Disruption of P. falciparum amino acid transporter elevates intracellular proline and induces resistance to Prolyl-tRNA synthetase inhibitors

SUMMARY

Plasmodium falciparum evades the antimalarial activity of proline-competitive prolyl-tRNA synthetase (PfProRS) inhibitors, such as halofuginone (HFG), by a resistance mechanism termed the adaptive proline response (APR). The APR is characterized by a marked elevation of intracellular proline following drug exposure. Contrary to initial expectations, the APR is not mediated by alterations in canonical proline metabolic pathways involving arginase (P. falciparum arginase [PfARG]) and ornithine aminotransferase (P. falciparum ornithine aminotransferase [PfOAT]). Instead, we identified loss-of-function mutations in the apicomplexan amino acid transporter 2 (P. falciparum apicomplexan amino acid transporter 2 [PfApiAT2]) as the primary genetic driver of this resistance phenotype. Importantly, reversion of these mutations to wild type effectively suppresses the APR, establishing PfApiAT2 as the molecular determinant of this resistance mechanism.

The elucidation of the APR significantly advances our understanding of antimalarial drug resistance. By delineating the role of PfApiAT2 in this process, we establish critical insights for the development of strategies to circumvent PfProRS inhibitor resistance for future antimalarial therapies.

In brief

Bopp et al. demonstrate that malaria parasites can acquire resistance to prolyl-tRNA synthetase inhibitors through modulation of proline homeostasis. This resistance mechanism occurs via the loss of function of the putative amino acid transporter PfApiAT2, resulting in elevated levels of intracellular proline.

Graphical abstract

INTRODUCTION

There were nearly 249 million malaria cases with over 608,000 deaths in 2022, mostly in children under 5 years of age.¹ Small-molecule therapeutics remain an essential component for global malaria control efforts. Widespread resistance to mainstay drugs has heightened the malaria threat, and the need for novel antimalarials is recognized as one of the world’s most pressing public health challenges.

Protein translation is a validated target for anti-infective compounds in a wide range of pathogenic microbes that are dependent on efficient translation for fast-growing cells.^2,3 More recently, aminoacyl-tRNA synthetases (aaRSs) have emerged as promising targets for malaria drug development.^4,5 aaRSs comprise a family of enzymes that exist in all domains of life. In their canonical function, aaRSs catalyze the conjugation of proteinogenic amino acids to their cognate tRNAs, generating the building blocks for protein synthesis. Various aaRS isoforms, including alanyl (AlaRS⁶), lysyl (LysRS⁷), isoleucyl (IleRS⁸), phenylalanyl (PheRS^9,10), asparaginyl (AsnRS¹¹), tyrosyl (TyRS¹²), and threonyl (ThrRS^6,13) isoforms, have already been explored in Plasmodium falciparum as potential targets for drug development.

Our group previously identified the P. falciparum cytoplasmic prolyl-tRNA synthetase (PfcProRS, PF3D7_1213800) and the human ortholog HsGluProRS as the molecular targets of halofuginone (HFG) and the natural product febrifugine.^14,15 In our mechanistic studies, we found that short-term exposure of asexual blood-stage (ABS) P. falciparum to HFG-produced parasites with an ~20-fold increased HFG EC₅₀ compared to the parent strain,¹⁶ while continued long-term selection under increasing drug pressure resulted in parasites that are 200- to 300-fold less sensitive to HFG. Unlike highly resistant parasites, which shared point mutations conferring HFG resistance mapping to the same amino acid residue in the active site of PfcProRS, we did not identify genetic determinants in PfcProRS in these moderately resistant (adaptive proline response [APR]) parasites. In contrast, comprehensive metabolomic analysis identified 10- to 20-fold increased levels of intracellular proline, establishing a mechanistic rationale for the decreased activity of the proline-competitive inhibitor HFG.^15,16 We defined this rapid acquisition of HFG resistance without any apparent genetic changes in PfcProRS as the APR. Specific changes in amino acid metabolism and homeostasis in response to drug exposure constitute a previously unrecognized mechanism of resistance evolution in Plasmodium. The changes in proline homeostasis potentially explain the failure of HFG analogs to control recrudescence in vivo and their narrow therapeutic indices as antimalarials.¹⁷

In eukaryotes, the phosphorylation of the eukaryotic initiation factor 2-alpha (eIF2α) by the GCN2 kinase functions as the central mediator of the integrated stress response (ISR). Although we demonstrated that short-term treatment of P. falciparum with HFG results in eIF2α phosphorylation, we found that the APR operates independently of this eIF2α stress signaling pathway and that the P. falciparum GCN2 ortholog, PfeIK1, is not required for the modulation of proline homeostasis in response to HFG treatment. This rules out a compensatory mechanism that is mediated through the amino acid response arm of the ISR.¹⁸

To pinpoint the cellular pathways governing the APR and identify potential approaches to counter this mode of resistance, we aimed to determine the primary sources contributing to the excess proline. In this report, we show that most of the free intracellular proline in APR parasites is derived biosynthetically from arginine. However, surprisingly, the genetic disruption of the arginine pathway does not eliminate the ability of APR parasites to mount an APR. Further genetic analysis, facilitated by our enhanced sequencing methods and analysis pipelines, revealed that loss-of-function mutations in the P. falciparum apicomplexan amino acid transporter 2 (PfApiAT2, also known as PfMFR4)^19,20 are the genetic changes driving the APR.

RESULTS

Increased proline levels are derived from arginine

Conceptually, Plasmodium parasites can source proline by three distinct mechanisms: (1) direct uptake from the extracellular environment, (2) de novo biosynthesis through the arginine and possibly glutamate pathways, and (3) hemoglobin digestion, which can directly supply proline and provide other amino acids as metabolic precursors (Figure 1A). All necessary metabolic enzymes for the arginine pathway, including arginase (ARG), ornithine aminotransferase (OAT), and pyrroline-5-carboxylate reductase, have been identified in P. falciparum. More recently, metabolic labeling studies have demonstrated the conversion of arginine to proline in ABS parasites.^21–24 However, a functional glutamate pathway, which represents the dominant metabolic source of proline in most organisms, has not been reported to date and might be absent in ABS parasites.^25–27

Figure 1. Proline acquisition pathways in ABS P. falciparum.

(A) The schematic depicts the three main routes by which P. falciparum, directly and indirectly, obtains proline. Direct uptake (blue arrows): proline (Pro), arginine (Arg), and glutamine (Gln) can be imported directly from the extracellular environment. Hemoglobin catabolism (red arrows): these amino acids can also be acquired through the breakdown of hemoglobin (Hgb) within the parasite’s digestive vacuole (DV). Biosynthesis (green arrows): proline can be synthesized from arginine via ornithine (Orn) and potentially from glutamine/glutamate. Solid lines represent pathways experimentally demonstrated in P. falciparum or RBCs, while dotted lines indicate pathways confirmed in other organisms but not yet validated in P. falciparum.

