Multidimensional in vitro assay for antimalarial combination testing and pharmacodynamic modeling – the MULT-i² assay

Reviewer #1 (Public review):

Summary:

The authors describe a clever genetic system based on rapamycin-inducible expression of a beta-galactose reporter. The authors compare this spectrophotometer-based readout to the parasite reduction rate version 2 (PRRv2) recently described by some of the same authors and based on incorporation of [3H]-hypoxanthine. The results are generally comparable, with some differences for slower-acting compounds. The authors report that this format is better suited for higher-throughput studies and requires less time to quantify the time-dependent onset of parasiticidal action compared with the PRRv2.

Strengths:

This is a very well-executed and well-described body of work with a comprehensive set of analyses.

Weaknesses:

The authors should revise their text to also describe other methods used to quantify parasite growth. This method saves time compared to the PRRv2 but is too complex for simple screening of antiplasmodial activity of agents tested alone. Its value lies in assessing the speed of action of compounds tested in combination.

There are a number of areas for improvement:

(1) Many antimalarials have quite specific times of action. Are these MULTI-i2 assays, and the comparator PRRv2 assays, conducted with asynchronous cultures? This should be described in the methods and referred to in the text (apologies if I missed some references).

(2) The authors correctly state that flow cytometry-based readouts, such as with MitoTracker alone, can limit throughput and that MitoTracker alone can produce spurious results. The authors should cite work from other labs that combine MitoTracker with a nuclear dye, such as SYBR Green I. I think others have also been used, such as YoYo-1, which overcomes the limitations of using MitoTracker alone. Also, many labs use a nuclear dye such as SYBR Green I in a spectrophotometer-based format that enables rapid processing of plates at scale (96, 384, or even 1536 wells per plate). Luciferase-based screens have also been used in large-scale screening campaigns. The introduction should cite these various approaches, especially as the MULTI-i2 method is quite a complex screen with an initial period of drug exposure (up to 3 days) followed by a five-day phase initiated by rapamycin addition to induce expression of the beta-gal sensor.

(3) It would be helpful for authors to provide some indication of the cost comparison between the PPRv2 and MULTI-i2.

(4) Also, the authors should indicate whether these reagents will be deposited in a repository such as BEI Resources. They should also indicate conditions for other groups to request these materials, such as whether an MTA is required.

(5) The pharmacological models are interesting, but likely well out of the range of expertise of many labs. Has code been deposited into public repositories that make it possible for other labs to implement these analyses?

Reviewer #2 (Public review):

Summary

Antimalarial combination therapy is the standard of care for malaria, a disease that impacts hundreds of millions of people annually. Combination therapy is crucial for effectively treating the disease and delaying the emergence of drug resistance. Despite the importance of choosing appropriate partner antimalarials for combination therapy, drug interactions are typically evaluated late in the course of drug development. Standard in vitro assays that determine synergistic, antagonistic, or additive interactions between drug combinations rely on measuring inhibition of parasite proliferation, which is inadequate for translation to pharmacodynamic models for parasite clearance in the patient. Direct measurement of parasite viability under drug treatment has previously relied on methods that are labor and resource-intensive, limiting applications to single compounds and single concentrations. Here, Hellingman et al make use of an inducible chemiluminescence reporter to measure cell viability and apply this novel approach to quantify drug interactions. The methodology is a significant improvement upon prior methods, requiring significantly fewer resources, half the time, and substantially less handling than the standard PRRv2 assay, whilst maintaining high resolution and sensitivity.

