Analysis of the interaction of antimalarial agents with Plasmodium falciparum Glutathione Reductase through molecular mechanical calculations

doi:10.21203/rs.3.rs-3952252/v1

Download PDF

Research Article

Analysis of the interaction of antimalarial agents with Plasmodium falciparum Glutathione Reductase through molecular mechanical calculations

https://doi.org/10.21203/rs.3.rs-3952252/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 23 May, 2024

Read the published version in Journal of Molecular Modeling →

You are reading this latest preprint version

Malaria remains a significant global health challenge, with emerging resistance to current treatments necessitating the development of novel therapeutic strategies. P. falciparum Glutathione Reductase (PfGR) plays a critical role in the defense mechanisms of malaria parasites against oxidative stress. In this study, we investigate the potential of targeting PfGR with conventional antimalarial drugs and dual drugs combining aminoquinoline derivatives with GR inhibitors using molecular docking and molecular dynamics simulations. Our findings reveal promising interactions between PfGR and antimalarial drugs, with the naphthoquinone Atovaquone (ATV) demonstrating particularly high affinity and potential dual-mode binding with the enzyme active site and cavity. Furthermore, dual drugs exhibit enhanced binding affinity compared to reference inhibitors, suggesting their efficacy in inhibiting PfGR. Insights into their interaction mechanisms and structural dynamics are described. Overall, this research provides valuable insights into the potential of targeting PfGR and encourages further exploration of its role in the mechanisms of action of antimalarial drugs, including dual drugs, to enhance antiparasitic efficacy.

Malaria

Plasmodium falciparum

antimalarials

Glutathione Reductase

PfGR

Molecular Docking

Malaria is a condition resulting from the infection by protozoa of the genus Plasmodium (P). These microorganisms can affect birds, reptiles, and mammals, including humans, and there are approximately 200 species of Plasmodium in circulation [1]. Among these species, five are capable of infecting humans and causing the disease: P. falciparum, P. vivax, P. malariae, P. ovale, and, more recently, P. cynomolgi and P. simium. Although the latter two were initially reported in non-human primates, there are records of naturally acquired infection cases in humans [2–4].

According to the World Health Organization (WHO), the total number of disease cases reached 249 million in 2023. The majority of cases (82%) and deaths (95%) occurred in the African Region, followed by the Southeast Asia Region (cases 10% and deaths 3%). Recent records indicate that the major contributors to Malaria worldwide are the species P. falciparum and P. vivax, with the former responsible for the highest number of fatalities, particularly in the African continent [5]. Regarding Brazil, the Amazon region is indeed considered the endemic epicenter of Malaria in the country, accounting for 99% of recorded autochthonous cases. Outside this area, the majority of reported cases originate from endemic states or countries, representing over 80% of cases. In terms of analyses conducted by the Ministry of Health in 2023, there was a decline in Malaria notifications between 2010 and 2016. However, in 2017, there was a significant increase of 38.7% compared to the previous year, as highlighted in Fig. 1 below. In the subsequent years, a reduction in notifications was observed; however, the Ministry of Health raised alarms due to an unexpected increase of approximately 27.8% in 2020, in cases of Malaria associated with P. falciparum or mixed P. falciparum/P. vivax Malaria (red dots in Fig. 1) [6, 7]

Malaria is transmitted to the vertebrate host through the bite of female mosquitoes of the genus Anopheles. The sporozoites from the salivary glands infect the human host through the insect's bite. In the liver tissues, the sporozoites multiply through asexual reproduction, forming thousands of merozoites. Released into the bloodstream, these invade red blood cells and initiate the asexual erythrocytic phase, causing disease symptoms in the human host after 48 hours of infection. Some of the merozoites may develop into male or female gametocytes. When a mosquito bites an infected person, the gametocytes are ingested and mature in its digestive system in around 24 hours. Thus, mosquitoes become ready to infect a new host, completing the transmission cycle [8, 9]. During the asexual erythrocytic phase, the parasite digests between 60% and 80% of hemoglobin, transporting it to its digestive vacuole (DV) to process it into an inert and non-toxic material known as hemozoin, which does not affect its survival [10]. In this same phase, the parasite is exposed to oxidative stress generated by the host's immune system to combat the infection, as well as by the heme group and other decomposition products of hemoglobin [11].

Glutathione (GSH) plays a fundamental role in the defense of Malaria parasites against oxidative stress. This tripeptide acts as an antioxidant, aiding the parasite in neutralizing reactive oxygen species both directly and through reactions catalyzed by glutathione peroxidase and glutathione S-transferase [12]. Thus, high levels of reduced GSH are maintained in the erythrocytes of infected hosts (asexual erythrocytic phase) thanks to the action of the enzyme Glutathione reductase (GR) [13]. GR is a homodimer that functions as a disulfide oxidoreductase and utilizes the cofactors FAD (prosthetic group) and NADPH to reduce one molar equivalent of GSSG to two molar equivalents of GSH (GSSG + NADPH⁺ → NADP⁺ + 2GSH) [13, 14].

The crystal structure of P. falciparum GR (PfGR) was reported in 2003 by Sarma et al. Similar to human GR (hGR), each subunit of PfGR (Fig. 2) has two cysteine residues in the active site (Cys39 and Cys44) that mediate electron transfer [15]. In addition to the active redox pair Cys39/Cys44, the active site includes the cofactor FAD, which is reduced by NADPH after binding to the oxidized enzyme (2NADPH + FAD → 2NADP⁺ + FADH₂ + 2e^-). Once reduced, the isoalloxazine ring of FADH^- packs against the persulfide, forming a charge transfer complex, reducing the Cys39-Cys44 disulfide. Thus, oxidized glutathione (GSSG) binds to the active site to be reduced (GSH), and the cysteine redox pair reforms in the disulfide form. Isoalloxazine and 1,4-naphthoquinone derivatives, methylene blue, ajoene, among others, are considered strong competitive inhibitors of both human (hGR) and parasite (PfGR) enzymes, binding to their active sites [16–18]. Studies on the mechanisms of inhibitors such as methylene blue (MB) suggest that it acts as a "subversive substrate" in the enzyme's active site, oppositely altering its physiological function. The enzyme catalyzes the reduction of methylene blue by FADPH after reduction by NADPH. The corresponding resulting reduced species (Leucomethylene blue) is a more efficient auto-oxidant, oxidized by O₂. From a cellular pharmacological perspective, each reaction cycle catalyzed by the GR-MB complex leads to the consumption of NADPH and O₂, and the production of parasitotoxic reactive oxygen species, predominantly H₂O₂ [16].

The most pronounced differences between the two enzymes (hGR and PfGR) are found in the cavity region at the dimer interface (Fig. 2), causing changes in the electrostatic properties of this region [15]. In addition to the active site, this cavity is also considered a binding site for non-competitive enzyme inhibitors, known as the allosteric site. Unlike the active site, the interface cavity does not confer selectivity regarding ligand binding [19, 20]. In recent years, this region has also been studied for the selective design of drugs [21]. Inhibitors of PfGR such as Xanthane (6-hydroxy-3-oxo-3H-xanthene-9-propanoic acid) and Menadione (2-methyl-1,4-naphthoquinone) bind to this cavity, potentially causing non-competitive enzyme inhibition [19]. Some studies indicate that MB can also act as a non-competitive inhibitor of GR and PfGR, probably by binding to the cavitiy of the enzyme [22].

Studies have shown that erythrocytes deficient in GR were able to fulfill their physiological functions, albeit with a shorter lifespan. However, they did not serve as host cells for P. falciparum. In fact, the depletion of GSH by GR inhibitors in erythrocytes infected with P. falciparum produces a drastic antiplasmodial effect [12]. Thus, the search for effective inhibitors of PfGR has become a promising strategy for the development of new and effective antimalarial drugs, aiming to overcome current concerns about the disease, given the emergence of strains resistant to first-line treatment (combined therapy with artemisinin). In addition to the defense of Malaria parasites against oxidative stress, studies have highlighted the relationship between GSH and PfGR with the development of resistance to antimalarial drugs. This is supported by the fact that GSH levels increase in mice infected with CQ-resistant strains. Dubois et al demonstrated that, despite CQ treatment not influencing intracellular GSH levels in erythrocytes infected with resistant or sensitive strains, the activity of the enzymes GR and GPx (Glutathione peroxidase) significantly decreases [12]. Consistently, the depletion of GSH by L-buthionine-(S,R)-sulfoximine (BSO), an inhibitor of GSH synthesis, in resistant strains of P. falciparum, restores sensitivity to CQ and its analogs [23].

