Design, synthesis and antimalarial activity of novel bis{N-[(pyrrolo[1,2-a]quinoxalin-4-yl)benzyl]-3-aminopropyl}amine derivatives

Abstract Novel series of bis- and tris-pyrrolo[1,2-a]quinoxaline derivatives 1 were synthesized and tested for in vitro activity upon the intraerythrocytic stage of W2 and 3D7 Plasmodium falciparum strains. Biological results showed good antimalarial activity with IC50 in the μM range. In attempting to investigate the large broad-spectrum antiprotozoal activities of these new derivatives, their properties toward Leishmania donovani were also investigated and revealed their selective antiplasmodial profile. In parallel, the in vitro cytotoxicity of these molecules was assessed on the human HepG2 cell line. Structure–activity relationships of these new synthetic compounds are discussed here. The bis-pyrrolo[1,2-a]quinoxalines 1n and 1p were identified as the most potent antimalarial candidates with selectivity index (SI) of 40.6 on W2 strain, and 39.25 on 3D7 strain, respectively. As the telomeres of the parasite could constitute an attractive target, we investigated the possibility of targeting Plasmodium telomeres by stabilizing the Plasmodium telomeric G-quadruplexes through a FRET melting assay by our new compounds.


Introduction
Malaria remains as one of the most devastating infectious diseases of the developing world. Malaria remains a major cause of public health problem in about 95 countries mainly located in the tropical zone of the globe (notably Africa, South-East Asia and also Eastern Mediterranean region) 1 ; while approximately 3.2 billion people are at risk of being infected with malaria and developing disease. The latest figures on the incidence and mortality of malaria show that, despite progress in the implementation of preventive measures such as insecticide-treated mosquito nets and intermittent preventive treatments, this parasitic disease is still estimated to affect over 214 million people and to account for 438,000 deaths in 2015, of which approximately 80% are concentrated in just fifteen countries, mainly in Africa. The death toll is particularly high in children under five and pregnant women of the World Health Organization (WHO) African region 1 . Five species of protozoan parasites belonging to the Plasmodium genus, namely, falciparum, malariae, vivax, ovale and knowlesi cause malaria in human beings; from which, P. falciparum is the most dangerous of these species [2][3][4] .
The increasing prevalence of multiple drug resistant strains of P. falciparum in most malaria endemic areas has significantly reduced the efficacy of the current antimalarial drugs. Nowadays, the only fully effective antimalarial drugs utilize artemisinin and its derivatives, notably parenteral artesunate now recommended as the first line of treatment of severe malaria for at least 24 h and until oral medication could be tolerated 5 . Artemisinin and its derivatives are also used in combination with several different partner drugs (including lumefantrine, mefloquine, amodiaquine, sulfadoxine/pyrimethamine and piperaquine) in artemisinin-based combination therapies (ACTs), now recommended as the first line of treatment of uncomplicated Plasmodium falciparum malaria and as the second part of treatment for 3 days of severe malaria in endemic areas 5 . However, over the last decade evidence has grown that artemisinin resistance has emerged and spread within Southeast Asia, first on the Cambodia-Thailand border in 2009 6,7 , but now across a widening area of the Greater Mekong Subregion. Rapid scientific advances in understanding of this problem have taken place within the last five years [8][9][10] and defined mutations in "K13" gene of P. falciparum associated to reduced ring-stage susceptibility to artemisinins 11 . Therefore, new antimalarial agents with new mechanisms of action are required to overcome the emergence of resistance and to control an ever-increasing number of epidemics caused by the malaria parasite 12 . Although current efforts in antimalarial drug discovery are focused on identification of new biological targets, continued research on new 4-aminoquinoline derivatives is still warranted. This is because the hemoglobin degradation pathway in P. falciparum has a proven history as an excellent therapeutic target to which the parasite has difficulty in evolving resistance. In spite of resistance to chloroquine, previous work has shown that modification of the lateral side chain of chloroquine results in aminoquinoline derivatives that avoid the chloroquine resistance mechanism 13,14 .
