STRUCTURE-BASED DESIGN OF NOVEL PYRIMIDINE CARBONITRILES ANALOGS TARGETING THE CYSTEINE PROTEASE FALCIPAIN 2 OF PLASMODIUM FALCIPARUM (pfFP2) AT THE TROPHOZOÏTE STAGE WITH FAVORABLE ADME SPECIFICITIES

Yves Kily Hervé Fagnidi1,2 ,  Eric Ziki3 , Koffi N’Guessan Placide Gabin Allangba2,4,5,6,9 ,  Beguemsi Toï7,8 , Eugene Megnassan2,8,9* 

1Department of Science and Technology, University Alassane Ouattara, Ivory Coast. 2Fundamental and Applied Physics Laboratory, University Nangui Abrogoua, Ivory Coast. 3Laboratory of Material Sciences, the Environment and Solar Energy, University Felix Houphouët-Boigny, Ivory Coast. 4Laboratory of Environmental Science and Technology, University Jean Lorougnon Guédé, Ivory Coast.  5Laboratory of Biophysics and Nuclear Medicine (LBNM), University Félix Houphouët-Boigny, Ivory Coast.6Department of Medical Physics, University of Trieste, Trieste, Italy. 7Ecole Normale Supérieure, Abidjan, Ivory Coast. 8Laboratory of Structural and Theoretical Organic Chemistry, University Felix Houphouët Boigny, Ivory Coast.

9ICTP-UNESCO, QLS, Strada Costiera 11, I 34151 Trieste, Italy.

ABSTRACT 

Aim and Objective: Structure-based drug design (SBDD) of new antimalarials at the moment of resistance of the most causative agent, Plasmodium falciparum to the more valuable artemisinin combination therapy (ACT) is even more urgent. Carbonitriles pyrimidine derivatives (CNP) has emerged as potential inhibitors of the cysteine protease falcipain 2 of Plasmodium falciparum (pfFP2), so here we report virtual pharmacophore based screening of the CNP chemical subspace yielding novel CNP analogs with predicted high inhibitory potency against pfFP2.

Methods: A quantitative structure activity relationships (QSAR) complexation model has been developed from a series of fifteen carbonitriles pyrimidine derivatives to establish a linear correlation between the calculated Gibbs free energies (GFE: ΔΔGcom) of pfFP2-CNP complex formation and the experimental half-maximal enzymatic inhibition concentration ( ).The predictive power of the QSAR model was then validated with the generation of a 3D-QSAR-PH4 pharmacophore (PH4) model as CNP chemical subspace (exemplified as a virtual combinatorial library of more than 83.300 CNP analogs) explorer for novel predicted more potent CNP analogs. Finally the best PH4 hits were evaluated with the initial QSAR model for predicted potency ( ) and pharmacokinetic profile.

Results: The QSAR model linear correlation equation: p = -0.1025 x ∆∆Gcom + 7.2867, R2=0.94, the subsequent PH4 model linear correlation between experiment and PH4-estimated IC50: p = 0.9366 x p + 0.2849, R2=0.91 documents the high predictive power of this approach. Finally the screening of the virtual library of CNP analogs yielded 52 orally bioavailable candidates the best reaching a predicted potency ( ) of 14 pM and displaying favorable pharmacokinetic profile.

Conclusion: The combined use of one descriptor complexation QSAR model and 3D-QSAR Pharmacophore model performs well in identifying novel CNP analogs against pfFP2 and the handful of top predicted analogs are worth undergoing synthesis and biological evaluation.

Keywords: ADMET, CNP, FP2, Malaria, pharmacophore, QSAR, virtual screening.

The same is also true for the case of artemisinin, our latest line of defense and artemisinin-based combination therapies (ACTs) which are also experiencing the emergence of new resistance hindering current efforts to counter malaria4,5,6,7,8. Given that resistance covers a wide range of antimalarial drugs and that it spreads in populations “at risk of parasitic infection" around the world and that no protective vaccine is available9, there is an urgent need for a new therapy against malaria through the identification and development of new classes of drugs with high antimalarial potential10.The development of new inexpensive antimalarial drugs bioavailable by oral route, overcoming drug resistance is an urgent necessity. This is subject to the constraint of finding a new attractive potential target while proceeding to a rational drug design approach and filtering a large diversified library of components to finally obtain an almost perfect pharmacokinetic profile and a multi-target compound11. The digestion of 3/4 of the hemoglobin present in the infected erythrocytes, provides the parasite with amino acids which are necessary for its growth and survival12,13,14.The inhibition of parasitic proteases, in particular the cysteine proteases involved in the process of hemoglobin degradation, proves to be central to avoiding parasitic proliferation3,15.

