One of the earliest changes found in vessel walls at predilection sites for atherosclerotic plaque formation involves the retention and modification of lipoprotein particles. Macrophages and vascular smooth muscle cells (VSMCs) in the arterial wall respond by ingesting large quantities of the modified lipoproteins, such as oxidised low-density lipoproteins (oxLDLs), which are then metabolised to release cholesterol and cholesterol esters for use or recycling. Chronic uptake of modified particles eventually reduces the ability of macrophages and VSMCs to effectively break down and release the lipoprotein metabolites, causing intracellular accumulation of lipids. Existing studies have described potential foam cell-like populations using single cell RNA sequencing (scRNA-seq) of plaques, however due to known limitations of whole cell and more recent findings expanding our current knowledge of foam cell biology, much remains to be explored. Foam cells can be generated in vitro by lipid loading with modified lipoproteins or cholesterol, however, with a growing appreciation for the complexity of foam cells in vivo there remains uncertainty in which aspects of foam cell formation can be recapitulated with such models.

To overcome some of the known drawbacks of single cell isolation, single nuclei extraction was optimised and applied to the selected vascular samples. One novel single nuclei and two existing single cell plaque RNA-seq datasets were analysed, showing successful annotation of expected plaque cell types and identification of individual foam cell populations, which demonstrated that single nuclei sequencing can be applied to vascular samples to describe their cellular heterogeneity. Intriguingly, a population of potential VSMC-derived foam cells could be seen to localise with macrophages when visualising cell populations in the single nuclei dataset, which had not been observed in existing single cell datasets. To identify foam cells in early stages of plaque development, I assessed the lipid content and macrophage and VSMC content of organ donor aortic samples across a range of tissue phenotypes to selected foam cell-rich samples that were used to generate single nuclei RNA-sequencing libraries.

In order to compare gene profiles of in vivo foam cells to in vitro-generated foam cells, a bulk RNA-seq quality control and differential expression pipeline was established and applied to a large dataset that had been previously generated from a iPSC-derived macrophage foam cell model. OxLDL-treated macrophages expressed a number of expected foam cell gene markers, independent of M0, M1 and M2 polarisation states, however, TREM2 was notably not differentially expressed in treated conditions. TREM2 is commonly associated with foamy macrophages in vivo; this observation therefore highlighted that significant differences in expression profiles may preclude in vitro models from effectively modelling certain pathways in foam cell development. Comparison of in vivo foam cell gene profiles with in vitro macrophage foam cells showed unexpectedly little overlap. However, the analysis revealed that comparing bulk and single cell RNA-seq foam cell profiles was highly challenging, and required further evaluation to understand which aspects of foam cell formation can be explored in vitro. Future analysis of the data generated from arteries showing early signs of disease will serve to understand whether the in vitro conditions provide a more suitable model for initial foam cell formation.