A structure-based virtual screening study to identify potential inhibitors targeting the TKB domain of CBL-B

doi:10.21203/rs.3.rs-5160369/v1

Download PDF

Research Article

A structure-based virtual screening study to identify potential inhibitors targeting the TKB domain of CBL-B

https://doi.org/10.21203/rs.3.rs-5160369/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

CBL-B is an E3-ubiquitin ligase that serves a key role in modulating immune response by negatively inhibiting effector T-cell activation. Inhibitors of CBL-B would be an effective way for immune activation, making it a promising target for cancer immunotherapy and other related immune diseases. In this study, we sought to identify potential inhibitors of CBL-B through structure-based virtual screening, molecular docking, ADMET analysis, molecular dynamics simulation, and MM/GBSA calculations. A diverse set of compounds was screened against CBL-B using molecular docking augmented by a machine-learning scoring function (RF-score). The top leads were further evaluated for their ADMET properties resulting in three compounds that were subjected to a 100 ns MD simulation. MD results revealed the stability of protein-ligand complexes. Finally, MM/GBSA calculations resulted in higher free binding energy for the three compounds compared to the cocrystallized ligand, implying better affinity towards CBL-B and suggesting the use of these compounds as novel inhibitors of CBL-B.

The E3 ubiquitin-protein ligase and adaptor protein Casitas B-lineage lymphoma proto-oncogene-b (Cbl-b) is a member of the Cbl family, which also includes c-Cbl and Cbl-c, and it serves a variety of biological purposes. A key function of Cbl-b is to control the activity of effector T cells (Lutz-Nicoladoni, Wolf and Sopper 2015). It has been determined that, in the absence of CD28 co-stimulation, Cbl-b is a crucial inhibitor of T cell activation. Cbl-b increases immunological tolerance throughout innate and adaptive immunity by inhibiting T cell transcriptional activity through a complex interplay of signal transducers (Chiang, Kole et al. 2000, Liu, Zhou et al. 2014). Regardless of upstream checkpoint signaling, Cbl-b inhibition may offer a more focused and effective path toward widespread immune activation. To create clinical therapeutics that block Cbl-b, several groups have explored innovative platforms. Pharmaceutical strategies include siRNA knockdown, CRISPR genome editing in conjunction with adoptive cell treatment, and DNA encoded libraries and machine learning algorithms (Augustin, Bao and Luke 2023). NX-1607, a small molecule developed by Nurix that has been shown to significantly reduce tumor growth in triple negative breast and colon cancer in murine models, has started phase I clinical trials in advanced malignancies (Rountree, Cohen et al. 2021).

Cbl-b structure is composed of an N-terminal half encompassing its tyrosine-kinase-binding (TKB) domain, a linker helix region (LHR), really interesting new gene (RING) finger domain and a C-terminal region. The N-terminal is highly conserved among Cbl protein family, and the ability of Cbl protein to act as E3 ubiquitin ligase is largely dependent on its domains. The C-terminal region is less well conserved and is able to interact with SH2 (Src homology 2 ) and SH3 domain containing proteins (Thien and Langdon 2005). The TKB domain, in particular, determines the Cbl’s substrate specificity by interacting with certain phosphorylated tyrosine residues on proteins that would be deubiquitinated. It is composed of three interacting subdomains; a four-helix bundle, a calcium-binding EF hand and a variant SH-2 domain(Meng, Sawasdikosol et al. 1999). It has been revealed that compounds binding to the TKB domain induce a closed conformational state as observed in the co-crystal structure (PDB code: 8GCY) leading to Cbl-b inactivation while stabilizing it (Fig. 1) (Kimani, Perveen et al. 2023).

In this study, a structure based virtual screening approach was conducted to identify potential inhibitors of Cbl-b by targeting its TKB domain in its closed/inactive state. Molecular docking, molecular dynamics simulation, and free binding energy calculations were used to examine the binding ability of the lead candidates and probe for the conformational changes in the target protein upon inhibitor binding.