 

 

INTRODUCTION

 

The malaria disease mainly occurs in tropical regions with a unicellular eukaryotic parasites of the genus plasmodium as causative agent, the most virulent of which is Plasmodium falciparum1. The WHO 2022 malaria reports deplore 247 million clinical cases for 619,000 deaths mostly under 5 years. Almost 80% of malaria-related deaths in the WHO Africa region in 20212.The spread of resistance to the majority of existing drugs, even including chloroquine, which was one of the pillars of antimalarial treatment, constitute an obstacle3.The cysteine protease falcipain-2 (FP2) arouses great interest because of the central role it plays in the life cycle of P. falciparum through the degradation of hemoglobin16. FP2, the most expressed and best studied enzyme among falcipain, is a promising target for the development of new antimalarial drugs17. It is a logical target for antiparasitic chemotherapy and therefore we have been interested in the development of its inhibitors as antiparasitic18. FP2 has a fairly considerable active site comprising four pockets S1, S’1, S2 and S3 which can accommodate, for each of them, substituent’s of substrates P1, P'1, P2 and P3 respectively. Previous studies have shown the selectivity of these pockets; the S1 pocket having a high affinity with compounds bearing a nitrile group at P119 and the S2 pocket having a marked preference for compounds having a hydrophobic group at P220. Various non-peptide heteroarylnitriles compounds have been studied as potential antiparasitics inhibitors of falcipain. This is the chemical class of 5-substituted-2-cyanopyrimidine known as carbonitriles pyrimidine (CNP) which constitutes a powerful and promising head of series18.Their P1 lateral chain containing the pyrimidine nucleus presents various substitution possibilities which is a very useful advantage in the fight against the drug resistance of certain pathogens and constitutes a reliable means in the design of antimalarial drugs with powerful bioavailable and favorable oral pharmacokinetic properties. Having no crystallographic structure of FP2-CNP complexes available in the literature, we therefore used the results of a previous study on azadipeptides nitrile (ADPN) peptide inhibitors of FP221, by in situ modification to explore the active site of FP2 with CNPs. The study with the ADPNs provided valuable structural information that guided the simulations carried out with the CNPs and led to the design based on the structure of new and more powerful antimalarial agents. In this work, we start from a set of CNP molecules to build a QSAR model of FP2 inhibition with a single descriptor (Gibbs free energy, GFE during the formation of the FP2-Inhibitor complex) which has been correlated with the experimental biological activity  . Then a 3D-QSAR pharmacophore protocol was used to build a four-characteristic pharmacophore model (PH4) based on the conformations linked from the CNPs to the active site of FP2. In addition, the calculated enzyme-ligand interaction energy Eint correlates well with the experimental activities   making it possible to achieve its distribution for each residue of the active site. This latest structural information allowed us to select appropriate fragments P1 and P2 as a construction base for a virtual library (VL) of FP2 inhibitor. In order to access a good pharmacokinetic profile while avoiding toxicity problems, VL was targeted in priority on compounds with 0 property descriptors lying outside the range of values determined for 95% of the known drugs on 24 selected descriptors calculated by the QikProp. The aim of this study was to use the predictability of QSAR models obtained from binding the inhibitory enzyme to the pharmaceutical PH4 3D-QSAR model for VL screening. The best Hit Fits from the virtual PH4-based assay for VL were evaluated in silico MM-PB for predicted inhibitory activity up to the picomolar range for the most potent analogues.

 

MATERIALS AND METHODS 

 

Training set and validation set

The series of compounds used in this work is taken from the literature and belongs to the class of 5-substituted-2-cyanopyrimidine known as carbonitriles pyrimidine (CNP)18. Their activities cover a sufficiently wide range of activity to allow the construction of a reliable QSAR model. The training set containing 12 CNP ligands and the validation set 3 CNP ligands taken from the reference18. 

Model building and calculation of binding affinity

No crystallographic structures of the FP2-CNP complexes exist. The FP2:CNP complexes were built by in situ modification of the high-resolution crystallographic structure of the reference complex FP2:E64 (PDB code 3BPF at a resolution of 2.90 Å) using the Insight-II 2005 Molecular Modeling program22. No water molecules from the crystallographic structure were kept in the model. To identify the lowest-energy conformation bound of the modified inhibitor, an exhaustive conformation search of all freshly created residue bonds, coupled with minimization of the inhibitor energy of the protein's active site, was necessary. The structure of the resulting low-energy complex is carefully optimized by minimizing the overall complex. In practice, the in situ modifications generate variations in the torsion angles and the bond angles of the ligand substituents. Then, in order to avoid steric bumps and to take into account the flexibility of the lateral chains of the residues of the active site of the ligand receptor, a local minimization is carried out (within a radius of 5 Angstroms around the current modification), followed by a global minimization of the receptor-ligand complex to obtain stable structure. Calculation of the relative binding affinity of the (ΔΔGcom) ligand has been fully described and reported previously.

The enthalpic contribution relative to the GFE change related to intermolecular interactions in the E:I complex is described by ΔΔHMM and derived from molecular mechanics (MM), ΔΔGsol and ΔΔTSvib represent, respectively, the relative solvation GFE and the simplified relative vibrational entropy.

Molecular mechanics

The modeling of CNP ligands and their FP2-CNP complexes was carried out by molecular mechanics using the CFF force field as widely described previously24.

