The Role of Oxidants in Percutaneous Coronary Intervention-Induced Endothelial Dysfunction: Can We Harness Redox Signaling to Improve Clinical Outcomes?

^•NO and RNS: endothelial sources

First discovered as the endothelial-derived relaxing factor, ^•NO is a key mediator of vascular tone and endothelial function (Palmer et al, 1987). In mammals, ^•NO is synthesized by three key homodimeric nitric oxide synthase (NOS) enzymes: neuronal NOS (nNOS), inducible NOS (iNOS), and endothelial NOS (eNOS). These enzymes convert the amino acid l-arginine to ^•NO and citrulline via a two-step oxidation process that is dependent on four redox-dependent cofactors and calcium/calmodulin-mediated transfer of electrons derived from reduced nicotinamide adenine dinucleotide phosphate (NADPH) to NOS heme (Fig. 2A) (Stuehr and Haque, 2019).

^•NO has both a short half-life and diffusion distance, owing to its highly reactive nature with heme compounds or other reactive species. One of the fastest ^•NO reactions is with O₂^•−, which produces both ONOO⁻ and its acid ONOOH (Fig. 2A). Once formed, ONOO⁻ and ONOOH can oxidize thiolates (S⁻) to form S-nitrosothiols (RSNO), with a small proportion of ONOO⁻ and ONOOH degraded to form ^•OH, nitrogen dioxide (^•NO₂), and nitrate (NO⁻₃) (Ferrer-Sueta et al, 2018; Radi, 2018). RSNO also form through ^•NO reactions with ferrous heme, forming NO⁺, or O₂/ROS, via N₂O₄ or N₂O₃ intermediates (Fig. 2A). As RSNO can transfer their nitroso group via transnitrosation, they are considered by some as a form of RNS.

Cellular RNSO may be present within protein structures as post-translationally modified thiols (also known as S-nitrosation) or small molecules, for example, S-nitrosoglutathione (GSNO). Labile S-nitroso bonds rapidly react with free thiols to generate disulfide bonds impacting protein conformation and/or activity (Wolhuter et al, 2018).

^•NO and RNS: vascular tone

Under physiological conditions, ECs produce and release ^•NO in response to muscarinic receptor stimulation (Palmer et al, 1987), isometric force generation (Wallis and Martin, 2000), shear stress-mediated activation of mechanosensitive channels (Wang et al, 2016), and adrenergic-induced contraction (Fig. 3) (Smith et al, 2020).

^•NO initiates arterial relaxation/dilation by decreasing VSMC concentration of, and/or sensitivity to, intracellular calcium. This occurs via activation of ^•NO smooth muscle receptor soluble guanylate cyclase (sGC). ^•NO binds to the heme NO and O₂ binding domain, and potentially additional unknown sites, resulting in conformational changes of the coiled-coil domain and protein backbone increasing catalytic activity (Horst and Marletta, 2018). Catalytically active sGC increases the conversion of 5′-guanosine triphosphate to 3′,5′-cyclic guanosine monophosphate (cGMP) which canonically activates protein kinase G subunit 1α (PKG1α) (Kim and Sharma, 2021). The ^•NO/sGC/cGMP/PKG pathway has been shown to regulate arterial tone via inhibition of smooth muscle depolarization by suppressing inositol trisphosphate receptors (IP₃Rs), regulating myosin light chain phosphorylation (Kitazawa et al, 2009), and phosphorylation and activation of large conductance calcium-activated potassium channels (BK_ca) (Alioua et al, 1998).

In addition to the canonical activation of PKG1α, RNS oxidatively activate PKG1α via RSNO-mediated disulfide formation at Cys42 (Burgoyne et al, 2007). Furthermore, ONOO⁻ induces S-glutathiolation of Cys674 of sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) and stimulates vasodilation via a thiyl radical or sulfenic acid intermediate (Adachi et al, 2004). Finally, RNS are proposed to directly activate BK_ca channels through oxidation of redox-active thiols as a mechanism of arterial relaxation (Bolotina et al, 1994).

^•NO and RNS: coagulation and inflammation

ECs act as a physical barrier to separate procoagulatory factors in the blood from the reactive tissues in the blood vessel wall, secreting pro- and anti-clotting factors, and regulating platelet activation. ^•NO inhibits platelet aggregation through the ^•NO/sGC/PKG pathway with subsequent regulation of the IP₃ receptor through inositol 1,4,5-triphosphate IP₃ receptor-associated cGMP kinase substrate (Fig. 3) (Antl et al, 2007). However, cGMP-independent inhibition of platelet aggregation has also been reported. ^•NO-dependent, cGMP-independent mechanisms are associated with decreased exocytosis of platelet granules dependent on the presence of key cysteine residues and can be inhibited using N-acetyl-l-cysteine, which is indicative of protein S-nitrosation (Matsushita et al, 2003). It was concluded that S-nitrosation of platelet granule proteins inhibits exocytosis. However, N-acetyl-l-cysteine is effective at breaking disulfide bonds (Aldini et al, 2018), and therefore, RNS-mediated disulfide formation cannot be mechanistically ruled out.

A recent study has shown that ^•NO along with RNS, nitrate, and nitrite all have differing platelet inhibition efficacies in a manner dependent on the oxygenation state of the blood (Wajih et al, 2017).

Leukocyte rolling, capture, arrest, and extravasation are regulated by ^•NO as part of the classic inflammatory response. Acting through cGMP-independent mechanisms, ^•NO inhibits nuclear factor kappa B (NF-κB), monocyte chemoattractant protein 1, integrin expression, P-selectin, E-selectin, vascular cell adhesion molecule-1 (VCAM-1), and to a lesser extent intercellular adhesion molecule-1 (ICAM-1) (Lefer, 1997). Subsequent studies found that the effects of ^•NO on NF-κB could be reversed in the presence of dithiothreitol, suggesting the involvement of RNS and thiol oxidation (Marshall and Stamler, 2001). Later works investigating the short-term effects of ^•NO on leukocyte adhesion (5 min) after tumor necrosis factor α (TNF-α) exposure revealed that increased adhesion was associated with S-nitrosation of protein kinase C (Aguilar et al, 2021). These data suggest that both ^•NO and RNS mediate the vascular inflammatory response.

