Novel anthraquinone amino-derivatives as anticancer targeting human serine/threonine kinase PAK4

doi:10.21203/rs.3.rs-4158792/v1

Anthraquinone scaffold has attracted increasing attention recently as a source for effective drug synthesis targeting various diseases. Here we demonstrate the potential of nine anthraquinone amino-derivatives (AAD) as anticancer agents. We synthesized nine Schiff bases (1–9) using anthraquinone scaffold and amino acids. We performed molecular docking against the anticancer drug target human serine/threonine p21-activated kinase 4 (PAK4) and compared it with the recently developed PAK4 inhibitor PF-3758309. Compounds 5 and 8 showed comparable docking properties to that of the PF-3758309 at two binding sites located at the c-terminal domain of PAK4. Compound 5, a tryptophan-based anthraquinone derivative, exerts a docking score of -9.3 and − 8.7 kcal/mol at two different identified cavities, cav-1 and cav-4, respectively, exceeding that of the control PF-3758309 for both cavities. Additional optimization on the indole moiety of 5 gave rise to 5b, resulting in high drug likeness (0.74) and comparable binding properties to cav-1 with − 8.6 kcal/mol and the highest achievable binding affinity to cav-4 with an average of -9.7 kcal/mol, indicating a potential allosteric inhibitory effect via cav-4. Our molecular dynamics simulation analysis on the top hits and the control compound provides important insights into the mechanism of action of these drug candidates. Additionally, all compounds exhibited promising pharmacokinetics and toxicity properties as revealed by the ADMET profiles, urging the need for comprehensive in vitro and in vivo antitumor characterization. Our results also indicate that cav-4 may serve as a target for designing specific allosteric inhibitors against PAK4-overexpressed cancer cells, suggesting a potential application in adjuvant cancer treatment.

Biological sciences/Biochemistry

Biological sciences/Cancer

Biological sciences/Drug discovery

Ser/Thr Kinases

PAK4 inhibitors

Anticancer

Anthraquinones derivatives

drug discovery

Human serine/threonine kinases (STKs) play essential roles in signal transduction (also known as cell signaling), and therefore are considered among the most attractive drug targets ^1–3. Signal transduction, catalyzed by protein kinases, involves a series of molecular events including, most commonly, protein phosphorylation that leads eventually to various cellular response, i.e. transformation of a certain stimulus into a biochemical signal that triggers a specific cellular response such as cell division, apoptosis, proliferation and differentiation as well as immune response triggering and control ^4,5. One of those STKs is the p21-activated kinases (PAKs) family. The PAK proteins are downstream effectors of Rho GTPases Rac and Cdc42 G proteins. It consists of six members, which are categorized into two groups; i) Group I, including PAK1, PAK2, and PAK3, and ii) Group II, including PAK4, PAK5 and PAK6 ^6–8.

Dysregulation of PAKs proteins has been observed in many mammalian disorders including immunological and infectious diseases and cancers ^5,7,9. It has been linked with cancer initiation, promotion, progression as well as recurrence ⁹. Human PAK4 kinase was the first to be discovered in PAKs family and has been shown to be localized predominantly in the perinuclear region of various subcellular compartments owing probably to its nuclear import and export signals which control its subcellular localization ^8,10,11. PAK4 was also found in the cell periphery of the breast cancer cells as well as in the cell substratum adhesions in prostate cancer cells and migrating macrophages ¹². PAK4 has a notable high expression throughout development with somewhat lower levels in adult tissues ⁷. It has also been shown to be overexpressed in tumors including breast, prostate, gastric, lung, ovarian, and pancreatic cancers ^13,14. PAK4 plays important roles in the deregulation of proliferation, apoptosis, migration, adhesion and invasion in cancer cells, thereby promoting overall development and spread of cancers ^8,15,16. It has been proposed that PAK4 may function as a hub molecule thereby joining major cell signaling pathways, which may lead eventually in cases of cancer to their progression through invasion, metastasis, survival, proliferation and drug resistance ^8,17. The mechanisms by which PAK4 may act as a hub is simplified in Fig. 1 ⁸.

These essential roles played by PAK4 in cancer progression have made it an attractive drug target for cancer treatments ¹⁴. Many inhibitors against PAK4 have been developed in the past decade with concerted focus on the improvement of inhibitors selectivity, cellular potency and pharmacokinetics properties as achieving effective inhibitors with optimal pharmacokinetics properties has been challenging so far ^18,19. For example, several inhibitors such as PF-3758309, an ATP-competitive inhibitor that is selective for PAK4, were found to suppress pancreatic cancer in vitro and in vivo, however, none of them has passed the clinical trials owing likely to the off-target effects ²⁰. Such complexity of current PAK4 inhibitors warrants urgent development of new strategies to combat cancers via PAK4 pathways ¹⁸. Here, we utilized anthraquinone as a promising scaffold for synthesis of potential anti-cancer inhibitors targeting PAK4. Natural anthraquinones such as emodin, alchemix and many synthetic analogs have been used against different targets including telomerase, topoisomerases, kinases and metalloproteinases as potential anticancer drug targets ^21,22. We synthesized several anthraquinone amino-derivatives (AAD), characterized and investigated their binding affinity against PAK4 aided with molecular dynamic simulation and thorough pharmacokinetics analysis. Effective PAK4 inhibitors may act as ATP competitive or as allosteric inhibitors for the treatment of cancers as first line or as adjuvant cancer treatment. An example to such inhibitors is the novel small molecule HBW-008-A which has been recently reported as a potent, selective PAK4 inhibitor and adjuvant agent for PD-1 therapy in mice ²⁴.

Chemical synthesis of the Schiff base compounds

Nine amino acids (threonine, glutamine, leucine, asparagine, tryptophan, glycine, cysteine, arginine and methionine) were used for the synthesis of the Schiff bases with anthraquinone (a 9,10-anthracenedione with C₁₄H₈O₂ and 208 g/mol molecular weight). The compounds were prepared by dissolving an amount of 1.12 g of KOH in a 20 ml of CH₃OH in the presence of 10 mmol of each of the abovementioned amino acids (solution (I)). The mixture was then stirred vigorously until it reached homogeneity. A solution (II) composed of 2.08 g 9,10-anthraquinone in 50 ml CH₃CH₂OH was added to solution (I). The mixture of solution (I) and (II) was shaken well before being heated at 80⁰C on a wetted wheel until it evaporated to roughly a quarter of the quantity. A 1.0 ml of acetic acid was then added directly to the mixture, resulting in yellow to yellowish green powdered crystals in ~ 2 h.

