Abstract
Cyclotriazadisulfonamide (CADA) is a macrocyclic compound known for its unique mechanism in inhibiting HIV infection by downregulating the CD4 T-cell receptor, a crucial entry point for the virus. Unlike other antiretrovirals, CADA exhibits activity against a wide range of HIV strains, as all HIV variants require CD4 binding for infection. Furthermore, CADA has shown a synergistic effect with clinically approved anti-HIV drugs, offering potential for enhanced therapeutic strategies (Vermeire & Schols, [65]). One proposed mechanism for CADA’s inhibition of the CD4 receptor involves blocking the gates of the Sec61 channel, thereby preventing its translocation. However, CADA suffers from poor solubility and bioavailability. To address this, the study aimed to design CADA analogs with improved binding to the Sec61 channel, enhanced bioavailability, and reduced toxicity. The analogs were designed using SeeSAR, with Avogadro and Meeko used for 3D configuration and pseudoatom placement, respectively. AutoDock Vina version 1.2.4 was employed to predict the binding energies of these analogs. Of the 113 analogs designed, 93 demonstrated a more negative binding energy to the Sec61 channel compared to CADA. Structure-binding energy analyses were done to the top-binding analogs to show favorable structural modifications. Enzyme-ligand interactions were analyzed to elucidate the forces contributing to these binding energies. Additionally, 33 of the 113 analogs were deemed bioavailable using a bioavailability criteria specific for macrocycles. Toxicity predictions using PASS Online and StopTox identified analogs JGL023, JGL024, JGL032, and JGL047 as potential drug candidates. Molecular dynamics simulations using Gromacs-2020.4 revealed that JGL023 and JGL032 exhibited the most favorable binding to the Sec61 channel, as determined by evaluating ligand and residue flexibility, compactness, contact frequency, motion pathways, free energy, and other relevant parameters. Synthetic routes for these four analogs were proposed for future studies. The results of this study offer a new perspective on developing drugs to inhibit HIV entry.
Similar content being viewed by others
Introduction
In 1984, it was discovered that the primary causative agent of acquired immunodeficiency syndrome (AIDS) is the human immunodeficiency virus (HIV)1. HIV’s persistence in hidden reservoirs of infected CD4 (cluster of differentiation 4) cells, even after active antiretroviral therapy (ART), complicates efforts to cure the disease2. The CD4 T-cell receptor is crucial for HIV’s entry into host cells, making it a potential target for therapeutic intervention. Cyclotriazadisulfonamide (CADA), a synthetic macrocyclic sulfonamide, has garnered attention for its unique properties3. Initially synthesized as an intermediate in bicyclic triamine production4, CADA features a 12-atom ring, including three nitrogen atoms. Recent studies have shown that both CADA and eeyarestatin I (ESI), another synthetic compound, can bind to the Sec61 channel, further affecting the expression of cell surface proteins, cytokines, and viral membrane proteins5. Figure 1A illustrates the structure of CADA.
(A) Chemical structure of the CADA compound. (B) Structural model of the Sec61 protein channel. (C) Superimposed binding poses of CADA on the Sec61 protein channel. Left: Yellow (AutoDock Vina) and Gray (cryo-EM, RCSB ID: 8DO2). Right: Yellow (AutoDock Vina without Meeko treatment) and Gray (cryo-EM, RCSB ID: 8DO2).
CADA has demonstrated a unique mechanism in preventing HIV infection by inhibiting viral transmission between cells. This unique antiviral mechanism involves interaction with CD4 receptor at the post-translational level, and most recently, demonstrating client-specific inhibition through Sec61 channel binding. This is mainly exerted by blocking the protein translocation process, specifically altering the conformation of the Sec61 channel, thereby restricting CD4 protein translocation5. The CD4 T-cell receptor is a type I transmembrane protein (TMP) substrate of the Sec61 channel, predominantly expressed on T helper lymphocytes and thymocytes5. CADA’s antiviral activity is primarily attributed to its ability to downregulate the CD4 T-cell receptor in T cell lines by binding to a specific pocket in the Sec61 channel, located halfway across the cell membrane. CADA’s binding site significantly overlaps with the binding site of a client’s signal sequence, ultimately leading to its translocation inhibition. By reducing the number of CD4 receptors on the surface of CD4+ cells by nearly 90%, CADA limits HIV binding and entry without affecting other surface receptors6. The current antiretroviral therapy (ART) has decreased HIV-associated morbidity and transmission. However, it does not completely cure HIV, as the virus will persist in a state of chronic infection once ART is stopped. Moreover, Hunt et al.7 found that in their study on the prevalence of HIV drug resistance among infants in South Africa, the highest incidence of antiviral resistance occurred in infants whose mothers had received ART, despite the infants having no prior direct exposure to antiretroviral drugs. Hence, drugs that target HIV structure and enzymes have limited potential and are subject to antiviral resistance. Unlike other antiretrovirals, CADA exhibits activity against a wide range of HIV strains, as all HIV variants require CD4 binding for infection. CADA’s proposed mechanism prevents HIV entry by targeting CD4 expression, which is not subject to antiviral resistance. Additionally, CADA shows synergistic activities with clinically approved anti-HIV drugs such as zidovudine and lamivudine, offering potential for enhanced therapeutic strategies8.
According to the Chang9, the eukaryotic Sec61 translocon is a heterotrimeric highly conserved protein complex that consists of the three subunits: Sec61α, Sec61β, and Sec61γ. The translocon has a central pore that is sealed by a plug helix and a ring of hydrophobic amino acid region to prevent ion reflux. The Sec61α is located at the pore of the channel and consists of ten transmembrane helices (TM1-10). The TM5 and TM6 forms a hinge that connects to the Sec61γ while the gap formed by the TM2b and the TM7 provides a space for the signal peptide of secretory proteins to move to during co-translational translocation10. Two of the transmembranes of Sec61α is located on the lateral gate of the channel. These TMs receive the helices or signal peptide sequences of the newly-synthesized protein upon its biogenesis from the ribosome, and delivers it through the ER’s membrane11. The Sec61 channel has a hydrophobic pocket characterized mainly by the TM2b, TM3, and TM7 in the lateral gate of the Sec61α subunit wherein the insertion of transmembrane domains and signal peptides into the membrane is enabled. Moreover, it was found that CADA binds to the Sec61 channel in a client-specific manner at a hydrophobic pocket formed from a partially open gate, whereas other Sec61 inhibitors such as apratoxin and mycolactone acts in a broad-spectrum inhibition5. The mechanism for client-specific inhibition is that once the inhibitor binds to the plug at the partially open conformation, it is displaced only by ligands exhibiting strong signals, such as a transmembrane signal anchor. These strong signals allow the signal peptide to attach to the partially open lateral gate, which then changes the gate’s conformation to open further and displace the inhibitor, allowing successful translocation. Client-specific translocation inhibition is favorable in this case, as Sec61 channel inhibitors should not inhibit every single translocating protein. CADA’s client-specific inhibition means it allows the translocation of other proteins while selectively inhibiting certain clients. This fact hints that CADA may be less effective in transforming the Sec61 channel protein in conformations that inhibit the translocation of proteins with stronger signal peptide interactions.
Despite its promising mechanism, CADA’s therapeutic potential is hindered by its poor solubility and bioavailability, impairing its absorption and effectiveness. Recent work has concentrated on synthesizing various CADA analogs and evaluating their CD4 inhibition properties, serving as potential alternatives for antiretroviral therapy12. While cell-based assays are typically used to measure CD4 downmodulating activity, this approach does not guarantee binding to the intended site. However, it is generally assumed that analogs with lower IC50 values exhibit greater binding affinity. Efforts have primarily focused on modifying different groups of CADA to improve CD4 downmodulation, but these modifications often result in poor water solubility, limiting biological efficacy. Improving drug solubility is crucial for orally administered drugs. Hence, the development of CADA analogs fused with pyridine rings has been pursued to enhance water solubility13. Despite progress in synthesizing CADA derivatives, solubility and bioavailability remain challenges for therapeutic application.
CADA’s poor bioavailability and cellular toxicity in vitro are notable issues. Macrocyclic drugs often face challenges in meeting bioavailability criteria, which typically apply almost exclusively to small molecules. However, several FDA-approved macrocyclic drugs demonstrate favorable physicochemical properties and bioavailability. Villar et al.14 focused on utilizing macrocycles to target non-druggable protein targets and proposed specific guidelines for designing synthetic macrocyclic drugs, emphasizing the need for favorable physicochemical properties and good bioavailability. Computational approaches have been extensively used in drug discovery and protein interaction studies, providing insights into receptor stability, binding affinities, and mutational analysis, as well as in identifying selective inhibitors and addressing solubility challenges in therapeutic compounds15,16,17,18,19,20. This study utilized macrocycle-protein complexes, investigating their common binding features to guide the pharmacologic design of synthetic macrocycles intended to target proteins. Guidelines were proposed based on these findings and the physicochemical properties of approved macrocyclic drugs (Table 1).
The main issue with CADA’s bioavailability stems from its low PSA and molecular weight. This study focused on two goals: designing CADA analogs with enhanced binding affinity to the Sec61 channel and creating analogs with improved bioavailability and reduced toxicity. Both goals provide critical data on specific modifications that enhance CADA’s binding affinity and bioavailability. Modifications were introduced to the head, sidearm, and tail groups of the molecule, and potential inhibitory effects on CD4 T-cell receptor translocation were assessed using molecular docking, bioavailability criteria, and online toxicity prediction tools.
Materials and methods
Collection of Base compounds and protein structure
RCSB Protein Data Bank and UniProt
The structure of the Sec61 protein was obtained from the RCSB Protein Data Bank (RCSB PDB), which was used to determine active sites, 3D structural data, and resolution of microscopy-based sequences. This provided the exact binding site amino acid residues for CADA. This was used in conjunction with Swiss Model to select the most suitable Sec61 protein model. The amino acid sequence of Sec61 was retrieved from the UniProt database (ID P61619).
