Endothelial nitric-oxide synthase (type III) is activated and becomes calcium independent upon phosphorylation by cyclic nucleotide-dependent protein kinases.

Endothelial nitric-oxide synthase (NOS-III) is defined as being strictly dependent on Ca(2+)/calmodulin (CaM) for activity, although NO release from endothelial cells has been reported to also occur at intracellular free Ca(2+) levels that are substimulatory for the purified enzyme. We demonstrate here that NOS-III, but neither NOS-I nor -II, is rapidly and strongly activated and phosphorylated on both Ser and Thr in the presence of cGMP-dependent protein kinase II (cGK II) and the catalytic subunit of cAMP-dependent protein kinase (cAK) in vitro. Phosphopeptide analysis by mass spectrometry identified Ser(1177), as well as Ser(633) which is situated in a recently defined CaM autoinhibitory domain within the flavin-binding region of human NOS-III. Phosphoamino acid analysis identified a putative phosphorylation site at Thr(495) in the CaM-binding domain. Importantly, both cAK and cGK phosphorylation of NOS-III in vitro caused a highly reproducible partial (10-20%) NOS-III activation which was independent of Ca(2+)/CaM, and as much as a 4-fold increase in V(max) in the presence of Ca(2+)/CaM. cAK stimulation in intact endothelial cells also increased both Ca(2+/)CaM-independent and -dependent activation of NOS-III. These data collectively provide new evidence for cAK and cGK stimulation of both Ca(2+)/CaM-independent and -dependent NOS-III activity, and suggest possible cross-talk between the NO and prostaglandin I(2) pathways and a positive feedback mechanism for NO/cGMP signaling.

distinct NOS isoforms have been identified, originally isolated from cerebellum (nNOS, NOS-I), macrophages (iNOS, NOS-II), and endothelial cells (eNOS, NOS-III). All NOSs share an overall 50% amino acid sequence homology and have similar cofactor requirements. Each NOS monomer contains an Nterminal oxygenase domain with a binding site for the arginine as substrate, tetrahydrobiopterin, Ca 2ϩ /CaM, and zinc, as well as a C-terminal reductase domain with an autoinhibitory region, and one binding site each for FAD, FMN, and NADPH as electron donor (1)(2)(3)(4).
Only NOS-I and NOS-III, the constitutively expressed forms prominent in neurons, and skeletal muscle or endothelial cells, respectively, are activated by agonist-induced elevation of the intracellular free Ca 2ϩ concentration with subsequent binding of Ca 2ϩ /CaM to NOS (5). In contrast, NOS-II is regulated predominantly at the transcriptional level by endotoxins and cytokines in macrophages, hepatocytes, and vascular smooth muscle cells (for review, see Ref. 6), and binds CaM with such high affinity that it is already maximally activated at the Ca 2ϩ level of a resting cell (7). Additional regulatory mechanisms which have been proposed to regulate NOS include autoinhibition, protein-protein interactions, subcellular localization, acylation, changes in substrate supply, phosphorylation (for review, see Ref. 8), and enzyme monomerization (9,10). Interestingly, shear stress and isometric contraction have been shown to activate endothelial NOS-III synthase in a Ca 2ϩ / CaM-independent manner that is sensitive to tyrosine kinase inhibitors (11,12). This Ca 2ϩ -independent activation is not understood molecularly. For example, a tyrosine kinase-dependent step has been suggested, however, shear stress-dependent tyrosine phosphorylation of NOS-III could not be shown (13). Alternatively, tyrosine kinase-dependent phosphorylation of yet unknown NOS-III-binding proteins was suggested to play a role in Ca 2ϩ -independent endothelial NO release (14). Importantly, NOS-III has not been shown to have Ca 2ϩ independent activity in broken cell preparations or as a purified enzyme. Thus, the possibility remained that NOS-III is in fact Ca 2ϩ -dependent, but that membrane-localized increases in free Ca 2ϩ undetectable by whole cell or multiple cell Ca 2ϩ imaging (15), are important in enzyme regulation. NO formation triggers cGMP synthesis and cyclic nucleotide-dependent protein kinase activation. Thus we considered the possibility that NOS-III is positively regulated by direct phosphorylation and becomes Ca 2ϩ -independent upon Ser/Thr phosphorylation by cAMP-and cGMP-dependent protein kinases.
