The following T. gondii strains were used in this work: RH strain, RHΔKU80, and RH containing Lac Z (clone 2F1) expressing β-galactosidase, which were kindly provided by Prof. Vern Carruthers (University of Michigan, USA) and P. falciparum 3D7 strain procured from the Institut de Recherche et de Développement (IRD-Bénin). The following reagents were also used: DAPI, (D9542, Sigma-Aldrich), crystal violet (C0775, Sigma Aldrich), Kit RAL 555 (RAL Diagnostics, Labelians, Belgium), 5’-fluorodeoxyuridine (5’FUDR, Merck), pyrimethamine (Merck), mycophenolic acid (MPA, Calbiochem), Xanthine (XAN, Merck), chlorophenol red-β-D-galactopyranoside (CPRG, Merck), Hoechst-Thiazole Orange (Sigma Aldrich) and Tocriscreen Total library (Cat. No. 2884, Tocris Biotechne).

The tachyzoites of T. gondii strains used in the present study were routinely cultured on a human fibroblast foreskin (HFF) monolayer in Dulbecco’s Modified Eagles Medium (DMEM, PAN Biotech, Dutscher, France) supplemented with heat-inactivated 10% fetal bovine serum (PAN Biotech, Dutscher, France), 2 mM of glutamine and 50 µg/mL of penicillin/streptomycin (PAN Biotech, Dutscher, France). After three hours, the infected cells were washed with culture medium and the intracellular parasites were allowed to grow at 37°C for 2−3 days to ensure complete lysis. Freshly lysed tachyzoites were purified using 0.33 µm filter (Millipore) to remove cell debris before counting, and the obtained parasites were used for transfection or for drug assays.

Transgenic ROP1-mCherry parasites were obtained via a knock-in strategy using a 2283 bp DNA fragment as well as KI-ROP1 forward primer (TAC TTC CAA TCC AAT TTA ATG CTG GGC TCG CAC CAA TAG CAC) and KI-ROP1 reverse primer (TCC TCC ACT TCC AAT TTT AGC TTG CGA TCC ATC ATC CTG CTC). This DNA fragment was cloned into the pLIC-mCherry-HXGPRT (hypoxanthine-xanthine-guanine phosphoribosyltransferase selectable marker) plasmid (Huynh and Carruthers, 2009), and was linearized using BstBI restriction enzyme. Tachyzoites (5×10⁶ parasites) of the RHΔKu80 strain were transfected with 25 µg of linearized plasmids. After two selections with 25 µg/mL MPA and 50 µg/mL XAN, stable parasites were cloned and positive clones were screened by indirect immunofluorescence assay (IFA), whereby three individual clones were selected for further confocal imaging. One of the positive clones was then knocked out for the uracil phosphoribosyl transferase (UPRT) gene using a second plasmid containing TgSORT-GFP as previously described Sloves et al. (2015), whereby the plasmid containing the HA tag was replaced by GFP. All aforementioned plasmids were checked for accuracy by full DNA sequencing before use. Transgenic tachyzoites (5x10⁶) expressing rhoptry ROP1-mCherry were transfected with 25 µg of linearized promSORT-TgSORT-GFP plasmid and were subsequently transferred to the monolayer HFF for 4h, and were washed with culture medium. This was followed by a 24-hour growth, after which 5 µM of 5’FUDR was added. After three days, selection was repeated twice before cloning the emerging resistant parasites. After screening by IFA, one positive clone was selected for drug screening using the Tocris library as described below.

T. gondii RH-ΔKu80 expressing ROP1-mCherry/TgSORT-GFP (clone D6) tachyzoites were maintained by growth on HFF cell monolayers as previously described, whereby HFF cells at 12 sub-cultured dilution cycles kept in liquid nitrogen prior to their use, and were cultivated in complete DMEM medium after thawing in a 75-cm² flask. Three days later, the cells were trypsinized, resuspended to complete the DMEM medium, and transferred to a 175-cm² flask until confluency had been attained. Prior to infection, the culture medium was removed and HFF cells were washed twice with sterile phosphate buffered saline (PBS) after which 3 mL of trypsin-EDTA was added and the sample was incubated for 4 minutes at 37°C. After verifying cell detachment by microscope, 25 mL of growth medium was added and HFF cells were then counted using a TALI Cytometer (Invitrogen) and were infected at MOI (multiplicity of infection) 2.5 with freshly lysed sourced from a 25-cm² flask containing infected HFF cells and filtered through 3 µM membrane (polycarbonate, Whatman). After parasite counting using Malassez hematocytometer, the infected cells were grown for 4 hours at 37°C, and the HFF cell monolayer was washed with PBS, trypsinized, and resuspended with complete culture medium before counting. The contents were diluted to obtain a final concentration of 3.25×10⁵ cells per mL and these infected cells were immediately used for drug library screening.