Swiss model
The amino acid sequence of Sec61 was constructed into a 3D structure using Swiss Model. A high-resolution 3D structure of Sec61, with a resolution of less than 3Å (preferably less than 2Å), was obtained. The model with RCSB ID 8do2.1.C was chosen, which represents a cryo-EM structure of the Sec61 protein (human) inhibited by CADA (Fig. 1B). This model was selected due to a number of reasons: its resolution is less than 3Å, includes CADA as its co-crystallized ligand, and represents the most recent Sec61 protein model available on RCSB. The active site was chosen from the cocrystal pose of CADA and covering all the residues involved in the key interactions with CADA.
Minimization of Ligands and Protein Preparation
Avogadro
Ligand minimization was first performed using Avogadro, a molecular editing and visualization software. In the study of Rajendran et al.21, Avogadro was employed in the minimization of the energy of the structures in the behavioral analysis of compounds. CADA and its analogs were protonated at pH 7.4 to simulate physiological conditions. The protein was similarly treated. Geometric optimization of the ligands was performed using the MMFF94 force field to minimize intramolecular clashes. The MMFF94 force field is chosen for its accuracy in conformational searching in organic compounds compared to other force fields, like GAFF and UFF22. Ligands were optimized until energy levels stabilized and saved in .sdf format for further use.
AutoDock tools
AutoDock Tools was used for further protein optimization before docking, defining the grid box parameters for the active site, and visualizing the protein-ligand complex. The Sec61 model (8Do2) was loaded as a macromolecule, with non-polar hydrogens added in Avogadro made implicit for docking. Kollman charges were applied to the protein to simulate in-vivo conditions. The grid box, set to 24 × 24 × 24 Å with a resolution of 0.375 Å, was centred on the Sec61 channel’s active site. The exhaustiveness was set to 64 (optimized for macrocycles), with number of modes set to 10, and energy range set to 4. These parameters were saved into a .txt file for AutoDock Vina.
Software for molecular docking and analysis
AutoDock vina
AutoDock Vina was used for the molecular docking of CADA analogs. Vina operates without its own interface and was utilized using an Anaconda shell, utilizing dependencies including Meeko, Open Babel (as a Python package), NumPy, and RDKit. Specifically, AutoDock Vina version 1.2.4 was used to accommodate the use of Meeko, a vital python plug-in. At an exhaustiveness of 64, Vina’s average root mean square deviations (RMSD) for macrocycles is 1.2223. AutoDock Vina’s accuracy in predicting binding energies is comparable to AutoDock and other docking sofwares24.
AutoDock Vina’s scoring function is in the top 25% when compared to other free and paid docking softwares25. Moreover, in the most recent version of AutoDock Vina, macrocycles can be made flexible with the aid of Meeko which serves as a link between AutoDock Vina and RDKit23. There is no study yet comparing AutoDock Vina and other docking software with respect to the accuracy of the binding energy involving macrocycles. In the past Drug Design Data Resource (D3R) challenge 4, AutoDock Vina-based softwares was among the top 10 submissions for the free energy prediction challenge26. Ligands underwent a final modification by Meeko in the Anaconda shell, with issues from OpenBabel installation addressed by integrating a package from an online source. Meeko was used to manage cyclic structures and introduce pseudoatoms, which were presented as atom types G0 and CG0. The ligands were then saved in pdbqt format, and the docking parameters were fed into Vina for docking simulations. In the study of Eberhardt et al.23. Meeko was implemented in AutoDock Vina and was validated using the 19 macrocycles from the BACE-1 set of the D3R Grand Challenge 4. The resulting RMSD average in the redocking of macrocycles using AutoDock Vina is between 1 and 2 angstroms. Docking of macrocycles is a challenging task because of the difficulty of sampling the ring flexibility by modeling the correlated torsional changes resulting in different conformations. AutoDock Vina 1.2.4 has a specialized protocol to dock macrocycles while modeling their flexibility on-the-fly. Molecular docking results served as the first initial criterion for screening the designed analogs of their binding energies and investigating their protein-ligand interactions with Sec61.
UCSF chimera
UCSF Chimera was used to visualize the ligand-protein binding poses post-docking. The output file from AutoDock Vina, containing multiple binding poses, was analyzed in Chimera’s ViewDock tool. The pose that has the most negative binding energy was selected, added to the original protein structure, and exported as a .pdb file for further analysis.
BIOVIA discovery studio visualizer
BIOVIA Discovery Studio was used to analyze protein-ligand interactions of the top binding analogs, providing insights to their obtained binding energies as determined by previous molecular docking procedures. After the .pdb file was loaded into BIOVIA, pseudoatoms were manually removed, as BIOVIA cannot process non-single fragment ligands. After removing two pseudoatoms per ligand, a 2D interaction diagram was generated to visualize key interactions between the protein and ligand.
Designing of CADA analogs
SeeSAR and ChemDraw online
CADA analogs were designed using ChemDraw Online for initial template creation and SeeSAR for visualizing unoccupied pockets in the active site. SeeSAR was used by Singh and coworkers in the discovery of inhibitors for Nsp1 protein of SARS-CoV-227. SeeSAR was instrumental in guiding the creation of CADA analogs. SeeSAR directed the potential modification sites in the ligand to fill the active site’s unoccupied pockets, theoretically enhancing binding affinity and specificity. Existing CADA derivatives from previous literature that were known to have favorable CD4 downmodulating activity were modified while complexed to Sec61 in SeeSAR to produce the designed analogs. The final analogs were saved in .sdf format for further preparation and docking.
Bioavailability and toxicity prediction of ligands
Bioavailability parameter procurement
SwissADME was used to predict ADME parameters and pharmacokinetic properties of the designed analogs. However, its bioavailability prediction model is not suitable for macrocyclic compounds. Therefore, the bioavailability criteria proposed by Villar et al.14 was used to screen the bioavailability of the designed CADA analogs. Analogs meeting these criteria were then further assessed for toxicity.
Toxicity prediction
PASS online and stoptox
Toxicity predictions of the analogs were performed using PASS Online and StopTox. PASS Online predicts potential biological interactions and estimates the probability of toxicity with “active” (Pa) and “inactive” (Pi) values. StopTox provides toxicity predictions through validated QSAR models and machine learning algorithms28. Analogs passing the bioavailability criteria were subsequently screened for toxicity prediction using these tools.
Molecular dynamics simulation
Potential drug candidates were identified after screening and narrowing down all analogs based on their bioavailability and toxicity. These candidates were then subjected to molecular dynamics (MD) simulations to get deeper insights into their comparative binding affinities and ability of the each respective ligand to stabilize the Sec61 protein. The MD simulations of 100 ns duration were performed with the Gromacs-2020.429,30 program. The Gromacs-compatible input topologies of CADA and its analogs were prepared with Acpype31 based on the AMBER program32. Meanwhile, the input topology of the Sec61 protein was prepared using the Amber ff99SB protein force field33. The respective complexes of CADA analogs and CADA with Sec61 protein were placed in a dodecahedron unit cell and solvated with TIP3P water molecules34. Subsequently, the solvated systems were neutralized by adding sodium or chloride counter-ions. In order to relieve the steric strain in the Sec61 protein structure, each system was subjected to the energy minimization step with the combination of steepest descent and conjugate gradient minimization algorithms where the thresholds of Fmax less than 1000 kJ mol-1 nm-1 were set for during each minimization step. The energy minimized systems were then subjected to equilibration where the equilibration was first performed at constant temperature and volume (NVT) conditions where the modified Berendsen thermostat35 was employed to achieve the constant temperature of 300 Kelvin. After that, the equilibration was performed at constant pressure and volume (NPT) conditions employing the Berendsen barostat36 at 1 atm. Both the NVT and NPT equilibrations were performed for 1 ns each. The equilibrated systems were then subjected to production phase 100 ns MD simulations where the temperature conditions of 300 Kelvin was achieved with a modified Berendensen thermostat, and pressure conditions of 1 atm were achieved with the Parrinello-Rahman barostat37. The LINCS algorithm38 was employed to restrain the covalent bonds during the production phase MD simulations. The long-range electrostatic energies were computed with the Particle Mesh Ewald (PME) method39 at the cut-off distance of 1.2 nm. Post-MD simulations, the trajectories were treated for the periodic boundary conditions. The trajectories captured at each 10 ps were analyzed for the RMSD in the backbone atoms of Sec61 protein in each complex. The conformational flexibility in the ligand atoms relative to the backbone atoms of Sec61 protein was measured in terms of RMSD in ligand atoms where the Sec61 protein backbone atoms were selected for the most minor square fitting and ligand atoms were chosen for the RMSD calculations which in turn effectively showed insights into the relative positions of ligands compared to initial position during the simulation. The flexibility in the Sec61 protein residues was gauged regarding root mean square fluctuation (RMSF) in the side chain atoms. This especially provided important insights into the flexibility in the side chain atoms of the binding site residues. Further, the stability and overall compactness of each complex of Sec61 protein with CADA and CADA analogs were analyzed in terms of radius of gyration (Rg) measurement, where the rotation of Sec61 protein complexes around its center of mass was analyzed. The solvent-accessible surface area analysis was performed to get insights into the cavities and buried solvent-accessible surface area on the Sec61 protein. The hydrogen bonds formed between the CADA and CADA analogs with Sec61 protein were analyzed through the hydrogen bond analysis. The hydrogen bonds formed were further investigated in select trajectories at 0, 25, 50, 75, and 100 ns. The non-bonded interactions within the 3.5 Å distance, which majorly covers the hydrogen bond interactions, were analyzed through the contact frequency analysis, which was performed using the MDCiao40 program. The major path of motions in each complex of Sec61 with CADA and CADA analogs signifying the stable conformations were analyzed through the principal component analysis (PCA)41. In the PCA analysis, the covariance matrix for the complexes of Sec61 protein was constructed considering the backbone atoms of Sec61 protein using the gmx covar program. The covariance matrix was then diagonalized to obtain the eigenvectors signifying the path of motions and eigenvalues signifying the mean square fluctuations in the Sec61 protein backbone atoms. Employing the first two principal components (PC1 and PC2) as the reaction coordinates, Gibb’s free energy landscapes42 were constructed and analyzed for each complex. The trajectories from 50 to 100 ns at each 100 ps time step were subjected to the Molecular Mechanics General Born surface area and surface area solvation (MM/GBSA)43 calculations. During the MM/GBSA calculations the entropic energies were taken into consideration to obtain the binding free energies (ΔGbinding kcal/mol) for each complex. The structures of protein-ligand complexes were rendered in PyMOL44 while the plots of RMSD, RMSF, Rg, hydrogen bond, and solvent accessible surface area were obtained from XMGRACE(45, p. 19).
