Enantioselective Synthesis and in Vivo Evaluation of Regioisomeric Analogues of the Antimalarial Arterolane

Abstract

We describe the first systematic study of antimalarial 1,2,4-trioxolanes bearing a substitution pattern regioisomeric to that of arterolane. Conformational analysis suggested that trans-3″-substituted trioxolanes would exhibit Fe(II) reactivity and antiparasitic activity similar to that achieved with canonical cis-4″ substitution. The chiral 3″ analogues were prepared as single stereoisomers and evaluated alongside their 4″ congeners against cultured malaria parasites and in a murine malaria model. As predicted, the trans-3″ analogues exhibited in vitro antiplasmodial activity remarkably similar to that of their cis-4″ comparators. In contrast, efficacy in the Plasmodium berghei mouse model differed dramatically for some of the congeneric pairs. The best of the novel 3″ analogues (e.g., 12i) outperformed arterolane itself, producing cures in mice after a single oral exposure. Overall, this study suggests new avenues for modulating Fe(II) reactivity and the pharmacokinetic and pharmacodynamic properties of 1,2,4-trioxolane antimalarials.

Introduction

Despite recent progress in the control and treatment of malaria, this devastating disease still affects millions around the world and was estimated by the World Health Organization to have caused 429,000 deaths in 2015. The blood stage of infection is responsible for symptomatic disease, which is characterized by cyclical rounds of asexual replication of Plasmodium spp. parasites within host erythrocytes. To support its rapid proliferation in this stage of infection, the parasite depends on the catabolism of host hemoglobin as a source of amino acids. However, the consequent production of toxic free ferrous iron heme during this process represents an Achilles heel that is targeted in different ways by multiple classes of approved antimalarials, including quinolines and artemisinins.

The sesquiterpene artemisinin (qinghaosu, 1) and its analogues dihydroartemisinin (DHA) and artesunate are employed in artemisinin-based combination therapies (ACT), the current WHO recommended front line therapy for uncomplicated malaria caused by Plasmodium falciparum, the most virulent human malaria parasite. Artemisinins exhibit a novel pharmacology that requires initial activation of a hindered endoperoxide bond by reduced iron sources in the parasite. The iron-dependence and likely involvement of carbon-centered radical intermediates in artemisinin pharmacology were first revealed in the early 1990s in elegant mechanistic work involving synthetic artemisinin derivatives.^1,2 Soon, more synthetically accessible chemotypes were identified that, like the artemisinins, bore a peroxide bond in a sterically hindered environment. Most notably, Vennerstrom and co-workers identified the 1,2,4-trioxolane-based pharmacophore that produced both arterolane³ (OZ277, 2a) and artefenomel⁴ (OZ439, 3) (Figure 1). Clinical development of arterolane by Ranbaxy led to its approval in India and some African countries as a combination with piperaquine. Artefenomel was subsequently selected for clinical development on the basis of its much superior Fe(II) stability and in vivo PK/PD properties⁵ and is currently in Phase 2 clinical trials.⁶

Figure 1.

Structures of antimalarial endoperoxides.

Recent chemoproteomic studies⁷⁻⁹ from two different groups have confirmed the long-hypothesized pleiotropic action of the artemisinins and demonstrated⁷ a remarkable concordance in the proteins covalently targeted by artemisinins and synthetic 1,2,4-trioxolanes. Hence, artemisinin and 1,2,4-trioxolane-derived affinity probes were shown to irreversibly label numerous proteins in the parasite, including those involved in energy acquisition, antioxidant response, and protein and DNA synthesis. Yet despite having many potential targets in the parasite, resistance to artemisinins has emerged in Southeast Asia and is now a significant clinical problem, especially with concomitant resistance to partner drugs.¹⁰ The resistant phenotype results from mutations in the propeller domain of the PfKelch13 (K13) protein, which shares homology with the mammalian Keap1 protein.¹¹ These mutations appear to confer an enhanced stress response and a prolonged ring stage that allows K13 mutant parasites to better sustain the potent but short-lived insult conferred by artemisinins.¹²

