INTRODUCTION

TOP ABSTRACT INTRODUCTION FUNCTIONAL MORPHOLOGY AND... MOLECULAR CASCADE RECRUITED IN... TOPOLOGICAL FEEDBACK REGULATION... ROLE OF LIPID COMPONENTS... NO AND MEMBRANE FLUIDITY CAVEOLAE AND LOCOMOTION: A... CAVEOLAE, NO, AND VASCULAR... CONCLUDING REMARKS REFERENCES

THE CONTEMPORARY MODEL OF the plasma membrane views it as a combination of lipid-disordered and lipid-ordered microdomains, or lipid rafts (83), enriched in glycosphingolipids and cholesterol (9). The lipid composition of these latter domains, characterized by tight acyl chain packing, explains their liquid-ordered phase, intermediate between the fluid and gel phases (10, 45). Multiple functions have been proposed for these domains, such as anchorage of various receptors, docking of elements of intracellular signaling cascades and different enzymes, trafficking of cholesterol, and regulation of permeability, among others. A distinct subpopulation of these lipid rafts containing caveolin (64, 79), and referred to as caveolae, has been identified.

Caveolae, flask-shaped invaginations of the plasma membrane occupying up to 30% of cell surface in capillaries, represent an ostensibly predominant location of endothelial nitric oxide synthase (eNOS) in endothelial cells (36). They harbor the bradykinin B₂ receptor; epidermal growth factor (EGF) and platelet-derived growth factor (PDGF) receptors; mitogen-activated protein (MAP) kinase; src family nonreceptor tyrosine kinases; G proteins; protein kinase C (PKC); cationic arginine transporter-1; class B, type I scavenger receptor for high-density lipoprotein; H-Ras; Ca²⁺ ATPases; and inositol 1,4,5 triphosphate-dependent Ca²⁺ channels, among others (60, 71, 73). Caveolin, the caveolar coat protein (36, 79), forms high-molecular-weight, Triton-insoluble complexes through oligomerization mediated by interactions between NH₂-terminal residues 61 and 101. (80) eNOS is targeted to caveolae by cotranslational N-myristoylation and posttranslational palmitoylation (36). Caveolin-1 coimmunoprecipitates with eNOS; interaction with eNOS occurs via the caveolin-1 scaffolding domain (NH₂-terminal residues 81-101) and appears to result in the inhibition of NOS activity (52, 62). The inhibitory conformation of eNOS is reversed by the addition of excess Ca²⁺/calmodulin (26, 27) and by Akt-induced phosphorylation of eNOS (34). These aspects of caveolar-eNOS interaction have been comprehensively discussed in several recent reviews (23, 51). Here, we shall dissect the system using the classic paradigm of a reflex loop: 1) the action of afferent elements, such as fluid shear stress and its putative caveolar sensor, on caveolae; 2) the ways in which afferent signals may affect the central element, the activation of the eNOS-nitric oxide (NO) system; and 3) the several resultant well-established and novel physiologically important effector mechanisms, i.e., vasorelaxation, angiogenesis, endothelial permeability, deterrance of inflammatory cells, and prevention of platelet aggregation, to name a few (Fig. 1).

View larger version (25K): [in this window] [in a new window]   Fig. 1.   Schematic view of the afferent, central, and efferent elements of the regulatory mechanisms for shear stress-induced activation of endothelial nitric oxide synthase (eNOS) and its cellular consequences. BH4, tetrahydrobiopterin; BH2, dihydropterin; ILK, integrin-linked kinase; Akt, protein kinase B; + or , stimulation or inhibition, respectively; NO, nitric oxide.

