Identification of Regulatory Sites of Phosphorylation of the Bovine Endothelial Nitric-oxide Synthase at Serine 617 and Serine 635*

Endothelial nitric-oxide synthase (eNOS) is regulated by signaling pathways involving multiple sites of phosphorylation. The coordinated phosphorylation of eNOS at Ser1179 and dephosphorylation at Thr497activates the enzyme, whereas inhibition results when Thr497 is phosphorylated and Ser1179 is dephosphorylated. We have identified two further phosphorylation sites, at Ser617 and Ser635, by phosphopeptide mapping and matrix-assisted laser desorption ionization time of flight mass spectrometry. Purified protein kinase A (PKA) phosphorylates both sites in purified eNOS, whereas purified Akt phosphorylates only Ser617. In bovine aortic endothelial cells, bradykinin (BK), ATP, and vascular endothelial growth factor stimulate phosphorylation of both sites. BK-stimulated phosphorylation of Ser617 is Ca2+-dependent and is partially inhibited by LY294002 and wortmannin, phosphatidylinositol 3-kinase inhibitors, suggesting signaling via Akt. BK-stimulated phosphorylation of Ser635 is Ca2+-independent and is completely abolished by the PKA inhibitor, KT5720, suggesting signaling via PKA. Activation of PKA with isobutylmethylxanthine also causes Ser635, but not Ser617, phosphorylation. Mimicking phosphorylation at Ser635 by Ser to Asp mutation results in a greater than 2-fold increase in activity of the purified protein, whereas mimicking phosphorylation at Ser617 does not alter maximal activity but significantly increases Ca2+-calmodulin sensitivity. These data show that phosphorylation of both Ser617 and Ser635regulates eNOS activity and contributes to the agonist-stimulated eNOS activation process.

(NO), a key regulator of blood pressure, platelet function, and vessel remodeling. Endothelial NOS is regulated by multiple mechanisms involving both protein-protein interactions with several different proteins, including caveolin-1 and Hsp90 (1), and post-translational modifications that include Nmyristoylation, cysteine palmitoylation, and multisite phosphorylation. The two most thoroughly studied phosphorylation sites have been the activation site, human Ser 1177 (bovine Ser 1179 ) in the enzyme's C-terminal tail, and an inhibitory site, human Thr 495 (bovine Thr 497 ) located in the calmodulin (CaM)-binding sequence. Phosphorylation of eNOS at Ser 1177 reduces the enzyme's Ca 2ϩ -dependence (2), increases the rate of electron flux from the reductase domain to the oxygenase domain (3), and increases the rate of NO production (4 -6). In contrast, phosphorylation at Thr 497 decreases eNOS activity by increasing Ca 2ϩ -CaM dependence (2,7,8). Several laboratories have shown that Akt phosphorylation at Ser 1179 mediates an increase in eNOS activity in response to stimulation by vascular endothelial growth factor (VEGF) and insulin-like growth factor-1 (4,9) and fluid shear stress (5,6). Activation of eNOS can be mimicked by a S1179D mutation (4,5); this has been exploited recently by Scotland et al. (10) with the development of an eNOS S1179D adenoviral construct that was able to restore NO-mediated acetylcholine dilation of carotid arteries from eNOS null mice. Adenoviral knock-in with an eNOS S1179A adenoviral construct also restored the acetylcholine response, but this was substantially attenuated at lower concentrations of acetylcholine. Ser 1179 is the target of multiple protein kinases in addition to Akt, including AMP kinase (11), PKA and protein kinase G (PKG) (12), and CaM II protein kinase (8).
Bradykinin (BK) has been reported in one study (13) to attenuate eNOS activity via activation of ERK1/2 based on the use of the mitogen-activated protein kinase inhibitor, PD98059, but in another study PD98059 did not alter BKinduced changes in eNOS phosphorylation (8). Time course studies of eNOS phosphorylation following BK treatment of endothelial cells have revealed transient dephosphorylation of the inhibitory site, Thr 497 , accompanied by phosphorylation of Ser 1179 (8,14). The phosphorylation and dephosphorylation of both these sites is highly coordinated. Protein kinase C signaling acts to increase phosphorylation of Thr 497 and decrease phosphorylation of Ser 1179 , whereas PKA signaling does the reverse (7).
