Search

Copy number variants (CNVs) that delete or duplicate 30 genes within the 16p11.2 genomic region give rise to a range of neurodevelopmental phenotypes with high penetrance in humans. Despite the identification of this small region, the mechanisms by which 16p11.2 CNVs lead to disease are unclear. Relevant models, such as human cortical organoids (hCOs), are needed to understand the human-specific mechanisms of neurodevelopmental disease. We generated hCOs from 17 patients and controls, profiling 167,958 cells with single-cell RNA-sequencing analysis, which revealed neuronal-specific differential expression of genes outside the 16p11.2 region that are related to cell-cell adhesion, neuronal projection growth, and neurodevelopmental disorders. Furthermore, 16p11.2 deletion syndrome organoids exhibited reduced mRNA and protein levels of RBFOX1, a gene that can also harbor CNVs linked to neurodevelopmental phenotypes. We found that the genes previously shown to be regulated by RBFOX1 are also perturbed in organoids from patients with the 16p11.2 deletion syndrome and thus identified a novel link between independent CNVs associated with neuronal development and autism. Overall, this work suggests convergent signaling, which indicates the possibility of a common therapeutic mechanism across multiple rare neuronal diseases.

Several deep learning approaches have been proposed to address the challenges in computational pathology by learning structural details in an unbiased way. Transfer learning allows starting from a learned representation of a pretrained model to be directly used or fine-tuned for a new domain. However, in histopathology, the problem domain is tissue-specific and putting together a labelled data set is challenging. On the other hand, whole slide-level annotations, such as biomarker levels, are much easier to obtain. We compare two pretrained models, one histology-specific and one from ImageNet on various computational pathology tasks. We show that a domain-specific model (HistoNet) contains richer information for biomarker classification, localization of biomarker-relevant morphology within a slide, and the prediction of expert-graded features. We use a weakly supervised approach to discriminate slides based on biomarker level and simultaneously predict which regions contribute to that prediction. We employ multitask learning to show that learned representations correlate with morphological features graded by expert pathologists. All of these results are demonstrated in the context of renal toxicity in a mechanistic study of compound toxicity in rat models. Our results emphasize the importance of histology-specific models and their knowledge representations for solving a wide range of computational pathology tasks.

Physical contact between organelles is vital to the function of eukaryotic cells. Lipid droplets (LDs) are dynamic organelles specialized in lipid storage that interact physically with mitochondria in several cell types. The mechanisms coupling these organelles are, however, poorly understood, and the cell-biological function of their interaction remains largely unknown. Here, we discover in adipocytes that the outer mitochondrial membrane protein MIGA2 links mitochondria to LDs. We identify an amphipathic LD-targeting motif and reveal that MIGA2 binds to the membrane proteins VAP-A or VAP-B in the endoplasmic reticulum (ER). We find that in adipocytes MIGA2 is involved in promoting triglyceride (TAG) synthesis from non-lipid precursors. Our data indicate that MIGA2 links reactions of de novo lipogenesis in mitochondria to TAG production in the ER, thereby facilitating efficient lipid storage in LDs. Based on its presence in many tissues, MIGA2 is likely critical for lipid and energy homeostasis in a wide spectrum of cell types., (Copyright © 2019 Elsevier Inc. All rights reserved.)