}^{72\mathrm{h}}$ data was weaker but still significant (p = 0.02, r² = 0.4) (Figure 5E).

Tubulin is relatively conserved (60–80% identity and 80–90% similarity) between Leishmania and higher eukaryotes. In addition to the toxicity data that we reported for MMV676477 and analogs against host cell lines (Table S2, Table 4), we obtained purified porcine tubulin (>99% pure, Cytoskeleton, Inc.) to assess selectivity of MMV676477. Following the manufacturer’s instructions, 3 mg/mL porcine tubulin was exposed to a serial dilution of the drug (0 to 25 μM) and absorbance was measured at 340 nm over time. Representative turbidity curves in the presence of MMV676477 at different concentrations (0 to 25 μM) are shown in Figure 5F. We found that MMV676477 promoted tubulin polymerization in a concentration-dependent manner (estimated EC₅₀ value of MMV676477= 11 ± 3.2 μM). Paclitaxel was also used as a benchmark drug in these experiments, which stimulated porcine tubulin polymerization in a concentration-dependent manner (estimated EC₅₀ value of paclitaxel = 1.5 ± 0.2 μM is consistent with that reported by Cytoskeleton, Inc. (https://www.cytoskeleton.com/bk011p) (Figure S2)). The unrelated antileishmanial drug miltefosine has no known activity on tubulin and did not affect tubulin polymerization at 50 μM (Figure S3). At micromolar concentrations, some drugs self-associate into colloidal aggregates, resulting in nonspecific effects on the target protein.³⁴ Tubulin polymerization activity by MMV676477 was not affected by 0.01% Triton-X treatment, suggesting that MMV676477’s activity was not merely aggregation-based (Figure S4). In addition, we imaged the products of assembly using fluorescence microscopy^35,36 (see Methods for details). MMV676477-treated tubulin was indistinguishable from tubulin treated with the known stabilizer paclitaxel (Figure S5).

MMV676477 Directly Binds to Purified Leishmania Tubulin.

The above studies using purified tubulin implied that MMV676477 directly binds to tubulin in order to facilitate polymerization. To further investigate whether the MMV676477 scaffold directly bound tubulin, we used an analog, SW22, as a probe for binding experiments (Table 4). SW22 was modified to include (1) a benzophenone that facilitates cross-linking to binding partners under UV irradiation, which results in irreversible binding, and (2) an alkyne group that facilitates conjugation to azide-containing dyes by copper-catalyzed alkyne–azide cycloaddition (“click chemistry”; CuAAC). Purified tubulin was first treated with SW22 (probe, L. amazonensis axenic amastigotes EC₅₀ = 156 nM). The tubulin was then UV cross-linked to the probe, and Alexa Fluor 532-azide dye was conjugated to the probe via CuAAC. SDS-PAGE and fluorescence imaging allowed visualization of probe-bound tubulin (Figure 6A). We found a concentration-dependent effect on the intensity of a 50 kDa fluorescent band using probe concentrations from 1× to 10× the antiparasitic ${\text{EC}}_{50}^{72\mathrm{h}}$ (Figure 6A). Some samples were simultaneously treated with both probe compound and a 100-fold excess of competitors (MMV676477 and an inactive analog, SW10). Here, dimming of the fluorescent tubulin band is representative of competition between the cross-linker probe and MMV676477, which lacks a photo-cross-linker (Figure 6A). The fluorescence band was significantly dimmer at all four concentrations in samples treated with MMV676477 (Figure 6A,B; see full gels in Figure S6), whereas no apparent competition by the inactive compound SW10 was observed. We then measured the fluorescence intensity of bands through densitometry (Figure 6B), which revealed that the intensity of MMV676477 and SW22 (probe; P) treated-samples was approximately 2-fold less than those treated with the probe alone (Figure 6B; p < 0.01 compared to probe alone by ANOVA).

Figure 6.