Conformational search

Conformational research is a method for calculating the relative energy associated with the conformation of a molecule25. Its aim is therefore to find the minimum possible and to calculate Boltzmann's population, which gives us information on the population of occupied levels at a given temperature26. This method has been described earlier25.

Solvation Gibbs free energy

The biological medium is aqueous and proteins and ligands interaction mechanisms involved in the binding process take account of solvation phenomenon. Discovery Studio's DelPhi module calculates the electrostatic component of the solvation GFE, which includes the effect of the ionic strength by solving the non-linear Poisson-Boltzmann (PB) equation27,28,29. This method has been described fully earlier25.

Interaction energy

The non-bonded interactions (van der Waals and electrostatic interatomic potential terms) between two sets of atoms in all E:I complexes were calculated via the MM interaction energy (Eint) calculation protocol available in Discovery Studio 2.534 as described earlier25.

Pharmacophore generation

The Discovery Studio's Catalyst Hypogen algorithm program30 allowed us, based on the models of the various EI complexes used, to generate the hypotheses to construct a 3D-QSAR pharmacophore as described previously25.

ADMET-related properties

The Qikprop program31 based on the Jorgensen method32,33,34 as fully described previously25 was used to calculate the pharmacokinetic profile of CNPs.

Virtual library generation

The virtual library for CNPs analogues was generated according to the protocol described to the reference25.

ADMET based library focusing

The criterion selection drug-likeness used to focus the initial virtual library of CNPs analogs was fully presented earlier26. 

Pharmacophore based library searching

PH4 based library screen process was described earlier25.

In silico screening

The molecular structures corresponding to the best mapping of the pharmacophore model of 3D inhibition-QSAR PH4 Hypo1 have been selected and subjected to a screening of the QSAR model of complexation. The relative free enthalpy ΔΔGcom upon the formation of E:I complex was calculated for each new ligand selected from the focused virtual library, then used for the prediction of the predicted inhibitory potencies ( )  against FP2.

QSAR model

The relative GFE which describes the mutual affinity between the protein and the inhibitor is presented in Table 2 with its various components. The method of calculating its different quantities is explained extensively in Section 2-2. Calculated with an approximate approach, the relevance of the binding model was evaluated by a linear regression analysis, equation (2) which led to a linear correlation with the experimental activity data   18. Table 3 presents the correlation equation obtained for GFE ΔΔGcom, equation (3) with relevant statistical data. The regression coefficient R2=97% and the Fischer F test=363.4 of the correlation with relatively high values show a strong relationship between the binding model and the experimental inhibition power   of the CNP series as indicated in Figure 2.

Binding mode and interaction energy 

Binding mode

 

RESULTS

 

Training and validation set

The series of CNP compounds comprising 12 ligands for the training set and 3 ligands for the validation set is a homogeneous series of pfFP2 inhibitors with known and determined inhibitory activities from the same laboratory18. A variation of two positions R1 and R2 on the pyrimidine carbonitriles backbone was made to obtain the whole series (Table 1). Their experimental biological activities (1 ≤  ≤ 609 nM)18 from the literature extend sufficiently over a wide concentration range for the construction of a reliable QSAR model. The ratio between the sizes of training and validation sets remains a critical point of correct classification but is limited by the count of the set of homologous compounds available from the literature35.

The active conformation of the most active of the carbonitriles pyrimidine CNP9 from this QSAR is revealed in Figure 2. However, the side chain P1 of the carbonitriles pyrimidine family comprising the 2-cyanopyrimidine core offers a great possibility of substitution in the pocket S1. The grafting of the halogens, in particular the Br in position 5 of the nucleus, promotes an improvement in the interactions with the residues Trp43, Asn173 and His174. Also, an intensity of hydrophobic interactions is noticed in the S2 pocket, particularly with the residues Ser149, Leu172, Ala175 and Asp-234. Figure 3 presents in 2D and 3D the mode of binding to the active site of the best CNP9 supported by a hydrogen bond with Gly83.

Interaction Energy 

The distribution of the inhibitory enzymatic intermolecular Eint interaction energy of the key residues of the different pockets of the active site was subsequently calculated and listed in Table 4. Similarly, their correlation was determined as a function of the experimental activity   and represented in Figure 3. A comparative individual contribution between the most active CNP9 and the least active CNP11 compounds confirming the observed trend of experimental activities by pockets is presented in Table 5.

From the analysis of Table 4, as it has been noted for the azadipeptides nitrile (ADPN) in previous study20, here in the case of pyrimidine carbonitriles (CNP) a relevant stabilizing drop of S2 pocket residues contributions to Enzyme – Inhibitor interaction energy (Eint) from the least active TS CNP11 to the best active one CNP9 by almost 4 kcal.mol-1i.e. 50%, close to the 45% increase of their observed inhibitory potency   from 6.21 to 9 respectively (see Table 5 and Figure 3 Top). Therefore the preeminent role of the hydrophobic S2 pocket of PfFP2 active site in the design of potent molecules against PfFP2 is confirmed again as previously reported36 devoting a centralrole of Leu172 in that S2 pocket (Figure 3, and Figure 4).