NOS phosphorylation has been investigated in several studies, but the effects and physiological consequences are not completely understood. Purified NOS-I was shown to be phosphorylated by cAMP-and cGMP-dependent protein kinase (cAK, cGK), protein kinase C, and Ca 2ϩ /CaM kinase II, with each kinase phosphorylating a different Ser. In all cases, Ser phosphorylation resulted in an inhibition of NOS-I catalytic activity, suggesting a negative feedback regulation of neuronal NO formation (16,17). In rat heart, NOS-III is phosphorylated on Ser 1177 and stimulated by an AMP-activated protein kinase both in vitro and during ischemia (18). In cultured endothelial cells, NOS-III is phosphorylated on Ser in response to bradykinin and then translocates to the cytosol (19). Several studies have shown that NOS-III is targeted to plasmalemmal caveolae, and it has been suggested that the caveolin⅐NOS-III complex undergoes cycles of dissociation (activates NOS-III) and re-association (inactivates NOS-III) which are modulated by increases and decreases in Ca 2ϩ , respectively (20,21). However, treatment of bovine aortic endothelial cells with hydrogen peroxide increased tyrosine phosphorylation of NOS-III and inhibited its activity, but did not disturb its association with caveolin (22).
In the present study, we demonstrate that phosphorylation of NOS-III by cGMP-and cAMP-dependent protein kinases not only stimulates Ca 2ϩ /CaM-dependent NOS-III activity, but also causes a Ca 2ϩ -independent partial activation of NOS-III both in vitro and in vivo that does not appear to involve enzyme translocation to the cytosol.
Preparation of Recombinant Human (h)NOS-I, II, and III, and Native Porcine NOS-I-Recombinant human NOS-I, -II, or -III were purified from Sf9 cells using affinity chromatography (2Ј,5Ј-ADP-Sepharose for NOS-II, or CaM-Sepharose for NOS-I and III) (9); native porcine cerebellum NOS-I was purified as described (25). Specific NOS activities were (nmol/mg/min): 401 (hNOS-I), 125 (pNOS-I), 10 (partially purified hNOS-II), and 105 (hNOS-III). All NOS were stored at Ϫ80°C in 50-l aliquots containing 10% glycerol until the day of use. Protein concentrations were determined spectrophotometrically according to the method of Bradford (26) using bovine serum albumin as a standard and a SpectraMax 340 Microplate reader (Molecular Devices, Sunnyvale, CA). Protein purity was determined by densitometry of Coomassie-stained SDS-PAGE gels using a flatbed scanner and NIH Image software (National Institutes of Health, Bethesda, MD). NOS-III immunoreactive bands were examined by Western blot analysis using a NOS-III-specific antibody and enhanced chemiluminescence detection.
In Vitro Phosphorylation of NOS-I, -II, and -III-NOS-I, -II, and -III were incubated at 30°C for the times indicated with 10 mM HEPES, pH 7.4, 5 mM MgCl 2 , 1 mM EDTA, 0.2 mM dithiothreitol, and either C subunit of cAK or cGK (I␣, I␤ or II) and 5 M cGMP. The reaction was started by the addition of 100 M ATP. In control experiments, heatinactivated (10 min at 95°C) cGK and C subunit were used.
Determination of NOS Activity-Catalytic activity of NOS was assayed under steady state conditions by the Ca 2ϩ /CaM-dependent conversion of L- 2), and a total volume of 100 l. For assaying NOS-III activity in human umbilical vein endothelial cells (HUVEC), arginine was reduced to 2 M and tetrahydrobiopterin was increased to 10 M. The L-citrulline formed was separated by cation exchange chromatography and measured by liquid scintillation counting. NOS activity was expressed either as a percentage of the maximum rate of L-citrulline formation (V max ) in the presence of Ca 2ϩ /CaM, or as picomole or nanomole mg Ϫ1 min Ϫ1 .
