HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (78) Citing Articles via Google Scholar Google Scholar Articles by Mehta, D. Articles by Malik, A. B. Search for Related Content PubMed PubMed Citation Articles by Mehta, D. Articles by Malik, A. B.

Signaling Mechanisms Regulating Endothelial Permeability

Department of Pharmacology and Center of Lung and Vascular Biology, The University of Illinois College of Medicine, Chicago, Illinois

ABSTRACT

I. INTRODUCTION

II. STARLING AND KEDEM-KATCHALSKY EQUATIONS DESCRIBING ENDOTHELIAL PERMEABILITY

III. PERMEABILITY OF THE ENDOTHELIAL BARRIER

IV. HETEROGENEITY OF ENDOTHELIAL PERMEABILITY

V. STRUCTURAL DETERMINANTS OF ENDOTHELIAL BARRIER FUNCTION

A. Glycocalyx

B. Extracellular Matrix

C. Vesiculo-Vacuolar Organelles

D. Development of the Endothelial Barrier

VI. PHYSIOLOGICAL SIGNIFICANCE OF ALBUMIN PERMEABILITY

A. Albumin Regulation of Tissue Oncotic Pressure and Endothelial Barrier Integrity

B. Albumin as a Chaperone

C. Other Functions of Albumin

VII. ALBUMIN TRANSPORT PATHWAYS

A. Pore Theory Versus Transcellular Pathway

B. Paracellular Pathway

VIII. REGULATION OF ENDOTHELIAL PERMEABILITY VIA JUNCTIONS AND MATRIX INTERACTIONS

A. Functions of IEJ Proteins

1. AJs

2. TJs

3. GJs

B. Endothelial Cell-ECM Interactions

1. Integrins

2. Integrin-associated proteins

IX. SIGNALING MECHANISMS REGULATING PARACELLULAR PERMEABILITY

A. Paracellular Mechanisms of Increased Endothelial Permeability

1. Role of actin-myosin motor

2. Role of FAC

3. Role of cell-cell junctional complexes

4. Role of microtubules

5. Role of intermediate filaments

B. Mechanisms of Recovery of Endothelial Barrier Function

1. Role of FAK

2. Role of Cdc42 and Rac

3. Roles of PKA and adenylate cyclase

X. TRANSCELLULAR ENDOTHELIAL PERMEABILITY

A. Function of Caveolae

1. Formation of caveolae

B. Transcytosis: Caveolae Fission, Targeting, and Fusion

1. Role of dynamin

2. Role of intersectin

3. Role of SNARES

4. Role of Rab GTPases

5. Role of actin and myosin

6. Role of microtubules and molecular motors

C. Role of Endothelial Cell Surface Albumin-Binding Proteins in Transcytosis

D. Membrane Dynamics

XI. SIGNALING MECHANISMS REGULATING TRANSCYTOSIS

A. Role of Src Kinase

B. Role of PKC Isoforms

C. Role of PI 3-Kinase

D. Role of Ca2+ Signaling

E. Role of 60-kDa Glycoprotein (gp60)

XII. MEDIATORS OF INCREASED ENDOTHELIAL PERMEABILITY

A. Mediators of Increased Endothelial Permeability

1. Thrombin

2. Bradykinin

3. Histamine

4. Oxidants

5. VEGF

6. TNF-{alpha}

7. LPS

B. Endothelial Barrier Stabilizing Mediators

1. S1P

2. Ang-1

XIII. CONCLUSIONS AND FUTURE DIRECTIONS

GRANTS

ACKNOWLEDGMENTS

REFERENCES

	ABSTRACT

Top Next References

	I. INTRODUCTION

Top Previous Next References

 
The vascular endothelium lining the intima of the blood vessels regulates a variety of functions including vascular smooth muscle tone, host-defense reactions, angiogenesis, and tissue fluid hemostasis. The maintenance by the endothelium of a semi-permeable barrier is particularly important in controlling the passage of macromolecules and fluid between the blood and interstitial space. It is known that loss of this function results in tissue inflammation, the hallmark of inflammatory diseases such as the acute respiratory distress syndrome. The characteristic permeability of transported macromolecules is dependent on their molecular radii as well as the barrier properties of the particular endothelium. This size-selective nature of the barrier to plasma proteins is a key factor in establishing protein gradients (especially in the case of albumin) required for fluid balance of tissues. In addition, plasma proteins, such as albumin, act as circulating chaperones for hydrophobic substances, fatty acids, and hormones, molecules whose transport is crucial for cell functions vital to the organism. Thus the efficient transfer of many water insoluble substances from the blood into the interstitium relies on endothelial permeability, and often on specific carrier proteins. The requirements for continuous transendothelial protein flux and, at the same time, a steep albumin gradient imply that dynamic processes exist in the endothelium controlling protein flux between the vascular and extravascular spaces.

Estimates of the transvascular flux of solutes and fluid indicate that protein transport occurs by a mechanism distinct from that of small hydrophilic molecules. In this regard, the traditional view of the endothelium as a "static barrier" through which proteins leak via interendothelial junctions (IEJ) is an oversimplification. Estimates of the dimensions of unperturbed IEJs are insufficient to allow the unrestricted passage of proteins known to cross the barrier. Moreover, the notion of passive protein leakage through the endothelium has been questioned in light of recent ultrastructural, biochemical, and physiological studies of protein tracers in transit through the intact endothelium. These newer studies have emphasized the role of a vesicular pathway in the mechanism of transfer of proteins with their cargo of small hydrophobic solutes. In this review, we evaluate the evidence that the endothelium controls flux of fluid and solutes across the vessel wall through highly regulated transport pathways. An emerging general principle is that transport of protein and liquid in the undisturbed endothelium occurs via the transcellular pathway. However, in response to intrinsic and extrinsic stimuli, the endothelium also sets into motion additional signaling pathways that allow transport of solutes through IEJs. This review describes the current understanding of the signaling mechanisms activated in endothelial cells that regulate barrier function via both pathways and raises questions in areas where understanding and important details are in doubt.

