Multicomponent reaction-based synthesis and biological evaluation of tricyclic heterofused quinolines with multi-trypanosomatid activity

Human African trypanosomiasis (HAT), Chagas disease and leishmaniasis, which are caused by the trypanosomatids Trypanosoma brucei, Trypanosoma cruzi and Leishmania species, are among the most deadly neglected tropical diseases. The development of drugs that are active against several trypanosomatids is appealing from a clinical and economic viewpoint, and seems feasible, as these parasites share metabolic pathways and hence might be treatable by common drugs. From benzonapthyridine 1, an inhibitor of acetylcholinesterase (AChE) for which we have found a remarkable trypanocidal activity, we have designed and synthesized novel benzo[h][1,6]naphthyridines, pyrrolo[3,2-c]quinolines, azepino[3,2-c]quinolines, and pyrano[3,2-c]quinolines through 2–4-step sequences featuring an initial multicomponent Povarov reaction as the key step. To assess the therapeutic potential of the novel compounds, we have evaluated their in vitro activity against T. brucei, T. cruzi, and Leishmania infantum, as well as their brain permeability, which is of specific interest for the treatment of late-stage HAT. To assess their potential toxicity, we have determined their cytotoxicity against rat myoblast L6 cells and their AChE inhibitory activity. Several tricyclic heterofused quinoline derivatives were found to display an interesting multi-trypanosomatid profile, with one-digit micromolar potencies against two of these parasites and two-digit micromolar potency against the other. Pyranoquinoline 39, which displays IC50 values of 1.5 μM, 6.1 μM and 29.2 μM against T. brucei, L. infantum and T. cruzi, respectively, brain permeability, better drug-like properties (lower lipophilicity and molecular weight and higher CNS MPO desirability score) than hit 1, and the lowest AChE inhibitory activity of the series (IC50 > 30 μM), emerges as an interesting multi-trypanosomatid lead, amenable to further optimization particularly in terms of its selectivity index over mammalian cells.


Introduction
Neglected tropical diseases (NTDs) are a group of 17 infectious diseases that globally affect more than 1 billion people from 149 countries [1]. Not only do NTDs cause a huge health impact, both in terms of disability-adjusted life years (DALY, 26 million DALYs) and mortality (534,000 deaths annually), but they also have harmful effects on the overall economic productivity of developing countries where these diseases are endemic, which become inexorably trapped in an unbreakable cycle of poverty [2e4].
Among NTDs, vector-borne kinetoplastid diseases are particularly deadly, with leishmaniasis, Chagas disease (or American trypanosomiasis) and human African trypanosomiasis (HAT or sleeping sickness) ranking first, fifth and sixth, respectively, in number of associated deaths [2]. Their causative agents are trypanosomatid parasites that are transmitted to humans through the intervention of an infected insect vector: Trypanosoma brucei gambiense and T. brucei rhodesiense (accounting for 98% and 2% of cases of HAT, respectively) spread through the bite of blood-feeding tsetse flies; Trypanosoma cruzi (for Chagas disease) transmitted most commonly through contact with the faeces of a blood-sucking triatomine bug (the so-called kissing bug); and different Leishmania species, prominently Leishmania donovani and Leishmania infantum (for visceral leishmaniasis), transmitted by the bite of female phlebotomine sand flies. In the absence of treatment, these diseases are frequently fatal, with their mortality being associated to particular stages or forms of the disease. In the case of HAT, after an initial hemo-lymphatic stage characterized by nonspecific clinical symptoms, parasites can cross the bloodebrain barrier (BBB) and invade the central nervous system (CNS), giving rise to an array of severe neurological manifestations that include profound sleep disruptions and eventually coma and death. The initial phases of Chagas disease are usually asymptomatic or associated with non-specific symptoms of fever, malaise, or lymph node enlargement, but in about 30% of patients it evolves into a chronic phase, usually characterized by cardiomyopathy, and is a major cause of premature heart failure in Latin America. The most severe manifestation of leishmaniasis is visceral form, which leads to hepatosplenomegaly, progressive anaemia, and ultimately death in most cases. These diseases are usually confined to rural areas of endemic countries (sub-Saharan Africa for HAT, mainly Central and South America for Chagas disease, and Middle East and Asia, East Africa, Central and South America and Southern Europe for leishmaniasis). However, climate changes due to global warming, which may result in an extension of the insect vector habitats, as well as international travel and immigration patterns may expand the geographical impact of these infectious diseases, thereby increasing the population at risk [5]. Travel to and immigration from endemic countries have made Chagas disease and several forms of leishmaniasis emerging infections in the United States (both infections), and Spain and Japan (Chagas disease) [6,7].
