Potent Antiplasmodial Derivatives of the Antitussive Drug Dextromethorphan Reveal the Ent-Morphinan Pharmacophore of Tazopsine-Type Alkaloids

: The alkaloid tazopsine 1 was introduced in the late 2000's as a novel antiplasmodial hit compound active against Plasmodium falciparum hepatic stages, with potential to develop prophylactic drugs based on this novel chemical scaffold. However, the structural determi-nants of tazopsine 1 bioactivity, together with the exact definition of the pharmacophore, re-mained elusive, impeding further development. We found that the antitussive drug dextromethorphan (DXM) 3 , although lacking the complex pattern of stereospecific functionalization of the natural hit, was harboring significant antiplasmodial activity in vitro despite suboptimal prophylactic activity in a murine model of malaria, which precluded its direct repurposing against malaria. The targeted N -alkylation of nor -DXM 15 delivered a small library of analogues with greatly improved activity over DXM 3 against P. falciparum asexual stages. Amongst these, N - 2’ -pyrrolylmethyl- nor -DXM 16i showed a 2- to 36-fold superior inhibitory potency compared to tazopsine 1 and DXM 3 against parasite liver and blood stages, with 760 ± 130 nM and 2.1 ± 0.4 µM IC 50 values, respectively, as well as liver/blood phase selectivity of 2.8. Furthermore, cpd. 16i showed a 5 to 8-fold increase of activity relatively to DXM 3 against P. falciparum stages I-II and V gametocytes, with 18.5 µM and 13.2 µM IC 50 values, respectively. Cpd. 16i can thus be considered a promising novel hit compound against malaria in the ent morphinan series with putative pan-cycle activity, paving the way for further therapeutic development (e. g., investigation of its prophylactic activity in a mouse model of malaria). incubated serial dilutions of or 2 72 cell evaluated and by measuring luciferase activity for 1 s on plate The tests on 96-well plates were performed in tripli-cates.

old, 2 principally from the deadliest and Africa-prevalent Plasmodium falciparum parasite species. 3 Malaria begins with the bite of a Plasmodium-infected female Anopheles mosquito, which injects sporozoites into the skin of the mammalian host. Sporozoites readily travel into the bloodstream, traverse a few liver cells and finally home into a hepatocyte. Once inside the host cell, sporozoites actively replicate and turn into multinucleated hepatic schizonts. At the end of the hepatic phase (i. e., 2-14 days after initial invasion, depending on the Plasmodium species), schizont and host cell rupture, releasing thousands of merozoites into the blood stream. Merozoites invade red blood cells inside which they actively replicate leading to schizonts, releasing merozoites which re-infect other erythrocytes in an exponential fashion. The symptomatology of malaria is directly associated with the parasite developmental phase in the blood, which is the principal target of most antimalarial drugs. 4 Semisynthetic artemisinin derivatives (ARTDs), associated to longer half-life companion drugs in Artemisinin Combination Therapies (ACT), exert fast curative action against parasite blood stages and remain the frontline antimalarials prescribed worldwide. However, ARTDs are threatened by the rapid spread of artemisinin-resistant P. falciparum strains in South-East Asia 5 and their independent emergence recently in Africa, [6][7][8] which manifest a delayed clearance phenotype under conventional drug regimens. 9 This phenomenon is a worrying continuum of the history of chemoresistance by malaria parasites, which most exclusively affects drugs targeting the parasite blood phase. Indeed, Plasmodium erythrocytic phase is characterized by important parasitemia and high mutations rates, allowing the selection of drug-resistant mutants. 4,10 On the other hand, the initial asymptomatic hepatic phase of parasite development features lower parasitemia and consequently lower mutation events, being thus considered an attractive target for malaria chemoprophylaxis. Drugs possessing novel chemical scaffolds active against parasite hepatic stages, either selectively 11 or as part of a pan-active mode of action, [12][13] are therefore strongly pursued in drug discovery programs.
Tazopsine 1, a novel ent-morphinan alkaloid isolated from the endemic Malagasy plant Strychnopsis thouarsii (Figure 1), induces low micromolar inhibition of P. falciparum liver and blood stages in vitro, but is insufficiently prophylactic at subtoxic dose in mouse infected by P. yoelii (70 % protection at 100 mg/kg). On the other hand, the semisynthetic derivative N-cyclopentyltazopsine 2 is 10-fold less active but 15fold more selective than 1 towards liver stages in vitro. 2 is also less toxic than 1, enabling full prophylaxis at 200 mg/kg in the aforementioned malaria mouse model. 14 Despite patents filled in 2004 and 2006, these unprecedented antimalarial hits were not further investigated due to the difficult bio-sourcing of 1, its complex chemical structures from total synthesis viewpoints, limited structure-activity relationships (SAR) [14][15] as well as absence of identified pharmacophore and molecular targets in the series. Interestingly, antiplasmodial properties are shared by other entmorphinan alkaloids, [15][16] suggesting the existence of a common pharmacophore irrespective of substitution or stereochemical patterns. Based on this rational, we identified the generic antitussive drug dextromethorphan 3 (DXM, alias 3-methoxy-17methylmorphinan) as possibly integrating the essential functional features of these alkaloids (Figure 1), having in mind its direct repurposing against malaria or its use to access simplified mimics of bioactive ent-morphinan alkaloids. The present paper describes the results of these endeavors.