^•NO and RNS: vascular structure

Endothelial-derived ^•NO plays a vital role in regulating arterial structure and integrity. In otherwise healthy large- and small-animal studies, the addition of NO donors reduces VSMC proliferation after vascular injury (Pearce et al, 2008). Similarly, endogenous stimulation of ^•NO formation protects against neointimal formation after carotid ligation injury. This process was sensitive to NOS inhibition and was associated with increased EC survival, decreased inflammatory markers (CD45 antigen, ICAM-1 and VCAM-1), and suppression of VSMC proliferation (Mukai et al, 2006). At the smooth muscle cell level, ^•NO causes G₀/G₁ cell arrest inhibiting smooth muscle proliferation through regulation of the ubiquitination system, with recent reports suggesting that this is via RNS-dependent oxidation of redox-active cysteines (Valek et al, 2019).

ROS: EC sources

In ECs, many of the ROS-dependent processes center around O₂^•− and H₂O₂. A range of cellular proteins have been shown to produce ROS, with the most notable being the NADPH oxidase family of enzyme complexes (NOX1–NOX5 and DUOX1–DUOX2), which use NADPH to reduce oxygen to O₂^•− or H₂O₂ (Fig. 2B). These transmembrane proteins consist of NADPH, FAD, and heme binding domains along with additional activating proteins/factors, with p22^phox commonly required for NOX1–NOX4 activation and calcium required for NOX5 and DUOX1–DUOX2 (Bedard and Krause, 2007). In the vasculature, NOX1–NOX5 are expressed in both ECs and VSMCs, with NOX2 predominantly expressed in ECs, NOX1 predominantly expressed in VSMCs, and NOX4 expressed at high levels in both the cell types.

A second major source of O₂^•− and H₂O₂ is mitochondrial respiration. O₂^•− is produced as a by-product of NADH-dependent respiration at a range of sites along the electron transport chain. Under physiological conditions, the majority of O₂^•− is reduced by superoxide dismutase (SOD) within the mitochondrial matrix, releasing H₂O₂ (Fig. 2B). While O₂^•− is predominantly produced in the mitochondrial matrix at two sites, located in complex III and mitochondrial glycerol 3-phosphate dehydrogenase, half of the O₂^•− is produced directly into the cellular cytosol, thus contributing to cellular O₂^•− concentrations (Brand, 2020). Mitochondrial O₂^•− is implicated in a range of diseases and is considered a key contributor to ED (discussed below), however, a role for physiological O₂^•− signaling is now emerging (Sies et al, 2022).

Although the main role of NOS enzymes is to produce ^•NO, uncoupled NOS also serve as a significant source of O₂^•− (Fig. 2B). Uncoupling results in electrons being passed from NADPH to molecular oxygen instead of l-arginine, with O₂^•− formed instead of ^•NO (Kietadisorn et al, 2012). NOS uncoupling occurs via (i) oxidation of zinc/thiolate cluster leading to the release of zinc, (ii) decrease in the production, recycling, or oxidation of the NOS cofactor tetrahydrobiopterin (BH4), (iii) increased production of the endogenous NOS inhibitor asymmetric dimethylarginine, and (iv) G-protein-coupled receptor mediated phosphorylation associated with enzyme dysfunction (Wu et al, 2021).

Additional sources of vascular O₂^•− include xanthine oxidase (XO), an enzyme involved with purine metabolism. Circulating XO binds to the endothelium and produces O₂^•− and H₂O₂ via monovalent and divalent electron reduction of molecular oxygen, respectively (Bortolotti et al, 2021). In addition, xenobiotic metabolizing enzyme cytochrome P450 produces O₂^•− or H₂O₂ through key reaction stages in its catalytic cycle (Albertolle and Peter Guengerich, 2018).

ROS: vascular tone

Exogenous H₂O₂ relaxes conduit and resistance arteries with EC₅₀ values commonly in the midmicromolar ranges. The mechanisms involved in relaxation to exogenous H₂O₂ have previously been reviewed (Stanley et al, 2017). In brief, in ECs, exogenous H₂O₂ stimulates ^•NO production via phosphorylation and increased expression/stability of eNOS, increasing EC calcium leading to ^•NO formation and activation of EC ion channels causing hyperpolarization-mediated relaxation. In VSMCs, exogenous H₂O₂ causes arterial relaxation through catalase-mediated activation of sGC, oxidative activation of PKG1α, and indirect activation of a range of ion channels (Fig. 3).

Endogenous formation of H₂O₂ in response to muscarinic (Matoba et al, 2000) or bradykinin receptor stimulation (Matoba et al, 2002) and increased shear stress (Liu et al, 2011) have been shown to mediate arterial relaxation. In support of endogenous H₂O₂ production directly regulating vascular tone, relaxation can be inhibited by catalase or mutation of Cys42 in PKG1α (Matoba et al, 2000; Prysyazhna et al, 2012). However, with contradictory findings from studies utilizing genetic manipulation of the sources and metabolism of ROS (i.e., altering NOX expression), the field remains divided as to the extent of endogenous H₂O₂-mediated relaxation under physiological conditions.

ROS: vascular structure

ROS have a role in the embryonic development and maintenance of vascular structure (Sauer and Wartenberg, 2005). Once mature, arterial cells remain under physiological control by ROS. EC proliferation, morphological structure, and survival are regulated by O₂^•−/H₂O₂ from NOX2 and NOX4, which mediate extracellular-regulated kinase 1–2 phosphorylation, increased caspase activity, Akt phosphorylation, and phosphorylation of mitogen-activated protein kinase (MAPK) (Peshavariya et al, 2009). NOX-derived ROS also regulate EC migration in response to vascular endothelial growth factor via S-glutathiolation of SERCA2 (Evangelista et al, 2012).