Endothelial transport can be thought of in a general sense as occurring via paracellular and transcellular pathways (Fig. 1). The continuous endothelium (as found in pulmonary, coronary, skeletal muscle, and splanchnic vascular beds) is described as being restrictive because solutes with molecular radii of up to 3 nm move passively across the barrier via the paracellular route. The transcellular vesicular pathway is responsible for the active transport of macromolecules as shown for albumin (316, 595, 603, 719, 810, 811, 864). Paracellular permeability is regulated by a complex interplay of cellular adhesive forces balanced against counteradhesive forces generated by actinomyosin molecular motors. The unperturbed endothelial barrier has restrictive properties that are due primarily to closed IEJs. Evidence now suggests that integrin receptor binding to the extracellular matrix (ECM) can also contribute to the barrier function by stabilizing the closed configuration of IEJs. The inflammatory mediators thrombin, bradykinin, histamine, vascular endothelial growth factor (VEGF), and others upon binding to their receptors, disrupt the organization of IEJs and integrin-ECM complexes, thereby opening the junctional barrier (for review, see Refs. 225, 553). Thus the formation of minute intercellular gaps allows passage of plasma proteins including albumin and liquid across the endothelial barrier in an unrestricted manner. The signaling pathways regulating opening and closing of junctions are of great interest as they relate to the regulation of tissue fluid balance and the mechanisms of inflammation, and are discussed extensively in this review.

View larger version (22K): [in this window] [in a new window]   FIG. 1. Schematic of transport pathways in continuous endothelium. Under basal conditions, the transcellular pathway can mediate the transport of plasma proteins (>3 nm Mr) such as albumin by caveolae via an absorptive (receptor-mediated) or fluid-phase pathway. Transcellular channels can also form transiently in endothelial cells by fusion of multiple caveolae and allow albumin transport. Aquaporins form channels across the lipid bilayer that are highly selective for water molecules and allow their movement across the luminal or abluminal endothelial membrane, thus creating a transendothelial pathway for water. Small molecules including urea and glucose (<3 nm Mr) are transported around individual endothelial cells via the paracellular, i.e., interendothelial junction (IEJ) pathway. Mr, molecular radius.

There are several recent comprehensive reviews dealing with the general subject of endothelial permeability discussed from different perspectives (592, 966). Michel and Curry (592) in particular have elegantly addressed the role of the endothelial glycocalyx (termed the "fiber matrix") in regulating endothelial permeability. Tuma and Hubbard (966) in a thorough review of the subject have drawn attention to the still incompletely understood processes of caveolae- and clathrin-mediated endocytosis and transcytosis in various organs. Although both reviews are seminal with respect to their own emphases, neither focuses on the signaling mechanisms involved in the regulation of endothelial permeability, the thrust of our review. Our objectives when we set out on this venture were twofold: 1) to provide a fresh perspective to the field and 2) to highlight those areas where signaling pathways are beginning to be better understood as well as those areas where a great deal of work is needed. Thus we have evaluated the evidence concerning the signals that control transendothelial liquid and protein transport via the paracellular-junctional and transcellular-vesicular pathways. Because a great deal of work has recently been carried out using genetically modified mouse models, we have also discussed the data that bear on and inform the signaling mechanisms regulating vascular endothelial permeability in an in vivo setting. The reader should be made aware that we have not been timid about speculating, and wherever possible, critically analyzing the findings because we wish to draw attention to uncertainties in the field and, hopefully, to stimulate debate. In writing this review, we have attempted to be as inclusive as possible in citing the most significant work that addresses endothelial permeability regulation; however as is invariably the case given space limitations and the need to emphasize specific areas, we have not been able to refer to all of the published findings. For this, please accept our apologies in advance.

	II. STARLING AND KEDEM-KATCHALSKY EQUATIONS DESCRIBING ENDOTHELIAL PERMEABILITY

Top Previous Next References

 
A recent review covered extensively the theory of fluid and solute exchange across vessels (592). We do not attempt to recapitulate this review, but instead emphasize the critical importance of the two equations defined below as their understanding is necessary to the appreciation of the signaling mechanisms regulating endothelial permeability.

The magnitude of fluid movement is dependent on the net balance of Starling forces (896) across the endothelial barrier according to the relationship

(1)

where J_v is volume flux of fluid (ml/min); L_p is hydraulic conductivity (cm · min^–1 · mmHg^–1); S is capillary surface area (cm²); P_c and P_i are capillary and interstitial fluid hydrostatic pressures, respectively (mmHg); _c and _i are capillary and interstitial colloid osmotic (oncotic) pressures, respectively (mmHg); and is osmotic reflection coefficient of vessel wall ( = 0 if membrane is fully permeable to transported molecular species and = 1 if membrane is impermeable). Kedem-Katchalsky (450) later derived equations that represent the overall flux of solute molecules, which is the sum of convective and diffusive components

(2)

where J_s is transvascular solute flux (mg/min), P is permeability (cm/s), C is the difference in solute concentration across the endothelial cell monolayer, and C_s is mean concentration of the solute within the hypothetical "pore" (which in endothelial cells is likely to be the cleft formed at the IEJ); J_v, S, and are the same terms used in Equation 1.

Besides quantitatively describing vessel wall permeability to liquid and solutes and the relative contributions of diffusion and convection to overall solute transport, these equations have been useful in laying the foundations for the techniques used to experimentally measure endothelial permeability. Table 1 lists the key methods that rely on these principles and describes the measurement of the specific constants derived from these equations.

