Phthalimide Derivatives with Bioactivity against Plasmodium falciparum: Synthesis, Evaluation, and Computational Studies Involving bc 1 Cytochrome Inhibition.

We describe herein the design and synthesis of N-phenyl phthalimide derivatives with inhibitory activities against Plasmodium falciparum (sensitive and resistant strains) in the low micromolar range and noticeable selectivity indices against human cells. The best inhibitor, 4-amino-2-(4-methoxyphenyl)isoindoline-1,3-dione (10), showed a slow-acting mechanism similar to that of atovaquone. Enzymatic assay indicated that 10 inhibited P. falciparum cytochrome bc 1 complex. Molecular docking studies suggested the binding mode of the best hit to Qo site of the cytochrome bc 1 complex. Our findings suggest that 10 is a promising candidate for hit-to-lead development.

Introduction

Malaria is a severe parasitic disease endemic in 91 countries. According to the WHO, 216 million cases and approximately 445.000 deaths globally were reported in 2016.[1] The disease can be caused by five species of Plasmodium spp (Plasmodium falciparum, Plasmodium vivax, Plasmodium ovale, Plasmodium malariae, and Plasmodium knowlesi). P. falciparum is the major pathogen responsible for deaths, and it is associated with severe malaria.[2] The malaria parasite has developed a resistance to many of the current drugs, including emerging P. falciparum resistance to artemisinin derivatives (1), which is a component of artemisinin-based combination therapies that comprise the current first-line therapies.[3] In this context, new safe and affordable antimalarial drugs are desperately needed. Structure-based approaches have been used to develop new antimalarial candidates with well-defined mechanisms of action.[4−6] The cytochrome bc1 complex contains a coenzyme Q, a cytochrome c reductase, and the ligand ubiquinone. The enzymatic complex catalyzes electron transfer in the mitochondria of the parasite, which maintains a stable membrane potential for this organelle.[7,8] Atovaquone (2) (Chart ), a ubiquinone analogue, was approved as a first line drug for the treatment of malaria in 1992.[9] Atovaquone has a well-defined mechanism of action; it binds to the oxidation site of the cytochrome bc1 enzyme complex.[2,8,10] As a competitive inhibitor of ubiquinone, the inhibitor blocks cellular respiration because of the reduction of electron transfer in the mitochondria of Plasmodium spp, which consequently promotes a collapse in the maintenance of the mitochondrial membrane potential.[10] The ubiquinone binding mode indicates that the molecule interacts with a P center of the bc1 enzymatic complex, inducing a conformational change in the Rieske protein at Fe–S bond, which allows the electron transfer. The atovaquone mode of action consists of the inhibition of the bc1 complex, resulting in conformational changes that can affect the electronic transfer to the cytochrome c.[9,11]

Chart 1

Approved Drugs for the Treatment of Malaria: Artemisinin (1) and Atovaquone (2); 3 (SL-2-25) and 4 (WR249685): Potent Compounds with in Vitro Inhibitory Activity against Whole Parasite (1) or the bc1 Complex (2–4), and Their Similarities to 2 Are Highlighted in Red

Recent studies have shown that the inhibition of the bc1 complex also affects the electron acceptor enzyme known as dihydroorotate dehydrogenase (DHODH). This enzyme plays an important role in the biosynthesis of pyridines; therefore, inhibiting this enzymatic complex indirectly impacts the biosynthesis of pyridines in the parasite as well as the transport of electrons in the mitochondria.[9,12]

Genetic mutations causing atovaquone resistance have been described since the early 1990s.[13,14] The replacement of Tyr268 with serine (Y268S) or asparagine (Y268N) disrupts the chemical interactions between atovaquone and the active site residues of the bc1 complex. Therefore, the investigation of new inhibitors is necessary to reduce the potential for drug resistance. Compounds 3 (SL-2-25) and 4 (WR249685) are examples of atovaquone analogues that are potent bc1 complex inhibitors (Chart ).[15,16]

Here, we designed N-phenyl-substituted phthalimides (5–16) (Scheme ) as surrogates for the atovaquone core. The approach consisted of the maintenance of the two aromatic subunits in addition to the 1,4-naphthoquinone moiety (A) including a para-substituted phenyl ring (B) with different electronic and steric properties. The 3-hydroxy group on the 1,4-naphthoquinone moiety was replaced with a monosubstituted amine (C) to maintain the hydrogen-bonding abilities. Atovaquone has shown better oral bioavailability when the 3-hydroxyl substituent was replaced with a hydrophobic group (OCH3),[11] and a benzyl group (C) was introduced to investigate its impact on the biological properties. It is worth mentioning that the designed compounds (5–16) follow the rule of three for fragment-based drug discovery, that is, a molecular weight lower than 300 Da, up to three hydrogen bond acceptors and a calculated log P < 3.[17]

