Calcium dependent Nox5 NADPH oxidase contributes to vascular oxidative stress in human coronary artery disease

Abstract

Objectives

To examine the expression and activity of the calcium dependent NADPH oxidase in human atherosclerotic coronary arteries.

Background

The Nox based NADPH oxidases are major sources of reactive oxygen species (ROS) in human vessels. Several Nox homologs have been identified but their relative contribution to vascular ROS production in coronary artery disease (CAD) is unclear. Nox5 is a unique homolog in that it is calcium dependent and thus could be activated by vasoconstrictor hormones. Its presence has not yet been studied in human vessels.

Methods

Coronary arteries from patients undergoing cardiac transplant with CAD or without CAD were studied. Nox5 was quantified and visualized using Western blotting, immunofluorescence and quantitative real-time PCR. Calcium dependent NADPH oxidase activity, corresponding greatly to Nox5 activity was measured by electron paramagnetic resonance.

Results

Both western blotting and quantitative real time PCR indicated a marked increase in Nox5 protein and mRNA in CAD vs non CAD vessels. Calcium dependent NADPH driven production of reactive oxygen species in vascular membranes, reflecting Nox5 activity was increased 7 fold in CAD and correlated significantly with Nox5 mRNA levels among subjects. Immunofluorescence demonstrated that Nox5 was expressed in the endothelium in the early lesions and in vascular smooth muscle cells in the advanced in coronary lesions.

Conclusions

These studies identify Nox5 as a novel, calcium dependent source of reactive oxygen species in atherosclerosis.

Introduction

Atherosclerosis and its risk factors increase vascular production of reactive oxygen species (ROS), which in turn contribute to many aspects of atherosclerotic lesion formation (1). A major source of ROS in atherosclerosis is the NADPH oxidases (2). The catalytic subunits of the NADPH oxidases are the Nox proteins, which are involved in transfer of electrons from NADPH to heme groups and ultimately molecular oxygen to form superoxide (O₂^•−) (2). The Nox based NADPH oxidases are major sources of ROS in human vasculature. There are various homologs of Nox proteins, and their expression in atherosclerotic lesions varies depending on location and stages of development (3,4). Particularly, the role of Nox2 and Nox4 and to the lesser extent Nox1 have been described in the context of human atherosclerosis and coronary artery disease (3–5). The last discovered of these is Nox5, which is unique because it contains EF hand domains in the N-terminal region that bind calcium and permit activation of the enzyme by increases in intracellular calcium ([Ca]_i) (6). Nox5 has been identified in immune tissues, testes, cancer cells (6) and recently in human endothelial cells (7). Although Nox5 has been identified in endothelium, it’s activity and contribution to ROS production in native endothelial cells or human blood vessels remains unclear as all experiments were conducted on cells transfected with exogenous Nox5 (7). Thus the potential role of calcium-dependent reactive oxygen species (ROS) production in human coronary artery disease remains unknown. Interestingly, it has been reported that calcium channel antagonists reduce ROS production in human endothelial cells (8). These drugs have been demonstrated to be clinically beneficial in improving clinical outcomes in patients with vascular diseases, particularly coronary artery disease. While there are other potential calcium-dependent sources of ROS in endothelial cells, given a key role of NADPH oxidases, it is possible that Nox5 may contribute to oxidative stress in the setting of vascular disease. The aim of the present study was therefore to determine if expression and function of Nox5 is increased in human atherosclerotic coronary arteries compared to non-atherosclerotic vessels.

Methods

Patients and Blood Vessels

Segments of human coronary arteries were obtained from explanted hearts of 26 patients undergoing heart transplantation for end stage heart failure. Collection of tissues was performed as previously described (3). The Ethics Committee at the University of Cracow approved collection of tissues. Informed consent was obtained from each subject.

