Identification of Flow-dependent Endothelial Nitric-oxide Synthase Phosphorylation Sites by Mass Spectrometry and Regulation of Phosphorylation and Nitric Oxide Production by the Phosphatidylinositol 3-Kinase Inhibitor LY294002*

Endothelial cells release nitric oxide (NO) acutely in response to increased laminar fluid shear stress, and the increase is correlated with enhanced phosphorylation of endothelial nitric-oxide synthase (eNOS). Phosphoamino acid analysis of eNOS from bovine aortic endothelial cells labeled with [32P]orthophosphate demonstrated that only phosphoserine was present in eNOS under both static and flow conditions. Fluid shear stress induced phosphate incorporation into two specific eNOS tryptic peptides as early as 30 s after initiation of flow. The flow-induced tryptic phosphopeptides were enriched, separated by capillary electrophoresis with intermittent voltage drops, also known as “peak parking,” and analyzed by collision-induced dissociation in a tandem mass spectrometer. Two phosphopeptide sequences determined by tandem mass spectrometry, TQpSFSLQER and KLQTRPpSPGPPPAEQLLSQAR, were confirmed as the two flow-dependent phosphopeptides by co-migration with synthetic phosphopeptides. Because the sequence (RIR)TQpSFSLQER contains a consensus substrate site for protein kinase B (PKB or Akt), we demonstrated that LY294002, an inhibitor of the upstream activator of PKB, phosphatidylinositol 3-kinase, inhibited flow-induced eNOS phosphorylation by 97% and NO production by 68%. Finally, PKB phosphorylated eNOS in vitro at the same site phosphorylated in the cell and increased eNOS enzymatic activity by 15–20-fold.

Endothelial nitric-oxide synthase (eNOS 1 or type III NOS) is one of three isoenzymes that converts L-arginine to L-citrulline and nitric oxide (NO). Endothelial cells synthesize NO tonically and increase NO production in response to agonists and increased fluid shear stress (FSS). Endothelial NO contributes to blood vessel homeostasis by regulating vessel tone (1), cell growth (2), platelet aggregation (3), and leukocyte binding to endothelium (4). In vivo eNOS is both myristoylated and palmitoylated. These modifications increase eNOS compartmentalization to plasmalemmal caveolae and facilitate release of NO from cells (5)(6)(7). In caveolae, which are small plasmalemmal invaginations that sequester signaling proteins (8), eNOS specifically interacts with the scaffolding protein caveolin-1 through a caveolin (9, 10) binding motif (11), located near the domain that binds Ca 2ϩ /calmodulin. Recent studies suggest that the activity of eNOS is regulated in a reciprocal manner through caveolin-1 inhibition and Ca 2ϩ /calmodulin stimulation (12)(13)(14).
Increased FSS stimulates an increase in free intracellular calcium [Ca 2ϩ ] i from intracellular stores (15,16) leading to a Ca 2ϩ /calmodulin-dependent increase in eNOS activity. However, recent investigations show that increases in [Ca 2ϩ ] i do not fully explain the rapid rise in NO production in response to FSS (17). Exposure of bovine aortic endothelial cells (BAEC) to 25 dynes/cm 2 FSS for 30 s caused a 7-fold rise in NO production and a corresponding 2-fold increase in eNOS phosphorylation, whereas the calcium ionophore A23187 neither caused rapid NO production nor any net increase in eNOS phosphorylation (17). We therefore hypothesized that an increase in FSS could be transduced to increased NO production via activation of a protein kinase cascade resulting in phosphorylation and activation of eNOS.
The objectives of the present work were to elucidate the eNOS sites phosphorylated in response to increased FSS and to identify potential protein kinase mediators of the mechano-transduction of FSS to NO production. The amino acid residues phosphorylated in small proteins may be determined by fractionation of tryptic peptides by high pressure liquid chromatography (HPLC) and mass spectrometry (18). However, tryptic fragmentation of large proteins such as eNOS creates numerous peptides whose rapid elution from HPLC exceeds the scan rate of the triple quadrupole mass spectrometer. This prevents acquisition of sequences, by collision-induced dissociation (CID), of a substantial fraction of the peptides. Additionally, phosphopeptides from proteins isolated from a cell will nearly always exist as minor ions due to sub-stoichiometric phosphorylation compared with nonphosphorylated peptides and will not be scanned as major parent ions. In order to identify phosphorylation sites in eNOS, tryptic phosphopeptides were first enriched by immobilized metal affinity chromatography (IMAC). The peptides were then separated by peak parking solid phase extraction capillary electrophoresis (SPE-CE), which extends the peak analysis time to provide for both increased scan times of parent ions and increased MS/MS analyses of each selected peptide according to optimized CID parameters. The identity of two specific eNOS residues phosphorylated in response to flow was determined, and evidence was obtained linking FSS to eNOS phosphorylation and NO production via activation of phosphatidylinositol 3-kinase (PI3-kinase) and protein kinase B (PKB). We present evidence that PKB phosphorylates eNOS in vitro at the same site phosphorylated in the cell and that the phosphorylation increases eNOS enzymatic activity.
