3.1. Docking protocol validation
The accuracy and reliability of molecular docking simulations are critical for predicting the binding modes and affinities of ligands within protein targets. In this study, we validated our docking protocol by employing root-mean-square deviation (RMSD), a widely accepted metric for assessing the similarity between docked and experimentally determined protein-ligand complexes [38]. An RMSD score of 2.0 Å or less is commonly considered indicative of a reliable docking pose, suggesting a close agreement between the predicted and crystallographically resolved structures. To verify the accuracy of our docking approach, we superimposed the docked complex of V22 with the crystallographic structure of the METTL3-V22 complex (PDB ID: 7O2I, resolution: 3.0 Å), and calculated the RMSD using UCSF Chimera.
The re-docked V22 compound exhibited an impressively low RMSD value of 0.14 Å when compared to the crystallographic conformation. This minimal deviation indicates a high level of agreement between the predicted binding pose generated by our docking procedure and the experimentally resolved structure.
This validation step instills confidence in the reliability of our docking protocol, affirming its suitability for subsequent analyses and downstream investigations. It ensures that the predicted binding orientations of potential METTL3 inhibitors, derived from our docking simulations, accurately capture the likely binding modes within the METTL3 active site. Thus, our validated docking protocol serves as a robust foundation for the identification and characterization of novel METTL3 inhibitors through virtual screening and computational modeling approaches.
3.2. V22 binding stabilizes METTL3 domain with resultant structural perturbations
Proteins exhibit dynamic behavior in response to various stimuli, including the binding of small molecules, which can induce structural changes. In this study, we investigated the influence of V22 on the structural dynamics of METTL3 to gain insights into its binding mechanism and effects on protein stability. To assess the structural dynamics, we analyzed the root mean square deviation (RMSD) and root mean square fluctuation (RMSF) of the METTL3 domain, along with hydrophobicity and solvent accessible surface area (SASA) of the binding site. RMSD provides information about the stability of the system relative to the starting protein structure, while RMSF characterizes residual fluctuations in the amino acid sequence, reflecting protein flexibility [39]. Analysis of 500 ns molecular dynamics (MD) simulation trajectories revealed that the unbound METTL3 (apo) exhibited lower stability compared to the METTL3-V22 complex, as indicated by average RMSD values of 2.74 Å and 2.42 Å, respectively. Although both systems maintained average RMSD values below 3.0 Å, suggesting overall stability, the presence of V22 induced further stabilization of the system (Fig. 2A).
Furthermore, RMSF analysis revealed increased residue fluctuations in the METTL3-V22 complex compared to the unbound protein, with average RMSF values of 14.98 Å and 15.58 Å, respectively (Fig. 1B). This observation suggests that the binding of V22 enhances the flexibility of the protein structure, potentially facilitating its functional dynamics [40], [41]. Additionally, we probed the hydrophobicity of the binding site to elucidate the effects of V22 binding on local structural properties, although detailed results are not provided here.
Overall, our findings indicate that V22 binding stabilizes the METTL3 domain while inducing structural perturbations characterized by increased residue fluctuations. These insights contribute to a deeper understanding of the molecular interactions underlying METTL3 inhibition and provide valuable information for the rational design of novel inhibitors with improved therapeutic efficacy.
Hydrophobicity, a fundamental property of amino acids, plays a pivotal role in protein stability and folding. It reflects the affinity of amino acid side chains for water, with nonpolar side chains exhibiting higher hydrophobicity compared to polar or charged side chains (Fig. 2).
The hydrophobicity effect is crucial for the stabilization and folding of the protein structure [42]. In our study, hydrophobicity analysis of the binding site provided insights into the structural characteristics crucial for stabilizing the protein structure. As depicted in Fig. 2, the hydrophobicity scale ranged from negative (–3.0) to positive values, with residues in the negative range (blue) being less hydrophobic, and those in the positive range (brown) characterized as highly hydrophobic. Residues such as Phe534, Leu409, Ile378, Pro397, and the aliphatic portion of the side chain of Arg536 exhibited high hydrophobicity in the brown region, contributing significantly to the surface functionality. These hydrophobic residues are known to possess strong binding affinity with V22, as corroborated by subsequent investigations.
