(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .
MyosinA is a druggable target in the widespread protozoan parasite Toxoplasma gondii [1]
['Anne Kelsen', 'Department Of Microbiology', 'Molecular Genetics', 'University Of Vermont Larner College Of Medicine', 'Burlington', 'Vermont', 'United States Of America', 'Robyn S. Kent', 'Anne K. Snyder', 'Eddie Wehri']
Date: 2023-05
Toxoplasma gondii is a widespread apicomplexan parasite that can cause severe disease in its human hosts. The ability of T. gondii and other apicomplexan parasites to invade into, egress from, and move between cells of the hosts they infect is critical to parasite virulence and disease progression. An unusual and highly conserved parasite myosin motor (TgMyoA) plays a central role in T. gondii motility. The goal of this work was to determine whether the parasite’s motility and lytic cycle can be disrupted through pharmacological inhibition of TgMyoA, as an approach to altering disease progression in vivo. To this end, we first sought to identify inhibitors of TgMyoA by screening a collection of 50,000 structurally diverse small molecules for inhibitors of the recombinant motor’s actin-activated ATPase activity. The top hit to emerge from the screen, KNX-002, inhibited TgMyoA with little to no effect on any of the vertebrate myosins tested. KNX-002 was also active against parasites, inhibiting parasite motility and growth in culture in a dose-dependent manner. We used chemical mutagenesis, selection in KNX-002, and targeted sequencing to identify a mutation in TgMyoA (T130A) that renders the recombinant motor less sensitive to compound. Compared to wild-type parasites, parasites expressing the T130A mutation showed reduced sensitivity to KNX-002 in motility and growth assays, confirming TgMyoA as a biologically relevant target of KNX-002. Finally, we present evidence that KNX-002 can slow disease progression in mice infected with wild-type parasites, but not parasites expressing the resistance-conferring TgMyoA T130A mutation. Taken together, these data demonstrate the specificity of KNX-002 for TgMyoA, both in vitro and in vivo, and validate TgMyoA as a druggable target in infections with T. gondii. Since TgMyoA is essential for virulence, conserved in apicomplexan parasites, and distinctly different from the myosins found in humans, pharmacological inhibition of MyoA offers a promising new approach to treating the devastating diseases caused by T. gondii and other apicomplexan parasites.
Funding: This work was supported by the National Institutes of Health (AI139201 and AI137767 to GEW, each including salary support; GM141743 to DMW, including salary support; F31AI145214 to RVS, including predoctoral fellowship stipend support; and T32AI055402 to GEW, including predoctoral fellowship stipend support for AKS). The work was also supported by the Canadian Institutes of Health Research (148596 to MJB), the Canada Research Chair program (to MJB, salary support) and the American Heart Association (20POST35220017 to RSK, including postdoctoral fellowship stipend support). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
We describe here the identification and characterization of KNX-002, the first specific inhibitor of an apicomplexan class XIVa myosin. KNX-002 inhibits TgMyoA motor activity, parasite motility, and parasite growth in culture in a dose-dependent manner, with little or no effect on a variety of other vertebrate myosins. On-target activity in parasites was confirmed by identifying a mutation in TgMyoA that rescues parasite motility and growth in the presence of the compound. The availability of a specific inhibitor of TgMyoA, together with isogenic parasite lines that show differential sensitivity to the compound, enabled us to undertake the first direct test of whether pharmacological inhibition of MyoA can be used to alter the progression of disease caused by an apicomplexan parasite.
Depletion of TgMyoA results in severely reduced parasite motility, host cell invasion, and host cell egress [ 26 – 29 ]. As a result, parasites lacking TgMyoA are avirulent in an animal model of infection [ 28 ]. Despite its importance in the T. gondii life cycle, we know little about how TgMyoA generates the complex pattern of parasite motility seen in 3D or how TgMyoA function contributes to disease pathogenesis in an infected host. Because the parasite can compensate for the loss or reduced expression of proteins important to its life cycle [ 30 – 32 ], small-molecule inhibitors of TgMyoA would serve as valuable complementary tools for determining how different aspects of motor function contribute to parasite motility and the role played by TgMyoA in parasite dissemination and virulence. Specific inhibitors are also necessary to establish TgMyoA as a “druggable” target, i.e., a protein whose activity is amenable to inhibition by small molecules, and to test whether pharmacological inhibition of TgMyoA activity in vivo alters the course of an infection.
Within the brain and other tissues of a patient suffering from active toxoplasmosis, the tachyzoite stage of the parasite uses a unique form of substrate-dependent “gliding” motility to invade into and egress from the cells of its host, to spread throughout the patient’s tissues, and to migrate across biological barriers [ 12 – 15 ]. T. gondii tachyzoites move in the extracellular matrix along flattened corkscrew-like trajectories, with regularly repeating periods of acceleration and deceleration [ 16 ]. The mechanisms underlying gliding motility are thought to be conserved among apicomplexan parasites, including Plasmodium spp. (the causative agents of malaria) and Cryptosporidium spp. (which cause severe diarrheal disease). Motility is driven, at least in part, by an unusual class XIVa myosin motor protein, myosinA (MyoA), found only in apicomplexan parasites and a few ciliates. In T. gondii, this motor consists of the TgMyoA heavy chain and 2 associated light chains, TgMLC1 and either TgELC1 or TgELC2 [ 17 – 19 ]. TgMyoA is a single-headed motor with a very low duty ratio (i.e., it remains strongly bound to actin for only approximately 1% of its catalytic cycle) yet moves actin filaments at the same speed as skeletal muscle myosin, a double-headed motor with a 5-fold higher duty ratio [ 20 , 21 ]. TgMyoA lacks a conventional tail, as well as several residues conserved in other myosins that normally function to regulate actomyosin activity [ 22 , 23 ] and employs atypical strategies for chemomechanical coupling and force transduction [ 24 , 25 ].