	FUNCTIONAL MORPHOLOGY AND SUBCELLULAR LOCALIZATION OF ENOS

TOP ABSTRACT INTRODUCTION FUNCTIONAL MORPHOLOGY AND... MOLECULAR CASCADE RECRUITED IN... TOPOLOGICAL FEEDBACK REGULATION... ROLE OF LIPID COMPONENTS... NO AND MEMBRANE FLUIDITY CAVEOLAE AND LOCOMOTION: A... CAVEOLAE, NO, AND VASCULAR... CONCLUDING REMARKS REFERENCES

Although lipid-rich domains, present on the exoplasmic leaflet of the plasma membrane, are highly dynamic structures undergoing constant entropy-driven dispersal, they are nonetheless constrained and marginalized, in part, by the cortical actin cytoskeleton (48). Although caveolin-1 oligomeric lattices tend to stabilize caveolae, stabilization of the rafts is believed to be achieved through the annexin-dependent formation of membrane-cytoskeleton complexes (4). It is not known whether eNOS is sequestered in both of these lipid-rich domains or in caveolae only. Most recently, direct evidence for eNOS localization to rafts has been presented in a series of studies in cells lacking caveolin-1 or caveolae (87). Therefore, in the following discussion, although eNOS-caveolar relationships are described, it is quite possible that a broader sense is applicable, inclusive of eNOS-lipid-rich domains. A recent demonstration of an interchange between caveolae and rafts lends additional support to this proviso: experiments showed that by abolishing glycosylphosphatidylinositol-anchored proteins, a constituent of rafts, the number of rafts does not change, whereas the number of caveolae increases (1). The situation has become even more complicated since a recent description of a caveolin-1-enriched, noncaveolar lipid-rich domain, novel type of raft (93). In addition to caveolae, eNOS has been localized to fenestrae: immunoelectron microscopy of liver sinusoidal cells disclosed both eNOS and caveolin-1 colocalized in endothelial fenestrae (95). This finding may indicate that the biogenesis of fenestrae represents a step in the evolution of fused endocytosed caveolae (see CAVEOLAE, NO, AND VASCULAR PERMEABILITY).

The present consensus is that caveolae are dynamic structures constantly recycled between the plasma membrane, endosomes, and the trans-Golgi network. If this is the case, caveolae-harbored receptors, elements of signaling pathways, and eNOS may undergo parallel internalization and recycling, tracking the caveolin-1 pathway. Indeed, eNOS has been previously localized to the cytoplasmic vesicles and trans-Golgi network, suggesting that the above recycling path does exist. Using surrogate cells transfected with eNOS fused to green fluorescent protein (GFP), Sowa et al. (86) demonstrated that ~50% of the total eNOS is localized to the plasma membrane and 35% to the Golgi apparatus. Interestingly, a palmitoylation-deficient eNOS mutant showed no distribution to the plasma membrane and was concentrated in the Golgi apparatus and the cytoplasm. These findings are in concert with a broader view that protein palmitoylation targets it to lipid-rich domains, rafts, and caveolae. In this regard, palmitoyltransferase activity has been found to be highly enriched in low-density membranes. Depletion of cellular cholesterol results in the inhibition of this activity and translocation of palmitoyltransferase to high-density, noncaveolar membranes (21).

Site-directed mutagenesis of the putative caveolin-binding motif of eNOS (from FSAAPFSGW to ASAAPASGA; single-letter amino acid coding) blocked the ability of caveolin-1 to inhibit eNOS activity (35). However, the pharmacological consequences of caveolin-1 scaffolding domain peptides appear controversial. According to two groups, they, as well as caveolin-1 itself, are capable of inhibiting eNOS activity in vitro (37, 52). A similar peptide (residues 82-101) injected systemically into rats appeared to enhance the activity of eNOS (11). Bucci et al. (11) have demonstrated that injection of the scaffolding domain, residues 82-101, results in selective inhibition of acetylcholine-induced vasorelaxation and NO production. It is possible that these disparate effects may be explained on the basis of different activities of neutral sphingomyelinase, an enzyme catalyzing formation of ceramide and activation of eNOS, because it is inhibited by a caveolin scaffolding domain (92). The issue awaits resolution.

The fact that an excess of Ca-calmodulin could override these inhibitory interactions has established the foundation for a hypothesis that the caveolin-calmodulin cycle regulates eNOS activity (61), as schematically illustrated in Fig. 2. This hypothesis has been strengthened most recently with the independent demonstration by two groups that caveolin-1 knockout mice exhibit enhanced eNOS activity (17, 75).

