Introduction: An attractive approach to central nervous system (CNS) tissue regeneration involves the development of biomaterials that mimic ECM features while supporting cell viability and providing suitable topology for controlled organization of tissue. Hydrogels are ideal tissue engineering materials due to their physical similarities to native soft tissues, such as their high water content and mechanical properties [1],[2]. We have designed a functionalized laminin-mimetic elastin-like polypeptide (FLAME) fusion protein that conserves the entire laminin globular domain 5 (LG5) and utilizes elastin-like peptide (ELP) repeat units to modulate the fusion proteins’ intra- and intermolecular behavior. Classical, all-atom molecular dynamics (MD) simulations [3] are used to examine the conformational and structural dynamics of the engineered FLAME fusion protein.
Materials and Methods: The LG5 domain is derived from the c-terminus of the native laminin gamma chain. This domain is responsible for much of the cell-ECM interaction in the CNS, including integrin binding. The ELP is a 8X repeat of (VPGXG) where X is represented by L, I, or K residues. We utilized NAMD on the 6000+ core UVA Rivanna computational cluster. Our simulation system scales well up to 500 cores and thus was run on a parallel queue, with 500 cores for 10 ns, 5,000,000 time steps. NAMD is a well-established and feature-rich MD codebase for simulations of biomolecular systems [4]. NAMD is designed for use on supercomputers, as it requires near-constant neighbor-to-neighbor communication and frequent all-to-all communication.
Results and Discussion: FLAME modeled the backbone of the LG antiparallel β-sandwich framework as a rigid body, and treated the ELP sequences attached to the N- and/or C-terminal domain of LG5 as flexible unstructured coils. The ELP sequences remained free of the LG domain under all simulation conditions. We conducted simulations of temperature- and solvent-equilibrated polypeptides at temperatures above the predicted lower critical solution temperature (LCST; Tt = 22 °C). The MD trajectories of FLAME (310 K, liquid H2O) enable the frequency of occurrence of secondary structural motifs for each residue in the ELP region to be computed, as shown in Figure 1. The Pro2 and Gly are the most dynamic sites in this peptide, where they predominantly sample canonical type II β-turn (ϕ = -60°, ψ = 120°) and poly-proline helical (ϕ = -75°, ψ = 150°) regions of conformational space. In general, the observed (high) fraction of β-sheet structural content at high temperatures can be attributed to the conformational dynamics of the Pro2, Gly3 and Gly5 residues. As expected, we did not detect any β-sheet formation at 290K, however, residues begin to adopt β-sheet conformations at 310K (Figure 2) after 60 ns of simulation.
Conclusions: The work presented here relies on atomically-detailed MD simulations to provide a comprehensive, physics-based analysis of the temperature-dependence of both the tertiary and secondary structures of our engineered FLAME protein systems. The developing ELP β-sheet indicates intermolecular assembly will occur. The computational framework serves as a robust and extensible platform for future work (both experimental and computational) toward our overarching goal of synthesizing self-assembling, protein-based biomaterials as stem cell therapeutics.
Cam Mura and Charles McAnany of the Mura lab for assistance in developing NAMD code and project guidance
References:
[1] F. Brandl, F. Sommer, A. Goepferich. Biomaterials 28 (2007) 134–146.
[2] K.Y. Lee, D.J. Mooney. Chem. Rev. 101 (2001) 1869–1879.
[3] Philips J. C et al. J. Comp. Chem. 26 (2005) 1781-1802
[4] Mura, McCammon, Molecular Simulation, 40 (2014) 732-764