.............................................................. Artemisinins target the SERCA of Plasmodium falciparum U. Eckstein-Ludwig 1 , R. J. Webb 1 , I. D. A. van Goethem 2 , J. M. East 2 , A. G. Lee 2 , M. Kimura 3 , P. M. O’Neill 4 , P. G. Bray 5 , S. A. Ward 5 & S. Krishna 1 1 Department of Cellular and Molecular Medicine, St George’s Hospital Medical School, Cranmer Terrace, London SW17 0RE, UK 2 Department of Biochemistry and Molecular Biology, University of Southampton, Southampton SO16 7PX, UK 3 Radioisotope Centre, Osaka City University Medical School, Asahi-machi 1-4-3, Abeno-ku, Osaka 545-8585, Japan 4 Department of Chemistry, The Robert Robinson Laboratories, University of Liverpool, Liverpool L69 7ZD, UK 5 Molecular and Biochemical Parasitology Group, Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L3 5QA, UK ............................................................................................................................................................................. Artemisinins are extracted from sweet wormwood (Artemisia annua) and are the most potent antimalarials available 1 , rapidly killing all asexual stages of Plasmodium falciparum 2 . Artemisi- nins are sesquiterpene lactones widely used to treat multidrug- resistant malaria 1 , a disease that annually claims 1 million lives. Despite extensive clinical and laboratory experience 3–5 their molecular target is not yet identified. Activated artemisinins form adducts with a variety of biological macromolecules, including haem, translationally controlled tumour protein (TCTP) and other higher-molecular-weight proteins 6 . Here we show that artemisinins, but not quinine or chloroquine, inhibit the SERCA orthologue (PfATP6) of Plasmodium falciparum in Xenopus oocytes with similar potency to thapsigargin (another sesquiterpene lactone and highly specific SERCA inhibitor). As predicted, thapsigargin also antagonizes the parasiticidal activity of artemisinin. Desoxyartemisinin lacks an endoperoxide bridge and is ineffective both as an inhibitor of PfATP6 and as an antimalarial. Chelation of iron by desferrioxamine abrogates the antiparasitic activity of artemisinins and correspondingly attenuates inhibition of PfATP6. Imaging of parasites with BODIPY-thapsigargin labels the cytosolic compartment and is competed by artemisinin. Fluorescent artemisinin labels para- sites similarly and irreversibly in an Fe 21 -dependent manner. These data provide compelling evidence that artemisinins act by inhibiting PfATP6 outside the food vacuole after activation by iron. Some 3,7 , but not all 8–10 , studies suggest that artemisinins act by haem-dependent activation of an endoperoxide bridge occurring within the parasite’s food vacuole. However, localization of artemi- sinins to parasite and not food vacuole membranes 11 , and killing of tiny rings lacking haemozoin argue against the food vacuole being a major site for drug action 2 . Artemisinins show structural similarities to thapsigargin, which is a highly specific inhibitor of sarco/ endoplasmic reticulum Ca 2þ -ATPase (SERCA). We therefore hypothesized that when activated, artemisinins act by specifically and selectively inhibiting the SERCA of P. falciparum. PfATP6 is the only SERCA-type Ca 2þ -ATPase sequence in the parasite’s genome. We expressed PfATP6 in Xenopus laevis oocytes because it could not be functionally assayed in COS cells (J.M.E., unpublished work). Membrane preparations from oocytes expres- sing PfATP6 consistently gave higher Ca 2þ -dependent ATPase activities than water-injected control oocytes (mean ^ s.e.m. activities of 0.054 ^ 0.006 IU compared with 0.026 ^ 0.003 inter- national units (IU), n ¼ 25; P , 0.0001). These activities are comparable to those observed with PfATP4, another P. falciparum Ca 2þ -ATPase 12 . Figure 1a compares the Ca 2þ -activation profiles of PfATP6 and rabbit skeletal muscle SERCA1a (sarcoplasmic reticu- lum preparation) and shows that PfATP6 and SERCA1a are acti- vated by similar concentrations of Ca 2þ ions, [Ca 2þ ] free (Fig. 1a and ref. 12). PfATP6 (like SERCA1a) is inhibited by thapsigargin and cyclopiazonic acid, but not by ouabain, a specific Na þ /K þ -ATPase inhibitor (Fig. 1b). Vanadate, which is a more general inhibitor of P- type ATPases, also inhibits PfATP6. We then assessed the inhibitory properties of chloroquine (a 4-aminoquinoline), quinine (a cinchona alkaloid) and artemisinin on transporter proteins encoded by P. falciparum and expressed in Xenopus oocytes, as well as their effects on SERCA1a. Supra- therapeutic concentrations of these antimalarials did not inhibit glucose transport by the falciparum hexose transporter (PfHT), confirming that none of these structurally diverse compounds Figure 1 Functional characterization of PfATP6. a, Ca 2þ -dependency of PfATP6 activity (filled circles) compared with SERCA1a (SR) activity (open circles) is shown as a percentage of maximal activity (normalized to the maximum value, and not corrected for endogenous activity for PfATP6) at the indicated [Ca 2þ ] free values. p Ca ¼ 2log 10 [Ca 2þ ] free . Bars with each symbol represent standard errors for each assay (a minimum of three experiments for sarcoplasmic reticulum (SR) and seven for PfATP6). Half-maximal activation is achieved at [Ca 2þ ] free values of 0.64 mM for PfATP6 and 0.3 mM for SERCA. Maximal activation is at [Ca 2þ ] free of 48 mM for both PfATP6 and SERCA. b, Inhibitor profiles for PfATP6 and SERCA. ATPase activities were determined with inhibitors and are compared with control values: thapsigargin (grey bar, 0.8 mM, P , 0.0001; multivariate analysis of variance, MANOVA), sodium orthovanadate (white bar, 100 mM, P ¼ 0.043), ouabain (hatched bar, 100 mM, P ¼ 0.87), and cyclopiazonic acid (black bar, 1 mM, P ¼ 0.0033). Each bar represents a minimum of six experiments. c, Effects of antimalarials on two P. falciparum Ca 2þ -ATPases (PfATP6 and PfATP4), the P. falciparum hexose transporter (PfHT) and mammalian SERCA (SR). All activities are shown as a percentage of uninhibited activity and are compared with these activities: quinine (open bars, 10 mM, P . 0.1; MANOVA), chloroquine (hatched bar, 1 mM, P . 0.1), artemisinin (black bars; 50 mM, P . 0.1 for PfATP4, SR, PfHT; 1 mM for PfATP6, P , 0.0001). Ca 2þ -ATPase activity for PfATP6 is shown after correction of Ca 2þ -ATPase activity measured in water-injected oocytes, which was unnecessary for PfATP4, because there was no inhibition observed with any antimalarial. Each bar represents a minimum of four experiments. letters to nature NATURE | VOL 424 | 21 AUGUST 2003 | www.nature.com/nature 957 © 2003 Nature Publishing Group