Competition-sensitive fluorescent binding of MMV676477 analogs to tubulin. Following treatment with the SW22 probe (“P”) in the presence or absence of competitors, purified L. tarentolae tubulin was subjected to UV cross-linking. Alexa Fluor 532 azide dye was then conjugated to the probe via copper-assisted cycloaddition (CuAAC; “click chemistry”). (A, top) Fluorescence imaging of dye-labeled tubulin samples, with and without competition by MMV676477 (MV) and SW10. Competing compounds were added at 100× the probe concentration (e.g., 1× probe = antiparasitic ${\text{EC}}_{50}^{72\mathrm{h}}$ or 156 nM; competitors added at 15.6 μM). ${\text{EC}}_{50}^{72\mathrm{h}}$ data and structures are shown in Table 4. (A, bottom) Coomassie blue staining of the gel is used as a loading control. (B) Densiometric analyses of fluorescence band intensity from three independent biological replicates + SE; * denotes MMV676477-treated and probe-treated bands that are significantly dimmer than those treated with the probe alone (p < 0.01 by ANOVA). See full gels in Figure S6.

We also treated purified L. tarentolae tubulin that still included some additional proteins (see Methods for purification description) with SW22 and analogs. We found that there was minimal binding of our compound to other proteins in the preparation (Figure S7). Again, we found a concentration dependent effect on the intensity of a 50 kDa fluorescent band. In line with their antiparasitic activity, dimming of the fluorescent tubulin band was evident in the presence of active competitors SW41 and MMV676477, as well as the moderately active SW102 (to a lesser degree). No apparent dimming was observed in the presence of the inactive analogs SW10 and SW23. Overall, these results support our hypothesis that MMV67647 and active analogs directly bind to tubulin.

DISCUSSION

To search for tractable chemotypes for the neglected tropical disease leishmaniasis, we used a phenotypic screen to test the activity of compounds in the MMV “Pathogen Box” against L. amazonensis axenic amastigotes. We identified and validated a hit compound, MMV676477, which also kills other protozoan parasites: Leishmania donovani, Plasmodium falciparum, and Trypanosoma brucei. We showed that MMV676477 and active analogs affect Leishmania cell division and morphology and increase the percentage of polymerized microtubules in parasites. Finally, we determined that this scaffold directly binds to tubulin and selectively promotes polymerization of purified Leishmania tubulin. To our knowledge, we are the first group reporting that a particular scaffold preferentially promotes microtubule polymerization in parasites. Our initial SAR studies on this scaffold highlight potential for improvements in potency and selectivity for future lead optimization studies.

We optimized several assays for screening the MMV Pathogen Box compounds against L. amazonensis axenic amastigotes and identified several compounds with comparable in vitro potency to that of the gold-standard antileishmanial drug, amphotericin B. Most antileishmanial drug screening campaigns have utilized microscopy-based techniques;⁷⁰ however, one difficulty with these methods is that they rely on detection of parasite burden as a proxy for parasite viability. We therefore employed multiple more direct ways to report parasite viability. First, we used an optimized alamarBlue assay to measure metabolic activity of extracellular L. amazonensis axenic amastigotes. Second, as previously reported for other parasites,^37-40 we generated transgenic L. amazonensis parasites stably expressing a luciferase reporter gene, which provided a tractable parasite bioluminescent marker. This assay is rapid and sensitive, and allowed us to validate the activity of our hit compound against intracellular parasites. Finally, we used a washout assay to ensure that our compound’s activity was cytocidal rather than cytostatic. Using these improved techniques, we narrowed our focus to one verified hit.

Because it was identified via a phenotypic screen, the target of MMV676477 was initially unknown. Through a series of experiments using intact parasites and purified protein, we were able to narrow one likely target of our compound from one that affected Leishmania cell division and morphology to one that affected Leishmania microtubules, and then to one that facilitated polymerization of purified Leishmania tubulin. We then confirmed that our scaffold directly binds to tubulin. There are several plausible mechanisms that could underlie why a microtubule stabilizer would result in parasite death. One possibility is that these stabilizers may directly act as antimitotics in Leishmania, which would be similar to the mechanism by which paclitaxel exerts its effects on cancer cells. Another possibility is that parasite death is instead caused by affecting the subpellicular leishmanial microtubular array. If Leishmania is unable to remodel the subpellicular microtubule array during parasite cell division due to treatment with these compounds, the number of parasites that were arrested at some point during cell division would also increase, even though the compounds are not acting directly as an antimitotic. The methodologies employed in this manuscript do not distinguish between these possibilities. Furthermore, we cannot be certain whether tubulin polymerization is the primary or only mechanism that leads to parasite death, as it is impossible to completely rule out additional targets for MMV676477 based on existing data. However, several lines of evidence detailed in this manuscript indicate that MMV676477 enhances tubulin polymerization in Leishmania.