They assess the limit of detection for the improved method and cross-reference their results for single drugs at a single concentration with the currently standard PRRv2 assay. The authors next established analytical methods to characterize the impact of drug combinations on parasite viability using the GDPI pharmacodynamic model and compared their MULT-i2 assay to the prior cPRR approach. Their refined workflow allowed them to comprehensively evaluate the known synergistic combination between atovaquone and proguanil with greater resolution than the comparable cPRR assay and identified additional interaction parameters between the fast-acting antimalarials piperaquine and pyrimethamine. Overall, the authors demonstrate that their inducible lacZ system provides significant advantages compared with prior approaches to determine parasite viability. They convincingly demonstrate the strengths of their approach by characterizing two antimalarial combinations at much greater resolution than previously possible with prior methods. The system and methods established here will be particularly useful for evaluating novel antimalarial combinations with chemical series in preclinical evaluation and to optimize future antimalarial therapies.

Strengths:

The streamlined approach relies on induction of the lacZ enzyme only after drug washout. As opposed to when stably expressed, this allows the authors to estimate parasite viability without undergoing serial dilutions to estimate viable parasite titers. This innovation vastly reduced resource and time intensity, enabling greater throughput for parasite viability estimation. The established methodology and analysis pipeline enabled the testing of 49 drug combinations for parasite viability in the MULT-i2 assay compared to only 9 in the conventional cPRR assay. This provided improved resolution in the ability to estimate drug combination parameters in a pharmacodynamic model. The ability to comprehensively characterize combination pharmacodynamic properties in vitro will have important implications for downstream modelling of in vivo combinations, and for optimizing future antimalarial combination therapies.

The authors made good use of modelling and AICc for parametric estimation and model evaluation to demonstrate the advantages of the richer dataset afforded by the MULT-i2 assay.

Weaknesses:

The authors correctly identified a range of confounding effects that lead to artefacts in their assay results when compared to the cPRR assay. For instance, the authors observed reduced signal at high parasite density during recovery due to overgrowth and likely enzyme degradation, and suggested residual signal may remain from non-proliferating sexual stage parasites surviving drug treatment that would not be detected in the cPRR assay.

Measurement of parasite viability in the MULT-i2 assay was achieved by extrapolating the chemoluminescence signal to that of a serial dilution of parasites made at the initiation of drug treatment. How did the authors account for differing levels of enzyme expression at early (e.g., ring) vs late stage parasites (trophozoite or schizonts)? Were cultures synchronized prior to initiation of assays? Could differences in life-cycle progression following drug treatment be an additional confounding factor that may account for differences with the PRRv2 assay?

The addition of an inducible element is an improvement of their earlier lacZ/β-galSENSOR (PMID: 41575867); however, the authors fail to explain why this is an improvement and how this adds additional merit over the initial system. While the authors compare their new assay to the PRRv2, they fail to compare it to their own non-inducible lacZ/β-galSENSOR system. Their non-inducible system already showed superiority to the cPRR assays, and it would be good to show how they compare and what the advantages of the new system are over the old. e.g., how is the signal-to-noise improved? How does the sensitivity compare? How quickly does the can the signal be detected after induction? They show signal after 48h, but it would be very useful to the community to look at earlier timepoints as well and compare them to the uninduced line and a line that has been induced 48h earlier to match the expression patterns throughout the lifecycle (something like 2h,4h,6h, 12h, and 24h).

Is the chemiluminescence signal for the i-lacZ induced parasites comparable to the stably expressed lacZ parasites previously characterized by the group? If so, do the authors consider this inducible iteration a complete replacement for PRR assays?

Reviewer #3 (Public review):

In this manuscript, the authors strived to develop a highly efficient drug survival assay for in vitro cultured human malaria parasites P. falciparum. This was done by generating a transgenic P. falciparum line using a creLox strategy that allows detection of (presumably) viable parasites by a β-lactamase assay. To estimate the Limit of quantification of the recombined P. falciparum NF54i-lacZ, the authors ultimately designed a protocol in which viable parasites are detected by the luminescence of β-D-galactoside generated by β-lactamase within the transgenic parasites. For this, the parasite must be incubated with rapamycin for 120 hours to induce CreLox recombinase, which places β-lactamase under an active promoter. Using this assay, termed MULTI-i2, the author shows interactions between two antimalarial drug pairs that were previously demonstrated by another assay. In the case of pyronaridine and piperaquine pair, the NULT-i2 assay generated some additional insights compared to the previous assay, presumably by virtue of including more concentration datapoints. In conclusion, the authors argue that the MULTI-i2 assay is much less resource-intensive and time-consuming and can be applied on a large scale at a much lower cost and with the highest efficiency.

