Background
The burden of atherosclerosis, which contributes to strokes, heart attacks, and other cardiovascular diseases, remains tremendous despite significant advances in disease prevention and treatment [
1]. Despite an improved knowledge of atherosclerosis development and therapeutics capable of combatting the disease, mainly statins, the prevalence of atherosclerosis-derived events remains high. Therefore, improved therapeutics that more efficiently and widely prevent atherosclerosis development are desired.
Atherosclerosis disease mechanisms are multifactorial. Blood vessel geometry is one important factor, as indicated by the development of atherosclerosis in a vascular site-specific manner [
2,
3]. Most often, atherosclerotic plaques appear in vessel regions of high curvature or branches, such as the carotid bifurcation and aortic arch [
2‐
5]. The complex geometry in these regions is known to create disturbed flow (DF) profiles [
2‐
5]. In the carotid bifurcation the profile is characterized by low time-averaged shear stress magnitudes, high shear stress gradients, and occasional flow reversal [
2‐
5]. Atherosclerotic plaques almost never localize in long, straight arteries, such as the abdominal aorta, which exhibit uniform flow (UF) that is unidirectional with constant and relatively high time-averaged shear stress magnitudes [
2,
3]. These observations, taken together, are strong indications of the relationship between local hemodynamics and atherosclerotic lesion progression [
5‐
7].
Another widely recognized factor that contributes to atherosclerosis is pathological change in the vascular endothelium, which precedes disease development [
8‐
10]. Alterations in endothelium function, which can actually vary quite distinctly from vasculature of different origins [
11], include deactivation of anti-atherogenic signaling pathways and activation of pro-atherogenic signaling, impaired vasoregulation, transition from an anti- to a pro-inflammatory state, which includes increased expression of adhesion molecules, and elevated permissiveness to small molecules and whole cells crossing the vascular wall [
12‐
14]. Studies conducted over the last three or more decades have confirmed hemodynamic force, most interestingly shear stress, as an important regulator of endothelium behavior [
15]. Both in vitro and in vivo studies have demonstrated that UF characteristic of atheroprotective regions of the vasculature promotes endothelial homeostasis, while atheropermissive DF induces endothelial dysfunction.
Mechanosensitivity of the vascular endothelium, both of macro- and microvasculature [
16], and subsequent changes in phenotype have been attributed to a number of mechanotransduction pathways. Caveolae, for example, are well-known mechanosignaling invaginations of the cell membrane structures, that are rich in cholesterol, sphingolipids, and a variety of signaling molecules [
17‐
19]. To our knowledge, there is no prior information on atheropermissive DF regulation of caveolae and caveolin-1 (cav-1), the protein that coats caveolae. However, atheroprotective UF, in comparison to static flow conditions, has been shown to increase expression of caveolae and cav-1, while also increasing caveolae/cav-1 polarization to the portion of endothelial cells (ECs) that lies upstream of the flow direction [
20,
21]. In healthy conditions, caveolae formation at the EC membrane, caveolae ability to receive extracellular stimuli and induce intracellular responses, and, subsequently, diverse caveolae-derived signaling cascades, make the membrane invaginations essential signaling domains that assist in maintaining endothelial homeostasis. Improper expression of caveolae can therefore lead to endothelial dysfunction relevant to atherosclerosis development.
For example, in one study, loss of caveolae in genetically modified mice was found to impair endothelium-dependent regulation of vascular relaxation, contractility, and myogenic tone [
22]. Other studies have implicated caveolae in regulating the expression of the anti-atherogenic vascular relaxation factor, nitric oxide (NO), which is enzymatically produced by endothelial-type NO synthase (eNOS) [
23]. Lipid modification of eNOS, by myristoylation and palmitoylation, can target the enzyme to caveolae [
19,
24] where it is activated through Ca
2+ exposure or through Ca
2+ independent phosphorylation at Serine 1177 [
25‐
27]. Activated eNOS is then enabled to produce and release NO, upon uncoupling from the cav-1 scaffold protein [
28‐
31]. Prior studies have collectively demonstrated that UF conditions can sustain this anti-atherogenic, caveolae-derived signaling cascade [
32‐
35]. While eNOS and NO expression are known to be suppressed in atheroprone regions of the vasculature [
33‐
36], the direct connection between DF and the signaling cascade that starts with caveolae and cav-1 expression and ends with eNOS activation and NO release remains to be clarified.