BAEC Culture and Exposure to Fluid Shear Stress-BAEC cultures were established and maintained in culture in medium 199 (Life Technologies, Inc.) supplemented with fetal calf serum as described (17). Cells from passage 3-8 were seeded in 100-mm tissue culture dishes and used at 60 -80% confluency for labeling and shear stress experiments. Cells were exposed to laminar FSS in a cone and plate viscometer as described (19).
Cell Labeling, Lysis, and Immunoprecipitation of eNOS-BAEC at 60 -80% confluence in 100-mm dishes were washed twice with phosphate-free DMEM and labeled with 1 mCi/ml [ 32 P]orthophosphate for 3 h in phosphate-free DMEM containing 10% dialyzed fetal bovine serum. Na 3 VO 4 (200 M) was added to the culture medium for 10 min prior to harvest. Cells were maintained in static culture or exposed to 12-15 dynes/cm 2 FSS for the indicated times in a cone and plate viscometer (19). Cells were rapidly washed with ice-cold phosphatebuffered saline and lysed in 0.7 ml of phosphate-buffered saline containing 1% Triton X-100, 50 mM ␤-glycerophosphate, 200 M Na 3 VO 4 , 10 g/ml leupeptin, 10 g/ml soybean trypsin inhibitor, 2 mM benzamidine HCl, and 2 mM EDTA. Lysates were centrifuged for 5 min at 14,000 rpm at 4°C. Equivalent amounts of protein as determined by Bradford assay were used for all immunoprecipitations. The supernatants were precleared as described (20) and incubated for 4 h at a final NaCl concentration of 400 mM (21) on a rocker with 60 l of a 20% suspension (v/v) of protein A-agarose and 5 g of anti-eNOS monoclonal antibody. The immune pellets were washed four times with 0.8 ml of lysis buffer without protease inhibitors, boiled in sodium dodecyl sulfate (SDS) sample buffer, and the eluates from the immune pellets were stored at Ϫ20°C.
SDS Gel Electrophoresis, Western Blotting, and Transfer of Proteins to Membranes-Immunoprecipitated eNOS in SDS sample buffer was fractionated on a 9% SDS-polyacrylamide gel in a Bio-Rad Mini-PRO-TEAN II electrophoresis apparatus. The gel was soaked in 100 ml of transfer buffer (24 mM Tris base, 181 mM glycine) for 10 min, and the proteins were transferred to nitrocellulose at 100 V for 70 min. For Western blotting, proteins (30 g/lane) were fractionated on a 9% SDSpolyacrylamide gel and transferred to nitrocellulose. After blocking for 1 h in 100 mM NaCl, 100 mM Tris-HCl, pH 7.4, 0.1% Tween 20, and 1% bovine serum albumin, the membranes were reacted with anti-phospho-PKB or anti-PKB antibodies at a 1:1000 dilution. After washing, the blot was reacted with horseradish peroxidase-conjugated antibodies (Amersham Pharmacia Biotech), and chemiluminescence was performed. For blotting of proteins to PVDF Immobilon P membrane, the gel was transferred at 35 V for 5 h in transfer buffer containing 20% methanol.
Phosphopeptide Mapping and Phosphoamino Acid Analysis-Phosphopeptide mapping and phosphoamino acid analysis were performed as described (22,23).
Preparation of eNOS for Analysis by Tandem Mass Spectrometry (MS/MS)-eNOS was immunoprecipitated from six [ 32 P]orthophosphate-labeled 100-mm culture dishes and 12 non-labeled culture dishes of BAEC exposed to 15 dynes/cm 2 FSS for 1 min. The pooled immunoprecipitates were separated on a 0.75-mm thick, 9% SDS-polyacrylamide gel. The eNOS was located by silver staining (24), excised from the gel, and the incorporation of 32 P was determined by Cerenkov counting. The gel slices were washed in 1% ammonium bicarbonate, pH 8.3, shrunk by dehydration in acetonitrile, and dried in a vacuum centrifuge. eNOS was reduced in the gel slices by incubation in 1% ammonium bicarbonate, pH 8.3, containing 10 mM dithiothreitol and 2 mM EDTA under N 2 gas for 1 h at 56°C. For alkylation (25), 4-vinylpyridine was diluted to 2% (v/v) directly into the reduction buffer, and the gel slices were incubated with intermittent shaking for 1 h at 25°C in the dark. After S-pyridylethylation, the buffer was removed, and the gel slices were washed with 1% ammonium bicarbonate and shrunk with acetonitrile. Gel slices were dried in a vacuum centrifuge, and the washing, shrinking, and drying procedure was repeated (24). The gel slices containing reduced and alkylated eNOS were digested with 100 ng of sequencing grade trypsin as described (24). Ninety-five percent of the tryptic phosphopeptides were recovered from the gel slice, as determined by Cerenkov counting.
