Hydrogen peroxide regulation of endothelial function: Origins, mechanisms, and consequences

Abstract

Increased production of reactive oxygen species (ROS) has been implicated in the pathogenesis of cardiovascular diseases. Enzymatic systems such as the mitochondrial respiratory chain, vascular NAD(P)H oxidases, xanthine oxidase, and uncoupled endothelial nitric oxide synthase (eNOS) produce superoxide anion (O₂^.−) in vascular cells. While some O₂^.− rapidly degrades by reacting with nitric oxide (NO^.), the O₂^.− signal preserved by dismutation into hydrogen peroxide (H₂O₂) exerts prolonged signaling effects. This review focuses on patterns and mechanisms whereby H₂O₂ modulates different aspects of endothelial cell function including endothelial cell growth and proliferation, endothelial apoptosis, endothelium-dependent vasorelaxation, endothelial cytoskeletal reorganization and barrier dysfunction, endothelial inflammatory responses, and endothelium-regulated vascular remodeling. These modulations of endothelial cell function may at least partially underlie H₂O₂ contribution to the development of vascular disease.

Increased production of reactive oxygen species (ROS) has been implicated in the pathogenesis of cardiovascular diseases such as atherosclerosis, restenosis, hypertension, diabetic vascular complications and heart failure [1–7]. One electron reduction of molecular oxygen forms superoxide anion (O₂^.−), which serves as a progenitor for hydrogen peroxide (H₂O₂). By rapidly inactivating nitric oxide (NO^.), O₂^.− contributes to endothelial dysfunction [7]. H₂O₂ however modulates endothelial cell function via intricate mechanisms. Ambient production of O₂^.− and subsequently H₂O₂ at low levels, maintained by basal activity of pre-assembled endothelial NAD(P)H oxidases [3], or potential leakage from mitochondrial respiration [8], is necessary for endothelial cell growth and proliferation. Under pathological conditions, agonists-provoked activation of vascular NAD(P)H oxidases and subsequently activated xanthine oxidase or uncoupled eNOS produce H₂O₂ in large quantities [9–11], leading to detrimental consequences [12,13]. These dual-faced functions of H₂O₂ are similar to what is characteristic of NO^.: vasodilating and protective at physiological concentrations but microbicidal and apoptotic when produced in large quantities. The current review discusses enzymes responsible for endothelial production of ROS, patterns and mechanisms whereby H₂O₂ influences different aspects of endothelial cell function and H₂O₂ contribution to vascular disease development.

1 Production and metabolism of hydrogen peroxide

In mammalian cells, potential enzymatic sources of ROS include the mitochondrial electron transport chain, the arachidonic acid metabolizing enzymes lipoxygenase and cycloxygenase, the cytochrome P450s, xanthine oxidase, NAD(P)H oxidases, uncoupled nitric oxide synthase (NOS), peroxidases, and other hemoproteins. These systems primarily catalyze one electron reduction of molecular oxygen to form O₂^.− which rapidly inactivates NO^. to form peroxynitrite. Under ambient conditions, some O₂^.− is dismutated to H₂O₂ spontaneously or catalyzed by superoxide dismutase (SOD). Of interest, loss of NO^. could lead to enhanced formation of H₂O₂. Some enzymes, such as xanthine oxidase and glucose oxidase, can directly produce H₂O₂ by donating two electrons to oxygen. In the presence of heavy metals, H₂O₂ undergoes Fenton reaction to form highly reactive hydroxyl radical (HO^.). When bound to peroxidases such as catalase, H₂O₂ forms compound I which oxidizes NO^. to nitrogen dioxide anion (NO₂⁻) and react with NO₂⁻ to form nitrogen dioxide radical (NO₂^.). NO₂^. in turn participates in nitrating events such as formation of nitrotyrosines [14,15]. Fig. 1 illustrates biochemical pathways of H₂O₂ generation and metabolism. When produced transiently at high concentrations, H₂O₂ and HO^. are capable of directly oxidizing proteins and lipids, and causing DNA strand breaks. At relatively lower pathophysiological concentrations (nano- to micromolar range) however, H₂O₂ plays important signaling roles in endothelial cells, and these are the focus of the current review.

