In Vitro Selection Implicates ROP1 as a Resistance Gene for an Experimental Therapeutic Benzoquinone Acyl Hydrazone in Toxoplasma gondii

Toxoplasma gondii is a globally distributed apicomplexan parasite and the causative agent of toxoplasmosis in humans. While pharmaceuticals exist to combat acute infection, they can produce serious adverse reactions, demonstrating a need for enhanced therapies.

ABSTRACT

Toxoplasma gondii is a globally distributed apicomplexan parasite and the causative agent of toxoplasmosis in humans. While pharmaceuticals exist to combat acute infection, they can produce serious adverse reactions, demonstrating a need for enhanced therapies. KG8 is a benzoquinone acyl hydrazone chemotype identified from a previous chemical screen for which we previously showed in vitro and in vivo efficacy against T. gondii. However, the genetic target and mechanism of action of KG8 remain unknown. To investigate potential targets, we generated resistant T. gondii lines by chemical mutagenesis followed by in vitro selection. Whole-genome sequencing of resistant clones revealed a P207S mutation in the gene encoding rhoptry organelle protein 1 (ROP1) in addition to two lesser resistance-conferring mutations in the genes for rhoptry organelle protein 8 (ROP8) and a putative ADP/ATP carrier protein (TGGT1_237700). Expressing ROP1^P207S in parental parasites was sufficient to confer significant (10.3-fold increased half-maximal effective concentration [EC₅₀]) KG8 resistance. After generating a library of mutants carrying hypermutated rop1 alleles followed by KG8 pressure, we sequenced the most resistant clonal isolate (>16.9-fold increased EC₅₀) and found independent recapitulation of the P207S mutation, along with three additional mutations in the same region. We also demonstrate that a rop1 knockout strain is insensitive to KG8. These data implicate ROP1 as a putative resistance gene of KG8. This work further identifies a compound that can be used in future studies to better understand ROP1 function and highlights this novel chemotype as a potential scaffold for the development of improved T. gondii therapeutics.

INTRODUCTION

Toxoplasma gondii is an apicomplexan parasite capable of infecting nearly all mammals (1, 2). It is highly prevalent in humans and has infected approximately one-third of the global human population (1, 3, 4). Although often asymptomatic in otherwise-healthy patients, recrudescence of the chronic stage of this infection can manifest as severe disease leading to death in immunocompromised individuals, often through fatal toxoplasmic encephalitis (5, 6). Additionally, severe congenital defects can arise in prenatally infected infants (7–9). Limited treatment options exist for the acute stage of infection, with a combination treatment of pyrimethamine and sulfadiazine being the most widely implemented (10–12). However, much of the human population is allergic to sulfadiazine (13), and high doses of pyrimethamine can cause hematopoietic deficiencies, even when coadministered with folinic acid supplements (14). Furthermore, there are no clinically available therapeutics that are able to satisfactorily clear the chronic stage of this infection, leaving affected individuals vulnerable to reactivation of T. gondii infection if they later become immunocompromised (5, 13, 15). These unmet needs highlight the necessity of developing novel therapeutics to treat both acute and chronic infection (5, 11).

To this end, our laboratory has investigated antiparasitic compounds identified in a previously conducted high-throughput screen (16) that had shown initial in vitro efficacy against T. gondii. In a recent study, we assessed the specific activity of three compounds from the screen in T. gondii both in vitro and in vivo (17). The benzoquinone acyl hydrazone KG8 (synonym SJ000296485) demonstrated a half-maximal effective concentration (EC₅₀) similar to that of pyrimethamine in vitro with relatively low host cell toxicity of >28 μM (17).

Here, we present our efforts to identify a parasite molecular resistance gene of KG8, accomplished by a forward genetic screen from which we isolated KG8-resistant parasites. Specifically, we show that growth of chemically mutagenized T. gondii tachyzoites under KG8 pressure selects for ROP1 mutant parasites, that introduction of a rop1 mutant allele in the wild-type parent background confers KG8 resistance, and that a rop1 knockout strain is rendered insensitive to KG8. ROP1’s nonessentiality for Toxoplasma viability is interesting (18), as it perhaps suggests a mechanism of action differing from simple enzyme inhibition; however, ROP1 as a target is highly apicomplexan specific and presents a potential avenue for improved therapeutic regimens.

