A Heat Shock Protein 90 Binding Domain in Endothelial Nitric-oxide Synthase Influences Enzyme Function*

Previous reports suggest heat shock protein 90 (hsp90) associates with endothelial nitric-oxide synthase (eNOS) to increase nitric oxide (·NO) generation. Ansamycin inhibition of chaperone-dependent activity increases eNOS generation of superoxide anion (\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{2}^{{\bar{{\cdot}}}}\) \end{document}) upon enzyme activation. In the present study we identify where hsp90 binds to eNOS using overlapping decoy peptides based on the amino acid (aa) sequence of eNOS (291–420). B1, B2, and B3 peptides inhibited hsp90 association with eNOS in cell lysates from proliferating bovine aortic endothelial cells. B2 (aa 301–320), common to both B1 and B3, decreased stimulated ·NO production and hsp90 association in bovine aortic endothelial cells. The B2/B3 peptide was redesigned to TSB2 that includes a TAT protein transduction domain and shortened to 14 aa. TSB2 impaired vasodilation of isolated facialis arteries in vitro and in vivo and increased eNOS-dependent \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{2}^{{\bar{{\cdot}}}}\) \end{document} generation in native endothelial cells on mouse aortas, whereas a control peptide, TSB(Ctr), which has the four glutamic acids in TSB2 substituted with alanine, showed no such effects. Site-directed mutagenesis of eNOS at 310, 314, 318, and 323 Glu to Ala yields an eNOS mutant that exhibited reduced hsp90 association and generated \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{2}^{{\bar{{\cdot}}}}\) \end{document} rather than ·NO upon activation. Together, these data demonstrate that hsp90 associates with eNOS at aa 310–323. Moreover, a decoy peptide based on this sequence is sufficient to displace hsp90 from eNOS and uncouple eNOS activity from ·NO generation. Thus, Glu-310, Glu-314, Glu-318, and Glu-323 in eNOS, although each does not do much by itself, synergistically they increase “cooperativity” in the association step that is critical for maintaining hsp90-eNOS interactions and promoting coupled eNOS activity. Such chaperone-dependent signaling may play an important role in modulating the balance of ·NO and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{2}^{{\bar{{\cdot}}}}\) \end{document} generation from eNOS and, therefore, vascular function.

Previous studies showed that the association of heat shock protein 90 (hsp90) 3 with eNOS played an important role in the generation of nitric oxide (⅐NO) (1). Studies from this laboratory revealed that inhibition of hsp90 ATPase-dependent chaperone activity not only decreased stimulated ⅐NO generation but also increased eNOS-dependent O 2 . production (2)(3)(4). These reports indicated that inhibiting hsp90dependent signaling with eNOS allows eNOS to generate O 2 . upon stimulation rather than ⅐NO. Because ⅐NO plays a central role in vascular biology, such changes in enzyme function will likely have a major impact on endothelial cell (EC) and vascular physiology. An earlier report by Sessa and co-workers (5) showed, using a yeast two-hybrid system, that hsp90 interacted with eNOS at amino acid (aa) 300 -400 of eNOS. Previous studies by Pagano and co-workers (6) showed that small peptides corresponding to a portion of gp91 phox could act as decoy peptides to inhibit assembly of vascular NADPH oxidoreductase and, therefore, vascular O 2 . generation. On the basis of these studies we reasoned that small peptides derived from the primary aa sequence of eNOS might be useful to determine where hsp90 bound on eNOS. Such knowledge might help in developing new tools for dissecting the cellular mechanisms by which hsp90 modulates eNOS function and subsequently EC-dependent vasodilation. The objectives here were to determine where hsp90 binds to eNOS by developing decoy peptides that could disrupt hsp90 interactions with eNOS and determine the critical sites in eNOS that are responsible for hsp90-eNOS interactions and, thus, coupled eNOS activity. Here we identify the location where hsp90 binds to eNOS and show that decoy peptides derived from this site in eNOS are potent inhibitors of stimulated ⅐NO production and eNOS-dependent vasodilation.

