Association between atherogenic risk-modulating proteins and endothelium-dependent flow-mediated dilation in coronary artery disease patients

All participants provided written informed consent, and the study was approved by the Liverpool East NHS Research Ethics Committee (approval reference no. 13/NW/0088) and conformed to the Declaration of Helsinki.

Sixty-four patients undergoing prospective percutaneous transluminal coronary angiography (PTCA) and/or percutaneous coronary intervention (PCI; angioplasty) were recruited from Liverpool Heart and Chest Hospital (LHCH). Patients were excluded if they were unable to give informed consent or had undergone a trans-radial cardiac catheterization or acute coronary syndrome within the last 3 months. Eight patients had angiographically normal coronaries following PTCA and were retrospectively excluded. As such, data from 56 patients with established CAD were included in this study. A summary of patient characteristics, medications, hemodynamic variables, and previous catheterization is included in Table 1.

Vascular function measurements were assessed prior to catheterization (1–4 h). In brief, the catheterized radial artery was assessed using two non-invasive measurements: (a) flow-mediated dilation (FMD) followed by (b) vascular dilation to incremental handgrip exercise. The order of these tests was maintained in all participants. A standardized period of 10 min was observed between tests to allow artery function to recover to baseline levels. Patients then underwent PTCA and/or PCI and ECs were collected from the catheterized radial artery, as described below.

PTCA and/or PCI were performed predominantly via the right radial artery under local anesthesia, as previously described (Tryfonos et al. 2020). Briefly, the catheterized artery was punctured with a 21-gage needle and a 5-7F hydrophilic sheath introducer (PreludeEase, MeritMedical, UK) was inserted. All patients received a weight-adjusted dose of heparin. Thirty patients received an additional ~ 1.5 mg/ml of vasodilator isosorbide di-nitrate, as indicated by the interventional cardiologist. Once the procedure was finished, the J-shaped guidewire (3 mm J TEF 150 cm × 0.35″, KIMAL, UK) was transferred immediately to ice-cold dissociation solution (~ 30 ml) (0.5% bovine serum albumin, 2 mmol/L EDTA, and 100 ug/mL heparin in Dulbecco's Phosphate-Buffered Saline (DPBS)). An additional sterile J-shaped guidewire was advanced into the radial artery (~ 3–4 cm above the sheath), run back and forth to collect ECs and transferred immediately to the dissociation buffer. All introducer sheaths were removed at the end of the procedure.

Vascular function measurements were completed in a quiet room between 0800 and 1100 h. Patients were fasted (including caffeine and alcohol) and asked to abstain from exercise and cigarettes for 12 h before the visit (Thijssen et al. 2019, 2011). Part of the vascular data (33 patients) has been published elsewhere (Tryfonos et al. 2020), where the methods have been described in detail. Briefly, patients rested in the supine position for > 10 min before blood pressure (BP) and heart rate (HR) were measured using an automated sphygmomanometer (GE Pro 300V2, Dinamap, Tampa, FL, USA). FMD and arterial response to exercise were then measured in the catheterized radial artery (10–15 cm proximal from the scaphoid bone in the wrist), using a 12-MHz multi-frequency linear array probe attached to a high-resolution ultrasound machine (T3000; Terason, Burlington, MA, USA) (Tryfonos et al. 2020). For the exercise protocol, patients performed a 3 × 3-min bout of handgrip exercise, in a seated position, at 5%, 10% and 15% of their pre-determined maximal voluntary contraction (MVC) (Takei 5420 Grip-D Digital Hand Grip Dynamometer, Japan), with 1-min rest between bouts. Arterial diameter/velocity recordings were measured at baseline and during the 1-min rest. The percentage difference in diameter from the baseline to peak in response to handgrip exercise (the maximum value at either 5%, 10% or 15% MVC) was then calculated (HE%) and used for further analysis. Custom-designed edge-detection and wall-tracking software was used to analyze both the FMD and handgrip exercise to minimize investigator bias (Woodman et al. 2001; Thijssen et al. 2011).