RESULTS

In vitro selection generated KG8-resistant T. gondii strains.

We previously characterized the activity of the benzoquinone acyl hydrazone KG8 (Fig. 1A) against the type I RH-dTom and type II Pru-dTom strains of T. gondii and its selectivity over five different host cell lines (17). Given its in vitro potency and the novelty of its chemical scaffold, we set out to investigate a primary resistance gene for KG8. We chemically mutagenized parental RH-dTom parasites by ethyl methanesulfonate (EMS) treatment and obtained resistant mutants by in vitro selection with KG8. We isolated KG8-resistant (KG8^r) mutants through clonal isolation, identified the three most resistant nonsyngeneic clones, and denoted them KG8^r-1, -2, and -3. The top KG8^r clones had KG8 EC₅₀ values ranging from 6.4 to 8.7 μM, reflecting 3.7- to 5.1-fold increases in resistance to KG8 relative to parental RH-dTom (Fig. 1B and Table 1). No differences in fluorescence between the unmutagenized control strain and the KG8^r clonal isolates were found. The following analyses were completed using these KG8^r clonal populations, but five additional, less resistant, clones were also isolated, although they were not used for further analysis (see Table S1 in the supplemental material).

FIG 1.

Emergence of KG8 resistance in T. gondii by chemical mutagenesis and in vitro selection. (A) Structure of the benzoquinone acyl hydrazone KG8 used in in vitro resistance selection experiments. Previous work demonstrated high efficacy against T. gondii strain RH parasites and minimal cytotoxicity against human foreskin fibroblast (HFF) host cells (17). (B) RH-strain tachyzoites were mutagenized with ethyl methanesulfonate (EMS) and subjected to gradually increasing doses of KG8. Surviving mutagenized parasites were clonally isolated, and the EC₅₀ of each clone and the parental control was determined by relative fluorescence.

TABLE 1.

KG8 EC₅₀ analyses conducted in this study^a

Parasite strain	KG8 EC50 (μM)	EC50 fold change
Parental RH	1.7
Clonal KG8r lines
KG8r-1	7.9	4.6
KG8r-2	6.4	3.7
KG8r-3	8.7	5.1
Bulk resistant populations
ROP8D84V	4.8	2.8**
TGGT1_237700D17N	4.6	2.7**
ROP1WT	2.0	1.2
ROP1P207S	17.7	10.3**
Clonal hypermutated ROP1 line
ROP1mut4	>29.0	>16.9
Clonal Δrop1 line and isogenic parent (18, 20)
RHΔrop1 (4R2)	>29	>14.8
RHΔrop1 (4R2)-rop1C619T	14.0	7.2**

^aEC₅₀ analyses were completed for each mutant T. gondii strain as described in the text and compared to their respective parental RH strain. Fold change is presented as (mutant EC₅₀)/(parental RH EC₅₀). Significance was determined by t test. **, P < 0.01.

A mutant rop1 allele confers KG8 resistance.

To discover the genetic determinants of KG8 resistance, we isolated genomic DNA from the KG8^r mutants, subjected it to whole-genome sequencing on an Illumina platform, and subsequently aligned the sequenced reads to the type I T. gondii reference strain GT1 genome. We then used this alignment to identify single-nucleotide variants (SNVs) in the EMS-mutagenized clones, prioritizing nonsynonymous mutations present in more than one isolated clone and those occurring in protein-coding genes or promoters for further analysis (Fig. 2A). We further prioritized genes with codon-altering SNVs based on similarity to mammalian proteins, reducing in priority those displaying substantial homology to host proteins due to the observed specificity of KG8 for T. gondii over host cells (17).

FIG 2.

