Molecular exploration of host-pathogen interactions in severe Pseudomonas aeruginosa infection through a multi-level data integration approach

TL;DR

This paper explores how Pseudomonas aeruginosa interacts with the human host during severe infections using integrated data to better understand sepsis outcomes.

Contribution

The study provides a comprehensive dataset of PA-human interactions and a molecular network for predicting clinical phenotypes in severe infections.

Findings

A dataset of 189 PA-human interactions involving 151 proteins/molecules was created.

Proinflammatory pathways were overexpressed in PA-infected lung samples.

The molecular network offers a foundation for dynamic computational models of clinical phenotypes.

Abstract

Understanding host-pathogen interactions is crucial for explaining the variability in sepsis outcomes, with Pseudomonas aeruginosa (PA) remaining a significant public health concern. In this work, we explored PA-human host interaction mechanisms through a data integration workflow, focusing on protein-protein and metabolite-protein interactions, along with pathway modulation in affected organs during severe infections.

A scoping literature review enabled us to construct a domain-based infection network encompassing pathogenesis concepts, molecular interactions, and host response signatures, providing a wide view of the relevant mechanisms involved in severe bacterial infections.

Our analysis yielded a literature-based comprehensive description of PA infection mechanisms and an annotated dataset of 189 PA-human interactions involving 151 proteins/molecules (109 human proteins, 3 human metabolites, 34 PA proteins, and 5 PA molecules). This dataset was complemented with gene expression analysis from in vivo PA-infected lung samples. The results indicated a notable overexpression of proinflammatory pathways and PA-mediated modulation of host lung responses.

Our comprehensive molecular network of PA infection represents a valuable tool for the understanding of severe bacterial infections and offers potential applications in predicting clinical phenotypes. Through this approach combining omics data, clinical information, and pathogen characteristics, we have provided a foundation for future research in host-pathogen interactions and the mechanistic grounds to build dynamic computational models for clinical phenotype predictions.

Figures