Off-line IMAC and HPLC for Enrichment of eNOS Phosphopeptides-An ion metal affinity column (IMAC) was constructed and operated as described previously (22) with the following exceptions. To each end of a 10-cm-long piece of Teflon tubing (1/16 inches outer diameter ϫ 0.0001 inches inner diameter) a piece of polyimide-coated fused silica capillary (PlymicroTechnologies, Tucson, AZ) was inserted and held in place by a union (Valco, Houston, TX). Prior to fixing the second of the two polyimide capillaries in place the open Teflon end was placed in a slurry of POROS-MC (PerSeptive) inside a vessel pressurized by helium, and the IMAC column was packed to a length of 5 cm under 500 pounds/square inch pressure. The second piece of fused silica capillary was then fixed in place with a second union. For operation one of the two polyimide capillaries was placed in a helium pressure vessel and the other served as an outlet. The IMAC column was prepared for use by washing (5 min/wash at 5 pounds/square inch) sequentially with water, 0.1 M EDTA, water, 0.1 M acetic acid, 0.1 M FeCl 3 , and 0.1 M acetic acid. The sample, reconstituted in 0.1 M acetic acid, was loaded in entirety and followed by washing with 0.1 M acetic acid, water, 0.1% NH 4 H 2 PO 4 for elution of bound phosphopeptides, water, and 0.1 M EDTA. Peptides were concentrated after IMAC, reconstituted in 0.4% acetic acid, 0.005% heptafluorobutyric acid (solvent A) and injected onto the HPLC column. HPLC was carried out on a Michrome Bioresources instrument (Auburn, CA) equipped with a 0.5-mm C18 column. A linear gradient from 0 to 60% acetonitrile, which contained 0.4% acetic acid and 0.005% heptafluorobutyric acid wash, was used to elute peptides from the column. One-minute fractions were collected and 32 P content per fraction estimated by Cerenkov counting.
Solid Phase Extraction and Capillary Electrophoresis (SPE-CE)-Fractions from the HPLC separation intended for SPE-CE were concentrated slightly to remove excess acetonitrile, as described (26). The sample was pressure-injected on a C18 cartridge (1 mm ϫ 250 m) placed at the head of the CE capillary. A series of washes both desalted and prepared the sample for CE which began when a small plug of organic solvent was pressure-injected onto the SPE cartridge (27,28).
Mass Spectrometry-Peptides were sequenced on a Finnigan (San Jose, CA) TSQ 7000 triple quadrupole mass spectrometer equipped with a home-built electrospray ionization device described previously (26). Data acquisition was computer-controlled using a program written in instrument control language that automatically provided data-dependent ion selection and varied capillary electrophoretic voltage. Ion selection and CE voltage were dependent on ion intensity such that as an ion reached a given intensity the mass spectrometer simultaneously switched from full scan mode to MS/MS mode and the CE voltage decreased from Ϫ20 to Ϫ5 kV. When the ion signal decreased below a preset threshold, the mass spectrometer returned to initial scanning/ electrophoretic conditions. Measurement of Nitric Oxide-NO released by BAEC was measured as its nitrogen oxide (NOx) metabolites, using a chemiluminescence detector as described in detail (17).
Construction of the eNOS Expression Plasmid-pCeNOS, the plasmid for the expression of His 6 eNOS in Escherichia coli, was constructed using two sequential two-piece ligations. An NdeI-SfiI fragment including the initial 533 nucleotides of bovine eNOS cDNA (29) was created using the following sense and antisense primers: 5Ј-TGA-TTACCATATGGCC[CATCAC] 3 AACTTGAAGAGTGTGGGCCAGGAG and 5Ј-GCGCCAGGCCTGCTTGGCCCCGAAC. This fragment was purified and used as a template to create a HindIII-SfiI1 fragment using additional polymerase chain reaction, using this sense primer 5Ј-TATCCCAAGCTTGGGTGATTACCATATGGCCCA and the former antisense primer. This fragment (a HindIII-SfiI fragment with an internal NdeI site) was purified and cut with HindIII and SfiI. pBluescript eNOS (29) was similarly cut and dephosphorylated. Ligation products were transformed into DH5␣ cells, and colonies were screened for positive recombinants via restriction digest. A positive clone (pBlue-NOS NDE) was maxiprepped and then cut with NdeI and XbaI. pCW oriϩ (30) was cut similarly, and the two pieces ligated, transformed, and screened as before. A positive clone (pCeNOS) allowed isopropyl-1-thio-␤-D-galactopyranoside-induced eNOS expression at 23°C (31) as judged by immunoblot analysis. The clone was sequenced through the region that was amplified by polymerase chain reaction, and this sequence was found to be perfectly consistent with the published bovine eNOS sequence (29) in the region.