Reducing systems

If left unregulated, endogenously produced ROS/RNS can cause irreversible damage to proteins, lipids, and DNA. Therefore, ECs have multiple reducing systems in place to regulate intracellular oxidant concentrations (Tretter et al, 2021). Direct-acting enzymes, including catalase and glutathione peroxidase 1 (GPx1), work in conjunction with low-molecular-weight antioxidants (lipoic acid, ascorbic acid, and tocopherols) to limit the accumulation of oxidants and maintain reducing conditions within the cell. Elevations in reactive species stimulate the nuclear factor E2-related factor 2 (Nrf2) antioxidant response that transcriptionally regulates the expression of antioxidant systems (Tonelli et al, 2018).

This includes two major thiol-based reducing systems that limit the accumulation of oxidized thiol residues and actively restore oxidized sensor thiols to their reduced state: (i) the glutathione system comprising reduced glutathione (GSH), oxidized glutathione (GSSG), and glutathione reductase (Forman et al, 2009), and (ii) the thioredoxin (Trx) system consisting of peroxiredoxin, Trx, and Trx reductase (TrxR) (Yamawaki et al, 2003). It should be noted that both these families of proteins are more than just an antioxidant defense system. They are recognized as signaling entities in their own right with the capacity to directly regulate cellular signaling. Furthermore, disruption of these reducing systems has been implicated in driving ED (Espinosa-Diez et al, 2018; Kirsch et al, 2016).

Vascular tone

A key feature of ED is the loss of NO bioavailability resulting in loss of vascular tone. Interestingly, studies show that inhibiting or depleting NOS alone is sufficient to drive ROS production. Decreasing ^•NO production, by pharmacological inhibition of NOS using L-NAME or siRNA knockdown of eNOS, elevates intracellular H₂O₂ levels in bovine aortic ECs, in a catalase-sensitive manner (Jin et al, 2009). In ex vivo human coronary arteries, H₂O₂-mediated dilation in response to flow was demonstrated to be a hallmark of disease (Freed et al, 2014). In patients without CAD, flow-induced arterial dilation is predominantly dependent on the production of ^•NO. However, in CAD patients, the mechanisms of dilation switch to mitochondrial-derived H₂O₂, which induces arterial dilation through oxidant activation of PKG1α (Freed et al, 2014; Prysyazhna et al, 2012).

These studies indicate that ED causes a switch from ^•NO-mediated to H₂O₂-mediated regulation of vascular tone (Fig. 3). Intriguingly, redox-dead PKG1α mutant mice are hypertensive, suggesting a role for reactive species in the physiological control of blood pressure (Khavandi et al, 2016). Taken together, this suggests that there is a fine balance between oxidant production required to maintain physiology and oxidant production in ED. Recent work has shed light on this area demonstrating that endothelium-dependent dilator responses to sphingosine-1-phosphate, in patients without CAD, are dependent on H₂O₂-mediated formation of ^•NO. However, in patients with CAD, the ^•NO component of dilation is lost with only the H₂O₂ component remaining. From these data alone, any detrimental issue with H₂O₂ being the sole mediator of arterial relaxation is not obvious.

However, prolonged exposure of resistance arteries to vasoactive concentrations of H₂O₂ causes endothelial and smooth muscle cell apoptosis (Shaw et al, 2021). These data highlight that, while H₂O₂ is a key mediator of vascular tone, reliance on H₂O₂ during ED is detrimental to vascular health and prolonged H₂O₂ production may weaken or destroy the endothelium and surrounding arterial tissue.

The relationship between RNS and vascular tone is complex. Work in conduit arteries, including mouse aorta and rabbit carotid arteries, suggests that arterial relaxation in response to EC stimulation is partly through the release of ^•NO, and is mediated by OONO⁻ formation. OONO⁻ mediates a reversible S-glutathiolation of SERCA leading to increased calcium uptake and smooth muscle relaxation (Fig. 3) (Adachi et al, 2004). However, in disease models such as atherosclerosis, ^•NO-mediated relaxation is blunted by ROS/RNS-mediated irreversible oxidation of SERCA, rendering it unable to sequester calcium. RNS are further implicated in ED by reducing NO bioavailability through oxidation of BH4 to dihydrobiopterin (BH2) leading to eNOS uncoupling and altering the redox state of smooth muscle sGC, thus decreasing its sensitivity to ^•NO (Kuzkaya et al, 2003; Tawa et al, 2014). In addition, in an obese mouse model, ED increased iNOS activity and EC NOX1 activity.

The combination of elevated ^•NO and ROS increased OONO⁻ leading to the oxidation of the transient receptor potential vanilloid receptor 4 regulatory protein, thus inhibiting arterial relaxation and raising blood pressure (Ottolini et al, 2020). Overall, exacerbated RNS production drives ED by inhibiting the key vasodilatory pathways and creating oxidative damage to proteins.

Coagulation and inflammation

With ED, the ability of the circulatory system to maintain blood in its fluid state is compromised. Due to decreased ^•NO bioavailability, clotting risk is elevated due to loss of its antithrombotic effects (Fig. 3). Furthermore, studies have shown that ROS and RNS generation in ED increases inflammation and clotting risk. In oscillatory shear stress, as associated with atherosclerosis and endothelial disruption, deletion of NOX1 and NOX2 is protective by reducing monocyte and leukocyte adhesion, respectively (Chen et al, 2004; Sorescu et al, 2004). Important factors in this process are the expression of cellular adhesion molecules ICAM-1 and VCAM-1, and the selectins P- and E-selectin.

Additional studies have identified that deletion of the NOX regulatory subunit (p47^phox) reduces adhesion and selectin molecule expression (Vendrov et al, 2007). Ramen spectroscopy of human aortic ECs after 3 h of menadione exposure to generate O₂^•− detected increased expression of the von Willebrand factor, a key mediator of platelet adhesion and blood clotting. ED-induced RNS are similarly deleterious to coagulation. Elevations in ONOO⁻ during CAD significantly accelerate clot formation and factor XIII crosslinking via nitration of fibrinogen (Vadseth et al, 2004).

