INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

ONE MECHANISM by which hypercholesterolemia is believed to induce vascular dysfunction is by increasing O· generation. Although ample, indirect evidence exists to suggest that hypercholesterolemia increases O· production by the vessel wall (16, 17, 22), the cells involved and the mechanisms by which native low-density lipoprotein (nLDL) and its oxidized derivatives alter vascular O· production remain unclear. For example, hypercholesterolemia increases adherence of mononuclear cells to the vessel wall (19, 29). Recruitment and subsequent migration of mononuclear cells into the vessel wall introduces new cellular sources of O·. Even before recruitment, high plasma concentrations of LDL may promote hyperresponsiveness of circulating blood elements (12). Thus mononuclear cells entering the vessel wall may be predisposed to generate increased quantities of O· after transmigration.

Hypercholesterolemia also increases nLDL entry into the vessel wall, where it becomes trapped and subsequently oxidized (20, 21). Once oxidized it is capable of inducing a wide variety of potent proinflammatory responses in the vasculature leading to impair function. Many studies have shown that the effects of LDL on endothelial and vascular function depend strongly on the degree of oxidation. For example, one essential function of the endothelium is to limit the accumulation of lipid hydroperoxides (LOOH) in nLDL (26). When the content of LOOH in LDL exceeds a critical threshold, endothelial cells begin to participate in the oxidation process (26). In addition to endothelial cells, smooth muscle cells can also oxidize LDL by a O·-dependent process (13).

Whereas many studies have implicated O· as a key factor in the pathogenesis of atherosclerosis, the direct effects of LDL on vascular endothelial O· production have not been systematically examined. Canines for the most part are resistant to diet-induced atherosclerosis, although some colonies have been established that are sensitive to cholesterol if fed for an extended time (4-8 mo) (1). Recent studies found that inhibiting platelet function significantly reduced the progression but did not eliminate atherosclerosis (1). These observations suggest that high plasma concentrations of cholesterol are able to induce atherogenesis in the canine.

The goal of this study was to test the hypothesis that lipoproteins, in the absence of other confounding influences, increase O· generation by vascular endothelium in situ. Previously, we and colleagues (25) and others (32) incubated cultured endothelial cells with nLDL and/or oxidized LDL and found that these lipoproteins increased O· production by uncoupling endothelial nitric oxide synthase (eNOS) activity as an initial step in sorting out the mechanisms involved. Although both nLDL and oxidized LDL increased O· generation in cultured endothelial cells, it is unknown whether they would increase O· in situ, and, if so, by what cell type and oxidative enzyme. Answering these questions required an alternative analytic approach for assessing O· generation in vascular tissues.

In this investigation, we incubated isolated carotid arteries with purified nLDL and minimally oxidized LDL (mmLDL) to determine their effects on vascular O· generation. We used fluorescent confocal microscopy and hydroethidine (HEt), an oxidant-sensitive, intracellular fluorescent probe, to identify which cells were generating O·. Conversion of HEt to ethidium (Et) was examined with respect to radical specificity. Sources of oxidative enzyme activity were determined with specific and selective inhibitors. Finally, as nLDL and oxidized LDL were reported to uncouple eNOS activity, the effects of these atherogenic lipoproteins on eNOS in cultured endothelial cells were examined by Western blot analysis for changes in phospho-eNOS (serine-1179) and association of heat shock protein 90 (HSP90).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Preparation of nLDL. nLDL was isolated by sequential density ultracentrifugation (1.019-1.063 g/ml) from pooled human plasma (2-5 donors per isolation) as previously described (25). Butylated hydroxytoluene, a lipid-soluble antioxidant, and EDTA were added immediately to the plasma (final concentrations = 20 µM and 0.01%, respectively). The plasma was mixed for 15 min at room temperature before ultracentrifugation. Sterile techniques, reagents, and dialysis solutions were used for all isolation steps. Cholesterol concentrations were determined using the cholesterol oxidase kit from Sigma (catalogue number 352-20). nLDL was used within 1 wk of isolation.

Preparation of mmLDL. nLDL was isolated as outlined above. The LDL was placed in a sterile 50-ml conical tube and stored at 4°C for at least 3-6 mo before use for experiments as described (4, 8). mmLDL had a significantly higher content of LOOH than nLDL as measured by the tri-iodide assay (mmLDL = 73.8 ± 25.0 vs. nLDL = 0.3 ± 5.0 nmol LOOH/mg cholesterol, P < 0.01, n = 5) (26).

