Acute Mechanical Stretch Promotes eNOS Activation in Venous Endothelial Cells Mainly via PKA and Akt Pathways

In the vasculature, physiological levels of nitric oxide (NO) protect against various stressors, including mechanical stretch. While endothelial NO production in response to various stimuli has been studied extensively, the precise mechanism underlying stretch-induced NO production in venous endothelial cells remains incompletely understood. Using a model of continuous cellular stretch, we found that stretch promoted phosphorylation of endothelial NO synthase (eNOS) at Ser1177, Ser633 and Ser615 and NO production in human umbilical vein endothelial cells. Although stretch activated the kinases AMPKα, PKA, Akt, and ERK1/2, stretch-induced eNOS activation was only inhibited by kinase-specific inhibitors of PKA and PI3K/Akt, but not of AMPKα and Erk1/2. Similar results were obtained with knockdown by shRNAs targeting the PKA and Akt genes. Furthermore, inhibition of PKA preferentially attenuated eNOS activation in the early phase, while inhibition of the PI3K/Akt pathway reduced eNOS activation in the late phase, suggesting that the PKA and PI3K/Akt pathways play distinct roles in a time-dependent manner. Finally, we investigated the role of these pathways in stretch-induced endothelial exocytosis and leukocyte adhesion. Interestingly, we found that inhibition of the PI3K/Akt pathway increased stretch-induced Weibel-Palade body exocytosis and leukocyte adhesion, while inhibition of the PKA pathway had the opposite effects, suggesting that the exocytosis-promoting effect of PKA overwhelms the inhibitory effect of PKA-mediated NO production. Taken together, the results suggest that PKA and Akt are important regulators of eNOS activation in venous endothelial cells under mechanical stretch, while playing different roles in the regulation of stretch-induced endothelial exocytosis and leukocyte adhesion.


Introduction
The free radical nitric oxide (NO), produced by endothelial NO synthase (eNOS), is an important vasoactive substance in normal vascular biology and pathophysiology. In addition to its wellknown vascular functions such as vessel dilation and angiogenesis [1,2], NO also regulates some of the key steps in thrombosis and inflammation, including platelet aggregation and monocyte adhesion [3,4]. In endothelial cells (ECs), NO production by eNOS is stimulated by a variety of chemical substances such as vascular endothelial growth factor, thrombin, hydrogen peroxide and bradykinin, as well as by hemodynamic forces, including shear stress, transmural pressure, and mechanical stretch [5][6][7][8][9][10].
While the molecular mechanisms underlying eNOS activation and NO production in arterial ECs in response to chemical stimuli and shear stress have been studied extensively, little is known about the mechanism in venous ECs under continuous stretch. Actually, continuous stretch of venous ECs caused by the abrupt and sustained dilation of veins is frequently observed in patients with portal vein embolization, venous congestion due to acute heart failure, and venous-arterial grafts [11][12][13][14]. In addition, overstretch of venous ECs may be closely associated with venous thrombosis and inflammation [15]. Accumulated evidence suggests that mechanical stretch can induce an inflammatory response in endothelial cells [16,17]. Endothelial exocytosis of Weibel-Palade bodies (WPBs), which contain von Willebrand factor (vWF), interleukin-8 (IL-8) and P-selectin, appears to be one of earliest events in the process of vascular inflammation [18,19]. Recently, we showed that acute hypertensive stretch induces endothelial exocytosis and initiates the pro-thrombotic and proinflammatory responses of ECs [20]. On the other hand, NO production has inhibitory effects on venous thrombosis and inflammation [21,22]. A previous study indicated that NO inhibits the endothelial exocytosis of WPBs via S-nitrosylation of N-Ethylmaleimide-sensitive Factor (NSF) [23]. Our recent study demonstrated that NO is also involved in the inhibition of stretchinduced endothelial exocytosis and vascular inflammation [20]. However, it is still unclear how stretch activates eNOS.
It is known that Ser 1177 phosphorylation leads to increased eNOS activity and NO production [24]. So far, a series of protein kinases, including PKB/Akt, protein kinase A (PKA), PKG, AMPactivated protein kinase (AMPK), mitogen-activated protein kinase (MAPK) and calmodulin-dependent kinase II, has been shown to regulate the Ser 1177 phosphorylation of eNOS [25][26][27][28][29][30]. In addition to Ser 1177 , eNOS has several other potential phosphorylation sites, including Ser 615 and Ser 633 , the phosphorylation of which enhances the activity of eNOS. It has been shown that Ser 615 is phosphorylated in a PKB/Akt-dependent manner while Ser 633 in a PKA-dependent manner [31,32]. These results provide clues for investigating the regulatory pathways of stretch-induced eNOS activation and NO production in venous ECs.
Therefore, we set out to determine whether AMPK, Akt, PKA, and MAPK regulate the Ser 1177 phosphorylation of eNOS and NO production in human umbilical vein endothelial cells (HUVECs) under continuous stretch by using kinase-specific inhibitors and gene-specific shRNAs.

