Integral Solvent-Induced Protein Precipitation for Target-Engagement Studies in Plasmodium falciparum

Abstract

The limited understanding of the mechanism of action (MoA) of several antimalarials and the rise of drug resistance toward existing malaria therapies emphasizes the need for new strategies to uncover the molecular target of compounds in Plasmodium falciparum. Integral solvent-induced protein precipitation (iSPP) is a quantitative mass spectrometry-based (LC–MS/MS) proteomics technique. The iSPP leverages the change in solvent-induced denaturation of the drug-bound protein relative to its unbound state, allowing identification of the direct drug–protein target without the need to modify the drug. Here, we demonstrate proof-of-concept of iSPP in P. falciparum (Pf) lysate. At first, we profiled the solvent-induced denaturation behavior of the Pf proteome, generating denaturation curves and determining the melting concentration (C_M) of 2712 proteins. We then assessed the extent of stabilization of three antimalarial target proteins in multiple organic solvent gradients, allowing for a rational selection of an optimal solvent gradient. Subsequently, we validated iSPP by successfully showing target-engagement of several standard antimalarials. The iSPP assay allows the testing of multiple conditions within reasonable LC–MS/MS measurement time. Furthermore, it requires a minimal amount of protein input, reducing culturing time and simplifying protein extraction. We envision that iSPP will be useful as a complementary tool for MoA studies for next-generation antimalarials.

Introduction

Malaria is a global health challenge with an estimated incidence of 249 million cases in 2022 and a mortality of 608,000.¹ Over 95% of malaria deaths occur in Africa, and 78% of these in children under 5 years.¹ Malaria is caused by parasites of the genus Plasmodium, with the highest incidence and mortality attributed to Plasmodium falciparum.¹ The asexual blood stage form of the parasite is responsible for the disease and a possible fatal outcome if left untreated.² Therefore, this stage represents the primary target of first-line antimalarials.² An active drug against all lifecycle stages would provide a more effective treatment and interruption of transmission.³ Artemisinin-based combination therapies remain the first-line treatment in most malaria-endemic regions.¹ However, the emergence of resistance to artemisinins, along with the growing resistance to its partner drugs, underscores the need for new antimalarials with new modes of action.⁴⁻⁶ Identifying the mechanism of action (MoA) of antimalarial molecules is crucial for the rational design of new inhibitors and their optimization by medicinal chemistry. Moreover, understanding the MoA guides drug combinations and supports the management of drug resistance.

Various strategies are employed to elucidate the MoA of antimalarials. These include in vitro resistance evolution and whole-genome sequencing (IVIEWGA),⁷ metabolic profiling,^8,9 chemogenomic profiling,¹⁰ chemical screens,¹¹ and transcriptomic analysis.¹² In IVIEWGA, parasites are exposed to sublethal drug pressure until resistant mutants emerge. Whole-genome sequencing of the parent and resistant strains is then conducted to identify the genetic variants associated with the resistance phenotype.^13,14 These variants often manifest as single point mutations or copy number variations in the target protein.^13,14 Notable compounds whose targets were identified or validated with this method include fosmidomycin,¹⁵ DSM265,¹⁶ MMV390048,¹⁷ and NITD609.¹⁸ IVIEWGA has been one of the most successful methods for identifying drug targets. However, resistance selection can also result in genetic changes not directly related to the drug target, such as mutations in drug efflux transporters (e.g., pfcrt gene, pfcarl gene).^19,20 This makes it challenging to discern the direct MoA of the compound.^13,14 Additionally, the generation of resistant parasites is frequently a time-consuming process and not always successful.^13,14

Proteomics and chemical proteomics have emerged as alternative or complementary approaches to IVIEWGA, mainly through quantitative mass spectrometry (LC–MS/MS) readout.²¹ Conventional chemical proteomics utilize a compound analogue that can be linked to a resin or matrix through, e.g., click chemistry for subsequent chemical pulldown.²² This often requires the addition of functional groups to the small molecule.²³ For instance, the target of MMV390048, phosphatidylinositol 4-kinase (PI4k), was identified with this method.¹⁷ However, chemically modifying the small molecule can alter its efficacy and specificity to the target.²²

Proteome-wide target–engagement approaches have emerged which do not involve modification of the drug or the target protein. Instead, they exploit the physical or chemical stabilization of proteins upon binding to their ligands.²⁴⁻³⁰ An example is the drug affinity responsive target stability method, which utilizes the change in protease degradation of drug-bound proteins relative to their unbound form.³¹ For instance, the target of Torin 2, a compound displaying low nanomolar activity on the sexual stages of P. falciparum, was identified with this method.³¹ The cellular thermal shift assay (CETSA) and thermal proteome profiling (TPP) have been at the forefront of modification-free approaches.^28,29,32 They rely on detecting changes in the melting temperature (T_M) of the protein–drug complexes compared to the unbound, free state of the same proteins. CETSA and TPP have recently been established in both cell lysates and intact cells of P. falciparum, facilitating the identification of protein targets for quinine, mefloquine, and antimalarials with previously unknown MoA.^24,25,33

Solvent-induced protein precipitation (SPP) is a recent addition to this field. It leverages the principle of denaturing proteins through organic solvents.^27,34 Zhang et al. demonstrated in human cell lysates that drug-binding proteins can be identified upon exposure to organic solvents by comparing the denaturation curves of a drug-treated versus a non-treated cell lysate.²⁷ The throughput of the standard SPP approach was then improved by Van Vranken and co-workers³⁴ by pooling the soluble fractions after denaturation across the organic solvent gradient.^34,35 The soluble proteins in each pool are then quantified by quantitative MS-based proteomics. This compressed approach yields an estimation of the area under the solvent denaturation curve, and the fold change measured in vehicle-treated lysates vs. drug-treated lysates is then calculated to determine protein stability.^34,35 Recently, Bizzarri et al. extended the technique to Gram-negative and Gram-positive bacteria.³⁶ We refer to the compressed format of SPP as integral solvent-induced protein precipitation (iSPP).