Furthermore, our analysis revealed a spatial distribution of hydrophobicity within the binding pocket, with residues located deeper in the pocket exhibiting higher hydrophobicity (brown) compared to those in the outer regions, which were more hydrophilic (blue). This observation underscores the importance of hydrophobic interactions in stabilizing the protein-ligand complex. Additionally, we assessed the solvent accessible surface area (SASA) of the binding site, another crucial parameter influencing protein folding and stability [43]. SASA impacts surface area minimization, side chain packing density, and burial of hydrophobic side chains. The degree to which residues interact with the solvent and the core is proportional to the surface area exposed to these environments [44]. Residues exposed to solvent interactions (blue) exhibited higher SASA, while those buried within the protein core (green) displayed lower SASA, indicating limited solvent exposure (Fig. 2). Overall, our hydrophobicity and SASA analyses provide valuable insights into the structural features governing protein-ligand interactions. Understanding these characteristics enhances our comprehension of the molecular mechanisms underlying METTL3 inhibition and aids in the rational design of novel inhibitors with improved therapeutic efficacy.
3.3. V22 Binding Interactions within the binding pocket
The binding of a drug molecule or ligand to a specific receptor site is crucial for eliciting therapeutic effects, but improper binding may lead to adverse side effects, including cytotoxicity [26]. In this study, we investigated the interactions between the reference compound V22 and the binding site residues of METTL3 over a simulation period of 500 ns to elucidate the molecular mechanisms underlying their binding dynamics. Analysis of snapshots generated at distinct frames of the simulation revealed that V22 formed multiple types of bonds with the active site residues of METTL3. However, not all interactions persisted throughout the simulation period; some bonds underwent alterations or disappeared entirely as observed in Fig. 3. The sp2 hybridized OD1 atom of Asp395 formed salt bridges with \({NH}_{2}\) atom of V22, while the sp2 hybridized O atom of Pro396 consistently engaged in conventional hydrogen bonding (H-acceptor) with the \({NH}_{2}\) atom of V22. These interactions remained stable throughout the MD simulation period, as depicted in Fig. 3.
The sp2 hybridized O atom in Pro405 only formed conventional hydrogen (H- acceptor) bond at 300 ns, while the O atom of Ser511 formed a carbon-hydrogen bond (H-acceptor) with the ligand through H and NH2 throughout the simulation time. The sp2 hybridized OE1 atom of Gln550 interacted with the H atom of V22 via a conventional carbon-hydrogen bond (H-acceptor) with a distance of less than 3.5 Å.
Furthermore, the interaction between Thr408 and V22 transitioned from a conventional hydrogen bond to van der Waals forces at 200 ns, reverting back to carbon-hydrogen bonds thereafter. Similarly, Leu409 formed a carbon-hydrogen bond with V22 at 300 ns, while the interaction between Phe534 and V22 persisted through carbon-hydrogen bonds until 400 ns, transitioning back to conventional carbon-hydrogen bonding at 500 ns. Arg436 displayed diverse interactions with V22, including π sigma, π cation, van der Waals, and π alkyl bonds at different time points during the simulation. Additionally, van der Waals forces were observed on non-contact residues.
Overall, our findings highlight the dynamic nature of V22 binding interactions within the METTL3 binding pocket over the simulation period. Understanding these dynamic interactions provides valuable insights into the molecular mechanisms underlying METTL3 inhibition, which may inform the rational design of more effective inhibitors with reduced cytotoxicity and improved therapeutic outcomes.
These observations from the snapshots were supported by computing the hydrogen bond occupancy of the interactions. Since snapshots are a representation of what occurred at a specific frame of the simulation trajectory, this allowed us to ascertain the percentage of the 500 ns simulation occupied by the hydrogen bonds and their average distances. The results are presented in Table 1. As observed, Pro396, and Pro405 are among the residues that formed hydrogen bonds with V22 that endured longer over the simulation period.
Table 1
Hydrogen bond occupancy, distance, and the angle formed between the host residues and V22.