A possibility to overcome this multidrug-resistant mechanism is to design new quinoline-based drugs which will not be recognized by the protein system involved in drug efflux. In this regard, two series of compounds that show promise in this regard are the bisquinoline and bisacridine antimalarial drugs A and B ( Figure 1) [15][16][17][18] . These drugs show much lower resistance indices than chloroquine, indicating that the bisquinoline or bisacridine structures are less efficiently excluded by drug-resistant parasites.
Hence, we described here the synthesis of new bis-or trispyrrolo[1,2-a]quinoxaline derivatives 1a-t ( Figure 1) and reported on their in vitro antiplasmodial activity against the chloroquine-sensitive (3D7) and the chloroquine-resistant (W2) strains of the malaria parasite Plasmodium falciparum. As these new compounds were designed as quinoline-like bio-isosteres, and as the quinoline skeleton is the fundamental unit of many antiprotozoan drugs, these pyrrolo [1,2-a]quinoxaline derivatives were also investigated on another medically important protozoan, Leishmania donovani, in order to evaluate the specificity of their antiparasitic activity. Leishmaniasis is an infectious disease caused by protozoan parasites belonging to the genus Leishmania. Leishmania parasites exist in two major morphological stages: extracellular flagellated promastigotes in the digestive tract of their sandfly vector, which is the infective stage and immobile intracellular amastigotes in the cells of their host' mononuclear phagocytic system. Leishmaniasis presents various clinical aspects including cutaneous leishmaniasis, the most common form, muco-cutaneous leishmaniasis and visceral leishmaniasis, the most severe form, lethal in untreated patients. Leishmania donovani is one of the major causative agents of human visceral leishmaniasis which represents a public health problem: 0.2 to 0.4 million visceral leishmaniasis cases occur each year and more than 90% of global visceral leishmaniasis cases occur in just five countries: India, Bangladesh, Sudan, Brazil and Ethiopa 31 . The current treatment of the disease is based on a limited number of chemotherapeutic agents (meglumine antimoniate, sodium stibogluconate, pentamidine, amphotericin B and miltefosine) which present many limits, notably characterized by a high toxicity and cost 32,33 . In addition, the in vitro cytotoxicity of these new molecules was assessed on the human HepG2 cell line, in order to determine their selectivity index. Moreover, as the telomeres of the parasite could constitute an attractive target, we also investigated the possibility of targeting Plasmodium telomeres by stabilizing the Plasmodium telomeric Gquadruplexes through a FRET melting assay by our new synthesized compounds. Indeed, telomerase activity has been identified in gametocytes and during the transition to erythrocytic stage of P. falciparum 34 . The telomeric 3' G-overhang region of P. falciparum is comprised of repeated degenerate unit 5 0 GGGTTYA 3 0 (where Y may be T or C) 35 which can fold into intramolecular G-quadruplex 36 .

Chemistry
Commercial reagents were used as received without additional purification. Melting points were determined with an SM-LUX-POL Leitz hot-stage microscope (Leitz GMBH, Midland, ON) and are uncorrected. IR spectra were recorded on a NICOLET 380FT-IR spectrophotometer (Thermo Electron Scientific Instruments LLC, Madison, WI). NMR spectra were recorded with tetramethylsilane as an internal standard using a BRUKER AVANCE 300 spectrometer (Bruker BioSpin, Wissembourg, France). Splitting patterns have been designated as follows: Analytical TLC were carried out on 0.25 precoated silica gel plates (POLYGRAM SIL G/UV 254 ) and visualization of compounds after UV light irradiation. Silica gel 60 (70-230 mesh) was used for column chromatography. Microwave experiments were carried out using a focused microwave reactor (CEM Discover, Saclay, France). High resolution mass spectra (electrospray in positive mode, ESIþ) were recorded on a Waters Q-TOF Ultima apparatus. Mass spectra were recorded on an Ultraflex III TOF/TOF system (Bruker Daltonics, Bremen, Germany), equipped with 200 Hz smartbeam laser (355 nm) and operating in reflectron positive ion mode. Mass spectra were acquired over the m/z range 300-5000 by accumulating data from 1000 laser shots for each spectrum. The instrumental conditions employed to analyze molecular species were the following: ion source 1: 25.08 kV; ion source 2: 21.98 kV, lens: 11.03 kV, pulsed ion extraction: 30  To suspension of 4-chloropyrrolo[1,2-a]quinoxaline 7 (3.3 mmol), and Pd(PPh 3 ) 4 (0.164 mmol) in a mixture of toluene/EtOH (50/3 mL) under nitrogen were added K 2 CO 3 (3.6 mmol) and phenylboronic acid (3.6 mmol). The reaction mixture was refluxed for 24 h, and the cooled suspension was extracted with CH 2 Cl 2 (3 Â 70 mL). The organic layer was washed with a saturated solution of NaCl (90 mL), and the combined organic extracts were dried over sodium sulfate, filtered, and evaporated under reduced pressure. The crude residue was triturated in ethanol. The resulting precipitate was filtered, washed with ethanol, and purified by column chromatography on silica gel using dichloromethane as eluent gave the pure product 8.