In response to the challenges posed by parasite resistance and adverse effects, various antimalarial drugs have been developed, targeting different stages of the parasite's life cycle [24]. These drugs, classified into classes such as aminoquinolines (e.g., chloroquine (CQ), amodiaquine (AQ), primaquine (PQ), mefloquine (MQ)), naphthoquinones (e.g., atovaquone (ATV)), antifolates (e.g., proguanil), and artemisinin derivatives, aim to combat malaria effectively [25]. Existing research suggests that 4-aminoquinoline antimalarials and artemisinin derivatives function during the intraerythrocytic stage by binding to the heme group, preventing its sequestration into hemozoin (Hz) [26, 27]. 8-aminoquinolines like PQ are believed to induce oxidative stress by accumulating H₂O₂ in the liver, leading to Plasmodium parasite toxicity [28] Despite these valuable findings, details of this mechanism of action are not well understood to date. PfGR, a disulfide reductase enzyme, has not been extensively studied as a potential target for these drugs. Motivated by this gap, our study explores the in silico interaction of quinoline-derived drugs and 1,4-naphthoquinones with the PfGR enzyme through molecular mechanical calculations. Additionally, we investigate double-drugs proposed by Davioud-Charvet et al. [29], combining quinoline-based alcohols with derivatives of 1,4-naphthoquinones, to understand their efficacy in inhibiting PfGR activities. Our discussion, based on binding energy results and graphical analysis of drug-receptor interactions, provides insights valuable for designing potent inhibitors for this essential enzyme in Malaria parasites.

Molecular docking studies were conducted to determine the binding affinity and interactions between candidate inhibitors and the PfGR enzyme (Fig. 2) and some results were better discussed via a molecular dynamics simulation. The aminoquinolines CQ, AQ, PQ, MQ, and the naphthoquinone ATV were selected among commercial drugs available for malaria treatment as ligands for the present study. Additionally, we address the double-drugs strategy proposed by Davioud-Charvet et al. [29], which combines quinoline-based drugs comprising known antimalarial activity with 1,4-naphthoquinones (known as GR inhibitors), linked by a labile ester bond (Fig. 3). The structural characteristics were defined among the library of studied compounds that exhibited greater affinity with PfGR. The [2-(3-methyl)naphthoquinolyl]alcanoic acids of menadione and plumbagin had the strongest inhibitory activity, where the most active of this last series led to more than 80% inhibition of both hGR and PfGR [29]. In this way, the double-drugs 1–3 represented in Fig. 3 were also selected as ligands for the present study. The ligands 1c, 3a-c were designed by the authors of the present study. The proposed ligands 3a-c, combine the inhibitor Menadione (MD) directly with quinoline drugs, unlike ligands 1 and 2, which have an ester bridge with a variable number of carbon atoms in the aliphatic part (n) connecting both residues as depicted in Fig. 3.

The ligands represented in Fig. 3 were optimized, and characterized as minima on the potential energy surface in aqueous phase at the Hartree-Fock level, using the 3-21G basis set in the ORCA 5.0.3 program [30]. The solvent effect was accounted for using the CPCM method [31]. The choice of a more basic method is justified by the subsequent treatment by molecular mechanics methods since all structural properties are parameterized by the force field, and the initial structures only provide a reference for the program without requiring an advanced description of the electronic part.

The crystal structure of the PfGR enzyme was obtained from the Protein Data Bank (PDB) under the code 1NOF [32] and edited using Discovery Studio software [33]. Water molecules and the ligand were removed to prepare the system for molecular docking. Concerning the enzyme's cofactors, FADH₂ was kept near the active site of the biomolecule, which was reconstructed based on the incomplete structure present in the original file.

2.1 Molecular Docking

Molecular docking simulations were conducted for all ligands using the software package Autodock 4 version 4.2.6. [34]. Two regions were selected for structural testing, the enzyme's active site (site 1) and the intersubunit cavity (site 2), as depicted in Fig. 2. Both were determined at specific locations by fixing a rectangular box, measuring 60Å-70Å-60Å and centered at x:73.371; y:67.404; z:80.181 for site 1, and measuring 50Å-100Å-60Å and centered at x:62.671; y:38.441; z:88.822 for site 2, with a spacing of 0.4 Å. Lamarckian algorithms were employed to obtain 20 target-ligand interaction poses in a population of 250 structures. Images of the poses were generated using Discovery Studio Visualizer 2019 software [33] and Chimera 1.17.3. [35].

2.2 Molecular Dynamics

The GAFF2 [36, 37] force field was employed for the parameterization of biomolecules. It was used some explicit water molecules as solvent in a periodic boundaries condition of a truncated octahedron box, and the solvent was instantiated as TIP3P [38] with a 15.0 angstroms radius. The simulation protocol comprised two successive minimization steps: the initial step involved steepest descent cycles (1000) followed by cycles of conjugated gradients (1500). During the first minimization step, a restraint force constant of 500 kcal mol⁻¹ Å⁻² was applied to the solute, while in the second step, no restraint was imposed, allowing the entire system to undergo unconstrained minimization.

Subsequently, six intercalated heating and equilibrium steps were conducted, each involving a temperature rise of 50 K (but the last, which increase 60K all the way up to 310K). These steps were executed under constant volume periodic boundaries (NVT) over 2800 cycles (2fs in time), with frames of equilibrium (2fs) implemented under constant pressure periodic boundary conditions (NTP) at a mean pressure of 1 bar. The production phase extended over 100 ns, comprising frames with a temporal interval of 2 fs between each frame and a cutoff distance of 6.0 Å for non-bonded interactions. The simulation was conducted under constant pressure and temperature (NPT) conditions, with the temperature maintained at approximately 310K through the implementation of a Langevin thermostat, and the pressure regulated at 1 bar using the Barendsen barostat [39, 40], in order to simulate physiological condition. The SHAKE [41] algorithm was activated to restrict hydrogen stretching during the simulation. Molecular dynamics energy evaluations were performed using MM-GBSA profile, accessible through the MM-PBSA.py script [42]. All molecular mechanics simulations were performed in Amber 16 [43, 44].

In general, the geometric parameters of ligand structures (CQ, AQ, PQ, MQ, and ATV), optimized at the HF level were in agreement with the solid structures available on the CCDC Cambridge website. Bond lengths were overestimated in aqueous solution, with a maximum error of 1.7% and a maximum deviation of 0.15 Å in relation to the solid state. The error falls within the margin of the HF method, partly due to the solvent effect that tends to increase bond lengths in aqueous solution. Angular parameters also show satisfactory agreement with experimental values, with a maximum error of 5.8%.

3.1 Evaluation of conventional antimalarial drugs as PfGR Inhibitors

Initially, molecular docking studies were conducted with the drugs CQ, AQ, PQ, MQ, and ATV, compounds of pharmacological significance, particularly in the treatment of malaria. CQ and AQ are weak diprotic bases; therefore, at physiological pH (∼7.2), they can exist in non-protonated, monoprotonated, and diprotonated forms [45]. Thus, the effect of protonation on these two drugs was investigated.

3.1.1 Interaction studies in the active site (site 1).

The molecular docking data at the enzyme's active site are reported in Table 1. The binding energies and estimated constants of Table 1 indicate a higher affinity of the studied antimalarials when compared to reference inhibitors MD and MB. All the studied drugs exhibited a binding energy score (∆G_int) lower than − 6.0 kcal mol^− 1 (except for CQ in the neutral form). Despite the neutral CQ showing the lowest scoring of the (∆G_int =-5.89 kcal mol^− 1), among the series of antimalarials studied, this value is quite close to reference inhibitors MD (-5.53 kcal mol^− 1) and MB (-6.08 kcal mol^− 1).

