Peroxide Bond-Dependent Antiplasmodial Specificity of Artemisinin and OZ277 (RBx11160)

Abstract

Using nonperoxidic analogs of artemisinin and OZ277 (RBx11160), the strong in vitro antiplasmodial activities of the latter two compounds were shown to be peroxide bond dependent. In contrast, the weak activities of artemisinin and OZ277 against six other protozoan parasites were peroxide bond independent. These data support the iron-dependent artemisinin alkylation hypothesis.

The antimalarial artemisinin contains a pharmacophoric peroxide bond within its 1,2,4-trioxane heterocycle. A long-standing hypothesis (17, 24, 29) to account for the antimalarial specificity of artemisinin is that the peroxide bond undergoes reductive activation by heme released by parasite hemoglobin (Hb) digestion. Although most (95%) of the heme released from parasite digestion is incorporated into the relatively inert hemozoin (12), it is estimated (8) that there is some 100 μM of “free” heme/hematin in malaria-infected red blood cells, a more than adequate quantity to react with artemisinin. The irreversible redox reaction between artemisinin and heme produces carbon-centered free radicals or carbocations that alkylate heme (25, 32, 33) and membrane-associated parasite proteins (2), one of which is the translationally controlled tumor protein (PfTCTP) (4) and another is likely to be PfATP6, a SERCA-type Ca²⁺-ATPase (11). Eckstein-Ludwig et al. (11) studied the latter protein in some detail and provided evidence that it may be an important target of the artemisinins.

The structural diversity of semisynthetic artemisinins (41) and synthetic peroxides (37) is quite remarkable and, we suggest, has implications for the mechanism of action for this class of antimalarial drugs. In this respect, we thought it might be useful to compare the activities of artemisinin and synthetic peroxide (1,2,4-trioxolane) OZ277 (RBx11160) (40) and their nonperoxidic counterparts deoxyartemisinin (6) and carbaOZ277 (Fig. 1) against a range of protozoan parasites. CarbaOZ277 (1,3-dioxolane) was prepared as a mixture of cis-trans isomers by an acid-catalyzed reaction (9) between the requisite adamantane diol and amino amide cyclohexanone. In vitro antiprotozoal assays for Plasmodium falciparum, Trypanosoma brucei rhodesiense, Trypanosoma cruzi, Leishmania donovani, Giardia duodenalis, Babesia divergens, and Neospora caninum were performed as previously described (1, 5, 16, 28, 35, 38, 40) with standard drug controls (Table 1).

FIG. 1.

Structures of artemisinin, deoxyartemisinin, OZ277, and carbaOZ277.

TABLE 1.

Activity of artemisinin, deoxyartemisinin, OZ277, and carbaOZ277 against six protozoan parasites

Compound	IC50 (μg/ml) for indicated strain(s)a
	P. falciparum K1/NF54	T. b. rhodesiense STIB 900	T. cruzi Talahuen C4	L. donovani MHOM-ET-67/L82b	G. duodenalis WB/G1	B. divergens 1903B/4201
Artemisinin	0.002/0.004	25	>30	8.8	65/49	>10/>10
Deoxyartemisinin	>10/>10	34	>30	>30	91/28	>10/>10
OZ277	0.0010/0.00091	0.56	9.5	24	41/49	>10/>10
CarbaOZ277	2.5/4.9	0.77	13	20	38/18	6.7/7.5
Melarsoprol	ND	0.0030	ND	ND	ND	ND
Benznidazole	ND	ND	0.14	ND	ND	ND
Miltefosine	ND	ND	ND	0.20	ND	ND
Albendazole	ND	ND	ND	ND	0.019/0.018	ND
Atovaquone	ND	ND	ND	ND	ND	0.0091/0.0087

^aMean from n values of ≥2; individual measurements differed by less than 50%.

^bThe compounds were tested against L. donovani amastigotes in an axenic assay.

The data in Table 1 shows a clear demarcation in the antiplasmodial activities of artemisinin and OZ277 versus their nonperoxidic analogs deoxyartemisinin (6) and carbaOZ277, confirming the peroxide bond-dependent activity of the former. For artemisinin and OZ277, there were losses of activity greater than 3 orders of magnitude against all of the other protozoa compared to activities against P. falciparum. In contrast, for deoxyartemisinin and carbaOZ277, there were only small differences in activity between P. falciparum and the other protozoa. Only for T. b. rhodesiense was there a significant difference between artemisinin and deoxyartemisinin versus OZ277 and carbaOZ277, suggesting that the latter two compounds may be exerting a structurally specific, but peroxide-independent, inhibition of parasite growth by some unknown mechanism. None of the four compounds was active against N. caninum at concentrations of up to 1.0 μg/ml (data not shown).