To a solution of 1,11-diamino-3,6,9-trioxaundecane (0.56 mmol) in ethanol (12 mL To a solution of compound 9 (1.26 mmol) in methanol (40 mL) was added portion-wise at 0 C sodium borohydride (10.1 mmol; 8 equiv.). The reaction mixture was then heated under reflux for 4 h and then evaporated to dryness under reduced pressure. After cooling, the residue was triturated in water and extracted with dichloromethane (85 mL). The organic layer was separated, dried over sodium sulfate and evaporated to dryness. The residue were then purified by column chromatography on silica gel using dichloromethane/methanol (90/10, v/v) as eluent to give the pure product 1.                    13 13

In vitro antiplasmodial activity
The in vitro antiplasmodial activities were tested over concentrations ranging from 39 nM to 40 lM against culture-adapted Plasmodium falciparum reference strains 3D7 and W2. The former strain is susceptible to chloroquine (CQ) but displays a decreased susceptibility to mefloquine (MQ) while the latter is considered as resistant to chloroquine. The parasites were cultivated in RPMI medium (Sigma-Aldrich, Lyon, France) supplemented with 0.5% Albumax I (Life Technologies corporation, Paisley, United Kingdom), hypoxanthine (Sigma-Aldrich), gentamicin (Sigma-Aldrich), and human erythrocytes and were incubated at 37 C in a candle jar, as described previously 37  20 lL of MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) solution (5 mg/mL) were added to each well followed by incubation for another 4 h. The enzyme reaction was then stopped by addition of 100 lL of 50% isopropanol -10% sodium dodecyl sulfate 42 .Plates were shaken vigorously (300 rpm) for 10 minutes and the absorbance measured in a plate reader at 570 nm in a BIO-TEK ELx808 Absorbance Microplate Reader. Inhibitory concentration 50% (IC 50 ) was defined as the concentration of drug required to inhibit by 50% the metabolic activity of L. donovani promastigotes compared to the control. IC 50 were calculated by non-linear regression analysis processed on doseresponse curves, using TableCurve 2D V5.0 software. IC 50 values represent the mean value calculated from three independent experiments.

Cytotoxicity evaluation
A cytotoxicity evaluation was realized according to the method of Mosmann 41 with slight modificationsto determine the cytotoxic concentrations 50% (CC 50 ) and using doxorubicin as a cytotoxic reference-compound. These assays were performed toward the human HepG2 cell line (HepG2 CC 50 ). HepG2 (hepatocarcinoma cell line purchased from ATCC, ref HB-8065) is a commonly used human-derived hepatocarcinoma cell line that has shown characteristics similar to those of primary hepatocytes. These cells express many of the hepatocyte-specific metabolic enzymes, thus enabling the cytotoxicity of tested product metabolites to be evaluated. Briefly, cells in 100 lL of complete RPMI medium, [RPMI supplemented with 10% FCS, 1% L-glutamine (200 mM) and penicillin (100 U/mL)/streptomycin (100 lg/mL)] were inoculated into each well of 96-well plates and incubated at 37 C in a humidified 6% CO 2 . After 24 h incubation, 100 lL of medium with various product concentrations dissolved in DMSO (final concentration less than 0.5% v/v) were added and the plates were incubated for 72 h at 37 C. Duplicate assays were performed for each sample. Each plate-well was then microscope-examined for detecting possible precipitate formation before the medium was aspirated from the wells. 100 lL of MTT solution (0.5 mg/mL in medium without FCS) were then added to each well. Cells were incubated for 2 h at 37 C. After this time, the MTT solution was removed and DMSO (100 lL) was added to dissolve the resulting blue formazan crystals. Plates were shaken vigorously (300 rpm) for 5 minutes. The absorbance was measured at 570 nm with 630 nm as reference wavelength spectrophotometer using a BIO-TEK ELx808 Absorbance Microplate Reader. DMSO was used as blank and doxorubicin (Sigma Aldrich) as positive control. Cell viability was calculated as percentage of control (cells incubated without compound). The 50% cytotoxic concentration was determined from the dose-response curve by using the TableCurve 2D V5.0 software (Systat Software, San Jose, CA).  