Genetic screens reveal a central role for heme metabolism in artemisinin susceptibility

Abstract

Artemisinins have revolutionized the treatment of Plasmodium falciparum malaria; however, resistance threatens to undermine global control efforts. To broadly explore artemisinin susceptibility in apicomplexan parasites, we employ genome-scale CRISPR screens recently developed for Toxoplasma gondii to discover sensitizing and desensitizing mutations. Using a sublethal concentration of dihydroartemisinin (DHA), we uncover the putative transporter Tmem14c whose disruption increases DHA susceptibility. Screens performed under high doses of DHA provide evidence that mitochondrial metabolism can modulate resistance. We show that disrupting a top candidate from the screens, the mitochondrial protease DegP2, lowers porphyrin levels and decreases DHA susceptibility, without significantly altering parasite fitness in culture. Deleting the homologous gene in P. falciparum, PfDegP, similarly lowers heme levels and DHA susceptibility. These results expose the vulnerability of heme metabolism to genetic perturbations that can lead to increased survival in the presence of DHA.

Fig. 1. DHA kills wild-type T. gondii efficiently and its effect is influenced by mutation of K13.

a Treatment with increasing concentrations of dihydroartemisinin (DHA) or pyrimethamine (Pyr) over 24 h resulted in a reduction in parasite viability, measured as the number of vacuoles with two or more parasites normalized to untreated wells. Results are mean ± SEM for n = 4 independent experiments. b Viability was similarly assessed for extracellular parasites treated with 1 or 10 µM DHA for varying periods of time, then washed and allowed to invade host cells. Viable vacuoles were normalized to untreated parasites kept extracellular for equal periods of time. Results are mean ± SEM for n = 3 independent experiments, each performed in technical triplicate. c Quantification of host monolayer lysis after treatment with 1 or 10 μM DHA for the indicated time, normalized to uninfected monolayers. Results are mean ± SEM for n = 4 independent experiments. d Alignment of the mutated region of K13 in P. falciparum and T. gondii displaying the chromatogram for the K13^C627Y T. gondii line. e Extracellular dose–response curve for parental or K13^C627Y parasites treated with DHA for 5 h. Results are mean ± SEM for n = 7 or 5 independent experiments with parental or K13^C627Y parasites, respectively. f Monolayer lysis following infection with parental or K13^C627Y parasites for 1 h, prior to a 5 h treatment with varying concentrations of DHA. Results are mean ± SEM for n = 4 independent experiments. g Monolayer lysis following infection with parental or K13^C627Y parasites for 24 h, prior to 5 h treatment with varying concentrations of DHA. Results are mean ± SEM for n = 4 or 3 independent experiments with parental or K13^C627Y parasites, respectively. h Graph summarizing the fold change in DHA EC₅₀ between parental and K13^C627Y parasites, treated 1 or 24 h after invasion. Results are mean ± SD for n = 4 or 3 independent experiments, p-value calculated from two-tailed unpaired t-test.

Fig. 2. Genome-wide screen under sublethal DHA concentration reveals that loss of Tmem14c increases DHA susceptibility.

a Screening workflow. The drug score is defined as the log₂-fold change in the relative gRNA abundance between the DHA-treated and vehicle populations, where lower scores are indicative of genes that enhance drug susceptibility when disrupted. b Results of a genome-wide CRISPR screen (calculated as described in a) comparing treatment with 50 nM DHA (equivalent to EC₅) to vehicle-treated parasites. Guides against one gene, Tmem14c, were significantly depleted by the sublethal DHA concentration, but were retained in the vehicle control. c ΔTmem14c parasites formed plaques normally under standard growth conditions, but their plaquing ability was attenuated in the presence of 100 nM DHA. The parental strain proliferated normally in both conditions. d Extracellular ∆Tmem14c parasites had a decreased EC₅₀ compared to the parental line (p = 0.0003, extra-sum-of-squares F-test). Results are mean ± SEM for n = 7 independent experiments. e Monolayer lysis following infection with parental or ΔTmem14c parasites and treatment with varying concentrations of DHA during hours 1–6 (top panel) or 24–29 (bottom panel) post infection. Results are mean ± SEM for n = 4 or 3 independent experiments for parental or ΔTmem14c parasites, respectively. f Overexpression of Tmem14c-Ty co-localized with mitochondrial marker mtHSP70. Scale bar is 5 μm. g Relative abundance of selected amino acids from targeted metabolomics of intracellular parental or ∆Tmem14c parasites. Results are mean ± SD for n = 3 technical replicates, with p-values calculated by one-way ANOVA.