Results and discussion
Method validation
Validation via manual RMSD
To validate the docking protocol, a manual root mean square deviation (RMSD) check was performed. CADA, the co-crystallized ligand from the cryo-EM structure of Sec61 (RCSB ID: 8DO2), was redrawn in ChemDraw and minimized following the same protocol used for the analogs. The protonation of CADA at pH 7.4, followed by minimization using the MMFF94 force field was performed. The ligand was then prepared with Meeko before docking. The Sec61 protein was prepared using AutoDock Tools, where non-polar hydrogens were replaced with Kollman charges. The docking was performed, and the resulting protein-ligand complex was visualized using BIOVIA Discovery Studio. The docked complex was superimposed with the cryo-EM structure from RCSB to calculate RMSD, as shown in Fig. 1C.
Figure 1C demonstrates that the grid box accurately encompassed the active site, and the significant similarity overlap between the docked CADA (yellow) and the cryo-EM structure of CADA from RCSB ID 8DO2 (gray) confirmed the accuracy of the AutoDock Vina protocol. Despite the grid box being larger than the binding site, the positions of the docked CADA and cryo-EM structure of CADA were nearly identical. To further assess the significance of Meeko, CADA was docked without Meeko as a comparison. The ligand was prepared using AutoDock Tools with manually added Gasteiger charges and docked using AutoDock Vina with the same parameters. As shown on the right side of Fig. 1C, the docked CADA without Meeko exhibited a greater deviation from the cryo-EM CADA structure, validating the contribution of Meeko in improved ligand preparation.
Top binding designed CADA analogs
After method validation, we have designed a total of 113 CADA analogs for investigation of their binding energies, molecular interactions with the Sec61 channel, and possible drug candidates. Only the analogs that have a more negative binding energy than CADA will be considered for bioavailability and toxicity predictions. Out of 113 designed analogs, 93 displayed a more negative free binding energy compared to CADA. Supplementary Figures S24-S136 show the docking results and binding poses adopted by all 113 designed analogs on the Sec61 pocket. Figure 2 displays the binding poses of the top five CADA analogs with the most negative binding energies upon molecular docking. Modifications to CADA (binding energy, B.E. = -9.91 kcal/mol) resulted in analogs with more negative binding energies and distinct binding poses in the Sec61 pocket. The analog WM038 demonstrated the most negative binding energy (B.E. = -12.73 kcal/mol), a significant improvement compared to CADA (B.E. = -9.91 kcal/mol). This improvement indicates a successful affinity enhancement through SeeSAR-based design. WM038, derived from LAL036 (B.E. = -11.3 kcal/mol), shows a more negative binding energy post-modification from LAL036. Figure 3A–C shows the structural modifications of the top 10 CADA analogs and the resulting binding energy. The top three analogs, which demonstrated the most negative binding energies, were derived from LAL036. This suggests that analogs with fused-pyridine structures exhibit not only greater solubility but also enhanced binding affinity to Sec61. This increased affinity is likely due to the analog’s larger tail group interacting with more amino acid residues in the active site.
Binding poses of the top 5 CADA analogs (WM039, WM041, WM015, JGL034, and WM038) against the Sec61 protein channel, displayed in both ribbon and molecular surface representations. (A) WM039 (B) WM041 (C) WM015 (D) JGL034 (E) WM038
(A-C) Binding energies and structural representations of the top 10 designed CADA analogs. (D) 2D interaction diagram of LAL036 with the Sec61 protein. (E) 2D interaction diagram of VGD022 with the Sec61 protein.
Enzyme-ligand interactions
Literature derivatives
The 2D-diagrams of enzyme-ligand interactions for literature CADA derivatives were analyzed to guide the design of new analogs. Supplementary Figures S1-S23 show the binding poses adopted by these literature derivatives upon molecular docking. Among the top five designed analogs with the most negative binding energies (Fig. 3), four were derived from LAL036, while one was from VGD022.
LAL036, with the second-highest binding energy among literature derivatives (Table 2), was used as a template to design four of the top binding analogs. Figure 3D shows that LAL036 forms multiple hydrophobic interactions with the Sec61 protein, including van der Waals and alkyl/π-alkyl interactions. The strongest interaction is hydrogen bonding between the nitrogen of the fused-pyridine ring and Thr86 of Sec61. These interactions were either retained or enhanced in the top analogs.
VGD022, with the third highest binding energy among literature derivatives (Table 2), served as the template for one of the top-designed analogs. Figure 3E shows VGD022 interacting with Sec61 through multiple hydrophobic interactions, with significant van der Waals interactions with Leu89 and Asn300. The strongest interactions were the two hydrogen bonds between the sulfonamide oxygen and Gln127 and Ser82 of the protein. All interactions were either retained or enhanced in the top-designed analogs.
Designed CADA analogs
The main problem for CADA’s bioavailability mainly stems from its low PSA and molecular weight. Hence, designing procedures employed strategies that sought to accomplish two goals: to increase binding affinity of analogs and analyze the enzyme ligand interactions at play resulting from the modifications, and to improve solubility and bioavailability by adding more polar groups in the structure. As for the first goal, the top five designed CADA analogs with the most negative binding energies were evaluated for their enzyme-ligand interactions using BIOVIA to understand the factors contributing to their binding energy. Supplementary Table S1 lists the structures and binding energies of all 113 designed CADA analogs. Table 3 lists the top analogs derived from LAL036 (top 1, 2, 3, and 5), and exhibited more negative binding energies than the parent compound. The binding energy of LAL036 (B.E. = -11.30 kcal/mol) is attributed to various intermolecular forces, including van der Waals, π-sigma, π-sulfur, and alkyl/π-alkyl interactions with the amino acid residues of the Sec61 protein.
Figure 4A illustrates that even minimal differences in the binding poses of compounds can lead to significant changes in binding energy. Small changes, such as the addition or removal of a methyl group or the conversion of sigma bonds to π bonds, can substantially affect the ligand’s binding affinity. This effect is particularly notable when comparing LAL036 to WM039 (B.E. -12.21 kcal/mol), the analog with the second most negative binding energy (Fig. 4A). Among the analogs derived from LAL036, WM039 exhibited the least visual difference in binding pose compared to the other derivatives.
(A) Comparative structures: Left - WM039, Right - LAL036, and their superimposed structures in the center (WM039 in yellow). (B) Comparative structures: Left - WM015, Right - VGD022, and their superimposed structures in the center (WM015 in yellow). (C-D) Binding energies and structural representations of bioavailable and low-toxicity CADA analogs.
WM039 differed from LAL036 mainly by replacing the cyclohexane tail connected to the fused pyridine structure with a benzene ring. This change is also present in WM038 and WM041. In addition to this, WM039 has an ethyl group added to another benzene ring. While this addition increases intermolecular forces of attraction, particularly π-alkyl interactions, it is the introduction of the aromatic benzene that ultimately enhances the potency of the LAL036-based analogs shown in Fig. 3. The benzene ring makes the region previously occupied by cyclohexane more polarizable due to the π bonds in the aromatic ring, significantly strengthening the van der Waals interactions with the protein residues in that region. A similar effect is seen with JGL034, where the cyclohexane bound to the fused pyridine is replaced with another pyridine ring.
The top analog in terms of binding energy is WM038 (B.E. = -12.73 kcal/mol). This analog benefits from multiple hydrophobic interactions with the amino acid residues of the protein. The strongest interactions include two hydrogen bonds: a conventional hydrogen bond between Gln127 and the sulfonamide oxygen atom, and a carbon-hydrogen bond between Asn300 and a carbon atom in the ring. Additionally, an amide-π stacking interaction between the toluene group and Gly126 was observed. Gly126, located in an alpha helix of the lateral gate transmembrane protein, is a flexible residue that destabilizes the helix. The amide-π stacking interaction with WM038 stabilizes Gly126, enhancing the binding affinity. These interactions are key to WM038’s exceptionally negative binding energy.
In comparison, the second-best analog, WM039 (B.E. -12.21 kcal/mol), has only one hydrogen bonding interaction: a conventional hydrogen bond between Thr86 and the nitrogen atom of the fused pyridine. This bond is significant due to the donor being an OH group from threonine, which creates a stronger interaction than the -NH2 donor and oxygen acceptor bond found in WM041. WM039 also benefits from multiple hydrophobic interactions, including van der Waals and π-sigma interactions, which contribute to its negative binding energy.
For the third-best analog, WM041 (B.E. = -11.98 kcal/mol), the strongest interactions include two hydrogen bonds: a conventional hydrogen bond and a carbon-hydrogen bond between Gln127 and the sulfonamide oxygen atom, and Asn300 with a carbon atom in the ring. WM041 also exhibits an amide-π stacking interaction with Gly126, contributing to its very negative binding energy. However, WM039 has a more negative binding affinity compared to WM041 because the hydrogen bond in WM039 is stronger. The hydrogen bond donor in WM039 is an -OH group (threonine), while WM041 has an -NH2 donor (glutamine). Oxygen’s higher electronegativity makes the hydrogen bonded to it more positively charged, resulting in a stronger interaction compared to nitrogen-bonded hydrogen. Table 3 summarizes these BIOVIA results.
As for structure-binding energy studies in LAL036 derived analogs, it was found that modifications to the tail group had the greatest impact to the compound’s binding energy. Generally, LAL036 derived analogs with a less polar tail demonstrated a more negative binding energy, which agrees to the findings of Bell et al.6. Moreover, shortening the tail of LAL036 by 1 atom generally increases the compound’s binding affinity to Sec61, as demonstrated by analogs WM036 to WM041. As for attempts to make more bioavailable analogs by adding polar groups, it was generally found that making the head or tail into a carbonyl group decreases binding affinity as demonstrated by analogs JGL022 to JGL033 and WM031.