Protective or atherogenic fluid mechanotransduction regulation of caveolae/cav-1 expression and mechanosignaling function may involve the EC surface glycocalyx (GCX) [
37]. GCX is a polysaccharide mesh layer on the outer surface of the endothelial membrane. The GCX consists of sialic acid and glycosaminoglycans, such as heparan sulfate (HS), hyaluronic acid, and chondroitin sulfate, some of which bind to the syndecan and glypican membrane-attached proteoglycan core proteins [
38,
39]. HS is the most abundant GCX glycosaminoglycan and, for the present study, HS is of particular interest because it binds to the glypican-1 core protein that, along with eNOS, resides in mechanosignaling caveolae [
40,
41]. Due to its unique location at the interface between the EC surface and the extracellular environment, GCX has been implicated in sensing and converting physiological fluid shear stress and mechanical stretch into biochemical signals to support healthy endothelial phenotypes [
41‐
43]. Most interestingly, GCX degradation has been linked to endothelial-dependent atherosclerosis and a number of other diseases while a healthy GCX has been shown to promote proper endothelium function [
44‐
47]. For example, evidence from in vivo and in vitro studies confirm that UF-conditioned GCX is abundant [
37,
48‐
50] and its HS component, via increased glypican-1 in caveolae, is responsible for eNOS activation and flow-induced NO production [
51‐
56]. In contrast, atheroprone DF conditions have been shown to be less supportive of GCX expression [
37,
48‐
50]. This decreased GCX expression has also been associated with endothelial motility and proliferation [
57], increased vascular permeability [
58], and leukocyte adhesion [
59]. However, downstream in the mechanotransduction pathway, GCX-mediated DF regulation of caveolae and its eNOS/NO relevant mechanosignaling function has never been studied, to our knowledge.
The goal of the present study is to link well-defined physiologically accurate flow conditions to GCX-mediated flow-regulation of caveolae/cav-1 and activated eNOS, which are relevant to atherosclerosis prevention or onset. We hypothesized that DF patterns will lead to decreased cav-l expression, loss of cav-1 cellular compartment-specific localization, and decreased cav-1 colocalization with activated eNOS, all due to suppressed GCX. Our results may reveal a mechanism by which GCX integrity contributes to atherosclerosis prevention while GCX damage permits disease progression, suggesting that the GCX may be a viable target for next-generation atherosclerosis prevention and treatment approaches.
Materials and methods
All materials were purchased from Fisher Scientific, MA, USA unless otherwise noted.
Cell culture
Rat fat pad ECs (RFPECs) were previously isolated from rat epididymal fat pad, immortalized, and shared with us by Dr. Mia Thi of the Albert Einstein College of Medicine (Bronx, NY, USA). RFPECs were cultured at passages 20 to 60 in Dulbecco’s Modified Eagle Medium (DMEM) with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin (P/S). Cells were grown at 37 °C in a humidified incubator maintained with 5% CO2. For shear stress experiments (described below), cells were plated on fibronectin-coated (3 μg/mL) glass coverslips (#1.5 or 0.17 mm thickness) at a seeding density of approximately 3 × 103 cells/cm2. Cells were cultured in static conditions for 2 to 4 days until fully confluent. At confluence, cells were transferred to flow conditions in GCX-stabilizing medium containing reduced FBS (5%) along with bovine serum albumin (BSA; 0.5%).
It is important to note here that RFPECs are ideal for this study, because unlike many other EC types RFPECs express a robust GCX [
60,
61]. Furthermore, similar to other EC types, RFPECs properly respond to the application of shear stress [
62,
63] and have been shown to elicit atherosclerosis relevant GCX function [
64‐
67], F-actin expression, morphology, and alignment [
62,
68], intercellular communication [
62,
69], endothelium permeability [
70,
71], and NO production [
72]. Acknowledging that ECs are heterogenous due to their origins from different species and different vascular beds [
73], we wish to indicate that we are conducting parallel studies with primary endothelium originating from other species, including human, and from a variety of vascular beds, including the aorta and coronary arteries. The additional investigations, taken together with this one, will provide comprehensive information, are far beyond the scope of this report, and will be described in other companion publications.