The current therapies against HAT, Chagas disease and leishmaniasis suffer from important shortcomings. HAT first-stage treatments rely on pentamidine and suramin, which require parenteral administration and are ineffective against the second stage. Stage 2 HAT can be treated by painful intravenous administration of the arsenical drug melarsoprol, which may lead to fatal reactive encephalopathy in 5e10% of patients, or with eflornithine, which is much safer but requires intravenous administration and hospitalization [8]. Toxicity is also a major issue with the approved drugs against Chagas disease, the nitroderivatives benznidazole and nifurtimox, and with some of the drugs used for the treatment of visceral leishmaniasis (pentavalent antimonials, amphotericin B, paromomycin and miltefosine). Apart from complicated long courses of treatment, in most cases parenteral administration is required. In addition, the emergence of resistance to these drugs in areas of high transmission further challenges their clinical application [6,8e13].
In the absence of preventive or therapeutic vaccines and rigorous control of insect vectors [5,14], the development of novel chemotherapies against these infectious diseases, with appropriate efficacy and safety profiles, is desperately needed [15,16]. Besides combinations of approved antiprotozoan drugs or repurposing of known drugs with other indications [8,14,15], increasing research efforts are being made to design novel chemical entities that hit one or several biological targets which play a key role in the biology of the parasite and are sufficiently different from those in the mammalian host cells as to enable selective toxicity [5,9,17e23]. However, while we are gaining a better understanding of the relevant parasite targets, phenotypic whole cell screening of novel compounds or chemical libraries remains a very successful approach for anti-protozoan drug discovery [8,24,25]. Thus, antiprotozoan drug pipelines are being enriched through drug discovery campaigns involving the synthesis of novel chemical entities and their biological evaluation against the selected parasites [26e28]. Of particular interest are those compounds that can be active against several protozoan parasites [25,29e31], as several NTDs usually coexist in endemic countries [1]. The trypanosomatid parasites that cause HAT, Chagas disease and leishmaniasis are taxonomically related, have similar structural and biochemical features, and seem to share many of their metabolic pathways [14], thereby rendering them especially amenable to modulation by common drugs. Indeed, several structural families featuring a 4aminoquinoline moiety have been recently reported to display a multi-protozoan profile, namely trypanocidal and antiplasmodial activity [32e35].
The use of multicomponent reactions [36] appears to be a very useful strategy to rapidly build new hits in a modular manner. This approach is having a tremendous impact in modern medicinal chemistry. Apart from considerably speeding the process and manufacture of some drugs [37], it is especially relevant in drug discovery. It allows the preparation of new scaffolds and their straightforward decoration, and also facilitates the hit to lead transition and pharmacological issues [38,39]. We have recently reported the synthesis and acetylcholinesterase (AChE) inhibitory activity of a series of 1,2,3,4-tetrahydrobenzo[h] [1,6]naphthyridines such as 1 (Fig. 1), which are prepared using a multicomponent Povarov reaction as the key step [40]. This transformation requires a cyclic enamide as an activated olefin, which affords the ring A of 1 ( Fig. 1), an aromatic aldehyde, which affords the substituent at position 5, and an aniline, which affords the ring C with the substituent at position 9. Because the tricyclic scaffold of the benzo[h] [1,6]naphthyridine system of 1 contains a 4-aminoquinoline motif and a side chain with a second protonatable nitrogen atom, we inferred that this compound might display some anti-protozoan activity. Indeed, the outstanding IC 50 value of 3.33 mM against T. brucei that we have found for 1 has confirmed our initial assumption. In light of this, and relying on the synthetic versatility of the multicomponent Povarov reaction [41,42], which might enable the modification of ring A and the substituents at positions 1, 5 and 9 of benzonaphthyridine 1 by simply changing the starting materials, we planned the synthesis of a series of analogues of 1 and their evaluation against the trypanosomatids that cause HAT, Chagas disease and leishmaniasis.
Here, we report i) the synthesis of novel benzo[h] [1,6]naphthyridine-, pyrrolo [3,2-c]quinoline-, azepino [3,2-c]quinoline-, and pyrano [3,2-c]quinoline-based analogues of 1 with different substituents at rings A, B, and C, as well as some quinoline derivatives resulting from opening of ring A of some of these tricyclic scaffolds; ii) their evaluation against T. brucei, T. cruzi and L. infantum; iii) and the assessment of their cytotoxic activity against rat myoblast L6 cells and their AChE inhibitory activity. Also, to determine their potential usefulness for late-stage HAT, the ability of the novel compounds to cross the BBB has been evaluated in vitro using a parallel artificial membrane permeability assay (PAMPA-BBB).

Design and synthesis of the target compounds
Compound 1 is a rather quite lipophilic molecule, with a calculated log P value of 6.64 [43], hence clearly above the commonly accepted threshold for good oral bioavailability [44]. In order to decrease lipophilicity, we first envisaged the synthesis of compound 2, i.e. the N-deethylated derivative of 1, and its isomer 3 ( Fig. 1 and Schemes 1 and 2), in which the chloro and aminomethyl substituents at the para position of the phenyl substituent and at position 9, respectively, were interchanged. Interestingly, both analogues turned out to be more potent against T. brucei than hit 1 (Fig. 1, see Section 2.2), especially compound 3, which was 3-fold more potent, albeit still too lipophilic (log P ¼ 5.89).