WM015 is the only analog among the top five not derived from LAL036. The LAL036 compounds benefit from their extended fused-pyridine structures, which enhance binding affinity through van der Waals interactions and increased bioavailability. WM015, which retains the original ring structure of CADA and only modifies the tail groups, has a naphthalene structure in its side arm groups, contributing significantly to van der Waals forces due to π bonds. The structural difference between VGD022 and WM015 are: the exocyclic double bond in VGD022’s main ring is replaced with a carbonyl group in WM015, and the benzene tail in VGD022 was replaced by a cyclohexane in WM015. This change resulted in a significant alteration in WM015’s binding pose, as shown in Fig. 4B. The replacement of a carbon-carbon double bond with a carbonyl bond caused a roughly 180-degree rotation around an axis, allowing WM015 to form hydrogen bonds with Ser82 and Gln127 on the carbonyl head group. WM015 thus gained three hydrogen bonds compared to the two in VGD022, although it is less tightly bound to the protein than LAL036 analogs due to fewer intramolecular aromatic structures. WM015 achieved a more negative binding energy compared to JGL034, primarily because JGL034 lacks amide-π stacking with Gly126, a major contributor to the negative binding affinity of LAL036 analogs (Table 4).
In the structure-binding energy studies in VGD022 derived analogs, it was observed that adding methyl groups to the naphthalene ring demonstrated more favorable (negative) binding energies as demonstrated by multiple analogs (e.g., WM018 and WM019, WM012 and WM013, and WM017 and WM020), which changed the binding poses of the compound upon methyl addition. Moreover, introducing a carbonyl group to the heads of VGD022 derived analogs produced mixed outcomes, increasing binding affinities (WM) and decreased it (WM014 vs. WM016).
Bioavailability studies
Macrocyclic bioavailability
Modifications to CADA and the existing literature derivatives mainly focused on introducing more polar groups to increase the polar surface area (PSA) and molecular weight, with the goal of increasing bioavailability. Most changes involved adding hydroxyl, amine, and carbonyl groups to the structure, which successfully increased PSA but did not result in any significant improvement in binding affinity. Addition of pyridine to structures is known to increase solubility (Islam et al.49). Most CADA analogs are modified with pyridine-fused structures to increase the analog’s solubility as evidenced by the study of Lumangtad et al.13 on pyridine-fused structures. As adding hydrophobic groups in the tail increases the analog’s potency, they incorporated pyridine to the tail as well to somehow counteract the decrease in solubility due to the hydrophobic groups. Aside from pyridine groups being abundant in drugs, fused pyridine-groups enhance the analog’s water solubility due to its polarity. Its moderate pKa (5.23) allows it to undergo easy protonation and deprotonation during cell permeation, adapting to different cellular environments (St. Jean & Fotsch50). CADA’s bioavailability was assessed using the guidelines proposed by Villar et al.14 in the designing of bioavailable macrocyclic drugs, as detailed in Table 1. Out of 113 designed CADA analogs, 33 met the bioavailability criteria, wherein 24 of these were identified as orally bioavailable, while 9 were non-orally bioavailable. However, the lack of oral bioavailability does not preclude these compounds from being potential drugs, as they may be administered through alternative routes. A common feature among the bioavailable analogs was the presence of a carbonyl in the head group instead of an exocyclic double bond. This modification significantly increases the polar surface area (PSA) and reduces the logP of the compound, aligning with bioavailability requirements. Of the 33 bioavailable analogs, only 4 were considered potential drug candidates. Their structures and corresponding binding energies are illustrated in Fig. 4C–D, with their binding poses shown in Fig. 5.
Binding poses of bioavailable, low-toxicity CADA analogs (JGL023, JGL024, JGL032, JGL047) against the Sec61 protein channel, displayed in both ribbon and molecular surface representations. (A) JGL023 (B) JGL024 (C) JGL032 (D) JGL047.
Physicochemical Properties
Table 5 presents the physicochemical properties relevant to determining the bioavailability of macrocycles. The table highlights the properties of the four proposed drug candidates, which have been already screened of their predicted toxicities. Analogs JGL023, JGL024, and JGL032 are deemed orally bioavailable using the bioavailability criteria, while JGL047 is deemed non-orally bioavailable using the same criteria. CADA, however, did not meet multiple bioavailability requirements. For a macrocyclic compound to be classified as non-orally bioavailable, its molecular weight should be between 600 and 1,300 g/mol, PSA between 150 and 500 Ų, with no more than 17 hydrogen bond donors, 9–20 hydrogen bond acceptors, and no more than 30 rotatable bonds (NRT). CADA has a molecular weight of 581.79 amu and a PSA of 94.76, which is way less than the necessary value for a macrocyclic compound to be classified as bioavailable. JGL047 was classified as non-orally bioavailable due to its PSA value falling short of the minimum required 180 Ų, thus disqualifying it as an oral drug candidate. The complete data on the physicochemical properties of all 33 bioavailable analogs can be found in Supplementary Table S1.
Lipophilicity
Lipophilicity is a key parameter assessed during the screening of potential drug compounds. It refers to a compound’s ability to penetrate the lipid bilayer and enter cells. Lipophilicity is important for developing drugs with optimal absorption rates and metabolic uptake. The primary measure of lipophilicity is the octanol/water partition coefficient, or logP. This value indicates a compound’s tendency to accumulate in the body versus being excreted. Higher logP values suggest greater accumulation, while lower values indicate easier excretion. Table 5 displays the logP values for the 4 proposed drug candidates. Comprehensive data on the logP values of all 33 bioavailable analogs can be found in Supplementary Table S1.
Water Solubility
Water solubility is a critical factor in drug development, as it affects the concentration of a drug in the bloodstream and its subsequent distribution to tissues. This property influences the dosage and pharmacologic potential of a drug. Achieving balance between water solubility and lipophilicity is essential for optimal drug absorption and bioavailability. Table 6 presents the water solubility classifications of the proposed CADA drug candidates based on their logS values. A compound is classified as insoluble if the logS value is less than − 10, poorly soluble if between − 10 and − 6, moderately soluble if between − 6 and − 4, and soluble if between − 4 and − 2. CADA is identified as poorly soluble across all three solubility models. For small orally bioavailable molecules, a suitable logS value typically falls between − 6 and 0. Although there is no specific criterion for macrocyclic compounds, among the 24 orally bioavailable analogs, 19 are moderately soluble in at least one model, while 5 are poorly soluble in all models. Given CADA’s poor water solubility, developing analogs with improved water solubility could significantly advance the field. The complete data on the logS and solubility classes of all 33 bioavailable analogs can be found in Supplementary Table S2.
Toxicity studies
The toxicity of the designed analogs meeting bioavailability criteria was evaluated using PASS Online and StopTox to identify potential toxic and adverse effects. PASS Online is known to give decent predictions. StopTox is regarded as an alternative to animal testing and may provide additional information. This software is not 100% accurate. For example, PASS Online is reported to have an average accuracy just above 95%51. Other tools that may be used are ProTox-3.0 and Toxicity Estimation Software Tool. These tools helped narrow down the 33 bioavailable analogs to 4 potential drug candidates: JGL023, JGL024, JGL032, and JGL047. PASS Online predicts toxicity based on Pa values, where a Pa greater than 0.7 indicates a high likelihood of exhibiting the activity, a Pa between 0.5 and 0.7 suggests a moderate likelihood, and a Pa less than 0.5 implies a low likelihood. Supplementary Figures S137-S170 shows the complete PASS Online results of all 33 bioavailable analogs. On the other hand, StopTox assesses toxicity through “6-pack” in vivo assays, which include acute oral toxicity, acute dermal toxicity, acute inhalation toxicity, skin irritation and corrosion, eye irritation and corrosion, and skin sensitization. These assays are required by regulatory agencies and use machine-learning models based on validated QSAR models for prediction.
The complete toxicity prediction results for the 4 proposed drug candidates are available in Supplementary Table S3. These results are why the initial 33 bioavailable analogs were narrowed down to 4 candidates. As shown in Fig. 6A, these 4 analogs demonstrated Pa values below 0.7, indicating an improvement over CADA, which was predicted to likely exhibit toxic interactions. Among the 4 analogs, JGL024 and JGL047 had Pa values below 0.5, suggesting they are unlikely to exhibit toxic biological interactions upon metabolism. JGL023 and JGL032 had Pa values between 0.5 and 0.7, indicating possible side effects rather than outright toxicity. Notably, JGL023 is predicted to act as a kinase inhibitor, which could be beneficial in therapeutics if it has sufficient specificity, potentially treating diseases such as cancer and immune-mediated disorders. However, kinase inhibitors can also cause side effects like headaches and diarrhea.
Toxicity assessment of proposed drug candidates: (A) PASS Online toxicity results for CADA, with Pa values exceeding 0.7 indicated by the red line. (B) StopTox results indicating a high likelihood of acute oral toxicity for CADA.
Figure 6B shows the StopTox results for the proposed CADA analogs, with Supplementary Table S4 providing details for all 33 bioavailable analogs. The results indicate that none of the analogs are predicted to cause acute oral toxicity, acute dermal toxicity, acute inhalation toxicity, skin irritation and corrosion, or skin sensitization. However, all were predicted to cause eye irritation and corrosion. Compared to CADA, which is predicted to be orally toxic and cause eye irritation, the modified analogs show significantly reduced toxicity risk. The 4 proposed drug candidates would theoretically be safer as oral drugs while maintaining their potential as HIV treatments.
Observed trends in toxicity include that adding a lactone group to the tail of the analogs, as seen in JGL047, significantly reduces predicted toxicity. Conversely, adding an OH or OR group to the meta position of a benzene ring tends to increase predicted toxicity, as demonstrated by several analogs.
