Perspectives in Pharmacology Post-Translational Control of Endothelial Nitric Oxide Synthase: Why Isn’t Calcium/Calmodulin Enough?

Endothelial nitric oxide synthase (eNOS) is important for cardiovascular homeostasis, vessel remodeling, and angiogenesis. Given the impact of endotheliumderived nitric oxide (NO) in vascular biology, much work in the past several years has focused on the control of NO synthesis by regulatory proteins that influence its function. Indeed calcium-activated calmodulin is important for regulation of NOS activity. Herein we discuss why other proteins, in addition to calmodulin, are necessary for eNOS regulation and summarize the biology of negative and positive regulators of eNOS function in vitro, in cells, and in blood vessels. Endothelium-derived nitric oxide (NO), the classic relaxing factor discovered by Furchgott in 1980, is produced by the enzyme endothelial nitric oxide synthase (eNOS). Early observations by pharmacologists studying endothelium-dependent relaxations of blood vessels and the release of relaxing factor by cultured endothelial cells demonstrated that removal of extracellular calcium from media solutions blocked agonist-induced release of NO, suggesting that calcium was important for the release of endothelium-derived NO (Singer and Peach, 1982; Forstermann et al., 1991). Additional studies using broken cell systems documented that calcium removal or antagonism of calmodulin (CaM) with inhibitors blocked the generation of NO and NOS activity, suggesting that eNOS was a calcium-calmodulin-requiring enzyme (Busse and Mulsch, 1990; Forstermann et al., 1991). The requirement for calcium-calmodulin was proven upon purification of eNOS to homogeneity (Pollock et al., 1991) and rationalized by the presence of a calmodulin binding motif in the deduced amino acid sequence of the cloned eNOS cDNA. In the past fours years many laboratories have described proteins other than CaM that may negatively or positively impact eNOS function. Insights into the need for additional regulatory proteins important for NO production from endothelial cells stemmed from observations that eNOS was an N-myristoyl protein (Pollock et al., 1992).N-Myristoylation is important for the subcellular targeting of discrete microdomains of cells, and mutations that block N-myristoylation impede proper subcellular targeting and various aspects of signal transduction. Indeed, expression of a nonacylated form of eNOS did not affect enzymatic activity in broken cell lysates but prevented calcium ionophore-stimulated NO release, arguing that additional mechanisms other than CaM, per se, were important for the fidelity of signal transduction coupling to eNOS (Sakoda et al., 1995; Sessa et al., 1995). In addition, the hypothesis that eNOS had to be localized to proper intracellular membranes to be near to other regulatory proteins (scaffolds, chaperones, kinases) provided the rationale for the discovery of additional protein regulators of eNOS function. Described below are putative regulators of eNOS function that have been shown to inhibit or enhance eNOS activity and NO release (Table 1). This work is supported by grants from the National Institutes of Health (RO1 HL57665, HL61371, and HL64793 to W.C.S.; T32HL10183 to D.F.) and a Grant-in-Aid from the American Heart Association (National Grant to W.C.S.). J.P.G. is the recipient of a fellowship from the Canadian Institutes of Health Research. W.C.S. is an Established Investigator of the American Heart Association. ABBREVIATIONS: NO, nitric oxide; eNOS, endothelial NO synthase; ACh, acetylcholine; B2, bradykinin 2; CaM, calmodulin; Erk, extracellular signal-related kinase; GA, geldanamycin; GST, glutathione S-transferase; hsp90, heat shock protein 90; ID4, intracellular domain 4; IGF, insulin-like growth factor; NOSIP, nitric oxide synthase interacting protein; PI-3K, phosphatidylinositol 3-kinase; VEGF, vascular endothelial growth factor. 