Pleiotropic and Potentially Beneficial Effects of Reactive Oxygen Species on the Intracellular Signaling Pathways in Endothelial Cells

Endothelial cells (ECs) are exposed to molecular dioxygen and its derivative reactive oxygen species (ROS). ROS are now well established as important signaling messengers. Excessive production of ROS, however, results in oxidative stress, a significant contributor to the development of numerous diseases. Here, we analyze the experimental data and theoretical concepts concerning positive pro-survival effects of ROS on signaling pathways in endothelial cells (ECs). Our analysis of the available experimental data suggests possible positive roles of ROS in induction of pro-survival pathways, downstream of the Gi-protein-coupled receptors, which mimics insulin signaling and prevention or improvement of the endothelial dysfunction. It is, however, doubtful, whether ROS can contribute to the stabilization of the endothelial barrier.


Introduction
Cells of vasculature, red blood cells (RBCs), endothelial cells (ECs), and vascular smooth muscle cells (VSMCs), work in concert to match the oxygen supply with tissue oxygen demand [1]. Some oxygen molecules encounter free electrons (e − ) and free protons (H + ), resulting in the formation of reactive oxygen species (ROS). ROS include superoxide anion (O 2 •− ), hydrogen peroxide (H 2 O 2 ), and hydroxyl radical ( • OH). The reaction of superoxide anion with nitric oxide (NO) produces peroxynitrite (ONOO − ). There are multiple sources of ROS in ECs, including nicotinamide adenine dinucleotide phosphate (NADPH) oxidases, xanthine oxidases, nitric oxide synthase (NOS), cyclooxygenase (COX), cytochrome P450 monooxygenases, and mitochondria [2,3]. There are seven members of the NADPH oxidase (Nox) family-Nox1, Nox2 [also known as gp91phox (phox stands for phagocyte oxidase)], Nox3, Nox4, Nox5, Duox1 (dual oxidase), and Duox2 [4]. Nox1, Nox2, Nox4, and Nox5 isoforms are expressed in cells of the cardiovascular system [3,[5][6][7]. As discussed below, there is experimental evidence for activation of these pro-survival pathways by Gi-protein-coupled receptors (Gi-PCRs) via liberation and activation of the Gβγ dimers ( Figure 1). Activation of Gαi/o by ROS would also activate Gβγ [24,25]. This makes us suggest that ROS via activation of Gi proteins can promote a Gi-PCR-independent EC survival. Data on triggering of EC survival by some Gi-PCRs are presented in Table 1. However, activation of Gi-PCRs above a specific threshold may also induce apoptosis (Table 1). Table 1. Involvement of Gi proteins, activated by ligand binding to GPCRs, in regulation of EC survival, and apoptosis.

Gi/o Proteins and AC-cAMP-PKA Pathway
AC activation via production of cAMP and activation of PKA leads to phosphorylation and nuclear translocation of CREB [36]. CREB can mediate the pro-survival effect of the AC-cAMP-PKA pathway. For example, in mouse, the immortalized cerebral endothelial (b.End3) cells CREB is responsible for vascular endothelial growth factor A (VEGF-A)/VEGF receptor-2 (VEGFR-2)-mediated cell survival [37]. In human umbilical vein, endothelial cells (HUVECs) lipopolysaccharide (LPS)-induced apoptosis was inhibited by  Table 1. Involvement of Gi proteins, activated by ligand binding to GPCRs, in regulation of EC survival, and apoptosis.