First, MMV676477 promoted the partitioning of cellular tubulin toward the polymeric form in intact parasites, confirming that this scaffold enhances tubulin polymerization in parasites (Figure 4). One caveat is that we used 0.1% Triton X-100 in buffer to separate the polymeric form from the dimeric form in Leishmania amastigote compound-treated and untreated samples, as has previously been done in the field.⁴¹ However, based on our data (Figure 4) and the results in Jayanarayan et al.,⁴¹ Leishmania tubulin polymers seem to tolerate 0.1% Triton X-100 better than mammalian tubulin polymers do. Additionally, all comparisons were made relative to the untreated DMSO control, which was prepared in the same buffer. It is likely that the concentration of Triton X-100 in our buffers led us to underestimate rather than overestimate the degree to which tubulin was polymerized in parasites.⁴²

Second, the antiparasitic activity of MMV676477 and analogs (L. amazonensis axenic amastigote EC₅₀) strongly correlated with their effects on purified Leishmania tubulin as estimated by turbidity assays (Figure 5), further supporting the findings shown in intact parasites (Figure 4) that tubulin is the target of MMV676477. Turbidity measurements at A340 are affected by the tubulin polymer type in addition to the amount of polymer. Because we did not compare dimer to polymer concentrations after our turbidity measurements, the possibility that compounds may induce polymer shape changes (sheet polymers), which would also increase the A340 plateau, could not be completely ruled out. Regardless, our data suggest that MMV676477 and its analogs exert an effect on tubulin polymerization behavior. Furthermore, it should be noted that Leishmania tarentolae and porcine tubulin EC₅₀ values were estimated at 30 °C for leishmanial versus 37 °C for mammalian tubulin, respectively, as is typical in the literature.^32,33 Because increased temperature makes tubulin polymerization more likely and may affect the lag time, polymerization rate, and steady-state level of the A340 curve in turbidity assays, it is difficult to directly compare MMV676477’s selectivity between Leishmania and mammalian tubulin based on turbidity data only. However, the >20-fold selectivity of MMV676477 for purified Leishmania tubulin over porcine tubulin (0.5 vs 11 μM EC₅₀) is consistent with the in vitro selectivity of MMV676477 for L. amazonensis over mammalian cells (79 vs 2200 nM).

Finally, MMV676477 showed direct binding to tubulin in cross-linking experiments (Figure 6). We noticed that the 100-fold excess of the competitor MMV676477 (Figure 6B) resulted in only a 2-fold decrease in the degree of fluorescence of the probe band, as is typical in these studies.⁴³ The reason is that the probe compound SW22 is expected to initially bind to proteins reversibly. However, irradiation of the samples generates a covalent adduct between the probe and its binding partners, rendering the binding irreversible. By contrast, the competitors only bind reversibly. Accordingly, the interaction with the cross-linking probe is not done under equilibrium conditions, and large excesses of competitor are required to block specific interactions with biologically relevant binding partners, such as tubulin in this case. Therefore, the direct tubulin binding seen in our cross-linking studies that can be competed with active but not inactive analogs strongly suggests that tubulin is a target of MMV676477.

Our SAR studies suggest that the benzamide moiety on the MMV676477 scaffold is required for its ability to bind tubulin and promote polymerization. Replacing the substituted N-aryl ring with a simple acetate (SW23) or an unsubstituted benzamide (SW100) resulted in a loss of activity. Replacing the benzamide with a pyridyl amide (SW74) slightly elevated the EC₅₀ value relative to the initial hit, while the urea in compounds SW102 and SW101 decreased activity. The position of the benzoyl group appears important because the O-acylated isomer SW10 was largely devoid of activity, although this analog also lacks the ethyl group. Interestingly, by contrast, disubstituted benzamide (SW41) improved potency and therapeutic index relative to the initial hit. At the other extreme, SW10, which is an isomer of MMV676477, and the truncated analog SW23 are nearly inactive against the Leishmania parasite.