EC isolation and protein expression measurements were performed as previously described (Pierce et al. 2011; Donato et al. 2007). Briefly, ECs were recovered from the two J-shaped guidewires by washing and centrifugation. Cells were then fixed with 3.7% formaldehyde, plated on glass slides (VWR, REF number: 631-0705, USA) and frozen at − 80 °C until analysis. At the time of analysis, cells were re-hydrated and incubated for 1 h at room temperature with primary antibodies against eNOS (610,297, BD, USA), Phospho-eNOS Ser¹¹⁷⁷ (peNOS Ser¹¹⁷⁷) (07-428-I, Merck), NAD(P)H oxidase subunit NOX2 (kind gift from Prof Mark Quinn, Montana State University), tumor suppressor protein p53 (p53) (Ab26, Abcam, UK), cyclin-dependent kinase inhibitor 1 (p21) (Ab109520, Abcam, UK), cyclin-dependent kinase inhibitor 2A (p16) (Abc51243, Abcam, UK), nitrotyrosine (Ab7048, Abcam, UK), ET-1 (PA3-067, Thermo Fisher Scientific, USA), NFκB p65 (NB100-56,712, Novus, UK). Cells were also stained for vascular–endothelial (VE) cadherin (NB600-1409, Novus, UK) for positive identification of the endothelial phenotype and DAPI for nuclear integrity.

Images were acquired using an inverted confocal microscope (Zeiss LSM-710, Carl Zeiss, Germany) with a 63 × 1.3NA oil immersion objective. Positive staining for VE cadherin coupled with a single intact nucleus was used to reliably select ECs (Casey et al. 2017). DAPI was excited using the 405 nm line of the diode laser and detected with 371–422 nm emission. Alexa Fluor 488 was excited with the 488 nm line of the argon laser and detected with 493–559 nm emission. Alexa Fluor 546 and 633 fluorophores were excited with 543 nm and 633 nm lines of the helium–neon laser and 548–623 nm and 638–747 nm emission, respectively. The images were acquired at a resolution of 1,024 X 1,024 pixels and stored in 24-bit tagged image format file format.

All images were analyzed using ImagePro Plus 5.1 (Media Cybernetics Inc, Bethesda, MD, USA). To ensure only the cytosolic fraction was assessed, nuclear regions of the ECs, identified through the DAPI stain, were extracted from the rest of the EC image identified using the VE cadherin image. The resulting mask was then overlaid onto the corresponding protein of interest image. Mean fluorescence intensity of the protein of interest signal was then quantified within the endothelial cytosolic-specific area. EC protein expression data are reported as ratios to human umbilical vein EC (HUVEC) protein expression. Slides were systematically scanned and at least 25 consecutive ECs were captured and analyzed for each protein in each patient, as previously suggested (Colombo et al. 2002). A single technician analyzed each batch of slides. Technicians were blinded to subject identity during the staining and analysis procedures.

FMD was positively correlated with eNOS Ser¹¹⁷⁷ phosphorylation (r = 0.290, P = 0.037) (Fig. 1) and p16 (r = 0.426, P = 0.002) and p21 protein expression (r = 0.307 P = 0.027) (Fig. 2). When eNOS Ser¹¹⁷⁷ phosphorylation was normalized to eNOS content, the significant correlation was no longer present (r = 0.218, P = 0.121) (Table 2). Further, hierarchical multiple regression models suggested that p16 protein expression remained significantly associated with FMD (P = 0.008), despite accounting for MAP, BMI, sex and age. In contrast, p21 and eNOS Ser¹¹⁷⁷ phosphorylation associations with FMD appeared to be affected by the patients’ characteristics (P = 0.099 and P = 0.175, respectively), and the limiting factor in both cases was the MAP (P = 0.129 and P = 0.522). In contrast, BMI, sex and age appeared to not affect the associations between p21 and eNOS Ser¹¹⁷⁷ phosphorylation with FMD (P < 0.05). No significant association was observed between FMD and eNOS, NOX2, ET-1 and NFκB protein expression or nitrotyrosine abundance in patient ECs (Table 2). HE-mediated dilation was not correlated with any of the proteins examined (Table 2).

There was no correlation between FMD and the percentage change in diameter (%HE) during handgrip exercise (r = − 0.015, P = 0.985).