Stretch Stimulates eNOS Activation and NO Production in Venous ECs
We first confirmed the effect of stretch on the Ser 1177 phosphorylation of eNOS and NO production. Early reports suggest that under shear stress, sustained eNOS activation for as long as 30-60 min was detected [26,36]. Therefore, we examined eNOS activation in ECs under stretch for as long as 120 min. The result showed that stretch induced Ser 1177 phosphorylation of eNOS in a time-dependent manner without significant change in the amount of total eNOS protein and had no significant effect on cell death within 2 h ( Fig. 1A and Fig. S1). The Ser 1177 phosphorylation was apparent as early as 2 min after stretch and reached a peak at 30 min, detectable at 60 min and returned to the base level at 120 min. HUVECs were then subjected to mechanical stretch for 15 min for different magnitudes (20%-50%). We found that stretch induced the Ser 1177 phosphorylation in a magnitude-dependent manner ( Figure 1B). As the phosphorylation of eNOS Ser 1177 is critical for NO production, we used DAF-FM (an NO species indicator) to assess the NO levels in HUVECs. The NO levels were significantly increased after stretch compared with the control. In addition, L-NAME (a NOS inhibitor) significantly inhibited the stretch-induced NO production (Fig. 1C, 1D).
Stretch Stimulates Phosphorylation of AMPKa, Akt, Erk1/ 2 and the Activation of PKA Sustained eNOS activation by stretch as mentioned above prompted us to identify the kinases responsible for eNOS activation and their nature. Previous work has shown that AMPKa, Akt, Erk1/2 and PKA phosphorylate eNOS [24,25,29,33]. Therefore, we determined whether these kinases are activated by mechanical stretch. Under static conditions, the phosphorylation of AMPK on Thr 172 , Akt on Ser 473 and Erk1/2 on Thr 202 /Tyr 204 was relatively low. The phosphorylation of these kinases was evident after mechanical stretch but showed different time courses. The phosphorylation of AMPK and Erk1/2 increased as early as 2 min and peaked from 5 to 30 min, while that of Akt was delayed to 15 min, peaked at 30 min, and then returned to the control level at 2 h. In addition, we measured PKA activity using a phospho-PKA substrate antibody, and found that it peaked at 5 min and 15 min after mechanical stretch, returning to the control level at 1 h (Figure 2A AMPKa and ERK Pathways are not Involved in the Regulation of Stretch-induced Ser 1177 Phosphorylation of eNOS and NO Production The above results showing stretch-induced phosphorylation or activation of protein kinases prompted us to consider whether these kinases regulate the Ser 1177 phosphorylation of eNOS and NO production. We thus chose to use kinase-specific inhibitors and gene-specific shRNAs to investigate their regulation of eNOS phosphorylation. First, HUVECs were pretreated for 30 min with 5-50 mM Compound C, a highly-selective inhibitor of AMPKa, and then stretched for 15 min. Compound C had no significant effect on stretch-induced Ser 1177 phosphorylation of eNOS, while it inhibited stretch-induced phosphorylation of AMPKa in a dosedependent manner (Fig. 3A). Then we used specific shRNA targeting the AMPKa1 gene and found that knock-down of AMPKa1 had a similar effect on Ser 1177 phosphorylation of eNOS (Fig. 3B). In addition, PD98059, a selective inhibitor of MEK1/2, inhibited stretch-induced phosphorylation of Erk1/2 in a dose-dependent manner but had no significant effect on the Ser 1177 phosphorylation of eNOS (Fig. 3C). Furthermore, the inhibition of AMPKa and Erk1/2 had no significant effect on stretch-induced NO production (Fig. 3D). These results suggested that the AMPKa and ERK pathways are dispensable for regulation of stretch-induced Ser 1177 phosphorylation of eNOS.

