Glycosylphosphatidylinositol (GPI)-anchored proteins have been shown to cluster in microdomains enriched in glycosphingolipids and cholesterol and represent a relatively selective marker of lipid rafts. In recent years, several attempts have been made to use fluorescent probes to nondisruptively label these domains in living cells. Here, we have transfected endothelial cells with a GPI-anchored thermotolerant green fluorescent protein (ttGFP) to show colocalization of this fluoroprobe with another marker of lipid rafts, urokinase-type plasminogen activator receptor-1. ttGFP was used to quantify the cell surface area occupied by lipid rafts and to examine the effect of various proatherogenic signals on lipid rafts. Exposure of endothelial cells to asymmetric dimethylarginine and oxidized LDL (oxLDL), as well as oxidant stress, reduced the cell surface area occupied by lipid rafts. Next, the property of ttGFP to undergo a shift in absorbance depending on the clustering of these molecules was utilized to perform proximity imaging (PRIM). PRIM showed that nitric oxide (NO) increased the distance between GPI-anchored ttGFP molecules clustered in lipid-rich microdomains. This “unclustering” of GPI-anchored ttGFP was not reproduced by prooxidant signals and was due to reduction in membrane-cytoskeletal constraints on the lipid rafts. These findings suggested that two fundamentally different mechanisms modulate lipid rafts: 1) substance regulation of lipid rafts involving modification of cholesterol and sphingolipids and 2) structural regulation of lipid rafts through disruption of membrane-cytoskeletal interactions, switching off the spatial confinement of lipid rafts.
- nitric oxide
- urokinase-type plasminogen activator receptor
- oxidized low-density lipoprotein
during the past decade, the fluid-mosaic model of the plasma membrane was revised to incorporate specialized microdomains, i.e., lipid rafts (22). These domains are characterized by tight acyl chain packing, which explains their liquid-ordered phase (4, 10). These plasma membrane domains are endowed with signaling functions due to the clustering of various receptors and mediators of signal transduction on the outer and inner leaflets, respectively (17). A distinct subpopulation of these caveolin-containing lipid rafts (16, 18), i.e., caveolae, has been identified as plasma membrane invaginations harboring, e.g., various receptors, nonreceptor tyrosine kinases, GTPases, and endothelial nitric oxide (NO) synthase (eNOS). However, glycosylphosphatidylinositol (GPI)-anchored proteins do not appear to be enriched in caveolae (14, 15) and represent a relatively selective marker of lipid rafts (see below).
GPI-anchored proteins have been shown to cluster in these microdomains enriched in glycosphingolipids and cholesterol (5). GPI-anchored proteins have been reported to display, in addition to Brownian motion, a 7- to 9-s residence time within confined microdomains containing glycosphingolipids (21). A GPI-anchored folate receptor shows clustering of up to 50 molecules in confined (<70-nm) plasma membrane domains (25). However, the reported differences in the size of lipid rafts, depending on the methods used to study them, span a 100-fold scale (1), indicative of possible methodological errors. Because of dissatisfaction with the techniques used to study lipid rafts, in the past few years, there have been several attempts to use fluorescent probes to nondisruptively label these domains in living cells. Zacharias et al. (27) generated a series of monomeric green fluorescent protein (GFP) chimeras containing different lipid anchors to demonstrate that acyl modification promotes clustering in lipid rafts. Using DNA constructs encoding GFP probes with diverse membrane topology, Kenworthy et al. (11) showed that proteins diffuse over large distances, thus advancing the model of dynamic partitioning. GPI-anchored GFP were constructed and used as probes to “paint” lipid-rich domains expressed on the cell surface (12) and examine them microscopically.
