Resolving Inflammatory Networks in Atherosclerosis Using Proteomics and Bioinformatics

Student thesis: Doctoral ThesisDoctor of Philosophy


Network reconstruction is a crucial component of quantitative omics data analysis, allowing the study of molecular interactions and regulatory associations among genes, transcripts, and proteins. Biological networks have been widely used in atherosclerosis and cardiovascular diseases to identify causal genes and potential biomarkers and to design drug targets.

In this Thesis, we first describe the main computational techniques for reconstructing and analyzing different types of protein networks and summarize the previous applications of such techniques in cardiovascular diseases, particularly atherosclerosis. Existing tools were critically compared, discussing when each method is preferred and presenting examples of reconstructing protein networks of different types (regulatory, co-expression, and protein-protein interaction networks). We demonstrate the necessity to reconstruct networks separately for each tissue type and disease entity, with different cardiovascular diseases serving as examples. We then demonstrate and discuss how the findings of protein networks could be interpreted using single-cell RNA-sequencing data.

Networks of protein interactions in atherosclerosis and relevant phenotype, sex, and vasculature-specific mechanisms, have not been fully elucidated. As existing network analysis techniques typically use the same association threshold for all proteins when inferring positive interactions, they may not be able to identify and incorporate the study of negative associations and interactions among genes and proteins, as well as lacking directionality in the reconstructed networks. In the present Thesis, we also introduce a pipeline of directional regulatory network reconstruction with adaptive partitioning (DiRec-AP), which overcomes existing problems of reconstructing regulatory and co-expression networks. DiRec-AP was benchmarked against three representative network reconstruction methods, using golden standard datasets and networks, and significantly outperformed all of them. Then, it was applied to the reconstruction of atherosclerotic plaque extracellular matrisome networks and phenotype-, sex- and vasculature-specific networks. The reconstructed atherosclerotic-directed networks can be used to formulate new hypotheses for atherosclerosis mechanisms and to identify potential novel drug targets.Mass spectrometry-based proteomics offers balanced identification and quantification performance compared to antibody and aptamer-based techniques.

However, the complexity of data produced, high missingness, and low reproducibility restrict the application in clinical practice. In the third chapter of this Thesis, we introduce a novel multi-objective optimization framework, HOpTar-omics, based on a Pareto-based Evolutionary Optimization Algorithm, which is designed to streamline and optimize the preprocessing pipeline of three different types of targeted mass spectrometry techniques. The proposed optimization framework outperformed benchmark methods and tools when applied to three large-scale tissue and plasma samples datasets. The application of this pipeline to apolipoprotein measurements in the matrisome of atherosclerotic plaques enabled the largest absolute quantification study of apolipoproteins in the human plaques matrisome and facilitated insight into how apolipoproteins accumulate in human plaques and adjacent tissue.

In the final chapter of this Thesis, using proteomics, we try to reveal novel molecular subtypes of human atherosclerotic lesions, study their associations with histology and imaging and relate them to long-term cardiovascular outcomes. We use discovery and targeted proteomics analysis on plaque samples from 120 patients undergoing carotid endarterectomy, along with a combination of statistical, bioinformatics, and machine learning methods to perform differential expression, network, enrichment analysis, and train and evaluate prognostic models. This proteomics analysis doubles the coverage of the plaque proteome compared to the largest proteomics study on atherosclerosis so far. We identify proteomic signatures for plaque calcification, inflammation and sex, validate them using different types of RNA-sequencing data, and compare them with proteomic signatures of plaque ultrasound and histology measurements. Through proteomics, we define plaque subgroups. A signature of four proteins predicts cardiovascular endpoints with an Area Under the Curve of 75% in the discovery and 65% in the validation cohort, improving the prognostic performance of imaging, and histology. Finally, we develop a prototype of the relational database, the PlaqueMS database, to store our findings and provide easy access to atherosclerotic plaque proteomics datasets and results.
Date of Award1 Nov 2023
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorManuel Mayr (Supervisor) & Konstantinos Theofilatos (Supervisor)

Cite this