FT-IR and NMR spectroscopy

All spectroscopic measurements were conducted at the Central Laboratory Unit, Faculty of Science, Alexandria University. The FT-IR spectra of the Schiff base compounds (1–9) were recorded on the Nicolet 380 FT-IR spectrophotometer (Thermo Co., USA) using a KBr pellet technique in a spectral range of 4000 − 500 cm^− 1. For solution ¹H NMR spectral analysis, we used a JEOL JNM-ECA 500 MHz spectrometer (JEOL Ltd, Japan) operated at 400 MHz. All Schiff base compounds were prepared in DMSO for the NMR measurements and tetramethylsilane (TMS) as an internal standard. The IR spectra and the chemical shifts of the ¹H NMR for the Schiff base compounds (1–9) are described below: Compound (1), 1-hydroxy-2-[(10-oxo-9,10-dianthracene-9-ylidene) amino] butanoic acid: IR (KBr, cm^− 1) v: 2974 (OH), 1671 (C = O), 1587 (C = N), 1671 (C = C), 3400 (NH); ¹H NMR (400 MHz, DMSO-d6, δ ppm) 8.3 (s, 1H, CH = N), 7.2–7.8 (m, 5H, Ar-H), 2.1 (s, 3H, CH₃ not next to an oxygen). Compound (2), 5-amino-5-oxo-2-[(10-oxo-9,10-dianthracene-9-ylidene) amino] pentanoic acid: IR (KBr, cm^− 1) v: 3200 (OH), 1700 (C = O), 1625 (C = N), 1673 (C = C), 3400 (NH), 3500 (NH₂); ¹H NMR (400 MHz, DMSO-d6, δ ppm) 8.2 (s, 1H, CH = N), 7.2–7.9 (m, 5H, Ar-H), 2.8 (s, 3H, CH₃ not next to an oxygen), 4.0 (s, 2H, CH₂). Compound (3), 4-methyl-2-[(10-oxo-9,10-dianthracene-9-ylidene) amino] pentanoic acid: IR (KBr, cm-1) v: 3005 (OH), 1700 (C = O), 1573 (C = N), 1661 (C = C), 3401 (NH); ¹H NMR (400 MHz, DMSO-d6, δ ppm) 8.3 (s, 1H, CH = N), 7.2–7.8 (m, 5H, Ar-H), 2.1 (s, 3H, CH₃ not next to an oxygen). Compound (4), 4-amino-4-oxo-2-[(10-oxo-9,10-dianthracene-9-ylidene) amino] butanoic acid: IR (KBr, cm^− 1) v: 3012 (OH), 1700 (C = O), 1577 (C = N), 1661 (C = C), 3401 (NH₂); ¹H NMR (400 MHz, DMSO-d6, δ ppm) 8.2 (s, 1H, CH = N), 7.2–7.9(m, 5H, Ar-H), 3.3 (s, 2H, CH₂ not next to oxygen). Compound (5), 3-(¹H-indol-3-yl) -2-[(10-oxo-9,10-dianthracene-9-ylidene) amino] propanoic acid: IR (KBr, cm^− 1) v: 3011 (OH), 1700 (C = O), 1570 (C = N), 1558 (C = C), 3400 (NH); ¹H NMR (400 MHz, DMSO-d6, δ ppm) 8.2 (s, ¹H, CH = N), 7.0-7.9(m, 5H, Ar-H), 2.4 (s, 2H, CH₂ not next to an oxygen). Compound (6), 2-[(10-oxo-9,10-dianthracene-9-ylidene) amino] ethanoic acid: IR (KBr, cm^− 1) v: 1760 (C = O), 1572 (C = N), 1760 (C = C); ¹H NMR (400 MHz, DMSO-d6, δ ppm) 8.3 (s, 1H, CH = N), 7.2–7.8(m, 5H, Ar-H), 2.1 (s, 2H, CH₂ next to a nitrogen). Compound (7), 3-sulfohydryl-2-[(10-oxo-9,10-dianthracene-9-ylidene) amino] propanoic acid: IR (KBr, cm^− 1) v: 2657 (OH), 1573 (C = N), 1617 (C = C); ¹H-NMR (400 MHz, DMSO-d6, δ ppm) 3.7 (s, 2H, CH₂). Compound (8), 5-guanidino-2-[(10-oxo-9,10-dianthracene-9-ylidene) amino] pentanoic acid: IR (KBr, cm^− 1) v: 3075 (OH), 1670 (C = O / C = C)), 1571 (C = N), 3318 (NH), 3196 (NH₂); ¹H-NMR (400 MHz, DMSO-d6, δ ppm) 8.2 (s, 1H, CH = N), 7.9 (m, 5H, Ar-H), 2.5 (s, 2H, CH₂ not next to oxygen), 3.5 (s, 2H, CH₂ next to a nitrogen). Compound (9) 4-thiomethyl-2-[(10-oxo-9,10-dianthracene-9-ylidene) amino] butanoic acid: IR (KBr, cm^− 1) v: 2917 (OH), 1671 (C = O), 1559 (C = N), 1619 (C = C); ¹H-NMR (400 MHz, DMSO-d6, δ ppm) 8.3 (s, 1H, CH = N), 7.2–7.8(m, 5H, Ar-H), 2.1 (s, 2H, CH₂ not next to a nitrogen).

Molecular Docking experiment and ADMET analysis

A blind docking of the nine synthesized AADs and the PAK4 inhibitor PF-3758309 against the human PAK4 kinase (PDB: 4JDH) was performed using the Cavity-detection guided blind Docking (CB-Dock2) package (https://cadd.labshare.cn/cb-dock2/index.php) ²⁴. The protein model was prepared by removing the water molecules (70 water molecules), AKTpeptide linker and 10 other hetatoms prior to the docking experiment. In addition, the protein chains are broken at the position Lys473-Leu475. We also retained the linker chain of the PAK4 which is treated as chain B in the deposited structure. CB-Dock server performs protein-ligand blind docking by enabling cavity detection on the protein target based on clustering of solvent-accessible surface, molecular docking using AutoDock Vina and fitting based on the patterns of homologous templates. All ligands and their SMILES were generated and edited using the web-based chemical sketch tool in the RCSB PDB server (https://www.rcsb.org/chemical-sketch). CB-Dock2 identified five cavities (binding pocket) for the ligand on PAK4, of which two cavities (cav-1 and cav-3) with at least 249 and 615 Å³, for cav-1 and cav-4, respectively, which also resulted in optimal binding affinities for all compounds were selected for further analysis. CB-Dock2 displays all possible protein-ligand molecular interactions including the Hydrogen bonds, hydrophobic interactions, van der Waals, and ionic interactions. Each docking experiment was repeated at least three times and the results are averages of the values recorded. All results of CB-DOCK2 were saved as PDB and MOL2 and visualized using PyMOL available at https://www.pymol.org/pymol ²⁵. For the absorption, distribution, metabolism, excretion, and toxicity (ADMET) analysis was carried out using the admetSAR (http://lmmd.ecust.edu.cn/admetsar2), the pKCSM servers (https://biosig.lab.uq.edu.au/pkcsm/prediction) and the SwissADMET ^26–28. Furthermore, and to investigate the molecular structure of the compounds and their suitability for further optimization, we applied AdmetOpt2 from AdmetSAR server ²⁶. For the prediction analysis SMILES of the compounds were submitted in these servers and results were downloaded and analyzed based on the limited values of descriptors and compared with that of the control PF-3758309.