Fig. 3. Genome-wide CRISPR screen under high DHA pressure identifies TCA and heme biosynthetic pathways as determinants of DHA susceptibility.

a Aggregated drug scores from two independent genome-wide CRISPR screens performed with a lethal dose of 500 nM DHA. Drug scores defined as the log₂-fold change in the relative gRNA abundance between the DHA-treated and vehicle populations, where higher scores are indicative of genes that reduce drug susceptibility when disrupted. Genes with consistently high drug scores are highlighted, indicating that their disruption favored parasite survival under otherwise lethal concentrations of DHA. b Diagram of key metabolic pathways highlighting genes with high drug scores in the screens from a and a third screen performed with 10 µM DHA, analyzed separately. Predicted localization of enzymes within organelles for the TCA cycle and heme biosynthesis pathways. 2-DG 2-deoxyglucose, ɑ-KG ɑ-ketoglutarate, ALA δ-aminolevulinic acid, CPIII coproporphyrinogen III, Gly glycine, HMB hydroxymethylbilane, NaFAc sodium fluoroacetate, PGB porphobilinogen, PPIX protoporphyrin IX, PPGIX protoporphyrinogen IX, SA succinyl acetone. Hits found in two or more replicates (blue) or only one replicate (yellow) are indicated. c Pretreatment with 500 mM NaFAc for 2 h significantly increased the EC₅₀ of DHA (extra-sum-of-squares F-test, p < 0.0001). Results are mean ± SEM for n = 7 or 4 independent experiments with vehicle or NaFAc treatment, respectively. d Pretreatment with 10 mM SA significantly increased the EC₅₀ of DHA (extra-sum-of-squares F-test, p < 0.0001). Results are mean ± SEM for n = 7 or 5 independent experiments with vehicle or SA treatment, respectively. e Pretreatment with 5 mM 2-DG did not significantly affect the EC₅₀ of DHA (extra-sum-of-squares F-test, p = 0.6). Results are mean ± SEM for n = 7 or 3 independent experiments with vehicle or 2-DG treatment, respectively. f Porphyrin levels (combination of heme and PPIX) in parasites exposed to various compounds, measured by fluorescence and normalized to untreated parasites (NT). Results are mean ± SD for n = 3 independent experiments, each performed in technical duplicate; p-values from a one-way ANOVA with Tukey’s test.

Fig. 4. Strains generated to study the function and localization of DegP2.

a, b The DegP2 coding sequence was replaced with YFP to generate the ∆DegP2 line, which was subsequently complemented with an HA-tagged copy of DegP2 (dark green) under the regulation of the TUB1 promoter (a). Separately, the endogenous locus of DegP2 was first tagged with the Ty epitope, and a point mutation was introduced at the catalytic serine (b). DegP2-HA, DegP2-Ty, and DegP2^S569A-Ty co-localized with the mitochondrial marker mtHSP70. Merged image additionally displays YFP (green) for ∆DegP2 strains and DNA stain (blue) for the HA-stained samples. Scale bar is 5 μm. c A DegP2-inducible mutant constructed using the U1 system. Staining for the HA tag appended to the DegP2 cKD locus showed the expected localization of the protein to the mitochondrion. Individual parasites are visualized by staining for CDPK1. Scale bar is 5 μm. d Immunoblotting for DegP2’s HA tag showed robust depletion of DegP2 48 h after a 2 h rapamycin treatment. CDPK1 was used as a loading control.