Tissue extraction. Canine carotid arteries were obtained from adult mongrel dogs of either gender. Vessels were quickly excised and transferred to physiological saline solution at room temperature, and surrounding advential tissue was quickly cleaned away. The artery was sectioned into segments of at least 2 cm long and placed in RPMI 1640 containing 10% fetal bovine serum (FBS), 20 mM HEPES, and antibiotics-antimycotics (termed RPMI medium hereafter). After being washed twice with RMPI 1640 to remove adherent blood elements, the vessels were placed into organ culture.

Organ culture. Canine carotid artery segments were placed in a 60-mm dish containing either control RPMI medium or RPMI medium containing nLDL (80, 160, or 240 mg/dl) or mmLDL (240 mg/dl). The arterial segments were incubated at 37°C, 5% CO₂-95% air and 100% H₂O for 24 h.

Superoxide anion detection. Radical specificity of HEt conversion to Et was examined in phosphate buffer (250 mM, pH 7.45) containing HEt (10 µM). Fluorescence was monitored with respect to time on a Perkin Elmer LS50B (excitation wavelength 488 nm, emission wavelength 620 nm, 10-nm slit widths). Superoxide anion was generated with xanthine/xanthine oxidase (30, 31). First, xanthine was added (final concentration = 1 mM) and mixed, and fluorescence was recorded. Xanthine oxidase was then added (final concentration = 14 U/ml), the solution mixed, and fluorescence recorded. Finally, superoxide dismutase (SOD) was added (final concentration = 1,000 U/ml) to inhibit O·-dependent increases in Et fluorescence. The solution was mixed and fluorescence recorded. The effect of Tiron, a well-known cell-permeable, O·-dismutating agent (34) was determined in a separate incubation by substituting Tiron (final concentration, 10 mM) for SOD.

Inhibitor studies. To determine the effects of nLDL and mmLDL on non-O·-dependent conversion of HEt into Et, arterial segments were preincubated with MnTBAP (10 µM) for 30 min, and Et fluorescence in the isolated arterial segment was recorded by confocal microscopy. Accordingly, the images obtained from control, nLDL, and mmLDL pretreated with MnTBAP represented the extent to which non-O·-dependent mechanisms converted HEt to Et.

Effects of nLDL and mmLDL on the association of HSP90 with eNOS. Bovine aortic endothelial cells (BAEC) were incubated with nLDL (240 mg/dl) or mmLDL (240 mg/dl) in RPMI medium for 24 h. After incubation, the cultures were washed three times with HBSS and then incubated in HBSS containing L-arginine (10 µM) with and without A-23187 (5 µM) for 10 min. The HBSS buffer was removed by aspiration, and cell proteins were harvested in modified RIPA buffer containing 1:100 dilution of phosphatase inhibitor cocktail I (P2850, Sigma Chemical; St. Louis, MO) and other inhibitors as described (24). To determine the effects of nLDL and mmLDL on eNOS activation, eNOS was immunoprecipitated from lysates of the cultures using established protocols (24). Coprecipitated proteins were separated by SDS-PAGE (7.5%), transferred to nitrocellulose, and blotted for phospho-eNOS (serine-1177/serine-1179), eNOS, and HSP90 as described (24).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Detection of O· generation. A cell-free system was used to determine the specificity of reactive oxygen species conversion of HEt to Et. Some of the reagents in these experiments were dissolved in NaOH to increase their stability. To ensure that reactions took place in constant pH, the HEt was dissolved in 250 mM phosphate buffer (pH 7.45). Xanthine had little effect on the baseline levels of Et fluorescence (Fig. 1). Upon addition of xanthine oxidase, however, marked increases in Et fluorescence were observed (rate = 0.0458 ± 0.005 U/s) that were completely blocked by SOD. Previous cell-free studies indicated that hydrogen peroxide (200 µM) increased conversion of HEt to Et to ~10% of the fluorescence induced by potassium superoxide (200 µM) (5). If hydrogen peroxide contributed significantly to the conversion of HEt to Et under these conditions, then SOD should have partially reduced this rate rather completely reducing it to zero. Accordingly, we interpret these data to mean that hydrogen peroxide formed from dismutation of O· contributes little to the conversion of HEt to Et.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Conversion of hydroethidine (HEt) to ethidine (Et). Increases in fluorescence due to xanthine (X)/xanthine oxidase (XO) generation of superoxide anion are shown. Freshly prepared solutions of HEt (10 µM) in phosphate buffer (250 mM, pH 7.45) possess low levels of fluorescence in the presence of xanthine alone. Addition of XO to the cuvette causes a marked increase conversion of HEt to Et, which is highly fluorescent. Addition of superoxide dismutase (SOD, 1,000 U/ml) blocks further increases, whereas addition of 2-(N,N-diethylamino)-diazenolate-2-oxide sodium salt (DEA NONOate, 14 µM) attenuates the rate of increase in Et fluorescence. These line graphs show that the conversion of HEt to Et is superoxide anion dependent.

