Importance of cholesterol-rich microdomains in the regulation of Nox isoforms and redox signaling in human vascular smooth muscle cells

Vascular smooth muscle cell (VSMC) function is regulated by Nox-derived reactive oxygen species (ROS) and redox-dependent signaling in discrete cellular compartments. Whether cholesterol-rich microdomains (lipid rafts/caveolae) are involved in these processes is unclear. Here we examined the sub-cellular compartmentalization of Nox isoforms in lipid rafts/caveolae and assessed the role of these microdomains in VSMC ROS production and pro-contractile and growth signaling. Intact small arteries and primary VSMCs from humans were studied. Vessels from Cav-1−/− mice were used to test proof of concept. Human VSMCs express Nox1, Nox4, Nox5 and Cav-1. Cell fractionation studies showed that Nox1 and Nox5 but not Nox4, localize in cholesterol-rich fractions in VSMCs. Angiotensin II (Ang II) stimulation induced trafficking into and out of lipid rafts/caveolae for Nox1 and Nox5 respectively. Co-immunoprecipitation studies showed interactions between Cav-1/Nox1 but not Cav-1/Nox5. Lipid raft/caveolae disruptors (methyl-β-cyclodextrin (MCD) and Nystatin) and Ang II stimulation variably increased O2− generation and phosphorylation of MLC20, Ezrin-Radixin-Moesin (ERM) and p53 but not ERK1/2, effects recapitulated in Cav-1 silenced (siRNA) VSMCs. Nox inhibition prevented Ang II-induced phosphorylation of signaling molecules, specifically, ERK1/2 phosphorylation was attenuated by mellitin (Nox5 inhibitor) and Nox5 siRNA, while p53 phosphorylation was inhibited by NoxA1ds (Nox1 inhibitor). Ang II increased oxidation of DJ1, dual anti-oxidant and signaling molecule, through lipid raft/caveolae-dependent processes. Vessels from Cav-1−/− mice exhibited increased O2− generation and phosphorylation of ERM. We identify an important role for lipid rafts/caveolae that act as signaling platforms for Nox1 and Nox5 but not Nox4, in human VSMCs. Disruption of these microdomains promotes oxidative stress and Nox isoform-specific redox signalling important in vascular dysfunction associated with cardiovascular diseases.

Of the Cav isoforms, Cav-1 is the most important for caveolae formation and function in VSMCs as demonstrated in Cav-1 −/− mice where VSMC caveolae are absent and caveolae-associated signalling is altered 6 . In particular cell proliferation is increased, eNOS is upregulated and vascular tone is impaired 7 . Cav-1 contains a scaffolding domain that interacts with several proteins in microdomains and regulates their activation or inhibition 8 . It also influences Ang II/ Ang II type 1 receptor (AT1R) signalling and trafficking, critically important in the regulation of VSMC function and vascular contraction 9,10 .
In addition to acting as a platform for classical signalling pathways, growing evidence indicates that lipid rafts/caveolae play a regulatory role in ROS generation. Cav-1 knock-down or cholesterol depletion by methylβ-cyclodextrin (MCD) in human lung fibroblasts and mouse macrophages leads to increased O 2 − and H 2 O 2 generation 11 . Cav-1 knockdown by siRNA in bovine aortic endothelial cells promotes increased mitochondrial ROS generation and in aortic endothelial cells from Cav-1 null mice H 2 O 2 levels are increased 12 . Cav-1 silencing in VSMCs enhanced both basal and Ang II-induced mitochondrial ROS generation 13 . These phenomena are especially important in VSMCs, where redox-dependent signalling is involved in almost every functional vascular response. Molecular mechanisms regulating vascular redox signalling are complex because different types of ROS influence different downstream pathways. For example, O 2 − is important in pro-contractile 14,15 and pro-inflammaory signalling 16 , whereas H 2 O 2 promotes vasodilation 17,18 . Processes controlling these differential responses may relate to sub-cellular compartmentalization of Noxs, the oxidases responsible for ROS generation in the vascular system.