PKA Pathway Mediates Stretch-induced Ser 1177 Phosphorylation of eNOS and NO Production in the Early Phase
To determine whether stretch-induced Ser 1177 phosphorylation of eNOS and NO production is regulated by a PKA-dependent mechanism, HUVECs were pre-treated for 1 h with 10-100 mM H89, a PKA-specific inhibitor, and then stretched for 15 min. Treatment of the cells with H89 significantly attenuated the Ser 1177 phosphorylation of eNOS in a dose-dependent manner, while slightly increasing the phosphorylation of Akt (Fig. 4A). Furthermore, specific shRNAs targeting both PKA catalytic subunits a and b greatly reduced the stretch-induced Ser 1177 phosphorylation of eNOS and slightly increased the phosphorylation of Akt compared with the scrambled control (Fig. 4B), confirming a role of PKA in regulating eNOS activation and the interplay between PKA and Akt. In accord with these results, NO production was significantly inhibited in HUVECs pretreated with 50 mM H89 or expressing shRNAs targeting both PKA catalytic subunits after stretch for 15 min, compared with scrambled control (Fig. 4C). These results suggested that stretch induces Ser 1177 phosphorylation of eNOS and NO production in the early phase (#15 min) in a PKA-dependent manner. Interestingly, after stretch for .15 min Ser 1177 phosphorylation of eNOS and NO production were not completely abolished but still slightly increased in the presence of 50 mM H89 or expressing the shRNAs targeting both PKA subunits, compared with unstretched cells (Fig. 4C-D). These results suggested that stretch induces Ser 1177 phosphorylation of eNOS and NO production in the late phase in a PKA-independent manner.

PI3K/Akt Pathway Mediates Stretch-induced Ser 1177 Phosphorylation of eNOS and NO Production in the Late Phase
The above results showed that the stretch-induced phosphorylation of Akt was relatively delayed. Therefore we speculated that Akt would regulate the late phase of stretch-induced Ser 1177 phosphorylation of eNOS and NO production. First, HUVECs were pretreated for 30 min with 10-100 mM LY294002 (a PI3K inhibitor), and then subjected to stretch for 15 min. The LY294002 treatment abolished the stretch-induced phosphorylation of Akt in a dose-dependent manner but had no significant effect on the phosphorylation of eNOS under these conditions (Fig. 5A). However, at 30 min of stretch, LY294002 attenuated the stretch-induced Ser 1177 phosphorylation of eNOS in a dosedependent manner (Fig. 5B). Then, specific shRNAs targeting both Akt1 and Akt2 significantly reduced the Ser 1177 phosphorylation of eNOS at 30 min of stretch but had no significant effect at 15 min of stretch compared with scrambled control (Fig. 5C). Furthermore, NO production did not change after stretching for 15 min but decreased after stretching for 30 min in HUVECs pretreated with 50 mM LY294002 or expressing the shRNAs targeting Akt1 and Akt2, compared with scrambled control (Fig. 5D). The above results suggested that stretch induces the late phase of eNOS Ser 1177 phosphorylation and NO production in an Akt-dependent manner.

Stretch Induces Ser 633 Phosphorylation of eNOS in a PKAdependent Manner and Ser 615 Phosphorylation in a PI3K/ Akt-dependent Manner
Previous work has shown that there are other potential phosphorylation sites in eNOS, including Ser 633 and Ser 615 [31,32]. Therefore, we determined whether these two sites are phosphorylated by mechanical stretch. Under continuous stretch, the Ser 633 phosphorylation increased as early as 2 min and peaked at 15 min, while the Ser 615 phosphorylation was relatively delayed, peaked from 15 min to 60 min (Fig. 6A). Then, we used the above kinase-specific inhibitors to determine the role of these kinases in regulation of stretch-induced Ser 633 and Ser 615 phosphorylation of eNOS. We found that inhibition of the PKA pathway using H89 abolished the stretch-induced Ser 633 phosphorylation but did not attenuate the stretch-induced Ser 615 phosphorylation (Fig. 6B). Furthermore, inhibition of the PI3K/Akt pathway using LY294002 abolished the stretch-induced Ser 615 phosphorylation but did not affect the stretch-induced Ser 633 phosphorylation (Fig. 6C). The above results demonstrated that the PKA pathway mediates stretch-induced Ser 633 phosphorylation of eNOS while the PI3K/Akt pathway mediates stretch-induced Ser 615 phosphorylation.