G i/o Proteins and AC-cAMP-PKA Pathway
AC activation via production of cAMP and activation of PKA leads to phosphorylation and nuclear translocation of CREB [36]. CREB can mediate the pro-survival effect of the AC-cAMP-PKA pathway. For example, in mouse, the immortalized cerebral endothelial (b.End3) cells CREB is responsible for vascular endothelial growth factor A (VEGF-A)/VEGF receptor-2 (VEGFR-2)-mediated cell survival [37]. In human umbilical vein, endothelial cells (HUVECs) lipopolysaccharide (LPS)-induced apoptosis was inhibited by cilostazol, a selective phosphodiesterase 3 inhibitor, via increase in cAMP, activation of MEK1/2-ERK1/2 and p38, and activation of CREB [38].
There can be signaling from the G i -coupled receptors to CREB activation, although the Gα i subunits inhibit AC. For instance, the relaxin family peptide receptor 1 (RXFP1) is coupled with Gα i3 and can activate the Gβγ-PI3K-PKCζ (protein kinase Czeta)-AC pathway [39] (Figure 1). Survival of human pancreatic islet microvascular ECs was enhanced by gastrointestinal ghrelin gene products acting via their receptor GHS-R1a (growth hormone secretagogue receptor 1a), which induced activation of the AC-cAMP-PKA pathway [27] ( Table 1). GHS-R1a is mainly coupled to Gα q/11 but also to G i [33].

G i/o Proteins and Pro-Apoptotic Pathways in ECs
It should be noted here that activation of the G i/o -coupled receptors may also lead to EC apoptosis (Table 1). In HUVECs, stimulation of cannabinoid receptors (CB1 and CB2 are both coupled to G i/o proteins [30]) by anandamide, induced apoptosis via activation of the JNK and p38 MAPK [31]. In human coronary artery ECs (HCAECs), stimulation of the CB1 receptor led to apoptosis via increase in ROS generation and activation of JNK and p38 MAPKs [32].

ROS Can Mimic Insulin Signaling
Non-insulin-dependent diabetes mellitus (NIDDM) is regarded as a significant contributor to the development of endothelial dysfunction [50]. There are paradoxical interrelationships between ROS and insulin signaling. ROS are known to participate in insulin signal transduction, as well as to evoke insulin resistance [51][52][53][54].

Reversible Inhibition of Protein Tyrosine Phosphatase 1B (PTP1B) by ROS
Insulin induces activation of NADPH oxidases [53,54], and the generated ROS transiently inhibit protein tyrosine phosphatase 1B (PTP1B) [55]. Increased insulin sensitivity was observed in mice deficient of PTP1B [56,57]. In a mouse model of pancreatic islet transplantation into eye, deletion of PTP1B, elevated revascularization of the graft islet, increased graft survival, facilitated recovery of normoglycemia, and improved glucose tolerance [58]. These effects of loss of PTP1B were mediated via an increase in the expression of VEGF-A by β cells, following activation of the peroxisome proliferator-activated receptor γ coactivator 1α (PGC1α) and the estrogen-related receptor α [58].

Indirect Activation of the PI3K-Akt Pathway
There are several points where ROS can enhance the signal transduction through the PI3K-Akt module ( Figure 2). Ras GTPases can be directly activated by ROS via oxidation of Cys118 [59]. In BAECs, peroxynitrite activated p21Ras, which in turn activated the PI3K-PDK1-Akt pathway [60]. Furthermore, CaMKII can directly phosphorylate and activate Akt [61]. CaMKII itself can be activated through a reversible oxidation of methionines 281/282 [62].

Reversible Inhibition of Protein Tyrosine Phosphatase 1B (PTP1B) by ROS
Insulin induces activation of NADPH oxidases [53,54], and the generated ROS transiently inhibit protein tyrosine phosphatase 1B (PTP1B) [55]. Increased insulin sensitivity was observed in mice deficient of PTP1B [56,57]. In a mouse model of pancreatic islet transplantation into eye, deletion of PTP1B, elevated revascularization of the graft islet, increased graft survival, facilitated recovery of normoglycemia, and improved glucose tolerance [58]. These effects of loss of PTP1B were mediated via an increase in the expression of VEGF-A by β cells, following activation of the peroxisome proliferator-activated receptor γ coactivator 1α (PGC1α) and the estrogen-related receptor α [58].
In addition, active Akt can be deactivated through dephosphorylation of Ser473 and Thr308 by protein phosphatase 2A (PP2A) [61,69,70]. PP2A can be reversibly inhibited by ROS through disulfide bond formation in the catalytic subunit of PP2A [69], which would enhance the signal transduction through the PI3K-Akt pathway ( Figure 2).

