Machine learning-driven optimization of hepatocyte functional maturation from hPSC

Abstract

Background & Aim

Producing large quantities of functionally mature hepatocytes remains a significant challenge for clinical applications, including liver failure treatment and preclinical drug testing. Differentiating human pluripotent stem cells (hPSCs) into hepatocyte-like cells (HLCs) offers a promising solution, with numerous protocols developed for this purpose. However, these methods typically yield immature cells lacking the full repertoire of functionalities characteristic of primary human hepatocytes.

We developed machine learning (ML) algorithms capable of predicting hepatocyte maturation directly from live-cell, label-free phase contrast (PC) images. These algorithms were successfully deployed to analyze phenotypic screening data, enabling the optimization of differentiation protocols to produce HLCs with improved functional maturation.

Methodology

hPSC differentiation into HLCs was performed following an 18-day protocol described in Ang et al. 2018. ML models were trained on paired PC and immunofluorescence (IF) images of six markers involved in metabolism, detoxification, and protein synthesis. Models took solely PC images as input and produced a multi-dimensional hepatocyte maturation score as output.

Validated ML models were used to rank and select hits in phenotypic screening experiments where differentiating hPSC-derived HLCs were exposed to small molecules and biologics at varying concentrations, durations, and combinations in iterative cycles. The top-ranking hits were compared to controls in a follow-up validation experiment using gold standard in vitro assays of hepatocyte function – CYP3A4 enzymatic activity, albumin secretion, and urea secretion. A mixed-effect model was used to assess statistical significance.

Results

ML-predicted hepatocyte maturation scores showed strong correlations (Pearson r > 0.97) across a panel of IF markers. Leveraging these ML predictions in live cultures, we rapidly and cost-effectively screened for combinations of molecules that significantly enhanced hepatocyte maturation. This led to generational improvements in protocols in a matter of months, resulting in statistically significant increases in CYP3A4 activity (+55%), albumin secretion (+40%), and urea secretion (+130%) over the baseline.

Conclusion

Our approach highlights a novel, ML-driven approach to optimizing differentiation protocols for improved functional maturation. More generally, this method holds promise for applications in upstream process development as well as downstream quality control.