Here, we apply the iSPP assay to P. falciparum. First, we characterized the solvent-induced denaturation behavior of the Pf proteome, followed by iSPP experiments in which multiple organic solvent gradients were employed on three antimalarials. This enabled the selection of the optimal solvent gradient for iSPP target validation. To demonstrate the potential of iSPP as target deconvolution tool for Pf, we validated the approach with six antimalarials and successfully observed target-engagement of the designated target for four of them as well as identified potential secondary targets for fosmidomycin and MMV390048.

Results

iSPP Workflow in P. falciparum

The iSPP approach was conducted on cell lysates of Pf (NF54 strain) derived from asynchronous culture. Because some antimalarial compounds are stage-specific,³⁷ asynchronous cultures were necessary to ensure the presence of the presumed target proteins. First, the cells were treated with saponin, resulting in erythrocyte lysis, followed by several wash steps to get rid of excess hemoglobin. The parasite pellet was then resuspended in lysis buffer and subjected to freeze–thaw cycles to extract proteins. The lysates were incubated with drug or vehicle, and then exposed to eight increasing concentrations of acetone/ethanol/acetic acid (50:50:0.1 v/v, abbreviated as A.E.A.). These chemical reagents are common, cost-effective, and readily available in most laboratories. Organic solvents such as acetone, ethanol, methanol, and acetonitrile are frequently used to precipitate proteins and remove contaminants. Zhang and colleagues demonstrated that increasing concentrations of the A.E.A. mixture effectively induce protein precipitation, a property that can be exploited to identify ligand-binding events.²⁷ This approach has consistently proven effective for various ligand–target pairs in human cell lysates.^26,34

The iSPP protocol used in this study requires only 20 μg total protein input material per data point (each sample across the gradient), representing 0.5 mg total protein per condition (drug or vehicle, in triplicates). This renders the process more applicable to compound screening campaigns in Pf considering the relatively low protein yield after saponin lysis. After centrifugation, the soluble fractions were collected and pooled for iSPP. The resulting proteins underwent bottom-up proteomics sample preparation for LC–MS/MS with label-free quantification in data-independent acquisition mode (LFQ-DIA) (Figure 1). The readout of an iSPP assay was the calculated fold change of drug-treated samples over control samples. These data were used to generate volcano plots, plotting fold change versus p-value, which had been derived from statistical testing of the biological replicates.

Figure 1.

Schematic workflow of iSPP in Plasmodium falciparum (created with BioRender.com).

Solvent Profile of the P. falciparum Proteome

Before employing the iSPP workflow for target-engagement studies, we investigated the solvent-induced denaturation of the complete Pf proteome. Thus, we first assessed the effect of increasing A.E.A. concentrations (0 to 50% v/v) on Pf proteins. The soluble fractions were first resolved by SDS-PAGE. The SDS-PAGE readout showed that most of the proteins responded in a gradient-dependent manner with a substantial fraction of proteins precipitating after exposure to 20% of A.E.A. (v/v) (Figure S1). Subsequently, the denaturation behavior was assessed by quantitative mass spectrometry.

We quantified approximately 3400 proteins in each biological replicate (n = 2), achieving more than 50% coverage of the theoretical P. falciparum NF54 proteome (UniProt ID: UP000030673). This represents a substantial improvement over the reported proteome coverage achieved by previous MS-based TPP and CETSA in Pf, which had reported approximately 2600 proteins.^24,25,38

Each protein has a denaturation profile, which can be modeled similarly to thermal denaturation.³⁹ This profile is essentially described by the melting point, the slope, and top and bottom plateaus of the curve.³⁴ The melting concentration, C_M, represents the solvent concentration at which there is an equal distribution between the folded and the unfolded state of the protein. It provides a measure for a protein’s susceptibility toward organic solvents.³⁴ As described by Van Vranken et al.,³⁴ the assigned curves for each protein were evaluated for quality of the fitting according to two parameters: the coefficient of determination (R²) which is the measure of goodness of fit, and the bottom plateau of the curve. We then selected the curves having a high-quality fit using the following criteria: R² ≥ 0.8 and Plateau < 0.3, which allowed us to confidently assign C_M values for 2712 proteins. The C_M value of each protein is provided in Supporting Information file Table S1. This will be useful in determining appropriate gradients for further iSPP approaches with P. falciparum.

Consistent with the SDS-PAGE readout, we observed a decreasing protein abundance with increasing A.E.A. % (v/v), reaching a bottom plateau at approximately 30% (v/v) (Figure 2A). The calculated median C_M of the Pf proteome in the experiment was 15% (v/v) A.E.A. (R² = 0.99). This median value was marginally lower than the median C_M measured by SPP for Escherichia coli K12 cell using the same protocol by Bizzarri et al. (C_M = 18% (v/v) A.E.A, Figure 2B).³⁶ These findings suggest that the Pf proteome overall exhibits a similar tolerance to organic solvent-induced denaturation as to the E. coli proteome. In contrast, the distribution profiles of C_M values for Pf and E. coli exhibited a notably different pattern. The distribution in Pf displays a narrow peak, suggesting a high degree of uniformity among C_M values (Figure 2B). This indicates that a large portion of Pf proteins share a similar denaturation profile, resulting in a concentrated peak in the distribution. Notably, this distribution pattern closely resembles the one obtained by Van Vranken et al.³⁴ in their solvent profiling of the Human HCT116 proteome, as well as the T_M distribution profile of human proteome obtained by Jarzab et al.⁴⁰ in a TPP meltome atlas study. This similarity in denaturation profiles may be attributed to the fact that the parasite lives intracellularly within the human host at this particular stage. Consequently, the parasite proteins are shielded against external stressors akin to human proteins. In contrast, the distribution of C_M values of E. coli proteins is broader with a less pronounced peak, implying greater variability among proteins in terms of their solvent stability.

Figure 2.

Solvent profiling of the Plasmodium falciparum proteome. (A) Denaturation curve of the P. falciparum proteome. For each data point, the median value among all quantified proteins is shown. The fold change of the relative abundance was calculated relative to 0% A.E.A. The curve was obtained for proteins with high-quality fitting curves (2712; R² ≥ 0.8, Plateau < 0.3, C_M ≥ 4). The inset shows the calculated C_M, R², Hillslope and bottom plateau; (B) distribution profiles of high confidence C_M values calculated in P. falciparum and Escherichia coli proteomes. (C) Correlation between C_M values (y-axis) vs protein abundance (x-axis, log₁₀ intensity), displaying a weak relationship between the two variables (r = 0.26); (D) heatmap representation of all proteins quantified (n = 3492). For each protein, the relative abundance (fold-change) was calculated relative to 0% A.E.A. Protein fold-changes were then clustered.

