Machine learning for metabolic engineering: A review.

Machine learning provides researchers a unique opportunity to make metabolic engineering more predictable. In this review, we offer an introduction to this discipline in terms that are relatable to metabolic engineers, as well as providing in-depth illustrative examples leveraging omics data and improving production. We also include practical advice for the practitioner in terms of data management, algorithm libraries, computational resources, and important non-technical issues. A variety of applications ranging from pathway construction and optimization, to genetic editing optimization, cell factory testing, and production scale-up are discussed. Moreover, the promising relationship between machine learning and mechanistic models is thoroughly reviewed. Finally, the future perspectives and most promising directions for this combination of disciplines are examined.

Amyloid Peptide Scaffolds Coordinate with Alzheimer's Disease Drugs.

Functional amyloid materials can combine the self-assembly of peptide scaffolds into amyloid fibrils with binding capacities for ions or compounds of pharmaceutical interest, endowed by mutable non-ß-sheet-forming residues at the termini. Herein, we report the first to our knowledge amyloid materials, encompassing a GAIIG amyloidogenic core, which bind to Alzheimer's disease (AD) drugs, by mimicking the mechanism by which the same AD drugs bind to enzymes according to experimentally resolved structures, including the target enzyme acetylcholinesterase (AChE). The computationally designed amyloid scaffolds are experimentally shown to coordinate with AD drugs, using two techniques, both in dilute solutions and at higher peptide concentrations, with a higher binding capacity for donepezil and tacrine compared to that for memantine and galantamine. The binding for some of the AD drugs is strong and stable even after extensive subsequent aqueous washings, denoting high capturing efficiency by the designed biomaterials, even after incubation under physiological conditions. Our findings constitute starting points to design novel drug delivery carriers binding to one or combinations of AD drugs (e.g., NMDA and cholinesterase inhibitors).

Designer Amyloid Cell-Penetrating Peptides for Potential Use as Gene Transfer Vehicles.

Cell-penetrating peptides are used extensively to deliver molecules into cells due to their unique characteristics such as rapid internalization, charge, and non-cytotoxicity. Amyloid fibril biomaterials were reported as gene transfer or retroviral infection enhancers; no cell internalization of the peptides themselves is reported so far. In this study, we focus on two rationally and computationally designed peptides comprised of ß-sheet cores derived from naturally occurring protein sequences and designed positively charged and aromatic residues exposed at key residue positions. The ß-sheet cores bestow the designed peptides with the ability to self-assemble into amyloid fibrils. The introduction of positively charged and aromatic residues additionally promotes DNA condensation and cell internalization by the self-assembled material formed by the designed peptides. Our results demonstrate that these designer peptide fibrils can efficiently enter mammalian cells while carrying packaged luciferase-encoding plasmid DNA, and they can act as a protein expression enhancer. Interestingly, the peptides additionally exhibited strong antimicrobial activity against the enterobacterium Escherichia coli.

Elucidating the multi-targeted anti-amyloid activity and enhanced islet amyloid polypeptide binding of ß-wrapins.

ß-wrapins are engineered binding proteins stabilizing the ß-hairpin conformations of amyloidogenic proteins islet amyloid polypeptide (IAPP), amyloid-ß, and α-synuclein, thus inhibiting their amyloid propensity. Here, we use computational and experimental methods to investigate the molecular recognition of IAPP by ß-wrapins. We show that the multi-targeted, IAPP, amyloid-ß, and α-synuclein, binding properties of ß-wrapins originate mainly from optimized interactions between ß-wrapin residues and sets of residues in the three amyloidogenic proteins with similar physicochemical properties. Our results suggest that IAPP is a comparatively promiscuous ß-wrapin target, probably due to the low number of charged residues in the IAPP ß-hairpin motif. The sub-micromolar affinity of ß-wrapin HI18, specifically selected against IAPP, is achieved in part by salt-bridge formation between HI18 residue Glu10 and the IAPP N-terminal residue Lys1, both located in the flexible N-termini of the interacting proteins. Our findings provide insights towards developing novel protein-based single- or multi-targeted therapeutics.