In Vitro Phosphorylation of eNOS and Arginine-Citulline Conversion Assay-Kinase reactions using PKB to phosphorylate eNOS were performed according to the manufacturer's instructions. For determination of stoichiometry 150 ng of PKB was used to phosphorylate 10 g of eNOS with [␥-32 P]ATP (90 M) at a specific activity of 2800 cpm/pmol. At the indicated time points, aliquots of the reaction were boiled in SDS sample buffer, and eNOS was separated by SDS-PAGE and then localized by silver staining. The eNOS bands were excised, and incorporation was determined by Cerenkov counting. For activation of eNOS by PKB, 300 ng of PKB (specific activity 560 nmol/min mg using eNOS as a substrate) was incubated with 2.5 g of recombinant eNOS in a 20-min protein kinase reaction at 30°C. Reactions without PKB were run in parallel. After phosphorylation, 1-g aliquots of eNOS from the kinase reactions with and without PKB were diluted into 100-l eNOS assays containing 20 M L-[U- 14 (17,35) have shown that eNOS is phosphorylated in BAEC in static conditions and that the phosphorylation of eNOS is enhanced in response to flow (17). BAEC were labeled with [ 32 P]orthophosphate in phosphate-free DMEM containing 200 M Na 3 VO 4 . Cells were maintained in static condition or subjected to laminar FSS (19). The cells were lysed; eNOS was immunoprecipitated with a monoclonal antibody, and the immunoprecipitate was fractionated on an SDS-polyacrylamide gel. The proteins in the gel were transferred to nitrocellulose and detected by autoradiography (Fig. 1A). eNOS was basally phosphorylated in cells maintained in static culture (Fig. 1A, lane 1), and exposure to laminar FSS caused a rapid, nearly 2-fold increase (17) in eNOS phosphorylation (see Fig. 1A and legend).

Time Course and Phosphoamino Acid Analysis of eNOS Phosphorylation in Response to FSS-Previously we and others
Previous investigators have labeled endothelial cells under static conditions with [ 32 P]orthophosphate in the presence of phenylarsine oxide (36) or pervanadate (37) and have shown that eNOS contains predominantly phosphoserine and a small amount of phosphotyrosine. BAEC were labeled with [ 32 P]orthophosphate in the presence of Na 3 VO 4 (see "Experimental Procedures") and maintained under static condition or exposed to FSS at 15 dynes/cm 2 for 1 min. Fig. 1B shows that Two-dimensional Tryptic Phosphopeptide Maps of eNOS under Static and Flow Conditions-The eNOS transferred to nitrocellulose (shown in Fig. 1A) was subjected to tryptic cleavage and the tryptic peptides were spotted onto a thin layer plate. The peptides were fractionated in the first dimension by HVE and in the second dimension by TLC. Autoradiograms of the thin layer plates are shown in Fig. 2. Under static conditions eNOS contained 8 -10 phosphopeptides ( Fig. 2A), two of which, F1 and F2, increased with time when BAEC are exposed to 15 dynes/cm 2 FSS (Fig. 2, B-D).
The relative volumes of each of the basal and flow-stimulated phosphopeptides were quantified by PhosphorImager analysis in two independent experiments examining phosphate incorporation as a function of shear stress duration (Table I). Whereas the incorporation of phosphate into eNOS during a 10-min time course was approximately 2-fold, the specific incorporation into the flow-dependent phosphorylation sites F1 and F2 was more pronounced. Flow-dependent phosphorylation of peptide F1 was stimulated by severalfold to 30-fold, and the stimulation of phosphorylation into peptide F2 ranged from 3-to 6-fold (Table  I). The substantial experimental variation is due to the fact that the peptide mapping procedure is semi-quantitative. Additionally, the fold stimulation of phosphorylation depends upon the basal level of phosphate incorporation, which is significantly modulated by the state of BAEC growth. 2 Nonetheless, substantial increases in phosphorylation occurred in both peptides F1 and F2 as early as 30 s and more than 90% of the incorporation of phosphate into the peptides occurred by 2.5 min. These data ( Fig. 2 and Table I) demonstrate that flowinduced eNOS phosphorylation is very rapid and that flow enhances the phosphorylation of two specific eNOS tryptic peptides.
Preparation of eNOS for Mass Spectrometry and Rationale for Enrichment of eNOS Phosphopeptides Prior to Mass Spec-trometric Analysis-To obtain enough eNOS for mass spectrometric analysis of the phosphopeptides, eNOS was immunoprecipitated from 18 100-mm tissue culture dishes of BAEC. Six of the dishes were labeled with [ 32 P]orthophosphate; 12 dishes were unlabeled, and all dishes were subjected to 15 dynes/cm 2 for 1 min. Pooled lysates were immunoprecipitated; eNOS was fractionated by SDS-polyacrylamide gel electrophoresis, and the gel was silver-stained (Fig. 3, inset). Comparison of intensities of silver-stained eNOS to molecular weight markers of known amount on the gel provided an estimation of 10 -20 pmol of eNOS recovered from 18 culture dishes (data not shown). The silver-stained eNOS was excised from the gel and digested in situ with trypsin. Typically 95% of the phosphopeptides were recovered. The phosphopeptides were bound to and eluted from an immobilized metal (Fe 3ϩ ) affinity column (IMAC) with cpm recoveries of ϳ70%. The peptides were further fractionated by reversed phase-HPLC, and the radioactivity in the individual fractions was determined by Cerenkov counting (Fig. 3, bar  graph).
