In silico analysis of the different variable domain oriented single-chain variable fragment antibody-antigen complexes

Abstract

Single-chain variable fragment (scFv) antibodies hold great potential as diagnostic tools and therapeutic agents, especially for tumor cells. Since these applications require their production with improved properties, the design strategy of scFvs is crucial for their active, soluble, and high yield expression with high affinity towards their antigens. The order of V_L and V_H domains is one of the important parameters that affect the expression and binding affinity properties of scFvs. In addition, the optimum order of V_L and V_H domains could change for each scFv. In the present study, we used computer simulation tools to evaluate the effect of variable domain orientation on structure, stability, interacting residues of scFvs, and binding free energies of scFv-antigen complexes. We selected anti-HER2 scFv, which is specific for human epidermal growth receptor 2 (HER2) overexpressed in breast cancer, and anti-IL-1β scFv against IL-1β which is an important inflammatory biomarker, as model scFvs. Molecular dynamics simulations of the scFv-antigen complexes for 100 ns resulted in stability and compactness for both scFv constructs. Interaction and binding free energies calculated by the Molecular Mechanics–Poisson–Boltzmann Surface Area (MM-PBSA) approach suggested that the relative binding energies of anti-HER2 scFv-V_LV_H and anti-HER2 scFv-V_HV_L constructs had similar binding affinity towards HER2, while a relatively more negative binding free energy obtained between anti-IL-1β scFv-V_HV_L and IL-1β pointed to a higher binding affinity. The in silico approach and the results obtained here could be applied as a guide for future experimental interaction studies for highly specific scFvs used as biotechnological tools.

Communicated by Ramaswamy H. Sarma

Keywords:

• Binding free energy
• GROMACS
• MD simulation
• scFv
• V_L/V_H orientation

Acknowledgments

This study was mostly the outcome of PhD thesis by I.K. The MD simulations reported in this paper were partially performed at TUBITAK ULAKBIM, High Performance and Grid Computing Center (TRUBA resources). Authors also acknowledge Assoc. Prof. Dr. Özge Yüksel Orhan for providing the high-performance computer to carry out some of the MD simulations.

Disclosure statement

The authors declare that there are no conflicts of interest.

Additional information

Funding

I.K. thanks to Scientific and Technological Research Council of Turkey (TUBITAK) for National Research Fellowship for PhD students (TUBITAK 2211-A).