Kinetics and Thermodynamics of Conformational Changes Upon Protein Association Studied by Molecular Dynamics Simulations

Principal Investigator:
Martin Zacharias

Lehrstuhl für Molekulardynamik, Physik-Department T38, TU München (Germany)

Local Project ID:

HPC Platform used:
SuperMUC of LRZ

Date published:

Leveraging the computing power of HPC system SuperMUC, researchers of the Technische Universität München investigated the free energy landscape for large-scale conformational changes coupled to the association of biomolecules. It allowed understanding the mechanism of substrate and inhibitor binding to the adenylate kinase (ADK) enzyme and helped to characterize the thermodynamics and kinetics of the propagation of Alzheimer Alzheimer Aβ9-40 amyloid fibrils.


Proteins and peptides are essential components of basically all biological processes. During association, protein molecules can undergo a variety of conformational changes essential for its function. Upon association such transitions can involve global domain motions for example of enzymes to switch between open accessible but inactive and closed active states. The enzyme, the adenylate kinase (ADK), consists of several semi-flexible segments or domains that rearrange from an open state to a closed enzymatically active state upon substrate binding (illustrated in Figure 1). The open conformation is easily accessible for the substrate and transitions to a closed state result in catalytic activation with the substrate largely buried in the protein molecule [1].

In order to elucidate how the lid-domain motion couples to binding of substrate and inhibitor molecules, the scientists employed Molecular Dynamics (MD) free energy simulations using the position of the two lid domains as independent reaction coordinates [1].

In addition to global motions, association of proteins and peptides can also involve refolding of peptide segments. For example, certain peptides form stable aggregates in solution called also fibrils or amyloids [2]. Such amyloid fibrils play a major role in several diseases of which the Alzheimer disease is probably the best known example. Although the structure of several peptide amyloids has been elucidated by experimental methods, the mechanism of formation is not very well understood. Using a similar methodology as for studying global motions in ADK, the researchers characterized the thermodynamics and kinetics of the propagation of Alzheimer Aβ9-40 amyloid fibrils.

Results and Methods:

Global domain motion in Adenylate Kinase

Apart from the two lid domains (termed ATP-lid and AMP-lid), the ADK enzyme contains a central core domain. In order to systematically investigate the lid domain mobility, the distance of each lid to the central domain served as reaction coordinate.

Using a two-dimensional (2D) umbrella sampling protocol coupled with replica exchanges, it was possible to calculate a complete 2D-free energy landscape for the lid domain motions in response to different bound substrates and inhibitors.

The coupling of umbrella sampling with replica exchange resulted in rapid convergence of the calculated free energy surface and showed very good scaling on a parallel supercomputer. The study is the first to investigate this free energy landscape for all possible substrate and inhibitor bound states of ADK and required >5 million SuperMUC core hours. It demonstrated a dramatic dependence of the lid-motion on the binding state (see Figure 1).

The simulation results helped to explain why for this type of enzyme two lid domains have evolved to efficiently bind and catalyze a reaction requiring two substrate molecules and to elucidate the most likely order of binding events [1]. In collaboration with an experimental single-molecule group (Rief group, TUM Physics Department) the study also indicated the stabilization of a half open ADK state adopted in the presence of certain inhibitor molecules [3].

Propagation of Alzheimer Aβ9-40 amyloid fibrils 

In a second subproject the researchers investigated the propagation step of the Alzheimer Aβ9-40 amyloid fibril formation (illustrated in Figure 2). Using extensive Molecular Dynamics simulations combined with umbrella sampling the scientists were able to study the thermodynamics and kinetics of fibril extension in good agreement with available experimental data [2].

The simulations identified the entropy change of water as a main driving force for association and interactions between the already formed fibril and a new monomer to overall disfavor association. The simulations also confirmed a dock/lock mechanism of association with a rapid docking phase, characterized also by many non-native contacts followed by a slow locking step which finally leads to the native peptide association mode (Figure3).

It was also possible to estimate the diffusion profile for the monomer approaching the fibril tip and to estimate the kinetics of fibril propagation.

In order to achieve sufficient convergence of calculated free energies and other physical properties, it was necessary to extend the simulations to several hundred nanoseconds per umbrella sampling window adding up to several microseconds total simulation time only feasible on the SuperMUC computer.

On-going Research/Outlook

The ADK enzyme is an excellent model system for studying the coupling between the enzymatic function of the protein and the global motion induced by substrate binding. The applied coupled umbrella sampling and replica exchange methodology shows excellent scaling on parallel supercomputers. In future research the scientists plan to study the influence of mutations on the energy landscape of domain motions in ADK. The simulation studies on amyloid formation and propagation will be extended to the study of inhibitors of amyloid propagation in collaboration with experimental groups at TUM. Both the systematic studies of global lid domain motions in the enzyme ADK but also the simulations of amyloid propagation were only possible by using the SuperMUC parallel computer facilities.


[1] Zeller F, Zacharias M. Substrate Binding Specifically Modulates Domain Arrangements in Adenylate Kinase. Biophys. J 109 (2015) 1978-85.

[2] Schwierz N, Frost CV, Geissler PL, Zacharias M. Dynamics of Seeded Aβ40-Fibril Growth from Atomistic Molecular Dynamics Simulations: Kinetic Trapping and Reduced Water Mobility in the Locking Step. J Am Chem Soc. 138 (2016) 527-39.

[3] Pelz B, Žoldák, Zeller F, Zacharias M, Rief M. Sub-nanometer enzyme mechanics probed by single-molecule force spectroscopy. Nature Commun. 7 (2016) 10848.

Research Team:

Rainer Bomblies, Christina Frost, Florian Kandzia, Alexander Knips, Manuel Luitz, Giuseppe La Rosa, Maria Reif, Christina Schindler, Nadine Schwierz, Martin Zacharias (PI), Fabian Zeller

Scientific Contact:

Prof. Dr. Martin Zacharias
Professor of Theoretical Biophysics
Physik-Department (T38) 
Technische Universität München 
James-Franck-Str. 1, D-85748 Garching (Germany)
e-mail: martin.zacharias [at]

LRZ Project ID: pr84ko

October 2016

Tags: Life Sciences Health and Medicine LRZ