One of the approaches used to study lipid-rich microdomains in vivo is proximity imaging (PRIM), a method based on the finding that the temperature-tolerant mutant of GFP (ttGFP) undergoes a shift in absorbance and fluorescence intensity at excitation wavelengths of 380–490 nm, depending on the distance between the neighboring molecules. This property makes ttGFP a convenient tool for the study of clustering of homologous proteins (7). Here we utilized GPI-anchored ttGFP as a probe to visualize lipid rafts and analyzed the changes in proximity of GPI-anchored ttGFP molecules expressed in human umbilical vein endothelial cells (HUVEC). Validation of this probe confirmed that it could be used to label lipid rafts. Several proatherogenic stimuli affected the distribution of lipid rafts, as did the cholesterol-depleting maneuvers, thus emphasizing the role of lipid structural components in their maintenance. We also found that NO donors or stimulation of endogenous NO production resulted in an increase in the distance between GPI-anchored ttGFP molecules, as assessed by PRIM. Analysis of the causes suggested that membrane-cytoskeletal interactions are targeted by NO. These findings emphasize another type of regulation by the cytoskeletal constraints of lipid rafts.
MATERIALS AND METHODS
HUVEC and endothelial cell growth medium (EGM-2) were purchased from Clonetics (Walkersville, MD). HUVEC were used between passages 4 and 7 and maintained at 37°C in 95% air-5% CO2. The cells (∼104/well) were seeded on circular microscope cover glasses (Fisherbrand) that were precoated with fibronectin (10 μg/ml final concentration) and sterilized under UV light. The slides were placed into 24-well tissue culture plates (Falcon). At 80% confluency, the cells were transfected with GPI-anchored ttGFP or wild-type ttGFP (wt-ttGFP) construct; after 24 h, they were incubated with 5 μM NG,NG-dimethyl-l-arginine dihydrochloride (ADMA), oxidized LDL (10 μg/ml, oxLDL), or LDL (10 μg/ml) for 24 h. The cells were treated with 2 mM methyl-β-cyclodextrin (MβCD) for 30 min and cholesterol oxidase (0.5 U/ml) for 60 min before fixation and analysis of GPI-anchored ttGFP distribution (19).
Antibodies and chemicals.
Mouse monoclonal eNOS/NOS type III and mouse monoclonal caveolin-1 antibodies (each at 2.5 μg/ml final concentration) were purchased from BD Transduction Laboratories; mouse monoclonal antibody against urokinase-type plasminogen activator receptor (uPAR) from American Diagnostica (CD87, 10 μg/ml); Alexa Fluor 594 goat anti-mouse IgG1 (2 μg/ml), Hoechst 33342, and Texas red-conjugated phalloidin from Molecular Probes; anti-annexin II antibody from Santa Cruz Biotechnology; filipin complex (derived from Streptomyces filipensis), fibronectin, sodium nitroprusside (SNP), calcimycin (A-23187), S-nitroso-N-acetylpenicillamine (SNAP, 20 mg/ml in DMSO), MβCD, and cholesterol oxidase (from Pseudomonas fluorescens) from Sigma-Aldrich; and bradykinin, ADMA, and NG-monomethyl-l-arginine from Alexis.
Construction of GPI-ttGFP expression vector and transfection of HUVEC.
GPI-ttGFP and wt-ttGFP plasmids were constructed as previously described. pGEX plasmid containing ttGFP was used to generate GPI-ttGFP constructs as described elsewhere (7). Plasmid purification was done with Qiagen Endofree Plasmid Maxi kit. The purified DNA was resuspended in endotoxin-free Tris-EDTA buffer at 2 μg/μl. Purity of the plasmid preparation was controlled by measurement of the ratio of absorbance at 260 nm to absorbance at 280 nm and agarose gel electrophoresis. Cells were passaged 1–2 days before transfection, which was carried out at 70–80% confluency using FuGENE 6 reagent (Roche). The optimal ratio of DNA to FuGENE was determined to be 1:3.
