A User's Guide to Machine Learning for Polymeric Biomaterials

The development of novel biomaterials is a challenging process, complicated by a design space with high dimensionality. Requirements for performance in the complex biological environment lead to difficult a priori rational design choices and time-consuming empirical trial-and-error experimentation. Modern data science practices, especially artificial intelligence (AI)/machine learning (ML), offer the promise to help accelerate the identification and testing of next-generation biomaterials. However, it can be a daunting task for biomaterial scientists unfamiliar with modern ML techniques to begin incorporating these useful tools into their development pipeline. This Perspective lays the foundation for a basic understanding of ML while providing a step-by-step guide to new users on how to begin implementing these techniques. A tutorial Python script has been developed walking users through the application of an ML pipeline using data from a real biomaterial design challenge based on group's research. This tutorial provides an opportunity for readers to see and experiment with ML and its syntax in Python. The Google Colab notebook can be easily accessed and copied from the following URL: www.gormleylab.com/MLcolab.

Tetrapeptide-based mimotope affinity monolith for the enrichment and analysis of anti-HER2 antibody and antibody-drug conjugate

Selective enrichment and analysis of therapeutic antibodies in biological fluids are crucial for the development of biopharmaceuticals. Recently, peptide-based affinity chromatography has exhibited fascinating prospects for antibody enrichment due to the high affinity and specificity of small peptides. However, the post-modification approach of peptide ligands on the material surface is complicated and time-consuming. In this study, a methacrylate modified tetrapeptide (m-EDPW) was firstly demonstrated as the affinity ligand of trastuzumab (K_d = 1.91 ± 1.81 μM). Next, the m-EDPW based affinity monolith was prepared using a facile one-step polymerization method, which could overcome the drawbacks of traditional post-modification preparation strategies. Based on the monolith as described above, a simple enrichment approach was developed under the optimal washing and elution conditions. Based on the excellent properties, such as high porosity (53.09%), weak electrostatic interaction and suitable affinity (1.00 ± 2.14 μM for anti-HER2 ADC), this novel monolith exhibited good specificity and recovery for antibodies (91.6% for trastuzumab, 98.37% for anti-HER2 ADC), and low nonspecific adsorption for human serum albumin (DBC_10% = 0.5 mg/g polymer). Particularly, this material was successfully applied to enrich trastuzumab and its related antibody-drug conjugate (ADC) from different cell culture medias. The dynamic tracking analysis of ADC in the critical quality attributes (e.g., charge variants, drug to antibody ratio and subunit conjugation ratio) was also achieved by combining the enrichment approach, capillary electrophoresis or reversed phase liquid chromatography. In summary, the exploited peptide-based mimotope affinity materials showed a great potential for the application in biopharmaceutical analysis.

Molecular Recognition of the Self-Assembly Mechanism of Glycosyl Amino Acetate-Based Hydrogels

The self-assembly of supramolecular hydrogels has attracted the attention of many researchers, and it also has a broad application prospect in biomedical fields. However, there are few studies on the intrinsic mechanism of molecular self-assembly of hydrogels. In this paper, the self-assembly process of glycolipid-based hydrogels is studied by combining quantum chemistry calculation and molecular dynamics simulation. Using quantum chemistry calculation, the stable stacking mode of gelator dimers was explored. Then, by varying the water content in the gelation system, three different morphologies of hydrogels after self-assembly were observed on the nanoscale. When the water content is low, the molecular chains were entangled with each other to form a three-dimensional network structure. When the water content is moderate, the system had obvious stratification, forming the typical structure of "gel-water-gel". The gelators can only form small micelle-like agglomerations when the water content is too high. According to the analysis of the interaction between gelators and that between gelators and water molecules, combined with the study of the radial distribution function and hydrogen bonding, it is determined that the hydrogen bonds formed between gel molecules are the main driving force of the gelation process. Our work is of guiding significance for further exploration of the formation mechanism of a hydrogel and developing its application in other fields.

Not only in silico drug discovery: Molecular modeling towards in silico drug delivery formulations

The use of methods at molecular scale for the discovery of new potential active ligands, as well as previously unknown binding sites for target proteins, is now an established reality. Literature offers many successful stories of active compounds developed starting from insights obtained in silico and approved by Food and Drug Administration (FDA). One of the most famous examples is raltegravir, a HIV integrase inhibitor, which was developed after the discovery of a previously unknown transient binding area thanks to molecular dynamics simulations. Molecular simulations have the potential to also improve the design and engineering of drug delivery devices, which are still largely based on fundamental conservation equations. Although they can highlight the dominant release mechanism and quantitatively link the release rate to design parameters (size, drug loading, et cetera), their spatial resolution does not allow to fully capture how phenomena at molecular scale influence system behavior. In this scenario, the "computational microscope" offered by simulations at atomic scale can shed light on the impact of molecular interactions on crucial parameters such as release rate and the response of the drug delivery device to external stimuli, providing insights that are difficult or impossible to obtain experimentally. Moreover, the new paradigm brought by nanomedicine further underlined the importance of such computational microscope to study the interactions between nanoparticles and biological components with an unprecedented level of detail. Such knowledge is a fundamental pillar to perform device engineering and to achieve efficient and safe formulations. After a brief theoretical background, this review aims at discussing the potential of molecular simulations for the rational design of drug delivery systems.