In this regard, we next envisioned a series of modifications around the structure of compound 3, aimed at improving drug likeness and deriving structureeactivity relationships first against T. brucei, and then against T. cruzi and L. infantum. Because removal of the benzyl group at position 1 of compound 3 should lead to a significant decrease of both lipophilicity and molecular weight and, hence, to improved drug likeness, we planned the synthesis of the N 1 -unsubstituted derivative 31 and its analogues resulting from ring contraction and ring expansion (30 and 32, respectively, Scheme 3), isomerization of the chlorine substituent from position 9 to position 8 (33, Scheme 3), and NH / O bioisosteric replacement (39, Scheme 5), using the corresponding nitriles as the immediate precursors. Moreover, to assess the relevance of the 4aminomethylphenyl substituent at the B-ring, we decided to study the biological activity of the nitrile precursors (i.e. 26e29 and 38, Schemes 3 and 5), as well as that of compounds 42 and 43, in which the 4-aminomethylphenyl group at the B-ring of compound 39 was replaced by a 2-furyl-or 2-thienyl-substituent (Scheme 6). During the synthesis of the target tricyclic heterofused quinoline compounds some byproducts arising from opening of the A-ring were obtained (see below) and, eventually, converted into additional target compounds featuring cyano, hydroxy or amino groups at the side chain (i.e. 35,44,45,47,49 and 50, Schemes 4, 6 and 7).
The synthesis of compound 2 was carried out by the four-step sequence depicted in Scheme 1. The multicomponent Povarov reaction between the known unsaturated lactam 4 as the activated olefin [45,46], and commercially available 4-chlorobenzaldehyde, 6, and aniline 5, bearing an N-Boc-protected aminomethyl side chain, under Sc(OTf) 3 catalysis in acetonitrile, followed by DDQ oxidation [47] of the resulting diastereomeric mixture of octahydrobenzonaphthyridines 7 afforded compound 8 in 47% overall yield, after silica gel column chromatography purification. N-Boc deprotection of 8, followed by reduction of the resulting lactam 9 with (EtO) 3 SiH under Zn(OAc) 2 catalysis [48], and silica gel column chromatography purification afforded the target benzonaphthyridine 2, in 29% overall yield for the last two steps (Scheme 1).
The synthesis of compound 3 required only 3 steps, starting with a multicomponent Povarov reaction between the enamide 4, 4-chloroaniline, 10, and the aromatic aldehyde 11, bearing a 4cyano group as the precursor of the aminomethyl side chain. DDQ oxidation of the resulting diastereomeric mixture of octahydrobenzonaphthyridines 12, followed by simultaneous LiAlH 4 reduction of the lactam and nitrile functionalities of compound 13 afforded the target benzonaphthyridine 3 in low overall yield, after silica gel column chromatography purification (Scheme 2).
For the synthesis of the N 1 -debenzylated analogues 30e33 we used a 4-step protocol that involved an initial multicomponent Povarov reaction between chloroanilines 10 or 17, cyano aldehyde 11 and commercially available N-Boc-protected cyclic enamines 14e16, followed by DDQ oxidation, N-Boc acidic deprotection, and final LiAlH 4 reduction of nitriles 26e29 (Scheme 3). After silica gel column chromatography purification, the target compounds 30, 31, 32, and 33 were obtained in 20%, 41%, 22%, and 39% overall yield, respectively.  Of note, during the synthesis of compound 25, a significant amount (17% yield) of ring-open by-product 34 (Scheme 4) was also formed. This compound was subjected to the standard acidic conditions for N-Boc deprotection, affording amine 35 in 90% yield (Scheme 4).
The synthesis of pyranoquinoline 39 was envisaged through a three-step sequence, analogous to that used for compound 3, but starting from 3,4-dihydro-2H-pyran, 36, as the activated olefin for the Povarov reaction instead of the enamide 4 (Scheme 5).
Because we wanted to assess the influence on anti-protozoan activity of the degree of oxidation of the B-ring of the heterofused quinoline compounds, at this point we decided to isolate the tetrahydroquinoline compound resulting from the Povarov reaction, before performing the oxidation to the final quinoline derivative. Thus, after the reaction between the cyclic enol ether 36, the aniline 10 and the aldehyde 11, the crude product was purified by silica gel column chromatography to obtain a 1:1 diastereomeric mixture 37 in 95% yield. After a second column chromatography from this material, a sample of the all-cis-diastereoisomer, all-cis-37, was isolated to be subjected to biological evaluation (see below). Eventually, the pyranoquinoline 39 was obtained in 23% overall yield by DDQ oxidation of the diastereomeric mixture 37 to the quinoline derivative 38, followed by LiAlH 4 reduction of the nitrile to an aminomethyl group (Scheme 5).
Analogously to the synthesis of 39, starting from the cyclic enol Finally, the Mitsunobu reaction of compound 45 with N-(tertbutyldimethylsilyloxy)-4-methylbenzenesulfonamide followed by treatment with CsF in acetonitrile afforded nitrile 47 in 84% overall yield, after silica gel column chromatography purification. Moreover, alcohol 45 was converted via mesylate into the corresponding amines 49 and 50 in 31% and 16% overall yield, respectively (Scheme 7).