Molecular Dynamics simulation

Molecular dynamics (MD) simulation based on GROMACS method was performed in the SiBioLead server (https://sibiolead.com/MDSIM) to investigate the binding stability, ligand-induced structural changes, and interaction modes of the top candidate compounds (5 and 8) ligand-protein complexes at two binding cavities (cav-1 and cav-4) and compared them with both the ligand-free protein and ligand-protein complex of the PAK4 control inhibitor PF-3758309. For ligand-free protein MD simulation, a crystal structure (PDB ID: 4JDH) was downloaded from the protein data bank and preprocessed using the BIOVIA Discovery Studio Visualizer (Dassault Systèmes BIOVIA, 2024) by removing 70 molecules of crystal water molecules as well as the AKTpeptide linker (labeled as chain B in the PDB file). The protein-ligand complexes were preprocessed in the CB-DOCK2 server prior molecular docking experiments as mentioned above and used directly in MD simulation after editing the ligand names manually in the PDB files. MD simulation systems of the protein and protein-ligand complexes were preprocessed by immersing the system in a cubic box containing Simple Point Charge water and the charged residues were neutralized with 0.15 M NaCl and the all-atoms optimized potentials for liquid simulations (OPLS/AA) forcefield was used to generate the molecular geometry of the systems ²⁹. The MD systems were then energy minimized (EM) with Steepest Descent as an EM integrator with 5000 steps and equilibrated using NVT/NPT equilibration protocol with constant temperature, volume and pressure at 300 K and 1.0 bar constant simulation pressure for 100 ps. MD simulation was run for 100 ns with 5000 frames per each MD simulation with a Leapfrog as an integrator. For the simulation analysis the trajectory files of the systems were used to calculate the root mean square deviation and fluctuations (RMSD and RMSF), the radius of gyration, hydrogen bonds, and the molecular mechanics Poisson-Boltzmann surface area (MM PBSA) binding-free energy calculation using the GROMACS utility GMX_MMPBSA. The equations used to calculate free energies (equations 1–7) are listed below:

∆G_MMPBSA or ∆G_bind = ⟨G_complex⟩ + ⟨G_protein⟩ + ⟨G_ligand⟩ (1)

G_x = ⟨E_MM⟩ + ⟨G_solv⟩ - ⟨TS⟩ (2)

∆G_bind = ∆H - T∆S (3)

∆H = ∆E_MM + ∆G_solv (4)

∆E_MM = ∆E_bonded + ∆E_nonbonded = (∆E_bond + ∆E_angle + ∆E_dihedral) + (∆E_GAS) (5)

∆G_solv = ∆G_polar + ∆G_nonpolar = ∆G_PB+ ∆G_nonpolar (6)

∆E_GAS = ∆E_ele + ∆E_vdW (7)

where ∆G_MMPBSA or ∆G_bind in (1) is the total energy of the complex, which can also be calculated using (3) from the enthalpy of binding ∆H and the entropy -T∆S after ligand binding. ∆H accounts for the energies of the molecular mechanisms systems including bonded and non-bonded in addition to the solvation free energies ∆G_solv (equations 4–7).

Synthesis and characterization of AAD compounds

Anthraquinone scaffold is a promising tool for the synthesis of inhibitors against various diseases including cancers, infectious diseases and neurodegenerative diseases as has attracted increasing attention in recent years ^21,30,31. We used 9,10-anthracenedione scaffolds to synthesize nine Schiff base compounds by condensation using nine different amino acids (see Materials and Methods). In principle, Schiff bases can be synthesized from anthraquinones and primary amines as previously described in several structures, in which their formation depends largely on the chemical nature of the amine ³². Figure 2 displays the nine Schiff base compounds (1–9) and the well-known PAK4 inhibitor PF-3758309. The IUPAC nomenclatures and the amino acid substrate used for the synthesis of the nine Schiff bases are listed in Table S1. The Schiff bases were validated by ¹H NMR and FT-IR spectroscopy. The ¹H NMR and FT-IR stretching frequencies of the Schiff bases are displayed in the supporting information (Supporting Figs. 1–18; Table S2). Here, the amino acid derivative Schiff bases were synthesized in the C9 position of the 9,10-anthracenedione by the reaction with the respective amines of the selected amino acids (Fig. 2). The bonding in C9 was previously found to be favored for Schiff base synthesis by 3 kJ/mol than in C10 of the anthraquinones ³². Similar drug with bonding in position C9 is the 10-(4-Acetamido benzylidene)-9-anthrone DK-V-47, which was found to be more effective by inhibiting the proliferation and transformation of HER-2/neu overexpressing human breast cancer cells when compared to the anthraquinone based inhibitor Emdoin, a 6-methyl-1,3,8-trihydroxyanthraquinone, that is known to be effective against human breast cancer ²¹.

Molecular docking analysis of AAD compounds against PAK4

To understand the molecular interactions of the Schiff bases and their effectiveness as potential anticancer agents, we chose the human PAK4 as a target. PAK4 kinase is upstream of the tumor protein p53 and has been implicated in several cancerogenesis processes in various cancers. For docking experiments, we used PAK4 crystal structure that was previously reported in its both ligand-free and a ligand (PF-3758309)-bound forms ²⁰. The structure used is the catalytic domain of PAK4 with a catalytic site flanked by an activation loop with a phosphorylation point at Ser474. The catalytic domains in PAK kinases are arranged into two-domains (NTD and CTD or N-lobe and C-lobe; Fig. 3) with well characterized activation segments. The ATP binding site is residing in the cleft between these two-domains in the vicinity of the catalytic site, implicating it in the designing of novel inhibitors for PAK4 kinase ³³.

We used CB-DOCK2 package to run the molecular docking experiments, which enabled us to identify 5 cavities for potential drug binding in PAK4 with volumes ranging from 140 to 615 Å³. Of these cavities, two are selected for further analysis, cav-1 and cav-4 (Fig. 3). Cav-1 is located in the vicinity of the catalytic site of PAK4 in the cleft between the NTD and CTD occupying a large volume of 615 Å³ and lined with residues that are essential for the molecular interactions and stabilization of ligands (Supporting Table 3; Fig. 3a). The PF-3758309 inhibitor was found to bind tightly in this cavity as shown by the co-crystallization experiment with PAK4 ²⁰. This prompted us to choose PF-3758309 as a control inhibitor for our study. Our docking experiment with PF-3758309 in cav-1 has reproduced the co-crystallization results to larger extend with high docking score − 8.5 kcal/mol (Table 1) with similar residues contributing to the molecular interactions with PF-3758309, indicating the validity of our approach. The second selected cavity was cav-4 (249 Å³) which is distant from the catalytic site and located in the peripheral region of the CTD domain (Fig. 3b). Such distant binding sites may entail an allosteric inhibition mode of action for these potential inhibitors, which is an important aspect to overcome drug resistance ³⁴.