2 Vascular NAD(P)H oxidases

Accumulating evidence has characterized a predominant role of vascular NAD(P)H oxidases in generating ROS in different vascular cells including endothelial cells, vascular smooth muscle and fibroblasts [1–6]. In endothelial cells, though there is much to be learned, it is clear that p47^phox modulates enzymatic activity by interacting with membrane and cytosolic components to form the active complex. The gp91^phox [also called Nox2, Nox for NAD(P)H oxidases, representing a family of novel NAD(P)H oxidases] is the catalytic core of this complex [16–19]. Studies using deficient mice or inhibitory peptide targeting Nox2 have established an essential role of Nox2 in producing ROS in endothelial cells [17,18,20,21]. Functions of the newly cloned gp91^phox homologues; Nox1, Nox4 and Nox5 however, remain obscure. Nox4 seems to express more abundantly in endothelial cells compared to other Nox proteins, representing the major catalytic unit of the endothelial NAD(P)H oxidase responsive to growth halting [22,23]. Nox1 on the other hand, was upregulated by oscillatory shear stress, mediating ROS-dependent leukocyte adhesion to endothelium [24]. Furthermore, VEGF receptor-dependent activation of Nox1 induced angiogenic tube formation of endothelial cells [25]. The observations that Nox1 mediates growth signaling while Nox4 is growth suppressive in endothelial cells seems similar to what have been observed in vascular smooth muscle [26,27]. It seems odd that the same ROS-producing Nox proteins mediate different cellular responses. It is possible that the function of each Nox protein is dependent on its distinctive subcellular localization, and is subject to specific regulations by selective agonists [4]. For example, in vascular smooth muscle cells, Nox1 localizes to caveolae while Nox4 is found in focal adhesions [28]. In endothelial cells however, Nox4 was found at endoplasmic reticulum [23] whereas Nox2 is localized to peri-nuclear cytoskeletal structure [17].

Novel homologues of Nox-regulating proteins p47^phox and gp67^phox have been identified in epithelial cells (p41^phox and p51^phox respectively), serving as potent positive regulators for Nox1 [29–32]. In the same cells longer Nox proteins with peroxidase tails, namely Duox1 and Duox2, have been cloned [33,34] and shown to be functional [35,36]. These proteins are yet studied for their presence and function in endothelium and vascular smooth muscle. Of note, p22^phox is the only other membrane component of the vascular NAD(P)H oxidases. Overexpression of p22^phox led to stabilization of Nox1 and Nox4 proteins [37]. Recent studies have elegantly characterized physical and functional interactions between p22^phox and Noxs (Nox1 and Nox4) in vascular smooth muscle [38,39]. Similar interactions might also occur in endothelial cells. In addition, a recent study reported that p22^phox expression correlates well with expression of Nox4 in human arteries and that of Nox2 in veins [40].

In addition, vascular NAD(P)H oxidase-derived H₂O₂ is able to amplify ROS production in vascular cells. To date at least five different mechanisms are involved in this propagation, including enhanced intracellular iron uptake, and activation of sources of mitochondria, NAD(P)H oxidases, xanthine oxidase and uncoupled eNOS [1]. These feed-forward mechanisms form a vicious circle to amplify and sustain H₂O₂ production in large quantities, contributing to prolonged pathological signaling.