2.1. Target preparation

The 3D structure of CBL-b in complex with an N-Aryl isoindolin-1-one inhibitor was retrieved from the protein data bank (https://www.rcsb.org/). The structure (PDB ID: 8GCY), composed of one chain, has a resolution of 1.8 Å and contains 376 residues. Protein was prepared by removing water molecules, impurities and the bound ligand from the structure using PyMOL (https://pymol.org/2/). Missing atoms and hydrogen atoms appropriate for pH 7.4 were added using PDBFixer tool (Eastman, Swails et al. 2017).

2.2. Ligand preparation

Discovery diversity set (DDS-50) comprising 50240 compounds was downloaded from Enamine database in 2D SDF format. Ligands were prepared by python script “RDKitGenerateConformers.py” from MayaChemTools (Sud 2016). One 3D conformer was generated for each compound, salts were removed, hydrogen atoms were added, and energy was minimized using MMFF94s forcefield resulting in 49930 valid compounds that were subjected to virtual screening.

2.3. Virtual Screening

Smina, a fork of autodock vina that focuses on improving minimization and scoring was used to perform molecular docking (Koes, Baumgartner and Camacho 2013). It has several built-in scoring functions that can estimate the protein-ligand binding affinity. Docking procedure was validated by re-docking the co-crystallized ligand into the protein active site using the different scoring functions. The RMSD value between the experimental and the redocked poses was calculated and the scoring function that resulted in the lowest RMSD was used for docking experiment. Subsequently, prepared ligands were docked into CBL-b in the place of the bound inhibitor occupying the TKB domain. The generated poses were ranked based on the docking scores and the top hits were rescored using RF-Score-Vs, a random forest scoring function that predicts binding affinity (Wojcikowski, Ballester and Siedlecki 2017). The predicted affinities are expressed in pK units, which means the higher the score, the greater the ligand’s affinity to the target protein.

2.4. Protein-Ligand interaction analysis

Protein-Ligand Interaction Profiler (PLIP) server (https://plip-tool.biotec.tu-dresden.de/plip-web/plip/index) was employed to analyze the interactions between the top lead compounds and the target protein (Adasme, Linnemann et al. 2021). It can detect various interaction types including hydrogen bonds, hydrophobic contacts, etc. The binding modes of the ligands in the target active site were visualized using PyMOL.

2.5. ADME-T analysis

ADMETlab 2.0 server (https://admetmesh.scbdd.com) was utilized to evaluate the ADME-T profile of the lead compounds. Ligands were submitted in SMILES format and several physiochemical properties, medicinal chemistry measures and ADME-T endpoints were calculated.

2.6. Molecular dynamics (MD) simulations

Apo protein and protein-ligand complexes were subjected to MD simulations using Gromacs v2020.4. Initially, protein topology was defined by CHARMM36 forcefield (Best, Zhu et al. 2012) and ligands’ parameters were generated by SwissParam server (Bugnon, Goullieux et al. 2023). A cubic box with TIP3P water model was used to enclose and solvate the systems which were neutralized by adding counterions. Particle Mesh Ewald scheme and LINCS algorithm were utilized for calculating long range electrostatic interactions and bond constraining respectively (Darden, York and Pedersen 1993, Hess 2008). Systems were then equilibrated at NVT ensemble (300°K) for 1 ns using Berendsen thermostat and for 2 ns at NPT ensemble (1 bar) utilizing Parrinello-Rahman barostat. Finally, MD simulations of 100 ns were executed with 2 fs integration time step. Simulation trajectories were analyzed using backbone root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (Rg) and number of hydrogen bonds employing Gromacs built-in tools to assess systems’ conformational stability.

2.7. Binding free energy calculations using MM-GBSA

Molecular mechanics-generalized born surface approach (MM-GBSA) was employed to evaluate the binding affinity of selected inhibitors using gmx_MMPBSA tool (Valdes-Tresanco, Valdes-Tresanco et al. 2021). The analysis was performed for the last 10 ns frames of the MD simulation trajectory with salt concentration maintained at 0.15 M and generalized born method (igb = 5). Free binding energy (DG _bind) was computed using the following equation:

DG _{bind =} DG _COM - DG _REC - DG _LIG

where terms to the right in the equation describe the Gibbs free energies of the complex, protein, and ligand respectively. Additionally, per-residue decomposition analysis was carried out to assess the contribution of individual residues within 5 Å of the bound ligand.

