Search

Abstract Integration of heterogeneous single-cell sequencing datasets generated across multiple tissue locations, time, and conditions is essential for a comprehensive understanding of the cellular states and expression programs underlying complex biological systems. Here, we present scDREAMER ( https://github.com/Zafar-Lab/scDREAMER ), a data-integration framework that employs deep generative models and adversarial training for both unsupervised and supervised (scDREAMER-Sup) integration of multiple batches. Using six real benchmarking datasets, we demonstrate that scDREAMER can overcome critical challenges including skewed cell type distribution among batches, nested batch-effects, large number of batches and conservation of development trajectory across batches. Our experiments also show that scDREAMER and scDREAMER-Sup outperform state-of-the-art unsupervised and supervised integration methods respectively in batch-correction and conservation of biological variation. Using a 1 million cells dataset, we demonstrate that scDREAMER is scalable and can perform atlas-level cross-species (e.g., human and mouse) integration while being faster than other deep-learning-based methods.

Lineage tracing studies combining CRISPR-Cas9 editing and scRNA-seq face several challenges and cannot integrate lineages from multiple individuals. Here the authors show that integration of mutation and expression leads to accurate lineage tree inference and enables the learning of a species-invariant lineage tree.

Abstract Interrupting the interplay between cancer cells and extracellular matrix (ECM) is a strategy to halt tumor progression and stromal invasion. Perlecan/heparan sulfate proteoglycan 2 (HSPG2) is an extracellular proteoglycan that orchestrates tumor angiogenesis, proliferation, differentiation and invasion. Metastatic prostate cancer (PCa) cells degrade perlecan-rich tissue borders to reach bone, including the basement membrane, vasculature, reactive stromal matrix and bone marrow. Domain IV-3, perlecan’s last 7 immunoglobulin repeats, mimics native proteoglycan by promoting tumoroid formation. This is reversed by matrilysin/matrix metalloproteinase-7 (MMP-7) cleavage to favor cell dispersion and tumoroid dyscohesion. Both perlecan and Domain IV-3 induced a strong focal adhesion kinase (FAK) dephosphorylation/deactivation. MMP-7 cleavage of perlecan reversed this, with FAK in dispersed tumoroids becoming phosphorylated/activated with metastatic phenotype. We demonstrated Domain IV-3 interacts with the axon guidance protein semaphorin 3A (Sema3A) on PCa cells to deactivate pro-metastatic FAK. Sema3A antibody mimicked the Domain IV-3 clustering activity. Direct binding experiments showed Domain IV-3 binds Sema3A. Knockdown of Sema3A prevented Domain IV-3-induced tumoroid formation and Sema3A was sensitive to MMP-7 proteolysis. The perlecan-Sema3A complex abrogates FAK activity and stabilizes PCa cell interactions. MMP-7 expressing cells destroy the complex to initiate metastasis, destroy perlecan-rich borders, and favor invasion and progression to lethal bone disease.