	III. PERMEABILITY OF THE ENDOTHELIAL BARRIER

Top Previous Next References

The monolayer of endothelial cells forming the innermost layer of the exchange vessels presents a cellular barrier to permeation of liquid and solutes. The vascular endothelium transports solutes with a range of molecular radii (M_r) from 0.1 nm (sodium ion) to 11.5 nm (immunoglobulin IgM) (reviewed in Ref. 757). On the basis of the seminal finding that the transport of lipid-insoluble solutes (M_r < 3 nm) across continuous endothelia decreased with increasing M_r of the permeating solute, it was concluded that the endothelial barrier behaves like a molecular sieve with an average pore radius of 3 nm (674). However, the relationship between M_r and transcapillary exchange of solutes is more complex. The permeability of the endothelial layer decreased by four orders of magnitude with an increase in M_r from 0.1 to 3.6 nm (M_r of albumin). Interestingly, with an increase in M_r from 3.6 to 6 nm, endothelial permeability was found to be independent of solute radius (749, 855). This suggested that the pathway for transport of macromolecules differs from that of small solutes (Fig. 2). The estimated area of the barrier made up of IEJ discontinuities of <3 nm could account for the permeability of small molecules, such as water, hexoses (glucose, mannitol, and fructose), amino acids, and urea (592). These findings are the basis for the generally accepted view that there is a paracellular permeation path that allows liquid and small solutes to move across the continuous endothelium. The transcytosis pathway (see sect. X) accommodates larger molecules such as plasma proteins regardless of their M_r and thus serves to extend the range of transported species. A significant amount of water (up to 40% of the total hydraulic pathway) (592) crosses the barrier through a transcellular route in vascular endothelial cells and epithelial cells (such as alveolar epithelial cells) by water-transporting membrane channels, the aquaporins (725, 1001). Thus, unlike macromolecules, water traverses the endothelial barrier both by paracellular and transcellular routes.

View larger version (17K): [in this window] [in a new window]   FIG. 2. Permeability of endothelial monolayer to molecules of different Stokes-Einstein radii. Permeability (P) of the endothelial cell layer decreases by a factor of 4 with a 36-fold increase in Mr from 0.1 to 3.6 nm. However, endothelial permeability plateaus when Mr increases from 3.6 to 12 nm, suggesting that the pathway for transport of these macromolecules differs from that of small solutes. [Modified from Siflinger-Birnboim et al. (855).]

	IV. HETEROGENEITY OF ENDOTHELIAL PERMEABILITY

Top Previous Next References

 
Even though endothelial cells from different vascular sites have many features in common and originate from the same embryonic precursor cells, the hemangioblasts, variations in permeability have been reported in experiments focusing on different regions of the endothelium (11, 29, 124, 173, 193, 194, 325, 339, 686, 861, 911, 935). These positional differences in endothelial permeability have been found in cultured cells obtained from these sites (98, 214, 561, 817, 854) as well as in intact vessels (15, 458, 492, 560, 607, 680, 864). Baseline vessel wall filtration coefficient (K_f,c) measurements of isolated lungs of various species indicated that total K_f,c is on average 19% arterial, 37% venous, and 42% microvascular (15, 607, 680) (Fig. 3A), with the exception of experiments from Khimenko and Taylor (458) where 96% of total K_f,c was reported to be the result of the venular segment. This discrepancy could be due to the stop-flow ischemic condition in their studies, as this may have directly increased venular permeability. Aside from this exception, these studies in general indicate that under baseline conditions the arterial segment of the lung is more restrictive to liquid flux than either the venular or capillary segment.

View larger version (16K): [in this window] [in a new window]   FIG. 3. Segmental heterogeneity of endothelial permeability. A: in situ differences in filtration coefficient (Kf,c) of three segments as percentage of total Kf,c in rabbit lungs. Kf,c values are calculated by multiplying hydraulic conductivity (Lp) by vascular surface area. [Modified from Parker and Yoshikawa (680).] B: permeability-diffusion coefficient ratios (PEC/D) of albumin across cultured bovine arterial or microvascular endothelial monolayers of equal surface area (37°C). [Modified from Siflinger-Birnboim et al. (855).]

Although the basis for segmental variations in liquid and albumin permeability is not completely clear, evidence suggests the involvement of both intrinsic and extrinsic factors (see Ref. 11). Chi et al. (163) using microarray analysis showed a marked variation in genes expressed in endothelial cells isolated from large and small vessels. ECM proteins collagen 41, collagen 42, and laminin were associated with microvessel endothelia (163, 917), whereas a greater contribution of fibronectin, collagen 51, and collagen 52 was seen with the large vessel endothelia (163). It is possible, therefore, that specific interactions of endothelial cell integrins with ECM are determinants of segmental permeability (208, 211, 730, 884, 934). Of particular note are the differences in genes regulating expression of ECM proteins, integrins, lin-1, isl-1, and mec-3 (LIM) kinase (LIMK), a guanine exchange factor (GEF) for Rho GTPase (Vav), and myosin light-chain kinase (MLCK) in large versus small vessel endothelia (163). Expression of LIMK, MLCK, Vav, and myosin was found to be higher in endothelial cells from microvessels (163), raising the possibility that extensive cytoskeletal remodeling and qualitative and quantitative differences in cell-ECM attachments in microvessel endothelia regulate the restrictiveness of their junctions. Thus epigenetically determined modifications in endothelial cells from different sites could contribute to the observed segmental variations in barrier properties.