Overall, the data generated in this manuscript are clear and well represented, and I am convinced that MULTI-i2 provides yet another of many drug assays for malaria parasites and could be put to good use. However, I struggle to fully appreciate the merit of his study, as the manuscript reads more like a technical document than a scientific study.

I particularly lack an understanding of the strengths and weaknesses/limitations of the MULTI-i2 methodology and, thus, its applicability. I also do not fully appreciate the need for such an elaborate luminescence-based experimental setup. It would be good if some of these issues were addressed.

Specifically:

(1) The whole assay is based on detecting parasites by luminescence after 120 hr (5 days) after drug exposure. During that time, presumably the parasites that survived the drug pressure regrow to a detectable level and, at the same time, perform efficacious CreLox-based recombination to produce β-D-galactoside for detection. Is this necessary? How superior is this detection method to other methods, such as Fluorescence-assisted Cell Sorting (FACS), etc? Moreover, the 5-day growth-CreLox-β-D-galactoside production could introduce a series of confounding effects. In my view, more studies (beyond comparisons with a single existing method) would be useful for understanding this entire process.

(2) Related to that above, how would MULTI-i2 perform in case of drugs that do not necessarily kill all parasites, such as artemisinin? In the case of artemisinin, it is becoming evident that at least a small fraction of the parasite revives after treatment via a temporary dormancy state. This has, in fact, also been shown for other drugs such as mefloquine, pyrimethamine, etc. Would such a situation produce a range of false readings? In general, in its current state, it is hard to see what the limitations of this method are, which makes it hard to decide whether to use it for a particular application.

(3) Given the stated cost and labor efficiency of MULTI-i2, it is disappointing to see only two applications for two drug pairs: atovaquone/proguanil and piperquine/pyronaridine, for both of which their interactions were already known. The manuscript would benefit greatly if the authors demonstrated more drug interactions and identified (and ultimately validated) new ones. This would certainly make MULT-i2 method more attractive. In particular, it would be nice to see if one could use MULTI-i2 for studies of triple combinations as enthusiastically suggested.

(4) Throughout the manuscript, the authors claim that MULTI-i2 is considerably less expensive and can be done much faster than previous methods. In my view, this is not exactly a scientific argument. The cost of an assay depends heavily on the cost of reagents and labor, which are subject to market price fluctuations. The efficiency and time consumption can very much depend on laboratory organization, etc. Unless the author could specifically demonstrate where and how these assays are cheaper and faster, I suggest not discussing this.

Author response:

Reviewer 1 (Public review):

Summary:

The authors describe a clever genetic system based on rapamycin-inducible expression of a beta-galactose reporter. The authors compare this spectrophotometer-based readout to the parasite reduction rate version 2 (PRR v2) recently described by some of the same authors and based on incorporation of [3H]-hypoxanthine. The results are generally comparable, with some differences for slower-acting compounds. The authors report that this format is better suited for higher-throughput studies and requires less time to quantify the time-dependent onset of parasiticidal action compared with the PRR v2.

Strengths:

This is a very well-executed and well-described body of work with a comprehensive set of analyses.

Weaknesses:

The authors should revise their text to also describe other methods used to quantify parasite growth. This method saves time compared to the PRR v2 but is too complex for simple screening of antiplasmodial activity of agents tested alone. Its value lies in assessing the speed of action of compounds tested in combination.

We thank reviewer 1 for the supportive feedback and for raising some important points.