Extended SAR in the tazopsine series.
We present here pharmacomodulation efforts towards novel derivatives of tazopsine 1. Altogether, these SARs drove the validation of DXM 3 as a general and simplified mimick of all natural ent-morphinan alkaloids, and subsequently guided its chemical diversification into optimized antiplasmodial derivatives. Tazopsine 1 was treated with excess diazomethane to give the 4-methyl phenol ether 6 with 39 % yield. The native alkaloid was in parallel submitted to reductive amination with various aldehydes in presence of sodium cyanoborohydride, to deliver tertiary amines 7a-g with 45-82 % yields. To assess the influence of a basic nitrogen on the antiplasmodial activity, N-acetyl-tazopsine 8 was produced by treating tazopsine 1 with acetic anhydride, with 46 % yield ( Figure 2). Primary mouse hepatocytes (PMH) infected by the murine parasite P. yoelii were used to assess the bioactivity of the generated tazopsine derivatives (Table 1), allowing direct comparison with previously generated SAR in the series. 14-15 Benzylic 10substitution (ring B) SAR: comparison of tazopsine 1, 10-epi-tazopsine 4 and sinococuline 5 to assess the effect of benzylic 10-substitution showed that the 10-(R)-hydroxy pattern of tazopsine 1 was optimal, its 10-epimer 4 being 5-fold less active. On the other hand, a completely unsubstituted pattern proved only slightly less beneficial, showed by the comparable bioactivities of tazopsine 1 and sinococuline 5. Aryl 4-O-substitution (ring A) SAR: alkylation of tazopsine 1 free phenol abolished the antiplasmodial activity, with 4-O-methyl-tazospine 6 showing unsignificant inhibitory effects against the parasite at 100 µM. 17-N-substitution (ring D) SAR: tertiary amine derivatives of tazopsine 1 showed loss of activity with increase of substituent size (N-methyl-tazopsine 7a already been 2-fold less active than the parent alkaloid) but this trend was mitigated by the beneficial N-4'-halo-benzyle substituents in analogues 7f and 7g, exhibiting similar levels of inhibition than tazospine 1. This suggests that specific substituents can favorably impact on the antiplasmodial activity of N-modified ent-morphinans in spite of relative bulkiness, a trend already observed for N-cyclopentyltazopsine 2 (IC50 = 3.5 ± 0.1 µM). 14 On the other hand, N-acetyl-tazopsine 8 was completely devoid of activity, suggesting that the presence of a nitrogen atom either basic (i. e., protonated at physiological pH values) or capable of engaging donating hydrogen bonds was important for the antiplasmodial properties. Cyclohexenol/enone/dienone (ring C) SAR: comparison of tazopsine 1 with sinoacutine 9 and sinomenine 10 revealed that the southern portion of these alkaloids exerted a profound influence on their bioactivity. Indeed, only the 6,7-dihydroxy-8,14-methylenol moiety of tazopsine 1 correlated with strong antiplasmodial effects, while the distinctive methoxy-enone/dienone systems present in 9 and 10 led to abolished activity ( Figure 2). In conclusion of these SAR in the ent-morphinane series, it appeared that the benzylic substitution at C-10 in ring B had to be either (R)-hydroxyl (as in tazopsine 1) or non-existent (as in sinococuline 5 and DXM 3). The ring A in tazopsine 1 seems to be a sensitive part to modify considering the loss of activity of 4-O-methyltazopsine 6, albeit this punctual variation should preclude premature conclusions. N-alkylation in the tazopsine series consistently appeared as a relevant pharmacomodulation at ring D with frequent conservation of antiplasmodial activity. Last, SAR regarding the southern ring C of ent-morphinans remains unconclusive except for the restricted benefit of that present in tazopsine 1, a fact corroborated by the previous description of bioactivity loss of the tazopsine-6,7-acetonide. 14 Table 1. Antiplasmodial activity of tazopsine derivatives 6-8 and of natural ent-morphinan alkaloids against P. yoelii liver stages. NA: non-applicable.

Cpd.