Expression of ROP1^P207S confers KG8 resistance to parental T. gondii parasites. (A) Venn diagram summarizing the SNVs identified in each clonally isolated KG8^r mutant through whole-genome sequencing and bioinformatic analysis. (B) Codon-altering SNVs were prioritized, and three mutated genes were selected for further study. Amino acid positions of each putative KG8^r mutation and the clonal isolates each mutation was identified in are indicated. (C) Candidate KG8 resistance-conferring alleles were transfected into wild-type RH-dTom tachyzoites and were evaluated for their EC₅₀ relative to the parental RH-dTom. EC₅₀ analyses were completed in technical triplicate for each strain, and error bars represent standard deviations.

Following these filtering steps, an initial set of mutated T. gondii genes were identified as high-priority candidate KG8^r genes. Of the 13 nonsynonymous mutations that were present in more than one KG8^r mutant (Fig. 2A), we selected three for further study: rhoptry protein 1 (ROP1^P207S; TGGT1_309590), rhoptry protein 8 (ROP8^D84V; TGGT1_215775), and a predicted mitochondrial ADP/ATP carrier protein (TGGT1_237700^D17N) (Fig. 2B).

To verify whether these mutations conferred KG8 resistance, we transfected constructs for the mutant alleles of these three genes into parental RH-dTom (Fig. 2C) and confirmed adequate rop1 transcript and ROP1 protein expression in transfected parasites by quantitative reverse transcription-PCR (qRT-PCR) and immunoblotting (Fig. S1). We subsequently selected transfected parasites with pyrimethamine, isolated clones, and determined their sensitivity to KG8 by fluorescent growth assays. Parasites episomally expressing ROP8^D84V or TGGT1_237700^D17N showed a shift in KG8 sensitivity (Fig. 2C); however, parasites expressing ROP1^P207S demonstrated a more substantial and significant shift in KG8 sensitivity (EC₅₀ of 17.7 μM, 10.3-fold increase relative to the parental), while episomal expression of ROP1^WT did not (Fig. 2C and Table 1). Of the alleles tested, the transfection of a mutant rop1 allele harboring a single-nucleotide substitution was sufficient to confer the greatest degree of KG8 resistance to RH-dTom T. gondii parasites; therefore, rop1 was selected for further analysis.

Hypermutation of rop1 leads to highly resistant clones.

It is known that while a single-amino-acid change in DHFR-TS (e.g., M3: T83N) mediates some resistance to pyrimethamine, the addition of a second mutation (e.g., M2M3: S36R + T83N) confers profound resistance to pyrimethamine, permitting T. gondii mutants to proliferate in the presence of high concentrations of compound at the same efficiency as wild-type parasites without pyrimethamine (19). Accordingly, we reasoned that while transfecting ROP1 with a single-amino-acid substitution (P207S) confers greater than 10-fold resistance to the compound, multiple mutations in the rop1 gene may enhance resistance and allow for the isolation of highly resistant clonal populations. To examine this, we generated a library of plasmids expressing hypermutated variants of the parental rop1 gene using error-prone PCR and transfected this library into parental RH-dTom tachyzoites (Fig. 3A). We exposed the transfected parasites to gradually increasing concentrations of KG8 up to 29 μM (16.9-fold higher than the parental EC₅₀) over 2 weeks. From a small population of parasites that survived the full drug exposure regimen, we then isolated individual resistant clones.

FIG 3.

Expression of hypermutated rop1 confers enhanced KG8 resistance to parental T. gondii parasites. (A) Hypermutated rop1 library preparation strategy. Wild-type rop1 was subjected to error-prone PCR, and the library of mutated PCR products was cloned via restriction enzymes into the pDHFR-TSc pyrimethamine selection plasmid. Parental RH-dTom parasites were transfected with the resulting plasmid library and selected with 2 μM pyrimethamine and increasing concentrations of KG8 up to 29 μM over 7 passages. Surviving parasites were clonally isolated by limiting dilution. (B) Plasmid curing strategy. To promote the selection of the highest resistance-conferring plasmid, clonal isolates were cultured with 12 μM KG8 over an extended period of 2 weeks. The plasmid carrying the mutant rop1 allele was rescued from resistant clones and sequenced. (C) Amino acid positions of the four nonsynonymous mutations identified in the ROP1mut4 strain (N200T, V201E, P205L, and P207S).

