Endothelial Cell–Specific Reactive Oxygen Species Production Increases Susceptibility to Aortic Dissection

Abstract

Background

Increased production of reactive oxygen species (ROS) throughout the vascular wall is a feature of cardiovascular disease states, but therapeutic strategies remain limited by our incomplete understanding of the role and contribution of specific vascular cell ROS to disease pathogenesis. To investigate the specific role of endothelial cell (EC) ROS in the development of structural vascular disease, we generated a mouse model of endothelium-specific Nox2 overexpression and tested the susceptibility to aortic dissection after angiotensin II (Ang II) infusion.

Methods and Results

A specific increase in endothelial ROS production in Nox2 transgenic mice was sufficient to cause Ang II–mediated aortic dissection, which was never observed in wild-type mice. Nox2 transgenic aortas had increased endothelial ROS production, endothelial vascular cell adhesion molecule-1 expression, matrix metalloproteinase activity, and CD45⁺ inflammatory cell infiltration. Conditioned media from Nox2 transgenic ECs induced greater Erk1/2 phosphorylation in vascular smooth muscle cells compared with wild-type controls through secreted cyclophilin A (CypA). Nox2 transgenic ECs (but not vascular smooth muscle cells) and aortas had greater secretion of CypA both at baseline and in response to Ang II stimulation. Knockdown of CypA in ECs abolished the increase in vascular smooth muscle cell Erk1/2 phosphorylation conferred by EC conditioned media, and preincubation with CypA augmented Ang II–induced vascular smooth muscle cell ROS production.

Conclusions

These findings demonstrate a pivotal role for EC-derived ROS in the determination of the susceptibility of the aortic wall to Ang II–mediated aortic dissection. ROS-dependent CypA secretion by ECs is an important signaling mechanism through which EC ROS regulate susceptibility of structural components of the aortic wall to aortic dissection.

Increased vascular reactive oxygen species (ROS) production is implicated in chronic endothelial dysfunction and the development of cardiovascular diseases, including hypertension, atherosclerosis, and aortic dissection.¹ However, large, randomized trials using antioxidants such as vitamin E and beta-carotene have not demonstrated significant reductions in cardiovascular events.^2,3 Lack of specificity may explain the failure of antioxidant therapy as ROS also have important signaling functions in the regulation of homeostasis.⁴ In animal models, knocking out ROS-producing enzymes such as NADPH oxidases reduced basal blood pressure,⁵ aortic plaque formation,⁶ and the development of aortic diseases.⁷ However, because ROS-producing enzymes are expressed in multiple vascular cell types, it is essential to determine the role and contribution of specific vascular cell types to vascular oxidative stress to develop effective cell-specific interventions.

Endothelial cells and vascular smooth muscle cells (VSMCs) are 2 of the main cell types in the vasculature. Much of the previous focus relating ROS to the development of structural disease in the vascular wall has been on VSMC ROS^8–10; however, the role of endothelial cell ROS, a key component of endothelial dysfunction characteristic of vascular disease states, is still largely unknown. This is an important question since endothelial ROS production is increased in vascular disease states, and endothelial dysfunction is predictive of cardiovascular events and long-term clinical outcome.¹¹ It is not clear whether increased endothelial cell ROS is a consequence of vascular disease pathogenesis or is sufficient to independently drive disease pathogenesis. If so, endothelial cell ROS production would be a rational target to identify new interventions for the prevention and treatment of structural vascular diseases.

Major sources of endothelial cell ROS are the NADPH oxidases, comprising of membrane-bound Nox and p22^phox subunits, and regulatory subunits, including p40^phox, p47^phox, p67^phox, and rac1. Of the 7 Nox isoforms identified (Nox1–5, duox1, and duox2), endothelial cells express mainly Nox2 and Nox4. In endothelial cells, Nox2 is an important source of ROS production in response to pathological stimuli such as angiotensin II (Ang II), whereas Nox 4 is more involved in basal ROS production.¹² Nox2 is upregulated in many diseases states such as hypertension and atherosclerosis.^6,13 Knockout of Nox2 reduced atherosclerotic plaque progression⁶ and the hypertensive response to Ang II.⁵ However, none of these previous studies used a cell-targeted approach, so it is unclear whether the increase in endothelial Nox2 and ROS production alone is sufficient to promote vascular disease pathogenesis or is merely a consequence of disease progression.

In this study, we generated a transgenic mouse model with endothelial cell–targeted Nox2 overexpression (Nox2 Tg), mimicking the increase in endothelial Nox2 observed in vascular diseases, to investigate the specific role of increased endothelial ROS in determining susceptibility to vascular diseases. We investigated how this selective increase in Nox2-dependent endothelial cell ROS production would alter the vascular response to Ang II, in the absence of atherosclerosis, to probe the mechanisms linking endothelial cell–specific ROS production to the development and progression of structural vascular disease.