Substitutions IC50 (Py265BY-PMH, µM) Tazopsine  DXM 3 is a well-known antitussive drug also used as pain reliever, as well as a dissociative anesthetic and hallucinogen in recreational uses. [17][18] Its pharmacology in the central nervous system (CNS) is well established. Despite being considered a synthetic opiate, DXM 3 does not act at the level of opioid receptors, binding instead with high affinity to sigma receptors as an agonist and to a lesser extent to the phencyclidine channel of N-methyl-D-aspartate (NMDA) receptors as an antagonist. 18 The relationship between the neuropharmacology and the antitussive effects of DXM 3 is poorly understood. 17 The similarity of DXM 3 with tazopsine 1 in term of entmorphinan backbone prompted us to evaluate its antiplasmodial properties both in vitro and in vivo, having in mind its possible direct repurposing against malaria. In a preliminary screen against primary human hepatocytes (PHH) infected by P. falciparum in vitro, DXM 3 exhibited an activity that was only 2-fold less than that of tazopsine 1 (Table 2). Noteworthily, the IC50 value of tazopsine 1 was slightly above that previously described of ca. 4 µM in this biological model. 14 This observation can be explained by the shift in the PHH used for P. falciparum culture from clinical samples in the previous study to standardized, commercially available cryopreserved PHH in the present work. The whole assay was validated by the expected submicromolar activity of the reference drug primaquine (PQ). 14 Following the exciting discovery of the antiplasmodial activity of DXM 3 against P. falciparum liver stages in vitro, we tested its prophylactic potential in a P. bergheiinfected mouse model of malaria. We used the transgenic parasite line P. berghei-Luc infecting BALB/c mice to follow parasitemia in situ based on the spontaneously emitted bioluminescence after injection of D-luciferine. 19 Following a preliminary study, we found that DXM 3 induced convulsive episodes and death in mouse at doses superior to 60 mg/kg administered daily (data not shown). Therefore, we decided to use a subtoxic regimen of 40 mg/kg DXM 3 administered daily starting 24 h before infection and further maintained for a 48 h period, corresponding to the duration of P. berghei liver phase. In addition, we chose to associate DXM 3 with quinidine (QND), a known inhibitor of CYP2D6-mediated O-demethylation of DXM 3 into its main hepatic metabolite dextrorphan 12 (DX, syn. 3-hydroxy-17-methylmorphinan), 18 as a way to increase the plasmatic half-life of DXM 3 20-21 and possibly increase its antimalarial effect in vivo. DXM 3 at 40 mg/kg and QND at 20 mg/kg were deprived of prophylactic activity in this in vivo model, with parasitemia levels not significatively different between these groups and the vehicle (Figure 3). While the combination of DXM 3 and QND at a daily regimen of 40 mg/kg and 20 mg/kg, respectively, exerted significative prophylactic inhibition of P. berghei-Luc growth, we found that this synergistic combination was far from eliciting complete parasite clearance as did the reference drug primaquine biphosphate at a daily regimen of 5 mg/kg ( Figure 3). The extend of DXM 3 metabolization in vivo into DX 12 (theoretically inhibited by QN) and to a lesser part into nor-DXM 15 (syn. 3-hydroxy-17-methylmorphinan) by CYP3A4-mediated N-demethylation 18,22 remains uncharacterized in our study. If effective and yielding inactive metabolites, its occurrence could explain the relatively poor prophylactic activity of DXM 3 in this mouse model of malaria. However, this outcome proved to be invalid as both DX 12 and nor-DXM 15 were later synthesized ( Figure 4) and found to have similar level of inhibitory potency in vitro than DXM 3 against P. berghei ( Figures 5 & 6). Despite the unlikeliness of its repurposing as a prophylactic drug against malaria, DXM 3 represents a readily accessible and flexible synthetic platform to explore the antimalarial potential of the ent-morphinan series and circumvents the inherent limitations of the tazopsines.

DXM pharmacomodulation towards improved antiplasmodial derivatives.