A number of groups have now explored whether K13 mutations confer resistance also to antimalarial 1,2,4-trioxolanes, employing in vitro assays that focus on the kinetics of ring-stage killing. Thus, Tilley and co-workers¹³ studied killing kinetics of DHA, 2a, and 3 against a clinical K13 mutant strain and a genetically matched K13 revertant strain bearing the wild-type K13 allele. For all three agents, the time to reduce early ring viability by 50% (t^∧₅₀) was more than doubled in the K13 mutant strain compared to that in the revertant parasites. Thus, DHA and 2 exhibited t^∧₅₀ values of ∼3 and ∼1 h for the K13 mutant and revertant strains, respectively, whereas 3 exhibited slower killing with t^∧₅₀ values of ∼5 and ∼2 h. Using a ring-stage survival assay and a larger panel of K13 mutants, Fidock and co-workers¹⁴ found that DHA and 2a were uniformly less effective against the K13 strains when compared to revertant or wild-type controls. In contrast, compound 3 showed cross-resistance to only one of the K13 mutants examined (I543T). Finally, Wittlin and co-workers¹⁵ studied clinical (2a, 3) as well as additional preclinical trioxolane analogues against a K13 clinical isolate, finding that the synthetic trioxolanes were superior to DHA in a ring stage survival assay. Although the findings of these three studies are in some cases at odds, they are in general agreement that structurally distinct endoperoxide antimalarials can exhibit divergent effects on K13 mutant rings and that the superior in vivo exposure profile of 3 in humans predicts for efficacy in patients with K13 mutant parasites. Early data⁶ from the clinical investigation of 3 suggests this is indeed the case.

The findings summarized above provide essential guidance for the optimization of endoperoxide antimalarials in the current environment of K13-mediated resistance. New agents should ideally confer rapid killing of wild-type and mutant K13 parasites while retaining a prolonged in vivo exposure profile that ideally enables single-dose therapy.^9,13 Among currently available agents, 3 meets this pharmacokinetic (PK) target, whereas 2a and current artemisinins do not due to shorter durations of exposure in vivo. The identification of novel 1,2,4-trioxolanes that combine rapid ring-stage killing kinetics with superior pharmacokinetics will be expedited by new approaches for the control of endoperoxide reduction/activation by Fe(II). Here, we provide data suggesting that trans-3″-substitution of the cyclohexane ring, like canonical cis-4″ substitution, modulates Fe(II) reactivity of 1,2,4-trioxolanes in a pharmacologically relevant regime in vivo. Comparing congeneric sets of trans-3″ and cis-4″ analogues, we furthermore demonstrate that trans-3″ analogues can exhibit in vivo PK/PD properties distinct from cis-4″ comparators. Finally, we describe an efficient and stereocontrolled synthesis of trans-3″-substituted 1,2,4-trioxolane analogues that will enable further lead optimization studies of this chemotype.

Results and Discussion

Design and Synthesis

The Fe(II) reactivity and resulting antimalarial effects of 2a and 3 can be rationalized in terms of the conformational dynamics of the cyclohexane ring. In ground state conformer I (Scheme 1), the axial endoperoxide lies in the concave face of an aliphatic surface with approach to the σ* orbital of the O–O bond effectively shielded by the four proximal axial hydrogen atoms of the adamantane and cyclohexane rings (Scheme 1). It is the peroxide-equatorial conformer II that exposes the O–O bond for inner-sphere coordination with Fe(II) and single-electron transfer leading to bond scission.¹⁶ Consistent with this interpretation is the highly regioselective nature of peroxide cleavage with reaction at the oxygen atom opposite the adamantane moiety predominating (O¹, Scheme 1).¹⁶ The crucial role of cyclohexane ring conformation on Fe(II) reactivity allows the latter to be finely tuned by varying the nature and stereochemistry of the 4″ substituent (i.e., R⁴ in Scheme 1).¹⁶ The optimal balance of Fe(II) reactivity and antiplasmodial effects was ultimately achieved in analogues bearing cis-R⁴ substitution that stabilizes unreactive conformer I. A focus on cis-R⁴ substitution during lead optimization efforts ultimately produced both of the clinical candidates from this class (2a and 3). Moreover, the improved stability of 3⁴ as compared to 2a toward endogenous Fe(II) sources can be understood as arising from more severe 1,3-diaxial interactions in conformer II of compound 3 due to the bulkier aryl R⁴ side chain. Importantly, 3 still retains sufficient reactivity with free Fe(II) heme in parasites to confer a potent antimalarial effect.