The basis for PRIM is the shift in the excitation ratio of ttGFP when two copies of the protein are brought into close proximity. Whereas monomeric proteins always display a constant excitation ratio, their homooligomerization or clustering results in an increase or a decrease in the excitation ratio of ttGFP (7). HUVEC grown to 80% confluency in Petri-35 dishes with glass bottom windows (MatTek) were transfected with GPI-GFP and wt-ttGFP plasmids. Fluorescence microscopy of HUVEC was performed using a dual-excitation–dual-emission spectrofluorometer (Photon Technology International) combined with an epifluorescence inverted microscope (Diaphot, Nikon) equipped with ×40 and ×60 objectives (1.3 and 1.4 numerical aperture, respectively). Cells were illuminated at alternating wavelengths of 410 and 470 nm at a frequency of 5 s−1. Fluorescence emission intensity at each excitation wavelength was recorded at 530 nm and analyzed as the ratio of fluorescence at 410 nm to fluorescence at 470 nm with the use of software from Photon Technology International. Alternatively, images were acquired every 5–15 s using an epifluorescence inverted microscope (Diaphot) equipped with an ×100 objective and a silicone-intensified-tube video camera (Hamamatsu). An automatic shutter (Lambda 10-2, Sutter Instruments) interfaced to MetaFluor software (Universal Imaging) was used to illuminate cells at alternating wavelengths of 390 and 480 nm. Data were analyzed with Origin 7 software.
Preparation and analysis of detergent-soluble and -insoluble fractions.
Cells were washed, lifted, and collected by centrifugation for 5 min at 1,000 g at 4°C. Pellets were resuspended and incubated for 30 min in 0.5 ml of buffer containing 125 mM NaCl, 20 mM MES, and 1% Triton X-100. The samples were homogenized (20 strokes) and centrifuged for 5 min at 16,000 g. The supernatant contained Triton X-100 (detergent)-soluble (DS) fraction. The pellet, i.e., the Triton X-100 (detergent)-insoluble (DI) fraction, was resuspended in a buffer containing 150 mM NaCl, 50 mM Tris·HCl, 1% Triton X-100, 1% sodium deoxycholate, and 0.1% NP-40. The proteins were separated on 4–20% Tris-glycine gel and transferred to a membrane, and annexin II (polyclonal antibody, 1:1,000 dilution; Santa Cruz Biotechnology) was detected by immunoblotting.
HUVEC were lysed in a buffer containing 500 mM sodium carbonate (pH 11.0) with protease inhibitors and homogenized with a Dounce homogenizer. The resulting suspension was centrifuged at 1,000 g for 10 min. The samples were adjusted to 45% sucrose by addition of 2 ml of 90% sucrose prepared in 25 mM MES (pH 6.5) and 0.15 M NaCl and then placed at the bottom of an ultracentrifuge tube. A 5–35% discontinuous sucrose gradient, formed at the top of the tube, was centrifuged at 44,500 rpm for 20 h in a rotor (model SW-T55i, Beckman Instruments). Fourteen fractions were isolated from each sample. Proteins were concentrated using the TCA method and separated in 4–20% gel. Annexin II was detected by immunoblotting.
HUVEC were transiently transfected with GPI-anchored ttGFP and subjected to different treatments (see results). The cells were washed with cold PBS, fixed with 4% paraformaldehyde for 30 min, washed twice with PBS, and then stained first with the anti-uPAR antibody overnight at 4°C and then with the secondary antibody for 1 h at 4°C. For eNOS staining, the cells were permeabilized for 10 min with 0.1% Triton X-100, and for staining with anti-caveolin-1 antibody, the cells were fixed with 1:1 cold methanol-acetone for 10 min on ice and then finally washed with cold PBS. For staining with filipin, the cells were incubated for 2 h at room temperature with filipin complex (0.05 mg/ml) in PBS-1% BSA. For staining with Texas red-conjugated phalloidin, the cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100 for 5 min. All slides were finally stained with Hoechst 33342 and mounted with Slowfade (Molecular Probes). The cells were imaged using a microscope (model TE-2000U, Nikon) equipped with a Spot Insight camera (Diagnostic Instruments). Images were digitally analyzed using MetaMorph software to quantify the proportion of clusters of cell membrane-associated GPI-anchored ttGFP (as percentage of cell perimeter) and colocalization with immunodetectable eNOS, uPAR, or caveolin-1.
NO measurement and fluorescence detection of superoxide.
HUVEC were cultured in 48-well cell culture dishes (Corning) and incubated with 5 μM ADMA, 5 μM ADMA + LDL, or oxLDL (10 μg/ml) for 24 h. The cells were loaded by incubation with 5 μM 4,5-diaminofluorescein fluoromethyl-diacetate (Molecular Probes) for 60 min at 37°C for NO measurement. After the cells were washed with PBS and incubated for an additional 15 min, fluorescence intensity was measured using a Bio-Tek fluorescence plate reader with 485-nm excitation and 528-nm emission filters.