H-Acceptor
|
H-Donor
|
Occupancy (%)
|
Average Distance (Å)
|
Asp395-OD1
|
V22-H2
|
92.
|
2.72
|
Pro396-O
|
V22-HN2
|
84.03
|
2.81
|
Pro405-O
|
V22-H1
|
18.87
|
2.82
|
Asp395-OD1
|
V22-NH2
|
3.90
|
2.78
|
V22-O1
|
Gly407-H
|
27.70
|
2.87
|
V22-O
|
Gly407-H
|
2.45
|
2.89
|
Ser511-O
|
V22-H2
|
2.07
|
2.84
|
Gln1550-OE1
|
V22-H
|
1.79
|
2.84
|
V22-O
|
Asn549-HD21
|
1.7
|
2.86
|
Pro396-O
|
V22-H2
|
0.66
|
2.83
|
Ser511-O
|
V22-HN2
|
0.65
|
2.82
|
Gly407-O
|
V22-H1
|
0.36
|
2.81
|
V22-O
|
Gln550-HE21
|
0.24
|
2.87
|
Gln550-OE1
|
V22-H1
|
0.22
|
2.81
|
3.4. Quantitative insights Per-Residue Binding Contribution Towards V22
Decomposing total free binding energies into individual contributions by residues offers insights into the basic nature of the interactions [45]. Binding free energy prediction provides knowledge on the prominent contributors in the interactions between receptor and the ligand. According to Wang, 2018, Molecular Mechanics Poisson-Boltzmann Surface Area (MMPBSA) method is an effective and sensible estimation for free energy simulations due to various key approximations used [46]. Therefore, it can be applied to the evaluations of binding affinities of small molecule ligands bound to a large receptor as a result, it has been used in several studies [35, 49, 51] Furthermore, it provides data on the stability of the complex.
Per-residue energy decomposition (PRED) analysis was conducted to estimate the energy contributions of the individual binding site amino acids toward the total binding energy of V22. It was estimated based on the total energies from electrostatic, van der Waals, polar solvation, and nonpolar solvation energies, and the residues with high total energy contributions were selected. Those with binding energies lower than − 1.5 kcal/mol were considered.
The energy contributions of the individual residues in the binding site, as computed by the decomposition of the total binding energy of the complex, are shown in Fig. 4. Residues that contributed significantly towards the formation of METTL3-V22 are Phe534 (–3.9 ± 0.6 kcal/mol), Pro397 (–3.9 ± 0.4 kcal/mol), Asp395 (–3.3 ± 1.0 kcal/mol), Ser511 (–2.9 ± 0.6 kcal/mol), Ille 378 (–2.1 ± 0.2 kcal/mol), and Arg536 (–1.7 ± 1.0 kcal/mol). In addition, we also assessed the functional groups of the ligand with which these amino acid residues interact.
3.5. Virtual screening and molecular docking
Virtual screening has shown to be a crucial tool in finding small molecule inhibitors that bind to the target receptor [50]. To search for small molecules which have the potential to inhibit the activity of METTL3 compared to the V22, the pharmacophore model in Fig. 5 below was developed and used as a 3D query compound to virtually screen the ZINC database. The model was generated based on the context of the per residue energy decomposition footprint approach where known features from the high contributing amino acid residues (Fig. 4). According to virtual screening results, 9 hit compounds were obtained, and these compounds were then subjected to molecular docking. Four compounds out of nine compounds that exhibited relatively equal or lower docking scores compared to the crystallographic ligand V22 were selected for further analysis.
3.6. Docking of the hit compounds
The docking scores by PyRx virtual screening software for the hit compounds into the target protein are presented in Table 6. Among all the hit compounds, 4 hit compounds with relatively compatible binding scores were selected for further analysis. ZINC67367742, ZINC76585975, ZINC76600653, and ZINC76603049 demonstrated good dock scores with values of − 10.0, − 9.1, − 9.1, − 9.1 kcal/mol, respectively, compared to the crystallographic V22 ligand (calc. − 9.1 kcal/mol) (Table 2). After we ascertained docking scores, we then characterized the physiological properties of the selected compounds. As well, the 2D structures of the selected compounds and the RMSD values of their poses in comparison to V22 after docking are illustrated in Fig. 6. Interestingly the compared poses have structural similarities, especially the amidic bond in all of them.