Endothelial Barrier: Redox Dependence of Some Intracellular Signaling Proteins Involved in Regulation of Endothelial Permeability
Generally, oxidative stress leads to an increase in endothelial permeability [75]. Here, we discuss data suggesting that ROS can also contribute to the stabilization of the endothelial barrier. Among multiple pathways implicated in the regulation of the endothelial barrier [76], some can be regulated by ROS.

ROS and G i/o Proteins
Hydroxyl radicals are likely to activate Gα i/o in a G-protein-coupled receptor (GPCR)independent manner [25]. It appears that GPCR-dependent activation of Gα i/o proteins has dual effects on endothelial permeability. For example, the interleukin 8 chemokine receptor CXCR2 is coupled to the G i/o proteins [77] and can increase pulmonary microvascular permeability in a murine model of LPS-induced lung injury [78]. Similarly, human cerebral microvascular ECs impaired their barrier integrity, upon activation of CXCR2 [79].
Moreover, there seems to be a cAMP-independent pathway leading from the activation of the G i proteins to the activation of PKA and endothelial barrier enhancement [85,86] ( Figure 3). In HPAECs, activation of P2YRs by ATP and its analogue adenosine 5 -[γ-thio]triphosphate (ATPγS), coupled to the G q and G i proteins, and led to an enhancement of the endothelial barrier [85]. It appears that G i proteins can activate PKA in an ACindependent manner via the PKA-anchoring proteins (AKAPs), and PKA phosphorylates VASP (vasodilator-stimulated phosphoprotein) and thus enhances the barrier [85]. Similarly, in human lung microvascular ECs, adenosine and ATPγS stabilized the endothelial barrier via P2Y 4 R (coupled to both G q and G i ) and P2Y 12 R (coupled to G i ), and an unconventional cAMP-independent activation of PKA [86].
[γ-thio]-triphosphate (ATPγS), coupled to the Gq and Gi proteins, and led to an enhancement of the endothelial barrier [85]. It appears that Gi proteins can activate PKA in an ACindependent manner via the PKA-anchoring proteins (AKAPs), and PKA phosphorylates VASP (vasodilator-stimulated phosphoprotein) and thus enhances the barrier [85]. Similarly, in human lung microvascular ECs, adenosine and ATPγS stabilized the endothelial barrier via P2Y4R (coupled to both Gq and Gi) and P2Y12R (coupled to Gi), and an unconventional cAMP-independent activation of PKA [86].

ROS and Some Branches of Signaling Downstream of Growth Factor Receptors
In bovine pulmonary artery endothelial cells (BPAECs), activation of the MEK-ERK pathway led to an increase in the endothelial permeability [41,87]. ROS may contribute to the activation of the Ras-Raf-MEK-ERK pathway via a direct activation of Ras [59,74]. In rat aortic vascular smooth muscle cells, ERK1/2 was activated by CaMKII [88], which itself could be activated by ROS [62].
The small GTPase RhoA can be directly inhibited by oxidants [89,90]. RhoA via activation of its downstream effector Rho-associated kinase (ROCK), induces phosphorylation of the myosin light chain kinase (MLCK), which activates non-muscle myosin II [91], resulting in EC contraction and increased endothelial permeability [76,83]. Inhibition of myosin phosphatase by ROCK also contributes to non-muscle myosin II activation [92].