Other factors that modulate vascular ROS bioavailabity in VSMCs are antioxidant systems such as superoxide dismutase (SOD), catalase and peroxidases. In addition, PARK7 (Parkinson's disease protein-7), also called DJ1, acts as an antioxidant and a signaling molecule. In its oxidized state, it activates nuclear factor erythroid-related factor 2 (Nrf2), a master regulator of anti-oxidant transcription factors 25,26 . DJ1 localizes in the cytoplasm and in conditions of oxidative stress translocates to the mitochondria and nucleus [27][28][29] . It has also been shown to associate with Cav-1 in rat astrocytes and ventricular cardiomyocytes [30][31][32] . However it is unknown whether DJ1 associates with Noxs and Ang II signaling in VSMCs.
To better understand subcellular mechanisms that coordinate redox signaling we questioned the role of cholesterol-rich microdomains in Nox isoform regulation and ROS production and sought to evaluate whether these processes influence downstream redox-sensitive targets and pro-contractile and proliferative signaling in human VSMCs and vessels.

Results
Expression of Nox isoforms and localization with Cav-1 in human small arteries. Cellular expression of Nox isoforms and co-localization with Cav-1 in intact human small arteries was assessed using double-labelling immunofluorescence. As shown in Fig. 1, Nox isoforms and Cav-1 are present in human arteries. Cav-1 localized mainly in the endothelium (intima) and vascular smooth muscle (media) layer. Nox5 was expressed both in the endothelium and vascular media, with greater abundance in the vascular smooth muscle layer. Nox4 was expressed both in the endothelium and vascular smooth muscle layer. As shown in Fig. 1A,B, there is a partial colocalization of Nox5 and Cav-1 as indicated by the orange colour. Nox4 does not seem to localize with Cav-1. We could not detect Nox1 in whole tissue in an isoform-specific manner by immunofluorescence despite multiple antibodies. Nox 1 and Nox5, but not Nox4, associate with cholesterol-rich microdomains in human VSMCs. To evaluate in greater detail the sub-cellular localization of vascular Nox isoforms in cholesterolrich microdomains and the impact of Ang II stimulation, we studied human VSMCs that were fractionated into cholesterol-rich and cholesterol-poor fractions using a detergent-free sucrose gradient centrifugation method. As shown in Fig. 2A, Nox5 was expressed in fractions 3-4, corresponding to cholesterol-rich fractions and which likely comprise lipid rafts/caveolae as confirmed by abundant Cav-1 expression. Nox5 was also present in the high density non-lipid raft fractions (fractions 7-12), possibly reflecting cytoplasmic localization. In subsequent studies, we combined lipid rich (fractions 3, 4) and non lipid-rich (fractions 7-12) fractions and showed that while Nox1 and Nox5 are expressed in both lipid-raft and high density non-lipid rafts fractions, abundance of Nox1 and Nox5 was greater in the high density versus the the low density fractions (Fig. 2B-D). Nox4 was abundantly expressed in high density non-lipid fractions and absent in low density fractions, suggesting that Nox4, while expressed in VSMCs, does not localize in cholesterol-rich microdomains (Fig. 2E). These findings support those of others who also failed to show Nox4 in lipid rafts in VSMCs and endothelial cells 3, 11 . Ang II stimulates lipid raft/caveoale trafficking of Nox1 and Nox5 but not Nox1. As shown in Fig. 2B,C, Ang II stimulation resulted in rapid translocation of Nox5 (within 5 min) out of the lipid-raft fractions into the high density non-lipid raft fractions. On the other hand, Ang II stimulation induced Nox1 trafficking into lipid raft fractions (Fig. 2D). Ang II did not have any effect on Nox4 lipid-raft trafficking.
Cholesterol-rich microdomains negatively regulate Nox-derived superoxide production in human VSMC. To 4A). Ang II-stimulated increase in H 2 O 2 levels was reduced only by the Nox4 inhibitor GKT137831 (Fig. 4B). These data further support our findings that Nox1 and Nox5, which localise in cholesterol-rich domains, play an important role in Ang II-stimulated O 2 − production, whereas Nox4, localised mainly in the cytoplasmic milieu, regulates H 2 O 2 generation in response to Ang II.
