Choline Analogues in Malaria Chemotherapy

Emerging resistance against well-established anti-malaria drugs warrants the introduction of new therapeutic agents with original mechanisms of action. Inhibition of membrane-based phospholipid biosynthesis, which is crucial for the parasite, has thus been proposed as a novel and promising therapeutic strategy. This review compiles literature concerning the design and study of choline analogues and related cation derivatives as potential anti-malarials. It covers advances achieved over the last two decades and describes: the concept validation, the design and selection of a clinical candidate (Albitiazolium), back-up derivatives while also providing insight into the development of prodrug approaches.


INTRODUCTION
With an estimated 250 million infected people worldwide, malaria is considered to be one of the most lethal human diseases. The majority of these cases concern Africans, including children under 5 years old, and pregnant women are most at risk in terms of both morbidity and mortality [1][2]. Prevention and treatment currently involves both vector control, using a physical barrier between the human host and mosquitoes, and chemotherapy. However, due to the presence of four species of Plasmodium parasites, their geographical distribution, and the potential for the development of resistance to marketed antimalarial drugs, new treatment strategies are required [3][4]. The potential emergence of resistance to artemisinin is one of the major threats, thus highlighting the need for new chemotherapeutic approaches with novel mechanisms of action to treat P. falciparum infections [5][6]. Vial et al. [7][8] developed a therapeutic strategy based on: i) the observation that, to be able to proliferate within human erythrocytes, the parasite must produce a high quantity of phospholipids (PL) to build its membranes, which are essentially composed of phosphatidylcholine (PC) and phosphatidylethanolamine (PE), and ii) the identification of parasite metabolic pathways involved in this PL biosynthesis [9][10]. Whereas the parasite possesses its own enzymatic machinery to synthesize the required PL ( Fig. 1), it relies on its host content for starting materials, including choline (mainly), ethanolamine, serine and fatty acids. It was also demonstrated that choline transport into the infected erythrocyte is one of the limiting steps in de novo PC biosynthesis [11]. Interfering with this crucial step through the use of choline analogues could thus lead to a novel and promising pharmaceutical approach for malaria treatment.

FROM PRIMARY AMINES TO QUATERNARY AMMO-NIUM DERIVATIVES: VALIDATION OF THE APPROACH
The overall project started 15 years ago with the building and study of a library of choline and ethanolamine analogues (some compounds were commercially available while others were obtained in house) [12][13]. Determination of the in vitro antiplasmodial activity against P. falciparum led to the establishment of structure-activity-relationships (SARs) thus facilitating the description *Address correspondence to these authors at the Institut des Biomolécules Max Mousseron (IBMM), UMR 5247 CNRS-UM1-UM2, Université Montpellier 2, place E. Bataillon, 34095 Montpellier, France; Tel: (+33) 4 6714 4964; (+33) 4 6714 3745; Fax (+33) 4 6704 2029; (+33) 4 6714 4286; E-mails: peyrottes@univ-montp2.fr, vial@univ-montp2.fr of essential features concerning molecular requirements to optimize our derivatives and improve their antiplasmodial potency. The screened library (about a hundred compounds) contained primary (PA), secondary (SA) and tertiary (TA) amines, and quaternary ammonium (QA) derivatives (Fig. 2). Most of the primary amines tested were commercially available. Secondary and tertiary amines, and quaternary ammonium derivatives were obtained in high yields by alkylation of primary, secondary, or tertiary amines using the desired alkyl halide in a polar solvent [12].
All choline and ethanolamine analogues were tested in vitro against the growth of P. falciparum, the most virulent human parasite (Tables 1 & 2). Compounds belonging to the family of primary, secondary and tertiary amines ( Table 1, except those including a long alkyl chain, see TA8) exhibited far lower antiplasmodial activities (IC 50 in the millimolar and micromolar range) than those containing a quaternary ammonium group (Table 2), with 50% inhibitory concentration (IC 50 ) ranging from 10 -4 to 10 -8 M. This first observation revealed that the positive charge is crucial for antiplasmodial activity.