Tested compounds were obtained from the Tocriscreen Total library (Cat. No. 2884, Tocris), which contains 1,120 biologically active compounds solubilized in pure DMSO at 10 mM. Compound solutions were stored in specific plates to allow Acoustic Droplet Emission transfers ( Supplementary Figure 1 ), which were performed using an Echo550^® (Labcyte), allowing us to dispense 40 nL of each compound to reach a 10 µM concentration in the 40 µL volume used for the test, with DMSO serving as negative control (1% final concentration in the assay). Initially, assays were performed using pyrimethamine (Pyr) as a positive control (at 0.6 µM and 10 µM concentrations), as this drug is known to efficiently and rapidly kill T. gondii at very low concentrations. After the initial identification, SB 203580 (Cpd1) and SB 208 (Cpd3) were used as positive controls. For compound screening, 40 µL (corresponding to 1.3×10⁵ cells) of infected-cell suspension was dispensed in each well of 384-well plates containing compounds, which were prepared and kept for a few minutes at room temperature before use. These plates were incubated at 37°C in 5% CO₂ and 95% humidity atmosphere for 24 hours before further processing.

After 24-hour of incubation, infected cells were subjected to drug effects, and the cell culture media and drugs were removed, after which 20 µL of PBS was added to each well using a washer-dispenser device (El406^® from Biotek), whereby 20 µL of 10% formalin solution was placed on top of the PBS. Plates were kept at room temperature for 20 minutes and were washed once with PBS before staining with DAPI prepared at a 2 µg/mL concentration in PBS containing 0.1% of Triton X-100, 100 mM of glycine and 5% fetal bovine serum (FBS). After incubating at room temperature for 20 minutes, plates were washed with 50 µL of PBS, and 40 µL of PBS containing 1% FBS was added to each well, after which the plate was sealed and stored at 4°C until required for image acquisition, performed using an InCell Analyzer 6000^® (GE Healthcare). For this purpose, six fields per well were acquired from the A1 well and were placed in the P24 well in a horizontal serpentine mode of acquisition with the 60× objective and the exposure parameters were indicated in Table 1 .

Images from all three channels (hereafter denoted as blue, green and red, respectively) were captured in confocal mode, using the closed aperture option in order to reduce background noise.

Images captured by the InCell Analyzer 6000^® were analyzed using Columbus software. Both cell- and parasite-based analyses were performed, and were supplemented by vacuole structure and organization assessments in certain cases. First, a local maxima detection algorithm was applied to the DAPI channel to detect the nuclei, which were then thresholded by size, roundness and DAPI mean intensity to avoid any false detection (as parasite vacuoles could be misidentified as nuclei). Next, cytosols were detected based on the DAPI background signal in this region, after which a spot detection on the DAPI channel was processed to detect parasites in the entire cell layer. Based on this parameter, parasite count per field or per HFF cell monolayer was obtained. In parallel, as parasites were gathered in vacuoles, the vacuole population was split into positive and negative groups. Given that positive vacuoles were considered to express the expected mislocalisation phenotype, they were selected using a multiparameter threshold, including the ratio of ROP1-mCherry intensity between the center and the border of the vacuole and the mean mCherry intensity in the vacuole. To be labelled as positive, vacuoles had to fulfill the following two criteria: (1) the ratio of intensity between the center and the border of the vacuole must exceed 1.4, and (2) the mean intensity in the vacuole must be below 500, as explained in detail in Supplementary Table 2 . The percentage of positive vacuoles was thus used as a parameter that reflects the vacuole disruption phenotype. The data were normalized using Z-score.

Intracellular parasites were fixed with 4% paraformaldehyde prepared in PBS for 30 minutes at room temperature, and were washed three times with PBS before staining with DAPI. Samples were observed with a Zeiss confocal microscope and the obtained images were processed using Image J software. When antibodies were utilized, the samples were permeabilized with 0.2% Triton X-100 prepared in PBS with 100 mM of glycine to block free aldehyde groups, incubations were done for 30 min at room temperature all along the experiment. After blocking with 10% of FBS in PBS containing 0.1% Triton X-100, monoclonal or polyclonal antibodies were added at 1:500 dilutions and incubated at 37°C for 30 min. After three washes, Alexa 488 nm or 560 nm secondary antibodies were added to the same buffer with DAPI and blue Evans, and after incubation at 37°C for 30 min, and three washes with PBS containing 0.1% of Triton X-100, the coverslips were mounted with Mowiol and dried at 37°C before confocal microscopy observations.