EXPERIMENTAL PROCEDURES
Peptide Synthesis-Twelve overlapping peptides (B1-B12, 20-mers) were designed to span the entire region where hsp90 was reported to associate with eNOS (aa 291-420, bovine) (5). TAT protein transduction domain (PTD) (6), PEP1 (7), and eNOS-derived peptides were synthesized using Fmoc (N-(9fluorenyl)methoxycarbonyl) chemistry in the Protein, Nucleic Acid Core Laboratory of the Medical College of Wisconsin. All peptides were made with C-terminal amide and N-terminal acetylation and HPLC-purified, and predicted molecular weights were confirmed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry.
Preparation of Proliferating EC Lysates and Disruption of hsp90 Interactions with eNOS-BAEC were expanded and maintained in RPMI 1640 media containing 10% fetal bovine serum, antibiotics, and mycotics. BAEC cultures were passaged with trypsin-EDTA and used for experiments between passage 5 and 7 (4). Previous studies from this laboratory revealed that proliferating EC had much higher levels of hsp90 association with eNOS than confluent, non-proliferating EC (4). To identify which eNOS-derived peptide disrupts hsp90 interactions with eNOS, the 12 different B peptides were incubated with cell lysates from proliferating BAEC cultures. Proliferating BAEC in 100-mm dishes (10 -20 dishes) prepared as previously described (4) were lysed in modified radioimmune precipitation assay buffer (50 mM Tris HCl, pH 7.5, 1% Nonidet P-40, 0.1 mM EDTA, 0.1 mM EGTA, 0.1% SDS, 0.1% deoxycholic acid, 1ϫ protease inhibitors (Sigma), 1ϫ phosphatase inhibitors (Sigma)). Cell lysates were transferred to 1. 5-ml microcentrifuge tubes, placed on ice, and sonicated 2 times (30 s), and cell debris was isolated by centrifugation (14,000 ϫ g, 10 min, 4°C) (4). Supernatants were removed from the cell debris and pooled, and cell protein was determined in the supernatants using BCA reagent (Pierce). An aliquot of the proliferating EC lysates (300 g in 0.5 ml) was precleared with protein A/G (20 l of a 50% slurry, 2 h at 4°C) and then incubated with each of the individual decoy peptides (final concentration 4.3 M) for 2 h at 4°C. Next, eNOS was immunoprecipitated from the incubations using H32 antibody from BioMol (1 g/100 g of cell lysate) as previously described (4). Protein A/G (50 l of a 50% slurry) was added to isolate the immunoprecipitates. eNOS and its associated proteins were separated by SDS-PAGE (7.5% gel), transferred to nitrocellulose, and immunoblotted for eNOS and hsp90 as described (4).
Effects of eNOS-derived Peptides on Stimulated ⅐NO Production and hsp90 Association with eNOS in BAEC Cultures-BAEC were cultured and maintained in 100-mm culture dishes until confluent. The eNOS-derived peptide B2, which possesses aa sequences that are common to B1 and B3, was incubated for 30 min with PEP1, a protein transduction domain peptide (7) (mole:mole ϭ 1:20) and then added to the BAEC cultures at a final peptide concentration of 5 nM. BAEC cultures were incubated overnight with either nothing (control), PEP1 alone (transduction control), or B2ϩPEP1. These pretreated BAEC cultures were washed and incubated in Hanks' balanced salt solution containing L-arginine (25 M) at 37°C for 15 min to obtain basal ⅐NO production or L-arginine (25 M) ϩ A23187 (5 M), a calcium-ionophore to stimulate eNOS-dependent ⅐NO production. The incubation buffers were removed and stored on ice for immediate nitrite ϩ nitrate analysis by ozone chemiluminescence as described (8). BAEC proteins were lysed in 400 l of MOPS lysis buffer (20 mM MOPS, pH 7.0, 2 mM EGTA, 5 mM EDTA, 30 mM sodium fluoride, 40 mM ␤-glycerophosphate, pH 7.2, 10 mM sodium pyrophosphate, 0.5% Nonidet P-40, 1ϫ protease inhibitors (Sigma), 1ϫ phosphatase inhibitors (Sigma)) and subjected to eNOS immunoprecipitation as described above. eNOS and the levels of associated hsp90 in the immunoprecipitates were determined by Western blot analysis as described above.