Shear stress apparatus
The confluent RFPECs were exposed to shear stress in a custom, parallel-plate flow chamber that was designed using SolidWorks and SolidWorks Flow Simulation. The chamber was designed to simulate atherosclerotic-prone and atherosclerotic-resistant flow patterns in adjacent neighboring spatial regions of the carotid artery (Additional file
1: Fig. S1a). Briefly, the device was an enclosed rectangular chamber with a height that was much smaller than the width and length. The chamber consisted of a top, a bottom, a rubber gasket to form a seal and shape the flow channel, and the coverslip coated with ECs on fibronectin (Additional file
1: Fig. S1b). To assemble the apparatus, the coverslip was placed on the chamber bottom, and the chamber top was then screwed to the chamber bottom with the rubber gasket in between. The flow apparatus was then connected to a peristaltic pump. Cells were exposed to 6-h of flowing DMEM containing 5% FBS, 1% P/S, and 0.5% BSA. Generally, the flow pattern was uniform (UF), as associated with atherosclerosis resistance. However, for our study, UF was interrupted by a protuberance at the bottom of the chamber to produce pro-atherosclerotic DF. This flow is characterized by separation followed by recirculation, stagnation, and eventual recovery to UF. During flow perfusion, the shear stress apparatus was maintained in an incubator at 37 °C with 5% CO
2 and humidity, an environment that the flow apparatus can withstand due to its fabrication out of 316 L stainless steel. The stainless steel is also biocompatible, which protects the ECs and DMEM from toxicity. An open media reservoir was placed upstream of the pump, to ensure proper CO
2 diffusion into the flow system’s DMEM. We also inserted a pulse dampener directly after the pump and upstream of the chamber inlet, to eliminate pulsatility generated by the peristaltic pump. Further details about the shear stress apparatus design and validation are elaborated upon later in this report.
Glycocalyx treatment
The RFPEC GCX was generally treated with the aforementioned flow conditions—DF versus UF—as generated by the flow apparatus. The RFPEC GCX was also treated by static flow, which served as a control condition. To confirm the role of the GCX in any observed RFPEC response to flow, in some experiments the GCX was also chemically treated with the enzyme heparinase III (Hep III; Ibex, Quebec, Canada). Hep III exclusively cleaves HS glycosaminoglycans, a major GCX component, thus impairing the GCX layer. RFPECs in both static and UF conditions were exposed to 62.5 micro-international units per milliliter (µIU/mL) of Hep III in DMEM for 6 h, allowing for simultaneous flow application and enzyme treatment. The Hep III concentration was determined via dose–response pilot experiments to degrade RFPEC HS by ~ 50% (data shown in Additional file
2: Fig. S2). For further clarification of when and why flow and Hep III were used to treat the GCX, refer to Table
1.
Table 1
GCX flow and/or chemical treatment with rationale, where applicable
Flow stimulus | Static | Yes | Yes | To assess whether GCX controls baseline EC behavior (as opposed to flow-induced EC behavior) |
Uniform | Yes | Yes | To assess whether EC behavior in UF only occurs due to endothelium coverage by the GCX |
Disturbed | Yes | | |
Rationale | To assess GCX expression and EC behavior in response to both pro- and anti-atherosclerotic flow patterns | | |
Animal studies
All animal studies were carried out using procedures approved by the Northeastern University Institutional Animal Care and Use Committee (IACUC). Male C57Bl/6 mice were obtained from Jackson Laboratories, fed a chow diet, and studied at 6–8 weeks of age.
In one experiment, acute DF was induced in vivo, by surgically performing a partial ligation of the left carotid artery (LCA), as previously described [
74]. This allowed for the investigation of DF effects on EC behavior in vivo. Briefly, mice were anesthetized with isoflurane and their LCAs were exposed through a 4–5 mm ventral midline incision followed by blunt dissection. Three of the four caudal branches of the LCA, including the left external carotid, internal carotid, and occipital artery, were ligated with 6–0 silk sutures, leaving the fourth branch (superior thyroid artery) intact (Fig.