Nearly one third of the world’s population is or has been infected with the apicomplexan parasite, Toxoplasma gondii. Although most infections are subclinical, acute toxoplasmosis can have severe consequences in neonates and immunocompromised individuals. Congenital infection can lead to spontaneous abortion or stillbirth, and even children born with subclinical infection frequently experience sequelae later in life including neurological damage and vision impairment [ 1 , 2 ]. Among the immunocompromised, toxoplasmic encephalitis is a particularly significant risk in AIDS patients who are unaware of their HIV status [ 3 – 5 ] and in the approximately 50% of HIV-infected individuals worldwide who do not have access to antiretroviral therapy [ 6 , 7 ]. In individuals with AIDS, adverse effects of the currently available drugs to suppress toxoplasmosis [ 8 ] cause the discontinuation of treatment in up to 40% of patients [ 4 , 9 – 11 ], highlighting the need for new, better-tolerated drugs to reduce toxoplasmosis-related morbidity and mortality.
When the same experiment was done with parasites expressing the T130A mutant myosin, the vehicle-treated mice did not survive beyond day 8 ( Fig 9 , bottom panel, black circles), similar to what we observed with wild-type parasites. In contrast to infection with wild-type parasites, KNX-002 treatment did not significantly improve the survival of mice infected with the T130A parasites ( Fig 9 , bottom panel, orange squares).
Four groups of mice were injected intraperitoneally on day 0 with either KNX-002 (20 mg/kg) or the equivalent volume of DMSO (vehicle). Thirty minutes later, the mice were infected with 500 wild-type parasites (RH; top panel) or 500 parasites expressing T130A mutant myosin (bottom panel). On day 2, the mice were treated again either with compound or DMSO. Differences between the survival curves were analyzed using a log-rank Mantel–Cox test (p = 0.004 for RH vs. RH + KNX-002 [top panel]; p = 0.922 for T130A vs. T130A + KNX-002 [bottom panel]) and p = 0.008 for RH + KNX-002 vs. T130A + KNX-002 (orange curves). The measurements underlying the data plotted in this figure can be found in S21 Data .
Finally, we tested whether KNX-002 (20 mg/kg, administered intraperitoneally on the day of infection and 2 days later) can alter the progression of disease in a mouse model of infection. Only 20% of the mice that were infected with wild-type parasites and treated with vehicle were still alive 8 days postinfection, and none survived to day 9 ( Fig 9 , top panel, black circles). In contrast, 90% of the mice treated with KNX-002 were still alive on day 8, 80% survived to day 9, and 40% were still alive on day 11 when the experiment was terminated ( Fig 9 , top panel, orange squares).
(A) Representative images of plaque assays using wild-type (WT; top row) and T130A mutant (bottom row) parasites in the presence of 0 (DMSO vehicle only), 20, 40, and 80 μM KNX-002, 7 days after inoculating confluent HFF monolayers with 200 parasites/well. An 11.3 × 11.3 mm area from the middle of the well is shown; scale bar = 2 mm. (B) Total number of plaques per well and average plaque area of HFF monolayers inoculated with 200 parasites/well (WT or T130A) and grown for 7 days in the presence of 0 (DMSO), 20, 40, or 80 μM KNX-002. Bars show the mean of 3 biological replicates ± SEM. Treatments were compared by Student’s two-tailed unpaired t test; only the statistically significant differences (p < 0.05) are indicated. The measurements underlying the data plotted in panel B can be found in S20 Data . HFF, human foreskin fibroblast; WT, wild-type.
Plaque assays were used to compare the growth of wild-type and mutant parasites in culture in the presence of KNX-002. Parasites expressing the mutant motor formed a significantly larger number of plaques in the presence of 20 μM compound than wild-type parasites, and even at 40 μM KNX-002 the mutant parasites were still able to generate small plaques ( Fig 8 ).
The displacement of both wild-type and mutant parasites is also inhibited by KNX-002 ( S7A and S12 Figs), but the mutants are again less sensitive to the compound, i.e., despite moving shorter distances in the absence of compound (compare the 2 sets of gray bars in S12 Fig ), the T130A parasites moved farther than wild type in the presence of 20 μM KNX-002 (compare the third and sixth bars in S12 Fig ). The decreased sensitivity of the T130A parasites to the effects of KNX-002 is apparent in the maximum intensity projections from the 3D motility assay, which show more T130A parasites than wild type moving in the presence of 20 μM KNX-002, with longer average displacements (compare right 2 sets of panels in Fig 7A ).