Direct sequencing of the rop1 cassette in the most resistant isolated clone resulted in mixed sequence signals, potentially resulting from the uptake of multiple plasmids following transfection. Spontaneous plasmid cure can be achieved in Toxoplasma by propagating parasites in the absence of selection pressure, as the parasites eliminate episomes for which maintenance is not required to survive. Similarly, we reasoned that resistant parasites harboring multiple plasmids would dispense with those that were not responsible for conferring resistance through extended KG8 selection pressure. Therefore, to ascertain the specific hypermutated rop1 allele driving high-level KG8 resistance, we maintained the mutant under extended drug selection with 2 μM pyrimethamine for 1 week and 12 μM KG8 for an additional week (Fig. 3B). Following curing, we obtained the remaining rop1 hypermutated allele by plasmid rescue. Sequencing revealed four nonsynonymous amino acid substitutions spanning positions 200 to 207 (Fig. 3C and Table S2, N200T, V201E, P205L, and P207S). As concentrations of KG8 above 29 μM demonstrated host cell toxicity, the KG8 EC₅₀ of this resistant clone, designated ROP1^mut4, could not be determined beyond >29 μM (Table 1). Interestingly, the independently generated P207S mutation identified by hypermutagenesis of the rop1 gene was identical to that identified through random chemical mutagenesis and in vitro selection (Fig. 2). These data indicate that the ROP1 residues in the 200–207 region are important for KG8 activity and further implicate ROP1 as a putative intracellular resistance gene for KG8.

Δrop1 mutants are insensitive to KG8.

Given that ROP1 is dispensable for both in vitro replication and in vivo virulence (18, 20), proposing a mechanism by which it is the definitive target of KG8 is challenging. To interrogate different possibilities and to rule out the more trivial ones, we used a previously constructed rop1 deletion strain, 4R2 (18, 20). We first confirmed the rop1 deletion by immunoblotting and PCR amplification with Sanger sequencing of the Δrop1 locus (Fig. S1B and S2). Next, we assessed the rop1 knockout strain’s sensitivity to KG8 compared to that of its isogenic parent by vacuole counting assay. We found that the RHΔrop1 strain was significantly less sensitive to KG8 (EC₅₀, >29 μM; q < 0.05 at and above 10 μM) than the isogenic parent strain, which showed sensitivity to KG8 similar to that of RH-dTom (Fig. 4 and Table 1). We also found that expression of ROP1^P207S (represented in the figure by its nucleotide base change, rop1^C619T) in the Δrop1 mutant restored KG8 sensitivity to the degree observed when expressing ROP1^P207S in the RH-dTom background (EC₅₀ of 14.0 μM) (Fig. 4B and Table 1). These data critically associate the presence of ROP1 with the anti-Toxoplasma properties we have observed for KG8.

FIG 4.

RHΔrop1 strain is resistant to KG8. The proliferative index of these non-fluorescence-expressing parasites in HFF monolayers was determined for the isogenic parent and the RHΔrop1 strain (A) or RHΔrop1 complemented with mutant ROP1^P207S protein (B) in the presence of increasing concentrations of KG8. After 5 days of KG8 treatment, the proliferation index (PI) was calculated for both strains at each KG8 concentration tested: PI = (percent infected host cells) × (average number of parasites/cell). The percentage of infected host cells was quantified over 20 fields of view, and tachyzoites in 40 randomly selected vacuoles were manually counted using phase-contrast microscopy at ×400 magnification in biological duplicate (each represented by data points). Significance was determined by multiple t tests with false discovery rate (FDR) correction using the two-stage step-up method of Benjamini et al. (49). *, q < 0.05.

DISCUSSION

Rhoptries are dedicated secretory organelles found in apicomplexan parasites such as T. gondii and secrete small rhoptry (ROP) proteins through the apical complex of T. gondii tachyzoites (21, 22). Rhoptries contain some of the machinery used by T. gondii to invade host cells and subvert normal host cell functions (21), although many ROP proteins, including ROP1, do not have a known biological function. Intriguingly, ROP1 is nonessential for both viability and virulence in the RH strain of T. gondii (18, 20).