Table 1 summarizes the molecular docking results obtained from the 9 Schiff bases and the control inhibitor in cav-1 and cav-4. The docking score here refers to the binding energy (ΔG) in kcal/mol (Vina score) ²⁴. The control inhibitor PF-3758309 scored − 8.6 and − 7.5 kcal/mol for cav-1 and cav-4, respectively. The molecular interactions contributed to the stability of the ligand-protein complex involve hydrogen bonding, weak hydrogen bonding, and ionic and hydrophobic interactions (Table 1, Figs. 4 and 5). PF-3758309 is a small-molecule pyrrolopyrazole inhibitor that was rationally-designed using a structure-based approach that exhibited a promising preclinical outcome ^20,35, however, it was withdrawn from the clinical trials due to unexpected low oral bioavailability in humans ^35,36. Based on the results obtained from PF-3758309 docking in cav-1 and cav-4 (Figs. 4a and 5a), we ranked the results of our Schiff bases, giving rise to compound 5 and 8 as the most promising candidates with the highest binding potential. Therefore, we subjected 5 and 8 to further investigations including molecular dynamics simulations. In cav-1, the PF-3758309 inhibitor interacts primarily via hydrophobic interactions (alkyl-π) with residues Val335, Ala402, Leu447, and Arg586 and with ionic interactions through Asp405 and Asp458 (Fig. 4, Table 1 & Supporting Table 3). These results corroborate well with those obtained from co-crystallization experiments of PAK4 with PF-3758309 ²⁰. Whereas in cav-3, the inhibitors interact through several interaction forces including hydrogen bonds with residues Tyr515, Arg534, and Leu535, hydrophobic interactions with Asp510, Glu512, Pro514, Tyr515, Lys536, and Leu538, and a cation-π interaction with Arg534.

Figures 4 and 5 show the molecular interaction of compound 5, 8 and PF-3758309, where the main residues involving these interactions in both cavities (cav-1 and cav-4) are highlighted. Compound 5, a tryptophan-based Schiff base, is the top identified compound based on binding potential with − 9.3 and − 8.7 kcal/mol for cav-1 and cav-3, respectively, which is higher than that of the control in both cavities. Interestingly, compound 5 docking score at cav-4 is comparable to that of the control in cav-1 and significantly higher than other compounds including the control in cav-4 (Figs. 4b and 5b). This result may implicate cav-4 in an allosteric inhibition since it is located distant from the catalytic cleft of PAK4. In cav-1, compound 5 interacts with hydrogen bond and weak hydrogen bond with Asp444 and via hydrophobic interactions (alkyl-π) with Ile327, Val335, Ala402, Asp444, and Leu447 (Fig. 4, 5 & Supporting Table 3). It is worth noting that PF-3758309 doesn’t interact with cav-1 through hydrogen bonding. Moreover, unlike PF-3758309, compound 5 is smaller (~ 395 Da) and perfectly sandwiched with its indole moiety in the pocket of cav-1 with more molecular interaction forces, which may explain the higher binding affinity.

By applying AdmetSAR and subsequently AdmetOpt2 for optimization based on the indole moiety by substituting the 1H-indole with 2,3-dihydro-1H-indole (compound 5b) we obtained an improved derivative with high drug-likeness score (0.74) with comparable synthetic accessibility (5.1) based on the AdmetOpt2/AdmetSAR (Supporting Figure S19) ²⁶. The docking score of 5b gave rise to a binding potential of -8.5 and − 9.2 kcal/mol in cav-1 and allosteric cav-4, respectively, which further supports the notion of allosteric inhibition through cav-4 (Table 1). In cav-4 the molecular interactions of compound 5 involve hydrogen and weak hydrogen bonding with Leu535, a cation-π interaction with Arg534 and hydrophobic interactions with Asp510, Tyr515, Glu518 and Lys536 (Fig. 4). These extensive interactions in this distal site exerted a strong binding affinity with − 8.7 kcal/mol and may likely involve an allosteric modulation of the active site. This is an important aspect for designing novel drugs especially in the context of combating multidrug resistance as non-competitive inhibitors ³⁴. Moreover, the cation-π interaction found in 5 and 8 has been shown to play important roles by interacting with positively charged residues in several approved anticancer drugs such as sorafenib and lapatinib ^37,38.

The molecular interactions of compounds 1–9 in both cavities are listed in Table 1 and the residues involving these interactions in Supporting Table 3. All of the molecules exhibit good interactions with the nearby residues in the binding pockets. The next candidate with high binding potential after 5 and 8 is compound 6, which scored − 7.8 and − 7.4 kcal/mol for cav-1 and cav-4, respectively, comparable to that of the control compounds. It is worth noting that for the optimized compound 5b an unfavorable donor-donor interaction is observed between the carboxyl group of the compound and that of the Glu329, which is expected to cause an electrostatic repulsion in cav-1 and similarly for compound 8 in cav-4 (Figs. 4d and 5c). However, both compounds interact well in the other binding cavities. Compound 5b, for instance, interacts well with the allosteric site by forming hydrogen bonds with Arg534 and Asn517 as well as a strong π-sigma covalent bond with Pro514 and π-alkyl with Lys536 (Fig. 5d). The π-sigma interaction is known for anthraquinone due to bending in the anthraquinone skeleton as a result of strong interaction thereby contributing to the protein-ligand stability ^39,40. In the case of compound 8 in cav-1, several π-sigma and π-alkyl interactions are formed with Leu447, Val335, and Ala348 beside conventional hydrogen bonds with Glu399 (Fig. 5c). Thus, it is safe to suggest that compound 8 acts competitively in cav-1, in the catalytic site vicinity, whereas 5b favors an allosteric mode of action via cav-4 (Fig. 5d).

Table 1 Molecular docking scores of compounds 1-9, 5b and PF-3758309 against PAK4 in two different binding pockets (cav-1 and cav-4).

Compd.	Binding affinity (ΔG), kcal/mol against PAK4 Cav-1	Binding affinity (ΔG), kcal/mol against PAK4 Cav-4	Types of molecular interaction in Cav-1 and/or Cav-3^#
1	-7.3	-7.1	H-bond; ionic & hydrophobic
2	-7.4	-7.6	H-bond; weak H-bond; hydrophobic
3	-7.4	-6.9	H-bond; weak H-bond; ionic & hydrophobic
4	-7.3	-7.6	H-bond; weak H-bond; ionic & hydrophobic
5	-9.3	-8.7	H-bond; ionic & hydrophobic
6	-7.8	-7.4	H-bond; weak H-bond; ionic & hydrophobic
7	-6.9	-7.1	H-bond; weak H-bond; ionic & hydrophobic
8	-8.7	-7.6	H-bond; weak H-bond; ionic & hydrophobic
9	-7.0	-7.0	H-bond; weak H-bond; hydrophobic
5b	-8.6	-9.7	H-bond; weak H-bond; ionic & hydrophobic
PF-3758309	-8.6	-7.5	H-bond; weak H-bond; ionic & hydrophobic

^#These interactions include direct and indirect molecular interactions with the respective ligand.

