Molecular Dynamics

The minicellulosome complex we use in our project contains a cellulose binding domain (CBD) which is thought to enhance activity of the enzyme complex. By binding the whole scaffold to cellulose, the efficiency of all enzymes is supposed to increase due to substrate proximity. However it is currently unknown what exactly the mechanism is for CBD binding to cellulose.

The CBD structure we are using was crystallized by Tormo et al. [1]. They hypothesize that the binding interface between CBD and cellulose must be at the planar strip of aromatic residues on one side of the beta-sheet sandwich. It is thought that the CBD binds by stacking these aromatic amino acids on the ring structure of the glucose monomers in cellulose, as this also happens with other known cellulose binding domains.

To verify the binding of CBD to cellulose, and to validate the proposed mechanism of binding, molecular dynamics modeling (MD) was used to simulate the binding of the CBD to cellulose. MD is ideally suited for subjects involving biomolecules as these are often difficult to visualize experimentally. In particular, version 2 of the MARTINI forcefield is used [2]. The forcefield in MD describes the interactions between all the different particles. Many different forcefields exist, but the coarse grained MARTINI forcefield is ideal for cheap, lengthy simulations of proteins. More recently polysaccharides have been properly parameterized for Martini 2 as well [3].

To start, the CBD structure from Tormo et al. was used for its structural information. The structure was then coarse grained using the martinize tool. An elastic network was added to preserve the secondary structure of the protein, at a cutoff of 0.5 nm. The cutoff was chosen to minimize artificial restraints, while keeping the structure of the beta sheets intact.

Next, the cellulose model from Lopez et al. [4] was used to model cellulose in MARTINI 2. An infinite fiber spanning over the periodic boundary condition was realized by adding bonds between the first and last glucose monomers, effectively realizing an infinite fiber.

Cellulose and CBD were put together in the same simulation and water and ions were added (150mM NaCl). As Martini 2 water is represented as one spherical bead for every 4 water molecules, it has a tendency to pack in a crystalline lattice structure. This system in particular had a high tendency to freeze, even at 30℃. Most likely this is due to the regular crystalline structure of the cellulose acting as a nucleation point for crystallization of the water beads. To prevent this, 10% of antifreeze beads were added. Antifreeze beads act just like regular water, except other water beads perceive them as slightly larger, thus disrupting crystalline packing. This should not impact the accuracy of the simulation.

Common parameters for martini simulations were used for dealing with electrostatics [5]. Temperature coupling was achieved by using the Berendsen thermostat at a ref-t of 303.15 K. Pressure coupling was done semiisotropically with Parenello-Rahman at a ref-p of 1.0 bar. While the compressibility in the xy plane was set to 3e-4, the compressibility in the z-direction was set to 0. This was done to keep the infinite fiber in tact and to keep the box from collapsing. The simulations were run for at least 400ns which should provide plenty of sampling. This took about 6 hours of wall time on the Groningen University’s Peregrine cluster.

4 simulations were setup where the CBD was inserted in different orientations around the cellulose fiber. Figure 1 shows the distance between CBD and cellulose over time. In all 4 simulations CBD binds to cellulose relatively quickly and never leaves.

Figure 1: Distance between CBD and cellulose over time. Every run is plotted separately, all bind to cellulose and never leave within the timeframe of the simulation.

With adequate sampling, a contact map between CBD and cellulose can be computed (Figure 2). By calculating a distance matrix between CBD and cellulose, and averaging this matrix, a visual representation of how CBD binds to cellulose on average. By taking the minimum distance along one axis of this matrix, a plot can be generated showing which residues of CBD are on average closer to cellulose (Figure 3).

Figure 2: Distance matrix between the backbone beads of CBD and all cellulose atoms. Distance (nm) is represented by yellow (close) to blue (far). Patches between indices 45-70 and 100-120 of CBD show the closest distance.

Figure 3: The average distance to cellulose for every backbone bead in the sequence of CBD. There is a clear contact patch around indices 56-67 and around index 112.

When these close residues are colored on the atomistic structure (Figure 4), it is striking to see that they correspond to planar strip that was hypothesized by Tormo et al. to bind to cellulose. Furthermore this binding patch contains many aromatic amino acids as was also hypothesized.

Figure 4: Structural visualization of the binding patch of CBD as predicted by Figure 3. Yellow beads are cellulose, the protein backbone of CBD is colored cyan. The space filling protein sequence represents the lowest valleys in the by-sequence distance data. This patch includes 4 of the 5 binding residues as hypothesized by Tormo et al. and contains many aromatic residues.

To gain more insight in the hypothetical molecular interactions, the atomistic structure of CBD was overlaid over the coarse grained structure, yielding Figure 5. When residues 118, 112, 56, 57 and 67 are visualized as “sticks”, it is striking to see that they would be in contact with the cellulose fiber exactly in the way as described by Tormo et al. This result strongly implies that these residues are indeed responsible for binding to cellulose, at the very least to straight, singular fibers. The planar patch of amino acids likely maximizes the contact area between CBD and cellulose, providing a strong binding affinity between the two.

Figure 5: Interface between cellulose and CBD. Non binding CBD residues are not visible. The backbone of the protein is in cyan, cellulose is in yellow. The binding patch of CBD consists of Trp 118, Arg 112, Asp 56, His 57 and Tyr 67.

Additionally a movie was rendered illustrating the strong binding between CBD and cellulose. The movie shows the separate CBD floating in aqueous solution. After some time it finds the cellulose fiber and binds at the binding site as hypothesized by Tormo et al.

In conclusion, CBD binding to cellulose was modeled using coarse grained molecular dynamics. The results verify that CBD indeed binds to cellulose using the mechanism described by Tormo et al. The binding seems to be very strong as the simulations show CBD never leaves cellulose after it has bound. This observation was verified by PMF calculations, which determined the binding free energy between CBD and cellulose to be -16.49 kCal/mol. These results justify keeping the CBD in our minicellulosome as the strong binding to cellulose will most likely enhance activity of the complex by proximity to cellulose.

References

[1] Tormo, J., Lamed, R., Chirino, A. J., Morag, E., Bayer, E. A., Shoham, Y., & Steitz, T. A. (1996). Crystal structure of a bacterial family-III cellulose-binding domain: a general mechanism for attachment to cellulose. The EMBO Journal, 15(21), 5739–5751. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8918451

[2] Monticelli, L., Kandasamy, S. K., Periole, X., Larson, R. G., Tieleman, D. P., & Marrink, S. J. (2008). The MARTINI coarse-grained force field: Extension to proteins. Journal of Chemical Theory and Computation, 4(5), 819–834. https://doi.org/10.1021/ct700324x

[3] Schmalhorst, P. S., Deluweit, F., Scherrers, R., Heisenberg, C. P., & Sikora, M. (2017). Overcoming the Limitations of the MARTINI Force Field in Simulations of Polysaccharides. Journal of Chemical Theory and Computation, 13(10), 5039–5053. https://doi.org/10.1021/acs.jctc.7b00374

[4] López, C. A., Bellesia, G., Redondo, A., Langan, P., Chundawat, S. P. S., Dale, B. E., … Gnanakaran, S. (2015). MARTINI coarse-grained model for crystalline cellulose microfibers. Journal of Physical Chemistry B. https://doi.org/10.1021/jp5105938

[5] De Jong, D. H., Baoukina, S., Ingólfsson, H. I., & Marrink, S. J. (2016). Martini straight: Boosting performance using a shorter cutoff and GPUs. Computer Physics Communications. https://doi.org/10.1016/j.cpc.2015.09.014