Tubulin is an excellent antiparasitic drug target, as evidenced by the first-line antihelminthic benzamidizoles (e.g., albendazole). Despite a high degree of amino acid sequence conservation for both the α- and β-subunits among eukaryotes,^44-47 certain tubulin-targeting compounds selectively bind to phylogenetically restricted tubulin subsets. Comparison of tubulin sequences from Tetrahymena, Plasmodium, Toxoplasma, Euglena, Trypanosoma, and Leishmania has demonstrated that the protozoan tubulins form a distinct group, and these tubulin proteins are more similar to tubulins from plants than vertebrate or fungal tubulins.⁴⁸ As such, plant antitubulin compounds, such as dinitroanilines, selectively disrupt microtubules from diverse protozoa and plants but not vertebrates and fungi.^49-53 Other studies have identified tubulin as an antiprotozoal target in the past.^51,53-56 However, all compounds that we found reported in the literature targeted trypanosomatid microtubules to prevent tubulin polymerization. By contrast, MMV676477 promotes tubulin polymerization, which provides a unique mechanism for this scaffold. Furthermore, our scaffold has potent antiparasitic activity against both kinetoplastids (Leishmania spp., T. brucei) and apicomplexans (Plasmodium), providing significant potential for eventual use as a broad-spectrum antiparasitic agent. In addition, MMV676477 was initially reported as an anti-mycobacterial agent,⁵⁷ and a recent study also identified MMV676477 as a hit against Staphylococcus aureus,²⁸ although its mechanism of action was not described. Thus, it is plausible that the MMV676477 scaffold could be employed as an antibiotic as well.

CONCLUSION

In conclusion, we have determined that a novel antileishmanial compound from the Pathogen Box binds to Leishmania tubulin, induces parasite microtubule polymerization, and affects Leishmania morphology and cell division. Such a mechanism of action would be unique among antiparasitic agents. Further genetic and biochemical studies using MMV676477 will help us understand differences in its mechanism of action compared to other tubulin-active drugs, as well as differences in tubulin structure, regulation, or dynamics between protozoa and other organisms that may be suggested by the selectivity differences for parasitic and mammalian tubulin seen here. We will also continue our ongoing iterative medicinal chemistry studies in an effort to identify a lead compound that is effective in a murine model of cutaneous leishmaniasis. Due to tubulin’s conservation across the protozoa and the MMV676477 scaffold’s activity against multiple parasites, there is significant potential for this scaffold to allow the development of a broad-spectrum antiparasitic agent that will treat a multitude of devastating protozoal infections.

METHODS

Compounds.

The Pathogen Box was generously provided by the MMV⁵⁸ as 10 mM stocks in DMSO (10 μL each) and stored at −20 °C. The antileishmanial reference drugs amphotericin B and miltefosine (Sigma) were prepared in deionized water and stored at −20 °C. The maximum final DMSO concentration was 0.2% v/v in all experiments.

Parasite Cultures.

Leishmania amazonensis promastigotes (strain IFLA/BR/67/PH8, provided by Norma W. Andrews, University of Maryland, College Park, MD) and L. tarentolae (Parrot strain, ATCC) were maintained at 26 °C in Schneider’s Drosophila medium supplemented with 15% heat-inactivated, endotoxin-free FBS and 10 μg/mL gentamicin.^20,21 L. amazonensis amastigotes were grown axenically at 32 °C in M199 (Invitrogen) at pH 4.5, supplemented with 20% FBS, 1% penicillin–streptomycin, 0.1% hemin (25 mg/mL in 50% triethanolamine), 10 mM adenine, 5 mM l-glutamine, 0.25% glucose, 0.5% trypticase, and 40 mM sodium succinate.^20,21

A transgenic luciferase-expressing line of L. amazonensis parasites (L. amazonensis^luc) was generated similar to the method in refs 39 and 40. Briefly, the 1.66 kb luciferase-coding region of pGL3-Basic (Promega) was cloned in the expression vector pLEXSY.hyg2 (Jenabioscience). The final construct containing the luciferase gene and hygromycin resistance marker was integrated into the 18S rRNA locus of the nuclear DNA of L. amazonensis using the Human T-Cell Nucleofector kit and the Amaxa Nucleofector electroporator (program U-033). Following transfections, after 24 h at 26 °C, transfectants were selected with 100 μg/mL hygromycin in Schneider’s Drosophila medium. Clones were isolated by limiting dilution. L. amazonensis^luc parasites were maintained as above, but the media was supplemented with 100 μg/mL hygromycin. L. amazonensis virulence was maintained by passage in C57B/6 mice.²⁰

