Hypoxia-Inducible Factor 1α drives endothelial metabolic changes and inflammation under atheroprone flow

To the Editor

We read with interest your recent paper “Mechanical Activation of Hypoxia-Inducible Factor 1α Drives Endothelial Dysfunction at Atheroprone Sites” in which Feng et al. elegantly show that exposure of endothelial cells to mechanical low shear stress activates hypoxia-inducible factor 1α (HIF-1α).¹ Since atherosclerosis develops near branches or bends of arteries where endothelial cells are exposed to low shear stress, these results suggest that HIF-1α activation in endothelial cells may play a causal role in the pathogenesis of atherosclerosis.

Feng and colleagues showed that upregulation of HIF-1α occurs via a dual mechanism involving transcriptional activation by nuclear factor-κB (NF-κB) and stabilization via the deubiquitinating enzyme Cezanne.

We were pleased to see that the study by Feng and colleagues largely recapitulated our recently published findings.² In our study, we took an unbiased approach and performed RNAseq in human arterial ECs exposed to either unidirectional flow (UF: athero-protective hemodynamics of high shear stress measured in human distal internal carotid artery) or disturbed flow (DF: athero-susceptible hemodynamics of low shear stress measured in human carotid sinus)³ to investigate the effects of shear stress on ECs. Analysis of our transcriptomic data showed that the dominant transcriptional events driven by DF were enriched with HIF-1α targets as well as glycolytic enzymes. We confirmed the DF-induced stabilization of HIF-1α at the protein level. Using bioenergetic studies, we also confirmed that DF induces glycolysis and reduces mitochondrial respiration similar to what is observed in cancer cells.⁴ Importantly, we discovered that HIF-1α was required and sufficient for DF-induced upregulation of glycolysis and downregulation of mitochondrial respiration, which played a key role in EC activation assessed by NF-κB activation and expression of pro-inflammatory cytokines and adhesion molecules. While our conclusions about HIF-1α playing a critical role in EC activation under DF agree with those of Feng et al¹, there are some differences between the two studies. We found that DF induced NADPH oxidase 4 (NOX4)-dependent generation of reactive oxygen species (ROS), which were required for HIF-1α stabilization and downstream effects on glycolysis and EC activation. Supporting a role for this mechanism in vivo, ECs in disturbed flow areas of porcine and mouse aorta showed increased expression of NOX4 and high levels of ROS in addition to HIF-1α and glycolytic enzymes. While we did not specifically investigate the mechanisms by which ROS mediate HIF-1α expression under DF, it is likely due to inhibition of the hydroxylation of two proline residues, mediated by a family of oxygen-dependent prolyl-4-hydroxylase domain enzymes.⁵ Although Feng et al state that HIF-1α accumulation under DF is not mediated via reduced hydroxylation or ubiquitination, they do not provide data on ubiquitination or hydroxylation of proline residues to support their conclusion. Instead, the authors provide an alternative mechanism that involves DF induced expression of deubiquitinating enzyme Cezanne, which rescues HIF-1α from degradation. ROS have been shown to inhibit deubiquitinases including Cezanne.^{6, 7} In fact, Dr. Evans’ group has previously shown that ROS reverses the Cezanne-induced negative regulation of NF-κB signaling.⁷ These data suggest that the regulation of DF-induced HIF-1α expression is complex involving ROS, and changes in ubiquitination and deubiquitination. Further studies will be needed to determine interaction between ROS, prolyl hydroxylases and Cezanne in HIF-1α expression under DF.

Another difference between two studies is the interaction between HIF-1α and NF-κB. In contrast to Feng et al who demonstrated that NF-κB is upstream of HIF-1α, we did not see this effect, as inhibition of NF-κB did not alter DF-induced HIF-1α expression. Instead, we found that HIF-1α drives NF-κB activation and consequently pro-inflammatory cytokine gene expression and EC activation. Since ROS can reverse the inhibitory effects of Cezanne on NF-κB⁷, it is possible that DF-induced NOX4/ROS may mediate its stimulatory effects on NF-κB via inhibition of Cezanne.

Our findings and the results from Feng and colleagues provide evidence for low shear stress inducing the expression of HIF-1α under normoxia, which is required for EC metabolic reprogramming and inflammation. Notably, both studies demonstrated that disturbed flow, when compared to static (no flow) conditions, significantly induces HIF-1α expression in vascular endothelium. These results suggest that cellular metabolism and related inflammation are quite distinct in vascular endothelium under flow and no flow conditions and moreover, caution should be taken to interpret results collected from vascular endothelium under static conditions. In summary, targeting DF-induced HIF-1α stabilization and/or EC metabolic changes may potentially lead to new therapies for atherosclerosis.