Roles of 3-Phosphoinositide-dependent Kinase 1 in the Regulation of Endothelial Nitric-oxide Synthase Phosphorylation and Function by Heat Shock Protein 90^*

Endothelial cells rely on the dynamic release of nitric oxide to maintain cardiovascular homeostasis (1). To cope with the continuously changing environment, endothelial cells need to control their nitric oxide production by various mechanisms. Hence, the function of endothelial nitric-oxide synthase (eNOS)¹ is regulated in an exceedingly complex fashion (2, 3). In the past, elevations of intracellular free Ca²⁺ concentrations were thought to play the principal role in eNOS activation and regulation (4). We now know that serine 1179-phosphorylated eNOS can fully function without the rise of cytosolic Ca²⁺ concentrations (5–7). Besides protein phosphorylation, eNOS function is critically modulated by a number of protein-protein interactions (2, 3, 8). For example, by interacting with caveolin-1, eNOS is targeted in the caveolae of endothelial cells and kept in an idle state (9, 10). eNOS was also reported to directly bind with the intracellular domain of certain G-protein coupled receptors such as angiotensin II AT1, bradykinin B2, and endothelin-1 ET_B (3, 11). Binding with these receptors was shown to cause reversible inhibition on eNOS activity. Recent yeast two-hybrid screen identified more eNOS-associated proteins (NOSIP, NOSTRIN, dynamin, etc.) (3, 8, 12, 13). Thus, it has become increasingly clear that either by influencing eNOS subcellular localization or by directly acting on its catalytic process, protein-protein interactions provide a particularly versatile way to modulate eNOS function.

Another intermediate protein that plays crucial roles in eNOS regulation is the 90-kDa heat shock protein (Hsp90) (14, 15). Hsp90 belongs to a group of highly conserved stress proteins expressed in all eukaryotic cells. As one of the most abundant cytosolic proteins (1–2% total cellular protein), Hsp90 functions as a molecular chaperone participating in protein folding and signal transduction (16, 17). Hsp90 was found to associate with eNOS in resting endothelial cells (14). The interaction between Hsp90 and eNOS can be further enhanced by a variety of stimuli such as histamine, bradykinin, vascular endothelial growth factor, shear stress, and estrogen (14, 18). Further structure analysis revealed that Hsp90 binds with the N-terminal oxygenase domain of eNOS (residues 300–400) (19). Binding with Hsp90 significantly increases eNOS activity. Inhibition of Hsp90 function attenuates endothelium-dependent vascular relaxation, suggesting that Hsp90 is coupled with eNOS in vascular tissues and that this coupling augments nitric oxide production (14). The effect of Hsp90 on eNOS is mediated, at least in part, by the enhancement of calmodulin binding affinity to eNOS (20–22). Hsp90 has also been shown to facilitate the ability of Ca²⁺/calmodulin to dissociate the interaction between caveolin-1 and eNOS, thereby reversing the inhibitory action of caveolin-1 on eNOS (23).

In addition to its allosteric action, Hsp90 was found to be crucial in eNOS serine 1179/1177 (bovine/human) phosphorylation (2, 8). Hsp90 was shown previously to associate Akt (24). Further studies demonstrated that Hsp90 serves as a module to recruit Akt to phosphorylate the eNOS serine 1179 residue (19). Indeed, Hsp90 inhibition resulted in decrease of eNOS serine 1179 phosphorylation (25). However, the details on how Hsp90 inhibition decreases eNOS serine 1179 phosphorylation are not fully understood. For example, whether the reduction of eNOS serine 1179 phosphorylation in Hsp90-inhibited cells was caused by the loss of Akt binding to eNOS or by other mechanisms remains unknown. Akt is activated by the phosphorylation of 3-phosphoinositide-dependent kinase 1 (PDK1) (26, 27). Whether the upstream kinase PDK1 involves in the actions of Hsp90 on eNOS phosphorylation and function is also unknown. In the present study, we demonstrate that the Hsp90 inhibition-induced decrease of eNOS serine 1179 phosphorylation is caused by Akt deactivation rather than loss of Akt binding to eNOS. We further reveal that depletion of PDK1 in Hsp90-inhibited cells is primarily responsible for Akt deactivation. We also provide evidence demonstrating that increase of PDK1 expression can prevent Hsp90 inhibition-induced loss of eNOS serine 1179 phosphorylation and accordingly preserve enzyme function.