QSAR Pharmacophore Model

The QSAR 3D Pharmacophore model generation extensively was detailed previousy21.  The model was generated from the active conformation of CNPA compounds in the FP2 active site. These compounds cover a wide range of experimental activity (1 – 609 nM). The results of the 10 best hypotheses are presented in Table 6 showing the cost, the RMSD and the correlation coefficient between predicted and experimental activities.

The CNPs series presents the costs of the 10 respective penalties of PH4 in the range between 68.812 (Hypo1) to 84.894 (Hypo10). The relatively small difference between the costs of the extreme hypotheses clearly reflects the homogeneity and the consistency of the training set used to produce them. The supreme indicator namely the gap ∆=280.644 between the fixed (32.305) and the null costs (312.949) >> 70 documents the quality and predictive character of the PH4 indicating the probability that the correlation between the IC50 values estimated by the PH4 and experimental ones is real at more than 90%34. The other standard statistical indicators such as the root mean square deviation (RMSD) of the various hypotheses, between 2.323 and 2.586 and the coefficient of determination (R2) between 0.94 and 0.91 (see Table 6) make it possible to retain the first hypothesis of PH4 for the screening of the virtual library of CNP analogues. Figure 6 presents the geometric characteristics of Hypothesis 1 (Hypo 1) of the FP2 inhibition pharmacophore.

The best-selected hypothesis Hypo1 represents with a probability of 91% a PH4 model with a similar level of predictive power as the complexation GFE enzyme-inhibitor binding QSAR model. The resulting regression equation expresses   as a function of  , estimated by Hypo1:  =0.9366 x   + 0.2849 are listed in Table 7 (n=12; R2=0.91; R2xv=0.91; F-test=114.34; σ=0.551; α > 95%) and its graph is presented in Figure 5 above. Once again, the predictive power of the PH4 model, like that of the QSAR, is confirmed and both present interesting predictive powers. The information obtained from the QSAR and PH4 inhibition models relating to the hierarchy of mechanisms governing the activity of CNPs will be useful for filling the S1 and S2 pockets.

Virtual screening

In silico screening of a virtual (combinatorial) library can lead to hit identification as it was shown in previous works on inhibitors design37. From the different groups listed in Table 8, we have created a virtual combinatorial library by substitution in positions R1, R2, R3 and R4 on the pyrimidine nucleus and its side chain, the size of which is R1 x R2 x R3 x R4=14 x 17 x 14 x 25=83300 CNPA analogs (CNPs analogs). This virtual combinatorial library has been focused to 32856 (39%) orally administrable compounds with very good predictive pharmacokinetic profile. 

It was then screened with the generated PH4 3D-QSAR model of Hypo1 to yield 127 analogues mapping to the PH4 and all were subjected to a last evaluation of their predicted  with the complexation method (  is calculated from the correlation equation (3) Table 3). Finally, 52 analogues with better scores are selected and listed in Table 8. The histograms of frequency of appearance of groups R1, R2, R3 and R4 for the 52 CNPA(Figure 6) reveal the fragments of greater occurrences (value in parentheses): 1(49), 2(1), 3(1), 4(1) in R1 for filling the S3 pocket ; 15(32), 31(20) in R2 and 38(14), 39(10), 40(12), 41(9) at R3 for filling the pocket S1 and finally preferentially the fragments 46(5), 49(4), 50(6), 51(4), 52(4), 53(5), 54(6), 55(6) and 56(6) in R4 for filling the pocket S2.

Novel CNPs analogs

The predictive activities  of the best 52 new analogs of CNPA calculated from the correlation equation (3) are better and more powerful than that of the most active compound CNP9 of the training set ( ) proposed by Cotereon et al.,20 are presented in Table 9.

The best CNP analogs (CNPA)with their predicted activity value (  ) in brackets are: 1-15-38-48 (23 pM); 1-15-38-54 (34 pM); 1-15-38-56 (37 pM); 1-15-39-54 (34 pM); 1-15-40-50 (40pM); 1-15-40-54 (26 pM); 1-15-40-55 (14 pM); 1-15-40-56 (27 pM); 1-15-41-49 (23 pM); 1-15-41-53 (25 pM); 1-31-41-53 (19 pM). The most active of the CNPAs, namely 1-15-40-55 (14 pM), has a predicted potency ( ) of approximately 70 times greater than that of the best CNP9 inhibitor of the training set.