3.1. Virtual screening

The present study aimed to identify compounds that could potentially inhibit Cbl-b protein by targeting its TKB domain using a structured based virtual screening approach. The structure of the target protein with a bound inhibitor was obtained and prepared as described. Prior to molecular docking, the ligand was redocked to the active site using built-in scoring functions of smina software. It was found that the redocked inhibitor, with vinardo scoring function, best reproduced the experimental binding pose with a docking score of -12.1 kcal/mol. The RMSD value between the original and redocked poses was 0.17 Å (<2) indicating the reliability of the docking procedure and the ability to use smina to perform molecular docking experiments. The alignment between the co-crystallized and redocked poses is shown in Fig. S1. A structurally diverse library of recently synthesized compounds was retrieved and prepared for molecular docking experiment. They have the following properties: 191 < molecular weight < 482, -4.4 < clogP < 4.9, 0 < HBA < 9 and 0 < HBD < 5, indicating their compliance with drug likeness rule (Lipinski Rule of 5). 49930 ligands were docked into the target active site and the generated poses were ranked based on their estimated binding energies. The docking scores ranged between -3.24 and -10.19 kcal/mol indicating that none of the docked compounds had better affinity than the original inhibitor (From here will be termed “Z3N”).

Scoring functions are used by molecular docking algorithms to evaluate the binding affinity of poses generated by the searching algorithm (Li, Fu and Zhang 2019). These “approximate” scoring functions don’t usually correlate well with the experimental binding affinities, increasing the likelihood of false-positives and false negatives appearing in the top ranked list (Rastelli and Pinzi 2019). To improve the accuracy of our virtual screening approach, by discriminating between binders and non-binders with high precision i.e., filter out false positives, a machine learning based scoring function (RF-Score-VS 1.0) was employed to re-score the top 1% docked poses (500 compounds, docking score < -8.61 kcal/mol). This scoring function was shown to outperform classical scoring functions such as X-score and Autodock vina (Li, Leung et al. 2015, Li, Lu et al. 2021) and has been employed successfully in other virtual screening studies (Ramesh, Shin and Veerappapillai 2021, Ramesh and Veerappapillai 2022, Krishnamoorthy and Karuppasamy 2023, Phengsakun, Boonyarit et al. 2023). 6 compounds had higher RF-Score than the reference inhibitor (6.34). Molecular docking and rescoring results together with the structures of the lead 6 compounds are depicted in Table 1. This indicated that both scoring functions complemented each other and provided more reliable screening results.

Table 1 Molecular Docking and rescoring results.

3.2. Protein-Ligand interaction analysis

To better understand the factors that would contribute to the binding of the potential inhibitors, their interactions with the targeted protein were analyzed and compared to the reference ligand. Fig. S2 shows the alignment of binding poses of the reference and lead compounds in the protein active site. Protein-ligand interactions are provided in Table S1 and represented in Fig. 2. The co-crystallized ligand Z3N, an N-aryl isoindolin-1-one inhibitor mediated three hydrogen bonds with Y224, F227 and Y312 at distances of 3.58, 3.04, and 3.97 Å respectively. Moreover, it formed hydrophobic interactions with T108, L112, T183, L186, F227 and T229. A parallel p-stacking and salt bridge were formed with Y312 and E232 respectively. Comparative analysis with the reference ligand revealed that the top hits mediated various hydrogen bonds and hydrophobic interactions with the active site residues. The polar aromatic Y312 was the most common residue involved in hydrogen bonding with the compounds (except for Z2077032497). It should be noted that the hydrogen bond donor-acceptor distances with Y312 were found to be shorter for the docked ligands, except for Z5011956847, compared to Z3N. The same was observed with F227 and Y224 suggesting a more favorable orientation inside the binding site. Z3396345178 and Z2077032497 formed the highest number of hydrogen bonds (five) which might be correlated with their high binding affinity (-9.72 and -10.06 kcal/mol respectively). The hydrophobic interactions are mediated most by F227 (except for Z5011956847) and K109 (except for Z1102107341 and Z5011956847). Z5011956847, having the lowest docking score (-8.86 kcal/mol) formed nine hydrophobic interactions and was the only compound to make a pi-stacking interaction with Y363, like Z3N. Results indicated that several residues contributed to ligands’ binding through formation of various interactions, mainly hydrogen bonding and hydrophobic interactions. The stability of these interactions needed to be further analyzed to better assess ligand binding efficiency.