Other sites of phosphorylation of eNOS have been reported, including Ser 116 in response to shear stress (6) and Ser 635 as a target of PKA and PKG in vitro (12). We have found that several protein kinases, including protein kinase C and AMP kinase, phosphorylate both Thr 497 and Ser 1179 in vitro, but only one site is accessible to each kinase in endothelial cells. In addition to Ser 635 , we have now found that PKA also phosphorylates Ser 617 in vitro; this latter site is also an Akt target in endothelial cells. Phosphorylation of these sites in response to BK, ATP, and VEGF shows distinct time courses with transient Ser 617 phosphorylation and more sustained phosphorylation of Ser 635 . Ser 617 phosphorylation is mediated in part by Akt, whereas Ser 635 phosphorylation in response to BK appears to be mediated by PKA. Mutation of the respective phosphorylation sites of eNOS from Ser or Thr to Asp shows that S1179D and S635D mutations increase maximal activity of purified eNOS, whereas a T497D mutation is inhibitory. A S617D eNOS mutant, on the other hand, has a significantly reduced Ca 2ϩ -CaM dependence but has a similar maximal activity to that of wild-type eNOS.
Phosphopeptide Extraction and Purification of Tryptic Digests-Tryptic peptides of eNOS gel bands were prepared as described previously (7) and extracted with consecutive washes in 2% trifluoroacetic acid (TFA), 0.1% TFA with 30% acetonitrile, then 0.1% TFA with 60% acetonitrile. The digest was dried, and peptides were separated by reversed phase chromatography on a Nucleosil C18 5-m, 300-Å (1 ϫ 250 mm) column on an Amersham Biosciences SMART system using a linear 60 min, 0 -80% acetonitrile gradient in 0.1% TFA at 40 l/min. ␥-32 P-Phosphopeptide Mapping-Phosphopeptides were either separated in two dimensions on thin layer cellulose plates by high voltage electrophoresis (HVE) in the first dimension and ascending chromatography in the second dimension or HVE only (17). Phosphopeptides were visualized by phosphorimage analysis.
MALDI-TOF Mass Spectrometry-Tryptic peptides were spotted onto the sample stage with ␣-cyano-4-hydroxy-cinnamic acid. Masses were analyzed using a linear Voyager DE (PerSeptive Biosystems) MALDI-TOF instrument operating in delayed extraction mode.
Dephosphorylation by -Phosphatase--Phosphatase (NEB) was used to dephosphorylate phosphopeptides purified by reversed phase chromatography and native eNOS. Dephosphorylation of eNOS tryptic peptides and protein was performed by incubation with 200 units of -phosphatase in phosphatase reaction buffer comprising 50 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 5 mM dithiothreitol, 0.01% n-octylglucoside, and 2 mM MnCl 2 . After 10 min the peptides were desalted by C18-ZipTip chromatography (Millipore, Bedford, MA) into 5 l of 60% acetonitrile and 0.1% trifluoroacetic acid prior to spotting on the MALDI-TOF mass spectrometry sample stage.
Antiphosphopeptide-specific Antibodies-Rabbit polyclonal antibodies were raised against phosphopeptides based on the amino acid sequence of human eNOS KIRFNpSISCSDPL (Ser 615 ), WRRKRKEpSS-NTDSAGALGTLRFC (Ser 633 ), CRIRTQpSFSLQER (Ser 1177 ), and GITRKKpTFKEVANC (Thr 495 ) in rabbits. Rabbits were immunized with phosphopeptides coupled to Keyhole limpet hemacyanin (KLH) as described previously (2). Antibodies were purified from serum using the corresponding phosphopeptide affinity columns after preclearing with dephosphopeptide affinity columns. Enzyme-liked immunosorbent assays and Western blot analyses confirmed that the purified antibodies did not recognize dephosphorylated eNOS. The sequences of human and bovine eNOS differ within the Ser 615 peptide at p ϩ 1 position with Ile in the human sequence and Val in the bovine sequence, but the phosphospecific antibody recognized both sequences.