(B) Multiplexed labeling strategy with stable isotope labeled amino acids to determine the source of proline (Pro): mono-labeled proline will result in Pro+1, dual-labeled arginine will result in Pro+2, quintuple-labeled glutamine will result in Pro+5, and proline from hemoglobin digestion (directly or indirectly through precursors) will result in unlabeled proline (Pro+0).

(C) Parasites were incubated with multiplexed isotope media for 10 h, after which metabolites were extracted and analyzed by LC-MS. Proline levels for the different proline species were normalized to total signal of all detected metabolites per sample and subsequently compared to the Dd2 parental line. Shown is the average and standard deviation of the fold change to Dd2 (at least three biological replicates). No Pro species were detected from glutamine metabolism (Pro+5). Statistical analysis: unpaired Student’s test between Dd2 and Dd2_APR for each species: *p < 0.05, ***p < 0.001.

See also Figure S1 for amino acid uptake over time and Table S1 for complete metabolomics dataset.

To establish a comprehensive experimental approach that would enable us to simultaneously quantify the contributions from all three principal sources, we designed a multiplexed metabolomics approach utilizing stable isotope-labeled amino acids. Specifically, we prepared a proline custom growth media (triplex media) replacing proline, arginine, and glutamine with [¹⁵N]-proline, [¹³C-¹⁵N]-arginine, and [¹³C5]-glutamine, respectively, to detect proline derived from each of the exogenous sources (Figure 1B). Proline originating from direct uptake from the growth medium will exhibit a +1 m/z shift ([¹⁵N]-proline, Pro+1), while proline derived from the metabolic conversion of arginine will correspond to the mass of the arginine [¹³C-¹⁵N] and display an m/z shift of +2 (Pro+2). Although the current literature suggests the absence of a functional glutamate pathway in Plasmodium, we decided not to discount this possibility, given the general importance of this pathway in other organisms. Therefore, we included glutamine [¹³C5], which, unlike glutamate, is efficiently imported byP. falciparum and rapidly converted to glutamate²⁶; proline derived from the [¹³C5]-glutamine would then have an m/z shift of +5 (Pro+5). Lastly, proline that is directly or indirectly (via arginine or glutamine/glutamate conversion) derived from hemoglobin digestion is expected to show no m/z shift (Pro+0).

First, we incubated synchronized P. falciparum Dd2-infected red blood cells (iRBCs) and uninfected RBCs (uRBCs) in triplex media for up to 10 h to determine uptake kinetics. Cells were harvested at different intervals and subsequently subjected to liquid chromatography-mass spectrometry (LC-MS) analysis. We found that the concentration of isotope-labeled amino acids in parasites reached equilibrium within 2 h during the experimental time frame and remained at a steady state for 10 h (Figure S1; Table S1).

Next, we incubated RBCs infected with wild-type (WT) Dd2 or Dd2 parasites displaying the APR (Dd2_APR parasites) in triplex media for 10 h, followed by the isolation of polar metabolites and isotopic analysis of proline to identify the source of the increased proline. Consistent with our published results, total proline levels in the Dd2_APR parasites were significantly higher than in Dd2 parasites (Figure 1C).¹⁶ Of the exogenously added isotopically labeled amino acids, Pro+2 was the dominant proline isotopologue, indicating that biosynthesis via the arginine pathway supplies excess proline for the APR (Figure 1C). A smaller portion of total intracellular proline was Pro+1, suggesting a contribution from the direct uptake of the amino acid, albeit much lower than the amount derived from arginine metabolism. In agreement with previous reports, we observed the conversion of glutamine to glutamate.²⁶ However, we did not detect Pro+5 in either the Dd2 or Dd2_APR parasites, providing strong experimental support for the absence of the glutamate pathway in ABS parasites. We also observed an increased level of Pro+0. This represents proline derived from hemoglobin digestion either directly as proline or through the metabolism of hemoglobin-derived arginine.

APR is independent of the arginine to proline pathway

We hypothesized that disrupting the arginine pathway by eliminating either P. falciparum ARG (PfARG) or P. falciparum OAT (PfOAT) activity would prevent parasites from mounting the APR. We followed a reverse genetics approach to disrupt PfARG (PF3D7_0906500) or PfOAT (PF3D7_0608800) individually using CRISPR-Cas9-mediated gene-editing approaches (Figure 2A; Figure S2) and isolated clonal transgenic parasites. We confirmed the genotype of the knockouts (KOs) by PCR and whole-genome sequencing (WGS). Neither KO line exhibited growth defects or altered susceptibility to HFG compared to the parental strain (Dd2 EC₅₀ = 0.472 nM, Dd2ΔARG clone 1 EC₅₀ = 1.4 nM, and Dd2ΔOAT clone 1 EC₅₀ = 0.758 nM), consistent with previous findings that have identified PfARG and PfOAT as non-essential for asexual parasites (Figures 2B and 2C; Table S2).^28,29

Figure 2. Successful disruption of the arginine to proline pathway.

(A) Schema of the arginine to proline pathway.

(B) Clonal parasites with an ARG (Dd2ΔARG) or ornithine-δ-aminotransferase (Dd2ΔOAT) deletion do not have a growth defect compared to the parental line (Dd2). Shown is the relative growth of Dd2ΔARG C1 (clone 1), Dd2ΔOAT C1 (clone 1), and Dd2 normalized to the initial signal measured by SYBR Green to stain DNA as a proxy for parasite growth. Shown are the average and standard deviation of three biological replicates.

(C) Dd2ΔARG C1 and Dd2ΔOAT C1 parasites show similar sensitivity to halofuginone (HFG) as the parental Dd2 line. Shown is a representative dose-response plot showing sensitivity to HFG for the three cell lines (HFG EC₅₀s: Dd2 = 0.472 nM, Dd2ΔARG C1 = 1.4 nM, and Dd2ΔOAT C1 = 0.758 nM). Individual data points represent the average response of technical triplicates plus/minus standard deviation and are plotted along with a non-linear regression curve fit for each cell line tested.

(D and E) Parasites were incubated with multiplexed isotope media for 10 h, after which metabolites were extracted and analyzed by LC-MS. Proline (D) and ornithine (E) levels for the different metabolite species were normalized to total signal of all detected metabolites per sample and subsequently compared to the Dd2 parental line. Shown is the average and standard deviation of the fold change to Dd2 (at least two biological replicates). Statistical analysis: unpaired Student’s test between Dd2 and KO parasites for each species: *p < 0.05, ***p < 0.001, ****p < 0.0001.

See also Figure S2 for cloning strategy and genetic confirmation of KO lines, Table S1 for complete metabolomics dataset, and Table S2 for dose-response data.

