Decrypting the Structural, Dynamic, and Energetic Basis of a Monomeric Kinesin Interacting with a Tubulin Dimer in Three ATPase States by All-Atom Molecular Dynamics Simulation

[slochower]

@doi:10.1021/bi501056h

We have employed molecular dynamics (MD)
simulation to investigate, with atomic details, the structural dynamics
and energetics of three major ATPase states (ADP, APO, and ATP
state) of a human kinesin-1 monomer in complex with a tubulin dimer.
Starting from a recently solved crystal structure of ATP-like kinesin−
tubulin complex by the Knossow lab, we have used flexible fitting of
cryo-electron-microscopy maps to construct new structural models of
the kinesin−tubulin complex in APO and ATP state, and then
conducted extensive MD simulations (total 400 ns for each state),
followed by flexibility analysis, principal component analysis, hydrogen
bond analysis, and binding free energy analysis. Our modeling and
simulation have revealed key nucleotide-dependent changes in the
structure and flexibility of the nucleotide-binding pocket (featuring a
highly flexible and open switch I in APO state) and the tubulin-binding
site, and allosterically coupled motions driving the APO to ATP transition. In addition, our binding free energy analysis has identified a set of key residues involved in kinesin−tubulin binding. On the basis of our simulation, we have attempted to address several outstanding issues in kinesin study, including the possible roles of β-sheet twist and neck linker docking in regulating nucleotide release and binding, the structural mechanism of ADP release, and possible extension and shortening of α4 helix during the ATPase cycle. This study has provided a comprehensive structural and dynamic picture of kinesin’s major ATPase states, and offered promising targets for future mutational and functional studies to investigate the molecular mechanism of kinesin motors.

[slochower]

This builds on their previous study but performs longer unconstrained MD simulations and uses a newly solved ATP-bound structure.

Methods

• Three simulations -- ADP, ATP, and apo kinesin -- with the apo structure being derived from the ATP structure. Tubulin was also included, but with restraints.
• Flexible fitting of crystal structures into cryo-EM maps using MDFF that applies two extra potential terms to help the fitting.
• CHARMM27 with TIP3P water.
• RMSF and PCA analysis performed, and the binding free energy was calculated.
• The binding free energy comes from snapshots of last 50 ns using a combination of non polar and electrostatics. The nonpolar comes from the vdW interaction energy between the kinesin and tubular dimer. The electrostatic binding energy contributions comes from the change in electrostatic energy comparing unbound kinesin and the kinesin-tubulin complex, with fitting parameters.
• The binding free energy was then partitioned among certain residues, using CHARMM.

Results

• There were some changes during MDFF fitting into the cryo-EM maps. I'm not sure if this is expected or how this relates to the goodness of fit. I skimmed this part.
• There are key (global) conformational changes associated with the three structures, including a counterclockwise rotation of kinesin on top of tubulin (Figure 1a).
• As expected switch I undergoes a large "open-to-close" transition from ADP-bound to apo to ATP-bound. This is consistent with the observation that switch II opens to enable complete releases of Mg and ADP.
• The contact surface area between kinesin and tubulin fluctuated during the simulation. Kinesin is most flexible in the ADP-bound state, when there is weaker tubulin binding.
• The central beta sheet undergoes a twist that may facilitate MT binding. I skimmed this part.
• The top two PC modes explained a decent amount of overall structure fluctuations.
• Their $\Delta G$ calculation is sensitive to how mobile the model is during the simulation. Here -- using a longer MD simulation than their previous paper -- $\Delta G$ changes a lot because the kinesin is more mobile and moves away from tubulin, decreasing their interaction energy in the ADP-bound state. However, they don't account for entropy in their $\Delta G$ and they note that.
• Beta sheet twist and alpha helix stability are still open questions, as well as neck linker influence on nucleotide binding and MD-activated ADP release.