Localization of caveolin-3 in the sinus endothelial cells of the rat spleen

The localization of caveolins in the sinus endothelial cells of the rat spleen has been demonstrated by confocal laser scanning and electron microscopy. Caveolin-3, a muscle-specific caveolin, was detected by Western blot analysis and immunofluorescence microscopy of isolated sinus endothelial cells and tissue cryosections of the spleen. During the immunofluorescence microscopy of isolated endothelial cells, both caveolin-3 and caveolin-1 were found. In tissue cryosections of the spleen, caveolin-3, as well as caveolin-1 and -2, was present in the contours and cytoplasm of the cells. Immunogold electron microscopy of tissue cryosections revealed caveolin-3, -1, and -2 to be present in caveolae in the apical, lateral, and basal plasma membranes and some vesicular profiles in the cytoplasm of sinus endothelial cells. Furthermore, caveolin-3 was colocalized with caveolin-1 in the same caveolae in the apical, lateral, and basal plasma membranes. Stress fibers and tubulovesicular structures were situated in the vicinity of caveolae labeled with anti-caveolin-3, anti-caveolin-1, and anti-caveolin-2 antibodies. It is speculated that caveolae in sinus endothelial cells play an important role in the constriction of stress fibers.     

Interactions of high-density lipoprotein 3 with brain capillary endothelial cells

High-density lipoprotein 3 (HDL3) binds to capillary endothelial cells when their lumen surfaces are exposed to 125I-HDL3 by post-mortem perfusion of whole brain. Kinetic studies of binding of HDL3 to isolated membranes show that HDL3 binds only to endothelial membranes with high affinity (Kd = 7 micrograms/ml). Trypsin treatment of membranes abolishes HDL3 binding. High-affinity binding sites for HDL3 were recovered when endothelial cells from bovine brain capillaries were maintained in culture (Kd = 13 micrograms/ml HDL3 protein). The characteristics of the binding were preserved up to the 6th passage. Competition experiments using isolated luminal membranes or cultured endothelial cells indicate that only HDL3 and not LDL or methylated LDL, are able to compete binding of 125I-HDL3. Furthermore, the inhibition of 125I-HDL3 binding by lipoprotein A-I and lipoprotein A-I:A-II strongly suggests that apolipoprotein A-I is implicated in the formation of HDL3-receptor complexes. The binding is increased by loading cells with free cholesterol or LDL cholesterol. In addition, surface-bound 125I-HDL3 remains sensitive to mild trypsin treatment after subsequent incubation of BBCE at 37 degrees C. HDL3 bound to the cell surface is not endocytosed, but rather rapidly released into the medium after binding (t1/2 = 5 min).     

Rapid activation of endothelial NO synthase by estrogen: evidence for a steroid receptor fast-action complex (SRFC) in caveolae

Estrogen has important atheroprotective and vasoactive properties related to its capacity to stimulate nitric oxide (NO) production by endothelial NO synthase. Previous work has shown that these effects are mediated by estrogen receptor (ER) alpha functioning in a nongenomic manner via calcium-dependent, MAP kinase-dependent mechanisms. Recent studies have demonstrated that estradiol (E(2)) activates eNOS in isolated endothelial plasma membranes in the absence of added calcium, calmodulin or eNOS cofactors. Studies of blockade by ICI 182,780 and by ER alpha antibody, and also immunoidentification experiments indicate that the process is mediated by a subpopulation of plasma membrane-associated ER alpha. Fractionation of endothelial cell plasma membranes has further revealed that ER alpha protein is localized to caveolae, and that E(2) causes stimulation of eNOS in isolated caveolae which is ER-dependent and calcium-dependent, whereas noncaveolae membranes are insensitive. Furthermore, in intact endothelial cells the activation of eNOS by E(2) is prevented by pertussis toxin, and exogenous GDP beta S inhibits the response in isolated plasma membranes. Coimmunoprecipitation studies have shown that E(2) exposure causes interaction between ER alpha and G(alpha i) on the plasma membrane, and eNOS activation by E(2) is enhanced by overexpression of G(alpha i) and attenuated by expression of a protein regulator of G protein signaling (RGS), RGS4. Thus, a subpopulation of ER alpha is localized to caveolae in endothelial cells, where they are coupled via G(alpha i) to eNOS in a functional signaling module. Emphasizing the dependence on cell surface-associated receptors, these observations provide evidence for the existence of a steroid receptor fast-action complex, or SRFC, in caveolae.     

