Team:Korea HS/Model


Designing a Hyperstable Antibody with
Cell-penetrating Peptide for Intracellular Targeting



Through access to the wet lab, our team designed and tested the validity of the hyperstable CPP tagged scFv(Ras). While this experiment in creating the antibody was successful, specific information about binding is needed to extend the experiment for future studies. CPP tagged scFv(Ras) mimics the X-ray structure of the 2vh5 antibody model. Therefore, the binding property is thought to be similar to each other. Comparison of the two antibody models 2vh5 and scFv(Ras) is an effective method to check the binding properties of scFv(Ras), and while the process can be done experimentally, we decided to use protein-protein docking modeling due to time contains in performing this experiment in the wet lab.

Modeling Plan / Objective

What is Protein-Protein Docking Modeling?

Molecular docking, or protein-protein docking, is a method to predict the structure of a protein-protein complex from the structure of individual proteins.

It predicts the preferred orientation of the binding of one molecule (ligand) to the other molecule (receptor) using computational simulations with scoring functions.

Search algorithms and scoring functions are used to determine the ‘best fit’ by calculating lowest energy configuration considering multiple factors such as rotational bonds and free-energy due to protein-ligand interactions.


Free online docking server, ClusPro

The ClusPro server is a widely used tool for protein-protein docking. It was developed by Boston University.

It only requires two protein (receptor & ligand) files in Protein Data Bank (PDB) format.

ClusPro offers advanced docking mode such as antibody mode with masking options to maximize accuracy. Antibody mode substantiates the modelling process by removing the assumption of symmetry from modeling, due to the fact that an antibody-antigen complex has a flat and asymmetrical surface compared to an enzyme-inhibitor’s general complementary surface. Additional masking of non-complementary determining regions (non-CDR) can be made to increase accuracy of results.

Docking Process

Two PDB files are provided for ClusPro Antibody Mode:
scFv(Ras) - Antibody
HRas(G12V) - Antigen


Our monoclonal antibody scFv(F8) is produced against the coat protein of the plant virus called AMCV. Previous sentence must be replaced with “scFv(Ras) was engineered from hyperstable scFv(F8) based on the HRas(G12V): antibody complex structure (PDB ID: 2vh5). HRas(G12V) is a version of human Ras protein that has a mutation on its 12th position. It was extracted from a known protein data bank of 2VH5 using AutoDockTools.

The scFv(Ras) with masked CPP was modeled using antibody mode. The CPP was masked because the peptide is considered ‘flexible’ which means there is likely to be minimal interference in the actual binding of scFv(Ras) and HRas(G12V).

The final results were derived from the ‘scFv with CPP (antibody mode, CPP masked)’ model. The model evaluated was cluster 0 with the largest members of 159 and lowest energy of -359.7 out of the 27 possible arrangements created on the platform. (Top 10 clusters are shown for simplicity)


The docking model comprised of scFv(Ras) and HRas(G12V)
scFv with CPP (antibody mode, CPP masked):
scFv without CPP (antibody mode)

Shown below are the receptor(scFv(Ras)), ligand (HRas(G12V)) and the docking model for scFv(Ras): HRas(G12V). The crystallographically observed complex structure of anti-Ras antibody:HRas(G12V) (PDB ID: 2vh5) was overlapped on the final docking model for comparison.

scFv(Ras) + HRas(G12V) Docking
scFv(Ras) + HRas(G12V) / 2vh5 Overlap


As shown on the results for the ClusPro protein-protein docking, our team was able to create a protein-protein docking model for scFv(Ras) and HRas(G12V) and compare it with the 2vh5 structure. The comparison revealed that while the antibody binds to the conventional antibody binding site, there were differences as to the exact positioning. This indicates that scFv(Ras) and HRas(G12V) does not have identical binding mode. Through the protein-protein modeling our team was able to create a basis for future study on further modification of our engineered antibody scFv(Ras).

Future Work

We believe that the protein-protein docking modeling created here is going to be our stepping stone in extending the possibility of using CPP tagged antibodies in the field of cancer therapeutics. The exact binding mode of engineered scFv(Ras): HRas(G12V) needs to be determined experimentally in the future. As much as the technology involved in cancer is meticulous, constant engineering and experimentation with the variation of initial creations are going to be crucial. Therefore by identifying the binding capacity and characteristics of scFv(Ras), an experimentally synthesized and verified antibody, our team hopes to make further contributions to CPP tagged antibodies in cancer therapeutics.


[1] Brenke R, Hall DR, Chuang G-Y, Comeau SR, Bohnuud T, Beglov D, Schueler-Furman O, Vajda S, Kozakov D. Application of asymmetric statistical potentials to antibody-protein docking. Bioinformatics. 2012 Oct; 28(20):2608-2614; pdf

[2] Kozakov D, Beglov D, Bohnuud T, Mottarella S, Xia B, Hall DR, Vajda, S. How good is automated protein docking? Proteins: Structure, Function, and Bioinformatics. 2013 Dec; 81(12):2159-66. pdf

[3] Kozakov D, Hall DR, Xia B, Porter KA, Padhorny D, Yueh C, Beglov D, Vajda S. The ClusPro web server for protein-protein docking. Nature Protocols. 2017 Feb;12(2):255-278. pdf

[4] Vajda S, Yueh C, Beglov D, Bohnuud T, Mottarella SE, Xia B, Hall DR, Kozakov D. New additions to the ClusPro server motivated by CAPRI. Proteins: Structure, Function, and Bioinformatics. 2017 Mar; 85(3):435-444. pdf