0022-3565/01/2993-818–824$3.00 THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS Vol. 299, No. 3 Copyright © 2001 by The American Society for Pharmacology and Experimental Therapeutics 900046/942728 JPET 299:818–824, 2001 Printed in U.S.A. 818 at A PE T Jornals on July 0, 2017 jpet.asjournals.org D ow nladed from Negative Regulatory Proteins Caveolin. Caveolin, being the major coat protein of caveolae, has several faces that may influence the biology of proteins that localize to cholesterol-rich plasmalemma caveolae. Indeed caveolin-1 is necessary for the biogenesis of caveolae through an unknown mechanism (Smart et al., 1999). In addition, caveolin-1 can serve as a cholesterol binding protein and traffic cholesterol from the endoplasmic reticulum through the Golgi to the plasma membrane. Finally, caveolin has the capacity to directly interact with other intracellular proteins such as c-Src and H-Ras through amino acids 82– 101, the putative scaffolding domain (Smart et al., 1999). Indeed, three groups independently demonstrated that eNOS could directly interact with caveolin-1 or caveolin-3 (Feron et al., 1996; Garcı́a-Cardeña et al., 1996; Ju et al., 1997). The primary binding region of caveolin-1 for eNOS is within amino acids 60–101 and, to a lesser extent, amino acids 135–178 (Garcia-Cardena et al., 1997; Ju et al., 1997). Furthermore, the caveolin-eNOS immunocomplex is disrupted in the presence of caveolin scaffolding peptides (amino acids 82–101) (Michel et al., 1997b). eNOS contains a consensus caveolin binding motif (Smart et al., 1999) located within amino acids 350–358. The importance of the caveolin interaction with eNOS has been most reproducibly demonstrated by the effects of caveolin scaffolding peptides and GST-caveolin on NOS activity. Incubation of pure eNOS with peptides derived from the scaffolding domains of caveolin-1 and -3 result in inhibition of eNOS activity (Garcia-Cardena et al., 1997). In cotransfection experiments, caveolin over-expression in COS-7 cells resulted in a reduction of eNOS activity (Michel et al., 1997b), and a reduction in NO release was also observed (Garcia-Cardena et al., 1997). Furthermore, mutagenesis of the predicted caveolin binding motif within eNOS blocked the ability of caveolin to suppress NO release in these latter experiments (GarciaCardena et al., 1997). The reduction of eNOS activity by caveolin peptides, or over-expressed caveolin, is reversed by exogenous addition of calmodulin, suggesting a reciprocal regulation of eNOS by calmodulin, an activator, and caveolin, an inhibitor (Michel et al., 1997a). Collectively, these overall results suggest that NO production is negatively regulated by interactions with caveolin and that for NO release to occur, the inhibitory clamp by caveolin must be overcome. CaM has been proposed to be solely responsible for the dissociation of eNOS from caveolin (Michel et al., 1997a). However, the relationship between caveolin as an inhibitor of eNOS and CaM as its allosteric modulator has not been examined in light of new findings demonstrating a role for other positive and negative regulators of eNOS activation. Another important issue is that there are no direct data showing more NO release from cells that do not express caveolins or that disruption of the eNOS-caveolin complex can lead to increased or prolonged NO release from cells, fundamental experiments if caveolin-1 truly negatively regulates eNOS and NO release. In Vivo Evidence Supporting the eNOS-Caveolin Interaction. To date, caveolin knockout mice are not available; therefore, examining endothelial function in these mice is not yet feasible. However, recent work using the caveolin scaffolding domain as a surrogate for caveolin has demonstrated that eNOS can be regulated in situ. Exposure of permeabilized cardiac myocytes to the caveolin-3 scaffolding domain peptide (amino acids 55–74), but not a scrambled version, antagonized the negative chronotropic actions of carbachol (Feron et al., 1998). Our group recently used a membranepermeable form of the caveolin-1 scaffolding domain (amino acids 82–101) by fusing it to a cell-permeable leader sequence (Bucci et al., 2000a). Exposure of the peptide to blood vessels resulted in uptake into the endothelium and adventitia and blockade of ACh-induced relaxations, with no effect on relaxant responses to sodium nitroprusside or the release of prostacyclin, showing that in an intact blood vessel, the caveolin peptide is a potent inhibitor of eNOS. In addition, the peptide also blocked