All the compounds to be subjected to biological evaluation, except 37, were transformed into the corresponding hydrochloride or dihydrochloride salts, and were chemically characterized through IR, 1 H and 13 C NMR spectra, HRMS and HPLC purity analysis and elemental analysis. As previously mentioned, we first evaluated the activity of 1, a compound that we had developed as an inhibitor of AChE [40], and the novel heterofused quinoline and ring-open analogues in vitro  against cultured bloodstream forms of T. brucei, using nifurtimox as a reference compound. We also investigated their brain permeability to assess their potential usefulness in late-stage HAT. To determine potential toxic effects, their cytotoxicity against rat skeletal myoblast L6 cells and AChE inhibitory activity, which might result in cholinergic side-effects, were also evaluated.
Within the most potent tricyclic heterofused quinoline derivatives, the best substitution pattern for activity against T. brucei seems to involve i) Bn-N > O > NH at position 1 of the A-ring, with compound 3 being 1.5-and 2.4-fold more potent than 39 and 31, respectively, ii) a five-membered A-ring, with compound 30 being 2.5-fold more potent than 31 and 32, bearing a six-and sevenmembered A ring, respectively, and iii) the presence of the chlorine atom at position meta relative to the quinoline nitrogen atom, with compound 33 being 2-fold more potent than 31.
As in the tricyclic heterofused quinoline derivatives, the presence of a protonatable nitrogen atom at the side chain of ring-open quinoline analogues seems to be play a role in the activity of these compounds against T. brucei, with amines 35, 49, and 50 exhibiting IC 50 values around 5 mM (Table 1), whereas compounds 44, 45, and 47, with neutral hydroxy or cyano groups at the side chain were clearly less potent (IC 50 > 25 mM).

Brain permeation
Brain penetration is a desirable property for novel drugs against HAT, as it will make them effective against the late-stage disease, when parasites have invaded the CNS. Thus, the brain passive permeability of the novel compounds was evaluated in vitro through the widely used PAMPA-BBB method [50], which is based on the use of a porcine brain lipid extract as an artificial BBB model. The novel compounds had in vitro permeabilities (P e ) clearly above the threshold established for a high BBB permeation, i.e. CNSþ with Pe (10 À6 cm s À1 ) > 5.25, with the exceptions of compounds 31, 33, and 42, for which an uncertain BBB permeation was predicted ( Table 1).
Apart from a high brain passive permeability, a low P-glycoprotein (P-gp) efflux is a highly desirable property to ensure target engagement in the CNS. A number of knowledge-based approaches have emerged to assist the early phases of the drug discovery process, by addressing different aspects related with pharmacokinetic and pharmacodynamic properties to enhance the likelihood of deriving candidates with appropriate drug-like properties [51]. For compounds intended to act within the CNS, some molecular descriptor guidelines recommend to maintain topological polar surface area (TPSA) < 90 Å 2 (preferably < 70 Å 2 ) and the number of hydrogen bond donors (HBD) < 2 to maximize the probability of evading P-gp efflux [52,53]. All of the novel compounds have TPSA values well below 70 Å 2 and most of them also have a HBD number 2 (Table S1, Supplementary Material), so that they should be expected to show a low P-gp efflux liability. To further support the lack of P-gp efflux liability in the novel compounds and gain additional insight into their CNS druglikeness, we applied the Pfizer CNS multiparameter optimization (CNS MPO) algorithm [54]. This algorithm is based on a set of six physicochemical parameters, namely lipophilicity (cLogP), calculated distribution coefficient at pH 7.4 (cLogD), molecular weight, TPSA, number of HBD, and the pK a value of the most basic centre, with all these parameters being weighted equally using a desirability score from 0 to 1. Increasing total CNS MPO scores for drugs has been correlated with increasing probabilities of appropriate pharmacokinetic and safety attributes, including high passive permeability and low P-gp efflux, amongst others. Particularly, drugs with CNS MPO desirability scores !4 (in a scale from 0 to 6) are expected to show a full alignment of the desired pharmacokinetic properties [54]. Interestingly, CNS MPO desirability scores ! 4 have been calculated for 12 out of the 21 novel compounds (Table S2, Supplementary Material), which supports their potential usefulness for treating late-stage HAT.

Cytotoxicity and acetylcholinesterase inhibitory activity
To assess potential toxic effects of the novel derivatives, all the compounds with IC 50 values against T. brucei below 10 mM were subjected to cytotoxicity studies using rat myoblast L6 cells. All the tested compounds were selective for T. brucei vs mammalian cells (selectivity indices (SI Tb ) in the range 2.1e6.8 (Table 1), being more selective than the hit 1 but less selective than nifurtimox (SI Tb ¼ 7.3).