For measurement of superoxide anions, the cells were incubated with 10 μM dihydroethidium (DHE; Molecular Probes) for 30 min at 37°C. A fluorescence plate reader (Bio-Tek) with 485-nm excitation and 620-nm emission filters was used for detection of DHE. Measurements were made immediately after removal of the incubation buffer. Background fluorescence from cell-free incubations was subtracted from the data.
Isolation of LDL and preparation of oxLDL.
LDL (1.019–1.063 g/ml) was isolated from normal human plasma by sequential ultracentrifugation. The LDL fraction was dialyzed against PBS containing 0.3 mM EDTA, sterilized by filtration through a 0.22-mm filter, and stored under nitrogen gas at 4°C. Protein content was determined by the method of Lowry. OxLDL was prepared by dialysis of LDL (500 mg/ml) in PBS containing 5 μM CuSO4 for 12 h at 37°C and then in PBS containing 0.3 mM EDTA twice for 12 h each. The purity and charge of LDL and oxLDL were evaluated by examination of electrophoretic migration in the agarose gel. The degree of oxidation of LDL and oxLDL was determined by measurement of the amount of thiobarbituric acid-reactive substances, which was 0.1 and 10–30 nmol/mg for LDL and oxLDL, respectively. Lipoproteins were used for experiments within 3 wk after preparation (9).
Values are means ± SE of at least three separate experiments. Student's t-test or ANOVA was used to determine statistical significance of differences. P < 0.05 was considered to be statistically significant.
Validation of the GPI-anchored ttGFP probe.
To test whether GPI-anchored ttGFP faithfully represents lipid-rich domains or rafts, we employed several maneuvers that are known to deplete cholesterol from the plasma membrane and, thus, would be expected to diminish the surface area occupied by lipid rafts. Application of cholesterol oxidase (0.5 U/l) for 60 min led to a significant decrease in total GPI clusters on the perimeter of HUVEC compared with untreated cells: from 1.56 ± 0.13 to 0.50 ± 0.08% (n = 25, P < 0.01; Fig. 1, A–C). Treatment for 30 min with 2 mM MβCD, another cholesterol-depleting agent, reduced the area occupied by lipid rafts on the cell surface to 0.83 ± 0.06% (n = 25, P < 0.01).
We costained HUVEC transfected with GPI-anchored ttGFP with antibodies against uPAR, a GPI-anchored protein known to partition preferentially to lipid rafts (6). The overall colocalization of uPAR and GPI-anchored ttGFP fluorescence was 25 ± 1.11% (Fig. 1, D–F), which corresponds to the previously reported fraction of uPAR (10–30%) at the cell surface in lipid rafts (6). This partitioning of uPAR did not change significantly after treatment with SNP, ADMA, oxLDL, or LDL (see below). The sterol-binding agent filipin, which was used to visualize the distribution of cholesterol and, thereby, lipid rafts in endothelial cells, strongly colocalized to GPI-anchored ttGFP fluorescence on the cell membrane (Fig. 1, G–I). Collectively, these data attest to the validity of GPI-anchored ttGFP as a probe for lipid rafts.
Distribution of lipid rafts on the cell surface: fluorescence microscopy.
In the next series of experiments, we used the GPI-anchored ttGFP to examine plasma membrane distribution and expression of lipid rafts and colocalization with other raft markers in HUVEC as affected by several proatherogenic stimuli. Endothelial cells transfected with GPI-anchored ttGFP were pretreated with the known proatherogenic stimuli for 24 h before fluorescence microscopy. In untreated cells, GPI-anchored ttGFP fluorescence occupied, on average, 1.31 ± 0.05% of the cell surface (Fig. 2). Pretreatment of HUVEC with oxLDL alone did not significantly change the percentage of GPI clusters on the perimeter of the cell membrane (1.37 ± 0.12%). Treatment of endothelial cells with LDL alone did not significantly change membrane distribution of lipid rafts compared with untreated cells. Incubation with 5 μM ADMA and oxLDL (10 μg/ml) decreased the expression of GPI-anchored ttGFP clusters at the perimeter of the plasma membrane to 0.97 ± 0.13%, which was significantly lower than in untreated cells (P < 0.01). In contrast, treatment with 5 μM ADMA + LDL (10 μg/ml) did not affect distribution of GPI-anchored ttGFP clusters on the cell surface (1.47 ± 0.16%, not significantly different from untreated cells).