4a). The incision was then closed. The mice were briefly monitored to confirm a successful recovery, and kept alive for a week to allow in vivo DF to induce endothelial dysfunction [
74]. To investigate UF effects on EC behavior in vivo, surgery was not performed and LCA was left non-ligated. One week after surgery, ligated and non-ligated LCAs were isolated from mice that were initially anesthetized using isoflurane and followed by xylazine (20 mg/mL) and ketamine (100 mg/mL). Mice vessels were then drained of blood by perfusion with a 1% BSA solution in phosphate buffered saline (PBS), administered via the left ventricle after severing the inferior vena cava. Following this, mice were perfusion fixed with a 2% paraformaldehyde solution in PBS. The ligated and non-ligated LCAs were isolated after removal of periadventitial fat. LCAs were then embedded in Tissue-Tek optimum cutting temperature (OCT) medium, frozen at − 80 °C, and 6-µm axial sections were made using a Leica CM3050 cryostat.
In another experiment, effects of chronic in vivo DF were assessed in aortas of chow diet fed 6–8-week-old male C57Bl/6 mice. As described by previous reports, ECs along the aortic arch and descending aorta, specifically along the inner curvature, are exposed to low shear stresses and DF that can predispose the vessel to atherosclerosis [
75,
76]. In contrast, the proximal abdominal aorta experiences fully developed UF, leading to atherosclerosis resistance. To extract mouse aortas, mice were processed as described above. Aortas were cleared of perivascular fat, dissected, kept in PBS at 4 °C for a maximum of 2 days, fluorescently labeled as described below, and longitudinally cut using curved 8-cm vannas scissors and a dissection microscope for
en face imaging [
77].
Fluorescent staining of caveolin-1, eNOS, and glycocalyx in vitro
For in vitro GCX staining immediately after exposure to static or flow conditions, RFPECs with untreated or Hep III-treated GCX were fixed using 2% paraformaldehyde and 0.1% glutaraldehyde in PBS for 30 min (PBS). For a general label of the GCX, RFPECs were stained using wheat germ agglutinin (WGA), which targets GCX constituents that include sialic acid and
N-acetylglucosamine, a component of HS and hyaluronic acid [
38,
78]. To accomplish WGA staining, RFPEC monolayers were blocked with 3% BSA for 30 min at room temperature and then incubated in biotinylated WGA (1:100, Vector Labs, CA, USA) for 1 h at 4 °C. For secondary labeling, Alexa Fluor 488-conjugated streptavidin was used at a concentration of 1:1000 for 30 min at 4 °C. To specifically label the HS component of the GCX, RFPECs were blocked in 2% goat serum for 30 min at room temperature and incubated in mouse anti-HS (1:100, 10E4 epitope, Amsbio, MA, USA) overnight at 4 °C. Secondary labeling was achieved by incubating cells with goat anti-mouse conjugated with Alexa Fluor 488 (1:1000) for 30 min at room temperature.
For in vitro cav-1 and eNOS staining, RFPEC monolayers were fixed with 4% paraformaldehyde in PBS for 20 min. RFPECs were then permeabilized using a 0.2% Triton X-100 solution for 5 min and blocked in 10% goat serum for 30 min at room temperature. RFPECs were co-incubated with rabbit anti-eNOS specific to eNOS phosphorylated on the 1177 serine residue (eNOS-pS1177) (1:500, Signalway Antibody, MD) and mouse anti-cav-1 (1:200, Santa Cruz Biotechnology, TX) overnight at 4 °C. This was followed by secondary incubation with goat anti-rabbit labelled with Alexa Fluor 647 (1:200, cav-1) and goat anti-mouse tagged with Alexa Fluor 488 (1:200, eNOS-pS1177), respectively, for 1 h at room temperature.
All fixed and stained RFPECs were then coverslipped using VectaShield anti-fade mounting medium with 4′,6-diamidino-2-phenylindole (DAPI) for nuclear staining (VWR, PA, USA).
Fluorescent staining of caveolin-1, eNOS, and glycocalyx in vivo
For in vivo staining of 6-µm cryosectioned LCAs, samples were post-fixed for 10 min with 4% PFA and permeabilized with 0.3% Triton X-100 for 10 min. Antigen retrieval was subsequently performed in 1.2 mM sodium citrate by pressure cooking for 10 min. To block endogenous peroxidase for future tyramide signal amplification, samples were incubated with 1% H2O2 for 10 min. Samples were then blocked using 10% goat serum for 1 h and incubated with either biotinylated WGA or rabbit anti-cav-1 (1:50) overnight at 4 °C. For fluorescent detection, samples were then labeled with horseradish peroxidase (HRP)-conjugated streptavidin or HRP-conjugated goat anti-rabbit (1:500) for 1 h at 4 °C. Tyramide signal amplification using the cyanine 3 (Cy3) fluorophore was then performed using a PerkinElmer Cy3 TSA kit following the manufacturers protocol. Samples were then mounted in VectaShield mounting medium with DAPI.