View larger version (34K): [in this window] [in a new window]   Fig. 2.   Simplified view of eNOS functional cycling. A: in unstimulated cells, eNOS appears to be associated with caveolin, which precludes calmodulin binding to the enzyme. The flow of electrons from NADPH to the heme moiety of eNOS (H) is therefore interrupted, and NO production is halted (top). On cell activation, eNOS dissociates from caveolin-1, calcium-calmodulin complex (CaM) binds to eNOS, and electron transport is restored, as is NO production (bottom). This NO generation results in the reversible dissociation of caveolin-1 scaffold and distancing of elements of signaling cascades, thus terminating signal transduction (right). In the process of cell activation, some caveolae may become endocytosed, others exocytosed, forming microparticles. FMN, flavin adenine mononucleotide. B: when the availability of the substrate, L-arginine (L-Arg), or BH4 becomes limited, eNOS functions in the uncoupled mode. CaM serves the same function of a switch as before, allowing electron transport to proceed to the heme moiety. However, in the uncoupled state electrons are not transferred to L-arginine to form NO; rather, they react with oxygen to form superoxide anions.

	MOLECULAR CASCADE RECRUITED IN SIGNAL TRANSDUCTION FROM THE CAVEOLAR SHEAR STRESS SENSOR TO ENOS STIMULATION

TOP ABSTRACT INTRODUCTION FUNCTIONAL MORPHOLOGY AND... MOLECULAR CASCADE RECRUITED IN... TOPOLOGICAL FEEDBACK REGULATION... ROLE OF LIPID COMPONENTS... NO AND MEMBRANE FLUIDITY CAVEOLAE AND LOCOMOTION: A... CAVEOLAE, NO, AND VASCULAR... CONCLUDING REMARKS REFERENCES

The constant adjustment of vascular tone to the changing blood flow and shear stress it exerts on the vascular endothelium is a requisite of physiological regulation of hemodynamics. This adaptation is governed by the shear stress-induced stimulation of eNOS (76). It has been demonstrated that activation of protein kinase B-Akt serves as an intermediary between a caveolar sensor and activation of caveolin-anchored eNOS (16, 33). However, the identity of the sensor remains unknown, as well as the detailed molecular mechanism activating Akt. Rizzo et al. (76) proposed that flow-induced mechanical strain on caveolae causes conformational changes in eNOS. Other potential coupling mechanisms include eNOS phosphorylation by Akt or allosteric modulation by calmodulin. It has recently been demonstrated that mechanosensitive ₁-integrins are compartmentalized in endothelial caveolae. We have confirmed this finding in human umbilical vein endothelial cells (HUVEC) transfected with GFP-caveolin and stained with antibodies to 1-integrins. Several potential kinases may be activated by shear stress and stimulate eNOS; these include integrin-linked kinase (ILK), which is a known activator of protein kinase B/Akt, integrin cytoplasmic domain-associated protein-1 (ICAP-1), as well as integrin-linked Ca²⁺ located in caveolae. Indeed, a recently described stretch-activated cation channel present in endothelial cells is activated by fluid shear stress in a tyrosine kinase- and protein kinase C-dependent manner (7). A cation channel with similar properties is also activated by sphingosine-1-phosphate in human endothelial cells (66). All of these may serve as plausible mechanisms interfacing the sensor with eNOS. Several integrins have been shown to be associated with Ca²⁺ channels (81). Both ₁- and ₃-integrins have been implicated in sensing shear stress in endothelial cells (50, 65). ILK is a 59-kDa serine-threonine kinase associated with ₁-integrins (44). ICAP-1 is associated with ₁-integrins, and their ligation enhances ICAP-1 phosphorylation, whereas removal of the ligand results in its dephosphorylation (12). Moreover, we would argue that the previously proposed mediators of caveolar shear stress- and flow-induced signaling to eNOS could be triggered by the ligation of integrins (Akt activation, activation of Ca²⁺ channels, and calmodulin), thus suggesting the role of these heterodimeric proteins in sensing shear stress. This possibility is further strengthened by the fact that Arg-Gly-Asp peptides, ligands for several integrins, have been found to induce vasorelaxation (63).