To specifically investigate the effects of Nox5 on Ang II-induced ROS production and redox signaling, cells were transfected with Nox5 siRNA (Supplementary figure 1A). In basal conditions, Nox5 silencing did not significantly alter O 2 − levels ( Fig. 7A), but increased H 2 O 2 levels (Fig. 7B, Nox5 siRNA 63.3 ± 27.69% increase vs. Control, p < 0.05) compared to control siRNA. In Nox5 downregulated cells, Ang II failed to increase O 2 − and had no effect on H 2 O 2 production.

Ang II influences redox regulation of DJ1 in human VSMCs.
To investigate whether other redoxsensitive targets important in vascular cell function are influenced by the lipid rafts/caveolae-Nox system, we investgated effects on DJ1, a dual antioxidant and signaling molecule, that we found to be abundantly expressed in human small arteries (Supplementary figure 2) and which interacts with Cav-1 and Nox1 (Supplementary figure 3). As shown in Supplementary figures 3A-2C, DJ1 physically associated with Nox1 but not with Nox5. Interactions between DJ1 and Nox1 were also observed by immunofluorescence, where Nox1 and DJ1 co-localized in VSMCs as shown in Supplemental figure 2D by the yellow fluorescence in merged images.
Having demonstrated that human vessels and VSMCs abundantly express DJ1, we questioned whether Ang II regulates redox-sensitive (oxidized) DJ1 through prcesses involving cholesterol-rich microdomains. Human VSMCs were probed with specific antibodies against total and irreversibly oxidised DJ1 (Cys 106-SO 2 H; Cys 106-SO 3 H). DJ1 expression was examined in isolated lipid-rafts and non-lipid raft fractions of VSMCs stimulated  www.nature.com/scientificreports/ with or without Ang II (10 −7 M) for 5 min. As shown in Fig. 9A, the irreversibly oxidised form of DJ1 was only present in non-lipid raft fractions even though total DJ1 was present in lipid-rafts. These findings indicate that either non-oxidised or only reversibly oxidised DJ1 is present in rafts. Ang II increased irreversible oxidation of DJ1 within 5 min (Ang II 5′: 125.8 ± 3.870% vs. Veh, Ang II 15′: 134.8 ± 10.17% vs. Veh; p < 0.005) as shown in Fig. 9B. Lipid-raft disruption by MCD or nystatin induced an increase of irreversible oxidation of DJ1, a trend that did not reach statistical significance. The irreversibly oxidized form of DJ1 represents increased activation of DJ1 and identifies a novel downstream signaling target of Ang II.

Cav-1 silencing increases ROS production and redox signaling in hVSMC.
To further investigate the relationship between lipid rafts/caveolae, ROS and Ang II-induced signaling, we investigated effects of Cav-1 silencing (with siRNA; Supplementary figure 4) on ROS generation and Ang II-mediated redox signaling in hVSMC. As shown in Fig. 10A, in basal conditions, Cav-1 silencing resulted in an increase in NADPH-dependent O 2 − levels compared to control siRNA (Cav-1siRNA 108.06 ± 33.98% increase vs. Control, p < 0.05). A similar response was observed for H 2 O 2 (Fig. 10B), since cells transfected with Cav-1 siRNA showed higher H 2 O 2 levels compared to control (Cav-1siRNA 165.3 ± 22.39% increase vs. Control, p < 0.05). Ang II did not cause a further increase in NADPH-dependent O 2 − or H 2 O 2 levels in Cav-1 siRNA-treated cells. In addition to impacting ROS production, Cav-1 silencing influenced signaling in hVSMCs as shown in Fig. 10C-F. Cav-1 downregulation resulted in an increase in basal levels of phospho-Ezrin/Radixin/Moesin and phospho-p53 but not ERK1/2. In Ang II-stimulated cells, phosphorylation of ERK1/2, but not Ezrin/Radixin/ Moesin or phospho-p53, was increased.
