EARLY PREDICTION OF HUMAN IPSC ASTROCYTE DIFFERENTIATION OUTCOMES USING SINGLE-CELL TRANSCRIPTOMIC SIGNATURES

The generation of human induced pluripotent stem cell (iPSC)–derived astrocytes is essential for neuroscience research and drug discovery, yet current growth factor–based differentiation protocols are time-consuming, resource-intensive, and often variable in outcome. To address this challenge, we used single-cell transcriptomics to profile multiple iPSC astrocyte differentiation experiments to identify transcriptomic differences that distinguish successful from unsuccessful differentiations. Cells were sampled across several differentiation time points to characterize temporal changes in gene expression and to determine the earliest stage at which predictive differences emerge. We found that in successful differentiations, neural progenitor cells progressed through distinct transcriptional states over time, whereas in unsuccessful differentiations, cells remained transcriptionally static and clustered within the same state across time points, despite showing a morphology reminiscent of astrocytes. Leveraging these data, we identified the most informative genes and optimal time points for early prediction. This approach enabled the design of a targeted gene expression panel comprising markers that are differentially up- or downregulated in successful versus unsuccessful differentiations. By enabling early prediction of successful differentiation outcomes, we will substantially reduce time, cost, and labor associated with iPSC astrocyte generation. Implementation of this platform will improve the efficiency and reliability of human iPSC astrocyte production for neuroscience research and drug discovery applications.