Organoids and 3D imaging techniques are crucial for studying human tissue structure and function, but traditional 3D reconstruction methods are expensive and time consuming, relying on complete z stack confocal microscopy data. This paper introduces VONet, a deep learning-based system for 3D organoid rendering that uses a fully convolutional neural network to reconstruct entire 3D structures from a minimal number of z stack images. VONet was trained on a library of over 39,000 virtual organoids (VOs) with diverse structural features and achieved an average intersection over union of 0.82 in performance validation. Remarkably, VONet can predict the structure of deeper focal plane regions, unseen by conventional confocal microscopy. This innovative approach and VO dataset offer significant advancements in 3D bioimaging technologies.
The human cerebral cortex is organized into distinct area-specific regions along the rostral–caudal axis, yet current human brain organoid models incompletely recapitulate this regional diversity. Here, we establish an area-specific brain organoid platform by leveraging transcription factors (TFs) identified through re-analysis of in vivo human cortical transcriptomic datasets. Publicly available single-cell RNA sequencing datasets from human developing cortex were re-analyzed to identify differentially expressed genes associated with rostral and caudal cortical identities. From this analysis, we identified SP9 (rostral-enriched) and DMRTA2 (caudal-enriched) as candidate TFs governing regional specification. To model cortical area identity, these TFs were overexpressed in an inducible manner during human cerebral organoid (hCO) generation. Overexpression of SP9 resulted in hCOs exhibiting rostral cortical characteristics, whereas DMRTA2 overexpression promoted caudal cortical features. The resulting hCOs showed distinct regional identities, reflected by differential expression of area-specific markers. In addition, these regional identities were accompanied by distinct functional phenotypes, as calcium imaging revealed divergent patterns of spontaneous neural activity between rostral and caudal hCOs. Altogether, our findings demonstrate that overexpression of TFs enables the controlled generation of area-specific hCOs. This approach provides a scalable and reproducible platform for studying human cortical regionalization and offers a framework for investigating region-specific mechanisms underlying neurodevelopmental and neurological disorders.
Neurodevelopmental disorders (NDDs), including autism spectrum disorder (ASD) and intellectual disability (ID), are genetically heterogeneous. DEAF1 has emerged as a key NDD risk gene, with pathogenic variants linked to DEAF1-associated neurodevelopmental disorder (DAND), but its role in human neurodevelopment remains unclear. Human cortical organoids (hCOs) provide a physiologically relevant model that recapitulates fetal brain development with an authentic human genetic background. Here, we show that a DEAF1 mutation in human embryonic stem cells disrupts chromatin accessibility at neuronal gene loci, leading to significant transcriptional alterations. In hCOs, this mutation results in aberrant progenitor proliferation, disrupted cortical lamination, and impaired neuronal differentiation. Furthermore, we identify WNT signaling, TGFβ superfamily signaling, and cell cycle regulation as commonly dysregulated pathways across multiple ASD-associated genetic perturbations. Pharmacological inhibition of WNT signaling with a Porcupine inhibitor partially rescues phenotypic defects in DEAF1-mutant hCOs. Our findings identify DEAF1 as a critical regulator of neurodevelopment and support pathway-targeted, mutation-independent therapeutic strategies for ASD and related disorders.