Methods

All animal experiments were conducted in accordance with the UK Home Office Animals (Science Procedures) Act 1986 (HMSO UK). Endothelium-targeted Nox2-overexpressing (Nox2 Tg) mice on C57BL/6 background were generated by our laboratory as described previously.¹⁴ For Ang II or noradrenaline infusion, mice anesthetized with isoflurane received osmotic mini-pumps (Alzet Corp) delivering saline, Ang II at 1 mg⋅kg⁻¹⋅d⁻¹, or noradrenaline at 10 mg⋅kg⁻¹⋅d⁻¹ for 3, 5, 14, or 28 days. Systolic blood pressure was measured with the Visitech tail-cuff system in conscious mice. Primary endothelial cells were isolated from lung, macrophages from peritoneal lavage, and VSMCs from aorta. Terminally anesthetized mice were injected with vascular cell adhesion molecule-1 (VCAM-1)/P-selectin or IgG-conjugated microparticles of iron oxide (MPIO) via the left ventricle as described previously.¹⁵ Ex vivo high-resolution magnetic resonance imaging (MRI) and 3-dimensional reconstruction were performed as described previously.¹⁶ ROS production was quantified with dihydroethidium high-performance liquid chromatography and lucigenin chemiluminescence.

See the online-only Data Supplement for a full description of materials and methods.

Statistics

Data with a normal distribution are expressed as mean±SEM and were tested for significance with the Student t test or ANOVA with the Bonferroni post hoc correction. Nonnormally distributed data are expressed as median with interquartile range and were tested for significance with the Kruskal-Wallis test with the Bonferroni post hoc correction. Comparison of interactions between treatment group, genotype, and treatment length were tested with 2- or 3-factor ANOVA followed by independent t tests corrected for multiple-comparison procedure with the Bonferroni post hoc correction. Nonnormally distributed data were transformed using Y=log(Y) before analysis with 2- or 3-way ANOVA. Categorical data were tested with the Fisher exact test. Values of P<0.05 were considered statistically significant.

Results

Increased Susceptibility to Aortic Dissection in Nox2 Tg Mice in Response to Ang II Infusion

We first tested the endothelial cell–specific expression of the Nox2 transgene in isolated primary endothelial cells, macrophages, and VSMCs from Nox2 Tg and wild-type (Wt) mice (Figure I in the online-only Data Supplement). Endogenous mouse Nox2 was expressed at similar levels in Nox2 Tg and Wt endothelial cells and macrophages, indicating that transgenic overexpression of human Nox2 did not alter the levels of endogenous mouse Nox2. However, human Nox2 transgene expression was detected only in endothelial cells, not in VSMCs or macrophages isolated from Nox2 Tg mice or in any cell types isolated from Wt mice (Figure I in the online-only Data Supplement and Figure 1A). Total Nox2 protein level was elevated in Nox2 Tg endothelial cells compared with Wt, whereas no difference in Nox4 protein was observed. There was no difference in either Nox2 or Nox4 protein level in VSMCs from Wt and Nox2 Tg (Figure 1B).

Figure 1.

Nox2 transgene and protein expression was greater in Nox2 transgenic (Tg) primary endothelial cells (ECs) compared with wild-type (Wt) ECs. A, mRNA expression of human Nox2 (hNox2), mouse Nox2 (mNox2), Nox4, platelet EC adhesion molecule (PECAM; an EC marker), and Myh11 (a vascular smooth muscular cell [VSMC] marker) detected by fluorescence quantitative real-time polymerase chain reaction. Fold change was expressed as ΔCt relative to GAPDH. Data are mean±SEM. ND indicates not detected. B, Western blot showing protein expression of Nox2, Nox4, F4/80 (a macrophage marker), CD102 (an endothelial marker), smooth muscle (SM) α-actin (a VSMC marker), and cyclophilin A (CypA). Mθ indicates primary bone marrow–derived macrophages.

We next evaluated the effects of transgenic Nox2 overexpression on O₂⁻ production in Wt and Nox2 Tg endothelial cells, VSMCs, and aortas. Primary cell O₂⁻ production was measured with 2 independent methods: quantification of 2-hydroxyethidine production with high-performance liquid chromatography¹⁷ and lucigenin chemiluminescence.¹⁸ In Nox2 Tg endothelial cells, O₂⁻ production was increased by 2-fold at baseline compared with Wt endothelial cells (Figure 2A), and this was abolished by the addition of apocynin (Figure IIA in the online-only Data Supplement). Ang II stimulation significantly increased superoxide production in endothelial cells (Figure 2A and Figure IIB in the online-only Data Supplement) and in intact aortas from Nox2 Tg mice compared with Wt, but this difference was not observed in endothelium-denuded aortas (Figure IIC in the online-only Data Supplement). There was no difference in the levels of O₂⁻ production between Wt and Nox2 Tg VSMCs, either at baseline or after Ang II stimulation (Figure 2B).

Figure 2.

