Abstract

Pairwise mapping of single-cell gene expression and electrophysiology in intact three-dimensional (3D) tissues is crucial for studying electrogenic organs (e.g., brain and heart)1-5. Here, we introduce in situ electro-sequencing (electro-seq), combining soft bioelectronics with in situ RNA sequencing to stably map millisecond-timescale cellular electrophysiology and simultaneously profile a large number of genes at single-cell level across 3D tissues. We applied in situ electro-seq to 3D human induced pluripotent stem cell-derived cardiomyocyte (hiPSC-CM) patches, precisely registering the CM gene expression with electrophysiology at single-cell level, enabling multimodal in situ analysis. Such multimodal data integration substantially improved the dissection of cell types and the reconstruction of developmental trajectory from spatially heterogeneous tissues. Using machine learning (ML)-based cross-modal analysis, in situ electro-seq identified the gene-to-electrophysiology relationship over the time course of cardiac maturation. Further leveraging such a relationship to train a coupled autoencoder, we demonstrated the prediction of single-cell gene expression profile evolution using long-term electrical measurement from the same cardiac patch or 3D millimeter-scale cardiac organoids. As exemplified by cardiac tissue maturation, in situ electro-seq will be broadly applicable to create spatiotemporal multimodal maps and predictive models in electrogenic organs, allowing discovery of cell types and gene programs responsible for electrophysiological function and dysfunction.