Rivaroxaban as a Protector of Oxidative Stress-Induced Vascular Endothelial Glycocalyx Damage via the IQGAP1/PAR1-2/PI3K/Akt Pathway

Endothelial cells (ECs) are surrounded by a negatively charged gel-like layer, the glycocalyx, which serves as a physical barrier between blood and vessel walls as well as a buffer region between cells and the extracellular matrix [1]. The vascular endothelial (VE) glycocalyx consists of glycosaminoglycan side chains (e.g., heparan sulfate, chondroitin sulfate, etc.), long-chain hyaluronic acid (HA) and proteoglycans (e.g., syndecan-1 and CD44) [2, 3]. Syndecan-1 is primarily involved in the structural stabilization of the glycocalyx, which serves as a barrier against vascular leakage and a regulator of cell adhesion and permeability [4]. Syndecan-1 interacts with coagulation factors [5, 6] like tissue factor and tissue factor pathway inhibitor (TFPI) [7], modulating coagulation and influencing the initiation of the extrinsic coagulation cascade. These interactions are crucial for preventing excessive coagulation and thrombosis on the endothelial surface. Additionally, syndecan-1’s interaction with protease-activated receptors (PARs) [8, 9], particularly PAR1 and PAR2, which are activated by thrombin and other proteases during coagulation and inflammatory responses, plays a significant role in modulating cellular responses such as barrier function, inflammation, and wound healing [10].

The glycocalyx is involved in vascular integrity and homeostasis by regulating many physiological processes, such as coagulation, inflammation, vasocontraction and relaxation, vascular permeability, and angiogenesis [11, 12]. Damage to the endothelial glycocalyx decreases vascular barrier function and leads to protein extravasation and tissue edema, loss of nutritional blood flow, and an increase in platelet leucocyte adhesion [13‒16]. VE glycocalyx shedding and degradation are known to be caused by a variety of cellular stressors [17]. Specifically, inflammatory mediators [18], hypoxia or excessive reactive oxygen species [19], ischemia/reperfusion injury [20], atrial natriuretic peptide, and endotoxin [21] are the cause of VE glycocalyx damage. Perturbed glycocalyx aggravates endothelial dysfunction, causing reduced nitric oxide bioavailability [22], inflammatory cytokine release [23], platelet adherence [24], coagulation [25], and leukocyte adhesion [26]-pathologies that are deeply associated with cardiovascular disease (CVD). VE glycocalyx degradation exacerbates inflammatory mechanisms that promote tissue degradation, including reactive activation of matrix metalloproteinases (MMPs), heparanases, and hyaluronidases. Regeneration of glycocalyx has emerged as a critical target for understanding and potentially treating CVD [27]. In addition to chronic CVD such as heart failure [28], atherosclerosis, diabetes [29] or Kawasaki disease [30], loss of the glycocalyx is implicated in acute shock states, such as acute coronary syndrome [31], hemorrhage and septic shock [32, 33]. Noting the potential for pharmacotherapy of both acute and chronic pathology might be useful.

On the other hand, we have clarified that rivaroxaban, an activated blood coagulation factor X (FXa) inhibitor, exerts its effects through multiple pathways [34‒36]. Primarily, as an anticoagulant, rivaroxaban directly inhibits FXa through FXa-PARs signaling, which is crucial in the coagulation cascade, thereby preventing the conversion of prothrombin to thrombin. This inhibition reduces thrombin generation and consequently diminishes thrombin-mediated activation of PARs [37]. We have previously demonstrated that rivaroxaban has not only an anticoagulant effect but also an antiapoptotic and anti-inflammatory effect by inhibiting nuclear-factor-kappa B signaling [34, 35]. A recent study reported that human ECs produce their own coagulation factors that can activate cell surface FX without the addition of exogenous proteins or phospholipids [38] suggesting that the modulation of the FX activity is an important biological process in ECs. As a consequence, we tested for the first time the hypothesis that rivaroxaban protects against endothelial glycocalyx damage.