Vasodilation Studies-Previous studies showed that vasodilation of facialis arteries of mice was an eNOS-dependent mechanism (3,4,9). To determine the effects of B2 on vasodilation, we redesigned B2 to contain a PTD. The B2 peptide was also shortened to 14 aa (ELVLEVPLEHPTLE) with an Ala added at the N terminus with the TAT PTD (RKKRRQRRR) added N-terminal to the Ala linker, as described by Pagano and co-workers (6). This redesign yielded a new decoy peptide (RKKRRQRRR-A-ELVLEVPLEHPTLE), which we called TSB2. In designing a control peptide, TSB(Ctr) (RKKRRQRRR-A-ALVLAVPLAHPTLA), we reasoned that if Glu were important for binding, then replacing Glu with Ala would result in a peptide that would fail to bind to hsp90 and, therefore, fail to impair vasodilation. To test these ideas, TSB2 and TSB(Ctr) (2.9 M) were incubated with separate facialis arteries from healthy C57BL/6 male mice, whose vasodilation is mediated 100% by eNOS (3,4,9). Ten min later, excess peptides were removed by changing the buffer, and ACh-dose response curves of the preconstricted and pressurized vessels were determined as before (3,4,9).
To determine whether TSB2 peptide blocks the direct actions of ⅐NO, vasodilation responsive curves to an ⅐NO donor, DETA-NONOate, were measured on preconstricted and pressurized facialis arteries from healthy C57BL/6 male mice. The vessels were preincubated with either TSB2 or TSBSCR (scrambled TSB2 peptide, RKKRRQRRR-A-EALTELPVHLEELPV) (2.9 M) for 10 min before the addition of DETA-NONOate.
To determine chronic effects of the decoy peptide on vasodilation in mice ex vivo, we injected C57BL/6 mice (The Jackson Laboratory (Bar Harbor, ME), male, 6 weeks) with TSB2 (1 mg/kg/day) for 2 weeks. Facialis arteries from these mice and PBS-treated C57BL/6 mice were isolated and examined for changes in ACh-dependent vasodilation as described above (3,4,9).

Effects of TSB2 on EC-and eNOS-dependent O 2 . Generation in
Situ-To determine whether TSB2 altered O 2 . generation in native EC on vascular tissues, anesthetized C57BL/6 mice were sacrificed by exsanguination, and aortas were perfused in situ at a rate of 2 ml/min with MOPS buffer containing TSB2 (2.9 M) and dihydroethidine (DHE, 10 M) in the absence and presence of L-NAME (200 M) for 10 min followed by washout of excess, unreacted DHE with MOPS buffer alone. DHE is a cell-permeable probe that upon reaction with O 2 . is converted to a fluorescent 2-hydroxyethidium (2-OH-E ϩ ) product that can be quantified by fluorescent microscopy or HPLC (2,10,11). Aortas were quickly excised and examined for nuclear 2-OH-E ϩ staining (an index of O 2 . generation) using confocal fluorescent microscopy as previously described (2,10). The intensity of the staining in each image taken from the confocal microscope was quantified by NIH image.
Determining Which Glutamic Acids and the Extent to Which These Glu Residues in eNOS Are Important for hsp90 Association-To determine which Glu residues are important to eNOS, we synthesized a series of decoy peptides that had Glu residues systematically replaced by Ala residues. The ability of these peptides to disrupt the eNOS-hsp90 interactions was analyzed by co-immunoprecipitation of eNOS as described before.