EXPERIMENTAL PROCEDURES

Materials—Cell culture materials were obtained from Invitrogen. Bovine aortic endothelial cells (BAECs) were from Cell Systems (Kirkland, WA). Geldanamycin and radicicol were purchased from Sigma. MG132 was obtained from BIOMOL Research Laboratories Inc. (Plymouth Meeting, PA). N-benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl ketone was obtained from Calbiochem. 2′,5′-ADP-Sepharose 4B was obtained from Amersham Biosciences. Antibody against eNOS was purchased from BD Transduction Laboratories. Antibodies against phospho-eNOS (serine 1179), Akt, and phospho-Akt (serine 473) were products of Cell Signaling Technology (Beverly, MA). Antibody against phospho-eNOS (threonine 497) was purchased from Upstate Biotechnology (Lake Placid, NY). Antibodies against PDK1, protein phosphatase 2A (PP2A)-Aα/β, PP2A-B56-α, PP2A-C, and β-tubulin were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibody against β-actin was purchased from Sigma. l-[¹⁴C]Arginine was purchased from PerkinElmer Life and Analytical Sciences (Boston, MA). The protease inhibitor tablet was obtained from Roche Applied Science. Calmodulin, NADPH, l-arginine, tetrahydrobiopterin, N-nitro-l-arginine methyl ester, and other reagents were purchased from Sigma unless otherwise indicated.

Cell Culture and Transfection—Human embryonic kidney (HEK) 293 cells (American Type Culture Collection, Manassas, VA) were grown in Dulbecco' modified Eagle' medium with 10% fetal bovine serum (Invitrogen). HEK 293 cells neither contain eNOS mRNA nor express eNOS protein (28). Wild-type bovine eNOS cDNA in mammalian expression vector pcDNA3 was transfected into HEK 293 cells using Lipofectamine and PLUS reagents (Invitrogen) according to the manufacturer' instruction. Transfected cells were cultured in selective media (complete Dulbecco' modified Eagle' medium containing 500 μg/ml G-418). G-418-resistant colonies were isolated with cloning cylinders, trypsinized, and proliferated in selective media containing 500 μg/ml G418. After two cycles of the colony selection process, eight G418-resistent cell lines were obtained, and further characterization showed that they stably expressed eNOS. One of the lines (eNOS.C4) was designated as the eNOS-HEK 293 cell and used in this study.

For some studies, eNOS-HEK 293 cells were transiently transfected with pCMV5-PDK1 or pCMV5-Akt vector (kindly provided by Dr. Brian A. Hemmings from Friedrich Miescher Institute) using FuGENE6 (Roche Molecular Biochemicals) according to the manufacturer' instructions. The expression of appropriate proteins was confirmed after 24 h of transfection.

siRNA—The siRNA oligonucleotides corresponding to human PDK1 (5′-GUCCGCCUGUAAGAGUUCATT-3′) were purchased from Dharmacon, Inc. A nonspecific oligonucleotide (5′-AUUGUAUGCGAUCGCAGACUU-3′) (Dharmacon) was used as a control. In 12-well plates, cells were plated the day before transfection and were grown to 30–50% confluence. siRNA oligonucleotides (100 nm) were transfected into cells with Oligofectamine reagent (Invitrogen). After 48 h of transfection, Western blottings were carried out to examine the knockdown of targeted proteins.