We next asked whether parasites expressing the T130A mutation were less sensitive to KNX-002 in the 3D motility assay. As shown above ( Fig 4C ), KNX-002 strongly inhibits the fraction of wild-type parasites that move in Matrigel ( Fig 7B , left 3 sets of bars). In contrast, the fraction of T130A parasites that move during the assay was insensitive to KNX-002, up to concentrations of 20 μM ( Fig 7B , right 3 sets of bars). The parasite speed data are more variable (Figs 7C and S7B ) but show that the mutant parasites retain some sensitivity to the compound ( Fig 7C , right 3 sets of bars). To capture the combined impact of changes to these 2 motility parameters and to facilitate comparison to the in vitro motility results with recombinant motor, we calculated the “parasite motility index” (the fraction of parasites moving multiplied by the average speed of those that do move) under each condition. The wild-type parasite motility index is more severely impacted by KNX-002 treatment than the motility index of T130A parasites ( Fig 7D ).
(A) Large panels: representative maximum intensity projections showing the trajectories of wild-type (WT) and T130A mutant parasites during 80 s of motility in Matrigel, in the presence of DMSO (vehicle control) or 20 μM KNX-002. The grayscale images were inverted to provide clearer visualization of the trajectories; scale bar = 100 μm. The boxed area in each large panel is magnified and shown to the right; scale bar = 10 μm. (B) Percent of WT and T130A mutant parasites moving in the presence of DMSO (vehicle), 10 μM KNX-002, or 20 μM KNX-002. (C) The mean speeds of the parasites analyzed in panel B. (D) The corresponding parasite motility index (fraction of parasites moving × mean speed). Each data point in panels B–D represents a single biological replicate composed of 3 technical replicates. Three of the biological replicates were collected on the same 3 days (circles), and 3 of the biological replicates were collected on a different 3 days (squares). Bars show the mean of the biological replicates ± SEM. Sets of biological replicates using the same parasite line and collected on the same days were compared by Student’s one-tailed paired t tests (significance indicated above the graphs). Sets of biological replicates comparing different parasite lines were analyzed by Student’s two-tailed unpaired t tests (significance indicated below the graphs). ns = not significant (p > 0.05). The measurements underlying the data plotted in Panels B–D can be found in S18 and S19 Data. WT, wild-type.
The resistant parasite line generated by mutagenesis and selection (clone R3, Fig 5B ) is expected to contain as many as 70 different mutations throughout its genome [ 40 ] in addition to the T130A mutation. To confirm that resistance to KNX-002 in this parasite line was due to the identified mutation in TgMyoA, we recreated the single T130A mutation in the wild-type background using CRISPR/Cas9 ( S10 Fig ). To quantitatively evaluate effects of the mutation on the parasite’s lytic cycle, we compared wild-type and T130A parasites in a plaque assay, which encompasses the parasite’s entire lytic cycle. The T130A mutation had no significant effect on the number of plaques formed or plaque area ( S11A and S11B Fig ), demonstrating little to no effect of the mutation on the parasite’s lytic cycle or growth in culture. Consistent with the plaque assay results, the T130A mutation also had little to no effect on the fraction of the parasite population moving in the 3D motility assay (compare the leftmost pair of panels in Fig 7A and 2 sets of gray bars in Fig 7B ) and only a minor effect on their mean trajectory speeds (compare the 2 sets of gray bars, Fig 7C ). Finally, the T130A mutation had no detectable effect on the virulence of the parasite in a mouse model of infection ( S11C Fig ).
In contrast to the wild-type motor, KNX-002 had little to no effect on the fraction of actin filaments moved by the motor containing the T130A mutation ( Fig 6B , top panel). Furthermore, although the basal filament sliding speed of the mutant motor was lower than that of wild type, this reduced sliding speed was also insensitive to compound ( Fig 6B , middle panel), as was the mutant motor’s overall filament motility index ( Fig 6B , bottom panel).
Including KNX-002 in the in vitro motility assay with wild-type motor inhibits both the fraction of actin filaments moving and the speed of those filaments that do move, each in a dose-dependent manner ( Fig 6A , top and middle panels). Combining these 2 effects into a single “filament motility index” (i.e., the fraction of filaments moving multiplied by the mean speed of the filaments that do move) provides a better measure of the overall effect of the compound in the assay. KNX-002 induces a large decrease in the filament motility index ( Fig 6A , bottom panel).
In vitro motility assays, showing the fraction of actin filaments moving, their sliding speed, and the corresponding filament motility index (fraction moving × mean speed) of TgMyoA treated with various concentrations of KNX-002, as indicated (0 = DMSO vehicle only). (A) Wild-type TgMyoA, (B) T130A TgMyoA. Bars show the mean of 3 independent experiments ± SEM. For each motility parameter, compound treatments were compared to DMSO by one-way ANOVA with Dunnett’s test for multiple comparisons. Only the statistically significant differences (p < 0.05) are indicated. The measurements underlying the data plotted in this figure can be found in S13 Data .