Many antimalarials have quite specific times of action. Are these MULT-i² assays, and the comparator PRR v2 assays, conducted with asynchronous cultures? This should be described in the methods and referred to in the text (apologies if I missed some references).

We thank the reviewer for this important comment. Both, the MULT-i² and PRR v2 assays were performed using asynchronous parasite cultures. This information is included in the Methods section together with the relevant references. To improve clarity, we will also explicitly state this in the main text.

The authors correctly state that flow cytometry-based readouts, such as with MitoTracker alone, can limit throughput and that MitoTracker alone can produce spurious results. The authors should cite work from other labs that combine MitoTracker with a nuclear dye, such as SYBR Green I. I think others have also been used, such as YoYo-1, which overcomes the limitations of using MitoTracker alone. Also, many labs use a nuclear dye such as SYBR Green I in a spectrophotometer-based format that enables rapid processing of plates at scale (96, 384, or even 1536 wells per plate). Luciferase-based screens have also been used in large-scale screening campaigns. The introduction should cite these various approaches, especially as the MULT-i² method is quite a complex screen with an initial period of drug exposure (up to 3 days) followed by a five-day phase initiated by rapamycin addition to induce expression of the beta-gal sensor.

We thank the reviewer for this helpful suggestion. In the Introduction we will mention and describe alternative approaches for assessing parasite viability. This will also include the work by Maiga et al., which combines MitoTracker with a nuclear dye to improve the reliability of flow cytometry-based readouts. We will revise the text to explicitly mention the use of dual straining to make this discussion more explicit.

We agree that several additional methods, such as luciferase-based reporter systems, have been successfully applied in antimalarial screening. However, these approaches are primarily designed to assess parasite growth inhibition rather than directly measuring parasite viability after drug exposure, which is the focus of the present study. Readout methods used to assess parasite viability in a PRR assay setup are so far based on HRP2-ELISA (de Carvalho et al.), MitoTracker and SYBR green staining (Maiga et al.) and [³H]-hypoxanthine incorporation (Sanz et al.; Walz et al.) as cited in the manuscript. Many other readout methods to assess parasite growth have other limitations as briefly discussed in Hellingman et al, 2024. A comprehensive comparison and review of all available readout methods would therefore be beyond the scope of this manuscript.

It would be helpful for authors to provide some indication of the cost comparison between the PPR v2 and MULT-i².

We thank the reviewer for this valuable suggestion. We agree that a comparison of the costs associated with the PRR v2 and MULT-i² assays would be informative, but while the consumable costs provide one measure of assay expense, we consider the reduction in hands-on time and the simplified workflow to be the main contributors to the overall cost advantage of the MULT-i² assay. These reductions in labor requirements are subject to large regional differences and impossible for us to access. Nevertheless, together with the increased throughput and the reduced labor, make the MULT-i² assay more cost-effective for larger-scale applications compared with the PRR v2 assay.

Also, the authors should indicate whether these reagents will be deposited in a repository such as BEI Resources. They should also indicate conditions for other groups to request these materials, such as whether an MTA is required.

We thank the reviewer for this important suggestion. The engineered parasite line will be made available for non commercial use to other researchers upon request. An MTA will be required excluding commercial use of the provided strains. The detailed code used for data analysis is available upon request, and an example code file has already been included as a Supplementary File.

The pharmacological models are interesting, but likely well out of the range of expertise of many labs. Has code been deposited into public repositories that make it possible for other labs to implement these analyses?

We thank the reviewer for this valuable comment. We agree that implementation of pharmacological modeling approaches can represent a barrier for laboratories without prior experience in pharmacometric analysis, particularly due to the requirement for specialized software such as NONMEM. To facilitate implementation, an example code is provided in the Supplementary File. The final model was developed using a forward–backward selection approach for parameter estimation and model refinement as described in the Methods section. These additions should help other researchers adapt the approach to their own datasets.