Table 1

Binding affinity scores of commercial antimalarials with the active site of PfGR enzyme. The residues involved in interactions and antiplasmodial activities are also included.
Compound		∆G_int (kcal/mol)	K_i (µM)	#contacts/residues involved in the interaction			IC₅₀ (nM)		Ref
Compound		∆G_int (kcal/mol)	K_i (µM)	Hydrogen bonds	Hydrophobic	Electrost.	3D7	W2
CQ CQH CQH₂	-5.89 -6.31 -6.71		48.1 23.9 12.0	1/Val383 1/Pro381 2/Pro381,Asp458	11/FAD,Cys44, Val45,Lys48, Leu352,Pro354, Pro381,Val383. 7/ Pro381,Val383, Ile393,Val461, Ala465. 6/Tyr185,Leu352,Pro381,Thr382, Val383, Phe385.	0 0 1/Glu459	18 12–15	459 571	[46] [47]
AQ AQH AQH₂	-6.96 -6.85 -7.39		7.89 9.45 3.80	2/Asp458,Gln462 4/Gln462,Asp458 4/Pro381,Val383,Asp458,Glu459	6/Tyr185,Leu352,Pro354,Val383, Asp458,Glu459, Gln462. 7/FAD,Leu352, Pro354,Val383, Pro381 6/Tyr185, Leu352,Val383, Phe385.	0 1/Glu459 2/Lys48, Glu459	18 9.7	86.2 6.4	[46] [47]
PQ	-6.33		104.1	3/Glu32,Val383,Asp458,Glu459	7/ FAD,Leu352, Val461,Pro381, Val383.	0	104.1	1117	[46]
MQ (R,S)	-6.59		14.7	2/Gln462, FAD	11/ FAD,Tyr185, Val461, Val383, Pro354, Leu352, Pro381.	0	15	3.5	[48]
ATV	-8,10		1.15	3/Val383, FAD	9/ FAD, Cys44, Val45, Lys48, Ile49,Leu352, Pro354.	0	2.4	2.1	[49]
MD	-5.53		88.6 (82.2)^*	1/Gln462	4/Pro354,Pro381,Val383.	0	‒	‒	[50]
MB	-6.08		35.01 (42.2)^*	1/FAD	0	0	‒	‒	[50]

*Experimental K_m values for MD and MB reduction by PfGR enzyme. In this case, both inhibitors act as redox-cyclers subversive substrates. 3D7 is a chloroquine susceptible strain, whereas W2 is a chloroquine resistant strain.

In its most stable state, neutral chloroquine (CQ) (in orange; Fig. 4a) adopts an extended conformation. The quinoline ring is perpendicular to the isoalloxazine ring, engaging in a π-π T-shaped interaction (4.6 Å) and an N-H···Ph hydrogen bond (2.5 Å) with the drug's pyridine π electron cloud (Fig. 4a). H-bonds with Val383 assist in positioning the quinoline ring toward FADH₂ (Fig. 4a). Moreover, the diethyl-pentane-amine group approaches Cys44, essential for the enzyme's electron transfer intermediate, through carbon-hydrogen bonds and hydrophobic interactions (Fig. 4a). As expected, increasing protonation in CQ leads to more H-bond formation, reflected in the elevated score (Table 1, Figure S1). In CQH and CQH₂ forms, a more closed conformation is observed, especially in the monoprotonated form with a higher RMSD value of 3.23 Å compared to the neutral (1.90 Å) and diprotonated (1.94 Å) forms. In the diprotonated form (green; Fig. 4a), the quinoline ring points toward the enzyme surface, interacting with residues in the α-helix. Conversely, in the CQH form (navy blue), the structure inverts, directing the quinoline ring toward the interface and interacting with residues in the β-sheets (Fig. 4a).

Like CQH₂, AQ's quinoline ring approaches the FADH₂ while its diethylamino-phenol group moves toward the interface region (Fig. 4b). Only the monoprotonated form interacts with FADH₂, establishing weak π-π stacked interactions between the flavin and quinoline rings and a π-alkyl interaction involving the drug's chloride (Figure S2).

The diprotonated form AQH₂ scores slightly better than the neutral and monoprotonated forms, displaying the lowest binding energy (-7.39 kcal mol-1). This is attributed to the formation of strong H-bonds between the diethylamino-phenol group with residues Asp458, Glu459, and Gln462 (⁓2.0 Å) of the first α-helix in this interface (Fig. 4b; Table 1). Moreover, the electrostatic contribution increased, resulting from interactions of the diethylaminomethyl-phenol group with residues Lys48 and Glu459 in the interface (Table 1; Fig. 4b) and a better fit into the cavity of this region (Fig. 4b). Protonation has a more pronounced effect on the lead drug CQ than its derivative AQ, causing significant conformational changes and larger scoring alterations (Table 1). The 8-aminoquinoline PQ adopts a conformation similar to CQ and AQ. The methoxy group of the quinoline ring interacts weakly with FADH₂ through a π-alkyl bond (4.1 Å). The alkyl portion with the amine function heads towards the protein interface, forming H-bonds and hydrophobic interactions with the same residues involved in interactions with CQ and AQ (Table 1).

The MQ has two stereogenic centers, but its erythro racemic mixture ((11S,12R) and (11R,12S)) is used clinically [51]. The drug's stereochemistry was considered for the docking study. Both MQ enantiomers exhibited similar interaction energy values at the PfGR active site. However, (R,S)-MQ showed a slightly better score, boasting an interaction energy of -6.59 kcal mol-1 and an inhibition constant of 14.74 µM (Table 1). In the lowest energy orientation, the piperidine approaches FADH₂, forming a robust H-bond (1.89 Å) between the oxygen of FADH₂ and the nitrogen of the piperidine ring of MQ. Simultaneously, the quinoline portion with two –CF₃ groups addresses the enzyme interface, showcasing notable halogen bonding interactions F∙∙O (2.4 Å) and F∙∙N (2.9 Å), especially with Pro381(Figure S3).

The hydroxy-1,4-naphthoquinone ATV, in docking analysis, exhibits a ∆Gint value of -8.10 kcal mol-1 (best scoring), showcasing the highest affinity for the enzyme's active site. The estimated inhibition constant value (1.15 µM) is in agreement with reported values for 1,4-naphthoquinone in PfGR inhibition (2.2 µM) and hGR (1.3 µM) [17]. Similar to neutral CQ, ATV's naphthoquinone ring is oriented inward, perpendicular to the isoalloxazine ring (Fig. 4c). In fact, the same interactions are found: an H-bond N-H···Ph and a π-π (T-shaped) interaction of 4.63 Å between the isoalloxazine and quinoline rings. However, an extra weak amino-hydroxide H-bond (N-H∙∙OH; 3.08 Å) stabilizes the ligand-receptor arrangement (see Fig. 4c). More H-bonds with Val383 (Table 1) are formed in this system, aiding in positioning the NQ ring toward FADH₂. Moreover, the chlorophenyl-cyclohexyl group adopts a conformation against the plane of the NQ ring, weakly interacting with initial residues of the α-helix, including Cys44. This arrangement strengthens interactions at site 1, positioning the ligand close to FADH₂ and the active redox pair Cys39/Cys44, with ATV showing stronger interactions than CQ.

In the docking analysis for the MB inhibitor (Fig. 5), the best-scoring orientations are closely similar to the positioning of the quinoline rings of CQ and the 1,4-NQ of ATV in the FADH₂ binding site (Fig. 4a and 4c). Similar aromatic ring interactions (π-π, T-shaped) with FADH₂ and Val383 from the β-sheet are observed (Figs. 4a, c, and 5). In contrast, MD positions itself farther from FADH₂ (⁓4.9 Å) and weakly interacts with Val383, resulting in a lower score than MB (-5.65 kcal mol^-1). The Michaelis constants (K_m) for PfGR inhibition, obtained by following NADPH oxidation (42.2 µM for MB and 82.2 µM for Menadione), are in agreement with those estimated in the docking study (35.5 µM and 88.6 µM, respectively; Table 1) [50].

According to the vdW surface of the enzyme in Figs. 4a and 4c, it is noted that CQ and ATV dock similarly to the active site of the enzyme. It is observed that neither of the two ligands penetrate the cavity above FADH₂ (Figs. 4a and 4c). In the case of CQ, the presence of a substituent bulkier larger than chloride could fit better into this cavity, potentially providing extra stability to the drug-receptor complex. Analogously for ATV, with a substituent in positions 7 or 8 of the NQ ring. This cavity is partially filled by the dimethylamino substituent in the reference inhibitor MB, according to the second-best binding pose in this active site (pose 2; Fig. 5).

It is worth highlighting that 1,4-naphthoquinones are oxidant redox cyclers and can act as acceptors of electrons from different flavoproteins like glutathione-disulfide reductase. The reduction of these compounds by these latter enzymes results in the formation of semiquinone radicals or quinone dianion. These species lead to the generation of superoxide and peroxide through oxygen reduction and, ultimately, the regeneration of naphthoquinone [52]. Nevertheless, previous studies have emphasized that the reduction potential of ATV is low for efficient two-electron reduction under intracellular conditions (⁓ -0.26 V vs NADP) [18]. In contrast, the two-electron reduction potential for methylene blue is estimated at ⁓ -0.01 V and ⁓ -0.25 V for MD at pH 7.

In order to estimate the tendency to reduce the antimalarial drugs, the standard reduction potential (E^o) relative to the NADPH/NADP⁺ redox couple at pH = 7 was calculated for the antimalarial conventional drugs together with the inhibitors MD and MB (Fig. 3). The E^o was calculated in aqueous solution at level B3LYP/6–31 + G(d) [53, 54] (see more details in Supplementary Materials).

