Keywords

DFG, Kinase, Model, Structure-based Drug Design, Type-II Inhibitors, Vector

Reference

DOI: 10.1021/cb500696t

Abstract

Protein kinases shift between active and inactive states, involving conformational changes in the aspartate-phenylalanine-glycine (DFG) motif, essential for their function.
Type-I inhibitors target the ATP-binding site in the active state, while type-II inhibitors exploit an allosteric pocket formed adjacent to the ATP site in the inactive DFG-out state.
However, the lack of inactive kinase structures limits structure-based drug design for type-II inhibitors.

This paper presents DFGmodel, a computational tool that:

  • Generates inactive kinase models from either known active structures or sequences.
  • Yields accurate models (RMSD ≤ 1.5 Å) validated through multiple structural metrics (Z-DOPE, TM-score).
  • Distinguishes type-II inhibitors from non-binders (AUC > 0.70), suitable for virtual screening.

They also provide case studies where DFGmodel predicts off-target interactions and expands structural understanding for drug design.

Notes

1. General Summary

  • DFGmodel fills a critical gap by predicting kinase inactive (DFG-out) conformations, enabling structure-based type-II inhibitor discovery.
  • Uses comprehensive analysis of kinase structures to inform model building.
  • Applies to both existing active structures and kinase sequences lacking any structure.
  • Achieves high accuracy, making it useful for virtual screening and understanding DFG-flip mechanics.

2. Structural and Modeling Insights

  • DFG-flip involves a 180° flip of the Asp and Phe residues in the DFG motif.
  • DFG-out state shows greater flexibility, especially in the N-lobe and αC-helix, while DFG-in is more compact.
  • Interestingly, the β-hairpin preceding DFG remains structurally conserved in DFG-out conformations across kinases.
  • The position of the DFG motif may be linked to A-loop conformation, but not necessarily controlling A-loop secondary structure.
  • Gatekeeper and (D–1) residues do not significantly differ between kinases with and without known DFG-out structures, suggesting DFG-out accessibility may be an intrinsic property modulated by other factors (e.g., phosphorylation, binding partners).

3. Methodological Points

  • Vector-based methods are used to characterize DFG orientation, but RD noticed:
    • They used cross product to define vector relations but did not employ dot product, which could also be useful for angle measures.
  • Something new and notable:
    • Z-DOPE and TM-score are used alongside RMSD for structural evaluation — interesting metrics RD may explore further.

4. Application and Performance

  • The models achieved RMSD ≤ 1.5 Å, demonstrating accuracy and structural relevance.
  • Successfully distinguish type-II inhibitors from non-binders, with AUC > 0.70 in virtual screening tests.
  • Captures inhibitors’ off-target activities through modeling — useful for drug design and safety assessments.

5. RD’s Thoughts and Learnings

  • RD finds the description of DFG-flip clear and useful.
  • References 21 and 22 offer valuable background (worth checking).
  • The vector approach focused only on DFG; rotation consideration was possibly missing — dot product might improve orientation analysis.
  • RD also noted novel metrics (Z-DOPE, TM-score), could be worth integrating into RD’s own pipeline.
  • The idea that DFG-out accessibility is intrinsic and widespread (not limited by specific residues) shifts perspective on kinase dynamics.
  • RD is considering trying DFGmodel for ongoing kinase-related work.

Take-home Messages

  • DFGmodel is a powerful computational approach for predicting kinase inactive (DFG-out) structures, facilitating type-II inhibitor discovery.
  • The DFG-flip mechanism is conserved and linked to broader kinase flexibility.
  • Structural flexibility and dynamic pockets are essential features of kinases, relevant for allosteric drug design.
  • Kinase modeling benefits from integrating multiple structural evaluation metrics beyond RMSD.
  • Intrinsic kinase flexibility and DFG-out accessibility may be more widespread than previously thought, independent of specific residue patterns.