Phosphoamino Acid Analysis-For phosphoamino acid analysis, 32 Pphosphorylated NOS-III was digested in gel with trypsin (2:1, w/w) and then subjected to acid hydrolysis with 6 N HCl at 110°C for 3 h in a sealed tube. The samples were dried, dissolved in water, and analyzed for phosphoamino acids by one-dimensional electrophoresis (Multiphor II, Amersham Pharmacia Biotech) on cellulose plates in 1.8% formic acid, 7.3% acetic acid (pH 1.9), at 1400 V, 15°C for 2 h. Phosphoserine and phosphothreonine were used as internal standards and visualized by staining with ninhydrin. 32 P-Labeled amino acids were visualized by autoradiography.
Mass Spectrometric Analysis of NOS Phosphopeptides-Maximally 32 P-phosphorylated NOS-III (20 pmol) was digested and the tryptic fragments were separated by high pressure liquid chromatography (HPLC) using a 0.3 ϫ 150-mm PL-SAX (8 m, 1000 Å) anion exchange column (LC Packings, Amsterdam Pharmacia Biotech, Netherlands). Peptides were eluted at a flow rate of 4 l/min with a linear gradient from 95% solvent A (20 mM ammonium acetate, pH 7.0) to 99% solvent B (0.5 M KH 2 PO 4 , 25% acetonitrile, pH 4.0) over 55 min. The eluate was monitored by UV absorbance at 214 nm and fractions of 4 l were collected. Fractions containing radioactivity were further separated using a 300 m ϫ 250 mm (5 m, 1000 Å) reversed phase C 18 column (LC Packings, Amsterdam Pharmacia Biotech). Peptides were eluted at a flow rate of 16 l/min with a complex gradient which started at 5% solvent B (0.08% trifluoroacetic acid, 84% acetonitrile) in solvent A (0.1% trifluoroacetic acid) for 30 min, was increased to 50% solvent B at 120 min and 99% solvent B at 136 min, and ended with 99% solvent B at 146 min. The eluate was monitored by UV absorbance at 214 nm. Fractions containing 32 P were analyzed by matrix-assisted laser-desorption/ionization-time of flight-mass spectrometry (MALDI-TOF-MS) (Bruker Reflex III, Bruker Daltonik, Bremen, Germany) and Edman degradation (Procise cLC 494 system, Applied Biosystems), commonly using 0.3 l of an HPLC fraction. As MALDI matrix, a saturated solution of 4-hydroxy-␣-cinnamic acid (Sigma, Deisenhofen, Germany) in 0.1% trifluoroacetic acid/acetonitrile was used. Samples were mixed on target with the matrix solution.
Size Exclusion Chromatography-Purified human recombinant NOS-III (20 g) was incubated as described above for the NOS activity assay after in vitro phosphorylation, except that in these experiments the concentration of L-arginine, tetrahydrobiopterin, and CaM were increased to 2 mM, 50 M, and 2 M, respectively. The reaction was stopped by the addition of 20 l of ice-cold EGTA (30 mM) and the samples were immediately snap frozen in liquid nitrogen. After thawing, the samples were centrifuged (10,000 ϫ g, 4°C for 10 min), and then a 100-l aliquot (equivalent to 17 g of NOS protein) was analyzed by fast protein liquid chromatography using a Superose 6 HR 10/30 gel filtration column (Amersham Pharmacia Biotech, Freiburg, Germany) that was equilibrated with 20 mM triethanolamine buffer (pH 7.5) containing 150 mM NaCl and 5% (v/v) ethylene glycol. Proteins were eluted at a flow rate of 0.25 ml/min and monitored by their absorbance at 280 nm.