	TOPOLOGICAL FEEDBACK REGULATION OF SIGNALING BY NO

TOP ABSTRACT INTRODUCTION FUNCTIONAL MORPHOLOGY AND... MOLECULAR CASCADE RECRUITED IN... TOPOLOGICAL FEEDBACK REGULATION... ROLE OF LIPID COMPONENTS... NO AND MEMBRANE FLUIDITY CAVEOLAE AND LOCOMOTION: A... CAVEOLAE, NO, AND VASCULAR... CONCLUDING REMARKS REFERENCES

The idea of the topological proximity of elements of signaling cascades playing an important role in the regulation of eNOS activation-inactivation, as well as in regulating the flow of information along signaling elements, is further supported by our recent finding that NO, produced endogenously or added to cultured endothelial cells, can dissociate oligomeric caveolin-1 (57). Using caveolin-1 fused to a temperature-tolerant GFP mutant, the fluorescence of which is modulated by the distance between adjacent GFP molecules, we demonstrated that NO leads to a reversible dissociation of caveolin-1 platforms, the appearance of caveolin-1 monomers in higher abundance, and the temporary silencing of signal transduction initiated via receptors harbored in caveolae. These data implicate NO generated in the process of endothelial cell activation in the termination of some signaling cascades, perhaps as a result of the distancing of the sequentially coupled elements (Fig. 2). This endothelial paradigm of topological forward and feedback regulation of signaling cascades has its neuronal equivalent in the activation of the N-methyl-D-aspartate (NMDA) receptor, leading to the generation of NO and suppression of the NMDA receptor (53).

	ROLE OF LIPID COMPONENTS OF CAVEOLAE IN ENOS SIGNALING

TOP ABSTRACT INTRODUCTION FUNCTIONAL MORPHOLOGY AND... MOLECULAR CASCADE RECRUITED IN... TOPOLOGICAL FEEDBACK REGULATION... ROLE OF LIPID COMPONENTS... NO AND MEMBRANE FLUIDITY CAVEOLAE AND LOCOMOTION: A... CAVEOLAE, NO, AND VASCULAR... CONCLUDING REMARKS REFERENCES

Compartmentalization of eNOS to lipid-rich domains like caveolae engenders remarkable substrate parsimony and biochemical plasticity. This is illustrated by the finding that components of these domains, sphingolipids, participate in the Ca²⁺-independent activation of eNOS through the generation of ceramides. Igarashi et al. (47) have demonstrated that the ceramide analog N-acetylsphingosine increases eNOS activity and NO production in bovine aortic endothelial cells, the effect of which is sustainable even in the presence of an intracellular Ca²⁺ chelator. Furthermore, the authors showed that bradykinin, acting via its B₂ receptor, induces ceramide generation by endothelial cells. These findings establish the participation of a lipid mediator, derived from the topologically proximal lipid-rich membrane, in the regulation of eNOS. In turn, NO produced may inhibit the rate of ceramide generation; it has been demonstrated that tumor necrosis factor--induced generation of ceramide in a monocytic cell line is suppressed by NO, thus preventing cell apoptosis (15). Another example of a lipid-rich membrane component, cholesterol, participating in signal transduction has been provided by Liu et al. (58). By introducing oxidized cholesterol into caveolae, these investigators were able to disrupt PDGF receptor-mediated tyrosine kinase phosphorylation of downstream substrates but not activation of the receptor itself.