Pharmacokinetic profile of the best analogs

The pharmacokinetic profile of the best designed CNPs has also been calculated and compared with those of the drugs used for the treatment of malaria alone or in combination with Artemisinin (CTA) or in clinical trials (Table 10): (c) molar mass: 300 £ MW £ 500  g.mol-1; (d) total solvent-accessible molecular surface, (probe radius 1.4 Å): 300 £ Smol £ 1000 Å2; (e) hydrophobic portion of the solvent-accessible molecular surface, (probe radius 1.4 Å): 0 £ Smolhfo  £ 750 Å2);(f) total volume of molecule enclosed by solvent-accessible molecular surface (probe radius 1.4 Å): 500 £ Vmol £ 2000 Å3; (g)number of non-trivial (not CX3), non-hindered (not alkene, amide, small ring) rotatable bonds: 0 £ #rotB £ 15; (h)estimated number of hydrogen bonds that would be donated by the solute to water molecules in an aqueous solution, values are averages taken over a number of configurations, so they can assume non-integer values: 0.0 £ HBdon £ 6.0; (i)estimated number of hydrogen bonds that would be accepted by the solute from water molecules in an aqueous solution, values are averages taken over a number of configurations, so they can assume non-integer values: 2.0 £ HBacc £ 20.0; (j)logarithm of partitioning coefficient between n-octanol and water phases: -2 £ LogPo/w £ 6.5; (k)logarithm of predicted aqueous solubility: logS,S in [mol.dm–3] is the concentration of the solute in a saturated solution that is in equilibrium with the crystalline solid:‑6.0 £ LogSwat £ 0.5; (l)logarithm of predicted binding constant to human serum albumin: ‑1.5 £ LogKhsa £ 1.5;(m)logarithm of predicted brain/blood partition coefficient:‑3.0 £ Log B/B £ 1.2; (n) predicted apparent Caco-2 cell membrane permeability in Boehringer-Ingelheim scale in [nm s-1]: BIPcaco < 25 poor, BIPcaco > 500 nm.s-1 great; (o) number of likely metabolic reactions: 1 £ #metab £ 8; (p)predicted inhibition constants IC50pre. IC50pre in pM was predicted from computed ∆∆Gcom using the regression Equation B shown in Table 3; (q) HOA: human oral absorption: 1=low, 2=medium, 3=high; (r)%HOA: percentage of human oral absorption in gastrointestinal tract: ³ 80%=high.

The value of the drug likelihood descriptor #stars (Table 10, column 2) indicates how through 24 descriptors that comply with those of 95% of drugs the CNPs analogs perform to have a better profile than the majority of ACTs. These results also present the percentage of absorption by oral route (%HOA) in response to WHO recommendation for new antimalarials: %HOA ⩾ 80%.

 

DISCUSSION

 

The study of the binding mode of PfFP2:CNP from its QSAR model of complexation with a single descriptor and the 3D-QSAR PH4 pharmacophore model generated allowed us to access major structural information on the molecular complementarities of the enzyme and the inhibitor. The visual analysis and the calculation of the interactions between PfFP2 and CNPs in the active site of the enzyme guided us in our efforts to design a virtual combinatorial library of new CNP analogues with four substitutions on the CNP scaffold at positions R1, R2, R3 and R4. A resulting targeted library filtered by a set of descriptors linked to ADME and screened by mapping the analogues to the PH4 pharmacophore, allowed the selection of a subset of libraries of CNPs bioavailable orally.

QSAR model

The robustness of this QSAR model with a descriptor is evaluated through the components of the GFE (ΔΔGcom), namely the contribution of the ΔΔHMM enthalpy, the ΔΔGsolv solvation and the loss of vibrational entropy ΔΔTSvib during the binding of the CNPs. The enthalpy contribution to GFE, then taking into account the effect of the solvent in order to get closer to the biological medium maintains the level of strong relationship between the experimental data and the simulation results. Finally, the likelihood of the model is increased by the loss of the inhibitory vibrational entropy TSvib to explain approximately 97% of the variations of the   by that of the GFE.

This last contribution is one of the most reliable indicators of the predictive power of the QSAR model as reported by Freire et al.,38. Consequently, the correlation equation (3) and the calculated quantities ΔΔGcom can be used respectively to predict the  inhibitory potencies with respect to pfFP2 for new CNP analogues, since they share the same binding mode as the compounds of the corresponding validation set. The quality of the fixation model is also confirmed by the ratio between the calculated inhibitory activities   and the values of the experimental activities   and documents the considerable predictive power of the QSAR model of complexation ( / ,  being calculated by equation 3, Table 2 which is close to 1 for the CNPs of the validation set see Table 2). 