Further, ROP1 does not possess conserved domains within its primary protein structure. However, previous studies have provided evidence that ROP1 is secreted rapidly and early during invasion directly into the host cytoplasm in a manner analogous to the delivery of effectors into host cells by prokaryotic type III secretion systems (23, 24). This secretion then leads to the formation of rhoptry-derived secretory vesicles, termed evacuoles, which fuse with parasitophorous vacuoles (23). ROP1 then briefly accumulates within the parasitophorous vacuole lumen and subsequently disappears (23, 25). These facts, as well as its divergence from other ROP proteins, pose a challenge in learning the mechanism of action of KG8 (26).

In this work, we show that the benzoquinone acyl hydrazone KG8 requires wild-type ROP1 in T. gondii for activity. Through chemical mutagenesis and in vitro selection, we were able to isolate KG8-resistant clones. Sequencing these clones revealed single-amino-acid substitutions in ROP1 and other proteins; however, expressing the mutant rop1 P207S allele conferred high-level KG8 resistance to parental strain parasites. Further, the deletion of rop1 resulted in >14.8-fold resistance to KG8.

Through independent hypermutation of the parental rop1 allele, we were able to identify four nonsynonymous rop1 mutations that together conferred a >16.9-fold increase in EC₅₀ compared to that of the wild type when expressed in the parental background. This finding was particularly interesting given that the P207S mutation identified in our mutagenized and selected clones independently arose in the hypermutated rop1 selection and that all four mutations occurred within a seven-amino-acid window, providing evidence that this specific region in ROP1 is important in mediating the activity of KG8. The four mutations we identified occur within a region of several tandem repeats first noted by Ossorio and colleagues (27). These tandem repeats have the consensus sequence (QELPPPXX)₈ and are embedded within a highly acidic and hydrophobic domain of ROP1, forming a border between this domain and the following basic domain (27). The four mutations we identified in ROP1^mut4 resulted in a change from NVQELPPP into TEQELLPS (Fig. 3C). Although we can only speculate on the importance of these mutations at this time, further investigation into this proline-rich domain could yield insight into how KG8 acts through ROP1 and, perhaps, into ROP1’s intracellular function as well. Beyond drug development, KG8 and its interaction with ROP1 and its mutants may be exploited in future studies of ROP1’s function.

It is worth noting that ROP1 may not be the only parasite KG8 resistance gene, as we detected resistance conferral by expression of the ROP8^D84V and TGGT1_237700^D17N mutant proteins, albeit at reduced levels of resistance. With this in mind, we propose two models that might explain how KG8’s interaction with this nonessential protein can result in the observed parasiticidal effect. KG8 could be a prodrug and metabolized into its active form by a complex formed by ROP1 and an additional, unidentified factor. In this model, both ROP1 and this unidentified factor would be necessary for the conversion of the prodrug into its active form and would physically interact. If so, KG8 would be metabolized by the active proteins and exerts its antiparasitic effect by inhibiting the bona fide intracellular target. In parasites overexpressing the mutant ROP1^P207S or ROP1^mut4 protein, the excess mutant protein would outcompete the wild-type protein for binding to the unidentified factor, yielding less productive complexes that metabolize KG8 at a reduced rate and subsequently reducing the intracellular concentration of KG8’s active form. This would then result in the observed KG8 resistance. In Δrop1 parasites, the lack of ROP1 likely prevents KG8 from being metabolized into its active form at all, potentially resulting in the observed KG8 insensitivity. This is the mechanism of action for two of the first-line drugs used to treat drug-sensitive Mycobacterium tuberculosis infection. Both isoniazid and pyrazinamide are prodrugs that are metabolized by KatG and PncA, respectively (28, 29), into their active forms. Notably, both katG and pncA are dispensable for viability in M. tuberculosis (30). One drawback to this explanation is that it requires an unidentified partner protein. Future work could include conducting a ROP1 pulldown to identify potential interactors. Additionally, mass spectrometry could be done on lysates from KG8-treated wild-type T. gondii to identify KG8 metabolites, if any exist.