Because hit 1 was developed as an AChE inhibitor with potential application against Alzheimer's disease, this kind of activity might result in unwanted cholinergic side effects if the novel compounds were used as anti-protozoan agents. We therefore evaluated their AChE inhibitory activity using Electrophorus electricus AChE (EeAChE), a widely used and affordable enzyme source for screening this activity. All the structural modifications carried out around the structure of hit 1 led to a decrease in EeAChE inhibitory activity (3e110-fold). However, despite their decreased AChE inhibitory activity relative to hit 1, most tested compounds were more potent against EeAChE than against T. brucei. Only the benzonaphthyridine 3, the pyranoquinoline 39 and the ring-open quinoline derivative 49 turned out to be more potent against T. brucei than against EeAChE (4-, >20-and~8-fold, respectively).
Of note, hit 1 is 6-fold less potent against human recombinant AChE (hAChE) than against EeAChE (IC 50 ¼ 0.15 mM compared with IC 50 ¼ 0.94 mM) [40]. Even though this might also be the case for the novel compounds reported here, future lead optimization should focus on decreasing AChE inhibitory activity and increasing selectivity indices.

In vitro activity against T. cruzi and L. infantum
The occurrence of common metabolic pathways in trypanosomatid parasites makes it potentially feasible to develop antiprotozoan agents endowed with a multi-trypanosomatid profile [14,55]. Thus, after having confirmed the significant activity against T. brucei of all the novel target heterofused quinoline derivatives and some of the ring-open analogues and assessed their brain permeation, cytotoxicity and AChE inhibitory activity, we undertook the evaluation of all the novel compounds against epimastigote forms of T. cruzi (strain MHOM/ES/2203/BCN590 (Tcl)) and promastigote forms of L. infantum (strain MCAN/ES/92/ BCN722), using benznidazole and potassium antimony (III) tartrate hydrate as reference compounds.
Most of the tested compounds exhibited activity against T. cruzi, albeit with two-digit micromolar IC 50 values (Table 2). Interestingly, there was a similar SAR profile as had been found with T. brucei. First, isomerization of the chlorine and aminomethyl substituents from compound 2 to 3 results in increased potency against T. cruzi (4-fold). Also, higher potencies were observed for those tricyclic heterofused quinoline analogues bearing a protonatable 4-(aminomethyl)phenyl group at the B-ring, relative to those bearing a 4-cyanophenyl, 2-furyl, or 2-thienyl groups. Thus, amines 30, 31, 32, and 39 were 2e6-fold more potent than their nitrile precursors 26, 27, 28, and 38, respectively, and amine 39 was also 3-and 10-fold more potent than the 5-(2-furyl)-and 5-(2thienyl)-substituted derivatives 42 and 43, respectively. The sole exception was amine 33, which turned out to be 2-fold less potent than its nitrile precursor 29. Also, within the aminomethylphenylsubstituted analogues, the order of potencies related to the substitution at position 1 of the A-ring was: Bn-N > O > NH, with the N 1 -benzylated benzonapththyridine 3 being 3-and 5-fold more potent than pyranoquinoline 39 and N 1 -unsubstituted benzonapththyridine 31. Regarding the size of the A-ring, again the presence of a fiveor a seven-membered ring A led to increased potency relative to the derivatives with a six-membered A-ring, with the pyrroloquinoline 30 and the azepinoquinoline 32 being 1.5-fold more potent than the benzonaphthyridine 31. For this activity, unlike against T. brucei, a higher potency seems to arise from the presence of a chlorine atom at position para relative to the quinoline nitrogen atom, with compound 31 being 1.5-fold more potent than 33. Overall, the most potent analogue of the series against T. cruzi was compound 3, which exhibited an IC 50 value of 9.47 mM, 4-fold more potent than the reference compound benznidazole. In addition, six other derivatives turned out to be slightly more potent or equipotent to benznidazole, with IC 50 values around 30 mM. However, all the novel compounds were clearly less selective than benznidazole for T. cruzi vs mammalian cells, with SI Tc < 1 for the novel compounds and 14.1 for benznidazole ( Table 2). All of the novel compounds turned out to be leishmanicidal agents, with IC 50 values in the low micromolar range in most cases, being more potent (up to 5-fold) than the reference compound potassium antimony (III) tartrate hydrate (Table 2). Thus, these compounds were more active against L. infantum than against T. cruzi, with the sole exception of compound 3, the most potent antichagasic derivative of the series. Even though the novel compounds were 2e25-fold more selective for L. infantum vs mammalian cells than potassium antimony (III) tartrate hydrate (SI Li ¼ 0.1), their selectivity indices were rather low (SI Li in the range 0.2e2.5) ( Table 2).