eNOS Activation and NO Production Negatively Regulate Stretch-induced WPB Exocytosis and Leukocyte Adhesion
NO has been shown to inhibit endothelial WPBs exocytosis, an early event in leukocyte adhesion [23]. We thus examined the effect of eNOS activation and NO production on stretch-induced endothelial WPB exocytosis and leukocyte adhesion by using chemical inhibitors and gene-specific RNA knockdown ( Fig. 7A and 7C). Stretch significantly enhanced endothelial WPB exocy- tosis and leukocyte adhesion, consistent with our recent finding [20]. The increase in NO production by pretreatment of HUVECs with SNAP, an NO donor that provides exogenous NO, significantly attenuated the stretch-induced exocytosis and leukocyte adhesion. In contrast, the decrease in NO production by pretreatment of HUVECs with L-NAME, an inhibitor of NO production, intensified both of these processes (Fig. 7A, C). Consistently, the levels of stretch-induced exocytosis and adhesion were significantly intensified in HUVECs expressing the shRNAs of eNOS compared with scrambled control (Fig. 7B, D). Taken together, these results suggested that eNOS activation and NO production exert negative feedback on stretch-induced WPB exocytosis and leukocyte adhesion.

Inhibition of the PI3K/Akt Pathway Increases Stretchinduced WPB Exocytosis and Leukocyte Adhesion While Inhibiting the cAMP/PKA Pathway has Opposite Effects
Using chemical inhibitors and gene-specific RNA knockdown, we further studied the role of PKA-and PI3K/Akt-mediated eNOS activation and NO production in stretch-induced WBP exocytosis and leukocyte adhesion in endothelial cells. As expected, inactivation of Akt by both LY294002, an inhibitor of the PI3K/Akt pathway, and the shRNAs targeting Akt1 and Akt2, significantly intensified the stretch-induced exocytosis (Fig. 8A, B) and adhesion (Fig. 8C, D). Unexpectedly, the inhibition of PKA by both H89, a PKA-specific inhibitor, and the shRNAs targeting the catalytic subunits (a and b) of PKA significantly attenuated the stretch-induced exocytosis and adhesion (Fig. 8). The cAMP/PKA pathway has been shown to mediate WPB exocytosis [34]. We thus speculated that the exocytosis-promoting effect of cAMP/ PKA overwhelms the inhibitory effect of PKA-mediated NO production. To confirm this, we further examined the roles of the cAMP/PKA pathway in eNOS activation and in WPB exocytosis of ECs under stretch. First, HUVECs were pretreated with 5-50 mM Rp-cAMP (a cAMP competitor) for 1 h, and then stretched. We found that Rp-cAMP inhibited the stretch-induced exocytosis even though the Ser 1177 phosphorylation of eNOS was attenuated (Fig. 9A, B). Consistently, 0.1-10 mM 8-Br-AMP, an agonist for raising cAMP, significantly increased exocytosis as well as the Ser 1177 phosphorylation of eNOS (Fig. 9C, D). These results indicated thatthe exocytosis-promoting effect of cAMP/PKAis stronger than the exocytosis-inhibiting effect of PKA-mediated NO production in ECs under stretch.

