点击显示 收起
【摘要】
Objective- We have previously shown in COS-7 cells that targeting of endothelial NO synthase (eNOS) to the Golgi or plasma membrane (PM) regulates the mechanism and degree of eNOS activation. However, little is known about the functional significance of eNOS targeting in endothelial cells (ECs). The goal of the current study was to isolate these 2 pools of enzyme in ECs and determine their functional significance in response to agonist stimulation and manipulation of membrane cholesterol levels.
Methods and Results- Using an RNA interference strategy, 90% inhibition of eNOS expression and lacked the ability to produce NO. Reconstitution of these eNOS "knockdown" ECs with Golgi- and PM-targeted eNOS restored the ability of ECs to produce NO. Calcium-dependent agonists were the more efficient stimulus for the PM-restricted eNOS in ECs. In contrast, Golgi eNOS was less responsive to both calcium- and Akt-dependent agonists. eNOS restricted to the PM was more sensitive to manipulation of membrane cholesterol levels and was significantly attenuated by modified low-density lipoprotein.
Conclusions- Within ECs, the PM is the most efficient location to produce NO but is more vulnerable to cholesterol levels and modified low-density lipoprotein.
To determine the functional significance of subcellular targeting in endothelial cells, endogenous eNOS was depleted using virally delivered RNAi and reconstituted with Golgi- and PM-restricted eNOS. We found that the PM is the most efficient location to produce NO but is more vulnerable to cholesterol levels and modified LDL.
【关键词】 eNOS cholesterol Golgi plasma membrane RNAi
Introduction
Endothelium-derived NO is an important regulator of cardiovascular homeostasis. The amount of NO produced by the endothelium is tightly controlled by transcriptional and post-translational control of endothelial NO synthase (eNOS). Although numerous studies have shown that eNOS has a protective role in vascular remodeling and atherosclerosis, 1,2 others have shown that dysregulation of eNOS can exacerbate atherosclerotic lesions. 3,4 See page 959
The activity of eNOS is regulated by numerous post-translational modifications, including protein-protein interactions, phosphorylation, and subcellular location. 5 The phosphorylation of eNOS, and in particular the phosphorylation of S1179 by the protein kinase Akt, increases enzyme activity by enhancing reductase activity and calcium sensitivity. 6-8 Within endothelial cells (ECs), eNOS is localized to the plasma membrane (PM) caveolae and lipid rafts 9,10 and is also found on intracellular membranes such as the Golgi complex. 11-13 Mislocalization of eNOS attenuates both agonist-stimulated NO release and eNOS phosphorylation, suggesting that the proper subcellular localization is critical for stimulus-dependent phosphorylation and activation of the enzyme. 12,14-16 Within the endothelium of blood vessels, eNOS is also present at both the Golgi and PM, 13,17 and across vascular beds, differences in eNOS abundance between Golgi and PM suggest a functional role of subcellular targeting.
In a reconstituted system in COS-7 cells, we have recently shown that the subcellular location of eNOS has a profound effect on its ability to produce NO. 12 When restricted to the PM, eNOS is constitutively phosphorylated and highly active. In contrast, Golgi eNOS is hypophosphorylated and produces less NO in response to calcium-dependent agonists. Coexpression of active Akt increases the phosphorylation and activity of the Golgi eNOS but does not influence the activity of the constitutively phosphorylated PM eNOS. These results suggest that in ECs, eNOS at the PM would be more responsive to agonists that elicit large calcium transients and that agonists selectively activating the kinase Akt could preferentially activate the Golgi pool of eNOS. However, the functional significance of eNOS targeting in ECs is not known.
The modulation of membrane cholesterol and exposure of ECs to oxidized low-density lipoprotein (LDL) have been shown to reduce eNOS activity by displacing eNOS from the PM to intracellular sites. 18,19 Therefore, our goals for the current study were to determine the functional significance of eNOS targeting in ECs and to establish whether manipulation of membrane cholesterol or exposure to oxidized LDL differentially influences eNOS activity in the Golgi or PM.
Methods
Cell Culture and Transfection
COS-7 cells and bovine aortic ECs (BAECs) were grown in DMEM containing 100 U/mL penicillin, 100 mg/mL streptomycin, and 10% FCS. BAECs were harvested in-house and passaged from primary cultures. COS-7 cells and BAECs were transfected using Lipofectamine 2000 (Invitrogen).