Angiotensin II (Ang II)–induced aortic dissection in Nox2 transgenic (Tg) mice. O₂⁻ production as measured by 2-hydroxyethidium (2OH E) quantified with high-performance liquid chromatography at baseline and after Ang II stimulation in (A) primary endothelial cells and B primary vascular smooth muscle cells (VSMCs). *P<0.05; n=8 per group. C, Number of aortic dissections in wild-type (Wt) and Nox2 Tg mice after 28 days of Ang II infusion. *P<0.05; n=11 per group. D, Representative photographs of Wt and dissected Nox2 Tg aortas after Ang II treatment for 28 days. White bar=1 cm. White dotted box indicates the region scanned by magnetic resonance imaging (MRI). E, A 3-dimensional reconstruction (left and middle) using MRIs of dissected Nox2 Tg suprarenal aorta (diaphragm to left renal artery). Pink indicates aorta; red, hematoma; and yellow dot/white arrow, point of dissection. LRA indicates left renal artery; and RRA, right renal artery. Right, Original MRIs. White arrow shows the point of dissection. H indicates hematoma; and L, aortic lumen. F, Transverse (10-μm) sections of suprarenal aorta from Wt and Nox2 Tg mice after 28 days of Ang II treatment stained with hematoxylin and eosin (left) at ×10 magnification or elastin–Van Gieson (right) at ×40 magnification. Black arrow shows the point of dissection. G, Measurement of total number of elastin breaks per length of suprarenal vessel circumference. Data are mean±SEM. *P<0.05; n=6 to 7 per group. H, Measurement of suprarenal vessel media area. Data are mean±SEM. *P<0.05; n=6 to 7 per group. I, Mean systolic blood pressure over 14 days of noradrenaline, Ang II, or saline infusion. Data are mean±SEM. n=3 to 6 mice per group.

Having demonstrated that O₂⁻ production was specifically increased in Nox2 Tg endothelial cells, we next examined the effect of this increased endothelial ROS production on the vascular response to Ang II infusion (1 mg⋅kg⁻¹⋅d⁻¹). Striking features of aortic dissection were observed in 45% (5 of 11) of Nox2 Tg mice, whereas no aortic dissection (0 of 11) was observed in any of the Wt littermates (P<0.05; Figure 2C and 2D). Aortic dissections in Nox2 Tg mice were characterized using ex vivo MRI with 3-dimensional reconstruction (Figure 2E), revealing that the dissection typically tracked along the entire length of the descending aorta, from a point of rupture in the suprarenal segment of aorta between the diaphragm and the renal arteries (Movie I in the online-only Data Supplement). Histological sections of aorta from Nox2 Tg mice demonstrated greater fragmentation of the elastic lamina and significantly reduced medial area compared with Wt littermates (P<0.05; Figure 2F–2H).

We have previously shown that Nox2 Tg mice had increased responses to Ang II at a low (subpressor) dose of 0.4 mg⋅kg⁻¹⋅d⁻¹.¹⁴ To establish whether the increased aortic dissection in Nox2 Tg mice might be due to increased blood pressure, we monitored blood pressure during Ang II infusion (1 mg⋅kg⁻¹⋅d⁻¹). No difference was observed in body weight or heart rate between Wt and Nox2 Tg mice (Table I in the online-only Data Supplement). Both Nox2 Tg and Wt mice responded to Ang II (1 mg⋅kg⁻¹⋅d⁻¹) with an increase in blood pressure. However, there was no difference in the blood pressure response to Ang II between Nox2 Tg and Wt mice either between 1 and 14 days (Figure 2I) or in the sustained effect of Ang II through 28 days (Figure IIIA in the online-only Data Supplement). In separate experiments, noradrenaline (10 mg⋅kg⁻¹⋅d⁻¹) was administered to Nox2 Tg and Wt mice to achieve an increase in systolic blood pressure similar to that observed with Ang II (Figure 2I). In contrast to Ang II infusion, no aortic dissection was observed after noradrenaline infusion (Figure IIIB in the online-only Data Supplement), demonstrating that increased endothelial cell Nox2-dependent ROS production does not drive structural changes in the aortic wall through altered blood pressure responses.

Increased Endothelial VCAM-1 Expression Precedes Ang II–Induced Aortic Dissection in Nox2 Tg Mice

To investigate the relationship between endothelial cell activation and aortic dissection in Nox2 Tg mice, we treated Wt and Nox2 Tg mice with either saline or Ang II (1 mg⋅kg⁻¹⋅d⁻¹) by osmotic mini-pump for 3 or 5 days and examined endothelial expression of VCAM-1 by immunohistochemistry. There was a significant increase in VCAM-1 expression in Nox2 Tg aortas after only 3 days of Ang II treatment compared with Wt and saline-infused controls, with further increases after 5 days (Figure 3).

Figure 3.