In fact, a remarkable cooperativity exists between cholesterol synthesis and expression of caveolin on the one hand and activity of eNOS on the other. Cellular cholesterol homeostasis is dependant on the balance among cholesterol biosynthesis, efflux, and salvage through the low-density lipoprotein receptor pathway (49). Newly formed free cholesterol (FC) is destined mainly to caveolae and, subsequently, to pre-high-density lipoprotein (HDL). It has been demonstrated that caveolin-1 trafficks from the endoplasmic reticulum to caveolae as a cytosolic complex containing chaperone proteins and cholesterol. Thus caveolin-1 effectively delivers newly synthesized cholesterol from endoplasmic reticulum to caveolae, which may also serve as sites of cholesterol efflux (29, 30, 85, 90).

The physiological significance of caveolae in cholesterol homeostasis is further supported by the observation that an HDL receptor, type I, class B scavenger receptor, as well as a multifunctional receptor, CD36 (74), are harbored in caveolae (54). Furthermore, the bidirectional flow of cholesterol between HDL particles and the cell occurs in caveolae (3, 41). The high concentration of cholesterol in caveolae and the tight association of cholesterol with caveolins (67, 88) underscore their mutual coregulation. Cells transfected with caveolin-1 cDNA have increased caveolin mRNA and protein levels, caveolar FC, and FC efflux. Cells transfected with caveolin-1 antisense DNA have 50% of the caveolin levels of control cells, and the FC efflux from these cells is proportionately reduced (28, 96).

On the other hand, caveolae are considered to be sensors of the FC content of the cell (30, 89). Depletion of caveolar cholesterol leads to the downregulation of caveolin-1 at the protein and mRNA levels, resulting in a decreased expression of caveolae on the cell surface (28, 43). Depletion of cellular cholesterol with either nystatin or cholesterol oxidase causes retrieval of caveolin-1 from the plasma membrane lipid-rich domains (14, 84). Restoring cellular cholesterol causes caveolin-1 to return to plasma membrane caveolae.

	NO AND MEMBRANE FLUIDITY

TOP ABSTRACT INTRODUCTION FUNCTIONAL MORPHOLOGY AND... MOLECULAR CASCADE RECRUITED IN... TOPOLOGICAL FEEDBACK REGULATION... ROLE OF LIPID COMPONENTS... NO AND MEMBRANE FLUIDITY CAVEOLAE AND LOCOMOTION: A... CAVEOLAE, NO, AND VASCULAR... CONCLUDING REMARKS REFERENCES

It has been established that NO is at least six times more soluble in lipid bilayers than in the aqueous phase (31, 59). This partitioning of NO in lipid-rich biomembranes appears to have regulatory functions. We have recently found that an elevation of ambient NO increases membrane fluidity, consistent with temporal dispersal of lipid microdomains. The high solubility of NO in lipids and proximity of the NO source to lipid-rich domains provide an ideal microenvironment for the observed changes in the biophysical properties of lipid rafts. The implications of this finding may be far-reaching.

Deformability of the plasma membrane of the formed elements and the endothelium has been recognized as a key contributor to blood rheology and microvascular perfusion (6, 19, 82). Fluid shear stress per se has been found to increase membrane fluidity in endothelial cells (42). These data are consistent with the observed NO-induced reversible increase in membrane fluidity of endothelial cells.

Constitutive NO production is an important regulator of vascular tone: inhibition of eNOS by nitroarginine inhibitors or eNOS gene deletion invariably results in the development of hypertension (46, 72). This has been attributed to the reduction of the endothelium-derived relaxing factor-induced vasorelaxation. Our recent observations (Li H and Goligorsky MS, unpublished observations) may provide a mechanistic explanation for these findings (Fig. 3). States of deficient NO production by the endothelial cells may result in increased membrane rigidity, thus leading to deceleration and trapping of formed elements in the capillary circulation (82), increased peripheral resistance to blood flow, and the development of hypertension. Hence, we propose that apart from the well-established NO regulation of vascular smooth muscle tone at the level of conduit and resistance vessels, it may exert its effects at the luminal boundary between the endothelium and circulating erythrocytes at the level of capillaries.