Cell Culture-Bovine aortic endothelial cells (BAECs) were passaged from primary cultures and used for experiments during passages 2-4. Cells were maintained in M199 medium supplemented with 10% fetal bovine serum, 5% iron-supplemented calf serum, 20 g/ml L-glutamine, 0.6 g/ml thymidine, 500 international units/ml penicillin, and 500 g/ml streptomycin. Serum-containing medium was replaced by serumfree medium 16 h prior to each experiment.
Immunoblotting-BAECs were washed twice with ice-cold phosphate-buffered saline containing 1 mM Na 3 VO 4 . Cells were then lysed in ice-cold lysis buffer containing 50 mM Tris-HCl, pH 7.4, 10 mM NaF, 1.5 mM Na 4 P 2 O 7 , 1.5 mM Na 3 VO 4 , and 1 mM phenylmethylsulfonyl fluoride. eNOS was partially purified by affinity binding to 2Ј,5Ј-ADP-Sepharose. The phosphorylation status of Ser 617 and Ser 635 of eNOS was then analyzed by immunoblotting with the phosphospecific antibodies.
Expression and Purification of Wild-type and Mutant eNOS Proteins-Construction of a baculovirus pVL1393 plasmid transfer vector containing the wild-type bovine eNOS cDNA was described previously (18). The wild-type construct was used as a template to create point mutations in the eNOS coding sequence by splicing by the overlap extension technique (19). T497D, S617D, S635D, and S1179D mutations were created by replacing the Thr 497 , Ser 617 , Ser 635 , and Ser 1179 codons with Asp codons. The sequences of all mutated constructs were confirmed by the Molecular Biology Core Facility of the Medical College of Georgia. Mutated DNAs in transfer vector were used to prepare high titer baculovirus stocks in Sf9 insect cells. Wild-type and mutant virus stocks were then used to express and purify the various forms of eNOS protein as described previously (18). Enzymes were purified to Ͼ95% homogeneity in buffers containing 2 mM EGTA to remove bound Ca 2ϩ -CaM.
Assay of eNOS Activity-Activities of purified wild-type and mutant forms of eNOS (1 g) were assayed by monitoring the rate of conversion of L-[ 14 C]arginine to L-[ 14 C]citrulline as previously described (18).

Identification of a Novel Akt Phosphorylation Site of eNOS-
Previously, we observed modest phosphorylation of eNOS by Akt in vitro at an unidentified site (9). Here we have identified this site as Ser 615 in the human sequence. Recombinant human eNOS (2.5 M) was phosphorylated by GST-Akt in vitro. Activated GST-Akt was isolated from HEK-293 cell lysates treated with vanadate (100 M) as described under "Experimental Procedures" (15,16). Phosphorylation of eNOS with activated GST-Akt (Akt-V) revealed three phosphopeptides (Fig. 1A). The major spot corresponded to the previously characterized Ser 1177 site (results not shown), but the minor spots corresponded to uncharacterized phosphopeptides. The total tryptic phosphopeptides derived from eNOS phosphorylated with either Akt-V ( 32 P incorporation, 123.2 pmol) or Akt-W ( 32 P incorporation, 4 pmol) were subjected to high voltage electrophoresis (Fig. 1B, lanes 1 (V) and 2 (W)).