Vascular cell adhesion molecule-1 (VCAM-1) expression is significantly increased in Nox2 transgenic (Tg) aortas after angiotensin II (Ang II) treatment. A, Immunofluorescence detection of VCAM-1 (red) and cell nuclei (DAPI, blue) in frozen aortic sections after 3 or 5 days of infusion with either saline or Ang II (1 mg⋅kg⁻¹⋅d⁻¹) presented at ×40 magnification with the lumen (L) facing up. White bar, 20 μm. B, Fluorescence intensity presented as mean area of fluorescence/length of vessel. Data are median±interquartile range. Wt indicates wild-type. *P<0.05; n=6 to 8 per group.

We next evaluated the spatial distribution of VCAM-1 in Nox2 Tg aortas by 3-dimensional MRI using dual-conjugated VCAM-1/P-selectin–targeted MPIO. Nox2 Tg mice were injected with MPIO conjugated with either anti-mouse VCAM-1/P-selectin or IgG after 5 days of Ang II infusion. Specificity of MPIO binding to VCAM-1 was confirmed with a combination of immunohistochemistry and MRI (Figure IV in the online-only Data Supplement). Increased binding of VCAM-1/P-selectin MPIO was observed in the aorta of Nox2 Tg mice between the renal arteries and the diaphragm, consistent with the site of aortic rupture (Figure 4A and 4B). Binding of VCAM-1/P-selectin MPIO was further increased in animals with the development of aortic dissection (Movie II in the online-only Data Supplement), but there was no significant increase in binding of control IgG-MPIO, demonstrating specificity for VCAM-1/P-selectin (Figure 4C). Together, these data indicate that endothelial activation in Nox2 Tg aortas is an early event after Ang II infusion and is spatially localized to the suprarenal segment of aorta, conferring local susceptibility to aortic dissection.

Figure 4.

Distribution of vascular cell adhesion molecule-1 (VCAM-1) expression in suprarenal Nox2 transgenic (Tg) aortas as demonstrated by magnetic resonance (MR) scans and 3-dimensional reconstruction. Mice were injected with VCAM-1/P-selectin– or IgG-conjugated microparticles of iron oxide (MPIO). Black bar, 1 cm. Black dotted box indicates the area of the MR scan; pink, aortic vessel; red, surrounding hematoma; green, MPIO; white arrow and yellow marker, point of dissection. In the next column, the white dotted box shows the area of magnification. Right, Original MR images. Black arrow shows the bound MPIO in (A) dissected aorta from Nox2 Tg mice injected with VCAM-1/P-selectin MPIO, (B) nondissected aorta from Nox2 Tg mice injected with VCAM-1/P-selectin MPIO, and (C) nondissected aorta from Nox2 Tg mice injected with IgG MPIO.

Increased Inflammatory Cell Recruitment and Matrix Metalloproteinase Activity in Nox2 Tg Aortas After Ang II Infusion

To investigate the effect of increased VCAM-1 expression in Nox2 Tg aortas on inflammatory cell recruitment, we evaluated inflammatory cells after 3 or 5 days of saline or Ang II infusion. Using immunofluorescence and confocal microscopy, we observed that CD45⁺ leukocyte recruitment was significantly increased in Nox2 Tg aortas after 5 days of Ang II treatment compared with either Wt or salinetreated controls. In contrast to VCAM-1 expression, there was minimal difference in CD45⁺ leukocyte recruitment after 3 days of Ang II infusion (Figure 5A and 5B), suggesting that leukocyte recruitment occurred after endothelial VCAM-1 upregulation.

Figure 5.

Increased CD45⁺ inflammatory cell recruitment in Nox2 Tg aortas after Ang II infusion. A, Immunofluorescence detection of inflammatory cells (CD45, red) and cell nucleus (DAPI, blue) in frozen aortic sections after 3 or 5 days of infusion with either saline or angiotensin II (Ang II; 1 mg⋅kg⁻⋅d⁻¹) presented at ×40 magnification and with the lumen (L) facing up. White bar=20 μm. B, Quantification of aortic adventitial CD45⁺ cells expressed as cells per adventitial area. Data are median±interquartile range. *P<0.05; n=6 to 8 per group. Number of (C) CD45⁺, (D) CD14⁺/CD11b⁺, and (E) CCR2⁺ cells in digested whole aortas after 5 days of either saline or Ang II infusion. Data are median±interquartile range. Wt indicates wild-type. *P<0.05; n=4 to 8 per group.

Because CD14b⁺ and CD11b⁺ macrophages are thought to play a major role in aortic dissection,¹⁹ we isolated inflammatory cells by enzymatic digestion of aortas after 5 days of in vivo Ang II infusion and quantified macrophages using flow cytometry. There was no difference in CD45⁺ cell numbers at baseline between Wt and Nox2 Tg mice, but the increase of CD45⁺ cells in Nox2 Tg aortas was significantly greater after Ang II treatment compared with Wt (Figure 5C). Furthermore, an increase in CD14/CD11b- and CCR2-positive cells contributed to this increase in CD45⁺ cells in Nox2 Tg aortas (Figure 5D and 5E).