RNA Interference Screening
RNA interference (RNAi) targets were selected, and small interfering RNA expression cassettes were generated by polymerase chain reaction (Ambion). Six targets were selected using the following primers: (1) 234: Sense: 5'-AACCTACACAAAGTTCTTCACGCGAGG-GAACCGGTGTTTCGTCCT TTCCACAAG -3', Antisense : 5'- CGGCGAAGCTTTTTCCAAAAAAGTTCCCTC GCGTGAAGAACCTACACAAAGTTC -3'; (2) 1545: Sense: 5'- ATGCTACACAAACATGAGTGAGGCAGAGATCCGGTGTTTCGTCCT TTCC-ACAAG -3', Antisense : 5'- CGGCGAAGCTTTTTCCAAAAAAGATCTCTG CCTCACTCATGCTACACAAACATG -3'; (3) 1590: Sense: 5'- GCCCTACACAAAGGCGTACAGGATGGTTGCTCGGTGTTTCGTCCT TTCCACAAG -3', Antisense : 5'- CGGCGAAGCTTTTTCCAAAAAAAGCAACCA TCCTGTACGCCCTACACAAAGGCG -3'; (4) 2304: Sense: 5'- GTCCTACACAAAGA-CTGTGGCCTGAAACATCCGGTGTTTCGTCCT TTCCACAAG -3', Antisense : 5'-CGGCGAAGCTTTTTCCAAAAAAGATG-TTTCAG GCCACAGTCCTACACAAAGACT-3'; (5) 3122: Sense: 5'-GCATGACATTGAGAGCAAAGGCGAACCTTTGCTC-TCAATGTCA TGC-3', Antisense :5'-GCATGACATTGAGA-GCAAAGGTTCGCCTTTGCTCTC AATGTCATGC -3'; (6) 3493: Sense: 5'- TTTCTACACAAAAAATGTCCTCGTGATAGCGC-CGGTGTTTCGTCC TTTCCACAA G -3', Antisense : 5'-CGG-CGAAGCTTTTTCCAAAAAACGCTATCA CGAGGACA-TTTCTACACAAAAAAT-3'.
The negative control (NEGATIVE) was obtained from Ambion (Silencer validated Negative small interfering RNA Control 1).
Generation of Subcellular eNOS-Targeting Fusion Proteins
The cDNAs encoding the eNOS fusion proteins that target specifically the Golgi or the PM have been described previously. 12 To generate an eNOS construct impervious to the RNAi 3122, several silent mutations, which do not change the amino acid sequence, were generated in the wild-type (WT)-eNOS using the following primers: Sense: 5'-GCACGATATCGAAAGCAAAGGGCTGCAGCCCGCCCCCATG-3'; Antisense: 5'-CATGGGGGCGGGCTGCAGCCCTTTGCTTTCG-ATATCGTGC-3'.
The mutated eNOS fragment was subsequently subcloned into the subcellular targeting constructs, and replication-deficient adenoviruses encoding the various eNOS constructs were generated using the Ad-Easy system. DNA sequences were verified by automated DNA sequencing (Genomics Core Facility; Medical College of Georgia).
Ca 2+ Measurements Using Aequorin
These measurements were obtained using methods described previously. 20 Cytosolic- and PM-targeted (CD8-aequorin) aequorin proteins were expressed in knockdown BAECs via adenovirus, and the relative exposure of these probes to calcium over time was calculated as a ratio to the total amount of aequorin remaining in each well under the different conditions (Lmax). Aequorin and its cofactor coelenterazine react in the presence of calcium to emit a photon. This reaction results in the permanent oxidation of coelenterazine to coelenteramide, and therefore the emission of light in response to calcium is a once-only reaction. Thus, the size of the available pool of luminescence decreases constantly over time in direct proportion to the amount of calcium present. We therefore interpreted the decrease in Lmax as being indicative of greater exposure to calcium over time.
Generation of Golgi- and PM-Targeted Calcium-Insensitive eNOS Fusion Proteins
Calcium-insensitive eNOS constructs were based on constructs ( 45/ 14eNOS) identified by Chen et al. 21 Autoinhibitory control elements present on eNOS (594-606/614-645 [ 45] and 1165-1178 [ 14]) but not inducible NOS (iNOS) were deleted. cDNA fragments containing the mutated regions were generated by polymerase chain reaction using the following primers: 5'-GAAG-CGGATCTTGTAACTCTTCATCTCCATCAGGGCAG-3' and 5'-GAGTTACAAGATCCGCTTCAGGTTCTGTGTGTT-CGGACTGGGC-3' for the 45 mutation; 5'-GATGCCGCTCC-TGCAGGGAAAACGTGAGGCGAAAATGTC-3' and 5'-CCTCACGTTTTCCCTGCAGGAGCGGCATCTG-3' for the 14 mutation.
Each fragment was sequentially subcloned into WT, S17, or CAAX eNOS.
Retroviral Generation and Transduction
Retroviruses were generated encoding either an irrelevant sequence (Negative) or 3122 (Clonetech). BAECs were seeded at a density of 1.3 x 10 6 cells/100-mm dish and infected with 10 5 cfu/mL of active retrovirus. Twenty-four hours later, cells were selected for viral uptake using the antibiotic puromycin (0.8 µg/mL) for 10 to 14 days. During the process of puromycin selection, BAECs were maintained in a subconfluent state.