Lectin binding and uptake and glycoprotein characterization of isolated porcine aortic endothelial and smooth muscle cells

Endothelial and smooth muscle cells were isolated from porcine aorta and kept in short-term culture. To determine the terminal carbohydate composition of the plasma membranes from both cell populations, the cells were incubated with a panel of fluorescein-labelled lectins. Both cell populations shared a number of terminal carbohydrates, but the N-galactosamine specific lectin Wistaria floribunda agglutinin labelled only the endothelial cells. A lectin which selectively labelled smooth muscle cells was not found. Western blot analysis of isolated endothelial cell membrane glycoproteins indicated that most membrane glycoproteins are labelled by Wistaria floribunda agglutinin.     

Growth of endothelial cells on microfabricated silicon nitride membranes for an<Emphasis Type="Italic">in vitro</Emphasis> model of the blood-brain barrier

The blood-brain barrier (BBB) is composed of the brain capillaries, which are lined by endothelial cells displaying extremely tight intercellular junctions. Several attempts at creating anin vitro model of the BBB have been met with moderate success as brain capillary endothelial cells lose their barrier properties when isolated in cell culture. This may be due to a lack of recreation of thein vivo endothelial cellular environment in these models, including nearly constant contact with astrocyte foot processes. This work is motivated by the hypothesis that growing endothelial cells on one side of an ultra-thin, highly porous membrane and differentiating astrocyte or astrogliomal cells on the opposite side will lead to a higher degree of interaction between the two cell types and therefore to an improved model. Here we describe our initial efforts towards testing this hypothesis including a procedure for membrane fabrication and methods for culturing endothelial cells on these membranes. We have fabricated a 1 μm thick, 2.0 μm pore size, and ∼55% porous membrane with a very narrow pore size distribution from low-stress silicon nitride (SiN) utilizing techniques from the microelectronics industry. We have developed a base, acid, autoclave routine that prepares the membranes for cell culture both by cleaning residual fabrication chemicals from the surface and by increasing the hydrophilicity of the membranes (confirmed by contact angle measurements). Gelatin, fibronectin, and a 50/50 mixture of the two proteins were evaluated as potential basement membrane protein treatments prior to membrane cell seeding. All three treatments support adequate attachment and growth on the membranes compared to the control.     

Functional interaction of caveolin-1 and eNOS in myocardial capillary endothelium revealed by immunoelectron microscopy.

Immunogold labeling on samples of isolated perfused rat hearts embedded by an innovative low-temperature LR White procedure provided detailed insight into the interaction of caveolin-1 and endothelial NOS in myocardial capillary endothelium at the subcellular level. Separately, the localization of caveolin-1 and eNOS at caveolae under steady state conditions was visualized. A double-labeling experiment supported their close co-localization. Short-term bradykinin stimulation caused a detectable dissociation of eNOS from caveolin and its redistribution to different cell compartments, whereas caveolin itself remained stationary at caveolae. Morphometric analysis revealed that more than 80% of detectable eNOS was co-localized with caveolin-1 at caveolae under control conditions. After brief stimulation for 2 min with 10(-7) M bradykinin, only 26% of the eNOS signals were associated with caveolin-1 and randomly distributed over the endothelial cells. After stimulation, eNOS was found at the plasmalemmal and intracellular membranes, freely in the cytoplasm, and at outer mitochondrial membranes.     

