Investigating the multifaceted host contribution to COVID-19 disease risk, progression and treatment: an integrative multi-omics network-based approach study

Agamah, Francis Adem

Investigating the multifaceted host contribution to COVID-19 disease risk, progression and treatment: an integrative multi-omics network-based approach study

Thesis / Dissertation

2024

Publisher

University of Cape Town

Department

Department of Integrative Biomedical Sciences (IBMS)

Faculty

Faculty of Health Sciences

Abstract

Coronavirus Disease-2019 (COVID-19) is a contagious respiratory disorder caused by Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), a newly emerged β coronavirus belonging to the Coronaviridae family. Since its discovery in Wuhan, China, in December 2019, COVID-19 disease has transformed into a devastating global pandemic that has created disruptions across healthcare, economic, and social systems with approximately seven hundred and seventy million reported cases and close to seven million reported deaths as of January 2024. The clinical presentation of COVID-19 is very heterogeneous, ranging from mild disease states to severe disease states, associated with varying transcription, protein expression, lipid synthesis, and metabolic profiles. As a result of the heterogeneity of COVID-19 disease, the efficacy of drugs used to treat it may vary depending on the disease states when the medicines are administered. Molecular biology studies investigating COVID-19 clinical heterogeneity and the drugs that might be used to treat the disease have varied in terms of the approaches implemented. These approaches have focused on single features of SARS-CoV-2 infected cells such as gene transcription levels or protein expression levels (so-called “single omics” approaches) to approaches that attempt to simultaneously examine multiple molecular features of infected cells (so-called “multi-omics approaches”). This project adopts the latter approach, implementing a network-based method that integrates multi-omics data and drug-related data to investigate the contribution of host physiology to COVID-19 disease progression and identify promising drug repurposing candidates as potential treatment options. This involved evaluating existing computational integrative multi-omics network-based methods, determining their strengths and limitations, and adapting them to fit the aims of this project: identifying and characterizing biosignatures associated with various COVID-19 disease phases, as well as identifying drug repurposing candidates tailored for mild, moderate, and severe COVID-19 disease phases. The World Health Organization (WHO) Ordinal Scale (WOS) was used as a disease severity reference to harmonize COVID-19 patient metadata across two studies. A unified COVID-19 knowledge graph was constructed by assembling a disease-specific interactome from the literature and databases. We leveraged a multi-omics network-based approach to construct disease-state and omics-specific graphs by integrating proteomics, transcriptomics, metabolomics, and lipidomics data with the unified COVID-19 knowledge graph. We used an adapted random walk with restart algorithm, called multiXrank, to explore the disease-state and omics-specific graphs, and drug data to search for and prioritize not only biosignatures associated with COVID-19 disease phases but also drug candidates with the potential for treating mild, moderate, and severe COVID-19. The network analysis identified critical biosignatures for each COVID-19 phase. Mild cases displayed unique signatures like CCL4 and IRF1, potentially driving chemotaxis and interferon signaling. The moderate phase was characterized by biosignatures like HGF, MMP12, IL-10, and NFKB1, implicating enhanced inflammation, matrix remodeling, and immune regulation. In severe cases, biosignatures such as lysophosphatidylcholines, diglyceride, and sphingomyelin appeared, suggesting profound tissue damage, dysregulated lipid metabolism, and disrupted repair pathways. As expected, the abundance of shared chemokine and cytokine biosignatures in severe and moderate COVID-19 disease phases as compared to either mild vs moderate or mild vs severe disease phases suggests a closer molecular relatedness between these phases. This finding, along with biosignatures that discriminate between the disease states, and interactions between biosignatures that are either common between or associated with COVID-19 disease phases sheds light on the nuanced progression of the illness. We further investigated the differential influence of interleukin-6 (IL6) and interleukin 6 receptor (IL6R) on disease progression. We found that IL6 interaction with features of different omics data types increased with disease severity, thus indicating the differential association of IL6 with the different disease states. More specifically, IL6 interaction with proteins (e.g., IFNB, IFIT3), transcripts (e.g., CXCL1, CXCL2, CCL3), and metabolites (e.g., 1-(1-enyl-palmitoyl)-GPC, 1-(1-enyl-palmitoyl)-2-oleoyl-GPC (P-16:0/18:1)) may contribute to its major role in disease severity. We also observed that IL6R interactions mainly with proteins and transcripts increase more clearly with disease severity than do interactions with metabolites and lipids. We present a multilayered visualization tool (hosted at http://cytoscape.h3africa.org, last accessed on February 6, 2024) to navigate and analyze complex interactions across different biological layers, offering a valuable resource for uncovering key drivers of disease severity. Interestingly, cross-layer interactions between different omics profiles increased with disease severity. These potential association patterns could be useful for providing insights into the underlying molecular causes and consequences of the clinical heterogeneity of COVID-19, enabling early disease diagnosis, and optimal treatment prediction. The network-based integration of drug data and multi-omics data assisted in drug prediction. Most importantly, we prioritized twenty Food and Drug Administration approved agents with potential utility for mild, moderate, and severe COVID-19 disease phases. For mild COVID-19, stimulating immune cell recruitment and activation is key. Drugs like histamine, curcumin, and paclitaxel show potential in this regard due to their ability to stimulate immune cell recruitment, potentially mitigating disease progression. Similarly, non-steroidal anti-inflammatory drugs like indomethacin and diclofenac may offer symptomatic relief in mild cases. In mild to moderate COVID-19, drugs like omacetaxine, crizotinib, and vorinostat, known for their antiviral properties, can potentially hinder viral replication and offer additional treatment options. Moreover, glutathione, a potent antioxidant, could be valuable in moderate cases due to its potential to counteract inflammation and potentially prevent the dangerous "cytokine storm" seen in patients with antioxidant deficiencies. In severe COVID-19, the excessive immune response triggers a dangerous "inflammatory cascade." To combat this, drugs with strong anti-inflammatory effects, including anti-inflammatory drugs (sarilumab, tocilizumab), corticosteroids (dexamethasone, hydrocortisone), and immunosuppressives (sirolimus, cyclosporine), emerge as potential candidates for controlling this harmful process. Moreover, we further explore the interactions involving the drug repurposing candidates and key biosignatures. The findings could be useful for personalized treatment options tailored to individual patients based on their disease severity level. This project identified both biosignatures of different omics types (proteins, transcripts, metabolites, and lipids) enriched in disease-state pathways and their associated interactions that are either common between, or unique to mild, moderate, and severe COVID-19. These biosignatures include molecular features that underlie the observed clinical heterogeneity of COVID-19 and emphasize the need for disease-phase-specific treatment strategies. In addition, we explored the potential of multi-omics and drug-related data to predict therapeutics for different phases of COVID-19. This project demonstrates that the integrative analysis of drug data and multi-omics data enables the prioritization of biosignatures and potential drug candidates for COVID-19 disease phases. These findings hold promise for guiding future experimental studies towards potential clinical applications, requiring further investigation to definitively translate into therapeutic advancements.