Some of the SARs of the leishmanicidal activity of the novel tricyclic heterofused quinoline analogues were very similar to those found for anti-trypanosome activities, with the best substitution pattern involving the presence of: i) a protonatable 4-(aminomethyl)phenyl group at the B-ring, with amines 30, 31, 32, 33, and 39 being 1.5 to >31-fold more potent than their nitrile precursors 26, 27, 28, 29, and 38, respectively, and amine 39 being 6-and 8fold more potent than the 5-(2-furyl)-and 5-(2-thienyl)substituted derivatives 42 and 43, respectively; ii) a fiveor a seven-membered A-ring, with the pyrroloquinoline 30 and the azepinoquinoline 32 being approximately 2-fold more potent than the benzonaphthyridine 31, bearing a six-membered ring A; and iii) an aromatic B-ring, with compound 38 being >9-fold more potent than the saturated analogue 37. For this activity, the presence of an oxygen atom a position 1 of the A-ring led to increased potency relative to an NH group, with compound 39 being 2-fold more potent than 31. However, a reverse SAR trend relative to those found for the anti-trypanosome activities was observed regarding the presence of a Bn-N group at position 1 of the A-ring and the isomerization of the chlorine and aminomethyl substituents. Thus, the presence of a Bn-N group at position 1 of the A-ring was detrimental for leishmanicidal activity, with compound 3 being 3and 5-fold less potent than the NH-and O-substituted counterparts 31 and 39, respectively, whereas the interchange of the chlorine and aminomethyl substituents from compound 2 to 3 resulted in decreased potency against L. infantum (4-fold). The position of the chlorine substituent at the C-ring had no influence on the leishmanicidal activity of the novel compounds. Of note, as found when measuring T. brucei activities, the three ring-open analogues featuring a protonatable amino group at the side chain, i.e. 35, 49, and 50, exhibited significant leishmanicidal activity, with IC 50 values around 15 mM ( Table 2).
The most potent analogues of the series against L. infantum were benzonaphthyridines 2, 30, and 32, and pyranoquinoline 39, with IC 50 values in the range 5e8 mM.

Conclusion
We have synthesized a series of tricyclic heterofused quinolines, namely benzo[h] [1,6]naphthyridines, pyrrolo[3,2-c]quinolines, azepino [3,2-c]quinolines, and pyrano[3,2-c]quinolines, through 2e4-step synthetic sequences that involve as the key step an initial Povarov multicomponent reaction between a cyclic enamide or enol ether as an activated olefin and a properly substituted aniline and aromatic aldehyde. The novel compounds have been designed from benzonaphthyridine 1, a submicromolar inhibitor of AChE previously developed in our group that has been found to display a significant in vitro activity against T. brucei. Initial structural modifications around hit 1, including N-dealkylation of the side chain at position 9 and isomerization by interchange of the chlorine atom at the para position of the 5-phenyl group and the aminomethyl substituent at position 9, have led to benzonaphthyridine 3, which has turned out to be 3-fold more potent than hit 1 against T. brucei. The structure of compound 3 has been further modified by N 1 -debenzylation, A-ring contraction and expansion, bioisosteric NH / O replacement at position 1, and substitution of the 5-(4aminomethyl)phenyl group by 5-(2-furyl) and 5-(2-thienyl). During the synthesis of the target tricyclic compounds, some quinoline derivatives with a side chain at position 3, arising from opening of the A-ring, were obtained. To further expand the SAR studies, the structure of these ring-open derivatives was subsequently modified by introduction of a protonatable amino group or a neutral cyano group at the end of the side chain.
Trypanosomatid parasites responsible for HAT, Chagas disease and visceral leishmaniasis seem to share common metabolic pathways [14,55], thereby being potentially amenable to treatment by common drugs. Consistent with this, we have found some common SAR trends related to the activities of the novel tricyclic heterofused quinoline analogues against T. brucei, T. cruzi and L. infantum, with several of these compounds being moderately potent against two or three of these parasites. Thus, the presence of an oxidized B-ring featuring a protonatable 4-(aminomethyl) phenyl group and the bioisosteric NH / O replacement at position 1 led to higher potencies against the three parasites. Benzonaphthyridines 2, 30 and 32, and pyranoquinoline 39 exhibit an interesting multi-trypanosomatid profile, with single digit micromolar IC 50 values against T. brucei and L. infantum and IC 50 around 30 mM against T. cruzi, whereas benzonaphthyridine 3 exhibits single digit micromolar IC 50 values against T. brucei and T. cruzi and IC 50 around 30 mM against L. infantum. Interestingly, all of these multitrypanosomatid compounds have been predicted to be able to cross the BBB, which is of utmost importance for the treatment of late-stage HAT, and have better drug-like properties than hit 1, both in terms of lower lipophilicity and molecular weight. They also display higher CNS MPO desirability scores, and hence, are expected to be endowed with more appropriate and aligned pharmacokinetic attributes, including low P-gp efflux. A significant AChE inhibitory activity in most of these compounds, albeit lower than that of hit 1, and significant toxicity to rat L6 cells are their main drawbacks, which should be addressed in further lead optimization. To this end, the dual trypanocidal and leishmanicidal pyranoquinoline 39, which displays the lowest AChE inhibitory activity, and hence, the lowest potential for cholinergic side effects, is likely to be the best starting point.