Table 2
Computed docking scores by UCSF Chimera and PyRx software
Compound
|
Chimera Docking Score (kcal/mol)
|
PyRx Docking Score (kcal/mol)
|
ZINC67367742
|
–8.7
|
–10.0
|
ZINC76585975
|
–8.3
|
–8.6
|
ZINC76600653
|
–8.3
|
–9.0
|
ZINC76603049
|
–8.3
|
–9.1
|
V22
|
–9.0
|
–9.1
|
3.7. Hit compounds induce structural conformational changes to METTL3 upon binding
Molecular dynamics (MD) simulations offers insights into the time-dependent behaviour of the molecular target system at the atomic level [51]. In our study, MD simulations were employed to investigate the conformational changes in the protein structure upon binding of the selected hit compounds. To assess the stability of the compounds within the binding site over the course of a 500 ns simulation period, several metrics were utilized, including root mean square deviation (RMSD), root mean square fluctuation (RMSF), and radius of gyration (RoG) at Cα atoms.
RMSD analysis was performed to evaluate the stability of each compound in the binding site throughout the simulation period. This metric provides insights into the deviation of the compound's position relative to its initial conformation. Figure 7 illustrates the RMSD profiles of the compounds, allowing us to discern their stability within the binding site over time. It is observed that the METTL3 underwent conformational changes due to the binding of the hit compounds. Free METTL3 (apo) displayed an average RMSD of 2.74 Å. In contrast, complexed systems displayed average RMSD values of 2.42 Å for V22; 2.55 Å for ZINC67367742; 2.66 Å for ZINC76603049; 1.95 Å for ZINC76600653; and 3.22 Å for ZINC76585975. ZINC76600653 system showed the least RMSD value compared to other systems, including free METTL3 protein, suggesting it induced the highest stability of METTL3, while ZINC76585975 induced the lowest stability over the simulation period. To accurately reflect the dynamic changes of the METTL3 structures over time, protein structure superimposition was conducted (Fig. 7).
The average of the coordinate differences was calculated for all the atoms in the structures as the measure of similarity of the structures at the period simulation. Superimposition of V22 in complex with METTL3 with reference to free apo yielded RMSD of 1.034 Å. ZINC76585975-, ZINC76600653-, ZINC76603049-, and ZINC67367742- METTL3 complexes displayed an average RMSD of 1.139, 1.012, 1.25, and 1.005 Å, respectively. This alignment of all atoms resulted in a decreased fit with significantly high RMSD values compared to backbone RMSD, as shown in Fig. 7. Probing further to the dynamic changes, RMSF was then conducted.
In addition to RMSD, RMSF analysis was conducted to examine the flexibility of the protein structure upon compound binding. RMSF measures the fluctuation of individual residues within the protein, providing valuable information about dynamic behavior. The parts of the structure with high RMSF values constantly diverge from the average, indicating high flexibility. ZINC76585975 system showed to have least average RMSF value of 12.82 Å in comparison to ZINC76600653, ZINC76603049, ZINC67367742, and V22 systems with RMSF average values of 15.38, 15.87, 15.17, and 14.98 Å, respectively. The unbound apo protein has an average RMSF value of 14.98 Å, which is relatively lower than other systems except for the ZINC76585975 system, which indicated lower residue fluctuations throughout MD simulation. This implies that the molecular flexibility of METTL3 decreased upon binding of ZINC76585975 in the active site. As observed in Fig. 7, the amino acids at 100–120 position displayed the highest mobilities.
The molecular spatial packing of residues is a crucial aspect of protein stability. As such, we computed the RoG of the Cα atoms to ascertain the impact of compound binding on the compactness of the protein. This metric is indicative of the protein structure compactness. The RoG obtained from the MD simulation period revealed less packing density in the 4 systems compared to the free protein due to the relatively high RoG in the complex systems. ZINC76600653, which had the lowest RMSD, displayed the highest compactness amongst the complexed systems with an average RoG value of 16.90 Å in comparison to 17.18 for V22; 17.26 for ZINC76585975; 17.20 for ZINC76603049, and 16.95 Å for ZINC67367742. While the free apo protein exhibited an average RoG of 16.86 Å, which is relatively lower than the other systems. Thus, the stable, least flexible, and compact system among them all was the ZINC76600653 complex system. Due to the stimuli responsiveness of proteins, these characteristics and induced changes could be the starting point towards achieving better therapeutic inhibition. Furthermore, the impacts of the dynamic changes were investigated at the atomic level as the binding site bind to each compound.