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Hypothetical consequences of NO-induced increase in plasma membrane fluidity. Effects of NO on the membrane fluidity predictably would increase the compliance of microvasculature. When NO generation in response to shear or mechanical stress is defective, the compliance of microcirculation decreases and the passage of red blood cells would require increasing propulsive force. This mechanism may be responsible for development of hypertension in states with defective NO production.

	CAVEOLAE AND LOCOMOTION: A ROLE PLAYED BY NO IN ANGIOGENESIS

TOP ABSTRACT INTRODUCTION FUNCTIONAL MORPHOLOGY AND... MOLECULAR CASCADE RECRUITED IN... TOPOLOGICAL FEEDBACK REGULATION... ROLE OF LIPID COMPONENTS... NO AND MEMBRANE FLUIDITY CAVEOLAE AND LOCOMOTION: A... CAVEOLAE, NO, AND VASCULAR... CONCLUDING REMARKS REFERENCES

Growing in vitro and in vivo evidence suggest that functional eNOS in endothelial cells plays a permissive role in the acquisition of migratory phenotype in response to stimuli like vascular endothelial growth factor (VEGF) or endothelin (ET) (69). Studies from our laboratory and by others provided convincing evidence that NO production is a prerequisite for endothelial cell migration, wound healing, and angiogenesis (39, 71). On the basis of these observations, we propose that endothelial cell migration and angiogenesis require simultaneous action of at least two signals, functional eNOS generating NO and guidance cues, e.g., ET-1 or VEGF, to secure the necessary direction of movement.

Studies of the mode(s) of NO action on endothelial cell migration have revealed that NO stimulates micromotion in cultured endothelial cells, the process that we termed "podokinesis," indicative of accelerated turnover of focal adhesions (71). Second, when endothelial cells were cultured on a flexible silicon rubber substratum and a wrinkling phenomenon was recorded using time-lapse videomicroscopy, it was possible to document that addition of NO donors resulted in the resolution of tractional forces on the substratum whereas NOS inhibition enhanced them (39). When one considers the fact that both eNOS and ₁-integrins are localized to caveolae in endothelial cells, one can envisage a possible link between caveolae and eNOS and cell adhesion, locomotion, and angiogenesis. Several lines of evidence provide circumstantial support for this link. First, double-labeling studies of endothelial cells revealed colocalization of caveolin-1 and ₁-integrins (94). Second, migrating endothelial cells dislocate caveolin-1 and ₅₁-integrins to the leading and trailing edges (Li H and Goligorsky M, unpublished observations). Third, both cholesterol and sphingolipids directly interact with ₅₁-integrins, and cholesterol depletion renders _V3 integrins dysfunctional (2, 40).

Hence, in this model, NO production provides the cell with a fork-tuning mechanism to adjust the tightness of cell-matrix adhesions to the physiologically required rate of cell migration. For instance, in the resting state, the endothelial cell phenotype is that of a spread, tightly attached cell that, on activation and NO generation, develops an intermediate strength of adhesion and is prepared for vectorial locomotion. When such a cell is presented with guidance cues (ET or VEGF), their chemical gradient results in a gradient of tyrosine phosphorylation/dephosphorylation of focal adhesion kinase and paxillin, thus determining the leading and trailing edges. Indeed, when cells are migrating toward a VEGF source, application of a tyrosine kinase inhibitor, genistein, significantly inhibits vectorial movement. These data further emphasize the relationship between scalar and vectorial motion in endothelial cells. When the level of NO production becomes supraphysiological, cells tend to detach from the matrix, whereas subphysiological levels of NO generation lead to the excessively tight cell-matrix adhesion; in both instances, endothelial cells are unable to respond appropriately to angiogenic stimuli. In fact, recent findings of defective angiogenesis in eNOS knockout mice are consistent with the above contention (68). It is important to emphasize that the requirement for podokinetic motion should be fulfilled for vectorial cell movement along the gradient of angiogenic substances to occur. This working hypothesis may explain some pathological situations accompanied by the inappropriate generation of NO.

