Identification of intima-to-media signals for flow-induced vascular remodeling using correlative gene expression analysis

Figure 1. WSS at the BA and BT before and after carotid ligation. (A) 3D renderings showing the distribution of WSS at the luminal surface of the BA and BT. WSS was calculated by CFD as described in Materials and Methods based on the indicated flow rates, which were measured by TCD sonography. (B) Average WSS values are shown for each location and condition. WSS in the BA was averaged over the length of the vessel, the value for the BT was averaged over the vessel surface within a 1 mm diameter sphere centered at the apex of the bifurcation (arrow). WSS is higher at the BT than the BA, and is increased at both locations after ligation, while sham surgery had no effect.

Figure 2. Principal component analysis of gene expression levels in the 16 samples, and differential expression analysis. (A, B). Principal component analysis was performed on the gene expression levels for all genes that had a count sum > 0 across all samples. The first two principal components for the collective gene expression in each sample are plotted for samples from (A) straight vessel segments and (B) bifurcations. In both cases, intima samples (circles) are spatially separated form media samples (squares). In addition, sham tissues (solid symbols) are distinguished from tissues taken from ligated animals (hatched symbols). Colored ellipses indicate paired duplicates from 2 different rabbits. (C–F) Gene expression in tissues from ligated and unligated rabbits was compared by RNA-seq. Volcano plots show fold change and p-value for comparisons at BA intima (C), BT intima (D), BA media (E), and BT media (F). Each dot represents one gene, and red dots indicate DEGs, i.e. detectable genes with q-value < 0.05.

Figure 3. Biological processes associated with medial correlates of flow-sensitive intimal genes. For all the medial genes whose expression correlated with the 18 flow-sensitive, extracellular-coding intimal genes that had > 200 medial correlates, significantly (q-value < 0.05) over-represented GOs were identified using g:Profiler. The diagrams summarize the biological processes terms that were present in > 50% of GOs enriched in (A) genes positively correlated with intimal genes exhibiting higher expression in ligated animals compared to sham, and (B) genes negatively correlated with intimal genes exhibiting lower expression in ligated animals than in sham. Symbol sizes represent the frequency of the ontology term, and lines between symbols indicate relationships. Groups of closely related terms are denoted by matching color and dashed ellipses. Note that organonitrogen compound metabolism is the largest symbol in (A), and one of the largest in (B). Both diagrams show multiple terms associated with leukocyte activation/immune responses (red) and secretion/exocytosis (green). Terms related to wound healing and cell movement are also present in both summaries.

Figure 4. Gene expression as a function of local WSS. The expression levels for the 18 intimal genes with > 200 medial correlates (as determined by RNA-seq) are shown as a function of the WSS at the sample’s location (as determined by CFD). Expression levels of SERPINE6 (A), SERPINE1 (B), ADAMTS9 (C), TIMP1 (D), and IL15 (E) were all positively correlated with WSS over the full range of observed values (Pearson r > 0.9). GLIPR1 (F) increased with increasing WSS, but plateaued at WSS > 40. The remaining genes displayed very non-linear “dose–response” curves, shown in (G) CYTL1, (H) LUM, CST6, SSP1, CYR61, ABI38P and OMD, and (I). NDP, C1QTNF7, LYPD6, PLA2G7, and MFAP5. The maximal expression values for these 12 genes are all at WSS below 40 Pa, and expression is dramatically lower at WSS > 50 Pa.