Although the estimated E^o values of ATV, MD and MB were underestimated (Fig. 6), the reduction trend is in agreement with the reported data. Overall, the reduction potential increased in the order PQ < AQ < AQH < MQ < CQH₂ < AQH₂ < MD < ATV < CQH < CQ < MB. MB exhibits the highest reduction potential in the studied series (-0.34 V), followed by CQ (-0.63 V) and ATV (-0.76 V) whose E^o value is pretty close to that estimated for the inhibitor MD (-0.81 V) as showed experimentally [18]. MQ exhibited a redox stability similar to CQH₂, while PQ was the most stable drug with the lowest potential in the series (Fig. 6).

The chloroquine reduction potential decreases significantly for the diprotonated form (CQH₂) (-1.38 V), where the electron density is more concentrated over the quinoline ring for this full protonation form, as evidenced by the molecular electrostatic potential maps (MEP) of Figure S4. A similar reduction is observed for AQH₂, with a slightly higher E^o (-1.13 V). For both reduced species, the greatest charge variation is observed on the 4-amino pyridine group, confirming its active participation in the reduction. The N-C² bond length increases while C²-C³ decreases, showing the loss of aromaticity and the formation of a structure with a localized double bond on the pyridine ring as in the quinolidine anion structure. Distortions in the planar quinoline ring geometry are evident in these reduced structures (Figures S4 and S5).

CQ and CQH exhibited a greater tendency to reduction (E^o⁓ 0.64 V) than their analogs AQ and AQH (E^o < -2.0 V). After CQ and CQH reduction, the C⁷‒Cl bond is broken, observing a high charge density on the quinoline ring's C⁷ and the leaving chloride, where the minimum MEP is found (Figures S4). In contrast, chlorine remains attached to the quinoline ring after the reduction of AQ and AQH, with no significant distortion of planarity observed. The electron density is distributed mainly on the p-hydroxyanilino and pyridine aromatic rings (Figure S5), disfavoring the reduction in relation to CQ and CQH₂.

Concerning ATV and MD, the formation of the quinone dianion occurs after a two-electron reduction, leading to structural modifications over the 1,4-dione ring. Similar structural changes over the quinoline ring are observed after the reduction of MQ and PQ (Figure S6).

Such electrochemical results suggest that ATV and CQ may not be as efficient "subversive substrates" as MB; however, they exhibit slightly higher E^o values to MD, indicating similar redox features to this inhibitor, which is considered a moderate redox-cycler drug [17]. Their similar redox characteristics to the inhibitor MD, along with their observed docking mode in the enzyme's active site, are interesting. The ligand-receptor arrangement promotes contacts necessary to position the ligand in a more lipophilic region near FADH₂ and the active redox pair Cys39/Cys44 (Fig. 4a,c). These last residues are crucial for the enzyme's redox reactions. If the ligand interacts in this vital pathway, it could potentially inhibit enzyme activity, possibly competing with GSSG. However, interactions with FADH₂ and Cys44 are weaker for CQ, and protonation disfavors its reduction, impacting their efficacy as redox-cyclers.

3.1.2 Interaction studies in the intersubunit cavity (site 2).

In addition to its active site, the interface region's cavity is considered another binding site of the PfGR enzyme. The interaction of the same ligands in Fig. 2 with the intersubunit cavity (site 2) was analyzed, and the results are presented in Table 2. The docking evaluation results again reveal ATV to have the highest affinity for this site 2, with a free binding energy of -9.28 kcal mol^-1 and an inhibition constant of 156.51 nM (Table 2). It is noteworthy that all the ligands analyzed increased their interaction energy values compared to site 1 and outperformed the scores for the inhibitors MD, Xanthane, and MB, except for PQ (Table 2). Nevertheless, a consistent trend in the stability of the drug-PfGR complex at the active site is observed in the cavity (ATV > AQ > MQ⁓CQ > PQ). These results suggest a higher affinity of antimalarial drugs for the cavity when compared to the active site.

Table 2

Binding affinity scores of commercial antimalarials with the homodimer intersubunit cavity of PfGR. The residues involved in interactions and antiplasmodial activities are also included.
Compound	∆G_int (kcal/ mol)	K_i (µM)	#contacts/residues involved in the interaction				IC₅₀ (nM)		Ref
Compound	∆G_int (kcal/ mol)	K_i (µM)	Hydrogen bonds		Hydrophobic	Eletrost.	3D7 -S	W2 -R	Ref
CQ CQH CQH₂	-6.38 -6.78 -6.34	20.90 10.66 22.41		1/Glu432. 3/Ser55, Asn456,Glu432. 3/Asp58,Asn62, Glu432.	8/ Ile426, Lys431,Leu455, Pro389,Pro388, His387,Phe421. 6/ Phe51, Ile59, Pro389 Leu455. 5/Asp58,Phe421,Try424,Leu45.	1/Glu432 3/Asp58, His387, Glu432. 2/ Asp58, Glu432.	18 12–15	459 571	[46] [47]
AQ AQH AQH₂	-7.83 -7.27 -8.61	1.83 4.69 0.49		4/Asp225, Lys228 2/Ser55, Glu432 3/Ser55, Asn62,Glu432.	7/His387,Pro389,Phe421,Tyr424,Ile426,Leu45. 2/ Ile426,Lys431. 3/ Ile426,Lys431, Tyr424.	2/Glu432 3/Asp58, Glu432 4/Asp58, Glu432	18 9.7	86.2 6.4	[46] [47]
PQ	-4.66	380.8		4/Ser55, Asn456,Glu432.	7/Phe51,Ile59, Leu455,His387, Pro388, Pro389.	3/His387, Glu432	104.1	1117	[46]
MQ (S,R)	-6.82	9.95		7/ Ser55,Ile59, Glu432,Asn456.	4/His387, Pro389, Leu455.	2/His387	15	3.5	[48]
ATV	-9.28	0.16		4/Asp58, Arg196,Asn229.	3/Phe51,His387, Leu455.	2/ Asp58, Glu432	2.4	2.1	[49]
MD	-5.65	72.74		1/Asn456	4/Phe51, His387, Leu455.	3/His387, Glu432	‒	‒	‒
Xantane	-5.81	54.72		1/Leu419	6/Leu419, Pro485,Thr486, Ala487.	0	‒	‒	‒
MB	-6.41	19.87		1/Leu419	3/Leu419.	0	‒	‒	‒

IC₅₀ data for PfGR inhibition by conventional antimalarial drugs are unavailable in the literature. Davioud-Charve et al. explored this property for ATV and analogs, reporting capabilities below 25 µM [55]. However, precise IC₅₀ values were hindered by compound precipitation in solution at doses exceeding 25 µM. To compensate for this absence of data, in vitro antiplasmodial activities against CQ-susceptible and resistant strains were included in Tables 1 and 2. These results demonstrate a notable correlation with docking interaction energies, where drugs with stronger antiplasmodial activity exhibit lower binding energy (better score), particularly in their interaction with the enzyme cavity (site 2). The correlation in this second binding site reveals a correlation coefficient, R² ⁓0.75 (Figure S7), observed in both CQ-sensitive and resistant strains.

All antimalarial drugs, including the MD inhibitor, dock within the same region of the cavity (Fig. 7). The docking takes place in more hydrophilic between the final residues of the enzyme's largest α-helix from the FADH₂ binding domain, linking the cavity with the active site (in blue, Fig. 7), and the sequence of parallel β-sheets in the interface (in red, Fig. 7). ATV and AQ have the ability to reach the innermost part of the enzyme by interacting with residues in the NADPH-binding domain region (in gray, Fig. 7) and the connecting loop between the last two β-sheets of the interface. Conversely, the inhibitor Xantane and MB are positioned at the monomer interface, proximate to the last α-helix and the β-sheet of this region (Figure S8). A similar binding position of inhibitor Xantane is found at the hGR dimer crystal structure [15].

In general, the complexation of the drugs with the cavity is generally stabilized by a greater number of H-bonds and electrostatic interactions due to the hydrophilic character of the region (Table 2). Concerning chloroquine, the monoprotonated form CQH demonstrates the most favorable interaction with the cavity (∆G_bind= -6.78 kcal mol^-1; Table 2). In contrast to the active site, CQH positions its quinoline ring towards the enzyme surface, engaging in H-bonds with residues from the β-sheets of the interface (Asn456) and the largest α-helix (Ser55) (Fig. 7). Π-sigma interactions with Leu455 and π-cation interactions with His387 contribute to the quinoline ring stabilization. Although the aliphatic chain has minimal interaction with the cavity, charge attractions occur between Glu432 and the protonated nitrogen of the tertiary amine within the cavity (Fig. 8a). Notably, the monoprotonated form establishes the highest number of contacts with the enzyme (Table 2), predominantly through its well-fitted quinoline ring on the cavity's surface (Fig. 8a), enhancing the stability of the PfGR-CQH complex.