Scheme 1. Conformational Effects on Fe(II) Reactivity in Antimalarial 1,2,4-Trioxolanes.

The ability to predict Fe(II) reactivity on conformational grounds is very appealing from a design perspective and may prove essential in finding the right balance of in vivo stability toward endogenous Fe(II) sources (principally labile iron) and reactivity with Fe(II) heme necessary to target both susceptible and resistant parasites. We considered that trans-3″ substitution, nearly¹⁷ unexplored previously, should modulate conformational dynamics in an analogous fashion as canonical cis-4″ substitution (R³ vs R⁴, Scheme 1). In the case of trans-R³ analogues, however, 1,3-diaxial interactions in conformer II would involve both O and H atoms as opposed to only H atoms in cis-R⁴ analogues. Depending on the nature of the R³/R⁴ side chain then, distinct conformational dynamics of the cyclohexane ring might be expected, and this would confer distinct in vitro and in vivo activities. Other important properties such as solubility, metabolism, and clearance were also expected to be affected by the switch to trans-R³ substitution, which eliminates the internal symmetry present in 2a and 3, producing a distinct topology.

The arterolane (2a) scaffold was selected for initial studies because a late-stage amide coupling reaction would allow ready access to the matched R³/R⁴ analogue pairs. As comparators, we selected several R⁴-side chain derivatives described in patent applications¹⁸ and reported to possess reasonable in vivo efficacy with oral dosing. Thus, 2a and nine additional R⁴-side chain analogues (2b–j) were synthesized¹⁹ from acid intermediate 4(17) and commercially available amines (Scheme 2). We then turned our attention to construction of the analogous R³-side chain variants.

Scheme 2. Synthesis of Arterolane and Related cis-4″ Analogues from Key Intermediate 4.

Reagents and conditions: (a) ethyl chloroformate, Et₃N, CH₂Cl₂, −10 °C, 75 min; R(R′)NH, CH₂Cl₂, −10 °C to rt, 2–24 h, 71–95%. Reaction conditions adapted from ref (19).

Unlike in analogues 2a–j, which possess internal symmetry and are achiral, R³ substitution leads to four possible stereoisomers. Because trans-R³ stereochemistry was desired based on our conformational analysis, this left the trans-(R,R) and trans-(S,S) enantiomers as potential candidates for evaluation. We arbitrarily selected the trans-(R,R) stereoisomers for this initial study. Previously, we found²⁰ that the Griesbaum co-ozonolysis reaction of 3-substituted cyclohexanones proceeds with high (∼90:10 dr) intrinsic diastereoselectivity and favors the desired trans stereoisomer. Preparing the desired trans-(R,R) stereoisomer then required starting from an appropriate nonracemic 3-cyclohexanone. Toward this end, cyclohexanone 5 was prepared in 87% yield via catalytic asymmetric Michael addition of dimethyl malonate to 2-cyclohexen-1-one (Scheme 3).²¹⁻²³ Notably, this very practical synthesis of 5 has been performed previously on kilogram scales in a premanufacturing setting.²³ We confirmed the high enantiomeric purity of 5 (95% ee) by conversion to ketal 6,^24,25 the two diastereomers of which could be readily distinguished by ¹³C NMR spectroscopy.

Scheme 3. Enantiocontrolled Synthesis of Ketone 7.

Reagents and conditions: (a) dimethyl malonate, (R)-ALB (1 mol %), t-BuOK (0.9 equiv relative to ALB), 4 Å MS, THF, rt, 68 h, 87%; (b) NaOH, H₂O/THF (11:1), 0 °C, 2 h; (c) DMSO, 160 °C, 4 h, 85% over two steps.