Reviewer 2 (Public review):

Summary

Antimalarial combination therapy is the standard of care for malaria, a disease that impacts hundreds of millions of people annually. Combination therapy is crucial for effectively treating the disease and delaying the emergence of drug resistance. Despite the importance of choosing appropriate partner antimalarials for combination therapy, drug interactions are typically evaluated late in the course of drug development. Standard in vitro assays that determine synergistic, antagonistic, or additive interactions between drug combinations rely on measuring inhibition of parasite proliferation, which is inadequate for translation to pharmacodynamic models for parasite clearance in the patient. Direct measurement of parasite viability under drug treatment has previously relied on methods that are labor and resource-intensive, limiting applications to single compounds and single concentrations. Here, Hellingman et al make use of an inducible chemiluminescence reporter to measure cell viability and apply this novel approach to quantify drug interactions. The methodology is a significant improvement upon prior methods, requiring significantly fewer resources, half the time, and substantially less handling than the standard PRR v2 assay, whilst maintaining high resolution and sensitivity.

They assess the limit of detection for the improved method and cross-reference their results for single drugs at a single concentration with the currently standard PRRv2 assay. The authors next established analytical methods to characterize the impact of drug combinations on parasite viability using the GDPI pharmacodynamic model and compared their MULT-i² assay to the prior cPRR approach. Their refined workflow allowed them to comprehensively evaluate the known synergistic combination between atovaquone and proguanil with greater resolution than the comparable cPRR assay and identified additional interaction parameters between the fast-acting antimalarials piperaquine and pyrimethamine. Overall, the authors demonstrate that their inducible lacZ system provides significant advantages compared with prior approaches to determine parasite viability. They convincingly demonstrate the strengths of their approach by characterizing two antimalarial combinations at much greater resolution than previously possible with prior methods. The system and methods established here will be particularly useful for evaluating novel antimalarial combinations with chemical series in preclinical evaluation and to optimize future antimalarial therapies.

Strengths:

The streamlined approach relies on induction of the lacZ enzyme only after drug washout. As opposed to when stably expressed, this allows the authors to estimate parasite viability without undergoing serial dilutions to estimate viable parasite titers. This innovation vastly reduced resource and time intensity, enabling greater throughput for parasite viability estimation. The established methodology and analysis pipeline enabled the testing of 49 drug combinations for parasite viability in the MULT-i² assay compared to only 9 in the conventional cPRR assay. This provided improved resolution in the ability to estimate drug combination parameters in a pharmacodynamic model. The ability to comprehensively characterize combination pharmacodynamic properties in vitro will have important implications for downstream modelling of in vivo combinations, and for optimizing future antimalarial combination therapies.

The authors made good use of modelling and AICc for parametric estimation and model evaluation to demonstrate the advantages of the richer dataset afforded by the MULT-i² assay.

We thank reviewer 2 for her/his appreciation of our work.

Weaknesses:

The authors correctly identified a range of confounding effects that lead to artefacts in their assay results when compared to the cPRR assay. For instance, the authors observed reduced signal at high parasite density during recovery due to overgrowth and likely enzyme degradation, and suggested residual signal may remain from non-proliferating sexual stage parasites surviving drug treatment that would not be detected in the cPRR assay.

Measurement of parasite viability in the MULT-i² assay was achieved by extrapolating the chemoluminescence signal to that of a serial dilution of parasites made at the initiation of drug treatment. How did the authors account for differing levels of enzyme expression at early (e.g., ring) vs late stage parasites (trophozoite or schizonts)? Were cultures synchronized prior to initiation of assays? Could differences in life-cycle progression following drug treatment be an additional confounding factor that may account for differences with the PRR v2 assay?

We thank the reviewer for raising this important point. All, the MULT-i² and PRR v2 assay were performed using asynchronous parasite cultures. We will clarify this in the revised manuscript.

