Advanced high-content chemical imaging tools for phenotypic drug screening in 3D culture models

Abstract: The discovery and development of new pharmaceuticals are hindered by high attrition rates, long development cycles, and escalating costs. On average, launching a new drug takes 12 years and costs more than $2 billion, with approximately 90% of clinical drug development failing mainly due to inadequate efficacy and unforeseen toxicity. Traditional cell cultures and in vivo animal models frequently lack predictive power for assessing efficacy and toxicity, leading to costly late-stage failures. In response, there has been a paradigm shift towards using more sophisticated model systems in drug screening, such as 3D culture models and organ-on-a-chip. These models require sophisticated assays that capture the heterogeneity and dynamic nature of drug response at the single-cell level and provide insights into mechanisms of action (MOA). High-resolution chemical imaging, particularly Stimulated Raman Scattering (SRS) microscopy, could potentially fulfill the role of multiparametric phenotypic imaging in complex 3D models. SRS leverages vibrational bond information to generate detailed maps of cellular components and compounds without the need for fluorescent tags that may perturb cellular functions or introduce staining artifacts. My group is at the forefront of applying SRS microscopy to imaging of small molecules such as drugs. In recent years, we have developed various techniques and strategies for quantitative SRS imaging of complex samples, applying these techniques to study single-cell drug uptake and response. These advancements have positioned us uniquely to leverage SRS as a phenotypic imaging tool in high-content drug screening for both efficacy and toxicity. This project has three synergistic directions: 1) developing innovative tools for longitudinal, multiparametric phenotypic imaging of single cells, 2) applying phenotypical imaging tools to drug efficacy screening and mechanism of action prediction using tumor spheroids and cuboids as model systems, and 3) multiparametric imaging, classification, and prediction of drug-induced liver injury in physiologically relevant tissue models. Together, these directions will enable the development and validation of new tools and methods to significantly improve drug efficacy and toxicity testing. By integrating SRS imaging and machine learning into phenotypic drug discovery workflows, our project promises to accelerate the selection of promising drug candidates and illuminate the pathways through which they exert their effects.