Site-directed Mutagenesis of eNOS and Establishing HEK-293 Mutant eNOS Cell Lines-The cDNA for bovine eNOS was mutated at 310, 314, 318, and 323 such that when translated Glu residues were substituted by Ala residues. Four glutamic acid site mutations of eNOS (E310A, E314A, E318A, and E323A) were generated by overlap extension PCR as described by Fulton and Sessa and co-workers (12, 13) using the primers 5Ј-AAGCTTGTCCGCCCCAGCGACATGGGCAACTTGAAG-AGTG-3Ј, 5Ј-TGGGTGCTCAGCCGGTACCTC-3Ј, 5Ј-CAG-TGCCCCTGGCCCACCCCACACTGGCCTGGTTCGCGG-CCCTGGGCCTG-3Ј, and 5Ј-TGGGCCAGGGGCACTGCA-AGGACCAGCGCGGGGGGCAGAACAAAGAGC-3Ј and subcloned via HindIII and KpnI sites. The region of mutation was verified by automated sequencing in the Protein and Nucleic Acid Core Laboratory of the Medical College of Wisconsin. Mutant eNOS 310, 314, 318, and 323 Glu to Ala (eNOS-4A) in pcDNA3 was transfected into HEK-293 cells using Lipofectamine and G418-selected, and individual clones were isolated and examined for levels of mutant eNOS expression.
NADPH Consumption with Mutant eNOS-eNOS protein was extracted from transfected HEK-293 cells according to Osuka et al. (14). Briefly, WT-eNOS and eNOS-4A HEK-293 cells were lysed in Nonidet P-40 lysis buffer (50 mM Tris⅐Cl, 0.2 mM EDTA, 0.2 mM EGTA, 1% Nonidet P-40, 0.1 mM dithiothreitol, 5 nM BH 4 , 1ϫ protease inhibitors, 1ϫ phosphatase inhibitors, pH 7.5) by incubating on ice for 30 min followed with sonicating at 10% power for 30 s twice. Supernatants of the lysates from centrifugation (13,000 rpm, 15 min, 4°C) were incubated with 2Ј-5Ј-ADP-agarose beads (Sigma) overnight at 4°C. The next day the beads were washed twice with the above buffer but without Nonidet P-40, and then eNOS was eluted from the beads using 10 mM NADPH. Eluents were then centrifuged through 30-kDa cutoff Microcon cups twice to remove NADPH. The eNOS crude extracts from this preparation were used for the NADPH consumption assay as described by Vasquez-Vivar et al. (15). Basically, 2.5 g of purified protein in 50 mM HEPES buffer, pH 7.4, containing either nothing else or 0.2 mM CaCl 2 , 20 g/ml calmodulin, 1 mM L-Arg, and 10 M BH 4 were incubated with 300 M NADPH (final concentration). The absorbance at 340 nm was immediately monitored at 1-min intervals for 10 min. The NADPH consumption rate (nmol⅐min Ϫ1 ⅐mg Ϫ1 ⅐protein) was calculated using the molar extinction coefficient of 6.22 mM Ϫ1 ⅐cm Ϫ1 .
⅐NO Generation and hsp90 and Calmodulin (CaM) Association with Mutant eNOS-WT eNOS HEK-293 cells and the mutant eNOS-4A HEK-293 cells were incubated at 37°C in Hanks' balanced salt solution containing L-arginine (25 M) or A23187 (5 M) ϩ L-arginine (25 M) to assess basal and stimulated eNOS-dependent ⅐NO production. The incubation buffers were removed and stored on ice for immediate nitrite analysis by ozone chemiluminescence as described (16). HEK-293 cell proteins were lysed in 400 l of MOPS lysis buffer and subjected to eNOS immunoprecipitation as described above. eNOS and the levels of associated hsp90 and CaM in the immunoprecipitates were determined by Western blot analysis as described above.

