Defining the Active Conformation of Typical Protein Kinase Domains from Substrate-Bound PDB Structures Enables...

Area of Science:

• Structural bioinformatics
• Computational biology
• Protein kinase structural dynamics

Background:

• Humans possess 437 catalytically competent protein kinase domains, typically adopting a fold similar to Protein Kinase A (PKA).
• The active conformation of a kinase is essential for binding ATP, magnesium, and substrate, yet many are not structurally characterized in this state.
• Understanding active kinase conformations is crucial for deciphering substrate specificity and the impact of mutations in diseases like cancer.

Purpose of the Study:

• To derive structural criteria defining the active form of protein kinases based on substrate-bound structures.
• To assess the proportion of human protein kinases currently represented in their active form within the Protein Data Bank (PDB).
• To generate computational models of all 437 human protein kinases in their active conformation using AlphaFold2.

Main Methods:

• Structural bioinformatics analysis of 248 substrate-bound kinase crystal structures to establish active form criteria.
• Application of derived criteria to PDB kinase structures to identify those in active conformations.
• Utilized AlphaFold2, employing PDB templates, sequence alignments, and active-kinase criteria, to model all 437 human protein kinases.

Main Results:

• Established structural criteria for active protein kinases, including requirements for the DFG motif and activation loop segments.
• Identified that only 130 out of 437 (30%) human catalytic protein kinases are structurally represented in their active form in the PDB.
• Generated high-accuracy AlphaFold2 models for all 437 human protein kinases in their active state, validated against PDB structures (71% < 1.0 Å RMSD).

Conclusions:

• A significant majority of human protein kinases are not available in their active form in structural databases.
• The generated AlphaFold2 models provide a valuable resource for understanding kinase function, substrate specificity, and mutations in disease.
• These models can serve as templates for studying mutations causing constitutive activity in cancer and for modeling substrate and inhibitor binding.