Recently, in parallel with the acceptance of an idea that angiogenesis is an integral function of vascular endothelium, effects of various pharmacological maneuvers on endothelial cell motility have been examined. Along the lines discussed above, hypercholesterolemia has been implicated in the reduced angiogenesis, whereas treatment with statins was found to accelerate it (8). Similarly, constitutively active Akt stimulates angiogenesis, and the opposite effect can be achieved with the dominant-negative form of Akt (55). All of these maneuvers are ultimately mediated via modulation of eNOS activity and NO generation by migrating endothelial cells.

	CAVEOLAE, NO, AND VASCULAR PERMEABILITY

TOP ABSTRACT INTRODUCTION FUNCTIONAL MORPHOLOGY AND... MOLECULAR CASCADE RECRUITED IN... TOPOLOGICAL FEEDBACK REGULATION... ROLE OF LIPID COMPONENTS... NO AND MEMBRANE FLUIDITY CAVEOLAE AND LOCOMOTION: A... CAVEOLAE, NO, AND VASCULAR... CONCLUDING REMARKS REFERENCES

The morphological route(s) for increased endothelial permeability has been an issue of lasting debate (22, 24, 77, 78). At least three different mechanisms regulating microvascular permeability have been described. First, nonselective permeability occurs due to gap formation between endothelial cells (5), which may be responsible for the leakage of intravascular macromolecular and cellular components into the interstitium under inflammatory conditions. Second, fenestrae form in endothelial cells subjected to the paracrine effect of elevated local VEGF production (78). This phenomenon may be operative in tumor neovasculature but could occur in other conditions, resulting in the stimulation of neovascularization (therapeutic neovascularization is one of the areas requiring further study). Because of the size of fenestrae, increased permeability of the vasculature to macromolecules ensues. Third, VEGF results in the formation of vesiculovacuolar organelles, which represent invaginated caveolae fused in a vectorial fashion to create transcellular channels (25).

Roberts and Palade (77, 78) consider caveolae as structures, which are plausibly involved in the increase of endothelial permeability. Indeed, several investigators have demonstrated in fixed cells that caveolae, if studied by serial sectioning, extend far beyond the plasmalemmal vesicles to form extensive invaginations. In an attempt to resolve some of the existing problems in reconstructing the three-dimensional organization of caveolae, we have generated a GFP-caveolin-1 vector to enable intravital microscopy of endothelial cells subjected to VEGF (13). Using the caveolin-1-GFP vector, we were able to intravitally monitor the dynamics of caveolin in endothelial cells stimulated with VEGF. Confocal microscopy disclosed that the probe decorates transcellular channel-like structures, which become conspicuous after exposure to VEGF. These data demonstrate in vivo that caveolin is organized into elongated cell-spanning structures in cells exposed to VEGF. Electron microscopic studies confirmed and further extended these observations by demonstrating the enrichment in caveolae, their fission, and their fusion after an application of VEGF. An alternative route for increased permeability via fenestrae could not be detected in HUVEC or renal microvascular endothelial cells at early time points after the application of VEGF. However, 36 h after the addition of VEGF, HUVEC and renal microvascular endothelial cells exhibited diaphragmed fenestrae. Vasile et al. (91) have recently provided additional evidence of VEGF-induced clustering of caveolae, resulting in formation of vesiculovacuolar organelles in bovine microvascular endothelial cells cultured on floating Matrigel-collagen gels. It is conceivable that VEGF elicits a rapid increase in vascular permeability via mobilization of caveolae, whereas the long-term effect requires formation of fenestrae. Recently, eNOS and caveolin have been found in sinusoidal fenestrae (95), further supporting the idea that fusion of caveolae eventually leads to the biogenesis of fenestrae (Fig. 4).