Within quaternary ammonium derivatives, the effect of increasing the length of the alkyl chain from 2 to 18 carbon atoms was studied in two series, i.e. containing an hydroxyethyl moiety as in choline or not. Interestingly, the best antiplasmodial activity was observed with an alkyl chain of 12 methylene groups (IC 50 below 1 M) and further addition of methylene groups did not lead to a significant improvement in the IC 50 value ( Table 2, comparison of QA6 with QA7-9). Thus, a long alkyl chain appeared crucial for antiplasmodial activity in the sub-micromolar range and the presence of an aromatic ring in the vicinity of the ammonium head did not markedly influence this activity. This latter observation shows that the targeted site is able to accommodate non-flexible substituents, but the affinity is not improved by using aromatic residues able to generate -interactions instead of lipophilic ones (i.e. in the presence of simple alkyl chains).
Besides this, for compounds with the same alkyl chain length, the presence of an hydroxyethyl group on the nitrogen atom was not essential for high antiplasmodial activity.
For active compounds of the QA series, we also decided to bulk the substituents around the nitrogen atom. Replacement of the three methyl substituents by ethyl or propyl residues thus significantly improved the activity, indicating that an increase in steric hindrance and/or lipophilicity around the nitrogen atom is beneficial for antiplasmodial potency. In the choline analogue series, N,N-diethyl  substitution did not modify the efficacy. Moreover, the presence of a second long alkyl chain (12 carbon atoms) grafted to the nitrogen atom did not improve the IC 50 compared to a methyl substituent, and in some cases was even unfavorable.
SAR analyses have shown that in vitro antiplasmodial activities against the human parasite P. falciparum were related to the shape, electronegativity and lipophilicity of the compound. A positive charge located on the nitrogen and adjustments of shape and size appear to be crucial features for the optimization of these competitive reversible inhibitors. The targeted site likely contains an anionic or a high electron density region able to accept a polar and cationic head (N,N,N-trimethyl, N,N-dimethyl-N-hydroxyethyl, or N,N-diethyl-N-hydroxyethyl). Analogues bulkier than choline (N,N,N-triethyl, N,N,N-tripropyl groups) likely highly interfere with the binding site, however compounds including a second long alkyl chain (C12) appeared too large and are consequently less potent. Clearly, this highlighted that the nature of substituents around the nitrogen atom should be carefully chosen and are essential for optimal interactions. We also hypothesized that a large and non-polar region is located close to the polar head binding site and is able to accept only one long alkyl chain.
HO-(CH2)2--(CH2)5-350 * Overall in vitro biological data were reported in reference [12] and measured after contact with the blood stage for one full cycle (48 h) * Overall in vitro biological data were reported in references [12,14]. Whereas most QA derivatives were isolated as bromine salt, compounds marked with a or b were obtained as chlorinate or iodinate salt, respectively.
In summary, the most potent in vitro activities were obtained for Ndodecyl-substituted ammonium derivatives.

BIS(QUATERNARY AMMONIUM) DERIVATIVES: PHARMACOPHORE ESTABLISHMENT
With the aim of improving the antiplasmodial activity of the previous quaternary ammonium derivatives, a new series of compounds was envisaged by designing duplicated molecules incorporating two polar heads linked by an alkyl chain. Indeed, by combining two pharmocophores as parts of a single molecule, one could expect to obtain a more effective drug that should be capable of interacting simultaneously with more than one targeted site. Furthermore, the chemical bridge (linker in between the two polar heads) itself might also participate in the biological effect. Therefore, symmetrical bis quaternary ammonium salts with various lipophilic substituents on the nitrogen atom and different alkyl chains lengths (3 to 21 methylene groups) were synthesized (Scheme 1) and then studied for their antimalarial potential [12,15].
We were pleased to observe that the studied derivatives exhibited a wide range of antiplasmodial activities (Table 3), with IC 50 values ranging from 10 -4 M to 10 -12 M, i.e. up to 4 orders of magnitude lower than mono quaternary ammonium salts (see above). As previously, we investigated the effects of various modifications on in vitro antiplasmodial activity, such as lipophilicity by increasing the inter-nitrogen chain length or by adding electronegative or electron-rich N-substitutions.