NO Release
Thirty-six hours after transfection or viral transduction, cells were incubated with serum-free medium for 45 minutes (iNOS- and calcium-insensitive eNOS constructs). BAECs expressing different eNOS constructs were stimulated with the agonists thapsigargin (100 nmol/L), angiopoietin (50 ng/mL), or insulin (200 nmol/L) for 25 minutes. Net NO release was calculated by NO-specific chemiluminescence after subtracting unstimulated basal release as described previously. 7
Live Cell Imaging
Retroviral "knockdown" BAECs were seeded at a density of 2.5 x 10 5 cells/3.8 cm 2 well dish and transfected with cDNAs encoding fusion proteins of enhanced green fluorescent protein (EGFP) and monomeric RFP (Roger Tsien, UCSD) as described above. Twenty-four to 48 hours later, cells were replated onto glass-bottomed culture dishes (MatTek). All imaging was performed using the LSM 510 Meta 3.2 Confocal Microscope (Zeiss). Magnification power was set at x 40 with oil.
Cholesterol Extraction and Supplementation
Cholesterol was prepared as a 50 mmol/L stock in 100% ethanol. Methyl-ß-cyclodextrin (CD) was dissolved in warm DMEM, and cholesterol dissolved in ethanol was added to make a final concentration of 4.5 mmol/L CD +0.5 mmol/L cholesterol (9:1) solution. COS cells or BAECs were seeded at a density of 2.5 x 10 5 cells/3.8 cm 2, and 36 hours after transfection or after viral transduction, cells were incubated in serum-free medium containing 5 mmol/L CD to extract membrane cholesterol or CD cholesterol complex (4.5 mmol/L CD +0.5 mmol/L cholesterol; 9:1) to load cholesterol into membranes for 1 hour. 22 After incubation, cells were washed with warm medium and stimulated with ionomycin (1 µmol/L) as described above.
LDL Oxidation
Modified LDL (modLDL) was prepared by incubating fresh human LDL (EMD Biosciences) with 10 µmol/L of CuSO 4 at 37°C for 16 hours as described previously. 18 BAECs were seeded at a density of 2.5 x 10 5 cells/3.8 cm 2. Thirty-six hours after viral transduction, cells were incubated in serum-free medium containing 50 µg/mL modLDL for 40 minutes, then washed with warm medium and stimulated with ionomycin (1 µmol/L) as described above.
Statistical Analysis
Data are expressed as means±SEM. Comparisons were made using ANOVA with a post hoc test. Differences were considered significant at P <0.05.
Results
Screening and Specificity of RNAi Sequences
To determine the most effective sequence for silencing eNOS expression, we screened 6 potential RNAi sequences. COS-7 cells were transfected with a fixed concentration of eNOS (100 ng) and increasing concentrations of RNAi expression cassettes (10 to 300 ng). The relative inhibition of eNOS expression was determined via Western blotting. As shown in Figure 1 A, 1 sequence (3122) was significantly more effective at silencing eNOS expression. This sequence at a concentration of only 30 ng inhibited eNOS 50% and was comparable to 300 ng of the other effective RNAi sequences (234, 1545, and 2304; data not shown). The 3122 sequence ( Figure 1 B) is also compatible with published human and porcine eNOS sequences. To control for possible off-target effects of this sequence, we performed a series of basic local alignment search tool ( BLAST ) searches using the full-length 21-nt sequence and truncated 5' and 3' ends and found that virtually all of the sequences returned were eNOS genes from other species. To generate an eNOS construct impervious to RNAi-mediated gene silencing, several silent mutations were generated in the open reading frame of WT-eNOS. These mutations, while changing the nucleotide sequence, do not alter the amino acid coding for eNOS ( Figure 1 B). To verify that the most effective anti-eNOS RNAi sequence (3122) does not inhibit the expression of mutated eNOS, the 2 constructs were coexpressed as described above. As shown in Figure 1 C, the expression of the mutated eNOS was not inhibited by 10 to 300 ng of RNAi 3122. Therefore, having shown that it is feasible to circumvent the inhibition of endogenous eNOS with RNAi, our next goal was to restore eNOS expression in BAECs.
Figure 1. Inhibition of eNOS expression by RNAi. RNAi sequences designed to specifically recognize endogenous eNOS mRNA were screened in COS-7 cells. A, Cells were cotransfected with 10 to 300 ng of RNAi vs a fixed concentration of eNOS cDNA (100 ng/3.8 cm 2 ). Relative expression of eNOS and GAPDH was determined via Western blotting. B, Strategy for bypassing RNAi inhibition: silent mutations were generated in the eNOS cDNA, which do not the change the amino acid sequence. C, COS cells were cotransfected with RNAi 3122 (10 to 300 ng) vs a fixed amount of exogenous eNOS cDNA containing the silent mutations (100 ng/3.8 cm 2 ).