Monoclonal antibody to CD9 inhibits platelet-induced human endothelial cell proliferation

Platelets are anucleate cytoplasmic fragments derived from bone marrow megakaryocytes, and endothelial cells constitute the barrier between bloodstream and adjacent tissues. Although platelets are thought to regulate the biological functions of endothelial cells, the molecular mechanisms involved are poorly understood. With human umbilical vein endothelial cells and freshly isolated platelets, we established an in vitro model of platelet-induced endothelial cell proliferation. Platelets stimulated endothelial cell proliferation in a dose-dependent manner and transwell experiments with semi-permeable membranes suggested that direct cell-to-cell contacts were required. We developed mAbs against platelets and selected a mAb that blocks their proliferative effect. We purified the antigen by immunoprecipitation and identified it by Q-TOF MS analysis as the tetraspanin CD9. Since both platelets and endothelial cells expressed CD9 strongly on their surfaces we carried out a pre-treatment experiment that showed that CD9 molecules on the endothelial cells participate in the mitogenic effect of the platelets. The inhibitory effect of our mAb was comparable to that of a well-known functional anti-CD9 mAb. These results suggest that the tetraspanin CD9 plays an important role in endothelial regeneration.     

The relation between connective tissue cells and intercellular substances, including basement membranes.

Basement membranes are distributed widely in the body forming an extracellular matrix for epithelial and endothelial cells. The collagenous and glycoprotein constituents of basement membranes are synthesized by these two cell types. Disturbance of the interactions between basement membranes and their associated epithelial and endothelial cells can lead to the pathological changes seen in diseases involving basement membranes. These changes are illustrated here by reference to glomerulonephritis induced by the deposition of immune complexes in the glomerulus of the kidney, and chronic inflammatory changes occurring in the lung after inhalation of asbestos. In these diseases basement membrane changes can occur in several ways. Hydrolytic enzymes released from inflammatory cells degrade basement membranes while other constituents by epithelial and endothelial cells. Alternatively the physical separation of epithelial and endothelial cells from their basement membrances by space-occupying substances such as immune complexes can interfere with feedback mechanisms leading to synthesis of basement membrane constituents and cell proliferation. Studies of these pathological changes at a cellular level should shed new light on the ways in which cells interact with their pericellular environment.     

Demonstration of a Na(+)/H(+) exchanger NHE1 in fresh bovine corneal endothelial cell basolateral plasma membrane.

Apical and basolateral plasma membranes of fresh bovine corneal endothelial cells were isolated using positively charged polyacrylamide beads. Marker enzyme assays demonstrated that the isolated apical and basolateral plasma membrane domains could be isolated and separated with relative purity. Western blotting with a polyclonal anti-NHE1 antibody detected a protein of 70 kDa in the basolateral plasma membrane isolate. NHE1 immunoreactivity was not detected in the apical membrane sample. This suggests that the Na(+)/H(+) exchanger, NHE1, is strictly localised to the basolateral membrane of fresh bovine corneal endothelial cells.     

Identification and characterization of human endothelial cell membrane binding sites for tissue plasminogen activator and urokinase

