Poster L29 - Integrating Pathway Optimization and Omics Analyses to Unveil Mechanisms on Pulmonary Arterial Hypertension

Introduction: Pulmonary arterial hypertension (PAH) is a multifactorial disease associated with complex biological pathways, but actionable treatment mechanisms remain elusive. An increasing knowledge that diets can affect the progression of PAH provides promises for studies on intestinal changes for mediating PAH. Thus, we pivoted quickly to using intestinal data. High-throughput “omics” technologies enable the broad-spectrum analysis of PAH, however, any single omics measurement often reveals incomplete pictures of disease pathology. Pathway optimization can complement omics datasets by revealing hidden regulators that provide a more holistic picture of disease pathology. Yet they haven’t been considered sufficiently in PAH. Here, our approach used the pathway optimization algorithm, Prize Collecting Steiner Forest (PCSF), and Gene Ontology (GO) enrichment analysis to analyze the high throughput RNA sequencing data in multiple tissues and cell types under simulated PAH-induced and treated conditions.

Materials and

Methods: We first prepared RNA-seq data and interaction networks for PCSF modeling. To narrow the pathway search, we filtered differentially-expressed genes from multiple RNA-seq data sets in mice: enterocyte 15-Hydroxyeicosatetraenoic acid (15-HETE) vs control, 15-HETE vs sodium butyrate, sodium butyrate vs control, and jejunum 15-HETE vs empty vector, 15-HETE vs TGF6 peptide, and 15-HETE vs empty vector. The jejunum and enterocyte samples represented a multi-cell-type and absorptive cell type sample, respectively. In all cases, 15-HETE was the disease-induction condition, and TGF6 or sodium butyrate represented the treatment conditions. Again, to speed the network search, we used the top 10% of mouse protein-protein interactions from the STRING database. We used the filtered genes and PPI network as inputs and generated six PCSF graphs, one for each condition. Lastly, we used Gene Ontology (GO) enrichment, implemented with GOATools, to assess functional terms associated with the input gene lists, and the filtered gene lists prioritized by PCSF.

Results, Conclusions, and Discussions:

Results: Prior to network analysis, we discovered that each dataset generally contained distinct sets of differentially expressed genes, suggesting that treatments did not work by directly reverting disease-induced gene changes. Further, PCSF networks for each condition generally contained distinct genes. However, after using GO enrichment of PCSF-selected genes, we discovered that diverse treatments in jejunum and enterocyte, shared biological functions more than the input datasets alone (Figure. 1). Enrichment of PCSF networks discovered processes, such as lipid transport and regulation of lipase activity, that were not present among the highest-ranked differentially expressed genes, which were associated with immune system processes and responses to interferon. These findings underscore the intricate nature of gene regulation in response to different treatments in PAH and highlight the need for comprehensive network analyses to decipher the complex biological interactions at play.
Discussions: In our study, we leveraged a network method algorithm combined with gene ontology enrichment analysis to interpret complex RNA-seq data from multiple tissues and conditions in pulmonary arterial hypertension (PAH). This approach revealed that contrary to expectations, conditions with opposing effects—such as treatment versus disease—did not revert gene expression patterns. However, by using a pathway-based search, we discovered shared gene ontology functional terms. The study highlighted the nuanced interaction of genes, showing that distinct genes can trigger similar biological processes, thus broadening our understanding of gene functionality and pathology in PAH.

Conclusions: Our approach aims to understand PAH disease pathology by identifying underlying biological networks and relevant pathways. Our approach might be a complementary way forward as compared to other techniques, such as gene knock-down experimental validations.

Acknowledgements (Optional): Thanks to Dr. Ellen O'Connor and Dr.Srinivasa Reddy from UCLA David Geffen School of Medicine for providing the data.