Vascular structure

Potentiated oxidant production within the endothelium impacts vascular remodeling through the regulation of VSMC proliferation, extracellular matrix composition, and cell-to-cell connections. CAD induces a significant upregulation of O₂^•− from NOX1 and NOX4 (Szocs et al, 2002). However, the detrimental role of NOX4 has been contested in a recent publication evaluating neointimal cells using scRNAseq in Nox4^−/− mice. It was concluded that vascular remodeling occurs irrespective of NOX4 expression, with similar changes detected between WT and Nox4^−/− mice (Buchmann et al, 2021). The role of NOX1 appears less contentious. In response to injury, Nox1^y/− mice had decreased VSMC proliferation, migration, and apoptosis (Lee et al, 2009). Vascular injury increases NOX2 expression leading to increased RNS formation, as detected by elevated protein nitrotyrosine, and ablation of NOX2 decreased VSMC proliferation and reduced leukocyte adhesion (Chen et al, 2004).

In an atherosclerotic model, deletion of the regulatory subunit for NOX1 and NOX2 (p47phox^−/−) resulted in a 50% reduction in lesion size (Barry-Lane et al, 2001). This was associated with a decrease in O₂^•− and an increase in basal ^•NO (Judkins et al, 2010). These data highlight the deleterious effects of NOX-derived ROS and their ability to potentiate vascular remodeling.

Mechanically induced oxidants

The ability of mechanical stresses from PCI to induce damage via induction of ROS is well established. Early canine studies demonstrated that balloon angioplasty results in an acute elevation of oxidant-dependent vasoconstriction (Laurindo et al, 1991). Pretreatment with SOD, but not catalase or ^•OH scavengers, prevented pathologically enhanced vasoconstriction and the formation of large thrombi. These data indicate that mechanical damage to the endothelium during coronary angioplasty induces pathological levels of O₂^•−. PCI with stent implantation further mechanically damages the vessel wall resulting in denudation of the endothelium, exposing blood, VSMCs, and extracellular matrix to the stent, thus increasing inflammatory signaling and risk of thrombosis (Fig. 1).

PCI and stent placement also disrupt the flow-generated forces within the vessel, which in turn regulate EC oxidant production (Fig. 4). Oscillatory shear stress generated post-PCI has been shown to inhibit re-endothelialization, thus decreasing the ability of the endothelium to regenerate and increasing the likelihood of thrombosis and neointimal hyperplasia. In ECs exposed to oscillatory shear stress, there is increased O₂^•− production via NOX directly and NOX-dependent elevations in XO activity (Godbole et al, 2009). Elevations in O₂^•− decrease ^•NO bioavailability and push ECs further into a dysfunctional state (Fig. 4). Depletion of ^•NO and elevated ROS result in increased expression of adhesion molecules and weaker cell–cell junctions, therefore, elevating vascular permeability (Wang et al, 2010).

Drug-induced oxidants

Second-generation DES have drastically reduced the rates of restenosis; however, they have failed to significantly improve MACE outcomes when compared with BMS. In clinical trials, sirolimus-, paclitaxel-, and, to a lesser extent, zotarolimus-DES have all been associated with prolonged ED (Hofma et al, 2006; Kim et al, 2009; Togni et al, 2007). Clinical trials have shown that 6 months post-paclitaxel-DES implantation, the segment of vessel adjacent to the stented site experienced exercise-induced paradoxical coronary vasoconstriction (Togni et al, 2005). These same vessels had preserved vasodilatory responses to nitroglycerin, indicating they were not insensitive to ^•NO but that paclitaxel exposure was driving the underlying ED.

The initial elution of drugs from DES has been observed to elevate oxidants, predominantly NOX-mediated O₂^•− production. A 7-day infusion with sirolimus was found to reduce the endothelium-dependent vasodilatory response of intact rat aorta to acetylcholine (ACh), while VSMC responses to nitroglycerin were unaffected (Jabs et al, 2008). Loss of ACh-mediated vasodilation was coupled with a 40% reduction in ^•NO production and a significant increase in ROS production from aortic ring segments. Mechanistically, exposure to sirolimus elevated ROS production via increased NOX expression. Similarly, a 7-day infusion of paclitaxel in rats led to reduced flow-mediated dilation and impaired ACh-mediated vasodilation, indicative of ED (Serizawa et al, 2012). These effects are due to paclitaxel-induced increases in NOX expression (subunits p47^phox and gp91^phox).

Vascular inflammation

The extent of PCI-induced inflammation correlates with short- and long-term prognosis (Gach et al, 2007). The ability of PCI to induce inflammatory signaling cascades has been extensively reviewed (Tucker et al, 2021). In brief, in response to injury, ROS/RNS mediate the upregulation of proinflammatory cytokines, chemokines, and adhesion molecules, resulting in platelet activation and the recruitment of inflammatory cells (Fig. 4). This, combined with PCI-mediated elevations in vascular permeability, results in the recruitment of inflammatory cells within the arterial wall. In addition, the release of microthrombi and the lipid-rich contents from the stented plaque triggers the activation of NOD-like receptor protein 3 inflammasomes, ultimately leading to downstream activation of MAPK and NF-κB signaling pathways resulting in elevated ROS generation (Gordon et al, 2011).

PCI with stent implantation also increases the recruitment of neutrophils containing myeloperoxidase (MPO), resulting in elevated HOCl production in ECs. Interestingly, in the short-term, neutrophil-derived MPO in the subendothelial space activates eNOS. In the presence of H₂O₂, MPO-derived ROS elevate cytosolic Ca²⁺, which drives beneficial eNOS phosphorylation and increases ^•NO production (Thai et al, 2021). However, these beneficial effects are short lived as MPO consumes endothelial-derived ^•NO, resulting in an overall reduction in NO bioavailability (Fig. 4). In addition, MPO-derived HOCl contributes to lipid peroxidation and endothelial damage further driving ED (Eiserich et al, 2002).