Cultured human endothelial cells synthesize and secrete two types of plasminogen activator, tissue plasminogen activator (t-PA) and urokinase (u-PA). Previous work from this laboratory (Hajjar, K.A., Hamel, N. M., Harpel, P. C., and Nachman, R. L. (1987) J. Clin. Invest. 80, 1712-1719) has demonstrated dose-dependent, saturable, and high affinity binding of t-PA to two sites associated with cultural endothelial cell monolayers. We now report that an isolated plasma membrane-enriched endothelial cell fraction specifically binds 125I-t-PA at a single saturable site (Kd 9.1 nM; Bmax 3.1 pmol/mg membrane protein). Ligand blotting experiments demonstrated that both single and double-chain t-PA specifically bound to a Mr 40,000 membrane protein present in detergent extracts of isolated membranes, while high molecular weight, low molecular weight, and single-chain u-PA associated with a Mr 48,000 protein. Both binding interactions were reversible and cell-specific and were inhibitable by pretreatment of intact cells with nanomolar concentrations of trypsin. The relevant binding proteins were not found in subendothelial cell matrix, failed to react with antibodies to plasminogen activator inhibitor type 1 and interacted with their respective ligands in an active site-independent manner. The isolated t-PA binding site was resistant to reduction and preserved the capacity for plasmin generation. In contrast, the isolated u-PA binding protein was sensitive to reduction, and did not maintain the catalytic activity of the ligand on the blot. The results suggest that in addition to sharing a matrix-associated binding site (plasminogen activator inhibitor type 1), both t-PA and u-PA have unique membrane binding sites which may regulate their function. The results also provide further support for the hypothesis that plasminogen and t-PA can assemble on the endothelial cell surface in a manner which enhances cell surface generation of plasmin.     

Separation of luminal and abluminal membrane enriched domains from cultured bovine aortic endothelial cells: monoclonal antibodies specific for endothelial cell plasma membranes

Two kinds of membrane (luminal and abluminal membrane domains) fractions have been isolated from bovine aortic endothelial cells by fractionation of whole cell homogenate on discontinuous sucrose density gradients. The luminal membrane domain was enriched 12-16-fold for angiotensin-converting enzyme activity and 8-10-fold in alkaline phosphatase activity. The abluminal membrane domain displayed an enrichment of 8-fold in (Na+ + K+)-ATPase activity. Both of the membrane domains were minimally contaminated with mitochondria, microsomes and Golgi bodies, as assessed by their corresponding marker enzyme activities. 125I-labeling of endothelial cell monolayers by the Enzymo-Bead lactoperoxidase-catalyzed iodination procedure, followed by isolation of membranes, revealed that the radioactivity was predominantly associated with membranes enriched in angiotensin-converting enzyme activity, corresponding to the luminal membrane domain. However, when cells were radioiodinated in suspension culture, radioactivity was found equally associated in both the luminal and abluminal membrane fractions. Electron microscopy of freeze-fractured and sectioned material showed both luminal and abluminal membrane domains to be in the form of vesicles varying in size from 100 to 400 nm in diameter. To characterize the separation of endothelial cell membrane domains, we have attempted to prepare monoclonal antibodies specific for endothelial cells. Several clones were obtained, producing antibodies which bound to endothelial cells of arterial, venous and capillary origin. Two antibodies of these clones, XIVC6 and XVD2, were studied in more detail. In the ELISA assay, these antibodies reacted with bovine vascular endothelial cells, but not with human umbilical cord endothelial cells, nor with bovine corneal endothelial cells, smooth muscle cells or fibroblasts. Both of these antibodies are directed against an antigen of approximately 130 kDa, under reducing and non-reducing conditions, as assayed by the immunoprecipitation method. Western blot analysis of luminal and abluminal membrane fractions revealed that only MAb XVD2 reacted with an antigen, indicating that the antibody XIVC6 is directed against an epitope which is denatured by SDS. Moreover, MAb XVD2 preferentially reacted with the luminal membrane compared to the abluminal membrane domain of the endothelial cell. These monoclonal antibodies do not react with platelet membrane proteins, indicating that this 130 kDa membrane antigen is not common to both endothelial cells and platelets.(ABSTRACT TRUNCATED AT 400 WORDS)     

Stimulation-dependent recruitment of cytosolic phospholipase A2-alpha to EA.hy.926 endothelial cell membranes leads to calcium-independent association.