3.8. Thermodynamics profile of hit compounds
The binding energy of the complexes was also investigated to compare the binding energies and stability of the complexes using MMPBSA. The change in the binding energy of a mechanism indicates the effects of enthalpy and entropy to estimate whether or not the mechanism is energetically favorable High binding energies suggest favorable energetic interactions [52]. The binding energies were averaged and presented in Table 3. V22 showed average binding energy (ΔGbinding) of − 53.0 ± 4.1 kcal/mol against ZINC76603049, ZINC76600653, ZINC67367742, and ZINC76585975 which displayed total binding energies (ΔGbinding) of − 56.9 ± 3.3, − 50.1 ± 4.1, − 45.3 ± 3.3, and − 40.1 ± 4.2 kcal/mol, respectively. The calculated binding energies of hit compounds are lower than that of V22 (–53.0 ± 4.1 kcal/mol) except ZINC76603049. The energies presented by the compounds suggest spontaneous complexing of the compounds with METTL3. Van de Waals and electrostatic energies contributed significantly to the total binding energies of the compounds.
Table 3
MMPBSA binding energy profiles of V22, ZINC67367742, ZINC76585975, ZINC76600653, and ZINC76603049 complexed with METTL3.
MMPBSA Binding Energy components (kcal/mol)
|
Systems
|
\({\varvec{\Delta }\varvec{E}}_{\mathbf{v}\mathbf{d}\mathbf{W}}\)
|
\({\varvec{\Delta }\varvec{E}}_{\mathbf{e}\mathbf{l}\mathbf{e}\mathbf{c}}\)
|
\({\varvec{\Delta }\varvec{G}}_{\mathbf{g}\mathbf{a}\mathbf{s}}\)
|
\({\varvec{\Delta }\varvec{G}}_{\mathbf{s}\mathbf{o}\mathbf{l}}\)
|
\({\varvec{\Delta }\varvec{G}}_{\mathbf{b}\mathbf{i}\mathbf{n}\mathbf{d}\mathbf{i}\mathbf{n}\mathbf{g}}\)
|
V22
|
–63.2 ± 3.9
|
–142.1 ± 10.0
|
–205.3 ± 10.1
|
152.3 ± 9.2
|
–53.0 ± 4.1
|
ZINC67367742
|
–57.3 ± 3.0
|
–127.8 ± 9.1
|
–185.0 ± 9.1
|
139.8 ± 9.0
|
–45.3 ± 3.3
|
ZINC76585975
|
–45.5 ± 2.9
|
–121.9 ± 14.2
|
–167.4 ± 14.8
|
127.3 ± 12.6
|
–40.1 ± 4.2
|
ZINC76603049
|
–57.9 ± 2.6
|
–145.2 ± 9.0
|
–203.1 ± 8.9
|
146.2 ± 8.7
|
–56.9 ± 3.3
|
ZINC76600653
|
–50.4 ± 3.1
|
–145.8 ± 12.9
|
–196.2 ± 12.6
|
146.0 ± 11.3
|
–50.1 ± 4.1
|
The stability of a bioactive molecule with an active site is critical for its viable effect [41]. Accordingly, the degree of ligand stability within the binding site of METTL3 was investigated for the simulation period. The average RMSD values of the compounds and V22, which are informative on their stability, are presented in Fig. 8. These results suggest the compounds were generally stable within the binding site. Hit compound molecules in the binding site of METTL3 displayed average RMSD values of 2.50, 1.44, 1.88, and 2.18, for ZINC67367742, ZINC76603049, ZINC76600653, and ZINC76585975 respectively, and 2.24 Å, for V22.