ROS generation and redox-sensitive signalling are increased in arteries from caveolin-1 −/− mice. To test proof of concept from our human studies, we investigated whether our in vitro findings in , p53 (C) and ERK1/2 (D) were analysed by western blot. β-Actin was used as loading control. Bar graphs are means ± SEM from 5-7 experiments. Control was taken as 100% and data are presented as the percentage changes relative to control conditions. *P < 0.05 vs. Ctl. Ang II angiotensin II, Ctl control, MCD methyl-b-cyclodextrin, Nys nystatin, Chol cholesterol, p-MLC20 phosphorylated MLC20, p-ERM phosphorylated Ezrin-Radixin-Moesin, p-p53 phosphorylated p53, p-ERK1/2 phosphorylated ERK1/2. www.nature.com/scientificreports/ human VSMCs are recapitulated in intact arteries in vivo. We studied isolated arteries from wild-type, and Cav-1 −/− mice and probed for Nox-dependent O 2 − production and activation of redox-sensitive signalling pathways, ERK1/2 and Ezrin-Radixin-moisin. As shown in Fig. 11A, mesenteric arteries isolated from Cav-1 −/− mice had significantly higher levels of NADPH-dependent O 2 − levels compared to wild-type mice (Cav-1 −/− , 443.02 ± 131.70% increase vs. Control, p < 0.05). Aortas isolated from Cav-1 −/− mice had significantly greater expression of phospho-Ezrin/Radixin/Moesin compared with control mice while ERK1/2 phosphorylation was not significantly altered between groups (Fig. 11B). These findings suggest that in the absence of regulated caveolae (Cav-1 −/− ), Nox-dependent ROS production and activation of signalling pathways associated with VSMC actin cytoskeletal organisation are increased, indicating an important role for intact lipid rafts/caveolae (Cav-1) in the regulation of Nox activity, ROS generation and vascular signalling, findings recapitulated in human VSMCs.

Discussion
Vascular signalling is mediated in large part through Nox-derived ROS and activation of redox-sensitive pathways that regulate VSMC function. Sub-cellular mechanisms underlying these processes have not been fully elucidated but compartmentalization in cholesterol-rich microdomains may be important. This is especially relevant for Ang II since the AT 1 R and associated signalling molecules localize in lipid rafts/caveolae. In the present study, using a multidisciplinary approach, we identify these microdomains as an important structural element in Nox-ROS regulation and redox-dependent signalling in human VSMCs and show that disruption of these microdomains promotes oxidative stress and aberrant vascular signalling. Our findings demonstrate that these processes are highly regulated and Nox isoform-specific since Nox1 and Nox5, but not Nox4, localize in lipid rafts/caveolae, which when disrupted lead to increased Nox-derived ROS production and hyperactivation of signalling pathways important in VSMC function including, MLC20, Ezrin-Radixin-Moesin and p53 involved in contraction, cytoskeletal organization and apoptosis/cell cycle control respectively. In addition, we identified DJ1 as a redox-sensitive downstream target regulated by Ang II in a Nox-and lipid raft/caveolae-dependent manner. To recapitulate our findings in human VSMCs in an intact system, we studied mice deficient in Cav-1 and caveolae and demonstrated that Nox-derived ROS and vascular signalling are exaggerated. Together our findings identify an important role for cholesterol-rich microdomains that act as regulatory elements for Nox1 and Nox5 and redox-sensitive signalling platforms in VSMCs (Fig. 12). Loss of integrity of these microdomains promotes oxidative stress and altered signaling, important in vascular dysfunction associated with pathological processes.  