March 23, 2015. Blood gene expression represents a potentially exciting new approach to dynamically monitor health, first explored by 'omics pioneer Mike Snyder at Stanford. Last week's PLOS Genetics publication of blood RNAs linked to hypertension, "A Meta-analysis of Gene Expression Signatures of Blood Pressure and Hypertension" demonstrates the emerging use of RNA expression to characterize specific biological conditions. Led by Daniel Levy, MD, the authors identify 34 genes in blood RNA that correlate with blood pressure, mRNAs that in aggregate (top 55 SBP, 22 DBP) explain an estimated 5-9% of the variation in BP.
To better understand how circulating lymphocytes can sense hypertension, we analyzed the 34 gene signature with Molquant analytical tools. Using a large diverse set of expression data, networks are generated and visualized to identify co-expression relationships among genes--correlated networks are inferred to exhibit similar biology. Note that here the GTEX tissue dataset (2900 RNAseq samples from human tissues) was not used to form the networks, such that the GTEX plots shown here represent an independent assessment of the networks identified.
Figure 1 shows expression of the genes across the GTEX project tissue survey data. Consistent with the lymphocytic origin of the RNA, a strong enrichment for blood expression was seen in the signature genes.
To provide a more refined view of the 34 genes, Figure 2 shows the networks of each gene plotted (white) along with networks for several previously generated (1) hematopoietic lineages (red).
Two dominant, and related biological networks were observed, one tightly linked to T cell and macrophage lineages, another comprising genes previously associated with inflammation (COX-2/PTGS2, GADD34/PPP1R15A, DUSP1, FOS).
These networks bring a finer resolution to the findings of the authors who noted that GSEA analysis identified inflammation and apoptosis as top enriched pathways. Together, these observations enable a hypothesis that the identified signature represents an activated hematopoietic signature associated with inflammation. Inflammation has long been linked with cardiovascular conditions linked to hypertension (e.g. Inflammation in Hypertension).
The network analysis further suggests caution in linking signature genes to specific non-hematopoietic mechanisms of hypertension. For example, figure 2 identifies links between KCNJ2 and hematopoiesis (figure 2). While KCNJ2 is expressed in heart, the GTEX tissue expression survey identifies blood as the top expressing tissue for KCNJ2, consistent with our analysis. In addition, genes identified by the authors as representing a co-expression network in neutrophils exhibit limited linkage to neutropils in either figure 2 or the GTEX dataset (figure 3).
While this work represents an exciting addition to the use of genomics data to characterize the biology of hypertension, we would agree with the authors that one should be cautious about interpreting hypertension biology from hematopoietic cell transcriptomes. It appears that blood based signatures can sense medically relevant conditions, but perhaps they do so by reflecting the impact of a condition on the hematopoietic system, not by revealing the genes driving the underlying mechanisms. Transcriptomic analysis from less accessible tissues such as endothelial, cardiac or neural tissue may provide a more direct window to mechanism.