Pull-down Assay—Cells were harvested and lysed on ice for 30 min in lysis buffer (50 mm Tris-HCl, pH 7.4, 100 mm NaCl, 0.5% Nonidet P-40, 50 mm NaF, 1 mm Na₃VO₄, 5 mm sodium pyrophosphate, and protease inhibitor tablet). The cell lysates were centrifuged at 14,000 × g for 15 min, and the supernatants were recovered. Supernatants containing equal amounts of proteins were incubated with 2′,5′-ADP-Sepharose 4B resins (50 μl in 50% slurry) for 2 h at 4 °C.The resins were washed once with regular washing buffer (50 mm Tris-HCl, 100 mm NaCl, 1 mm EDTA, and 0.5% Nonidet P-40), twice with high-salt washing buffer (50 mm Tris-HCl, 500 mm NaCl, 1 mm EDTA, and 0.5% Nonidet P-40), and another time with regular washing buffer. Pulled-down proteins were then eluted by 5-min boiling of the beads in SDS/PAGE buffer and analyzed by Western blotting.

Western Blotting—Cells were lysed on ice for 30 min in modified radioimmunoprecipitation assay buffer containing 50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, 50 mm NaF, 1 mm Na₃VO₄, 5 mm sodium pyrophosphate, and protease inhibitor tablet. Cell lysates were centrifuged at 14,000 × g for 15 min, and the supernatant was recovered. The total protein concentrations were determined by using the detergent compatible protein assay reagent (Bio-Rad). The lysates were denatured by boiling in SDS sample buffer. The proteins were separated by SDS/PAGE on 4–20% gradient gels (Invitrogen) and then transferred to nitrocellulose membranes by using a semidry transfer cell (Bio-Rad). After blocking, the membranes were probed with the appropriate primary antibodies. Membrane-bound primary antibodies were detected using secondary antibodies conjugated with horseradish peroxidase. Immunoblots were developed on films using the enhanced chemiluminescence technique (SuperSignal West Pico; Pierce). Densitometry was performed by using an AlphaImager 3300 gel documentation and image analysis system.

Nitric-oxide Synthase Activity Assay—eNOS activity was measured by the l-[¹⁴C]arginine to l-[¹⁴C]citrulline conversion assay (29). To measure the activity of phospho-eNOS, the assay was performed in the presence of 10 nm Ca²⁺ as reported previously (6, 7). In brief, cells were harvested in homogenate buffer (50 mm Tris-HCl, pH7.4, 2 mm DTT, 50 mm NaF, 1 mm Na₃VO₄, and protease inhibitor mixture) and homogenated by pulse sonication. After centrifugation (14,000 g for 15 min at 4 °C), the pellets were recovered, washed, and resuspened in homogenate buffer. Cell lysates (45 μg of protein) was added to the reaction mixture containing 50 mm Tris-HCl, pH 7.4, 0.5 mm NADPH, 10 nm CaCl₂, 10 μg/ml calmodulin, 10 μm tetrahydrobiopterin, 5 μm l-[¹⁴C]arginine, and 45 μm l-arginine. After 45 min incubation at 37 °C, the reactions were terminated by ice-cold stop buffer. l-[¹⁴C]Citrulline was separated by passing the reaction mixture through Dowex AG 50W-X8 (Na⁺ form; Sigma) cation exchange columns and quantitated by liquid scintillation counting.

Statistics—Data were expressed as mean ± S.E. Comparisons were made using a two-tailed Student' paired or unpaired t test. Differences were considered statistically significant at p < 0.05.

RESULTS

To determine the roles of Hsp90 in maintaining eNOS phosphorylation, we treated the bovine eNOS stably transfected HEK 293 cells with Hsp90 inhibitor geldanamycin (1 μm). As shown in Fig. 1A, geldanamycin time-dependently decreased serine 1179-phophorylated eNOS. This was a highly specific effect because the phosphorylation status of another residue in eNOS, threonine 497, was unchanged. The total eNOS or Hsp90 level in cells was also not affected by geldanamycin. Corresponding to the decrease of serine 1179 phosphorylation, eNOS activity was markedly attenuated in Hsp90-inhibited cells (Fig. 1B). The specificity of the eNOS assay was evidenced by the fact that the classic nitric-oxide synthase inhibitor N-nitro-l-arginine methyl ester (1 mm) completely abolished the nitric-oxide synthase activity measured. Although geldanamycin had been used as an Hsp90 inhibitor to probe the effects of Hsp90 on eNOS, there was a concern that geldanamycin might redox cycle with eNOS (30, 31). To rule out this concern, we examined the effects of radicicol, a non-quinone Hsp90 inhibitor known to be redox insensitive (30). Similar to the effects of geldanamycin, radicicol (10 μm) also selectively reduced eNOS phosphorylation of serine 1179 but not that of threonine 497 (Fig. 1C). These two lines of evidence demonstrated the crucial roles of Hsp90 in maintaining eNOS serine 1179 phosphorylation and function.