Generation of a Stable Population of "Knockdown" ECs
To generate a stable population of BAECs with greatly diminished levels of endogenous eNOS ("knockdown"), we used a retroviral RNAi expression system to deliver RNAi 3122 to BAECs. As shown in Figure 2 A, ECs expressing the 3122 RNAi sequence exhibit significantly less eNOS expression relative to cells transduced with an irrelevant RNAi sequence (Negative). Consistent with these findings, the ability of thapsigargin to elicit NO release from these cells was virtually abolished ( Figure 2 A). These cells retained the cobblestone EC morphology (Figure I, left versus right panels, available online at http://atvb.ahajournals.org) and phenotype as determined by the equivalent presence of an EC marker, the Tie-2 receptor ( Figure 2 A, bottom panel). Furthermore, these cells retained the ability to respond to different endothelial specific agonists, as shown by the ability of both vascular endothelial growth factor (VEGF) and the Tie-2 ligand angiopoietin to stimulate equivalent Akt phosphorylation ( Figure 2 B).
Figure 2. Retroviral knockdown of endogenous eNOS in BAECs with retention of EC phenotype. A, BAECs were treated with retroviruses (10 5 cfu/mL) encoding RNAi sequences for eNOS (3122) or an irrelevant sequence (negative). Thapsigargin-stimulated NO release was measured via chemiluminescence (* P <0.05 vs negative; + P <0.05 vs unstimulated controls). The relative expression of eNOS, hsp90, and Tie-2 were determined via Western blot (bottom panels). B, Serum-starved ECs were exposed to angiopoietin and VEGF (50 ng/ml) for 10 and 5 minutes, respectively, and the phosphorylation level of Akt was determined via Western blot.
It has also been reported that RNAi sequences can induce the intracellular interferon-triggered Jak/Stat signaling pathway. 23 To address this, we determined the activation state of the Jak/Stat signaling axis by measuring the levels of Jak/Stat phosphorylation using Western blotting in ECs stably expressing RNAi. We found no difference in the phosphorylation of these molecules between BAECs exposed to a control retrovirus encoding GFP or RNAi 3122 (data not shown).
Subcellular Targeting and Activation of Golgi and Plasma eNOS in ECs
Subconfluent eNOS "knockdown" ECs were transfected with cDNAs encoding eNOS-GFP (WT), GFP-eNOS-CAAX (PM), and GFP-eNOS-S17 (Golgi) together with RFP-NLS (nucleus) cDNA. As shown in Figure 3 A, the WT eNOS is found at both the PM and perinuclear/Golgi region. The PM-targeted eNOS-CAAX is found predominantly at the cell periphery ( Figure 3 C), and the Golgi-targeted eNOS-S17 is exclusively perinuclear ( Figure 3 B). The ability of these 2 separate pools of eNOS (PM versus Golgi) to respond to different stimuli was determined by stimulating the reconstituted "knockdown" BAECs with calcium-elevating agonists (thapsigargin) or agonists that activate Akt (angiopoietin, insulin). The PM-targeted eNOS-CAAX was much more sensitive to transmembrane calcium fluxes compared with WT- or Golgi-targeted S17 ( Figure 3 D). However, the Akt-dependent agonists did not show a preference for the Golgi pool of eNOS and angiopoietin significantly activated the PM eNOS to a greater extent ( Figure 3 D). The ability of "knockdown" BAECs to respond to bradykinin and VEGF were consistent with that of the calcium-dependent agonist thapsigargin in that the PM eNOS generated the greatest amounts of NO (Figure II, available online at http://atvb.ahajournals.org). The equal expression of eNOS transgenes was confirmed by Western blot as shown in Figure 3 D (bottom panel).
Figure 3. Subcellular targeting and agonist activation of Golgi- and PM-targeted eNOS in "knockdown" BAECs. Retroviral eNOS "knockdown" BAECs were cotransfected with WT eNOS-GFP, GFP-eNOS-CAAX, or GFP-eNOS-S17 together with a nuclear-driven RFP fusion protein, and live cells were visualized for GFP or RFP using confocal microscopy (A through C). A, WT eNOS is present at the PM and Golgi (arrow). B, Golgi targeting of eNOS-S17. C, eNOS-CAAX displays prominent peripheral membrane staining. D, Cells were exposed to calcium-dependent (thapsigargin; 100 nmol/L) and Akt-dependent (angiopoietin 50 ng/mL and insulin 200 nmol/L) agonists for 25 minutes, and NO release was measured via chemiluminescence. Relative expression of eNOS constructs were determined via Western blot with hsp90 and caveolin-1 (cav-1) used as loading controls (bottom panel). The data are presented as mean±SE (n = 6). * P <0.05 vs the unstimulated control; + P <0.05 vs the WT enzyme.