A Krapcho decarboxylation was next considered as a means to produce the desired ester 7(26) from 5. However, for the possibility of epimerization via retro-Michael/Michael addition of dimethyl malonate to be avoided, a two-step procedure was utilized instead.²⁶ Thus, hydrolysis of 5 with NaOH at 0 °C afforded the mono-carboxylate intermediate, which was immediately heated to 160 °C to facilitate decarboxylation. This afforded 7 in 85% yield over two steps. Conversion of 7 to ketal 8 confirmed that the hydrolysis/decarboxylation sequence had proceeded without erosion of enantiomeric purity (94% ee for 7). Griesbaum co-ozonolysis of 7 with two equivalents of oxime 9 at 0 °C using our optimized conditions²⁰ afforded 1,2,4-trioxolane intermediate 10 in nearly quantitative yield. Hydrolysis of ester 10 with NaOH then furnished the desired carboxylic acid 11 in excellent yield. From 11, the desired 3″ amides 12a–k were prepared in 68–95% yields via formation of the mixed anhydride (ethyl chloroformate and Et₃N at −10 °C) and subsequent reaction with primary or secondary amines (Scheme 4). We were able to confirm that the Griesbaum co-ozonolysis of 7 and 9 had proceeded with good diastereoselectivity (91:9 dr) by the preparation of analogue 12k in which trans and cis diastereomers could be readily distinguished by NMR.²⁰

Scheme 4. Stereocontrolled Synthesis of 1,2,4-Trioxolane Analogues 12a–k.

Reagents and conditions: (a) 0.5 equiv of 7, O₃, CCl₄, 0 °C, 3 h, 95%; (b) NaOH, EtOH/H₂O, 50 °C, 4 h, 95%; (c) ethyl chloroformate, Et₃N, CH₂Cl₂, −10 °C, 75 min; R(R′)NH, CH₂Cl₂, −10 °C to rt, 2–24 h, 68–95%.

In Vitro Evaluation of Congener Pairs

The effect of R³ substitution on antiplasmodial activity was evaluated by comparing the regioisomeric analogues 2a–j and 12a–j (Chart 1). New trans-R³ analogues exhibited potent, low nM activities against the chloroquine-resistant W2 strain of P. falciparum with activity comparable to that of cis-R⁴ comparators. Moreover, structure–activity trends tracked remarkably closely in the regioisomeric scaffolds consistent with our hypothesis that trans-R³ substitution should confer similar conformational constraints and Fe(II) reactivity as cis-R⁴ substitution. In both scaffolds, piperidine amides (12e and 2e) were the least potent, approximately 10- to 20-fold weaker than piperazine analogues 12g and 2g. The most significant difference between regioisomer pairs was observed for morpholine amides 12f and 2f, where 3″ analogue 12f was 3-fold more potent than that of 2f. Overall, these findings confirmed that trans-R³ substitution modulates 1,2,4-trioxolane reactivity in a pharmacologically relevant range, producing effects on cultured parasites that are comparable to those of canonical cis-R⁴ substitution.

Chart 1. In Vitro Activity of Trioxolanes 12a–j and 2a–j against W2 P. falciparum Parasitesa.

^a In vitro activity of 12a–j and 2a–j against W2 P. falciparum parasites (EC₅₀ ± SEM). Reported EC₅₀ values are the means of three determinations ± SEM

A second hypothesis regarding trans-R³ substitution was that the resulting desymmetrized molecular topology would impact the in vivo PK/PD properties of these analogues. Specifically, we reasoned that the chiral trans-R³ analogues might interact differently with CYP enzymes and drug transporters (e.g., P-gp). To explore possible metabolic differences, we evaluated selected analogues with mouse liver microsomes and derived intrinsic clearance (CL_int) values (Table 1). All the analogues evaluated exhibited low or moderate CL_int values that would predict reasonable exposure in vivo. In four of the six analogue pairs evaluated, it was the trans-R³ analogues that exhibited higher CL_int values with trans-R³ analogues 12a and 12c showing lower intrinsic clearance than their cis-R⁴ comparators. Clearance appeared to be CYP mediated in all cases, with the possible exception of 2c, where chemical instability or non-CYP-mediated clearance was implicated. The kinetic solubilities of arterolane (2a) and its regioisomer 12a were evaluated and found to be comparable. Overall, the in vitro ADME data suggested that trans-R³ analogues were suitable for evaluation in animals and that specific analogues might be differentiated from their comparators in vivo.