Ultrastructural comparisons of confluent endothelial monolayers from microvessels, veins, and arteries of the lung vasculature have helped to explain the observed differences in endothelial permeability. Microvessel endothelial cells from lungs exhibited better developed IEJs than those in large vessels (817, 862, 863, 869). With the use of the microperoxidase tracer (M_r 2.0 nm), 25–30% of IEJs in postcapillary venules have intercellular gaps of 3.0 nm, whereas IEJs of arterioles and capillaries were impermeable over their entire length to molecules with diameters >2.0 nm (863, 866). There is also evidence that caveolar density is highest in capillaries (584, 816, 863, 868), which may be yet another factor contributing to the observed more restrictive nature of IEJs toward albumin (discussed in sect. XA).

Intracellular signals regulating the endothelial barrier may also vary at specific locales in the circulation. Endothelial cells from microvessels compared with cells from large vessels showed distinct profiles for thrombin-induced intracellular Ca²⁺ transients (170); higher basal cyclic nucleotide levels; responsiveness to cAMP-increasing agents isoproterenol, forskolin, and rolipram (192, 900, 1109); and oxidant production (340). While these in vitro differences are real, a potential concern is that these cultured endothelial monolayers may have undergone a phenotype drift and may no longer reflect their in situ characteristics.

Because the endothelium is continuously exposed to fluid shear force at its apical side and this force differs along the length of a single vessel, it is evident that in addition to the aforementioned intrinsic factors, mechanical stress represents an important extrinsic factor capable of modifying regional barrier properties. Shear stress is known to alter both the organization of IEJs and cell-ECM interactions (203, 412, 851). Mechanical forces can also alter barrier properties of the endothelium by activating intracellular signaling events. Exposure of cultured cells and lung capillaries to mechanical stress resulted in increased intracellular Ca²⁺ and the generation of inositol trisphosphate (IP₃) (485, 647, 826), activation of Rac (390, 970), RhoA-dependent reorganization of actin cytoskeleton (89, 851), and ₁-integrin-dependent increase in caveolin-1 phosphorylation (734; see Refs. 412 and 851). These intracellular signals induced by mechanical stress to which the vascular segments are differentially exposed may contribute to differentially modifying the barrier function of these segments.

	V. STRUCTURAL DETERMINANTS OF ENDOTHELIAL BARRIER FUNCTION

Top Previous Next References

 
The endothelial monolayer is embedded in a complex meshwork of interacting proteins, glycoproteins, proteoglycans, and glycolipids. We discuss below the importance of these cell surface structural components in modulating endothelial barrier function.

A. Glycocalyx

The glycocalyx is a negatively charged, surface coat of proteoglycans, glycosaminoglycans, and adsorbed plasma proteins lining the luminal surface of the endothelium (548) (see Ref. 726). The apparent thickness of the glycocalyx varies between 20 and 3,000 nm depending on the dye used (ruthenium red, alcian blue, or osmium tetroxide), detection method (electron or fluorescence microscopy), vessel type (capillaries, arterioles, or venules), and tissue (skeletal muscle or heart) (347, 400, 872, 980, 1005, 1006). The negative charge repels red blood cells (133, 200, 774, 827), suggesting that the glycocalyx can modulate oxygen delivery in a charge-dependent manner. The glycocalyx can be shed following exposure of postcapillary venules to formyl-methionyl-leucyl-phenylalanine (fMLP), a neutrophil-activating peptide (626, 627). This implies that the glycocalyx shields the endothelium from leukocyte attachment, although the role of the glycocalyx as a regulator of neutrophil-endothelial interactions has not been directly addressed.

The glycocalyx may also contribute to the overall function of the endothelial barrier by limiting the passage of macromolecules to the endothelial cell surface. The frog mesenteric capillary endothelium is twofold more permeable to the positively charged globular protein ribonuclease (molecular mass 13.7 kDa) than the same-sized negatively charged -lactalbumin (molecular mass 14.2 kDa) (6). This difference might be attributed to the anionic sites on the glycocalyx. Albumin (molecular mass 67 kDa) and fibrinogen (molecular mass 340 kDa) were found to permeate the glycocalyx at the same rate despite their threefold difference in molecular mass (1005), reflecting the albumin pI of 4.9 vs. pI of 6.1 for fibrinogen; thus the charge restriction imposed by the glycocalyx may determine accessibility of selected proteins. In electron micrographs of mouse capillaries perfused with cationic ferritin, the glycocalyx appeared to generate heterogeneous microdomains on the luminal cell surface due to a nonuniform distribution of negative charge (865). Together, these findings indicate that the glycocalyx could play a role in "gating" differentially charged macromolecules at specific regions of the endothelial cell surface. In addition, the possibility exists that the binding of macromolecules could itself alter the structure and charge distribution of the glycocalyx (398; see also Ref. 726) and thus influence the permeation of plasma proteins, such as the most abundant anionic protein albumin.

Further evidence to support the key contribution of the glycocalyx to endothelial barrier function comes from experiments disrupting the glycocalyx or alternatively neutralizing its negative charge. Degradation of the glycocalyx by pronase resulted in a 2.5-fold increase in capillary L_p in frog mesenteric arteries, independent of changes in intercellular cleft dimensions (4). A similar increase in endothelial permeability to macromolecules was reported in coronary arterioles or swine skeletal muscle following treatment with pronase (400) and/or heparinase (220). Glycocalyx disruption by photolysis (1006) or tumor necrosis factor (TNF)--activated proteolysis of hyaluronan (360) also led to increased permeability of the endothelium to macromolecules. In other studies, neutralization of the apical endothelial negative charge using cationic ferritin (561) or protamine (561, 909) increased the transendothelial permeability of ¹²⁵I-albumin, pointing to the key role that the glycocalyx plays in maintaining endothelial permselectivity by charge-selective exclusion of plasma proteins.