Phosphorylation and Activation of NOS-III in Intact HUVEC-HU-VEC (2 ϫ 10 6 cells/25 cm 2 ) were prepared as described previously (27) and cultivated in M199 medium with 20% fetal calf serum and 1% penicillin/streptomycin on 0.1% gelatin-coated flasks. All experiments were performed with subconfluent cells in passage 1. For phosphorylation and activation of NOS-III in intact HUVEC, cells were stimulated with 200 M Sp-5,6-DCl-cBiMPS for 10 min, or with 100 M of the prostaglandin I 2 analog, Iloprost, for 5 min. Cells were then harvested in 25 mM triethylamine (pH 7.5), 7 mM GSH, 1 M leupeptin, 1 M pepstatin, 0.2 mM phenylmethylsulfonyl fluoride, 1 M E-64, and 1 g/ml aprotinin, and subsequently analyzed for NOS activity. The cpm obtained using extracts from untreated cells assayed in the absence of CaCl 2 (presence of 1 mM EDTA) was defined as assay blank and subtracted from all values.
Statistics-Unless otherwise indicated, data shown represent mean Ϯ S.E. Statistical analysis was performed by Student's unpaired t test (two-tailed). p Ͻ 0.05 was considered statistically significant.

Consensus Phosphorylation Sites-
The key sequence for substrate recognition by cAK and cGK contains basic amino acids N-terminal to the phospho-acceptor serine or threonine, with a more stringent requirement for multiple basic residues in the case of cGK recognition. Most consensus phosphorylation sequences for cAK and cGK are described by the motif R(R/ K)X(S/T) and (R/K 2-3 )X(S/T), respectively (28). Using this information, we identified several putative cAMP-and cGMPdependent protein kinase phosphorylation sites in human NOS forms I (Thr 196 , Ser 631 , Ser 1002 , and Thr 1432 ) and III (Thr 495 and Ser 633 ). For NOS-II, no protein kinase recognition sequence motif was recognized. Fig. 1 shows a multiple sequence alignment of selected NOS sequences. A comparison of human NOS-III with mouse and bovine NOS-III revealed that the two possible phosphorylation sites are conserved within the CaMbinding site (Thr 495 ) and the CaM autoinhibitory sequence (Ser 633 ).
Effect of Cyclic Nucleotide-dependent Phosphorylation on Human NOS Activity-The putative phosphorylation sites of the NO synthases were further characterized by incubating purified recombinant NOS-I, -II, and -III with cGK I␣, I␤, II, or catalytic subunit of cAK in the presence of [␥-32 P]ATP. An autoradiogram of a representative SDS-PAGE is shown in Fig.  2B. Only marginal incorporation of phosphate into NOS-I was observed after 5 min treatment with any of the four kinases. In contrast, VASP, a well known (control) substrate for cAK and cGK (29), was highly phosphorylated by all four kinases, similar in degree to NOS-III phosphorylation by cGK II and C subunit, described below (Fig. 2B). To assess the activity of phosphorylated NOS, the conversion of [ 3 H]arginine to citrulline was measured. Phosphorylation caused no obvious change in NOS-I activity ( Fig. 2A). Similar negative results (weak NOS-I phosphorylation and no change in enzyme activity) were also observed with two other recombinant NOS-I preparations, as well as a native preparation from porcine cerebellum (data not shown). In agreement with our consensus sequence analysis showing no putative cAK and cGK phosphorylation sites for NOS-II, no effect of any of the four kinases on phosphorylation (Fig. 2B) or activity ( Fig. 2A) of NOS-II could be detected.
In contrast to NOS-I and NOS-II, endothelial NOS-III was strongly phosphorylated by cGK II and C subunit (Fig. 2B), however, cGK I␣ and I␤ showed much less 32 P incorporation into the enzyme. Phosphorylation of NOS-III by cGK II and C subunit increased NOS activity 4 -6-fold ( Fig. 2A), whereas no major change of NOS-III activity was observed after phosphorylation by cGK I␣ and I␤. The stoichiometry of NOS-III phosphorylation (Fig. 2C)  mol of phosphate/mol of NOS by cGK II, and 0.76 Ϯ 0.1 mol of phosphate/mol of NOS by C subunit. Half-maximal phosphorylation occurred at about 2 min. Further addition of kinases after 30 min did not increase phosphorylation, suggesting that the steady-state level of 32 P incorporation was not being underestimated due to time-dependent loss of protein kinase activity.