We next analyzed the metabolomic profiles of the Dd2ΔARG and Dd2ΔOAT cell lines. Pro+2 was undetectable in the Dd2ΔOAT line, confirming that PfOAT is required for the conversion of arginine to proline (Figure 2D). As expected, there was a concomitant accumulation of ornithine (Orn+2) derived from arginine (Arg+2) in the Dd2ΔOAT parasite (Figure 2E). Interestingly, Pro+2 levels were unchanged in Dd2ΔARG parasites compared to WT parasites (Figure 2D). Further analysis revealed the presence of arginine-derived ornithine in uRBCs, consistent with previous reports (Table S1).²¹ These results suggest that the basal host RBC ARG activity is sufficient to compensate for the lack of PfARG activity, while genetic ablation of PfOAT activity abolished the conversion of arginine to proline.

Surprisingly, when we performed two independent short-term selection studies with Dd2ΔOAT parasites under HFG pressure, we observed a decrease in sensitivity to HFG within 2 weeks (Dd2ΔOAT EC₅₀ = 0.758 nM, Dd2ΔOAT_APR1 bulk EC₅₀ = 21.0 nM, and Dd2ΔOAT_APR2 bulk EC₅₀ = 5.03 nM, Figure 3A; Table S2). This time frame and dose-response phenotype are similar to those observed in our previous studies with Dd2_APR parasites.¹⁶ We renamed these resistant parasites Dd2ΔOAT_APR1 and Dd2ΔOAT_APR2.

Figure 3. Parasites with a disrupted arginine to proline pathway can still mount an APR.

Dd2ΔOAT parasites were exposed to a low dose of HFG, and recrudescent parasites showed the hallmarks of the APR.

(A) Example of a dose-response curve to HFG for the parental Dd2ΔOAT C1, bulk selected Dd2ΔOAT_APR2, and three clonal lines of Dd2ΔOAT_APR2 (Dd2ΔOAT_APR2 C1, Dd2ΔOAT_APR2 C2, and Dd2ΔOAT_APR2 C3) (HFG EC₅₀s: Dd2ΔOAT = 0.758 nM, Dd2ΔOAT_APR2 bulk = 5.03 nM, Dd2ΔOAT_APR2 C1 = 5.4 nM, Dd2ΔOAT_APR2 C2 = 5.2 nM, and Dd2ΔOAT_APR2 C3 = 5.0 nM). Individual data points represent the average response of technical triplicates plus/minus standard deviation and are plotted along with a non-linear regression curve fit for each cell line tested.

(B) Parasites were incubated with multiplexed isotope media for 10 h, after which metabolites were extracted and analyzed by LC-MS. Amino acid levels for the different metabolite species were normalized to total signal of all detected metabolites per sample and subsequently compared to the Dd2ΔOAT parental line. Shown is the average and standard deviation of the fold change of each amino acid species to Dd2ΔOAT (six biological replicates). Statistical analysis: multiple unpaired Student’s test with two-stage step-up method of Benjamini, Krieger, and Yekutieli between Dd2ΔOAT and bulk and clonal Dd2ΔOAT_APR2 parasites for each species: ****p < 0.0001.

See Table S1 for complete metabolomics dataset and Table S2 for dose-response data.

To elucidate the mechanism of resistance, we isolated clones from the Dd2ΔOAT_APR2 parasite bulk population by limiting dilution. We then performed metabolomic analysis using our triplex isotopic labeling strategy (as described earlier) on the bulk population and three independent clones. Strikingly, all induced Dd2ΔOAT_APR2 parasites displayed elevated proline levels compared to the parental strain (10-fold change, Figure 3B), mirroring the phenotype observed in HFG-resistant Dd2_APR parasites. Based on these findings, we conclude that a functional arginine pathway is not required for the APR.

Loss of PfApiAT2 function leads to APR

To identify if the induced Dd2ΔOAT_APR parasites had acquired any mutations during the selection process, we performed WGS on the bulk Dd2ΔOAT_APR1, Dd2ΔOAT_APR2, and clonal Dd2ΔOAT_APR2 parasites and their parental line (see Table S3 for variants identified by WGS). While we did not detect any copy-number variations or mutations in PfcProRS, the molecular target of HFG, we found that the Dd2ΔOAT_APR2 clones had a nonsense mutation (1471G>T; Gly491*) in the putative PfApiAT2 (PF3D7_0914700, Table S3). In contrast, Dd2ΔOAT_APR1 parasites did not appear to have an obvious resistance-conferring mutation. However, upon closer inspection of the PfApiAT2 locus, we observed two nonclonal mutations in PfApiAT2 (Thr97Ile and Leu159fs; 25% and 56% allele frequencies, respectively), suggesting a mixed population (Table S3). Our previous WGS analysis of APR parasites prioritized variant calls with ≥90% of reads mapping to the alternate allele, causing us to initially miss the PfApiAT2 mutant alleles present in the Dd2ΔOAT_APR1 bulk parasite population. Given our discovery of mixed alleles in PfApiAT2 in the Dd2ΔOAT_APR1 bulk parasites, we reevaluated our previous WGS data to accommodate mixed calls and focused on mutations in PfApiAT2 in particular. We discovered that all APR parasite lines from previous selections (including different genetic backgrounds, e.g., Dd2 and 3D7) had similar HFG-resistance phenotypes and mutations in PfApiAT2 but at unique sites. Interestingly, the majority of mutations resulted either in a stop codon directly or in a frameshift, leading to a premature stop, suggesting a loss of function associated with the resistance (summarized in Figure 4A, see also Table S3).

Figure 4. Nonsense mutations in PfApiAT2 lead to APR.

(A) Schema of all mutations detected in P. falciparum apicomplexan amino acid transporter 2 (PfApiAT2) in parasite lines selected with HFG (blue boxes indicate predicted transmembrane helices, and red bars indicate the position of PfApiAT2 mutations).

(B) Example of a dose-response curve to HFG for the parental 3D7 line, the selected 3D7_APR2 C1 clone containing a premature stop at position G230fs, the bulk revertant 3D7_APR2REV population with restored PfApiAT2 function, and a 3D7_APR2REV C1 clonal line isolated from the bulk revertant population (HFG EC₅₀s: 3D7 = 0.599 nM, 3D7_APR2 = 6.97 nM, 3D7_APR2REV = 0.443 nM, and 3D7_APR2REV clone 1 = 0.402 nM). Individual data points represent the average response of technical triplicates plus/minus standard deviation and are plotted along with a non-linear regression curve fit for each cell lines tested.

(C) Log2 protein fold change from 3D7 parental line compared to 3D7_APR2 C1, the 3D7_APR2REV bulk culture, or the 3D7_APR2REV C1 clonal line. The colored dot depicts PfApiAT2, the dotted line on the y axis at 3.3 depicts a −log10(p value) of 0.0005, and the dotted lines on the x axis depict a fold change of ±2.

