Initiation and progression of atherosclerosis occurs focally at regions of bifurcations or high curvature where complex shear stress patterns exist. Shear stresses derived from regions of disease are low and reversing, and they promote a proinflammatory or "atheroprone" endothelial cell (EC) phenotype. Hemodynamics from regions absent of disease have comparatively higher levels of unidirectional shear stress that promotes an "atheroprotective" and anti-inflammatory phenotype.

To study how the hemodynamic environment differentially dictates endothelial phenotype, atheroprone and atheroprotective shear stresses derived directly from the human vasculature were recapitulated in vitro on human ECs and changes in cell signaling were assessed.

First, endoplasmic reticulum (ER) stress protein, GRP78, exists at every stage of lesion development, even prior to disease, and is regulated in the endothelium by several atherosclerotic stressors. Increased endothelial expression of GRP78 was observed in atheroprone versus atheroprotective regions of C57/BL6 mice, and, in vitro, GRP78 was significantly upregulated after 24 hours of atheroprone, but not atheroprotective flow. This response was dependent on the p38/a2ßl pathway. Increased GRP78 correlated with ATF6 activation ofthe ER stress sensing element (ERSEl) promoter by atheroprone flow signifying the activation of the UPR. This study supports a role of the hemodynamic environment in preferentially inducing GRP78 and the UPR in atheroprone regions, prior to lesion development.

Secondly, focal activation of NF-κB and deposition of the extracellular matrix protein, fibronectin (FN), are found in regions of atherosclerosis where they contribute to inflammatory signaling. We sought to elucidate the mechanism by which NF-κB and FN are regulated by local shear stress patterns, its dependence on PECAM-I mechanotransduction, and the role these pathways play in sustaining an atheroprone/proinflammatory phenotype. In ????"? mice and in vitro, atheroprone flow induced elevated levels of NF-κB activity compared to atheroprotective flow, which was attenuated by knock-out or siRNA-mediated depletion of PECAM-I. Additionally, atheroprone flow caused a steady and sustained increase in FN expression over time significantly greater than atheroprotective flow. Comparing FN staining in ApoE7" and ApOE7PECAM'7- mice showed that PECAM-I was essential for FN accumulation in atheroprone regions of the aortic arch. In vitro, siRNA against PECAM-I blocked the induction of FN and the activation of NF-κB by atheroprone flow, which was rescued by the addition of exogenous FN. Additionally, blocking NF-κB activation attenuated the flow-induced FN expression, and siRNA against FN significantly reduced NF-κB activity. Thus, atheroprone shear stress creates positive feedback to maintain inflammation.

Lastly, we hypothesize that the underlying mechanisms responsible for the induction of unique endothelial phenotypes by atheroprone and atheroprotective hemodynamics are rooted in distinct frequency spectrums. Steady flows of the same magnitude, but lacking frequency, did not activate NF-κB to the levels ofthe physiologic flows containing frequency;​ thus frequency content must contribute to endothelial phenotype. In order to investigate the roles of frequency, harmonics 0-8 were systematically swapped between atheroprone and atheroprotective shear stresses using Fourier Transforms to create new "mutated" waveforms. The protective waveform displayed a protective phenotype independent of frequency. For the atheroprone waveform, raising the amplitude of the 0th or 1st harmonic reduced NF-κB levels, where mutations in harmonics 2-8 resulted in elevated NF-κB similar to the control. First harmonic mutations also decreased pro-inflammatory genes and proteins, such as VCAM-I, FN, and E-Selectin, while activating higher levels of KLF2 transcription. Atheroprone flow was characterized by low 0th and 1st harmonics that promote NF-κB activation, but, swapping either elevated 0th or 1st harmonics from the atheroprotective waveform can promote an atheroprotective phenotype. Frequency-dependent activation of NF-κB and KLF2 were found to depend on PECAM-I, where its removal resulted in reversal of activity levels from the controls. The relationship between the 0th and 1st harmonics was quantified and validated in a mathematical model that was able to predict NF-κB levels and was more sensitive than other commonly used hemodynamic parameters. Therefore shear stress magnitude alone does not control endothelial phenotype, but the amplitude ofmultiple harmonics.