Using Multiscale Molecular Modeling to Analyze Possible NS2b-NS3 Protease Inhibitors from Philippine Medicinal Plants

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This work by Cordero et al., summarizes the use of in-silico ADMET screening coupled with molecular docking to first identify top 10 hits from a total of 2944 phytochemicals obtained from various sources followed by using molecular dynamics to first characterize the interactions of these hits with the dengue NS2b-NS3 protease. A binding-free energy analysis using MM-PBSA was used to compare these 10 compounds with known allosteric inhibitors of the viral NS2b-NS3 protease, idelalisib and nintedanib. 

I have some concerns that need to be addressed that will result in a much better presentation towards the general scientific community.

Major:

1. In the materials and methods section for docking there are two main issues that need clarification -  the first one deals with the structure used for docking -PDB 2FOM. Upon having a close look at the PDB header, it was seen that the structure lacked several residues which might interact with the ligands during the docking/MD simulations. Moreover, omitting a part of the protein also portrays an incomplete finding to the community. Can the authors show how they accounted for the missing residues?. The second issue pertains to the molecular docking. In this instance, It was clear that 1000 poses were generated for each run, with the top ranking pose with the most negative binding energy being selected. However, it is a widely known fact that molecular docking is approximate and more post-processing in the post-docking phase in the form of clustering and rescoring is warranted to get an unbiased pose selection. In the current case no such endeavor was observed. Can the authors perform RMSD-based clustering and rescoring with other open-source scoring functions like DOCK, DrugScoreX, or any machine-learning based score to ascertain if the top ranked pose from AutoDock is indeed the best ranking one. Alternatively, they can use DiffDock/Alphafold3/NeuralPlexer along with AutoDock and report their findings.

2. A closer look at the structures of the top 10 hits indicate striking structural similarities in the scaffold with negligible scaffold diversity. Can the authors discuss the reasons behind this?

3. The results in the table 4 do not show any standard deviation/SEM inspite of the results coming from 3 independent simulations. Please add the values and discuss them in the results.

Minor

1. Line 125 - A cartoon representation showing the active site and the catalytic triad would be much appreciated.

2.  A typo on line 158 - A ' after mechanics.

3.  Duplication of references 18 and 20.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

I have peer reviewed the manuscript entitled “Using multiscale molecular modeling to analyze possible 2 NS2b-NS3 protease inhibitors from Philippine medicinal 3 plants”. The authors aimed to find out whether various plant-derived ligands can be used as inhibitors of Dengue protease. After careful review, I find that the manuscript does not meet the required scientific standards for publication in this time.

The following are my major concerns:

1. Regarding writing style, there are significant structural issues in this manuscript. The results section is combined with the discussion, the discussion contains mostly results, and the conclusion reads like a discussion. I highly recommend that the authors thoroughly revise and restructure the manuscript. Also, there are many fundamental errors, especially in the table captions and figure legends. The authors should double-check these parts and also ensure consistent use of abbreviations throughout the manuscript.
2. The study lacks experimental validation for the computational predictions, docking, MD and MM/PBSA.
3. The docking section lacks clarity in the description of benchmarking docking procedures.
4. The MD simulation time is inadequate and should be extended from 500 ns to at least a microsecond.
5. Some analyses should be performed to check the reproducibility of data across three replicates.
6. To check the stability of ligands in the active site of the protease and to study 3-D structural changes during MD simulations in presence of ligands for both bound and unbound states, more analyses need to be done by authors.

Here are my minor concerns:

1. The introduction should have more information about the protein-receptor involved in virus invasion and how it may be inhibited through small molecule drugs.
2. Line 122: The protease should be optimized using MD before docking.
3. Line 154” The MM/PBSA method needs further explanation, including number of frames and trajectory editing details
4. The quality of all figures is very low and lack important details. They need be replaced with higher quality figures.
5. Line 231: Cleavage from the image is unclear, thus additional explanation or more detailed description would clarify what methods were used by authors in order to identify cleavage.
6. Line 292: The order of Figures 3 and 4 is incorrect and should be corrected.
7. The authors should mention which software was used for illustrations in the figure legends.
8. The docking data (binding free energy) should be moved to the supplementary section, as docking is usually used for obtaining the initial structure for MD simulations.
9. The structure of the compounds should be explained or provided as structural images in the supplementary materials.
10. Line 224: If molecular weight can impact interactions, the authors should provide a graph and calculate the R² value for the relationship between molecular weight and MM/PBSA binding free energy.
11. Figure 1 is unnecessary and should be replaced with an image of the last frame of the MD simulation.
12. Line 230-232: This explanation is unclear and needs to be more detailed

 

Comments on the Quality of English Language

English is OK in this manuscript.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Review (CIMB- 3068357):

 

The current research article by Cordero et al. implements computational approaches to identify the potential hits to target dengue virus NS2b-NS3 protease, involved in the replication process. To this end, authors performed various in silico approaches including drug-likeness, molecular docking, and molecular dynamics simulations. The authors considered 2,944 phytochemicals from various Philippine plants for anti-dengue activity and identified at least 6 potential compounds using the present study observations. I appreciate the author's efforts in these directions. The research was conducted in a well-structured manner, and the results were discussed thoroughly throughout the manuscript. However, the analysis was a bit of a concern, and below are some of my comments. I believe it will be helpful to maximize the potential of the manuscript. Below are some of my specific comments.

