Multi-scale Molecular Simulations on Long-Range Conformational Switching and ATPase Activity of Hsp90

Principal Investigator:
Ville R. I. Kaila

Department of Chemistry, Technical University of Munich (Germany)

Local Project ID:

HPC Platform used:
SuperMUC and SuperMUC-NG of LRZ

Date published:


In proteins, biological signals can be transferred over large molecular distances and can thereby enhance or inhibit the protein activity via long-range coupling mechanisms. Such allosteric effects regulate the function of all kinds of proteins, from molecular chaperones and kinases involved in cell signalling, to redox enzymes and ion pumps [1]. An input signal, e.g., a post-translational modification or binding of a ligand, can be propagated through the protein via both electrostatic and conformational changes of central amino acids. Mutation of the amino acids involved in the signal propagation pathway may change the protein’s activity.

The molecular chaperone Hsp90 (Figure 1) is involved in the maturation of hundreds of substrate proteins and regulates important cellular pathways in the eukaryotic cell. Amongst these client proteins of Hsp90 are transcription factors and protein kinases that control, e.g., cell growth. Some of these clients play a role in human diseases, including cancer and Alzheimer’s disease. Hsp90 has thus evolved into a promising target for drug development [2].

Hsp90 functions as a homodimer that undergoes large conformational changes from an inactive, open state to an active, closed state, constituting its chaperone cycle that is coupled to a slow ATPase activity. To gain insight into the signaling pathways within the chaperone Hsp90 and their effect on its catalytic activity, we performed both quantum chemical and classical atomistic simulations in combination with biophysical and biochemical experiments to elucidate the molecular principles responsible for the function of Hsp90.

Results and Methods

To investigate the connection between conformation and ATPase activity of Hsp90, we performed classical atomistic molecular dynamics (MD) simulations to explore the conformational dynamics, and we calculated free energy profiles for the ATP hydrolysis reaction in different conformational states by employing hybrid quantum mechanical/classical mechanical (QM/MM) umbrella sampling simulations.

We found that conformational changes of a glutamate/arginine ion-pair (R32/E33) in the ATP binding site of Hsp90 influence the free energy barrier of the ATPase reaction by electrostatic tuning effects (Figure 2a).Substitution of the computationally identified switching residue by a neutral amino acid (R32A) led to an Hsp90 variant that decouples the global conformational state from the catalytic activity (Figure 2b, c) [3].

We also studied a post-translationally modified lysine that controls the function of Hsp90. Methylation of the lysine affects the conformational state and catalytic activity of Hsp90, suggesting signal propagation across large distances. MD simulations showed that methylation of the lysine could induce conformational changes in a neighboring ion-pair (Figure 3a, b), which might propagate through the protein. Our studies indicated that the methyl-lysine can be mimicked by an isoleucine [4].

To computationally study the conformational dynamics of Hsp90, we performed MD simulations on microsecond time scales using the software package NAMD [5]. We employed periodic boundary conditions (PBC) and used a 2 fs integration time step. Starting from available crystal structures of the active state of Hsp90, we investigated structural effects of amino acid modifications in monomeric and dimeric models. All structural models were extracted from the Protein Data Bank (PDB) and embedded in a water-ion environment to mimic biological conditions. We typically used 800-1000 cores per classical MD simulation.

Multiple restrained QM/MM simulations of 1 ps each were performed to estimate free energy profiles for ATP hydrolysis in Hsp90 using umbrella sampling. The simulations were performed in parallel to each other, with each simulation using one node. QM/MM calculations were performed by coupling together the software packages CHARMM (for classical MD simulations) and TURBOMOLE (for quantum chemical calculations) using a Python interface. The simulations were propagated for maximum 48 hours and were restarted from the latest checkpoint until the desired simulation time was reached. All in all, we used 10 Mio CPUh for this project, requiring 2 TB of storage in the WORK partition of SuperMUC. The work resulted in two publications, both of which were published in Nature Communications [3,4].

On-going Research / Outlook

In this project, SuperMUC provided us with the necessary computing resources to study the structure, function, and dynamics of the Hsp90 chaperone. This allowed us to investigate important structural changes in different variants and models of the Hsp90 dimer in atomic detail, and to probe the connection between protein conformation and catalytic activity. Future work will focus on characterizing the structural ensemble of Hsp90 in different nucleotide states and with various point mutations of previously identified amino acid residues that stabilize the closed conformation of the dimer. In combination with biophysical experiments, we will investigate structural determinants involved in the large-scale conformational changes leading from the open to closed state of the Hsp90 dimer.

References and Links


[2] Biebl MM, Buchner J. Structure, Function, and Regulation of the Hsp90 Machinery. Cold Spring Harb. Perspect. Biol. 11: a034017 (2019). DOI: 10.1101/cshperspect.a034017

[3] Mader SL, Lopez A, Lawatscheck J, Luo Q, Rutz DA, Gamiz-Hernandez AP, Sattler M, Buchner J, Kaila VRI. Conformational dynamics modulate the catalytic activity of the molecular chaperone Hsp90. Nat. Commun.11, 1410 (2020). DOI: 10.1038/s41467-020-15050-0

[4] Rehn A, Lawatscheck J, Jokisch M-L, Mader SL, Luo Q, Tippel F, Blank B, Richter K, Lang K, Kaila VRI, Buchner J. A methylated lysine is a switch point for conformational communication in the chaperone Hsp90. Nat. Commun.11, 1219 (2020). DOI: 10.1038/s41467-020-15048-8

[5] Phillips JC, Hardy DJ, Maia JDC, et al. Scalable molecular dynamics on CPU and GPU architectures with NAMD. J Chem Phys. 2020;153(4):044130. doi:10.1063/5.0014475

Research Team

Ville R. I. Kaila (PI), Ana P. Gamiz-Hernandez, Qi Luo, Sophie L. Mader
All: Department of Chemistry, Technical University of Munich (TUM)

Scientific Contact

Prof. Dr. Ville R. I. Kaila
Department of Chemistry
Technical University of Munich (TUM)
Lichtenbergstr. 4, D-85747 Garching (Germany)
e-mail: ville.kaila [@]

Local project ID: pr53po

January 2021

Tags: TUM LRZ Life Sciences