ROS and Some Branches of Signaling Downstream of Growth Factor Receptors
In bovine pulmonary artery endothelial cells (BPAECs), activation of the MEK-ERK pathway led to an increase in the endothelial permeability [41,87]. ROS may contribute to the activation of the Ras-Raf-MEK-ERK pathway via a direct activation of Ras [59,74]. In rat aortic vascular smooth muscle cells, ERK1/2 was activated by CaMKII [88], which itself could be activated by ROS [62].
The small GTPase RhoA can be directly inhibited by oxidants [89,90]. RhoA via activation of its downstream effector Rho-associated kinase (ROCK), induces phosphorylation of the myosin light chain kinase (MLCK), which activates non-muscle myosin II [91], resulting in EC contraction and increased endothelial permeability [76,83]. Inhibition of myosin phosphatase by ROCK also contributes to non-muscle myosin II activation [92].

ROS and Ca 2+ -Dependent Mechanisms
Stimulation of Ca 2+ -dependent pathways can increase endothelial paracellular permeability [76]. For example, activation of CaMKII in bovine pulmonary artery ECs disrupted the endothelial barrier via phosphorylation of caldesmon and activation of the ERK [87].
Increase in intracellular Ca 2+ may occur via opening of the Ca 2+ channels of the plasma membrane, release of Ca 2+ from intracellular stores through ryanodine receptors and inositol-trisphosphate (IP3) receptors, or from mitochondria -all these mechanisms being regulated by ROS [93]. ROS can also activate protein kinases such as PKA type I [94], protein kinase C [95], and CaMKII [62]. These kinases regulate both the ion channels [93] and the endothelial permeability [76,84].

ROS and Tyrosine Kinases and Phosphatases
The effect of Src on the endothelial barrier appears to be biphasic-initially Src can enhance the barrier function, but prolonged action of Src impairs the endothelial barrier [96]. H 2 O 2 can activate Src tyrosine kinase via reversible sulfenylation (Cys-SOH formation) of two cysteine residues-Cys-185 and Cys-277 [97].
In HPAECs, c-Abl enhances the endothelial barrier, apparently via regulation of the actin cytoskeleton [98]. Exposure of COS7 cells to high concentrations (1 mM) of H 2 O 2 , induced a 5-fold increase in c-Abl tyrosine kinase activity [99].
In several model systems, including a mouse model of acute LPS-induced lung injury, HUVECs, and HPAECs, Lyn tyrosine kinase was revealed to stabilize the endothelial barrier [100]. Lyn can be directly activated by H 2 O 2 [101].
Inhibition of PTP1B can increase paracellular endothelial permeability via an increase in phosphorylation of vascular endothelial (VE)-cadherin and disruption of cell-cell adhesions [102]. PTP1B can be inhibited by ROS [55].

Endothelial Dysfunction
Endothelial dysfunction is characterized by 3 main features-impaired flow-induced endothelium-dependent vasodilatation, increased pro-inflammatory activity of endothelium, and enhanced pro-thrombotic status of endothelium [103]. Oxidative stress appears to be a key factor in the pathogenesis of endothelial dysfunction [103]. ECs are exposed to fluid shear stress (SS), which, depending on its pattern, can be either laminar shear stress (LSS), well-established as the anti-inflammatory and athero-protective factor, or disturbed shear stress (DSS), which is known to lead to vascular inflammation and atherosclerosis [104][105][106][107]. H 2 O 2 produced by ECs is itself regarded as an endothelium-derived hyperpolarizing factor, since it directly activates the PKGIα in vascular smooth muscle cells [108,109]. It is of interest to see if and how ROS could ameliorate endothelial dysfunction. For this purpose, we first discuss the potential involvement of ROS in the production of NO by ECs. Next, we regard the role of the mechano-as well as the redox-sensitive MEK5-ERK5 pathway, in the regulation of transcription factors that are known to suppress inflammation.
Since endothelial dysfunction comprises impairment of several EC functions, agents that possess a wide spectrum of actions are of interest as therapeutics. Phytochemicals can serve as agents that target multiple signaling mechanisms in ECs [110,111]. In vivo studies have shown such effects of indole-3-carbinol (I3C), a phytochemical found in cruciferous vegetables, and its derivative 3,3,'-diindolylmethane (DIM) on EC functions such as suppression of angiogenesis, prevention of thrombus formation, and alleviation of inflammation. Suppression of the inflammatory response by I3C and DIM are mediated via suppression of production and release of the inflammatory cytokines, modulation of ROS production, inhibition of leucocyte-EC interaction, and decrease in paracellular microvascular permeability. It appears that I3C and DIM can exert dual effects on ROS production; both stimulation and inhibition of ROS generation were reported [110]. Maslinic acid, a triterpene derivative from Olea europaea, was shown to suppress activation of NF-κB in human dermal microvascular ECs and human placenta-derived pericytes. Suppression of NF-κB by maslinic acid reduced the expression of adhesion molecules E-selectin, intercellular adhesion molecule 1, and vascular adhesion molecule 1 on EC and pericytes, and attenuated the development of inflammation [111].