Mechanisms Underlying Differences Between Golgi (S17)- and PM (CAAX)- Targeted eNOS in BAECs
To identify mechanisms that account for the differences between Golgi and PM eNOS, we first investigated whether there are differences in eNOS phosphorylation. As shown in Figure 4 A, the PM eNOS is more heavily phosphorylated on S1179, S617, and S116 compared with the Golgi eNOS. To assess whether Akt can preferentially activate the Golgi or PM eNOS, we transduced BAECs with an adenovirus expressing constitutively active Akt (myr-Akt). Coexpression of Akt stimulated the release of more NO from cells expressing the PM eNOS relative to the Golgi ( Figure 4 B). To determine whether the location of eNOS affects protein-protein interactions, we immunoprecipitated eNOS from "knockdown" BAECs expressing Golgi and PM eNOS. As shown in Figure 4 C, there was significantly more hsp90 bound to the PM eNOS versus the Golgi eNOS. However, there was no difference in the relative association of caveolin-1.
Figure 4. Mechanisms underlying the differences in activity of Golgi and PM eNOS. Retroviral eNOS "knockdown" BAECs were transduced with adenoviruses encoding Golgi (S17) or PM (CAAX), and the degree of eNOS phosphorylation on serines (S) 116, 617, 635 and S1179 and threonine (T) 497 was determined via Western blot as shown in A. B, "Knockdown" BAECs were transduced with viruses encoding Golgi and PM eNOS with or without constitutively active Akt or control LacZ (50 MOI). NO release was measured by chemiluminescence, and the relative expression of phosphorylated and nonphosphorylated proteins were determined via Western blot. The data are presented as mean±SE (n=5). * P <0.05. C, "Knockdown" BAECs expressing either Golgi (S17) or PM (CAAX) eNOS were lysed, and the various forms of eNOS were immunoprecipitated (IP). Immunoprecipitated proteins were immunoblotted with antibodies against eNOS, hsp90, or caveolin-1. Results are representative of 2 independent experiments.
Effect of Cholesterol and modLDL on the Activity of Golgi and PM eNOS
Retroviral eNOS "knockdown" ECs were reconstituted with adenoviruses encoding Golgi and PM eNOS fusion proteins. Thirty-six hours later, cells were treated with regular medium containing 5 mmol/L CD to extract membrane cholesterol or CD-cholesterol complexes to load more cholesterol into cellular membranes. Manipulation of cellular cholesterol levels had the greatest impact on the PM eNOS. As shown in Figure 5 A, CD significantly reduced NO release from the PM eNOS (40% inhibition), whereas cholesterol supplementation enhanced its activity 2-fold. In contrast, the Golgi pool of eNOS was not significantly affected. Golgi- and PM eNOS-reconstituted BAECs were also incubated in serum-free medium containing 50 µg/mL modLDL for 40 minutes followed by agonist stimulation (1 µmol/L ionomycin). As shown in Figure 5 B, NO production from the PM targeting eNOS construct (CAAX) was significantly decreased after short-term exposure to modLDL compared with the Golgi (S17)- targeted eNOS. To address potential mechanisms underlying the increased vulnerability of the PM eNOS, we determined the relative phosphorylation state of Golgi and PM eNOS under the various conditions. Manipulation of membrane cholesterol did not modify the phosphorylation of S1179 on either the Golgi- or PM-targeted eNOS in unstimulated cells or in the presence of active Akt (Figure IIIA and IIIB, available online at http://atvb.ahajournals.org). We also examined whether cholesterol modifies the relative binding of hsp90 to PM eNOS. As shown in Figure IIIC, there was no significant difference in the association of hsp90 with eNOS in cholesterol-loaded or depleted BAECs. In Figure IIID, we determined whether cholesterol affects the subcellular targeting of PM eNOS to cholesterol-enriched membrane microdomains using sucrose gradient centrifugation. However, there was no significant redistribution of PM eNOS to light or heavy membrane fractions.
Figure 5. Manipulation of membrane cholesterol levels selectively influences PM-targeted eNOS. A, "Knockdown" ECs were reconstituted with Golgi (S17)- and PM (CAAX)-restricted eNOS via adenoviruses (100 mois). Thirty-six hours after transduction, these cells were incubated in serum-free medium with 5 mmol/L methyl-ß-CD or a CD-cholesterol complex (cholesterol [Chol.], 4.5 mmol/L CD +0.5 mmol/L cholesterol) separately for 1 hour. Cells were stimulated with ionomycin (1 µmol/L; 25 minutes). NO release was measured by chemiluminescence, and relative expression of eNOS and hsp90 was determined via Western blot. The data are presented as mean±SE (n = 6). * P <0.05 vs the respective S17 response; + P <0.05 compared with control (Con) CAAX. B, Reconstituted ECs were exposed to modLDL (mLDL; 50 µg/mL) for 40 minutes and stimulated with ionomycin (1 µmol/L; 25 minutes). * P <0.05 vs the respective S17. + P <0.05 vs the control-treated CAAX cells.