(A, B) Left panels: tdTomato-expressing wild-type parasites (A) or clone R3 (B), which was generated by mutagenesis and selection in KNX-002, were preincubated with various concentrations of KNX-002 for 5 min and then added to HFF cells on a 384-well plate. Fluorescence was measured daily over the next 7 days to quantify parasite growth in the continuing presence of compound. RFU = relative fluorescence units. The data shown are from 3 independent biological replicates, each consisting of 2 to 3 technical replicates at all time points. Right panels: the data from day 5 (D5) of the growth assays were used to calculate the IC 50 and 95% CI of KNX-002 for parasite growth. Vertical bars indicate SEM (left panels) or 95% CI (right panels). (C) Multiple sequence alignment showing the high degree of conservation of T130 in apicomplexan MyoAs (Tg = T. gondii, Pf = Plasmodium falciparum, Cp = Cryptosporidium parvum) and MyoIc, Ie and II from Homo sapiens (Hs) and Dictyostelium discoideum (Dd). (D) Structure of the TgMyoA (PDB ID [6DUE]) motor domain, showing the position of T130 (red) relative to the converter, U50 and L50 subdomains. The measurements underlying the data plotted in panels A and B can be found in S11 and S12 Data. CI, confidence interval; HFF, human foreskin fibroblast.
To independently establish that the T130A mutation alters the sensitivity of TgMyoA to KNX-002, we expressed recombinant TgMyoA containing the T130A mutation and compared its motion-generating capacity to that of wild-type TgMyoA in an in vitro motility assay that measures the speed with which fluorescently labeled actin filaments are moved by myosin adsorbed to a glass coverslip [ 20 , 41 ]. In the absence of added compound, the wild-type motor moved 92.5 ± 1.2% of the filaments on the coverslip, with an average mean speed of 4.6 ± 0.3 μm/s ( Fig 6A , top and middle panels, gray bars). The T130A mutation had no effect on the fraction of filaments moving in the assay but decreased the basal speed of filament sliding by 77 ± 3% compared to wild-type ( Fig 6B , top and middle panels, gray bars). Accordingly, we tested whether the mutant was misfolded, prone to aggregation, or otherwise unstable by comparing its size-exclusion chromatography (SEC) elution profile and thermal stability, as measured by differential scanning fluorimetry (DSF), to that of the wild-type motor ( S9 Fig ; both proteins assayed in the near-rigor [SEC] and pre-powerstroke [DSF] states). No mutation-associated differences were seen in either assay, suggesting that the effects of the T130A mutation on motor function are not due to large structural changes in the protein caused by the amino acid substitution.
The results presented thus far show that KNX-002 is an inhibitor of TgMyoA activity, parasite growth in culture, and parasite motility. The data do not, however, directly link the compound’s effects on parasite motility and growth to inhibition of TgMyoA function; it remains possible that the inhibition of parasite motility and growth are due to off-target effect(s). As a first step towards validating the specificity of the compound, we chemically mutagenized parasites with N-ethyl-N-nitrosourea [ 40 ] and selected for parasites resistant to KNX-002 by growth in 40 μM KNX-002. Parasites that grew up after 10 to 14 days under selection were cloned and their sensitivity to KNX-002 determined using the plate-based growth assay. We isolated 26 such clones from 2 independent mutagenesis/selection experiments, 5 of which showed at least a 2.5-fold increase in IC 50 for KNX-002 ( S8 Fig ). cDNA was generated from the 5 clones, and each of the motor constituents (TgMyoA, TgMLC1, TgELC1, and TgELC2) was sequenced. Four of the five resistant clones contained no amino acid-altering mutations in the motor genes; the fifth (clone R3) contained a single A to G point mutation in the coding sequence of TgMyoA that changes threonine130 to alanine (T130A). This clone showed a 2.9-fold increase in IC 50 for KNX-002 in the growth assay ( Fig 5A and 5B ), from 14.9 μM (wild type; 95% CI = 10.2 to 22.6 μM) to 43.4 μM (T130A mutant; 95% CI = 37.0 to 51.2 μM). T130 is highly conserved in eukaryotic myosins ( Fig 5C ) and located on the first of 7 β-strands comprising the transducer subdomain ( Fig 5D ). The residues immediately flanking T130 are also well conserved between the MyoA homologs of other apicomplexan parasites, but diverge in the vertebrate myosins examined ( Fig 5C ).
To more quantitatively evaluate differences in parasite motility in the presence of KNX-002, we used a 3D motility assay that automatically analyzes hundreds to thousands of trajectories in a single experiment, providing robust statistical comparisons between populations of parasites [ 16 , 21 ]. KNX-002 treatment decreases the fraction of the parasite population moving in the 3D assay in a dose-dependent manner ( Fig 4B and 4C ; IC 50 = 6.2 μM, 95% CI 4.1 to 9.4 μM). Of those parasites that move, KNX-002 treatment also reduces their average displacement and speed ( S7 Fig ).