Thus, the structural requirements for antiplasmodial activity of bis(QA) salts were very similar to those of mono(QA) salts, i.e. polar head steric hindrance and lipophilicity around nitrogen (methyl, hydroxyethyl, ethyl, pyrrolidinium, etc…). Within the bis(QA) series, increasing the lipophilicity of the alkyl chain between the two nitrogen atoms (from 5 to 21 methylene groups) constantly improved the activity. Most of these duplicated molecules exhibited activity within the low or sub-nanomolar ranges, and the most lipophilic compound had an IC 50 as low as 3 pM (Table 3, n=21 methylene groups, compound G19). As the polymethylene chain linking the positively charged heads should contain more than 10 methylene groups to obtain bis(QA) derivatives substantially more potent than their corresponding mono(QA)salts, it was suggested that the two targeted sites may be vicinal and separated by an hydrophobic domain. Scheme 1. Generic structures and synthetic approach of the studied bis(QA) derivatives. R 1 = Alkyl or -(CH 2 ) 2 OH R 2 = Alkyl or Bn R 3 = Alkyl Tertiary Amines (TA) (CH 2 ) n X-R 1 = Alkyl or -(CH 2 ) 2 OH R 2 = Alkyl or Bn R 3 = Alkyl or Alkenyl or Alkynyl n= 3 to 21 X= Clor Br -bisQuaternary Amines, bis(QA) Concerning modification of the bulk of the cationic head, a significant difference was observed for N-methylpyrrolidinium derivatives in comparison to their corresponding trimethyl and triethyl analogues ( Table 3, see derivative named G25). This indicates that the pocket is large enough to accommodate such a large head.
On the basis of the overall data for both mono(QA) and bis(QA) salts, we estimated that the site targeted by the cationic head may be viewed as a globular volume of 200 to 350 Å 3 , which corresponds to a radius of ~4 Å. Our results were also consistent with the targeting of two vicinal sites. With alkyl chains longer than 14 methylene groups (20 Å), the distance between the two cationic heads appeared to be optimal for the concomitant interaction with two binding sites of the target (Fig. 3).
To the best of our knowledge, this new family of derivatives appears to be one of the most potent classes of antimalarials in the development stage [16][17][18]. Among currently marketed antimalarials, only halofantrine [19] and artemisinin [20] analogues have exhibited IC 50 values in the low nanomolar range.
Further in vitro biological studies [21][22] were performed and some selected derivatives were tested against various chemoresistant P. falciparum strains or isolates, and they showed the same high activity as against sensitive strains. More specifically, the G25 compound was equally active against the two chloroquine-resistant strains FCR3 and L1 (chloroquine IC 50 0.8 and 0.9 M, respectively) with an IC 50 of 3.6 and 1.4 nM, respectively, while also exhibiting potency (nanomolar range) against various human isolates (such as cycloguanil-, chloroquine-, or mefloquine-resistant isolates).
On the basis of these encouraging results, we then focused on determining the mechanism of action of such derivatives, which were designed as choline analogues in order to interfere with the parasite's phospholipid biosynthesis. In this respect, the relationship between the in vitro P. falciparum growth and the inhibition of the phospholipid metabolism was studied as well as the effect of the compounds on choline transport [21,23]. One prominent feature of the biscationic analog G25 is its ability to accumulate, by several hundredfold, within P. falciparum-infected erythrocytes, which is already significant at the ring stage and increased as the parasite * Overall in vitro biological data were from references [12,15].

Hydrophobic zone
matures. This nonreversible accumulation is likely important with respect to the mechanism of action and appears to determine its specificity and potency [24]. For all 50 derivatives studied, a close correlation between the PL metabolism impairment and the parasite growth inhibition was observed. Some evidence that these compounds act by PL metabolism inhibition was also obtained when an excess of choline was used, which led to reversal of the PC metabolism inhibition. As shown in (Fig. 4), the typical dose-response curves obtained with one bis(QA) salt, namely G25, confirmed that specific inhibition of choline incorporation into PC (PC 50 0.4 M, 4 h incubation) had occurred, whereas inhibition of nucleic acid and PE biosynthesis were observed only at higher concentrations. Furthermore, for the bis(QA) salts, the PC 50 s were 3 to 40 times lower than the NA 50 , indicating specificity with respect to de novo PC biosynthesis. Thus, a positive relationship was established between the ability of the drugs to inhibit parasite growth in culture and their capacity to specifically inhibit phosphatidylcholine biosynthesis of P. falciparum-infected erythrocytes.