View larger version (38K): [in this window] [in a new window]   Fig. 2.   En face confocal fluorescent microscopy of superoxide anion production in situ. Representative black and white images of en face fluorescent confocal micrographs of canine carotid arteries are shown. A: fluorescence at the edge of a cut arterial segment. B: fluorescence in the endothelium of an arterial segment incubated for 24 h with minimally oxidized low-density lipoprotein (mmLDL, 240 mg/dl). C: fluorescence in the smooth muscle cell layer of the same arterial segment as in B. D: pseudocolorized image of endothelium from enface fluorescent confocal micrograph of an arterial segment incubated for 24 h with native LDL (nLDL, 240 mg/dl). E: pseudocolorized image of endothelium from arterial segment incubated for 24 h with nLDL (240 mg/dl) pretreated with manganese-5,10,15,20-tetrakis(4-benzoic acid)porphyrin for 30 min before incubation with HEt.

nLDL increases vascular O· generation. nLDL significantly increased vascular O· production in a concentration-dependent fashion (Fig. 3A). When arterial segments were incubated with a single concentration of nLDL (240 mg/dl), O·-dependent Et fluorescence was significantly increased at 6 h and again at 24 h (Fig. 3B). L-NAME significantly inhibited nLDL-induced increases in O· (P < 0.01), suggesting that eNOS was a major source of O· in nLDL-treated arterial segments (Fig. 4). The role of xanthine oxidase and NADPH oxidoreductase in the nLDL-induced increase in O· production was not assessed. The rationale for this decision lies in the observation that L-NAME reduced O· generation by ~64% (Fig. 4). The remaining 36% of the O· was too small to dissect out the role of other oxidative enzymes pharmacologically within the statistical and discriminatory power of the HEt assay. Whereas we cannot discount a modest role for either xanthine oxidase or NADPH oxidoreductase in the production of O· in response to nLDL, it is clear that the majority of O· generated in the endothelium of arterial segments treated with nLDL appears to originate from eNOS. It should be noted that L-NAME reduced Et staining in the nLDL-treated arterial segments to essentially the same extent as DPI did in the mmLDL-treated arterial segments.

View larger version (25K): [in this window] [in a new window]   Fig. 3.   nLDL increases endothelial superoxide anion production in situ. Relative changes in superoxide anion, as assessed by fluorescence intensity of Et, in the endothelium of arterial segments incubated with nLDL (240 mg/dl) with respect to concentration (A) and time (B) are shown. Data represent means ± SE. Statistical analysis was two-tailed Student's t-test with Welch's correction (**P < 0.01, n = 5). NS, not significant.

View larger version (49K): [in this window] [in a new window]   Fig. 4.   nLDL and mmLDL differentially increase endothelial superoxide anion production in situ via different oxidative enzymes. Relative changes in superoxide anion, as assessed by fluorescence intensity of Et, in the endothelium of arterial segments incubated with nLDL (240 mg/dl, 24 h) and mmLDL (240 mg/dl, 24 h) in the absence and presence of inhibitors are shown. Data represent means ± SE. Statistical analysis was two-tailed Student's t-test with Welch's correction (*P < 0.01; **P < 0.001). nLDL- and mmLDL-treated arterial segments were incubated with N-nitro-L-arginine methyl ester (L-NAME, 500 µM), allopurinol (AL, 100 µM), or diphenyleneiodonium (DPI, 100 µM) for 15 min before and during incubation with HEt.