Scheme 1

Structural Design of the N-Phenyl Phthalimide Fragments from Atovaquone

Results and Discussion

In the synthesis of the N-phenyl-substituted phthalimide derivatives (5–16) (Scheme ), 3-nitrophalic acid (17) was refluxed in acetic anhydride for 16 h.[18] Then, the nitro group of anhydride 18 underwent catalytic hydrogenation with Pd/C (3%) resulting in 19.[19] In a divergent synthetic step, 3-aminophalic anhydride (19) was refluxed for 7.5 h in glacial acetic acid with the appropriate para-substituted aniline (20–25) to afford fragments 5–10.[20] Finally, these aminophthalimides (5–10) were dissolved in N,N-dimethylformamide with benzyl bromide (26) to form the N-benzylphthalimides (11–16).[21] See the Supporting Information for more details regarding the experimental procedures and the characterization data of the products and intermediates.

Scheme 2

Synthetic Route Used to Obtain the N-Phenylphthalimides (5–16)

Compounds 5–16 were evaluated in vitro against P. falciparum (3D7 strain—chloroquine sensitive) by the SYBR green assay.[22] The compounds were tested at single concentration (10 μM), and IC50 values were determined for compounds that showed inhibitions >60%. Compounds 5–9 and 11–15 were inactive as they presented IC50Pf values higher than 10 μM (Table ).

Table 1

Inhibitory and Cytotoxic Activities of the Phthalimide Derivatives against P. falciparum (3D7 Strain) and HepG2 Cellsa

Cpd	% of inhibition @ 10 μM	IC50Pf (CI95%) (μM)	IC50HepG2 (μM)	SI*
5	15 ± 2	>10	nd	nd
6	3 ± 2	>10	nd	nd
7	5 ± 3	>10	nd	nd
8	6 ± 3	>10	nd	nd
9	4 ± 2	>10	nd	nd
10	75 ± 4	4.2 (3.2–5.2)	>250	>59
11	6 ± 2	>10	nd	nd
12	5 ± 2	>10	nd	nd
13	7 ± 3	>10	nd	nd
14	23 ± 3	>10	nd	nd
15	8 ± 3	>10	nd	nd
16	65 ± 5	6.8 (5.6–8.0)	>250	>36
artesunate		0.006 (0.005–0.008)	305 (281–329)	50.833

The presented IC50 values are the average of two independent experiments. *SI = IC50HepG2/IC50Pf; CI95% = 95% confidence interval; nd = not determined.

Conversely, compounds 10 and 16 inhibited in vitro P. falciparum growth at low micromolar concentrations (IC50Pf values of 4.2 and 6.8 μM, respectively). These data suggest that a para-methoxy substituent is beneficial for the inhibitory activity in this series. Moreover, the bulky benzyl substituent at site C was tolerated, suggesting that structural variation at this position can be further explored. Next, the cytotoxic effects of the active compounds were evaluated against a human hepatoma cell line (HepG2) by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide viability assay.[23] Compounds 10 and 16 showed low cytotoxicities (IC50HepG2 > 250 μM) and considerable selectivity indices (SI values of >36 and >59, respectively) (Table ). It is worth mentioning that both compounds were soluble at the highest concentration tested (250 μM).

To better characterize the antiplasmodial activities and confirm the low cytotoxicities of the series, the most potent phthalimide derivative (10) was selected. The morphology of HepG2 cells was evaluated after 24 h of treatment with 250 μM of 10. At this concentration, we verified that compound 10 was not toxic to the cells as they showed a microscopic morphology similar to that of untreated cells (Figure A).

Figure 1

(A) Morphology of HepG2 cells before and after incubation with 10 at 250 μM. Insets: Expansion showing representative cell morphology; (B) microscopy of synchronized parasites continuously treated with 10 at 40 μM and the controls [dimethyl sulfoxide (DMSO), 10, atovaquone, and artesunate]. Representative images of two independent experiments.

Next, the inhibitory activity of 10 was evaluated against the P. falciparum K1 strain, a strain that is multidrug-resistant to the antimalarial drugs chloroquine, pyrimethamine, and sulfadoxine. Compound 10 was active in the low micromolar range against the resistant parasite with an IC50 value of 4.3 μM, which is approximately the same inhibitory activity observed against the sensitive 3D7 strain (Figure ). To elucidate the stage-specific inhibitory activity of 10, we incubated the compound with highly synchronized parasites at a concentration 10-fold greater than its IC50 value and observed the morphological changes over time. As seen in Figure B, compound 10 showed activity in the later ring stages (late rings or early trophozoites) and it induced alterations in P. falciparum morphology between 16 and 32 h after incubation. These data suggest a stage-specific effect that may be related to a slow-acting mechanism in which the latter forms of P. falciparum in the intraerythrocytic cycle are susceptible to the effects of the compound. These results are in good agreement with the stage-specific inhibitory activity of atovaquone, which is a slow-acting inhibitor of P. falciparum.[24]

Figure 2

Representative concentration–response curves of 10 against sensitive (3D7 strain) and resistant parasites (K1 strain). The reported IC50 values are the mean values of two independent experiments.