Identification of NOS-III Cyclic Nucleotide-dependent Phosphorylation Sites-Phosphoamino acid analysis of maximally 32 P-phosphorylated NOS-III by one-dimensional electrophoresis identified both serine and threonine as targets of cAK and cGK II (Fig. 3A). To directly identify the residues phosphorylated, purified NOS-III was incubated with C subunit or cGK II, digested with trypsin, and 32 P-phosphorylated peptides were identified by HPLC followed by MALDI-MS. For both kinases, two phosphopeptides (amino acids 631-646 and 1173-1183) with phosphoserine at positions 633 and 1177, respectively, were identified (Fig. 3, B and C). The sequences and the phosphorylated serines determined by MALDI-MS were confirmed by direct Edman sequencing of the radioactive peptides (not shown). The phosphothreonine which had been detected by one-dimensional electrophoresis was not recovered. Tryptic digestion of NOS-III resulted in most likely too small a fragment (KpTFK) not retarded by the anion exchange HPLC. In addition, the presence of an intense matrix background, extending up to 1000 Da, would have made the detection of this small peptide very difficult. To circumvent these problems, phosphorylated NOS-III (up to 200 pmol) was digested with endoproteinase Glu-C to obtain a theoretical 35-mer peptide containing Thr 495 . However, the digest was incomplete (less than 10%), and no radioactive peptides were recovered.
Effect of Phosphorylation on NOS-III Quaternary Structure-Since all NOS forms are only active as homodimers, we tested the effects of cGMP and cAMP-dependent protein kinase phosphorylation on NOS-III quaternary structure by size exclusion chromatography. Phosphorylated NOS-III was found to elute as two protein peaks, a main dimer peak at 13.8 ml of elution buffer and a second smaller monomer peak at 14.9 ml (Fig. 4). Both peaks contained NOS-III immunoreactive protein as analyzed by Western blot (not shown). Neither cAMP-nor cGMP-dependent protein kinase phosphorylation altered the ratio of dimeric to monomeric enzyme (Fig. 4, A and B), suggesting that changes in NOS-III catalytic activity resulting from phosphorylation are not due to changes in NOS-III quaternary structure or increased dimer stability.
Effect of Phosphorylation on NOS-III Flavin Content-A potential effect of phosphorylation on flavin binding was investigated since two cGK and cAK consensus phosphorylation sites (Ser 633 and the putative Thr 495 ) in the autoinhibitory and CaM binding sequences are immediately flanked by FMN-binding domains ( Fig. 1) (31). However, phosphorylation did not change the effect of increasing FMN concentration (0.1-10 M) on NOS-III activity (Fig. 5A). In a second experiment, the amount of flavin bound to the phospho-and dephospho-form of NOS-III was examined by HPLC. No differences in the amount of NOSassociated FMN and FAD was observed after phosphorylation by C subunit (Fig. 5B) or cGK II (data not shown).
Effect of Phosphorylation on Apparent Calmodulin Affinity of NOS-III-To determine the effect of phosphorylation on CaM binding, phosphorylated NOS-III was incubated with various CaM concentrations up to 1 M. As shown in Fig. 6A, phosphorylation did not change the half-maximal concentration of CaM (15 nM) that activated NOS-III, but considerably increased V max . Surprisingly, phosphorylated NOS-III was found to be active even under assay conditions in which Ca 2ϩ /CaM was omitted, although this NOS isoform is defined as being strictly dependent on CaM and physiological increases in free Ca 2ϩ . To further investigate this effect, NOS-III activity was examined either in the absence or presence of 100 nM CaM, in either the absence (0.5 mM EDTA) or presence of 1 mM Ca 2ϩ . When NOS-III was assayed in the absence or presence of Ca 2ϩ /CaM, both cGK II and C subunit activated NOS-III and caused a 10 -20% fraction (Fig. 6, B and C, three left black columns) of the maximal Ca 2ϩ /CaM stimulated NOS-III activity (Fig. 6, B and C, open column), or 5-10% of the maximal activity attained by Ca 2ϩ /CaM in the presence of active kinases (Fig. 6, B and C, black column at far right) to become independent not only of Ca 2ϩ but also of CaM. No activity was observed for nonphosphorylated NOS-III (denatured cGK II and C used) under the same Ca 2ϩ /CaM-free assay conditions (Fig. 6, B and C). As shown in Fig. 6, A-C, NOS-III activity in the presence of calmodulin (10 -1000 nM) was also considerably enhanced by C subunit of cAK or cGK II.