mLDL increases vascular O· generation. Incubation of arterial segments with mmLDL (240 mg/dl) significantly increased O· in vascular endothelium greater than two times the levels induced by nLDL and greater than eight times the levels at baseline (P < 0.01, Fig. 4). Optical dissection of the mmLDL-treated arterial segments revealed that most of the increase in Et fluorescence resided in the endothelium with minor and occasional involvement of smooth muscle cells (data not shown). This larger increase in O· production made it possible to examine the role of other oxidative enzymes with greater confidence. L-NAME significantly decreased Et fluorescence induced by mmLDL (P < 0.01) confirming observations that oxidized LDL uncouples eNOS activity (32). Interestingly, L-NAME reduced Et fluorescence in the mmLDL-treated segments to essentially the same extent as it did in the nLDL-treated arterial segments (delta change with L-NAME, mmLDL = 1,900 vs. nLDL = 2,172 arbitrary fluorescent units). Allopurinol and DPI significantly reduced Et fluorescence in mmLDL-treated segments by ~56% and ~80%, respectively. Thus mmLDL increases vascular O· generation by uncoupling eNOS and activating additional oxidative enzymes xanthine oxidase and NAD(P)H oxidoreductase. Representative pseudocolorized images of Et fluorescence in the vascular endothelium of arterial segments illustrate the prooxidant effects of nLDL and mmLDL and the effects of selective inhibitors on blocking O· generation from the different oxidative enzymes (Fig. 4).

Effects of nLDL and mmLDL on the activation state of eNOS. Typically, control cultures contain low levels of phospho-eNOS and low levels of HSP90 associated with eNOS. Upon stimulation, control cultures display marked increases in phosphorylation of eNOS at serine-1179 and association of HSP90 with eNOS (24). In Fig. 5, we see the typical pattern of phospho-eNOS and HSP90 association with eNOS in the control cells. In this blot, neither nLDL nor mmLDL appeared to alter basal or A-23187-stimulated levels of phospho-eNOS. These lipoproteins, however, did alter the association of HSP90 with eNOS. nLDL increased the association of HSP90 with eNOS under basal conditions, suggesting that low levels of oxidative stress activate cultured endothelial cells. In contrast, in cells incubated with mmLDL, the association of HSP90 with eNOS remained low. When the lipoprotein-treated endothelial cell cultures were stimulated with A-23187, the association of HSP90 with eNOS decreased to baseline levels in cells incubated with nLDL (Fig. 5A, lane 5) but failed to increase over control levels in cells treated with mmLDL (Fig. 5A, lane 4 vs. lane 6). Figure 5B illustrates the reproducibility of the effects of nLDL and mmLDL on phospho-eNOS, eNOS, and association of HSP90 in endothelial cultures exposed to nLDL and mmLDL for 24 h.

View larger version (40K): [in this window] [in a new window]   Fig. 5.   nLDL and mmLDL differentially modulate the activation state of enothelial nitric oxide synthase (eNOS). A: Western blots of coprecipitated proteins in an immunoprecipitation of eNOS from bovine aortic endothelial cell cultures incubated for 24 h with nLDL (240 mg/dl) and mmLDL (240 mg/dl). IP, immunoprecipitation; IB, immunoblot. Low levels of phospho-eNOS (p-eNOS) are observed in the IP of eNOS under basal conditions. A-23187 (5 µM) stimulation markedly increased phospho-eNOS levels, indicating an increase in electron flow through the enzyme. IB for eNOS shows that the level of eNOS in each lane is essentially equal. IB for heat shock protein 90 (HSP90) reveals the nLDL and mmLDL differentially modulate the association of HSP90 with eNOS. Increases in phospho-eNOS without sufficient HSP90 suggest that nLDL and mmLDL increase eNOS-dependent superoxide anion production by disrupting normal HSP90 and eNOS interactions. B: means ± SE for the relative densities for phospho-eNOS/eNOS and HSP90/eNOS under the treatment conditions indicated as pluses and minuses in A. Open bars, ratio of a control value of 1.0; solid bars, endothelial cell cultures treated with nLDL or mmLDL under basal (nonstimulated conditions); hatched bars, control, nLDL, and mmLDL cultures stimulated with A-23187.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In the present study, we used HEt to detect relative changes in vascular O· production in situ. With this oxidant-sensitive probe, we were able to identify which cells in the vessel wall generated O· and obtain semiquantitative data from the images using conservative analysis protocols. Our findings indicate that nLDL and mmLDL induced different oxidative enzymes to increase O· generation in the endothelium of arterial segments. En face optical dissection of the vessel wall revealed that nLDL increased O· by an eNOS-dependent mechanism. mmLDL increased O· primarily in the endothelium, although some vessels showed clear evidence of O· production in the smooth muscle cells. mmLDL uncoupled eNOS activity as well as increased O· generation from xanthine oxidase and NAD(P)H oxidoreductase. Such differences reinforce the notion that vascular disease develops from a series of distinct mechanisms that may or may not overlap, depending on the extent of oxidative modification of the LDL particle. These in situ data confirm previous findings by us and colleagues (25) and Malinski and associates (32) using cultured endothelial cells. Furthermore, our observations that A-23187 stimulation decreased the association of HSP90 with eNOS in nLDL-treated endothelial cells and failed to increase in mmLDL-treated cells, rather than the typical response of increased association, provide new insight into the cellular mechanisms governing uncoupled eNOS activity.