Pharmacokinetics and toxicity properties of AAD compounds

To get insights into the pharmacokinetics and physicochemical properties of the AAD Schiff base compounds, Lipinski's rule of five and toxicity profiles for these compounds were predicted and compared with that of the PAK4 inhibitor PF-3758309 ⁴¹. Two of the main challenges in the drug industry especially in cancer treatment are the toxicity profiles and the bioavailability of the potential drug ^42,43. For instance, the control PAK4 inhibitor used in this study (PF-3758309) and despite its very promising preclinical outcome in suppressing several tumors by targeting PAK4 kinase was unfortunately withdrawn from the clinical trials due to low human oral bioavailability and its toxicity profiles ^35,36. Therefore, we subjected our compounds (1–9) to extensive pharmacokinetics prediction analysis with three different softwares (SwissADME, AdmetSAR, and pKCSM) to better understand the drug likeness of these compounds (see the Materials & Methods). The ADME and the toxicity profiles of all compounds are listed in Tables 2 and 3. A compound is considered bioactive if it obeys the Lipinski’s rule of five based on the following criteria: i) molecular weight (≤ 500 g/mol), ii) cLogP which is the partition coefficient between n-octanol and water (≤ 5.0), iii) number of hydrogen bond donors (HBD) (≤ 5), iv) number of hydrogen bond acceptors (HBA) (≤ 10), and v) molar refractivity which measures the overall polarity of a drug (between 30–140) ⁴¹. Table 2 shows the ADME profiles including those of Lipinski's rule. All compounds (1–9) obey Lipinski's rule of five. All of the Schiff bases were at 100 g/mol smaller than the control PF-3758309 with compound 6 as the smallest molecule with only 265.27 g/ml MW. This is an important feature for these compounds that allows for further optimization to achieve higher druggability with preferred features ^41,44. We focused our analysis in compounds 5 and 8 which exert higher binding potential in both cavities than the control. Both compounds follow Lipinski’s rule of five with zero violation and are better than the control in their pharmacokinetics properties and drug likeness according to the rules developed by Lipinski, Ghose, Veber, Egan and Muegge (Table 2). The synthetic accessibility of all compounds is also high (an average of ~ 3.5) based on SwissADME ²⁷. Table 3 lists the toxicity profile of the compounds which shows that all compounds have no AMES toxicity with better profiles than the control PF-3758309. Importantly, our AdmetSAR analysis predicted higher human oral bioavailability for compound 5 and 8 than the control and are considered in domain based on the topological properties used by AdmetSAR ²⁶. This is an improved feature in comparison with the PF-3758309, which was withdrawn from the clinical trials due to poor oral bioavailability ³⁵.

Table 2

ADME and pharmacokinetics properties of AADs and PF-3758309 compounds
Compd.	MW, g/mol	LogP^a / MR^j	Surface area, Å²	Water solubility, mol/L	Caco2^b permeability, 10^− 6 cm/s	Intestinal absorption, %	Skin permeability, Kp	Pgp^c substrate	Pgp I inhibitor	Pgp II inhibitor	VDss^d (human), L/kg	BBB^e	CNS^f permeability	HBD^g/ HBA^h/ NROTBⁱ
1	309.32	1.90 / 85.2	132.48	-3.29	1.086	100	-2.74	No	No	No	-2.06	-0.15	-2.54	2/5/3
2	336.35	1.79 / 91.7	143.55	-3.76	-0.063	66.54	-2.73	No	No	No	-1.70	-0.66	-2.76	2/5/5
3	321.38	3.57 / 93.6	140.42	-3.83	0.935	99.25	-2.72	Yes	No	No	-1.49	-0.27	-2.08	1/4/4
4	322.32	1.4 / 86.9	137.19	-3.66	-0.108	60.76	-2.73	No	No	No	-1.71	-0.59	-2.76	2/5/4
5	394.43	4.25 / 115.6	172.27	-4.98	0.814	95.95	-2.73	Yes	No	Yes	-0.82	0.008	-1.94	2/4/4
6	265.27	2.15 / 74.4	114.96	-3.25	0.948	97.0	-2.58	No	No	No	-0.44	0.27	-2.30	1/4/2
7	311.36	2.45 / 87.1	132.48	-3.77	0.935	94.77	-2.63	No	No	No	-0.46	-0.38	-2.34	1/4/3
8	364.41	1.78 / 102.7	155.99	-3.49	-0.254	56.05	-2.74	Yes	No	No	0.177	-0.95	-2.95	4/5/7
9	339.42	3.27 / 96.4	144.2	-3.85	0.951	99.61	-2.72	Yes	No	No	-1.45	-0.30	-2.286	1/4/5
5b	396.44	3.59 / 118.7	173.86	-5.22	0.92	96.77	-2.71	No	No	Yes	-0.384	-0.33	-2.07	1/4/4
PF-3758309	490.64	4.53 / 143.0	208.13	-3.08	0.45	76.57	-2.74	Yes	Yes	Yes	0.818	-1.55	-2.468	3/5/8

^aLogP, the blood-brain permeability-surface area product; ^bCaco2 is a cell line composed of human epithelial colorectal adenocarcinoma cells; ^cPgp, the P-glycoprotein; ^dVDss (human), the theoretical steady state volume of distribution that a total dose of drug would need; ^eBBB is the blood-brain barrier; ^fCNS is the central nervous system; ^gHBD, the hydrogen bond donor; ^hHDA, the hydrogen bond acceptor; ⁱNROTB, number of rotatable bonds; ^jMR, molar refractivity.

Table 3

Toxicity profiles of the compounds based ADMET prediction.
Compd.	AMES toxicity	MRTD^a, mg/kg/day	ORA^b toxicity (LD50), mol/kg	ORC^c toxicity (LOAEL), Log mg/kg/day	Hepatotoxicity	Skin sensitization	T. pyriformis toxicity, Log µg/L	Minnow toxicity, Log mM
1	No	0.606	2.21	1.88	Yes	No	0.285	1.02
2	No	0.741	2.25	1.45	No	No	0.286	0.004
3	No	0.708	2.29	1.58	No	No	0.286	-0.228
4	No	0.702	2.26	1.42	No	No	0.286	0.646
5	No	0.43	2.74	0.73	Yes	No	0.287	-0.022
6	No	-0.157	2.40	1.70	Yes	No	0.818	0.577
7	No	-0.273	2.59	1.26	yes	No	0.727	0.629
8	No	0.535	2.22	1.79	Yes	No	0.285	6.091
9	No	0.734	2.33	2.12	No	No	0.287	-0.106
5b	No	0.254	2.80	1.22	Yes	No	0.322	-1.14
PF-3758309	No	0.711	2.71	2.17	Yes	No	0.285	3.62
^aMRTD is the maximum recommended tolerated dose (human); ^bORA toxicity is the Oral Rat Acute toxicity; ^cORC toxicity is the Oral Rat Chronic toxicity