We found a low correlation between C_M value and protein abundance (Pearson r = 0.26, Figure 2C), demonstrating that we can confidently assign C_M values to proteins regardless of their relative abundance. To confirm this, we calculated the percentage of low-abundant proteins for which we assigned high-quality C_M values. Proteins classified as low-abundance fell in the lowest 30th percentile based on their intensity. Out of the 1045 low-abundant proteins, we fit high-confidence curves and determined C_M values for 707 of them (68%, Figure S2).

To analyze the solvent denaturation of the Pf proteome, we performed a hierarchal clustering based on the relative abundance of proteins across the gradient. The gene ontology (GO) terms on the upper and bottom clusters were then analyzed using DAVID.^41,42 Hierarchal clustering and GO analysis revealed a group of proteins that displayed a high resistance to solvent-based precipitation (Figures 2D and S3). This cluster consists of proteins involved in proteasomal protein degradation such as the proteasome complex subunits (median C_M = 32.8%), as well as proteins involved in protein folding such as Hsp70/Hsp90 organizing protein, peptidyl-prolyl cis–trans isomerase, and prefoldin subunits. The cluster also contained a group of proteins functioning as surface-exposed antigens. For instance, the 6-cysteine protein family, which is the most abundant surface antigen present throughout all stages of Plasmodium,⁴³ has 6 proteins in this group (median C_M = 24.79%). In addition, proteins involved in the invasion of host cells are found in this cluster, such as the ring-infected surface antigen (RESA) N-terminal domain-containing protein family, which is exported to the host erythrocyte.⁴⁴ Moreover, the merozoite surface proteins MSP3 and MSP6, a potential vaccine development candidate⁴⁵ and chaperones such as heat shock proteins (Hsp110) and histone chaperones were also identified within this group.

GO analysis of protein clusters displaying low tolerance toward A.E.A. revealed 58 proteins belonging to ribosomal protein subunits (median C_M = 9.99%). Moreover, they contained proteins involved in ribosome biogenesis such as the nucleolar GTP-binding protein 1 (C_M = 9.7%), RNA cytidine acetyltransferase (C_M = 8.3%), the ribosome biogenesis proteins TSR1 (C_M = 9.39%), BOP1 homologue (C_M = 10.24%), and the ATP-dependent RNA helicase family (median C_M = 8.98%).

Selection of Gradients for iSPP Experiments in P. falciparum

After assessing the behavior of the Pf proteome when exposed to organic solvents, we examined the capability of iSPP to validate the known protein targets of the standard antimalarials pyrimethamine, DSM265, and fosmidomycin. To determine the optimal gradient for target-engagement studies, we evaluated the stabilization extent of the known targets by employing five different A.E.A. range windows. Previous studies in iSPP³⁴ and the compressed TPP format⁴⁶ have highlighted the importance of selecting the suitable solvent/thermal gradient and its impact on the observed stabilization (log₂ fold-change, log₂FC) for drug-target proteins.

Pf cell lysate was incubated with either the vehicle control (DMSO or ddH₂O, n = 3) or 50 μM of the drugs (n = 3). To save Pf input material, we performed a mixed drug approach by incubating the lysate with the three selected standard antimalarials as single condition.

Pyrimethamine is a well-known antimalarial and inhibitor of the bifunctional dihydrofolate reductase-thymidylate synthase (PfDHFR).^25,47,48 Multiple in vitro studies, including CETSA assay, and clinical studies support that PfDHFR inhibition is the cause of death of the parasite during pyrimethamine treatment.^25,47,48 DSM265 is a compound derived from a target-based high throughput screening of novel dihydroorotate dehydrogenase (PfDHODH) inhibitors,^49,50 which exhibits a low nanomolar activity against PfDHODH.⁵¹ Lastly, fosmidomycin is an inhibitor of the 1-deoxy-d-xylulose 5-phosphate reductoisomerase (PfDXR),⁵² which had been further validated through IVIEWGA¹⁵ and metabolic profiling.⁵³

After drug or vehicle incubation, the samples were exposed to the A.E.A. gradient with eight increasing concentrations, and the resulting soluble fractions were pooled. We selected gradients encompassing proteins with low C_M values relative to the median C_M (11–25%) and moderate-to-high C_M values (14–28%, 16–26.5% and 17–31%), also accounting for narrow and wide increments (i = 1.5%, i = 2%, i = 3%) within the gradient. Figure 3A illustrates the schematic diagram of the solvent gradients employed, and Table 1 shows the calculated C_M for each expected target. In the following iSPP data analysis, we applied a log₂FC > 0.3 and a p-value < 0.05 as inclusion criteria.

Figure 3.

Solvent gradient has an impact on the extent of stabilization for each expected target (A) schematic figure of the iSPP solvent gradients used and the different increments (i) 1.5% A.E.A, 2% A.E.A, and 3% A.E.A. (v/v) (B) log₂FC for the expected targets across solvent gradient used. Red inline displays the threshold log₂FC = 0.3.

Table 1. Calculated C_M of the Expected Drug Targets.

compound	designated target	CM (%A.E.A.)
fosmidomycin	PfDXRa	20.1
pyrimethamine	PfDHFRb	20.1
DSM265	PfDHODHc	21.4

^a1-deoxy-D-xylulose5-phosphate reductoisomerase.

^bBifunctional dihydrofolate reductase-thymidylate synthase.

^cDihydroorotate dehydrogenase.

The iSPP approach was able to quantify approximately 3000 proteins (n = 3) for all the conditions tested with a coefficient of variation < 8% between replicates (Figure S4A). We observed an inverse correlation between the %A.E.A. gradient and the number of proteins detected. Higher %A.E.A. gradients corresponded to lower numbers of identified proteins, as expected by a more pronounced precipitation. For instance, we detected 353 proteins using the gradient 11–25% A.E.A (n = 3414) which were not detected in the gradient 17–31% A.E.A. (n = 3061) (Figure S5).

