derivatives that inhibit Trypanosoma brucei glyceraldehyde-3-phosphate dehydrogenase ( Tb GAPDH) and Trypanosoma cruzi trypanothione reductase ( Tc TR) and display trypanocidal activity

Crassiflorone


M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT 4 In addition, NPs often act as plant and animal defense chemicals with an intrinsic multitarget mechanism of action [9,10]. Due to the processes that nature uses for their generation, NPs usually possess a complex structure with an array of functional groups. Thus, their biosynthetic pathways involve a range of enzymes, each with distinct topology and binding sites, to which the NP under construction needs to bind. As a result of this mode of generation, NPs have an inherited propensity to recognize multiple target proteins [10,11].
Based on the above considerations, as well as on our continuous interest in multitarget compounds directed against two anti-trypanosomatid drug targets [12], namely Trypanosoma brucei glyceraldehyde-3-phosphate dehydrogenase (TbGAPDH) [13] and Trypanosoma cruzi trypanothione reductase (TcTR) [14], we turned our attention to NP crassiflorone (1) (see Fig. 1). 1 is a pentacyclic furocoumarin naphthoquinone isolated from the African ebony Diospyros crassiflora [15], with reported anti-mycobacterial and anti-gonorrhoeal activities [16]. To note, following the total synthesis of 1, some doubt has been cast on the structure of crassiflorone, which might be a regioisomer of 1, although a direct comparison with the natural material was not possible [24].
We envisaged that 1 could act as a potential dual GAPDH/TR inhibitor. This proposal was based on the fact that its scaffold shares the same structural features of dual-targeted inhibitors previously identified by us [12]. In particular, starting from the dual-targeted 2-phenoxy-naphthoquinone fragment 2 (TbGAPDH IC 50 = 7.2 µM and TcTR IC 50 = 9.0 µM) [17,18], we had developed dual inhibitors exemplified by the quinone-coumarin hybrid 3 (TbGAPDH IC 50 = 5.4 µM and TcTR K i = 2.3 µM) [12]. Hence, we reasoned that 1, similarly featuring both the quinone and coumarin frameworks, could be a good starting point for the development of novel NP-inspired GAPDH/TR inhibitors (Fig. 1).

M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT 5 Building on this rationale, we aimed to verify whether 1 could bind at both enzyme active sites.
According to the significant similarity within the Trypanosomatid family of the two selected proteins and their accessibility [19,20], we performed docking analysis on TbGAPDH (PDB 2X0N) [21] and TcTR (PDB 1BZL) [22]. The docking results showed that 1 recognizes TbGAPDH by establishing favorable interactions (Fig. 1). In particular, both the coumarin and naphthoquinone frameworks of 1 are involved in the formation of a network of H-bonds, between the carbonyl of lactone ring with the side chain of Thr198 (4.0 Å), and both the carbonyl groups of quinone with the backbone of Cys165 (2.0 Å) and the side chain of Arg248 (2.1 Å). In addition, the pentacyclic aromatic scaffold establishes π-π interaction with His193 (4.2 Å) and favorable hydrophobic interactions with Tyr338. On the other hand, the wide and hydrophobic cavity of TR nicely accommodates 1, primarily anchored by π-π interactions with Tyr111 (2.9 Å). Hydrogen bonds could be identified between both the carbonyl groups of the coumarin and quinone frameworks of 1 and Ser110 (1.8 Å and 2.2 Å, respectively), and between the hydroxyl substituent of 1 and Glu19 (1.6 Å) (Fig. 1). Moreover, the hydrophobic interactions with Trp22, Val54, Val59, Ile107, Met114, Ile339, and Leu399′ further contribute to stabilize the binding of 1. Thus, docking studies predicted that such a bulky and rigid scaffold interacts with both targets of interest, confirming the design rationale.
On this basis, we designed a small series of 1-derived compounds, which, for reasons of synthetic accessibility, do not carry the methyl substituents on the pentacyclic core. Our goal was to identify