The Colchicine to Prevent Periprocedural Myocardial Injury in Percutaneous Coronary Intervention pilot study found a potential role for the antioxidant colchicine in the reduction of inflammatory markers post-PCI (Cole et al, 2021). Along with its anti-inflammatory properties, colchicine elevates serum GSH levels and thus may be protective by reducing ROS/RNS levels (Das et al, 2000). Similarly, treatment with the antioxidant and lipid-lowering drug nicanartine was found to reduce the inflammatory response to balloon angioplasty in a rabbit model (Wohlfrom et al, 1998). While antioxidant therapies have thus far failed to translate into clinical practice, these data highlight the central role of downstream oxidative signaling during PCI-induced inflammation, which drives ED and contributes to adverse outcomes.

Restenosis: the role of oxidants

Restenosis is typically defined as >75% narrowing of the cross-sectional luminal area by neointimal tissues, made up of VSMCs and proteoglycan-collagenous matrix. A major factor in the development of restenosis is the failure of the endothelium to functionally recover post-PCI, combined with a prolonged inflammatory response. Human studies show that 2 years after sirolimus-DES implantation, stent struts remained uncovered in 20% of patients, corresponding to increased neointimal hyperplasia and up to a 40% risk of subclinical thrombus formation (Takano et al, 2007). A first-in-human clinical trial in 2005 demonstrated that stents coated with anti-CD34 antibodies could capture circulating CD34⁺ endothelial progenitor cells and increase re-endothelialization post-PCI (Aoki et al, 2005).

However, a follow-up study determined that while early re-endothelialization was improved using this approach, there was no downstream improvement in neointimal thickness at 90 days post-PCI (van Beusekom et al, 2012). Although positive eNOS staining was observed in the regenerated endothelium, there was continuous adhesion of leukocytes to the endothelium, indicating a sustained inflammatory response and decreased NO bioavailability, equating to ED. These data highlight that endothelial coverage is not the only factor to consider when trying to combat restenosis and that endothelial function must be restored to improve clinical outcomes.

Although the initial burst of NOX activation rapidly subsides after stent deployment, low-level O₂^•− production persists during vascular repair. Combining elevated O₂^•− with a compensatory upregulation iNOS- and nNOS-derived ^•NO, to offset impaired eNOS activity, results in the increased formation of RNS that drives inappropriate protein oxidation and hinders endothelial function (Leite et al, 2004).

A recent clinical study has suggested an association between circulating oxidative burden and incidence of restenosis post-PCI when using sirolimus-DES (Ganjali et al, 2022). Elevated SOD is indicative of an increase in antioxidant enzymes in response to elevations in oxidants, thus suggesting that patients with poor prognoses have uncontrolled ROS production. However, upregulation of extracellular SOD post-PCI is counterintuitively beneficial due to its capacity to remove O₂^•−. Initial studies showed that reduced extracellular SOD expression promoted ROS/RNS production and impaired iNOS-derived NO bioavailability, resulting in loss of vascular tone (Leite et al, 2003). Furthermore, elevation of extracellular SOD by local gene therapy in a hyperlipidemic rabbit model significantly accelerated endothelial recovery and reduced neointimal formation (Brasen et al, 2007).

Attempts have been made to therapeutically reduce the oxidative burden associated with restenosis using third-generation β-blockers to reduce O₂^•− production via the inhibition of NOX expression and preventing NOS uncoupling (Mollnau et al, 2003). However, a 2-year follow-up from a clinical trial using carvedilol-loaded stents found no significant differences in the neointimal area or the incidence of MACE compared with BMS (10.5% vs. 30.0%, respectively, p = 0.132) (Kim et al, 2011).

Restenosis: the role of the reducing systems

Highlighting the central role that cellular reducing systems play in governing vascular remodeling, GPx1 was recently found to critically regulate neointimal hyperplasia and restenosis following PCI. GPx1 is an important antioxidant enzyme that reduces H₂O₂ to H₂O and, in turn, oxidizes two GSH molecules to GSSG. GPx1 also catalyzes the reduction of lipid peroxides and other organic hydroperoxides. Deficiency of GPx1 in mice was found to drive ED via elevations in O₂^•− (Dayal et al, 2002; Forgione et al, 2002). PCI-induced damage to the vessel wall has been found to reduce the expression of GPx1 causing knock-on effects on oxido-reductive signaling (Ali et al, 2014). PCI-induced injury in Gpx1^−/−Apoe^−/− mice resulted in a buildup of H₂O₂ and a time-dependent increase in GSH resulting in elevated protein S-gluathiolation, likely via sulfenic intermediates (Charles et al, 2007).

S-gluathiolation of the tyrosine phosphatase SHP-2 inactivates the protein and consequently results in the sustained phosphorylation and activation of orphan proto-oncogene receptor tyrosine kinase ROS1. Sustained activation of ROS1 amplifies its ability to activate downstream kinases involved in VSMC proliferation, migration, and apoptosis, resulting in pathological remodeling and restenosis. In addition, GPx1 is a critical regulator of MAPK and NF-κB proinflammatory pathways. Disrupted GPx1-mediated signaling in ECs from Gpx1^−/− mice resulted in elevated expression of VCAM-1 on ECs, corresponding to increased adhesion of inflammatory cells to the aortic wall in vivo (Sharma et al, 2016). These data demonstrate that ablation of GPx1 activates the endothelium and facilitates vascular inflammation resulting in ED, which in turn can drive restenosis.