For other parasites, L. donovani, strain MHOM/SD/62/1S-C12D, was kindly provided by Robert Duncan (Food & Drug Administration) and was grown as previously described.⁵⁹ Plasmodium falciparum parasites of the 3D7 strain (kindly provided by Margaret A. Phillips, UT Southwestern) were cultured in RPMI 1640 medium supplemented with 37.5 mM HEPES, 10 mM d-glucose, 2 mM l-glutamine, 100 μM hypoxanthine, 25 μg/mL gentamicin, 4% (v/v) human serum, and 0.25% (v/v) Albumax II, at a 2% hematocrit in an atmosphere of 1% O₂, 3% CO₂, and 96% N₂ as described previously.¹⁰ Staging and parasitemia of the in vitro culture were assessed by light microscopy of Giemsa-stained thin blood smears. The parasites were synchronized using sequential sorbitol lysis treatment,^22,60 with experiments carried out at least one intra-erythrocytic cycle later. T. brucei single marker (SM) cells were maintained at log phase growth (<1.5 × 10⁶ cells/mL) in HMI-19 media⁶¹ supplemented with 10% FBS and 2.5 μg/mL G418 (Life Technologies) at 37 °C and 5% CO₂.

Cell Cultures.

RAW 264.7 cells (ATCC TIB-71) were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Invitrogen) as described previously.^20,21

Concentration–Response Assays in L. amazonensis, T. brucei, and P. falciparum.

Leishmania amazonensis axenic amastigotes (100 μL, 2 × 10⁶ cells/mL) were added to 96-multiwell plates containing 100 μL of amastigote culture medium with an appropriate compound dilution series. DMSO at 0.1% (no drug) served as a positive control (100%), and 5 μM amphotericin B served as a negative growth control (0%). Only internal wells were used to minimize edge effects from evaporation. Axenic amastigotes were incubated at 32 °C with a compound or drug for 72 h prior to measurement. Similarly, the cytotoxicity of each compound was determined against RAW 264.7 cells, following 72 h incubation at 37 °C (promastigotes at 26 °C). AlamarBlue (10%, Thermo Fisher Scientific) was used to measure the growth of both parasite and mammalian cells. Conversion from a blue oxidized state to a pink reduced state was assessed visually at 6 h.^17,62,63 The fluorescence signal was measured with a BioTek Synergy H1 plate reader (530 nm excitation, 570 nm emission; BioTek Instruments). For bioluminescence assays using L. amazonensis^luc, relative luminescence units (RLU) were measured using Britelite Plus (PerkinElmer, USA).³⁸ Luminescence produced by this luciferase reaction is proportional to the amount of luciferase-expressing, viable L. amazonensis^luc.³⁸ Following 5 min incubation at room temperature, the luminescence was measured with a BioTek Synergy H1 plate reader. Compounds were tested against the blood-stage Trypanosoma brucei (48 h end point) and intra-erythrocytic asexual stages of Plasmodium falciparum (3D7) using CellTiter-Glo luminescent and Malaria SYBR green I fluorescence (MSF) assays, respectively, as described previously.^21,22,64

All experiments were conducted as three technical replicates on the same plate, with at least three independent biological repeats of each plate performed. For all assays, percent growth was expressed as a proportion of the untreated (positive) control (i.e., 100%) as described previously¹⁰ and plotted against drug or compound concentration. Concentrations were then log₁₀ transformed, and EC₅₀ values for each biological repeat were determined using nonlinear regression (sigmoidal dose–response/variable slope equation) in GraphPad Prism v5.0 (GraphPad Software, Inc.).¹⁰ Values from the three biological replicates were used to calculate the mean EC₅₀ values ± SE shown.

Intramacrophage L. amazonensis Assays.

Intracellular EC₅₀ values were estimated with L. amazonensis^luc parasites. Briefly, RAW 264.7 cells were starved overnight and then infected with metacyclic promastigotes at a multiplicity of infection (MOI) of 15 and incubated for another 24 h at 37 °C. The plates containing infected RAW 264.7 cells were washed five times with serum-free DMEM. Serially diluted compounds were added, and plates were incubated at 37 °C for 72 h. The bioluminescence signal was measured as described above.

We measured intracellular LD₅₀ values using adaptations of protocols described by Paape et al.¹⁷ and Jain and colleagues.⁶⁵ Briefly, RAW 264.7 cells were seeded to 96-multiwell plates at a density of 2 × 10⁵ cells/mL (200 μL) and starved overnight at 37 °C. The cells were infected with metacyclic promastigotes at an MOI of 15 and incubated for another 24 h at 37 °C. The plates containing infected RAW 264.7 cells were washed five times with serum-free DMEM. Serially diluted compounds were added, and plates were incubated at 37 °C for 72 h. Wells were washed five times with DMEM, and the RAW 264.7 cells were lysed with 100 μL of 2 mg/mL saponin in DMEM for 5 min at room temperature,¹⁷ and further lysis was stopped with 100% FBS. After centrifugation, 200 μL of acidic promastigote media was replaced, and plates were incubated at 26 °C for 96 h. Fluorescence intensity (alamarBlue) was measured as described above.