DXM 3, possessing the same backbone as the natural hit tazopsine 1, constituted the synthetic starting point of our study. This compound combines two advantages for a SAR study: (i) straightforward and cheap access from various commercial suppliers in gram quantities, and (ii) easy functionalization on the C-2 and/or O-and/or N-positions ( Figure 4). To explore yet unraveled SAR on the aryl ring, DXM 3 was firstly o-iodinated using NIS to deliver 2-I-DXM 11 with a 91 % yield. DX 12 was obtained as described by Jakobsson et al. 23 by the O-demethylation of DXM 3 with 48 % aqueous HBr, then o-iodinated with NIS in the previous conditions to yield 2-I-DX 13 with a 87 % yield ( Figure 4). As previously observed with tazopsine 1, derivatization of the nitrogen position revealed to be a relevant modification with retainment of the antiplasmodial activity (Table 1) and possible improvement of the antimalarial profile. 14 Therefore, N-modification of nor-DXM 15 was decided to be fully explored. Towards this aim, DXM 3 free base was N-demethylated via the trichloroethylcarbamate intermediate 14 as described by Jozwiak et al. 24 to give nor-DXM 15. Reductive amination, particularly using sodium triacetoxyborohydride (STABH) as reductant, represents a mild and chemoselective method for the N-alkylation of primary and secondary amines. 25 Tertiary amines 16a-m were thus synthesized from nor-DXM 15 using n-alkyl, cycloalkyle, heteroaryle and bis-heteroaryle aldehydes in presence of STABH with moderate to good yields (33-99%) ( Figure 4). To decipher the influence of the protonation state of the nitrogen atom on the antiplasmodial activity of entmorphinans -taking into account that tazopsine 1, N-cyclopentyltazopsine 2, DXM 3, DX 12, nor-DXM 15 and its N-alkyle derivatives 16a-m are to be fully protonated at physiological pH values -two compounds 17 and 18 with a neutral or constitutively positive charge on the nitrogen atom, respectively, were synthesized. Cpd. 17, corresponding to the exact amide congener of the amine 16d for the sake of precise SAR comparison, was prepared by reacting nor-DXM 15 with cyclopropanecarbonyle chloride in presence of triethylamine, with excellent yield. On the other hand, the quaternary ammonium 18 was obtained in high yield by reacting DXM 3 free base with methyl iodide (Figure 4). In vitro pre-screening against P. berghei liver stages.
The antimalarial activity against liver stages of the obtained 21-compound library was first pre-screened using PHH infected by P. berghei expressing the Green Fluorescent Protein (PHH-Pb-GFP). This model has the advantage of being less expensive and more accessible than the hepatic stages of P. falciparum, thanks to routine production of P. berghei sporozoites in the laboratory, while using the same host cell than for the human parasite and therefore being representative of parasite-host cell interactions. Compound activity was assessed by two criteria, (i) the number of parasites developing within PHH (Exo-Erythrocytic Forms, EEFs, Figure 5A) and (ii) the size of parasites (µm 2 ) normalized on untreated controls ( Figure 5B). False positives, due to compound toxicity against host PHH, were excluded by normalizing parasite number on nuclei number of untreated controls. At the highest concentration (20 µM), viability of PHH was > 80 % for the less active compounds and 50-60 % for the most active ones. Four activity profiles could be categorized from the 21-compound library screening in terms of EEFs number ( Figure 5A), namely: (i) inactive compounds (2-I-DX 13, 14, 16f, 16h, 16j, 17, 18); (ii) low-activity compounds with similar inhibitory potency than DXM 3 (DX 12, Nor-DXM 15, 16b, 16c, 16d, 16e, 16k and 16m); (iii) active compounds (2-I-DXM 11, 16a, 16g, and 16l); (iv) one very active compound (16i). These activity ranges were respectively characterized by (i) a high EEFs number at 20 µM (the maximal concentration of the range), (ii) a low EEFs number at 20 µM, (iii) a low EEFs number at 10 µM and finally (iv) a low EEFs number at 1 µM. We observed the same trend in the plot of parasite size ( Figure 5B) with a delineation for cpd. 16i exhibiting an important size effect at 1 µM, followed by cpds. 11, 16c, 16e, 16g, and 16l at 10 µM. In the light of these results, we focused on molecules possessing an inhibitory activity between 1-10 µM and established an amplified cut-off test at the arbitrary concentration of 7 µM to validate the above-described categories of inhibitors ( Figure 6). This evaluation permitted to identify which compounds inhibited liver stage parasite development (e. g. leading to a 50 % reduction in EEFs number normalized on the drug-free controls. Five molecules were thus validated in the cut-off test, i. e. cpds. 16c, 16d, 16i, 16l and 17 ( Figure 6A). In addition, 16e was halfway and susceptible to have interesting activity. A low EEFs size effect was observed for cpds. 16g, 16h, 16i, 16l and 16m but none of these molecules inhibited parasite size by at least 50 % ( Figure 6B). Regarding substitution of DXM 3 or DX 12 with iodine on the C-2 position, only 2-I-DXM 11 exhibited a gain of potency relatively to the parent compound while 2-I-DX 13 was inactive ( Figure  5). Interestingly, amine 16d and its amide congener 17 displayed similar levels of inhibitory activity against P. berghei liver stages ( Figures 5 and 6). This suggested that the presence of an electro-donating doublet on the nitrogen atom rather than that of a protonated one was an important feature for the interaction of ent-morphinans with their biochemical target(s). However, this behavior was contradictory with that observed in the tazopsine series where N-acetylation abolished the activity (Table 1). Nevertheless, this trend in the DXM series was reinforced by a lower activity of the quaternary ammonium 18 compared to DXM 3 ( Figure 5). This showed that in spite of the intrinsically charged state of most bioactive DXM derivatives at physiological pH (with exception of cpd. 17), the presence of a permanently charged nitrogen atom was detrimental to the antiplasmodial activity.  In vitro screening against P. falciparum liver and asexual blood stages.