assay (B). Representaive images of phospho-Ezrin-Radixin-Moesin, phospho-p53, phospho-ERK1/2 and Nox5 detected by western blot. α-tubulin and total ERK1/2 were used as loading control (C). Protein quantification of Ezrin-Radixin-Moesin (D), p53 (E) and ERK1/2 (F) phosphorylation in hVSMCs. Data are means ± SEM from 4-7 experiments. Control was taken as 100% and data are presented as the percentage changes relative to control conditions. *P < 0.05 vs. control, + P < 0.05 vs. AngII Ctl siRNA. Ang II angiotensin II, Ctl control, H 2 O 2 hydrogen peroxide, RLU relative light units, p-MLC20 phosphorylated MLC20, p-ERM phosphorylated Ezrin-Radixin-Moesin, p-p53 phosphorylated p53, p-ERK1/2 phosphorylated ERK1/2. www.nature.com/scientificreports/ Figure 8. p22phox silencing effect on ROS production and Ang II-induced signaling in hVSMC. Cells were transfected with p22phox or control (Ctl) siRNA and stimulated with Ang II (100 nmol/L) for 5 min. NADPHderived O 2 − generation was measured by lucigenin assay (A). H 2 O 2 levels were measured by the Amplex Red assay (B). Representaive images of phospho-Ezrin-Radixin-Moesin, phospho-p53, phospho-ERK1/2 and p22phox detected by western blot. α-Tubulin and total ERK1/2 were used as loading control (C). Protein quantification of Ezrin-Radixin-Moesin (D), p53 (E) and ERK1/2 (F) phosphorylation in hVSMCs. Data are means ± SEM from 4-7 experiments. Control was taken as 100% and data are presented as the percentage changes relative to control conditions. *P < 0.05 vs. control, + P < 0.05 vs. AngII Ctl siRNA. Ang II angiotensin II, Ctl control, H 2 O 2 hydrogen peroxide, RLU relative light units, p-MLC20 phosphorylated MLC20, p-ERM phosphorylated Ezrin-Radixin-Moesin, p-p53 phosphorylated p53, p-ERK1/2 phosphorylated ERK1/2. www.nature.com/scientificreports/ Noxs are widely recognised as key sources of ROS in vascular cells that play a crucial role in the progression of cardiovascular diseases 36,37 . Despite advances in redox research, molecular mechanisms regulating Nox activity and their compartmentalization in the vascular system, especially in human VSMCs, is still poorly defined. This is particular relevant for Nox5, where progress has been slow due to the lack of research tools and experimental models, as NOX5 gene is not expressed in rodents, the major experimental model used to study cardiovascular Nox biology. Although it has been proposed that Nox and ROS compartmentalize in different cellular organelles, including mitochondria and endoplasmic reticulum 38,39 , where distinct redox signaling is tightly controlled, the role of cholesterol-rich microdomains remains unclear, particularly in the human context. Previous studies using qualitative approaches showed Nox1 in caveolae in human VSMCs 3 . In lung fibroblasts and heterologous expression systems in COS-7 expressing Cav-1/Nox5, Cav-1/Nox2, Cav-1/Nox4 11 , Nox2 and Nox5, but not Nox4, were found to associate with lipid-rafts through direct interactions with Cav-1. Using multiple approaches   14,15,17,18 . Moreover processes of Nox regulation may be important because Nox1 and Nox5, localised in cholesterol-rich domains, are activated by vasoactive agents, whereas Nox4, which does not associate with lipid rafts/caveolae, is constitutively active.

Figure 12.
