Illuminating tissue organization by imaging the spatial transcriptome

Abstract

Our bodies consist of a large collection of cells that each have their own function in the organ that they reside in. The cells are grouped by functionality in cell types that arise during development as the result of the gene regulatory network encoded in the genome. With the development of novel single cell technologies, we are starting to understand just how diverse our cells are. In the brain for instance there are at least 3,000 distinguishable types. However, we have little understanding of how all these cell types are spatially organized in the tissue, because conventional labeling and microscopy techniques are incapable of resolving such high complexity in a single experiment.

In this thesis I present the development of two methods that can resolve the cellular complexity and spatial organization of mouse and (developmental) human brain samples. These methods are built upon the concept of cyclic RNA labeling with single molecule Fluorescent in situ Hybridization (smFISH) to detect hundreds of gene targets in tissue samples. The resulting RNA localizations can then be used to study spatial gene expression and to identify the cell type of each cell in the sample. The cellular identity and position can then be used to study spatial relationships between cells to understand the tissue architecture.

To place the development of these two methods into context, I will first review the field of spatially resolved transcriptomics. I will discuss the methods that are based on microscopy and spatially tagged RNA sequencing, where I will compare their strengths and weaknesses.

Then I will present the two projects: Paper I presents the development of a cyclic smFISH protocol called osmFISH that leverages the high detection efficiency of smFISH to measure the gene expression of 33 cell type marker genes in the mouse somatosensory cortex at single cell resolution. We developed the labeling technology, instrumentation and analysis software to enable the study of cellular organization at multiple length scales.

Even though osmFISH and related microscopy-based methods generate high quality data they are limited by the spatial throughput so that only small tissue areas can be processed. In paper II I present another method called EEL FISH that uses electrophoresis to transfer the RNA from a 3D tissue section onto a flat surface. The collapsing of one dimension substantially reduces the time needed to image, while retaining the information, so that the complex spatial gene expression profiles of entire mouse brain sections, sub-structures of the human brain and human developmental tissues can be studied.

Lastly, I will discuss these results and look at the future of the field of spatially resolved transcriptomics.