, which displayed the best inhibitory effects in the P. berghei pre-screening, were selected for screening against both liver and blood stages of the human parasite P. falciparum and determination of their IC50 values. PQ and chloroquine (CQ) were used as reference drugs against liver and blood stages, respectively. Compound toxicity against host PHH was excluded by normalizing parasite number on nuclei number of untreated controls. At 10 µM, viability of PHH was ca. 100 % for the less active compounds and 60-70 % for the most active ones. The observed gain of antiplasmodial activity of 2-I-DXM 11 compared to DXM 3 against P. berghei liver stages was confirmed against P. falciparum, the latter being 4-fold more active than the parent compound (Table 3). Regarding the pre-screened N-modified analogues 16a, 16g, 16i and 16l, these were active against P. falciparum liver stages in the low micromolar/submicromolar range and displayed significantly lower IC50 values than both parent compounds, i. e. the natural hit tazopsine 1 and DXM 3 (Tables 1 and 3). The most active compound was the N-2'pyrrolylmethyl derivative 16i which strongly inhibited the development of parasite liver stages, being 10-fold more active than tazopsine 1 (Table 1) and 20-fold more active than DXM 3 (Table 3). Strikingly, cpd. 16i showed a superimposable inhibitory potency to the antimalarial drug PQ against P. falciparum liver stages (IC50 values of 0.76 ± 0.13 µM and 0.75 ± 0.15 µM, respectively). Cpd. 16i was also very active on the parasite blood stages (IC50 values of 2.1 ± 0.4 µM), again similarly to PQ with reported ca. 1-20 µM IC50 values against the P. falciparum 3D7 strain. [26][27] Besides 16i, only 16l showed activity in the low micromolar range against P. falciparum blood stages (IC50 of 6.5 ± 0.4 µM). Other N-alkylated derivatives exhibited low activity against blood stages, similarly to DXM 3 with IC50 values of 43-62 µM. However, all compounds were found to be selective for the hepatic phase of parasite development (2.8-36-fold selectivity) including DXM 3 (4.9-fold selectivity) ( Table 3). To explore the activity profile of the new lead cpd. 16i against other developmental stages of P. falciparum , we tested it against early (stage I-II) and late (stage V) gametocytes, the last being responsible for transmission of P. falciparum malaria. Cpd. 16i was found to be fairly effective against both sexual stages with activities in the high micromolar range, whereas DXM 3 was completely inactive (IC50 > 100 µM) (Figure 7).

Conclusions
This section is not mandatory but can be added to the manuscript if the discussion is unusually long or complex.
Despite the intensive search for novel antimalarial drugs, malaria remains a major disease in tropical regions. The rapid spread of artemisinin resistant P. falciparum strains in South-East Asia 5 and independent increase of k13 polymorphisms in Africa, [28][29] which culminated in clinical artemisinin-resistance recently detected in the continent, 6-8 is a dramatic continuum of the drug resistance history of malarial parasites. This situation underscores the need for alternative chemotherapeutic strategies beyond pursuing novel drugs to eliminate blood parasites, prone to readily acquire resistance mutations. In this context, the novel antimalarial alkaloid tazopsine 1, introduced in the late 2000's, is active against both pre-erythrocytic and erythrocytic P. falciparum stages and the precursor of the in vivo prophylactic compound N-cyclopentyltazopsine 2. However, development of the tazopsines by bio-sourcing or synthetic strategies appeared difficult. Aiming to overtake these limitations, we managed to simplify the natural alkaloids tazopsine 1 -the same principle applying to the related sinococuline 5 -and to extract a "naked" ent-morphinan antiplasmodial pharmacophore under the form of the antitussive drug DXM 3. In particular, substitutions at the level of rings B and C proved non-essential while those affecting ring A (particularly at C-2) of potential relevance. In spite of limited potential for a repurposing against malaria, DXM 3 exhibited a significant level of bioactivity in vitro that was only 2-fold lower than that of tazopsine 1 against P. falciparum liver stages. Capitalizing on strengthened ring D SARs, a rapid diversification of DXM 3 into N-modified derivatives readily led to improved analogues. Amongst those, the hit cpd. N-2'-pyrrolylmethylnor-DXM 16i exhibited a 10-fold and 20-fold increase of activity against P. falciparum liver stages relatively to tazopsine 1 and DXM 3, showing similar activity than the reference drug primaquine against parasite liver and blood stages. In the context of prophylactic antimalarial drug discovery, cpd. 16i was more selective for the parasite liver phase than tazopsine 1 (S = 2.8 vs 0.5) in addition to a stronger bioactivity than the natural alkaloid.