Schematic showing putative interactions between lipid rafts/caveolae, Nox isoforms, DJ-1 and vascular signaling in conditions where lipid rafts/caveolae are intact and when they are discrupted. (i) When cholesterol-rich microdomains are intact, likely in physiological conditions, Nox1 and Nox5 are regulated and ROS production is controlled. Nox4 which is constitutively active, localizes in the cytoplasm where it generates mainly H 2 O 2 , which may be vasoprotective. (ii) During lipid-raft disruption, which may occur in pathological conditions or when the Ang II system is upregulated, Nox1 and Nox5 are activated leading to excessive ROS production and oxidative stress. This promotes increased redox-dependent signaling through multiple downstream pathways that influence VSMC function. P phosphorylation, Ox oxidation, MLC20 myosin light chain, ERM Ezrin-Radixin-Moesin. www.nature.com/scientificreports/ Ang II signaling involves Nox-derived ROS as we previously demonstrated 1,37 . Considering the importance of lipid rafts/caveolae in Nox1/5-ROS regulation in VSMCs we questioned whether Ang II influences lipid raft/ caveolae trafficking of these Noxs. Within 5 min of Ang II stimulation, Nox1 shuffled from non-lipid raft fractions to lipid-rich fractions, while Nox5 moved in the opposite direction, indicating that Ang II regulates Nox trafficking in a highly organized and isoform-specific manner. The significance of this may relate to the differential functions of Nox1 and Nox5. Nox1 is important in vascular inflammation and expression/activation of pro-inflammatory adhesion molecules [44][45][46] , which associate with the cell membrane and lipid rafts/caveolae, while Nox5 is involved in VSMC contraction, proliferation and cytoskeletal organization 2, 23 . To dissect out distinct roles for Nox1 and Nox5 in human VSMCs we used a multidisciplinary approach including various pharmacological inhibitors (NoxA1ds, GKT137831 and mellitin) and siRNA targeted to Nox5 and p22phox-dependent Noxs, which includes Nox1-Nox4. Our findings suggest that Noxs regulate multiple signaling pathways and that Nox5 seems to be especially important in Ang II-stimulated ERK1/2 signaling whereas Nox1 is involved in Ang II-induced activation of p53. These phenomena may contribute, at least in part, to the diverse vascular actions of Ang II. It should be highlighted that we assessed acute signaling events and that different Nox-dependent responses may occur with chronicstimulation.
Lipid-rafts/caveolae seem to play an important regulatory role in maintaining basal ROS production, because disruption of these microdomains or depletion of cholesterol, caused excessive ROS generation, processes that lead to oxidative stress, cell injury and vascular dysfunction. This is exemplified by our findings that in human VSMCs exposed to MCD and nystatin, Nox-derived ROS generation was increased, findings that were recapitulated in Cav-1-silenced human VSMCs and vessels from Cav-1 −/− mice. Similar responses have been observed in MCD-treated mouse macrophages, macrophages from Cav-1 −/− mice and in human lung fibroblasts where Cav-1 was knocked down by siRNA 11 . These results together with our findings support the notion that cholesterol-rich microdomains are negative regulators of Nox-derived O 2 − generation. While we can not distinguish exactly which Nox isoform is involved in this process, it is likely to be Nox1 and/or Nox5, but not Nox4 because Nox4 does not seem to be abundant in lipid-rafts and Nox4 generates primarily H 2 O 2 .
Vascular signaling through lipid rafts/caveolae is highly selective. For example, phosphorylation of MLC20, Ezrin/Radixin/Moesin and p53, but not ERK1/2, was variably increased. ERK1/2 activation seems to be independent of cholesterol microdomains in human VSMCs. This was corroborated in VSMCs in which Cav-1 was downregulated, since phosphorylation of Ezrin/Radixin/Moesin and p53, but not ERK1/2, was increased. These in vitro results are in agreement with intact vessels isolated from Cav-1 −/− mice where phosphorylation levels of ERK1/2 were similar to vessels from wild-type mice. The importance of Cav-1 in ERK1/2 activation remains controversial, as previous studies showed that Cav-1 might have an inhibitory or stimulatory role depending on the cell type. In rat aortic VSMCs, cardiac fibroblasts, NIH 3T3 fibroblast and Rat-1 cells, regulation of ERK1/2 activity is Cav-1-dependent [47][48][49][50] , whereas in bovine aortic endothelial cells and in mouse embryonic fibroblasts it seems to be Cav-1-independent 7, 51 . On the other hand, signalling pathways associated with cell contraction and cytoskeletal organisation in human VSMCs seem to be controlled by microdomain integrity, because lipidraft disruption induced phosphorylation of MLC20 in human VSMCs, a response that is Nox5-dependent as we previously demonstrated 23 . The importance of lipid rafts/caveolae in the regulation of vascular contraction has been demonstrated in rat tail artery rings and rat cremaster arterioles 52,53 and confirmed in aortic rings from Cav-1 −/− mice, which exhibited reduced contractile responses to vasoconstrictive agents including Ang II, endothelin-1 and phorbol ester 6 .