Manipulation of Membrane Cholesterol Modifies Calcium-Dependent Activation of PM eNOS
Given that the PM eNOS is highly sensitive to transmembrane calcium fluxes, we next addressed whether cholesterol selectively modifies the activity of PM eNOS by altering the levels of intracellular calcium. BAECs were transduced with adenoviruses encoding cytosolic and PM-targeted aequorin calcium sensors. Cholesterol loading or depletion in BAECs did not influence the relative exposure of cytosolic or PM probes to calcium under basal conditions (data not shown) or after ionomycin stimulation ( Figure 6A and 6 B). To address whether the inherent calcium sensitivity of the PM eNOS had changed, "knockdown" BAECs were transfected with cDNAs encoding novel calcium/calmodulin-insensitive eNOS fusion proteins targeted to both the Golgi and PM. The calcium-insensitive eNOS produces NO constitutively, and activity is not significantly modified by chelation of extracellular calcium with EGTA or elevation with ionomycin. 20 These eNOS constructs also contained the silent mutations described in Figure 1 to avoid RNAi. Thirty-six hours after reconstitution of eNOS, CD or CD-cholesterol complexes did not significantly influence NO production from Golgi- or PM-targeted calcium-insensitive eNOS ( Figure 6 C). Identical results were obtained with the calcium-independent NOS isoform iNOS (data not shown).
Figure 6. Changes in membrane cholesterol do not affect cytosolic or membrane calcium levels or activity of a calcium/calmodulin-independent eNOS construct. A and B, CYT-AEQ (HA-AEQ) and PM-AEQ (CD8-AEQ) were expressed in BAECs via adenovirus. AEQ was reconstituted by incubation with coelenterazine (5 µmol/L; 1 hour) in the presence and absence of 5 mmol/L methyl-ß-CD or a CD-cholesterol complex (Cholesterol, 4.5 mmol/L CD +0.5 mmol/L cholesterol), and cells were stimulated with ionomycin (1 µmol/L) in DMEM containing 1.5 mmol/L free Ca2+. Depletion of the coelenterazine pool was measured by determining Lmax at 0, 5, or 30 minutes. Data are presented as means±SE (n=4 to 8). C, Retroviral "knockdown" ECs were transfected with Golgi- and PM-targeted calcium/calmodulin-independent ( 45/ 14) eNOS in the presence and absence of CD or cholesterol for 1 hour. NO release (45 minutes) was measured by chemiluminescence and the relative expression of eNOS and hsp90 determined by Western blotting. The data are presented as mean±SE (n=6).
Discussion
The presence of eNOS has been reported within PM caveolae and also the perinuclear/Golgi region both in vivo and in vitro. 11-13,17 Because of the existence of these 2 distinct enzyme pools, it is difficult to dissect the regulation and significance of each pool of eNOS to overall NO release in ECs. The central dogma is that for eNOS to be fully functional it must reside within caveolae or lipid rafts at the PM. The function, if any, of the perinuclear/Golgi pool of eNOS in ECs has remained an enigma. To address this hypothesis, the goal of this study was to functionally separate eNOS into Golgi and PM pools in ECs.
Our initial approach was to use cultured ECs from eNOS-/- mice. However, in our hands without immortalization, these cells do not grow well in culture and rapidly lose phenotypic markers and EC morphology. Therefore, we adopted an alternative approach by generating BAECs that were devoid or had greatly reduced levels of endogenous eNOS. To achieve this, we created stable populations of eNOS "knockdown" ECs using retrovirus-mediated integration of anti-eNOS RNAi. To replace the endogenous eNOS with a Golgi- or PM-restricted eNOS fusion protein, we created several silent mutations that circumvented the RNAi-mediated eNOS inhibition. The mutated eNOS constructs, which have the same amino acid sequence as the endogenous eNOS, were then expressed in "knockdown" ECs using adenovirus.
In these reconstituted ECs, we found that the PM-restricted eNOS was much more responsive to transmembrane calcium fluxes compared with the WT- and Golgi-restricted pools of eNOS. These results are consistent with those reported in COS cells 12 and support the concept that the PM is an optimal location for eNOS activity. However, the Akt-dependent agonists angiopoietin and insulin did not show a preference for the Golgi pool of eNOS and released the most NO from the PM eNOS. Previously in COS cells, we reported that constitutively active Akt preferentially activated only the Golgi pool of eNOS. To determine whether BAECs respond similarly to Akt, we coexpressed constitutively active Akt with Golgi and PM eNOS. We found that in the "knockdown" BAECs, Akt elicits the greatest amount of NO release from PM eNOS. Although these findings are at odds with those reported in COS cells, they provide an explanation for the greater ability of angiopoietin and insulin to activate the PM eNOS. Interestingly, we also found that PM-restricted eNOS binds to significantly more hsp90 than the Golgi eNOS.