Cytosolic phospholipase A2-alpha (cPLA2-alpha) is a calcium-activated enzyme involved in agonist-induced arachidonic acid release. In endothelial cells, free arachidonic acid is predominantly converted into prostacyclin, a potent vasodilator and inhibitor of platelet activation. As the rate-limiting step in prostacyclin production is the generation of free arachidonic acid by cPLA2-alpha, this enzyme has become an attractive pharmacological target and the focus of many studies. Following stimulation with calcium-mobilizing agonists, cPLA2-alpha translocates to intracellular phospholipid membranes via its C2 domain. In this study, the calcium-induced association of cPLA2-alpha with EA.hy.926 endothelial cell membranes was investigated. Subcellular fractionation and immunofluorescence studies showed that following stimulation with histamine, thrombin or the calcium ionophore A23187, cPLA2-alpha relocated to intracellular membranes. Treatment of A23187-stimulated cells with EGTA or BAPTA-AM demonstrated that a substantial pool of cPLA2-alpha remained associated with membrane fractions in a calcium-independent manner. Furthermore, immunofluorescence microscopy studies revealed that cells stimulated for periods of greater than 10 min showed a high proportion of calcium-independent membrane-associated cPLA2-alpha. Calcium-independent membrane association of cPLA2-alpha was not due to hydrophobic or cytoskeletal interactions. Finally, the recombinant C2 domain of cPLA2-alpha exhibited calcium-independent membrane binding to membranes isolated from A23187-stimulated cells but not those isolated from nonstimulated cells. These findings suggest that novel mechanisms involving accessory proteins at the target membrane play a role in the regulation of cPLA2-alpha. Such regulatory associations could enable the cell to discriminate between the varying levels of cytosolic calcium induced by different stimuli.     

Localization of claudin-5 and ZO-1 in rat spleen sinus endothelial cells

Splenic sinus endothelial cells, which adhere through tight and adherens junctions, regulate the passage of blood cells through the splenic cord. The objective of this study was to assess the localization of tight junctional proteins, claudin-5 and ZO-1 in the sinus endothelial cells of rat spleen and to characterize spatial and functional relationships between tight and adherens junctions. Immunofluorescence microscopy of tissue cryosections demonstrated that claudin-5, ZO-1, and α-catenin were distinctly localized in the junctional regions of adjacent endothelial cells. Immunogold electron microscopy demonstrated claudin-5 localized in the tight-junctional fused membranes of adjacent endothelial cells. Immunogold labeling for ZO-1 was localized not only in the tight-junctional-fused membranes of endothelial cells but also in the junctional membrane. α-Catenin was intermittently localized along the juxtaposed junctional membranes of adjacent endothelial cells. Double-staining immunogold microscopy for claudin-5 and ZO-1, claudin-5 and VE-cadherin, ZO-1 and VE-cadherin, and ZO-1 and α-catenin demonstrated that ZO-1 was closely localized to VE-cadherin and α-catenin in their juxtaposed membranes of endothelial cells. Thus, ZO-1 might play an important role in regulating the cell–cell junctions of sinus endothelial cells for blood–cell passage through splenic cords. This work was supported by a Grant-in-Aid for Scientific Research (C), Japan.     

Characterization of the major heparan sulfate proteoglycan secreted by bovine aortic endothelial cells in culture. Homology to the large molecular weight molecule of basement membranes

To determine the precise architecture and functional characteristics of the subendothelial basal lamina, detailed information of the molecules contained in this structure is required. To this end, we have studied low passage bovine aortic endothelial cells and have isolated the major heparan sulfate-containing proteoglycan from the growth medium of the cells maintained under static culture conditions. This large macromolecule consists of a core protein approximately 500,000 daltons in mass and two to three glycan side chains as revealed by carbon/platinum rotary shadow casting. Specific antibodies raised by immunization of rabbits with the native or deglycosylated bovine molecule could be isolated from an immunoadsorption column prepared with a preparation isolated from the murine Engelbreth-Holm-Swarm tumor. The antibodies purified by immunoaffinity react with basement membranes of blood vessels, lung, liver, or skin, and this reactivity is indistinguishable, at least for the organs studied, from the reactivity of antibodies specific for the Engelbreth-Holm-Swarm tumor-derived high molecular weight heparan sulfate proteoglycan isolated previously. Immunoelectron microscopy of frozen ultrathin tissue sections from the kidney indicates localization of the epitope(s) also in the basement membranes of the renal glomeruli and tubuli. The close structural relationship and homology between the aortic endothelial cell product can be demonstrated even more convincingly by two-dimensional peptide mapping procedures. The peptide patterns from the bovine and mouse products of approximately 500 kDa are nearly indistinguishable. Maps of polypeptides of molecular masses ranging from 400 to 150 kDa, which are found in the bovine as well mouse tumor preparation, are clearly related to each other and suggest that this proteoglycan is quite sensitive to degradation by tissue proteases. Thus the data presented here strongly suggest that the large proteoglycan previously isolated and described as a tumor cell product can be produced by normal cells.     