inflammation in two different models by influencing vascular permeability, suggesting that peptidomimetics may be useful therapeutically. With respect to disease mechanisms that may influence the caveolin/eNOS interaction, there is evidence that in a rat model of cirrhosis, caveolin-1 is over-expressed, more caveolin-1 interacts with eNOS, and the basal and stimulated production of NO is depressed (Shah et al., 1999a), suggesting that this interaction may increase portal pressures and contribute to the disease state. Intracellular Domains of G-Protein-Coupled Receptors. Work by Venema et al. (1996) has shown that the intracellular domain 4 (ID4) of the bradykinin 2 (B2) and the angiotensin II R1 receptors can negatively regulate eNOS activity in vitro (Ju et al., 1998). Indeed, eNOS coprecipitated with the B2 receptor and in vitro interacted with a GST fusion of ID4, and synthetic peptides from ID4 inhibited eNOS activity in a dose-dependent manner in vitro. Mechanistically, the ID4 peptide has been shown to affect NOS catalysis by interference with flavin to heme electron transfer (Golser et al., 2000). The concept that a receptor can directly interact with eNOS is extremely novel, suggesting that signaling, albeit negative signaling to eNOS, can occur in the absence of a G-protein intermediate. However, direct evidence supporting the physiological relevance of this interaction is presently unavailable. NOSIP. NOSIP is the newest protein to interact with eNOS (Dedio et al., 2001). NOSIP is a 34-kDa protein that was initially identified as an eNOS binding partner. The interaction between NOSIP and eNOS has been shown both in vitro and in vivo, and through deletion analysis, NOSIP was shown to bind eNOS between amino acids 366 and 486. Stimulation of cells with calcium ionophore does not change the association of NOSIP and eNOS; however, a peptide derived from the scaffolding domain of caveolin (82–101) is able to displace eNOS from NOSIP. NOSIP does not affect eNOS activity assays in vitro but, when coexpressed in cells, does reduce ionomycin-stimulated NO release. The ability of NOSIP to reduce NO release from intact cells is due to the redistribution of eNOS from the plasma membrane to intracellular compartments. The specificity of NOSIP to eNOS and the true function of NOSIP are not known. Positive Regulatory Proteins Calmodulin. The first protein shown to be involved in eNOS regulation was calmodulin (CaM). Early studies on neuronal NOS and eNOS (Bredt and Snyder, 1990; Forstermann et al., 1991), demonstrated that purified NOS utilized CaM as an activator of NO synthesis. Mechanistically, CaM binding to a canonical CaM binding motif can displace an Protein-Protein Modulation of NO Release 819 at A PE T Jornals on July 0, 2017 jpet.asjournals.org D ow nladed from adjacent autoinhibitory loop on eNOS and neuronal NOS, thus facilitating NADPH-dependent electron flux from the reductase domain of the protein through to the oxygenase domain. The terminal electron acceptor in the oxygenase domain is heme, which can bind oxygen for insertion into the NOS substrate, L-arginine. To date, there are no papers documenting that CaM can actually be recruited to eNOS in a stimulus-dependent manner and that the recruitment occurs contemporaneously with NO release. Pharmacological evidence using inhibitors of CaM or calcium-free buffers have indirectly shown the requirement for CaM. Recent work has shown that eNOS immunoprecipitated from human endothelial cells has immunoreactive CaM bound to it (Russell et al., 2000). Moreover, upon challenge of the cells with estrogen, the amount of CaM recovered in the eNOS immunocomplex does not change. This suggests that CaM may serve as a tightly bound prosthetic group, akin to CaM found in inducible NOS (Cho et al., 1992), and that regulation of the affinity of CaM interactions with NOS may occur through subtle changes in free calcium levels. Heat Shock Protein 90 (hsp90). The hsp90 family is a group of highly conserved stress proteins that are expressed in all eukaryotic cells (Pratt, 1997). Two genes encode hsp90, with the human gene products hsp90 and hsp90 having 86% sequence homology. The hsp90 is highly abundant in cells, accounting for 1 to 2% of cytosolic