ROS and Flow-Induced Release of Vasodilators by ECs
Here, we regard three scenarios in which, hypothetically, ROS can promote activation of eNOS and release of NO-via redox sensitive CaMKII-, PI3K-Akt-, and G i/omediated pathways.

ROS and Flow-Induced Release of Vasodilators by ECs
Here, we regard three scenarios in which, hypothetically, ROS can promote activation of eNOS and release of NO-via redox sensitive CaMKII-, PI3K-Akt-, and Gi/o-mediated pathways.
There can be several pathways from G i proteins to the activation of eNOS. For example, stimulation of G i -coupled S1P 1 receptor appears to activate the following signaling cascade: Gβγ-Src-Tiam1-Rac1-PI3K-Akt-eNOS [49] (Figure 5). G i proteins can be activated by both SS [122,123] and ROS [24,25], suggesting that ROS can enhance or even mimic SS-induced G i activation in ECs ( Figure 5).
Furthermore, some steroid hormone receptors are well-established to be coupled to G i/o proteins and activate eNOS [124,125]. Endogenous estrogens act via three different receptors-classical estrogen receptors α and β (ERα and ERβ) and G-protein-coupled estrogen receptor (GPER), also known as GPR30 [126]. In immortalized ovine pulmonary artery endothelial cells (iPAECs), the plasma membrane ERα was shown to localize to caveolae and to stimulate eNOS [124,127] via activation of Gα i [124]. In BAECs, membrane receptor of the adrenal dehydroepiandrosterone (DHEA) was shown to activate eNOS via coupling to Gα i2 and Gα i3 , but not Gα i1 or Gα o [125]. In this scenario, ROS can contribute to eNOS stimulation via direct activation of the G i/o proteins [24,25].
There can be several pathways from Gi proteins to the activation of eNOS. For example, stimulation of Gi-coupled S1P1 receptor appears to activate the following signaling cascade: Gβγ-Src-Tiam1-Rac1-PI3K-Akt-eNOS [49] (Figure 5). Gi proteins can be activated by both SS [122,123] and ROS [24,25], suggesting that ROS can enhance or even mimic SS-induced Gi activation in ECs ( Figure 5). Furthermore, some steroid hormone receptors are well-established to be coupled to Gi/o proteins and activate eNOS [124,125]. Endogenous estrogens act via three different receptors-classical estrogen receptors α and β (ERα and ERβ) and G-protein-coupled estrogen receptor (GPER), also known as GPR30 [126]. In immortalized ovine pulmonary artery endothelial cells (iPAECs), the plasma membrane ERα was shown to localize to caveolae and to stimulate eNOS [124,127] via activation of Gαi [124]. In BAECs, membrane receptor of the adrenal dehydroepiandrosterone (DHEA) was shown to activate eNOS via coupling to Gαi2 and Gαi3, but not Gαi1 or Gαo [125]. In this scenario, ROS can contribute to eNOS stimulation via direct activation of the Gi/o proteins [24,25].