Binding mode of inhibitors

In addition to the robustness of the QSAR model, an analysis of the interactions between FP2 and carbonitriles pyrimidine allowed us to reveal key interactions justifying the affinity of the FP2: CNPx complexes. As shown in Figure 2 in 2D and in 3D, the mode of binding to the active site of the best CNP9 is supported by a hydrogen bond with Gly83 and hydrophobic contacts. CNPs unlike ADPNs have a pyrimidine core in P118 offering a multiple substitution positions. The halogenation of the C5 favors strong interactions with the Trp43, Asn173 and His174 residues. The hydrophobic fragments P2 fills in the S2 pocket better in contact with Ser149, Leu172, Ala175 and Asp234 increasing affinity as reported by Löser et al.,20. The, pyrimidine carbonitriles exhibit the same similarity of interaction with the active site of FP2 as ADPNs20. For the design of new analogs, the interaction energy (Eint) between each residue of the active site (Table 4) were calculated in order to guide the search of substituent at R1 to R4 positions as structural characteristics of the binding affinity improvement by filling suitably the S1, S’1, S2 and S3 pockets. In this way, the key residues of these pockets contribute to the overall FP2–CNPx interaction energy. Specifically, the S2 pocket filling and Leu172 interactions, impact FP2 inhibition by more than 70% (Figure 3) as reported previously19,20. In contrast, Figure 4 comparing each active site residue contribution to the Eint for the best active TS CNP9 and the least active TS CNP11 confirms the observed trend but cannot justify the large gap in their  asin our recent work justifying successfully the observed 37.5% jump in the pfA-M17  between methylphosphonic arginine and the hydroxamic acid from the enzyme active site residues contribution to Eint at a level of 35%39.  Therefore, the essential structural information in the design of new powerful analogues of CNP will not only be derived from the Eint but also from a more predictive approach. The analogs will be selected by virtual screening from a diversified virtual library of analogues with the hydrophobic contact S2 as the central structural requirement displayed by the Pharmacophore model of inhibition of FP2 provided by the single descriptor QSAR model (GFE) (Table 3) (Figure 2)39.

Analysis of new inhibitors from in silico screening

The PH4-based screening of the virtual combinatorial library of CNP analogues has resulted in the identification of new compounds with better predicted activities:   =14 pM for 1-15-40-55, 70-fold the best TS CNP9 ( ). The representation of 1-15-40-55in Figure 8 shows the cyclohexane (R1=1) replacing in P3 the methylpropane of TS for better S3 pocket filling. In P1, two substitutions have been made at positions 4 and 5: the bromine (Br) and hydrogen (H) atoms were replaced by methyl (R2 =15) and pentyl (R3=40) respectively to substantially increase the hydrophobic contact at S1. The lipophilic S2 pocket contains a larger 2-But-4-Me(piperidine) (R4=55) in place of the cyclohexane fragment, for better hydrophobic contact beyond Structure Activity Relationship (SAR) results of carbonitriles pyrimidine inhibitors of FP2 and FP320 and similar to the strong potency increase in our previous study on nitrile dipeptides inhibitors of FP311.

ADMET-related properties

The properties related to ADMET described in section 2-8 and the results in Table 10 indicate that the CNP analogues designed possess a good level of drug likelihood since the descriptor #stars (column 2) which lies in the validation interval between 0 and 5, just like the available antimalarial drug. Moreover all the analogues’ human oral absorption in the gastrointestinal tract (HOA) is strong because none of them falls below the range of good oral bioavailability admitted as the WHO main requirements for new antimalarial drugs.

Further improvement to this work

The predictive results of this study are expected to undergo synthesis and biological evaluation. Since this last step is expensive an intermediary step would be a crosscheck of the stability of the FP2 – CNPAx conformation and the related MM-PB interaction patterns and metrics by time-consuming but relevant Molecular Dynamics (MD) runs. We’re operating to fulfill this achievement in shirt future.

 

CONCLUSION

 

The evaluation of key structural information on FP2 inhibition from FP2 – CNPx complexes or from the in situ modification of an existing generic FP2 inhibitor constitutes a reliable means for designing powerful, bioavailable and favorable oral pharmacokinetic antimalarial. From the series of 15 CNPs inhibitors (12 for the training set and 3 for the validation set), we have, through substitutions at the P1, P2 and P3 positions, established a QSAR model explaining the variation experimental activities  by that of complexation GFE(ΔΔGcom) of calculated during the FP2 – CNPx complex formation, shedding light on the main determinant of the activity. A subsequent 3D-QSAR pharmacophore (PH4) helped in screening a virtual combinatorial library 83300 CNPA analogues with the purpose to substantially increase the hydrophobic contact at in the S2 pocket and, at a lower level, S’1 and S3. The new 52 CNP analogues (CNPAs), which were crosscheck evaluated by QSAR complexation exhibit picomolar range predicted potency  and favorable predictive pharmaco-kinetic profile, (Table 10): 1-15-38-48 (23 pM); 1-15-38-54 (34 pM); 1-15-38-56 (37 pM); 1-15-39-54 (34 pM); 1-15-40-50 (40pM); 1-15-40-54 (26 pM); 1-15-40-55 (14 pM); 1-15-40-56 (27 pM); 1-15-41-49 (23 pM); 1-15-41-53 (25 pM); 1-31-41-53 (19 pM). They are recommended for the synthesis and biological evaluation of FP2 inhibitory activity.

 

ACKNOWLEDGEMENTS

 

Authors are thankful for Fundamental and Applied Physics Laboratory, University Nangui Abrogoua, Ivory Coast to provide necessary facilities for this work.

 

AUTHOR’S CONTRIBUTIONS

 

This work was carried out in collaboration among all authors. All authors read and approved the final manuscript.