In initial experiments, when individual fractions of eNOS tryptic phosphopeptides isolated by reversed phase-HPLC were analyzed on-line by MS/MS, no eNOS phosphopeptides could be identified by CID, even though tryptic peptides representing 40% coverage of the amino acid sequence of eNOS were identified (data not shown), and more importantly, the majority of the incorporated phosphate was recovered in the fractions analyzed. Typically the MS/MS instrument is programmed to isolate peptides for CID in order of decreasing abundance, thus limiting the numbers of peptides that are sequenced from each chromatographic peak in the case where several peptides coelute. Since each analytical cycle (consisting of peptide mass analysis, peptide ion selection, CID, and acquisition of MS/MS spectra derived from selected peptide ions) requires on the order of 3 s, only the most abundant peptide ions from a typical chromatographic peak can be analyzed. Because the phosphorylation sites in phosphoproteins may be modified only to low stoichiometry, many phosphopeptides from proteins isolated from cells will not be sequenced without technical innovations.
The failure to identify eNOS phosphopeptides of low abundance was overcome by intermittently extending the time available for peptide analysis by reducing the sample flow rate to the mass spectrometer (26), a procedure known as "peak parking" (38). Briefly, each HPLC fraction in Fig. 3, bar graph, containing significant cpm (labeled 1-7 above the peaks) was submitted to solid phase extraction-capillary electrophoresis (SPE-CE) for peptide concentration and separation. A program was written so that as each peak entered the mass spectrometer by SPE-CE the voltage dropped, providing for acquisition by CID of the seven most abundant peptides in each fraction. This procedure is described in more detail elsewhere (26).
Mass Spectra and Sequences of the Flow-dependent Phosphorylation Sites of eNOS-HPLC fractions containing radioactive counts above background were analyzed by automated SPE-CE-MS/MS. The CID spectra generated from these analyses were initially screened by SEQUEST for the presence of eNOSderived phosphopeptides using an 80-atomic mass unit tag for serine, threonine, or tyrosine. Fractions 18 and 25 were found to each contain a single, mono-phosphorylated peptide in addition to several non-phosphorylated peptides. These spectra were then analyzed manually (39) to confirm the SEQUEST results. The CID spectrum for the phosphopeptide, [M ϩ 2H] 2ϩ ϭ 558 m/z, from fraction 18 is shown in Fig. 4 and corresponds to bovine eNOS residues Thr-1177 through Arg-1185. Location of phosphate to Ser-1179 is unambiguously derived by observation of y series fragment ions, y1-y7, and the b2 ion (40). Both the y7 ion, 946 m/z, resulting from fragmentation be-2 B. Gallis and M. Corson, unpublished observations.

FIG. 2. Two-dimensional tryptic phosphopeptide maps of eNOS isolated from BAEC under static and flow conditions.
BAEC were labeled with [ 32 P]orthophosphate, maintained in static culture, or exposed to FSS at 15 dynes/cm 2 for the indicated times. The eNOS was immunoprecipitated, size fractionated by SDS-PAGE, and transferred to nitrocellulose as in Fig. 1A. The eNOS protein was excised from the blots and digested with trypsin; typically ϳ85% of the eNOS cpm were recovered after trypsinization. Tryptic digests were spotted onto thin layer plates and separated by HVE in the first dimension and by TLC in the second dimension, and autoradiography was performed as shown: A, static condition; B, FSS 1 min; C, FSS 2.5 min; and D, FSS 10 min. These data are representative of two similar experiments (see Table I).
tween Gln-1178 and Ser-1179 and loss of phosphate from the y7 ion, 848 m/z, are observed. Fig. 5 shows the CID spectrum for the phosphopeptide, [M ϩ 3 H] 3ϩ ϭ 784 m/z, from fraction 25 that corresponds to eNOS residues Lys-110 through Arg-130. Location of phosphate to Ser-116 is derived by observation of fragment ions b2 and b5 that indicate phosphate is not present on Thr-113 and the y series ions, y14 -y10, that preclude placement of phosphate at Ser-127. A complete series of y ions is not observed because of multiple internal cleavages, rather than sequential cleavages from carboxyl to amino terminus, caused by the five prolines directing fragmentation to proline imide bonds (41). Also preventing formation of a complete series of y ions is the presence of basic amino acids Arg-114 and Lys-110 due to incomplete proteolysis by trypsin.