To explain this protective effect, we studied the link between rivaroxaban activity and phosphatidylinositol-3-kinase (PI3K)/serine-threonine protein kinase (Akt) pathway. Indeed, it is already known that the PI3K/Akt pathway is linked to the VE glycocalyx as it essential for EC survival by activating HA/CD44 [39]. Moreover, it is also known that HA activates the HA/CD44 system by binding CD44 and regulates various signaling pathways through IQ motif-containing GTPase activating protein 1 (IQGAP1). IQGAP1, a scaffold protein that binds to VE growth factor receptor type 2 (VEGFR-2), modulates the actin cytoskeleton, cell-cell adhesion, cellular motility, and morphogenesis. HA-induced CD44-IQGAP1 complex is capable of binding F-actin and regulates cytoskeletal function [40]. Furthermore, IQGAP1 colocalizes and forms a complex with VE-cadherin at the site of cell-cell contact and link VEGFR-2 to the VE-cadherin/βcatenin complex at the human umbilical vein endothelial cell (HUVEC) adherens junctions which may contribute to angiogenesis [41]. IQGAP1 is essential to control the modulation of cytoskeletal architecture and other important physiological function in the vasculature [40].

According to previous study, glycocalyx layer around 1,5-2 μm is observed in cultured HUVECs even in static condition [42]. Hydrogen peroxide is a well-established inducer of oxidative stress in various cell types, including ECs. Numerous studies have demonstrated its efficacy in mimicking oxidative stress conditions that lead to VE dysfunction, including increased endothelial permeability, barrier dysfunction, and apoptosis [43, 44]. In the present study, we analyzed the effect of rivaroxaban and the relation with PI3K/Akt/IQGAP1 pathway in HUVECs under excessive oxidative-stress conditions.

HUVECs (Promo Cell C12203) were cultured at 37°C with 5% CO₂ in endothelial growth medium 2 (EGM2, Lonza CC4176). Cells between passages 3–5 were then seeded at 8,000 cells/cm², in collagen Type 1 (Sigma‒Aldrich C8919, St. Louis)-precoated 96-well plates, 10 cm plate dishes or Lab-Tek slides (Nunc 177437PK, Waltham) until confluence. HUVECs were starved pretreated with or without rivaroxaban (1 μg/mL, approximately 2.29 μm, unless indicated otherwise, Toronto R.C R538000, Toronto) in 0.5% FBS (Sigma F7524, St. Louis) endothelial basal media (EBM2, Lonza CC3156, Basel) for 1 h and stimulated by H₂O₂ (Wako 086–07445, Osaka) 1 mm (unless indicated otherwise) in 0.5% FBS in EBM for the designated hours.

To detect the level of syndecan-1 in the culture supernatant, cells were starved and treated as described in the previous section. Cell culture supernatants (150 μL/well of 96-well plate) were collected, centrifuged at 1,000 × g for 10 min, and was immediately use. The cell lysates were also collected using RIPA buffer (Wako 188-02453, Osaka) (150 μL/well of 12-well plate), followed by a ultrasonication technique, and stored at −80°C. The concentration of soluble syndecan-1 (sCD138) was measured by ELISA (Diaclone 950.640.192, Besançon) at the dilution of 1:1.

FX levels in ECs were measured as previously described [38]. Briefly, cell lysates were prepared from HUVECs grown in 10 cm dishes until confluence in EGM2. 24-h prior to lysate preparation, cells were washed 3 times with Ca⁺², Mg⁺²-containing PBS and were then starved for 24 h in serum-free medium (EBM, Lonza CC3156, Basel + insulin-transferrin-selenium, Thermo Fisher 41400045, Waltham) alone, stimulated with or without H₂O₂ 1 mm. To ensure that all traces of exogenous proteins are eliminated, after removing the media, cells were washed 3 times with 10 mL of cold, sterile Tris buffer (50 mm Tris, pH 7.3, 1% BSA, and 5 mm CaCl₂). Then, on ice, cells were lysed with a lysis buffer (pH 7.4 10 m Tris, 100 mm NaCl, 1 mm EDTA, 1 mm EGTA, 1 mm NaF, 20 mm Na4P2O7, 2 mm Na3VO4, 1% Triton X-100, 10% glycerol, 0.1% SDS, 0.5% deoxycholate) + protease inhibitor (Wako 165-26021, Osaka) 167 μL for 15 min with rocking. Cells were collected with a cell scraper, followed by a ultrasonication technique, place into a chilled tube and centrifuged at 12,000 g for 15 min at 4°C. The cell supernatants were collected and stored at −80°C until assay. The protein was quantified with undiluted lysates using FX ELISA kit (Abcam ab108832, Boston).