To ascertain that our findings from eNOS-HEK 293 cells occurred in native endothelial cells, we determined that the effect of Hsp90 inhibition on eNOS phosphorylation in BAECs. In accordance with the results from eNOS-HEK 293 cells, geldanamycin also selectively decreased serine 1179-phosphorylated eNOS in a time-dependent manner (Fig. 1D). These data confirmed that the eNOS-HEK 293 cell was a valid model to study the regulation of eNOS phosphorylation by Hsp90 in endothelial cells.

Because Hsp90 recruits Akt to phosphorylate eNOS serine 1179, we pulled down eNOS and determined whether alterations of eNOS-associated Akt was responsible for the decrease of serine 1119-phosphorylated eNOS in Hsp90-inhibited cells. As shown in Fig. 2A, eNOS-associated phospho-Akt, the active form of Akt, was markedly attenuated by Hsp90 inhibition. It is interesting that the total levels of Akt associated with eNOS remained unchanged. eNOS-associated Hsp90 was not altered by geldanamycin treatment either. These data indicated that Hsp90-induced decrease of eNOS serine 1179 phosphorylation was caused by Akt deactivation rather than the loss of Akt association with eNOS. We also measured the alterations of phospho-Akt and total Akt levels in the whole lysates of Hsp90-inhibited cells. Similar to the changes in eNOS-associated Akt, the levels of phospho-Akt were reduced, whereas the total Akt levels remained the same (Fig. 2B). In line with the decrease in phospho-Akt, the phosphorylation of cytosolic glycogen synthase kinase 3β was also attenuated (Fig. 2C). In addition to declined kinase activity, elevations of phosphatase can also give rise to the decrease of phosphoproteins. eNOS serine 1179 was reported to be selectively dephosphorylated by PP2A (32). We therefore monitored PP2A expressions in the absence and presence of the Hsp90 inhibitor. PP2A consists of a structure subunit A, a regulatory subunit B, and a catalytic subunit C (33). As shown in Fig. 2D, the expression levels of the three PP2A subunits were all not significantly altered by Hsp90 inhibition. Thus, loss of eNOS serine 1179 phosphorylation in Hsp90-inhibited cells was not caused by the elevated PP2A expression.

Akt is activated by the phosphorylation of PDK1 (26, 27). Because neither total Akt levels in cells nor the amounts of Akt bound with eNOS were changed by Hsp90 inhibition, we then sought to determine whether alterations of PDK1 resulted in Akt deactivation in Hsp90-inhibited cells. As shown in Fig. 3A, both geldanamycin and radicicol time-dependently depleted cytosolic PDK1 concentrations. PDK1 reduction synchronized the decrease of eNOS serine 1179 phosphorylation. Similar effects were also seen in BAECs (Fig. 3B). We then determined by what mechanism PDK1 was depleted. Because the ubiquitin-proteasome system is responsible for the majority of protein degradation in cells, we reasoned that the PDK1 in Hsp90-inhibited cells was degraded by proteasome. Indeed, proteasome blockade by MG132 (10 μm) prevented the loss of PDK1 in Hsp90-inhibited cells. In contrast, treating the cells with the broad-spectrum caspase inhibitor N-benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl ketone had no effect on Hsp90 inhibition-PDK1 depletion (Fig. 3C). Together, these data strongly suggested that Hsp90 inhibition promoted PDK1 degradation in proteasome and lack of PDK1 resulted in Akt deactivation and subsequent loss of eNOS serine 1179 phosphorylation.