We agree that parasite developmental stages may influence the MULT-i² readout, as LacZ expression levels differ between parasite stages, with differences observed between ring stages and more mature trophozoite/schizont stages as published by Hellingman et al., 2024. This represents a potential source of variability, as the MULT-i² assay quantifies the amount of expressed reporter enzyme rather than directly measuring parasite numbers at the time of readout. The use of asynchronous cultures minimizes the impact of stage-specific effects by providing a mixed parasite population representative of the natural distribution of developmental stages. Nevertheless, we acknowledge that differences in parasite stage progression following drug exposure may contribute to variation in the extrapolated parasite numbers and may partially explain differences observed between the MULT-i² and PRR v2 assay measurements. We will add this consideration to the Discussion.

The addition of an inducible element is an improvement of their earlier lacZ/β-gal^SENSOR (PMID: 41575867); however, the authors fail to explain why this is an improvement and how this adds additional merit over the initial system. While the authors compare their new assay to the PRR v2, they fail to compare it to their own non-inducible lacZ/β-gal^SENSOR system. Their non-inducible system already showed superiority to the cPRR assays, and it would be good to show how they compare and what the advantages of the new system are over the old. e.g., how is the signal-to-noise improved?

We thank the reviewer for this important comment. The main improvement provided by the inducible system is the temporal separation of parasite growth/drug exposure from reporter expression. In the original non-inducible lacZ/β-gal^SENSOR system, reporter expression occurs continuously throughout the assay, resulting in accumulation of β-galactosidase during parasite growth/drug exposure and therefore an increasing background signal. Consequently, quantification relies on endpoint reporter levels and does not allow the reporter expression window to be standardized independently of parasite exposure history.

In contrast, in the MULT-i² system, reporter expression is initiated only after addition of rapamycin post antimalarial drug washout. This prevents reporter accumulation during the drug exposure window and ensures a defined reporter enzyme accumulation window after drug exposure. Importantly, this allows parasite numbers to be extrapolated from a calibration curve generated at the time of induction, which would not be possible with the non-inducible system because reporter expression would continue after drug removal and would depend on the previous culture history.

We will revise the manuscript to more clearly describe these advantages and to emphasize that the key benefit of the inducible system is not simply an increase in signal intensity, but improved control of reporter expression, reduced background accumulation, and the ability to perform quantitative parasite reduction rate measurements.

How does the sensitivity compare? How quickly does the can the signal be detected after induction? They show signal after 48h, but it would be very useful to the community to look at earlier timepoints as well and compare them to the uninduced line and a line that has been induced 48h earlier to match the expression patterns throughout the lifecycle (something like 2h,4h,6h, 12h, and 24h).

We thank the reviewer for this important suggestion. We acknowledge that the sensitivity of the MULT-i² readout depends on both the initial parasite density and the duration of the induction period and that a detailed characterization of the induction kinetics, including earlier time points after rapamycin addition, would provide additional information on the sensitivity and temporal resolution of the MULT-i² system.

In the present study, we focused on the time window relevant for application of the assay in a PRR assay workflow and routine drug screening setting. Earlier time points (<24 h after induction) were therefore not systematically evaluated. The selected time points were chosen based on the expected kinetics of the loxP-DiCre recombination system, which has previously been reported to achieve high recombination efficiency within one asexual parasite cycle, (Collins et al., 2013) and shown with own data in this study, as well as on practical considerations for implementation in routine workflows.

Is the chemiluminescence signal for the i-lacZ induced parasites comparable to the stably expressed lacZ parasites previously characterized by the group? If so, do the authors consider this inducible iteration a complete replacement for PRR assays?

We thank the reviewer for this question. The chemiluminescence signal obtained with the inducible lacZ (i-lacZ) parasites is comparable to that observed with the previously characterized constitutively expressing lacZ parasites. However, the inducible system provides an important additional advantage by avoiding continuous β-galactosidase production and accumulation during parasite growth, thereby reducing background signal and enabling a controlled reporter expression window.