Because matrix metalloproteinases (MMPs) play key roles in vessel remodeling, aortic aneurysm formation, and aortic dissection,^20,21 we next examined MMP activity in aortic sections using in situ zymography (Figure 6). MMP activity was significantly increased in Nox2 Tg aortas after 5 days of Ang II infusion compared with Wt. Specificity was confirmed with the MMP inhibitor phenanthroline (1 mmol/L). Taken together, these observations indicate that a selective and cell-specific increase in endothelial cell ROS production, through Nox2 overexpression, is able to induce aortic dissection by increased endothelial cell activation, leading to inflammatory cell recruitment and increased MMP activity throughout the aortic wall.

Figure 6.

Matrix metalloproteinase (MMP) activity was significantly increased in Nox2 transgenic (Tg) aorta after 5 days of angiotensin II (Ang II) treatment. A, MMP activity was detected using in situ zymography in fresh frozen sections of suprarenal aorta harvested after 3 or 5 days treatment with either saline or Ang II. Images are presented at ×40 magnification with the lumen (L) facing up. Blue indicates DAPI; and bright green, MMP. White bar, 20 μm. Phenanthroline (1 mmol/L) was used as the MMP inhibitor. B, Fluorescence intensity was quantified and presented as mean fluorescence per vessel area. Data are median±interquartile range. ND indicates not detected; and Wt, wild-type. *P<0.05; n=6 to 8 per group.

Primary Nox2 Tg Endothelial Cells Secrete a Factor That Activates VSMCs

To examine the mechanism by which endothelial ROS confers susceptibility of the aortic wall to aortic dissection, we isolated primary endothelial cells and primary VSMCs from Nox2 Tg and Wt mice and tested whether a factor secreted from endothelial cells in response to increased ROS production could mediate changes in VSMCs. We exposed VSMCs to conditioned media from either Wt or Nox2 Tg endothelial cells (without Ang II stimulation) and measured VSMC Erk1/2 phosphorylation as a marker of VSMC activation relevant to human aortic disease.²²

We found that Erk1/2 phosphorylation was significantly increased in VSMCs stimulated with conditioned media from Nox2 Tg endothelial cells compared with conditioned media from Wt endothelial cells or medium alone (Figure 7A), indicating the presence of an ROS-dependent secreted factor from Nox2 Tg endothelial cells. Erk1/2 phosphorylation in VSMCs was associated with an increase in ROS production that was inhibitable by U126 and polyethylene glycol superoxide dismutase (Figure 7B and 7C).

Figure 7.

Endothelial cells (ECs) activate vascular smooth muscle cells (VSMCs) through the secretion of reactive oxygen species (ROS)– dependent cytokine, cyclophilin A (CypA). Nox2 transgenic (Tg) ECs, but not VSMCs, have greater basal secretion of CypA and Erk1/2 phosphorylation. A, Western blot and densimetric analysis of phosphorylated (p) Erk1/2 and total (t) Erk1/2 in cell lysates of VSMCs stimulated with control media, wild-type (Wt) endothelial cell conditioned media, or Nox2 Tg endothelial cell conditioned media. Data are presented as normalized ratio of phosphorylated/total Erk1/2. *P<0.05; n=3 independent experiments. B, Western blot and densimetric analysis of phosphorylated Erk1/2 and total Erk1/2 in cell lysates of VSMCs preincubated with either control media or U126 (10 μmol/L) before stimulation with purified human CypA (50 nmol/L). Data are presented as normalized ratio of phosphorylated/total Erk1/2. *P<0.05; n=3 independent experiments. C, O₂⁻ production in CypA-stimulated primary VSMCs preincubated with control media, U126 (10 μmol/L), or polyethylene glycol superoxide dismutase (100 U/mL) as measured by 2-hydroxyethidium (2OH E) quantified with high-performance liquid chromatography. Data are presented as mean±SEM. *P<0.05 vs control, #P<0.05 vs CypA-treated cells; n=3 independent experiments. D, Western blot and densimetric analysis of primary endothelial-secreted CypA and phosphorylated and total Erk1/2 before and after Ang II stimulation. Data are presented as normalized ratio of phosphorylated/total Erk1/2. *P<0.05; n=4 independent experiments. E, Western blot and densimetric analysis of concentrated (×10) VSMC-secreted CypA and phosphorylated and total Erk1/2 in cell lysates before and after Ang II stimulation. Data are presented as normalized ratio of phosphorylated to total Erk1/2. *P<0.05; n=4 independent experiments.

