Haem-activated promiscuous targeting of artemisinin in Plasmodium falciparum

Abstract

The mechanism of action of artemisinin and its derivatives, the most potent of the anti-malarial drugs, is not completely understood. Here we present an unbiased chemical proteomics analysis to directly explore this mechanism in Plasmodium falciparum. We use an alkyne-tagged artemisinin analogue coupled with biotin to identify 124 artemisinin covalent binding protein targets, many of which are involved in the essential biological processes of the parasite. Such a broad targeting spectrum disrupts the biochemical landscape of the parasite and causes its death. Furthermore, using alkyne-tagged artemisinin coupled with a fluorescent dye to monitor protein binding, we show that haem, rather than free ferrous iron, is predominantly responsible for artemisinin activation. The haem derives primarily from the parasite's haem biosynthesis pathway at the early ring stage and from haemoglobin digestion at the latter stages. Our results support a unifying model to explain the action and specificity of artemisinin in parasite killing.

Figure 1. The chemical proteomics approach to study artemisinin's mechanism of action.

(a) Chemical structures of artemisinin (Art), artesunate (Arts) and the alkyne-tagged-clickable probe (AP1). (b) General workflow of the chemical proteomics approach. Fluorescence labelling was used to study the activation mechanism of artemisinin, while biotin pull-downs coupled with LC–MS/MS were used to identify protein targets of artemisinin. (c) The killing effect of AP1 is comparable to that of artemisinin and artesunate on P. falciparum 3D7. (d) In situ parasite labelling with AP1. The labelling was dose dependent and specific to parasite proteins. Healthy RBC cytosolic proteins were not labelled. (e) The AP1 in situ parasite labelling was artemisinin specific as the excess Arts can largely compete with AP1-target labelling. (f) Free-radical scavenger (Tiron, 1 mM; Trolox, 400 μM; TEMPO, 1 mM) co-treatment reduces the level of parasite protein alkylation by AP1. Fluo, fluorescence scanning; Coo, Coomassie staining. Error bars represent s.d. in three independent replicates in c. Full-gel images for panels d and f are shown in Supplementary Fig. 13.

Figure 2. Artemisinin targets are involved in multiple biological processes essential for parasite survival.

GO analysis conducted using Cytoscape with Cluego plugin revealed that the AP1 targets are involved in many essential biological processes of the parasite, including the metabolism of carboxylic acids, cellular biogenic amines and nucleosides, as well as ribonucleoside biosynthesis. This highlights the multi-targeting ability of artemisinin, which exerts numerous effects on the physiological state of the malaria parasite.

Figure 3. In vitro binding and functional validation of artemisinin targets.

(a) Artemisinin specifically interacts with OAT, PyrK, LDH, SpdSyn, SAMS and TCTP as the unlabelled artesunate (25 ×) can compete with the AP1 binding. Heat denaturation reduces the AP1-labelling level of OAT, suggesting that the interaction of artemisinin with OAT is activity based. (b) Dose-dependent labelling of OAT with AP1 (4 h treatment). (c) Time-dependent labelling of OAT with AP1. (d) The interaction of artemisinin with OAT may involve thiol and amine groups as IAA (blocking thiol, 30 mM) and NEM (blocking amine, 10 mM) pretreatment (20 min) can reduce binding. (e,f) Activated artesunate inhibits the activities of PyrK (e) and LDH (f) in vitro. Δ, heat denaturation; IAA, iodoacetamide; NEM, N-ethylmaleimide; Conc., concentration. Error bars represent s.d. in three independent replicates in e and f. Full-gel images for panels a–d are shown in Supplementary Fig. 13.

Figure 4. Artemisinin activation is haem dependent.

(a) Artemisinin's interaction with OAT depends mainly on haemin (200 μM) or haem (200 μM haemin reduced by L-ascorbic acid (Vc, 200 μM), Na₂S₂O₄ (200 μM) or glutathione (GSH, 200 μM)) and much less on ferrous iron (FeSO₄, 200 μM). Pretreatment (30 min) with DFO (200 μM) slightly reduced the activation of artemisinin. (b) For the live parasite, pretreatment (1 h) with the cysteine protease inhibitor ALLN markedly abolished AP1 labelling of the parasite proteins in situ, whereas pretreatment with DFO (30 min) had little effect. (c) Modulation of the endogenous haem biosynthesis pathway affects the AP1 labelling intensity in HCT116 colon cancer cells. Pretreatment (1 h) with the haem synthesis precursor ALA (1 mM) enhances the labelling intensity. Conversely, pretreatment with the haem synthesis inhibitor SA (500 μM) reduces the labelling intensity. ALA and SA co-treatment did not enhance the labelling intensity as SA inhibition occurs downstream of the haem synthesis pathway. Free iron (FeSO₄, 200 μM) did not affect probe activation. DFO (200 μM) partially reduces the labelling signal intensity, likely due to its role in chelating the free iron, thus impeding the final step of haem synthesis. (d) The percentage of HCT116 cells killed by AP1 treatment or pretreatment with various haem synthesis modulators. (e) Correlation between the AP1 labelling intensity of the total HCT116 target proteins and the percentage of cancer cells killed by AP1 in the presence of various haem synthesis modulators. Error bars represent s.d. in three independent replicates in d and e. Full-gel images for panels a–c are shown in Supplementary Fig. 13.

Figure 5. Artemisinin activation in P. falciparum relies on two distinct haem sources.

(a) AP1 fluorescence-labelling intensity is much lower at the early ring stage compared with the latter stages (trophozoite and schizont stages) of the parasite. The cysteine protease inhibitor ALLN reduced the labelling intensity at the latter stages but could not reduce the labelling intensity at the early ring stage. (b) Modulation of the haem biosynthesis pathway of the parasite affected the labelling intensity of AP1 in the early ring stage. Pretreatment (1 h) with the haem synthesis precursor ALA (1 mM) enhanced the labelling signal intensity. Conversely, pretreatment with the haem synthesis inhibitor SA (500 μM) partially reduced the labelling intensity. (c) A model for artemisinin's mechanism of action. Artemisinin activation relies on haem generated in the parasite from both biosynthesis and haemoglobin digestion. In the early ring stage, biosynthesized haem was primarily responsible for drug activation; at the latter stages, both pathways co-existed, with haem derived from haemoglobin digestion playing the major role. COPROGEN, coproporphyrinogen III; CRT, chloroquine resistance transporter; ER, endoplasmic reticulum; Hb, haemoglobin; MDR, multidrug resistance protein; PBG, porphobilinogen; PPIX, protoporphyrin IX; PROTOGEN, protoporphyrinogen IX; UROGEN, uroporphophyrinogen III. Full-gel images for panels a and b are shown in Supplementary Fig. 13.