Inducible lacZ expression in the P. falciparum NF54i-lacZ.

(A) Schematic representation of the P230p locus in parental NF54DiCre and NF54i-lacZ parasites before and after rapamycin-induced recombination. Arrows indicate primers used for diagnostic integrations PCRs. Striped boxes represent homology regions used for CRISPR/Cas9-based gene editing. The gRNA binding site is indicated. loxPint, loxP-intron. (B) Diagnostic PCR confirmed the integration of the loxPInt-gfp-lacZ expression cassette in P. falciparum NF54i-lacZ, and DiCre-induced recombination. Diagnostic hdhfr-yfcu (1595 bases) PCR demonstrates loss of selectable markers in the NF54i-lacZ (marker control). Primers used are shown in Figure 1A and Figure 1 – figure supplement 1 (#1-#6). wt: wild type; int: integration; pos. ctrl: positive control (1252 bp). (C) GFP and Hoechst fluorescence signals, and (D) chemiluminescence emitted from NF54i-lacZ parasites before (-Rapa) or after (+Rapa, measured after 48 h) rapamycin-induced recombination, and controls (vehicle control (+DMSO), uRBC control treated with Rapa); error bars represent the standard error of the mean (SEM) of 5 biological replicates conducted in at least technical duplicates. Rapa: rapamycin; GFP: green fluorescent protein; DIC: differential interference contrast; uRBC: uninfected red blood cells. The β-galSENSOR probe was used at 10 µM and rapamycin at 100 nM.

Limit of quantification of recombined P. falciparum NF54i-lacZ parasites.

Serially diluted NF54i-lacZ parasites were incubated for varying durations (24-168 hours) with 100 nM rapamycin or the corresponding concentration of DMSO before chemiluminescence measurement using 10 μM β-galSENSOR probe. The signal linearly correlated with the initial parasite inoculum, improving the limit of quantification over time from ∼300 initial parasites per well to as low as ∼1 initial parasite(s) per well (indicated by the vertical dotted line). Controls (uRBC treated with rapamycin or DMSO) are indicated. The LOQ was tested in ≥ three biological replicates with at least technical duplicates; error bars represent the standard error of the mean (SEM) of biological replicates and may not be visible where smaller than the plotted symbols. The LOQ was set to the final initial parasite number per well for which the generated signal against the negative control (uRBC treated with Rapa) was considered significantly different in an unpaired t-test (p < 0.05) and a linear relationship was given. uRBC: uninfected red blood cells; Rapa: rapamycin.

Schematic overview of the MULT-i2 assay workflow for dual drug combination testing.

A 0 h calibration plate containing a parasite serial dilution, starting at 10⁵ parasites, is prepared as a reference. In parallel, checkerboard plates with the desired drug concentrations and combinations are assembled, and 10⁵ parasites are added per condition. The 0 h calibration plate is immediately treated with rapamycin and incubated for 120 h to induce lacZ expression and β-galactosidase accumulation. Drug-treated plates are incubated under drug pressure for the desired duration, after which drugs are removed by extensive washing. Rapamycin is then added for 120 h to enable surviving parasites to express the reporter enzyme β-galactosidase. Plates are subsequently frozen for later chemiluminescent signal measurement. The 0 h calibration plate establishes the relationship between chemiluminescence signal intensity and initial parasite number, allowing quantification of viable parasites after drug exposure. In this configuration, the assay supports testing of two compounds (A and B) across seven concentrations each, yielding 49 dual combinations. The resulting data are used to model pharmacodynamic parameters using the GPDI framework, including interaction parameters (INT), lag phases, maximal effect (Emax), and half-maximal effective concentration (EC50).

© 2026, BioRender Inc. Parts of this image were created with BioRender are not made available under the same license as the Reviewed Preprint.

Comparison of time-killing profiles of four reference compounds between the MULT-i2 assay and the published PRR v2.

Artemisinin, chloroquine, pyrimethamine and atovaquone were tested in the MULT-i2 assay using the same compound concentration as in the published PRR v2 (Walz et al., 2023). Data points represent the mean of ≥ three biological replicates (two for the 120-hour time point in the MULT-i2 assay) in four technical replicates, error bars represent the standard error of the mean (SEM) of the biological replicates.

Comparison of pharmacodynamic parameters calculated with the published R pipeline (Walz et al., 2023), based on data from ≥ three biological replicates in one (MULT-i2 assay) in four technical replicates.

Log10(PRR): log10(parasite reduction ratio); PCT99.9%: 99.9% parasite clearance time (h); Emax: maximal drug effect – parasite killing rate (h-1); hypo.: hypoxanthine. The 95% confidence interval is shown in square brackets.

Model fits based on cPRR assay or MULT-i2 assay data for the tested combination atovaquone-proguanil.

Model predictions are shown as orange (BI null-interaction) or blue (BI interaction) line, points represent original data (cPRR - one biological replicate with four technical replicates, three tested timepoints; MULT-i2 - three biological replicates with one technical replicate, five tested time points), the LLOQ is indicated by a dashed brown line. Dark grey boxes indicate the concentrations (nM) of the drugs used: ATO: atovaquone; PRO: proguanil; BI: Bliss Independence; LLOQ: lower limit of quantification.

Model fits based on cPRR assay or MULT-i2 assay data for the tested combination, pyronaridine-piperaquine.

Model predictions are shown as orange (BI null-interaction) or blue (BI interaction) line, points represent original data (cPRR - one biological replicate with four technical replicates, three tested time points; MULT-i2 - three biological replicates with one technical replicate, five tested time points), the LLOQ is indicated by a dashed brown line. Dark grey boxes indicate the concentrations (nM) of the drugs used: PYR: pyronaridine; PIP: piperaquine; BI, Bliss Independence; LLOQ, lower limit of quantification.

Overview of the two-plasmid CRISPR/Cas9-based gene editing strategy.

Schematic of the P230p locus in NF54DiCre parasites and the plasmids used for CRISPR/Cas9-mediated gene editing (P230p-loxPInt-gfp-lacZ and pHF-gC-P230p) to generate the NF54i-lacZ parasites.

Susceptibility to the antifolate drug WR99210 and parasitized erythrocyte infection rate of the novel NF54i-lacZ compared to NF54WT.

(A) IC50 values of WR99210; error bars represent the standard error of the mean (SEM) of three biological replicates conducted in technical duplicates and (B) parasitized erythrocyte infection rates (within a period of 48 hours) are comparable between NF54i-lacZ parasites, either rapamycin-treated or untreated, and the NF54WT strain; error bars represent the SEM of three biological replicates. Rapa: rapamycin.

Compartmental representation of the developed pharmacometric model.

Assay signal was translated to parasitemia. Drug effects were modelled using a sigmoidal maximum effect model and Bliss Independence was used as null-interaction model with Emax scaled. For atovaquone, drug effect was modeled with a lag time which is implemented on Emax with the term (1-e-klag× t). A, compound A; B, compound B; N0: initial parasitemia, kg: growth rate, Emax: maximal drug effect - maximum killing rate, EC50: concentration stimulating 50% of Emax, Hill: Steepness of the concentration-effect relationship, klag: first-order delay rate for atovaquone until full killing rate is achieved.