ROS, MEK5-ERK5 Module and Transcription Factors (TFs) That Can Alleviate Endothelial Dysfunction
Inflammation is regarded as a key factor in the pathogenesis of atherosclerosis [128][129][130]. ROS are well-known to promote inflammatory response of ECs [131]. LSS-dependent EC survival and quiescence are mediated by TFs, such as KLF2 (Krüppel-like factor 2), KLF4, and Nrf2 [106]. The DSS-induced pro-inflammatory response of ECs on the other hand is evoked by TFs, such as NF-κB, AP-1, YAP/TAZ, and HIF-1α [106]. Let us regard several hypothetical scenarios where ROS may contribute to mechanisms that prevent endothelial dysfunction via regulation of athero-protective TFs ( Figure 6).

ROS, MEK5-ERK5 Module and Transcription Factors (TFs) That Can Alleviate Endothelial Dysfunction
Inflammation is regarded as a key factor in the pathogenesis of atherosclerosis [128][129][130]. ROS are well-known to promote inflammatory response of ECs [131]. LSS-dependent EC survival and quiescence are mediated by TFs, such as KLF2 (Krüppel-like factor 2), KLF4, and Nrf2 [106]. The DSS-induced pro-inflammatory response of ECs on the other hand is evoked by TFs, such as NF-κB, AP-1, YAP/TAZ, and HIF-1α [106]. Let us regard several hypothetical scenarios where ROS may contribute to mechanisms that prevent endothelial dysfunction via regulation of athero-protective TFs ( Figure 6).

Redox-Sensitive Elements in SS-Induced Signaling Pathway Leading to ERK5 Activation
To illustrate how ROS can influence the signal transduction on athero-protective TFs, we have chosen the ERK5 [also known as big mitogen-activated protein kinase 1 (BMK1)] and its upstream kinase MEK5 (MAPK/ERK kinase 5) as a signaling module that integrates both SS-induced athero-protective signaling [132,133] and redox sensitivity [134,135]. In BAECs, SS (12 dynes/cm 2 ) activated ERK5 in a Ca 2+ -dependent and -Src-tyrosine-kinaseindependent manner [136]. Mechano-sensitivity of the MEK5-ERK5 module was also demonstrated in HUVECs where LSS (12 dynes/cm 2 ) activated ERK5 [133,137]. Since Ca 2+ -dependent mechanisms are involved in the activation of ERK5 in BAECs [136], ROS can interfere in these signaling proteins via effects on the Ca 2+ -permeable channels [93] and via direct activation of CaMKII [62].

MEF2 Family of Transcription Factors
The MEF2 family of transcription factors includes four members-MEF2A, MEF2B, MEF2C, and MEF2D [141]. Among these factors MEF2A, MEF2C, and MEF2D, but not MEF2B, can be phosphorylated and activated by BMK1 [141]. In human retinal ECs and HUVECs, MEF2C was shown to inhibit tumor necrosis factor alpha (TNF-α)-induced activation of NF-κB, to suppress the expression of pro-inflammatory genes and to decrease leukocytes adhesion to ECs [142]. In PC12 cells, H 2 O 2 activated the c-Src-MEK5-ERK5-MEF2C pathway [134], suggesting that in ECs, H 2 O 2 is also likely to activate MEF2C and contribute to the inhibition of inflammation.
At physiological ROS levels, Nrf2 is kept inactive and directed to proteasomal degradation via association with the redox sensor KEAP1 (Kelch-like Ech-associated protein 1) [147]. ROS induce dissociation of the Nrf2-KEAP1 complex and Nrf2 is then translocated into the nucleus [147]. Nrf2 is a key TF mediating cyto-protection against oxidative stress in HUVECs [133]. In HUVECs, laminar SS (12 dynes/cm2) was shown to activate Nrf2 via activation of MEK5α and its effector ERK5 [133].
Normal level of ROS is well-known to provide antioxidant defense in ECs via stimulation of Nrf2, which activates antioxidant response element (ARE)-dependent expression of antioxidant genes. Nrf2-dependent vascular protection also includes the suppression of the NF-κB-dependent pro-inflammatory pathway and improvement of the mitochondrial function [148]. Recently, in an in vitro model of atherosclerosis-exposure of human coronary artery endothelial cells (HCAECs) to oxidized low-density lipoprotein (ox-LDL)activation of Nrf2 by prenyldiphosphate synthase subunit 2 (PDSS2) was reported. Activation of Nrf2 decreased ferroptosis and promoted proliferation of HCAECs [149]. In HUVECs, activation of Nrf2 by 1,25 dihydroxyvitamin D3 was shown to protect against high glucose-induced injury [150].