In the protonated forms, the amodiaquine quinoline ring deepens into the cavity's inner part with the diethylamino-phenol group towards the interface's last two β-sheets (see Figs. 7 and 8b). In contrast, the neutral form of amodiaquine assumes a different orientation, placing its quinoline ring on the enzyme's surface. Simultaneously, the diethylamino-phenol group shifts towards the cavity's interior, establishing H-bond interactions with Lys228 and Asp225 (see Table 2 and Figure S9).

The AQH₂ form maintains its extended conformation upon interacting with the cavity, exhibiting minimal structural deformations. Its RMSD value is significantly smaller at 1.78 Å compared to AQH (4.70 Å) and AQ (4.48 Å). Both protonated structures' OH and secondary NH groups act as H-bond acceptors with residues Ser55 and Glu432 (⁓2.1 Å). Moreover, the additional proton from the quinoline ring in AQH₂ engages in an H-bond with Asn62 (2.3 Å), a terminal residue of the α-helix (Fig. 8b).

Protonation considerably increases electrostatic energy, resulting in a 0.81 kcal mol^− 1 increase between the neutral and di-protonated forms. Notably, protonated states exhibit a higher number of charge attractions between the diethylamino-phenol group and Asp58 and Glu432 (Table 2 and Fig. 8b). These interactions significantly contribute to the better binding score of AQH₂ (∆G_bind = -8.61) compared to the neutral form (∆Gint = -7.83). The effect of protonation on chloroquine and amodiaquine becomes more pronounced in this secondary binding site, characterized by a more lipophilic nature.

For this second site, the (S,R) enantiomer of MQ exhibited high score than (R,S) counterpart. The conformation adopted by (S,R)-MQ enhances contact and forms numerous hydrogen bonds with cavity residues, significantly contributing to the free interaction energy. Similar to CQH and neutral AQ, the quinoline ring of (S,R)-MQ, with its two –CF₃ groups, is oriented towards the surface (Fig. 7). Here, fluorine acts as a hydrogen bond acceptor with Asn456 of the interface and Ser55 of the α-helix (Fig. 8c). Additionally, the –CF₃ substituents engage in halogen interactions. The 2-piperidyl-methanol group also contributes to stabilizing the conformation through hydrogen bonds with Ser55 and Glu432 (Fig. 8c). However, the ligand structure of (S,R)-MQ does not conform to the cavity shape or align with the site compared to CQH and ATV (see vdW surface in Fig. 8c).

Docking results highlight ATV's tighter binding to the enzyme's cavity compared to the other studied drugs. The 1,4-NQ ring of ATV penetrates the cavity, reaching its innermost part (Fig. 7). The carbonyl oxygen serves as a hydrogen bond acceptor with internal residues Arg196 and Asn229, while the hydroxyl is directed towards the major α-helix, forming a robust hydrogen bond with Asp58 (OH∙∙∙O = CO; 2.0 Å) (Fig. 8d). Notably, residues such as Asp58, Asn62, His65, and Arg196 in PfGR, as identified by Sarma et al., contribute to altering the electrostatic properties of the cavity compared to the analogous human enzyme (hGR) [15]. Moreover, the NQ ring is stabilized by electrostatic interactions (π-ion), particularly with residues Asp58 (4.87 Å) and Glu432 (2.97 Å) (Fig. 8d), influencing the drug's positioning within the cavity. The same interactions are observed for the naphthoquinone ring of the reference drug Menadione; However, the charge attraction is weaker with Glu432 (3.67 Å) (Figure S10), and the lack of the extra polar group hydroxide as in ATV disfavor significantly the H-bond formation inside the cavity. Furthermore, the ATV's chlorophenyl-cyclohexyl group stretches across site 2, occupying a substantial cavity region and contributing to its heightened stability (see vdW surface Fig. 8d).

ATV stood out as the drug with the highest score for both investigated binding sites, showcasing as the best candidate to inhibit PfGR between the series of conventional antimalarial drugs studied. The potent antimalarial activity of ATV against both sensitive and resistant parasite strains might involve the inhibition of PfGR, complementing its established action mechanism, specifically, inhibiting the mitochondrial bc1 complex of P. falciparum and disrupting the mitochondrial electron transport chain [56]. This inhibition is likely attributed to the drug's binding in both the active site and the cavity within the interface region, exhibiting a dual docking mode.

3.2 Antimalarial dual drugs based on inhibitors of PfGR

3.2.1 Analysis via molecular docking simulations

Table 3 presents the energetic data obtained from molecular docking simulations, along with experimental ED₅₀ and inhibition percentage data for dual drug compounds [29]. Compounds 1a, 1b, and 1c are all naphthoquinones that share the same number of carbons in the 3-ester linker, comprising different O-R substituents. While in 1a, naphthoquinone position 3 is linked to a pentoic acid, 1b and 1c are occupied by chloroquine and amodiaquine, respectively. An amodiaquine-like substituent is employed in compounds 1(d-f), varying the length of the linking ester between the naphthoquinone motif and the aminoquinoline ring. The influence of an -OH group (Plumbagin-based compounds) in the naphthoquinone skeleton was also evaluated through compounds 2a and 2b, which are direct analogs of 1e and 1f, respectively. Additionally, the direct naphthoquinone-aminoquinoline dual drugs are under the names of 1(a-c), which comprise (3a) chloroquine, (3b) amodiaquine, and (3c) the amodiaquine-like compound. These dual drugs showed high stability toward chemical hydrolysis in aqueous solutions and under physiologic conditions [29].

Table 3

Estimated binding energy estimated by the molecular mechanics analysis for the different inhibitors in each studied site along with calculated inhibition constant and experimental ED₅₀ and inhibition percentage.
Compound	Site 1		Site 2		ED₅₀ FcB1R (µM)	% inhibition of P. falciparum^*
Compound	Binding energy (kcal/mol)	Inhibition constant (Ki/µM)	Binding energy (kcal/mol)	Inhibition constant (Ki/µM)	ED₅₀ FcB1R (µM)	% inhibition of P. falciparum^*
1a	-6.74	11.53	-7.22	5.11	3.5 [0.5]^a	-
1b	-4.43	564.28	-6.34	22.69	0.107	-
1c	-7.41	3.67	-6.72	11.82	-	-
1d	-9.77	0.069	-9.11	0.210	0.144	54
1e	-9.09	0.217	-9.62	0.088	0.047	82
1f	-7.93(-57.42)	1.54	-9.58(-54.82)	0.096	0.023	87
2a	-9.00	0.253	-8.65	0.455	0.0287	-
2b	-7.74	2.12	-8.32	0.799	0.056	-
3a	-7.94	1.52	-7.56	2.88	-	-
3b	-8.79	0.361	-8.55	0.542	-	-
3c	-8.67(-32.74)	0.437	-9.28(-45.62)	0.157	-	-
MD	-5.53(-18.68)	88.6	-5.65(-15.40)	72.74	-	-

( ) indicates average values of the binding free energies obtained from the 100 ns of production in the dynamics simulation, via MM-GBSA profile.

^aExperimental IC₅₀ value for PfGR inhibition [29]

*Inhibition assays at 25 µM of compound [29]

The best molecular docking binding energy achieved for site 1 comprises compound 1d (n = 2 / -9.77 kcal mol^− 1), with the shortest alkane chain among the ester-linkers. In the same group, the binding energy reduced with the increase of carbons in the ester bridge: compound 1e (n = 4 / -9.09 kcal mol^− 1) and 1f (n = 5 / -7.93 kcal mol^− 1).

On the other hand, 1e and 1f showed a slightly better score than 1d, whose binding energy did not vary significantly (⁓ 9.60 kcal mol^− 1) for site 2. A similar tendency is observed experimentally, where both P. falciparum inhibition and ED₅₀ increase with the ester length; nonetheless, such an increase of n = 4 for n = 5 produces slight biological effects (Table 3). This might me explained with the differences in the permeability in the cellular membrane as hydrophobicity may increase with the length of the chain allowing more of the compound to reach the active target thus presenting higher toxicity to the parasite cell.

Furthermore, among the compounds with five carbons in the alkane chain, 1b [the double drug containing chloroquine] presented an ED₅₀ of 0.107 µM. However, the docking interaction energy was only − 4.43 kcal mol^− 1 in the protein's active site. In this instance, the discordance suggests that the conjugation strategy involving an NQ alkanoic acid might instigate a synergistic effect, leading to the accumulation of the ester in the parasite's food vacuole. This accumulation amplifies the interaction with heme groups, thus impeding the formation of inert hemozoin crystals, rather than obstructing the biological function of PfGR. 1b presented − 6.34 kcal mol^− 1 interaction energy in the allosteric site, which is in the mean of the other compounds in the group.

