Towards genome wide RNA imaging in single cells

Abstract

Quantitative and spatial analysis of full transcriptomes in single cells is crucial for our understanding of the interplay between cellular and even subcellular heterogeneity and spatial context in driving a variety of biological processes, including development and disease pathogenesis. Multiplexed error-robust fluorescence in situ hybridization (MERFISH) allows individual RNA molecules to be identified in their native spatial context. Previously, MERFISH has been applied to profile up to a thousand unique RNA species in single cells. Further increase in gene throughput would extend the range of biological problems than can be addressed. The first part of this thesis reports our efforts to image RNA transcripts from over 10,000 genes in individual cells with high accuracy and detection efficiency. We investigate the RNA spatial distribution at both subcellular level and cellular level. At the subcellular level, we conduct MERFISH measurements with counterstaining of cellular structures, and identify the subcellular compartmentalization of RNAs on ER and in nucleus. By leveraging the nuclear versus cytoplasmic RNA counts, we develop an approach to determine RNA velocity in situ, apply this approach to determine pseudotime ordering of cells, and infer cells at different cell-cycle states. At cellular level, we are able to identify spatial heterogeneity of transcriptionally distinct cell populations with cell-cycle-dependent and cell-cycle-independent manners. The second part of this thesis describes our efforts to develop an improved form of branched DNA (bDNA) amplification that dramatically increases the signal of individual molecules without increasing spot size and variation in brightness from spot to spot. We apply this amplification approach to MERFISH measurements with fewer FISH probes, resulting in improved MERFISH performance. This potentially allows MERFISH to measure shorter RNAs, increase the imaging speed, and image tissue samples with high degrees of background. We envision that by combining advances described in both parts, we can further increase the gene coverage of spatially-resolved single-cell transcriptomic analysis, which would facilitate a variety of biological applications.