(D) Example of a dose-response curve to HFG for the parental Dd2 line and three clonal PfApiAT2 KO lines (Dd2ΔApiAT2 C1, Dd2ΔApiAT2 C2, and Dd2ΔApiAT2 C3) (HFG EC₅₀s: Dd2 = 0.472 nM, Dd2ΔApiAT2 bulk = 12.7 nM, Dd2ΔApiAT2 C1 = 14.3 nM, Dd2ΔApiAT2 C2 = 13.7 nM, and Dd2ΔApiAT2 C3 = 16.3 nM). Individual data points represent the average response of technical triplicates plus/minus standard deviation and are plotted along with a non-linear regression curve fit for each cell line tested.

(E) Parasite metabolites from the Dd2 parent and three Dd2ΔApiAT2 clonal lines were extracted and analyzed by LC-MS. Amino acid levels for the different metabolite species were normalized to total signal of all detected metabolites per sample and subsequently compared to the Dd2 parental line. Shown is the average and standard deviation of the fold change of each amino acid species relative to Dd2 (six biological replicates). Statistical analysis: multiple unpaired Student’s test with two-stage step-up method of Benjamini, Krieger, and Yekutieli between Dd2 and bulk and clonal Dd2ΔApiAT2 parasites for each species: ****p < 0.0001. All dose-response curves were run with technical triplicates and repeated at least twice.

See Figure S3 for cloning strategy and genetic confirmation of KO lines, Table S1 for complete metabolomics dataset, Table S2 for dose-response data, Table S3 for variants identified in whole-genome sequencing, and Table S4 for complete proteomics dataset.

To further validate that loss of function of PfApiAT2 drives the APR phenotype, we reverted one of the mutations discovered in the induced parasites 3D7_APR2 (G230fs mutation) using CRISPR-Cas9 to its WT genotype (fs230G) and named them 3D7_APR2REV parasites. This reversion successfully restored WT sensitivity to HFG, as demonstrated by dose-response analysis of the 3D7_APR2REV bulk parasite population as well as individual 3D7_APR2REV clones (Figure 4B, 3D7 EC₅₀ = 0.599 nM, 3D7_APR2 EC₅₀ = 6.97 nM, 3D7_APR2REV bulk EC₅₀ = 0.443 nM, and 3D7_APR2REV clone 1 EC₅₀ = 0.402 nM, see also Table S2). Multiplexed quantitative proteome mapping^30,31 of the original 3D7, 3D7_APR2, and 3D7_APR2REV lines confirmed the loss of PfApiAT2 protein in the APR parasites and the restoration of PfApiAT2 protein in the 3D7_APR2REV bulk and clonal parasites (Figure 4C; Table S4). These results provide further evidence supporting the hypothesis that the G230fs mutation resulted in a loss of PfApiAT2 function.

To test the hypothesis that loss of function of PfApiAT2 is sufficient to cause the APR, we disrupted PfApiAT2 in Dd2 parasites, creating a functional KO (Dd2ΔApiAT2, Figure S3). These parasites were not previously exposed to HFG. When we tested the dose-response phenotype of Dd2ΔApiAT2 bulk and clonal parasites to HFG, we observed a >20-fold reduction in sensitivity compared to the parental line (HFG EC₅₀: Dd2 = 0.472 nM, Dd2ΔApiAT2 bulk = 12.7 nM, and Dd2ΔApiAT2 clones = 13.7–16.3 nM, Figure 4D; Table S2). Moreover, metabolomic profiling of Dd2ΔApiAT2 parasites revealed a >20-fold increase in proline compared to the WT Dd2 parent (Figure 4E; Table S1). Together, these data conclusively demonstrate that loss of PfApiAT2 function alone is sufficient to induce intracellular proline accumulation, thereby recapitulating the hallmark metabolic phenotype of the APR and conferring HFG resistance.

DISCUSSION

In this work, we employed a chemogenomic approach to establish a mechanistically conclusive model of the APR and identify potential vulnerabilities for therapeutic intervention. By integrating multiplexed metabolomic flux analysis, global proteomics, genetically modified parasite lines, and improved WGS analysis, we systematically dissected the metabolic pathways contributing to proline accumulation in HFG-resistant P. falciparum.

Our multiplexed metabolomics studies revealed a distinct metabolic profile in P. falciparum. Quantitative metabolite flux analysis revealed that WT parasites source proline to comparable extent from culture media, hemoglobin digestion, and arginine-derived biosynthesis. In the APR parasites, proline from media-derived arginine was increased, as was proline derived from hemoglobin. We cannot differentiate between proline sourced directly from hemoglobin digestion and proline synthesized intracellularly from hemoglobin-derived arginine.

Consistent with prior reports, our studies also confirmed that [¹³C5]-glutamine is efficiently taken up by ABS parasites and rapidly converted to [¹³C5]-glutamate, accounting for virtually all intraparasitic glutamate. However, despite the glutamate pathway constituting the primary route for de novo proline biosynthesis in most organisms, our findings demonstrate that it does not contribute to the intracellular proline pool in P. falciparum, regardless of the genetic background tested. These observations underscore differences in the metabolic features of this parasite compared to other eukaryotes and further substantiate the absence of a functional glutamate pathway for proline biosynthesis in ABSs.

While initial findings hinted at the arginine pathway’s involvement in the APR, genetic disruption of key enzymes (PfARG and PfOAT) demonstrated that this pathway is dispensable for HFG resistance. Interestingly, host erythrocytic ARG activity compensated for the loss of PfARG, highlighting the interplay between host and parasite metabolism. The ability of ΔOAT parasites to develop HFG resistance despite an inability to synthesize proline from arginine suggests that alternative proline sources, such as hemoglobin digestion, can be utilized. Additional studies will be needed to determine if this involves increased hemoglobin catabolism in ΔOAT parasites or if the contribution of hemoglobin-derived proline is effectively diluted by the abundant metabolic flux of arginine-derived proline in parasites with a functional ARG pathway. These observations highlight the complexity of proline homeostasis in P. falciparum and warrant further investigation into the interplay between different proline sources.

Having comprehensively investigated and excluded an immediate metabolomic response as the basis for the altered proline homeostasis observed in APR parasites, we turned our attention to reexamining the potential contribution of genetic factors. Although our previous studies did not identify a genetic basis for the APR,¹⁶ advances in our analytical pipeline, now capable of detecting SNPs, insertion-deletions, and copy-number variations even with heterozygous calls, allowed for a more refined analysis of WGS data. The comparison of the parental Dd2ΔOAT and Dd2ΔOAT_APR genomic sequences, along with the re-examination of nine other independently selected APR lines, revealed a striking pattern, showing that all lines harbored a total of 14 distinct mutations in PfApiAT2 (see Table S3). These included multiple independent nonsense and frameshift mutants predicted to result in premature termination consistent with a loss-of-function mechanism of resistance. We directly demonstrated that the G230fs mutation resulted in significant reduction of PfApiAT2 protein, and repair of that mutation restored protein levels in the parasite.