Computational Design of Functional Amyloid Materials with Cesium Binding, Deposition, and Capture Properties.

Amyloid materials are gaining increasing attention as promising materials for applications in numerous fields. Computational methods have been successfully implemented to investigate the structures of short amyloid-forming peptides, yet their application in the design of functional amyloid materials is still elusive. Here, we developed a computational protocol for the design of functional amyloid materials capable of binding to an ion of interest. We applied the protocol in a test case involving the design of amyloid materials with cesium ion deposition and capture properties. As part of the protocol, we used an optimization-based design model to introduce mutations at non-ß-sheet residue positions of an amyloid designable scaffold. The designed amino acids introduced to the scaffold mimic how amino acids bind to cesium ions according to experimentally resolved structures and also aim at energetically stabilizing the bound conformation of the pockets. The optimum designs were computationally validated using a series of simulations and structural analysis to select the top designed peptides, which are predicted to form fibrils with cesium ion binding properties for experimental testing. Experiments verified the amyloid-forming properties of the selected top designed peptides, as well as the cesium ion deposition and capture properties by the amyloid materials formed. This study demonstrates the first, to the best of our knowledge, computational design protocol to functionalize amyloid materials for ion binding properties and suggests that its further advancement can lead to novel, highly promising functional amyloid materials of the future.

A novel amyloid designable scaffold and potential inhibitor inspired by GAIIG of amyloid beta and the HIV-1 V3 loop.

The GAIIG sequence, common to the amyloid beta peptide (residues 29-33) and to the HIV-1 gp120 (residues 24-28 in a typical V3 loop), self-assembles into amyloid fibrils, as suggested by theory and the experiments presented here. The longer YATGAIIGNII sequence from the V3 loop also self-assembles into amyloid fibrils, of which the first three and the last two residues are outside the amyloid GAIIG core. We postulate that this sequence, with suitably selected modifications at the flexible positions, can serve as a designable scaffold for novel amyloid-based materials. Moreover, we report the single crystal X-ray structure of the beta-breaker peptide GAIPIG at 1.05 Å resolution. The structural information provided in this study could serve as the basis for structure-based design of potential inhibitors of amyloid formation.

Self-Assembled Amyloid Peptides with Arg-Gly-Asp (RGD) Motifs As Scaffolds for Tissue Engineering.

Self-assembled peptides gain increasing interest as biocompatible and biodegradable scaffolds for tissue engineering. Rationally designed self-assembling building blocks that carry cell adhesion motifs such as Arg-Gly-Asp (RGD) are especially attractive. We have used a combination of theoretical and experimental approaches toward such rational designs, especially focusing on modular designs that consist of a central ultrashort amphiphilic motif derived from the adenovirus fiber shaft. In this study, we rationally designed RGDSGAITIGC, a bifunctional self-assembling amyloid peptide which encompasses cell adhesion and potential cysteine-mediated functionalization properties through the incorporation of an RGD sequence motif and a cysteine residue at the N- and C- terminal end, respectively. We performed replica exchange MD simulations that suggested that the key factor determining cell adhesion is the total solvent accessibility of the RGD motif and also that the C-terminal cysteine is adequately exposed. The designer peptides self-assembled into fibers that are structurally characterized with Transmission Electron Microscopy, Scanning Electron Microscopy and X-ray fiber diffraction. Furthermore, they supported cell adhesion and proliferation of a model cell line. We consider that the current bifunctional properties of the RGDSGAITIGC fibril-forming peptide can be exploited to fabricate novel biomaterials with promising biomedical applications. Such short self-assembling peptides that are amenable to computational design offer open-ended possibilities toward multifunctional tissue engineering scaffolds of the future.