Molecular dynamics simulation analysis

To determine the binding stability, conformational changes and interaction modes of AAD Schiff bases with PAK4, we used GROMACS-based whole molecule MD simulation up to 100 ns simulation time. Our top candidates from the molecular docking and ADMET analysis (compounds 5 and 8), the optimized compound 5b and the positive control compound PF-3758309 were chosen for the MD simulation analysis. Figure 6 shows detailed MD analysis by plotting the RMSD, RMSD distribution and RMSF changes of the protein-ligand complexes over 100 ns. The RMSD analysis of the protein-ligand complex evaluates the stability of the protein upon ligand binding across the given simulation time ^45,46. Our analysis indicates that PAK4 kinase is mostly stable in its interaction with these compounds at least during the simulation time (100 ns) as indicated by the low RMSD (0.2 to 0.3 nm). In few cases the RMSD jumps sharply to above 0.4 nm displaying some sorts of instability in the c𝛼 backbone upon interaction with ligand, however these jumps are remained for few ns only. This is the case for the protein-compound 8 and 5b interactions with cav-1 (Fig. 6). Compound 5 interaction exerts more stability on the protein backbone in the first 80 ns, whereas later in the simulation time the RMSD of the protein increased gradually to reach ~ 0.35 nm at the end of the simulation. We observed that 5 was more stable through the course of simulation with an RMSD of 0.05–0.15 nm, which is also supported by the high binding affinity (-9.3 kcal/mol) (Table 4).

Figures 6 (e-f) show the RSMF and RG fluctuations of the amino acid residues from the MD trajectories recorded during the simulation upon ligand binding. The RMSF describes the fluctuations in the protein backbones upon ligand binding ⁴⁷. We observed four regions of interest (named R1-4) which displayed significant fluctuations, whose changes exceeded 0.35 nm upon interactions with at least one of the compounds (5, 8, or 5b), whereas no such changes were noticeable in the control PF-3758309 (Fig. 6e). The spike RMSF fluctuation of PAK4-PF3758309 complex is predominant as it was observed at the four different regions indicating the significant conformational changes imposed by the inhibitor (PF-3758309). In R1 residue Ser331 in the vicinity of cav-1 displays the highest RMSF fluctuation upon binding with PF-3758309 and compound 5b. Ser331 is located nearby residues Ile327, Glu329, and Glu329 that contribute to the molecular interactions with compounds 5, 5b and 8, which may explain its high RMSF. In R2 residues Glu358, Arg360 for 5b and Met370 for compound 8 were shown to exert higher RMSF, whereas in R3 Glu468 and Arg471 are the most fluctuated upon interaction with 8 and PF-3758309, respectively. Moreover, we also observed a long-range effect on the RMSF upon interaction with compound 8 and 5 in cav-1, which affected cav-4 residues Phe516 and Asn530, respectively. This shows another example of the interplay between cav-1 and the allosteric site cav-4. To investigate the structural compactness of the protein upon ligand binding, we analyzed the RG changes of the MD trajectories (Fig. 6f). The RG variations of all compounds range between 1.5 nm to 2.1 nm. The lowest RG score was observed upon the interaction of PAK4 with PF-3758309 with RG value below 2.0 nm, indicating a structural compactness with less conformational changes upon ligand binding, whereas compounds 8 and 5b exhibited high conformational changes on PAK4 especially after 50 ns of simulation to reach RG values higher than 2.0 nm (Fig. 6f). These results indicate that PAK4 is more rigid during simulation when forming complexes with these compounds without significant differences. These results were further confirmed with the MM-PBSA binding free energy calculations which showed that all compounds were stable in complex with PAK4 with free energies ranging between − 27.20 to -39.22 kcal/mol (Table 4). The analysis indicates that the main interaction forces contributing to the binding are van der Waals and electrostatic interactions as revealed by the molecular docking experiments (Fig. 4).

Table 4

MM-PBSA analysis of the binding free energies upon protein-ligand interactions.
Ligand-protein interaction at cav-1 (kcal/mol)^#
MD system	ΔE_vdW	ΔE_ele	ΔG_solv	ΔE_non−polar	ΔG_GAS	ΔE_PB	ΔG_binding
PF-3758309	-50.34	-0.70	11.82	-4.24	-51.04	16.06	-39.22
Compd. 5	-42.80	-1.81	12.47	-3.80	-44.61	16.27	-32.14
Compd. 8	-36.70	-0.76	10.26	-3.27	-37.46	13.53	-27.20
Compd. 5b	-48.25	-2.58	13.51	-3.34	-50.83	16.85	-37.32
Ligand-protein interaction at the allosteric site cav-4 (kcal/mol)
Compd. 5	-34.38	-4.21	11.29	-2.70	-38.59	13.98	-27.30
Compd. 5b	-33.88	-2.12	9.76	-2.65	-35.99	12.41	-26.23

^#ΔE_vdW, van der Waals free energy; ΔE_ele, electrostatic interactions calculated using Coulomb's law with atomic charges taken from the MM force field; ΔG_solv, solvation free energies which is composed of a nonpolar and polar free energy; ΔE_non−polar, non-polar free energies; ΔG_GAS, the Gibbs free energies derived from the electrostatics and van der Waals interactions free energies; ΔE_PB, Poisson-Boltzmann free energies which considers only the polar component of the solvation free energies.

Ligand interaction analysis at the allosteric site cav-4

To understand the pattern of interactions of the ligand at the allosteric site we run MD simulation analysis on our top compounds 5 and 5b that were selected based on their high binding affinity to the allosteric site (cav-4) -8.7 and − 9.7 kcal/mol, respectively. Figure 7 shows the RMSD, RMSF, and RG values of the MD trajectories during 100 ns simulation. Similar to that of cav-1, MD simulation of PAK4 when interacting with 5 and 5b via cav-4 shows stable complexes throughout the simulation as indicated by the low RMSD values ( ~ > 0.3 nm) with few exceptions; i) in 5b the RMSD value of the protein backbone have increased sharply after 10 ns to reach ~ 0.4 nm before it decreased gradually to ~ 0.3 nm at ~ 35 ns and remained stable to the end of simulation, and ii) compound 5b shows the highest stability of all compounds in both cavities with only ~ 0.13 nm throughout the simulation, suggesting a very strong interaction with the allosteric site cav-4, which agrees well with the high docking score (-9.7 kcal/mol) achieved for 5b (Table 1). We observed, however, that the total free binding energies of 5 and 5b (an average total ΔG_binding of -26.80 kcal/mol) at cav-4 is higher than all of those interacting with cav-1 (Table 4). This may be attributed to the smaller volume of cav-4 in comparison with cav-1, roughly 2.5× less volume, and/or the differences in the polarity between cav-1 and cav-4 as shown previously in other systems ^48–51. The stability of the protein-ligand interaction was further confirmed by the individual residual flexibility which showed low RMSF fluctuation values along the 100 ns simulation with few exceptions when some residues fluctuate strongly beyond 0.35 nm as shown in R1 and R2 (Fig. 7c). R1 and R2 regions are located near the cav-1 binding site at residues located between 350–375 and 450–475, respectively. No significant fluctuations were observed at the cav-4 site, implying that ligand binding at cav-4 may affect stability at the catalytic site of PAK4 (cav-1). These results indicate that the protein-ligand complexes with the allosteric site are more stable than that of the cav-1 cavity. This is also consistent with the RG values which indicate that the protein is more compact during simulation when it interacts with the ligand at the allosteric site (Fig. 7d). Our MD simulation analysis also indicates that compounds 5 and 5b form more hydrogen bonds with PAK4 at the allosteric site than in the cav-1 site, indicating higher stability at the allosteric site. Overall, the docking and MD simulation results of ligand binding at cav-4 suggest that this site is a potential allosteric site, increasing thereby the possibility to develop specific inhibitors for PAK4 that may overcome the challenges associated with drug resistance ³⁴.