Previous work in individuals with CVD risk factors but without established CAD suggests that impaired endothelium-dependent dilation in humans is largely related to elevated oxidative stress (Donato et al. 2009; Seals et al. 2011; Gates et al. 2007; Lavi et al. 2008; O'Driscoll et al. 1999; O'Driscoll et al. 1997a, b; Cheetham et al. 2000) rather than impaired NO production (Donato et al. 2009; Pierce et al. 2011). In contrast with these observations in pre-clinical populations, we report that markers of oxidative stress (expression of NOX2 and nitrotyrosine content) were not associated with endothelium-dependent vasodilation in patients with established CAD. However, total eNOS Ser¹¹⁷⁷ phosphorylation, reflecting the overall activation of eNOS, was positively associated with endothelium-dependent dilation in these patients. Together with the lack of association between FMD and eNOS content or peNOS Ser1177/eNOS ratio may suggest that both lower eNOS content and ability to phosphorylate the available eNOS combine to reduce overall eNOS phosphorylation. The overall phosphorylation of eNOS is relevant (Hambrecht et al. 2003) as this appears to determine NO production and not the content or ability to phosphorylate the available eNOS (Mount et al. 2007).

Taken together the above evidence may indicate a difference in the regulation of endothelium-dependent dilation between patients with established CAD and pre-clinical subjects. Our data seem to suggest that further elevations in oxidative stress do not contribute to progressive declines in endothelium-dependent vasodilation in patients with diagnosed CAD. On the other hand, the activation of eNOS by Ser¹¹⁷⁷ phosphorylation becomes increasingly important to maintain NO production. In support of this, Hambrecht et al. (2003) reported a positive association between increased eNOS Ser¹¹⁷⁷ phosphorylation and endothelium-dependent dilation in the arteries of CAD patients. Furthermore, a number of drugs used to treat CVD and improve endothelium-dependent dilation act by stimulating eNOS Ser¹¹⁷⁷ phosphorylation (Huang 2009). Estrogens (Hisamoto et al. 2001), statins (Kureishi et al. 2000) and peroxisome proliferator activated receptors agonists (Aikawa et al. 2004) are all examples of such drugs.

In our ECs, we also demonstrated a positive association between endothelial eNOS Ser¹¹⁷⁷ phosphorylation and oxidative stress (NOX2 protein content). It has previously been shown that reactive oxygen species can modulate NO production by influencing eNOS activity (Li et al. 2014). Specifically, H₂O₂ has been shown to increase phosphorylation of eNOS at Ser¹¹⁷⁷ (Thomas et al. 2002). As such, the association between eNOS Ser¹¹⁷⁷ phosphorylation and NOX2 could reflect an attempt to increase NO production and overcome elevated NO scavenging by superoxide, as it has been hypothesized in aging (Donato et al. 2007, 2015).

Cellular senescence, a stress-response resulting in irreversible growth arrest of a cell, is emerging as a potential driver of endothelial dysfunction and the development of atherosclerosis (Katsuumi et al. 2018). The accumulation of these cells alongside the pro-inflammatory and pro-oxidative phenotype of these cells have been proposed as mediators of the adverse effects of senescence (Childs et al. 2015; Minamino et al. 2002). Direct evidence associating senescence with endothelial function in humans is limited. However, recent work suggests that age-associated reductions in endothelium-dependent dilation were negatively associated with endothelial p53, p21 and p16 protein content (Rossman et al. 2017). In contrast to this, we report that FMD was positively associated with endothelial protein content of p21 and p16 in patients with established CAD. These positive associations were unexpected. Given the lack of human studies investigating vascular senescence in CAD, we cannot fully explain these observations. To explore the above paradoxical associations, we ran hierarchical multiple regressions accounting for patients’ characteristics. The results showed that endothelial senescence (at least p16 protein expression) was associated with higher endothelial function in these patients independently of individual characteristics. Mean arterial pressure appeared to influence the association between FMD and p21 expression. Endothelial (dys)function and arterial pressure are mechanistically (Konukoglu and Uzun 2017) and statistically related (Holder et al. 2021). Thus, it is difficult to differentiate the effect of arterial pressure on FMD, especially in our study which included uncontrolled hypertensives, controlled hypertensive and non-hypertensive patients. The prevailing hypothesis is that endothelial senescence leads to vascular dysfunction. However, studies in murine models have shown that knockout of p53 (Mercer et al. 2005) and p21 (Khanna 2009) resulted in greater severity of atherosclerotic lesions. As such, the role of senescence within the vasculature may depend on the stage of the progression of atherosclerosis (Childs et al. 2015, 2016). It is possible that senescence initially plays a protective role in atherosclerosis. Indeed, it has been proposed that senescent cells may delay the atherosclerotic process in the beginning as they prevent cell growth by activating cell arrest and/or perhaps apoptosis pathways (Mercer et al. 2005). Future studies should test the causal contribution of senescence at different stages of CAD pathology.