Moreover, cpd. 16i showed a significant effect on the viability of stage I-II and V P. falciparum gametocytes compared to the completely inactive DXM 3. These results warrant further mechanistic investigation of the new hit cpd. 16i regarding causal prophylaxis (i. e., early sterilization of sporozoites) and possibly pan-activity against multiple parasite stages. In addition, the benefit of C-2 substitution suggests an entry to optimized derivatives in the hit series by means of ring A modifications.

Reagents, solvents and equipment.
Reagents and anhydrous solvents were purchased from Merck-Sigma and were of the highest grade available. DXM 3 hydrobromide monohydrate, sinoacutine 9, sinomenine 10, primaquine biphosphate (PQ), chloroquine biphosphate (CQ) and D-luciferin were purchased from Merck-Sigma. Tazopsine 1, 10-epi-tazopsine 4 and sinococuline 5 were isolated from the plant Strychnopsis thouarsii as described previously. 14-15 DX 12 and Nor-DXM 15 were synthesized according to literature procedures. [23][24] Column chromatography was performed using silica gel 60 (9385 Merck). Thin layer chromatography (TLC) was performed on aluminum plates coated with silica gel 60F254 (554 Merck), visualized with UV light (254 and 366 nm) and revealed with sulfuric vanillin or phosphomolybdic acid reagents. NMR spectra ( 1 H and 13 C) were recorded on an Advance Bruker 400 MHz spectrometer or an Oxford Instruments 600 MHz spectrometer equipped with a BBI 600 MHz probe, using solvent signal as an internal standard General procedure for the reductive animation of tazopsine 1 (cpds. 7a-g). N-alkyltazopsine derivatives were obtained from tazopsine using classical reductive amination of 37 % aqueous formaldehyde or pure aldehydes by NaBH3CN. 30 Briefly, a stirred solution of tazopsine 1 free base (34 mg, 0.097 mmol) in anhydrous MeOH (600 L) was primed by a gentle stream of argon for 15 s. To this solution were added the aldehyde (0.107 mmol) at r. t. followed after 10 min by NaBH3CN (95 %, 6.4 mg, 97 µmol). The mixture was stirred at r. t. under argon for 24 h. After removal of the solvent under reduced pressure, the residue was acidified with 1 M HCl, then basified with aqueous NH3, and dried under vacuum. The residue was purified by silica gel column chromatography eluted with CH2Cl2/MeOH (0-10 v/v containing 20 % aqueous NH3).

DXM 3, C18H25NO (Generation of DXM 3 free base).
A solution of NaOH (2.16 g, 5.4 mmol in 4 mL of H2O), was added at r.t. to a suspension of dextromethorphan 3 hydrobromide monohydrate (2 g, 5.4 mmol) in 8 mL CHCl3 and the resulting mixture was stirred for 30 min at r.t. The organic layer was separated, dried over MgSO4 and filtered. The solvent was evaporated under reduced pressure to give 3 free base as a dense and viscous off-white oil which solidified upon standing (1.45 g, 5.3 mmol, 99 % yield). The analytical data were in accordance with the literature. 24 2-I-DXM 11, (9,13,14)-2-iodo-17-methyl-3-methoxymorphinan, C18H24INO. To a solution of DXM 3 free base (21 mg, 0.077 mmol) in MeCN (1 mL) protected from light at 0 °C was added N-iodosuccinimide (20 mg, 0.089 mmol) and p-toluenesulfonic acid monohydrate (27 mg, 0.14 mmol). The mixture was allowed to warm up to r.t. and stirred overnight. The reaction mixture was treated with water (1.5 mL) and Na2S2O3 1 M (1 mL) and basified with a saturated solution of Na2CO3 to pH 10. The aqueous phase was extracted with CH2Cl2 (4 x 3 mL) and the combined organic phases were dried over Na2SO4 and evaporated to dryness under reduced pressure. The crude product was purified by column chromatography on silica gel using CH2Cl2/MeOH (100:0 to 95:5 v/v containing 1 % NEt3) as eluent. 11 was obtained as a pale orange solid (28 mg, 0.070 mmol, 91 % yield). 1 06 (m, 1H). The analytical data were in accordance with the literature. 23
Parasite maintenance and inhibition assays (by order of appearance in the manuscript).