Alternatively, KG8 could bind to ROP1 and cause a toxic conformational change in the protein. In wild-type parasites, KG8 binds ROP1 and causes the toxic altered conformation that builds up in the cell, resulting in cell death. In parasites overexpressing the mutant ROP1^P207S or ROP1^mut4 protein, the mutated ROP1 proteins would not form the toxic conformation when bound to KG8. The excess mutant protein would outcompete for binding to KG8 and reduce the concentration of ROP1 in a toxic conformation within the cell, resulting in the observed KG8 resistance. In Δrop1 parasites, the lack of ROP1 prevents KG8 from binding and producing the toxic ROP1 conformation, resulting in the observed KG8 insensitivity. Although simpler than the first model in that it does not require an unidentified binding partner, this explanation relies on the assumption that the mutant ROP1 does not form a toxic conformation when bound to KG8. Future work could include in vitro thermostability assays that compare the shift in melting temperature of wild-type and mutant ROP1 in the presence of KG8 (31). If the ROP1 mutants are still able to bind KG8 and show thermostabilities different from that of the wild-type protein, it would support this hypothesis.

Despite its poor selectivity over some host cell lines and its unfavorable pharmacokinetic characteristics, KG8 and its unique structure presents a scaffold to develop derivatives based on structure-activity relationships. Accordingly, in a recent follow-up study, we assessed the in vitro and in vivo efficacy of a small library of KG8 analogs, along with their ADME and physiochemical properties (32). Two of these derivatives, KGW44 and KGW59, showed higher in vitro potency, much lower cytotoxicity in host cells, and improved metabolic stability. Furthermore, in the murine in vivo infection model, treatment with KGW59 at 5 mg/kg of body weight following intraperitoneal infection with 5 × 10³ type II ME49 tachyzoites resulted in a significant increase in survival compared to the vehicle control (n = 10) (32).

Conclusions.

Through a forward genetic screen employing the use of random chemical mutagenesis and subsequent whole-genome sequencing, we could identify a series of high-priority SNVs potentially conferring resistance to the benzoquinone acyl hydrazone KG8. We further evaluated individual SNVs and found that a P207S mutation in the rop1 gene yielded a significant and substantial increase in resistance to KG8 compared to the parental controls. Further, we were able to generate an increase in this observed resistance through error-prone PCR hypermutagenesis of the rop1 gene, through which we identified several mutations within the same region of this gene. One SNV from this set represented an independent recapitulation of the P207S mutation obtained through random mutagenesis. Taken together, these data suggest that further development of benzoquinone acyl hydrazone as a novel antiparasitic scaffold is promising, that continued medicinal chemistry-based derivatization can yield enhanced molecules with even greater in vivo efficacy, and that KG8 is valuable as a chemical probe in further studies of ROP1.

MATERIALS AND METHODS

Cell lines and maintenance.

Type I RH strain T. gondii tachyzoites, which constitutively express an integrated fluorescent dimerized Tomato protein, RH-dTom, were used in the chemical mutagenesis and in vitro selection experiments. The RHΔrop1 strain 4R2 (18, 20) and its isogenic parent (kind gifts of Dominique Soldati-Favre, University of Geneva) were used in plaque assays. Parasites were maintained by serial passage through human foreskin fibroblast (HFF) monolayers (ATCC, Manassas, VA). All cells and parasites were cultured at 37°C with 5% CO₂ in D10 medium: Dulbecco’s modified Eagle’s medium (DMEM) (Lonza, Walkersville, MD) supplemented with 20% Medium-199 (Corning, Manassas, VA), 10% heat-inactivated bovine calf serum (GE Healthcare Life Sciences, Logan, UT), 2 mM l-alanyl-l-glutamine (Corning, Manassas, VA), 100 μg/ml penicillin-streptomycin (Corning, Manassas, VA), and 20 μg/ml gentamicin sulfate (Corning, Manassas, VA). Compounds used in this study all were solubilized in dimethyl sulfoxide (DMSO) (Fisher, Hampton, VA).

Random chemical mutagenesis and in vitro selection.