After treatments, cells were fixed with 1% paraformaldehyde for 10 min and permeabilized with 0.2% Triton +50 mm NH₄Cl for 10 min. Cells were then incubated overnight with IQGAP1 conjugated to Alexa Fluor 488 (1:50, Santa Cruz 3760211, Dallas), VE-cadherin conjugated to Alexa Fluor 546 (1:100, Santa Cruz 9989, Dallas) or mouse control IgG (1:20,000, Jackson Immuno Research 015-000-003, West Grove). Thereafter, the samples were incubated with secondary antibody for the control IgG and DAPI nuclear stain (Thermo Scientific 62248, Waltham), covered with mounting medium (Diagnostic BioSystems K024, Pleasanton). Multichannel images were acquired with an inverted confocal microscope using a Plan-Apochromat 63X/1.4 Oil DIC objective. The images were Z-projected, analyzed with ImageJ software (NIH, Bethesda). Data are expressed as the means of the fold change +/− SE of the means of the fluorescence intensity from the protein of interest.

The MTT assay was performed using a kit (Roche 11465007001, Basel) as previously described [45] in the presence of rivaroxaban 1 μg/mL followed by 1 mM-H₂O₂ stimulation for 20 min. Control experiments were performed in parallel. To explore the participation of key signaling modulators, cells were pretreated with pharmacological inhibitors for 1–2 h before H₂O₂ stimulation. Specifically, the phosphatidylinositol 3-kinase (PI3K) inhibitor LY294002 at 20 μm (Sigma‒Aldrich L9908, St. Louis), the PAR1 antagonist SCH79797 at 200 nm (Tocris Bioscience 1592, Bristol) and the PAR2 antagonist AZ3451 at 200 nm (Tocris Bioscience 6806, Bristol).

Confluent HUVECs cultured in 10 cm dishes (4.0 × 10⁶ cells/dishes) were treated with or without rivaroxaban followed by stimulation with H₂O₂. Cells were lysed and total protein was extracted with lysis buffers (R&D Systems, Minneapolis). Finally, proteins were analyzed on cell lysates with the Proteome Profiler Human Protease Array kit (R&D Systems ARY009, Minneapolis) and Proteome Profiler Human Apoptosis Array kit (R&D Systems ARY021B, Minneapolis) according to the manufacturer’s instructions. Spots were quantified with IQ800 (Cytvia, Marlborough), and values were reported as the mean pixel density. For data analysis, two-way ANOVA (factor1 = treatment, factor2 = protein) was first performed, multiple comparison was only made for treated cells versus untreated cells (control), followed by a Dunnet post hoc analysis. Multiple unpaired t test was then performed for each protein (parametric test, post hoc test = Holm-Sidak method) using no treated cells as control each time to define a list of protein affected by the treatment. Identified significant proteins were analyzed on reatome.org, followed by network analysis using STRING database, which allowed for the clustering of the selected proteins into the main relevant protein networks (signaling pathway or protein functions) in which they are involved [46].

The MMP and hyaluronidase activities in cell culture media from the cells treated as described above were measured using the Sensolyte 390 generic MMP activity kit (AnaSpec 71158, Fremont) and the hyaluronidase activity ELISA kit (Echelon Biosciences K-6000, Salt Lake City) following the manufacturer’s instructions. For MMP activities, the fluorescence intensity of Mca fluorescence standard and enzyme reaction was measured at Ex/Em = 300 nm/390 nm. Mca fluorescence reference standard curve was plot as RFU (relative fluorescent unit) versus concentration.