The iSPP approach revealed the stabilization of PfDHFR and PfDHODH in at least one of the A.E.A. gradients (Figure 3B). PfDXR did not result in a significant stabilization or destabilization in any of the gradients employed. Notably, aspartate tRNA ligase (PfAspRS) was identified among the top hits, as already observed in the iSPP study by Bizzarri et al. in E. coli, Klebsiella pneumonia (Kp), and Pseudomonas aeruginosa (Pa) cell lysates upon fosmidomycin incubation.³⁶ Therefore, AspRS was also considered as a secondary target of fosmidomycin in the current iSPP approach in Pf. We observed a more pronounced stabilization for most of the designated target proteins when using the higher A.E.A. (v/v) concentration gradients 14–28% and 17–31% (Figure 3B). This is consistent with their denaturation profiles, as they demonstrated high tolerance to chemical denaturation, with C_M values greater than the median C_M of the Pf proteome.

There was a larger number of stabilized proteins when utilizing the broadest concentration increment (12–33%, i = 3%, Figure 3B). In this scenario, a substantial number of proteins unrelated to the employed drugs were stabilized (n = 154, Figure 4B). Conversely, the narrowest concentration increment (16–26.5%, i = 1.5%) resulted in a compressed magnitude of change in protein stability, with only a few proteins exhibiting stabilization (n = 15, Figure 4D).

Figure 4.

The solvent gradient 14–28% stabilized the majority of the expected protein targets with the least nonspecific proteins (de)stabilized. P. falciparum lysates were incubated with vehicle control or 50 μM of pyrimethamine, DSM265, and fosmidomycin (mixed drug approach). Lysates were then exposed to either of the five % A.E.A. gradient v/v: (A) 11–25%, (B) 12–33%, the broadest gradient which resulted in the most (de)stabilization of unexpected proteins, (C) 14–28%, (D) 16–26.5%, the narrowest increment and (E) 17–31%. Data is shown as volcano plots where the threshold criteria for identification of proteins exhibiting statistically significant changes in response to the compound treatment was set to a log₂ fold change (log₂FC) > |0.3| and p-value < 0.05. Red shows stabilized proteins with log₂FC > 0.3 and p-value < 0.05; Blue shows destabilized proteins with log₂FC < −0.3 and p-value < 0.05; gray shows proteins with −0.3 < log₂FC < 0.3 and p-value < 0.05 and proteins with p-value > 0.05.

This result guided the rational selection of the solvent gradient for target validation studies in the succeeding experiments. The 14–28% A.E.A. (v/v) gradient successfully identified two out of three designated protein targets as stabilized proteins in addition to the secondary target of fosmidomycin, AspRS. This gradient maintained a relatively low number of nonspecific targets (de)stabilized (Figure 4C). Thus, it was selected as the optimal gradient for subsequent experiments. Moreover, this gradient encompasses the median C_M value of the Pf proteome, making it the most suitable choice for target validation studies of the majority of Pf proteins.

iSPP Validation for Target-Engagement Studies in P. falciparum

We proceeded to verify the target validation of fosmidomycin, pyrimethamine, and DSM265 using iSPP by individually incubating each drug with Pf lysate. We also included the three structurally diverse antimalarials MMV390048 (inhibiting PI4k¹⁷), sulfadoxine (targeting dihydropteroate synthase⁵⁴), and mefloquine (likely targeting the cytoplasmic ribosomes^25,55). The drugs were tested at 100 μM, to ensure maximum target occupancy.

Consistent with our previous data, we detected approximately 3200 proteins for each condition. Pyrimethamine stabilized its known target, PfDHFR (Figure 5A; log₂FC = 0.56). The compound DSM265 stabilized PfDHODH (log₂FC = 0.31; Figure 5B). Interestingly, we also detected a similar level of stabilization of transcription elements such as RNA polymerase Rpb4/RPC9 core domain-containing protein (log₂FC = 0.62; Table S2) and transcription initiation factor subunit 10 (log₂FC = 0.37; Table S2). Fosmidomycin did not lead to the stabilization of PfDXR but resulted in the identification of PfAspRS among the top hits (Figure 5C). For MMV390048, we observed a significant stabilization for its protein target PfPI4k¹⁷ (p-value <0.05, log₂FC = 0.86, Figure 5D). Interestingly, we also observed a stabilization of CDP-diacylglycerol-inositol 3-phosphatidyltransferase (PfPIS), an enzyme also implicated in inositol phosphate metabolism (p-value < 0.05, log₂FC = 0.54) (KEGG pfa00562, 2.7.8.11). Moreover, we detected significant stabilization for proteins involved in protein trafficking and signal transduction such as the signal peptidase complex subunit SPC1 (log₂FC = 0.53), protein phosphatase inhibitor 2 (log₂FC = 0.53), osmiophilic body protein (log₂FC = 1.2), and EMP1-trafficking protein (log₂FC = 0.37). As for sulfadoxine, its validated target, dihydropteroate synthase⁵⁴ (PfDHPS), was significantly stabilized (p-value < 0.05, log₂FC = 0.31, Figure 5E). However, it is notable that several other proteins were stabilized similarly (n = 43). Of note are several ribosomal protein subunits: 60 ribosomal protein L7 (log₂FC = 0.58), 60S ribosomal protein L22 (log₂FC = 0.39), 40S ribosomal protein S2 (log₂FC = 0.37), 60S ribosomal protein L34 (log₂FC = 0.34) and 60S ribosomal protein L7a (log₂FC = 0.33). For mefloquine, it was previously suggested that the compound inhibits protein translation by interacting with the 80S ribosome.⁵⁵ In our iSPP experiments, we detected a destabilization of two ribosomal subunits: 60S ribosomal protein L18a (Pf60RPL18a, p-value <0.05 and log₂FC = −0.44) and ribosomal protein L28 (PfRPL28, p-value <0.05, log₂FC = −0.66) (Figure 5F). However, it should be noted that additional proteins were stabilized and destabilized (n = 40 and n = 49, respectively).

Figure 5.