Generally, lower RMSD values regarding actual poses are associated with high binding affinities and represent a good reproduction of the correct pose. Among the inhibitor compounds, ZINC76603049 showed comparatively high stability. There could be potential to use RMSD scores to optimize ligand pose for better orientation and improved binding affinity. From MMPBSA results in Table 3, V22 has a high binding energy while displaying a high RMSD value. Nevertheless, V22 is quite large compared to other hit compounds, giving a great chance of having a high possible number of interactions, possibly more binding modes, and high binding affinity.
3.9. Physiochemical properties and Drug likeness evaluation of hit compounds
An ideal drug should undergo appropriate administration within the body, ensuring effective delivery to various organs and tissues. Moreover, it should be metabolized in a manner that does not immediately compromise its therapeutic activity, while also being eliminated from the body in a timely and safe manner. This comprehensive process encompasses aspects of pharmacokinetics and pharmacodynamics, crucial for ensuring both efficacy and safety of the drug [36]. Lipinski's Rule of Five is a set of criteria used to assess the likelihood of a compound being orally active in humans based on its physicochemical properties. The rule evaluates five key conditions [53]. These criteria encompass molecular weight below 500, estimated octanol-water partition coefficient exceeding 5, fewer than 5 hydrogen bond donors, and fewer than 10 hydrogen bond acceptors [54, 55]. All compounds adhere to Lipinski's rule of five, indicating their potential as biologically active molecules, as demonstrated in Table 4 and Table 7. The partition coefficient (LogP) serves as a descriptor for lipophilicity, representing the relative affinity of a compound for lipid and aqueous phases in its non-ionized state [55]. The hydrophobicity of a ligand contributes to the entropic aspects of molecular interactions, influencing the binding affinity of the ligand-protein complex. Changes in the hydrophobicity of the ligand binding site can lead to alterations in the conformation of the complex [58]. The LogP value serves as an indicator of how a drug should be formulated. As evidenced in Table 4, all four selected hit compounds exhibit LogP values ranging from 1.61 to 2.65, which are lower than the LogP value of V22, which is 3.4. Ideally, drug molecules with LogP values between 1.35 and 1.8 are considered suitable for oral and intestinal absorption. [56].
The size of the drug particles plays a pivotal role in their transport from the gastrointestinal (GI) tract to the active site [57]. Larger particle sizes tend to decrease solubility due to their inverse relationship with surface area [57]. As indicated in Table 4, all compounds possess molecular weights below 500 g/mol, including V22 with a molecular weight of 443.54 g/mol, slightly higher than the selected hit compounds ranging from 326.40 to 386.88 g/mol. This implies their potential for oral distribution.
In conjunction with molecular weight, the count of rotatable bonds significantly impacts both bioavailability and binding efficacy [53]. A rotatable bond is a nonrigid bond connected to a nonterminal, non-hydrogen atom. It is an essential factor in compound permeability since high molecular flexibility causes poor permeability. The number of rotatable bonds of selected hit compounds is nine, making it more flexible than V22 with six rotatable bonds. Also, hydrogen bond donors and hydrogen bond acceptors play a crucial role in the ability of a molecule to permeate the membrane bilayer [58]. A large number of hydrogen bond donor (HBD) groups can reduce a molecule's permeability, while hydrogen bond acceptors (HBA) influence permeability by interconnecting favorably with hydrogen bonding solvents [53]. All selected hit compounds have an HBD of 2, while V22 has an HBD of 3. However, all compounds together with V22 have an HBA of 4. This suggests a good chance of binding potency.
Generally, a well-balanced drug molecule should be adequately hydrophilic to be soluble in aqueous biological fluids and buffer solutions and also adequately lipophilic to penetrate biological membranes [59]. LogS, the logarithm of the molar solubility in water [35] is among the factors affecting absorption. It is the largest concentration of the drug molecule dissolved in the solvent under certain conditions of temperature, pressure and pH [57]. All selected hit compounds including V22 have logS values ranging from − 2.7 to -5.7 l. Compounds with 0 and higher solubility value are highly soluble, those in the range of 0 to − 2 are soluble, those falling within the range of − 2 to − 4 are slightly soluble and insoluble if less than − 4 [60].