Statistical analyses (except for Protein Array analysis which is described above) were performed using JMP 16 (SAS Institute). Data are presented as the mean +/− SE. Continuous variables were compared using the one-way ANOVA and Dunnett’s post hoc analysis. p value of <0.05 was considered statistically significant.

VE glycocalyx degradation occurs in the presence of excessive ROS [22]. Evidence suggests that HUVECs viability was at 70% in 0.2 mm and 40% at 1 mm H₂O₂ after 0, 5-2 h of exposure [47, 48]. We therefore tested 1–100 mm as range of H₂O₂ concentration. As expected, the concentration of soluble syndecan-1 (sCD138, a marker of VE glycocalyx damage) after 20 min of H₂O₂ stimulation in cultured ECs was significantly higher in a dose-dependent manner, accompanied with the decrease of cell viability than that in unstimulated cells (Fig. 1a, b).

The sCD138 levels in cell lysates were also measured to detect the shed syndecan-1, which could be taken up by cells [49] or which remains in the cell membrane. There was no change in sCD138 concentration in all conditions whether with or without H₂O₂ stimulation and with or without rivaroxaban pretreatment (mean 22.1+/− 1.05 pg/mL), suggesting that the role of shed syndecan-1 in the cells or its expression in the cell membrane may be insignificant in our study model.

To examine the drug effect of FXa inhibition, we pretreated cells with or without rivaroxaban, having confirmed ourselves that HUVECs produced their own FX and that this production was 1.9-fold higher when treated with H₂O₂ (Fig. 1c). Rivaroxaban suppressed H₂O₂-induced syndecan-1 shedding from the cell surface in a dose-dependent manner (Fig. 1d).

Since we observed the protective effect of rivaroxaban on the inhibition of glycocalyx impairment in the acute phase under oxidative stress, we examined whether cell viability was also preserved. Analysis of the MTT (3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide) reduction assay 20 min after H₂O₂ stimulation, showed an increase in cell viability in rivaroxaban-treated ECs compared to untreated cells (Fig. 2). Because FXa regulates the activation of extracellular proteases, such as thrombin, activation of the protease-activated receptors (PARs) family can be involved in the protective effect of rivaroxaban. Thus, the implication of PAR1 and PAR2 receptors was explored in H₂O₂-induced cell damage with or without the presence of rivaroxaban. As shown on Figure 2, the inhibition of PAR1 or PAR2 alone was not sufficient to recover the H₂O_2-induced cell damage but the addition of rivaroxaban was able to recover the cell viability in both conditions. Blocking both PAR1 and PAR2 was required to protect from H₂O₂ damage, while neither antagonist alone had this effect (Fig. 2). These data suggest that the inhibition of both PAR1 and PAR2 is required to rescue H₂O₂-induced cell viability.

To further decipher the mechanism underlying the protective effect of rivaroxaban, Proteome Profiler Array kits were used to quantify changes in proteins related to apoptosis and cell stress in H₂O₂-stimulated cells, with or without rivaroxaban pretreatment. The treatment with H₂O₂ induces significant changes in the abundance of the tested proteins, suggesting that H₂O₂ treatment induces apoptosis. In contrast, no significant differences were observed with rivaroxaban alone or when H₂O₂ treated cells were also treated with rivaroxaban (Fig3a, b; see online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000542419 for the originally saved images). This strongly suggests that rivaroxaban protects against apoptosis. There was no significant change in multiple unpaired t test for each protein.

For cell stress-related proteins, the treatments with H₂O₂ or rivaroxaban alone did not induce significant differences in the abundance of the tested proteins (Fig. 3c). However, significant differences were observed when cells were treated with both H₂O₂ and rivaroxaban. This emphatically suggests that co-treatment with both agents may induce changes in the abundance of proteins related to stress. COX-2, p21, SOD2, Thioredoxin, and BCL2 were identified in multiple t test (Fig. 3d).

These proteins were analyzed (overrepresentation analysis) on reactome.org (including interactors, used in 2024-05-30). The Table 1 is the 10 most significant pathways found. Among them, the PI3K/AKT pathway seems to be important (2nd, 3rd, and 9th pathways in Table 1). The protein networks are shown in Figure 3f.