Prior to SDS-PAGE of 32 P-phosphorylated eNOS, acrylamide (1%) was added to the sample buffer to form acrylamide adducts (propionamide) of the cysteine residues. Tryptic peptides of eNOS gel bands were separated by reversed phase chromatography and the fractions analyzed by high voltage electrophoresis (Fig. 1B). Fractions containing 32 P-phosphopeptides (fractions 22, 23 and 32, 33) were then analyzed by MALDI-TOF mass spectrometry. Because phosphorylated peptides are 80 mass units larger than unphosphorylated peptides, the spectra were analyzed for 80-unit differences between masses of the tryptic peptides derived from eNOS phosphorylated with either Akt-V or Akt-W. Fraction 33 contained a phosphopeptide of mass 1851.6 present in the tryptic peptides derived from Akt-V-phosphorylated eNOS that was not present in Akt-W-phosphorylated material. The corresponding dephosphopeptide of 1771 mass units was a major peak in fraction 33 of the peptides from Akt-W-phosphorylated eNOS but not the Akt-V-phosphorylated material (Fig 1B). Phosphate release sequencing of fraction 33 revealed phosphorylation in the third cycle, consistent with a Ser at the third position, whereas in a corresponding Lys-C digest, phosphorylation in the fifth cycle was detected, again consistent with the surrounding sequence, for Ser 615 , KIRFNS (results not shown).
The calculated masses of the theoretical eNOS tryptic peptides were generated using Promac software. The theoretical mass for a human eNOS tryptic peptide FNSISCSDPLVSSWR (613-627) was 1775 units (propionamide cysteine adduct). Because of the apparent 4 mass unit discrepancy with the dephosphorylated peptide (1771 observed versus 1775 expected), we analyzed the corresponding synthetic peptide following phosphorylation. The synthetic peptide YKIRFNSISCSDPLVSSWR (609 -627) (2416.6 mass units) cysteine was modified by 1% acrylamide and then subjected to C18-SepPak purification. The modified peptide was phosphorylated, separated from unbound [␥-32 P]ATP by C18-SepPak, dried, and digested with trypsin. Two tryptic phosphopeptides, IRFNpSISZSDPLVSSWR (611-627) (2120 mass units) and FNpSISZSDPLVSSWR (613-627) (1851 mass units), where Z represents a propionamide cysteine and pS represents phosphoserine, were produced from the same site. The phosphopeptide map of the synthetic peptide were resolved in one dimension by HVE and visualized by phosphorimage analysis. C, the tryptic digests were purified by reversed phase chromatography, and an 80-unit mass difference was detected in fraction 33 between the Akt-vanadate (V) and Akt-wortmannin (W) peptides by MALDI-TOF mass spectrometry indicating a phosphopeptide. A peptide of 1851.6 mass units was present in the Akt-vanadate condition but not in the Akt-wortmannin condition; instead, a peptide of 1771.1 mass units was detected that corresponds to the dephosphopeptide. Solid and dashed arrows indicate the phosphopeptide and dephosphopeptide, respectively. D, a synthetic peptide corresponding to the sequence surrounding the human Ser 615 phosphorylation site (609 -627) was phosphorylated by PKA and subjected to trypsin digestion. The peptides were analyzed by phosphopeptide mapping and MALDI-TOF mass spectrometry (E). The phosphopeptides of 1851 and 2120 mass units, indicated by arrows, correspond to the peptides FNpSISCS-DPLVSSWR and IRFNpSISCSDPLVSSWR following cysteine-acrylic acid modification, where pS represents a phosphoserine. (Fig. 1D) closely matched the phosphopeptide maps generated by digestion of phosphorylated eNOS protein (Fig. 1A). Furthermore, the peptide masses obtained by MALDI-TOF mass spectrometry from the synthetic peptide digests (Fig. 1E) were the same as the masses obtained from digested eNOS, confirming that the 4 mass unit difference was due to instrument error. also a PKA phosphorylation site in vitro, shown by phosphopeptide mapping ( Fig. 2A) with phosphopeptides corresponding to 611-627 and 613-627. In addition, another major site of phosphorylation was detected. Recombinant eNOS was used for the phosphorylation site analysis because the phosphopeptide maps were identical to native eNOS. By a combination of phosphopeptide mapping, phosphate release sequencing, and MALDI-TOF mass spectrometry we found that this site corresponded to the human Ser 633 phosphorylation site described previously by Butt et al. (12). Reversed phase chromatography fractions from a tryptic digest of human recombinant eNOS phosphorylated by PKA in the presence of 100 M CaCl 2 and 1 M CaM were analyzed by MALDI-TOF mass spectrometry. In fraction 23, peptides of 1688 and 1844 mass units (Fig. 2B) closely matched the theoretical masses of the related Ser 633 -containing phosphopeptides corresponding to KEpSSNTDSA-GALGTLR (631-646) (1687 units) and RKEpSSNTDSAGAL-GTLR (630 -646) (1843 units), respectively. These peptides fit the phosphate release data with a Ser at the third and fourth positions, respectively, and the expected Lys or Arg at the N terminus of the longer peptide resulting from incomplete tryptic digestion (results not shown). The Ser 615 peptide FNpSISZSDPLVSSWR (1771 units) and Ser 1177 peptide TQpS-FSLQER (1175 units) were also present in fractions 32 and 22, respectively (data not shown). To confirm the presence of phosphopeptides, dephosphorylation with -phosphatase was performed. eNOS tryptic peptides were incubated with and without -phosphatase for 10 min and then ZipTip-purified prior to spotting on the MALDI-TOF mass spectrometer sample stage. Dephosphorylation of fraction 23 generated peptides of 1608 and 1764 mass units derived from Ser 633 (Fig. 2B), whereas in fractions 22 and 32, peptides of 1096 and 1770 mass units were generated corresponding to the Ser 1177 and Ser 615 peptides, respectively (results not shown).