(A) Representative images from a 2D trail assay of parasites treated with DMSO (vehicle) only, 10 μM KNX-002, or 25 μM KNX-002. The parasites and trails were visualized by indirect immunofluorescence using an antibody against TgSAG1 after 15 min of gliding on the coverslip. Scale bar = 20 μm. (B) Representative maximum intensity projections showing parasite trajectories during 60 s of motility in Matrigel in the absence (DMSO) or presence of 5 μM KNX-002. Scale bar = 40 μm. The grayscale images were inverted to provide clearer visualization of the trajectories. (C) Fraction of the parasite population moving during a 60-s assay in the presence of the indicated concentrations of KNX-002 (IC 50 = 6.2 μM, 95% CI = 4.1–9.4 μM). Each data point represents a single biological replicate consisting of 3 technical replicates. Sets of data captured on the same days, indicated by similar symbol shapes, were compared by Student’s one-tailed paired t tests. The measurements underlying the data plotted in panel C can be found in S9 and S10 Data. CI, confidence interval.
To test whether KNX-002 inhibits parasite motility, we first used a standard immunofluorescence-based assay that visualizes the protein “trails” left behind the parasites as they glide on a glass coverslip (e.g., [ 38 , 39 ]). Untreated parasites deposit a variety of helical, meandering, and large diameter circular trails in this qualitative 2D assay ( Fig 4A ). Treatment with 10 μM KNX-002 results in a different motility pattern, consisting almost entirely of small diameter circles, and at 25 μM KNX-002 most parasites do not move at all ( Fig 4A ).
A series of additional analogs containing the p-OMe aryl substituent were also assessed. Substitution of the thiophene ring in KNX-002 for a phenyl ring (in VEST6, highlighted in blue, Fig 3 ) led to a drop in activity in both the in vitro ATPase and growth inhibition assays. However, the retention of some biological activity by VEST6 may provide a way to assess rapidly the importance of additional substitution in this region of the molecule in future studies through the incorporation of substituents on the new phenyl ring. No tolerance for replacement of the pyrazole (highlighted in green) and thiophene (blue) rings in KNX-002 was observed, with analogs containing only a phenyl, thiazole, or furan substituent, instead of the thiophene-linked pyrazole, all being inactive in both assays ( VEST7 , VEST8 , and VEST9 ). Substitution of a pyrrole into KNX-002 to give VEST10 led to loss of activity in the in vitro assay but retention of some activity in the growth inhibition assay. Finally, the importance of the cyclopropyl group in KNX-002 (highlighted in purple, Fig 3 ) was confirmed by the lack of biological activity in either assay of the des-cyclopropyl analog VEST11 . The growth inhibition activity associated with VEST5 was also shown to require the cyclopropyl group as analogs VEST12 (CH 2 ) and VEST13 (CMe 2 ) were both less active in the growth assay. Attempts to incorporate a stereogenic center through the replacement of the cyclopropyl group with a single methyl substituent led to analogs with decreased biological activity and mild cytotoxicity against HFF cells ( VEST14 and VEST15 ).
The structures of KNX-002 and 15 analogs (VEST1-15) are shown at the top. The table summarizes the activity of each of the compounds in assays measuring recombinant TgMyoA ATPase activity, parasite growth, and mammalian cell toxicity; the data on which the table is based are shown in S3 – S5 Figs. ATPase assay data correspond to relative luminescence units × 10 −6 (mean from 3 independent replicates +/-SEM) in the presence of 20 μM compound ( S3 Fig ). Green = less than 20% inhibition, yellow = 20%–40% inhibition; orange = 45%–70% inhibition; red = greater than 70% inhibition compared to vehicle (DMSO) controls. Growth assay data correspond to calculated IC 50 and 95% CI from a 5 to 6 day growth assay ( S4 Fig ). Red = IC 50 less than 20 μM; orange = IC 50 of 20–35 μM; yellow = IC 50 of 35–50 μM; green = no detectable growth inhibition relative to vehicle (DMSO) controls. Toxicity assay entries summarize the results from 72-h CellTox green toxicity and CellTiter-Glo viability assays on both human foreskin fibroblasts and HepG2 cells. Thresholds for mild and moderate toxicity and loss of viability compared to vehicle (DMSO) controls are indicated by the colored bars on the right-hand side of S5 Fig . CI, confidence interval; SAR, structure-activity relationship.
Next, we used the ATPase and parasite growth assays to carry out a preliminary structure-activity relationship (SAR) study, with the goal of determining which parts of KNX-002 are important for its biological activity. We began by modifying the p-OMe substituted aryl ring in KNX-002 (colored red in KNX-002 structure, Fig 3 ). Replacement of this structural unit in full by a hydrogen atom (in VEST1 ) confirmed the importance of this part of the molecule for biological activity (Figs 3 and S3 and S4 ). In terms of the preferred substituent on this aryl ring, inhibition of both in vitro ATPase activity and parasite growth were retained when the p-OMe substituent in KNX-002 was replaced by p-Cl ( VEST2 ), and partially retained in analogs containing p-Me ( VEST3 ) or m,p-dichloro ( VEST4 ) substituents. VEST5 , an analog in which the p-OMe in KNX-002 was replaced by a hydrogen, was significantly less active than KNX-002 in the purified protein assay but a strong inhibitor of parasite growth. VEST5 was not toxic to host cells, extracellular parasites, or intracellular parasites ( S5 and S6 Figs); further studies will be required to understand the activity of this analog in the parasite growth assay. The mild to moderate toxicity levels shown by VEST2 (p-Cl) against HepG2 and human foreskin fibroblast (HFF) cells, compared to the lack of toxicity in either cell line demonstrated by KNX-002, led to the preliminary conclusion that the p-OMe aryl substituent present in the starting KNX-002 analogue was preferred.
