Joint profiling of cell morphology and gene expression during in vitro neurodevelopment

AbstractDifferentiation of induced pluripotent stem cells (iPSC) towards neuronal lineages has enabled diverse cellular models of human neurodevelopment and related disorders. However, in vitro differentiation is a variable process that frequently leads to heterogeneous cell populations that may confound disease-relevant phenotypes. Here, we report a thorough characterization of cortical neuron differentiation from iPSCs using joint profiling of neuronal morphology and transcriptomics in single cells. We assayed 60,000 developing neurons across three timepoints with CellPainting, a high-content imaging method, and single-cell RNA-sequencing (scRNA-seq). By modeling the relationship between morphological features and gene expression within our differentiation system, we annotated image-based features with biological functions and found that while phenotypes related to cell morphology capture broader neuronal classes than scRNA-seq, they enhance our ability to quantify the biological processes that drive neuronal differentiation over time, such as mitochondrial function and cell cycle. Further, we found that while over 60% of the cells match those seen in the fetal brain, 28% represented metabolically abnormal cell states and broader neuronal classes specific to in vitro cells. Finally, we show that specific subtypes of iPSC-derived cortical neurons are a relevant model for a range of brain-related complex traits, including schizophrenia and bipolar disorder, highlighting the potential of multi-modal single cell phenotyping for disease modeling.