The topological surface area (TPSA) molecular descriptor measures the surface area of hetero atoms, nitrogen, oxygen, and their hydrogens [61]. It checks whether the molecule has potential for cell membrane permeability and if a compound possesses > 140 Å2. ZINC67367742, ZINC76585975, and ZINC76600653 have the STMe TPSA value of 74 Å2, which is lower than the one observed for ZINC76603049 (79.26 Å2), which is also lower than that of V22 (91.29 Å2).
Table 4
Physicochemical properties of the hit compounds selected.
Physiochemical Properties
|
ZINC67367742
|
ZINC76585975
|
ZINC76600653
|
ZINC76603049
|
V22
|
Formula
|
C20H23ClN4O2
|
C18H22N4O2
|
C18H21ClN4O2
|
C19H27N5O2
|
C25H29N6O2
|
MW (g/mol)
|
386.883
|
326.4
|
360.845
|
357.458
|
444.53
|
Log Po/w
|
2.65
|
1.61
|
2.20
|
1.74
|
3.48
|
LogS (Ali) (mol/L)
|
–4.20
|
–2.70
|
–3.70
|
–2.99
|
–5.77
|
TPSA (\({\AA }^{2})\)
|
74.33
|
74.33
|
74.33
|
79.26
|
91.29
|
HA
|
27
|
24
|
25
|
26
|
33
|
H-bond acceptors
|
4
|
4
|
4
|
4
|
5
|
H-bond donors
|
2
|
2
|
2
|
2
|
4
|
Rotatable bonds
|
9
|
9
|
9
|
9
|
8
|
Lipinski violations
|
No; 0
|
No; 0
|
No; 0
|
No; 0
|
No; 0
|
3.10. Toxicity of selected hit compounds
In silico toxicity, evaluation is an important method for appropriate lead compound selection before a drug candidate goes through a clinical trial [40]. The toxicity classes are defined by their corresponding \({LD}_{50}\) values in milligrams per kilogram (mg/kg), specifically focusing on the fatal and toxic ranges. These classes range from "Fatal if swallowed" for \({LD}_{50}\) values less than or equal to 55 mg/kg, to "Toxic if swallowed" for \({LD}_{50}\) values between 50 and 300 mg/kg [62]. According to the ProTox II webserver, all selected 4 hit compounds belong to class 4 for acute oral toxicity, where ZINC67367742, ZINC76585975, and ZINC76603049 have the \({LD}_{50}\) values of 1000 mg/kg, while ZINC76600653 has an \({LD}_{50}\) value of 590 mg/kg. These would be dangerous if taken orally. However, the predicted median lethal dose associated with the original ligand V22 shows to be more harmful as it belongs to class 4 with \({LD}_{50}\) value of 250 mg/kg as shown in Table 5.
Table 5
Toxicity of the selected hit compounds.
|
ZINC67367742
|
ZINC76585975
|
ZINC76600653
|
ZINC76600653
|
V22
|
\({\varvec{L}\varvec{D}}_{50}\) (mg/kg)
|
1000 mg/kg
|
1000 mg/kg
|
1000 mg/kg
|
590 mg/kg
|
250 mg/kg
|
Toxicity class
|
Class 4
|
Class 4
|
Class 4
|
Class 4
|
Class 3
|
Table 6
Canonical smiles for other hit compounds
Zinc ID
|
Canonical smile
|
Chimera Docking Score (kcal/mol)
|
PyRx Docking Score (kcal/mol)
|
ZINC67367742
|
CN1CCN(CC1) c1ccc(cc1) [C@@H] (NC(=O) N[C@H]1CCO[C@H]1C) C
|
–8.7
|
–10.0
|
ZINC76585975
|
CN(CCc1ccc(cc1) NC(=O) C(=O) N[C@@H] (c1cn(nc1C) C) C) C
|
–8.3
|
–8.6
|
ZINC76600653
|
CN(CCc1ccc(cc1) NC(=O) C(=O) NCc1ccc(nc1) CI) C
|
–8.3
|
–9.0
|
ZINC68316170
|
CN1CCN(CC1) c1ccc(cc1) [C@@H] (NC(=O) N[C@@H]1CCO[C@H]1C) C
|
–7.8
|
–8.4
|
ZINC76158525
|
CN(CCc1ccc(cc1) NC(=O) C(=O) Nc1c[nH]nc1) C
|
–8.7
|
–8.6
|
ZINC92651254
|
COc1cccc(c1) OCC(=O) Nc1ccc(cc1) CCN(C)C
|
–8.4
|
–8.7
|
ZINC95448669
|
Coc1cc (OC)ccc1NC(=O) c1ccnc(c1) N1CCN(CC1) C
|
–7.7
|
–8.0
|
ZINC95448671
|
Clc1ccc(cn1) CNC(=O) C(=O) Nc1ccc(cc1) CCN1CCCC1
|
–7.4
|
–7.9
|
ZINC76603049
|
CN(CCc1ccc(cc1) NC(=O) C(=O) NCc1cccnc1) C
|
–8.3
|
–9.1
|
Table 7
Physiological properties of the hit compounds retrieved.