Intracellular RH-dTom tachyzoites grown in triplicate T-175-cm² flasks were mutagenized with 10 mM ethyl methanesulfonate (EMS) (Sigma-Aldrich, St. Louis, MO) in D10 medium for 4 h at 37°C as previously described (33, 34). Mutagenesis with 10 mM EMS resulted in ∼40% survival after treatment compared to the DMSO controls. Following EMS treatment, medium was removed and the infected HFF monolayers were washed three times with 1× phosphate-buffered saline (PBS). Mutagenized parasites were released by scraping of host cell monolayer and Dounce homogenization of the cell suspension, followed by filtration through a 5-μm filter to remove host cell debris. Parasites were permitted to recover in fresh HFF monolayers for 24 h, and then KG8 (synonym SJ000296485 [ChemDiv, San Diego, CA]) was added. Mutagenized and DMSO-treated parasites were passaged in the presence of 1.15 μM KG8 (0.5× parental EC₅₀) for one passage, 2.30 μM KG8 for four passages (1.0× parental EC₅₀), and 3.45 μM KG8 (1.5× parental EC₅₀) for two passages, after which no surviving parasites were observed in the DMSO control flasks. Individual resistant clones were isolated by limiting dilution in 96-well plates of HFF and transferred to T-25-cm² flasks. Following isolation, individual clones were maintained in the absence of KG8.

Growth assays.

Parasite growth in the presence of drug selection was determined by measuring fluorescence in a BioTek Synergy HT reader as previously described (17, 35). Briefly, HFF monolayers in a 96-well plate were infected with 2,000 tachyzoites of each parasite strain and allowed to infect for 12 h, and then KG8 was added in various concentrations and fluorescence was read 5 days postinfection. Half-maximal effective concentration (EC₅₀) was then calculated from the fluorescence readings. The EC₅₀ for each parasite line was calculated using the equation of the line between the two data points nearest to 50% inhibition.

The RHΔrop1 (4R2), ROP1^P207S complement, and isogenic parent RH strains lack constitutive fluorescent markers, so growth assays were performed through the use of vacuolar counts as described previously (36, 37). HFF monolayers in a 24-well plate were infected with 2,000 tachyzoites of each strain and allowed to infect for 12 h, and then KG8 was added in various concentrations as before. After 5 days, the proliferation index (PI) was calculated for parasites treated at each KG8 concentration as discussed in reference 37: PI = (percent infected host cells) × (average number of parasites/cell). The percentage of infected host cells was quantified over 20 fields of view, and tachyzoites in 40 vacuoles were manually counted using phase-contrast microscopy at ×400 magnification in biological duplicate. The EC₅₀ value was calculated from the resulting PI curves for the parental strain as described above (EC₅₀ was not reached for the RHΔrop1 strain).

Whole-genome sequencing.

Mutagenized parasites were grown in HFF cells until they lysed the host monolayer and were released by scraping and Dounce homogenization, followed by filtration through 5-μm filters (three times) and 3-μm filters (two times) to remove host cell debris. Genomic DNA was isolated with the Wizard Genomic DNA purification kit (Promega, Madison, WI) using the manufacturer’s tissue culture protocol. Host DNA contamination was determined to be <25% by qRT-PCR and by read counts (data not shown). Sequencing libraries were prepared from genomic DNA by the University of Nebraska Medical Center High-Throughput DNA Sequencing and Genotyping Core Facility, and libraries were sequenced on an Illumina HiSeq2500 high-throughput sequencer, yielding 101-bp paired-end reads and 70 to 100× coverage for each genome.

Paired-end reads were trimmed of low-confidence bases using PrinSeqLite (38). The trimmed reads were then aligned with Bowtie2 v2.2.9 (39) using the type I parasite GT1 reference genome. From this alignment, mapped reads were sorted and indexed through the use of SAMTools v1.2 (40), and then FreeBayes v1.1.0 (41) was used to identify single-nucleotide variants. Next, SNVs found in the parental RH-dTom strain were manually removed, and the remaining SNVs were annotated with SNPEff v4.1 (42). For each program listed, any parameters differing from the default settings are described in the supplemental material. We developed an analysis pipeline to identify SNVs using these existing bioinformatics tools, which can be accessed on GitHub (https://github.com/matthew-martens/SNVVerificationPipeline). SNVs occurring in nonjunctional intronic regions, untranslated regions, or hypothetical proteins lacking sufficient annotation were deprioritized. Filtered results were then compiled, and the remaining candidate KG8 resistance-conferring SNVs that were present in multiple KG8^r clones were validated by conventional Sanger sequencing of PCR products amplified with described primers (Table S3).