We do not consider the MULT-i² assay to be a replacement for classical PRR assays. Rather, we consider it a complementary approach that enables more efficient screening and characterization of drug combinations, particularly by providing information on the time-dependent onset of parasiticidal activity in a higher-throughput format. Promising combinations identified using MULT-i² assay can subsequently be investigated in more extensive PRR assays.

Reviewer 3 (Public review):

In this manuscript, the authors strived to develop a highly efficient drug survival assay for in vitro cultured human malaria parasites P. falciparum. This was done by generating a transgenic P. falciparum line using a creLox strategy that allows detection of (presumably) viable parasites by a β-lactamase assay. To estimate the Limit of quantification of the recombined P. falciparum NF54i-lacZ, the authors ultimately designed a protocol in which viable parasites are detected by the luminescence of β-D-galactoside generated by β-lactamase within the transgenic parasites. For this, the parasite must be incubated with rapamycin for 120 hours to induce CreLox recombinase, which places β-lactamase under an active promoter. Using this assay, termed MULT-i², the author shows interactions between two antimalarial drug pairs that were previously demonstrated by another assay. In the case of pyronaridine and piperaquine pair, the NULT-i² assay generated some additional insights compared to the previous assay, presumably by virtue of including more concentration datapoints. In conclusion, the authors argue that the MULT-i² assay is much less resource-intensive and time-consuming and can be applied on a large scale at a much lower cost and with the highest efficiency.

Overall, the data generated in this manuscript are clear and well represented, and I am convinced that MULT-i² provides yet another of many drug assays for malaria parasites and could be put to good use. However, I struggle to fully appreciate the merit of his study, as the manuscript reads more like a technical document than a scientific study.

We thank reviewer 3 for her/his appreciation of our work.

I particularly lack an understanding of the strengths and weaknesses/limitations of the MULT-i² methodology and, thus, its applicability. I also do not fully appreciate the need for such an elaborate luminescence-based experimental setup. It would be good if some of these issues were addressed.

Specifically:

(1): The whole assay is based on detecting parasites by luminescence after 120 hr (5 days) after drug exposure. During that time, presumably the parasites that survived the drug pressure regrow to a detectable level and, at the same time, perform efficacious CreLox-based recombination to produce β-D-galactoside for detection. Is this necessary? How superior is this detection method to other methods, such as Fluorescence-assisted Cell Sorting (FACS), etc? Moreover, the 5-day growth-CreLox-β-D-galactoside production could introduce a series of confounding effects. In my view, more studies (beyond comparisons with a single existing method) would be useful for understanding this entire process.

We thank the reviewer for raising this important point regarding the rationale, applicability, and limitations of the MULT-i² methodology.

Quantification of viable parasites after drug exposure remains challenging, particularly when surviving parasites are present at low frequencies or require extended recovery periods. Current approaches, such as the parasite reduction ratio (PRR) assay based on [³H]-hypoxanthine incorporation, provide sensitive measurements of replicating parasites but are labor-intensive, require specialized infrastructure, and are not easily scalable for large numbers of drug combinations. Alternative approaches based on HRP2 detection no longer rely on radioactive readouts but generally provide lower sensitivity, particularly when quantifying low levels of surviving parasites within a shorter time frame.

The MULT-i² assay was developed to address these limitations by combining a highly sensitive chemiluminescent β-galactosidase readout with an inducible reporter system. The 5-day induction period after drug exposure serves as a controlled gene expression step, allowing surviving parasites to recover and produce sufficient reporter signal for sensitive quantification using a standard plate reader. This approach enables higher-throughput assessment of parasiticidal activity while avoiding radioactive readouts and reducing the need for labor-intensive dilution-based approaches.