Cyclophilin A Is the ROS-Dependent Endothelial Secreted Factor That Mediates VSMC Activation

Because cyclophilin A (CypA) may be secreted by VSMCs in the development of aortic dissection and aneurysm formation,²³ we next investigated whether CypA might also play a role as an ROS-dependent secreted factor from endothelial cells. We compared CypA levels in conditioned media from primary Nox2 Tg and Wt endothelial cells using Western blotting. We found that CypA secretion by Nox2 Tg endothelial cells was increased 2-fold at baseline compared with Wt (Figure 7D). This increase in secreted CypA in Nox2 Tg endothelial cells was associated with increased cellular Erk1/2 phosphorylation (Figure 7D). After 24 hours of Ang II stimulation, CypA secretion and Erk1/2 phosphorylation were increased in Wt endothelial cells to levels similar to basal levels in Nox2 Tg endothelial cells (Figure 7D). In contrast to endothelial cells, there was no difference in CypA secretion or Erk1/2 phosphorylation in VSMCs between Wt and Nox2 Tg either at baseline or after Ang II stimulation (Figure 7E). To confirm that the active secreted factor in endothelial cell conditioned media was CypA, we knocked down CypA gene expression in sEND.¹ endothelial cells (Figure 8A). Knockdown of CypA in endothelial cells was sufficient to abolish the ability of endothelial cell conditioned medium to activate Erk1/2 phosphorylation in VSMCs (Figure 8B). Moreover, purified human CypA was able to directly induce VSMC Erk1/2 phosphorylation in both a dose- and time-dependent manner (Figure 8C and 8D). To test the priming effect of CypA, primary VSMCs were stimulated with CypA for 20 minutes as “priming,” followed by exposure to Ang II for an additional 20 minutes. Pretreatment with CypA caused significant augmentation of Ang II–induced VSMC superoxide production as measured by dihydroethidium high-performance liquid chromatography, which demonstrated the priming effect of CypA (Figure 8E).

Figure 8.

Knockdown of cyclophilin A (CypA) abolished endothelial cell conditioned media (EC CM)–induced vascular smooth muscle cell (VSMC) Erk1/2 phosphorylation, and prestimulation with CypA potentiated angiotensin II (Ang II)–induced superoxide production. A, Western blot of CypA knockdown with SiRNA. CL indicates cell lysate; G, GAPDH SiRNA positive control; M, transfection agent only; S, scrambled SiRNA; and UT, untreated. B, Western blot and densimetric analysis of phosphorylated (p) and total (t) Erk1/2 in VSMCs treated with Dulbecco modified Eagle medium control (DMEM), untreated sEND.1 endothelial cell conditioned media, and CypA SiRNA knockdown in sEND.1 endothelial cell conditioned media. Data are presented as ratio of phosphorylated to total Erk1/2. *P<0.05; n=3 independent experiments. C, Representative Western blot and densimetric analysis of phosphorylated Erk1/2 and total Erk1/2 in cell lysates of VSMCs stimulated with purified human CypA at 0, 10, or 50 nmol/L for 20 minutes. Data are presented as the normalized ratio of phosphorylated to total Erk1/2. *P<0.005; n=3 independent experiments. D, Western blot and densimetric analysis of phosphorylated and total Erk1/2 in cell lysates of VSMCs stimulated with purified human CypA (50 nmol/L) for 0, 5, 10, or 20 minutes. Data are presented as ratio of phosphorylated to total Erk1/2. *P<0.01; n=3 independent experiments. E, Superoxide production as measured by 2-hydroxyethidium (2OH E) detection with high-performance liquid chromatography in primary VSMCs preincubated for 20 minutes with control media, CypA (50 nmol/L), or polyethylene glycol superoxide dismutase (100 U/mL) and then stimulated with Ang II for 20 minutes. Data are presented as mean±SEM. Peg-SOD indicates polyethylene glycol superoxide dismutase. *P<0.05; n=3 independent experiments. F, Representative Western blot and densimetric analysis of aorta-secreted CypA (4-hour incubation). Data are presented as median±interquartile range. *P<0.05; n=4 to 8 animals per group.

From these observations from cell culture, we next determined the secretion of CypA from whole aortas from Nox2 Tg or Wt mice after in vivo infusion of Ang II for 5 days. Media from explanted aortas were collected after 4 hours of incubation and analyzed for the presence of CypA by Western blot. Nox2 Tg aortas from animals infused with saline had a 4-fold increase in CypA secretion compared with Wt control aortas (Figure 8F). Ang II treatment increased CypA secretion in both Wt and Nox2 Tg aortas, but CypA secretion remained 3-fold higher in Nox2 Tg compared with Wt. Taken together, these data indicate that increased endothelial ROS production and CypA secretion in Nox2 Tg endothelial cells and aorta activate and prime medial VSMCs through Erk1/2 phosphorylation.

Discussion

In this study, we investigated the effect of a constitutive, endothelial cell–specific increase in ROS production on the response of the vascular wall to Ang II through targeted Nox2 overexpression in the endothelium. We report that increased endothelial cell ROS production alone is sufficient to promote susceptibility to aortic dissection in response to Ang II through increased endothelial VCAM-1 expression, CD45⁺ inflammatory cell recruitment, and MMP activity. Crucially, Nox2 Tg primary endothelial cells, which had greater Nox2 gene and protein expression, increased ROS production, and Erk1/2 phosphorylation, secreted to a greater extent a factor that was able to activate VSMCs through Erk1/2 phosphorylation and augment VSMC ROS production in response to Ang II stimulation. This ROS-dependent endothelium-secreted factor was identified as CypA. No difference in secreted CypA levels was observed in VSMCs, where the Nox2 transgene was not expressed. Collectively, our data identify endothelial cell ROS production and endothelial cell secretion of CypA as critical mechanisms that regulate the vascular inflammatory response to Ang II and the susceptibility to aortic dissection (Figure V in the online-only Data Supplement).