Microparticle formation after co-culture of human whole blood and umbilical artery in a novel in vitro model of flow

Cardiovascular disease (CVD) is now the largest killer in western society, and the importance of interactions between vascular endothelium and circulating blood components in disease pathogenesis is well established. Microparticles are a heterogeneous population of <1 μm blood borne particles that arise from blebbing or shedding of cell membranes. The microparticle population includes several classes of apoptotic bodies; however, increased numbers of procoagulant microparticles have been described in plasma from people with CVD. We have previously demonstrated that interactions of monocytes and platelets with isolated inflamed endothelial cells lead to production of pro-coagulant tissue factor bearing microparticles under laminar flow conditions. Here we have investigated microparticle production after perfusion of human whole blood through intact inflamed human umbilical artery. When blood was perfused through umbilical arteries which had been pre-stimulated with tumour necrosis factor (TNFα) for 18 h under flow conditions, there was significantly increased production of microparticles from both platelet and non-platelet sources, in particular from erythrocytes. To determine whether microparticles generated during interactions with inflamed endothelium could induce a pro-inflammatory response in trans, we isolated microparticles by centrifugation after co-culture and incubated with isolated quiescent endothelial cells followed by measurement of reactive oxygen species formation. Microparticles derived from co-culture with inflamed endothelium induced significantly enhanced levels of reactive oxygen species (ROS). These data suggest that presence of an inflamed endothelium causes release of pro-inflammatory microparticles from circulating blood cells, which could contribute to prolonged endothelial activation and subsequent atherosclerotic changes in blood vessels subjected to inflammatory insult.     

Isolation and characterization of basement membrane and cell proteoheparan sulphates from HR9 cells

The mouse teratocarcinoma cell line HR9 was investigated for proteoheparan sulphate production. Four species of proteoheparan sulphate molecules were isolated and purified to homogeneity. The proteoheparan sulphate isolated from the tissue-culture medium contains four heparan sulphate side-chains of 25 kDa each, and its core protein has an approximate molecular mass of 50 kDa. The proteoheparan sulphates associated with the cells were separated into three individual species: cell proteoheparan sulphate I exhibits structural characteristics which are very similar to the proteoheparan sulphate isolated from the tissue culture medium; cell proteoheparan sulphates II and III contain one heparan sulphate chain of 25 kDa and 20 kDa, and core proteins of approximately 30 kDa and 25 kDa respectively. Antisera, raised against the medium form, react specifically with basement membranes in various tissues by immunofluorescence. This staining pattern was compared to the pattern observed with an antiserum which we have obtained to a proteoheparan sulphate species isolated from the plasma membrane of bovine aortic endothelial cells. The structural and immunological data suggest that basement membrane and plasma membrane proteoheparan sulphates are different biosynthetic products and are not directly related to each other.     