Purified tachyzoites (2×10⁵ parasites) of the transgenic T. gondii RH (clone 2F1) strain were used to infect confluent HFF cell monolayer in 24 cm²-well plates for 4 hours and, after washing once with culture medium, the infected cells were incubated with compounds at different concentrations for 48 hours. After recovery by scraping, materials were centrifuged at 5,000 rpm for 10 minutes at 4°C. The pellets were washed once with PBS and lysed for 60 minutes at 50°C in 150 µL of a buffer containing of 1% Triton X-100, 5 mM DTT, 1 mM MgSO₄ and 100 mM HEPES as described previously (Seeber and Boothroyd, 1996; McFadden et al., 1997). After centrifugation at 10,000 rpm for 5 minutes at 4°C, 50 µL of the supernatant was diluted to 50% with lysis buffer. After adding 100 µL of 2 mM chlorophenol red-β-D-galactopyranoside (CPRG), the resulting mixtures were incubated at 30°C overnight and the optical density was measured at λ = 570 nm.

Plaque assays were performed using 24-well plates containing at least 5 day-confluent HFF cells infected with 10³ parasites per well in media, with DMSO as negative control, or in DMSO combined with drugs at 25 µM or 50 µM concentration. Nine days post-infection, the controls and infected HFF cells were stained with crystal violet, as previously described (Sloves et al., 2012; Sangaré et al., 2016). Two independent experiments were performed with identical results.

P. falciparum 3D7 strain was used at the Institut de Recherche et de Développement (IRD-Bénin) to obtain parasite cultures from group O+ red blood cells (RBCs) of healthy adult volunteers. RBCs were washed twice in PBS 1X and then in complete malaria culture medium (CMCM) composed of RPMI 1640 (PAN Biotech, Dutscher, France), 0.8% Albumax, 25 mM HEPES, 0.4 mM hypoxanthine, 0.05 mg/mL gentamicin and 2 mM L-glutamine before storage at 4°C. RBCs infected with P. falciparum 3D7 were grown at 5% hematocrit at 37°C in an atmosphere comprising 5% CO₂, 1% O₂ and 94% N₂. The culture medium was renewed daily and samples were checked for evidence of parasitemia. Prior to drug testing, the infected RBCs were synchronized with 5% sorbitol for 15 minutes at room temperature, and a second synchronization was performed after 48 hours to eliminate persistent mature forms. Compounds were then tested in 96-well flat-bottom plates at a hematocrit of 5% and an initial ring parasitemia of 1% in 250 µL of CMCM. The following concentrations were tested: 5 µM, 25 µM, 50 µM, 100 µM and 150 µM drugs solubilized by DMSO and prepared in CMCM. The experiments were done in quadruplicate and DMSO was used as the negative control. Chloroquine, a well-known anti-plasmodial drug used at 0.005−150 µM served as the positive control. Plates were incubated for 48 hours at 37°C under the same conditions. Thin smears were created by spreading 6 µL of parasitized RBCs (pRBCs) on a glass slide, which was fixed and stained using the RAL 555 kit before using the AxioCam MRc (color) CCD Rev3 camera of ZEISS microscope at X63 to observe and photograph the RBC smears and process the data using ImageJ software.

Both untreated and drug-treated infected RBCs (2.5 µL) were collected from each well and were incubated at 37°C for 60 minutes in a Hoechst 33258-Thiazole Orange mixture at 0.001 mg/mL and 0.005 mg/mL final concentrations, respectively (Grimberg et al., 2008). Samples were analyzed using BD FACS Canto flow cytometer, whereby FlowJO was used for data analysis and statistical analyses were conducted using GraphPad Prism.

Statistical analyses involved two-way ANOVA followed by Tukey’s multiple comparisons test, which were performed using GraphPad Prism version 8.3.0 (GraphPad Software, San Diego, California USA), with p < 0.05 indicating statistical significance. The results are reported as mean of parasite growth percentage ± SD as a function of the logarithm of inhibitor concentration.