36,43 . The oligonucleotides were prefolded in 10 mM lithium cacodylate buffer (pH 7.2), with 10 mM KCl and 90 mM LiCl (K þ condition). The FAM emissions were recorded at 516 nm using a 492-nm excitation wavelength in the absence and presence of a single compound as a function of temperature (25 to 95 C) in 96-well microplates by using a Stratagene MX3000P real-time PCR device at a rate of 1 C.min À1 . The data obtained were normalized between 0 and 1, and the temperature required for half-denaturationof oligonucleotides corresponded to the emission value of 0.5 was calculated as DT m . Each experiment was performed in duplicate with 200 nM of labeled oligonucleotide and 0, 1, 2, or 5 lM of compound under K þ condition. For each compound, two independent experiments were carried out. The data were plotted using OriginPro 9.1 software (OriginLab Corporation, Northampton, MA).
All these quinoxaline compounds 1a-u were then converted into their ammonium oxalate salts by treatment with oxalic acid in refluxing isopropanol (Table 1).

Biological activity
In vitro antimalarial activity All new substituted bis-and trispyrrolo[1,2-a]quinoxaline derivatives 1a-u were evaluated for their antimalarial activity in vitro on P. falciparum CQ-resistant strain W2 (IC 50 CQ ¼0.20 lM) and the CQ-sensitive and MQ decrease susceptibility strain 3D7 (IC 50 CQ ¼0.08 lM). As shown in Table 2, they were found to have IC 50 between 0.05-4.20 lM on W2, and 0.04-8.96 lM on 3D7 P. falciparum strains, respectively.
Against the CQ-sensitive strain (3D7)and in the subseries of the pyrrolo[1,2-a]quinoxalines unsubstituted on their heterocyclic moieties (compounds 1a-c), compound 1c joined by a bis-(3-aminopropyl)piperazine linker was found more active than its structural analogs with bis-(3-aminopropyl)amine (compound 1a) or bis-(3-aminopropyl)methylamine linkages (compound 1b): i.e. IC 50 ¼ 0.80 lM for 1c versus 3.42 and 1.20 lM for 1a and 1b, respectively. Surprisingly, most of the pyrrolo[1,2-a]quinoxalines 1d-g and 1i bearing a methoxy substituent on the heterocyclic skeleton exhibited moderate antimalarial activity against 3D7 strain with IC 50 ranging from 1.28 to 5.85 lM, in the exception of compound 1 h which showed an interesting IC 50 of 0.25 lM against this Plasmodium strain. As against the W2 strain, replacement of the methoxy function by an ethoxy one in position 3 of the pyrrolo[1,2-a]quinoxaline moieties (compounds 1j-k) increased the anti P. falciparum activity against the 3D7 strain (IC 50 ¼ 0.26 and 0.19 lM for 1j and 1k, respectively). The introduction of a phenyl ring in position 2 of the pyrrolo[1,2-a]quinoxaline moiety (compound 1 l) reduced considerably the antimalarial activity on the 3D7 strain (IC 50 ¼ 8.96 lM). Similar observations could be made with the structural analogs 1q-r which showed moderate activity against 3D7 strain (IC 50 ¼ 1.00-7.22 lM). In the subseries in which the polyaminoalkyl side chain was anchored in position 3 of the benzyl ring, the piperazine substituted bis-pyrrolo[1,2-a]quinoxaline 1p was found to be two times more active than the reference drugs chloroquine and mefloquine (IC 50 ¼ 0.040 lM for 1p versus IC 50 ¼ 0.080 lM for the references). The tris-pyrrolo[1,2-a]quinoxalines 1s-t showed similar antimalarial activities when compared with their methoxy substituted bis-homologs 1o-p (IC 50 ¼ 0.20-0.24 lM for 1s-t versus 0.04-0.18 lM for 1o-p). Compound 1u bearing diaminotrioxaundecane linkage exhibited similar activity against the 3D7 strain than those observed for chloroquine and mefloquine (IC 50 -0.08 lM).