3.3. ADME-T analysis

Evaluation of the absorption, distribution, metabolism, excretion, and toxicity is very important to assess the quality of a potential drug candidate. The pharmacokinetic and toxicity profiles of the top hits were determined using ADMETlab server and compared with the reference compound. Detailed results are listed in Table S2. Z3N had a QED score of 0.52, passed Lipinski rule but violated other drug likeness rules (GSK and Pfizer). In addition, it showed a Caco-2 permeability score of -5.47 which is not optimal and low HIA score indicating poor intestinal absorption. Pgp inhibition and BBB penetration scores were also high. Toxicity analysis revealed that it could potentially cause respiratory toxicity and block hERG, in addition to low probability of causing human hepatotoxicity. The QED score of the lead compounds ranged between 0.34 and 0.71 with Z5011956847 having the highest score indicating its attractiveness. All compounds passed Lipinski and Pfizer filters in addition to Z1102107341 which also satisfied GSK rules. They exhibited excellent Caco2-permeability and low HIA. Only Z3396339411 had higher HIA score than Z3N. Z5011956847 and Z2077032497 had the lowest Pgp inhibition probability, and the BBB penetration score was low with all top hits. Evaluation of potential toxicity showed that the lead compounds had high hERG blocking capabilities (except for Z5011956847). Z3396345178, Z2077032497 and Z5011956847 were predicted to be carcinogenic. Results indicated that the potential ligands had good pharmacokinetic properties, but caution must be taken because of possible side effects that would need to be monitored carefully. Based on molecular docking scores, binding interaction analysis and ADMET evaluation, the three compounds Z5011956847, Z2077032497 and were selected for further analysis.

3.4. MD simulations

MD simulations of the CBL-B protein in complex with the three lead compounds were performed alongside the reference ligand to better assess the system stability and conformational changes of CBL-B induced by potential inhibitor binding. Initially, RMSD plots of the protein backbone over the 100 ns were generated and depicted in Fig. 3A. Apo CBL-B was stable throughout the simulation run with average RMSD of 0.29 nm. CBL-B in complex with Z2077032497 had an average RMSD value of 0.38 nm with several increments above 0.4 nm indicating less stability. Other complexes exhibited similar RMSD pattern with average values ranging between 0.27 and 0.32 nm showing stability like the apo protein. RMSD of the ligands from their initial positions were also analyzed and shown in Fig. 3B. The reference ligand Z3N was stable for 50 ns with RMSD < 0.3 nm. The RMSD started to increase gradually and didn’t become stable until it reached 0.71 nm at the simulation end. Z3396339411 which had the highest docking score, exhibited major drift in RMSD at the beginning of the simulation from 0.2 to 0.8 nm. Till the end of the simulation, the ligand kept fluctuating around 0.8 nm. Z5011956847 was stable for 40 ns at 0.3 nm then fluctuates till it reaches around 0.68 nm at 50 ns. It then remained stable till the end of the run. Z2077032497 was somewhat similar to Z5011956847. It showed stability till 30 seconds at 0.3 nm after which RMSD increased till 0.5 nm at 50 ns. Stability was observed till the simulation end with an average RMSD of 0.55 nm. In general, RMSD analysis suggested that the selected ligands had no significant effect on CBL-B stability except for Z2077032497 which exhibited the highest RMSD value. The observed fluctuations in ligands’ RMSD plot indicated multiple conformations adopted by the ligands inside the binding pocket allowing possibly for different interactions with surrounding residues. This was evident from the snapshots taken for the ligands at 100 ns where different orientations from the initial positions were clearly observed (Fig. S3). Visual inspection also revealed the stability of the potential inhibitors inside the protein active site.