The effect of G25 on choline transport was also assessed [21][22] in P. knowlesi-infected RBC and in Saccharomyces cerivisiae cells. The inhibition pattern was sigmoidal, occurring over 1 to 2 orders of magnitude. The G25 concentration leading to 50% inhibition of choline influx was 0.8 M, which corresponds (assuming that inhibition was competitive) to a calculated Ki of 0.4 M [25], which is exactly the PC 50 value. Pretreatment of healthy RBC with parasite-lethal concentrations of G25 did not change the susceptibility of RBC to P. falciparum invasion and their ability to support parasite growth, indicating that the G25 toxic effect did not occur at the host cell level. The time course of in vitro P. falciparum growth inhibition as a function of parasite development was also determined. For the G25 model compound, the best activity was observed for the trophozoite stage in the 48 h erythrocytic cycle (IC 50 0.75 nM), whereas the schizont and ring stages were 12-and 213fold less susceptible. Of upmost importance, the compounds exerted a rapid nonreversible cytotoxic effect with complete clearance of parasitemia after 5 h of contact with the mature stages.
In the meantime, in vivo experiments were carried out [26] in several murine malarial models to demonstrate the full potential of this family of derivatives ( Table 4). Tolerance after administration to mice was first investigated after intraperitoneal (i.p.) administration, and the acute 50% lethal dose (LD 50 ) generally ranged from 0.15 to 50 mg/kg. Toxicity increased with steric hindrance around the nitrogen molecule or lengthened the alkyl chain from 12, 16 to 21 methylene groups.
Three distinct murine models were used to determine in vivo antimalarial activities, i.e. P. berghei-, P. chabaudi-, or P. vinckeiinfected mice. The two latter murine strains differ from the former by their marked preference for invading mature erythrocytes and their high degrees of synchronization [27][28]. Most of the tested compounds were able to clear parasitemia, which is evidence of their antimalarial potencies. Against P. berghei, the ED 50 after i.p. administration (once-daily for 4 days) ranged from 0.04 to 0.3 mg/kg for bis-QA compounds, and even lower against P. chabaudi.
The in vivo selectivity (therapeutic index, TI) increased slightly as the polar head volume decreased. Similar TI values were observed for derivative G25 when the in vivo antimalarial activity was investigated after i.p., s.c., or oral administration. However, there was a 90-fold difference in LD 50 (or ED 50 ) between the i.p. and oral modes, which indicates that the oral absorption of this compound is very low. In fact, oral uptake for such bis cationic derivatives was expected to be weak due to charges.
In conclusion, PL metabolism of P falciparum-infected erythrocytes, especially de novo PC biosynthesis from choline, is thus a quite realistic target for our derivatives. The efficacies of bis(QA) salts appeared to be consistent with the presence of two anionic sites on the choline carrier separated by a large lipophilic domain. This impressive and selective in vitro toxicity against P. falciparum, the high therapeutic index observed in infected mouse models and the absence of recrudescence strongly highlights the clinical potential of these bis(quaternary ammonium) salts for malarial chemotherapy.

MONO-AND BIS(THIAZOLIUM) SALT STUDY: SELEC-TION OF ALBITIAZOLIUM AS CLINICAL CANDIDATE
In the bis(QA) series, the G25 derivative (Fig. 5) appeared to be one of the most active compounds against P. falciparum (IC 50 0.65 nM) and was able to cure, without recrudescence, P. falciparuminfected Aotus monkeys at the very low dose of 0.03 mg/kg [24]. However, this is hampered by the very limited oral bioavailability and toxicity at doses higher than 1.5 mg/kg due to the effect on the cholinergic system (H. Vial, unpublished data). To overcome this problem, we envisaged the replacement of the quaternary ammonium group by a thiazolium ring. This heterocycle presents two main advantages, it is much less toxic and present in a naturally occurring thiazolium derivative, i.e. vitamin B1, and allows the design of neutral prodrugs to improve their bioavailability, as already described for vitamin B1 [29][30][31].