Cholesterol-rich microdomains interact with the cell cytoskeleton and are intricately involved in the organisation of the actin and its associated structural proteins 54 . Cholesterol depletion promotes F-actin polymerisation and stress fibre formation in tumour cells 55 , mesenchymal and epithelial cells while in fibroblasts MCD induces actin disassembly and reduction of stress fibres 56 . In our study we showed that cholesterol-rich microdomains influence activation of the actin cytoskeleton-associated proteins, Ezrin/Radixin/Moesin, which act as crosslinkers between plasma membranes and actin filaments. Phosphorylation of these proteins is necessary for its binding to the F-actin cytoskeleton 57,58 . We found that lipid-raft disruption and Cav-1 downregulation in human VSMCs was associated with increased phosphorylation of Ezrin/Radixin/Moesin indicating the importance of intact microdomains for cytoskeleton organization. These findings were further confirmed in intact vessels from Cav-1 −/− mice.
Another signaling system regulated by lipid rafts/caveoale involves p53, a master transcriptional factor controlling genes involved in cell cycle arrest, senescence or apoptosis. Phosphorylation of p53 at Ser15 and Ser20 disrupts the interaction between p53 and its negative regulator MDM2, leading to the accumulation and activation of p53 in response to cellular stress-induced DNA damage 59,60 . We found that Ang II increased phosphorylation of p53 (Ser 15) in human VSMCs, similar to what has been shown in neonatal rat cardiomyocytes, H9c2 cells 61 and rat aortic cells 62,63 . We also found that MCD increased phosphorylation of p53 (Ser 15) supporting a role for cholesterol-rich fractions in p53 regulation, similar to what was demonstrated in chicken myogenic cells 64,65 . To further support this notion, ROS-induced activation of p53 in fibroblasts from Cav-1 −/− mice was altered 66,67 and transient overexpression of Cav-1 in mouse NIH 3T3 fibroblasts or stable transgenic Cav-1 expression in fibroblasts caused cell cycle arrest and senescence 68 .
In addition to identifying a regulatory role for microdomains in Nox-ROS-redox signalling, we found that DJ1 is a caveolar-resident protein that interacts with Cav-1 and Nox1 in human VSMCs. DJ1 is increasingly being recognized as an important redox-sensitive molecule involved in vascular regulation [66][67][68] . DJ1 acts as a ROS scavenger and also as a signaling protein that activates transcriptional factors, such as nuclear factor erythroid 2-related factor 2 (Nrf-2) 26,69 . The anti-oxidant function of DJ1 depends on its oxidation state of Cys-106, which is oxidised by ROS from sulfenic acid (Cys-SOH) to sulfinic acid (Cys-SO 2 H) to sulfonic acid (Cys-SO 3 H). Having observed that DJ1 is abundantly expressed in human arteries and VSMCs, we questioned whether DJ1 may be another redox-sensitive target regulated by Ang II through Nox-lipid raft/caveolae mechanisms. We found that www.nature.com/scientificreports/ Ang II rapidly increased irreversible oxidation of DJ1 (Cys 106-SO 2 H; Cys 106-SO 3 H), but only in the non-lipid raft fraction. To our knowledge these are the first studies showing that Ang II regulates DJ1 in VSMCs through processes involving rafts and oxidation. The functional significance of this awaits clarification. Similar to many other studies 70-74 , we used a pharmacological approach to disrupt lipid rafts/caveolae in human VSMCs. However this approach does have limitations because MCD has been shown to disrupt cholesterol-rich domains beyond the plasma membrane 70,71 . In particular MCD influences membranes of subcellular organelles, including mitochondria, and accordingly we can not exclude the potential role of other sources, such as mitochondria, for ROS in our experimental paradigm. To mitigate some of these limitatons, we used a second disrupter, nystatin to interrogate lipid rafts. Moreover we studied VSMCs in which Cav-1 was downregulated using siRNA approaches.