In conclusion, we synthesized nine anthraquinone amino-derivatives (Schiff bases) using nine different amino acids based on the 9,10-anthraquinone scaffold. Our molecular docking experiments with CB-Dock2 found five cavities, of which two were found to possess higher binding energies than that of the known PAK4 inhibitor PF-3758309. Two promising compounds (5 and 8) were identified and characterized in comparison with the known PAK4 inhibitor PF-3758309. In addition, optimizing the indole moiety of compound 5 resulted in a pharmacologically favored compound 5b with high drug likeness and synthesis accessibility. Both 5 and 5b have high effectiveness targeting PAK4 through the newly identified allosteric site with − 8.7 and − 9.7 kcal/mol docking scores. We further validated our docking experiments at the allosteric site (cav-4) with MD simulations with both compounds (5 and 5b) which shows comparable stability to that of cav-1 when compared with the control and compounds 5, 8 and 5b indicating that cav-4 is a potential allosteric site. Therefore, these compounds are promising adjuvant therapies for the treatment of PAK4-overexpressed cancer cells. These findings urge further comprehensive in vitro and in cellulo evaluation on the cytotoxicity and druggability of these top compounds as novel anticancer agents.

Acknowledgements

We thank the staff member of the central laboratory of the Faculty of Science, Alexandria University, Egypt for technical support. We acknowledge the staff members of the Faculty of Science and Faculty of Pharmacy, Omar Al-Mukhtar University for useful discussion during the preparation of this manuscript. A.I., and H.H. acknowledge the financial support of Omar Al-Mukhtar University.

Disclosure statement

The authors declare no potential conflict of interest.

Author contributions

F.K. and A.I., designed and conceptualized research; H.H. and H.B., performed all chemical synthesis and characterization; F.K. and R.E., conducted molecular docking and pharmacokinetics characterization; F.K., performed MD simulations and analyzed the data accordingly; F.K., prepared Figures and the initial draft of the manuscript; F.K., and A.I., reviewed and edited the manuscript with contribution from all authors. All authors have read and approved the submitted version of the manuscript.

Data availability

The datasets used and/or analyzed during the current study available from the corresponding authors on reasonable request.