As a follow-up to our previous studies on mono-and bis(QA) compounds, we built-up and studied an oriented library of monoand bis cationic derivatives incorporating thiazolium moiety(ies). All the synthesized derivatives were evaluated for their in vitro antiplasmodial activity on P. falciparum. Among the most active compounds (IC 50 in the low namolar range), several were tested for in vivo inhibition of P. vinckei growth in mice ( Table 5) [32].
The conclusions of the structure-activity relationship (SAR) study were: (i) As for QA-based compounds, duplication of the polar head, when using a dodecyl chain, led to an impressive improvement, i.e. by 2-to 3-log scales of the antiplasmodial activity in comparison to the corresponding monothiazolium derivatives, (ii) the length of the alkyl chain (from C8 to C16) is a crucial element for high antiplasmodial activity. It indicate an optimum for 12 methylene groups (as compared to more than 16 observed for bis(QA)), (iii) the nature of the R 2 substituent at the C-5 position of the thiazolium ring (Fig. 5) also has a strong influence. Briefly, molecules including a small size group (R 2 = Me, Et, (CH 2 ) 2 OH, (CH 2 ) 2 OMe) were more active than those unsubstituted (R 2 = H) or bearing a bulky substituent [R 2 = (CH 2 ) 2 OiPr, (CH 2 ) 2 Cl, (CH 2 ) 2 OCO(CH 2 ) 2 CO 2 Me]. The optimal substituent at the C-5 position was a methoxyethyl group.

Radioactive incorporation
The two most potent compounds (named by us T3/SAR97276 and T4, respectively) were selected for further development on the basis of their in vivo activity by the i.p. route against P. vinckei in infected mice (ED 50 0.2 and 0.14 mg/kg, respectively), their superiority to chloroquine (ED 50 1.1 mg/kg), and their good tolerance (therapeutic index > 50). These compounds were thus evaluated for their pharmacological potency in very severe conditions such as  high parasitemia or in short course treatments with a single injection [33]. After a daily administration of T3/SAR97276 for 4 days, mice were completely cured, with an ED 50 below 0.3 mg/kg using ip, im or iv routes. Remarkably, the ED 50 of T3/SAR97276 was in the same range at low and high initial parasitemia: 0.3 mg/kg and 0.5 mg/kg, respectively. ED 90 and complete cure without recrudescence were obtained at 2-to 4-times the ED 50 . A single injection of T3/SAR97276 at a dosage of 6.75 mg/kg led to a complete cure of P. vinckei-infected mice infected at low or moderate parasitemia, thus highlighting the potency of the compound. T3/SAR97276 was also evaluated in comparison to artesunate, whose remarkable antimalarial activity is associated with a high parasite killing rate [34]. Our lead compound exhibited higher activity by the i.p. route (ED 50 = 0.5 mg/kg) than artesunate (ED 50 >2.5 mg/kg). Moreover, a total cure was obtained with T3/SAR97276 at a daily dose of 0.5 mg/kg for 4 days, this was not obtained with an artesunate dosage of even 10 mg/kg.
In addition, the bisthiazolium salts appeared to be specific inhibitors of malarial phosphatidylcholine biosynthesis [33] and also accumulated to high extent in hematozoan-infected erythrocytes (Plasmodium and Babesia) [35][36]. A substantial share of the accumulated drugs appeared to be localized within the parasite's digestive vacuole and their interaction with heme, the non-protein part of haemoglobin, was suggested to contribute to their potent antimalarial activity [36].
In the light of the overall pharmacological, biological, ADME and toxicological data generated in collaboration with an industrial partner (unpublished), the findings of solubility and formula studies, and the low cost and ease of synthesis, the bis-thiazolium salt T3/SAR97276 was selected as a clinical candidate. Regulatory preclinical and Phase I clinical studies were successfully carried out by Sanofi-Aventis. In 2008, phase II studies were initiated for the treatment of severe P. falciparum malaria via the parenteral route in several research centers in Africa. This open-label, nonrandomized, non-comparative Phase II study was aimed at assessing the antimalarial activity, safety, and pharmacokinetic profile of T3/SAR97276, renamed Albitiazolium, following single and repeated administrations via intravenous (i.v.) and intramuscular (i.m.) routes.