3 Hydrogen peroxide regulation of endothelial cell growth and proliferation

H₂O₂ generated by xanthine oxidase or glucose oxidase stimulated endothelial cell growth [41]. Consistently, scavenging H₂O₂ by adenoviral overexpression of catalase or cytosolic glutathione peroxidase inhibited endothelial cell proliferation [42,43]. These data indicate a role of H₂O₂ in growth signaling. Indeed, H₂O₂ is downstream of Flk1/KDR, proceeding activation of ERK1/2 [44]. The growth regulating p90RSK, an important downstream effector of ERK1/2, is activated by H₂O₂ in endothelial cells [45]. Furthermore, H₂O₂ feed-forwardly increases endothelial VEGF expression and secretion via activation of PKC and NFκB [46], and regulates expression and function of other growth factors [47]. Growth signaling induced by agonists other than growth factors may also require H₂O₂. For instance, cyclic strain induces early growth factor 1 (Egr-1) expression and this requires H₂O₂-dependent activation of ERK1/2 [48].

Furthermore, H₂O₂ increases activity of eIF4E, an eukaryotic initiation factor of protein synthesis [49], and is effective in promoting endothelial cell tube formation [50,51]. Hypoxia induced activation of ERK1/2 and p38 MAPK in lung endothelial cells is mediated by H₂O₂[52]. This may partially underlie hypoxia-induced angiogenesis. Indeed, endogenously produced H₂O₂ upregulates VEGF expression and angiogenesis in vivo [53]. The signaling events involved in H₂O₂ modulation of endothelial cell growth, and later discussed endothelial apoptosis are illustrated in Fig. 2. Whereas a modest increase in endothelial cell growth and angiogenesis is beneficial in repairing ischemic damage [54], excessive growth contributes to atheroma formation post injury [53]. In addition, when H₂O₂ diffuses to adjacent vascular smooth muscle, it could induce hypertrophy [13]. Indeed, endogenously overproduced H₂O₂ augmented angiotensin II induced vascular hypertrophy [55].

4 Hydrogen peroxide regulation of endothelial apoptosis

Cultured umbilical vein endothelial cells are particularly prone to oxidative damage and often used to study H₂O₂ induced apoptosis. Of note, aortic endothelial cells undergo no significant cell death with exogenous addition of H₂O₂ less than 200 μmol/L [56–58]. However, 50–100 μmol/L H₂O₂ was apoptotic for umbilical vein endothelial cells. Endogenous H₂O₂, stimulated by oxidized LDL, induced endothelial cell apoptosis via JNK activation [59]. Mitochondrion-dependent ROS propagation is likely involved because mitochondria-targeted antioxidant abrogated H₂O₂ activation of JNK and apoptosis [60]. Studies by other groups also support a role for JNK/c-Jun in H₂O₂ induced apoptosis of endothelial cells [59,61,62]. Of note, Chen et al. previously characterized upstream events of JNK activation by H₂O₂, which involve Src-dependent activation of EGFR [63]. Additionally, H₂O₂ upregulates Fas expression [64], enhances intracellular ion uptake [65], and induces mitochondrial DNA damage [66], all of which may directly trigger apoptosis.

Importantly, endothelial cell apoptosis has been implicated in atherosclerosis [67,68] although molecular mechanisms underlying H₂O₂ induced endothelial cell apoptosis remain to be fully elucidated. It is however interesting to note that H₂O₂ induced apoptosis is inhibitable by unidirectional laminar shear stress, which exerts this effect by attenuating JNK activation and enhancing glutathione reductase-dependent glutathione redox-cycling [69,70]. Furthermore, S-nitrosylation of thioredoxin 1 [71] and ERK5 activation [72] was found protective of endothelial cells from apoptosis.