In Vivo Studies of NOS-III Activation in HUVEC-In order to investigate NOS phosphorylation and activity in intact cells,

FIG. 2. Effects of cAMP-and cGMP-dependent protein kinases on in vitro phosphorylation and activity of human NOS-I, -II, and -III.
A, regulation of Ca 2ϩ /CaM-dependent NOS-I, -II, and -III catalytic activity by C subunit of cAK and cGK phosphorylation of NOS. NOS was phosphorylated by C subunit or cGK as described for B below. Activity was measured as described under "Experimental Procedures" and is expressed as percent of the respective denatured-kinase control  Fig. 7A). Moreover, a significant fraction (29%) of total NOS-III activity was Ca 2ϩ independent (black column in Fig. 7A) in Iloprost-treated samples. These results correlated with those obtained upon co-incuba- tion of purified NOS-III and cGK II or cAK in vitro. To investigate whether phosphorylation by cAK affected the subcellular distribution of NOS-III, we performed Western blot analysis of the 100,000 ϫ g pellet (P100) and supernatant (S100) subcellular fractions of HUVEC. However, the membrane association of NOS-III remained unchanged (89% in untreated, 86% in Iloprost treated cells, Fig. 7B).

DISCUSSION
The present study establishes that endothelial NOS-III is phosphorylated in vitro on Ser 633 , Ser 1177 , and most likely Thr 495 , by cAMP-and cGMP-dependent protein kinase II. Furthermore, this phosphorylation can activate NOS-III in the absence of Ca 2ϩ /CaM which is unusual for NOS-III since this isoform is classified as a constitutively expressed, strictly Ca 2ϩ / CaM-dependent enzyme. Recent experiments have, however, suggested that Ca 2ϩ -independent NO release can occur in response to fluid shear stress or isometric contraction (11,12) and a parallel rapid Ser/Thr phosphorylation of NOS-III was suggested (13). In addition, general tyrosine kinase activation upon shear stress was observed (13). However, it remained unclear whether NOS-III activity indeed became Ca 2ϩ /CaMindependent or whether intracellular increases in free Ca 2ϩ occurred as highly localized Ca 2ϩ spikes in compartments rich in NOS-III (e.g. caveolae) which were not detected by whole cell Ca 2ϩ analysis. Moreover, the protein kinases involved in these effects were not characterized. Our present results show for the first time that cAMP-and cGMP-dependent protein kinase II-evoked Ser/Thr phosphorylation of purified NOS-III causes a partial Ca 2ϩ /CaM-independent activation of the enzyme.
Besides evoking Ca 2ϩ /CaM-independent activation, cAK and cGK also enhanced the V max of NOS-III activity in the presence Ca 2ϩ /CaM in vitro. Furthermore, stimulation of cAK in intact HUVEC also increased both Ca 2ϩ /CaM-dependent and -independent activation of NOS-III. This could account for an earlier observation by Graier et al. (30) that forskolin-induced elevation of endothelial cell cAMP levels increased NO release, an effect that was abolished by inhibitors of cAK.
With respect to the possible molecular mechanism underlying phosphorylation-dependent activation of NOS-III, we considered previous evidence presented by Salerno et al. (31) for an autoinhibitory control element, a 45-amino acid insert located near the CaM-binding region of only constitutive NOS forms (NOS-I and III, see Fig. 1), not NOS-II. The autoinhibitory region is postulated to stabilize an inhibited NOS conformation. Binding of CaM could cause displacement of the autoinhibitory element, thereby evoking a conformational change of NOS-I and -III to support activation. The autoinhibitory domain is notably rich in positively charged amino acid residues, especially in the case of NOS-III which contains the sequence RRKRK immediately before Ser 633 , one of the cAK/cGK II phosphorylation sites (Fig. 1). Phosphorylation of Ser 633 by cAK/cGK II would increase the negative charge in the autoinhibitory domain, and could thereby facilitate displacement of this domain from the site with which it interacts, thus partially activating NOS in the absence of Ca 2ϩ /CaM, as well as facilitating Ca 2ϩ /CaM binding and enhancing maximal NOS activity.