 

COMPETING INTERESTS

 

The authors declare there is no conflict of interest in relation with the work presented herein.

 

REFERENCES

 

  1. Wongsrichanalai C, Meshnick SR. Declining artesunate–mefloquine efficacy against falciparum malaria on the Cambodia–Thailand border. Emerg Infect Dis 2008;14(5):716–719. https://doi.org/10.3201/eid1405.071601
  2. World malaria report 2022. Geneva: World Health Organization; 2022. Licence: CC BY-NC-SA 3.0 IGO.
  3. Hyde JE. Drug–resistant malaria–an insight. FEBS J. 2007; 274(18):4688–4698.https://doi.org/10.1111/j.1742-4658.2007.05999.x
  1. Sutherland CJ, Lansdell PL, Sanders M, et al. Pfk13-Independent treatment failure in four imported cases of Plasmodium falciparum malaria treated with artemether lumefantrine in the united kingdom. Antimicrob. Agents Chemother 2017, 61, No. e02382.https://doi.org/10.1128/aac.02382-16
  1. Ashley, E. A.; Dhorda, M.; Fairhurst, R. M.; Amaratunga, C.; Lim, P.; Suon, S et al.. Spread of artemisinin resistance in Plasmodium falciparum N. Engl. J. Med. 2014, 371, 411-423. https://doi.org/10.1056/NEJMoa1314981.
  2. Wongsrichanalai C, Meshnick SR. Declining artesunate–mefloquine efficacy against falciparum malaria on the Cambodia–Thailand border. Emerg Infect Dis 2008;14(5):716–719. https://doi.org/10.3201/eid1405.071601
  3. Dondorp AM, Nosten F, Yi P, et al. Artemisinin resistance in Plasmodium falciparum Eng J Med 2009;361(15):455–467. https://doi.org/10.1056/NEJMoa0808859
  4. Garcia Linares GE, Rodriguez JB. Current Status and Progresses Made in Malaria Chemotherapy. Curr Med Chem. 2007;14(3): 289–314.https://doi.org/10.2174/092986707779941096
  1. Juliane Wunderlich, Petra Rohrbach, John Pius Dalton. The malaria digestive vacuole. Front Biosci 2012;4:1424–1448.https://doi.org/10.2741/s344
  1. Rosenthal PJ. Protease Inhibitors. In: Antiparasitic Chemotherapy: Mechanisms of Action, Resistance, and New Directions in Drug Discovery. New York: Springer; 2001:325–345.
  2. Esmel A, Keita M, Megnassan E, et al. Insight into binding mode of nitrile inhibitors of Plasmodium falciparum Falcipain–3, QSAR and Pharmacophore models, virtual design of new analogues with favorable pharmacokinetic profiles. J Comput Chem Molec Model 2017;2–1:1–21. http://dx.doi.org/10.25177/JCMP.2.1.5
  3. Francis SE, Sullivan DJ Jr, Goldberg DE (1997) Hemoglobin metabolism in the malaria parasite Plasmodium falciparum. Annu Rev Microbiol 51: 97-123.https://doi.org/10.1146/annurev.micro.51.1.97
  1. Rosenthal PJ. Hydrolysis of erythrocyte proteins by proteases of malaria parasites. Curr Opin Hematol 2002 Mar;9(2):140-5.https://doi.org/10.1097/00062752-200203000-00010
  1. Goldberg DE, Slater AF, Cerami A, Henderson GB. Hemoglobin degradation in the malaria parasite Plasmodium falciparum: an ordered process in a unique organelle. Proc Natl Acad Sci USA 1990 Apr; 87(8):2931-5.https://doi.org/10.1073/pnas.87.8.2931
  1. Kerr ID, Lee JH, Pandey KC, Harrison A, Sajid M, Rosenthal PJ, Brinen LS. Structures of falcipain-2 and falcipain-3 bound to small molecule inhibitors: Implications for substrate specificity. J Med Chem 2009 Feb 12; 52(3):852-7. https://doi.org/10.1021%2Fjm8013663
  2. Boris DB, Fidele NK, Luc COO, Eugene M. (2016). Targeting cysteine proteases from Plasmodium falciparum: A general overview, rational drug design and computational approaches for drug discovery. Current Drug Targets 2016; 17, 1-26.http://dx.doi.org/10.2174/1389450117666161221122432
  1. Bekono BD, Ntie-Kang F, Owono Owono LC, Megnassan E. Targeting cysteine proteases from Plasmodium falciparum: A general overview, rational drug design and computational approaches for drug discovery. Curr Drug Targets 2018; 19(5):501-526. https://doi.org/10.2174/1389450117666161221122432
  2. Cotereon JM, Catterick D, Castro J, et al. Falcipain Inhibitors: Optimization studies of the 2–Pyrimidine–carbonitrile lead series. J Med Chem 2010; 53(16):6129–6152. http://dx.doi.org/10.1021/jm101228f
  3. Löser R, Schilling K, Dimmig E, Gütschow M. Interaction of papain-like cysteine proteases with dipeptide-derived nitriles. J Med Chem 2005; 48(24):7688–7707.https://doi.org/10.1021/jm050686b