Peroxisome Proliferator-Activated Receptors (PPARs)
The family of nuclear receptors PPARs are presented by three subtypes-PPARα, PPARβ/δ, and PPARγ [151]. PPARγ plays important roles in the regulation of lipid metabolism, insulin resistance, vascular inflammation, and arterial hypertension [151]. Expression of a constitutively active mutant of PPARγ1 in HUVECs, suppressed the activation of pro-inflammatory TFs AP-1 and NF-κB, resulting in a reduced expression of markers of inflammation, such as ICAM-1, VCAM-1, and E-selectin [152]. In HUVECs, ERK5 mediated flow-induced activation of PPARγ1 [137], suggesting that H 2 O 2 -via activation of the c-Src-MEK5-Erk5 pathway [134,135]-can also activate PPARγ1. Furthermore, ROS may induce activation of PPARγ via generation of oxidized fatty acids, which were shown to bind to and activate PPARγ [153].

Role of NADPH Oxidase-Derived ROS in Angiogenesis
There is increasing evidence that ROS derived from NADPH oxidases significantly contribute to signaling mechanisms regulating angiogenesis, a process of formation of new blood vessel from the pre-existing vessel. ECs' proliferation, migration, differentiation, and capillary tube formation constitute the main features of angiogenesis. Angiogenesis is important for embryonic development, wound healing, and post-ischemic neovascularization. Endothelial NADPH oxidases, particularly NOX2 and NOX4, produce ROS that play important role in the regulation of angiogenesis [7,121,[154][155][156][157][158][159][160][161][162]. Endothelial ROS generation itself is subject to tight spatial and temporal regulation [156,163,164]. Endothelial NADPH oxidases can be activated by growth factors, cytokines, ligands of GPCRs, mechanical forces, and metabolic factors. Downstream, NADPH-derived ROS participate in several cellular processes, such as regulation of self-renewal, survival, proliferation, and differentiation of mesenchymal stem cells. Moreover, commitment of stem cells to adipogenic, osteogenic, or myogenic lineage, and endothelial-mesenchymal stem cell transition are also ROS-dependent [7]. Upstream inducers of pro-angiogenic NOX4 include hypoxia, ischemia, VEGF, TNF-related apoptosis-inducing ligand (TRAIL), and transforming growth factor-β1 (TGF-β1) [160].
ROS derived from NOX1 and NOX4 are important regulators of proliferation, hypertrophy and apoptosis in human pulmonary artery endothelial and smooth muscle cells, which can lead to airway and vascular remodeling. Lung airway and vascular remodeling lead to disorders such as pulmonary artery hypertension, chronic obstructive pulmonary disease, asthma, and neonatal bronchopulmonary dysplasia. Factors inducing NADPH oxidases and increased ROS generation in lung include hyperoxia, hypoxia, LPS, allergens, angiopoietin-2, EGF, TGF-β, bone morphogenetic proteins, interleukins, and S1P. Downstream, ROS activate transcription factors such as NF-κB and AP-1, which results in the development of inflammation and vascular cell proliferation [158]. In case of peripheral artery disease, physiological levels of NADPH oxidase-derived ROS are proangiogenic and can stimulate EC proliferation, sprouting, migration, and tubule formation. Additionally, ROS contribute to the stability of newly-formed vessels via regulation of pericytes [159]. Pathogenesis of pulmonary arterial hypertension involves endothelial dysfunction accompanied by smooth muscle cell proliferation, inflammation, and fibrosis. The main underlying cause may be excessive production of ROS by NADPH oxidases (NOX1, NOX2, NOX4) and mitochondria [121].
