In this research project, different computational techniques were combined to gain insights into the fluctuations of photosynthetic proteins and their impact on the efficiency of photosynthesis. The major aims of the project were:
1. To build coarse-grained models of the protein machineries involved in plant photosynthesis.
2. To study and characterize the structural fluctuations of the protein machineries in their natural environment – the thylakoid membrane – on the microsecond time scale.
3. To gain new insights about the impact of the proteins’ structural fluctuations on their photosynthetic properties.
With respect to (1), I optimized a protein model for the antenna protein light-harvesting complex II (LHCII) in the framework of the coarse-grained Martini model. The Martini model still operates at almost atomistic resolution but offers an about 1000-fold speed-up compared to atomistic simulations. To improve the capability of the model, I used a combination of the Martini model with an approach to describe contact-specific dynamics in proteins, so-called Gō models. To make this method applicable for the large-scale simulations of this research project, I developed two major improvements. These improvements were crucial to efficiently use the method on supercomputers which are indispensable for this kind of research.
The new protein model of light-harvesting complex II allowed studying in detail its structural fluctuations in order to achieve aim (2). I could show that the chlorophylls embedded in the antenna protein are sensitive to their position within the protein: The closer they are to the membrane interface, the higher are their structural fluctuations. Moreover, the chlorophylls are also sensitive to the environment of the protein. If the antenna protein interacts with other proteins, e.g. proteins to which the light energy should be transmitted, the chlorophylls close to the interface get stabilized. The simulations also unravelled the preferred lipid environment of LHCII, its lipid fingerprint.
The diversity of the chlorophyll arrangements obtained from the performed coarse-grained simulations served as basis to work towards aim (3). Using tools from machine learning, I extracted representative arrangements. These arrangements representing the full diversity present in the coarse-grained simulations were used to calculate the light absorption properties of the proteins by means of quantum mechanical simulations. The employed method relies on excitonic Hamiltonians which take into account the chlorophyll arrangements. My simulations have shown that the different chlorophyll arrangements exhibit different light absorption which potentially could be measured in experiments.
The obtained results were disseminated in different ways: First, I presented my results at 14 scientific conferences in form of five talks and nine poster presentations. This included national meetings like the “DutchBiophysics” as well as leading international conferences like the “XXI International Conference on Ultrafast Phenomena”. Second, parts of the results are already published in two peer-reviewed articles, one preprint, and two conference papers. More peer-reviewed articles are in progress. Third, I will present part of my research results to a non-scientific broader audience in a public talk at the Nationalpark-Haus in St. Peter-Ording, Germany.