Recombinant human eNOS was phosphorylated by PKA and Akt, Akt-V, or Akt-W and analyzed by immunoblotting with antiphosphopeptide antibodies directed against the Ser 615 , Ser 633 , and Ser 1177 phosphorylation sites, respectively. Blots were also probed with nonphospho-specific anti-eNOS antibody. In eNOS phosphorylated by PKA and Akt-V, the Ser 615 and Ser 1177 residues were detected, whereas Ser 633 was only detected in eNOS phosphorylated by PKA, as expected (Fig. 3).
BK-and ATP-stimulated Phosphorylation of eNOS at Ser 617 and Ser 635 in BAECs-Because BK is one of the most potent eNOS-activating agonists known, we tested its effects on Ser 617 and Ser 635 phosphorylation and showed that both sites were phosphorylated in response to BK treatment. BAECs were treated with BK (1 M) for various times, and cell lysates were prepared. eNOS was partially purified by affinity chromatog- raphy on 2Ј,5Ј-ADP-Sepharose and immunoblotted with the corresponding antiphosphopeptide-specific antibodies. In most experiments, no basal phosphorylation of either site was detected. However, BK stimulated a rapid and transient increase in Ser 617 phosphorylation between 1 and 2.5 min (Fig.  4A). In contrast, Ser 635 phosphorylation was delayed; phosphorylation appeared at 2.5 min and was maintained at maximal levels between 5 and 30 min (Fig. 4B). In this and all other experiments with BAECs described in this study, the lack of effects of treatment on amounts of total eNOS was confirmed by immunoblotting with nonphospho-specific anti-eNOS antibody (Fig. 4C). To determine whether other Gprotein-coupled receptor agonists, such as ATP, have a similar effect to BK on eNOS phosphorylation, BAECs were incubated with ATP (10 M) for various times, and eNOS phosphorylation was analyzed. An almost identical eNOS phosphorylation pattern was shown for Ser 617 , whereas Ser 635 phosphorylation was comparatively delayed; Ser 635 phosphorylation peaked at 15 min and was maintained for 60 min (the longest time tested) (Fig. 5).
VEGF-stimulated Phosphorylation of eNOS at Ser 617 and Ser 635 in BAECs-We also examined the effects of VEGF stim-ulation on phosphorylation of Ser 617 and Ser 635 . BAECs were treated with VEGF (20 ng/ml) for various times, and eNOS phosphorylation was analyzed by immunoblotting. VEGF stimulated a transient phosphorylation of Ser 617 between 2.5 and 10 min, a slightly delayed onset compared with BK signaling, whereas phosphorylation of Ser 635 was maintained at maximal levels between 10 and 30 min (Fig. 6), similar to that observed with BK.