Meanwhile, we have a global research plan to obtain an orally available formulation of T3/SAR97276 and/or a related analogue that could be developed for the treatment of uncomplicated malaria. Indeed, even though bis(thiazolium) salts have already been shown to achieve a complete cure after oral administration, their intestinal absorption appears too low to warrant pharmaceutical development and this aspect needs to be optimized. In this respect, two approaches have been developed in parallel (see below): the first one is based on the design of neutral prodrugs or bioprecursors of the parent drugs, while the second deals with the synthesis and study of Albitiazolium analogues with more favorable molecular parameters (i.e. molecular weight, flexibility and lipophilicity).

EXPLORING PRODRUG APPROACHES FOR ORAL DELIVERY
The derivatives described above exhibit a low oral bioavailability mainly due to the presence of permanent cationic charges, which is crucial for antiplasmodial activity. This prompted us to develop neutral precursors of the corresponding bis-thiazolium salts (Scheme 4), which should be quantitatively and rapidly converted in vivo to the active parent drug after crossing the gastrointestinal barrier [33,37]. The prodrugs studied thus incorporated thioester, thiocarbonate or thiocarbamate pro-moieties, which are expected to be hydrolyzed in vivo to the parent drugs after enzymatic transformation involving plasmatic esterases.
The synthetic pathways developed take advantage of the particular reactivity of the thiazolium ring which can be opened in basic aqueous conditions [37]. Thioester, thiocarbonate or thiocarbamate derivatives were thus obtained in one step by reaction of an alkaline solution of the parent drug (T3 or T4) with the appropriate activated acyl group. A series of 37 neutral precursors was obtained using a low cost synthetic pathway, with yields ranging from 15% to 96%, depending on the nature of the reagents used (i.e. acyl-and arylchloride, alkyl chloroformate or activated carboxylic acid, etc.). Structural variations (lipophilicity, steric hindrance, electronic effect, etc.) affecting the physicochemical properties were made in order to study their influence on oral activity.
All prodrugs were evaluated for their antimalarial activity in vitro against P. falciparum and in vivo against P. vinckei (Tables 6  & 7). Twenty-five of them exhibited potent in vitro antiplasmodial activity, with an IC 50 below 10 nM, indicating quantitative conversion of the prodrugs into their parent drugs. Furthermore, four derivatives were as potent as the parent drugs T3 and T4, by either i.p. and per os (p.o) administration.
In the thioester series (Table 6), the steric hindrance (Me, Et, iPr, tBu, etc.) and/or lipophilicity of the alkyl residue demonstrated a modest effect on antimalarial activity. Introduction of a phenyl group was tolerated, but the presence of hindered substituents on the aryl dramatically increased the IC 50 (from nM to M range) and this lack of in vitro activity was attributed to a slower prodrug-drug conversion rate. Significant differences were observed when decreasing the pro-moiety lipophilicity by introducing polar functional groups (ether, ketone or ester). Notably, when R= (CH 2 ) 2 COMe, the prodrug exhibited an IC 50 of 2.2 nM and an ED 50 of 3 mg/kg per os.
Within thiocarbonate derivatives ( Table 6), the minimal substituent (methoxy group) appeared to be the best one, with an IC 50 of 1.8 nM, and this compound was the most powerful after oral administration, with an ED 50 of 1.3 mg/kg. When using T3 as parent drug ( Table 7), only thioester prodrugs were available due to synthetic problems. Thus similar effects of the nature of the acyl substituents upon biological activity were observed as for the corresponding T4 prodrugs. More interestingly, to generate low molecular weight prodrugs the cyclic thiocarbonate and dithiocarbonate (that should be more stable towards esterase hydrolysis) prodrugs were designed and prepared. Whereas the dithiocarbonate derivative was not active (IC 50 of 4400 nM and ED 50 ip>30 mg/kg), the cyclic thiocarbonate compound showed an in vitro IC 50 = 2.2 nM and the oral activity was enhanced (ED 50 = 5 mg/kg) in comparison to T3. Unfortunately, pharmacokinetics data showed that this compound is rapidly converted into T3 in the gastrointestinal tract. Structural variations of the pro-moieties were designed to affect the lipophilicity, molecular weight and enzymatic stability. These results show that the common features of the most active compounds are a low molecular weight and a low clog P compared to weaker compounds.
