A new approach for a physicochemical characterization of nanoparticles in complex media: a pilot study

Current techniques used to measure the phys-icochemical characteristics of nanoparticles in simple media are poorly predictive of their behavior observed during in vivo experi-ments1. As a consequence, some pharmaco-kinetic or toxicokinetic issues are detected too late. In this contribution, we are proposing an innovative approach to tackle this challenge2. The goal is to develop a generic characteriza-tion process that could be used in any labora-tory with different categories of measurement technologies. The proposed solution is com-posed of three main steps: 1.Sample preparation; 2.Measurement phase; 3.Statistical analysis. The sample preparation relies on a set of n serum-free media initially designed for cell culture but used herein to mimic heterogenei-ty of biological context. Each culture medium is composed of a large number (p), around a hundred, of biological compounds (proteins, vitamins, mineral salts, etc.), which may indi-vidually and synergistically interact with the nanoparticle surface. The nanoparticle to be characterized is added to each medium of the kit with the same concentration. The resulting mixtures are then analyzed by an appropriate technology compatible with complex media to measure the size distribu-tion of constitutive nano-objects. In a third step, all the experimental data com-ing from the n series of measurement are used to solve a “ large p small n” regression problem. This statistical analysis informs about the most likely medium compounds to affect the size distribution of nanoparticles compared to their initial dimensions. This communication presents the first results of a pilot study in which the proposed approach was tested on gold nanoparticles mixed in n=8 cell free culture media provided by Thermo Fisher Scientific. The nanoparticle size distributions were measured by a Dy-namic light scattering system (Nanosight, Malvern). A partial least squares method was used to solve the “ large p small n” regression problem. Preliminary results confirms signifi-cant changes of the size distribution between the culture media and the feasibility of the statistical method to identify a set of medium compounds that may explain those variations.