Co-migration of In Vivo Labeled Flow-dependent Phosphopeptides F1 and F2 with Synthetic Phosphopeptides-Whereas F1 and F2 were among the most prominent phosphopeptides found in eNOS from BAEC exposed to brief laminar FSS, eNOS peptide maps may have as many as 10 discrete labeled spots. To determine whether the phosphopeptide sequences obtained by mass spectrometric analysis (Figs. 4 and 5) were in fact the major flow-dependent phosphopeptides, phosphopeptides with the sequences RIRTQpSFSLQER and KLQTRPpSPGPPPAEQLLSQAR were synthesized. Both phosphopeptides were digested with trypsin, and the digest was mixed together with tryptic digests of eNOS from [ 32 P]orthophosphate-labeled BAEC exposed to FSS of 15 dynes/ cm 2 for 2 min. The tryptic peptide mixture was separated by HVE and TLC, and the plates were stained with ninhydrin and autoradiographed. The phosphopeptide maps showed that the two synthetic phosphopeptides co-migrated with the two flowdependent phosphopeptides F1 and F2 of eNOS isolated from BAEC (Fig. 6A). Two-dimensional separation of the tryptic phosphopeptide TQpSFSLQER alone showed that it was identical to peptide F1 (Fig. 6B). Likewise, two-dimensional separation of the tryptic phosphopeptide KLQTRPpSPGPPPAE-QLLSQAR demonstrated that it was identical to peptide F2 (Fig. 6C). This confirmed the identity and sequences of the in vivo flow-regulated phosphopeptides found by mass spectrometry.

Analysis of Candidate Kinases as Mediators of eNOS Phosphorylation--
The flow-dependent phosphopeptide F2 contains a phosphoserine (serine 116) with proline at the n ϩ 1 position to this site. This sequence most closely resembles a MAP kinase consensus substrate site (42). Since MEK1 is an upstream activator of MAP kinase, we used the MEK1 inhibitors PD98059 (50 M, n ϭ 4) (43) and U0126 (10 M, n ϭ 1) (44) to inhibit flow-dependent phosphorylation of F2, but we failed to  observe a decrease in flow-dependent phosphorylation at this site even though MAP kinase was inhibited by 95% in an immune complex kinase assay (data not shown). Other inhibitors and activators of protein kinases were used to modulate phosphorylation levels of eNOS tryptic phosphopeptides. Since laminar FSS causes a rapid increase in BAEC [Ca 2ϩ ] i , cells were incubated prior to flow with the group-specific inhibitor of the Ca 2ϩ /calmodulin kinases (45) KN-62 (10 M, n ϭ 4), but no inhibition of phosphorylation of any phosphopeptide was observed (data not shown). Because NO increases levels of cyclic GMP and may activate the cyclic GMP-dependent protein kinase, we used the NO donor 2,2Ј-(hydroxynitrosohydrazine)bisethanamine-NONOate (100 M, n ϭ 3) and stable cyclic GMP analogue 8-(chlorophenylthio)-guanosine 3Ј:5Ј-cyclic monophosphate (500 M, n ϭ 3) to determine whether the level of phosphorylation of any eNOS tryptic peptide could be enhanced by these agents. No increases in the level of phosphorylation of any tryptic peptide were observed (data not shown). However, three novel eNOS phosphopeptides, not observed under static or flow conditions, appeared in the presence of these compounds (data not shown). These findings suggest that under the conditions used for these studies eNOS is neither a sub-strate for MAP kinases, Ca 2ϩ /calmodulin-dependent kinase, nor a kinase indirectly activated by increases in NO.
Inhibition of Flow-dependent Phosphorylation of eNOS and PKB by the PI3-kinase Inhibitor LY294002-The flow-dependent eNOS phosphorylation site, F1, identified by mass spectrometry is the sequence (RIR)TQpSFLQER which contains the PKB consensus substrate phosphorylation site RXRXXSX (where X is F, I, or V) (46). Because PKB is activated by PI3-kinase (46), which itself is rapidly activated by shear stress in BAEC (47), we used the PI3-kinase inhibitor LY294002 (48) to determine whether inhibition of an upstream activator of PKB would decrease flow-dependent phosphorylation of eNOS. [ 32 P]Orthophosphate-labeled BAEC were incubated for 30 min with vehicle (0.1% Me 2 SO) or LY294002 (20 M) and then maintained in static culture (Fig. 7A, 1st and 2nd lanes) or exposed to laminar FSS at 15 dynes/cm 2 for 2 min (Fig. 7A, 3rd  and 4th lanes). An autoradiogram of the size-fractionated eNOS immunoprecipitates and a corresponding eNOS Western blot are shown (Fig. 7A). LY294002 inhibited the flow-dependent increase in eNOS phosphorylation by 97% (n ϭ 3).
