Exploring the Lung–Liver Axis in Pulmonary Arterial Hypertension

Figures

• 

  FIGURE 1. Differences in gene expression from bulk RNA sequencing of pulmonary artery endothelial cell biopsies from PAH participants in the High‐ versus Low‐MELD clusters were examined using three modeling approaches. (A) The number and proportion of genes generated by three independent models (linear discriminant analysis (LDA) with principal component analysis (PCA), LDA with weighted gene co‐expression analysis (WGCNA), and a random forest (RF) model) and their intersections was reviewed and the intersection of LDA with PCA and WGCNA was used for all subsequent analysis. (B) Pathways analysis demonstrates enrichment for pathways related to apelin signaling, YAP/TAZ, as well as immunity and inflammation.

• 

  FIGURE 2. Protein–protein interaction (PPI) network analysis of differentially expressed genes in pulmonary artery endothelial cell biopsies from High‐ versus Low‐MELD‐Na PAH patients identifies IL‐6 as a primary hub of inflammatory activation. (A) PPI network of 480 genes (degree ≥ 5) constructed from 2458 genes with nominal p < 0.05 by DESeq2 using STRING (combined confidence score ≥ 400). Node size reflects betweenness centrality; node color reflects Louvain community membership (16 modules, modularity Q = 0.554). IL‐6 emerged as the primary network hub (betweenness centrality = 0.072, degree = 132, log2FC = +1.30), co‐localizing in Module 4 (n = 242 genes, 81% upregulated in High‐MELD) with genes involved in inflammatory signaling (COX2, MAPK3), vascular remodeling (LOX), non‐canonical Wnt activation (WNT5A), and TGFβ/Activin signaling (SMAD3, INHBA). (B) GO Biological Process enrichment of Module 4 demonstrates significant enrichment for leukocyte chemotaxis, myeloid leukocyte migration, positive regulation of cytokine production, and regulation of inflammatory response (all FDR < 0.05, Benjamini–Hochberg correction), consistent with an IL‐6‐centered transcriptional program associated…

• 

  FIGURE 3. Increased hepatic fibrosis in SuHx rats compared to controls. Control and SuHx rat livers were stained with Masson's Trichrome which resulted in increased perivascular and parenchymal fibrosis (A), 20.8 versus 16.6% area stained, p = 0.09 (B). n = 5 rats per group. SuHx = Sugen‐hypoxia. All images taken at 4×. Scale bar = 500 μm. Collagen‐positive (blue) area was quantified using the Otsu algorithm in Fiji by a reviewer blinded to group assignment.

• 

  FIGURE 4. Bulk RNA sequencing of MCT (A) and SuHx (B) male rat livers compared to controls demonstrates increased gene expression in pathways related to inflammation and TGB beta signaling and decreased expression in pathways related to cellular metabolism. (C) A regulatory network built from differentially expressed genes in both MCT and SuHx rat livers demonstrates activation of pathways related to activation and recruitment of leukocytes.

• 

  FIGURE 5. Differential gene expression in SuHx and MCT rat lung endothelial cells from a publicly available single‐cell RNA sequencing dataset was compared to pulmonary artery endothelial cell (PAEC) transcriptomes from High‐MELD participants. (A) Venn diagram demonstrating 775 genes overlapping between human High‐MELD PAECs and rat lung endothelial cells across both models. (B) GO pathway enrichment of the 775 overlapping genes demonstrated significant enrichment for pathways related to cell survival, proliferation, and cancer biology, including EGR1, ECM1, and MIF. SuHx = Sugen‐Hypoxia. MCT, monocrotaline; MELD, model for end state liver disease.

• 

  FIGURE 6. Masson's trichrome staining of liver tissue from humans with and without pulmonary arterial hypertension (PAH) demonstrates that livers from PAH patients without liver disease (non‐PoPH PAH n = 4 of which CTD‐APAH n = 3, idiopathic PAH n = 1) demonstrated quantitatively intermediate fibrosis between controls (CTRL) and portopulmonary hypertension (PoPH). All images taken at 10×. Scale bar = 100 μm. *p = 0.03, ns = not significant.