We present a computational study that links the design of collagen-like peptide (CLP) heterotrimers with ‘sticky’ ends (i.e., free ends with hydrogen-bonding groups) to the thermal stability of the heterotrimeric triple helix in aqueous solutions at low CLP concentrations and their self-assembly into fibrillar aggregates and percolated networks. Unlike CLP homotrimeric helices, the CLP heterotrimeric triple helices are made of CLP strands of different chain lengths that result in ‘sticky’ ends with available hydrogen bonding groups. These ‘sticky’ ends then facilitate inter-helix hydrogen bonding leading to higher order assembly into fibrils and percolated networks. To quantify thermal stability of the CLP heterotrimeric helices, we use coarse grained (CG) molecular dynamics (MD) simulations at low CLP concentration in implicit water to generate melting curves for CLP heterotrimers as a function of sticky end length, number of sticky ends, and number of repeat units in the CLP. We find that increasing sticky end length destabilizes the triple helix, resulting in lower melting temperatures. At higher CLP concentrations, for CLP heterotrimers with one sticky end, we observe non-monotonic trends in cluster sizes with increasing sticky end length; in contrast, for CLP heterotrimers with two sticky ends we observe a monotonic trend. The nonmonotonicity stems from the formation of turn structures stabilized by hydrogen bonds at the single, sticky end which result in smaller CLP assemblies. At higher concentrations, CLP heterotrimers form percolated networks with increasing sticky end length such that a minimum sticky end length of four repeat units was required to observe percolation. Moreover, two sticky ended designs formed percolated networks with a greater number of helices per cluster as compared to a one sticky ended design. Using these results from CD MD simulations we then train a machine learning model so we can predict the finite cluster formation vs. percolated networks for a given CLP triple helix design without having to do extensive CG-MD simulations. Overall, this computational work highlights the predictive capabilities of simulations and machine learning for guiding the design of thermoresponsive, peptide-based biomaterials with unique, yet complex morphologies.

Computational design of collagen-like-peptides (CLP) for desired CLP triple helix melting transition and assembled structure

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.