The autoinhibitory domain insert is present only in NOS-I and NOS-III and, interestingly, Ser 633 is conserved only in human, mouse, and bovine NOS-III, but not in NOS-I. Human NOS-I contains four possible cAK/cGK phosphorylation sites, however, our experiments detected only weak phosphorylation and no significant change in catalytic activity of hNOS-I after 5 min incubation with either C subunit or cGK. At the same time, NOS-III was highly (maximal 0.82 mol of phosphate/mol of NOS-III) phosphorylated. Previously, phosphorylation of rat neuronal NOS-I (rNOS-I) by cGK I␣ or C subunit of cAK was shown to result in a 25-40% decrease in rNOS-I activity (17) and translocation of rNOS-I from the membrane to the cytoplasm (16). However, incorporation of only 0.1 mol of phosphate/mol of rNOS-I was observed after treatment with cGK I␣ for 5 min, and half-maximal phosphorylation was evident only after 30 min. The lack of a cAK/cGK effect on phosphorylation and activity of hNOS-I in our experiments may be explained by the presence of a consensus sequence for cAK/cGK phosphorylation in the autoinhibitory domain of rat NOS-I (SRKSSGD) that is absent in human NOS-I (SQKSSGD) (see Fig. 1). It remains to be shown whether such low levels and slow rates of NOS-I phosphorylation ever become physiologically relevant, e.g. for negative feedback inhibition of neuronal NO synthesis. NOS-III, in contrast, is phosphorylated with much faster kinetics.
FIG. 6. Effects of cAMP-and cGMP-dependent protein kinase II on the Ca 2؉ /CaM dependence of NOS-III activity. Purified recombinant NOS-III (0.25 M) was phosphorylated by either active (f) or denatured, inactive (Ⅺ) C subunit, as well as by active (q) or denatured, inactive (E) cGK II (0.05 M each) as described under Fig. 2B, and subsequently NOS-III activity was determined as described under "Experimental Procedures." A, in the presence of 1 mM CaCl 2 , the halfmaximal CaM concentration for NOS-III activation was unchanged by kinase treatment, however, the activation V max was increased. B and C, when NOS-III was assayed in the absence or presence of Ca 2ϩ /CaM, both cGK II (B) and C subunit (C) activated NOS-III and caused a 10 -20% fraction (three left black columns) of the maximal Ca 2ϩ /CaM stimulated NOS-III activity (open column), or 5-10% of the maximal activity attained by Ca 2ϩ /CaM in the presence of active kinases (black column at far right), to become independent not only of Ca 2ϩ but also of CaM. Data represent mean Ϯ S.E. (n ϭ 6) FIG. 7. In vivo activation and subcellular localization of NOS-III. A, intact HUVEC were preincubated with 100 M Iloprost for 5 min, and lysed for NOS activity measurement as described under "Experimental Procedures." EDTA (1 mM) was added to the samples assayed in the absence of CaCl 2 . Data represent mean Ϯ S.E. (n ϭ 6); asterisks indicate significant difference in comparison to respective experiments without Iloprost (p Ͻ 0.05). B, shown is a representative Western blot analysis of NOS-III in total cell homogenate (H, 20 g of protein), and in the 100,000 ϫ g pellet (P100, 40 g of protein) and 100,000 ϫ g supernatant (S100, 15 g of protein) fractions of Iloprost-stimulated HUVECs from the above experiment. No change in NOS-III subcellular distribution was observed. The experiment shown was repeated twice with similar results.