Molecular dynamics simulation
Based on the docking studies and bioavailability studies, four designed compounds, namely JGL023, JGL024, JGL032, and JGL047, from amongst 113 designed CADA analogs were selected for MD simulations to get further insights into their binding affinity, their propensity in stabilizing the Sec61 protein, and their ability to adopt the biologically relevant binding poses. The standard drug CADA was also selected for the MD simulation studies to obtain comparative insights. The respective docked complexes of these ligands with Sec61 were taken up for the MD simulation studies. MD simulation setup creates a biologically relevant environment with solvent and neutrality with biologically relevant counter-ions. It helps find ligands binding affinity, binding free energies, and influence on protein’s structural flexibility and stability52.
Root mean square deviation
The RMSD in the backbone atoms of Sec61 protein in complex with CADA showed significant deviations during 40 to 80 ns simulation duration (Fig. 7A), and the overall average RMSD was 0.384 nm (Table 7). The complex with JGL023 showed slightly lower RMSD with an average of 0.376 nm with significant deviations during 60 to 80 ns simulation duration. The complex with JGL024 showed significant structural change, which was evident from the major deviations in the RMSD during the 60 to 80 ns simulation duration. The complexes with JGL032 and JGL047 showed reasonably converged RMSD with averages of 0.404, and 0.464 nm, respectively. The RMSD in ligand atoms relative to the Sec61 protein backbone atoms gave an insight into the dynamical changes in the ligand poses and the position of ligands relative to the binding site. The results showed that the structure of CADA remained stable with fewer deviations at the binding site with average RMSD in CADA atoms of 0.230 nm (Fig. 7B). The structure of JGL024 also remained stable in the binding site with an average RMSD of 0.295 nm, with few deviations at the 40 to 80 ns simulation period. The structure of JGL023 remained stable in terms of reasonably converged RMSD with slightly higher RMSD with an average of 0.440 nm. The RMSD in JGL032 and JGL047 was almost similar, with the average RMSD of 0.366 and 0.350 nm, respectively.
The RMSD, RMSF, Rg, and SASA analysis. A) RMSD analysis in Sec61 protein backbone atoms, B) RMSD analysis in ligand atoms relative to Sec61 backbone atoms, C) RMSF analysis in Sec61 protein residues, D) Rg analysis, and E) Solvent accessible surface area analysis.
Standard deviations in average values are given in parentheses.
The RMSD in the protein backbone atoms is an important measurement to get insightful clues about the stability of the resulting protein-ligand complex and to quantify the degrees of conformational freedom of protein53. Ideally, if the RMSD in protein backbone atoms is below 0.3 nm, the protein or protein-ligand complex conformational change is considered stable54. The RMSD in Sec61 backbone atoms suggests that the stable conformations of complexes bound to JGL023 were lower than those bound to standard drug CADA, indicating better stability of the Sec61 protein JGL023 complex. Among other complexes, the complex with JGL047 is seen as slightly unstable as the RMSD rises above 0.5 nm after around 50 ns. The complexes of Sec61 protein with JGL024 and JGL032 are reasonably stable as the RMSD remained without any significant deviations, except in the complex with JGL024, showing significant deviations during the 60 to 80 ns simulation period. The analysis of RMSD in ligand atoms, mainly when estimated relative to protein backbone atoms, gives insights into the position and conformational freedom ligands in the binding site during MD simulation, compared to the starting docked poses52. In such a measurement, if the ligand stays in the binding site in a stable conformational state, the consequent RMSD is lower, suggesting the consequent stability of the protein-ligand complex. It is seen that CADA remained bound to the Sec61 protein throughout the simulation period in a stable conformational state and thus has the lowest RMSD in its atoms relative to the Sec61 protein’s backbone atom.
Similarly, the JGL024 also adopts a relatively stable conformational state at the binding site of the Sec61 protein, which is comparable to CADA. Although the RMSD in Sec61 protein backbone atoms in complex with JGL047 is seen higher, the ligand remained bound in a reasonably stable conformational state as the RMSD in its atom is also lower. JGL023 and JGL032 also remained bound and reasonably stable. However, the RMSD in ligand atoms in these complexes is higher, indicating they might have adopted a different conformational state than the initially docked poses.
Root mean square fluctuation
The RMSF analysis in the Sec61 protein residues showed that the fluctuations in the residues in the range 90–110, 190–240, and 250–280 were common in all the complexes (Fig. 7C). The RMSF in the case of Sec61 protein residues in the complex with CADA was lower than the JGL023, JGL024, and JGL032 complexes. The Sec61 protein residues in the complex with JGL023 showed significant fluctuations in the residues in the range 190–250, whereas the complex with JGL024 showed significant fluctuations in the residues in the range 250–290. The complex with JGL032 showed significant fluctuation in the range 80–110, 200–250, and 300–350. The complex with JGL047 showed lower fluctuations comparable to that with CADA. The detailed analysis of the higher fluctuation in the residues suggested that significantly lower fluctuations were observed in the residues Ile95, Met202, Thr219, Gln270, Glu351, and Arg400 in the case of the complex with CADA. JGL023 caused the highest fluctuations in the residue Gln270 and Arg400. JGL024 caused major fluctuations in the residue Met202, whereas JGL032 caused major fluctuations in the residue Glu351. JGL047 caused the major fluctuations in the residue Ile95. These residues are part of loop regions in the active site and the flexibility and adaptation of respective ligands might have resulted in the significant fluctuations in the side chain atoms of these residues.
The RMSF analysis further provides insights into the side chain flexibility of protein residues during simulation, which is important in understanding the structural stability of the protein-ligand complex and the binding site adaptation55. Furthermore, the residues from the unstructured loop regions usually show higher RMSF due to the higher flexibility in their side chain atoms56. The residues in the range 90–110, 190–220, and 360–420 with RMSF reaching 0.5 nm are quite flexible. Most of the residues in these ranges belong to the binding site and belong to the loop regions. However, in the complexes with CADA and the ligand JGL047, the RMSF in the residues in these ranges is lower, indicating stable protein-ligand complexes. The complexes with other ligands are also reasonably stable but show significant fluctuations in the range of residues at the binding site.
Radius of gyration
The Rg analysis, which provided insights into the compactness of the complex of Sec61 protein with ligands, showed that the complex with JGL023 was stable and reasonably compact with an average Rg of 2.473 nm compared to the complex with CADA with an average of 2.537 nm (Fig. 7D). The Rg in Sec61 protein complex with JGL024 showed deviations after around 40 ns simulation period and had an average Rg of 2.508 nm. The complexes of JGL032 and JGL047 showed deviations after around 25 ns simulation period and had almost similar Rg with averages of 2.520 and 2.516 nm, respectively.
The radius of gyration is a measure of protein rotation around its center of mass. It provides insights into the compactness of protein-ligand complexes and the stability of corresponding protein-ligand complexes57. Lower values of Rg and converged plots of Rg suggest compact and stable protein-ligand complex structures58. Larger Rg values and any significant Rg deviations indicate the functional impairment of the protein-ligand complex59. The lowest Rg in the Sec61 protein complex with JGL023 indicates good stability and minimal secondary structural changes. The complex with JGL024 also suggests a compact and stable complex. Compared to the complex of CADA, the complex with ligand JGL032 also suggest reasonably good stability of corresponding protein-ligand complexes.
Solvent accessible surface area
The solvent accessible surface area analysis, which provides the measure of the cavities at the surface and deeply buried cavities, showed that the extent of SASA in the case of complex with CADA and JGL023 was almost similar to the average SASA of 249.821 and 247.127 nm56 (Fig. 7E). The complex with JGL024 had slightly higher and almost similar SASA with an average of 254.388 nm56. The complex with JGL032 had significantly higher SASA than the complex with CADA, with an average of 258.980 nm56.The complex with JGL047 had slightly higher SASA, with an average of 251.774 nm56.
The solvent-accessible surface area estimation is another key measure to investigate the stability of protein-ligand complexes and conformational changes in proteins upon binding respective ligands7. The larger SASA indicates more residues exposed to the solvent, indicating protein unfolding events with lower stability60. The complex of Sec61 protein with JGL023 having the lowest average SASA suggests the most compact and stable complex compared to the complex with CADA. The complexes with other ligands showed comparably good stability. However, the complex with ligand JGL032 might have major unfolding events indicating slightly lower stability of the corresponding system.
Hydrogen bond analysis
The number of hydrogen bonds formed between the Sec61 protein and each ligand during the MD simulation was analyzed. The results showed that CADA occasionally formed a maximum of 3 hydrogen bonds during the first 40 ns simulation period (Fig. 8). However, after that, the frequency of hydrogen bond formation decreased to around 1 frequent hydrogen bond. Further, the MDciao module measured the contact frequency between the CADA and residues within 3.5 Å to identify the key residues that possibly formed the hydrogen bonds. It was found that the residue Asn295 and Met60 formed hydrogen bonds with 50% contact frequency, indicating a 50% probability of these hydrogen bonds in the overall hydrogen bond interactions. At the same time, the residues Ser77, Val80, and Gln122 formed hydrogen bonds with around 40 to 50% contact frequency. The trajectories extracted at 0, 25, 50, 75, and 100 ns simulation time further confirmed that almost all the trajectories had the key hydrogen bonds with either Asn295 or Ser77 or both the residues Asn295 and Ser77.
The hydrogen bond and contact frequency analysis for Sec61 protein complex with CADA.
The ligand JGL023 formed very occasional hydrogen bonds during the first 40 ns simulation period, and the frequency rose after that, forming around 2 consistent and a maximum of 3 occasional hydrogen bonds (Fig. 9). The contact frequency analysis suggested that the residues Asn295 and Gln122 had more than 75% contact frequency, while Ala291, Leu84, and Thr81 had a contact frequency of around 55%. The trajectories at different time intervals showed that most of the trajectories had the typical hydrogen bond interaction with Asn295 residue. In contrast, some trajectories showed a hydrogen bond interaction with Thr81, Tyr126, or Gln122 residue.
The hydrogen bond and contact frequency analysis for Sec61 protein complex with JGL023.