To exclude the possibility that the effect of KNX-002 on parasite growth was due to a deleterious effect on the host cells, we also tested the effect of the compound on human foreskin fibroblasts (the host cells used in the growth assays) and human HepG2 liver cells (sensitive indicators of cytotoxicity and genotoxicity [ 37 ]) in 72-h multiplexed cytotoxicity and cellular viability assays. KNX-002 showed no detectable deleterious effect on either cell line in either assay at the highest concentration tested (80 μM; Fig 2C ).
Parasite growth in culture involves repeated lytic cycles comprised of host cell invasion, intracellular replication, lysis of and egress from the infected cell, and migration to new host cells. While invasion, egress, and migration are all TgMyoA-dependent processes, intracellular replication does not require TgMyoA [ 26 , 28 ]. To test whether any portion of the observed growth inhibition in culture is due to off-target effect(s) of KNX-002 on parasite replication, we scored the number of parasites per parasitophorous vacuole over time during the lytic cycle. The compound had no detectable effect on parasite replication in this assay ( S2 Fig ).
(A) tdTomato-expressing parasites were preincubated with various concentrations of KNX-002 for 5 min (0 = DMSO vehicle only) and then added to HFF cells on a 384-well plate. Fluorescence was measured daily to quantify parasite growth, in the continued presence of compound, over the next 7 days. RFU = relative fluorescence units. The data shown are the mean ± SEM from 2 independent biological replicates, each consisting of 3 technical replicates at all time points. (B) The data from day 5 (D5) of the growth assay shown in panel (A) were used to calculate the IC 50 of KNX-002 for parasite growth (16.2 μM, 95% CI = 13.0 to 20.5). See Fig 5A for a repeat of these growth assays and IC 50 calculations. (C) Subconfluent HepG2 and HFF cells were cultured in the presence of 0 (DMSO vehicle only), 40, or 80 μM KNX-002 for 72 h. Cell viability/growth was measured using Promega CellTiter-Glo (top panel) and toxicity was measured using Promega CellTox green (bottom panel), and 0.1% w/v NaN 3 served as a positive control for cytotoxicity. Results are plotted relative to the DMSO controls. Bars show the mean of 3 independent experiments ± SEM. KNX-002 shows no evidence of growth inhibition or cytotoxicity in either cell type at either 40 or 80 μM. The measurements underlying the data plotted in this figure can be found in S2 Data . HFF, human foreskin fibroblast.
The 3 lipid bilayers that comprise the pellicle of T. gondii, i.e., the plasma membrane and subjacent double bilayer of the inner membrane complex [ 34 , 35 ], represent a potential barrier to the penetration of externally applied drugs. To determine if KNX-002 can access and inhibit the motor in live parasites, we determined the compound’s effect on parasite expansion in culture in a modified plate-based growth assay (see Methods and ref. [ 36 ]). KNX-002 does indeed inhibit parasite growth, in a dose-dependent manner ( Fig 2A ), with an IC 50 of 16.2 μM (95% CI 13.0 to 20.5 μM; Fig 2B ).
(A) The effect of various concentrations of KNX-002 on TgMyoA ATPase activity, normalized to the activity with an equivalent amount of DMSO (vehicle) only. Inset shows the structure of KNX-002. Each data point represents mean ATPase activity of 3 different batches of KNX-002 ± SEM. (B) The inhibitory effect of KNX-002 on TgMyoA and various vertebrate myosins (IC 50 and 95% CI). The dose-response curves for the individual vertebrate myosins are shown in S1 Fig and the measurements underlying the data summarized in this figure can be found in S1 Data . CI, confidence interval.
To identify inhibitors of TgMyoA motor activity, we used our previously described method for producing large amounts of functional motor, in which the TgMyoA heavy chain is co-expressed in insect cells with its 2 light chains, TgMLC1 and TgELC1, and a myosin co-chaperone protein [ 33 ]. We developed a miniaturized coupled enzyme assay to measure actin-dependent ATPase activity of the purified recombinant motor and screened 50,000 compounds from the compound library at Cytokinetics Inc. for inhibitors of this ATPase activity. Hit follow-up and characterization included dose-response analysis using resupplied compound that was determined to be >95% pure by LC/MS analysis and control assays demonstrating the compound was inactive against an unrelated ATPase, hexokinase. The most potent hit was 1-(4-methoxyphenyl)-N-((3-(thiophen-2-yl)-1H-pyrazol-4-yl)methyl)cyclopropan-1-amine ( Fig 1A , inset), subsequently named KNX-002. KNX-002 inhibits TgMyoA ATPase activity in a dose-dependent manner with an IC 50 of 2.8 μM (95% confidence interval (CI) 2.4 to 3.2 μM; Fig 1A ). In contrast, KNX-002 has little to no detectable effect on the actin-activated ATPase activity of bovine cardiac myosin, chicken gizzard smooth muscle myosin (SMM), rabbit fast skeletal muscle myosin, or bovine slow skeletal muscle myosin, up to the highest concentrations tested (40 μM; Figs 1 and S1 ). These data identify KNX-002 as a potentially specific inhibitor of T. gondii class XIVa myosin activity.