Calcium dependent NADPH oxidase activity

Calcium-dependent NADPH oxidase activity was measured from membrane fractions of human coronary arteries using electron paramagnetic spin resonance by measuring production of reactive oxygen species as described previously (9) upon stimulation with 100uM NADPH; in the presence and absence of 1mM Ca⁺⁺. Membrane fractions were prepared by a 2-hour ultracentrifugation of vascular homogenates at 200,000 g. Twenty μg of membrane fractions were suspended in calcium free media A with 1 mmol/L AAP, 5 U/ml HRP, 50 U/ml Cu/Zn-SOD, and 1mmol/L PPH. The reaction mixture placed in 100 μl ESR capillary tube with 100 □M NADPH and accumulation of the PP· nitroxide was measured using a Bruker EMX ESR spectrometer and a super-high Q microwave cavity as previously described (9). All measurements were performed at 25°C using a Bruker Nitrogen Temperature Controller system with the following ESR settings: microwave frequency 9.46 GHz, modulation amplitude 2 G, microwave power 10 dB, conversion time 1.3 s, time constant 5.2 s. Measurements were performed in both Ca⁺⁺ free media and also in media A containing 1 mM Ca⁺⁺. The NADPH-driven ROS production in the presence of Ca⁺⁺ was used to determine total NADPH oxidase activity and the difference between total and Ca⁺⁺ independent oxidase activity was calculated as Ca⁺⁺ dependent NADPH oxidase activity.

Detection of Nox5 protein

Twenty □g of total protein from vascular homogenates was separated by 12% SDS-PAGE and transferred to nitrocellulose membranes. Nox5 protein was detected using a rabbit polyclonal antibody raised against the Nox5 second extracellular loop peptide EASPFQFWELLLTTRPGIG. This is common to all Nox5 splice variants. Bands were detected using chemiluminescence as previously described and analyzed using NIH Image software.

For detection of Nox5 protein in histological sections, immunofluorescence was performed on frozen 7-μm OCT-embedded tissue sections as described previously. Nox5 was visualized using the rabbit polyclonal anti-Nox5 (1:100 dilution). Endothelial cells were identified using a mouse monoclonal anti-CD31 (1:200 dilution). Appropriate secondary antibodies were employed. Sections treated with secondary antibodies alone did not show specific staining. Staining was visualized on a confocal microscope.

Detection of Nox5 mRNA by quantitative real-time RT-PCR

For detection of Nox5 mRNA, segments of coronary arteries were snap-frozen in Tri-reagent and total RNA isolated using the RNA easy kit with DNAse digestion (4). cDNA was synthesized using High Capacity cDNA Reverse Transcription kit (Applied Biosystems) and was subjected to quantitative PCR using the TaqMan ddCT method (Applied Biosystems) and ABI 7500 Fast Real Time PCR system (Applied Biosystems). All primers and probes (Nox5: Hs00225846_m1 gene expression assay as well as an 18S endogenous control) were from Applied Biosystems.

Suppression of Nox5 using small interfering RNA

Nox5 siRNA OnTargetPlus^™ (Dharmacon, US) and OnTarget Plus^™siControl was pre-incubated in Eppendorf-tubes containing 0.572 ml Optimem and 24-μl Oligofectamine per sample. siRNA was added at a final volume of 3 ml to pre-confluent human aortic endothelial cells (HAEC; Cambrex, MA) (final concentration of siRNA- 100 nmol/L). After 4–6 hours of incubation at 37°C and 5% CO₂ 6 ml EBM2 media was added. This procedure was repeated after 48 hours.

Statistical analysis

All data are expressed as means ± SEM with n equal to the number of patients. Comparisons between groups of patients or treatments were made using student t-test or Mann-Whitney U test. Correlation between oxidase activity and Nox5 expression was assessed by Spearman correlation coefficient. Values of p < 0.05 were considered statistically significant.

Results

Characteristics of patients studied

Subjects included in this study had end-stage heart failure and were undergoing heart transplantation. Fourteen individuals had coronary artery disease (CAD) and twelve patients had either dilated (n=11) or hypertrophic cardiomyopathy (n=1), but did not have CAD. The diagnosis of CAD was based on either a history of MI or on coronary angiography. As expected, demographic and clinical characteristics indicated that CAD patients had more risk factors for atherosclerosis than non-CAD patients and were more likely taking statins. Similarly, the presence of CAD was associated with a greater incidence of prior MI, transient ischemic attack (TIA) and peripheral arterial disease, and typical CAD. The degree of left ventricular dysfunction, as reflected by the ejection fraction, was similar between the two groups (Table 1).

Table 1.