Microscopy.

Promastigotes or amastigotes were treated with the indicated compound concentrations and allowed to adhere to poly(l-lysine) coated plates. To estimate effects on cell division or morphology, we used the estimated ${\text{EC}}_{50}^{72\mathrm{h}}$ concentrations of compounds but treated parasites only for 48 h. For studies of flagellar length, we used the estimated ${\text{EC}}_{50}^{72\mathrm{h}}$ concentrations of compound but treated parasites for only 24 h. Thus, for each experiment, each active compound was used at its respective ${\text{EC}}_{50}^{72\mathrm{h}}$ concentration; for inactive compounds, the values shown in Table 4 were used. All cells were fixed with 4% paraformaldehyde and permeabilized and blocked with 0.01% Triton X-100 and 2% BSA in PBS. Promastigotes and amastigotes were incubated with mouse anti-GP46 or anti-P8, respectively (both kind gifts from Diane McMahon-Pratt, Yale University) at 1:50 or 1:1000 and rat anti-α-tubulin monoclonal YL1/2 antibody (cat. no. MA1-80017, Invitrogen) at 1:1000. Samples were then probed with A568 anti-mouse and A488 anti-rat secondary antibodies (Molecular Probes). DNA was labeled with Hoechst 33342. For representative images, samples were visualized on a Zeiss LSM 880 inverted confocal Airyscan microscope at 63×; full Z-thicknesses through parasites were obtained. Maximal intensity projections were formulated and quantified with ImageJ (1.52a, http://imagej.nih.gov/ij). Parasites in representative images were selected from the maximal intensity projections and linearly processed in Adobe Photoshop CS6 (version 13.0.6).

For quantification, images were obtained with a BioTek Cytation 5 confocal imaging reader and analysis was performed by an observer blinded to experimental condition. For the promastigotes shown in Figure 3C, parasites in 2F1K1N, 2F2K1N, and 2F2K2N were all considered to be undergoing cell division (mid-cell division). For the amastigotes shown in Figure 3C, parasites that appeared to be in 2K1N or 2K2N were considered to be in mid-cell division. For both categories, at least 200 parasites were analyzed per condition per experiment after 48 h incubation at compounds’ ${\text{EC}}_{50}^{72\mathrm{h}}$, and the mean percentage in mid-cell division plus standard error for three experiments was calculated. To calculate the average flagellar length shown in Figure S1, the flagellar length was measured in the gp46 channel in ImageJ for at least 50 parasites per experimental condition for each of 3 experiments after 24 h incubation at compounds’ ${\text{EC}}_{50}^{72\mathrm{h}}$; mean lengths for each experimental category (normalized to DMSO control flagellar lengths) plus SE are shown. To calculate the percentages of pear-shaped or rounded parasites shown in Figure S1, parasites with cell bodies that had a width of ≥70% their length were counted. At least 200 parasites were analyzed per condition per experiment after 48 h incubation at compounds’ ${\text{EC}}_{50}^{72\mathrm{h}}$, and the mean percentage of pear-shaped parasites per experimental category plus SE for three experiments was calculated.

Polymerized versus Unpolymerized Tubulin in Parasites.

Soluble (unpolymerized) and insoluble (polymerized) tubulin fractions were separated by high-speed centrifugation, as described previously,^41,66,67 with minor modifications. Amastigotes or promastigotes of L. amazonensis were seeded at a density of 2 × 10⁷ cells in 2 mL and treated with the specified compounds for 24 h at their EC₅₀ concentration. Parasites were lysed in buffer containing 100 mM 2-(N-morpholino) ethanesulfonic acid (MES, pH 6.9), 1 mM EGTA, 1 mM MgSO₄, 0.1% Triton X-100, and protease inhibitors (1 mM benzamidine, 0.5 mM phenylmethylsulfonyl fluoride, 25 μg/mL leupeptin) and centrifuged at 100 000g to separate the polymerized and unpolymerized tubulin fractions. The pellets were dissolved in 0.5% SDS in 25 mM Tris, pH 6.8, in a volume equal to that of the supernatant.

