Multi-scale in-silico toxicological assessment of Ti₃C₂ and Cr₂C MXene nanoparticles at human biological interfaces.

Despite their huge technological interest, practical application of MXene nanomaterials is somehow hampered by their potential toxicity. In this regard, the present work showcases the design and implementation of a full in-silico methodology for the toxicological analysis of nanoparticles (NPs) of two MXenes within a human cell environment, employing cost-effective computational approaches to achieve a preliminary hazard estimation in the absence of experimental data. This study combines computational methodologies of very different nature, from Density Functional Theory (DFT) calculations for a full quantum mechanics optimization of the system of interest and thermodynamical approximations like the COSMO-RS methodology for the analysis of the behaviour in membranes, to other, simpler methods like classic electrostatic Monte-Carlo calculations for the interaction with proteins, and machine-learning based methods for the development of simpler yet accurate predictive tools for a broader use. This multi-scale workflow was applied to a compendium of NPs derived from two MXenes of interest, Ti3C2 and Cr2C, to understand the possible harm such materials could cause in the human body. As a result, an extremely low permeability and membrane-crossing potential was observed for the Ti3C2 NPs, and very weak and superficial interaction was observed for both MXenes with a library of critical human proteins. Two predictive docking energies models with R2 of 0.85 were developed, and a low toxicological potential via the considered mechanisms was concluded as a result, thus paving the way of reaching accurate toxicological predictions for a wide range of MXene nanomaterials.