Expression of candidate KG8 resistance-conferring mutations.

The candidate alleles, rop1, rop8, and TGGT1_237700 (putative ADP/ATP carrier protein), were PCR amplified from genomic DNA extracted from resistant mutant and parental RH-dTom parasites using Herculase II fusion DNA polymerase (Agilent Technologies, Santa Clara, CA) with described primers (Table S3). PCR products were cloned into the pDHFR-TSc3 plasmid using restriction enzyme digestion and ligation. The pDHFR-TSc3 plasmid carries the mutant M2M3 DHFR-TS allele, permitting pyrimethamine selection of transfected parasites (43). Inserts were sequence verified by conventional Sanger sequencing.

Parental RH-dTom or RHΔrop1 (4R2) tachyzoites were transfected with 50 μg of plasmid DNA in cytomix buffer as described previously (44–46). For each transfection, 1 × 10⁷ RH-dTom parasites were resuspended in cytomix buffer and mixed with plasmid DNA. Parasites were electroporated at 1.5 kV, 50 Ω, and 50 μF with a Gene Pulser Xcell (Bio-Rad, Hercules, CA) in a 2-mm-gap electroporation cuvette. The tachyzoites were allowed to infect HFF for 12 h, after which the culture medium was changed and 1 μM pyrimethamine was added to select for transfected parasites. Following pyrimethamine selection for 7 days and clonal isolation, KG8 dose-response growth assays for the transfected parasites were completed as described above. In Fig. 2C, two technical replicates from the parental RH-dTom dose-response curve at the 50 μM KG8 concentration were treated as outliers and excluded.

PCR hypermutagenesis of rop1.

Hypermutagenesis of the wild-type rop1 allele was completed through error-prone PCR using a method previously described (47) to generate a library of hypermutagenized rop1 PCR products. The library was confirmed to be a mixed population by Sanger sequencing, cloned into the pDHFR-TSc3 plasmid (43), and transformed into 10β competent cells (New England Biolabs, Ipswich, MA). These cells were inoculated directly for a maxiprep, and the resultant heterogeneous plasmid product was transfected into parental RH-dTom parasites as described above. After a 12-h recovery, transfected parasites were selected with 1 μM pyrimethamine for 24 h, followed by successive passages with increasing concentrations of KG8 over 7 passages up to 29 μM (12.61× parental EC₅₀). Parasites were maintained in 29 μM KG8 for 2 weeks, and individual resistant clones were isolated by limiting dilution as described above. To cure the parasites of plasmids dispensable for KG8 resistance, resistant clones were grown in the presence of 2 μM pyrimethamine for 1 week, followed by selection with 12 μM KG8 for an additional week. Following curing, resistant clones were again isolated by limiting dilution.

Plasmid rescue and sequencing of hypermutated rop1.

The plasmid carrying the hypermutated rop1 allele was rescued from integration in the ROP1^mut4 parasite genome using a method described previously (45, 48). Briefly, genomic DNA was isolated from the KG8-resistant ROP1^mut4 clonal isolate and digested with HindIII and EcoRI, and cut fragments were purified by column cleanup. These fragments were ligated into the pDHFR-TSc3 plasmid (43), and the ROP1^mut4 hypermutated rop1 allele was sequenced by conventional Sanger sequencing.

Statistical analyses.

All statistical tests were completed using GraphPad Prism (ver. 8.4.2) unless otherwise noted, with statistical significance defined as a P value of <0.05 or q value of <0.05 following multiple testing correction.

Data availability.

The raw sequencing reads for the mutagenized KG8^r clones described in this study have been deposited in the NCBI SRA under accession numbers SRR11678606, SRR11678607, and SRR11678608.