In vitro antileishmanial activity against promastigote forms In order to increase the biological profile of molecules 1a-u, complementary analyses were performed. Notably, P. falciparum belonging to coccidian protozoan parasites, their in vitro biological  activity on flagellate protozoan parasites like Leishmania donovani was evaluated ( Table 2). In comparison with amphotericin B and pentamidine used as reference drugs (IC 50 ¼ 0.1 lM and 5.5 lM, respectively), most of the tested compounds 1 were found active against the promastigote forms of L. donovani. It must be noticed that a majority of these bis-pyrrolo[1,2-a]quinoxaline derivatives 1 (compounds 1a, 1c, 1d-j and 1m-q)with IC 50 ranging from 1.0 to 4.26 lM were found slightly more potent than pentamidine, one of the reference compounds (IC 50 ¼ 5.5 lM). In particular, bis{N-[4-(indolo[1,2-a]quinoxalin-6-yl)benzyl]-3-aminopropyl}piperazine 1q was found to be 9.65 times more active than pentamidine (IC 50 ¼ 0.57 lM versus IC 50 ¼ 5.5 lM). In our previously described antileishmanial series 21,22 , we have noticed that introduction of a methoxy group in positions 7 and 8 of the pyrrolo[1,2-a]quinoxaline moiety generally increased the antileishmanial activity. In this new series, we also decided to introduce a methoxy or an ethoxy to evaluate the influence of such groups for their in vitro antileishmanial activity upon the L. donovani strain. Meanwhile, the presence of these alkoxy groups in the 7, 8, 9 or 3 position seems not to be detrimental for the activity against L. donovani as illustrated by results obtained for compounds 1d-k and 1o-p (IC 50 ¼ 1.07-4.26 lM) in comparison with unsubstituted compounds 1a, 1c and 1m-n (IC 50 ¼ 1.0-3.71 lM). The same observation could be done for the nature of the polyalkylamine linker between the two benzylpyrrolo[1,2-a]quinoxaline moieties that seemed also not to be crucial, with the exception of piperazine substituted 1 g, which showed a better antileishmanial activity in vitro compared to its structural methylamine analog 1f (IC 50 ¼ 4.25 lM versus 1.07 lM for 1 g) upon the L. donovani strain. Moreover, in the subseries in which the polyaminoalkyl side chain was anchored in position 3 of the benzyl ring, the various bis-pyrrolo[1,2-a]quinoxalines 1m-p were found less beneficial for the in vitro antiparasitic activity with an IC 50 of 2.80-3.71 lM in comparison with their analogs 1a-c and 1f-g. In addition, the trisubstitution of the tris(3-aminopropyl)amine linker by the pyrrolo[1,2-a]quinoxaline core (compounds 1s-t) led to a decrease in the antileishmanial activity (IC 50 > 10 lM) in comparison with their bis-pyrrolo[1,2-a]quinoxaline analogs 1d-e and 1o-p.

Cytotoxicity and selectivity index
In order to assess their selectivity of action, the cytotoxicity of the antimalarial compounds 1a-u was also evaluated in vitro upon one human cell line. The cytotoxic concentrations 50% (CC 50 ) on the metabolizing human hepatocyte HepG2 cell line allowed access to the corresponding selectivity indexes (SI) defined as ratio of cytotoxic and antimalarial activities (SI ¼ CC 50 /W2 or 3D7 IC 50 ). The SI could validate their real potential as selective antiparasitic agents. The results concerning the cytotoxicity and SI data are presented in Table 2. As expected, most of the active bis-and trispyrrolo[1,2-a]quinoxalines 1 showed significant level of cytotoxicity against the HepG2 cell line (IC 50 ¼ 1.25-3.38 lM). Considering the W2 P. falciparum strain, the selectivity indexes varied between 0.37 and 40.6, and SI ranged from 0.53 to 39.25 by taking into account the 3D7 strain. This SI led to the identification of compound 1n with SI of 40.6 on W2 strain, and SI of 39.25 for derivative 1p on 3D7 strain. We could notice that these last bioactive compounds 1n and 1p were bis-pyrrolo[1,2-a]quinoxalines linked by a 1,4-bis(3-aminopropyl)piperazine moiety in position 3 of their benzyl ring. On the other hand, the bis-pyrrolo[1,2-a]quinoxaline 1u is also interesting with SI ¼20.26, on the 3D7 malaria strain. However, according to the SI values, these compounds appeared to be toxic; thus, new modulations by replacing the benzyl substituted heterocyclic moiety to synthesize new candidates could be developed for forthcoming pharmacological investigations.