Markers of senescence (p53, p21 and p16) were associated with oxidative stress (EC nitrotyrosine abundance and NOX2 protein content). These associations are supported by previous work showing that oxidative stress leads to senescence through the activation of p53/p21 and p16 pathways (Bhayadia et al. 2016; Childs et al. 2015; Muñoz-Espín and Serrano 2014). Surprisingly, endothelial inflammation (NFκB) was not associated with any marker of senescence. Importantly, NFκB is known to initiate and maintain the senescence associated secretory phenotype via upregulating inflammatory mediators and reactive oxidative species (Salminen et al. 2012; Childs et al. 2015).

In addition, p21 protein content was associated with eNOS Ser¹¹⁷⁷ phosphorylation. This finding is also in contrast to previous data surrounding the role of senescence in ECs. In vitro data from human aortic ECs showed that senescence reduced eNOS protein content and activity (Matsushita et al. 2001). However, the positive association may provide a link between elevated expression of senescence markers and improved endothelium-dependent dilation, given that eNOS Ser¹¹⁷⁷ phosphorylation was also positively associated with endothelium-dependent dilation. The apparent disconnect between in vitro and in vivo data stresses the need for future human studies to evaluate the role of senescence at different stages of CAD development.

NFκB is a key transcription factor in the regulation of pro-inflammatory markers (Kempe et al. 2005). Endothelial expression NFκB had a positive correlation with NOX2 protein content. This association suggests that NFκB may be an important link between inflammation and oxidative stress (Donato et al. 2009; Marchio et al. 2019). Indeed, previous work demonstrates the potential for NAD(P)H oxidase-dependent induction of NFκB (Clark and Valente 2004). In support of this, NFκB inhibition reduced NOX2 expression and improved endothelium-dependent dilation in obese subjects (Pierce et al. 2009). This suggests a vicious cycle between inflammation and oxidative stress.

The expression of the potent vasoconstrictor ET-1 had a strong positive correlation with NFκB expression. ET-1 has been proposed to affect endothelial function via inflammatory pathways (Bohm and Pernow 2007; Bohm et al. 2007). Cardiac overexpression of ET-1 in mice is associated with increased activation of NFκB (Yang et al. 2004). In turn, NFκB stimulates ET-1 expression (Bohm and Pernow 2007; Virdis and Schiffrin 2003). As such, our data further highlight the critical role of NFκB as a key regulator in the development of endothelial dysfunction in CAD.

FMD was not correlated with the dilation induced during handgrip exercise. Exercise-induced dilation was also not associated with any of the atherogenic risk-modulating proteins assessed. Our team previously showed impaired FMD but preserved exercise-induced dilation in CAD patients following catheterization-induced damage (Tryfonos et al. 2020). This suggests different mechanisms may be responsible for dilation in these two tests. FMD is determined by a large and transient increase in shear stress. This is opposed to a more gradual increase in shear stress observed during exercise (Tremblay and Pyke 2018). FMD response is also known to be endothelium-dependent and mediated to a large extent though endothelial NO production (Green et al. 2014). In contrast, regulation of blood flow during exercise is a more complex process that involves a number of mechanisms (e.g., transmural pressure and vasoactive compounds) with multiple interactions and redundancy (Hellsten et al. 2012; Schrage et al. 2004, 2007; Green et al. 2017; Padilla et al. 2006). We believe the lack of association between FMD and exercise response in CAD patients add weight to the argument of different mechanisms underlined among these physiological responses.

A limitation of this study is that vascular measurements were taken on the day of catheterization. As such, although guidelines for FMD were followed (Thijssen et al. 2019), patients continued their medications as instructed by their consultant. Such medications could have affected the in vivo assessments of vascular function but are unlikely to affect the protein expression of ECs. Previous studies in CAD patients suggest that our radial FMD data are within the normal range (Dawson et al. 2010, 2012). Handgrip exercise was conducted 10 min after FMD assessment. To our knowledge, the influence of prior FMD assessment on response to handgrip exercise has not been established. Therefore, it is possible that the response to the handgrip exercise was influenced by the earlier hyperaemic stimulus. However, previous work has shown that FMD response was not influenced by earlier FMD assessment when the vessel was allowed to rest for ~ 5 min (Barton et al. 2011). Finally, the current study only measured total EC expression of NFκB p65. Given that translocation of NFκB to the nucleus is required to exert its actions on gene transcription, future work should expand this preliminary assessment and investigate nuclear abundance of NFκB p65 (Donato et al. 2008).