Table 1. In Vitro ADME Data for Selected Trioxolane Analogues and Controls.

cmpd	T1/2 (min)a	CLintb	T1/2 (min) no NADPH	solubility (μM)c
2a	128	5.4	stable	433
12a	169	4.1	stable	429
2b	124	5.6	stable
12b	37.3	18.6	stable
2c	64.2	10.8	70
12c	84.5	8.2	136
2d	161	4.3	stable
12d	64.2	10.8	stable
2g	48.1	14.4	88.9
12g	25.5	27.2	stable
2i	277	2.5	stable
12i	84.5	8.2	stable
midazolam	1.65	420
diclofenac	55.5	12.5
amiodarone				<3
testosterone				315

^aHalf-life when incubated with mouse liver microsomes.

^bClearance in units of μL min^–1 mg^–1 of protein calculated as CL_int = ln(2)*1000/T_1/2/protein concentration, where protein concentration is in mg/mL; midazolam and diclofenac served as controls.

^cKinetic solubility in PBS (pH 7.4) at a final DMSO concentraion of 5% as an average of three determinations; amiodarone and testosterone served as controls.

The suppressive 4-day “Peters test”²⁷ involving P. berghei infected mice provides a convenient and cost-effective means to assess the overall in vivo performance of preclinical antimalarials. Thus, groups of five P. berghei infected female Swiss Webster mice were treated via oral gavage with four daily (QD) doses of either 2a (as the tosylate salt), regioisomeric comparator 12a (both tosylate and free base), or chloroquine control (Table 2). Mice were followed for 30 days and judged to have been cured of the infection based on the lack of detectable parasitemia at the end of the study. In this initial study, trans-R³ analogue 12a as either tosylate salt or free base exhibited efficacy comparable to arterolane tosylate, producing cures in all five animals.

Table 2. In Vivo Efficacy of Trioxolanes 12a and 2a and Controls in P. berghei-Infected Mice Treated for 4 Days^a.

treatment	salt form	dose (mg kg–1 day–1)	mice curedb (%)
2a	tosylate	13.6	100
12a	tosylate	13.6	100
12a	free base	9.5	100
chloroquine		30	80
vehicle treated			0
untreated			0

^aBeginning 1 h after infection, cohorts of five P. berghei-infected female Swiss Webster mice were treated once a day for 4 days by oral gavage.

^bMice were considered cured if there was no detectable parasitemia at 30 days postinfection.

This result indicated that 12a was orally absorbed and achieved systemic exposure sufficient to produce a robust pharmacodynamic response. Prior to evaluating additional trans-R³ analogues, we sought to determine the 50% curative dose for both 12a and 2a in this model and use this dose as a benchmark for future studies. Thus, compounds 12a and 2a (both as free bases) were administered to mice over 4 days at doses of 10, 6, 4, or 1 mg kg^–1 day^–1, and survival and parasitemia were monitored until 30 days postinfection. This study revealed a clear dose–efficacy relationship and allowed us to estimate 50% curative dose (PD₅₀) values of ∼6 mg kg^–1 day^–1 for 12a and ∼4 mg kg^–1 day^–1 for 2a (Table 3). With the trans-R³ chemotype validated in vivo and a relatively stringent dose/efficacy benchmark established, we set out to more broadly evaluate the matched pairs of regioisomeric analogues in animals.

Table 3. In Vivo Efficacy of 12a and 2a in P. berghei-Infected Mice Following Four Daily Doses^a.

treatment	dosea (mg kg–1 day–1)	mice curedb (%)
12a	1	0
	4	0
	6	60
	10	100
2a	1	0
	4	80
	6	100
	10	100
vehicle		0
untreated		0

^aBeginning 1 h after infection, cohorts of five P. berghei-infected female Swiss Webster mice were treated once a day for 4 days by oral gavage at the indicated daily dose.

^bMice were considered cured if there was no detectable parasitemia at 30 days postinfection.