On ultrastructural observation, the glycocalyx appears as a meshlike or "fiber matrix" structure with regular spacing of 20 nm (892). Michel and colleagues (196, 892) have used mathematical modeling to relate the fiber matrix geometric properties to measured microvascular permeability. Since the glycocalyx (termed "fiber matrix" in these studies) extends between adjacent endothelial cells and lies in series with IEJs, it has been argued that the fiber matrix represents the primary barrier to fluid and solute exchange across the vascular endothelium (196; see Ref. 592). Adamson et al. (7) showed that a sufficient oncotic pressure gradient to oppose net filtration could develop across the glycocalyx. In this model, the glycocalyx (rather than IEJs) offers the highest resistance to diffusion of solutes through the endothelial barrier. IEJs are treated as insignificant barriers even to macromolecules such as albumin, based on reconstructions of electron micrographs showing that IEJs possess discontinuous regions or breaks in their structure (see Ref. 592). However, experimental evidence for this model is still lacking. Detailed studies of various microvascular beds by electron microscopy using electron-dense macromolecular tracers, such as gold- or haptenized (dinitrophenyl-labeled) albumin, have shown that albumin reaches the IEJ cleft, but does not pass through it (719). This contrasting evidence supports the hypothesis that IEJs rather than the glycocalyx are the primary limiting factor with respect to macromolecular permeability. However, the concern remains that fixatives used in these studies may have altered or even destroyed the glycocalyx; thus these studies by themselves do not provide the most robust test of this hypothesis.

The endothelial glycocalyx may also have another function; it may function as a fluid shear-stress sensor (30, 268, 446, 1042), which may regulate the production of nitric oxide (NO) (268, 608). This finding was dependent on the heparin sulfate and hyaluronic acid constituents of the glycocalyx (268). The released NO derived from endothelial NO synthase (eNOS) may stabilize the endothelial barrier through activation of focal adhesion kinase and recruitment of additional focal adhesion complexes to the basal endothelial cell surface (920, 1094). This assertion is reinforced by findings that the number of such complexes is directly related to increased monolayer electrical resistance (539, 1040). Dull et al. (227) showed that yet another function of the endothelial cell heparan sulfate proteoglycan constituent of glycocalyx is to provoke cytoskeletal reorganization leading to barrier dysfunction. They observed that clustering of cell surface proteoglycans induced by arginine-lysine polymers, a model for neutrophil cationic peptide, increased endothelial permeability. Thus the glycocalyx is implicated in the regulation of endothelial barrier function both by the fiber matrix model (see Ref. 592) and by data demonstrating activation of intracellular signals and regulation of NO production (227, 268, 608, 920, 1094).

B. Extracellular Matrix

The ECM consists of collagen IV, fibronectin, entactin, laminin, chondroitin sulfate, and heparan sulfates, perlecan and syndecan (reviewed in Ref. 443). It appears in cross-section as a fuzzy band 40–60 nm thick (560). In addition, ECM contains "matricellular" proteins (or matrix-associated proteins): thrombospondin (TSP) and secreted protein acidic and rich in cysteine (SPARC) (reviewed in Ref. 443). The first step in ECM assembly is the secretion by endothelial cells of laminin polymers, which bind predominantly to ₁-integrins (14, 502). Collagen IV polymers (also produced by the endothelium) then interact with laminin polymers in the extracellular space to form a scaffold onto which other ECM proteins are assembled to produce a basement membrane of unusually high elasticity and tensile strength (443). Endothelial cells synthesize and secrete these ECM constituents during angiogenesis and vasculogenesis (12, 269, 410, 443, 1009) and have the capacity to continuously remodel ECM in mature vessels (443, 716, 1008, 1080). The interaction of the ECM with endothelial cell surface integrins (described in sect. VIIIB) generates signals that inhibit endothelial cell proliferation and migration, but stimulate cell-cell and cell-ECM adhesion (271). The latter signals are crucial for the formation of the intact endothelial barrier. Thus it is surmised that endothelial cells adherent to the normal ECM are quiescent.

A few studies have addressed the permeability characteristics of the ECM barrier itself. Qiao et al. (729) showed that treatment of the endothelium with the lectin Ricinus communis agglutinin (RCA) reduced transendothelial albumin permeability by strengthening the ECM barrier. This finding was attributed to lectin modification of the ECM, which reduced ECM permeability to albumin. A similar concept appears to apply in vivo (177, 202, 1003). A 40% decrease in plasma fibronectin induced by infusing sheep with trypsin, and reflecting degradation of the ECM fibronectin, resulted in a sustained threefold increase in lung transvascular liquid clearance (177). Because fibronectin can stabilize ECM, this finding is indicative of increased pulmonary transvascular protein permeability resulting from degradation of ECM. Likewise, infusion of thrombin into sheep (202) or hydrogen peroxide (H₂O₂) into isolated-perfused rabbit lungs (1003) caused release of fibronectin fragments into the plasma or lymph (202) and perfusate (1003) in association with increased endothelial permeability (202, 1003). The release of fibronectin from ECM coupled with increased permeability provides correlative evidence that remodeling of ECM may result in endothelial barrier dysfunction.

In cultured endothelial monolayers, the release of fibronectin from ECM was also associated with two- to threefold increases in endothelial permeability to albumin (684, 751). In some cases, the increases in endothelial monolayer permeability were prevented by reincorporation of plasma fibronectin into ECM (199, 751, 1049, 1050), demonstrating that ECM fibronectin confers a barrier-protective property by maintaining the integrity of ECM. Other ECM protein constituents were also shown to support endothelial barrier function. Degradation of hyaluran by hyaluronidase increased permeability of endothelial monolayers as the result of increased L_p of ECM, suggesting that loss of ECM hyaluran can disrupt the endothelial barrier to liquid (730) (Fig. 4A), but "add-back" experiments have not been carried out as they have for fibronectin. Disruption of proteoglycans of the ECM by elastase has been shown to produce pulmonary edema (687). Enzymatic degradation of collagens, proteoglycans, and fibronectin by the zinc-dependent metalloproteases (MMPs), gelatinase A (MMP-2) and gelatinase B (MMP-9), increased permeability of cultured endothelial cell monolayers (685) (Fig. 4B) and also induced edema in rabbit lungs (688). These findings collectively support the role of ECM as being crucial in regulating barrier properties of the endothelium.