Fig. 5. Deletion of DegP2 alters porphyrin concentrations and DHA susceptibility.

a ∆DegP2 and ∆DegP2/DegP2-HA parasites formed smaller plaques than their parental strains, indicating a growth defect. b Endogenously Ty-tagged DegP2 (DegP2-Ty) parasites and a derived line bearing a mutation in the protease domain (DegP2^S569A-Ty) formed plaques normally. c Total porphyrin concentrations (pmol/mg protein) showing reduced porphyrin concentrations for both ∆DegP2 and DegP2^S569A-Ty parasites. Results are mean ± SD for n = 3–4 independent experiments; p-values are from one-way ANOVA with Sidak’s multiple comparison test. d DHA Mean EC₅₀ values ± SD were calculated by curve-fitting 4–7 independent experiments together. e Porphyrin levels show a negative correlation with DHA EC₅₀ (r² = 0.75) in the strains and conditions tested. f DegP2 cKD parasites formed plaques similarly to DiCre, both in the presence or absence of rapamycin. g Total porphyrin concentrations for DiCre and DegP2 cKD in the presence or absence of rapamycin. Results are mean ± SEM for n = 4 independent experiments; p-values are from a two-way ANOVA. h DegP2 cKD parasites outcompeted DiCre parasites in the presence of DHA. The fraction of the population composed of mutant parasites was calculated at each time point by flow cytometry. Results are mean ± SEM for n = 3 independent experiments.

Fig. 6. DegP2 influences the intersection of heme biosynthesis, the TCA cycle, and the ETC.

a TPP revealed that depleting DegP2 reliably changed the melting temperatures of 13 proteins (highlighted in green, p < 0.2 by z-test in each of two replicates). Three mitochondrial proteins are indicated by their gene IDs. b Melting curves for three mitochondrial proteins identified as hits by TPP. c Dose–response curve for parasites treated with the complex II inhibitor TTFA. Results are mean ± SEM for n = 4 or 3 independent experiments, for the parental and ∆DegP2 or ∆DegP2/DegP2-HA strains, respectively. ∆DegP2 is significantly more resistant to TTFA compared to the parental; p < 0.0001 from extra-sum-of-squares F-test. d Dose–response curve for parasites treated for 5 h with increasing concentrations of atovaquone (ATQ). Results are mean ± SEM for n = 5 or 3 independent experiments for the parental or ∆DegP2 strains, respectively. Both strains displayed similar susceptibility to ATQ; p = 0.59 from extra-sum-of-squares F-test. e Summary of changes in metabolites between parental and ∆DegP2 parasites in the TCA cycle and closely related pathways. Full results can be found in Supplementary Data 6. Asp aspartate, α-KG α-ketoglutarate, Gln glutamine, Glu, glutamate. Asterisks indicate significant change from parental, *p < 0.05, **p < 0.005, ***p < 0.0005, by two-way ANOVA.

Fig. 7. Mutating the ortholog of DegP2 in P. falciparum reduces susceptibility to DHA.

a Diagram showing strategy to knockout PfDegP in the Cam3.II^WT background. b PCR confirmation of successful KO of PfDeg2. PCR 1 (using primers P7 and 8) confirmed the loss of exons 3 and 4 in the two DegP KO clones; clone G10 was used for the remainder of this study. PCR 2 (using primers P7 and 5) demonstrated a shorter product upon successful deletion of two exons. c Ring-stage survival assay, performed as above, using the Cam3.II^WT, Cam3.II^ΔPfDegP, and the K13 mutant Cam3.II^R539T lines. Results are mean ± SEM for n = 3 independent replicates; p-values derive from two-tailed, unpaired t-tests. d Schizont-stage survival assay following a 4 h pulse of 700 nM DHA. Results are mean ± SEM for n = 4 independent replicates; p-values derived from two-tailed unpaired t-test. e–g Heme measured from the free (e), hemoglobin-associated (f), and hemozoin (g) pools from trophozoites for various strains. Heme concentrations were calibrated to a standard curve and normalized to the number of parasites per sample to calculate fg/cell. Results are mean ± SD for n = 4 independent replicates; p-values from one-way ANOVA.