Gene Coexpression Networks in Whole Blood Implicate Multiple Interrelated Molecular Pathways in Obesity in People with Asthma

Abstract

Objective: Asthmatic children who develop obesity through adolescence have poorer disease outcomes compared with those who do not. This study aimed to characterize the biology of childhood asthma complicated by adult obesity.

Methods: Gene expression networks are powerful statistical tools for characterizing human disease that leverage the putative coregulatory relationships of genes to infer relevant biological pathways. Weighted gene coexpression network analysis of gene expression data was performed in whole blood from 514 adult asthmatic subjects. Then, module preservation and association replication analyses were performed in 418 subjects from two independent asthma cohorts (one pediatric and one adult).

Results: A multivariate model was identified in which three gene coexpression network modules were associated with incident obesity in the discovery cohort (each P < 0.05). Two module memberships were enriched for genes in pathways related to platelets, integrins, extracellular matrix, smooth muscle, NF-κB signaling, and Hedgehog signaling. The network structures of each of the obesity modules were significantly preserved in both replication cohorts (permutation P = 9.999E-05). The corresponding module gene sets were significantly enriched for differential expression in subjects with obesity in both replication cohorts (each P < 0.05).

Conclusions: The gene coexpression network profiles thus implicate multiple interrelated pathways in the biology of an important endotype of asthma with obesity.

Figure 1:. Analysis overview.

In this schematic of this gene expression study of obese asthma, cylinders represent inputs and outputs, and boxes represent analysis steps. More detail for each analysis stage can be found in the Methods section and in the Supporting Information. QC, quality control; WGCNA, weighted gene co-expression network analysis; GSEA, gene set enrichment analysis.

Figure 2:. Hierarchical relationships of 17 modules identified in CAMP.

Top panel shows eigengene dissimilarity clustering. Bottom panel shows pair-wise Pearson correlations among the eigengenes, ranging from –1 (blue) to 1 (red). The order of the colored columns corresponds to the color names in the top panel. Black boxes denote those modules belonging to the larger and smaller meta-modules. Asterisks denote modules associated with obese asthma in multivariate regression modeling (Table 2). ME, module eigengene.

Figure 3:. Associations of four eigengenes with obese asthma in CAMP.

Panels show violin plots of eigengene residuals by obesity status. Residuals include adjustments for demographic characteristics and the other three eigengenes (see Table 2). Each dot represents an individual CAMP subject. Horizontal lines indicate 25^th, 50^th, and 75^th percentiles. In this multivariate context, the black, midnightblue, and grey60 eigengenes are significantly different between obese and non-obese asthmatic subjects (P < 0.05), while the lightcyan eigengene is not.

Figure 4:. Preservation of obese asthma modules in CHS and GACRS.

The first (top) panel shows a heatmap of pair-wise correlations among the genes comprising the black (1), lightcyan (2), grey60 (3), and midnightblue (4) modules [see ‘cor.cor’ and ‘avg.cor’ metrics defined in Table S6]. The second panel shows a heatmap of the edge weights (connections) among the genes comprising the four modules [‘avg.weight’]. The third panel shows the distribution of scaled weight degrees (relative connectedness) among the genes comprising the four modules [‘cor.degree’]. The fourth panel shows the distribution of node contributions (correlation to module eigengene) among the genes comprising the four modules [‘cor.contrib’ and ‘avg.contrib’]. The ‘coherence’ metric is not summarized in this figure. Genes are ordered from left to right based on their weighted degree in CAMP (the discovery cohort) to highlight the consistency of the network properties in CHS and GACRS (the replication cohorts).