With the aim of screening small molecules on a large scale, we created a transgenic T. gondii strain that simultaneously expresses ROP1-mCherry (red fluorescent signal) and TgSORT-GFP (green fluorescent signal) as schematically depicted in Figures 1A , C . For this purpose, we chromosomally appended an encoded mCherry to the rhoptry protein 1 (ROP1) using the plasmid mCherry-LIC-HXGPRT (as explained in the Materials and Methods section and shown in Figure 1A ). This knock-in strategy was adopted, as it ensures steady-state levels of epitope-tagged protein expression via homologous promoters. As indicated in Figure 1B , it resulted in a clonal T. gondii strain expressing ROP1-mCherry at the apical end of intracellular dividing tachyzoites. Next, we transfected a second vector that expresses TgSORT-GFP, thereby disrupting the uracil phosphoribosyl transferase (UPRT) gene by double homologous recombination and conferring the resistance to 5’FUDR ( Figure 1C ). The UPRT gene was chosen because it is non-essential and its knockout has no phenotypic consequences for the parasite, allowing it to salvage uracil from the infected host cells (Fox and Bzik, 2002).

As shown in Figure 1D , we also selected a positive clone for TgSORT-GFP (green signal) in the background of the transgenic T. gondii already expressing ROP1-mCherry (red signal). These intracellular parasites dividing transgenic tachyzoites of T. gondii simultaneously exhibited TgSORT-GFP in the post-Golgi/endosomal-like compartment and ROP1-mCherry at the apical end of the parasites. It is worth noting that these transgenic parasites showed identical rate of multiplication to the parental or wild type parasites growing inside the human foreskin fibroblasts under our experimental in vitro conditions.

For the assay, HFF cells were first infected with a pre-culture of T. gondii expressing both TgSORT-GFP and ROP1-mCherry, as shown in Figure 2 depicting assay workflow. The infected cells were subsequently harvested onto a compound containing 384-well plates and were incubated for 24 hours, after which the samples were fixed and stained with nuclear label-DAPI. Next, confocal images were acquired using an automated microscope and well-based analysis was performed using Columbus software ( Supplementary Table 1 ; Figure 3 ). To establish the assay, the number of HFF cells, the multiplicity of infection (MOI), and the DMSO impact have to be determined to ensure a homogenous T. gondii-infected HFF layer in each well at read-out ( Supplementary Figure 2 ). We have chosen pyrimethamine (Pyr) for this purpose, as this well-known drug is frequently used in toxoplasmosis treatment (Hill and Dubey, 2002; Halonen and Weiss, 2013; Dunay et al., 2018) and is an effective parasite growth inhibitor in assays ( Supplementary Figure 2 ). A dose response of Pyr was determined and the number of parasites per field was quantified. For DMSO, 18 parasites were noted on average per field, declining to 5 parasites in the presence of 0.6 µM Pyr.

As a part of the present study, a pre-screen optimization was also performed by filling two 384-well plates with two Pyr doses (0.6 µM and 10 µM) in a checkerboard style ( Figure 4A ). For the DMSO negative control, 20 parasites were noted on average per HFF cell, declining to about 2 parasites in cells containing 10 µM Pyr. Next, the first 384-well plate from the Tocris library in Supplementary Table 2 was analyzed under the conditions described above. For this purpose, we generated a data processing script utilizing the signal emitted by the ROP1-mCherry and correlating it with the mislocalisation of the rhoptry signal. We first tested 300 compounds of Tocris library and the initial results pertaining to the wells corresponding to a compound yielding positive results were visually validated to concur with the findings related to the rhoptry marker mislocalisation. For the DMSO negative control, the rhoptry fluorescence signal was in the apical form of very distinct spots, while it was largely diffuse for the SB 239063 (Cpd2) and SB 208 (Cpd3) compounds, which correlated with a disruption of the classical and apical pattern of rhoptries ( Supplementary Figure 3 ). Cpd2 and Cpd3 were identified as p38-MAP kinase and TGF-β type I receptor inhibitors, which were reported to reduce T. gondii growth by several authors (Wei et al., 2002; Wiley et al., 2010), indicating that our screen is robust. Thus, Cpd2 and Cpd3 were used as positive controls for the automatic screening of the 900 remaining small molecules distributed in three additional 384-well plates, as shown in Figure 5 . During the analysis script creation, the dual GFP and mCherry labelling approaches were tested but it appears that ROP1-mCherry signal was sufficient to determine the effect of drugs on the parasite morphology inside the parasitophorous vacuole. Indeed, the TgSORT-GFP fluorescence was concentrated and created puncta, whereas the ROP1-mCherry created larger structured area in which the vacuole disruption can be captured in a more visible manner. Therefore, the percentage of positive vacuoles that correlate with the disruption of rhoptry signal was chosen as the key parameter, and the corresponding data plot is shown in Figure 6 . These automatic image-based screenings identified nine further compounds that exhibited the expected phenotypic disruption fluorescence, as demonstrated in Figure 7 . Unlike the negative DMSO control showing viable well-shaped intracellular tachyzoites with distinct apical red signal corresponding to normal rhoptries, these drug-treated wells contained necrotized intracellular parasites. When our visual and automatic screening methods were combined, 12 such compounds were identified, and their chemical structures are shown in Figure 8 , revealing inhibitors of one CB₂ receptor, two ATP-competitive TGF-βRI, one Ca_V3.x pump and one estrogen receptor. The remaining seven compounds are kinase inhibitors, five of which inhibit p38-MAP kinase, and two inhibit ERK and Src family kinases, respectively ( Table 2 ).