Akt Mediates the BK-stimulated Phosphorylation of Ser 617 but Not Ser 635 -Previously we showed that Akt is activated by BK stimulation of endothelial cells (14). To determine whether BK-stimulated phosphorylation of Ser 617 or Ser 635 involves Akt, we tested the effects of the phosphatidylinositol 3-kinase inhibitor, LY294002, (20) on phosphorylation. Incubation of BAECs with LY294002 (20 M) for 30 min partially suppressed Ser 617 phosphorylation in response to BK (1 M), indicating that Akt phosphorylates this site in response to BK treatment (Fig. 7). In contrast, LY294002 had no effect on BK-stimulated phosphorylation of Ser 635 (data not shown). Similar results to those shown for LY294002 were also obtained with a structurally distinct phosphatidylinositol 3-kinase inhibitor, wortmannin (21) (not shown).
PKA Mediates the BK-stimulated Phosphorylation of Ser 635 but Not Ser 617 -Because Akt is not the protein kinase responsible for BK-stimulated Ser 635 phosphorylation in BAECs, we tested the effects of PKA activation. The phosphodiesterase inhibitor, IBMX (isobutylmethylxanthine), prevents the conversion of cAMP to AMP, leading to accumulation of cAMP and PKA activation. Incubation of BAEC with IBMX (300 M) stimulated phosphorylation of Ser 635 only slightly from 5 min but reached a maximum after 60 min. In contrast, IBMX treatment did not result in phosphorylation at Ser 617 between 1 and 60 min (Fig. 8). To confirm the involvement of PKA in BK signaling to Ser 635 , we tested the effect of the PKA inhibitor, KT5720 (22). BAECs were treated with BK (1 M) for various times with and without pretreatment with KT5720 (500 nM for 30 min). KT5720 completely blocked BK-stimulated Ser 635 phosphorylation, consistent with PKA mediating the BK-dependent phosphorylation of Ser 635 (Fig.  9). KT5720 had no effect on BK-stimulated Ser 617 phosphorylation (data not shown). BK-stimulated Phosphorylation of Ser 617 but Not Ser 635 Is Ca 2ϩ -dependent-Insulin-stimulated phosphorylation of eNOS at Ser 1179 is Ca 2ϩ -independent (23). To examine whether phosphorylation of Ser 617 or Ser 635 is Ca 2ϩ -dependent or -independent, we utilized the intra-and extracellular Ca 2ϩ chelator, BAPTA-AM (10 M for 30 min) prior to stimulation with BK (1 M) for various times. Phosphorylation was analyzed as before. As shown in Fig. 10, BAPTA-AM completely blocked BK-stimulated phosphorylation of eNOS at Ser 617 . In contrast, no effect of BAPTA-AM on Ser 635 phosphorylation was observed (data not shown).
Effects of Mimicking Phosphorylation of eNOS at Ser 617 and Ser 635 on eNOS Catalytic Activity and Ca 2ϩ -CaM-dependence-The effects of phosphorylation of a Ser or Thr residue on activity of an enzyme can be mimicked by mutation of the uncharged Ser or Thr to a negatively charged Asp where the effects of phosphorylation on enzyme conformation and activity are mainly due to the introduction of a negatively charged group. For example, an ϳ2-fold increase in eNOS activity due to phosphorylation at Ser 1179 has previously been shown to be mimicked by creation of a S1179D mutation (3)(4)(5). To determine the effects of phosphorylation of eNOS at Ser 617 and Ser 635 on eNOS catalytic activity and Ca 2ϩ -CaM-dependence, we expressed and purified wild-type eNOS, and Ser 617 3 Asp (S617D) and Ser 635 3 Asp (S635D) mutants of eNOS from a baculovirus expression system by procedures described previously (18,24,25). For comparative purposes, we also expressed and purified S1179D and T497D eNOS mutants. Enzymes were purified to Ͼ95% homogeneity in buffers containing 2 mM EGTA to remove bound Ca 2ϩ -CaM. The activities of equal quantities of equally purified wild-type and mutant forms of eNOS were then determined by monitoring the rate of conversion of L-[ 14 C]arginine to L-[ 14 C]citrulline in the presence of excess cofactors and CaCl 2 (2 mM) and in the presence of either no CaM or increasing concentrations of exogenously added CaM. As shown in Fig. 11A, mimicking phosphorylation of Ser 635 resulted in a greater than 2-fold increase in eNOS activity at saturating concentrations of Ca 2ϩ -CaM, as well as a small increase in Ca 2ϩ -CaM sensitivity. Mimicking phosphorylation of Ser 617 , in contrast, increased Ca 2ϩ -CaM sensitivity but did not significantly affect maximal eNOS activity. Mutation of Ser 1179 increased both Ca 2ϩ -CaM sensitivity and maximal activity (Fig. 11B), as has been shown previously with cell lysates from cells transfected with this mutant (4,5) or for the enzyme purified from an Escherichia coli expression system (3). Mutation of Thr 497 , in contrast, resulted in a large reduction in eNOS activity, consistent with the known role of phosphorylation of this residue in blocking CaM binding to eNOS (2,7,8).