protein, and is localized to the cytoplasm, with a small amount found in the nucleus and cytoskeleton (Pratt, 1997). The main function of hsp90 has been its involvement in a multicomponent chaperone system that is responsible for the proper folding of proteins such as steroid receptors and cell cycle-dependent kinases (Pratt, 1997) However, the abundance of hsp90 associated with newly synthesized proteins suggests that this may not be its only function. There is increasing evidence that hsp90 may be an integral part of signal transduction in all cells. Indeed, hsp90 orthologs are important for tyrosine kinase signaling in Drosophila and receptor-G-protein signaling in yeast. Previously it had been shown that eNOS coprecipitated with a 90-kDa tyrosine-phosphorylated protein, later shown to be hsp90 (Venema et al., 1996; Garcia-Cardena et al., 1998). Indeed, hsp90 was associated with eNOS in resting endothelial cells; and treatment of cells with four distinct stimuli that cause NO release, vascular endothelial growth factor (VEGF), histamine, fluid shear stress, and estrogen, all enhanced the interaction between hsp90 and eNOS in a time frame consistent with NO release (Garcia-Cardena et al., 1998; Russell et al., 2000). The rapid stimulus-dependent formation of the hsp90-eNOS hetero-complexes suggests that it occurs simultaneously with other signaling events such as the mobilization of intracellular calcium and/or protein phosphorylation. The mechanism of how hsp90 regulates eNOS function is less clear. Previously we have shown that hsp90 can directly activate eNOS in vitro (Garcia-Cardena et al., 1998), and coexpression of eNOS with hsp90 in COS cells increased NOS activity in broken cell lysates. These results suggest that hsp90 may act as an allosteric modulator of eNOS by inducing a conformational change in the enzyme that results in increased activity or possibly stabilize the “activation” complex. A recent paper has shown that hsp90 increases CaM affinity for neuronal NOS (Song et al., 2001). Alternatively, hsp90 may act as a scaffold for the recruitment of other regulatory molecules including kinases and phosphatases that may influence eNOS function. Interactions Between Caveolin, CaM, and hsp90. As mentioned previously, CaM has been proposed to be exclusively responsible for the dissociation of eNOS from caveolin. Recently the relationship between caveolin as an inhibitor of eNOS and CaM as its allosteric modulator has been examined in light of hsp90 as an additional regulatory protein. Labeling of endothelial cells with [S]methionine followed by immunoprecipitation of eNOS resulted in the appearance of several co-associated radiolabeled proteins interacting substoichiometrically with eNOS (Gratton et al., 2000). Western blotting of these immunoprecipitated proteins demonstrated the presence of eNOS, caveolin-1, and hsp90 in the same complex. Moreover, the addition of exogenous CaM weakly displaced caveolin from CaM. Reconstitution of the heterotrimeric complex in vitro showed the eNOS interaction with both hsp90 and caveolin, but the latter proteins did not interact with each other, demonstrating that eNOS was the bridge holding the complex together. Interestingly, the binding of caveolin to eNOS was displaced by the caveolin scaffolding domain peptide, but not by calcium-activated CaM, demonstrating that CaM cannot physically disrupt the eNOS-caveolin complex in vitro. However, hsp90, per se, did not influence the eNOS/caveolin interaction but facilitated the ability of CaM to displace caveolin from eNOS. These data are consistent with two potential models: 1) perhaps the “recruitment or activation” of hsp90 and CaM to eNOS results in weak physical displacement of eNOS from caveolin, TABLE 1 Potential eNOS interacting proteins ProteinAgonistAssociationEffect of BoundProteinReferences CaveolinA23187/bradykininDecreased/increasedInhibitoryJu et al., 1997; Michel et al.,1997bB2 receptorBradykinin, ionophoreDecreasedInhibitoryGolser et al., 2000NOSIPIonomycinNo changeInhibitoryDedio et al., 2001Calmodulin??Stimulatoryhsp90VEGF, histamine, shearstress, estrogenIncreasedStimulatoryVenema et al., 1996; Garcia-Cardena et al., 1998Dynamin-2IonomycinIncreasedStimulatoryCao et al., 