Aiming to elucidate the effects of compound 10 on P. falciparum mitochondrial respiration, an enzymatic assay was conducted to measure bc1 complex decylubiquinol-cytochrome c oxidoreductase activity.[25] Mitochondrial fractions were extracted from P. falciparum 3D7 parasites, and the assay was performed using a compound 10 concentration of 70 μM, in parallel with negative (DMSO) and positive (atovaquone at 6 μM) controls. Our results showed that 10 inhibited cytochrome bc1 enzymatic activity (74% at 70 μM) and demonstrated that cytochrome bc1 complex is a molecular target of 10 antiplasmodial activity (Figure ).

Figure 3

Inhibition of bc1 complex activity after incubation with atovaquone (ATV) at 6 μM (84 ± 1% of inhibition) or compound 10 at 70 μM (74 ± 2% of inhibition). DMSO was used as a negative control (1 ± 3% of inhibition). **p < 0.005 by Mann–Whitney test.

The mode of atovaquone binding to the cytochrome bc1 complex indicated that the carbonyl group alpha to the hydroxyl group is important for ligand binding and recognition. In addition, the hydroxyl group is involved in a salt bridge with the nitrogen atom of the imidazole moiety of histidine in the Rieske protein.[10,11] Finally, hydrophobic interactions and van der Waals forces are important for stabilizing atovaquone binding. For example, the cyclohexyl subunit and the 4-chlorophenyl substituent interact with the side chains of the V146, I269, P271, M139, L275, M295, and F296 residues from the Saccharomyces cerevisiae cytochrome bc1 complex.[26] Accordingly, on the basis of the structural similarities and in vitro cellular findings, which indicated that phthalimide derivative 10 shares a slow-acting inhibitory mechanism with atovaquone, we modeled the binding mode of 10 to the cytochrome bc1 complex.

The crystal structure of the S. cerevisiae cytochrome bc1 complex co-crystallized with stigmatellin was solved at 1.9 Å resolution (PDB ID: 3CX5)[27] and has been used as a molecular target for modeling studies. The observed binding mode of atovaquone co-crystalized with a cytochrome bc1 complex (PDB ID: 4PD4, resolution: 3.04 Å)[26] was used as a reference. Superposition of the two crystal structures showed no significant differences in the orientation of the amino acid residues within the protein binding sites, suggesting that the higher resolution structure is adequate for docking studies. The cytochrome bc1 complex from S. cerevisiae was prepared for rigid docking using GOLD software (v5.4).[28] The hydrogen atoms were added to the structure, and the water molecules and ligands were removed. The 3D structure of the hydroxyl group of atovaquone was built in the ionized state because previous studies showed that this microspecies is responsible for the activity.[26]

To evaluate the accuracy of the method, including the ChemPLP scoring function,[29] we redocked atovaquone into the cytochrome bc1 complex binding site. The best-ranked binding mode was compared to the binding mode of atovaquone in the crystal structure, and both the orientation and interactions with key residues were reproduced (RMSD = 0.4 Å; PLPscore = 63.3), which confirms the accuracy of this strategy for the system under investigation (see the Supporting Information for details). On the basis of that, we used the validated protocol to predict the binding mode of 10 within the cytochrome bc1 complex binding site. According to the model, the phthalimidic moiety of 10 establishes hydrogen-bonding interactions with the side chains of E272 and H181. Additionally, this moiety is in close contact with the side chains of I147, V146, and P271 and with the backbone atoms of W142. The methoxyphenyl substituent forms hydrophobic interactions with the side-chain atoms of M295 and L275. The ChemPLP score for this pose is 63.6, which is close to the calculated score for the atovaquone binding mode (Figure ).