An elevation of GSH-synthesizing enzymes was observed in human restenosis, indicating that elevated reductive signaling post-PCI may be a clinically relevant process (Ali et al, 2014). When combined with evidence that PCI reduces GPx1 expression, these data suggest that elevations in reductive signaling may contribute to the deleterious effects of stents and restenosis. However, an alternative explanation could be that GSH production is simply upregulated in response to exacerbated oxidant production. A failure to upregulate GPx1 may lead to dysregulated oxidation of proteins, via the accumulation of S-glutathiolated thiols, leading to altered redox signaling and progression of neointimal hyperplasia.

In support of this, the presence of a single-nucleotide polymorphism within the GPx1 gene (599C/T) that decreases GPx1 activity correlated to an increased risk of restenosis post-PCI with BMS (OR = 2.9; 95% CI: 1.23 to 6.82) (Shuvalova et al, 2012). The complexities of antioxidant signaling pathways may partially explain the failure of antioxidant therapies in clinical trials and highlight that when targeting redox-mediated processes, there is a need for targeted rather than broad-spectrum therapies.

The inducible protein heme oxygenase-1 (Hmox1) is a key player in the vascular reducing system during conditions of inflammation as it plays a protective role against vascular injury by inhibiting neointimal hyperplasia and promoting re-endothelialization (Brouard et al, 2000; Duckers et al, 2001). Hmox1 elicits these opposing effects on VSMC and EC growth through regulation of the yin yang-1 transcription factor (Beck et al, 2010). Hmox1 catalyzes the degradation of heme to Fe²⁺, CO, and biliverdin, which reduces the pool of free heme and limits ROS production (Fig. 5). Biliverdin is quickly reduced, by biliverdin reductase, to the antioxidant bilirubin, which can improve endothelial function, while the released CO protects ECs from apoptosis (Brouard et al, 2000; Maruhashi et al, 2019). Together, these effects contribute to the protective effects of Hmox1 in the endothelium.

Using the antioxidant probucol to induce Hmox1 expression in rabbits, its systemic administration was found to promote functional re-endothelialization, inhibit VSMC proliferation, and reduce macrophage infiltration resulting in reduced restenosis (Lau et al, 2003; Tanous et al, 2006). A 10-year follow-up of the ISAR-TEST-5 randomized clinical trial, comparing a sirolimus-probucol-DES to a zotarolimus-DES, found that the stents had comparable MACE and very late-stage stent thrombosis, suggesting no added benefit of probucol (Kufner et al, 2020). However, the recent FRIENDLY-OCT randomized trial found an ∼8% increase in stent coverage on a probucol-coated sirolimus-DES compared with a bioresorbable sirolimus-DES (Otaegui Irurueta et al, 2022). This suggests that therapeutically targeting cellular reducing systems may be one mechanism to alleviate PCI-induced oxidative damage and limit restenosis.

Neoatherosclerosis

PCI-induced ED plays a critical role in the formation of neoatherosclerosis, which is characterized by the accumulation of lipid-laden foam cells within the neointima, with/without calcification, and/or a necrotic core. Although the advent of DES reduced the rates of restenosis, the incidence of neoatherosclerosis is significantly greater with first-generation DES than with BMS (30% vs. 16%, p < 0.001) (Nakazawa et al, 2011). Critically, neoatherosclerosis typically develops within a year of DES implantation, whereas in BMS it occurs >1 year post-PCI, indicating that the presence of a DES accelerates neoatherosclerosis formation that may lead to late stent failure (Cui et al, 2016).

Theoretically, re-endothelialization of the vessel wall post-PCI should limit inflammation and prevent neoatherosclerosis. However, in a rabbit model comparing EC function between DES and BMS, while there was comparable EC coverage at 1 month, eNOS expression was significantly lower with DES (sirolimus, zotarolimus, or everolimus) (Torii et al, 2018). Diminished eNOS expression suggests that the regenerated endothelium within these arteries was dysfunctional. As discussed above, DES are known to elevate ROS/RNS within ECs, and therefore, DES may drive neoatherosclerosis via prolonged increases in intracellular oxidant levels.

As with atherosclerosis, neoatherosclerosis is an inflammatory process driven by ED that results in the uptake and accumulation of oxidized lipids, mainly oxidized low-density lipoprotein (oxLDL), within the subendothelial space (Fig. 6). Within ECs, lipid transport is largely regulated by caveolin-1 (Cav-1) (Fernandez-Hernando et al, 2009). Cav-1 expression is upregulated in response to ROS via increased binding of Sp1 to the Cav-1 promoter (Fig. 6). Furthermore, ROS induce c-Abl-mediated phosphorylation of Cav-1 and Src-mediated phosphorylation of dynamin-2, promoting Cav-1/dynamin-2 association and driving the formation of caveolae and lipid uptake (Cui et al, 2016). The accumulation of oxidized lipids further drives NOX expression and ROS production, which perpetuates oxidant production within ECs via their effects on eNOS and XO. A buildup of oxLDL potentate inflammatory signaling increases NOX expression, decreases SOD, and limits iNOS expression through inhibition of arginase I/II signaling (Alexander et al, 1991; Barreto et al, 2021).

ROS-induced Cav-1 expression also hinders cellular antioxidant responses (Fig. 6). Elevations in Cav-1 inhibit TrxR, consequently limiting Trx-mediated reduction and allowing oxidant accumulation (Volonte and Galbiati, 2009). In addition, Cav-1 inhibits Nrf2, decreasing the transcription of a plethora of antioxidant proteins (Li et al, 2012). Elevated oxidative signaling combined with reduced removal of oxidants ultimately exacerbates ED, VSMC proliferation, and foam cell formation.

Heme-dependent peroxidases also regulate redox signaling and impact neoatherosclerosis. Indoleamine 2,3-dioxygenase (IDO) signaling within atherosclerotic plaques is cardioprotective and has been shown to regulate atherogenesis and inflammation (Cole et al, 2015). CAD patients with higher IDO activity also trend toward a lower 1-year mortality rate (p = 0.081) (Wongpraparut et al, 2021). In the presence of H₂O₂, IDO1 produces the tryptophan-derived hydroperoxide cis-WOOH, which regulates vasodilation via the oxidation of PKG1α Cys42 (Stanley et al, 2019). This alternative oxidant-mediated vasodilatory response may preserve vascular tone when NO bioavailability is compromised. Conversely, elevated MPO activity correlates with decreased plaque stability due to reduced NO bioavailability and HOCl-mediated accumulation of oxidized lipid species (Rashid et al, 2018).