A new series of prodrugs are currently being developed with the aim of preventing early prodrug-drug bioconversion and/or increasing the aqueous solubility, with both parameters affecting oral absorption. A new prodrug in the thioester family has thus shown encouraging preliminary results and deserves further attention. Findings of preliminary studies on the efficacy and tolerance of these compounds in mice and in primates appear promising, indicating that this approach may be applicable to human malaria.

BACK-UP DERIVATIVES: MODIFICATIONS OF THE LINKER
Bis-thiazolium salts exhibited weak oral bioavailability (< 5%) so there was no further clinical development for oral administration. Nevertheless, such derivatives could be administered as neutral precursors that are expected in vivo to revert back to the parent drugs (see above). The use of a cyclic prodrug approach led to im-provement of the absolute oral bioavailability of T3 (reaching 15%) but it remained weak.
Careful analysis of the previous results showed that the reduction in the overall flexibility of the molecule, in the case of neutral prodrugs, led to an increased ip/po ratio. Moreover, it is well established that physico-chemical properties (hydro solubility, flexibility, etc.) are critical parameters for designing oral-drug like compounds. Improving such properties for bis-thiazolium salts, and corresponding prodrugs, has been envisaged through modification of the linker between cationic heads which is highly flexible (11 C-C bonds), lipophilic and thus may not have favorable parameters for oral bioavailability.
As an example, the aromatic ring has been introduced to decrease the flexibility, and polar atoms (oxygen ones) have been incorporated to decrease the lipophilicity of the related prodrugs (Scheme 5) [39]. For each aromatic ring, different methylene arm lengths (n= 3, 4, 5) have been tested and modulation of the anchoring position (ortho, meta, para) has been explored.
All compounds were then evaluated on P. falciparum-infected erythrocytes and against P. vinckei in mice for their antimalarial properties ( Table 8).
The in vitro activity ranged from 9 nM to 800 nM. Lengthening of the linker led to an increased antiplasmodial activity and the presence of oxygen atoms was somehow detrimental to the antiplasmodial activity. Para and meta compounds exhibited lower IC 50 values for compounds with the same chain length but anchored at different positions on the aromatic ring.
The structure-activity relationship suggested that the optimal linker-construct was an aromatic moiety branched with two n-butyl chains in the para orientation. Two promising compounds incorporating modified linkers were identified and were able to cure  malarial infection in mice at very low doses (i.e. IC 50 20 nM and ED 50 ip < 5 mg/kg). A prodrug approach will soon be applied to evaluate the effect of the linker modifications on oral absorption enhancement.

CONCLUDING REMARKS
Phospholipids are crucial for the development of intracellular malaria parasites. Consequently, proteins involved in their biosynthetic pathways often appear essential [40][41] and thereby represent potential targets for antimalarial drugs. Our main goal is to develop new antimalarial drugs that will affect their biosynthetic pathways with an innovative mechanism of action.
Based on i) crucial biosynthetic pathways, ii) limiting steps within the metabolic pathways, and iii) specificity regarding the host, we have identified phospholipid-related transporters or enzymes suitable for drugability and for pharmacological targeting.
We rationally designed choline analogs and optimized them for their antiplasmodial activity. The end products are compounds able to block P. falciparum asexual blood stages, at single digit nanomolar concentrations, and able to cure malarial infection in rodents or non-human primates. The potency and specificity of these antiphospholipid effectors are likely due to their unique ability to accumulate in a nonreversible way inside the intraerythrocytic parasite. These compounds are thought to inhibit choline transport, thus preventing PC synthesis, and also to interact with plasmodial haemoglobin degradation metabolites in the food vacuole. This multiple mode of action, distinct from current antimalarial agents, is the