Structure-Based Discovery of Potential HPV E6 and EBNA1 Inhibitors: Implications for Cervical Cancer Treatment
, Carolyn N. Ashley 1, Miriam Velazquez 1,2, Patrick O. Sakyi 3,4, Samuel K. Kwofie 5,6 and Whelton A. Miller III 1,2,*Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript "Structure-Based Discovery of Potential HPV E6 and EBNA1 Inhibitors: Implications for Cervical Cancer Treatment" is dedicated to the brilliant research of drugs against cervical cancer. It is well known that cervical cancer is the most common form of female genital cancer, claiming thousands of lives around the world. It is gratifying to see that research in this area is actively underway, and that molecular drug design can significantly reduce the search for active substances and thereby facilitate the creation of new drugs. The team of authors of the article submitted for review carried out extensive and meticulous work on the design of the connections. Of course, I would very much like to see the synthesis and study of the biological activity of the presented compounds. The reviewer really hopes that this work will continue and that the simulated drugs will see the light of day.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsIn this manuscript, the authors have aimed to identify potential inhibitors of HPV16 E6, HPV18 E6, and EBNA1 utilizing a comprehensive molecular modeling approach. They have screened the ZINC database following docking, ADMET and biological activity predictions, and molecular dynamics simulations to propose six potential hits that can be potential hits towards therapeutics for cervical cancer patients with HPV/EBNA1 co-infection.
1. The rationale for discovering compounds targeting both proteins is missing. Although this is interesting, the authors should provide justifications as to why it is important to develop therapeutics targeting both proteins vs either one for cervical cancer therapeutics.
2. Details of the workflow for virtual screening of the ZINC library are entirely missing.
3. The authors have used the Chinese Medicine catalog library of the ZINC15 database and prefiltered the compounds using the molecular weight cutoff. Details of how many compounds did they begin with and how many after the cutoff need to be mentioned. How was the actual virtual screening carried out to provide the 26 hits shown in Table 1?
3. Figure 1 showing the schematic of the workflow is confusing and should be redone indicating at every stage how many compounds were moved to the next tier.
4. Figure 3 shows the ligand-protein interactions of the control as well as one of the hits (ZINC000085568136). Why was this compound chosen? This is not the highest-scored compound against EBNA1.
5. Same is the case with Figures 4 and 5.
6. In the bioactivity prediction section, the authors should define what the terms Pa and Pi mean for better readability.
7. Additionally, the bioactivity prediction section discussion should be represented in a tabular form for better understanding.
Overall, the hypothesis of targeting both HPV E6 and EBNA1 is novel and provides a great basis for the application of polypharmacology. Although the topic and techniques are of great interest to the journal and the scientific novelty is good, the manuscript lacks critical data and hence is not suitable for publication in its current form. The reviewer however strongly encourages the authors to address the following comments to strengthen the paper for resubmission.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe manuscript describes an effort of docking compounds to HPV16 E6, HPV18 E6 and EBNA1. The authors also performed classical unbiased molecular dynamics simulation and computational biological activity predictions. While I appreciate the effort the authors have put in, the work has serious deficiencies. I will try to point out only few of the many major deficiencies I encountered.
The predicted docking affinities (compared to reference compounds) are quite low and within margin of error of docking program used in this study (+-2 kcal/mol). Not even one of the compounds predicted was tested in vitro to reinforce confidence in computational predictions. At least some computational ligand optimization of predicted top compounds could have been performed to inspire confidence in results.
Authors mention in line 234 "charges were computed using UCSF Chimera" and in line 235 mention "The ligands were then submitted to LigParGen [81] to generate their respective topology files (“gro” and “itp” files)". LigParGen tool calculates partial charges for the compound hence there was no need to use UCSF Chimera for partial charge generation. Additionally, method used to generate partial charge should be explicitly mentioned. e.g. am1-bcc, Gasteiger charges etc.
Line 240 authors mention "chlorine or sodium ions were then added". Authors should be clear whether Chlorine or Sodium ions were added.
The solvated box was not brought up to the physiological salt molarity of ~0.15 mM, which is quite standard in MD simulations.
The authors do not describe which thermostat, barostat was used for MD in the methods section. Details on methodology used to minimize and equilibrate the solvated complex also seem missing from methods MD section. Example how many steps of minimization was performed (steepest decent and/or conjugate gradient) and how long did the equilibration last.
The MD was run for only 100ns which is quite short by today's standards. I understand that the authors performed lot of simulations for this study, but only 100 ns does not inspire confidence in the results.
Line 289 "docking scores ranging from 4.33 to 8.46 kcal/mol.." the docking scores are positive. Positive docking scores are concerning as it means that the complex formation is endothermic. Similarly in line 290 the values are positive.
Line 295 "Herein, the shortlisted compounds" table number should be mentioned.
Authors mention consensus docking. I think they want to imply multi target docking throughout the manuscript.
The authors refer to MMPBSA obtained numbers as binding free energy. MMPBSA calculates change in enthalpy. MMPBSA does not evaluate entropy. Binding free energy estimation requires both these terms (and temperature). Hence using term binding free energy is misleading throughout the manuscript whenever used describing MMPBSA results.
Only one docking tool was employed throughout the docking study. The docking should be performed with at least 3-4 docking tools. If the docking for all 25K compounds is not possible with 3-4 docking tools due to lack of resources, then at least the top compounds should be docked with docking tools with different evaluation methodologies. I would suggest Dock (docking tool) from USCF as authors are familiar with UCSF tools and it is free for academic use. There are many other free docking tools to choose from. I suggest authors to visit Clik2drug (https://www.click2drug.org/) website to get list of a variety of tools that authors might find useful not only for this study but for future reference too.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have incorporated all major concerns the reviewer has. The manuscript is fit to be published now.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for Authors
I appreciate the efforts authors have put in to revise the manuscript. They have addressed most of my concerns. The only concern left is on convergence of MD simulations. I understand that extending simulations for all complexes may require investment of significant resources which might not be possible, but inferences derived out of analysis of single trajectory MD simulation are prone to under sampling and/or over interpretation. I would suggest authors show convergence of their MD using some statistical measure on already performed simulation data. Only RMSD as a function of simulation time does not guarantee convergence by itself. For example: authors could cluster first 80% their simulation frames and show that rest 20% frames sample conformations already found in first 80% data. Authors can use any other convergence criteria of their choice too.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