Based on our observations in protein profiling, we then verified if the PI3K/Akt pathway can modulate the effect of rivaroxaban. Improvements in cell viability (Fig. 4a) and sCD138 release (Fig. 4b) mediated by rivaroxaban in ECs were downregulated by the addition of a PI3K inhibitor (LY29002) indicating that PI3K/Akt seems to be involved in the protective effect of rivaroxaban.

To further clarify the chronological changes in H₂O₂-stimulated syndecan-1 release, we analyzed the sCD138 concentration in the culture supernatant and other EC glycocalyx-associated factors that could contribute to the chronic phase. After 12 and 24 h of 1 mm H₂O₂ stimulation, sCD138 was no longer detectable in the culture supernatant, suggesting the degradation of the shed EC glycocalyx by oxidative stress (data not shown). However, ECs pretreated with rivaroxaban accompanied an amelioration of cell viability (Fig. 5a). Rivaroxaban pretreatment also demonstrated a greater increase in hyaluronidase activity (Fig. 5b). In the vasculature, hyaluronan is mainly incorporated into the glycocalyx [2, 3]. The degradation of a small amount of hyaluronan (10–100 mg/day in the blood of adult humans) can be considered as its turnover, and this process is highly dependent on the activity of hyaluronidases, a family of enzymes that degrade HA [29]. This enhanced remodeling process could help to clear damaged components of the glycocalyx, thus maintaining its barrier function and protective properties against oxidative stress and inflammatory mediators. By facilitating the regulated turnover of hyaluronan and possibly other glycosaminoglycans within the glycocalyx, rivaroxaban could help preserve the essential permeability barrier and mechanotransduction functions of the endothelial layer. Furthermore, the removal of old or oxidatively damaged hyaluronan through increased hyaluronidase activity may prevent the accumulation of dysfunctional glycocalyx components, which could otherwise lead to increased vascular permeability and leukocyte adhesion, exacerbating inflammatory processes. Taken together, our results suggest that rivaroxaban could also have a protective role during the turnover of the EC.

A previous report has shown that the loss of syndecan-1 is associated with suppressed MMP activity, which is regulated by sphingosine 1-phosphate (S1P)-dependent pathways [50]. However, in our study, although H₂O₂-induced MMP activities, these activations were not significant despite the addition of rivaroxaban (Fig. 5c), suggesting that a mechanism other than the regulation of MMPs modulates the protective effect of rivaroxaban on the EC glycocalyx.

Our observations indicate that rivaroxaban-augmented endothelial glycocalyx protection is regulated by a process other than the activation of MMPs but is modulated by the PI3K/Akt pathway. It has been shown that the PI3K/Akt pathway is essential for EC survival by activating HA/CD44 [39]. Moreover, HA activates the HA/CD44 system by binding CD44 and regulates various signaling pathways through IQ motif-containing GTPase activating protein 1 (IQGAP1). For example, IQGAP1 interacts with CD44 and promotes cell proliferation, cell migration, angiogenesis, cell survival, and cell differentiation [51]. Thus, we then focused on this scaffold protein, IQGAP1. As a scaffold protein involved in different signaling pathways (actin cytoskeleton [51], cellular adhesion [52], cell cycle, and transcription), IQGAP1 has a unique ability to potentially couple diverse cellular functions. For instance, IQGAP1 is associated with actin dynamics through direct binding of actin and indirect regulation through Cdc42/Rac1 [53].

We examined the subcellular localization of IQGAP1 and VE-cadherin before and after 5 min of H₂O₂ stimulation with or without rivaroxaban pretreatment in a confluent monolayer of HUVECs using confocal microscopy. As shown in Figure 6a, in unstimulated ECs, IQGAP1 was mainly found at cell-cell contacts, where it colocalized with VE-cadherin. H₂O₂ stimulation reduced the staining of both VE-cadherin and IQGAP1 at cell-cell contacts, consistent with the loss of cell adhesions, whereas it increased the staining of IQGAP1 and VE-cadherin at the perinuclear area. Strikingly, rivaroxaban inhibited H₂O_2-induced IQGAP1 translocation from the cell membrane to the perinucleus and was diffusely distributed within the cytosol (Fig. 6a, b). Interestingly, in unstimulated ECs, IQGAP1 was relocalized in the nucleus when treated with rivaroxaban (Fig. 6a, c). There was no specific binding in cells incubated with control IgG (Fig. 6a). Taken altogether, our results strongly suggest, that the protective effect of rivaroxaban is mediated by the PI3K/Akt/IQGAP1 pathway under excessive ROS-induced vascular cell damage conditions.