Discussion
The most important finding in the current study was that in venous ECs, mechanical stretch induced Ser 1177 phosphorylation of eNOS and NO production via the PKA and PI3K/Akt pathways in a time-dependent manner. The PKA pathway regulates Ser 1177 phosphorylation of eNOS and NO production in the early phase and the PI3K/Akt pathway in the late phase. In addition, the PKA pathway mediates stretch-induced Ser 633 phosphorylation of eNOS, while the PI3K/Akt pathway mediates stretch-induced Ser 615 phosphorylation in ECs. Although stretchinduced NO production acts as a negative feedback on stretchinduced WBP exocytosis and leukocyte adhesion, the PKA pathway shows overwhelming positive regulation while the PI3K/Akt pathway still shows negative regulation.
Our conclusion that PKA and Akt kinases play distinctive roles in the activation of eNOS in a time-dependent manner is based on the following lines of evidence: (1) stretch-induced activation of PKA pathway occurred relatively earlier than Akt pathway (Fig. 2); (2) inhibition of PKA pathway by kinase-specific inhibitor as well as gene-specific shRNAs significantly attenuated Ser 1177 phosphorylation of eNOS by stretch in the early phase (#15 min), whereas inhibition of Akt pathway by the inhibitor as well as genespecific shRNAs decreased Ser 1177 phosphorylation of eNOS in the late phase (.15 min) (Fig. 4, 5 ); (3) PKA-mediated Ser 633 phosphorylation of eNOS by stretch occurred relatively earlier (from 2 to 15 min), compared to Akt-mediated Ser 615 phosphorylation of eNOS (from 15 to 60 min) (Fig. 6). Taken together, stretch-induced activation of eNOS in ECs is controlled through such time-dependent coordinated regulation of PKA and Akt.
One possible explanation for the biphasic response is that the difference of the two signaling pathways is caused different rates of activation by upstream pathways. Previous work has shown that VEGFR2 and GPCRs are components of the mechanosensor complex in ECs [35,36]. We showed that a cAMP antagonist attenuates stretch-induced eNOS activation, and it is conceivable that the cAMP level might increase under continuous stretch via GPCRs. Some reports have shown that VEGFR2-mediated activation of the PI3K/Akt pathway by mechanical stress is relatively delayed (within minutes) [37,38], compared with the GPCR-mediated activation of the PKA pathway that occurs relatively quickly (within seconds) [39,40]. Another possible explanation is that PKA may have higher or prior affinity for eNOS than Akt for Ser 1177 phosphorylation of eNOS. A slight increase of Ser 1177 phosphorylation of eNOS can be detected under stretch for 15 min in HUVECs after inhibition of the PKA pathway, while inhibition of the PI3K/Akt pathway does not attenuate stretch-induced Ser 1177 phosphorylation of eNOS at this time point (Figs. 4 and 5). This result suggested that the activation of PKA plays dominant role in phosphorylating eNOS-Ser 1177 in the early phase (#15 min) and when PKA is inhibited or deactivated (.15 min), Akt replaces PKA and plays the main role in maintaining Ser 1177 phosphorylation of eNOS until the recruitment of Hsp90 in the eNOS complex [26,41]. It should be noted that stretch-induced eNOS activation and NO production by the PKA and Akt pathways were not absolutely separate. In fact, there was still some overlap between activation of the two pathways (Fig. 2).
In addition, our work showed that acute stretch induced Ser 1177 phosphorylation in an AMPK-independent manner, although phosphorylation of AMPK was also significantly increased (Figs. 2  and 3). Accumulating evidence demonstrates that AMPK directly phosphorylates eNOS Ser 1177 [42]. It has been reported that AMPK is also involved in shear stress-dependent eNOS activation [33]. The discrepancy might be due to the different features of shear stress and stretch. Previous work has demonstrated that shear stress is a protective stimulus [43], but over-stretch of ECs injures blood vessels. Thus, there might be different signaling pathways at play, due to the availability and kinetics of competing eNOS-activating kinases, and their reaction rates. In addition, it has also been suggested that AMPK activation alone might not be sufficient to trigger Ser 1177 phosphorylation of eNOS under certain conditions. For example, Thors et al found that eNOS becomes AMPK-responsive under conditions of ATP depletion but not when cellular ATP is high [44]. Thus, care should be taken when interpreting the role of AMPK in eNOS activation.
The present study suggested that stretch-induced eNOS activation and NO production attenuated but did not abolish stretch-induced WPB exocytosis and leukocyte adhesion and acted as an auto-negative feedback for stretch-induced vascular inflammation. Thus we explored the role of cAMP/PKA-and PI3K/ Akt-mediated NO production in stretch-induced WPB exocytosis and leukocyte adhesion. Interestingly, inhibition of the PI3K/Akt pathway increased the stretch-induced exocytosis and adhesion, while inhibition of the PKA pathway unexpectedly had the opposite effect (Fig. 8). Previous work has shown that the cAMP/ PKA pathway also positively regulates WPB exocytosis via RalGDS [34]. It is most likely that the exocytosis-promoting effect of cAMP/PKA overwhelms the exocytosis-inhibiting effect of PKA-mediated NO production. Thus, the effect of PKAmediated NO production is not evident in the regulation of stretch-induced WPB exocytosis and leukocyte adhesion. However, a negative role of PKA-mediated NO production is not excluded. In addition, NO has other physiological actions, such as short-term vessel dilation and long-term apoptosis, which need further investigation.
In conclusion, the current study demonstrated that timedependent coordinated regulation of PKA and Akt kinase pathways is critical for the regulation of eNOS activation and NO production. Our results may provide a novel insight into the protective mechanism against vascular inflammation by mechanical stretch under pathological conditions in the early stage.