RMSF was computed to assess the fluctuations in CBL-B residues upon ligand binding, with lower values indicating lower flexibility and higher stability (Fig. 4A). Residues lining the binding site (within 5 Å of bound ligand) showed low RMSF values in the apo CBL-B implying high stability. High fluctuations were only observed in the far C-terminal region (residues 350-360) with RMSF reaching 0.3 nm. This is not surprising given its loop conformation. Similar RMSF pattern was exhibited in other complex systems indicating that all inhibitors did not influence the stability of the protein active site residues. Only, the region between residues 300-310 showed an increase in RMSF when bound to the reference ligand Z3N and Z2077032497. Additionally, the impact of inhibitor binding on protein compactness and folding nature was examined by measuring Rg. CBL-B in the apo state showed average Rg value of 2.24 nm indicating tightly packed structure (Fig. 4B). Similar pattern was observed in ligand-bound CBL-B with average Rg values ranging between 2.27 and 2.29 nm. In case of Z3N complex, fluctuations were observed in the last 30 ns with Rg increasing to 2.34 nm suggesting a less compact structure.

Overall, results indicated that bound inhibitors had no significant effects on protein overall stability or compactness. Notably, RMSD analysis indicated changes in ligands orientation enabling the formation of new interactions not observed in the initial docked poses.

Hydrogen bonds contribute significantly to the protein-ligand interaction, and hence the complex stability. Time evolution formation of hydrogen bonds for the studied complexes was monitored and depicted in Fig. 5. The reference compound Z3N exhibited a maximum number of 8 bonds with an average value of 3 bonds. The highest number of H-bonds was observed for Z2077032497 (10) followed by Z5011956847 (9) and Z3396339411 (8). It is worth mentioning that for 40 ns, Z5011956847 formed an average of 2 hydrogen bonds. After this, the number of hydrogen bonds increased till the simulation end. This can be attributed to the change in the compound conformation previously detected in the RMSD plot (Fig. 3B). In general, H-bond analysis indicated tight interactions of the lead compounds with the protein and hence more stability inside the binding pocket.

3.5. Post MD simulations interaction analysis

The last frames were retrieved from the 100 ns MD trajectories and the inhibitors’ interactions with CBL-B were analyzed (Fig. S4). Z3N formed one hydrophobic interaction with T229, one hydrogen bond with R105 and one salt bridge with E232. R105 interacted with fluoride atom of Z3N through halogen bonding which was not observed previously. Z3396339411 mediated one hydrophobic interaction with Y224 and three hydrogen bonds, two with R105 and one with F227. Six hydrogen bonds were established between Z5011956847 and CBL-B through R105, K109, S182, F227 and Y312. In addition, Z5011956847 formed four hydrophobic interactions with I113, A226, F227 and L228. The largest number of interactions was observed with Z2077032497 which interacted with CBL-B through five hydrogen bonds, with T229, Y230 and R250. It also formed five hydrophobic interactions with R105, T108, F227, L228 and Y230. Finally, a pi-stacking and pi-cation interactions with Y230 and R105 respectively were observed. Post MD simulation analysis indicated that all ligands exhibited interactions different to some extent from those detected after molecular docking procedure. This underlies the importance of MD simulations to examine protein-ligand interactions in a more dynamic context given the static nature of the docked poses. These observed changes might affect the affinity of the lead compounds towards CBL-B. Hence, it was crucial to estimate the binding energy of lead compounds using a more accurate way. This was performed in the next step.