The iSPP assay in Plasmodium falciparum was able to identify the majority of the expected targets of the standard antimalarials. P. falciparum lysates were incubated with vehicle control or 100 μM of the compounds. The lysates were then exposed to 14–28% A.E.A. gradient (v/v). The threshold criteria for identification of proteins exhibiting statistically significant changes in response to the compound treatment was set to a log₂ fold change (log₂FC) > |0.3| and p-value < 0.05. Data is shown as volcano plots that highlights changes in abundance vs statistical significance for treatment with compounds (A) pyrimethamine (B) DSM265 (C) fosmidomycin (D) MMV390048, (E) sulfadoxine, and (F) Mefloquine. Red shows stabilized proteins with log₂FC > 0.3 and p-value <0.05; Blue shows destabilized proteins with log₂FC < −0.3 and p-value < 0.05; gray shows proteins with −0.3 < log₂FC < 0.3 and p-value < 0.05 and proteins with p-value > 0.05.

Discussion

Elucidating the MoA is an important aspect in antimalarial drug discovery and development. The iSPP assay represents a modification-free approach that can be used as a complementary tool for existing target validation studies. The iSPP principle relies on detecting changes in the solvent denaturation behavior of protein targets upon binding with the compound of interest. In this work, we have adapted iSPP to P. falciparum lysate and validated its potential for target-engagement. We used six chemically diverse antimalarials with validated protein targets. These proteins have demonstrated druggability and their inhibitors often exhibited activity against multiple Plasmodium life-cycle stages. Furthermore, there has been a rise of novel chemotypes with inhibitory effects on these targets. For instance, several PfDHODH and PfDHFR inhibitors have progressed into the drug development pipeline.⁵⁶ Our iSPP assay identified PfDHODH and PfDHFR as main stabilized proteins by DSM265 and pyrimethamine, respectively (Figures 4–5). DHFR has previously been identified as main target of methotrexate in Gram-negative bacteria through the iSPP approach.³⁶ With isothermal dose response (ITDR)-CETSA approach, PfDHFR was also identified as the main target in pyrimethamine-treated Pf lysates, demonstrating a dose-dependent stabilization.²⁵

The list of proteins (de)stabilized by each compound treatment provides insights into potential off-targets or cytotoxicity associated with these antimalarials (Supporting Information file Table S2). For instance, the consistent stabilization of AspRS observed in iSPP experiments in lysates across multiple organisms could be an indication of an additional target for fosmidomycin. The AspRS belongs to the amino-acyl-tRNA synthetases (AaRS) family, with proven druggability and relevant protein targets across organisms.⁵⁷ Moreover, inhibitors of AaRS in Plasmodium are also active in other pathogens.⁵⁸ The AaRS family catalyzes the attachment of amino acids to their cognate tRNAs to produce the aminoacyl tRNAs with the corresponding amino acid code thereby playing a vital role in preventing translational errors.⁵⁹ The enzymes have also multiple binding sites for small inhibitors.⁵⁷ However, additional experiments are needed to validate the binding of fosmidomycin to aspartate tRNA ligase in vitro. Similarly, iSPP showed stabilization of PfPIS in MMV390048-treated lysates, besides the main target PfPI4k.¹⁷ Both enzymes are involved in signal transduction and synthesis of phosphatidylinositol. PfPI4k is involved in the synthesis of phosphatidylinositol 4-phosphate⁶⁰ whereas PfPIS catalyzes the reaction of CDP-diacylglycerol and myo-inositol.⁶¹ PfPIS stabilization can be a result of MMV390048 direct binding, considering the shared similarity of its catalytic site with PfPI4k, given that 1,2-diacyl-sn-glycero-3-phospho-(1D-myo-inositol) is a substrate of PfPI4k and a product of PfPIS (KEGG pfa00562, 2.7.8.11).

As for mefloquine, we detected the destabilization of two ribosomal protein subunits out of the 40 proteins destabilized. For years, the target of mefloquine has been debated. However, it has been suggested that mefloquine is a putative inhibitor of the cytoplasmic ribosome.⁵⁵ Furthermore, ITDR-CETSA in lysate showed the stabilization of four ribosomal subunits upon mefloquine treatment in Pf lysates. However, stronger stabilization was observed for other proteins.²⁵ It was not the first time that a destabilization occurred for inhibitors of protein–protein interactions. Zhang et al. report a destabilization of proteins interacting with Hsp90 as a result of geldanamycin treatment in human cell lysates.²⁷ Therefore, the destabilization of the ribosomal subunits in our iSPP experiments could be explained by mefloquine inhibition of ribosomal protein–protein interactions.

The C_M values obtained from the solvent denaturation profile could be valuable for conducting a targeted iSPP profiling of a potential target especially in accommodating for proteins e.g., with extremely high or low C_M values by selecting an appropriate gradient. The extent of drug-induced protein stabilization in iSPP is influenced by the solvent gradient as demonstrated by the extent of stabilization observed for each target protein (Figure 3B). The proteasome complex, which is very stable against solvent-induced denaturation, represents an extreme case (median C_M = 32.8%). Proteasomes are large protein complexes that play a role intracellular protein turnover.⁶² Proteasomes have proven druggability and several inhibitors are active not only in Plasmodium but also in Trypanosomes,⁶³ Leishmania⁶³ and cancer cells.⁶⁴ In the case of a proteasome inhibitor profiled by iSPP, a higher solvent gradient is recommended due to their high resistance to solvent-induced precipitation.