McConnell, J.L. & Wadzinski, B.E. Targeting protein serine/threonine phosphatases for drug development. Mol. Pharmacol. 75, 1249-1261 (2009).
Kumar, R., Sanawar, R., Li, X. & Li, F. Structure, biochemistry, and biology of PAK kinases. Gene 605, 20–31 (2017).
Bhullar, K.S. et al. Kinase-targeted cancer therapies: progress, challenges and future directions. Mol. Cancer 17, 48 (2018).
Bononi, A. et al. Protein kinases and phosphatases in the control of cell fate. Enzyme Res. 2011, 329098 (2011).
Li, Z. et al. Effects of PAK4/LIMK1/Cofilin-1 signaling pathway on proliferation, invasion, and migration of human osteosarcoma cells. J. Clin. Lab. Anal. 34, e23362 (2020).
Arias-Romero L. E., Chernoff J. A tale of two Paks. Biol. Cell 100, 97-108 (2008).
Minden, A. PAK4-6 in cancer and neuronal development. Cell. Logist. 2, 95-104 (2012).
Won, S.-Y., Park, J.-J., Shin, E.-Y. & Kim, E.-G. PAK4 signaling in health and disease: defining the PAK4-CREB axis. Exp. Mol. Med. 51, 11 (2019).
Liu, H., Liu, K. & Dong, Z. The Role of p21-Activated Kinases in Cancer and Beyond: Where Are We Heading? Front. Cell Devel. Biol. 9,641381 (2021).
Dart, A.E. & Wells, C.M. P21-activated kinase 4--not just one of the PAK. Eur. J. Cell Biol. 92,129-38 (2013).
Sun, X., Su, V.L. & Calderwood, D.A. The subcellular localization of type I p21-activated kinases is controlled by the disordered variable region and polybasic sequences. J. Biol. Chem. 294,14319-14332 (2019).
Wells, C.M., Whale, A.D., Parsons, M., Masters, J.R. & Jones, G.E. PAK4: a pluripotent kinase that regulates prostate cancer cell adhesion. J. Cell Sci. 123(Pt10),1663-1673 (2010).
Blankenstein, L.J., Cordes, N., Kunz-Schughart, L.A. & Vehlow, A. Targeting of p21-Activated Kinase 4 Radiosensitizes Glioblastoma Cells via Impaired DNA Repair. Cells 11, 2133 (2022).
Yuan, Y. et al. PAK4 in cancer development: Emerging player and therapeutic opportunities. Cancer Let. 545, 215813 (2022).
Radu, M., Semenova, G., Kosoff, R. & Chernoff, J. PAK signalling during the development and progression of cancer. Nat. Rev. Cancer 14,13-25 (2014).
Yu, X., Huang, C., Liu, J., Shi, X. & Li, X. The significance of PAK4 in signaling and clinicopathology: A review. Open Life Sci. 17, 586-598 (2022).
Li, Z. et al. Inhibition of neuroblastoma proliferation by PF-3758309, a small-molecule inhibitor that targets p21-activated kinase 4. Oncol. Rep. 38, 2705-2716 (2017).
Li, Y., Lu, Q., Xie, C., Yu, Y. & Zhang A. Recent advances on development of p21-activated kinase 4 inhibitors as anti-tumor agents. Front. Pharmacol. 13, 956220 (2022).
Wang, C. et al. Synthesis and biological evaluation of 7H-pyrrolo [2,3-d] pyrimidine derivatives as potential p21-activated kinase 4 (PAK4) inhibitors. Bioorg. Med. Chem. 60,116700 (2022).
Murray, B.W. et al. Small-molecule p21-activated kinase inhibitor PF-3758309 is a potent inhibitor of oncogenic signaling and tumor growth. Proc. Nat. Acad. Sci. USA 107, 9446–9451 (2010).
Siddamurthi, S., Gutti, G., Jana, S., Kumar, A. & Singh, S.K. Anthraquinone: a promising scaffold for the discovery and development of therapeutic agents in cancer therapy. Future Med. Chem. 12, 1037-1069 (2020).
Malik, M.S. et al. Journey of anthraquinones as anticancer agents - a systematic review of recent literature. RSC Adv. 11,35806-35827 (2021).
Lee, N., Li, Y., Liu, G. & Yang, M. HBW-008-A, a Novel, Potent, Selective, Safe and Bioavailable p21-Activated Kinase 4 (PAK4) Inhibitor, is an Efficacious Adjuvant Agent for PD-1 Therapy in Mice. J. Pharmacol. Exp.Ther. 385 (S3), 92 (2023).
Liu, Y. et al. CD-Dock2: improved protein-ligand blind docking by integrating cavity, detection, docking and homologous template fitting. Nucleic Acids Res. 50, W159-W164 (2022).
Schrödinger, L. & DeLano, W. (2020). PyMOL. Available from: https://www.pymol.org/.
Yang, H. et al. admetSAR 2.0: web-service for prediction and optimization of chemical ADMET properties. Bioinformatics 35, 1067-1069 (2019).
Daina, A., Michielin, O. & Zoete, V. SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules. Sci. Rep. 7, 42717 (2017).
Pires, D.E.V., Blundell, T.L. & Ascher, D.B. pkCSM: predicting small-molecule pharmacokinetics and toxicity properties using graph-based signatures. J. Med. Chem. 58, 4066-4072 (2015).
Jorgensen, W.L., Maxwell, D.S. & Tirado-Rives, J. Development and testing of the all-atom force field on conformational energies and properties of organic liquids. J. Am. Chem. Soc. 118, 11225-11236 (1996).
Volodina, Y.L. et al. (2021). Thiophene-2-carboxamide derivatives of anthraquinone: a new potent antitumor chemotype. Eur. J. Med. Chem. 221, 113521 (2021).
Chua, H.M., Moshawih, S., Goh, H.P., Ming, L.C. & Kifli, N. Insights into the computer-aided drug design and discovery based on anthraquinone scaffold for cancer treatment: A protocol for systematic review. PLoS One 18, e0290948 (2023).
Falke, H., Müller, N. & Oberreiter, M. Concerning the question of covalent bonding in hypericin-chromoproteins: Schiff base formation? Monatshefte für Chemie 125, 313-323 (1994).
Baskaran, Y. et al. An in cellulo-derived structure of PAK4 in complex with its inhibitor Inka1. Nat. Commun. 6, 8681 (2015).
Aboukameel, A. et al. Novel p21-Activated Kinase 4 (PAK4) Allosteric Modulators Overcome Drug Resistance and Stemness in Pancreatic Ductal Adenocarcinoma. Mol. Cancer Ther. 16, 76-87 (2017).
Lin, A. et al. Off-target toxicity is a common mechanism of action of cancer drugs undergoing clinical trials. Sci. Transl. Med. 11, eaaw8412 (2019).
Semenova, G. & Chernoff, J. Targeting PAK1. Biochem. Soc. Trans. 45, 79–88 (2017).
Simard, J.R. et al. Development of a fluorescent-tagged kinase assay system for the detection and characterization of allosteric kinase inhibitors. J. Am. Chem. Soc. 131, 13286-13296 (2009).
Qiu, C. et al. Mechanism of activation and inhibition of the HER4/ErbB4 kinase. Structure 16, 460–467 (2008).
Gleiter, R. Pi-sigma interactions: experimental evidence and its consequences for the chemical reactivity of organic compounds. Pure Appl. Chem. 59, 1585-1594 (1987).
Singh, R.K., Tiwar,i M.K., Kim, I.W., Chen, Z. & Lee, J.K. Probing the role of sigma π interaction and energetics in the catalytic efficiency of endo-1,4-β-xylanase. Appl. Environ. Microbiol. 78, 8817-8821 (2012).
Lipinski, C.A., Lombardo, F., Dominy, B.W. & Feeney, P.J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 46, 3-26 (2001).
Basak, D., Arrighi, S., Darwiche, Y. & Deb, S. Comparison of Anticancer Drug Toxicities: Paradigm Shift in Adverse Effect Profile. Life 12, 48 (2022).
Ioele, G. et al. Anticancer Drugs: Recent Strategies to Improve Stability Profile, Pharmacokinetic and Pharmacodynamic Properties. Molecules 27, 5436 (2022).
Hughes, J.P., Rees, S., Kalindjian, S.B. & Philpott, K.L.Principles of early drug discovery. Br. J. Pharmacol. 162,1239-49 (2011).
Durrant, J.D. & McCammon, J.A. Molecular dynamics simulations and drug discovery. BMC Biol. 9, 71 (2011).
Schreiner, W., Karch, R., Knapp, B. & Ilieva, N. (2012). Relaxation estimation of RMSD in molecular dynamics immunosimulations. Comput. Math. Methods Med. 2012, 173521 (2012).
Ghahremanian, S., Rashidi, M.M., Raeisi, K. & Toghraie D. Molecular dynamics simulation approach for discovering potential inhibitors against SARS-CoV-2: A structural review. J. Mol. Liq. 354,118901 (2022).
Wei, B.Q., Baase, W.A., Weaver, L.H., Matthews, B.W. & Shoichet, B.K. A model binding site for testing scoring functions in molecular docking. J. Mol. Biol. 322,339-355 (2002).
Mobley, D.L. et al. Predicting absolute ligand binding free energies to a simple model site. J. Mol. Biol. 371, 1118-1134 (2007).
Deng, Y. & Roux, B. Computations of standard binding free energies with molecular dynamics simulations. J. Phys. Chem. B 113, 2234-2246 (2009).
Quillin, M.L., Breyer, W.A., Griswold, I.J. & Matthews, B.W. Size versus polarizability in protein-ligand interactions: binding of noble gases within engineered cavities in phage T4 lysozyme. J. Mol. Biol. 302, 955-977 (2000).

No competing interests reported.

SIPAK4Kouaetal.docx

Novel anthraquinone amino-derivatives as anticancer targeting human serine/threonine kinase PAK4

Status:

Version 1

Abstract

Figures

Introduction

Materials & Methods

Chemical synthesis of the Schiff base compounds

FT-IR and NMR spectroscopy

Molecular Docking experiment and ADMET analysis

Molecular Dynamics simulation

Results & Discussion

Synthesis and characterization of AAD compounds

Molecular docking analysis of AAD compounds against PAK4

Pharmacokinetics and toxicity properties of AAD compounds

Molecular dynamics simulation analysis

Ligand interaction analysis at the allosteric site cav-4

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1