Colocalization of caveolin-1 and eNOS with GPI-anchored GFP.
In the next series of experiments, we used a GPI-anchored ttGFP probe to address colocalization of eNOS and caveolin-1 with lipid rafts. Colocalization of caveolin-1 and GPI-anchored GFP was 27 ± 2.18% and did not change significantly after 24 h of pretreatment with ADMA, oxLDL, or LDL (Fig. 3, A–C). Colocalization of eNOS and GPI-anchored GFP averaged 32 ± 2%. Treatment of HUVEC with 5 μM ADMA + oxLDL (10 μg/ml) decreased (P < 0.05) colocalization of eNOS with lipid rafts (Fig. 3, D–F).
Exogenous and endogenous NO affects GPI-anchored GFP clustering.
As noted above, spectral shifts of the ttGFP mutant, depending on the proximity of neighboring molecules, give a ratiometric index, which represents the extent of self-association and can be used to monitor the clustering of GFP-tagged proteins in live cells. In HUVEC transfected with GPI-anchored ttGFP, the PRIM fluorescence ratio increased after application of SNP in a dose-dependent manner (Fig. 4). Immediately after application of 100 μM SNP, the ratio increased by Δ0.32 ± 0.01 (P < 0.05, n = 5). Treatment with 10 μM SNP increased the Δfluorescence ratio by 0.19 ± 0.02 (P < 0.05, n = 3), and treatment with 1 μM SNP had no effect. In contrast, cells transfected with wt-ttGFP showed a stable fluorescence PRIM ratio throughout the experiments and exhibited no shift in response to 100 μM SNP.
Cells transfected with ttGFP tagged with palmitoylation and isoprenylation consensus sequences showed stable fluorescence ratios of 0.47 ± 0.04 and 0.3 ± 0.03, respectively, throughout the experiments (13).
Next, HUVEC transiently transfected with GPI-anchored ttGFP or ttGFP were treated with NO donors (SNAP or SNP), 0.1–1.0 μM bradykinin, or 20 nM A-23187 (Fig. 5) and studied by PRIM (13). The basal ratio of fluorescence at 410 nm to fluorescence at 470 nm averaged 0.299 ± 0.07 (n = 25). Application of SNAP resulted in a dose-dependent increase in the 410 nm-to-470-nm fluorescence ratio, which did not occur in HUVEC transfected with ttGFP (Fig. 5A). After application of 0.1 μM bradykinin, the PRIM ratio increased to 0.48 ± 0.05 (P < 0.05, n = 5), consistent with an increase in the distance between GPI-anchored ttGFP probe molecules clustered in the lipid rafts (Fig. 5B). When the same stimulation was performed after pretreatment of HUVEC with 0.5 mM NG-monomethyl-l-arginine for 10 min, the 410 nm-to-470 nm fluorescence ratio was reduced to 0.34 ± 0.03, which is not different from baseline (n = 5). Similar results were obtained with the calcium ionophore A-23187 (Fig. 5D).
In the next series of experiments, clustering of GPI-anchored ttGFP in lipid rafts was tested by cross-linking GFP molecules with a monoclonal anti-GFP antibody. Pretreatment of HUVEC expressing GPI-ttGFP with anti-GFP antibodies (1:100 dilution) resulted in a significant shift of the bradykinin-elicited PRIM ratio from 0.57 ± 0.04 to 0.36 ± 0.02 (P < 0.01, n = 5; Fig. 5C). Under similar experimental conditions, pretreatment of HUVEC with an isotypic IgG was not associated with a statistically significant decline in the PRIM ratio. These findings further suggest that GPI-anchored ttGFP is clustered in unstimulated HUVEC, whereas cell stimulation with bradykinin or the calcium ionophore and subsequent generation of NO are accompanied by an increase in the distance between the molecules of the fluoroprobe.