As noted previously, the cis-R⁴ amide comparators were selected in part based on in vivo efficacy data in patents and other reports²⁸ from the Vennerstrom group. Therefore, all nine cis-R⁴ analogues were expected to have good in vivo prospects, whereas the in vivo performance of the trans-R³ comparators was less certain. In the next study then, all 20 test compounds (2a–j and 12a–j) were evaluated under 4 and 6 mg/kg daily dosing protocols in groups of five Swiss Webster mice. Controls 12a and 2a again showed good efficacy at these low doses with analogue 12a marginally more effective than 2a in this study (Table 4). Among the other nine analogue pairs, four pairs failed to cure any mice at either dosing level. These analogues included the glycinamide (12b and 2b), piperidine (12e and 2e), morpholine (12f and 2f), and N-methyl piperazine (12h and 2h) analogues. Interestingly, whereas the piperidine and piperazine analogues failed in these studies, the structurally very similar 4-aminopiperidine derivatives 2i and 12i were among the most efficacious analogues examined, curing all animals at the lowest dose of 4 mg kg^–1 day^–1 and proving superior to 12a and 2a. This finding regarding 2i is consistent with a previous report²⁸ demonstrating the superiority of 2i over 2a in a similar murine model with 3 days oral dosing at 3 mg/kg. Thus, for many of the analogue pairs, in vivo efficacy seemed to track between the cis-R⁴ and trans-R³ comparators.

Table 4. In Vivo Efficacy of Matched Analogue Pairs in P. berghei-Infected Mice^a.

compd	dose (mg kg–1 day–1)	mice curedb (%)
12a	4	60
	6	100
2a	4	20
	6	80
12b	4	0
	6	0
2b	4	0
	6	0
12c	4	100
	6	100
2c	4	0
	6	20
12d	4	0
	6	0
2d	4	40
	6	100
12e	4	0
	6	0
2e	4	0
	6	0
12f	4	0
	6	0
2f	4	0
	6	0
12g	4	0
	6	20
2g	4	100
	6	100
12h	4	0
	6	0
2h	4	0
	6	0
12i	4	100
	6	80
2i	4	100
	6	100
12j	4	60
	6	100
2j	4	80
	6	80
chloroquine	30	40
vehicle		0

^aBeginning 1 h after infection, cohorts of five P. berghei-infected female Swiss Webster mice were treated once a day for 4 days by oral gavage.

^bMice were considered cured if there was no detectable parasitemia at 30 days postinfection.

Of greatest interest were those analogue pairs for which in vivo efficacy varied between the trans-R³ and cis-R⁴ comparators. Most strikingly, cis-R⁴ analogues 2d and 2g were fully effective at 6 mg kg^–1 day^–1, whereas the analogous trans-R³ comparators 12d and 12g failed entirely or were minimally effective (Table 4). In the case of the analogue pair 12c and 2c, however, it was the novel trans-R³ analogue 12c that produced full cure at either dose, whereas the cis-R⁴ derivative 2c was essentially ineffective. Thus, regioisomeric trioxolane analogues can in some cases exhibit starkly different efficacy in vivo despite having similar intrinsic potency against cultured parasites. These results confirmed our hopes that trans-R³ substitution might offer new avenues for the optimization of PK/PD properties in preclinical 1,2,4-trioxolanes. In the case of analogue pairs 12c and 2c, 12d and 2d, and 12g and 2g, CL_int values from the mouse liver microsome assay correctly predicted the superior analogue in each pair (compare Tables 1 and 4). However, the magnitude of differences in CL_int values seems insufficient to fully explain the differences observed, nor do the CL_int values alone explain the failure of 12b and 2b. It appears that additional in vivo PK/PD studies will be required to fully understand the relative merits of 3″ and 4″ substitution within the context of specific side chain types.

Given that the novel trans-R³ analogues 12c and 12i demonstrated complete cures at 4 mg kg^–1 day^–1, we next explored whether even lower doses of these agents might be effective in mice. Although 12c and 12i again cured all animals at a 4 mg kg^–1 day^–1 dose, neither produced cures with reduced doses of 2, 1, or 0.5 mg kg^–1 day^–1 (Table 5). We next asked whether a full 4 days of treatment at 4 mg kg^–1 day^–1 was necessary to achieve cures with 12c and 12i, cognizant that current antimalarial target product profiles seek agents with shortened dosing regimens. When 12c and 12i were administered at 4 mg/kg for 4, 3, 2, or 1 day, a predictable decline in efficacy with the number of doses was observed (Table 6). Thus, both compounds were again fully effective at four daily doses of 4 mg/kg, and 12i was shown to be superior to 12c with three daily doses of 4 mg/kg (60 vs 20% of animals cured for 12i and 12c, respectively). Neither compound was curative with one or two daily doses of 4 mg/kg, although under these regimens mouse survival was extended ∼3–6 days compared to that of vehicle-treated control animals.