RNA Interference
To silence eNOS, AMPKa1, Akt1/2 and PKA (catalytic subunits a and b), we used a commercial lentiviral system from Sigma to deliver short hairpin RNAs (shRNAs). The target and

Cell Culture
The origin and the culture of human umbilical vein endothelial cells (HUVECs) used in this manuscript have been described in the literature [45,46]. The cells were cultured at 37uC in a humidified atmosphere of 5% CO 2 . Cells were used from passages 3 to 6. 293T and HL-60 cells were obtained from the ATCC and cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum.

Biaxial Stretch of Cultured Cells
HUVECs were stretched on a device established for the study of static continuous stretch, which has been described in detail [47]. Serum-starved ECs were cultured on rat-tail collagen-coated silicone elastic membrane (Specialty MFG, MI) in a single-well device and were uniformly stretched by vertical indentation, resulting in sustained homogeneous strain of 20-50% (static control as 0%). All experiments were performed under 50% stretch unless otherwise noted.

ELISA Analysis
Confluent HUVECs were starved in serum-free M199 medium supplemented with 2% BSA for 4 h, and then stimulated by stretch or other factors. The supernatant was harvested and centrifuged at 3600 rpm at 4uC. The concentration of vWF was assessed by the standard sandwich ELISA procedure according to the manufacturer's instructions.

Annexin V/PI Staining
Confluent ECs were washed twice with cold PBS, fixed with 4% PFA supplemented with 0.2 mol/L sucrose for 1 hr at 4uC. After washing three times with PBS, ECs were stained with Annexin V/ PI kits (Invitrogen V13241) according to the protocol and incubated at room temperature for 15 min and then washed another three times. ECs were also stained with DAPI.

Virus Preparation and Infection
Preparations of lentiviruses were made in 293T cells. Fortyeight hours after the cells were transfected, the virus-containing supernatant was harvested. Three milliliters of supernatant, mixed with 3 mL fresh M199 medium containing a final concentration of 8 mg/mL polybrene, was used to infect ECs as previously described [37]. The medium was replaced with normal M199

Measurement of Nitric Oxide Production
Confluent HUVECs were starved in serum-free M199 medium supplemented with 2% BSA for 3 h, and incubated with 5 mM DAF-FM diacetate in phenol red-free DMEM for 30 min at 37uC. Then the cells were washed gently with PBS and stimulated by stretch. The real-time changes in DAF fluorescence were recorded though a 406oil-immersion lens and analyzed by laser scanning confocal microscopy (Zeiss, Germany).

In vitro Leukocyte Adhesion Assay
Confluent HUVECs were starved in serum-free M199 medium supplemented with 2% BSA for 4 h, and then stimulated by stretch for 30 min. Then the supernatant was discarded and 2610 6 HL-60 cells were added to the surface of the endothelial cells. After 30-min incubation, the unbound cells were washed off 3 times with M199 medium, the bound cells were fixed with 4% paraformaldehyde, and images were captured on a Nikon inverted microscope with a 206objective lens. The numbers of bound HL-60 cells in each field were counted and analyzed by ANOVA. Representative results of three independent experiments are shown.

Statistics
Results are expressed as mean 6 S.D. on the basis of triplicate experiments. Statistical analysis was done using ANOVA with Bonferroni's correction. A value of P,0.05 was considered statistically significant.