Similar to other methods for measuring target-engagement, the iSPP approach has limitations. For instance, along with the designated drug targets, multiple proteins are displayed as (de)stabilized. This effect can be attributed to their low signal-to-noise ratio, which results in false-positives. Moreover, since the iSPP approach does not generate full protein denaturation curves, it inevitably leads to false-negatives when proteins have extreme C_M values, and are thus not represented within the selected profiling window of solvent concentrations. Finally, there might be protein-specific reasons why target–engagement might not be detectable. For PfDXR, for instance, we did not observe notable stabilization in the presence of fosmidomycin, although PfDXR was detected and abundant in all replicates and had a calculated C_M that fell within the tested gradients. One explanation could be that it might have been partially unfolded during protein extraction given that it resides within the four membrane-bound organelle–the apicoplast.⁵²

In this work, we used an asynchronous culture of Pf. However, iSPP can be extended to synchronized cultures to accommodate for compounds with stage-specific activities and stage-specific target expression. For instance, the antifolates pyrimethamine and sulfadoxine have a higher activity against late trophozoites and early schizonts.^65,66 Moreover, synchronized P. falciparum at the midtrophozoite stage are the most metabolically active and might be optimal for target deconvolution studies.⁶⁷ The iSPP in Pf was designed to use a low input protein amount, adapting to the protein yield in Pf. This adaptation simplifies the protein extraction workflow as enough lysates are produced without the need for magnetic enrichment or separation of infected erythrocytes. This can be beneficial, especially in extending iSPP for profiling compounds targeting the gametocytes (the transmission stages), or the liver stages which are more challenging to culture.^68,69 We highlighted the potential of iSPP for target-engagement for compounds with known protein targets. However, iSPP could also be extended for target deconvolution studies of compounds with unknown MoA, e.g., from target-based discovery projects and phenotypic-based drug screening attempts. In that case, we would recommend measures to increase the likelihood of success and the specificity of the approach: (i) Structurally related negative control compounds to filter out artifacts. (ii) The use of multiple solvent gradients to enhance the likelihood of success by broadening the accessible C_M range. (iii) Concentration-dependent iSPP profiling to allow differentiating targets based on the concentration-dependent saturation of drug binding. These control measures are also recommended in validating a possible off-target for the standard antimalarials.

Conclusion

In our study, the iSPP assay was adapted and extended to P. falciparum. The minimal protein input amount required from our protocol accounts for the low protein yield streamlining the protein extraction process. Thus, permitting iSPP to be extended to stages that are more labor-intensive to cultivate.

We validated iSPP in P. falciparum through the identification of the expected drug targets and potential secondary targets of six standard antimalarials. The compounds used in this study and their corresponding expected targets are pyrimethamine (PfDHFR), DSM265 (PfDHODH), fosmidomycin (PfDXR and PfAspRS), MMV390048 (PfPI4k), sulfadoxine (PfDHPS), and mefloquine (cytoplasmic ribosomal subunits). We provide a list of stabilized and destabilized proteins from our iSPP workflow that could be useful for potential exploration of off-targets or cytotoxicity studies of these compounds or their chemical analogues. Moreover, the high-confidence C_M value list for 2712 proteins that we provide will be useful for profiling potential drug targets by iSPP. The log₂FC observed for protein targets are influenced by the solvent gradient used, and we recommend the gradient 14–28% A.E.A. v/v as an initial iSPP profiling because it accommodates for the C_M of the majority of the Pf proteome. However, the C_M values of each protein provided in the Supporting Information Table S1 will be useful in determining an appropriate gradient for targeted iSPP profiling.

We envision that the iSPP would play an important role as a complementary and/or alternative tool for MoA studies in the development of potential antimalarials.

Methods

The compounds used in this study were purchased from the following: Pyrimethamine (MedChemExpress, HY-18062), Fosmidomycin sodium salt (MedChemExpress, HY-112853), DSM265 (MedChemExpress, HY-100184), MMV390048 (MedChemExpress, HY-106005), Sulfadoxine (Sigma, S7821), and Mefloquine (Sigma, PHR1705).

P. falciparum Culture

P. falciparum asynchronous NF54 wild-type strain parasites were cultured at 3% hematocrit until 3–5% parasitemia was reached. RPMI 1640 was used as a culture media supplemented with 25 mM HEPES, 0.36 mM hypoxanthine, 24 mM sodium bicarbonate (pH 7.3), 0.5% Albumax II and 100 μg/mL neomycin as described in ref (70). The cultures were maintained at 37 °C with the mixed gas containing 3% O₂, 4% CO₂, and 93% N₂ and were kept in an incubator under atmospheric pressure.

P. falciparum Saponin Lysis

Saponin lysis was performed to lyse the erythrocytes. Briefly, the cultures were transferred to 50 mL falcon tubes and centrifuged at 1200 g for 5 min at room temperature (RT). The supernatant was then discarded. Then, 10 volumes of 0.1% (w/v) saponin (Calbiochem 558,255, dissolved in PBS, and filtered with 0,22 μM) were added and incubated on ice for 10 min. During the incubation step, the falcon tubes were shaken every 60 s. A centrifugation step was done at 4000 g for 15 min at 4 °C, which resulted in a dark red supernatant corresponding to the lysed erythrocytes and a brown pellet corresponding to the intact parasite. A washing step was done with ice-cold PBS and centrifuged at 4000g for 5 min at 4 °C. This was repeated until the supernatant was transparent. The parasite pellet was redissolved in 1 mL ice-cold PBS and transferred to a 1.5 mL Eppendorf tube which was stored at −80 °C or proceeded for protein extraction.

P. falciparum Protein Extraction

The parasite pellet was centrifuged at 4 °C for 20,000g for 10 min. Once the supernatant was aspirated, 2.5 volumes of the ice-cold lysis buffer was added to the parasite pellet. The lysis buffer contained 50 mM Tris/HCl pH 7.5 (Avantor), 5% glycerol (Sigma-Aldrich), 150 mM NaCl (Carl Roth), 1.5 mM MgCl2 (Sigma-Aldrich), 1 mM DTT (Carl Roth), 0.8% IGEPAL CA-630 (Sigma-Aldrich), 1X Halt Protease and Phosphatase Inhibitor-Cocktails EDTA-free (Thermo Fisher Scientific). The lysate was then subjected to three cycles of flash freeze–thawing using liquid nitrogen and ddH₂O at RT. The lysate was centrifuged at 20,000g for 20 min at 4 °C. The supernatant, which contains the soluble protein fraction, was collected and transferred to a new 1.5 mL Eppendorf tube. The protein content was quantified using the BCA assay (Pierce BCA Protein Assay Kit, Thermo Fisher Scientific). The lysate was stored at −80 °C until further use.