L-NAME-inhibitable O 2 . Generation from Mutant eNOS-4A-
To assess L-NAME-inhibitable O 2 . generation from WT and mutant eNOS-4A, we quantified 2-OH-E ϩ production by HPLC (2,10,11). Briefly, HEK-293 cells in 35-mm dishes were washed 2ϫ with Dulbecco's PBS and then incubated in 2 ml of DMEM containing 2% fetal bovine serum, DHE (10 M, final concentration), and A23187 (5 M) with or without L-NAME (1 mM) at 37°C for 30 min. The HEK cells cultures were washed 2ϫ with Dulbecco's PBS and scraped into Dulbecco's PBS, and cells were pelleted by centrifugation (3000 rpm, 5 min) in an Eppendorf microcentrifuge. The cell pellet was lysed and sheared in 0.25 ml of Dulbecco's PBS solution containing 0.1% Triton X-100 using a 0.5-inch 26-gauge needle on a 1-ml syringe. After mixing, an aliquot was removed for protein analysis. Next, 0.5 ml of n-butanol was added to the microcentrifuge tubes and vortexed vigorously for 15 min. The microcentrifuge tubes were centrifuged at 10,000 rpm for 10 min. An aliquot of 450 l of the butanol phase was transferred to a new microcentrifuge tube and dried under a stream of argon gas at 37°C. The dried residue was dissolved in water and quantified by HPLC (HP1100, Agilent Technologies) with fluorescence detection (excitation 510 and emission 595 nm) using authentic standards (2-10 M) in the Free Radical Research Center as described (2,10,11). Care was taken to minimize exposure of reagents and samples to light during incubations and analysis.
Mice-Male C57BL/6 mice were purchased from The Jackson Laboratory. All animal protocols were approved by the Institutional Animal Care and Use Committee of the Medical College of Wisconsin.
Effects of B2 on Basal and Stimulated eNOS-dependent ⅐NO Generation and hsp90 Association with eNOS-Because B2 contained aa sequences that were common to B1 and B3, we examined the effects of B2 on stimulated EC ⅐NO generation using the PTD peptide PEP1 to facilitate B2 uptake (for uptake efficiency, see the supplemental data). The B2:PEP1 mixture significantly inhibited basal and A23187-stimulated ⅐NO production compared with BAEC cultures treated with PEP1 alone or nothing (control) (Fig. 2A, p Ͻ 0.05, n ϭ 3). Immunoprecipitates of eNOS from these test groups reveal the B2:PEP1 mixture markedly decreased hsp90 association not only under basal conditions but also when the cultures were stimulated with A23187 (Fig. 2B, p Ͻ 0.05, n ϭ 3). B2 decreased hsp90 associa-

An eNOS-derived Decoy Peptide Inhibits Vasodilation
DECEMBER 28, 2007 • VOLUME 282 • NUMBER 52 tion to half that of the levels in BAEC cultures treated with PEP1 alone or nothing.
Effects of Acute Disruption of hsp90 Association on EC-and eNOS-dependent Vasodilation-To further pinpoint the region in eNOS that is responsible for association with hsp90, we redesigned a peptide from B1 to B3 regions in eNOS that contain the four Glu residues. This peptide, SB2, with the sequence of ELV-LEVPLEHPTLE, retained the ability to disrupt hsp90-eNOS interaction in BAEC lysates (Fig. 3A). Then a TAT PTD was added to SB2 to increase cellular uptake (for uptake efficiency, see supplemental data). The resultant peptide, TSB2, significantly decreased ACh-induced vasodilation by Ͼ50%, whereas the modified TSB(Ctr) had no effect on vasodilation compared with vehicle control (Fig. 3B, p Ͻ 0.02, n ϭ 6). Failure of TSB2 to inhibit ⅐NO-mediated vasodilation from DETA-NONOate demonstrated that TSB2 disrupts cell signaling at a point upstream of ⅐NO (Fig. 3C, n ϭ 4).
Effects of Chronic Disruption of hsp90 Association on EC-and eNOS-dependent Vasodilation ex Vivo-If hsp90 association with eNOS is important for vascular function in vivo, then chronic TSB2 treatments should inhibit EC-and eNOSdependent vasodilation in the C57BL/6 mice. TSB2 treatments of C57BL/6 mice decreased ACh-induced vasodilation of facialis arteries (Fig. 3D, p Ͻ 0.01, n ϭ 6). During isolation we noted that the facialis arteries from control C57BL/6 mice had thin layers of connective tissue on the advential side of the vessel. When facialis arteries from TSB2-treated mice were examined, connective tissue on the surface of the vessel appeared as large, thick fibers crisscrossing the adventitia (data not shown).