Table 5. In Vivo Efficacy of 12c and 12i in P. berghei-Infected Mice at Various Dosing Levels^a.

compd	dose (mg kg–1 day–1)	mice curedb (%)
12c	0.5	0
	1	0
	2	0
	4	100
12i	0.5	0
	1	0
	2	0
	4	100

^aBeginning 1 h after infection, cohorts of five P. berghei-infected female Swiss Webster mice were treated once a day for 4 days by oral gavage.

^bMice were considered cured if there was no detectable parasitemia at 30 days postinfection.

Table 6. In Vivo Efficacy of 12c and 12i in P. berghei-Infected Mice with Less Frequent Dosing^a.

compd	dose (mg/kg)	number of doses	mice curedb (%)
12c	4	4	100
	4	3	20
	4	2	0
	4	1	0
12i	4	4	100
	4	3	60
	4	2	0
	4	1	0

^aBeginning 1 h after infection, cohorts of five P. berghei infected female Swiss Webster mice were treated once a day by oral gavage for either four, three, two, or 1 day as shown above.

^bMice were considered cured if there was no detectable parasitemia at 30 days postinfection.

In a final in vivo study, we evaluated the efficacy of 12c and 12i following a single dose of 40 or 80 mg/kg or two doses of 40 mg/kg (Table 7). We included as a positive control in this study artefenomel (3),^4,29 which is known to be highly effective in these models. Indeed, a single dose of 3 was remarkably effective, curing all animals at both doses examined. Although the new analogues 12c and 12i were completely effective with two doses of 40 mg kg^–1 day^–1, a single dose of 40 mg/kg was only partially effective for 12i and ineffective for 12c. Both compounds were partially effective with a single 80 mg/kg dose and again 12i proved superior to 12c (60 vs 20% cures). Thus, multiple PD studies consistently revealed analogue 12i to be the most promising of the novel trans-3″ analogues explored herein, although it was inferior to artefenomel (3), which has demonstrated⁴ single-dose cures at 20 mg/kg in similar models.

Table 7. Efficacy of 12c, 12i, and Artefenomel in P. berghei-Infected Mice Following a Single or Repeated Dose^a.

compd	dose (mg/kg)	number of doses	mice curedb (%)
3 (artefenomel)	40	1	100
	80	1	100
12c	40	2	100
	40	1	0
	80	1	20
12i	40	2	100
	40	1	20
	80	1	60

^aBeginning 1 h after infection, cohorts of five P. berghei-infected female Swiss Webster mice were treated by oral gavage for either 1 or 2 days at the indicated dose.

^bMice were considered cured if there was no detectable parasitemia at 30 days postinfection.

Conclusions

The unusual Fe(II)-dependent pharmacology of antimalarial 1,2,4-trioxolanes necessitates an equally unusual approach to their optimization, namely, one that recognizes and exploits the strong connection between conformational dynamics, chemical reactivity, and antiparasitic effects.¹⁶ Clinical experience with 2a and 3 perfectly illustrates this point. Thus, early clinical studies of 2a revealed inferior drug exposure and half-life in malaria patients when compared to healthy volunteers. This effect was traced to reaction of 2a with endogenous Fe(II) sources in infected individuals. Nevertheless, by modifying the nature of the cis-4″ side chain, a superior drug candidate (3) with much improved Fe(II) stability and in vivo PK properties was identified.⁴ Now, an improved understanding¹³⁻¹⁵ of how emergent K13 mutant parasites escape endoperoxide exposure provides rationale for identifying new endoperoxides that more rapidly kill P. falciparum K13 mutant ring forms while retaining stability toward endogenous Fe(II) sources in the host. The result of just such an effort focused on a tetraoxane chemotype has appeared very recently.³¹

Here, we present the first systematic study comparing the canonical cis-4″-substituted pharmacophore of 2a and 3 with regioisomeric trans-3″-substituted analogues. On the basis of their in vitro and in vivo properties, we conclude that the trans-3″ side chain modulates peroxide reactivity in a pharmacologically relevant regime as hoped. In this preliminary study, we examined just ten side chains employed previously in cis-4″ analogues. Even with this limited survey, two novel analogues were identified that exhibited in vivo properties superior to 2a in the P. berghei model, and one that afforded single-dose cures at higher doses. The ultimate potential of the new 3″-substituted chemotype will only be revealed with the examination of a more diverse set of side chains, and in particular, those designed to exploit steric and conformational effects unique to this substitution pattern. Efforts in this direction are well underway in our laboratories and will be described in due course.