View larger version (55K): [in this window] [in a new window]   Fig. 4.   Potential transcellular routes for endothelial permeability. A and B: vascular endothelial growth factor induces recruitment of caveolae into vesiculovacuolar-like structures. A: human umbilical vein endothelial cells (HUVEC) transfected with caveolin-1-green fluorescent protein fusion construct were analyzed by confocal microscopy. This image demonstrates the appearance of elongated transcellular structures decorated with caveolin-1. Inset: higher magnification of elongated transcellular structure. B: transmission electron microscopic image of an endothelial cell after stimulation with vascular endothelial growth factor. Note the recruitment of caveolae and formation of multiple endocytic vesicles (arrows). C: late effect of vascular endothelial growth factor consists in the formation of diaphragmed fenestrae (arrowheads), as illustrated on this electron micrograph. D: schematic summary of vascular endothelial growth factor-induced fission and fusion of caveolae, resulting initially in formation of vesiculovacuolar structures and, at later time points, possibly leading to formation of fenestrae.

NO appears to regulate these processes. In vivo, N^G-nitro-L-arginine methyl ester application resulted in a rapid increase in albumin extravasation, adherence and immigration of leukocytes, and increased generation of reduced oxygen intermediates in postcapillary venules (56). In addition, N^G-nitro-L-arginine methyl ester potentiated a thrombin-induced increase in macromolecular permeability of endothelial cell monolayers, an effect that was counteracted by exogenous 8-bromo-cGMP or stimulation of endogenous cGMP production (sodium nitroprusside or atrial natriuretic peptide), further suggesting that activation of eNOS and generation of NO may reduce endothelial permeability stimulated by agents like thrombin (18). However, it is important to emphasize that the concurrent increase in the production of reduced oxygen intermediates in proinflammatory states, for instance, could reverse the above effect of NO, resulting in peroxynitrite-induced elevation of endothelial permeability. Our scanning electron microscopic data on the integrity of endothelial monolayers subjected to sodium nitroprusside or peroxynitrite demonstrate a remarkable effect of the latter on gap formation (38). In conclusion, basal NO generation by endothelial cells is necessary for the maintenance of the barrier functions of these cells. Inhibition of eNOS results in increased vascular permeability. Overproduction of NO, especially when inducible NO synthase (iNOS) is expressed, similarly leads to increased vascular permeability. It is not known, however, whether the cellular routes for the increase in endothelial permeability to macromolecules under conditions of deficient or excessive NO generation are uniform or dissimilar. Recent studies of vascular permeability in eNOS- and iNOS-deficient mice provide some insight into this question. Fukumura et al. (32) have demonstrated that the VEGF-induced increase in vascular permeability was blunted in eNOS-deficient mice but not in iNOS knockout mice. These mice, together with caveolin knockout animals, should provide the field of vascular permeability with invaluable models.

	CONCLUDING REMARKS

TOP ABSTRACT INTRODUCTION FUNCTIONAL MORPHOLOGY AND... MOLECULAR CASCADE RECRUITED IN... TOPOLOGICAL FEEDBACK REGULATION... ROLE OF LIPID COMPONENTS... NO AND MEMBRANE FLUIDITY CAVEOLAE AND LOCOMOTION: A... CAVEOLAE, NO, AND VASCULAR... CONCLUDING REMARKS REFERENCES

This brief overview of eNOS interactions with lipid-rich domains, including caveolae, highlights the plethora of functional implications. Incoming signals, signals emanating from these organelles, as well as their modulation and termination, utilize these platforms for integrating information and sorting the elements of signaling cascades. The richness of such interactions is underscored by the versatility of functional outputs: from the regulation of NO synthesis to the modulation of cell adhesion, migration, and permeability. Among the unresolved questions are problems related to the possible alternative sorting of eNOS to caveolae or rafts and the functional competence of eNOS in the Golgi apparatus, in caveolae that underwent fission and fusion, in fenestrae, and in mitochondria. When one takes into consideration the possibility that at each of these sites eNOS can function as a coupled or an uncoupled enzyme, thus resulting in the generation of NO or superoxide, respectively, the number of such questions doubles. From the therapeutic standpoint, it remains to be established whether different peptides or peptidomimetics, i.e., peptides derived from the caveolin-1 scaffolding domain, have beneficial effect(s) on hemodynamics. Thus again a protean molecule, NO, and the enzyme producing it are occupying the center stage of endothelial cell biology and pose challenges.