Our observations suggest that rivaroxaban has a protective effect against oxidative stress-induced glycocalyx damage, improving cell viability and preserving the integrity of VE-cadherin interactions. To the best of our knowledge, this is the first work to propose a comprehensive mechanism for endothelial glycocalyx protection by rivaroxaban.

As damage to the endothelial glycocalyx reduces vascular barrier function and results in several physiological perturbations, preserving, and restoring the endothelial glycocalyx is a potential target to prevent CVD [27]. Here, we propose rivaroxaban, as a novel protector. Previous studies have shown that metformin, a hypoglycemic drug, or sulfoxide, a heparin sulfate-like compound, improve he endothelial glycocalyx in diabetic mice or in septic mice, respectively [54, 55]. Hydrocortisone inhibits mast-cell degranulation, tumor necrosis factor-a inhibitor direct antagonizes glycocalyx degradation, avoidance of high-volume loading prevents the liberation of natriuretic peptides, and infusion of human plasma albumin maintains stability of the EC surface layer and has been also reported to help reduce oxygen radical stress on glycocalyx [56]. Additionally, antithrombin III, a physical inhibitor of serine proteases such as thrombin and elastase, which inhibits coagulation abnormalities, has also been reported to protect ECs by reducing susceptibility to enzymatic attack [57]. However, our new findings on another anticoagulant drug, rivaroxaban, another anticoagulant drug, are quite particularly significant, as antithrombin is typically administered with heparin or heparinoids, a combination unfavorable for the binding of antithrombin to the VE glycocalyx.

In our study, rivaroxaban was shown to be a safe agent for HUVECs at concentrations up to 10 μg/mL, protecting the EC glycocalyx from damage induced by H₂O₂. This protective effect was accompanied by an improvement in cell viability. Control over the activity of the PI3K/AKT pathway appears to be crucial in these effects; PI3K-dependent EC viability, apoptosis, and endothelial glycocalyx synthesis-related proteins were elevated in our observations, consistent with former reports [58‒60]. According to a previous study, edoxaban, another FXa inhibitor, stabilizes ECs, counteracts the antiangiogenic effects of FXa, and has remarkable anti-inflammatory action, preventing peripheral blood mononuclear cell adhesion and transmigration through the endothelium. Its main molecular mechanism seems to be the regulation of the PAR1-2/PI3K/ nuclear-factor-kappa B pathway under the stimulation of FXa [61]. In agreement with this study, in our experiments, the inhibition of PI3K by a specific pharmacological inhibitor (LY294002) underscored the important role of rivaroxaban in improving glycocalyx damage and cell viability. Furthermore, interestingly, we were able to demonstrate the direct protective effect of rivaroxaban without the addition of exogenous FXa. This could be explained by the physiological function of human ECs, which produce their own coagulation factors that can activate cell surface FX without the addition of exogenous proteins or phospholipids (Fig. 1b) [38].