Comparably, JGL024 formed fewer hydrogen bonds with Sec61 protein’s binding site, where one hydrogen bond was occasionally formed and reached a maximum of 2 occasionally (Fig. 10). The contact frequency analysis of these hydrogen bonds suggested that the residues Gln122 and Asn295 were predominantly formed with around 75% contact frequency. At the same time, the residues Met60, Ala291, and Leu64 showed a contact frequency between 30 and 60%. The trajectories at 0, 25, 50, and 100 ns showed a hydrogen bond with Asn295, whereas the trajectory at 0 ns also showed a hydrogen bond with Gln122. No hydrogen bond was formed in the trajectory at 75 ns.
The hydrogen bond and contact frequency analysis for Sec61 protein complex with JGL024.
The ligand JGL032 formed one consistent hydrogen bond throughout the simulation period and a maximum of 2 frequently formed hydrogen bonds (Fig. 11). Out of these, the hydrogen bonds with Gln122 had a higher occurrence as the contact frequency was around 75%. The hydrogen bonds with Asn295 and Tyr126 had more than 50 to 60% contact frequency, while the residue Ser294 had around 50% contact frequency. The trajectory at 0 and 25 ns showed a hydrogen bond with Thr81, while the trajectories at 75 and 100 ns showed a hydrogen bond with Asn295.
The hydrogen bond and contact frequency analysis for Sec61 protein complex with JGL032.
Around two hydrogen bonds were formed between the Sec61 protein and JGL047 (Fig. 12). Interestingly, the bonds with residues Thr81 had almost 90% contact frequency out of these hydrogen bonds, while the residue Asn295 had around 75% contact frequency. Gln122, Val80, and Ala291 residues showed between 30% and 55% contact frequency. All the extracted trajectories at 0, 25, 50, 75, and 100 ns showed the hydrogen bond with the residue Thr81. The trajectory at 0 and 75 ns additionally showed a hydrogen bond with Asn295, and the trajectory at 100 ns additionally showed a hydrogen bond with Ser294.
The hydrogen bond and contact frequency analysis for Sec61 protein complex with JGL047.
Non-bonded interactions such as van der Waals, electrostatic, and hydrogen bond interactions between ligand and protein stabilize the protein-ligand complexes. Amongst these, the hydrogen bond interaction is one of the significant non-bonded interactions that affect the binding affinity of ligands to the protein binding sites prominently61. Here, the number of hydrogen bonds is directly proportional to the binding affinity and stability of the system, where the higher the number of hydrogen bonds formed between protein and ligand, the better the binding affinity of the corresponding ligand62. The ligands JGL023, JGL032, and JGL047 formed consistent and around two hydrogen bonds throughout the simulation period. Comparably, the standard CADA, which showed consistent hydrogen bonds during only the first 50 ns simulation period, suggests that these ligands may have better binding affinity than the CADA. The ligand JGL024, which showed less frequent hydrogen bonds, suggests it might have slightly less favorable interactions and less binding affinity than other ligands. Here, the hydrogen bond analysis gave the impetus of only the number of hydrogen bonds formed on the time scale of MD simulation. In order to get further insights into the involvement of critical residues that predominantly participated in the hydrogen bond formation, a contact frequency analysis using the module MDciao was performed. All the ligands showed hydrogen bond interaction with the residue Asn295 with varied degrees of contact frequency. Notably, the ligand CADA, JGL023, JGL024, and JGL047 establishes more than or around 75% contact frequency with the residue Asn295, while JGL032 establish around 50% contact frequency. The contact frequency of around 75% for the ligand CADA with the residue Met60 is not seen with other ligands.
The ligands JGL023, JGL024, and JGL032 additionally show a contact frequency of around 75% with the residue Glu122. The contact frequency with the residue Thr81 is identified as a key interaction having more than 75% contact frequency in the complexes with the ligands JGL032 and JGL047. These results highlight that the hydrogen bond with the residues Asn295, Gln122, and Thr81 is critically important. The trajectories isolated at different time intervals further supported that most of the trajectories have key hydrogen bond interactions with the residues Asn295, Gln122, and Thr81. Additionally, the interactions with the residues Met60 in the complex with CADA is notable.
Principal component analysis
The PCA-based Gibb’s free energy analysis showed that the complex of Sec61 with CADA had the lowest energy conformations in the unique energy basin, occupying the reaction coordinates between 0 and 5 on PC1 and − 1 to 2.5 on PC2 (Fig. 13). The representative lowest energy conformation of Sec61 with bound CADA showed the interaction with Asn295. In comparison, the complex of Sec61 protein with JGL023 showed two lowest energy conformations where the more significant number of conformations occupy the energy basin between − 12 to -6 on PC1 and − 2.5 to 2.5 on PC2. The representative lowest energy conformation from this energy basin showed that JGL023 formed a hydrogen bond with Asn295, identified as the residue with the highest contact frequency. In contrast, the smaller ones occupy the energy basin between 5 and 8 on PC1 and 4 to 6 on PC2. The representative conformations from these energy basins showed that the ligand JGL023 occupied the same binding site with a significant conformational change in the loop regions of Sec61. Interestingly, this energy basin’s representative lowest energy conformation showed that JGL023 also formed a hydrogen bond with Asn295. In the case of the complex with JGL024, the lowest energy conformations were observed in the energy basin, occupying between − 2 and 4 on PC1 and − 2 to 2 on PC2. This energy basin’s representative lowest energy conformation showed that JGL024 formed a hydrogen bond only with the residue Asn295. However, the hydrogen bond with Gln122 was not observed which is shown as the residue with the highest contact frequency. The complex of Sec61 protein with JGL032 showed a unique lowest energy basin between − 6 to -2 on PC1 and − 6 to -2.5 on PC2 (Fig. 14). The representative lowest energy conformation from this energy basin showed that JGL032 formed a hydrogen bond with the residue Asn295, and no hydrogen bond was observed with the residue Thr81. In the case of the complex with JGL047, two energy basins, where the larger one occupied the coordinates − 6 to -4 on PC1 and − 4 to 0 on PC2, and the smaller one occupied the coordinates between 0 and 2.5 on PC1 and 3 to 5 on PC2 (Fig. 14). The conformations from these energy basins differ in the loop regions while the ligand is bound at the same binding site. The representative lowest energy conformation from both the energy basins showed JGL047 forming a hydrogen bond with the residue Thr81, which was also identified as the residue with the highest contact frequency.
Gibb’s free energy landscape analysis. Sec61 protein complex with A) CADA, B) JGL023, and C) JGL024. (The lowest energy conformations of the Sec61 protein complex with respective bound ligands are shown for larger energy basins. The bond conformations of ligands are shown in red stick representations, while the gray stick representations show the stable conformations from contact frequency analysis).
Gibb’s free energy landscape analysis. Sec61 protein complex with A) JGL032, and B) JGL047 (The bond conformations of ligands are shown in red stick representations, while the gray stick representations show the stable conformations from contact frequency analysis).
The PCA analysis using the covariance matrix constructed for the protein backbone atoms in the protein-ligand complexes gives important insights into the path of motions in proteins from substantially large data of MD simulation trajectories63. The lowest energy conformations could be further identified using the selected principal components as the reaction coordinates through Gibb’s free energy analysis64. The lowest energy conformations for the complex of Sec61 protein with CADA, JGL024, and JGL032 occupied a single large energy basin, suggesting a unique protein-ligand complex conformation. At the same time, the complexes with JGL023 and JGL047 occupy the two lowest energy basins. The conformations in these two energy basins differ majorly in the loop regions of the Sec61 protein. It is imperative that the protein-ligand complex cross the energy barrier to stabilize the loop regions and adopt the lowest energy conformations. The results of PCA analysis, especially Gibb’s free energy landscape analysis, point out that the lowest energy conformations almost replicate the lowest energy poses of respective ligands to form hydrogen bonds with the residues having the highest contact frequencies. The residues Asn295 and Thr81 have the most influencial role in stabilizing the motions in the Sec61 dynamics and stabilizing the resulting protein-ligand complexes.
MM-GBSA calculations
The results of MM-GBSA calculations, where the entropic energies are taken into account to calculate the ΔGbinding, are shown in Table 8. The trajectories from 50 to 100 ns at each 100 ps time step were selected for the MM-GBSA calculations as the protein-ligand complexes were reasonably stabilized during this MD period. The ligand JGL023 also showed a comparably favorable ΔGbinding of -50.62 kcal/mol. The other ligands, JGL024 and JGL047, showed ΔGbinding -42.81, and − 41.65 kcal/mol, respectively. Notably, the ligands CADA and JGL047 have shown high entropic and van der Waals energies compared to the other ligands, resulting in slightly higher ΔGbinding. The ligand JGL032 had lower entropic energy and the lowest van der Waals and electrostatic energies than all other ligands, resulting in the lowest ΔGbinding of -55.95 kcal/mol. The plots of framewise total binding energy (ΔTOTAL) are shown in Fig. 15.
MM-GBSA total binding energy (ΔTOTAL) for frames extracted from 50 to 100 ns simulation period at each 100 ps time step. Sec61 protein complex with A) CADA, B) JGL023, C) JGL024, D) JGL032, and E) JGL047.
The overall binding affinity of the ligands is calculated through the MM-GBSA calculations, where the energetics between the ligand, protein, and protein-ligand complex are separately calculated for the energies such as van der Waals, electrostatic, polar solvation, and solvent-accessible surface areas to estimate the total binding energy (ΔTOTAL) as relative binding free energy without taking into account the entropic energy (-TΔS) and binding free energy (ΔGbinding) by taking entropic energy into account65. The entropic energy estimates for CADA and JGL047 were substantially higher, indicating significant conformational changes in the system compared to the complexes with other ligands. These higher entropic energies affect the relative binding energy and the ΔGbinding significantly higher than those with other ligands. The ligands JGL032 and JGL023 emerged as the most favorable regarding the lowest relative binding free energy and ΔGbinding of -55.95 and − 50.62 kcal/mol, respectively. The van der Waal’s and electrostatic energies also favorably contributed to these ligands’ lowest ΔGbinding.
The results of MD simulations identified the ligands JGL023 and JGL032 as having the most favorable binding affinities and Sec61 protein stabilizing propensities compared to CADA and other studied ligands.
