Self-assembling peptide nanostructures for controlled cell-material interactions
Loading...
Date issued
Authors
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Reuse License
Abstract
The ability of peptides to self-assemble into complex hierarchical supramolecular structures is the basis for their diverse structural functions in Nature. This inspired researchers to design peptides from bottom-up to employ them as versatile scaffolds for cell-interactive materials in biomedical contexts. An example for applying this concept is utilizing peptide nanofibrils for therapeutic retroviral gene delivery. However, it is difficult to a priori design the bioactivity of a peptide, as even small changes in the peptide’s sequence can drastically impact its self-assembly behavior and physicochemical properties on multiple length scales. The combination of experimental and computational approaches holds great promise to overcome these challenges in the quest for functional materials design for biomedical applications.
This thesis provides fundamental mechanistic insights for the interactions between peptide nanofibrils with cells and viruses together with strategies for peptide materials design in the context of clinical applications. Viral infectivity as a biological readout was chosen not only because of its clinical relevance but also because it offers a relatively simple readout to unravel the multi-parameter, multi-scale challenges for the identification and optimization of biologically active supramolecular peptide assemblies. The investigation of the peptide–cell–material interaction in this work can be divided into four main parts. The thesis begins with (I) optimizing the sequence of the enhancing factor C (EF-C) peptide that binds virions and facilitates gene delivery applications. This is followed by a (II) comprehensive investigation of essential physicochemical features, which are subsequently employed in (III) machine-learning assisted discovery of novel peptides. Finally, (IV) the application of peptides is explored with a focus on translational clinical aspects.
First, a peptide library was established by designing and synthesizing 163 short peptides based on the known infectivity-enhancing self-assembling peptide EF-C. To systematically study the structure-property-activity relationship of the obtained bioactive sequences, a data-mining strategy was applied to correlate various physicochemical, bioinformatic and biological properties of the peptides. Thereby, it was shown that peptides forming µm-sized -sheet-rich aggregates are essential for promoting interactions of retroviruses with cells. These key-properties are also observed in short amyloidal peptides from diverse pathological and functional origin which emphasizes the universality of this requirement. To gain mechanistic insights of the peptide aggregation process and supramolecular fibril formation, molecular dynamics simulations were performed and revealed that fibril–fibril supramolecular networks are observed for non-dynamic, stable fibrils with low solvent accessible surface areas, i.e. high fibril surface hydrophobicity. Within the EF-C based library, peptides with an alternating amphiphilic sequence and a high ratio of hydrophobic amino acids exhibit a strong propensity to form µm-sized aggregates. This sequence pattern proved to be a reliable strategy for optimizing the sequence of EF-C and rationally design novel bioactive peptide fibrils.
Next, a machine learning workflow was performed on the EF-C based library to inverse-engineer de novo minimalistic peptides via continuous vector representations. The conceptual idea behind this approach is that relevant sequence information underlying infectivity enhancement can be extracted and decoded from a higher-dimensional vector space and this information can be applied for the design of new sequences. With this approach, a new sequence space of self-assembling peptides was discovered. The results show that infectivity enhancement is governed by both electrostatic and hydrophobic contributions. As these newly found peptides display unexpected sequences that are not related to the original EF-C peptide, nor the derived peptide library and they were also not found in Nature, this machine-learning approach presents an efficient strategy to discover entirely new self-assembling peptide sequences with aggregation propensity and high bioactivity.
In the next step, these newly acquired insights about the impact of molecular design on aggregation properties were employed to develop biodegradable peptide fibrils suitable for ex vivo clinical applications. Specifically, peptide amphiphiles for viral gene delivery were evaluated. In contrast to the metabolically very stable amyloid-type peptides that contain an alternating amphiphilic sequence pattern, peptide amphiphiles are characterized by their block-type design and flexibility to customize fibril surface properties. By systematically studying a library composed of 35 peptide amphiphiles, it was found that peptide amphiphiles that form small aggregates can promote virus-cell interactions while being biodegradable. Via molecular dynamics simulations it was shown that their aggregation is dependent on the mobility of monomers in fibrils and surface hydrophobicity that is governed by hydrophobicity, flexibility, and order of amino acids within the peptide sequence. The findings from the PA library are well in line with amyloid-type peptides and emphasize the general importance of the key-physicochemical properties for promoting viral infectivity, i.e. positively charged, µm-sized aggregates formed by hydrophobic, -sheet rich structures.
Finally, to showcase how these properties can be effectively harnessed to regulate interactions between fibrils and cells, a photo-decomposable peptide was engineered. The decomposition was achieved by incorporating a photosensitive linker into the peptide backbone that causes fragmentation of the alternating amphiphilic peptide sequence. This, in turn, results in decomposition of the β-sheet ordered supramolecular assembly and loss of its cell-adhesive properties. Due to the external trigger light, this process can be precisely controlled in space and time, allowing the manufacturing of cell-adhesive regions on surfaces.
Overall, the powerful combination of physicochemical and computational methods can circumvent the individual limitations of both approaches. The findings presented in this thesis showcase the bottom-up design and discovery of new functional peptide nanomaterials for biomedical applications.