3.6. Binding free energy calculations using MM-GBSA

Table 2 lists the various energy terms and the binding free energies calculated through MM-GBSA approach. Z2077032497 exhibited the highest affinity, showcasing a binding energy of -50.78 kcal/mol, followed by Z5011956847 (-42.48 kcal/mol) and Z3396339411 (-23.03 kcal/mol). In all complexes, electrostatic interactions provided the largest favorable contribution to the protein-ligand binding, followed by van der Waals forces. Notably, the binding energy of the positive control Z3N was significantly lower than other compounds, which can be attributed to the highly unfavorable polar solvation energy resulting in the overall decreased binding energy and hence lower affinity. These findings support the idea of utilizing these compounds as potential inhibitors of CBL-B. Moreover, per-residue decompositions analysis was performed to delineate the key residues contributing favorably to the overall binding free energy (Fig. 6). In general, amino acids with energy contributions greater than -1 kcal/mol can be considered as hot spot residues. For Z2077032497, R250 was the major contributor to the ligand binding with an energy of -38.1 kcal/mol followed by T229 (-16.8 kcal/mol) and Y230 (-8.4 kcal/mol). Other residues include C253, R105, L228, T108, F227, P35, K109, T254, L112 and S252 with energies ranging between -3.5 and -1.5 kcal/mol. From previous analysis, Z2077032497 formed hydrogen bonds with R250, T229 and Y230, which explains these favorable interaction energies. Concerning Z5011956847, the major contribution came from K109 with a binding energy of -32.8 kcal/mol. It was shown to mediate hydrogen bond with the ligand during MD simulation. Through VdW and electrostatic interactions, F227 contributed with an energy of -4.6 kcal/mol. Other key residues included M315, Y312, D185, L112, L186, S182, A226, M225 and I113. The interaction between Z3396339411 and CBL-B was primarily contributed by M315 and K109 with binding energies of -5.7 and -5.3 kcal/mol, respectively caused mainly by VdW forces. Notably, electrostatic interactions contributed favorably to K209-Z3396339411 binding with energy of -20 kcal/mol. However, polar solvation energy (19 kcal/mol) opposed this effect leading to a decreased binding energy. F227, A226, M225, L112, Y224 and T108 can be regarded also as hot spot residues (DG: between -2.5 and -1.2 kcal/mol). In conclusion, decomposition analysis revealed the significant role of several amino acids towards maintaining stable interactions with the lead compounds and explains the disparities in binding energies, and hence in binding affinities towards CBL-B observed among them. R250, T229, Y230, K109, F227 and M315 seemed to be the major hot spot residues, and development of CBL-B inhibitors is suggested to focus on mediating interactions with these amino acids.

Table 2 Binding free energies of CBL-B complexes computed via MM-GBSA method. Units are expressed in kcal/mol.

Inhibitor	DE_ele	DE_VdW	DE_gb	DE_surf	DG_gas	DG_solv	DG_total
Z3N	-103.30	-25.22	125.97	-4.78	-128.53	121.19	-7.33
Z3396339411	-129.34	-35.92	147.08	-4.85	-165.25	142.23	-23.03
Z5011956847	-247.13	-32.49	243.03	-5.89	-279.62	237.14	-42.48
Z2077032497	-206.06	-40.75	201.75	-5.72	-246.81	196.03	-50.78

This study describes the virtual screening of a compounds set from Enamine against CBL-B protein with the aim of identifying promising inhibitors. Molecular docking followed by rescoring of the docked poses with a machine learning RF-score resulted in six compounds that showed higher binding affinity than the crystal structure bound ligand. Three compounds were subsequently chosen for further analysis based on their interactions with the target protein and their ADMET profile. MD simulation results indicated the stability of the lead compounds in the protein active site and the formation of tight interactions that enforce this stability. The free binding energies calculated by MM-GBSA approach demonstrated higher affinity for the three compounds compared to the bound ligand. Together, the results highlight the possible use of these compounds as novel inhibitors of CBL-B, which needs to be validated experimentally.

Data availability

All data used for this study is included in the article.

Declaration of Competing Interest

The author declares no competing interest exists.

Funding

No funding was received to conduct this research.