Targeted gene disruption and CRISPR-Cas9-mediated gene editing experiments confirmed the central role of PfApiAT2 in the APR. PfApiAT2 KO parasites phenocopied the APR while genetic correction of a PfApiAT2 mutation restored WT sensitivity to HFG. These complementary loss-of-function and gain-of-function experiments unequivocally demonstrate that PfApiAT2 is the central mediator of the APR, directly linking PfApiAT2 dysfunction to proline accumulation and HFG resistance. This discovery resolves the mystery of the APR and implicates PfApiAT2 as a proline transporter. Future exploration is required to determine if PfApiAT2 is a proline transporter and the directionality of that transport. This raises intriguing questions about its role in P. falciparum biology and drug resistance, particularly for strategies targeting prolyl-tRNA synthetase (ProRS).

PfApiAT2 is a member of the apicomplexan amino acid transporter (ApiAT) family of putative amino acid transporters that are present in all apicomplexan parasites. ApiATs comprise 11 subfamilies and have the closest homology to the major facilitator superfamily class.^32–34 A recent publication identified five ApiATs in P. falciparum (PfApiAT2, 4, 8, 9, and 10); all except PfApiAT9 were localized to the parasite plasma membrane during the asexual stage and in gametocytes.²⁰ Interestingly, Kenthirapalan et al. report that several members of the ApiAT family (ApiAT2, ApiAT4, and ApiAT8) are not essential for blood stage growth but are essential for development in the mosquito.¹⁹

Our data clearly show that loss of function of PfApiAT2 results in higher intracellular proline concentrations, implying that native PfApiAT2 is exporting proline. This is consistent with previous works by others demonstrating that several amino acids, including proline, are exported from infected cells.³⁵ It is estimated that the parasite utilizes only 16% of the amino acids derived from hemoglobin and exports the excess.³⁶

While previous concerns regarding the rapid emergence of resistance to HFG and its analogs cast doubt on the viability of ProRS as a clinically viable drug target, our recent development of proline-uncompetitive inhibitors that exhibit very low propensity for resistance and are insensitive to the APR, coupled with the essentiality of PfApiAT2 in mosquito stages, has renewed optimism.³⁷ Further investigations into PfApiAT2 function may uncover additional therapeutic vulnerabilities and could inform the development of transmission-blocking strategies.

Limitations of the study

While this study provides evidence that loss of PfApiAT2 function is associated with increased intracellular proline and resistance to ProRS inhibitors in P. falciparum, some limitations should be noted. First, our conclusions regarding the transporter function of PfApiAT2 are based on genetic and metabolomic association rather than direct measurement of substrate transport; we did not perform reconstitution assays or direct uptake/efflux studies to demonstrate proline movement across membranes mediated by PfApiAT2. Second, all experiments characterizing resistance and metabolic phenotypes were performed exclusively in cultured ABS parasites; therefore, our findings may not fully reflect proline homeostasis or transporter function in other parasite life cycle stages or in in vivo conditions where host and environmental factors could differ. Third, our metabolic tracing approach cannot inform on the relative contributions of proline derived directly from hemoglobin versus proline derived indirectly from arginine.

RESOURCE AVAILABILITY

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Dyann F. Wirth (dfwirth@hsph.harvard.edu).

Materials availability

All plasmids, cell lines, and small molecules generated in the study will be made available from the lead contact with a completed materials transfer agreement.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

P. falciparum cell lines and culture conditions

Dd2 is a chloroquine-resistant in vitro cultured parasite cloned from the Laos-derived W2-MEF parasite (BEI Resources, MRA-156). Dd2-2D4 is a genetically homogenous line that was cloned from Dd2 by limiting dilution in the Wirth lab. 3D7 is a chloroquine-sensitive in vitro cultured parasite cloned from the NF54 parasite (BEI Resources, MRA-151). The parent NF54 isolate was derived from a patient living near Schipol Airport, Amsterdam, who had never left the Netherlands. 3D7-A10 (aka clone IG06) is a genetically homogenous line that was subcloned from 3D7 by limiting dilution in the Goldberg lab and used in previous in vitro drug selection and sequencing efforts.³⁸ Cultures were cultured in 5% human O+ hematocrit (Grifols Bio Supplies, Inc.) and maintained under standard conditions [RPMI 1640 (Life Technologies) supplemented with 28 mM NaHCO₃ (Sigma), 25 mM HEPES (Sigma), 50 mg/ml hypoxanthine (Sigma), 25 μg/ml gentamycin (Sigma), and 0.5% AlbuMAX II (Life Technologies)], except where stated.⁴² Synchronized cultures were obtained by sorbitol treatment, as previously reported.⁴³ In brief, mixed stage cultures were centrifuged to isolate packed iRBCs then resuspended in 0.5x culture volume of a 5% sorbitol solution for 5-10 minutes at 37°C. Samples are then pelleted by centrifugation, the sorbitol solution aspirated, and then washed with incomplete RPMI media. After washing, the iRBC pellet is resuspended to the original culture volume with complete media and returned to the incubator. Cultures were monitored by Giemsa-stained thin smears and split as needed to maintain a parasitemia between 0.5-5%. Cell line nomenclature: Dd2 (parental line), Dd2_APR (Dd2 parasites selected with HFG and demonstrate the APR), Dd2ΔARG (Dd2 parasites with ARG genetically knocked out), Dd2ΔARG_APR (Dd2ΔARG parasites selected with HFG and demonstrate the APR), Dd2ΔOAT (Dd2 parasites with OAT genetically knocked out), Dd2ΔOAT_APR (Dd2ΔOAT parasites selected with HFG and demonstrate the APR), Dd2ΔApiAT2 (Dd2 parasites with ApiAT2 genetically knocked out), 3D7-A10 (parental line), 3D7_APR (3D7-A10 parasites selected with HFG and demonstrate the APR), 3D7_APRREV (3D7_APR parasites genetically edited to revert their ApiAT2 locus to wild type). All parasite cultures tested negative for mycoplasma using the LookOut Mycoplasma PCR Detection Kit (Sigma).

METHOD DETAILS

LC-MS analysis of amino acids and polar metabolites

Highly synchronous (within 4 h) early schizonts were magnetically purified with MACS CS columns (Miltneyi Biotec Inc., San Diego, CA, USA). Each sample was washed twice in PBS and then suspended in 10 μL PBS (Life Technologies, Carlsbad, CA, USA). Polar metabolites were extracted using nine volumes of 74.9:24.9:0.2 (v/v/v) acetonitrile/methanol/formic acid containing stable isotope-labeled internal standards (0.2 ng/μL valine-d8 (Sigma Aldrich, St. Louis, MO, USA); and 0.2 ng/μL phenylalanine-d8 (Cambridge Isotope Laboratories, Tewksbury, MA, USA) and stored at −80°C prior to the metabolite profiling assays.