2000Erk 1/2BradykininDecreasedInhibitoryBernier et al., 2000Raf-1BradykininDecreasedInhibitoryBernier et al., 2000AktBradykininDecreasedN.D.Bernier et al., 2000 N.D., not determined.a It is presumed that more CaM is bound.820 Fulton et al. atAPETJornalsonJuly0,2017jpet.asjournals.orgDownladedfrom but the complex remains in caveolae; or 2) hsp90 and calci-um-activated CaM coexist with eNOS bound to caveolin, anda slight change in eNOS conformation, in the absence of bulktranslocation away from caveolin, allows for efficient stimu-lus-response coupling.In Vivo Evidence Supporting the eNOS-hsp90 Inter-action. To study the relationship between hsp90-mediatedsignaling and NO production, a specific inhibitor of hsp90,the ansamycin antibiotic geldanamycin (GA) was used. GAbinds to the unique ATP site of hsp90 and a related protein,GRP94, and influences the conformational stability of hsp90binding to its substrates (Pratt, 1997). GA attenuated hista-mine and VEGF-stimulated cGMP production in culturedendothelial cells and blocked ACh-induced vasorelaxation ofrat aortic rings (Garcia-Cardena et al., 1998), middle cerebralartery (Khurana et al., 2000), and flow-induced dilation(Viswanathan et al., 1999), indicating that hsp90 signalingwas crucial for NO release and endothelial function. Furthersupport for the relevance of hsp90/eNOS interactions in vivowas demonstrated in a model of portal vein ligation in rats(Shah et al., 1999b) and in a model of inflammation (Bucci etal., 2000b). In the former study, the physical interaction ofhsp90 with eNOS isolated from the mesenteric microcircula-tion was documented, and GA attenuated ACh-dependentvasodilatation to the same extent as conventional NOS in-hibitors. In portal hypertensive rats, eNOS protein levels arenot changed compared with control rats, but NOS activity ismarkedly enhanced in the mesenteric tissue of hypertensiverats. The enhanced activity correlated with hyporesponsive-ness to the vasoconstrictor methoxamine, and GA potenti-ated the methoxamine-induced vasoconstriction after portalvein ligation, partially reversing the hyporeactivity to thisagent, indicating that hsp90 can act as a signaling compo-nent leading to NO-dependent responses in the mesentericmicrocirculation. In the latter study, GA inhibited inflamma-tion in a dose-dependent manner, an effect as potent as asteroid. Because GA blocks NO release and NOS inhibitorsreduce edema formation, it is possible that drugs that spe-cifically inhibit hsp90 will be good anti-inflammatory drugs.Dynamin-2. Dynamin-2 belongs to the family of largeGTPases and is believed to be involved in vesicle formation,receptor-mediated endocytosis, caveolae internalization andvesicle trafficking in and out of the Golgi. Dynamin-2 hasbeen shown by confocal microscopy to colocalize with eNOSin the Golgi membranes of endothelial cells and to bind eNOSdirectly, both in vivo and in vitro (Cao et al., 2000). Previouswork by others has documented dynamin-2 in the plasmamembrane of endothelial cells, suggesting that, like eNOS,dynamin-2 exists in the plasma membrane and Golgi. Basedon extensive in vitro studies, dynamin-2 has an affinity foreNOS in the nanomolar range. In cells, the association be-tween eNOS and dynamin-2 is increased by calcium iono-phore, and in in vitro activity assays, dynamin has beenshown to directly augment NOS activity. The physiologicalrole of this interaction has yet to be explored.Signaling Kinases: Akt, RAF, and Erk. Immunoprecipi-tation of eNOS from endothelial cells also results in thecoprecipitation of kinases and related proteins including Akt,Erk, and RAF (Michell et al., 1999; Bernier et al., 2000). Theinteraction of these proteins with eNOS distinguishes themfrom other protein/protein interactions in that it is the effectof the kinase (phosphorylation) rather than the presence ofthe protein that influences eNOS activity. The relevance of Akt binding is clear based on in vitro and in vivo evidence describing the phosphorylation site (see below), whereas the phosphorylation by Erk has not been characterized. However, the association of these proteins to eNOS following agonist activation is indicative of a dynamic multiprotein signaling complex influencing eNOS function (i.e., NOS-os-