Chemistry
For the synthesis of 15-23, the same retrosynthetic approach reported for 1, starting from 4hydroxycoumarin and 2-bromonaphthoquinone derivatives as precursor structures, was exploited [23,24]. Accordingly, the preparation of intermediate coumarin 6 and quinones 7-14 was required.
As shown in Scheme 1, the nitration of the commercially available 4 with potassium nitrate and concentrated sulfuric acid at 0 °C gave compound 5 [25], which was reduced by using iron powder in a mixture of ethanol/acetic acid/hydrochloric acid/water, furnishing 6 in a quantitative yield. 2- Bromonaphthoquinones 7-14 were synthesized as previously described [26,27]. With both intermediates 4, 6, 7-14 in hand, we followed the reported coupling reaction in the presence of potassium carbonate and copper(II) acetate to easily afford 15-23 in moderate to good yields (16-57%, Scheme 2) [23,24]. This one-pot procedure allowed the efficient coupling between the two precursor structures under mild reaction conditions (i.e. weak base and room temperature). The nucleophilic attack of enolate anions formed through the base-mediated deprotonation of the hydroxyl groups of 4 and 6 onto the more electrophilic carbon atom of 7-14 is followed by the copper(II) mediated-oxidative cyclization of the resulting intermediate adducts, which represents the thermodynamic driving force for the ring closure and subsequent aromatization to directly provide the pentacyclic scaffold. Compounds 15-23 thus obtained were characterized by 1 H-NMR and high-resolution mass spectra (see the Experimental Section).

Results and discussion
To investigate the anti-trypanosomal profile, crassiflorone derivatives 15-23 were characterized for their enzymatic (TbGAPDH and TcTR) and whole-cell parasite (L. infantum, T. brucei and T. cruzi) activities. In addition, assessment of their in-vitro early-toxicity profile was carried out.

TbGAPDH inhibition studies
We tested the inhibitory activity of 15-23 against TbGAPDH at a concentration of 10 µM, following a previously reported protocol [28]. The collected percentages of enzymatic inhibition are graphically reported in Fig. 2A, in comparison with 2. Notably, among the synthesized compounds, 18, 19 and 23 showed inhibition percentages higher than 50% and higher to that observed for 2.
Particularly, the most active compounds were the 8-OMe-substituted derivative (88%) (18) and the amino-coumarin derivative 23 (75%). In spite of the promising results, the poor solubility of 18 prevented further measuring its inhibitory activity at different (higher) concentrations, due to compound precipitation in the assay. The same applied to 19. For the more soluble 23, the doseresponse curve was determined and the IC 50 value was found to be 2.1 µM (Fig. 2B). Importantly, 23 turned out to be more active than the corresponding "open analogue" 3 (IC 50 TbGAPDH = 5.4 µM) [16], demonstrating that the molecular rigidification strategy applied resulted in a better TbGAPDH recognition. Moreover, these values place 23 as the most active TbGAPDH inhibitor reported so far, being more active than other identified natural and unnatural compounds [29][30][31].
Notably, docking studies suggest that 23 interacts at the TbGAPDH active site through additional H-bonds between the -OH group and the backbone of Asn334 (1.9 Å) and between the -NH 2 and the side chain of Thr166 (1.7 Å), which are not present in 1/TbGAPDH complex ( show π-π interactions.

TcTR inhibition studies
In parallel, we determined the TcTR inhibition profile of 15-23 at the highest concentrations soluble in the assay buffer, as previously described [32]. At a concentration of 100 µM or 40 µM of trypanothione disulfide (TS 2 ), which correspond to a substrate concentration of 6 × K m and 2.4 × K m , respectively [33], 10 µM of antraquinone 16 and -OMe derivative 18 showed negligible inhibitory activity (< 20%), whereas 23 turned out to be totally inactive, even at a concentration of 100 µM (Table 1) (Table 1). This was a first indication that the compounds did not act as purely competitive inhibitor of the enzyme.

Whole-cell parasite assays
Compounds 15-23 were tested against T. brucei bloodstream, T. cruzi intracellular trypomastigote, and L. infantum intracellular amastigote forms (Fig. 6). The compounds were tested at the highest

