During this project, I established a new, high-throughput technology platform which simultaneously measures both physical properties and gene expression levels of single cells. This technology is based on microfluidics, a technique which creates microscale channels to control fluid flows, akin to circuits on a computer chip. Using microfluidics, we can generate of microscale droplets: aqueous droplets of 1 nL volume (1 billionth of 1 L) are stably suspended in oil, so that each droplet acts as an individual microreactor. A key advance was the use of such droplets for single cell analysis. By encapsulating individual cells in such droplets, the contents of each cell can be measured. However, a crucial limitation of droplets is that every droplet looks identical, meaning that we cannot easily track individual droplets between measurements. Here, I developed a method to uniquely tag droplets, allowing us to link imaging and sequencing data for single cells.
My droplet tagging technology introduces physical barcodes into each droplet with the cells. Each barcode is identifiable by both imaging and DNA sequencing, so that we can link the two data types for the accompanying cells. I developed methods to synthesise these barcodes: to make them optically identifiable, I edited the parameters of size, shape, and colour. In order to make them also identifiable by sequencing, I incorporated DNA molecules with controllable release mechanisms.
Having established a way to tag large numbers of individual droplets, I then developed the systems to read barcodes in droplets. I optimised a microfluidic chip design which carries out the whole workflow, comprising 1) characterisation of single cells by imaging, 2) production of droplets containing both barcodes and characterised cells, and 3) identification of barcodes within droplets with their accompanying cell. The barcoded droplets can then be collected and sequenced, to determine gene expression levels for each cell. In addition to these experimental aspects, I wrote software to analyse the resulting datasets, extracting cell properties from images and pairing these physical parameters (e.g. size, mechanical stiffness) with their gene expression profiles.
I validated the whole platform by application to known samples, including a mixture of two cell types. These cell types have known differences in both their physical properties (one cell type is larger than the other), and their gene expression patterns. This cell mixture thus demonstrated that the images and gene expression profiles are correctly linked for each single cell. I also applied the platform to investigate several cancer cell lines, to identify genes associated with physical properties and their universality across cell types.