Mouse aortas were labeled for WGA, cav-1, and total eNOS using a protocol similar to the in vitro fluorescent staining procedure. For WGA, in 24-well plates, 3–4 mm aortic segments were blocked with 3% BSA, incubated in biotinylated WGA for 1 h at 4 °C, and then incubated in Alexa Fluor 488-conjugated streptavidin for 30 min at room temperature. For cav-1 and total eNOS, samples were blocked with 5% goat serum for 1 h at room temperature, separately incubated with either rabbit anti-cav-1 (1:300, Thermo Fisher Scientific, MA) or rabbit anti-total eNOS (1:200, Cell Signaling Technology, MA) overnight at 4 °C, and then incubated with Alexa Fluor 488 conjugated goat anti-rabbit secondary at 1:500 (cav-1) or 1:1000 (total eNOS) for 1 h at room temperature. All steps at room temperature were performed on a rocker. Aortic segments were placed in VectaShield mounting medium containing DAPI and then longitudinally cut using a dissection microscope and mounted on glass coverslips as previously described [
77].
Confocal microscopy
All RFPECs, LCA cryosections, and en face aorta vessel sections were imaged using a Zeiss LSM 710 laser scanning confocal microscope at 40× or 63× magnification (with an oil immersion lens) with 8-bit pixel values for each color channel. Gain and laser power levels were set for each experiment to provide adequate signal levels below saturation point. Microscope settings remained the same throughout data collection for each experiment. Images of XY-plane slices were collected and then stacked to create a Z-projection image.
Data acquisition
All data acquisition was performed using ImageJ, unless otherwise specified.
First, to determine overall expression levels of cav-1 and eNOS in RFPECs, ImageJ was used to calculate the mean fluorescence intensity (MFI) of the respective immunofluorescence signals from en face images of the Z-projections. Over a specified field of view, MFI provides a measure of the average pixel value, ranging from 0 to 255, for each color channel. In all cases, a field of view was taken as an entire 63× magnification en face image.
It was previously noted that cav-1 localization correlates to direction of pressure drop and corresponding flow [
20]. Therefore, cav-1 distribution with respect to flow direction was examined, as previously described [
20]. In the studies performed in vitro, randomly selected RFPECs were identified to have preferential cav-1 localization in one of four zones of relatively equal size. These zones were defined as upstream (towards the direction of flow), downstream (away from the direction of flow), upper lateral region, and lower lateral zones. An additional, fifth “zone” was also considered, in which cav-1 was evenly dispersed.
To quantify cav-1 localization at cell membranes, where the protein can contribute to the formation of caveolae rafts, kurtosis analyses were performed in the in vitro studies of RFPECs. The kurtosis analysis provides a coefficient indicative of the degree to which immunofluorescence data is centralized to a specific point. Therefore, performing kurtosis analyses between two neighboring ECs provided us with a measure of the density of cav-1 immunofluorescence at the cell-to-cell appositions in comparison to the intracellular compartments of ECs. Kurtosis coefficient values for various experimental conditions were normalized to the kurtosis values of the static control condition. Normalized kurtosis values below 1.0 reflected decentralization of cav-1 from the cell membranes, while values above 1.0 reflected increased cav-1 centralization to cell membranes.
Colocalization of caveolae-associated signaling molecules, represented by activated eNOS, to caveolae was assessed after eliminating background noise in Z-projection
en face images of RFPECs. To eliminate background noise, a signal intensity threshold was determined by measuring fluorescence intensity within the field of view at positions above the cell surface. Pixels with signal intensities at or below the threshold were removed. With background noise subtracted, colocalization was determined using the Colocalization Threshold tool on ImageJ. The colocalization values were then analyzed using the Mander’s overlap coefficient approach to quantify the extent to which activated eNOS overlaps cav-1. Mander’s method was employed in lieu of other colocalization analysis techniques, such as the Pearson’s method, because of the localization of eNOS in multiple areas of the cell and the large variation in the ratio of fluorescent probe signal intensity [
79].
