UC San Diego

The structure and function of tissues are highly intertwined, necessitating an understanding of their architecture to comprehend their normal and pathological states. While traditional histology and immunostaining offer high spatial resolution, they are limited in molecular resolution. Conversely, single-cell sequencing provides molecular resolution but lacks spatial context due to tissue dissociation. Spatial transcriptomics bridges this gap by combining microscopy and DNA technology, enabling high resolution molecular readouts with spatial information.

In the first part of this dissertation, we utilize a spatial RNA capture technology to map cell types in the human kidney at a near-cellular resolution, revealing cellular neighborhoods across different anatomical regions of the kidney, identifying niches for novel populations, and uncovering strong mitochondrial gene expression patterns in a specific epithelial subtype.

While sequencing-based methods have a fixed spatial resolution, multiplexed in situ detection methods offer cellular resolution but face challenges in sensitivity, throughput, and cost which severely limit their application in human sections that are large and have low quality. In aim 2, I developed a novel multiplexed in situ RNA detection method, DART-FISH, capable of profiling hundreds of genes in large human tissue sections with ease of implementation. I further developed an accompanying suite of computational tools for designing and processing the output of in situ experiments.

Finally, in aim 3, the utility of DART-FISH is demonstrated by applying it to multiple tissue types. This includes the human brain where the layered organization of excitatory neurons was recapitulated and a rare subclass of inhibitory neurons uncovered. In the human kidney, I profiled healthy and pathological cell states in the cortex, and identified interactions between disease-altered epithelial cells and myofibroblasts. Finally, I showed how DART-FISH can be utilized for organ-scale measurements by imaging a cross section of a mouse kidney where all segments of the nephron can be visualized and pathological neighborhoods can be systematically analyzed with single-cell resolution.

In summary, this dissertation provides several experimental and computational solutions to advance the field of spatial transcriptomics and lower its cost. It also shows the application of spatial transcriptomics to map the architecture of human and mouse tissues.