In-vitro compound mediated early-toxicity assessment
In light of the poor drug-like properties and the scarce safety profile of the currently available antitrypanosomatid drugs [1] as well as the liabilities of quinones [36], the in-vitro early-toxicity profiles [37] of 15-23 were determined for cytotoxicity (A549, human lung adenocarcinoma epithelial cell line), cytochrome P450 inhibition (CYP1A2, CYP2C9, CYP2C19, CYP2D6 and CYP3A4), mitochondrial toxicity (786-O, human renal carcinoma cell line) and hERG inhibition [38]. The output of these studies together with the phenotypic ones, are useful to make an informed decision as to which compounds should be progressed in further studies. The compounds were screened in each of the in-vitro early-toxicity assays at 10 µM and a traffic light system was employed to rank the compounds such that green = no liability, yellow = medium liability and red = significant liability (Table 2).
The rationale for implementing the cytotoxicity assay was to establish whether the observed phenotypic effects were related to specific anti-parasitic activity and not merely due to toxicity to the host cell on which the parasites depend. Remarkably, none of the tested compounds inhibited the growth of the human A549 cell line at 10 µM. This information added confidence that for the best-performing compounds, namely 18 and 21, a certain degree of host/parasite selectivity could be reached.
The inhibition of the cytochrome P450 (CYP) enzymes is a major cause of drug-drug interactions [39] and in the case of anti-trypanosomatid compounds this is an issue, with co-morbidity (e.g. HIV infection) as a major problem for many diseased patients [40]. In this respect, we determined the inhibitory profiles of 15-23 at 10 µM on five isoforms of CYP enzymes (CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4) [41]. Although 15-20 and 23 had no or medium liability against CYP3A4, the other compounds were associated with significant cytochrome P450 liabilities and this needs to be addressed in any further compounds of this class that are designed and synthesized.
Blockade of hERG K + channel is important as can lead to cardiotoxicity, especially in the case of cardiomyopathy caused by Chagas disease [38], therefore it is important for any drug not be associated with this effect [42]. Notably, among the tested compounds at 10 µM, only 19 and 21 were associated with significant hERG liability.
As mitochondrial toxicity by way of swelling and membrane potential collapse have been reported as a possible mode of trypanocydal action of naphthoquinone analogues [28,43], and nonselective off-target mitochondrial toxicity [44] is a major contributor to the failure of quinone drugs, we tested whether 15-23 at 10 µM could induce mitochondrial toxicity. Remarkably, all the synthesized compounds display no sign of mitochondrial toxicity at the tested concentration.
The above in-vitro early-toxicity assessment has demonstrated some liabilities of the crassiflorone scaffold and has led to the identification of those that need to be addressed when designing the next generation of derivatives.

Pan Assay Interference Compounds (PAINS) and solubility analyses
Due to the presence in 15-23 of chemical substructures that belong to the PAINS category [45], we have assessed their potential PAINS behavior using the FAF-Drugs4 filter (available on http://fafdrugs3.mti.univ-paris-diderot.fr) [46]. Not unexpectedly, our compounds were assessed to include high-risk (quinone) and low-risk (coumarin) structural alerts (data not shown). However, we are quite confident that 15-23 should not act as PAINS and their activity is not an artifact, because they showed no activity when tested in two parallel HTS assays against Leishmania and Trypanosoma pteridine reductase-1 (PTR-1) enzymes (see Supplementary Information, Table S1).
In addition, to rule out that solubility issues might have affected the accuracy of the performed assays, target compounds 15-23 were evaluated for their aqueous solubility (Table S2). Solubility data included in silico calculations of LogS, performed by means of two different programs (i.e., MarvinSketch 17.13 and FAF-Drugs4), and experimental aqueous solubility determined by nephelometry. Encouragingly, all the compounds were tested at concentrations below the solubility limit.

Conclusions
Herein, for the first time the natural compound crassiflorone (1) has been proposed as a potential dual TbGAPDH/TcTR inhibitor and novel synthetic crassiflorone derivatives (15)(16)(17)(18)(19)(20)(21)(22)(23) have been designed and synthesized. In this respect, 15-23 were envisaged as rigid analogues of previously reported coumarin-quinone hybrids exemplified by 3 [12]. Among the tested compounds, only 19 displayed a balanced dual profile against the selected targets (% inhibition at 10 µM TbGAPDH = 64% and TcTR = 65%). Unfortunately, this profile did not give rise to the expected superior antitrypanosomal cellular activity. In fact, in the phenotypic assay, the best performing compounds of the series were single-targeted inhibitors 18 and 21, which were more active than 19. However, solubility concerns clearly affected the anti-trypanosomatid profile of these compounds.
Although we have identified liabilities of the crassiflorone scaffold in terms of solubility and CYP1A2, CYP2C9, CYP2C19, and CYP2D6 inhibition, the lack of cytotoxicity against human cells as well as mitochondria, points towards a possible developability of this class. In fact, we anticipate that it will be possible to overcome the current drawback in the next generation of compounds, as these problems are not likely to be insurmountable. For instance, with regards to the low solubility of the compounds, formulation technology (i.e. particle size reduction, spray drying and hot melt extrusion) can be investigated as this has been successfully implemented for class IV (low permeability, low solubility) compounds based upon the Biopharmaceutics Classification System [47].