Effects of Disrupting hsp90 Association on eNOS-dependent O 2 . Generation in Native EC-To test the hypothesis that disruption of hsp90 interactions with eNOS uncouples eNOS activity in native vascular EC, we perfused aortas of mice in situ with TSB2 and DHE under a physiological flow rate of 2 ml/min and then rapidly removed the vessels and analyzed the fluorescent intensity in the native EC by fluorescent confocal microscopy. TSB2 markedly increases 2-OH-E ϩ staining, an index of O 2 . generation (Fig. 4, upper right compared with upper left), by a mechanism that could be inhibited in part by L-NAME (Fig. 4,  lower right). On the basis that L-NAME is a substrate-specific inhibitor that blocks both ⅐NO and O 2 . generation from eNOS  (15), these data confirm that acute exposure to TSB2 uncouples eNOS activity to increase eNOS-dependent O 2 . generation in native EC on aortas in situ. The roles of Glu-310, Glu-314, Glu-318, and Glu-323 on eNOS-hsp90 Interactions-When TSB2 had all Glu residues replaced by Ala residues, it lost the ability to block eNOS-dependent vasodilation. Recognizing that the negative charge on each Glu might contribute uniquely to interactions between eNOS and hsp90, we designed 16 peptides with Ala residues systematically replacing Glu residues. The peptides were incubated with lysates from proliferating EC and eNOS-hsp90 interactions determined by Western blot analysis of eNOS immunoprecipitates as above. Lysates without any peptide incubation or with TSBSCR peptide were used as controls. Results showed that the more Glu residues replaced by Ala residues, the less the peptide disrupted eNOS-hsp90 interactions (Fig. 5).

Effects of Glu to Ala Mutations of eNOS (at 310, 314, 318, and 323) on eNOS Association with hsp90 and ⅐NO and O 2 .
Generation-Transfection and selection of HEK-293 cells yielded several clones for analysis. Clones were selected that expressed similar levels of WT-eNOS or mutant eNOS-4A for a balanced comparison in physiological responses. Western blot analysis reveals that clone 4 of the Glu to Ala mutant eNOS contained essentially the same level of eNOS protein as the WT-eNOS clone 2 (Fig. 6A). On the basis of similar protein levels, these clones were selected for further analysis of eNOS-hsp90 association, eNOS-CaM association, NADPH consumption, and stimulated ⅐NO and O 2 . generation. WT-eNOS and Ala substitutions had no effect on vasodilation. C, 10 min exposure of isolated, pressurized facialis arteries to TSB2 (2.9 M) has no effect on DETA-NONOateinduced vasodilation compared with 2.9 M TSBSCR-treated vessels (n ϭ 4). D, C57BL/6 mice were treated with TSB2 (1 mg/kg) or PBS (100 l) for 2 weeks. At the end of this treatment period facialis arteries from TSB2-treated and untreated C57BL/6 mice were isolated, pressurized, and examined for responses to ACh as previously described (9). This line graph shows that TSB2 treatments alter vascular responses to ACh. Not only is ACh-induced vasodilation reduced, but other mechanisms of vasodilation are developing in vessels from TSB2-treated mice based on the fact that L-NAME fails to reduce vasodilation to base line. DHEϩL-NAME in the TSB2ϩL-NAME group as described under "Experimental Procedures." TSB2 (2.9 M) markedly increases 2-OH-E ϩ staining in the nuclei of native vascular EC on TSB2 aortas compared with the control aortas. Although L-NAME does not alter 2-OH-E ϩ staining in EC on controlϩL-NAME aortas compared with the control aortas, L-NAME does decrease 2-OH-E ϩ staining in native EC on TSB2ϩL-NAME aortas compared with the TSB2 aortas. (n ϭ 12). mutant eNOS-4A showed no significant differences in rates of NADPH consumption and CaM association (Fig. 6, B and C). In contrast, the mutant eNOS-4A bound little to no hsp90 in basal or stimulated states, whereas hsp90 was associated with WT-eNOS at low levels under basal conditions and at high levels under stimulated conditions (Fig. 6D). Measurements of basal and stimulated ⅐NO production (measured as nitrite) also revealed that the mutant eNOS-4A is incapable of generating ⅐NO even with stimulation (Fig. 6E).