The activation of PKB (in this case PKB␣) requires phosphorylation of Thr-308 and Ser-473 (46). In order to determine whether FSS may enhance the phosphorylation of PKB, BAEC were incubated for 30 min with vehicle (0.1% Me 2 SO) or LY294002 (20 M), then maintained in static culture (Fig. 7B,  1st and 2nd lanes), or exposed to laminar FSS at 15 dynes/cm 2 for 2 min (Fig. 7B, 3rd and 4th lanes). Lysates were transferred to membranes and reacted with anti-phospho-PKB antibody or with anti-PKB antibody to verify equivalent loading.  5. Tandem mass spectra of eNOS phosphopeptide F2. Flowdependent phosphopeptide F2 was sequenced as described in the legend for Fig. 4. The scale on the y axis has been expanded (4ϫ) to better show key fragment ions.
FIG. 6. Co-migration of in vivo labeled flow-dependent phosphopeptides F1 and F2 with synthetic phosphopeptides. A, BAEC were labeled with [ 32 P]orthophosphate and exposed to FSS at 15 dynes/cm 2 for 2 min. The eNOS was immunoprecipitated, and tryptic phosphopeptides were separated by HVE and TLC (as in the legend of Fig. 2). Prior to separation, 10 g each of synthetic phosphopeptides corresponding to F1 and F2 were mixed with the in vivo labeled tryptic phosphopeptides. After two-dimensional separation the plates were stained with ninhydrin to localize the unlabeled synthetic phosphopeptides. Circles representing the migration location of the stained peptides F1 and F2 are superimposed on the autoradiogram. B, ninhydrin stain of a two-dimensional phosphopeptide map of a tryptic digest of the synthetic phosphopeptide RIRTQpSFSLQER. C, ninhydrin stain of a two-dimensional phosphopeptide map of a tryptic digest of the synthetic phosphopeptide KLQTRPpSPGPPPAEQLLSQAR. These data are representative of three similar experiments. PI3-kinase Inhibitor LY294002-Laminar FSS induces both rapid eNOS phosphorylation and NO production in BAEC. If flow regulates the production of NO by altering phosphorylation levels in eNOS, it is possible that phosphorylation regulates eNOS enzymatic function. Since LY294002 inhibited flowdependent phosphorylation of eNOS, we determined whether the compound would inhibit NO production in BAEC. Cells were preincubated for 30 min with either vehicle or LY294002 (10 or 20 M) and then exposed to FSS at 12 dynes/cm 2 for 2 or 5 min. LY294002 (10 M) inhibited NO production by 62 and 48% after 2 and 5 min of flow, respectively (n ϭ 3). A higher concentration of LY294002 (20 M) inhibited NO production by 68 and 61% after 2 and 5 min of flow, respectively (n ϭ 3, Fig.  7C). The inhibition by LY294002 of PKB phosphorylation, eNOS phosphorylation, and NO production suggested that laminar FSS regulates BAEC NO production by altering levels of eNOS phosphorylation via a pathway involving PI3-kinase and PKB.
Phosphorylation and Activation of eNOS by PKB-Because the flow-dependent phosphorylation site F1 is a consensus site for PKB (46) and LY294002 inhibits flow-dependent NO production in BAEC, we determined whether PKB could phosphorylate and activate eNOS in vitro. Recombinant PKB phosphorylated bacterially derived recombinant eNOS to a stoichiometry of 1.1 mol of P/mol of eNOS (Fig. 8A). The eNOS from this reaction was separated by SDS gel electrophoresis, excised from the gel, and digested in situ with trypsin. The tryptic phosphopeptides were separated by two-dimensional peptide mapping. An autoradiogram (Fig. 8B) shows that the major phosphopeptide migrates to a position similar to peptide F1 isolated from BAEC. All of the phosphopeptides seen in Fig.  8B were extracted from the cellulose and subjected to MS/MS analysis. Only one sequence, TQpSFSLQER, was detected with a mass ϭ 1175.8 Da, close to the average mass of 1175.2 Da predicted for the F1 peptide. Thus the major site in eNOS phosphorylated by PKB in vitro is identical to the site designated as F1 that is phosphorylated in response to flow in BAEC. After determining conditions necessary to achieve stoichiometric phosphorylation of eNOS with PKB, eNOS was incubated in protein kinase reactions with and without PKB. The kinase reactions were diluted into eNOS enzyme assays, and conversion of [ 14 C]arginine to [ 14 C]citrulline was measured. Stoichiometric phosphorylation of eNOS by PKB activated eNOS by 15-20-fold (Fig. 8C). DISCUSSION Through the use of innovative modifications of mass spectrometry, we have identified two flow-dependent phosphorylation sites in eNOS. We have confirmed the identities of the tryptic phosphopeptides by co-migration of in vivo labeled tryptic phosphopeptides of eNOS with synthetic peptides based on the sequences discovered by mass spectrometry. One tryptic phosphopeptide, designated F1, TQpSFSLQER, cleaved from the sequence, RIRTQpSFSLQER, is consistent with the PKB substrate consensus phosphorylation sequence RXRXXSX (where X is F, I, or V) (46). Because PI3-kinase is an upstream activator of PKB (46), we used the PI3-kinase inhibitor, LY294002 (48), to partially inhibit flow-dependent PKB activity. When BAEC were pretreated with LY294002 prior to flow, the inhibitor decreased flow-dependent phosphorylation of PKB by 80%, eNOS phosphorylation by 97%, and NO production by 68%. These data suggest that flow-dependent regulation of NO production in BAEC occurs in part by regulation of the level of phosphorylation of Ser-1179 in eNOS. Indeed, in vitro phosphorylation of eNOS by PKB occurred at Ser-1179 and activated the enzyme by 15-20-fold.
