Supervised Learning Model Predicts Protein Adsorption to Carbon Nanotubes

Abstract

Engineered nanoparticles are advantageous for numerous biotechnology applications, including biomolecular sensing and delivery. However, testing the compatibility and function of nanotechnologies in biological systems requires a heuristic approach, where unpredictable biofouling often prevents effective implementation. Such biofouling is the result of spontaneous protein adsorption to the nanoparticle surface, forming the "protein corona" and altering the physicochemical properties, and thus intended function, of the nanotechnology. To better apply engineered nanoparticles in biological systems, herein, we develop a random forest classifier (RFC) trained with proteomic mass spectrometry data that identifies which proteins adsorb to nanoparticles. We model proteins that populate the corona of a single-walled carbon nanotube (SWCNT)-based optical nanosensor. We optimize the classifier and characterize the classifier performance against other models. To evaluate the predictive power of our model, we then apply the classifier to rapidly identify and experimentally validate proteins with high binding affinity to SWCNTs. Using protein properties based solely on amino acid sequence, we further determine protein features associated with increased likelihood of SWCNT binding: proteins with high content of solvent-exposed glycine residues and non-secondary structure-associated amino acids. Furthermore, proteins with high leucine residue content and beta-sheet-associated amino acids are less likely to form the SWCNT protein corona. The classifier presented herein provides an important tool to undertake the otherwise intractable problem of predicting protein-nanoparticle interactions, which is needed for more rapid and effective translation of nanobiotechnologies from in vitro synthesis to in vivo use.