Clinical characteristics of studied patients

	No CAD	CAD
N	12	14
Age (mean ± SEM)	46.3±5.6	50.3±2.9
Sex (M:F)	10:2	13:1
Risk factors
Smoking (n; %)	3 (25%)	4 (28%)
Hypertension (n; %)	2 (16%)	12 (86%)*
Diabetes (n; %)	0 (0%)	2 (15%)
Hypercholesterolemia (n; %)	2 (16%)	9 (65%)*
BMI	22.5±1.2	27.2±0.9*
Cholesterol (mmol/L; mean±SEM)	4.9±0.7	5.2±0.6
Fasting glucose (mmol/L; mean ±SEM)	5.0±0.3	5.6±0.2*
MI/CABG (n; %)	0 (0%)	11 (78%)*
PAD, TIA (n; %)	0 (0%)	12 (86%)*
Ejection fraction (%)	17.0 ± 7.2	18.2 ± 2.3
Medications
Aspirin	3 (25%)	4 (28%)
β-blocker	6 (50%)	7 (50%)
HMG-CoA reductase inhibitor	0 (0%)	8 (55%)*
Digoxin	4 (33%)	6 (42%)
ACEI	9 (75%)	10 (70%)
Ca++ channel blockers	0 (0%)	0 (%)

CAD- coronary artery disease; BMI - body mass index; PAD –peripheral arterial disease; TIA – transient ischemic attacks;

^*- indicates p<0.05 vs no CAD group

Calcium-dependent NADPH oxidase activity in human endothelial cells and coronary arteries

As an initial attempt to quantify Nox5 activity and protein levels in vascular segments, we studied human endothelial cells to validate our assay and antibody. Western blots using the rabbit polyclonal antibody showed a single band at a molecular weight of ca. 70 kDa, in keeping with the reported molecular the □ isoform of Nox5. Importantly, siRNA against Nox5 reduced this band by 75%, while scrambled siRNA had no significant effect. Using ESR, we detected NADPH-driven ROS production in membranes of human endothelial cells that was greater in the presence of calcium than in its absence. The difference between these two, i.e. the calcium-dependent NADPH oxidase activity, was reduced by siRNA against Nox5. These data strongly suggest that studies of membranes from human endothelial cells in the presence and absence of calcium can reflect Nox5 activity. We recognize that calcium can activate protein kinase C, which can contribute to activation of Nox2 and perhaps other Nox proteins, by phosphorylation of p47^phox (ref). In addition, calcium can serve to activate ROS from xanthine oxidase and the mitochondria (ref), however these events require the presence of cytoplasmic proteins that were absent in this assay. The fact that siNox5 reduced the calcium-dependent NADPH oxidase activity to approximately the same degree as it reduced Nox5 protein levels suggests that Nox5 is the source of ROS under the conditions of this assay.

As observed in endothelial cells, membrane preparations of all of studied coronary arteries demonstrated NADPH-driven ROS production both in the absence and in the presence of calcium. ROS production in the absence of calcium, reflecting the activity of Nox1, Nox2 and Nox4 was significantly greater in membranes from CAD than in non-CAD subjects (Figure 2A). Interestingly, the increase in calcium-dependent NADPH oxidase activity was significantly more pronounced, averaging 7–8 fold more in vascular membranes from CAD subjects compared to non-CAD subjects (Figure 2A; right panel).

Figure 2. Calcium dependent NADPH oxidase activity and Nox5 expression in coronary artery disease.

Panel A. Calcium independent (left; panel A) and calcium dependent (right; panel A) NADPH oxidase activity in human coronary arteries in relation to the presence of coronary artery disease (CAD). NADPH oxidase activity was measured by ESR as described in methods in membranes isolated from coronary arteries of subjects with (n=8) and without (n=8) CAD. Panel B. Nox5 mRNA expression in coronary arteries from patients with (n=13) and without (n=11) CAD. TaqMan real time PCR analysis was performed using commercially available Gene expression assays. Panel C. Relationship between Ca++ dependent NADPH oxidase activity and Nox5 mRNA expression in studied coronary arteries. Data were expressed as mean+/−SEM. *-p<0.05 vs non CAD; **-p<0.01 vs non CAD.

Nox5 mRNA expression and correlation with calcium-dependent NADPH oxidase activity

Real time PCR demonstrated the presence of Nox5 mRNA in all vascular segments, however its levels were much greater in homogenates of vessels from CAD as compared to non-CAD subjects (Figure 2B). Importantly there was a very significant relationship between these levels of Nox5 mRNA and the calcium-dependent NADPH oxidase activity (Figure 2C). These findings strongly suggest that, as in the case of cultured endothelial cells, in human coronary arteries, Nox5 contributes to calcium-dependent NADPH oxidase activity.