The data in Table 1 are consistent with the weak activities reported for artemisinin against protozoa other than plasmodia. For example, artemisinin has 50% inhibitory concentrations (IC₅₀s) ranging from 8 to 35 μg/ml against Leishmania major amastigotes (42) and L. donovani promastigotes (3, 27). In the study by Avery et al. (3), several artemisinin derivatives had IC₅₀ values as low as 0.79 μg/ml; in comparison, their nonperoxidic counterparts had no activity at concentrations of up to 50 μg/ml. This result is consistent with our data showing that deoxyartemisinin was completely inactive against L. donovani amastigotes, suggesting some role for the peroxide bond in the activity of artemisinin against this protozoa species. Artemisinin has an IC₅₀ of 2.3 μg/ml against Toxoplasma gondii in cell culture (18); at 4.0 μg/ml, artemisinin completely eliminates the parasite (30). That this very weak activity of artemisinin against T. gondii is peroxide-bond dependent is suggested by data (18) in which one semisynthetic artemisinin was shown to be 64-fold more potent than its deoxy derivative. Artemisinin at 0.14 μg/ml inhibits the growth of cultured Pneumocystis carinii by 55% (23) and has IC₅₀ values of 3.8 and 5.8 μg/ml against T. cruzi and T. b. rhodesiense in cell culture (27). At doses as high as 400 mg/kg of body weight, artemisinin had no efficacy against Cryptosporidium parvum infections in a neonatal mouse model (14). Nonprotozoans are similarly unaffected by the artemisinins. For example, the semisynthetic artemisinin, artemether, had almost no inhibitory activity against 86 bacterial and 72 fungal strains with MICs ranging from 64 to >512 μg/ml (13).

The iron-dependent artemisinin alkylation hypothesis is supported by the peroxide bond-dependent activity of artemisinin and OZ277 against plasmodia and the minimal and peroxide bond-independent activity of these two compounds against other protozoa. The signal biochemical difference between plasmodia and all other protozoa is that the former degrade Hb and the latter do not. This is most clearly evident for the intraerythrocytic protozoan Babesia, which unlike Plasmodium, does not catabolize Hb (31). It has also been suggested (21) that artemisinin may undergo reductive activation in the mitochondria of P. falciparum and thereby inhibit the growth of the parasite. Indeed, in addition to Hb (19, 26, 34, 43), artemisinin alkylates other hemoproteins, such as cytochrome c (43). However, if the mitochondria were the primary plasmodial target of artemisinin (21), it is hard to explain why all other protozoa are so much less sensitive to artemisinin.

The iron-dependent artemisinin alkylation hypothesis is also supported by the antiplasmodial stage specificity of these drugs, that is, young ring stage malaria parasites are slightly less sensitive to the artemisinins and OZ277 than are the more mature parasites in the trophozoite and schizont stages (22). Hb digestion has been underway for only a short time in the ring stage compared to that in the mature parasite stages, and consequently less heme has been generated. Even so, hemozoin can already be detected in young ring stage parasites (15). Similarly, young gametocytes are more susceptible (20) to artemisinin than are more mature gametocytes, presumably a reflection of the greater heme content (7) of the former. Only the sporozoite, bereft of hemozoin (36), is unaffected by artemisinin (10).

Whether heme or protein alkylation is the more important pharmacodynamic event in the antimalarial activity of artemisinin and OZ277 is unknown. In this regard, it is interesting to note that artemisinin inhibits the putative target PfATP6 with an IC₅₀ of 79 nM, a value 2 orders of magnitude more potent than the one reported for OZ277 (IC₅₀ = 7,700 nM) (39). Deoxyartemisinin at concentrations up to 50 μM does not inhibit PfATP6 (11). It will be informative to ascertain whether PfATP6 homologs in other protozoan parasites are inhibited by artemisinin and deoxyartemisinin. Future studies are likely to reveal new plasmodial targets of the artemisinins and synthetic peroxides.