To further prove that loss of PDK1 is responsible for Akt deactivation as well as decrease of eNOS serine 1179 phosphorylation, we investigated whether down-regulation of PDK1 can recapitulate the effects of Hsp90 inhibition on Akt and eNOS serine 1179 phosphorylation. siRNA was used to knock down PDK1. As shown in Fig. 4, transfection of PDK1 siRNA dramatically reduced the PDK1 content in cells. As a result, phospho-Akt and serine 1179-phosphorylated eNOS were decreased. Consistent with the effects of Hsp90 inhibition on eNOS, PDK1 knockdown did not change the phosphorylation of eNOS threonine 497. Transfection of the nonspecific control siRNA had no effect on PDK1, phospho-Akt, or eNOS phosphorylation in cells. These results demonstrated that PDK1 knock-down was sufficient to induce Akt deactivation as well as reduction of eNOS serine 1179 phosphorylation.

Finally, we investigated whether overexpression of PDK1 can reverse Hsp90 inhibition-induced loss of eNOS serine 1179 phosphorylation. As shown in Fig. 5A, transfections of the control empty vectors neither affected the normal levels of serine 1179-phosphorylated eNOS nor changed the effects of Hsp90 inhibition on eNOS serine 1179 phosphorylation. Overexpression of PDK1 significantly reversed the reduction of phospho-Akt in Hsp90-inhibited cells. As a result, the levels of serine 1179-phosphorylated eNOS were largely retained. Parallel to levels of serine 1179 phosphorylation, eNOS activity in Hsp90-inhibited cells was significantly improved by PDK1 overexpression (Fig. 5C). In contrast, overexpression of wild-type Akt failed to improved eNOS serine 1179 phosphorylation in Hsp90-inhibited cells (Fig. 5B). Therefore, decreased eNOS activity in cells was not recovered by Akt overexpression (Fig. 5C). These data further confirmed that depletion of PDK1 was the primary cause of the loss of eNOS serine 1179 phosphorylation and function under conditions of Hsp90 inhibition.

DISCUSSION

The early study by co-immunoprecipitations demonstrated that Hsp90 and Akt formed a complex inside cells (24). The interaction between Hsp90 and Akt seemed to be protective to cells because disrupting their association with truncated Akt constructs increased cell sensitivity to apoptotic stimuli. The critical role of Hsp90 in Akt-mediated human eNOS serine 1177 phosphorylation was first revealed by using geldanamycin as an Hsp90 inhibitor (25). With cultured human endothelial cells, geldanamycin was found to block vascular endothelial growth factor-induced eNOS serine 1177 phosphorylation. In this prior study, serum-starved cells were pretreated with geldanamycin and then the stimulating effect of vascular endothelial growth factor on eNOS phosphorylation was detected within a 30-min period. Thus, these studies addressed the crucial role of Hsp90 in the initiating phase of eNOS phosphorylation. To determine whether Hsp90 is also required in maintaining eNOS phosphorylation in the long term, our experiments were performed with bovine eNOS-transfected cells cultured in regular media containing growth factors (10% fetal bovine serum). Under this condition, part of eNOS in cells is already phosphorylated. We found that Hsp90 inhibition by both geldanamycin and radicicol attenuated eNOS serine 1179 phosphorylation in a time-dependent manner. These results demonstrate that Hsp90 is not only critical in initiating eNOS phosphorylation but also indispensable for the maintenance of serine 1179-phosphorylated eNOS.