Solvent Proteome Profiling for SDS-PAGE Readout

The P. falciparum lysate was thawed on ice and diluted to 0.8 mg/mL. The solvent proteome profiling workflow was performed as described.³⁶ Briefly, the lysate was transferred to a 96-well plate distributing 20 μg protein per well. Then, the samples were exposed to increasing acetone/ethanol/acetic acid (A.E.A.) concentration from 0 to 50% (v/v) using 12 aliquots (0, 8, 11, 14, 17, 20, 23, 26, 32, 35, 40, and 50%). This step was performed on the Bravo Automated Liquid Handling Platform (Agilent) in a final reaction volume of 50 μL. The samples were incubated at 37 °C and mixed at 750 rpm for 20 min (ThermoMixer C, Eppendorf). Centrifugation was then performed at 4402 g for 35 min (centrifuge 5920R, Eppendorf) to remove precipitated proteins. Upon supernatant collection, the soluble fractions were dried using a concentrator plus (Eppendorf). The samples were resuspended to 1.0 mg/mL final protein concentration with 2x NuPAGE LDS Sample Buffer (Thermo Fischer Scientific) which contained 25 mM DTT. Proteins were then resolved on NuPAGE 4–12% Bis-Tris Protein Gels (Thermo Fischer Scientific). The gels were stained using Coomassie-staining solution (ROTI Blue, Carl Roth) and the image was acquired with ChemiDoc MP Imaging System (BIO-RAD).

Solvent Proteome Profiling for LC–MS/MS Readout

The P. falciparum cell lysate was thawed on ice and distributed in a 96-well plate (Greiner Microplate, 96-well plate, V-Bottom) with 20 μg total protein per well as described.³⁶ The lysates were exposed to increasing A.E.A. concentrations from 0 to 40% (v/v) in 12 solvent concentrations: 0, 5, 8, 11, 14, 17, 20, 23, 26, 29, 35, and 40% in a final reaction volume of 50 μL using the Bravo Automated Liquid Handling Platform. The samples were incubated at 37 °C and mixed at 750 rpm for 20 min (ThermoMixer, Eppendorf). Centrifugation was then performed at 4402 g for 35 min. The soluble fractions were collected (Bravo Automated Liquid Handling Platform) and prepared for LC–MS/MS analysis.

iSPP Assay

P. falciparum cell lysate was thawed on ice and distributed in aliquots. Each aliquot was incubated with either vehicle control (DMSO or ddH₂O) or compound (dissolved in DMSO or ddH₂O) (50 μM or 100 μM). Next, the samples were incubated at RT for 30 min with a rotating mixer. Once the samples were distributed into eight wells, the samples were treated with an increasing A.E.A. (v/v) concentration depending on the gradient used: from 11 to 25% (11, 13, 15, 17, 19, 21, 23, and 25%), 12 to 33% (12, 15, 18, 21, 24, 27, 30, and 33%), 14 to 28% (14, 16, 18, 20, 22, 24, 26, and 28%), 16 to 26.5% (16, 17.5, 19, 20.5, 22, 23.5, 25, and 26.5%), or 17 to 31% (17, 19, 21, 23, 25, 27, 29, and 31%). After an incubation time (37 °C at 750 rpm for 20 min), centrifugation was done (4400g for 35 min) and precipitated proteins were removed. The soluble fractions were pooled in equal volumes into a single sample (Bravo Automated Liquid Handling Platform). Pooled samples were then prepared for LC–MS/MS analysis.

Sample Preparation for LC–MS/MS Analysis

The soluble fractions were dried down and then resuspended to a final protein concentration of 1.0 mg/mL with 5% SDS which contains 50 mM TEAB pH 7.5. The samples were then diluted to a final protein concentration of 0.4 mg/mL with 10% SDS buffer (1:1). Then, 10 mM DTT was added to all samples to reduce disulfide bonds and incubated at 35 °C for 30 min at 700 rpm (ThermoMixer). Chloroacetamide (55 mM, CAA, Merck) was added for protein alkylation and incubated for 30 min at RT in the dark. Samples were acidified with phosphoric acid to a final concentration of 2.5% and then diluted 7-fold with 90% methanol in 100 mM TEAB pH 7.5. The samples were transferred onto an S-trap column (ProtiFi) and subjected to five wash cycles using the same buffer. The Sequencing grade Modified Trypsin (Promega) in TEAB pH 8.5 was added to the S-trap column at a ratio of 1:10 (trypsin/protein), and the digestion reaction was carried out overnight at 37 °C. Peptides were eluted with 50 mM TEAB pH 8.5, 0.1% formic acid (FA, Th. Geyer), then 50/50 acetonitrile (ACN, Sigma-Aldrich)/water with 0.1% FA. Once the samples were dried down, peptides were resuspended with 0.5% FA. The peptides underwent desalting on the Bravo Automated Liquid Handling Platform using C18 cartridges (5 μL bed volume, Agilent) by using the standard AssayMAP peptide cleanup v2.0 protocol. First, the C18 cartridges were primed with 100 μL of 50/50 ACN/water with 0.1% FA and then equilibrated with 50 μL of 0.1% FA at a flow rate of 10 μL/min. Next, the samples were loaded at 5 μL/min, followed by an internal cartridge wash with 0.1% FA at a flow rate of 10 μL/min. The peptides were eluted using 50 μL of 60/40 ACN/water with 0.1% FA at a flow rate of 5 μL/min. The eluted samples were then dried and stored at −80 °C until further use.

Liquid Chromatography and Mass Spectrometry Data Acquisition

The samples were solubilized in 0.1% FA and injected in a volume equating to 1 μg to an Dionex UltiMate 3000 nano System (Thermo Fisher Scientific) coupled online to a Q Exactive Plus (Thermo Fisher Scientific) equipped with an Orbitrap mass analyzer. The peptides were delivered to a trap column (75 μm × 2 cm, packed in-house with ReproSil-Pur 120 ODS-3 resin, Dr. Maisch). The samples were separated on an analytical column (75 μm × 55 cm) packed in-house with Reprosil-Gold 120C18, 3 μm resin, Dr. Maisch. The flow rate was set at 300 nL/min using a 100 min gradient, ranging from 2–32% solvent B (0.1% FA, 5% DMSO in acetonitrile) in solvent A (0.1% FA, 5% DMSO in HPLC grade water) wherein the column oven temperature was set at 50 °C. The QE plus instrument was operated in DIA, in positive ionization mode. Full scan spectra (m/z 400–1000) were acquired in centroid mode at an Orbitrap resolution of 70,000, an AGC target set to 3 × 10⁶, a maximum injection time of 20 ms. Subsequently, DIA scans were collected utilizing 30 windows, with a 1 Da window overlap. HCD collision was set to 27%, loop count of 30, Orbitrap resolution of 35,000, AGC target set to 3 × 10⁶, and a maximum injection time was set to automatic.