Previous studies have shown that depletion of membrane cholesterol impairs caveolae structure, induces eNOS translocation from caveolae to intracellular locations, and inhibits eNOS activity. 18 In addition, oxidized LDL can also modify eNOS subcellular targeting and inhibit NO release. 18,19 However, the mechanisms by which modification of membrane cholesterol and oxidized LDL influence eNOS activity are not fully understood, particularly in light of the ability of Golgi or intracellular eNOS to produce equivalent amounts of NO. 12 Therefore, eNOS "knockdown" ECs were reconstituted with Golgi- and PM-restricted eNOS and were exposed to CD or CD-cholesterol complexes and modLDL. We found that the PM location of eNOS was most sensitive to changes in membrane cholesterol. Removal of cholesterol with CD greatly attenuated NO release, whereas cholesterol supplementation increased activity 2-fold. In contrast, the Golgi-targeted S17 eNOS was not affected by manipulation of membrane cholesterol.
To address the mechanisms by which cholesterol modification influences the activity of the PM eNOS, we investigated a number of different post-translational controls on eNOS activity. Alterations of membrane cholesterol did not influence the phosphorylation state, protein-protein interaction, or overall membrane association of PM eNOS. Because the increased activity of PM-restricted eNOS is highly dependent on transmembrane calcium fluxes, our next hypothesis was that cholesterol was influencing the calcium-dependent activation of PM eNOS. To test this, we compared the effect of cholesterol modification on cytosolic versus PM calcium concentrations. We found no change in the basal or ionomycin stimulated calcium levels with either the PM or cytosolic aequorin probe. However, because we were seeing clear changes in the calcium-dependent activation of the PM eNOS, we next targeted a calcium/calmodulin-insensitive eNOS construct to the PM and Golgi and treated ECs with CD or CD-cholesterol complexes. No significant difference in NO production was observed in the calcium-independent eNOS mutants or iNOS constructs targeted to either the Golgi or the PM. Therefore, the mechanism by which cholesterol modifies PM eNOS activity relates to changes in the sensitivity of the PM eNOS to calcium/calmodulin. Because we detected no significant difference in calcium levels under these conditions, we hypothesize that changes in the association of calmodulin must account for these differences.
In summary, we successfully used an RNAi strategy to generate a stable population of eNOS "knockdown" ECs and reconstituted these cells with eNOS fusion proteins that specifically target the Golgi and PM. In response to the Akt-dependent agonists insulin and angiopoietin, ECs produced very little NO and did not show a preference for the Golgi pool of eNOS. In response to thapsigargin, which elevates intracellular calcium, the PM eNOS produced substantially more NO than WT- or Golgi-localized eNOS. However, the PM eNOS was more vulnerable to changes in membrane cholesterol and to modLDL. CD significantly reduced and CD-cholesterol complexes significantly increased NO release from PM eNOS but did not affect the activity of Golgi eNOS. Similarly, modLDL substantially reduced the activity of PM eNOS but not Golgi eNOS. The molecular mechanisms by which eNOS activity is impaired in the endothelial dysfunction associated with atherosclerosis are not fully understood. However, the ability of cholesterol and modLDL to selectively influence the activity of PM eNOS elevates the significance of eNOS subcellular targeting in vivo and raises the important question of whether a Golgi-targeted eNOS would offer more vascular protection than a PM eNOS in models of endothelial dysfunction.
Acknowledgments
This work was supported by National Institutes of Health grant HL74279 (D.F.).
【参考文献】
Rudic RD, Shesely EG, Maeda N, Smithies O, Segal SS, Sessa WC. Direct evidence for the importance of endothelium-derived nitric oxide in vascular remodeling. J Clin Invest. 1998; 101: 731-736.
Kuhlencordt PJ, Gyurko R, Han F, Scherrer-Crosbie M, Aretz TH, Hajjar R, Picard MH, Huang PL. Accelerated atherosclerosis, aortic aneurysm formation, and ischemic heart disease in apolipoprotein E/endothelial nitric oxide synthase double-knockout mice. Circulation. 2001; 104: 448-454.
Ozaki M, Kawashima S, Yamashita T, Hirase T, Namiki M, Inoue N, Hirata K, Yasui H, Sakurai H, Yoshida Y, Masada M, Yokoyama M. Overexpression of endothelial nitric oxide synthase accelerates atherosclerotic lesion formation in apoE-deficient mice. J Clin Invest. 2002; 110: 331-340.
Shi W, Wang X, Shih DM, Laubach VE, Navab M, Lusis AJ. Paradoxical reduction of fatty streak formation in mice lacking endothelial nitric oxide synthase. Circulation. 2002; 105: 2078-2082.
Fulton D, Gratton JP, Sessa WC. Post-translational control of endothelial nitric oxide synthase: why isn?t calcium/calmodulin enough? J Pharmacol Exp Ther. 2001; 299: 818-824.