5 Hydrogen peroxide regulation of endothelial barrier function and actin cytoskeleton

Relatively high concentrations of H₂O₂ (>200 μmol/L) increase endothelial permeability [73], and this seems mediated by activations of PKC [74], phosphodiesterase [75], small G protein Rho [76], Src tyrosine kinase [77], p38 MAPK [78], and an increase in intracellular calcium [79]. Tight junction protein occludin was found downstream of ERK1/2 in mediating H₂O₂-induced barrier dysfunction [80]. In addition, H₂O₂ downregulation of intracellular cAMP content appears important in inducing barrier dysfunction [75,81]. Decreased PKA activity can itself account for many of the activating effects that result in barrier dysfunction or at least permit those changes [82]. Crosstalks or spatial regulations among above-described signaling events however remain to be revealed. Caveolin-1 is expressed abundantly in lipid rafts of endothelial cell membrane, where it plays an important role in modulating barrier function [83]. Whether caveolin-1 is required for H₂O₂ modulation of endothelial barrier function however is unclear although our unpublished data suggest that H₂O₂ induces rapid Src-dependent phosphorylation of caveolin-1. Of note, at various concentrations, NO^. is able to enhance or offset H₂O₂-induced loss of endothelial barrier function [84,85].

It has been shown that cytoskeletal reorganization modulates vascular permeability [86]. Likewise, H₂O₂ regulation of barrier function is linked to its effects on endothelial actin cytoskeleton. p38 MAPK phosphorylation of actin binding protein HSP27 mediates reorganization of actin cytoskeleton in H₂O₂-stimulated endothelial cells [87]. In keeping with this, H₂O₂ induced formation of actin stress fibers was found to be mediated by CaMKII-dependent activation of p38 MAPK/HSP27 and ERK1/2 [88]. Whereas HSP27 phosphorylation was completely prevented by blockade of p38 MAPK, inhibition of ERK1/2 only transiently attenuated HSP27 phosphorylation [88]. These data suggest that ERK1/2 has different downstream effectors in modulating actin cytoskeleton.

It was previously shown that endothelial MLCK is involved in H₂O₂ induced actin reorganization [89]. MLCK also plays an important role in thrombin induced endothelial barrier dysfunction [86]. Thrombin has been shown to stimulate endothelial NAD(P)H oxidase activation and subsequent H₂O₂ production [90]. Recent studies further demonstrated that thrombin upregulates p22^phox expression via p38 MAPK and PI-3K/AKT [91]. Given these observations, it is interesting to speculate that MLCK lies downstream of ERK1/2 in mediating H₂O₂ induced actin reorganization. In addition, H₂O₂ phosphorylation of tropomyosin-1 is ERK1/2-dependent, preceding its co-localization with developing actin filaments [92]. Of note, H₂O₂ activation of phospholipase D is also involved in actin reorganization in endothelial cells [93], likely serving as upstream activator of ERK1/2 [94]. The signaling events described above are summarized as part of the Fig. 3. Importantly, H₂O₂ regulation of endothelial cytoskeleton and barrier function may not only mediate ischemia–reperfusion induced acute vascular injury [73], but also potentiate vascular inflammation to contribute to chronic vascular disease [95].

6 Hydrogen peroxide regulation of endothelium-dependent vasorelaxation and eNOS expression

Earlier studies demonstrated that H₂O₂ produces endothelium-dependent and independent vasorelaxation [96,97]. The endothelium-dependent vasorelaxation caused by H₂O₂ in large vessels depends on eNOS, as L-NAME (NOS inhibitor) abolishes it [98,99]. Indeed, H₂O₂ can acutely stimulate eNOS production of NO^. via PI3-K/AKT and ERK1/2, [58,100]. When eNOS was uncoupled to produce O₂^.− rather than NO^. in hypertensive or atherosclerotic large vessels, endothelium-derived H₂O₂ mediated compensatory relaxation via unknown mechanisms [10,101]. One possibility is direct polarization of vascular smooth muscle.