Synthesis of the CADA analogs
To advance the 4 proposed CADA analogs as potential drug candidates, a step-by-step synthesis plan is required for further studies. Supplementary Scheme 1 outlines a proposed synthetic route for one of these analogs, specifically JGL047, which is considered a non-oral drug candidate. This route is based on the synthesis of unsymmetrical CADA compounds47 but has been modified to accommodate the structure of JGL047. The synthetic routes proposed for the remaining three analogs: JGL023, JGL024, and JGL032, were based on the established synthesis of pyridine-fused CADA derivatives13. The proposed synthetic route for JGL024 involves the formation of hydrazine intermediates, which is done to allow the synthesis of the tail group as outlined in Supplementary Scheme 2. Finally, the synthetic pathways for JGL023 and JGL032, outlined at Supplementary Scheme 3, begins similarly and are largely analogous to one another, as both share similar piperidine tails and only differs primarily in the side arm groups introduced.
Conclusion
CADA is known to have a unique antiviral mechanism that prevents the entry of HIV to CD4 cells by preventing the expression of the CD4 receptor. This macrocyle has an HIV activity against an extensive variety of HIV strains unlike other antiretrovirals. This is because all HIV viruses need to bind to the CD4 receptor to cause infection. Furthermore, CADA has shown a synergistic effect with clinically approved anti-HIV drugs8. By inhibiting the CD4 receptor expression, CADA associated therapies would theoretically be advantageous than antiretroviral therapy, which does not completely eradicate HIV in the cells and is susceptible to antiviral resistance development. This study highlights the successful design of CADA analogs with enhanced Sec61 binding and improved bioavailability using tools like SeeSAR, AutoDock Vina 1.2.4 with Meeko, etc. A total of 113 analogs were designed and predicted of their molecular docking scores, toxicity, and bioavailability. The top binding analogs were analyzed for structure-binding energy trends, revealing several derivative-specific favorable and unfavorable modifications. Through the integration of molecular docking, and toxicity and bioavailability predictions, analogs JGL023, JGL024, JGL032, and JGL047 were identified as promising drug candidates, each demonstrating favorable docking, toxicity, and bioavailability profiles. JGL023, JGL024, and JGL032 are predicted to be orally bioavailable while JGL047 is predicted to be non-orally bioavailable. Molecular dynamics simulations further confirmed that all 4 drug candidates exhibited a more negative binding energy than CADA, with JGL023 and JGL032 exhibiting the most negative binding energy and Sec61 protein stabilizing propensities. Interestingly, the initial molecular docking scores from AutoDock Vina indicate that JGL023 and JGL032 had the more positive binding energy. This discrepancy may be attributed to molecular dynamics simulations accounting for dynamic interactions and fluctuations between the ligand and Sec61 protein. This is supported by the RMSD analysis, which indicated moderate stability and minimal deviations from the initial docked poses in Sec61. This research highlights the potential of these analogs to advance HIV/AIDS treatment. Synthetic routes for the 4 analogs are developed, with future work focusing on further optimizing these analogs, validating their efficacy in clinical trials, and exploring their broader therapeutic potential.
Data availability
Data is provided within the manuscript or supplementary information files.
Change history
24 January 2025
The original online version of this Article was revised: In the original version of this Article, the author name Jay Gabriel B. Larga was incorrectly indexed. The original Article has been corrected.
References
Jameson, J. et al. Human immunodeficiency virus disease: AIDS and related disorders. In: Harrison’s Principles of Internal Medicine. 20th ed. The McGraw-Hill Companies. (2018).
Chomont, N. et al. HIV reservoir size and persistence are driven by T cell survival and homeostatic proliferation. Nat. Med. 15 (8), 893–900 (2009).
Yang, L. et al. HeLa cells apoptosis induced by 1,7-dimethyl-1,4,7,10-tetraazacyclododecane. 17(6),1818–1822. (2007).
Bell, T. W., Choi, H. J., Harte, W. & Drew, M. G. B. Syntheses, conformations and basicities of bicyclic triamines. J. Am. Chem. Soc. 125, 12196–12210 (2003).
Itskanov, S. et al. A common mechanism of Sect. 61 translocon inhibition by small molecules (Nat Chem Biol, 2023).
Bell, T. W. et al. Synthesis and structure – activity relationship studies of CD4 down-modulating Cyclotriazadisulfonamide (CADA). Analogues. 49 (4), 1291–1312 (2006).
Hunt, G. M. et al. Prevalence of HIV-1 drug resistance amongst newly diagnosed HIV-infected infants age 4–8 weeks, enrolled in three nationally representative PMTCT effectiveness surveys, South Africa: 2010, 2011-12 and 2012-13. BMC Infect. Dis. 19 (Suppl 1), 787 (2019).
Vermeire, K. & Schols, D. Cyclotriazadisulfonamides: promising new CD4-targeted anti-HIV drugs. J. Antimicrob. Chemother. 56 (2), 270–272. https://doi.org/10.1093/jac/dki208 (2005).
Chang, Z. Biogenesis of Secretory Proteins, Encyclopedia of Cell Biology, Academic Press. 535–544. ISBN 9780123947963. (2016).
Bousfield, G. R. Biosynthesis and Posttranslational Processing of Peptide Hormones. Reference Module in Biomedical Sciences (Elsevier, 2019).
Pfeffer, S., Dudek, J., Zimmermann, R. & Förster, F. Organization of the native ribosome-translocon complex at the mammalian endoplasmic reticulum membrane. Biochim. Biophys. Acta. 1860 (10), 2122–2129 (2016).
Lumangtad, L. A. et al. Syntheses and Anti-HIV and Human Cluster of Differentiation 4 (CD4) Down-Modulating Potencies of Pyridine-Fused Cyclotriazadisulfonamide (CADA) Compounds (Bioorganic & Medicinal Chemistry, 2020).
Lumangtad, L. A. Pyridine-fused Cyclotriazadisulfonamide (CADA) Compounds: Synthesis and CD4 Down-Modulation Potency. [Doctoral Dissertation, University of Nevada, Reno]. ScholarWorks. (2018).
Villar, E. A. et al. How proteins bind macrocycles. Nat. Chem. Biol. 10 (9), 723–731 (2014).
Bhardwaj, V., Singh, R., Singh, P., Purohit, R. & Kumar, S. Elimination of bitter-off taste of stevioside through structure modification and computational interventions. J. Theor. Biol. 486, 110094. https://doi.org/10.1016/j.jtbi.2019.110094 (2020). Epub 2019 Nov 26.
Gupta, A. & Purohit, R. Identification of potent BRD4-BD1 inhibitors using classical and steered molecular dynamics based free energy analysis. J. Cell. Biochem. 125 (3). https://doi.org/10.1002/jcb.30532 (2024). Epub 2024 Feb 5.
Rajasekaran, R. et al. Effect of deleterious nsSNP on the HER2 receptor based on stability and binding affinity with herceptin: a computational approach. C R Biol. 331 (6), 409–417 (2008). Epub 2008 Apr 24.
Kamaraj, B., Rajendran, V., Sethumadhavan, R., Kumar, C. V. & Purohit, R. Mutational analysis of FUS gene and its structural and functional role in amyotrophic lateral sclerosis 6. J. Biomol. Struct. Dyn. 33 (4), 834–844 (2015). Epub 2014 May 14.
Singh, R., Bhardwaj, V. K., Sharma, J., Das, P. & Purohit, R. Identification of selective cyclin-dependent kinase 2 inhibitor from the library of pyrrolone-fused benzosuberene compounds: an in silico exploration. J. Biomol. Struct. Dyn. 40 (17), 7693–7701 (2022). Epub 2021 Mar 22.
Singh, R., Manna, S., Nandanwar, H. & Purohit, R. Bioactives from medicinal herb against bedaquiline resistant tuberculosis: removing the dark clouds from the horizon. Microbes Infect. 26 (3), 105279. https://doi.org/10.1016/j.micinf.2023.105279 (2024 Mar-Apr). Epub 2023 Dec 19.
Rajendran, P., Rathinasabapathy, R., Kishore, C., Bellucci, S. & S., & Computational-Simulation-based behavioral analysis of Chemical compounds. J. Compos. Sci. 7, 196. https://doi.org/10.3390/jcs7050196 (2023).
Lewis-Atwell, T., Townsend, P. A. & Grayson, M. N. Comparisons of different force fields in conformational analysis and searching of organic molecules. Rev. Tetrahedron. 79, 131865 (2021).
Eberhardt, J., Santos-Martins, D., Tillack, A. F. & Forli, S. AutoDock Vina 1.2.0: new docking methods, expanded force field, and Python Bindings. J. Chem. Inf. Model. 61 (8), 3891–3898. https://doi.org/10.1021/acs.jcim.1c00203 (2021).
Nguyen, N. T. et al. Autodock Vina adopts more accurate binding poses but Autodock4 forms better binding affinity. J. Chem. Inf. Model. 60 (1), 204–211. https://doi.org/10.1021/acs.jcim.9b00778 (2019).
Gaillard, T. Evaluation of AutoDock and AutoDock Vina on the CASF-2013 Benchmark. J. Chem. Inf. Model. 58 (8), 1697–1706. https://doi.org/10.1021/acs.jcim.8b00312 (2018).
Parks, C. D. et al. D3R grand challenge 4: blind prediction of protein-ligand poses, affinity rankings, and relative binding free energies. J. Comput. Aided Mol. Des. 34 (2), 99–119. https://doi.org/10.1007/s10822-020-00289-y (2020).
Singh, R., Bhardwaj, V. K., Das, P. & Purohit, R. A computational approach for rational discovery of inhibitors for non-structural protein 1 of SARS-CoV-2. Comput. Biol. Med. https://doi.org/10.1016/j.compbiomed.2021.104555 (2021). Epub 2021 Jun 8. PMID: 34144270; PMCID: PMC8184359.
Borba, J. V. B. et al. STopTox: an in-Silico Alternative to Animal Testing for Acute systemic and topical toxicity. Environ. Health Perspect., 130(2). (2022).
Abraham, M. J. et al. High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX. GROMACS, 19–25. https://doi.org/10.1016/j.softx.2015.06.001 (2015).