It was of interest to determine if GCX correlated to and was involved in cav-1 function. Therefore, overall levels of GCX expression in RFPECs were determined using ImageJ by calculating MFI of GCX immunofluorescence in the same manner that was used for determining eNOS and cav-1 expression. In addition to examining GCX expression, GCX thickness was also assessed. Measurements of GCX thickness were performed from cross-sectional views of the confocal z-projection images. On the cross-section view, randomized areas of the cell membrane were chosen using the “Grid” tool. Lines at the predetermined randomized positions were then drawn perpendicular to the apical cell membrane, which contained immunofluorescence signal representing the GCX (Additional file
3: Fig. S3). The fluorescent profile along the line was then plotted (Additional file
3: Fig. S3). On the fluorescence profile plots, the width of fluorescence signal above background signal was identified and taken to indicate GCX thicknesses (Additional file
3: Fig. S3).
Confocal microscope fluorescence images of LCAs and aortas stained for cav-1, total eNOS, and WGA-labeled GCX were qualitatively assessed for the most part. Quantitative analysis of LCA and aorta images, although limited, were performed in accordance with RFPEC analysis as described above.
Statistical analysis
Prior to statistical analyses, ImageJ data points were normalized with respect to static samples within each experiment. Normalization within experiments eliminated any bias caused by microscopic and other variations between experiments. Therefore, experimental data acquired from ImageJ quantification are presented in the “
Results” section as normalized means ± standard error of the means (SEMs). The total number of experiments, along with counts of the RFPEC monolayer replicates and sizes of data sets within each experiment, are also presented in the “
Results” section. These experimental data were statistically analyzed using GraphPad Prism software. Upon confirming uniform distributions of the data sets collected from static flow, DF, UF, static flow with Hep III treatment, or UF with Hep III treatment conditions, one-way ANOVA analyses were performed. In addition, post hoc Tukey’s multiple comparison tests were completed with an alpha value of 0.05, unless otherwise specified.
Discussion
Our goal was to link well-defined physiologically accurate UF and DF conditions to EC phenotype that is atherosclerosis-resistant or atherosclerosis-permissive, respectively, as mediated by the GCX mechanotransducer. In the end, we confirmed that in the presence of DF patterns and shear stress levels that are characteristic of atheroprone areas of the vasculature, GCX expression is weak and is a key factor underlying inhibition of EC cav-1 expression and eNOS activity, which were taken as indicators of pro-atherosclerotic EC phenotype.
Specifically, we found that in vitro exposure of ECs to anti-atherosclerotic UF led to increased expression of the principal caveolae marker, cav-1, preference for cav-1 localization to the upstream zone of the cellular body, and cav-1 distribution to EC-to-EC appositions. The response of cav-1 to atheroprotective UF, compared to static conditions, is somewhat established. Rizzo et al. [
21] previously found that exposure of bovine aortic ECs to UF increased cav-1 by fivefold, compared to static controls. Sun et al. [
20] showed a significant increase in cav-1 localization at the upstream side of cell bodies in UF compared to static conditions. Our observed significant UF-induced increases in cav-1 expression (Fig.
1a, c, f) and in cav-1 localization at the upstream side of cell bodies (Fig.
1j) agree with the previous findings [
20,
21]. To our knowledge, while decreased expression of cav-1 has been observed in human atherosclerotic plaques and is associated with atheroma formation [
86‐
88], there has been no direct evidence of how cav-1 is affected by DF, which is viewed as an initiator of endothelial dysfunction in the onset of atherosclerosis. Here, we found that exposure of RFPECs to DF, compared to UF, results in decreased cav-1 expression (Fig.
1b, c, f). Furthermore, using our novel atherosclerosis-relevant parallel plate flow chamber, we observed de-localization of cav-1 from RFPEC borders and the upstream side of cell bodies compared to UF (Fig.
1g–f). These findings were corroborated by in vivo data that demonstrated decreased cav-1 expression in ligated LCAs affected by acute DF (Fig.
4b) and the inner curvature of the distal aortic arch affected by chronic DF (Fig.
4g) when compared to UF conditioned controls that include the non-ligated LCA and proximal abdominal aorta, respectively (Fig.