Dimmeler S, Fleming I, Fisslthaler B, Hermann C, Busse R, Zeiher AM. Activation of nitric oxide synthase in endothelial cells by Akt-dependent phosphorylation. Nature. 1999; 399: 601-605.
Fulton D, Gratton JP, McCabe TJ, Fontana J, Fujio Y, Walsh K, Franke TF, Papapetropoulos A, Sessa WC. Regulation of endothelium-derived nitric oxide production by the protein kinase Akt. Nature. 1999; 399: 597-601.
McCabe TJ, Fulton D, Roman LJ, Sessa WC. Enhanced electron flux and reduced calmodulin dissociation may explain "calcium-independent" eNOS activation by phosphorylation. J Biol Chem. 2000; 275: 6123-6128.
Shaul PW, Smart EJ, Robinson LJ, German Z, Yuhanna IS, Ying Y, Anderson RG, Michel T. Acylation targets endothelial nitric-oxide synthase to plasmalemmal caveolae. J Biol Chem. 1996; 271: 6518-6522.
Sowa G, Pypaert M, Sessa WC Distinction between signaling mechanisms in lipid rafts vs. caveolae. Proc Natl Acad Sci U S A. 2001; 98: 14072-14077.
Fulton D, Fontana J, Sowa G, Gratton JP, Lin M, Li KX, Michell B, Kemp BE, Rodman D, Sessa WC. Localization of endothelial nitric-oxide synthase phosphorylated on serine 1179 and nitric oxide in Golgi and plasma membrane defines the existence of two pools of active enzyme. J Biol Chem. 2002; 277: 4277-4284.
Fulton D, Babbitt R, Zoellner S, Fontana J, Acevedo L, McCabe TJ, Iwakiri Y, Sessa WC. Targeting of endothelial nitric-oxide synthase to the cytoplasmic face of the Golgi complex or plasma membrane regulates Akt- versus calcium-dependent mechanisms for nitric oxide release. J Biol Chem. 2004; 279: 30349-30357.
O?Brien AJ, Young HM, Povey JM, Furness JB. Nitric oxide synthase is localized predominantly in the Golgi apparatus and cytoplasmic vesicles of vascular endothelial cells. Histochem Cell Biol. 1995; 103: 221-225.
Sessa WC, Garcia-Cardena G, Liu J, Keh A, Pollock JS, Bradley J, Thiru S, Braverman IM, Desai KM. The Golgi association of endothelial nitric oxide synthase is necessary for the efficient synthesis of nitric oxide. J Biol Chem. 1995; 270: 17641-17644.
Sakoda T, Hirata K, Kuroda R, Miki N, Suematsu M, Kawashima S, Yokoyama M. Myristoylation of endothelial cell nitric oxide synthase is important for extracellular release of nitric oxide. Mol Cell Biochem. 1995; 152: 143-148.
Jagnandan D, Sessa WC, Fulton D Intracellular location regulates calcium-calmodulin-dependent activation of organelle-restricted eNOS. Am J Physiol Cell Physiol. 2005; 289: C1024-C1033.
Andries LJ, Brutsaert DL, Sys SU. Nonuniformity of endothelial constitutive nitric oxide synthase distribution in cardiac endothelium. Circ Res. 1998; 82: 195-203.
Blair A, Shaul PW, Yuhanna IS, Conrad PA, Smart EJ. Oxidized low density lipoprotein displaces endothelial nitric-oxide synthase (eNOS) from plasmalemmal caveolae and impairs eNOS activation. J Biol Chem. 1999; 274: 32512-32519.
Nuszkowski A, Grabner R, Marsche G, Unbehaun A, Malle E, Heller R. Hypochlorite-modified low density lipoprotein inhibits nitric oxide synthesis in endothelial cells via an intracellular dislocalization of endothelial nitric-oxide synthase. J Biol Chem. 2001; 276: 14212-14221.
Church JE, Fulton D. Differences in eNOS activity because of subcellular localization are dictated by phosphorylation state rather than the local calcium environment. J Biol Chem. 2006; 281: 1477-1488.
Chen PF, Wu KK. Structural elements contribute to the calcium/calmodulin dependence on enzyme activation in human endothelial nitric-oxide synthase. J Biol Chem. 2003; 278: 52392-52400.
Furuchi T, Anderson RG. Cholesterol depletion of caveolae causes hyperactivation of extracellular signal-related kinase (ERK). J Biol Chem. 1998; 273: 21099-21104.
Sledz CA, Holko M, de Veer MJ, Silverman RH, Williams BR. Activation of the interferon system by short-interfering RNAs. Nat Cell Biol. 2003; 5: 834-839.
作者单位：Vascular Biology Center and Department of Pharmacology (Q.Z., J.E.C., D.J., J.D.C., D.F.), Medical College of Georgia, Augusta; and Department of Pharmacology and Molecular Cardiobiology Division (W.C.S.), Boyer Center for Molecular Medicine, Yale University School of Medicine, New Haven, Conn.