M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT
21 diluted with water (20 mL) and extracted with dichloromethane (3x50 mL). The collected organic fractions were washed with brine, dried over Na 2 SO 4 and evaporated under reduced pressure. The resulting residue was purified by filtration through a pad of silica, using a mixture of CH 2 Cl 2 /Toluene/ EtOAc (7.5:2:0.5). Compound 18 was obtained as a yellow solid in 16% yield.   TbGAPDH activity was assayed spectrophotometrically by following NAD + reduction at 340 nm in triethanolamine (TEA) buffer (10 mM TEA, 1.7 mM NaHCO 3 , 100 mM KCl, 5 mM MgSO 4 , and 1 mM ethylenediaminetetraacetic acid (EDTA)) pH 7.6 at 25 °C, as reported by Wiggers et al [48].
The reaction mixtures included TEA buffer, pH 7.6, TbGAPDH (ranging from 200 to 300 µg of protein), 400 µM NAD + , 500 µM KH 2 PO 4 , 300 µM GAP, and varying concentrations of inhibitors in a total volume of 1 mL. Stock solutions of the inhibitors were prepared in 100% DMSO (v/v).
Inhibition data are presented as percentage of GAPDH activity. Experiments were carried out in triplicate.

TcTR inhibition assays and kinetic analysis
Trypanothione disulfide (TS 2 ) and recombinant TcTR were prepared according to published procedures [32,49]. Stock solutions of the inhibitors were prepared in DMSO (v/v). The kinetic analyses were performed by using a Jasco V650 spectrophotometer. The activity of TcTR was measured at 25 °C in a total volume of 1 mL assay buffer (40 mM HEPES, 1 mM EDTA, pH 7.5 [33]) containing 100 µM NADPH and 5-10 mU enzyme in the absence and presence of the inhibitor. Each assay contained a total of 5% DMSO (v/v). The reaction was started by adding TS 2 , and NADPH consumption was followed at 340 nm. The type of inhibition was derived from the respective Lineweaver-Burk plots. The activity of TcTR was measured in the absence and presence of two fixed concentrations of inhibitor varying the concentration of TS 2 (20, 40, 60, 100 and 200 µM). The inhibitor constants were calculated from the direct plot using non-linear least-squares data fitting in Microsoft Excel [50,51].

M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT 27 acquisition, data processing and normalization were performed as described for the L. infantum assay above. The host toxicity value was calculated based on the following equation: host toxicity = 1 -(CRS/CRN) x 100; where CRS = cell ratio in compound/sample treated well and CRN = average cell ratio in negative control wells.
6.3.6. In-vitro compound mediated early-toxicity assessment 6.3.6.1. Cytotoxicity assay against A549 cells The assay was accomplished using the Cell Titer-Glo TM assay from Promega as described in the literature [54,55]. Briefly, the assay detects cellular ATP content with the amount of ATP being directly proportional to the number of the present cells (A549 cells were obtained from DSMZ (German Collection of Microorganisms and Cell Cultures), Braunschweig, Germany. Paclitaxel was used as a positive control at 10 µM.

hERG assay
The hERG inhibition assay was performed using the Invitrogen Predictor TM hERG Fluorescence Polarisation Assay. Briefly, the assay uses a membrane fraction containing hERG channel (Predictor TM hERG Membrane) and a high-affinity red fluorescent hERG channel ligand, or "tracer" (Predictor TM hERG Tracer Red), whose displacement by test compounds can be determined in a homogenous, Fluorescence Polarization format, as described in the literature [54,55].

Cytochrome P450 1A2, 2C9, 2C19, 2D6 and 3A4 assays
The CYP inhibition assays were carried out by means of the Promega P450-Glo TM assay platform, as described in the literature [54,55]. Briefly, action of the CYP450 enzymes upon each substrate results in the generation of light and a light decrease was indicative of inhibition of the enzymes.