In conclusion, we demonstrate that Nox1 and Nox5 localize and traffic through cholesterol-rich microdomains, which act as negative regulators for Nox-induced ROS generation and redox signaling in human VSMCs. Loss of integrity of cholesterol-rich microdomains promotes oxidative stress and alters signaling important in vascular dysfunction associated with cardiovascular disease. Our findings suggest that lipid rafts/caveolae are discrete sub-cellular compartments involved in Nox1 and Nox5-derived ROS generation and signaling in human VMSCs. This regulatory system is Nox isoform-specific because Nox4-derived H 2 O 2 seems to be independent of lipid rafts/caveolae.

Methods
Detailed methodology is provided in the Online Data Supplement.
Human vascular tissue and primary human vascular smooth muscle cell culture. All studies related to accessing small arteries and vascular smooth muscle cells (VSMCs) from humans were approved by the West of Scotland Research Ethics Service (WS/12/0294). Written informed consent was received from all study participants in accordance with the Declaration of Helsinki (1997). Human small arteries were obtained from patients undergoing elective craniofacial surgeries (n = 12) at the Queen Elizabeth University Hospital, Glasgow. A small piece of vascular tissue was fixed in 4% paraformaldehyde (PFA) overnight and used for immunofluorescence. The remaining tissue was used to isolate VMSCs by enzymatic digestion, as previously described 39,75 . Experiments were performed on low-passage cells (passage 4-6).

Cholesterol depletion/sequestration and cholesterol reloading.
To disrupt caveolae in VSMCs, free cholesterol was depleted or sequestrated from the plasma membrane by using two different agents, MCD or Nys, respectively as previously described 74 . Cholesterol was depleted by treating VSMCs with mM MCD for 45 min at 37 °C. Cholesterol was sequestrated by treating VSMCs with 50 μg/mL Nys for 30 min. In some experiments, after MCD treatment, cholesterol was reloaded with cholesterol:MCD (1-10 mmol/L) complex.
Immunoblotting. Total or fractionated proteins from VSMCs were separated by electrophoresis on a polyacrylamide gel, transferred onto a nitrocellulose membrane and probed with primary antibodies. Horseradish peroxidase-conjugated secondary antibodies were visualised by Azure c300 Western Blot Chemiluminescent Blot Imaging System and fluorescent-conjugated secondary antibodies were visualised by infrared laser scanner. Bands were quantified densitometrically with either ImageJ (https ://image j.nih.gov/ij/) or Image Studio Lite software from LI-COR.
Immunocytochemistry. Human VSMCs were cultured on sterile glass coverslips, fixed in ice-cold 100% methanol for 5 min, blocked with 3% Bovine Serum Albumin (BSA) and incubated with primary antibodies overnight at 4 °C. Proteins were detected with Alexa Fluor secondary antibodies and slides were mounted in ProLong Gold anti-fade mounting media containing DAPI (Life technologies) overnight at RT. Fluorescence imaging was recorded in an Axiovert 200M microscope with a laser scanning module LSM 510 (Carl Zeiss AG, Heidelberg, Germany).
Hydrogen peroxide measurement-Amplex Red Assay. Hydrogen peroxide production in VSMC lysate was measured using the horseradish persoxidase-linked Amplex Red™ Hydrogen Peroxide/Peroxidase Assay Kit according to manufacturer's instructions (Life Technologies, Paisley, UK).
Lucigenin-enhanced chemiluminescence. The lucigenin-enhanced chemiluminescence assay was used to assess NAD(P)H-dependent superoxide anion (O 2 − ) production in VSMC homogenates as previously described 39 . Statistical analysis. Statistical analysis was performed using GraphPad Prism 8.0 (GraphPad Software Inc, San Diego, CA, USA). All data are expressed as mean ± SEM. Statistical comparisons were made with one-way ANOVA followed by Newman-Keuls test or 2-tailed Student's t test as appropriate. P < 0.05 was considered statistically significant.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.