Chemistry. General methods
Melting points were determined in open capillary tubes with a MFB 595010M Gallenkamp melting point apparatus. 400 MHz 1 H/ 100.6 MHz 13 C NMR spectra were recorded on a Varian Mercury 400 spectrometer. The chemical shifts are reported in ppm (d scale) relative to solvent signals (CD 3 OD at 3.31 and 49.0 ppm in the 1 H and 13 C NMR spectra, respectively; CDCl 3 at 7.26 and 77.0 ppm in the 1 H and 13 C NMR spectra, respectively), and coupling constants are reported in Hertz (Hz). Assignments given for the NMR spectra of the new compounds have been carried out by comparison with the NMR data of 3,9,28,37,42,43,44,45,47, and 50, which in turn, were assigned on the basis of DEPT, COSY 1 H/ 1 H (standard procedures), and COSY 1 H/ 13 C (gHSQC or gHMBC sequences) experiments. IR spectra were run on a PerkineElmer Spectrum RX I or on a Thermo Nicolet Nexus spectrophotometer. Absorption values are expressed as wavenumbers (cm À1 ); only significant absorption bands are given. Column chromatography was performed on silica gel 60 AC.C (35e70 mesh, SDS, ref 2000027). Thin-layer chromatography was performed with aluminium-backed sheets with silica gel 60 F 254 (Merck, ref 1.05554), and spots were visualized with UV light and 1% aqueous solution of KMnO 4 . NMR spectra of all of the new compounds were performed at the Centres Científics i Tecnol ogics of the University of Barcelona (CCiTUB), while elemental analyses and high resolution mass spectra were carried out at the Mycroanalysis Service of the IIQAB (CSIC, Barcelona, Spain) with a Carlo Erba model 1106 analyser, and at the CCiTUB with a LC/MSD TOF Agilent Technologies spectrometer, respectively. The HPLC measurements were performed using a HPLC Waters Alliance HT apparatus comprising a pump (Edwards RV12) with degasser, an autosampler, a diode array detector and a column as specified below. The reverse phase HPLC determinations were carried out on a YMC-Pack ODS-AQ column (50 Â 4.6 mm, D S. 3 mm, 12 nm).
Solvent A: water with 0.1% formic acid; Solvent B: acetonitrile with 0.1% formic acid. Gradient: 5% of B to 100% of B within 3.5 min. Flux: 1.6 mL/min at 50 C. The analytical samples of all of the compounds that were subjected to pharmacological evaluation were dried at 65 ºC/2 Torr for at least 2 days (standard conditions) and possess a purity !95% as indicated by their elemental analyses and/or HPLC measurements.
To a stirred solution of p-chlorobenzaldehyde, 6 (1.16 g, 8.25 mmol) and aniline 5 (1.83 g, 8.23 mmol) in anhydrous CH 3 CN (30 mL), 4 Å molecular sieves and Sc(OTf) 3 (0.81 g, 1.65 mmol) were added. The mixture was stirred at room temperature under argon atmosphere for 5 min and then treated with a solution of enamine 4 (1.50 g, 8.01 mmol) in anhydrous CH 3 CN (16 mL). The resulting suspension was stirred at room temperature under argon atmosphere for 3 days. Then, the resulting mixture was diluted with sat. aq. NaHCO 3 (150 mL) and extracted with EtOAc (3 Â 200 mL). The combined organic extracts were dried over anhydrous Na 2 SO 4 and evaporated under reduced pressure to give a solid residue (4.71 g), mainly consisting of a diastereomeric mixture of octahydrobenzonaphthyridines 7, which was used in the next step without further purification.
To a solution of crude diastereomeric mixture 7 (4.58 g of a crude of 4.71 g) in anhydrous CHCl 3 (150 mL), DDQ (4.85 g, 21.4 mmol) was added. The reaction mixture was stirred at room temperature under argon atmosphere overnight, diluted with CH 2 Cl 2 (150 mL) and washed with sat. aq. NaHCO 3 (3 Â 250 mL). The combined organic extracts were dried over anhydrous Na 2 SO 4 and evaporated under reduced pressure to give a solid residue (5.33 g), which was purified through column chromatography (35e70 mm silica gel, hexane/EtOAc mixtures, gradient elution  [1,6]naphthyridin-9-yl}methanamine 9 Compound 8 (1.94 g, 3.67 mmol) was dissolved in 4M HCl/ dioxane solution (24 mL) at 0 C. The mixture was stirred at room temperature overnight and concentrated in vacuo. The solid residue was diluted in water (20 mL), treated with 1N NaOH (20 mL), and the aqueous phase was extracted with a 10% MeOH/CHCl 3 mixture (3 Â 50 mL). The combined organic extracts were dried over anhydrous Na 2 SO 4 and evaporated at reduced pressure to give a crude product (1.42 g), which was purified through column chromatography (35e70 mm silica gel, EtOAc/MeOH/50% aq. NH 4 1 mL). The reaction mixture was stirred at 65 C for 48 h in a sealed vessel. The resulting mixture was cooled to room temperature, poured onto 1N NaOH (5 mL), stirred for 10 min, and then extracted with EtOAc (3 Â 15 mL). The combined organic extracts were dried over anhydrous Na 2 SO 4 and concentrated under reduced pressure to give a solid (313 mg), which was purified through column chromatography (35e70 mm silica gel, CH 2
From this crude (4.67 g) and DDQ (5.39 g, 23.7 mmol), a brown solid residue ( [56]. Trypanocidal activity was assessed by growing parasites in the presence of various concentrations of the novel compounds and determining the levels which inhibited growth by 50% (IC 50 ) and 90% (IC 90 ). T. brucei in the logarithmic phase of growth were diluted back to 2.5 Â 10 4 mL À1 and aliquoted into 96-well plates. The compounds were then added at a range of concentrations and the plates incubated at 37 C. Each drug concentration was tested in triplicate. Resazurin was added after 48 h and the plates incubated for a further 16 h and the plates then read in a Spectramax plate reader. Results were analysed using Graph-Pad Prism.