Experimental Section

Known compounds were prepared according to literature procedures as cited in the main text. The syntheses of new compounds 12a–k are described in the Supporting Information. All compounds tested in parasites or mice were judged to be of >95% purity as assessed using a Waters Micromass ZQ 4000 equipped with Waters 2795 Separation Module, Waters 2996 Photodiode Array Detector (254 nm), and Waters 2424 ELS detector. Separations were carried out with an XBridge BEH C18, 3.5 μm, 4.6 × 20 mm column, at ambient temperature (unregulated) using a mobile phase of water–methanol containing a constant 0.10% formic acid.

Plasmodium falciparum EC₅₀ Determinations

The growth inhibition assay for P. falciparum was conducted as described previously³⁰ with minor modifications. Briefly, Plasmodium falciparum strain W2 synchronized ring-stage parasites were cultured in human red blood cells in 96-well flat bottom culture plates at 37 °C, adjusted to 1% parasitemia and 2% hematocrit under an atmosphere of 3% O₂, 5% CO₂, and 91% N₂ in a final volume of 0.1 mL per well in RPMI-1640 media supplemented with 0.5% Albumax, 2 mM l-glutamine, and 100 mM hypoxanthine in the presence of various concentrations of inhibitors. Tested compounds were serially diluted 1:3 in the range 10000–4.6 nM (or 1000–0.006 nM for more potent analogues) with a maximum DMSO concentration of 0.1%. Following 48 h of incubation, the cells were fixed by adding 0.1 mL of 2% formaldehyde in phosphate buffered saline, pH 7.4 (PBS). Parasite growth was evaluated by flow cytometry on a FACsort (Becton Dickinson) equipped with AMS-1 loader (Cytek Development) after staining with 1 nM of the DNA dye YOYO-1 (Molecular Probes) in 100 mM NH₄Cl and 0.1% Triton x-100 in 0.8% NaCl. Parasitemias were determined from dot plots (forward scatter vs fluorescence) using CELLQUEST software (Becton Dickinson). EC₅₀ values for growth inhibition were determined from plots of percentage control parasitemia over inhibitor concentration using GraphPad Prism software.

Plasmodium berghei Mouse Malaria Model

Female Swiss Webster Mice (average of 20 g body weight) were infected intraperitoneally with 10⁶Plasmodium berghei-infected erythrocytes collected from a previously infected mouse. Beginning 1 h after inoculation, the mice were treated once daily by oral gavage for 1–4 days with 100 μL of solution of test compound formulated in a vehicle comprised of 10% DMSO, 40% of a 20% solution of 2-hydroxypropyl-β-cyclodextrin in water, and 50% PEG400. There were five mice in each test arm. Infections were monitored by daily microscopic evaluation of Giemsa-stained blood smears starting on day 7. Parasitemia were determined by counting the number of infected and uninfected erythrocytes. Body weight was measured over the course of the treatment. Mice were euthanized when parasitemia exceeded 50% or when weight loss of more than 15% occurred. Animal survival and morbidity were closely monitored for up to 30 days postinfection when the experiment was terminated.

Glossary

Abbreviations Used

DHA

dihydroartemisinin

ACT

artemisinins combination therapy

WHO

World Health Organization

PK

pharmacokinetics

PD

pharmacodynamics

CL_int

intrinsic clearance

MLM

mouse liver microsome assay

QD

once-daily dosing

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jmedchem.7b00699.

• Synthetic procedures and compound characterization for new compounds (PDF)

• Molecular formula strings (CSV)

Author Contributions

B.R.B., J.G., and A.R.R conceived experiments. B.R.B. devised synthetic routes and synthesized compounds. B.R.B. and J.G. acquired data. All authors analyzed data. A.R.R. supervised the project. B.R.B. and A.R.R wrote the manuscript. All authors reviewed, edited, and commented on the manuscript. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.