Indeed, H₂O₂ was found to be an endothelium-derived hyperpolarizing factor in small arteries [102,103] and an activator of potassium channel in large cerebral arteries [104]. In human coronary arterioles, flow-induced vasodilatation is mediated by endothelium-derived H₂O₂[105]. In these arterioles, the enzymatic source of H₂O₂ appears to be the mitochondrial respiratory chain whereas in large vessels, vascular NAD(P)H oxidases are responsible for H₂O₂ production [6,8]. This seems consistent with previous observations that NO^. plays a lesser role in vasodilatation of small arteries. Of note, in coronary arterioles, H₂O₂ induced endothelium-dependent vasorelaxation is NO^.-independent involving activation of cyclooxygenase 1 and smooth muscle potassium channel [106]. Therefore, it is clear that H₂O₂ is capable of mediating endothelium-dependent vasorelaxation, NO^.-dependently or independently. The underlying mechanisms however are diverse depending on the size of the blood vessels and the availability of functional eNOS to produce NO^.[107,108].

H₂O₂ can also stimulate endothelium-independent vasorelaxation with unknown mechanisms [109–111]. Early studies demonstrated that H₂O₂ can directly activate cyclic GMP via a compound I/H₂O₂ complex [109,110]. Additional recent studies suggest that this might be true [112]. Known and hypothetical mechanisms underlying H₂O₂ induced vasorelaxation are summarized schematically in Fig. 4.

Of note, H₂O₂ potently upregulates eNOS expression in vitro and in vivo [37,56,113]. Oscillatory shear stress upregulation of eNOS requires H₂O₂ activation of CaMKII [114]. These observations are consistent with previous findings that eNOS is upregulated in animal models of atherosclerosis, diabetes or aging, where vascular H₂O₂ production is increased [115,116]. Rapidly accumulating evidence suggests that eNOS can become uncoupled to produce O₂^.− under conditions such as hypertension and atherosclerosis, likely consequent to tetrahydrobiopterin deficiency and oxidation. Thus upregulation of eNOS could be less beneficial or even detrimental while functioning as an oxidase. Indeed, our recent study indicates that endothelial NAD(P)H oxidase-derived H₂O₂ downregulates tetrahydrobiopterin salvage enzyme dihydrofolate reductase (DHFR) and DHFR/eNOS ratio in response to angiotensin II, leading to tetrahydrobiopterin deficiency and subsequently uncoupling of eNOS [9].

7 Hydrogen peroxide regulation of endothelial inflammatory responses

Exposure of endothelial cells to H₂O₂ (50–100 μmol/L) increased surface expression of intercellular adhesion molecule-1 (ICAM-1) [117]. Furthermore, endogenously produced H₂O₂ mediates ischemia–reperfusion induced upregulation of ICAM-1 [118] and MCP-1 (monocyte chemoattractant protein-1) [119]. H₂O₂ is also required for TNFα induction of ICAM-1 and VCAM-1 (vascular cell adhesion molecule-1) [120], and upregulation of MCP-1 by TNFα [121] or hyperglycemia [122]. Additionally, activation of NFκB by H₂O₂ was found responsible for TNFα induction of ICAM-1 [123] and angiotensin II upregulation of VCAM-1 [124].

Endothelial expression of platelet activating factor (PAF) [125] and P-selectin [126] is inducible by H₂O₂. All these molecules including ICAM-1, VCAM-1, MCP-1, PAF and P-selectin have been shown to mediate neutrophil adhesion to endothelium (Fig. 5). PAF and P-selectin also mediates platelet activation and endothelium–platelet interaction. Upregulation of Mac 1 (CD11b and CD18) in neutrophils [127] by H₂O₂ stimulates more H₂O₂ release from activated neutrophils [128], forming a vicious circle. Taken together, these data strongly suggest that H₂O₂ mediates endothelial expression of inflammatory proteins and enhanced platelet–endothelium interaction, which in turn contributes to development of atherosclerotic vascular disease. The events described above are summarized in Fig. 5.