Salerno et al. (31) also described a basal activity of purified NOS-I or NOS-III (5% of maximal) in the presence of EGTA and absence of CaM which they speculate may arise from a low steady-state concentration of the disinhibited NOS conformer. However, our recombinant NOS-III preparations from Sf9 cells did not display Ca 2ϩ /CaM-independent activity, this appeared only after cAK/cGK II-dependent phosphorylation. Chen et al. (18) recently reported that isolated rat heart NOS-III displayed already detectable phosphorylation of Thr 495 and Ser 1177 (the cAK/cGK phosphorylation sites) and that phosphorylation of Ser 1177 by an AMP-activated protein kinase would further activate the enzyme in concert with Ca 2ϩ /CaM. While our article was under review, several publications appeared demonstrating the phosphorylation and activation of NOS-III by Akt (protein kinase B) (32)(33)(34)(35). All reports demonstrated that phosphorylation of Ser 1177 enhanced NOS-III enzyme activity and NOS sensitivity to low Ca 2ϩ concentrations. However, none of the reports demonstrated Ca 2ϩ -independent activation of purified NOS-III like that we have shown for cAMP-and cGMP-dependent protein kinase. Interestingly, protein kinase B and cAK not only have overlapping substrate phosphorylation site consensus sequences, as shown here for NOS-III Ser 1177 and elsewhere for cAMP response elementbinding protein Ser 133 (36), but also a recent report has shown cAK activation of protein kinase B via a phosphatidylinositol 3-kinase-independent pathway (37). One of the reports of protein kinase B phosphorylation of NOS-III on Ser 1177 (35) also stated (data was not shown) that 8-(para-chlorophenylthio)-cGMP, a selective activator of cGK, did not cause phosphorylation of NOS-III in bovine aorta endothelial cells. HUVEC do not contain detectable amounts of cGK, and other endothelial cells which have been shown to contain cGK, rapidly lose it upon passaging (38), possibly explaining their data.
In addition to substrate specificity, factors such as cellular and subcellular co-localization may have a role in determining kinase interaction with NOS-III in intact cells. The cAMP-dependent protein kinase is fairly ubiquitous and therefore may co-localize with NOS-III in a number of cells, such as endothelial cells, cardiac myocytes, endocardium, cells of conduction tissue (39), hippocampal pyramidal cells (40), and bone cells of osteoblastic lineage (41). cGK II is present in high concentrations in intestinal epithelial microvilli, and in certain cell types of kidney, brain, lung, and bone (42)(43)(44)(45)(46). Detailed cellular co-localization of NOS-III and cGK II needs further investigation, potential sites of co-localization certainly being brain and bone.
Regarding subcellular localization, NOS-III is modified by N-terminal myristoylation and palmitoylation which targets it to plasmalemmal caveolae in endothelial cells and cardiac myocytes (47,48). Likewise, cGK II is localized via N-terminal myristoylation to the membrane, whereas cGK type I␣ and I␤b are primarily soluble proteins (for review, see Ref. 49). cAK is targeted to distinct intracellular sites via association of its regulatory subunit with anchoring proteins (for review, see Ref. 50). Therefore it is tempting to speculate that, in addition to substrate specificity, co-localization of cGK II/cAK and endothelial NOS-III at the membrane may contribute to rapid phosphorylation and regulation. For example, in intestine, cGK II and cAK, but not cGK I, regulate the integral membrane Cl Ϫ channel CFTR and intestinal secretion (46,51), and a major determinant of the differential effects of cGK II versus cGK I appears to be the membrane versus soluble subcellular localizations of cGK (52).
In conclusion, our data demonstrate a selective activating effect of cGK II and cAK on NOS-III but no other NOS forms. Moreover, they extend the mechanisms for regulating NO syn-thesis to include direct induction of a Ca 2ϩ -independent state of NOS-III by phosphorylation. Additionally, phosphorylationevoked Ca 2ϩ /CaM-independent or -dependent NOS-III activity may constitute, in the case of cGK II, an example of internal positive feedback enhancement of NO production, or, in the case of cAK, an example of signaling cross-talk between the cAMP and NO/cGMP pathways.