It has recently been demonstrated that a tyrosyl residue in the fluorophoric pocket of GFP is sensitive to nitration, resulting in a decrease in fluorescence (8). In our experiments, the fluorescence intensity of ttGFP controls was unchanged during the experiments with NO donors. Hence, it is most likely that the recorded shift in the excitation ratio of GPI-anchored ttGFP on application of SNP, SNAP, A-23187, or bradykinin reflects transient changes in the intermolecular distance and/or the angle separating individual molecules.
Fluorescent microscopy shows that addition of 100 μM SNP 5 min before fixation leads to a significant decrease in the surface area occupied by GPI-anchored GFP clusters on the plasma membrane compared with untreated cells: 1.56 ± 0.1 vs. 0.41 ± 0.08% (n = 25 for each treatment, P < 0.01). This value is in good agreement with the above-mentioned results of intravital imaging. Interestingly, chronic inhibition of eNOS in transfected HUVEC (for 24 h with 5 μM ADMA) per se did not affect the extent of GPI clustering, but an additional 5 min of treatment with 100 μM SNP again showed a significant, but less dramatic, decrease in GPI clustering on the cell surface: 1.89 ± 0.2 vs. 1.11 ± 0.2% (n = 25, P < 0.05; Fig. 6).
Immunocytochemical analysis of GPI-anchored ttGFP colocalization with F-actin showed that the actin-rich web decorating the cytoplasmic face of the GPI-anchored ttGFP in control cells was conspicuously absent in HUVEC subjected to 5 min of SNP treatment (Fig. 7). It has been previously demonstrated that lipid-rich domains are constrained by the binding of annexin II and formation of membrane-actin cytoskeletal complexes (2, 3). Therefore, we inquired whether NO can regulate the stability of membrane-cytoskeletal complexes. First, we confirmed the similar membrane localization of annexin II in HUVEC. Isolation of membrane fractions via a nondetergent method (23) resulted in recovery of 13.2% of annexin II in light fractions 4–6 and 81.2% in heavy fractions 9–14 (Fig. 8A). After pretreatment with 10 μM SNP, only 2% of annexin II remained in light fractions 4–6. A cholesterol-modifying drug, cyclodextrin (MβCD), which depletes cholesterol from the plasma membrane, served as a positive control. After 30 min of treatment with MβCD, almost all detectable annexin II was recovered in heavy fractions 9–14.
Because Triton X-100 insolubility is one of the hallmarks of proteins residing in glycolipid- and cholesterol-rich domains (5, 20), DI and DS fractions of HUVEC were isolated and immunoblotted with a polyclonal anti-annexin II antibody. In resting HUVEC, ∼60% of annexin II resided in the DI fraction and 40% in the DS fraction (Fig. 8B). After 10 μM SNP treatment, the proportion of DI fraction-associated annexin II dropped to 44%, with redistribution to the DS fraction. This effect was reproduced with MβCD (Fig. 8B). These data suggest that NO induces dissociation of the F-actin cytoskeleton from the lipid-rich domains of the plasma membrane.
To test whether the NO-induced increase in the distance between GPI-anchored ttGFP molecules was due to oxidant stress, HUVEC expressing GPI-ttGFP were treated with H2O2 and analyzed by PRIM. There was no detectable shift in the PRIM fluorescence ratio, suggesting that this mechanism was not responsible for the effect of NO on lipid rafts (Fig. 9). At the same time, HUVEC treatment with cholesterol oxidase or MβCD resulted in a significant shift in the PRIM ratio, indicating that cholesterol depletion was accompanied by an increase in the distance between GPI-anchored ttGFP probe molecules in the lipid raft.
Release of NO and reactive oxygen species.
To examine the impact of proatherogenic stimuli on the lipid raft-anchored proteins, as exemplified by eNOS, in the next series of experiments, HUVEC production of NO and superoxide anion was monitored. Incubation of HUVEC with 15 μM ADMA and 5 μM ADMA + oxLDL (10 μg/ml) maximally suppressed NO production, as detected by 4,5-diaminofluorescein fluorescence (Fig. 10). On the other hand, pretreatment of HUVEC with 5 or 15 μM ADMA, LDL, oxLDL, 5 μM ADMA + oxLDL (10 μg/ml), or 5 μM ADMA + LDL (10 μg/ml) increased production of superoxide measured by DHE fluorescence, reaching statistical significance for treatment with 5 μM ADMA, 15 μM ADMA, and 5 μM ADMA + oxLDL (10 μg/ml).
