Considerations on the mechanism of action of artemisinin antimalarials: part 1--the 'carbon radical' and 'heme' hypotheses.

The isolation of artemisinin from the traditional medicinal herb qīng hāo (Artemisia annua), its characterization as a peroxide and preparation of the derivatives dihydroartemisinin, artemether and artesunate in the 1970s and 1980s by Chinese scientists under the umbrella of Project 523 collectively represents one of the great events in medicine in the latter third of the 20(th) Century. Artemisinins have become the most important component of chemotherapy of malaria: although used initially in monotherapy, they are now used in combination therapies or ACTs with longer half-life quinolines or arylmethanols. Nevertheless, the recent emergence of artemisinin-tolerant strains of the malaria parasite as reflected in increased clearance times of parasitaemia in patients treated with ACTs represents the greatest threat to control of malaria since resistance to chloroquine was first reported over 55 years ago. Importantly, the event brings into sharp focus the realization that relatively little is precisely understood, as opposed to widely assumed, for the mechanism of drug action of artemisinins and their synthetic peroxide analogues. Thus, we review here their antimalarial activities, the use of artemisinins in combination therapies, drug-drug interactions with the quinolines and arylmethanols, and metabolism of the artemisinins and synthetic peroxides. The mechanism of action of quinolines and arylmethanols, in particular their ability to induce redistribution of heme into the parasite cytosol, is also highlighted. This collective information is then used as a counterpoint to screen the validity of two of the prevailing hypotheses of drug action of artemisinins and synthetic peroxides, namely i. 'the C-radical hypothesis' wherein the peroxide undergoes 'bioactivation' by ferrous iron to generate C-radicals that are held to be the cytotoxic agents and ii. the 'heme hypothesis' wherein ferrous heme may generate either the same type of 'cytotoxic' C-radical, or the peroxide forms heme adducts that apparently inherit the exquisite cytotoxicities of the parent peroxide in one way or another. In a subsequent review, we screen the third and fourth hypotheses: the SERCA hypothesis wherein artemisinins modulate operation of the malaria parasite sarcoendo plasmic reticulum calcium pump SERCA Ca(2+)-ATPase ATP6 and the co-factor hypothesis wherein artemisinins act as oxidant drugs through rapidly oxidizing reduced conjugates of flavin cofactors, or those of flavin cofactor precursors such as riboflavin, and other susceptible endogenous substrates that play a role in maintaining intraparasitic redox homeostasis. For the C-radical hypothesis, details of in vitro chemical studies in the context of established chemistry of C-radicals and their ability to react with radical trapping agents such as nitroso compounds, cyclic nitrones, persistent nitroxyl radicals and atmospheric oxygen (dioxygen) are summarized. Overall, there is no correlation between antimalarial activities and abilities of the derived C-radicals to react with trapping agents in a chemical flask. This applies in particular to the reactions of C-radicals from artemisinins and steroidal tetraoxanes with the trapping agents vis-a-vis those from adamantyl capped systems. In an intraparasitic medium, it is not possible to intercept C-radicals either through use of a vast excess of a nitroxyl radical or dioxygen. The lack of correlation of antimalarial activities also applies to the Fe(2+)-mediated decomposition of artemisinins and synthetic peroxides, where literature data taken as indicating otherwise are critically assessed. The antagonism to antimalarial activities of artemisinins exerted by desferrioxamine (DFO) and related Fe(3+)-chelating agents is due to formation of stable chelates with bioavailable Fe(3+) that shuts down redox cycling through Fe(2+) and the subsequent generation of reactive oxygen species (ROS) via the Fenton reaction. The generation of ROS by Fe(2+) complements the action of artemisinins, to be discussed in Part 2; there is no need to posit a reaction of Fe(2+) with the artemisinins to account for their antimalarial activity. The ability of artemisinins and synthetic peroxides to elicit membrane damage is examined in the light of established processes of autoxidation. The oxidant character of the intraparasitic environment is incompatible with the reducing conditions required for generation of C-radicals, and in contrast to the expectation raised by the C-radical hypothesis, and indeed by the heme hypothesis outlined below, antimalarial activities of artemisinins are enhanced under higher partial pressures of dioxygen. Structure-activity data from a wide variety of artemisinins and synthetic peroxides cannot be accommodated within the bounds of the C-radical hypothesis. Finally, the