Having demonstrated that the mutant eNOS-4A does not make ⅐NO and does not associate with hsp90, the next step in testing our hypothesis was to determine whether eNOS-4A generated O 2 . upon stimulation. 2-OH-E ϩ is an oxidized product of the reaction between DHE and O 2 . . The rate of 2-OH-E ϩ accumulation is considered to be an index of the rate of O 2 .
generation. HPLC quantification of 2-OH-E ϩ revealed that A23187 stimulation of cells expressing WT-eNOS in the presence of L-NAME increases O 2 . , whereas A23187 stimulation of cells expressing mutant eNOS-4A in the presence of L-NAME decreases O 2 . production (Fig. 7). Such reciprocal flips in O 2 .
generation in the presence of L-NAME have been used to characterize uncoupled eNOS for more than 10 years (18,19). Taken together, these data (Figs. 6E and 7) demonstrate that WT-eNOS is an enzymatic source of ⅐NO, whereas eNOS-4A is an enzymatic source of O 2 . upon activation.

DISCUSSION
Here we report that hsp90 associates with eNOS at 310 -323 in eNOS. This critical interaction site was identified using eNOS-derived decoy peptides and site-directed mutagenesis. We have shown that both small peptides derived from eNOS and site-specific Glu to Ala mutations in eNOS can disrupt chaperone-dependent signaling with eNOS to inhibit eNOSdependent ⅐NO generation and, in the case of peptides, also vasodilation by a mechanism that actually changes this enzyme function from an ⅐NO synthase into a O 2 . -generating NADPH oxygenase.
One of the first pieces of evidence indicating the importance of 310 -323 in eNOS was the fact that B1, B2, and B3 are all capable of disrupting hsp90 association with eNOS in cell lysates. We used cell lysates from proliferating BAEC to screen the eNOS-derived peptides for two important reasons. First, proliferating BAEC are known to possess a high level of hsp90 association with eNOS that is essential for maintaining a high level of BAEC ⅐NO generation and EC proliferation (3). Second, conducting the studies with lysates from proliferating BAEC rather than intact EC cultures removes confounding variables of cell physiology and metabolism. Thus, findings from these studies are justifiably restricted to protein-protein interactions.
After determining where hsp90 bound to eNOS, we reasoned that it might be advantageous to reduce the size of the B2 peptide and include a PTD to improve cellular uptake. TAT, a protein transduction domain of human immunodeficiency virus Tat protein, has been used to increase translocation of other small decoy peptides for inhibiting NADPH oxidoreductase activity in vascular tissues (6). In addition, it has been reported that TAT facilitates translocation of peptides or proteins through a variety of cells (20 -23). Thus, we decided to use TAT as the PTD for generating decoy peptides that would be rapidly taken up.
Although B1, B2, and B3 all inhibit hsp90-eNOS association, there was no significant difference in the levels of inhibition among the three decoy peptides. It is interesting to note that SB2 (aa 310 -323 of bovine eNOS) has an amino acid sequence that is highly conserved among different species including bovine, human, and mouse. Protein sequence alignment analysis by ClustalW revealed that the 14 amino acids in SB2 region of eNOS are almost identical among the three species, with only 1 amino acid difference. The second amino acid in SB2, which is amino acid 311 in bovine, amino acid 318 in mouse, and amino acid 309 in human, is Leu in bovine and human but Met in mouse. The high level of conservation in the SB2 region indicates the importance of this binding domain to eNOS activity and function.
According to the three-dimensional structure of the dimer of the arginine oxygenase domain of eNOS (Protein Data Bank code 1d0c), the SB2 region is exposed to the aqueous environment and is composed of a short helix followed by a short ␤-pleated sheet, which terminates in another short helix on the crown of the arginine oxygenase domain (Fig. 8). This location likely provides ample access for the chaperone to modulate enzyme function. The calmodulin binding domain is located at aa 490 -510 (24). This is sufficiently distant from the SB2 region that it is unlikely to interfere with calmodulin uptake or activity of eNOS. Likewise, the SB2 region is ϳ535 aa from the autoinhibitory loop (aa 830 -870) (25). Taken together these reports (24,25) and our data suggest that the SB2 region represents a novel site for post-translational regulation of enzyme function by hsp90.