The data presented here confirmed the usefulness of the GPI-anchored ttGFP beacon for locating lipid rafts and demonstrated that the plasma membrane surface area occupied by lipid rafts is reduced after prolonged exposure of endothelial cells to oxLDL and the eNOS inhibitor ADMA. In addition, the data established the use of GPI-anchored ttGFP in PRIM-specific spectral shifts characteristic of a change in the degree of clustering of this fluorophore within lipid-rich domains. Using this approach, we demonstrated that acute exposure to NO donors, as well as stimulation of eNOS, results in reversible dissociation of lipid rafts. These findings suggested that two fundamentally different mechanisms modulate lipid rafts: 1) modification of cholesterol and sphingolipids, i.e., substance regulation of lipid rafts (Fig. 11B), and 2) membrane-cytoskeletal interactions resulting in disruption of the spatial confinement of lipid rafts, i.e., structural regulation of lipid rafts (Fig. 11A). OxLDL and ADMA appear to affect lipid rafts through mechanism 1, as predicted Blair et al. (3). In contrast, acute NO effects appear to act via mechanism 2, by destabilizing raft-cytoskeletal constraints. The cellular consequences of these disparate modes of regulation may be also different. The depletion of lipid rafts on the cell surface may reduce the signaling from these platforms, on which receptors, G proteins, eNOS, and tyrosine kinases are clustered. The implications of structural regulation may be related to the reversible changes of biophysical properties of endothelial cells. The finding that NO resulted in an increase in the distance between GPI-anchored ttGFP molecules indicates that this marker of lipid-rich microdomains may undergo a reversible transition from a clustered form to a more dispersed organization. Given that endothelial cells in vivo are continuously stimulated by the changing fluid shear stress and mechanical deformation to generate NO, one could infer that lipid rafts in endothelial cells may be constantly remodeled, exhibiting a high rate of turnover. The high solubility of NO in lipids and the proximity of the NO source to lipid-rich domains (24) would provide an ideal microenvironment for the observed changes in the biophysical properties of lipid rafts.
There is substantial evidence that cholesterol depletion, with cholesterol oxidase or MβCD, promotes dissociation of lipid rafts (19). Using GPI-anchored ttGFP as a beacon for lipid rafts, we were able to confirm these observations; both maneuvers leading to cholesterol depletion produced an unambiguous decrease in the surface area on the plasma membrane occupied by lipid rafts in HUVEC. In addition, we demonstrated that oxLDL + ADMA produces a similar shrinkage of lipid-rich domains. Because several messengers and enzymes, including eNOS, are anchored to lipid-rich domains, any perturbations of their structure should modify the function of these previously lipid-anchored molecules. We measured endogenous NO production as a readout of such modifications. The results obtained with H2O2 treatment of HUVEC, leading to the depletion of lipid rafts, justified the use of a second readout system: superoxide generation. Although oxLDL and ADMA suppressed basal and ionomycin-stimulated NO production and stimulated superoxide anion generation, their combined action showed sensitization of eNOS to the lower concentrations of ADMA. This effect may be reflective of enhanced uncoupling of eNOS through the oxidative stress induced by oxLDL.
These studies were supported in part by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-45462, DK-54602, and DK-52783 (to M. S. Goligorsky), a National Kidney Foundation Fellowship (to H. Li), and an American Heart Association Fellowship (to S. Brodsky).
We are indebted to Dr. E. London (Dept. of Biochemistry, SUNY) for preliminary studies utilizing multilamellar vesicles. We thank Michael S. Wolin for allowing us to perform several experiments in his laboratory and Christopher J. Mingone for technical support. We also thank Steven S. Gross (Dept. of Pharmacology, Weill Medical College) for providing LDL and oxLDL.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 by the American Physiological Society