Identification of a gene network driving the attenuated response to lipopolysaccharide of monocytes from hypertensive coronary artery disease patients

Abstract

Introduction: The impact of cardiovascular disease (CVD) risk factors, encompassing various biological determinants and unhealthy lifestyles, on the functional dynamics of circulating monocytes-a pivotal cell type in CVD pathophysiology remains elusive. In this study, we aimed to elucidate the influence of CVD risk factors on monocyte transcriptional responses to an infectious stimulus.

Methods: We conducted a comparative analysis of monocyte gene expression profiles from the CTMM - CIRCULATING CELLS Cohort of coronary artery disease (CAD) patients, at baseline and after lipopolysaccharide (LPS) stimulation. Gene co-expression analysis was used to identify gene modules and their correlations with CVD risk factors, while pivotal transcription factors controlling the hub genes in these modules were identified by regulatory network analyses. The identified gene module was subjected to a drug repurposing screen, utilizing the LINCS L1000 database.

Results: Monocyte responsiveness to LPS showed a highly significant, negative correlation with blood pressure levels (ρ< -0.4; P<10^-80). We identified a ZNF12/ZBTB43-driven gene module closely linked to diastolic blood pressure, suggesting that monocyte responses to infectious stimuli, such as LPS, are attenuated in CAD patients with elevated diastolic blood pressure. This attenuation appears associated with a dampening of the LPS-induced suppression of oxidative phosphorylation. Finally, we identified the serine-threonine inhibitor MW-STK33-97 as a drug candidate capable of reversing this aberrant LPS response.

Conclusions: Monocyte responses to infectious stimuli may be hampered in CAD patients with high diastolic blood pressure and this attenuated inflammatory response may be reversed by the serine-threonine inhibitor MW-STK33-97. Whether the identified gene module is a mere indicator of, or causal factor in diastolic blood pressure and the associated dampened LPS responses remains to be determined.

Keywords: circulating monocytes; coronary artery disease; gene regulatory network; hypertension; inflammatory responses.

Figure 1

Genes’ responses to LPS were weakened in monocytes of the CAD patients with high blood pressure. (A) Schematic diagram of the experimental set up. (B) Volcano plot showing the correlation between CVD risk factors and the average LPS response signature. A positive correlation of a risk factor means the increase of this trait is associated with a stronger LPS response. (C) The correlations among 5 CVD risk factors based on the values from clinical records. P-values of correlation coefficients are shown in the boxes. FDR adjusted pvalue are denoted by **Padj < 0.01, ***Padj < 0.001. (D) Heatmap shows that the LPS responses of 13 typical genes in TLR4 pathway and their correlation with DBP and SBP are in reverse. Genes were color-coded green through to magenta to indicate value of correlation coefficients. Bar plots displayed the genes’ log₂FCs. (E) Box plots compared the LPS responses of CAD patients with normal (DBP<80 or SBP<130) and high (DBP>=80 or SBP>=130) blood pressure levels, on down and up regulated genes respectively. A point on a boxplot represents the median of the most significantly up/down regulated gene’s LPS responses of a CAD patient. All correlations were calculated using Pearson’s product moment correlation coefficient. Statistical significances were calculated using Student t-test. LPS: lipopolysaccharide, BMI: Body Mass Index. HDL, High-density lipoprotein; BP, blood pressure; SBP, systolic blood pressure; DBP, diastolic blood pressure; TG, triglyceride.

Figure 2

WGCNA network analysis. (A) Schematic diagram of the co-expression network build-up. (B) The correlations between eigengenes of 10 gene modules and DBP and SBP. P-values of correlation coefficients are shown in the boxes. FDR corrected pvalue are denoted by *Padj < 0.05. (C) Volcano plot showing each gene’s average LPS response and its association with DBP in module salmon and cyan. Genes were color-coded green through to magenta to indicate value of correlation coefficients. (D) A dot plot visualized the significant levels of top enriched GO terms of 2 DPB-related gene module (salmon and cyan). Significant levels were evaluated by Fisher’s exact test. Dot plots were color-coded in red (log₂FC>0) and blue (log₂FC < 0). Top 5 significantly enriched GO terms of module salmon were highlighted in bold. Significant levels are shown by using log10-transformed adjusted p-values. (E) A Bayesian network to infer the causal relations between 10 gene modules and DBP. LPS, lipopolysaccharide; SBP, systolic blood pressure; DBP, diastolic blood pressure; GO, Gene Ontology; BP, blood pressure.

Figure 3

WGCNA and gene regulatory network of module salmon. (A) Co-expression network of module salmon. A node in the network represents a gene. The colors of node borders indicate the correlations between LPS responses and DBP. The color of a node stands for the average LPS response of this gene. Genes with red symbols are genes in the GO term: oxidative phosphorylation. (B) Scatters displaying the association between DBP levels and LPS response of gene ATP5J and COX7C. (C) Two networks showing the targets of ZNF12 and ZBTB43 in sub-gene regulatory network of module salmon. An ellipse represents a gene, and a diamond node stands for a transcription factor. A node’s color stands for the average LPS response of this gene. The color of edges of the network indicates the degree of a transcription factor repression (in blue) or activate (in red) its targets. (D) Association of the LPS response between ZNF12 and ATP5J, and between ZBTB43 and COX7C respectively. LPS, lipopolysaccharide; DBP, diastolic blood pressure; TF, transcription factor.

Figure 4

Drug reproposing based on the hub genes from module salmon. (A) Schematic diagram of the pursued drug repurposing pipeline. (B) Bar plot showing the 20 most significant drugs that can enhance the LPS responses in the result of drug reproposing. Drugs were ranked based on the Connectivity Score (or Enrichment Score). Bars of drugs were color-coded from white to red based on -log10(P).