In the plumbagin compounds (2a and 2b), both performed well in the molecular docking tests interacting with site 2 with energies of -8.65 and − 8.32 kcal mol^− 1, respectively. 2a was the most favored in relation to the active site (-9.00 kcal mol^− 1) what might infer the lower ED₅₀ concentration in relation to 2b. Again, there is a direct connection between the reported values of ED₅₀ and the calculated K_i for the compounds.

Compounds 1f and 3c differ only by the linking chain that connects the naphthoquinone group with the quinolinic condensed rings. While the first is connected through an ester functional, the second is bound via a simple alkane chain. 1f demonstrated activity in the 0.0023 µM of ED₅₀, which is concise to the 0.096 µM inhibition constant based on the − 9.58 kcal mol^− 1 interaction with the allosteric site 2. Thus, for compound 1f the simulations agreed with experimental in vitro and in vivo tests [29] that demonstrated 1f as the most promising inhibitor among the ones studied.

Besides, 3d displayed an interaction energy with site 2 of -9.28 kcal mol^− 1, and is indicated, thus, as a possible candidate to biological tests to investigate its antiparasitic properties further. Compared to the currently employed inhibitor MD, all studied compounds (but 1b in site 1) displayed lower binding energy, indicating they might be more effective PfGR inhibitors. This highlights the potential of combining drugs to design better analogs that positively impact some treatments' efficiency.

Molecular dynamics simulations of the dual drugs 1f and 3c were performed for the best pose in docking, in order to explore the binding modes at thermodynamic conditions and validate the docking results.

3.2.2 Complementary molecular dynamics simulations

The molecular dynamics simulations allowed a more detailed analysis of compounds 1f and 3c, mainly about the reference MD. The estimated average binding free energies were also reported in Table 3 in parenthesis, where data point out the dual drugs with a significantly greater binding affinity than inhibitor MD. The thermodynamic analysis in the realm of molecular docking had predicted a better affinity of 3c than 1f for site 1. This binding energy was corrected by the molecular dynamics GBSA profile which predicted an interaction energetics of -57.42 kcal mol^− 1, which is almost twice as favored as 3c (-32.74 kcal mol^− 1) (Table 3). However, regarding site 2, this interaction energy difference lowers to a 9.2 kcal mol^− 1 (1f: -54.82 kcal mol^− 1; 3c: -45.62 kcal mol^− 1), presenting a more modest variation, which is closer to the data obtained from molecular docking analyses.

Along the molecular simulation trajectory, some insightful conformations were observable. At first, 1f interaction on site 1 was predicted to occur mainly via the residue Val383 (as observed for the conventional antimalarials and the inhibitor MB), which acts as an arm that appropriately allocates the ligand in the binding position. That supposition was confirmed by the molecular dynamics, which described the ester oxygen interacting uninterruptedly with the amine group in the protein residue. In the first 3ns of the simulation, the conformation obtained in the molecular docking for this binding mode rotated in the esters' C-O bond, resulting in a new conformation. Figure 9a indicates the interaction poses on both molecular docking and dynamics. Whereas in the docking-obtained structure, the naphthoquinone part of the ligand went towards the NADPH-binding domain, and the amodiaquine part was found near the FADH₂ binding domain, in the molecular dynamics, after 3ns, the rotation of the ester group lead both drug skeleton to a more hydrophobic region.

Differently, ligand 3c keeps the interaction profile observed in the molecular docking simulation on site 1, and the only observable contrast in the molecular dynamics is a movement away from the FADH₂. This movement allows the ligand to move to a more hydrophilic a part of the protein surface, which is stabilized by an H-bonds between the carboxylate in NQ and the residue GLN462. Figure 9b demonstrates the binding mode described above. That difference in binding mode between 1f and 3c may infer the importance of the ester group, which interacts directly with FADH₂ in the case of 1f and keeps the drug positioned optimally in the active site of the PfGR. The lack of a hydrogen acceptor group in the alkane chain of 3c makes it translate to a less effective area, thus reducing the GBSA interaction energy (1f: -57.42 kcal mol^− 1; 3c: -32.74 kcal mol^− 1; Table 3).

On the other hand, 1f interacted with the cavity of the enzyme (site 2) mainly via the interface area. Here, the linking chain was merely a better tool to optimally position both terminal drug structures. The long chain allows both terminal drugs to interact better with the interface region and fit more suitably with the enzyme shape (Fig. 10a,b). The idea for testing the interaction in this cavity is to study the viability of a protein dimerization blockage once this is the area that connects both monomers in the dimer unit. Additionally, as the cavity is more hydrophilic than the active site, the increase in the number of polar groups (such as the ester) may favor the binding energy as is the case of 1f over 3c (Fig. 10b,c). This energetic behavior on site 2 is also shown in Table 3 (1f: -54.85 kcal mol^− 1; 3c:-45.62 kcal mol^− 1). Besides, 3c interaction geometry did not change in relation to the molecular docking simulation. In the case of site 2, the shorter linking chain of 3c did not affected significantly in the interaction mode, and a longer linker (1f) only led to a slightly better proximity with the protein residues.

Some deviation analysis through an RMSD approach indicates that, in general, the protein residues have a reduction in structural freedom when interacting with the drugs compared to a simulation comprising only the protein and its cofactor. 1f interaction in the active site demonstrates an RMSD mean of 3.84 Å, whereas 3c in the same location accounted for a mean of 4.80 Å. This difference infers the binding effectiveness of each compound in pfGR and may indicate the importance of a more flexible and polar linking chain (1f: 5 carbons-ester long; 3c: 2 carbons-alkane long). Similarly, a RMSD analysis in the MD and reference resulted in 4.61Å and 5.95 Å, respectively. Regarding the interaction influence over the FADH₂ RMSD, 1f causes both the major deviations and for longer than 3c (Figure S11). This might be caused by the shorter proximity between 1f and the cofactor, once 3c, as discussed earlier, tends to move away from the FADH₂ domain.

Furthermore, when interacting in the cavity (site 2), a contrary behavior was observed: 1f reduced the protein RMSD by 4.29 Å compared to a reduction of 3.61 Å achieved by 3c. MD had a slightly better performance than the one for the active site and displayed a protein RMSD of 4.45 Å. All RMSD graphs are displayed in Figure S11 in the supplementary information. Similarly, a solvent-accessible surface area analysis (SASA) demonstrated a mean area available to the solvent of 25767.91 Å2 on the protein during the reference simulation. This average was lower for the compound interacting on site 2 (1f: 23859.96 Å2; 3c: 23097.26 Å2) than for site 1 (1f: 24075.94 Å2; 3c: 23954.61 Å2). However, MD behaved differently and the SASA for site 1 (24210.24 Å2) was lower than the one achieved interacting with site 2 (24435.59 Å2). All SASA graphs are displayed on Figure S12 in the supplementary information.

A root mean square of atomic fluctuation (RMSF) was also evaluated, and the analysis of the residues along both sites of interest demonstrates the binding effect of the drugs based on the fluctuation of atoms in the protein. All drugs can decrease RMSF in the protein in comparison to the free protein simulations. The highest decrease was, in general, achieved by both compounds 1f and 3c, which was better than MD in most studied frames. This very same pattern appeared on site 2 RMSF analysis. Although dual-drugs 1f and 3c performed quite similarly in both sites, 1f binding was slightly more effective in the active site than 3c. On the other hand, 3c performed slightly better on site 2, according to the RMSF results. All RMSF graphs are displayed in Figures S13-14 in the supplementary information.

In summary, it was not observed a significative change in the interacting pose, which demonstrates the affinity of those simulated dual-drugs and PfGR. Figure S13 in the supplemental material illustrates that even though there is a fluctuation in both CQ condensed rings and MD motifs, the overall binding conformation is kept, and the drugs do not rotate inside the binging sites.

The molecular docking assessment of conventional antimalarial drugs as PfGR inhibitors revealed that all drugs exhibited higher affinity than reference inhibitors MD and MB, particularly at the enzyme's active site. ATV emerged as a promising candidate, displaying the highest affinity for both the active site and the cavity in the interface region, suggesting a dual docking mode. The docking results also showed a moderate correlation with in vitro antiplasmodial activities, suggesting that the potent antimalarial activity of ATV might involve the inhibition of PfGR. ATV and CQ demonstrated a binding mode similar to inhibitor MB at the enzyme's active site, interacting with the cofactor FADH₂ through H-bonds and π-π interactions. Valine383 plays an essential role assisting the drug binding to the active site, acting as an "anchor" residue and keeping it close to the cofactor FADH₂. Electrochemical analysis suggested that ATV and CQ, despite being less efficient than MB, could act as moderate "subversive substrates" at the active site.

