Enhanced metastable state models of TAM kinase binding to cabozantinib explains the dynamic nature of receptor tyrosine kinases

Abstract

Receptor tyrosine kinases (RTKs) are essential proteins in the regulation of cell signaling. Tyro3, Axl and Mer are members of TAM RTKs and are overexpressed in several cancer forms. Kinase inhibitors such as cabozantinib, foretinib are reported to inhibit TAM kinases at nanomolar concentrations. The atomistic details of structure and mechanism of functional regulation is required to understand their normal physiological process and when bound to an inhibitor. The docking of cabozantinib into the active state conformations of TAM kinases (crystal structure and computational models) revealed the best binding pose and the complex formation that is mediated through non-bonding interactions involving the hinge region residues. The alterations in the conformations and the regions of flexibility in apo and complexed TAM kinases as a course of time are studied using 250 ns molecular dynamics (MD) simulations. The post-MD trajectory analysis using Python libraries like ProDy, MDTraj and PyEMMA revealed encrypted protein dynamic motions in active kinetic metastable states. Comparison between Principal component analysis and Anisotropic mode analysis deciphered structural residue interactions and salt bridge contacts between apo and inhibitor bound TAM kinases. Various structural changes occurred in αC-helix and activation loop involving hydrogen bonding between residues from Lys-(β3 sheet), Glu-(αC-helix) and Asp-(DFG-motif) resulting in higher RMSD. Mechanical stiffness plots revealed that similar regions in apo and cabozantinib bound Axl fluctuated during MD simulations whereas different regions in Tyro3 and Mer kinases. The residue interaction network plots revealed important salt bridges that lead to constrained domain motions in the TAM kinases.

Communicated by Ramaswamy H. Sarma

Keywords:

• TAM kinase
• molecular dynamics simulations
• mechanical stiffness
• principal component analysis
• residue interaction networks

Acknowledgements

NGKRS thanks University of Hyderabad for UGC Non-NET fellowship. The authors thanks DST-PURSE and UGC UPE2 for funding and CMSD for computational facilities.

Disclosure statement

The authors declare no competing interests.