Ethics and Consent to Participate declarations

Not applicable

Adasme, M. F., K. L. Linnemann, S. N. Bolz, F. Kaiser, S. Salentin, V. J. Haupt and M. Schroeder (2021). "PLIP 2021: expanding the scope of the protein-ligand interaction profiler to DNA and RNA." Nucleic Acids Res 49(W1): W530-W534.
Augustin, R. C., R. Bao and J. J. Luke (2023). "Targeting Cbl-b in cancer immunotherapy." J Immunother Cancer 11(2).
Best, R. B., X. Zhu, J. Shim, P. E. Lopes, J. Mittal, M. Feig and A. D. Mackerell, Jr. (2012). "Optimization of the additive CHARMM all-atom protein force field targeting improved sampling of the backbone phi, psi and side-chain chi(1) and chi(2) dihedral angles." J Chem Theory Comput 8(9): 3257-3273.
Bugnon, M., M. Goullieux, U. F. Rohrig, M. A. S. Perez, A. Daina, O. Michielin and V. Zoete (2023). "SwissParam 2023: A Modern Web-Based Tool for Efficient Small Molecule Parametrization." J Chem Inf Model 63(21): 6469-6475.
Chiang, Y. J., H. K. Kole, K. Brown, M. Naramura, S. Fukuhara, R. J. Hu, I. K. Jang, J. S. Gutkind, E. Shevach and H. Gu (2000). "Cbl-b regulates the CD28 dependence of T-cell activation." Nature 403(6766): 216-220.
Darden, T., D. York and L. Pedersen (1993). "Particle mesh Ewald: An N⋅log(N) method for Ewald sums in large systems." The Journal of Chemical Physics 98(12): 10089-10092.
Eastman, P., J. Swails, J. D. Chodera, R. T. McGibbon, Y. Zhao, K. A. Beauchamp, L. P. Wang, A. C. Simmonett, M. P. Harrigan, C. D. Stern, R. P. Wiewiora, B. R. Brooks and V. S. Pande (2017). "OpenMM 7: Rapid development of high performance algorithms for molecular dynamics." PLoS Comput Biol 13(7): e1005659.
Hess, B. (2008). "P-LINCS: A Parallel Linear Constraint Solver for Molecular Simulation." J Chem Theory Comput 4(1): 116-122.
Kimani, S. W., S. Perveen, M. Szewezyk, H. Zeng, A. Dong, F. Li, P. Ghiabi, Y. Li, I. Chau, C. H. Arrowsmith, D. Barsyte-Lovejoy, V. Santhakumar, M. Vedadi and L. Halabelian (2023). "The co-crystal structure of Cbl-b and a small-molecule inhibitor reveals the mechanism of Cbl-b inhibition." Commun Biol 6(1): 1272.
Koes, D. R., M. P. Baumgartner and C. J. Camacho (2013). "Lessons learned in empirical scoring with smina from the CSAR 2011 benchmarking exercise." J Chem Inf Model 53(8): 1893-1904.
Krishnamoorthy, H. R. and R. Karuppasamy (2023). "A multitier virtual screening of antagonists targeting PD-1/PD-L1 interface for the management of triple-negative breast cancer." Med Oncol 40(11): 312.
Li, H., K. S. Leung, M. H. Wong and P. J. Ballester (2015). "Improving AutoDock Vina Using Random Forest: The Growing Accuracy of Binding Affinity Prediction by the Effective Exploitation of Larger Data Sets." Mol Inform 34(2-3): 115-126.
Li, H., G. Lu, K. H. Sze, X. Su, W. Y. Chan and K. S. Leung (2021). "Machine-learning scoring functions trained on complexes dissimilar to the test set already outperform classical counterparts on a blind benchmark." Brief Bioinform 22(6).
Li, J., A. Fu and L. Zhang (2019). "An Overview of Scoring Functions Used for Protein-Ligand Interactions in Molecular Docking." Interdiscip Sci 11(2): 320-328.
Liu, Q., H. Zhou, W. Y. Langdon and J. Zhang (2014). "E3 ubiquitin ligase Cbl-b in innate and adaptive immunity." Cell Cycle 13(12): 1875-1884.
Lutz-Nicoladoni, C., D. Wolf and S. Sopper (2015). "Modulation of Immune Cell Functions by the E3 Ligase Cbl-b." Front Oncol 5: 58.
Meng, W., S. Sawasdikosol, S. J. Burakoff and M. J. Eck (1999). "Structure of the amino-terminal domain of Cbl complexed to its binding site on ZAP-70 kinase." Nature 398(6722): 84-90.
Phengsakun, G., B. Boonyarit, T. Rungrotmongkol and W. Suginta (2023). "Structure-based virtual screening for potent inhibitors of GH-20 beta-N-acetylglucosaminidase: Classical and machine learning scoring functions, and molecular dynamics simulations." Comput Biol Chem 104: 107856.
Ramesh, P., W. H. Shin and S. Veerappapillai (2021). "Discovery of a Potent Candidate for RET-Specific Non-Small-Cell Lung Cancer-A Combined In Silico and In Vitro Strategy." Pharmaceutics 13(11).
Ramesh, P. and S. Veerappapillai (2022). "Designing Novel Compounds for the Treatment and Management of RET-Positive Non-Small Cell Lung Cancer-Fragment Based Drug Design Strategy." Molecules 27(5).
Rastelli, G. and L. Pinzi (2019). "Refinement and Rescoring of Virtual Screening Results." Front Chem 7: 498.
Rountree, R., F. Cohen, A. Tenn-McClellan, A. Borodovsky, M. Gallotta, J. Stokes, J. G. Romo, C. Karim, G. M. Hansen, C. Guiducci, A. Sands and J. Gosling (2021). "Abstract 1595: Small molecule inhibition of the ubiquitin ligase CBL-B results in potent T and NK cell mediated anti-tumor response." Cancer Research 81(13_Supplement): 1595-1595.
Sud, M. (2016). "MayaChemTools: An Open Source Package for Computational Drug Discovery." J Chem Inf Model 56(12): 2292-2297.
Thien, C. B. and W. Y. Langdon (2005). "c-Cbl and Cbl-b ubiquitin ligases: substrate diversity and the negative regulation of signalling responses." Biochem J 391(Pt 2): 153-166.
Valdes-Tresanco, M. S., M. E. Valdes-Tresanco, P. A. Valiente and E. Moreno (2021). "gmx_MMPBSA: A New Tool to Perform End-State Free Energy Calculations with GROMACS." J Chem Theory Comput 17(10): 6281-6291.
Wojcikowski, M., P. J. Ballester and P. Siedlecki (2017). "Performance of machine-learning scoring functions in structure-based virtual screening." Sci Rep 7: 46710.