antimalarial Cradical construct sharply contrasts with that of the potently antitumour-active ene-diyne antibiotics such as neocarzinostatin. In an iron-free process, these compounds generate highly reactive aryl C-radicals that abstract H atoms from deoxyribose units in DNA to generate alkyl C-radicals. The last do react with dioxygen in a normal intracellular environment to initiate DNA strand cleavage. Overall, it must be concluded that the C-radical hypothesis as the basis for antimalarial activities of artemisinins and synthetic peroxides is untenable. Heme has been intensively studied as an 'activator' of artemisinins and other antimalarial peroxides, and indeed the hypothesis seemingly has become firmly embedded in the underlying brickwork of the scientific edifice. The locus of activity of the peroxides interacting with the heme is considered to be the parasite digestive vacuole. The basis for the nanomolar activities of artemisinins and synthetic peroxides is variously ascribed to heme-Fe(2+)-mediated generation of C-radicals from the peroxides, formation of heme-artemisinin adducts that are held either to engage in redox cycling with concomitant generation of ROS or to inhibit formation of hemozoin. In the last case, just like the aminoquinolines and arylmethanols, the peroxides are not the active agents, but exert their parasiticidal effects through allowing the build-up of free heme-Fe(3+), the ultimate cytotoxic entity. We assess the literature relating to generation of heme by hemoglobin digestion, and the stage at which this process becomes significant in the intraerythrocytic parasite. The claims of production of heme and conversion into hemozoin occurring in a lipid environment may have to be put aside based on recent literature data that indicates crystallization of hemozoin must take place an aqueous interface; association of lipids with the heme/hemozoin is likely to be a reflection of attractive van der Waals interactions involving the hydrophobic surface of the heme or hemozoin aggregates. In addition, the observation leading to the claim that hemozoin manufacture commences at the mid-ring stage cannot be independently verified. That the quinoline and arylmethanol antimalarials have essentially no activities on the ring stage parasites and exert greatest efficacy at the trophozoite stage where heme production is maximal is consistent with this. Conversely, artemisinins, and indeed redox active drugs such as methylene blue and others, are highly active against early ring stage parasites. Thus, there is a prominent disconnect between stage specificities of artemisinins vis-a-vis those of 4-aminoquinolines and arylmethanols suggesting that heme is not the target of the former class of drug. Further, the ability of the Fe(3+) chelate DFO to antagonize antimalarial activities of artemisinins, but not the activities of 4-aminoquinolines, cannot be explained by involvement of heme as a target for artemisinins. We critically examine the basis for formation of products obtained from reaction of heme with artemisinins and synthetic peroxides under conditions ranging from biomimetic - reactions employing catalytic reagents under aqueous or semi-aqueous conditions - to those conducted under highly reducing and eminently artificial conditions, usually in the solvent dimethyl sulfoxide (DMSO) that both forms well characterized complexes with heme-Fe(2+) and actually assists in driving single electron transfer processes. It is noted that alkylated products tend to form in high yields under the last conditions, and this aspect is readily explained. Irrespective of product yields obtained under various conditions, an overarching correlation between facility of the reaction of the peroxide with heme and their antimalarial activities does not exist. The is underscored by the reproducible outcomes of reactions conducted under biomimetic conditions indicating adducts cannot form in physiologically meaningful concentrations and that heme is a recalcitrant reaction partner to artemisinins in general. Again, as in the case of the C-radical hypothesis, structure-activity data from a wide variety of artemisinins and synthetic peroxides is difficult to reconcile with the heme hypothesis. This applies in particular to dimeric and trimeric artemisinin derivatives where the ascribing of biological activity to reactions of the derived radicals or to the vastly encumbered artemisinin-heme adducts is physically unrealistic. Finally, the facile metabolism and induction of metabolism of the current clinically used artemisinins by members of the CYP superfamily - heme proteins that require an intimate interaction of the heme with the artemisinin for metabolism to occur - is incompatible with the oft-cited proclivity of the peroxide to associate via complex formation with heme as a prelude to its 'activation' as an antimalarial agent within the malaria parasite. (ABSTRACT TRUNCATED)