Concomitant medical therapeutics: Hmox1

Hmox1 expression can be induced by the naturally occurring antioxidants resveratrol and quercetin, which have been shown, both alone and in combination, to inhibit neointimal hyperplasia after carotid balloon injury (Fig. 5) (Huang et al, 2009; Khandelwal et al, 2012; Khandelwal et al, 2010). An early rabbit study found that the use of a resveratrol-quercetin-DES accelerated re-endothelialization by 50%, and significantly reduced in-stent stenosis by ∼65% and macrophage infiltration by ∼60%, compared with a BMS control (Kleinedler et al, 2012). In a 2020 study, oral delivery of resveratrol and atorvastatin in combination with a rapamycin (RAPA)-DES improved strut coverage, with areas of high endothelialization in a rabbit model, however, the combined resveratrol/atorvastatin treatment also had increased neointimal thickness, which could potentially develop into restenosis (Chen et al, 2020).

Further in vivo studies found that a 3D-printed resveratrol-loaded stent protected ECs from H₂O₂-induced apoptosis by promoting ^•NO release and reducing apoptosis, proinflammatory cytokine (e.g., TNF-α) release, and ROS production (Veerubhotla and Lee, 2022). Resveratrol may also improve vascular tone through oxidative activation of PKG1α (Prysyazhna et al, 2019). Although these studies appear promising, currently no clinical trials are investigating the effect of utilizing resveratrol and/or quercetin in PCI for CAD.

Concomitant medical therapeutics: apolipoprotein A1

Apolipoprotein A1 (apoA1), the major protein constituent of high-density lipoprotein (HDL), has been previously considered a potential treatment target for restenosis. Systemic apoA1 infusions in a murine model of stent deployment resulted in enhanced endothelialization, inhibition of platelet activation, and an overall reduction in neointimal hyperplasia (Vanags et al, 2018). However, apoA1 is susceptible to HOCl-mediated oxidative modifications that result in loss of the atheroprotective properties of HDL. To overcome the oxidant-induced loss of function, oxidation-resistant apoA1 (4WF) was combined with local delivery by an adeno-associated viral-coated stent (Hooshdaran et al, 2022). Preliminary ex vivo studies with 4WF apoA1 were promising, as 4WF maintained its ability to export cholesterol in the presence of HOCl and significantly inhibited VSMC proliferation. However, in subsequent studies using an in vivo porcine model, there was a loss in luminal area with 4WF compared with controls, with increased neointimal thickness and no mitigation of restenosis.

Advances in PCI biomaterials

There has been a shift in recent years to attempt to create stents using materials that are not only robust enough to keep open a blocked artery but that are multifunctional and aim to restore healthy vascular function while reducing stent failure.

Multiple laboratories have sought to achieve improved endothelial function and increased NO bioavailability by harnessing the properties of redox-active molecules to reduce labile nitroso bonds in circulating RSNO, namely GSNO, to increase local ^•NO concentrations at the site of stent placement (Fig. 7A–C). Liberated ^•NO is proposed to improve endothelial function, promote healthy vascular tone via ^•NO/cGMP signaling, and reduce thrombus formation by decreasing platelet adhesion. One material is a sulfur-mediated polycarbonate polyurethane stent that relies on the reaction of GSNO with disulfide bonds to release NO and in turn S-glutathiolate thiols on the stent surface (Fig. 7A) (Li et al, 2022). The reducing capacity of stent material can be regenerated via reaction with GSH, producing GSSG and a reduced thiol on the stent surface (Fig. 7D).

Similarly, an NO-eluting hydrogel has been developed that can be used in conjunction with standard stents (Chen et al, 2021). The hydrogel relies on the reducing potential of selenium in catalyzing GSNO decomposition, which is advantageous as selenium is a better nucleophile with a lower pKa resulting in more efficient reduction of nitroso bonds (Fig. 7B). In a porcine model, the NO-eluting hydrogel improved re-endothelialization, reduced neointimal hyperplasia, and upregulated genes involved in the regulation of vascular tone, including HMOX1. The use of selenium-conjugated stent coatings has also been shown to accelerate endothelial repair and decrease injury-induced inflammation, highlighting the potential for this redox-active molecule to improve endothelial function post-PCI (Zhang et al, 2022b).

However, the liberated “NO” from GSNO will be RNS, likely in the form of HNO, not authentic ^•NO. As such, this will lack the ability to influence classical ^•NO/cGMP-mediated vasodilation and may instead add to the oxidative burden in ECs (Fig. 7E). In addition, it is perhaps unlikely that these materials will be efficient at generating NO species in vivo as blood plasma GSH concentrations are typically below 6 μM, thus limiting the capacity of these stent materials to restore their intrinsic reducing capacities (Michelet et al, 1995).

In an effort to promote re-endothelialization and recovery of endothelial function, a functional stent coating that mimics the endothelium has been created (Zhang et al, 2022a). This stent coating utilizes a thiol-ene “click” reaction combining the bioactive Arg-Glu-Asp-Val (REDV) peptide with catechol-grafted chitosan (CS-C) and zinc sulfate. Once in place within the vessel, the REDV peptide promotes re-endothelialization by capturing circulating endothelial progenitor cells (Duan et al, 2019). The CS-C creates a large number of zinc sulfate adsorption sites, which in turn catalyze the release of ^•NO/RNS, again via the decomposition of circulating RNSO and relying on the redox properties of Zn²⁺ to coordinate the reduction of the nitroso bond (Fig. 7C).