Discussion
The intense interest in motility among those who study apicomplexan parasites reflects both the unique nature of the process and its importance in parasite biology and virulence. These parasites move without cilia, flagella, or a protruding leading edge that drives the substrate-dependent motility of most other eukaryotic cells. The class XIV myosins play a central role in motility, but they are unusual myosins in many respects [17,23], and precisely how these myosins and their interacting proteins function to drive motility remains controversial [27,42–47]. The mechanisms underlying motility are therefore of fundamental cell biological interest. Since MyoA is essential for virulence, highly conserved in apicomplexan parasites, and different in several respects from the myosins found in humans, the MyoA motor complex also represents a potentially attractive target for drug development. We describe here the identification and characterization of a novel small molecule, KNX-002, which will be a useful chemical probe for future studies of the function of TgMyoA. Importantly, we also identified a mutation in TgMyoA that reduces the sensitivity of the motor to KNX-002. We used isogenic parasite lines expressing either the wild-type or mutant myosin to demonstrate that TgMyoA is a biologically relevant target of the compound in parasites and to show that pharmacological inhibition of TgMyoA can slow the progression of disease in mice infected with a lethal dose of T. gondii.
KNX-002 as a chemical probe for studying class XIV motor function Most previous studies of TgMyoA motor function involved phenotypic characterization of parasites in which the genes encoding TgMyoA or its associated proteins were either disrupted or their expression down-regulated (e.g., [26,27,46,48]). Such reverse genetic approaches, particularly gene disruptions, are powerful but blunt tools for studying the mechanisms underlying processes important to parasite survival, such as motility. Pharmacological inhibitors that interact specifically with a target of interest provide a useful complementary approach for studying motility and other facets of parasite biology [49,50], since different aspects of the target’s function can be perturbed at specific times and in a controlled manner. Because the parasites are exposed to the compound for only short periods of time, compensatory adaptations (e.g., through mutation or changes in gene expression) that can confound data interpretation in experiments involving gene disruption [31,32] are unlikely to occur. Even experiments involving controlled protein depletion or transcriptional down-regulation of a gene of interest, which can be achieved in hours to days, run the risk of inducing compensatory changes in transcription, translation, protein stability, posttranslational modifications, or the localization of proteins with overlapping function (e.g., [30–32]). Determining the effect of a pharmacological inhibitor of motility can, in contrast, be accomplished within minutes (e.g., Figs 4 and 7). There have been only 2 previously reported attempts to pharmacologically manipulate the function of TgMyoA. In the first, the general myosin inhibitor 2,3-butanedione monoxime (BDM) was shown to inhibit parasite motility [38]; however, BDM is now recognized as an inhibitor with so many off-target effects that little can be concluded from its use in cells [51,52]. The other study used computational methods and a homology model of TgMLC1 bound to TgMyoA to identify 2 predicted inhibitors of TgMLC1-TgMyoA interaction [53]. These 2 compounds (C3-20 and C3-21) inhibit T. gondii growth with sub-micromolar IC 50 s [53] but whether this effect on parasite growth is due to inhibition of TgMyoA has not been established. The compound we have identified and characterized here, KNX-002, inhibits the ability of recombinant TgMyoA to propel actin filaments in the in vitro motility assay, and decreases the speed of the filaments that move (Fig 6A). Similar effects are seen in the 3D motility assay: compound treatment reduces both the number of parasites moving and, to a lesser extent, the speeds with which these parasites move (Figs 7B and 7C and S7B). To provide a better picture of the combined impact of changes to these 2 motility parameters, we multiplied the fraction of actin filaments or parasites moving by the mean actin filament/parasite speed to generate the filament/parasite motility index. In the in vitro motility assay with wild-type TgMyoA, fraction moving and speed contributed approximately equally to the filament motility index in both the presence and absence of the compound. In wild-type parasites, the compound-induced changes to fraction moving are larger than to speed and therefore the predominant contributor to the changes in parasite motility index upon compound treatment. Precisely how the effects of the compound on motor function in the in vitro motility assay relate to the observed changes in parasite motility will require further study. Nevertheless, the data presented demonstrate the value of KNX-002 as a new chemical probe for studying the mechanism(s) underlying the unusual form of eukaryotic cell motility used by T. gondii. Excitingly, KNX-002 and a recently discovered analog named KNX-115 were shown to also inhibit PfMyoA, the TgMyoA homolog of the malaria parasite P. falciparum [54,55], and to block the growth of blood stage P. falciparum in culture and the invasion of hepatocytes by liver stage parasites [54,55]. Structure-function studies revealed that the compound sequesters PfMyoA in a state of low affinity for actin [54], consistent with a decrease in duty ratio we observe in wild-type TgMyoA treated with KNX-002 (S13 Fig).