8 Hydrogen peroxide regulation of endothelium-mediated vascular remodeling

Emerging evidence seems to suggest that H₂O₂ plays an important role in mediating vascular remodeling. First of all, H₂O₂ increases endothelial MMP-2 expression [12]. TGF-β1 and macrophages are important regulators of tissue fibrosis and remodeling. TGF-β1 induction of macrophage colony-stimulating factor, and IL-1 induction of plasminogen activator inhibitor-1 (PAI-1) in endothelial cells were found dependent on H₂O₂ activation of NFκB [129,130]. PAI-1 plays an important role in vascular remodeling induced by thrombosis [131] or chronic eNOS inhibition [132]. Recent studies demonstrated that PAI-1 is also critically important for angiotensin II or salt-induced vascular remodeling in vivo [133]. Thus by mediating TGF-β1 and PAI-1 signaling and MMP release, H₂O₂ contributes to endothelium-regulated vascular remodeling, which has been shown to regulate stability of atherosclerotic plaques [134].

9 Contribution of hydrogen peroxide to vascular disease

To date, the specific, individual ROS that is most relevant to vascular signaling pathophysiologically is yet identified. Nevertheless, selectively overproducing or removing H₂O₂ significantly altered atherogenesis in animal models. Mice overexpressing p22^phox had markedly increased atheroma formation in a carotid ligation model [53]. This response was associated with enhanced H₂O₂ production in the vessel wall, and was abolished by scavenging H₂O₂ with ebselen, implicating a critical role of H₂O₂[53]. Parallel studies from another group confirmed the same notion [135]. Yang et al. cross bred transgenic mice overexpressing Cu/Zn-SOD or catalase, with mice deficient in apolipoprotein E (apoE^−/−), to examine a specific role of H₂O₂ versus O₂^.− in atherogenesis [135]. They found that whereas overexpressing Cu/Zn-SOD had no effect on atherosclerotic lesion formation, overexpression of catalase, or co-overexpression of catalase and Cu/Zn-SOD markedly retarded atherosclerosis in all aspects [135]. These observations were consistent with the findings by Tribble et al. that overexpression of Cu/Zn-SOD failed to prevent atherosclerosis in high fat diet-fed apoE^−/− mice [136]. Taken together, these data suggest that H₂O₂ is more atherogenic than O₂^.−. It is true that Cu/Zn-SOD is intracellular and overexpression of Cu/Zn-SOD can not protect NO^. during its trafficking to vascular smooth muscle. Thus extracellular SOD might be more important in determining NO^. bioavailability and be more anti-atherosclerosis. Indeed, evidence gained from ecSOD-null mice and adenovirus-mediated overexpression of ecSOD supports that ecSOD is the main determinant of NO^. bioavailability in the vessel wall, and is thus involved in blood pressure regulation [137,138]. However, impact of ecSOD overexpression on atherosclerosis is yet to be reported although mice deficient in ecSOD developed similar atherosclerotic lesions compared to wild-type mice [139]. Therefore, whereas O₂^.− is important in directly modulating NO^. bioavailability and serving as the precursor for H₂O₂[7], relatively lasting H₂O₂ seems more important in mediating atherogenic signaling.

In summary, O₂^.− and H₂O₂ are produced in vascular cells by multiple enzymatic systems including vascular NAD(P)H oxidases, mitochondrion, xanthine oxidase and uncoupled eNOS. While some O₂^.− spontaneously degrades by reacting with NO^., O₂^.− signal preserved by dismutation into H₂O₂ exerts prolonged signaling effects. This may explain why direct scavenging H₂O₂ but not O₂^.− is more effective in athero-protection. Although there is much to be learned, the impact of H₂O₂ on different aspects of endothelial cell function as discussed throughout the review may at least partially underlie H₂O₂-mediated development of atherosclerotic vascular disease.

Acknowledgements

The author's work is supported by an American Heart Association Scientist Development Grant (#0435189N), an American Diabetes Association Research Award, a Career Development Award from the Schweppe Foundation, and a Start-up Fund from the University of Chicago.

References

Author notes