Detection of Nox5 protein in human coronary arteries

To quantify Nox5 protein, we performed Western blots on homogenates of coronary arteries from CAD and non-CAD patients. As in cultured human endothelial cells, western blotting demonstrated a single band at a molecular weight of 70 kDa, compatible with the beta isoform of Nox5, identical size to the band observed in endothelial cells and DU145 cells (Figure 3A). More importantly, this was increased by four fold in CAD compared to non-CAD segments.

Figure 3. Nox5 expression in human coronary arteries in relation to coronary artery disease.

Example Western blots (Panel A.) average data showing Nox5 protein expression in non-CAD (n=7) and CAD (n=7) coronary arteries (Panel B). Lysates of DU 145 prostate cancer cells were used as positive control. Bars represent mean+/− SEM. *-p<0.01.

To confirm this and to attempt to localize Nox5 in atherosclerotic lesions, we performed immunofluorescence staining utilizing the antibody we developed. These studies corroborated that Nox5 in expression was very low in coronary arteries from subjects without CAD (Figure 4A), but that CAD vessels demonstrated specific staining. In the vessels from CAD subjects but in the absence of plaque, Nox 5 expression seemed most prominent in endothelial cells, as evidenced by co-staining with anti-CD31 (Figure 4B and C). As more progressed lesions, with moderate neointima, Nox5 staining was most striking in the neointima although punctate staining could also be observed in the inner aspects of the media (Figure 4D). The most complex lesions demonstrated intense co-staining of Nox5 with smooth muscle underlying advanced plaques (Figure 4E). In no instance did we observe co-localization of Nox5 with T cell marker CD3 or macrophage marker CD68 (data not shown).

Figure 4. Nox5 localization in human coronary artery disease at different stages of atherosclerosis.

Immunofluorecent localization of Nox5 in human coronary arteries. Nox5 (red) was studied in control, non CAD coronary arteries (panel A); in coronary artery segments from CAD patient showing no atherosclerosis (panels B and C), in coronary artery segments showing neointimal hyperplasia (panel D) and in coronary artery segment with severe complex lesion (panel E). Panel C presents a magnification of panel B showing presence of Nox5 in endothelium (double staining; arrows). Green staining represents endothelial cell marker CD31 in panels A–D and smooth muscle cell alpha actin in panel E. Micrographs show representative staining of at least 5 independent experiments.

Discussion

In this report, we show that Nox5 protein and mRNA expression are markedly increased in vessels of patients with CAD compared to non-CAD subjects. This is associated with increased calcium-dependent membrane NADPH oxidase activity, characteristic of Nox5. Indeed, we find that the level of calcium-dependent NADPH oxidase activity correlates significantly with Nox5 mRNA when compared between individual patients. Our immunostaining confirmed the increase in Nox5 protein and shows that its presence seems to vary depending on the stage of atherosclerosis.

Our study is the first to describe the presence of calcium dependent NADPH oxidase activity in human vasculature. We employed an electron spin resonance-based assay to detect calcium-dependent NADPH oxidase activity based on a modification of recently published methodology (9). The conditions of this assay allow detection of hydrogen peroxide by taking advantage of the formation of compound II from horseradish peroxidase, which in turn reacts with AAP, forming a radical that reacts with the nitroxide spin probe. In numerous preliminary experiments, we were unable to detect calcium-dependent superoxide formation from membrane fractions of endothelial cells or human coronary arteries using ESR. This was surprising, because Nox5, like other Nox proteins, contains heme groups that should perform a one-electron reduction of oxygen, and should therefore form superoxide. It is possible however, that Nox5 shares similarity with Nox4, which seems to predominantly release hydrogen peroxide (10). This might be due to positive charges near the heme groups, which retain the negatively charged superoxide until it undergoes spontaneous dismutation to hydrogen peroxide. It is also likely that the membranes we prepared contained extracellular superoxide dismutase, which is known to be present in large amounts in the vasculature (11). This enzyme binds to components of the extracellular matrix, which was almost certainly included in our membrane preparations and could rapidly convert superoxide to hydrogen peroxide, preventing detection of the former. Prior reports have reported that Nox5 produces both hydrogen peroxide and superoxide using fluorescent methods, however these studies used overexpression of the enzyme and did not quantitate the relative amounts of hydrogen peroxide vs. superoxide formed (7,12). Hydrogen peroxide plays an important role in cell signaling and also contributes to atherosclerosis, as transgenic mice that overexpress catalase are protected from lesion development (13). Thus, the formation of hydrogen peroxide by Nox5 could have important implications for the genesis of vascular disease.