Surprisingly, both ROS generation by NOX4 and ROS scavenging by thioredoxin 2 (TRX2) can promote angiogenesis. TRX2, a key mitochondrial ROS scavenger, promote EC survival and proliferation via elimination of ROS and thus enhancement of NO availability, and also via inhibition of apoptosis signaling kinase-1 (ASK1). The mechanism of cross-talk between NOX4 and TRX2 may depend on regulation by common angiogenic factors, such as hypoxia, ischemia, and VEGF [160], or spatial presence of NOX4-derived ROS and absence of mitochondrial ROS.
Effects of NADPH-derived ROS on angiogenesis depend on isoform of NADPH oxidase. NOX4-generated ROS can promote vascular restoration after hypoxic and ischemic injuries and can inhibit vascular inflammation. However, in case of oxygen-induced retinopathy, NOX4 promotes pathological angiogenesis. NOX2 appears to be involved not only in normal physiological angiogenesis but also in pathological angiogenesis, in cases of choroidal neovascularization, retinopathy, and tumor growth. Moreover, NOX2generated ROS were shown to inhibit physiological angiogenesis via activation of apoptotic signaling in the retina and the brain, but promotes vascular restoration in ischemic hindlimb. Additionally, activation of NOX2 by proinflammatory TNFα can contribute to vascular inflammation [161]. NOX2-generated ROS were shown to significantly contribute to diabetes-induced premature senescence of retinal ECs [165].
NADPH-produced ROS exert their actions on angiogenesis via regulation of intracellular signaling and gene expression. ROS regulate signaling proteins via reversible oxidation of cysteine residues to sulfenic acid (-SOH), sulfinic acid (-SO 2 H), and sulfonic acid (-SO 3 H). Protein tyrosine phosphatases contain a conserved cysteine residue in their catalytic domain and oxidation of this cysteine to sulfenic acid and sulfonic acid reversibly and irreversibly, respectively, inhibit tyrosine phosphatases [162].
Finally, ROS trigger mobilization of bone marrow progenitor cells in response to ischemic injury. In a mouse model of hindlimb ischemia, Nox2-derived ROS were shown to regulate the mobilization of progenitor cells from the bone marrow. Effects of hindlimb ischemia-induced Nox2-derived ROS included increase in expression of HIF-1α and VEGF throughout the bone marrow, elevated survival and proliferation of bone marrow Linprogenitor cells, Akt phosphorylation, activation of matrix metalloproteinase-9 (MMP-9), and membrane type 1-MMP (MT1-MMP) [172].

Conclusions
The aim of this review was to analyze the literature evidence to test our hypothesis that ROS can evoke beneficial effects on ECs. Analysis of the available experimental data has shown that ROS can mimic, to some extent, signaling through the G i -protein coupled receptors, as well as insulin receptor and other growth factor receptors. Although, the prevalent view is that ROS are responsible for an increase in paracellular endothelial permeability, accumulating data suggest that physiological ROS may be actively contributing to endothelial barrier stabilization or maintenance. Development of endothelial dysfunction involves numerous signaling proteins, and a few of them can be either activated or inhibited by ROS. In particular, vascular inflammation is one of primary conditions leading to vascular diseases. Of great interest would be to explore the causal relationships between ROS and the activities of transcription factors that regulate inflammatory response. Further investigations are needed to better delineate the boundary between normal physiological and pathological ROS levels. As ROS are tiny and short-lived, their live-cell, real-time observation is difficult. Nevertheless, the progress in the field depends on studies of complexes between sources of ROS and targets of ROS. Funding: This research was funded by grant from "Fondo di Ateneo per la ricerca 2020", University of Sassari, Italy.
Data Availability Statement: Data is contained within the article.

Conflicts of Interest:
The authors declare no conflict of interest.