The inhibitory activity of the compounds identified through image-based screening was validated by another enzymatic method and the findings revealed a highly significant difference (p < 0.0001) in parasite growth depending on the inhibitor presence and concentration used, reflecting an important effect of inhibitors on the intracellular replication of T. gondii using β-galactosidase assays ( Figures 9A, B ). As compound Cpd5 failed to produce inhibitory effect ( Figure 9A ), it was excluded from the validation process. When the positive compounds were placed in groups of three, their joint IC₅₀ decreased from 15−60 µM to 1.5−2.5 µM. The concentrations below the IC₅₀ of each individual compound was mostly chosen and these compounds were combined according to their ability to inhibit different putative targets or pathways ( Figure 9C ). Each compound of the combined drug mixtures was used at a range of concentrations between 1.0 to 3.0 µM. Nonetheless, the difference observed between the results obtained by applying inhibitor mixes to T. gondii and the untreated parasites was significant (p = 0.0024), indicating beneficial synergistic effect. Moreover, when infected confluent HFF cells were subjected to drug inhibition for nine days to allow plaque formation, which corresponds to several rounds of invasion, egress and reinvasion, parasite growth was completely abolished for all six compounds tested, i.e., Cpd2, Cpd3, Cpd4, Cpd7, Cpd10 and Cpd11 ( Figure 9D ). Even at high concentrations (25 and 50 µM), the confluent monolayer of HFF cells remained intact, as shown by homogenous crystal violet staining at these doses. Furthermore, we determined the effects of these drugs on the classical fluorescence patterns of few intracellular organelles of the parasites using specific antibodies and confocal imaging. Figure 9E showed that intracellular tachyzoites of parental RH strain treated with Cpd2, 3, 4, 10 and 11 led to detrimental effects on the apical signal of rhoptries with from more fragmented patterns until a total disappearance of the fluorescence signals in some parasites. These drugs also severely affected the apical pattern of micronemes except for Cpd 6 as compared to the four other drugs. The compounds Cpd2 and Cpd10 were less active on the pattern of secretion dense granule proteins into parasitophorous vacuoles ( Figure 9E ). The inner complex membranes of drug-treated parasites appears normal, except that the parasites were shorter in sizes compared to parasites treated with DMSO alone ( Figure 9E ). As expected, these parasites treated with DMSO alone showed typical fluorescence signals in the apical end for rhoptries and micronemes. Altogether, these data confirms the detrimental effects observed on ROP1-mCherry containing rhoptries in the intracellular parasites during our automatic and miniaturized drug screening.

To test the inhibitory effects of eight compounds identified during our image-based screen on P. falciparum growth, we used infected red blood cells (RBCs) synchronized twice with sorbitol and subjected them to flow cytometry analysis. The findings indicated that four compounds (Cpd3, Cpd4, Cpd10 and Cpd11) were active against P falciparum and the infected RBC cultures were cleared of ring stages ( Figure 10A ). We estimated the IC₅₀ of these drugs between 1.0 to 5.0 µM. Moreover, increasing inhibitor concentrations resulted in a significant (p < 0.0001) decrease in P. falciparum growth in RBCs. When trophozoite and schizonte forms of P falciparum were counted by flow cytometry ( Figures 10B, C ), identical inhibitory effect of these four compounds was affirmed. The remaining four compounds inhibited P falciparum growth, but their activity was below that noted for Cpd3, Cpd4, Cpd10 and Cpd11 ( Figure 10 ). In all cases, inhibitory activity and absence of ring stage in the infected RBCs were confirmed by direct observation of stained red blood smears, as shown in Figure 10D . Only the negative DMSO controls exhibited numerous ring stages, confirming that these compounds are capable of directly targeting parasites, since P. falciparum was grown in non-nucleated red blood cells characterized by reduced metabolism and lack of intracellular organelles.