4c, h). Taken together with previous studies indicating a vasculoprotective role of cav-1 and caveolae [
22,
82‐
84], our in vitro and in vivo findings suggest that atheroprone flow can lead to endothelial dysfunction through decreased cav-1/caveolae expression and localization.
To determine the functional consequences of differential UF versus DF regulation of cav-1, we considered that endothelial caveolae have been implicated in a variety of cellular signaling processes including NO signaling [
22,
82‐
84,
89,
90]. Therefore, we investigated the effects of DF and UF on eNOS-pS1177, which is an active form of eNOS that resides in caveolae and synthesizes and releases NO to promote endothelial health. In agreement with previous findings, we found eNOS-pS1177 levels and eNOS-pS1177 colocalization with cav-1 to be significantly elevated in regions of UF [
21]. Our studies confirmed our expectation that in DF conditions, where cav-1 expression is disrupted, eNOS-pS1177 expression and colocalization with cav-1 are reduced. In comparison to UF, eNOS-pS1177 expression in regions of DF decreased near twofold. Our observation of eNOS-pS1177 expression in RFPECs and total eNOS in mouse vessels agree with previous studies [
35,
74,
91,
92]. For example, Won et al. [
35] found that the application of 72 h of UF significantly increased eNOS-pS1177 expression compared to DF. While the study performed by Won et al. exposed cells to an extended time period compared to our study, it has been shown that flow-induced changes in NO can occur within several minutes of flow application [
53,
93]. We exposed RFPECs to a 6-h period and found that changes in eNOS-pS1177 expression occur within the 6-h time frame. We also observed a similar upward trend in expression of total eNOS in UF-conditioned mouse abdominal aortas compared to DF conditioned mouse aortic arches (Fig.
4i, j). More specifically, we found a 28-fold decrease in eNOS expression in areas of DF (aortic arch/descending aorta) than in regions of UF (proximal abdominal aorta).
Rizzo et al. [
21] previously found that the application of UF resulted in increased residence of eNOS-pS1177 within caveolar compartments of ECs. However, this study did not probe caveolae in regions of DF. Here, we expanded on the findings of Rizzo et al., demonstrating that colocalization of cav-1 with eNOS-pS1177 is significantly higher in cells exposed to UF when compared to regions of DF (Fig.
2). Due to its role in eNOS activation, the observed increase in cav-1 overlap with eNOS-pS1177 suggests that decreases in eNOS-pS1177 in DF are at least partially a result of the flow regulation of cav-1-eNOS colocalization. Decreased expression of cav-1 in areas of DF may also further exacerbate this phenomenon. In regards to these findings, it is important to note that cav-1 has been shown to inhibit eNOS activity [
28‐
31]. Despite this, in the present study we do not expect direct, molecular interaction between the two molecules in vitro because such studies investigated the active, phosphorylated form of eNOS, which does not bind to cav-1 [
94]. Instead, colocalization of eNOS-pS1177 with cav-1 is an indicator of its presence within caveolae, which agrees with previous studies. However, we cannot speculate on the in vivo interaction between cav-1 and eNOS as we did not perform
en face co-staining and used an antibody targeting total eNOS. Additionally, while this study only addressed the impact of cav-1 deregulation on NO production, it is possible that DF-induced cav-1 deregulation impacts additional cav-1 signaling pathways. Nevertheless, the collective data from this study indicates that the loss of caveolae/eNOS-pS1177 colocalization and subsequent decreased eNOS activation in areas of DF may further increase the risks for atherosclerosis development.
In this study, immunoprecipitation and proximity ligation assays were not performed due to our interest in identifying eNOS localization within caveolae without direct interaction between eNOS and cav-1, which is believed to inhibit eNOS activity [
28‐
31]. Isolation of endothelial plasma membranes and subsequent isolation of the caveolae fraction of the plasma membrane to then detect eNOS within the caveolae, a technique established by Rizzo et al. [
95], were also not performed. Localization of eNOS within caveolae was evaluated using immunofluorescence in place of these other methods because previous immunofluorescence findings were corroborated using transmission electron microscopy (TEM) and western blotting of subfractionated luminal plasma membrane [
21]. In addition, other similar correlative studies using immunofluorescence, Western blotting, immunoprecipitation, and TEM validated colocalization of caveolae with additional cellular components including dynamin [
96] and the transforming growth factor-β (TGF-β) type I receptor [also known as activin receptor-like kinase (ALK)] [
97]. These studies, all taken together, suggest that localization of eNOS in caveolae can be accurately determined using the immunofluorescence technique.