Table 1 is available in the Supplementary Files section

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

A structure-based virtual screening study to identify potential inhibitors targeting the TKB domain of CBL-B

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials and methods

2.1. Target preparation

2.2. Ligand preparation

2.3. Virtual Screening

2.4. Protein-Ligand interaction analysis

2.5. ADME-T analysis

2.6. Molecular dynamics (MD) simulations

2.7. Binding free energy calculations using MM-GBSA

3. Results and discussion

3.2. Protein-Ligand interaction analysis

3.3. ADME-T analysis

3.4. MD simulations

3.5. Post MD simulations interaction analysis

Inhibitor

DEele

DEVdW

DEgb

DEsurf

DGgas

DGsolv

DGtotal

Z3N

-103.30

-25.22

125.97

-4.78

-128.53

121.19

-7.33

Z3396339411

-129.34

-35.92

147.08

-4.85

-165.25

142.23

-23.03

Z5011956847

-247.13

-32.49

243.03

-5.89

-279.62

237.14

-42.48

Z2077032497

-206.06

-40.75

201.75

-5.72

-246.81

196.03

-50.78

4. Conclusion

Declarations

References

Table

Additional Declarations

Supplementary Files

Status:

Version 1

DE_ele

DE_VdW

DE_gb

DE_surf

DG_gas

DG_solv

DG_total