Profiles of amino acids were measured using LC-MS as described previously.⁴⁴ Briefly, positive ionization, multiple reaction mode (MRM) data were acquired using a Q-Exactive Focus mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) coupled to a Nexera X2 U-HPLC (Shimadzu Corporation, Kyoto, Japan). Cell extracts (10μL) were injected onto a 150×2.1mm Atlantis HILIC column (Waters, Milford, MA, USA). The column was eluted isocratically at a flow rate of 250μL/minute with 5% mobile phase A (10 mM ammonium formate and 0.1% formic acid in water) for 1 minute followed by a linear gradient to 40% mobile phase B (acetonitrile with 0.1% formic acid) over 10 minutes. The ion spray voltage was 3.5kV and the source temperature was 350°C.

LC-MS data extraction and statistical analysis

TraceFinder 3.3 software (Thermo Fisher Scientific, Waltham, MA, USA) was used for automated peak integration, and metabolite peaks were manually reviewed for quality of integration and compared against a known standard to confirm identity. Stable isotope-labeled internal standards valine-d8 and phenylalanine-d8 were used to eliminate samples with poor data quality.

Each metabolite signal was normalized to the average of the internal standard peak signals in the same sample to control for variation between runs. Graphical representations and statistical analysis were performed in GraphPad Prism version 10 (GraphPad Software, La Jolla, CA, USA). A one-way analysis of variance was performed for multiple statistical comparison t-tests. The metabolite data generated in this study are available at the NIH Common Fund’s National Metabolomics Data Repository (NMDR) website, the Metabolomics Workbench, https://www.metabolomicsworkbench.org with Project ID: PR002663.

CRISPR/Cas9 genome editing

We used atwo plasmid CRISPR/Cas9 system. pDC2-Cas9-U6-hDHFR contained the Cas9 endonuclease and the guide RNA (gRNA) and pGEMhdhfr+ plasmid contained the donor homology region used by the parasite for repair. To generate the basic template vector pGEMhdhfr+, the hdhfr positive selectable marker cassette was PCR amplified from the vector pL-6_eGFP³⁹ with primers hdhfr 5’ F and hdhfr 3’ R and cloned into pGEM-3Z (Promega, Madison, WI) digested with HincII. Two flanking regions corresponding to the gene of interest (GOI) were then cloned into this basic vector to generate a template for homologous recombination and disruption of the GOI. We individually knocked out PfARG, PfOAT, and PfApiAT2 in the Dd2 parasite parental line. Gene sequences were obtained from PlasmoDB (plasmodb.org). Potential gRNAs for CRISPR/Cas9 editing of PfARG, PfOAT, and PfApiAT2 were identified using Benchling online software (https://benchling.com/). 1-2 gRNAs were designed per gene. gRNA oligos (Table S5) with overhang BbsI sites were synthesized by Integrated DNA Technologies, annealed, and ligated into the BbsI-digested pDC2-Cas9-U6-hDHFR plasmid. This previously described plasmid includes a U6 snRNA polymerase III promoter as well as regions for the expression of the Cas9 enzyme and the human DHFR drug resistance cassette.⁴⁰ Homology regions for the crossover were PCR amplified from Dd2 wild-type genomic DNA (Table S5). The 3’ homology regions were digested with PstI and SphI and ligated into the appropriately digested pGEMhdfr+ basic template vector. The coordinating 5’ homology regions were digested with AflII and KpnI and ligated into the pGEMhdfr+ vector already containing the 3’ homology region.

Generation of allelic replacements with CRISPR/Cas9

Two kinds of plasmids were generated for transfection of the 3D7_APR2 line to revert the G230fs mutation in PfApiAT2 back to the wildtype G230. The pDC2-cam-Cas9-U6-hDHFR vector⁴⁰ was used to generate a specific double-strand break. gRNAs were designed using Benchling (Benchling, Inc.) and ordered as primers from Integrated DNA Technologies (IDT, Coralville, Iowa, USA) with overhangs compatible with the BbsI overhangs in pDC2-cam-Cas9-U6-hDHFR. The pDC2-cam-Cas9-U6-hDHFR plasmid was digested with BbsI and ligated with the annealed gRNAs. To generate the repair template plasmid, approximately 500 base pairs surrounding PfApiAT2 G230 were amplified from gDNA of 3D7 parasites and ligated into pGEM-3z vector (Promega, Madison, WI, USA) using HincII. The scrambling of the gRNAs was introduced with the Q5^® site-directed mutagenesis kit (New England Biolabs, Ipswich, MA, USA) following manufacturer’s instructions. Sorbitol-synchronized ring-stage parasites were electroporated using a Bio-Rad Gene Pulser (Biorad, Hercules, CA) and conditions of 0.31 kV and 960 μF with a total of 100 μg plasmid template and two pDC2-cam-Cas9-U6-hDHFR plasmids with two different gRNAs in incomplete cytomix using a 0.2-cm cuvette. Resulting transfected parasite lines were cloned by limiting dilution and sequenced by Sanger sequencing to confirm successful gene editing. Sequences of gRNAs and primers used are indicated in Table S5.

Transfection of parasites

50μg each of the CRISPR/Cas9 and donor plasmids were resuspended in Cytomix media (120 mM KCl, 0.15 mM CaCl₂, 2 mM EGTA, 5 mM MgCl₂, 10 mM K₂HPO₄/KH₂PO₄, pH 7.6, 25 mM Hepes, pH 7.6) and transfected into red blood cells infected with sorbitol-synchronized ring-stage Dd2 parasites at 5% parasitemia. Bio-Rad Gene Pulser at 0.34 kV and 950 μF was used to electroporate iRBCs, as described previously.⁴⁵ Transfected iRBCs were plated at 5% hematocrit in RPMI complete medium to recover for 24 h before the addition of 5 nM WR99210. All transfected parasites were selected under continuous drug pressure except the revertant parasites, and their recovery was monitored by microscopy. Revertant parasites were kept under drug pressure for 4 days before removal of drug pressure. Transfected parasites were cloned by limiting dilution.⁴⁶ PCR screening for all genes and Southern blot for PfApiAT2 were used to confirm successful integration of hdhfr at the desired locus.

DNA extraction

Genomic DNA (gDNA) was isolated from parasites using QIAamp DNA Blood kits (Qiagen) following the manufacturer’s instructions with the following modifications: Prior to lysis of the samples, parasite cultures were pelleted and lysed by 0.05% Saponin in 1x PBS on ice for 5 minutes. Pellets were washed twice in 1x PBS to remove hemoglobin and then resuspended in 200μl PBS and the QIAamp protocol instructions were followed to obtain gDNA samples.

Southern blot

10μg gDNA of each clone and of the homology and donor plasmids was digested overnight with EcoRI and BstBI. DNA was resuspended, run on a 1.5% agarose gel, transferred to nitrocellulose membrane, and blot was carried out using Amersham ECL Direct Labeling and Detection System (Cytiva). 454bp PCR product of HR2 (Figure S3) was labeled with horseradish-peroxidase and served as the molecular probe.