Berendsen, H. J. C., van der Spoel, D., van Drunen, R. & GROMACS A message-passing parallel molecular dynamics implementation. Comput. Phys. Commun.91, 43–56. https://doi.org/10.1016/0010-4655(95)00042-E (1995).
Sousa Da Silva, A. W. & Vranken, W. F. ACPYPE - AnteChamber PYthon Parser interfacE. BMC Res. Notes. 5, 367. https://doi.org/10.1186/1756-0500-5-367 (2012).
Wang, J., Wang, W., Kollman, P. A. & Case, D. A. Automatic atom type and bond type perception in molecular mechanical calculations. J. Mol. Graph. Model. 25, 247–260. https://doi.org/10.1016/j.jmgm.2005.12.005 (2006).
Lindorff-Larsen, K. et al. Improved side‐chain torsion potentials for the Amber ff99SB protein force field. Proteins. 78, 1950–1958. https://doi.org/10.1002/prot.22711 (2010).
Jorgensen, W. L. & Madura, J. D. Solvation and conformation of methanol in water. J. Am. Chem. Soc. 105, 1407–1413. https://doi.org/10.1021/ja00344a001 (1983).
Bussi, G., Donadio, D. & Parrinello, M. Canonical sampling through velocity rescaling. J. Chem. Phys. 126, 014101. https://doi.org/10.1063/1.2408420 (2007).
Berendsen, H. J. C., Postma, J. P. M., van Gunsteren, W. F., DiNola, A. & Haak, J. R. Molecular dynamics with coupling to an external bath. J. Chem. Phys. 81, 3684–3690. https://doi.org/10.1063/1.448118 (1984).
Parrinello, M. & Rahman, A. Polymorphic transitions in single crystals: a new molecular dynamics method. J. Appl. Phys. 52, 7182–7190. https://doi.org/10.1063/1.328693 (1981).
Hess, B., Bekker, H., Berendsen, H. J. C. & Fraaije, J. G. E. M. LINCS: a linear constraint solver for molecular simulations. J. Comput. Chem. 18, 1463–1472. (1997).
Petersen, H. G. Accuracy and efficiency of the particle mesh Ewald method. J. Chem. Phys. 103, 3668–3679. https://doi.org/10.1063/1.470043 (1995).
Pérez-Hernández, G. & Hildebrand, P. W. mdciao: Accessible Analysis and Visualization of Molecular Dynamics Simulation Data. Bioinformatics; doi: (2022). Jul https://doi.org/10.1101/2022.07.15.500163
Sittel, F., Jain, A. & Stock, G. Principal component analysis of molecular dynamics: on the use of cartesian vs. internal coordinates. J. Chem. Phys. 141, 014111. https://doi.org/10.1063/1.4885338 (2014).
Maisuradze, G. G., Liwo, A. & Scheraga, H. A. Relation between Free Energy landscapes of proteins and dynamics. J. Chem. Theory Comput. 6, 583–595. https://doi.org/10.1021/ct9005745 (2010).
Valdés-Tresanco, M. S., Valdés-Tresanco, M. E., Valiente, P. A., & Moreno, E. gmx_MMPBSA: A New Tool to perform end-state Free Energy calculations with GROMACS. J. Chem. Theory Comput. 17, 6281–6291. https://doi.org/10.1021/acs.jctc.1c00645 (2021).
DeLano, W. L. et al. DeLano, W.L. (2002). J.C.N.o.p.c. DeLano. Pymol: an open-source molecular graphics tool. CCP4 Newsletter Pro Crystallogr.
Turner, P. & XMGRACE Version 5.1. 19. Center for Coastal and Land-Margin Research2 (Oregon Graduate Institute of Science and Technology, 2005).
Chawla, R. et al. Tuning side Arm Electronics in Unsymmetrical Cyclotriazadisulfonamide (CADA) endoplasmic reticulum (ER) translocation inhibitors to improve their human cluster of differentiation 4 (CD 4) receptor down-modulating potencies. J. Med. Chem. 59 (6), 2633–2647 (2016).
Demillo, V. G. et al. Unsymmetrical cyclotriazadisulfonamide (CADA) compounds as human CD4 receptor down-modulating agents. J. Med. Chem. 54 (16), 5712–5721 (2011).
Jin, Q. Synthesis of CADA Analogs as Potential Antiviral Agents. Ph.D. thesis, University of Nevada, Reno, NV. (1997).
Islam, M. B. et al. Recent Advances in Pyridine Scaffold: Focus on Chemistry, Synthesis, and Antibacterial Activities. BioMed research international, 2023, 9967591. (2023). https://doi.org/10.1155/2023/9967591
St. Jean, D. J. Jr & Fotsch, C. Mitigating heterocycle metabolism in drug discovery. J. Med. Chem. 55 (13), 6002–6020. https://doi.org/10.1021/jm300343m (2012).
Filimonov, D. A. et al. Prediction of the Biological Activity Spectra of Organic compounds using the pass online web resource. Chem. Heterocycl. Comp. 50, 444–457. https://doi.org/10.1007/s10593-014-1496-1 (2014).
Magala, P., Klevit, R. E., Thomas, W. E., Sokurenko, E. V. & Stenkamp, R. E. RMSD analysis of structures of the bacterial protein FimH identifies five conformations of its lectin ___domain. Proteins. 88, 593–603. https://doi.org/10.1002/prot.25840 (2020).
Pavan, M., Menin, S., Bassani, D., Sturlese, M. & Moro, S. Qualitative estimation of protein–ligand Complex Stability through Thermal Titration Molecular Dynamics simulations. J. Chem. Inf. Model. 62, 5715–5728. https://doi.org/10.1021/acs.jcim.2c00995 (2022).
Rampogu, S., Lee, G., Park, J. S., Lee, K. W. & Kim, M. O. Molecular Docking and Molecular Dynamics simulations Discover Curcumin Analogue as a plausible dual inhibitor for SARS-CoV-2. IJMS. 23, 1771. https://doi.org/10.3390/ijms23031771 (2022).
Martínez, L. Automatic Identification of Mobile and Rigid Substructures in Molecular Dynamics Simulations and Fractional Structural Fluctuation Analysis. Kleinjung J, editor. PLoS ONE. ;10: e0119264. doi: (2015). https://doi.org/10.1371/journal.pone.0119264
Arnittali, M., Rissanou, A. N., Amprazi, M., Kokkinidis, M. & Harmandaris, V. Structure and Thermal Stability of wtRop and RM6 proteins through all-Atom Molecular Dynamics Simulations and experiments. IJMS. 22, 5931. https://doi.org/10.3390/ijms22115931 (2021).
Lobanov MYu, Bogatyreva, N. S. & Galzitskaya, O. V. Radius of gyration as an indicator of protein structure compactness. Mol. Biol. 42, 623–628. https://doi.org/10.1134/S0026893308040195 (2008).
Noguchi, H. & Yoshikawa, K. Morphological variation in a collapsed single homopolymer chain. J. Chem. Phys. 109, 5070–5077. https://doi.org/10.1063/1.477121 (1998).
Blache, D., Gautier, T., Tietge, U. J. F. & Lagrost, L. Activated platelets contribute to oxidized low-density lipoproteins and dysfunctional high‐density lipoproteins through a phospholipase A2‐dependent mechanism. FASEB j. 26, 927–937. https://doi.org/10.1096/fj.11-191593 (2012).
Durham, E., Dorr, B., Woetzel, N., Staritzbichler, R. & Meiler, J. Solvent accessible surface area approximations for rapid and accurate protein structure prediction. J. Mol. Model. 15, 1093–1108. https://doi.org/10.1007/s00894-009-0454-9 (2009).
Bissantz, C., Kuhn, B. & Stahl, M. A Medicinal Chemist’s guide to molecular interactions. J. Med. Chem. 53, 5061–5084. https://doi.org/10.1021/jm100112j (2010).
Majewski, M., Ruiz-Carmona, S. & Barril, X. An investigation of structural stability in protein-ligand complexes reveals the balance between order and disorder. Commun. Chem. 2, 110. https://doi.org/10.1038/s42004-019-0205-5 (2019).
David, C. C. & Jacobs, D. J. Principal component analysis: a method for determining the Essential Dynamics of Proteins. In: (ed Livesay, D. R.) Protein Dynamics. Totowa, NJ: Humana; 193–226. doi:https://doi.org/10.1007/978-1-62703-658-0_11 (2014).
Dalal, V. et al. Structure-Based Identification of Potential Drugs Against FmtA of Staphylococcus aureus: Virtual Screening, MM-GBSA QM/MM Protein J. ;40: 148–165. doi:https://doi.org/10.1007/s10930-020-09953-6 (2021).
Genheden, S. & Ryde, U. The MM/PBSA and MM/GBSA methods to estimate ligand-binding affinities. Expert Opin. Drug Discov. 10, 449–461. https://doi.org/10.1517/17460441.2015.1032936 (2015).
Author information
Authors and Affiliations
Contributions
J.G.B.L. and W.T.M. conceptualized the study, acquired and curated the data, and contributed to the methodology. A.T.M. administered the project, contributed to the methodology, and generated the figures. J.G.B.L., W.T.M., A.T.M., M.S.J., M.S.N., M.A.A.M., A.A., A.P., and M.S.Y. were involved in writing the original manuscript and in the review and editing process. R.M.B. and R.B.P. contributed to the methodology, generated figures, and participated in writing, reviewing, and editing the manuscript. M.C.B. contributed to the conceptualization, data acquisition, curation, and methodology, and provided supervision. All authors reviewed and approved the final manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Larga, J.G.B., Munabirul, W.T., Moin, A.T. et al. Cutting-edge Bioinformatics strategies for synthesizing Cyclotriazadisulfonamide (CADA) analogs in next-Generation HIV therapies. Sci Rep 14, 29764 (2024). https://doi.org/10.1038/s41598-024-77106-1
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-024-77106-1
Keywords
This article is cited by
-
Ginsenoside Rg1 alleviates chronic stress-induced depression in rats by targeting Cx43-YAP axis
Acta Pharmacologica Sinica (2025)