Aortic dissection is a disease with a high mortality,²⁴ so it is crucial to develop strategies that will prevent or reduce the progression of this disease. Recently, an important role for vascular NADPH oxidase–derived ROS has been identified in the development of aortic dissection and aneurysm formation.^7,25,26 However, with multiple Nox homologs expressed simultaneously in different vascular cell types, it is difficult to establish which cell type or which Nox isoforms are crucial in the pathogenesis of aortic dissection. This is important because nonspecific ROS inhibition with vitamins has failed to show any benefit in the prevention of cardiovascular events.^2,27 Two studies have recently shown an increase in Nox2 subunits in human aortic aneurysm tissues.^28,29 The endothelium-specific Nox2 Tg mouse therefore provides a powerful tool to test whether increased endothelial Nox2 expression and the consequent increase in endothelial cell ROS production are sufficient to modify the vascular response to Ang II. The specificity of the endothelial Nox2 transgene was confirmed by quantitative real-time polymerase chain reaction, which demonstrated that the Nox2 transgene was detected only in endothelial cells, not in VSMCs or in macrophages isolated from Nox2 Tg mice. Moreover, we showed an increase at baseline in O₂⁻ production in endothelial cells (but not VSMCs) isolated from Nox2 Tg mice compared with endothelial cells isolated from Wt mice using 2 independent methods. This difference in endothelial ROS production was also observed in aortas, where endothelial denudation completely abolished the difference in O₂⁻ production at baseline¹⁴ and in response to Ang II.

Ang II is a potent activator of vascular inflammation, and its importance in aortic dissection has been highlighted by studies in which Ang II receptor blockers have been shown to slow the progression of aortic dissection in Marfan syndrome in both humans and mice.^30,31 Ang II–induced aortic aneurysm formation has been studied extensively in hyperlipidemic mice such as ApoE^−/− mice.^32–34 In these mice, increased ROS production and vascular inflammation caused by hyperlipidemia led to marked acceleration of atherosclerosis and aortic aneurysm formation in response to Ang II. However, we used Nox2 Tg mice on C57BL/6J background in our study. Treatment with Ang II (1 mg⋅kg⁻¹⋅d⁻¹) in Wt C57BL/6J mice did not lead to the development of aortic dissection. However, in Nox2 Tg mice, the same dose of Ang II induced a 45% incidence of aortic dissection. In previous studies using C57BL/6J, much higher doses of Ang II (1.44–3 mg⋅kg⁻¹⋅d⁻¹)^19,35 have been required to induce aortic dissection in Wt mice. Thus, endothelium-specific Nox2 overexpression is sufficient to markedly increase the susceptibility to aortic dissection, highlighting the role for endothelial ROS in the initiation and development of the disease. We have previously shown that Nox2 Tg mice on ApoE^−/− background have greater macrophage recruitment to the aortic root.³⁶ Interestingly, this susceptibility to aortic dissection was not reported in mice with overexpression of Nox1 in VSMCs despite having similar increases in basal aortic ROS,^10,37 further supporting the hypothesis of a specific role for endothelial ROS in Ang II models of inflammation.

Vascular inflammation and inflammatory cell recruitment are critical steps in the development of aortic dissection. In the ApoE^−/− mice, inflammatory cells were recruited to the aorta after only 48 hours of Ang II infusion.³⁴ Leukocyte adhesion is mediated primarily by endothelial surface expression of adhesion molecules, that is, VCAM-1. Partial knockout of VCAM-1 significantly attenuated the development of inflammatory vascular diseases such as atherosclerosis.³⁸ In this study, we demonstrated by immunohistochemistry that endothelial VCAM-1 expression was significantly increased in Nox2 Tg aortas after only 3 days of Ang II infusion. Increased VCAM-1 expression was followed by significant increases in CD45⁺ cell recruitment and MMP activation after 5 days of Ang II infusion. Furthermore, through immunohistochemistry, MRI, and 3-dimensional reconstruction, we found a gradual increase in the density of VCAM-1 expression toward the rupture site in the suprarenal aorta, which then gradually reduced as the diaphragm was reached. These findings indicate that endothelial activation is an early event of aortic dissection, preceding changes in the VSMCs and inflammatory cell infiltration. Therefore, endothelial activation driven by Nox2-derived ROS production could be the first step in the development of Ang II–induced aortic dissection. This novel finding adds to the conventional understanding that VSMCs are the main initiators of pathways leading to the development of aortic dissection in response to Ang II.^9,19,25,23