Exploring the enzyme's cavity highlighted distinct binding modes, with ATV and AQ exhibiting notable affinity. Their polar groups of naphtoquinone/quinoline rings play a crucial role as robust H-bond acceptors, while their substituent groups contribute significantly to the ligand's precise adjustment within the site.

Moreover, according to molecular mechanism analysis, dual drugs combining aminoquinoline derivatives and GR inhibitors exhibited significantly greater binding affinity than the reference MD. Results provided insights into their interaction mechanisms, where the presence of the aliphatic ester bond (linker) is essential for effective binding with the enzyme's active site. Analysis of protein structural dynamics indicated reduced freedom of protein residues when interacting with the dual drugs 1f and 3c, highlighting their effectiveness in stabilizing the protein structure. The linker ester with a long chain does not play an important role in the dual-drug binding to the enzyme's cavity; nonetheless, it provides flexibility for both terminal drugs to interact strongly with the interface region and better fit the enzyme shape. Overall, the study confirms the strong affinity of these dual-drugs for PfGR, with stable binding poses maintained throughout the molecular dynamics simulations. In summary, 1f showed to be a good alternative as a dual drug due to the good binding modes reported for both enzyme sites. The study encourages further experiments to investigate the role of PfGR or other disulfide reductases in the mechanisms of action of conventional antimalarial drugs and dual-drugs and their contribution to antiparasitic efficacy.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Author Contribution

This study was conceptualized and written by Dr. GYSD. FHCF, BAO, LVD, and LRP conducted the molecular dockings. FHCF also performed molecular dynamics calculations and contributed to writing. MN supervised the work and revised the manuscript. All authors approved the final version.

Acknowledgement

The authors thank the Minas Gerais State Agency for Research and Development FAPEMIG (BPD-00777-22), the National Council for Scientific and Technological Development CNPq and the Coordination of Superior Level Staff Improvement CAPES (Finance Code 001) for supporting this work. The authors also thank SDumont/LNCC and NEQC(UFJF) for the computational resources for calculations and D. Quintanilha for helping generate some images. FHCF would additionally like to thank CeMEAI/EULER and Repesq/UFJF for the computational resources.