The non-hemostatic functions of FXa are largely dependent on signaling pathways mediated by PARs [62]. PARs are a family of related G-protein-coupled receptors that are activated upon cleavage of their extracellular domain [63]. For PAR1, PAR2, and PAR4, these cleavage sites generate various tethered ligands, which can alter cellular responses [64]. FXa promotes conversion from prothrombin to thrombin and subsequent activation of PAR1 [65]. In addition to promoting thrombin activation, FXa can directly signal through PAR1 and PAR2 and may vary in many different pathophysiological states [66]. According to previous studies, both PAR1 and PAR2 promote activation of the PI3K/Akt pathway, which is related to a wide range of cellular biological processes [67, 68]. These receptors play significant roles in modulating inflammatory responses by influencing cytokine production and leukocyte recruitment and are integral to maintaining endothelial barrier integrity. Activation of PAR1 and PAR2 can lead to endothelial barrier disruption [69, 70], which facilitates immune cell transmigration during inflammatory responses [71]. In the context of therapeutic intervention, the use of specific antagonists such as SCH79797 for PAR1 and AZ3451 for PAR2 has been shown to mitigate these effects. SCH79797 acts as a competitive antagonist, directly blocking the binding of agonists like thrombin to PAR1, thereby inhibiting the downstream signaling cascades that lead to inflammation and barrier disruption [72]. Alternatively, AZ3451 represents a novel approach to receptor inhibition through its allosteric mechanism of action. Unlike competitive antagonists, AZ3451 binds to a site distinct from the ligand-binding domain on PAR2, inducing a conformational change that reduces the receptor’s ability to engage in its natural signaling pathways [73]. This allosteric inhibition is particularly valuable because it can potentially modulate the receptor’s function in a more subtle and regulated manner, potentially reducing side effects associated with complete receptor blockade.

In the present study, the inhibition of PAR1+PAR2 using their specific antagonists (SCH79797 and AZ3451) improved H₂O₂-induced cell damage, although PAR1 or PAR2 inhibition alone was not able to rescue cell viability. This is consistent with the functional mechanism of rivaroxaban and has implications, suggesting that FXa inhibitors can be more effective than selective PAR1 inhibitors, such as thrombin inhibitors, in terms of cellular protection. Moreover, as AZ3451 is an allosteric antagonist, our observations indicate that subtle modulations of PAR2 activity (rather than complete inhibition) are sufficient to protect against oxidative damage induced by H₂O₂.

IQGAP1 is a VEGFR-2 binding protein that interacts with actin, cadherin, and β-catenin, thereby regulating cell functions [74]. Moreover, it has been shown that IQGAP1 and α-catenin compete for binding to β-catenin and that IQGAP1 dissociates α-catenin from the β-catenin-α-catenin complex in vitro [75]. In line with earlier research, our study demonstrated that IQGAP1 was localized at the cell membrane with VE-cadherin in unstimulated ECs, possibly stabilizing the cadherin-catenins complex, which results in strong cell-cell adhesion [74]. In contrast, oxidative stress induced the loosening of stable cell-cell contact, suggesting that IQGAP1 interacts with β-catenin, dissociating α-catenin from the cadherin-catenins complex, which may lead to weak cell-cell adhesion. From our findings, we propose a model in which rivaroxaban prevents IQGAP1 from interacting with β-catenin and maintains strong cell-cell adhesion under oxidative stress. Interestingly, when rivaroxaban was added to unstimulated ECs, IQGAP1 translocated to the nucleus, where the complex may modulate the transcription of several genes [76, 77]. The underlying mechanism remains to be investigated. IQGAP1 functions as a scaffold, assembling crucial components of PI3K-Akt pathway and works as a key controlling cell growth or survival [78]. In HUVECs, it has been shown that statin preparations repairs lipopolysaccharide-stimulated endothelial injury through the PI3K/Akt/IQGAP1 pathway activity and platelet-derived growth factor receptor β [79]. In summary, our results suggest that PAR1-2/PI3K/IQGAP1 pathway is an important regulator in the protective process induced by H₂O₂ injury.

The glycocalyx is also able to directly interact with different plasma proteins and circulating cells, regulating inflammation and coagulation [27]. For instance, an intact glycocalyx acts as a barrier to leukocyte adhesion. Increased adhesion of leukocytes is observed only upon degradation of the glycocalyx. On the other hand, coagulation FXa is known to stimulate inflammatory changes, including MCP1-expression or IL-8-release, in ECs and leukocyte under conditions of excessive oxidative stress [80‒82]. Our study reveals a possibility that rivaroxaban may also protect against monocyte EC-adhesion by reducing the production of inflammation mediators, and that the regulation of glycocalyx can certainly be the key to this mechanism.