In recent years HEt has gained popularity as an O· probe because it is sensitive and easy to use, despite the possibility it may actually underestimate production (3). Arnal and associates (2) reported that for measurements of extracellular O·, SOD-inhibitable increases in Et fluorescence detected similar variations in O· production as ferricytochrome c and lucigenin. Recent studies combining confocal microscopy with HEt successfully detected intracellular O· in the myocytes of heat-stressed diaphragms from mice (35). Here, we used a similar approach to identify which cells were generating O· in the vessel wall and specific enzyme inhibitors to determine which oxidative enzymes were induced after exposure to nLDL and mmLDL. Identifying which cells were generating O· was made easier by the fact Et partitions in the nuclei after conversion. Such partitioning makes the fluorescent probe less available to reducing agents or reagents that quench. For example, MnTBAP, which is a metal heme compound that may quench, had no effect on the median fluorescence per cell in endothelial cells prelabeled with Et. Thus, when cells or tissues are incubated with MnTBAP or Tiron, another O· scavenger that is accessible to intracellular compartments (35), one can conclude that the observed decreases in Et fluorescence are more likely due to O· scavenging than quench.

The cell-free studies using SOD indicate that the conversion of HEt to Et is relatively specific for O·. We found that when ·NO scavenged O·, the conversion of HEt to Et was attenuated in a concentration-dependent fashion by >90%. Freshly synthesized peroxynitrite did not significantly promote HEt conversion to Et, especially in the presence of glutathione. In vascular tissues, MnTBAP (a cell-permeable SOD mimetic) reduced Et fluorescence in the endothelium of isolated arterial segments to background levels. Our observation that MnTBAP reduced O·-dependent increases in Et fluorescence was consistent with findings obtained with Tiron (35). The decrease in Et fluorescence in the nLDL- and mmLDL-treated arteries mediated by L-NAME suggests that both atherogenic lipoproteins induced eNOS to generate O·. The decreases in Et fluorescence in mmLDL-treated arteries mediated by allopurinol and DPI indicate that mmLDL increased O· generation by a xanthine oxidase- and a NADPH oxidoreductase-dependent mechanism. These conclusions are based on the fact that these inhibitors are well recognized for blocking O· generation from their targeted enzymes in other systems (23, 31, 33). Although conversion of HEt to Et may underestimate actual O· production (3), relative increases in endothelial generation of O· can be assessed with sufficient precision in situ to allow for statistical discrimination among test groups.

With appropriate protocols, en face fluorescent confocal microscopy provided precise and reproducible measurements of Et fluorescence in vascular tissues. After the endothelium was located by adjusting the microscope through different focal planes, the surface of the lumen was quickly assessed for endothelial integrity and fluorescent intensity. Moistening coverslips with PSS before placing arterial segments endothelial side down helped maintain the integrity of the endothelium. Marked increases in Et fluorescence were noted in the blood vessels in regions of the lumen that were damaged due to cutting carotid arteries into segments. For our analyses, we cut the arteries into segments 2 cm in length before incubation. This step increased the surface area of endothelium available for analysis, making it possible to exclude damaged regions. En face fluorescence microscopy revealed that the arterial segments also possessed low levels of nonspecific fluorescence. Accordingly, background measurements of fluorescence in regions between nuclei were obtained and subtracted from Et fluorescence in the nuclei of the vascular endothelium. By avoiding damaged regions and correcting nuclear pixel intensity for nonspecific fluorescence, we were able to obtain fluorescent intensity data that revealed that nLDL increased O· by an endothelium-, eNOS-dependent mechanism and that mmLDL increased O· predominantly by an endothelium- (with some involvement of vascular smooth muscle cells), eNOS-, xanthine oxidase-, and NAD(P)H oxidoreductase-dependent mechanism.