View larger version (19K): [in this window] [in a new window]   fig. 4. Endothelial-derived extracellular matrix (ECM) as a determinant of endothelial barrier function. A: degradation of hyaluronan with Streptomyces hyaluronidase (6 U/ml for 10 min) significantly increases ECM hydraulic conductivity (Lp) of bovine endothelial monolayer. The Lp response was greater in cells from microvascular segment. Also note that microvascular monolayer forms a tight barrier as revealed by lower basal Lp compared with arterial monolayer. However, treatment with chondroitinase C, which cleaves chondroitin, did not alter ECM Lp. These results indicate that the ECM hyaluronan provides a significant barrier to macromolecules. *Significantly different. [Adapted from Qiao et al. (730).] B: ECM stripped from bovine pulmonary endothelial monolayer (BPMVE) as well as total cell monolayer displayed an increase in 125I-albumin permeability after incubation with gelatinase (membrane metalloprotease, MMP-9) compared with untreated BPMVE. The response was similar to that induced by tumor necrosis factor (TNF)- exposure. Inhibition of gelatinase with 1,10-phenanthroline (1,10-Phe) reversed the increased permeability of ECM as well as cell monolayer to 125I-albumin, thus indicating a role of MMP-9 in regulating endothelial monolayer integrity *,#Significant increase above basal value for total cell monolayer and ECM, respectively. [Modified from Patridge et al. (685).]

ECM is also capable of remodeling the endothelium in response to shear stress acting on the endothelial cell surface. Jalali et al. (426) showed that endothelial cell alignment in response to shear stress required endothelium-derived ECM as a growth substrate. Effective mechanotransduction required the ECM constituents, vitronectin and fibronectin, and the association of Src homology containing (Shc) protein with integrins. Shc activation is probably important in reorganizing the endothelial cell cytoskeleton and focal adhesion complexes in response to shear stress (see Ref. 412). Thus ECM contributes to regulating endothelial integrity and barrier function by orchestrating signaling cues that favor cell adhesion over cell proliferation. It will be important to address whether specific cell-ECM attachments also have a role in triggering IEJs assembly, thus promoting the expression of adherens junctional proteins and contributing to junctional integrity.

C. Vesiculo-Vacuolar Organelles

Vesiculo-vacuolar organelles (VVOs), described by Dvorak and colleagues (470), are grapelike clusters of interconnected, uncoated vesicles, and vacuoles present in the continuous endothelia lining venules, small veins, and tumor vessels (Fig. 5A) (470, 546; see also Ref. 257). VVOs are enormous cytoplasmic structures that can vary from 80 to 140 nm in diameter (258). VVOs consist of 79–362 vesicles or vacuoles that are 1–2 µm at their longest dimension and collectively occupy 16–18% of the venular endothelial cytoplasm. The individual vesicles and vacuoles are substantially larger than caveolae (diameter of 70 nm on average) of capillary endothelial cells. Unlike caveolae, however, VVOs are sessile structures that can assemble into transcellular membranous channels in some instances. These channels open into the luminal, abluminal, as well as the lateral endothelial cell surface (257), but their function in regulating junctional permeability is unknown. VVOs are thought to contain caveolin-1, but VVOs appear normal at the ultrastructural level in caveolin-1-deficient mice (see Refs. 257 and 601), suggesting that the assembly of VVOs probably does not require the presence of caveolin-1.

View larger version (59K): [in this window] [in a new window]   FIG. 5. Vesiculo-vacuolar organelles (VVOs) form transcellular channels. A: 14-nm-thick serial sections through a single VVO in normal mouse skin endothelia shows two sequences (a and b) of interconnecting VVOs. A closed IEJ is also visible at the left side of panel (arrow), supporting transport via the transcellular route. R, red blood cell; L, lumen; PVS, perivascular space. Bar = 200 nm. [Adapted from Feng et al. (258).] B: amount of tracer ferritin in VVOs and immediate subendothelial space as percent of plasma concentration in normal unstimulated (None) mouse skin compared with skin injected with Hanks' balanced salt solution (HBSS), serotonin, or vascular endothelial growth factor (VEGF). Injection of either serotonin or VEGF increased ferritin concentration in VVOs and subendothelial space, indicating VVOs can provide a route for plasma proteins to cross the endothelium in response to permeability increasing mediators. *,#Significant increase above none or HBSS for VVOs and subendothelial space, respectively. [Adapted from Feng et al. (258).]

Continuous vesicles within VVOs make contact at stomata (230, 895) that are either patent or closed by diaphragms comprising the glycoprotein PV1 (895). Stan et al. (894) showed by electron microscopy that diaphragms have a central density or "knob." These stomatal diaphragms may act as a barrier between the luminal and abluminal fronts of endothelial cells (258). Permeability-increasing mediators, VEGF, serotonin, and histamine, caused the tracer ferritin to accumulate in VVOs with open diaphragms and in the subendothelial space (258). Diaphragms of VVOs are apparently subject to regulation and may be involved in increased albumin permeability, but the mechanisms of their opening and closing as well as their role in regulating transcellular permeability via VVOs have yet to be identified.