In this study, we have elaborated on the fundamental role of the endothelial glycocalyx in vascular health, underlining its composition and function in maintaining endothelial integrity and mediating cellular responses. However, it is crucial to consider the limitations inherent in our experimental model which utilizes a static, monolayer setup of ECs. In vivo, the endothelial glycocalyx is subjected to constant hemodynamic forces, particularly shear stress from blood flow, which significantly influences its structure, function, and the cellular behavior of ECs. Shear stress has been shown to affect the production of critical components of the glycocalyx and modulates its protective functions against oxidative stress and inflammation [83]. The absence of flow conditions in our static cell culture model means that the dynamic interactions between blood components and the glycocalyx, crucial for its physiological and pathological responses, are not replicated. This limitation could affect the generalizability of our findings to in vivo conditions, where fluid shear stress plays a pivotal role in glycocalyx maintenance and function. In addition, the use of ECs cultured in a monolayer does not fully capture the complex three-dimensional architecture and multicellular interactions present in blood vessels. In vivo, ECs interact not only with each other but also with other types of cells such as smooth muscle cells and pericytes, which contribute to the overall vascular function and stability. The simplified model of a cell monolayer might not adequately reflect these interactions or the spatial organization of the glycocalyx and its associated cellular components. Given these limitations, while our findings provide valuable insights into the protective effects of interventions like rivaroxaban on the endothelial glycocalyx under oxidative conditions, caution must be exercised when extrapolating these results to more complex in vivo settings. Future studies could benefit from incorporating dynamic flow models, such as microfluidic systems or organ-on-a-chip technologies, which allow the simulation of blood flow and its mechanical effects on ECs. Additionally, employing advanced 3D culture systems or co-culture models could help in better mimicking the in vivo vascular environment, thereby enhancing the physiological relevance of the findings [84].

Here, we demonstrate a new insight into an in vitro model of oxidative stress-induced vascular damage in which a single use of rivaroxaban may have a protective effect on ECs. This suggests potential applications beyond its primarily use in thrombosis prevention, including benefits in managing or preventing conditions like atherosclerosis or vascular inflammation where glycocalyx degradation plays a key role. Furthermore, it could improve outcomes in conditions of oxidative stress such as diabetes, sepsis, or ischemia/reperfusion injury. These findings could also stimulate further research into new therapeutic strategies that target glycocalyx preservation.

Rivaroxaban protects against oxidative stress-induced glycocalyx damage via IQGAP1 scaffolding to maintain vascular permeability, through the PI3K/Akt cell survival signaling pathway.

We are grateful to Dr. Emiyu Ogawa assisted us to compose this article. We thank Dr. Yoshinobu Hara and Mr. Manabu Tanaka for providing technical assistance with confocal microscopic imaging. Images were acquired at the imaging platforms at the Kitasato University School of Medicine. Dr. Pierre Joanne and Dr. Genjiro Suzuki provided support, encouragement, and scientific advice.

HUVECs used in our experiments were purchased from Promo Cell, and no animal experiment or human’s tissue were included in our study. According to the “3rd Scope” of the Ethical Guidelines for Medical and Biological Research Involving Human Subjects, these guidelines do not apply to specimens and information that have already been established as having academic value, are widely used for research purposes, and are generally available. Therefore, ethical approval was not required under national guidelines or the guidelines of the Clinical Research Review Board of the Kitasato Institute.

The authors declare that they have no competing interests.

This work was supported by JSPS (Japan Society for the Promotion of Science) Kakenhi Grant No. JP19K11371 (M.Y.-T.) and JP20K17094 (L.K.), SRL Research grant for young doctors (L.K.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

L.K. conceived, performed all the experimental work, and wrote the manuscript with input from all authors. T.I., Y.M., T.H., M.S., N.M., and R.K. provided technical assistance on experiments, J.O. supervised the quantitative work. Y.I., R.K., M.Y.-T., and F.O. supervised the work gave their expertise, revised the manuscript. J.A. directed the project.

All data generated or analyzed during this study are included in this article and its online supplementary material files. Further enquiries can be directed to the corresponding author.