Identification of Hypoxia Induced Metabolism Associated Genes in Pulmonary Hypertension

Abstract

Objective: Pulmonary hypertension (PH) associated with hypoxia and lung disease (Group 3) is the second most common form of PH and associated with increased morbidity and mortality. This study was aimed to identify hypoxia induced metabolism associated genes (MAGs) for better understanding of hypoxic PH. Methods: Rat pulmonary arterial smooth muscle cells (PASMCs) were isolated and cultured in normoxic or hypoxic condition for 24 h. Cells were harvested for liquid chromatography-mass spectrometry analysis. Functional annotation of distinguishing metabolites was performed using Metaboanalyst. Top 10 enriched metabolite sets were selected for the identification of metabolism associated genes (MAGs) with a relevance score >8 in Genecards. Transcriptomic data from lungs of hypoxic PH in mice/rats or of PH patients were accessed from Gene Expression Omnibus (GEO) database or open-access online platform. Connectivity Map analysis was performed to identify potential compounds to reverse the metabolism associated gene profile under hypoxia stress. The construction and module analysis of the protein-protein interaction (PPI) network was performed. Hub genes were then identified and used to generate LASSO model to determine its accuracy to predict occurrence of PH. Results: A total of 36 altered metabolites and 1,259 unique MAGs were identified in rat PASMCs under hypoxia. 38 differentially expressed MAGs in mouse lungs of hypoxic PH were revealed, with enrichment in multi-pathways including regulation of glucose metabolic process, which might be reversed by drugs such as blebbistatin. 5 differentially expressed MAGs were displayed in SMCs of Sugen 5416/hypoxia induced PH rats at the single cell resolution. Furthermore, 6 hub genes (Cat, Ephx1, Gpx3, Gstm4, Gstm5, and Gsto1) out of 42 unique hypoxia induced MAGs were identified. Higher Cat, Ephx1 and lower Gsto1 were displayed in mouse lungs under hypoxia (all p < 0.05), in consistent with the alteration in lungs of PH patients. The hub gene-based LASSO model can predict the occurrence of PH (AUC = 0.90). Conclusion: Our findings revealed six hypoxia-induced metabolism associated hub genes, and shed some light on the molecular mechanism and therapeutic targets in hypoxic PH.

Keywords: hypoxia; metabolism associated genes; metabolomics; pulmonary hypertension; transcriptomics.

FIGURE 1

Main analysis flowchart. Identification of distinguishing metabolites of rat pulmonary arterial smooth muscle cells (PASMCs) after hypoxia (1% O₂) exposure for 24 h versus PASMCs in normoxia condition. Enrichment metabolite sets were analyzed by Metaboanalyst (v 5.0). Top 10 enriched pathways were revealed, and 1,259 metabolism associated genes (MAGs) with a relevance score ≥8 of top 10 enriched pathways in Genecards were identified. Next the shared MAGs and differentially expressed genes (DEGs) in mouse lung tissues of hypoxia induced pulmonary hypertension (PH) were identified (gene set 1), and potential small molecular compounds (drugs) to reverse the altered expression of gene set 1 were revealed by Connectivity Map (CMap) analysis. Dataset of single cell RNA sequencing of rat lung tissues of Sugen 5416/hypoxia (SuHx) induced PH were obtained and shared genes of MAGs and DEGs in smooth muscle cells of SuHx versus control rats were listed in gene set 2. Gene set 1 and gene set 2 were combined and protein protein interaction were constructed and visualized. Hub gene were identified by cytoHubba or MCODE add-ins in Cytoscape. LASSO model was constructed with the expression profiles of hub genes and ROC curves generated to evaluate the ability of LASSO model to identify PH in dataset from GSE117261. Common hub genes were validated in human PH lung tissues versus controls in dataset from GSE117261 or combined datasets from GSE117261 and GSE24988.

FIGURE 2

Identification of metabolites distinguishing rat PASMCs under hypoxia from normoxic controls and enriched metabolite sets. (A) Partial least squares discriminant analysis (PLS-DA) demonstrated a well separated sample distribution of rat PASMCs under hypoxia (Hx) and control PASMCs (Nor) for 24 h and visualized in scatter plot (n = 6). (B) 29 upregulated metabolites (red dots) and 121 down-regulated metabolites (dark green) were identified and visualized in volcano plot (Fold change >1.5 or <0.67 and p < 0.05). (C) Expression of distinguishing metabolites in (B) with VIP score >1 were visualized in heatmap. (D) Top 25 enrichment metabolite sets of distinguishing metabolites were identified in Metaboanalyst (v 5.0).

