In the early development of antibody drugs, the identification of antibody-antigen binding epitopes is the basic prerequisite for revealing the pharmacodynamic mechanism of antibody drugs. It is of great significance for the subsequent humanization of antibody drugs and antibody affinity maturation, and can accelerate the process of antibody drug development. However, it is still difficult to obtain accurate antigen-antibody complexes. The main obstacle lies in the error of antibody CDR-Loop prediction of homology modeling. In the process of conventional rigid protein-protein docking, these modeling errors have not been optimized or alleviated, resulting in incorrect models for subsequent antigen-antibody docking. Unlike rigid docking, Rosetta's flexible docking can mimic the "conformational selection" effect during the formation of antibody-antigen complexes by introducing multiple conformations, but it still cannot resolve the conformational changes that occur during antigen-antibody binding ("induced fit effect"). In 2009, Aroop Sircar of the GrayLab laboratory proposed a new flexible docking scheme SnugDock for the antibody-antigen system. The tutorial is detailed below.
The calculation of SnugDock requires an initial antibody-antigen complex conformation, which is usually derived from rigid docking knots. In the absence of experimental information, the combination strategy of ZDOCK/iPatch or PIPER/ADARS can be used to obtain a more reasonable docking conformation. If there is certain experimental information, you can use HADDOCK or SwarmDock for preliminary docking. If you are more fortunate, when the antibody has a similar sequence complex structure, you can also use the complex homology template modeling to obtain the initial conformation! Since the initial structure sources are diverse, different programs have different chain sequence processing, so after obtaining the preliminary complex PDB structure. The chain sequence needs to be adjusted to meet Rosetta's requirements: the chain sequence should be light chain (L)->heavy chain (H)->antigen (*).
2.1 Prepack stage
With the initial guess of the structure of the complex, it is necessary to pre-optimize the side chains of the two docking molecules to ensure that the amino acids outside the binding interface are in the lowest energy state, so as to avoid their interference with subsequent scoring and ranking.
2.2 SnugDock officially launched
SnugDock is currently a mature Rosetta application, which can be invoked by users using a simple command line.
2.3 SnugDock result analysis
After running, the program will enter the score-snugdock.sf file, where you need to pay attention to the scores of total_score, I_sc, sc_value, dG_separated, and dSASA_int.
total_score: total energy of the complex
I_sc: interface score (interface_score)
sc_value: shape complementarity
dG_separated: binding free energy change
dSASA_int: solvent accessible surface area embedded in the interface
With multiple index references, you can further filter the conformation and select the final output model through some empirical parameters, such as:
The embedding surface area of the antibody-antigen complex is 1100~1600Å^2.
The shape complementarity of a good composite model is roughly between 0.6-0.8
If you still have difficulty using Rosetta, you can also directly use the SnugDock module in GrayLab's ROSIE server to submit your own tasks online.
Note: Online snugdock cannot use ensemble mode, and the input structure also needs to be adjusted according to the chain sequence described in this article.
After entering the job description, the input PDB file, Docking partner, SnugDock method control, etc. according to the prompts on the page, you can submit it to the server for calculation.