From Microscale to Macroscale: Nine Orders of Magnitude for a Comprehensive Modeling of Hydrogels for Controlled Drug Delivery

Because of their inherent biocompatibility and tailorable network design, hydrogels meet an increasing interest as biomaterials for the fabrication of controlled drug delivery devices. In this regard, mathematical modeling can highlight release mechanisms and governing phenomena, thus gaining a key role as complementary tool for experimental activity. Starting from the seminal contribution given by Flory-Rehner equation back in 1943 for the determination of matrix structural properties, over more than 70 years, hydrogel modeling has not only taken advantage of new theories and the increasing computational power, but also of the methods offered by computational chemistry, which provide details at the fundamental molecular level. Simulation techniques such as molecular dynamics act as a "computational microscope" and allow for obtaining a new and deeper understanding of the specific interactions between the solute and the polymer, opening new exciting possibilities for an in silico network design at the molecular scale. Moreover, system modeling constitutes an essential step within the "safety by design" paradigm that is becoming one of the new regulatory standard requirements also in the field-controlled release devices. This review aims at providing a summary of the most frequently used modeling approaches (molecular dynamics, coarse-grained models, Brownian dynamics, dissipative particle dynamics, Monte Carlo simulations, and mass conservation equations), which are here classified according to the characteristic length scale. The outcomes and the opportunities of each approach are compared and discussed with selected examples from literature.

Understanding hydrogelation processes through molecular dynamics

Molecular dynamics (MD) is currently one of the preferred techniques employed to understand hydrogelation processes for its ability to include large amounts of atoms in computational calculations, since substantial amounts of solvent molecules are involved in gel formation.

Efficient synthesis of amino acid polymers for protein stabilization

Poly-l-glutamate exerts substantial protein stabilization during lyophilization by preventing protein aggregation.

ε-Polylysine-based thermo-responsive adsorbents for immunoglobulin adsorption-desorption under mild conditions

Thermo-responsive adsorbents for immunoglobulin G (IgG) employing ε-polylysine (EPL) as a polymer backbone were developed. The introduction of mercaptoethylpyridine (MEP) as an IgG-binding ligand and hydrophobization of side chains afforded thermo-responsive IgG adsorbents, whose thermo-responsive IgG desorption ratio was up to 88% (EPL/MEP derivative 3m). The changes in surface densities of active MEP groups, which are caused by thermal conformational changes of the adsorbents, play key roles for IgG desorption. Although a trade-off of IgG adsorption capacity and IgG desorption ratio was observed, the present study offers a novel molecular design for thermo-responsive adsorbents with high synthetic accessibility and potentially low toxicity.

Molecular modeling to predict peptide accessibility for peptide-functionalized hydrogels

Peptide-functionalized (PF) hydrogels are being widely investigated by the tissue engineering and regenerative medicine communities for a broad range of applications because of their unique potential to mimic the natural extracellular matrix and promote tissue regeneration. In order for these complex material systems to perform their intended bioactive function (e.g., cell signaling), the peptides that are tethered to the hydrogel matrix must be accessible at the hydrogel surface for cell-receptor binding. The factors influencing the surface accessibility of the tethered peptide mainly include the length of the tethers, the loading (i.e., concentration) of the peptide, and the association between the tethered peptide and the hydrogel matrix. In the present work, the authors developed coarse-grained molecular models based on the all-atom polymer consistent force field for a type of poly(ethylene glycol)-based PF hydrogel and conducted molecular simulations to investigate the distribution of the peptide within the hydrogel and its surface accessibility as a function of tether length and peptide concentration. The calculated results of the effects of these design parameters on the surface accessibility of the peptide agree very well with corresponding experimental measurements in which peptide accessibility was quantified by the number of cells attached to the hydrogel surface per unit area. The developed modeling methods are able to provide unique insights into the molecular behavior of PF hydrogels and the distribution of the tethered peptides, which can serve as a guide for hydrogel design optimization.

Investigation of the morphological transition of a phospholipid bilayer membrane in an external electric field via molecular dynamics simulation

Elucidating the mechanisms for morphological transitions of the phospholipid bilayer membrane during cellular activity should lead to greater understanding of these membrane transitions and allow us to optimize biotechnologies such as drug delivery systems in organisms. To investigate the mechanism for and the dynamics of morphological changes in the phospholipid membrane, we performed molecular dynamics simulation of a phospholipid membrane with and without membrane protein under the influence of electric fields with different strengths. In the absence of membrane protein, it was possible to control the transition from one lamellar membrane morphology to another by applying a strong electric field. The strong electric field initially disordered the lipid molecules in the membrane, leading to the formation of a hydrophilic pore. The lipid molecules then spontaneously fused into a new lamellar membrane morphology. In the presence of membrane protein, a morphological transition from lamellar membrane to vesicle under the influence of a strong electric field was observed. Studying the complex transition dynamics associated with these changes in membrane morphology allowed us to gain deep insight into the electrofusion and electroporation that occur in the presence or absence of membrane protein, and the results obtained here should prove useful in work aimed at controlling membrane morphology. Graphical Abstract Memebrane morphological transition under the electric field of 0.6 V/nm with the membrane protein (down) and without membrane protein (up).

Overcoming T. gondii infection and intracellular protein nanocapsules as biomaterials for ultrasonically controlled drug release

One of the pivotal matters of concern in intracellular drug delivery is the preparation of biomaterials containing drugs that are compatible with the host target.

Mesoscale modelling of environmentally responsive hydrogels: emerging applications

We review recent advances in mesoscale computational modeling, focusing on dissipative particle dynamics, used to probe stimuli-sensitive behavior of hydrogels.