Rich SM, Ayala FJ (2006) Evolutionary Origins of Human Malaria Parasites. In: Malaria: Genetic and Evolutionary Aspects. Springer US, Boston, MA, pp 125–146
Singh B, Daneshvar C (2013) Human Infections and Detection of Plasmodium knowlesi. Clin Microbiol Rev 26:165–184. https://doi.org/10.1128/CMR.00079-12
Ta TH, Hisam S, Lanza M, et al (2014) First case of a naturally acquired human infection with Plasmodium cynomolgi. Malar J 13:68. https://doi.org/10.1186/1475-2875-13-68
Imwong M, Madmanee W, Suwannasin K, et al (2019) Asymptomatic Natural Human Infections with the Simian Malaria Parasites Plasmodium cynomolgi and Plasmodium knowlesi. J Infect Dis 219:695–702. https://doi.org/10.1093/infdis/jiy519
World Health Organization (2023) World malaria report 2023. https://www.who.int/teams/global-malaria-programme/reports/world-malaria-report-2023.
Boletim epidemiológico da malária vol 55: Dia da Malária nas Américas – um panorama da malária no Brasil em 2022 e no primeiro semestre de 2023. https://www.gov.br/saude/pt-br/centrais-deconteudo/publicacoes/boletins/epidemiologic epidemiologicos/edicoes/2024/boletim-epidemiologico-volume-55-no-01/
Epidemiológico B, Especial N Biblioteca Virtual em Saúde do Ministério da Saúde Boletim Epidemiológico.
Cowman AF, Healer J, Marapana D, Marsh K (2016) Malaria: Biology and Disease. Cell 167:610–624. https://doi.org/10.1016/j.cell.2016.07.055
Biot C, Castro W, Botté CY, Navarro M (2012) The therapeutic potential of metal-based antimalarial agents: Implications for the mechanism of action. Dalton Transactions 41:6335. https://doi.org/10.1039/c2dt12247b
de Villiers KA, Egan TJ (2021) Heme Detoxification in the Malaria Parasite: A Target for Antimalarial Drug Development. Acc Chem Res 54:2649–2659. https://doi.org/10.1021/acs.accounts.1c00154
Vasquez M, Zuniga M, Rodriguez A (2021) Oxidative Stress and Pathogenesis in Malaria. Front Cell Infect Microbiol 11:. https://doi.org/10.3389/fcimb.2021.768182
Dubois VL, Platel DFN, Pauly G, Tribouleyduret J (1995) Plasmodium berghei: Implication of Intracellular Glutathione and Its Related Enzyme in Chloroquine Resistance in vivo. Exp Parasitol 81:117–124. https://doi.org/10.1006/expr.1995.1099
Jortzik E, Becker K (2012) Thioredoxin and glutathione systems in Plasmodium falciparum. International Journal of Medical Microbiology 302:187–194. https://doi.org/10.1016/j.ijmm.2012.07.007
Huber PC, Almeida WP, Fátima Â de (2008) Glutationa e enzimas relacionadas: papel biológico e importância em processos patológicos. Quim Nova 31:1170–1179. https://doi.org/10.1590/S0100-40422008000500046
Sarma GN, Savvides SN, Becker K, et al (2003) Glutathione Reductase of the Malarial Parasite Plasmodium falciparum: Crystal Structure and Inhibitor Development. J Mol Biol 328:893–907. https://doi.org/10.1016/S0022-2836(03)00347-4
Buchholz K, Schirmer RH, Eubel JK, et al (2008) Interactions of Methylene Blue with Human Disulfide Reductases and Their Orthologues from Plasmodium falciparum. Antimicrob Agents Chemother 52:183–191. https://doi.org/10.1128/AAC.00773-07
Morin C, Besset T, Moutet J-C, et al (2008) The aza-analogues of 1,4-naphthoquinones are potent substrates and inhibitors of plasmodial thioredoxin and glutathione reductases and of human erythrocyte glutathione reductase. Org Biomol Chem 6:2731. https://doi.org/10.1039/b802649c
Ehrhardt K, Davioud-Charvet E, Ke H, et al (2013) The Antimalarial Activities of Methylene Blue and the 1,4-Naphthoquinone 3-[4-(Trifluoromethyl)Benzyl]-Menadione Are Not Due to Inhibition of the Mitochondrial Electron Transport Chain. Antimicrob Agents Chemother 57:2114–2120. https://doi.org/10.1128/AAC.02248-12
Iribarne F, González M, Cerecetto H, et al (2007) Interaction energies of nitrofurans with trypanothione reductase and glutathione reductase studied by molecular docking. Journal of Molecular Structure: THEOCHEM 818:7–22. https://doi.org/10.1016/j.theochem.2007.04.035
Iribarne F, Paulino M, Aguilera S, Tapia O (2009) Assaying phenothiazine derivatives as trypanothione reductase and glutathione reductase inhibitors by theoretical docking and Molecular Dynamics studies. J Mol Graph Model 28:371–381. https://doi.org/10.1016/j.jmgm.2009.09.003
Tyagi C, Bathke J, Goyal S, et al (2015) Targeting the intersubunit cavity of Plasmodium falciparum glutathione reductase by a novel natural inhibitor: Computational and experimental evidence. Int J Biochem Cell Biol 61:72–80. https://doi.org/10.1016/j.biocel.2015.01.014
Färber PM, Arscott LD, Williams CH, et al (1998) Recombinant Plasmodium falciparum glutathione reductase is inhibited by the antimalarial dye methylene blue. FEBS Lett 422:311–314. https://doi.org/10.1016/S0014-5793(98)00031-3
Ginsburg H, Famin O, Zhang J, Krugliak M (1998) Inhibition of glutathione-dependent degradation of heme by chloroquine and amodiaquine as a possible basis for their antimalarial mode of action. Biochem Pharmacol 56:1305–1313. https://doi.org/10.1016/S0006-2952(98)00184-1
Shibeshi MA, Kifle ZD, Atnafie SA (2020) Antimalarial Drug Resistance and Novel Targets for Antimalarial Drug Discovery. Infect Drug Resist Volume 13:4047–4060. https://doi.org/10.2147/IDR.S279433
Tibon NS, Ng CH, Cheong SL (2020) Current progress in antimalarial pharmacotherapy and multi-target drug discovery. Eur J Med Chem 188:111983. https://doi.org/10.1016/j.ejmech.2019.111983
Sullivan DJ, Gluzman IY, Russell DG, Goldberg DE (1996) On the molecular mechanism of chloroquine’s antimalarial action. Proceedings of the National Academy of Sciences 93:11865–11870. https://doi.org/10.1073/pnas.93.21.11865
Buller R, Peterson ML, Almarsson Ö, Leiserowitz L (2002) Quinoline Binding Site on Malaria Pigment Crystal: A Rational Pathway for Antimalaria Drug Design. Cryst Growth Des 2:553–562. https://doi.org/10.1021/cg025550i
Camarda G, Jirawatcharadech P, Priestley RS, et al (2019) Antimalarial activity of primaquine operates via a two-step biochemical relay. Nat Commun 10:3226. https://doi.org/10.1038/s41467-019-11239-0
Davioud-Charvet E, Delarue S, Biot C, et al (2001) A Prodrug Form of a Plasmodium falciparum Glutathione Reductase Inhibitor Conjugated with a 4-Anilinoquinoline. J Med Chem 44:4268–4276. https://doi.org/10.1021/jm010268g
Neese F (2022) Software update: The ORCA program system—Version 5.0. WIREs Computational Molecular Science 12:. https://doi.org/10.1002/wcms.1606
Barone V, Cossi M (1998) Quantum Calculation of Molecular Energies and Energy Gradients in Solution by a Conductor Solvent Model. J Phys Chem A 102:1995–2001. https://doi.org/10.1021/jp9716997
Larson SB, Day J, Barba de la Rosa AP, et al (2003) First Crystallographic Structure of a Xylanase from Glycoside Hydrolase Family 5: Implications for Catalysis. Biochemistry 42:8411–8422. https://doi.org/10.1021/bi034144c
(2019) Dassault Systèmes BIOVIA. Discovery Studio Visualizer v.20.1.0.19295: San Diego.
Morris GM, Huey R, Lindstrom W, et al (2009) AutoDock4 and AutoDockTools4: Automated docking with selective receptor flexibility. J Comput Chem 30:2785–2791. https://doi.org/10.1002/jcc.21256
Pettersen EF, Goddard TD, Huang CC, et al (2004) UCSF Chimera—A visualization system for exploratory research and analysis. J Comput Chem 25:1605–1612. https://doi.org/10.1002/jcc.20084
Vassetti D, Pagliai M, Procacci P (2019) Assessment of GAFF2 and OPLS-AA General Force Fields in Combination with the Water Models TIP3P, SPCE, and OPC3 for the Solvation Free Energy of Druglike Organic Molecules. J Chem Theory Comput 15:1983–1995. https://doi.org/10.1021/acs.jctc.8b01039
He X, Man VH, Yang W, et al (2020) A fast and high-quality charge model for the next generation general AMBER force field. J Chem Phys 153:. https://doi.org/10.1063/5.0019056
Jorgensen WL, Chandrasekhar J, Madura JD, et al (1983) Comparison of simple potential functions for simulating liquid water. J Chem Phys 79:926–935. https://doi.org/10.1063/1.445869
Berendsen HJC, Postma JPM, van Gunsteren WF, et al (1984) Molecular dynamics with coupling to an external bath. J Chem Phys 81:3684–3690. https://doi.org/10.1063/1.448118
Uberuaga BP, Anghel M, Voter AF (2004) Synchronization of trajectories in canonical molecular-dynamics simulations: Observation, explanation, and exploitation. J Chem Phys 120:6363–6374. https://doi.org/10.1063/1.1667473
van Gunsteren WF, Berendsen HJC (1977) Algorithms for macromolecular dynamics and constraint dynamics. Mol Phys 34:1311–1327. https://doi.org/10.1080/00268977700102571
Miller BR, McGee TD, Swails JM, et al (2012) MMPBSA.py: An Efficient Program for End-State Free Energy Calculations. J Chem Theory Comput 8:3314–3321. https://doi.org/10.1021/ct300418h
Case DA, Cheatham TE, Darden T, et al (2005) The Amber biomolecular simulation programs. J Comput Chem 26:1668–1688. https://doi.org/10.1002/jcc.20290
Salomon-Ferrer R, Case DA, Walker RC (2013) An overview of the Amber biomolecular simulation package. WIREs Computational Molecular Science 3:198–210. https://doi.org/10.1002/wcms.1121
Yayon A, Cabantchik ZI, Ginsburg H (1984) Identification of the acidic compartment of Plasmodium falciparum-infected human erythrocytes as the target of the antimalarial drug chloroquine. EMBO J 3:2695–2700. https://doi.org/10.1002/j.1460-2075.1984.tb02195.x
De Souza Pereira C, Quadros HC, Aboagye SY, et al (2022) A Hybrid of Amodiaquine and Primaquine Linked by Gold(I) Is a Multistage Antimalarial Agent Targeting Heme Detoxification and Thiol Redox Homeostasis. Pharmaceutics 14:1251. https://doi.org/10.3390/pharmaceutics14061251
Friebolin W, Jannack B, Wenzel N, et al (2008) Antimalarial Dual Drugs Based on Potent Inhibitors of Glutathione Reductase from Plasmodium falciparum. J Med Chem 51:1260–1277. https://doi.org/10.1021/jm7009292
P Villareal WJ (2017) Complexos fosfíncos de Platina(II) e Paládio(II): atividade farmacológica e interação com o DNA e com a Ferriprotoporfirina
Daniel L, Karam A, Hebert C, et al (2023) Metal(triphenylphosphine)-Atovaquone complexes: Synthesis, antimalarial activity, and suppression of heme detoxification. [Manuscript submitted for publication]
Müller T, Johann L, Jannack B, et al (2011) Glutathione Reductase-Catalyzed Cascade of Redox Reactions to Bioactivate Potent Antimalarial 1,4-Naphthoquinones – A New Strategy to Combat Malarial Parasites. J Am Chem Soc 133:11557–11571. https://doi.org/10.1021/ja201729z
Basco L, Gillotin C, Gimenez F, et al (1992) In vitro activity of the enantiomers of mefloquine, halofantrine and enpiroline against Plasmodium falciparum. Br J Clin Pharmacol 33:517–520. https://doi.org/10.1111/j.1365-2125.1992.tb04081.x
Belorgey D, Antoine Lanfranchi D, Davioud-Charvet E (2013) 1,4-Naphthoquinones and Other NADPH-Dependent Glutathione Reductase- Catalyzed Redox Cyclers as Antimalarial Agents. Curr Pharm Des 19:2512–2528. https://doi.org/10.2174/1381612811319140003
Becke AD (1993) Density-functional thermochemistry. I. The effect of the exchange‐only gradient correction. J Chem Phys 98:5648–5652. https://doi.org/10.1063/1.462066
W. J. Hehre, R. Ditchfield and J. A. Pople (1972) Self–Consistent Molecular Orbital Methods. XII. Further Extensions of Gaussian–Type Basis Sets for Use in Molecular Orbital Studies of Organic Molecules. J. Chem. Phys. 56, 2257. https://doi.org/10.1063/1.1677527
Lanfranchi DA, Belorgey D, Müller T, et al (2012) Exploring the trifluoromenadione core as a template to design antimalarial redox-active agents interacting with glutathione reductase. Org Biomol Chem 10:4795. https://doi.org/10.1039/c2ob25229e
Birth D, Kao W-C, Hunte C (2014) Structural analysis of atovaquone-inhibited cytochrome bc1 complex reveals the molecular basis of antimalarial drug action. Nat Commun 5:4029. https://doi.org/10.1038/ncomms5029

No competing interests reported.

SIpapermanuscript2024finalversion.docx

Download PDF

Journal Publication

published 23 May, 2024

Read the published version in Journal of Molecular Modeling →

Editorial decision: Revision requested
14 Feb, 2024
Editor assigned by journal
14 Feb, 2024
Submission checks completed at journal
13 Feb, 2024
First submitted to journal
12 Feb, 2024

You are reading this latest preprint version

Analysis of the interaction of antimalarial agents with Plasmodium falciparum Glutathione Reductase through molecular mechanical calculations

Status:

Journal Publication

Version 1

Abstract

Figures

1. Introduction

2. Methodology

2.1 Molecular Docking

2.2 Molecular Dynamics

3. Results and discussion

3.1 Evaluation of conventional antimalarial drugs as PfGR Inhibitors

3.1.1 Interaction studies in the active site (site 1).

3.1.2 Interaction studies in the intersubunit cavity (site 2).

3.2 Antimalarial dual drugs based on inhibitors of PfGR

3.2.1 Analysis via molecular docking simulations

3.2.2 Complementary molecular dynamics simulations

Conclusions

Declarations

Declaration of Competing Interest

Author Contribution

Acknowledgement

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1