P. yoelii growth inhibition assays in vitro.
Parasite culture. P. yoelii (265 BY strain) sporozoites were obtained by dissection of infected Anopheles stephensi salivary glands. Primary mouse hepatocytes were isolated as previously described 31 and seeded in eight-well Lab-Tek plastic chamber slides (VWR, Fontenay-sous-Bois, France) previously coated with rat-tail collagen I (BD Biosciences, Le Pont de Claix, France) at a density of 105 cells per well. Mouse hepatocytes were cultured at 37 °C in 5 % CO2 in William's E medium supplemented with 10 % fetal calf serum, 1 % L-glutamine, 1 % sodium pyruvate, 1 % insulin-transferrin-selenium, 1 % nonessential amino acids, and 1 % penicillin-streptomycin (all from Invitrogen, Cergy-Pontoise, France), for 24 h before inoculation of P. yoelii sporozoites (105 per well). Test cpds. were solubilized in DMSO, further diluted in culture medium (equal DMSO concentrations of < 0.3 % per well), and then added to hepatocyte cultures at the time of sporozoite inoculation. 3 h later, after sporozoite penetration into hepatocytes, cultures were washed and further incubated in the presence of each test cpd. Culture medium containing the appropriate cpd. concentration was changed daily until 48 h.
IC50 measurement. Parasites were quantified on the last day of incubation by immunofluorescence analysis following fixation of the cultures with cold methanol and parasite-specific staining by a mouse polyclonal serum raised against the P. yoelii heat shock protein 70 and revealed with FITC-conjugated goat anti-mouse immunoglobulin (Sigma). Parasite numbers were counted under a fluorescence microscope with a 25× light microscope objective. IC50 values, the compound concentration at which a 50 % reduction in the number of parasites was observed, as compared to the number in the DMSO control cultures, were calculated by linear regression using Excel software and derived from three independent experiments, where each concentration was tested in triplicates.

P. berghei growth inhibition assays in vivo.
This study was performed according to Bosson Vanga et al. 32 Six to eight week-old BALB/c female mice were obtained from Janvier CERJ (Le Genest-Saint-Isle, France) and housed in CEF (UMS28, La Pitié-Salpêrière) under pathogen-free conditions with food and water ad libitum. All animal work was conducted in strict accordance with the Directive 2010/63/EU of the European Parliament and Council "On the protection of animals used for scientific purposes". Protocols were approved by the Ethical Committee Charles Darwin N°005 (approval #01736.02). Five mice were used per treatment group. Drugs were administrated on days − 1, 0, +1, and mice were challenged on day 0 by retro-orbital injection of P. berghei (GFP-luc strain) sporozoites (10,000 per mouse). In vivo imaging was performed 44 h post-infection to assess liver stage development. IVIS Spectrum (Caliper Life Science, Hanover, MD, USA) was used to measure luminescence expressed as total flux photon/seconde. Prior to analysis, mice were injected ip with D-luciferin (100 mg/kg), anesthetized with isoflurane and imaged 10 min post-injection. Images were analysed using the living Image 3.0 software (Capiler Life Science, Hanover, MD, USA). Data analysis and statistical analysis using a one-way ANOVA test for multiple comparisons were done with GraphPad Prism 8 statistical Software (GraphPad. Software, San Diego, CA, USA). P. falciparum and P. berghei liver stages growth inhibition assays in vitro.
Parasite culture. Plasmodium liver stages were cultured as described elsewhere (Baron et al. manuscript in preparation). Briefly, cryopreserved primary human hepatocytes were purchased from Lonza Bioscience and Biopredic International (Saint-Grégoire, France). Cells were thawed and seeded in 384-well plates (Greiner Bio-One, Germany) pre-coated with rat-tail collagen I (BD Bioscience, USA). Human hepatocytes were maintained at 37°C in 5 % CO2 in William's E medium (Gibco) supplemented with 10 % fetal clone III serum (FCS, Hyclone), 100 u/mL penicillin and 100 ug/mL streptomycin (Gibco), 5 × 10−3 g/L human insulin (Sigma-Merck), 5 × 10−5 M hydrocortisone (Upjohn Laboratories SERB, France). The next day, cells were overlaid with matrigel (Corning) and medium was then renewed every two days. Four days later, sporozoite were isolated by aseptic hand dissection of salivary glands of P. berghei-GFP 33 or P. falciparum-infected mosquitoes (P. falciparum NF54 strain, obtained from Department of Medical Microbiology, University Medical Centre, St Radboud, Nijmegen, The Netherlands). Matrigel was then removed from hepatocyte culture and 5,000 or 30,000 sporozoites of P. berghei-GFP or, P. falciparum were inoculated to cells before centrifugation at 560 xg for 10 min at RT and incubation at 37°C, 5 % CO2. Drugs were tested in quadruplicate, starting from time of sporozoite addition. 3 h later, infected cultures were covered with matrigel prior to addition of fresh cell culture medium containing the appropriate drug dilutions. Medium, containing drug or not, was renewed on a daily basis, until cell fixation, 48 h and 6 days postinfection with P. berghei and P. falciparum sporozoites respectively.