Peptide and Protein Identification and Quantification

The raw LFQ-DIA files were processed with DIA-NN (v. 18.1). Analysis was conducted in library-free mode, utilizing the UniProt FASTA file for P. falciparum NF54 (taxon identifier: 5843); canonical version, not older than five months prior to MS measurements. Raw files were digested with Trypsin/P enzyme specificity, allowing for a maximum of two missed cleavages. Peptide length was constrained within 7 to 30 peptides, and the precursor m/z range was set from 300 to 1800. Cysteine carbamidomethylation was set as a fixed modification, while variable modifications included methionine oxidation and N-terminal acetylation. The maximum number of variable modifications was set to three, and ‘match between runs’ functionality was enabled. All other parameters remained at default settings, including the precursor FDR set at 1%. Cross-run normalization (RT-dependent) was enabled for raw files of iSPP experiments.

Curve Fitting and C_M Value Calculation for Solvent Proteome Profiling

The raw LFQ-DIA values of each replicate were normalized to the median abundance and expressed as a ratio to the lowest A.E.A. concentration sample (0%) with Excel. The sigmoidal denaturation curves were generated using a nonlinear regression model with GraphPad prism (v. 8.3.0) which calculated the C_M value for all unique protein IDs. High-quality denaturation curves were filtered based on the following criteria in RStudio (v. 4.3.2) with the dplyr package (v. 1.1.4):⁷¹ (i) curves must reach a bottom plateau of ≤0.3; (ii) coefficient of determination (R²) must be ≥ 0.8; (iii) a valid slope. The distribution profile of P. falciparum NF54 C_M values and E. coli K12 C_M values were plotted in RStudio using the ggplot2 package (v. 3.5.0).⁷²

Heatmap Generation and GO Analysis

The previously normalized LFQ-DIA values were used to cluster the quantified proteins. This was performed in Perseus (v2.0.10.0) by using complete linkage with Euclidean distance and number of clusters set to 3 without any constraints. The web tool Database for Annotation, Visualization and Integrated Discovery (DAVID) (https://davidbioinformatics.nih.gov/) was used to analyze the GO terms in each cluster.⁴²

iSPP Data Analysis

The raw LFQ-DIA intensity values of biological replicates in all conditions were normalized to the median abundance (Excel). Then, the normalized values were log₂ transformed in Perseus (v2.0.10.0). Missing values were imputed from a normal distribution (width 0.3, down shift 1.5) and p-values obtained after a two-sample t-test over replicates with a permutation-based false discovery rate correction (FDR 0.05). The volcano plots were generated in RStudio using the EnhancedVolcano package (v. 1.20.0)⁷³ plotting the proteins by statistical significance where y-axis is presented as – Log₁₀p-value vs magnitude of change where x-axis is presented as log₂ fold change of the protein intensities for each compound condition over vehicle control.

Glossary

Abbreviations

MoA

mechanism of action

iSPP

integral solvent-induced protein precipitation

SPP

solvent-induced protein precipitation

LC-MS/MS

liquid chromatography–mass spectrometry/mass spectrometry

ACTs

artemisinin-based combination therapies

IVIEWGA

in vitro resistance evolution and whole-genome sequencing

pfmdr1

plasmodium falciparum multidrug resistance 1

pfcarl

plasmodium falciparum cyclic amine resistance locus

PI4k

phosphatidylinositol 4-kinase

DARTS

drug affinity responsive target stability

TPP

thermal proteome profiling

CETSA

cellular thermal shift assay

A.E.A.

acetone/ethanol/acetic acid

LFQ-DIA

label-free quantification in data-independent acquisition

SDS-PAGE

sodium dodecyl sulfate–polyacrylamide gel electrophoresis

Pf

plasmodium falciparum

C_M

melting concentration

T_M

melting temperature

R²

coefficient of determination

E. coli

escherichia coli

DAVID

the database for annotation, visualization and integrated discovery

GO

gene ontology

Hsp

heat shock protein

RESA

ring-infected surface antigen

MSP

merozoite surface protein

GTP

guanosine triphosphate

RNA

ribonucleic acid

ATP

adenosine triphosphate

PfDHFR

bifunctional dihydrofolate reductase-thymidylate synthase

PfDHODH

dihydroorotate dehydrogenase

PfDXR

1-deoxy-d-xylulose 5-phosphate reductoisomerase

PfAspRS

Aspartate tRNA ligase

CV

coefficient of variation

log₂FC

log₂ fold change

Kp

Klebsiella pneumonia

Pa

Pseudomonas aeruginosa

PfPIS

CDP-diacylglycerol--inositol 3-phosphatidyltransferase

Pf60RPL18a

60S ribosomal protein L18a

PfRPL28

ribosomal protein L28

(ITDR)-CETSA

(isothermal dose response)- cellular thermal shift assay

AaRS

amino-acyl-tRNA synthetases

tRNA

transfer ribonucleic acid

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsinfecdis.4c00418.

• The Supporting Information contains the solvent proteome profiling of the P. falciparum (1–3); quality control of the iSPP experiments (4–5); Denaturation curves for the expected target proteins of the antimalarials tested (6) (PDF)

• This contains the list of all the stabilized and destabilized proteins in the iSPP experiments (1); The high-confidence C_M values for the P. falciparum proteome (2) (XLSX)

Author Contributions

^‡ P.B. and L.B. contributed equally.

This project received funding from the European Union’s Horizon 2020 Research and Innovation Program under Marie Skłodowska-Curie Grant Agreement 860816. P.B. was also funded by the Emilia-Guggenheim-Schnurr Foundation of the NGiB.

The authors declare the following competing financial interest(s): H.H. is a co-founder and shareholder of OmicScouts GmbH, a proteomics and chemical company.