In addition to studying cav-1 expression and colocalization with eNOS, as an indicator of EC phenotype, we sought to also identify an upstream mechanism of flow-mediated cav-1 expression and function. Previous research has shown that the endothelial GCX is weakly expressed in areas of atherosclerosis [
5‐
7], implicating the endothelial GCX in pro-atherosclerotic EC dysfunction, for example, by deregulating vasculoprotective molecules such as NO [
51,
53,
55]. Previously, Koo et al. [
49] used a cone-and-plate flow apparatus to study GCX regulation after exposure to 24 h of UF or DF. They found that GCX expression is increased in regions of UF but decreased in areas of DF. However, their study isolated cells in DF from those in UF, preventing communication between the two cell phenotypes as is experienced in vivo. This current study uses a more physiologically relevant parallel-plate flow chamber with juxtaposed regions of DF and UF, allowing for proper EC communication. From our RFPEC studies we also demonstrated that changes in GCX expression can occur in as little as 6 h of flow exposure, suggesting that the GCX may help regulate rapidly changing endothelial processes. In the present study, we specifically investigated the effects of pro-atherosclerotic flow on the sialic acid, hyaluronic acid, and HS components of the GCX, labeled using WGA lectin and HS antibody. The specific GCX components of interest were chosen due to previous research implicating them in NO regulation [
52,
54,
56,
98]. We found that RFPEC exposure to DF resulted in statistically significant decreases in GCX coverage and thickness (Fig.
3). Similarly, mouse LCAs and aortas demonstrated similar decreases in GCX expression in the presence of DF (Fig.
4). Furthermore, we observed that the extent of GCX decrease (as well as cav-1 and eNOS differences) was spatially variable when we compared the proximal versus distal aortic arch (data not presented), indicating the DF vascular regions are areas of transitional endothelial phenotype.
By applying Hep III enzyme to HS, a major component of the GCX, we were able to ascertain whether decreased GCX expression in DF regions has a causal relationship with cav-1 and eNOS-pS1177. Interestingly, we found that HS degradation eradicated the effects of UF on cav-1 expression and distribution (Fig.
1). Under UF conditions with an intact GCX, cav-1 expression was significantly increased as was upstream and appositional localization. In contrast, no changes in expression or upstream localization were observed in Hep III-treated UF-conditioned cells when compared to Hep III controls, while appositional localization was retained. The findings of unchanged appositional localization of cav-1 in the absence of HS suggest that the localization of caveolae to cell–cell borders is not dependent on HS. However, all of our other evidence points to the fact that the extent and upstream nature of cav-1 expression are dependent on the presence of the HS component of the GCX. While previous research has suggested that caveolae serve as primary mechanosensors to shear stress and assist in eNOS activation [
99,
100], the work presented here indicates that caveolae act as indirect mechanosensors that are mediated by HS. Therefore, we propose that decreased cav-1 and caveolae polarization in ECs associated with atherosclerotic plaques is due to poor expression of HS. Additionally, other GCX components, such as hyaluronic acid, may similarly regulate cav-1 expression and localization, as the hyaluronic acid receptor CD44 is localized to caveolae [
101]. However, this was outside the scope of the current study.
We also found that HS is necessary for UF-mediated eNOS activation, showing significantly less eNOS-pS1177 expression in Hep III-treated UF-conditioned cells than UF-conditioned cells with an intact GCX (Fig.
2). These findings agree with one of our previous reports, which shows HS as a necessary GCX component for eNOS activation [
54] and proposes a GCX-caveolae-eNOS pathway in which under UF, the HS glycosaminoglycans, bound to the glypican core protein GCX, activate eNOS through their association with caveolae [
54,
102]. More interestingly, we found, for the first time, that HS degradation resulted in decreased cav-1 and eNOS-pS1177 colocalization (Fig.
2). These results suggest that decreased eNOS activation in DF regions is a result of inadequate caveolae-eNOS colocalization due to decreased HS expression. Furthermore, our data imply that HS can trigger post-translational modifications of eNOS or cav-1, such as eNOS acetylation, which is implicated in cav-1-eNOS colocalization [
31]; although, confirming acetylation is beyond the scope of the current study.