The relative Ca 2ϩ -CaM sensitivities of wild-type and phosphomimetic forms of eNOS were also assessed under conditions in which suboptimal Ca 2ϩ -CaM concentrations were produced by the combination of excess CaM (2 units/l) and limiting concentrations of Ca 2ϩ . When wild-type eNOS was assayed in the presence of EGTA (2 M)-buffered CaCl 2 (1 M), the enzyme possessed only 20% of the catalytic activity detected in the absence of EGTA (data not shown). Equal quantities of wildtype and mutant forms of eNOS were therefore assayed under these conditions of limiting Ca 2ϩ , and relative activities were compared. As shown in Fig. 12, significant differences in Ca 2ϩ -CaM sensitivities were observed for the S617D, S1179D, and T497D phosphomimetics. These results further confirm that phosphorylation of Ser 617 and Ser 1179 increases the Ca 2ϩ -CaM sensitivity of eNOS and that Thr 497 phosphorylation decreases Ca 2ϩ -CaM sensitivity. DISCUSSION In this study, we have identified two novel phosphorylation sites, human Ser 615 and Ser 633 (equivalent to bovine Ser 617 and Ser 635 ), in the putative "CaM autoinhibitory sequence (586 -641)" within the FMN binding domain of eNOS. This sequence is proposed to retain eNOS in an inhibited state that is reversed upon CaM binding (26). Based on these observations, phosphorylation or dephosphorylation of Ser 635 or Ser 617 might influence the interaction of the insert sequence with the CaM-binding domain, thereby regulating the autoinhibition of eNOS. A similar insert is present in neuronal NOS but is not shared with inducible NOS. However, the insert from neuronal NOS lacks the RRKRK motif thought to be important for autoinhibition in eNOS (26,27). This motif is located immediately N-terminal to the Ser 635 site in eNOS.
Ser 615 in human eNOS corresponds to Ser 847 in human neuronal NOS with high sequence conservation surrounding this site (see Sequence 1).
This study is the first to identify the Ser 617 and Ser 635 residues of eNOS as important regulators of eNOS activity. Butt et al. (12) showed that the cGMP-dependent protein kinase II and PKA phosphorylate and activate eNOS, but the relative contribution of Ser 635 phosphorylation to this process was not determined. Previous studies with a S635A mutant showed that this site is not responsible for Akt-dependent activation of eNOS (4). Our data suggest that phosphorylation of Ser 617 and Ser 635 residues of eNOS have distinct roles in the overall agonist-stimulated eNOS activation process. This process appears to involve multiple phosphorylation events as well as changes in eNOS protein-protein interactions (1). In the initial stage of eNOS activation, Ser 617 is phosphorylated, rendering eNOS significantly more susceptible to activation by Ca 2ϩ -CaM. Subsequently, Ser 635 is phosphorylated, increasing eNOS maximal activity to an extent equal to that produced by phosphorylation of Ser 1179 . We and others (28,29) have shown previously that BK stimulation of NO release from endothelial cells peaks at about 5 min but is sustained at a lower level significantly above baseline for at least 20 -25 min. Ser 635 phosphorylation may be responsible for the longer term potentiation of eNOS activation that persists beyond peak activation.