 

Major Comments:

 

1. The introduction section needs to be improved. The authors could highlight the need for anti-dengue agents; discuss the limitations of current treatment options and outline the specific advantages of using in-silico approaches; It would also be beneficial to discuss how well such methods have been successfully implemented in previous studies.  
2. ADME properties: It appears to me that the authors solely focused on the drug-likeness (i.e., using criteria <= 5 violations based on Lipinski, Ghose, Veber, Egan, and Muegge rules) and toxicity parameters without providing any information regarding the ADME properties that correspond to the absorption, distribution, metabolism, and excretion. Could authors mention how the estimated ADME properties were utilized to screen the studied compounds? Additionally, mention the inclusion and exclusion criteria based on the ADME properties as well.
3. Details on Molecular docking: Have the authors considered using flexible docking? It would be helpful if you could specify the type of docking implemented in this study. If the authors think that the catalytic triad plays an important role in the substrate binding, I think the flexible docking flexible docking might provide better predictions for the studies docking results.
4. Details on Molecular dynamics simulations: I believe that force field parameters play an important role in the case of molecular dynamics simulations. Improper parameters could introduce artifacts in the simulations and lead to misleading results. I am curious to know whether the authors evaluated the force field parameters obtained from the CGenFF web server? It would be beneficial to provide the force field parameters to the community for the studied ligands in supporting the information.
5. I think it would be interesting to see the final conformations of ligands within the protein active site. Could the authors also provide the snapshots of protein-ligand complexes at the end of the 100 ns simulation, similar to Figure 1?
6. Page 7: Lines 257-259. Statements like these might not be appropriate. I believe the systems should be sufficiently equilibrated before proceeding to the MD production run and to appropriately estimate quantitative properties. In this case, a 100 ps equilibration might be too short to be considered for equilibrating the protein-ligand complexes. It would be better to check and provide the equilibrium properties in the supporting information to ensure the proper equilibration. Further, it could be better to use the final 40 ns for the present study calculations (as some of the systems were found to be stable around 50-60 ns time), considering the initial 60 ns as part of the equilibrium simulation. I hope these adjustments do not affect the outcomes of the study.
7. Page 8: Lines 292-298: Defining RMSD here might be irrelevant. It would be better to define it in the methods sections or before discussing Figure 2; Additionally, Figure 4 was discussed before addressing Figure 3.  The same lines were repeated on Page 10. Try to arrange these sections appropriately.
8. Figure 3; I am a bit surprised to see such large RMSD values (~100-200 Å) for systems D and F. It looks like there might be something wrong with the system or calculations. Try to examine the trajectory one more time and verify your RMSD calculation as well.
9. I would suggest moving Figures 2 and 3 to the supporting information.
10. Figure 4, suggests that all the studied ligands were very well stabilized within the active site of NS2b-NS3, except the system ‘D’, i.e., ISO complex. Moreover, it would be interesting to see the protein alone RMSD in addition to the shown ligand RMSD values, i.e., Figure 4. Additionally, to better understand the protein dynamics authors should have conducted the MD simulations for the ‘Apo’ system as well.
11. Table 4: Could the authors provide error bars? I think the authors have performed the 3 MD runs, considering average free energy estimates along with error bars would improve statistical reliability. Additionally, some of the binding energy values did not add up to the total, especially for IDE and HON systems. 
12. Figure 7: Could the authors highlight the active site and gate residues in the shown plot? It appears to me that the residues with higher decomposition were different among the studied systems. Additionally, could the author estimate the hydrogen bonding interaction details?

 

Other Comments:

1. In the case of MD simulations, authors added 0.15 M NaCl ions. Could the authors also mention about the neutralizing ions used? Additionally, include the 2D structural information with the charge states in the supporting information (only for reference and selected candidates).
2. It looks like the MM/PBSA equation was not completed. It would be helpful if the authors could provide the complete equation.
3. Figur1 and 6. The amino acid residue labels appear unclear and difficult to read.
4. It appears to me that the plots in Figure 2, specifically, A, D, and F are different from all the other graphs. Is there any difference in the calculations?
5. Missing punctuation and spaces were observed.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Although the authors tried hard to revise and improve the manuscript, several issues remain unaddressed. The structure of manuscript still has major problems, especially the combination of the discussion with the results, which negatively affects the flow. In the conclusion, the authors should restate the hypothesis and very briefly explain the main findings. Also the authors have not provided any validation for their methods. The quality of some figures is not clear, like Figure 7, completely unreadable.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

Review_V2 (CIMB- 3068357):

 

Thank you for all your answers.

 

Minor Comments:

1. Some of the Figure and Figure 9 residue labels were overlapped. Make sure that they are clearly visible.
2. Figure 3D, Y-axis has been cropped.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

I have no further comments.

Author Response

We thank the reviewer for all their comments and suggestions.