Electrophoresis, Western Blotting, and Protein Assays.

SDS-PAGE was performed using 12% polyacrylamide gels as described previously.^32,54 Protein purity and concentration were assessed using Coomassie Blue staining and the bicinchoninic acid (BCA) (Pierce Biotechnology) assay, respectively, following the manufacturers’ protocols. Bovine serum albumin (BSA) was included to generate a standard curve. Western blotting was performed as described.^41,61 Total proteins were resolved by polyacrylamide gel electrophoresis and transferred to PVDF membranes using a Mini Trans-Blot Cell (BioRad). The membranes were blocked in 5% nonfat dry milk in Tris-buffered saline (TBS) (20 mM Tris (pH 7.6), 150 mM NaCl). The membranes were incubated with mouse anti-α-tubulin antibody (monoclonal antibody B-5-1-2, cat. no. 32-2500, Invitrogen) at 1:1000 in 5% BSA and TBS-T overnight at 4 °C. The membranes were then incubated with a 1:8000 dilution of goat anti-mouse HRP-conjugated secondary antibody in TBS-T (5% nonfat dry milk) for 1 h. Finally, membranes were incubated in SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) for 5 min and visualized by the ImageQuant LAS 4000 (GE Healthcare). GAPDH antibodies (Santa Cruz) were used at a 1:1000 dilution as a loading control.

Tubulin Purification from Leishmania tarentolae.

L. tarentolae (Parrot strain from ATCC) tubulin was purified as originally described by Yakovich et al.³² and others.⁵⁴ Promastigotes of L. tarentolae were grown to a high density (~1 × 10⁸ cells/mL), harvested, and resuspended in PME + P buffer containing 100 mM piperazine N, N′-bis (2-ethanesulfonic acid) (PIPES) buffer (pH 6.9), 1 mM glycol ether diamine tetraacetic acid (EGTA), 1 mM MgCl₂, 1 mM benzamidine, 0.5 mM phenylmethylsulfonyl fluoride, and 25 μg/mL leupeptin. The resulting suspension was lysed using an Emulsiflex C5 homogenizer (Avestin) or extensively sonicated on ice with a probe sonicator (Misonix), cooled on ice for 30 min, and centrifuged at 40 000g for 1 h at 4 °C using an ultracentrifuge (Beckman). The resulting supernatant was filtered through a glass wool or 0.45 μm filter. Using the peristaltic pump, the sample was loaded onto an equilibrated DEAE-Sepharose Fast Flow matrix (Amersham Biosciences). The column was washed with two column volumes of PME + P and subsequently four column volumes PME + P containing 0.1 M KCl and 0.25 M glutamate (pH 6.9). Tubulin that still contained some additional proteins was then eluted with two column volumes of PME + P containing 0.3 M KCl and 0.75 M glutamate (pH 6.9). Samples for studies shown in Figure S7 were removed at this point. For full tubulin purification for the reminder of studies shown, the column was connected to the AKTA fast performance liquid chromatography system (GE), and tubulin was eluted with two column volumes of PME + P containing 0.3 M KCl and 0.75 M glutamate (pH 6.9). The purified fractions were confirmed using SDS-page and Coomassie blue staining. The tubulin-rich fractions were confirmed by tubulin polymerization assays as described previously. The assembly competent tubulin fractions were pooled together and subjected to dialysis overnight in 1× PME buffer. Tubulin was concentrated by using Amicon ultracentrifugal filters (Millipore Sigma), flash frozen in liquid nitrogen, and stored at −80 °C.

Tubulin Polymerization Assays.

Tubulin polymerization assays were adapted from protocols from Cytoskeleton, Inc., and others^14,32,54,68 using 96-well half-area microplates (Costar) in a final volume of 100 μL. Leishmanial or porcine tubulin (>99% pure, cat. no. T240, Cytoskeleton, Inc.) in 50 μL volume was pretreated with drugs on ice for 5 min before adding 50 μL of ice cold buffer to provide a final concentration of 3 mg/mL tubulin in 80 mM PIPES (pH 6.9), 2 mM MgCl₂, 0.5 mM EGTA, 1 mM GTP, and 1% DMSO unless specified otherwise. The absorbance at 340 nm was recorded in a Synergy H1 microplate reader (BioTek) for up to 45 min at 37 °C (porcine tubulin) or 30 °C (L. tarentolae tubulin). To estimate the EC₅₀, tubulin was exposed to 8 concentrations of drug as above. A control sample with 10% DMSO was included to create the V_max 100% standard (positive control, maximal tubulin polymerization). Untreated sample was used as a negative control (0%) as described in refs 14, 32, 54, and 68. The V_max data at each drug concentration used were converted into percent of the control V_max. Percent tubulin polymerization was then expressed as a proportion of the untreated control and plotted against log₁₀ transformed drug or compound concentration. EC₅₀ values were estimated using nonlinear regression (sigmoidal dose–response/variable slope equation) in GraphPad Prism v5.0.¹⁰

Fluorescent Microtubule Assembly Assays.