Effects of the T130A mutation The T130A mutation that confers partial resistance to KNX-002 was identified by chemical mutagenesis and selection in KNX-002. While the mutagenesis/selection approach is commonly used to generate resistant parasites in T. gondii and other apicomplexan parasites (e.g., [56–58]), the next step after isolating resistant clones is typically to do whole-genome sequencing on the most promising of the resistant clones to identify potentially relevant mutations. We took a more direct and scalable approach in this study, by sequencing only the motor protein loci (TgMyoA, TgMLC1, TgELC1, TgELC2) in the resistant parasites, since (a) the goal of this study was to identify a direct inhibitor of motor function; and (b) we had already shown that KNX-002 can inhibit the activity of recombinant motor consisting only of TgMyoA, TgMLC1, and TgELC1 (note: since TgELC1 and TgELC2 are functionally redundant [19], a mutation in the gene encoding either could confer compound resistance in parasites). Of the 5 resistant clones whose IC 50 for KNX-002 shifted 2.5-fold or more relative to wild type in the growth assay, only one had a mutation in the coding sequence of one of the motor protein genes, T130A in TgMyoA. The source of resistance in the other 4 clones is unknown. T130, which lies in the first β-strand at the extreme N-terminus of the transducer domain [24], is not predicted to be directly involved in the binding of KNX-002 [54], even though other aspects of the transducer may be involved [54]. Thus, the mechanism by which the T130A mutation confers partial resistance to the compound is likely to be indirect. For example, the T130A mutation could induce long range structural changes that alter the KNX-002 binding pocket, thereby mitigating the effect of the compound. However, myosin motors depend on a complex and highly interconnected network of allosteric interactions to perform their function [59], making it difficult to predict precisely how the T130A mutation reduces the sensitivity of TgMyoA to KNX-002. While the T130A mutation reduces the sensitivity of the motor to KNX-002, the mutation is not without its effects on motor performance, causing a 75% to 80% reduction in basal filament sliding speed in the in vitro motility assay (Fig 6B). However, when expressed in parasites the mutation had a much less pronounced effect on parasite motility (compare the first and fourth sets of gray bars in Fig 7B–7D). We explored several possible explanations for this differential effect of the T130A mutation on the activity of the recombinant motor versus motor-driven motility of the parasites. First, the mutation might alter the folding or stability—and therefore the activity—of the recombinant TgMyoA used in the in vitro assays. This seems unlikely, since the mutant motor was indistinguishable from the wild-type motor by SEC and DSF (S9 Fig). Second, the partially functional motor might be overexpressed in the mutant parasites. This possibility was ruled out by quantitative western blotting (S14A Fig). Third, the parasite might compensate for the loss of TgMyoA activity in the T130A mutants through changes in the transcription of other myosins, myosin light chains, glideosome components, actin or actin regulatory proteins. No such changes to known motility-associated proteins were observed by RNAseq analysis (S14B Fig, right panel). Analysis of all differentially regulated transcripts revealed that 57% of the genes up/down-regulated more than log2-fold were annotated as hypothetical proteins, and the remainder were largely associated with trafficking and metabolism (S14B Fig, left panel) so cannot readily explain a compensation mechanism in parasites with impaired motor function, although further studies would be required to definitively rule out this possibility. The compound might have less of an effect on motor function in the parasite due to some protein or factor present in the parasite (but not the insect cell system used to produce recombinant motor) that can counteract the deleterious effects of the mutation. Alternatively, the density of TgMyoA motors might be higher in the parasite pellicle than on the glass coverslips used in the in vitro motility assays, and therefore require higher concentrations of compound to achieve equivalent levels of inhibition. Regardless of the reason for this differential impact of the mutation on in vitro and parasite motility, it is notable that the fraction of actin filaments moving in vitro and the fraction of parasites moving within Matrigel were better correlated than the speeds of actin filament and parasite movement. The fraction of filaments moving in the in vitro motility assay may therefore be more relevant to motor-driven parasite motility than filament sliding speed. The mutagenesis protocol used to generate the T130A mutation kills 70% of the initial parasite population and results in an average of 63 mutations per parasite across the genome [40]. It is unclear how these harsh experimental mutagenesis conditions relate to what occurs during a natural infection and therefore how readily parasites showing reduced sensitivity to KNX-002 (or, more likely, an optimized compound based on the KNX-002 scaffold—see below) would arise during treatment of a human infection. From the drug development perspective, it would be useful to know what other resistance mechanisms are available to the parasite for this class of compounds. Further characterization of the 4 partially resistant mutants identified here that have no mutations in the TgMyoA motor (S8 Fig) would be a good place to start. TgMyoA knockout parasites are still capable of a small degree of 3D motility (13% of wild-type levels [46] under the same assay conditions used here). Although controversial [46], it has been proposed that TgMyoC can to some extent functionally compensate for the loss of TgMyoA [30,31], so mutations in TgMyoC would be of particular interest. Because humans are a dead-end host for Toxoplasma, even if resistance were to emerge within a treated individual through mutations in TgMyoA or some other locus in the genome, the chances of this resistance spreading human-to-human or into the zoonotic reservoir are exceedingly small.
[END]
---
[1] Url:
https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002110
Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.
via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/