Transcriptomics comparing aortas from a rabbit model with control stents versus the CS-C/Zn/REDV-coated stents showed an upregulation in cell adhesion markers combined with a decrease in markers for platelet activation and lipid deposition, suggesting that the REDV peptide and ^•NO/RNS release provide a synergistic effect resulting in the creation a functional endothelium. However, this study had inconsistencies in the impact of the CS-C/Zn/REDV on the expression of matrix metalloproteases, with an apparent upregulation in vivo, which may impact the composition of the vasculature matrix resulting in the weakening of the arterial wall.

It should be noted that previous efforts to create NO-eluting stents have thus far failed to make it into the clinic. Increasing ^•NO levels in the endothelium has clear benefits in reducing thrombosis and restenosis. However, deleterious by-products of ^•NO production and metabolism, namely disulfides via RSNO intermediates, must be considered when developing stent technologies (Wolhuter et al, 2018). Moreover, simply increasing ^•NO may not be enough to improve the long-term function of ECs if intracellular oxidant production is uncontrolled.

To tackle increased ROS production post-PCI, a novel stent was created with an electrospun core-shell nanofiber coloaded with 4-hydroxy-2,2,6,6-tetramethylpiperidine 1-oxyl (TEMPOL) and RAPA that correspondingly serves as an ROS scavenger and VSMC proliferation inhibitor (Wang et al, 2023). Scanning electron microscopy of the luminal side of stented arteries identified significantly higher re-endothelialization with the RAPA/TEMPOL-loaded stent, compared with standard DES or BMS, and ECs were elongated and orientated in the direction of flow. Furthermore, there was a reduction in endothelial adhesion factors (VCAM-1, E-selectin) indicating that RAPA/TEMPOL confers anti-inflammatory and antithrombogenic properties. Notably, neointimal hyperplasia was comparable between the DES and RAPA/TEMPOL-loaded stent at 1 month. Long-term studies are required to determine if this novel stent confers clinical advantages over current DES.

Bioresorbable scaffolds

Bioresorbable stents (BRS), also known as bioabsorbable or biodegradable stents, have the main advantage that they clear from the body within a few years and can theoretically reduce the long-term adverse effects of conventional metallic stents. First-generation BRS came to the market in 2016, however, meta-analysis of randomized trials comparing commercially available BRS to second-generation DES found BRS to be inferior with no clinical advantage (Ke et al, 2020). Nevertheless, while these first-generation BRS failed to positively impact outcomes, the prospect of being able to clear the body of a stent is still an attractive possibility and research into novel BRS is ongoing. The inherent physical and mechanical properties of specific metals, namely Fe, Mg, and Zn, make them attractive candidates for BRS and a great deal of research in the last decade has focused on creating the ideal alloy (Mani et al, 2007).

Fe-based materials are one option for BRS due to their high strength, high elastic modulus, and high ductility. In vivo studies in naïve rabbits found that these stents can be safely implanted and do not exacerbate inflammation, neointimal proliferation, or thrombotic incidents at 18 months postprocedure (Peuster et al, 2001). Pure iron peripheral stents deployed in naïve minipigs were found to be effective and safe, with no signs of iron overload or iron toxicity 1 year postimplantation (Peuster et al, 2006). However, recent studies have shown that corrosion of iron has the strong potential to generate HO^•, likely via the Fenton reaction (Scarcello et al, 2020a). HO^• is a potent oxidizing agent with significant proinflammatory properties that can affect ECs and the surrounding tissue (Ahsan et al, 2003).

Exposure of isolated rat aortic rings to Fe wire was associated with elevations in Hmox1 expression, uncoupling of eNOS, and decreased NO production within ECs, resulting in impaired endothelium-dependent relaxation (Scarcello et al, 2020b). Concurrent treatment with catalase protected the endothelium, likely by reducing H₂O₂ and decreasing HO^• formed via the Fenton reaction. These data suggest that Fe-based BRS may be deleterious to endothelial function and the effect of HO^• production on ECs should be considered when evaluating the biocompatibility of these alloys.

While pure Mg degrades rapidly, biodegradable magnesium alloys are favorable due to their slower degradation rates, biocompatibility, and low thrombogenicity. The release of Mg²⁺ has been proposed to restore endothelial function by counteracting the pro-oxidative effects of common biodegradable polymers, namely PLLA, by attenuating the expression of inflammatory markers and decreasing mitochondrial ROS production (Ko et al, 2021). Early work showed that an Mg-based BRS had significantly higher re-endothelialization and reduced infiltration of foam cells 35 days post-PCI compared with DES in a rabbit model of neoatherosclerosis (Nicol et al, 2020). In support of this, the use of Mg-BRS was associated with increased vasoconstriction to Ach at 1 year post-PCI (vs. SES, p = 0.003), indicating improved endothelial function (Sabate et al, 2019).

However, a separate 3-year follow-up trial found a higher incidence of target lesion revascularization associated with Mg stents, indicative of increased restenosis and stent failure (CI: −20.7 to −1.2; p = 0.030) (Ortega-Paz et al, 2022).

Perhaps one of the most intriguing studies to come out in 2022 was the development of a hybrid material composed of a biodegradable polymer substrate and double anodic/cathodic metallic layers that comprised Mg and Zn aiming to harness the signaling potential of ROS to improve clinical outcomes (Jeong et al, 2022). This work leads on from the development of an H₂O₂-generating biomaterial that couples nitinol stents with biodegradable Mg/Zn to reduce VSMC proliferation (Park et al, 2019). While these ROS-producing biomaterials suppressed the proliferation of cultured VSMCs, neither study determined the ability of stents made from these materials to improve in vivo re-endothelialization or endothelial function.

Furthermore, the rationale for the use of these biomaterials to improve PCI outcomes is based on the ability of ^•NO, not ROS, to suppress VSMC proliferation and improve endothelial function. ROS generated by these composite biomaterials are perhaps even less likely to have beneficial effects on the vasculature, and it is questionable whether increasing ROS production in a state of ED will have beneficial consequences.