In vitro drug sensitivity and dose-response analysis

Parasite growth was determined using an endpoint fluorescence assay based on the SYBR Green I method as reported previously.^47–49 P. falciparum parasites were seeded in 384-well plates at 1% hematocrit and 1% starting parasitemia. Growth was assessed by SYBR Green staining of parasite DNA after 72-h exposure to compound. All dose-response assays were carried out with 12-point dilutions in technical triplicate. Compounds were dispensed with an HP D300 Digital Dispenser (Hewlett Packard, Palo Alto, CA, USA). Fluorescence intensity measurements were performed on a SpectraMax M5 (Molecular Devices, Sunnyvale, CA, USA) and analyzed in GraphPad Prism version 10 (GraphPad Software, La Jolla, CA, USA) after background subtraction and normalization to control wells. EC₅₀ values were determined using a four-parameter nonlinear regression curve fit and are represented as mean ± standard deviation from at least three bioreplicate assays (Table S2). Statistical significance was determined by the Mann-Whitney test.

HFG-induction of KO parasites

Two independent 25mL cultures of mixed-stage Dd2ΔOAT parasites at 4% parasitemia were exposed to media containing HFG (4x EC₅₀) for 144h (three intra-erythrocytic developmental cycles). The cultures were then maintained in compound-free complete RPMI growth medium with regular media exchange until healthy parasites reappeared.

DNA extraction and WGS

gDNA was obtained from P. falciparum samples by washing iRBCs with 0.1% saponin and extracting gDNA using a QIAamp DNA Blood kit (Qiagen). Sequencing libraries were prepared with the Nextera XT kit (cat. no. FC-131-1024, Illumina) using the standard dual index protocol, and whole-genome sequencing was performed on an Illumina NovaSeq 6000 in RapidRun mode with 100–base pair paired-end reads. Reads were aligned to the P. falciparum 3D7 reference genome (PlasmoDB v. 13.0), and variants were called using a GATK-based analysis pipeline as previously described.³⁸ Previously, only variant calls with ≥90% reads mapping to the alternate allele were considered further in mutational analyses. In the reanalysis of the WGS data, variant calls were expanded to include cases where all standard quality metrics were met and the GATK HaplotypeCaller⁴¹ readout was “0/1” (indicating a mixed population of the alternate allele was present). Whole genome sequencing data have been deposited at the National Center for Biotechnology Information Sequence Read Archive (NCBI SRA: www.ncbi.nlm.nih.gov/sra) under the BioProject Accession number: PRJNA1219695.

Proteomics sample extraction

Infected RBCs were pelleted and lysed by 0.05% Saponin in 1x PBS on ice for 5 minutes to isolate freed parasites. Parasite pellets were then washed twice in cold 1x PBS to remove hemoglobin and stored at −80°C until time of extraction. Protein lysates were extracted from parasite pellets by resuspension in room temperature lysis buffer containing 75mM NaCl, 50mM HEPES pH 8.5, 10mM Sodium pyrophosphate, 10mM Sodium Fluoride, 10mM β-glycerophosphate, 10mM sodium orthovanadate, 3% SDS, 10mM PMSF, and ROCHE cOmplete^™ EDTA-free protease inhibitors at a 5:1 ratio of lysis buffer volume to parasite pellet volume. Cells were further lysed by passing the resuspended cells ten times through a 21-gauge needle. To remove insoluble cellular debris, the samples were centrifuged at 16,000 x g at room temperature and the supernatant transferred to a fresh microfuge tube and stored at −80C.

Protein LC-MS and analysis

The lysate was reduced and alkylated following the previously published method.⁵⁰ Subsequently, 25 μg of the resulting peptides were labeled using TMTpro reagents from Thermo Scientific, as per the manufacturer’s instructions. The labeled samples were then combined and subjected to fractionation using high pH reversed-phase HPLC.⁵⁰

12 of the resulting fractions were analyzed using a 3-hour reversed-phase LC-MS2/MS3 run on an Orbitrap Fusion Lumos instrument. For quantification, MS3 isolation employed Simultaneous Precursor Selection (SPS), as described in earlier studies.^30,51,52 Protein identification was based on MS2 spectra using the sequest algorithm, which searched against a Plasmodium falciparum (isolate 3D7) database (uniprot 2022) on an in-house-built platform.⁵³

To ensure high confidence in protein identifications, a target-decoy database-based search strategy was utilized to filter against a false-discovery rate (FDR) of protein identifications of less than 1%.⁵⁴ Proteomics raw files can be found on MassIVE data repository (massive.ucsd.edu) under the accession number MSV000096697.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses

Statistical analysis was performed using GraphPad Prism v10. A four-parameter non-linear regression was used to generate dose-response curves and EC₅₀ values. Metabolomics values were compared by multiple unpaired Student’s test with two-stage step-up method of Benjamini, Krieger, and Yekutieli between the parental line and the tested lines if six biological replicates were tested and by two-tailed student’s t test if only three or less biological replicates were compared. No methods were used to determine whether the data met assumptions of the statistical approach.

Adjusted p-values for the proteomics data set were inferred from multiple t tests followed by correction of the False Discovery Rate after Benjamini-Hochberg.

Reporting conventions

The center was defined by the arithmetic mean and the dispersion by the standard deviation. Generally, all experiments were repeated at least twice (3 biological replicates) and the individual data points are represented in the Figures. For all drug EC₅₀ data, all the mean, standard deviation, and number of biological replicates are given in Table S2. Exact n and what n represents (e.g., independent parasite cultures, technical wells) are specified in every figure legend and corresponding Results text.