From our findings, we hypothesized that endothelial cell activation, governed by endothelial Nox2–derived ROS, determined the degree of aortic vascular inflammation in response to Ang II and the subsequent development of aortic dissection. For this hypothesis to be true, endothelial cells will have to interact and influence the rest of the aorta, including VSMCs. Recently, CypA has been identified as an ROS-sensitive secreted cytokine that plays a critical role in Ang II–induced aortic aneurysm formation.²³ CypA is an 18-kDa protein belonging to a family of highly conserved and ubiquitous proteins called immunophilins. When secreted, it acts as a potent chemoattractant for inflammatory cells and increases VSMC MMP activity. However, previous studies have focused on VSMCs as the main cellular source of vascular CypA.^22,39 We now identify endothelial cells as important sources of vascular CypA that are capable of driving the development of vascular disease because endothelial cell secretion of CypA can modify the overall vascular response to Ang II and lead to striking differences in the susceptibility to aortic dissection. We demonstrated that Nox Tg endothelial cells had greater ROS production and increased CypA secretion. This association between ROS and CypA secretion could also be seen at the level of the whole vessel, where increased basal production of ROS in the Nox2 Tg aorta was associated with increased CypA secretion. Interestingly, CypA secretion from primary endothelial cells was >10-fold higher than that from VSMCs, suggesting that endothelial cells may be the predominant source of CypA in the vessel wall.

The effects of endothelial CypA were apparent through VSMC activation via Erk1/2 phosphorylation. Erk1/2 phosphorylation was used as a marker of VSMCs activation because there is increasing evidence that it could be more relevant than other MAPKs in aortic diseases.^40–42 In keeping with these studies, we found that inhibition of VSMC Erk1/2 phosphorylation abolished the increase in superoxide production in response to CypA. We have also shown in this study that VMSCs stimulated by conditioned media from Nox2 Tg endothelial cells had significantly greater Erk1/2 phosphorylation compared with VSMCs stimulated with Wt endothelial conditioned media. The effect of endothelial CypA on Erk1/2 phosphorylation was striking when CypA was removed from endothelial conditioned media. Conditioned media contains many different secreted factors, and by removing a single factor, CypA, we were able to entirely abolish the effect on Erk1/2 phosphorylation, providing further evidence for the central role of endothelial CypA in VSMC activation. Furthermore, prestimulation of VSMCs with CypA significantly potentiated Ang II–induced O₂⁻ production, which demonstrated the priming effect of CypA on VSMCs in response to Ang II. In our study, Ang II–dependent O₂⁻ was derived predominantly from NADPH oxidase; however, there was also an l-N^G-nitroarginine methyl ester–dependent component, indicating possible cross-talk between the different ROS pathways,^37,43 including uncoupled eNOS.

Conclusions

We have shown for the first time a specific and pivotal role for endothelial ROS in determining the susceptibility to aortic dissection in response to Ang II stimulation. Endothelial cell–specific ROS production and secretion of CypA are signaling mechanisms that are sufficient to induce vascular inflammation, MMP secretion, and structural changes in the vascular wall that lead to dissection. Endothelial cell–specific ROS production is a rational and valid therapeutic target in the prevention of aortic dissection.

Limitations

The findings of this study in a mouse transgenic model have intrinsic limitations. Transgenic overexpression of endothelial Nox2 may not reflect the level of Nox2 in human disease states. However, Nox2 is elevated in disease states, as observed in human aortic aneurysm tissue,^28,29 and the use of the overexpression model provides important insights into determining whether increased Nox2 is causative or consequential. The in vitro studies were done in endothelial cells derived from lungs, not the aorta, although it is likely that under tissue culture conditions the cells would provide a similar model system.

Supplementary Material

The online-only Data Supplement is available with this article at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/CIRCULATIONAHA.113.005062/-/DC1.

Clinical Perspective.

Increased vascular reactive oxygen species (ROS) production is implicated in endothelial dysfunction and the development of cardiovascular diseases. However, it is unclear whether an increase in endothelial cell ROS production alone is sufficient to promote susceptibility to vascular disease. In this study, we generated an endothelial cell–targeted transgenic mouse to model the increase in endothelial Nox2 expression observed in vascular diseases. We found that increased endothelial cell ROS production is sufficient to promote susceptibility to aortic dissection in response to angiotensin II through increased endothelial vascular cell adhesion molecule-1 expression, CD45⁺ inflammatory cell recruitment, and matrix metalloproteinase activity. Crucially, the increase in ROS production in Nox2 transgenic endothelial cells increased the secretion of cyclophilin A, a novel cytokine that primes vascular smooth muscle cells and augments vascular smooth muscle cell ROS production, in response to angiotensin II stimulation. Our findings provide evidence for a specific and pivotal role for endothelial cells in promoting susceptibility to aortic dissection in response to angiotensin II. These data extend our understanding of the pathways leading to aortic dissection, identifying endothelial cell ROS and cyclophilin A secretion as potential therapeutic targets in the prevention of aortic dissection.