Flow-dependent phosphorylation of eNOS occurs near the amino terminus, at Ser-116, and near the carboxyl terminus, at Ser-1179. The amino acid residues surrounding Ser-1179 suggest that eNOS may be phosphorylated by PKB as well as possibly other protein kinases (49). Unlike Ser-1179, the sequence surrounding Ser-116 contains a proline residue at the n ϩ 1 position, suggesting that a proline-directed protein kinase may phosphorylate this site. Thus FSS appears to activate at least two different protein kinases that phosphorylate eNOS. The significance of these modifications for regulation of eNOS function will be better understood when a crystal structure of the enzyme or site-directed eNOS mutants become available. Recently the crystal structure of an eNOS fragment (amino acid residues 39 -482) revealed that pairs of cysteine residues (96 and 101) at the dimer interface tetrahedrally coordinate a zinc ion that stabilizes the tetrahydrobiopterin-binding site that interacts with Ser-104 and Val-106 (50). We speculate that incorporation of phosphate at Ser-116 may alter the structure of the tetrahydrobiopterin-binding site to enhance eNOS catalytic activity.
The two phosphorylation sites identified in this study and a third site described elsewhere (26) are the first in vivo phosphorylation sites identified for any isoenzyme of nitric-oxide synthase. The other two isoenzymes of nitric-oxide synthase, inducible nitric-oxide synthase (iNOS) and neuronal nitricoxide synthase (nNOS), share 50 -60% amino acid identity (51-53) with eNOS. The mechanism of regulation of nNOS remains unclear, but this enzyme contains the conserved PKB consensus sequence (51) we have identified as being phosphorylated in eNOS. nNOS lacks the F2 site and iNOS contains neither the F1 nor the F2 sites we have identified as being phosphorylated in eNOS in response to flow. Indeed, levels of NO produced by iNOS are believed to be regulated not by phosphorylation of iNOS but by the levels of mRNA for this protein (53). In contrast to the sequence divergence between eNOS and the other isoforms of NOS, the primary amino acid sequences of mammalian eNOS are more than 90% conserved (52) and contain the phosphorylation sites we have described. Previous investigators have shown that eNOS contains predominantly phosphoserine with phosphotyrosine as a minor modification (36,37). In these studies, either phenylarsine oxide or pervanadate was used. Both of these protein tyrosine phosphatase inhibitors greatly increase the overall tyrosine phosphate content of many cellular proteins (54,55). We have found only phosphoserine in eNOS under both static and flow conditions in BAEC. Tandem MS/MS analysis of eNOS tryptic peptides derived from cells under static or flow conditions yielded CID spectra that covered 40% of the amino acid sequence of eNOS (data not shown). These spectra were screened and matched against the known sequence of eNOS using the SEQUEST (56) program, for addition of phosphate to serine, threonine, and tyrosine. No phosphothreonine or phosphotyrosine residues were found. Whereas such modifications may exist and may not be detected by the mass spectrometer because of low stoichiometry, the SEQUEST search did confirm the phosphoamino acid analysis in this study by identifying three phosphoserine-containing peptides (26).
Repeated attempts to identify tryptic phosphopeptides of eNOS by MS/MS analysis after conventional means of enrichment (IMAC) and separation (HPLC) were not successful, despite a 40% coverage of the eNOS primary amino acid sequence. In this study, the data-dependent modulation of the rate at which peptide ions entered the MS/MS, produced by a peptide peak-generated drop in the electric field of the CE, expanded the analytical window for each peptide to be analyzed. This peak parking procedure (26,38) should be applicable to identification of in vivo phosphorylation sites in other large phosphoproteins or in cases in which the phosphoprotein cannot be purified to homogeneity.
It has recently been shown (57) that the AMP-activated kinase phosphorylates Ser-1179 on eNOS in vitro and increases eNOS activity. While this manuscript was under review, two groups (58,59) showed that Ser-1179 is phosphorylated in response to VEGF and shear stress, that PI3-kinase inhibitors decrease agonist-induced phosphorylation of eNOS and NO production, and that PKB phosphorylates and activates eNOS in vitro.