We acknowledge that the recovery and reporter expression period introduces additional biological steps compared with direct parasite detection methods and may therefore represent a potential source of variability. The MULT-i² assay is not intended to replace all existing viability measurements but rather to provide a complementary screening tool for investigating larger numbers of drug combinations. More detailed comparisons with additional detection platforms, including fluorescence-based approaches such as flow cytometry, would be valuable; however, a comprehensive comparison of all available parasite viability readouts was beyond the scope of this study. We will add more explanations to the Discussion including the strengths and limitations.

(2) Related to that above, how would MULT-i² perform in case of drugs that do not necessarily kill all parasites, such as artemisinin? In the case of artemisinin, it is becoming evident that at least a small fraction of the parasite revives after treatment via a temporary dormancy state. This has, in fact, also been shown for other drugs such as mefloquine, pyrimethamine, etc. Would such a situation produce a range of false readings? In general, in its current state, it is hard to see what the limitations of this method are, which makes it hard to decide whether to use it for a particular application.

We thank the reviewer for raising this important point regarding the interpretation and applicability of the MULT-i² assay. We agree that distinguishing between growth inhibition assays and viability-based assays is essential when interpreting the response to drugs that induce temporary parasite dormancy or delayed recovery.

The MULT-i² assay was specifically developed as a viability-based approach and therefore differs fundamentally from conventional IC50 assays, which primarily measure inhibition of parasite growth during drug exposure and may not capture parasites that survive treatment through temporary growth arrest or dormancy. Similar to the PRR assay, the MULT-i² assay measures the ability of surviving parasites to recover and proliferate after drug exposure. Therefore, parasites that temporarily enter a dormant state but subsequently resume replication are expected to contribute to the measured signal rather than representing false-positive or false-negative results.

This is illustrated by the artemisinin experiments presented in this study, where the MULT-i² assay captures the recovery of surviving parasites following treatment as it does the PRR v2 assay.

(3) Given the stated cost and labor efficiency of MULT-i², it is disappointing to see only two applications for two drug pairs: atovaquone/proguanil and piperquine/pyronaridine, for both of which their interactions were already known. The manuscript would benefit greatly if the authors demonstrated more drug interactions and identified (and ultimately validated) new ones. This would certainly make MULT-i² method more attractive. In particular, it would be nice to see if one could use MULT-i² for studies of triple combinations as enthusiastically suggested.

We thank the reviewer for this valuable suggestion. We agree that demonstrating additional applications, including triple-drug combinations, would further highlight the potential of the MULT-i² assay.

The primary aim of this study was to validate the MULT-i² methodology against the established PRR v2 assay and to demonstrate that the new platform can reproduce known parasiticidal interaction profiles while providing a more scalable workflow. For this reason, we selected well-characterized drug combinations, including atovaquone/proguanil and piperaquine/pyronaridine, which provide suitable benchmark systems for comparison with previous PRR data.

Although evaluation of a larger number of novel combinations and triple-drug regimens would be highly valuable, generating corresponding PRR datasets for direct comparison was beyond the scope of the current study.

(4) Throughout the manuscript, the authors claim that MULT-i² is considerably less expensive and can be done much faster than previous methods. In my view, this is not exactly a scientific argument. The cost of an assay depends heavily on the cost of reagents and labor, which are subject to market price fluctuations. The efficiency and time consumption can very much depend on laboratory organization, etc. Unless the author could specifically demonstrate where and how these assays are cheaper and faster, I suggest not discussing this.

We thank the reviewer for this important comment. We agree that absolute assay costs can vary depending on local reagent prices, labor costs and laboratory infrastructure.

When comparing both methods under the same laboratory conditions, the total assay duration of the MULT-i² assay is shorter than that of the PRR assay (11 days (MULT-i²) compared with approximately 21–28 days (PRR) according to published protocols). In addition, the MULT-i² assay reduces labor-intensive processing steps and enables higher-throughput measurements using a plate reader for readout. These factors contribute to reduced workload and improved scalability, independent of fluctuations in individual reagent or personnel costs.