We observed that acutely disrupting hsp90 association with eNOS inhibited eNOS-dependent vasodilation in vitro (i.e. isolated vessels). Interestingly, chronic inhibition of hsp90 association with eNOS in vivo also impaired eNOS-and EC-dependent vasodilation ex vivo (i.e. isolated vessels from a treated mouse). Because TSB2 induces eNOS to switch from generating ⅐NO to O 2 . (Fig. 4), a free radical whose physiological effects are diametrically opposed to those of ⅐NO, disrupting hsp90 association, should have profound physiological effects on vascular function as we showed in Figs. 3, A and C. The fact that substituting Ala residues for Glu residues in TSB2 prevented the peptide from inhibiting hsp90 association and impairing vasodilation (Figs. 3A and 5) underscores the importance of the Glu residues at 310 -323 in eNOS to this critical protein-protein interaction. Systematic replacement of Ala residues for Glu residues in TSB2 reveals Glu-314, Glu-318, and Glu-323 exert the greatest influence with Glu-310 exerting the least influence (Fig. 5). The importance of this domain to eNOS function was confirmed in the mutant eNOS-4A studies (Figs. 6 -7). Results from these experiments not only show where the hsp90 binding domain is located on eNOS but also which Glu residues are important in regulating eNOS hsp90 protein-protein interactions.
Glu to Ala mutations of eNOS at 310, 314, 318, and 323 induce a major change in electrostatic charge in eNOS. Because loss of four Glu residues decreases the overall negative charge of the region, we cannot exclude the possibility that the loss of four Glu residues may directly alter the balance of ⅐NO and O 2 . generation from eNOS-4A in addition to attenuating interactions with hsp90. However, having said this, it should be . production in A23187-stimulated WT-eNOS-transfected HEK-293 cells. MT, HEK-293 cells transfected with eNOS-4A. WT, HEK-293 cells transfected with wild-type eNOS. (n ϭ 8) A, superoxide anion generation in terms of 2-OH-E ϩ (pmol/mg protein) from mutant eNOS-4A and wild-type eNOS-transfected HEK-293 cells stimulated with A23187 and treated with or without L-NAME. B, relative changes in superoxide anion generation ϭ (A23187 ϩ L-NAME) Ϫ (A23187).

An eNOS-derived Decoy Peptide Inhibits Vasodilation
DECEMBER 28, 2007 • VOLUME 282 • NUMBER 52 noted that our data clearly indicate TSB2 disruption of hsp90 interactions with WT-eNOS was sufficient to uncouple eNOS activity. Moreover, the substitution of Ala for Glu does not inhibit the ability of the enzyme to bind CaM and consume NADPH.
Exactly how hsp90 interacts with eNOS to modulate enzyme function is unclear. However, because chaperones are involved in changing protein conformation, hsp90 may hold eNOS in a conformation that favors ⅐NO production, whereas its absence allows the enzyme to take on a slightly different conformation that favors O 2 . generation upon activation. In this way the ability to generate both ⅐NO and O 2 . with which to modulate a full range of ⅐NO activity and oxidative stress resides in a single enzyme. Indeed, eNOS has been shown to generate ⅐NO, O 2 . , and H 2 O 2 and, with ⅐NO reacting with O 2 . , even peroxynitrite (17). It is interesting to speculate that the ability of eNOS to generate such an array of reactive oxygen species may provide unique advantages in regulating EC functions in response to physiological and pathophysiological stimuli.
In summary, we have shown that aa 310 -323 in bovine eNOS, especially Glu-310, -314, -318, and -323, are critical sites for hsp90-eNOS interactions and coupled eNOS activity. Peptide sequences derived from eNOS can be used to design specific decoy peptides for studying the cell biology of this unique protein-protein interaction. Finally, our findings may have importance for understanding the role of other chaperone proteins and how their interactions with eNOS may influence vascular function.