FIGURE 3

Identification of metabolism associated DEGs in mouse lungs of hypoxia induced PH and enriched pathways. (A) Principal component analysis (PCA) demonstrated a well separated sample distribution of mouse lung tissues of hypoxia at day 14 and at day 21 (Hx) and those of hypoxia at day 1 (Con) from GSE 1909 (n = 4 per condition). (B) 204 upregulated genes (red dots) and 118 down-regulated metabolites (dark green) were identified and visualized in volcano plot (Fold change >1.5 or <0.67 and p < 0.05).(C) Overlap of metabolism associated genes (MAGs) and DEGs between Hx and Con were visualized in Venn diagram; 38 overlapped genes was listed as gene set 1. (D) Expression of 38 overlapped genes were visualized in heatmap at indicated time point after hypoxia; Hx_21: day 21 after hypoxia; Hx_14: day 14 after hypoxia; Hx_1: day 1 after hypoxia. (E) The enriched GO ontology (biological process) were identified by functional annotation tool DAVID and visualized in bar plot.

FIGURE 4

Identification of metabolism associated DEGs in SMC of SuHx rats and expressions in mouse lungs. (A) Uniform manifold approximation and projection (UMAP) was generated based on Seurat_umap.coords data from an open-access online platform (http://mergeomics.research.idre.ucla.edu/PVDSingleCell/). (B) Overlap of metabolism associated genes (MAGs) and DEGs in smooth muscle cells between Sugen 5416/hypoxia induced PH (SuHx) and control rats were visualized in Venn diagram; 5 overlapped genes was listed as gene set 2. (C) Expression of 4 overlapped genes (Aldh6a1 was not detected in lung tissues from GSE1909 dataset) were visualized in box plot at indicated time point after hypoxia; Hx_1: day 1 after hypoxia, Hx_14: day 14 after hypoxia, Hx_21: day 21 after hypoxia. Data represent mean ± SEM. *p < 0.05, as analyzed by one-way ANOVA with Tukey’s multiple comparisons or Kruskal-Wallis test respectively, as appropriate.

FIGURE 5

Identification of hypoxia induced metabolism associated hub genes and the expressions in hypoxic lungs. (A) Overlap of gene set 1 and gene set 2 were visualized in Venn diagram; 42 genes in union set of both gene sets were regarded as hypoxia induced metabolism associated genes. (B) Protein protein interaction among the 42 genes were analyzed and visualized by online tool STRING (v11.0). (C) Identification of 10 hub genes by cytoHubba plugin with MCC algorithm in Cytoscape (v 3.8.2); each circle represents unique gene and the redder the color is, the higher the MCC score is. (D) Identification of 6 hub genes by MCODE plugin in Cytoscape (v 3.8.2); each circle represents unique gene. (E) Expression of 6 shared hub genes were visualized in box plot at indicated time point after hypoxia or recovery to normoxia; Hx_1: day 1 after hypoxia, Hx_14: day 14 after hypoxia, Hx_21: day 21 after hypoxia, Re_14: day 14 of normoxia recovery after 21 days of hypoxia exposure. Data represent mean ± SEM. *p < 0.05, **p < 0.01, as analyzed by one-way ANOVA with Tukey’s multiple comparisons or Kruskal-Wallis test respectively, as appropriate.

FIGURE 6

Validation of hub genes in lungs of PH patients and a predicting model for PH. (A) Expression of six hypoxia induced metabolism associated hub genes in lung tissues of 58 patients with pulmonary hypertension (PH) and 25 control subjects from GSE117261. (B) ROC curve analysis of training set (GSE117261) using six hub genes. (C) ROC curve analysis of test set (GSE117261) using six hub genes. (D) PCA analysis demonstrated the distribution of data sets GSE117261 (red) and GSE24988 (green) before (upper panel) and after (lower panel) removal of batch effect. The distribution was visualized in scatter plot. (E) Expression of six hypoxia induced metabolism associated hub genes in lung tissues of 120 patients with pulmonary hypertension (PH) and 47 control subjects from GSE117261 and GSE24988 after correction of batch effect. Data represent mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 compared to corresponding control subjects, as analyzed by unpaired t test or Mann-Whitney U test respectively, as appropriate.