Washed human platelets prevent ischemia-reperfusion edema in isolated rabbit lungs

Washed human platelets prevent edema formation in isolated rabbit lungs infused with xanthine oxidase, an enzyme that injures endothelial membranes by generating extracellular oxidants. We hypothesized that platelets would similarly preserve membrane permeability in isolated lungs exposed to ischemia-reperfusion injury, a model that perturbs endothelial cells by the generation of intracellular oxidants. Isolated perfused rabbit lungs (IPL) were exposed to warm ischemia-reperfusion to cause lung edema. The infusion of washed human platelets (1.05 +/- 0.02 x 10(10) cells) prevented edema formation as measured by lung weight gain, wet-to-dry lung weight ratios, histological edema, and preservation of paraendothelial cell tight junctions. Inhibition of the platelet glutathione redox cycle with 1,3-bis(2-chloroethyl)-1-nitrosourea, dehydroepiandrosterone, or 1-chloro-2,4-dinitrobenzene interfered with platelet protective effects. In contrast, inhibition of platelet catalase with aminotriazole and H2O2 had no effect on platelet protection. Lung tissue malonyldialdehyde concentrations were similar in isolated lungs exposed to ischemia-reperfusion with or without the infusion of platelets. These results indicate that platelet attenuation of ischemia-reperfusion lung edema depends on platelet glutathione redox cycle antioxidants but not platelet catalase.     

Direct proteomic mapping of the lung microvascular endothelial cell surface in vivo and in cell culture

Endothelial cells can function differently in vitro and in vivo; however, the degree of microenvironmental modulation in vivo remains unknown at the molecular level largely because of analytical limitations. We use multidimensional protein identification technology (MudPIT) to identify 450 proteins (with three or more spectra) in luminal endothelial cell plasma membranes isolated from rat lungs and from cultured rat lung microvascular endothelial cells. Forty-one percent of proteins expressed in vivo are not detected in vitro. Statistical analysis measuring reproducibility reveals that seven to ten MudPIT measurements are necessary to achieve > or =95% confidence of analytical completeness with current ion trap equipment. Large-scale mapping of the proteome of vascular endothelial cell surface in vivo, as demonstrated here, is advisable because distinct protein expression is apparently regulated by the tissue microenvironment that cannot yet be duplicated in standard cell culture.     

Plasma membrane estrogen receptors are coupled to endothelial nitric-oxide synthase through Galpha(i)

Estrogen causes rapid endothelial nitric oxide (NO) production because of the activation of plasma membrane-associated estrogen receptors (ER) coupled to endothelial NO synthase (eNOS). In the present study, we determined the role of G proteins in eNOS activation by estrogen. Estradiol-17beta (E(2), 10(-8) m) and acetylcholine (10(-5) m) caused comparable increases in NOS activity (15 min) in intact endothelial cells that were fully blocked by pertussis toxin (Ptox). In addition, exogenous guanosine 5'-O-(2- thiodiphosphate) inhibited E(2)-mediated eNOS stimulation in isolated endothelial plasma membranes, and Ptox prevented enzyme activation by E(2) in COS-7 cells expressing ERalpha and eNOS. Coimmunoprecipitation studies of plasma membranes from COS-7 cells transfected with ERalpha and specific Galpha proteins demonstrated E(2)-stimulated interaction between ERalpha and Galpha(i) but not between ERalpha and either Galpha(q) or Galpha(s); the observed ERalpha-Galpha(i) interaction was blocked by the ER antagonist ICI 182,780 and by Ptox. E(2)-stimulated ERalpha-Galpha(i) interaction was also demonstrable in endothelial cell plasma membranes. Cotransfection of Galpha(i) into COS-7 cells expressing ERalpha and eNOS yielded a 3-fold increase in E(2)-mediated eNOS stimulation, whereas cotransfection with a protein regulator of G protein signaling, RGS4, inhibited the E(2) response. These findings indicate that eNOS stimulation by E(2) requires plasma membrane ERalpha coupling to Galpha(i) and that activated Galpha(i) mediates the requisite downstream signaling events. Thus, novel G protein coupling enables a subpopulation of ERalpha to initiate signal transduction at the cell surface. Similar mechanisms may underly the nongenomic actions of other steroid hormones.