Our present observations have implications for mechanisms underlying alterations of vascular function in CAD. Vasoactive agonists increase [Ca]_i both in the endothelial and vascular smooth muscle cells. In endothelial cells, this promotes release of nitric oxide (NO), while in vascular smooth muscle cells, increases in intracellular calcium trigger the contractile apparatus (14). The induction of Nox5 in the endothelium could negate release of NO from the endothelium by permitting the simultaneous production of superoxide and hydrogen peroxide. The former reacts with NO at a diffusion-limited rate, leading to loss of NO and formation of peroxynitrite (15) and these mechanisms are enhanced in human atherosclerosis (16). Hydrogen peroxide can also participate in consumption of NO via reactions with peroxidases (17). The induction of Nox5 in the vascular smooth muscle could also lead to loss of NO as it diffuses into these cells. Moreover, reactive oxygen species such as peroxynitrite can alter function of the sarcoplasmic reticulum, leading to increases in vascular smooth muscle [Ca]_i (18). Moreover, recent studies in Drosophila have shown that the fruit-fly homolog of Nox5 can contribute to smooth muscle cell contraction in response to calcium flux (19). Together, these events would promote vasoconstriction and could alter vascular responses to vasoactive hormones.

The localization of Nox5 is of interest in human vessels. In early lesions, Nox5 seemed in many cases to co-localize with endothelial cells. It is conceivable that the cytokine milieu present in early atherosclerosis could promote Nox5 expression in endothelial cells. Studies of cytokine regulation of Nox5 in endothelial cells would be informative in this regard. In moderately advanced lesions, endothelial staining was less evident, however a large amount of Nox5 co-localized with vascular smooth muscle cells in sub-intimal regions. Complex regions show extensive Nox5 staining in the area of plaque. The presence of Nox5 in early lesions and its loss in advanced lesions is reminiscent of the expression of the endothelial nitric oxide synthase in these settings, as it is present in early lesions and is lost in endothelial cells overlying advanced plaques (20).

In prior studies, xanthine oxidase, Nox4 and Nox2 were found to be expressed in coronary arteries with CAD (3). In the conditions of our assays of isolated vascular membranes, it is unlikely that these enzymes would be activated by addition of calcium. In keeping with this, siRNA against Nox5 markedly inhibited calcium-dependent NADPH oxidase activity, while calcium-independent activity remained unchanged. As evident from figure 2A, the calcium-dependent activity was similar to the calcium-independent activity in CAD vessels, while in non-CAD membranes its contribution was much less. This is in keeping with Nox5 induction in CAD. Thus, calcium-independent NADPH oxidase activity is increased approximately 2.5-fold, while the increase in calcium-dependent activity is 7-fold. This would indicate that the atherosclerotic milieu provides a potent stimulus for Nox5.

Our findings could have implications for therapy in the setting of CAD. Calcium channel antagonists are commonly employed in the treatment of angina and coronary artery disease and in general have been associated with improved symptoms and outcome (21). One of their beneficial effects would be to reduce activation of Nox5 in cells harboring L-type calcium channels, including vascular smooth muscle cells in lesions. Other commonly employed vasodilators, including nitrovasodilators and phosphodiesterase inhibitors also reduce intracellular calcium. In this fashion, these agents might reduce activation of Nox5 in diseased vessels and thus prevent oxidant injury.

Figure 1. Contribution of Nox5 to calcium-dependent NADPH oxidase activity in human endothelial cells.

Panel A. Example ESR spectra of nitroxide adduct formation by membranes prepared from human aortic endothelial cells (HAEC) in Ca⁺⁺ free media and in the presence of 1 mM Ca⁺⁺ (left) and the effects of small interfering RNA (siRNA) against Nox5 on calcium dependent signal (right). Panel B. Effects of siRNA Nox5 on Nox5 protein in HAECs; n=4 experiments; Panel C. Average Ca++ dependent (top) and Ca⁺⁺ independent (bottom) NADPH oxidase activity in the presence of control siRNA (black bars) and Nox5 siRNA (open bars); n=4; values are presented as mean +/−SEM; *-p<0.01 vs control siRNA.