ZINC ID
|
Formula
|
Mw (g/mol)
|
LogP
|
LogS
|
TPSA
|
HBA
|
HBD
|
Rotatable bonds
|
GI
|
Lipinski Drug likeness
|
ZINC67367742
|
C19H30N4O2
|
346.47
|
1.71
|
- 2.49
|
56.84 A2
|
3
|
2
|
6
|
High
|
Yes; 0 violation
|
ZINC68316170
|
C19H30N4O2
|
346.47
|
1.72
|
-2.49
|
56.84 A2
|
3
|
2
|
6
|
High
|
Yes; 0 violation
|
ZINC76158525
|
C15H19N5O2
|
301.34
|
0.95
|
-2.44
|
90.12 A2
|
4
|
3
|
8
|
High
|
Yes; 0 violation
|
ZINC76585975
|
C19H27N5O2
|
357.45
|
1.74
|
-2.99
|
79.26 A2
|
4
|
2
|
9
|
High
|
Yes; 0 violation
|
ZINC76600653
|
C18H21CIN4O2
|
360.84
|
2.20
|
-3.70
|
74.33 A2
|
4
|
2
|
9
|
High
|
Yes; 0 violation
|
ZINC76603049
|
C18H22N4O2
|
326.39
|
1.61
|
-2.70
|
74.33 A2
|
4
|
2
|
9
|
High
|
Yes; 0 violation
|
ZINC92651254
|
C19H24N2O3
|
328.41
|
2.79
|
-3.73
|
50.80 A2
|
4
|
1
|
9
|
High
|
Yes; 0 violation
|
ZINC95448669
|
C19H24N4O3
|
356.42
|
1.87
|
-2.93
|
66.93 A2
|
5
|
1
|
6
|
High
|
Yes; 0 violation
|
ZINC95448671
|
C20H23CIN4O2
|
386.88
|
2.65
|
-4.20
|
74.33 A2
|
4
|
2
|
8
|
High
|
Yes; 0 violation
|
3.11. Intermolecular interaction profiling of hit compounds within METTL3 binding pockets.
The impact of compounds on the protein is determined by their interactions, influencing both the binding site and the overall structure of the protein [63]. Analysis of snapshots extracted from molecular dynamics (MD) simulation trajectories of the METTL3-ZINC complexes revealed diverse bonding patterns between the compounds and residues within the binding pocket. These interactions encompassed conventional hydrogen bonds, carbon hydrogen bonds, Van der Waals forces, alkyl interactions, salt bridges, pi-alkyl interactions, pi-cation interactions, pi-sigma interactions, and pi-stacking interactions. While certain interactions remained stable throughout the simulation, others exhibited temporal fluctuations.
Key residues participating in these interactions were identified, including Arg471, Arg536, Thr510, Phe534, Asp395, Tyr406, Pro397, Trp457, Lys459, Val507, Ser511, His512, Lys513, and Trp457, highlighting their crucial role in ligand binding. Visualization of the interaction patterns is provided in Fig. 9 and Fig. 10.
Notably, several interactions observed in the compounds mirror the binding modes observed with the reference compound V22. These similarities suggest a high binding affinity and specificity of the compounds towards METTL3, as they engage multiple residues essential for the catalytic function and substrate recognition of METTL3[64].