Immunostaining of P. berghei and P. falciparum liver stages.
Infected cultures were fixed using 4% paraformaldehyde (PFA) for 15 min at r.t. and liver stage parasites were immune-labeled with polyclonal anti-PfHSP70 murine serum revealed with Alexa-Fluor 488-conjugated goat anti-mouse immunoglobulin (Invitrogen). DAPI was used to visualize nuclei.
Parasite enumeration and toxicity assessment using high-content imaging. Upon fixation and immunostaining, cell culture plates were analyzed in order to determine the number and size of the parasites using a CellInsight High Content Screening platform equipped with the Studio HCS software (Thermo Fisher Scientific). Parasite size reduction was calculated on the average object area using the total surface area of each selected object (µm 2 ) using the high content imaging approach described previously. 28 Analysis of compound cytotoxicity was performed by counting host cell nuclei on the DAPI channel. Parasite culture. Chloroquine-sensitive (3D7) P. falciparum strain was obtained from the Malaria Research and Reference Reagent Resource Center (MR4). Parasites were maintained in human erythrocytes (O + , provided by Etablissement français du sang, EFS, France) at 5 % haematocrit, suspended in complete culture medium RPMI 1640 medium supplemented with 25 mM HEPES, 20 mM D-glucose, 25 mM sodium bicarbonate, 0.4 mM hypoxanthine, 5 mM L-Glutamine and 10 % AB human serum. Parasite cultures were kept at 37 °C in a gaseous environment composed of 5 % CO2, 10 % O2 and 85 % N2. The culture medium was changed daily, with control of parasitaemia using light microscopy (Axioskop microscope, ZEISS, Germany) under oil immersion, after fixing thin blood smears with methanol and staining with Diff-Quik TM stain Set (RAL Diagnostics, France).
IC50 measurement. 50% inhibitory concentrations (IC50) determination test was carried out by isotopic 42 h 3 H-hypoxanthine incorporation assays as previously described 34 with minor modifications. Briefly, P. falciparum cultures at ring-stage were highly synchronized by two consecutive treatments with 5 % sorbitol (Merck-Sigma) in PBS (v/v) at 40 h intervals and diluted down to 0.3-0.5 % parasitaemia and 2 % haematocrit. Parasites were dispensed into 96-well plates containing 14 serially diluted concentrations of drug ranging from 0 to 240 µM, and incubated as described above in presence of 5 % 3 H-hypoxanthine (Perkin Elmer, USA) for 42 h. 3 H-hypoxanthine uptake was then evaluated by scintillation counting (Top Count NXT, Perkin Elmer, USA) and results were expressed as the inhibitory concentrations IC50 defined as drug concentrations at which 50 % of 3 Hhypoxanthine incorporation was inhibited compared with drug-free controls. IC50 values were established by non-linear regression with ICEstimator software (http://www.antimalarial-icestimator.net/). [35][36] The tests on 96-well plates were done in triplicates. P. falciparum sexual blood stages.
Parasite culture and gametocyte production. The P. falciparum transgenic line NF54-cg6-Pfs16-CBG99 has been described elsewhere. [37][38] Parasites were cultivated in vitro under standard conditions using RPMI 1640 medium supplemented with 10 % heat-inactivated human serum and human erythrocytes at a 5 % hematocrit. To obtain synchronous asexual stages, parasites were synchronized by the isolation of schizonts by magnetic isolation using a MACS depletion column (Miltenyi Biotec) in conjunction with a magnetic separator, then placed back into culture. After invasion of merozoites, a second magnetic isolation was used for the selection of ring-stage parasites to obtain a tighter window of synchronization. Synchronous production of specific gametocytes stages was achieved by treating synchronized cultures at the ring stage (10-15 % parasitemia, day 0) with 50 mM N-acetylglucosamine (NAG) for 5 days to eliminate asexual parasites. Gametocyte preparations were enriched in different experiments by magnetic isolation.
IC50 measurement. To calculate the IC50 for DXM 3 and N-2'-pyrrolylmethyl-nor-DXM 16i on early and mature gametocytes, 2 × 10 5 MACS-purified early GIE (day 2 post NAG treatment) and mature GIE (days 7 post NAG treatment) from the NF54-cg6-Pfs16-CBG99 line were incubated with serial dilutions of inhibitors or 2 % DMSO for 72 h. After 72h, GIE were washed and cell viability was evaluated by adding a non-lysing formulation of 0.5 mM D-luciferin substrate 33 and by measuring luciferase activity for 1 s on a plate Reader Infinite 200 PRO (Tecan®). The tests on 96-well plates were performed in triplicates.
Author Contributions: The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. ‡These authors contributed equally.

Funding:
We thank the Doctoral School of Paris University (ED 563, MTCI) for financial support (PhD fellowship to AK).