AlphaFold2 models of the active form of all 437 catalytically-competent typical human kinase domains.

Humans have 437 catalytically competent protein kinase domains with the typical kinase fold, similar to the structure of Protein Kinase A (PKA). Additionally, there are 57 pseudokinases with the typical kinase domain but without phosphorylation activity. Only 268 of the 437 catalytic typical protein kinases are currently represented in the Protein Data Bank (PDB) in various functional forms. The active form of a kinase must satisfy requirements for binding ATP, magnesium, and substrate. From the structures of 40 unique substrate-bound kinases, as well as many structures with bound ATP, we derived several criteria for the active form of protein kinases. These criteria include: 1) the DFGin position of the DFG-Phe side chain; 2) the " " conformation based on the backbone and side-chain dihedral angles of the XDFG motif which we previously characterized as required for ATP binding (Modi and Dunbrack, PNAS, 2019); 3) the existence of an N-terminal domain salt bridge between a conserved Glu residue of the C-helix and a conserved Lys of the N-terminal domain beta sheet; 4) backbone-backbone hydrogen bonds of the sixth residue of the activation loop (DFGxxX) and the residue preceding the HRD motif ("X-HRD"); and 5) a contact (or near contact) between the Cα atom of the APE9 residue (9 residues before the C-terminus of the activation loop) and the carbonyl oxygen of the Arg residue of the HRD motif. These last two requirements underscore the structural interplay between the activation loop and the catalytic loop containing the HRD motif that serve to construct a groove capable of binding substrate. With these criteria, only 155 of 437 catalytic kinase domains (35%) are present in the PDB; only 130 kinase domains (30%) are in the PDB with complete coordinates for the activation loop. Because the active form of catalytic kinases is needed for understanding substrate specificity and the effects of mutations on catalytic activity in cancer and other diseases, we used AlphaFold2 to produce models of all 437 human protein kinases in the active form. We used active structures we identified from the PDB as templates for AlphaFold2 (AF2) as well as shallow sequence alignments of orthologous kinases from Uniprot (>50% sequence identity to each query) for the multiple sequence alignments required by AF2. We select models for each kinase based on the pLDDT scores of the activation loop residues, demonstrating that the highest scoring models have the lowest or close to the lowest RMSD to 22 non-redundant substrate-bound structures in the PDB. A larger benchmark of 130 active kinase structures with complete activation loops in the PDB shows that 80% of the highest-scoring AlphaFold2 models have RMSD < 1.0 Å and 90% have RMSD < 2.0 Å over the activation loop backbone atoms. We show that several of the benchmark structures from the PDB may be artifacts that are not likely to bind substrate and that the AlphaFold2 models are closer to substrate-bound structures of closely related kinases. Models for all 437 catalytic kinases are available at http://dunbrack.fccc.edu/kincore/activemodels . We believe they may be useful for interpreting mutations leading to constitutive catalytic activity in cancer as well as for templates for modeling substrate and inhibitor binding for molecules which bind to the active state.