D. Development of the Endothelial Barrier

Embryonic precursor cells (EPCs), also known as angioblasts, and hematopoietic cells originate from bipotential stem cells known as hemangioblasts (reviewed in Ref. 33). The origin of these precursor cells is still a matter of debate (344, 802, 977). However, it is known that they are formed extra-embryonically in the yolk sac mesoderm from cell clusters, termed blood islands, probably in response to the endodermally-derived signals VEGF-A (604) and Indian Hedgehog (233) (for reviews see Refs. 100 and 642). Pluripotent stem cells differentiate into hemangioblasts giving rise to an intermediate preendothelial cell type that can differentiate into either a committed cell of the hematopoeitic lineage or an endothelial cell (Fig. 6). The molecular determinants of the fate of hemangioblasts are not yet fully elucidated. Evidence thus far indicates that fibroblast growth factor (FGF)-2 is an important mediator responsible for induction of endothelial precursor cells from the mesoderm (see Ref. 714).

View larger version (29K): [in this window] [in a new window]   FIG. 6. Model depicting development of endothelial barrier. Pluripotent stem cells form blood islands in the extraembryonic yolk sac splanchnic mesoderm in response to endodermally derived signals, VEGF-A and Indian Hedgehog (IHH). These cells differentiate into bipotential stem cells known as hemangioblasts. Hemangioblasts, in response to fibroblast growth factor (FGF)-2, give rise to an intermediate preendothelial cell type that can differentiate into either committed hematopoeitic lineage cells or endothelial precursor cells (EPCs) also known as angioblasts. Expression of fms-like tyrosine kinase-1 (Flt-1), tyrosine kinase with immunoglobulin like hoops and epidermal growth factor homology domain-2 (Tie-2), the receptor for angiogenic growth factors angiopoietins (Ang-1 and Ang-2), VE-cadherin, and integrins occurs later in development on these committed cells. In response to VEGF, the activation of Flt-1, Tie-2, and Ang-1 receptors signal proliferation. After proliferation, endothelial cells migrate within ECM and establish cell-cell contacts that inhibit proliferation (–) while permitting formation of the characteristic endothelial barrier. In response to wound healing or angiogenesis, endothelium is remodeled by activation of MMPs, which degrade ECM resulting in cell detachment and disruption of IEJs. ECM degradation leads to liberation of matrix-bound VEGF, again stimulating proliferation. Platelets also release angiogenic factors such as tumor growth factor-, thrombin, and sphingosine-1-phosphate. These mediators send signaling cues necessary for endothelial cell migration and proliferation. This is followed by laying down of new ECM, which results in wound repair or formation of new vessels (angiogenesis). Proliferating endothelial cells may also generate tumor vessels in response to specific pathological stimuli (described in text). EC, endothelial cells; FGF-R, fibroblast growth factor receptor. See text for details. [Modified from Bohnsack and Hirschi (100).]

The high proliferation rate of EPCs may distinguish them from endothelial cells that are shed from the vessel wall (33). The proliferative capacity of endothelial cells can be reactivated in response to certain physiological stimuli (e.g., acute wound healing, angiogenesis, cycling endometrium, pregnancy) or pathological stimuli (e.g., tumor growth, rheumatoid arthritis) (759) (Fig. 6). Under these conditions, EPCs can also contribute to vessel growth. However, the relative contributions of EPCs and preexisting endothelial cells to the repair of damaged blood vessel are still unclear (see Ref. 556). Because EPCs have the potential to differentiate in situ into endothelial cells, they may induce vasculogenesis resulting in neovascularization (33). Therefore, this may be an important strategy for reannealing an injured endothelial barrier resulting from inflammation and could thereby restore the vascular integrity and basal level of permeability (165, 365, 556, 1002).

Angiogenic and coagulation factors, proteinases and their inhibitors, junctional adhesion molecules, and ECM proteins and their receptors contribute in a complex way to the mechanism of angiogenesis (see Ref. 556). For example, MMPs and plasminogen activators (urokinase-type plasminogen activator and its inhibitor plasminogen activator inhibitor-1) signal formation of vessels by degrading ECM, which results in the liberation of matrix-bound VEGF and proteolytic activation of chemokines such as interleukin-1 (47, 67, 888). Angiogenesis was inhibited in mice lacking plasminogen or MMP-2 (47, 423). Factors released during intravascular coagulation also were shown to induce angiogenesis (140, 259, 663, 753). Fibrin deposition signaled the migration of endothelial cells, thus inducing angiogenesis (754). Activation of platelets also led to a release of angiogenic factors VEGF, PDGF, tumor growth factor-, thrombin, and sphingosine-1-phosphate (S1P) (reviewed in Ref. 122). These mediators may send the signaling cues necessary for angiogenesis by inducing the disruption of IEJs and ECM (212, 634, 780, 1060, 1111; see also Ref. 212).

Interestingly, S1P also has an important endothelial barrier protective effect (300, 511, 780, 794). When administered along with thrombin, S1P suppressed thrombin's effect of increasing endothelial permeability and restored barrier function (300, 794). Similarly, VEGF and Ang-1 cooperate in the formation of blood vessels (see Refs. 293 and 910). However, unlike VEGF, which is known to increase endothelial permeability, Ang-1 is implicated in inducing the formation of a restrictive barrier (295, 943). In the adult vasculature, Ang-1 opposed VEGF-induced increase in endothelial permeability (295, 942). Overexpression of Ang-1 in skin of adult mice stimulated the growth of nonleaky vessels (943). These findings raise the possibility that S1P or Ang-1 may link angiogenesis with the formation of a stable endothelial barrier. Furthermore, these results point to the potential role of these and other mediators [such as ATP (472)] in reducing endothelial permeability and acting as anti-inflammatory agents (discussed in sect. XIIB).