Figure 4

Docking pose of 10 (orange stick) in the binding site of the S. cerevisiae cytochrome bc1 complex (PDB ID 3CX5). Main interacting residues are indicated as green sticks. Hydrogen bonds are represented as yellow dotted lines. Residues I269, V270, P271, E272, L275, Y279, and M295 (black label) are located at the cytochrome b subunit, whereas H181 (red label) is located at the Rieske protein subunit. Residues W142, V146, I147, and all hydrogens atoms were omitted for clarity. Figure generated using PyMol (v 0.99).[30]

As highlighted in the sequence alignment (Figure S52),[31,32] all of the mentioned residues interacting with the phthalimidic moiety are conserved in the P. falciparum cytochrome bc1 complex (Uniprot Q02768/Q8IL75), apart from M295 and L275. Residues M295 and L275 from bc1 of S. cerevisiae are replaced with V284 and F264, respectively. In P. falciparum, both substitutions can establish hydrophobic interactions with the methoxyphenyl substituent of 10. Moreover, the alignment of S. cerevisiae, P. falciparum, and Mycobacterium tuberculosisbc1 sequences shows that the binding site of 10 is conserved within these species. The residues H181, I269, and P271 from bc1 complex of S. cerevisiae are conserved in P. falciparum and M. tuberculosis, whereas V146 and I147, conserved in P. falciparum, are conservative replaced by alanine and leucine, respectively, in M. tuberculosis. Most of nonconserved residues within the three species stabilizes compound 10 by hydrophobic and van der Waals interactions. Because a potent antimycobacterial activity is observed for phthalimide-containing compounds whose target is the bc1 complex of M. tuberculosis,[33] the binding site identity is an indicative that the bc1 complex might be the molecular target of compound 10.

Conclusions

Phthalimide-based derivatives are attractive molecules for drug discovery and development. Because of the potential biological activity and pharmaceutical use, the privileged phthalimide scaffold was designed to be used in many therapeutic treatments, including antituberculosis as cytochrome bc1 inhibitor,[33] anti-inflammatory compound developed as a phosphodiesterase-4 inhibitor[34] and for antimalarial as plasmepsin II & IV inhibitor.[35] In this work, the best phthalimide derivative, compound 10, exhibited antimalarial activity against sensitive and resistant P. falciparum strains in the low micromolar range (IC50 < 10 μM), no cytotoxicity against HepG2 cells, a considerable selectivity index (SI > 30), and a slow-acting mechanism, which is consistent with the inhibitory stage of action of atovaquone. The modeled binding mode of 10 suggested the molecular determinants that might be related to the inhibitory activity of this series. Therefore, our findings indicate that 4-amino-2-(4-methoxyphenyl)isoindoline-1,3-dione (10) is a new hit for the development of lead compounds with superior properties.

Computational Methods

Initially, all molecular structures of the p-phenyl-functionalized phthalimidic compounds were generated with Discovery Studio Visualizer free software (DSV/version: 17.2.0)[36] and were optimized using the semiempirical method PM6 in MOPAC free software (version 2016)[37] through the Mercury CSD software interface (v3.9), which is included in the CCDC GOLD Suite computational package. The commands used during optimization were: MMOK, XYZ, CHARGE = −1 [for docking of atovaquone (2)] and GNORM = 1. The format of the optimized compound file was *.mol2. Atovaquone, the reference ligand for this work, was modeled in the ionized state.[26] In parallel, the 3D file (*.pdb) of the S. cerevisiae cytochrome bc1 crystal structure in complex with stigmatellin was obtained from Protein Data Bank (PDB ID: 3CX5) at 1.9 Å resolution.[27] The structure was chosen because atovaquone is a high affinity inhibitor of S. cerevisiae complex III (Ki = 5 nM), a nonpathogenic surrogate model for studying biological activities of compounds that target the parasite through this mechanism. Although the crystal structure contains a ligand other than atovaquone, we chose it because the cytochrome bc1 crystal structure in complex with atovaquone (PDB ID: 4PD4)[26] lacks the Rieske protein and has lower resolution in comparison with the 3CX5 structure. Superposition of 3CX5 and 4PD4 indicates no significant differences in the position of the atoms in amino acid residues of the binding site, thereby suggesting that the structure is suitable for the docking calculations using atovaquone’s binding mode in 4PD4 as a reference for evaluation of the model. The 3CX5.pdb file was then prepared for docking using GOLD software (v5.4),[28] where the hydrogen atoms were added to the structure. All water molecules and ligands were removed. The rigid docking was performed in GOLD using coordinates x = 13.0858, y = 5.3280, and z = 17.4635 as the center of a 10 Å search radius. GOLD was set to consider possible ligand internal hydrogen bonds in the calculations and score 50 binding modes using the ChemPLP scoring function. The selection of this scoring function was based on a previous article by Li et al. that indicates best accuracy for ChemPLP in comparison with several other score functions.[29] The best GOLD docking results for each ligand were analyzed using Discovery Studio Visualizer free software to check the ligand–cytochrome bc1 complex interaction types. The figures were generated in Pymol (v0.99).[30]