Although the studies with Hsp90 inhibitor suggested the important roles of Hsp90 in eNOS phosphorylation, the mechanisms underlying the actions of Hsp90 were not revealed until another study elucidating the scaffolding role of Hsp90 for Akt to phosphorylate eNOS (19). By using yeast two-hybrid analysis, Fontana et al. (19) determined that the M region of Hsp90 interacted with the amino terminus of eNOS and Akt. They further proposed that this modular domain brought Akt and eNOS to proximity to facilitate protein phosphorylation. The aim of the present study was to understand the mechanisms underlying Hsp90 inhibition-induced eNOS dephosphorylation. According to the model in which Hsp90 serves as a module to scaffold Akt and eNOS, it seemed natural to assume that Hsp90 inhibition would disrupt Akt-eNOS association and the loss of Akt binding caused the decrease of eNOS serine 1179 phosphorylation. Our findings did not support this assumption. We found that the amounts of Akt bound to eNOS were not significantly changed by Hsp90 inhibition. Different from cancer cells in which geldanamycin treatment was reported to reduce Akt (34), the total Akt levels in eNOS-HEK 293 cells were not affected by Hsp90 inhibitors. Indeed, further increasing Akt levels by overexpression also failed to reverse the down-regulation of eNOS phosphorylation by Hsp90 inhibitors. Instead, we found that eNOS-associated phospho-Akt, the active form kinase, was dramatically reduced by Hsp90 inhibition. Thus, loss of eNOS serine 1179 phosphorylation in Hsp90-inhibited cells was caused by Akt deactivation rather than lack of Akt association to eNOS.

Protein phosphorylation is determined by the integrated actions of kinases and phosphatases (35). Phospho-Akt is known to be dephosphorylated by PP2A (36). Our finding that Hsp90 inhibition had no effect on intracellular PP2A expressions led us to focus on the Akt upstream kinase. Akt is activated by the phosphorylation of PDK1, a serine/threonine kinase ubiquitously expressed in human cells (26, 27). PDK1 seems to exist in an active and phosphorylated configuration under basal physiological conditions. Hence, the overall PDK1 activity in cells is often in accord with its intracellular concentrations (27). Hsp90 inhibition dramatically depleted intracellular PDK1, suggesting that loss of PDK1 may be the primary reason for Akt deactivation and eNOS serine 1179 dephosphorylation. Indeed, PDK1 knockdown by siRNA was sufficient to recapitulate the effects of Hsp90 inhibitors on phospho-Akt and eNOS serine 1179 phosphorylation. Furthermore, overexpression of PDK1 markedly reversed the effects of Hsp90 inhibition on phospho-Akt and eNOS phosphorylation and preserved enzyme function. On the other hand, overexpression of wild-type Akt failed to reverse the effects of Hsp90 inhibition on eNOS phosphorylation. This was expected because there was no shortage of Akt binding to eNOS in Hsp90-inhibited cells. Moreover, under the condition of PDK1 depletion, overexpressed Akt cannot be phosphorylated and therefore remains dormant. The fact that Akt overexpression failed to protect eNOS phosphorylation and activity in Hsp90-inhibited cells further highlighted the key role of PDK1 in the down-regulation of eNOS phosphorylation by Hsp90 inhibition. We noticed that residual amounts of eNOS serine 1179 phosphorylation remained in PDK1 siRNA-treated cells. This suggested that in addition to the dominant role of PDK1-Akt pathway, other kinases might also involve in eNOS serine 1179 phosphorylation under these experimental conditions (3, 8).

Proteasomal inhibition but not caspase blockade reversed PDK1 depletion in Hsp90-inhibited cells, suggesting that Hsp90 stabilized PDK1 by protecting it from proteasomal degradation. This result is consistent with the actions of Hsp90 in other signaling pathways. As a chief intracellular chaperone, Hsp90 has been shown to protect ErbB-2 and receptors of insulin-like growth factor, insulin, and epidermal growth factor from ubiqutin-dependent proteolysis (37, 38). Hsp90 was previously reported to be important in maintaining PDK1 stability and solubility (39). Our studies confirmed the role of Hsp90 in stabilizing PDK1 and further extend the importance of Hsp90-PDK1 interaction in the context of eNOS phosphorylation and regulation.

In summary, the findings described in this study uncovered a previously uncharacterized role of PDK1 in the effects of Hsp90 on eNOS phosphorylation and function. In addition to serving as a module to scaffold Akt to eNOS, Hsp90 also functions to stabilize PDK1 so that sustained Akt activity can be introduced to eNOS. Thus, the maintenance of eNOS serine 1179 phosphorylation and function requires the synergistic actions of Hsp90 on both Akt and PDK1. Hsp90 inhibition depletes PDK1, which leads to Akt deactivation and subsequent loss of eNOS phosphorylation and function.