  1. Löser R, Gut J, Philip J, et al. Antimalarial activity of azadipeptides nitriles. Bioorg Med Chem Lett 2010; 20 (1):252–255. https://doi.org/10.1016/j.bmcl.2009.10.122
  2. Fagnidi YKH, Toi B, Megnassan E, et al. In silico design of Plasmodium falciparum cysteine protease falcipain 2 inhibitors with favorable pharmacokinetic profile. J Anal Pharm Res 2018; 7(3):298‒309.https://doi.org/10.15406/japlr.2018.07.00244
  1. Discovery Studio molecular modeling and simulation pro­gram, version 2.5, Accelrys, Inc., San Diego, CA 2009; California
  2. Kouassi AF, Kone M, Keita M, et al. Computer–aided design of orally bioavailable Pyrrolidine carboxamide inhibitors of Enoyl–Acyl carrier protein reductase of Mycobacterium tuberculosis with favorable pharmacokinetic profiles. Int J Mol Sci 2015; 16(12):29744–29771.https://doi.org/10.3390/ijms161226196
  1. Keita M, Kumar A, Dali B, et al. Quantitative structure–activity relationships and design of thymine–like inhibitors of thymidine monophosphate kinase of Mycobacterium tuberculosis with favourable pharmacokinetic profiles. RSC Advances 2014;4(99):55853–55866.https://doi.org/10.1039/C4RA06917J
  1. Dugas H. Basic principles in molecular modeling, theoretical and practical aspects. 4th ed, University of Montreal Bookstore; 1996.
  2. Bartol J, Comba P, Melter M, Zimmer M. Conformational searching of transition metal compounds. J Comput Chem 1999; 20(14):1549-58. https://doi.org/10.1002/(sici)1096-987x(19991115)20:14%3C1549::aid-jcc8%3E3.0.co;2-f
  3. Gilson MK, Honig B. The inclusion of electrostatic hydration energies in molecular mechanics calculations. J Comput Aided Mol Des. 1991; 5(1):5–20.https://doi.org/10.1007/bf00173467
  1. Rocchia W, Sridharan S, Nicholls A, et al. Rapid grid–based construction of the molecular surface and the use of induced surface charge to calculate reaction field energies: Applications to the molecular systems and geometric objects. J Comput Chem 2002; 23(1):128–137.https://doi.org/10.1002/jcc.1161
  1. Discovery Studio Molecular Modeling and Simulation Software. USA: San Diego; 2009.
  2. Böttcher CJF. Theory of electric polarization. Amsterdam, The Netherlands: Elsevier; 1973.
  3. QikProp, 6.5 (Release 139); Schrodinger LLC: New York, NY, USA, 2019.
  4. Duffy EM, Jorgensen WL. Prediction of properties from simulations: Free energies of solvation in hexadecane, octanol, and water. J Am Chem Soc 2000; 122:2878–2888.https://doi.org/10.1021/ja993663t
  1. Jorgensen WL, Duffy EM. Prediction of drug solubility from Monte Carlo simulations. Bioorg Med Chem Lett 2000;10(11):1155–1158.https://doi.org/10.1016/s0960-894x(00)00172-4
  1. Jorgensen WL, Duffy EM. Prediction of drug solubility from structure. Adv Drug Deliv Rev 2002; 54(3):355–366. https://doi.org/10.1016/s0169-409x(02)00008-x
  2. Frecer V, Miertuš S. Polarizable continuum model of solvation for biopolymers. Int J Quantum Chem 1992; 42(5):1449-68. https://doi.org/10.1002/qua.560420520
  3. Allangba KNPG, Keita M, Kre N’Guessan R, Megnassan E, Vladimir F, Miertus S. Virtual design of novel Plasmodium falciparum cysteine protease falcipain-2 hybrid lactone–chalcone and isatin–chalcone inhibitors probing the S2 active site pocket, J Enzyme Inhib Med Chem 2019, 34:1, 547-561.https://doi.org/10.1080/14756366.2018.1564288
  1. Krovat EM, Frühwirth KH, Langer T. Pharmacophore identification, in silico screening, and virtual library design for inhibitors of the human factor Xa. J Chem Inf Model 2005; 45(1):146-59. https://doi.org/10.1021/ci049778k
  1. Freire E. Do enthalpy and entropy distinguish first in class from best in class? Drug Discov Today 2008;13(19-20):869–874. https://doi.org/10.1016/j.drudis.2008.07.005
  2. N’Guessan H, Megnassan E. In silico Design of phosphonic arginine and hydroxamic acid inhibitors of Plasmodium falciparum M17 Leucyl Aminopeptidase with favorable pharmacokinetic profile. J Drug Design Med Chem 2017; 3(6):98–125. http://dx.doi.org/10.11648/j.jddmc.20170306.13