Fluorescence images of microtubule assembly were acquired as described.^32,33 Briefly, 1.5 mg/mL purified porcine tubulin (>99% pure, cat. no. T240, Cytoskeleton, Inc.) was treated for 40 min in assembly buffer containing 80 mM PIPES (pH 6.9), 1 mM EGTA, 1 mM MgCl₂, 1 mM GTP, and 1% DMSO at 37 °C. The mixture was cross-linked by diluting it 10-fold using assembly buffer containing 1% glutaraldehyde. After 3 min, the reaction was quenched by diluting 5-fold with assembly buffer containing 20 mM Tris, pH 6.8. Pedestals were inserted into centrifuge tubes, and a poly(l-lysine) coated coverslip was placed on top. Glycerol, 20% in assembly buffer with no GTP, was made, and the poly(l-lysine) coated coverslips were covered with 3 mL of cushion. The quenched, cross-linked reactions (50 μL) were gently layered on top and spun through the cushion onto the poly(l-lysine) coated coverslips using an ultracentrifuge (22 500g at 20 °C for 45 min or 4000g for 12 h at 20 °C). Coverslips were washed three times with assembly buffer (no GTP), fixed with ice-cold methanol, and stained for 20 min with FITC-DM1α (anti-α tubulin, cat. no. F2168 Sigma-Aldrich) diluted 250× in PBS + BSA. The coverslips were washed three times with assembly buffer and imaged by epifluorescence as described by Ayaz et al.³⁶

Chemical Cross-Linking Experiments.

A probe compound (SW22) with benzophenone and alkyne modifications was synthesized (for details, see the synthesis of N-(1-(5-ethyl-4-methyl-6-oxo-1,6-dihydropyrimidin-2-yl)-3-methyl-1H-pyra-zol-5-yl)-4-(4-(prop-2-yn-1-yloxy) benzoyl) benzamide). A click-chemistry protocol was adapted from Theodoropoulos et al.⁶⁹ Purified parasite tubulin at 10 μM (starting concentration) was plated in 96 well plates, treated with the probe in the presence or absence of highly active and less active competitors, and polymerized for 1 h at 30 °C. The samples were UV cross-linked by placing the 96 well plates on ice approximately 3–4 in. below the bulbs in a Stratalinker (Stratagene) and then exposing them to 15 min of UVB radiation. The samples were immediately solubilized in 1% SDS with benzonase (Sigma) diluted 1:20000 in buffer containing 50 mM HEPES, pH 7.4, 10 mM KCl, and 2 mM MgCl₂. The samples were normalized for protein concentrations using the BCA assay (Life Technologies). Equal amounts of sample were subjected to a click reaction with 100 μM TBTA (dissolved in 4:1 DMSO/t-butanol), 1 mM TCEP, 2 mM CuSO₄ and 25 μM Alexafluor-532 azide (see Theodoropoulos et al.⁶⁹ for synthesis of Alexafluor-532 azide) for 1 h at 25 °C with agitation. SDS sample buffer was then added to the samples to quench the reaction, and proteins were resolved by SDS-PAGE. Samples in Figure 6 and Figure S6A were run using a 4–12% gel (BioRad precast), while those in Figure S6B,C were run using a 12% gel (custom-made). A Typhoon scanner with a 532 nm excitation laser and a 555 nm emission filter was used to scan the gels for fluorescently labeled proteins.

Synthesis of MMV676477 Analogs.

A total of 11 analogs were synthesized, and all tested compounds have a purity of >95% as judged by HPLC analysis (UV detection at 210 nM) (see Supplemental Methods for full details).

ABBREVIATIONS USED

CuAAC

copper-catalyzed alkyne–azide cycloaddition

MMV

Medicines for Malaria Venture

Mϕ

macrophage

SI

selectivity index