The neovascularization arising in response to physiological events such as pregnancy differs from the processes forming microvessels in response to pathological stimuli in tumor-driven angiogenesis (556). Tumor vessels are described as "highly disordered, tortuous, dilated and leaky" (556), and it is apparent that the function of VEGF is upregulated in these vessels (231). The contribution of mediators involved in angiogenesis (discussed above) and the formation of tumor microvasculature is an area of intense investigation, and as such it is beyond the scope of this review. It has been extensively covered recently in an up-to-date review (556).

	VI. PHYSIOLOGICAL SIGNIFICANCE OF ALBUMIN PERMEABILITY

Top Previous Next References

 
The concentration of albumin in human plasma is 3 g/100 ml (60% of total protein), making albumin the most abundant plasma protein. In addition, the molecular structure and charge (pI = 4.9) of albumin facilitates the cotransport of a number of hydrophobic molecules, enzymes, and hormones across the endothelium. In this section, we review the significance of the permeability of the endothelium to albumin.

A. Albumin Regulation of Tissue Oncotic Pressure and Endothelial Barrier Integrity

The oncotic pressure generated by plasma proteins (_c = 25 mmHg) is a key factor in maintaining fluid balance across capillaries (499). _c plays an important role in fluid reabsorption across the capillary wall, as the plasma protein concentration is greater in vessels than the interstitial space. Plasma albumin accounts for 65% of _c, and other plasma proteins, e.g., globulins and fibrinogen, contribute according to their concentrations (71, 1044). Albumin has a plasma half-life of 15–19 days (reviewed in Refs. 701 and 702) and thus must be replaced via resynthesis to maintain _c. Interestingly, like IgG, the albumin was shown to bind the major histocompatibility complex-related Fc receptor (FcRn) at low pH and be shielded from degradation, which significantly prolonged the plasma half-life of both proteins (156). Extravasated albumin is recycled into the general circulation by lymphatic vessels, and albumin newly synthesized by hepatocytes is secreted into the circulation (at the rate of 15 g/day in humans) (703). Plasma albumin moves into the extravascular space by crossing the microvessel barrier and entering the interstitial tissue where it serves as the chief interstitial oncotic agent. The endothelial cell layer thus regulates the transport of albumin into the interstitium and in this manner controls the transendothelial oncotic pressure gradient (_c – _i), the difference between _c and tissue oncotic _i pressures, the principle Starling force responsible for fluid reabsorption.

Plasma albumin also has additional functions in mediating endothelial barrier stability (397, 399). This notion is supported by studies in which removal of albumin from the perfusate resulted in a 1.5-fold increase in capillary wall L_p (932). Albumin contributes to the maintenance of endothelial barrier function by interacting with the glycocalyx (397, 399) as shown by the finding that loss of adsorbed albumin from the glycocalyx increased the transport of tracer ferritin across the endothelium (808). The basis for this is not totally clear. Albumin's interaction with ECM proteins may also regulate endothelial barrier properties. Kajimura et al. (442) showed that removal of albumin from the perfusate increased L_p of microvessels and permeability to -lactalbumin, supporting a role for albumin in regulating endothelial barrier integrity. However, these may not be unique functions of albumin. Apparently, analbuminemic humans and rats have a normal fluid balance. They have normal _c and _i values and _c – _I gradient, since they compensate by increasing the production of other proteins (102, 435, 436, 633, 750, 907, 933, 1036). Ultrastructurally, microvessels from analbuminemic rats also appeared normal (D. Predescu, unpublished observations). Compensatory mechanisms increased the production of -macroglobulins, immunoglobulin G, and fibrinogen (239) and may thus be able to maintain the _c – _i gradient and restore fluid balance and endothelial barrier integrity.

B. Albumin as a Chaperone

Albumin has a cargo chaperone function as it binds to many substances in the plasma and facilitates their delivery across the vessel wall barrier. It is not clear whether albumin is cotransported with its cargo molecules in all cases or whether albumin is involved in the transfer of hydrophobic cargo molecules to specific binding proteins on the endothelial cell surface. In the case of fatty acids, evidence favors the latter mechanism (see Ref. 979). Permeation of free fatty acid-conjugated albumin via transcytosis was threefold greater compared with delipidated albumin (28, 294), perhaps on the basis of higher affinity binding of lipidated-albumin to the endothelial cell surface (294). Free fatty acids and other lipids such as S1P conjugated to albumin are required for many vital functions (600, 703, 1087); they serve as an energy source in muscle tissue, substrates in surfactant production and lipid synthesis in lung and adipose tissues, and provide cues in development and tissue patterning as in the case of S1P. Albumin also plays an important role in the transport of drugs such as digoxin to target organs (662). These phenomena have an important clinical impact on drug efficacy, particularly because they have relatively narrow therapeutic indexes (74). Albumin has also been shown to be a carrier protein for the amino acid tryptophan (675). There is some evidence that albumin also acts as a carrier for thyroid hormone, transporting it across the capillary endothelium (825). These studies showed that endothelial cells take up albumin-bound thyroxin, which after exocytosis, dissociated from its carrier in the pericapillary space, thereby providing free thyroxin to target organs (359).

C. Other Functions of Albumin

A recent study by Siddiqui et al. (852) using a protein alignment algorithm has shown a structural homology between regions of human albumin and human transforming growth factor (TGF)-1. In mature human TGF-1, a 112-residue peptide has marked homology with the human albumin amino acid sequence. A similar 26% homology exists with mouse albumin and mouse TGF-1 for a 53-amino acid stretch (852). The implications of this finding are not yet clear, but it is possible that albumin has an important "cytokine-like activity" at low levels, a function distinct from its role as a carrier protein and oncotic agent. Intriguingly, Tiruppathi et al