FRET-melting experiments
As these new compounds presented the same requirements as those required for G-4 stabilizing ligands, we have also investigated their ability of targeting P. falciparum telomeres as a potential strategy to interfere with human protozoan parasite infections. In fact, the telomeres of the parasite could constitute an attractive target. They are composed of repetitions of a degenerate motif ( 5 0 GGGTTYA 3 0 , where Y could be T or C), which is different from the human one ( 5 0 GGGTTA 3 0 ). This antimalarial approach based on targeting P. falciparum telomeres has been previously described by our team 36 . Thus, in this report, we investigated the stabilization of Plasmodium telomeric G-quadruplexes by these new compounds through a FRET melting assay. For this evaluation, the two Plasmodium telomeric sequences (FPf1T and FPf8T) and the human one (F21T), in potassium, were used (two different Plasmodium telomeric repeats were used to fit the degenerate consensus). Thus, we used this FRET melting assay to study the interactions of some of our most bioactive compounds 1 and also the reference drugs (chloroquine -CQ and mefloquine -MQ) with these three telomeric sequences. For better visualization, a plot of DT m values induced on FPf1T or FPf8T versus the DT m induced on F21T is presented in Figure 2, allowing us to classify the more active compounds 1 in various subsets. Almost all the tested compounds displayed roughly the same stabilization profiles for the two Plasmodium telomeric sequences FPf1T and FPf8T and slightly better than that of the human one F21T, except 1s and 1t for FPf1T and 1o and 1s for FPf8T.The reference compounds CQ and MQ, never evaluated on these parasitic sequences before our work, showed very slight stabilization on these three telomeric sequences. These preliminary results mean that most of the tested bis-pyrrolo[1,2-a]quinoxalines 1 have a moderate preference for Plasmodium over human telomeric quadruplexes, excepted bispyrroloquinoxaline 1o. Surprisingly, the tris-pyrrolo[1,2-a]quinoxalines 1s-t were found more active on the human telomeric sequence F21T in comparison with their bis-analogues1d-e and 1o-p. In contrast, the bis{N-[4-(methoxypyrrolo[1,2-a]quinoxalin-4-yl)benzyl]-3-aminopropyl}piperazines 1e and 1i were found to be more active on the Plasmodium sequences with the higher DT m .
These last data could benow considered as a promising starting point for the further development and optimization of new and potent antimalarial compounds as well as promising stabilizers for the telomeric Plasmodium DNA sequences. As very little is known about telomere regulation in P. falciparum, further studies should be now investigated both on parasitic and human sequences which presented different response or time-response to telomere perturbation. molecules was assessed on the human HepG2 cell line. Structureactivity relationships of these new synthetic compounds are here also discussed, as well as their relative ability of targeting P. falciparum telomeres as potential mechanism of action. Thus, as the telomeres of the parasite could constitute an interesting target, we have also established the possibility of targeting Plasmodium telomeres by stabilizing the Plasmodium telomeric G-quadruplexes through a FRET melting assay. These results led us to conclude that the two bis{N-[4-(methoxypyrrolo[1,2-a]quinoxalin-4-yl)benzyl]-3-aminopropyl}piperazines 1e and 1i seemed to be able to discriminate between Plasmodium and human telomeric quadruplexes. By taking into account the biological and biophysical results and also the structure-activity relationships of these new series and those previously described, our overall results could be now considered as a promising starting point for the further development and optimization of new and potent antiprotozoan compounds.