The results were deemed significant at 0.01 < *p ≤ 0.05, 0.001 < **p ≤ 0.01, highly significant at 0.0001 < ***p ≤ 0.001, and extremely significant at ****p ≤ 0.0001.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.chembiol.2025.09.007.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and virus strains
E. coli DH10Bac	Invitrogen	Cat# 10361012
E. coli XL1 cells	Agilent Technologies	Cat# 50125058
Biological samples
O+ CPDA-1 Diagnostic Red Blood Cell Units	Grifols Biosupplies, Inc.	N/A
Chemicals, peptides, and recombinant proteins
SYBR Green I fluorescent dye	Invitrogen	Cat# S7563
Blasticidin S hydrochloride	RPI Corp	Cat# B12150-0.1; CAS 3513-03-9
Atovaquone	AK scientific	Cat# G211; CAS 95233-18-4
Chloroquine diphosphate salt	Millipore-Sigma	Cat# C6628; CAS 50-63-5
Halofuginone	Biosynth, Ltd.	Cat# FB29543; CAS 55837-20-2
NCP26	Tye et al.37	N/A
ProSA	Tye et al.37	N/A
Borrelidin	Millipore-Sigma	Cat# B3061; CAS 7184-60-3
Amodiaquine	Millipore-Sigma	Cat# 1031004; CAS 6398-98-7
Artemether	Millipore-Sigma	Cat# 1042780; CAS 71963-77-4
Artesunate	Millipore-Sigma	Cat# 1042850; CAS 88495-63-0
Halofuginol	Tye et al.37	N/A
Mefloquine	Millipore-Sigma	Cat# 1379059; CAS 51773-92-3
Piperaquine	Millipore-Sigma	Cat# C7874; CAS 915967-82-7
[13C5]-glutamine	Cambridge Isotopes Laboratories, Inc.	Cat# CLM-1822
[13C-15N]-arginine	Cambridge Isotopes Laboratories, Inc.	Cat# CLM-2051
[15N]-proline	Cambridge Isotopes Laboratories, Inc.	Cat# NLM-835
Valine-d8	Cambridge Isotopes Laboratories, Inc.	Cat# DLM-311-1
Phenylalanine-d8	Cambridge Isotopes Laboratories, Inc.	Cat# DLM-372-PK
Critical commercial assays
Nextera XT kit	Illumina	Cat# FC-131-1024
QIAamp DNA Blood kit	Qiagen	Cat# 51104
Maxiprep system	Qiagen	Cat# 12163
LookOut Mycoplasma PCR Detection Kit	Millipore-Sigma	Cat# MP0035
Amersham ECL Direct Labeling and Detection System	Cytiva	Cat# RPN3000
TMTpro™ Label Reagent Set	Thermo Fisher Scientific	Cat# A44522
Q5® site-directed mutagenesis kit	NEBiolabs	Cat# E0554S
Deposited data
Whole Genome Sequencing data from parental and APR parasite lines	NCBI Sequence Read Archive	NCBI SRA: PRJNA1219695
Metabolomics Data from parental and APR parasite lines	National Metabolomics Data Repository	Project ID: PR002663
Proteomics Data from parental, APR, and revertant parasite lines	MassIVE data repository	MassIVE: MSV000096697
Experimental models: Cell lines
Dd2	BEI Resources	Cat# MRA-156
3D7-A10	Cowell et al.38	N/A
HFGRI	Herman et al.15	N/A
Dd2APR1	Herman et al.16	N/A
Dd2APR3	This paper	N/A
Dd2APR4	This paper	N/A
Dd2ΔARG Clone 1	This paper	N/A
Dd2ΔOAT Clone 1	This paper	N/A
Dd2ΔOATAPR1	This paper	N/A
Dd2ΔOATAPR1	This paper	N/A
Dd2ΔOATAPR2	This paper	N/A
Dd2ΔOATAPR2 Clone1	This paper	N/A
Dd2ΔOATAPR2 Clone2	This paper	N/A
Dd2ΔOATAPR2 Clone3	This paper	N/A
Dd2ΔARGAPR	This paper	N/A
Dd2ΔARGAPR	This paper	N/A
3D7APR2	Fagbami et al.18	N/A
3D7APR2 Clone1	This paper	N/A
3D7APR2 REV bulk	This paper	N/A
3D7APR2 REV Clone1	This paper	N/A
3D7APR2 REV Clone2	This paper	N/A
3D7APR2 REV Clone3	This paper	N/A
3D7Δelk1APR2	Fagbami et al.18	N/A
3D7Δelk1APR2	Fagbami et al.18	N/A
3D7Δelk1APR2	Fagbami et al.18	N/A
3D7Δelk1APR2	Fagbami et al.18	N/A
3D7Δelk1APR3	Fagbami et al.18	N/A
3D7Δelk1APR4	Fagbami et al.18	N/A
Dd2ΔApiAT2 Clone 1	This paper	N/A
Dd2ΔApiAT2 Clone 2	This paper	N/A
Dd2ΔApiAT2 Clone 3	This paper	N/A
Oligonucleotides
Oligos for CRISPR gene editing and plasmid construction (See Table S5)	Integrated DNA Technologies	N/A
Recombinant DNA
pL-6_eGFP	Ghorbal et al.39	N/A
pGEM-3Z	Promega	Cat# P2151
pGEMhdfr+	This paper	N/A
pDC2-cam-Cas9-U6-hDHFR	Ng et al.40	N/A
Software and algorithms
GATK HaplotypeCaller	McKenna et al.41	https://gatk.broadinstitute.org/hc/en-us
Benchling	Benchling	https://benchling.com/
Graphpad Prism 10.0	Graphpad Software Inc.	https://www.graphpad.com/scientific-software/prism/

Highlights.

• PfApiAT2 loss mediates an adaptive proline response to ProRS inhibitors

• Genetic reversion of PfApiAT2 mutations restores wild-type drug sensitivity

• PfApiAT2 knockout alone raises intracellular proline without prior drug exposure

• The adaptive proline response is independent of de novo proline biosynthesis

SIGNIFICANCE.

Understanding the molecular mechanisms of resistance to antimalarial drugs is essential for developing effective therapies to combat malaria, a disease responsible for an estimated 250 million cases and over 600,000 deaths each year. This study elucidates a mode of rapid antimalarial drug resistance, termed the adaptive proline response (APR), to halofuginone (HFG), a potent inhibitor of P. falciparum cytoplasmic prolyl-tRNA synthetase (PfcProRS). The APR operates independently of the canonical eIF2α-mediated stress response pathway and is characterized by a 10- to 20-fold increase of intracellular proline, which competes with HFG for binding to PfcProRS. However, the molecular basis of the APR has remained elusive. By employing a multiplexed metabolomics approach using stable isotope-labeled amino acids and CRISPR-edited KO parasite lines, we unraveled the metabolic pathways contributing to increased intracellular proline levels in resistant parasites. We showed proline biosynthesis from arginine as the dominant source of excess proline. Surprisingly, however, our findings reveal that the arginine pathway is not essential for the APR. Instead, we identified loss-of-function mutations in the P. falciparum apicomplexan amino acid transporter 2 (PfApiAT2) as the primary driver of the APR phenotype. This discovery not only resolves the long-standing mystery of the APR mechanism but also highlights an alternative mode of drug resistance in malaria parasites. The elucidation of PfApiAT2’s role in proline homeostasis and its potential importance in parasite transmission stages opens new avenues for antimalarial drug development. Furthermore, our study demonstrates the power of an integrated chemogenomic approach, combining advanced genomic, proteomic, and metabolomic approaches to unravel complex biological phenomena. These findings have broad implications for understanding amino acid metabolism in Plasmodium and potentially other apicomplexan parasites and for developing strategies to combat drug resistance. Ultimately, this work contributes significantly to our knowledge of malaria parasite biology and informs future efforts in antimalarial drug discovery.

Data and code availability

• WGS data have been deposited at the National Center for Biotechnology Information Sequence Read Archive (NCBI SRA: www.ncbi.nlm.nih.gov/sra) under the BioProject accession number: PRJNA1219695. The metabolite data generated in this study are available at the NIH Common Fund’s National Metabolomics Data Repository (NMDR) website, the Metabolomics Workbench, https://www.metabolomicsworkbench.org with Project ID: PR002663. Mass spectrometry proteomics data have been deposited at the Mass Spectrometry Interactive Virtual Environment repository (massive.ucsd.edu) under the accession number MSV000096697.

• This article does not report original code.

• Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.