T. cruzi and L. infantum culturing and evaluation of trypanocidal and leishmanicidal activity
Stock solutions of the novel compounds in DMSO were prepared at concentrations of 20 mg mL À1 , with the final DMSO concentration being lower than 2% for all experiments. Trypanocidal and leishmanicidal activity was assessed by growing parasites in the presence of various concentrations of the novel compounds and determining the levels which inhibited growth by 50% (IC 50 ). The IC 50 was determined from a least-squares linear regression of growth rate versus log drug concentrations.
For evaluation of T. cruzi activity serial dilutions of the novel compounds in LIT culture medium were aliquoted in 96-well microtiter plates (Costar 3596). Then 4 Â 10 6 mL À1 epimastigotes culture medium in the logarithmic growth phase were added to each well, incubating at 28 C for 72 h. Benznidazole was used as the reference drug at concentrations from 2.50 mM to 2.42 mM.
Parasite inhibition for each drug concentration was determined using an automated cell counter (TC20 BIO-RAD). All assays were performed in duplicate at least twice.
Potassium antimony (III) tartrate hydrate was used as the reference drug at concentrations from 815 to 0.80 mM. Growth was measured through the acid phosphatase activity [57]. All assays were performed in duplicate at least twice.

Cytotoxic activity against rat skeletal myoblast L6 cells
Cytotoxicity against mammalian cells was assessed using microtitre plates following a described procedure [58]. Briefly, rat skeletal muscle L6 cells were seeded at 1 Â 10 4 mL À1 in 200 mL of growth medium containing different compound concentrations.
The plates were incubated for 6 days at 37 C and 20 mL resazurin was then added to each well. After a further 8 h incubation, the fluorescence was determined using a Spectramax plate reader.

Acetylcholinesterase inhibitory activity
The inhibitory activity against E. electricus (Ee) AChE (Sigma-eAldrich) was evaluated spectrophotometrically by the method of Ellman et al. [59]. The reactions took place in a final volume of 300 mL of 0.1 M phosphate-buffered solution pH 8.0, containing EeAChE (0.03 u/mL) and 333 mM 5,5 0 -dithiobis(2-nitrobenzoic) acid (DTNB; SigmaeAldrich) solution used to produce the yellow anion of 5-thio-2-nitrobenzoic acid. Inhibition curves were performed in duplicates using at least 10 increasing concentrations of inhibitors and preincubated for 20 min at 37 C before adding the substrate. One duplicate sample without inhibitor was always present to yield 100% of AChE activity. Then substrate, acetylthiocholine iodide (450 mM; SigmaeAldrich), was added and the reaction was developed for 5 min at 37 C. The colour production was measured at 414 nm using a labsystems Multiskan spectrophotometer.
Data from concentrationÀinhibition experiments of the inhibitors were calculated by non-linear regression analysis, using the GraphPad Prism program package (GraphPad Software; San Diego, USA), which gave estimates of the IC 50 (concentration of drug producing 50% of enzyme activity inhibition). Results are expressed as mean ± S.E.M. of at least 4 experiments performed in duplicate.

Determination of brain permeability: PAMPA-BBB assay
The in vitro permeability (P e ) of the novel compounds and fourteen commercial drugs through lipid extract of porcine brain membrane was determined by using a parallel artificial membrane permeation assay [50]. Commercial drugs and the target compounds were tested using a mixture of PBS:EtOH 70:30. Assay validation was made by comparing experimental and described permeability values of the commercial drugs, which showed a good correlation: P e (exp) ¼ 1.583 P e (lit) e 1.079 (R 2 ¼ 0.9305). From this equation and the limits established by Di et al. for BBB permeation, three ranges of permeability were established: compounds of high BBB permeation (CNSþ): P e (10 À6 cm s À1 ) > 5.25; compounds of low BBB permeation (CNSe): P e (10 À6 cm s À1 ) < 2.09; and compounds of uncertain BBB permeation (CNS±): 5.25 > P e (10 À6 cm s À1 ) > 2.09.