The mechanisms by which LDL induces vascular dysfunction are complex. Previously, our laboratory focused on mechanisms of endothelial dysfunction induced by nLDL. When LDL crosses the endothelium and becomes trapped in the vessel wall, it can be oxidized to a proinflammatory LDL particle by endothelial cells, smooth muscle cells and monocytes (4, 6, 13). In this way, multiple mechanisms can be induced in the vessel wall by the initial insult of increased plasma concentrations of nLDL. Our findings suggest that the association of HSP90 with eNOS, which is required for ·NO production and activity (9, 10), might also be regulated by oxidative stress. The evidence supporting this statement comes from the fact that mmLDL contained significantly higher levels of LOOH than nLDL before incubation. Such changes in oxidation are similar to the levels in mmLDL reported earlier (8).

The finding that the amount of Et fluorescence inhibited by L-NAME in mmLDL-treated arterial segments was essentially equal to that inhibited in the nLDL-treated segments suggests that these atherogenic lipoproteins uncoupled eNOS to the same extent. Western blot analysis revealed that the activation state of eNOS was altered with respect to its association with HSP90. However, neither lipoprotein significantly altered phospho-eNOS levels under basal or A-23187-stimulated conditions. nLDL increased the association of HSP90 with eNOS under basal conditions, whereas mmLDL maintained this association at low levels. In A-23187-stimulated cells, nLDL decreased the level of HSP90 associated with eNOS compared with the levels in nLDL cells under basal conditions and the levels in A-23187-stimulated control cells. In mmLDL-treated cells, the association of HSP90 with eNOS failed to increase when stimulated with A-23187. Taken together, these data indicate that one mechanism may be a decrease in or failure of HSP90 to associate with eNOS when eNOS is activated. It should be noted that differences in the regulation of the association of HSP90 with eNOS appeared to be related to the levels of oxidative stress induced by nLDL and mmLDL. Furthermore, it is important to note that the lipoprotein with the highest level of oxidative change, mmLDL, impaired HSP90 interactions with eNOS under both basal and A-23187-stimulated conditions. Such differences in regulation of this protein-protein interaction suggest a role for oxidative stress.

Recent findings by Go et al. (11) also indicate that oxidized LDL uncouples eNOS activity. They report that oxidized LDL induced a nearly threefold increase in phospho-eNOS (serine-1179) levels in BAEC over a period of 30 min. In the initial incubations, citrulline conversion kept pace with phospho-eNOS. However, at a later time point (30 min), citrulline conversion decreased even though phospho-eNOS levels remained elevated (11). Our findings here are consistent with those of Go et al. (11) in that a decrease in citrulline production in the presence of increased levels of phospho-eNOS also suggests that eNOS generates O·. Evidence in support of this idea comes from our data showing that L-NAME decreased Et fluorescence in mmLDL-treated arterial segments.

In separate studies using diverse redox cycling agents (dichromate VI, adriamycin, and bis-N-methylacridinium nitrate) to induce oxidative stress, we found that low levels of oxidative stress appeared to increase, whereas high levels appeared to decrease HSP90 interactions with eNOS (data not shown). The changes in the association of HSP90 with eNOS in cells acutely treated with the redox cycling agents (15 min to 4 h) were similar to the changes induced by 24 h incubation with nLDL representing low levels, and mmLDL, representing high levels of oxidative stress. Taken together, such changes indicate that the effects of oxidative stress on the association of HSP90 with eNOS are both concentration and time dependent.

In summary, nLDL and mmLDL increase vascular endothelial O· generation in situ. nLDL increases O· by what appears to be an eNOS-dependent mechanism, whereas mmLDL increases O· by eNOS-, xanthine oxidase- and possibly NAD(P)H oxidoreductase-dependent mechanisms in vascular endothelium of intact arterial segments. Protein-protein interaction studies suggest that the cellular mechanisms by which these atherogenic lipoproteins increase eNOS-dependent O· generation may involve altering the association of HSP90 with eNOS leading to a change in enzyme function. As increases in O· generation are believed to play a central role in vascular dysfunction and premature development of atherosclerosis, therapeutic targeting of these oxidative pathways to limit O· generation may be a useful strategy for preserving endothelial and vascular function.