The GCW sieving coefficient, _gθ, is defined as the ratio of the Bowman's space to plasma concentration for any given molecule. Since the GCW prevents macromolecule filtration despite its large hydraulic conductivity (water permeability), _gθ for macromolecules is usually much lower than 1 and depends on size, charge, and shape of the macromolecule. It should also be noted that proximal tubule albumin reabsorption contributes to the low urinary albumin concentration. Hence, _gθ_Alb can only be determined from urinary albumin levels if the modification of urine by proximal tubule cells is blocked. Alternatively, the albumin/tracer concentration in Bowman's space or early proximal tubule must be quantified. Functional models of the GCW (20–24), derived from experimental sieving data for infused tracers like ficoll (25–28), dextran (29–31), albumin (32–34) or endogenous circulating proteins (28, 35, 36), suggest that the GCW functions as a composite gel-like mesh with a high density of pores having a radius in the 45–50 Å range, a few large pores with radii of 75–155 Å, and a negatively charged layer at the blood/endothelial interface with a charge density of ~35–45 mEq/L (22). Taking into account these experimentally derived parameters, a mathematical model predicted a _gθ_Alb of 2 × 10⁻³ (a ratio of ultrafiltrate: plasma albumin of 2: 1,000) (22). In fairly close agreement, the best measured estimate of _gθ_Alb obtained by micropuncture in rats was 6.2 × 10⁻⁴ (0.62: 1,000) (33), and _gθ_Alb derived from radiolabeled albumin tracer studies (34) was 6 × 10⁻⁴. More recent quantification of _gθ_Alb by intravital two-photon fluorescence microscopy in rats has varied more widely: 0.034 (37), 0.014 (38), 0.002–0.004 (39) and 0.00044 (40). It appears that technical limitations account for some of the higher values by this approach (39, 41). Norden et al. (36) studied humans with the Fanconi syndrome due to Dent's disease, in whom proximal tubule albumin reabsorption is negligible, and found that _gθ_Alb averaged 7.7 × 10⁻⁵. Similarly, when megalin and cubulin were conditionally deleted in mice (42, 43), _gθ_Alb was estimated at 7.5 × 10⁻⁵ and 1.7 × 10⁻⁵, respectively. In such mice, streptozotocin diabetes (43) or superimposed podocin (42) deletion resulted in a significant increase in _gθ_Alb. Since proximal tubule uptake of albumin was completely absent in megalin/cubulin deficient mice (42), these data, taken together with those from rats and humans, indicate that ~0.01–0.1% of plasma albumin passes through the GCW into Bowman's space. In normal humans therefore, an estimated 500–5,000 mg of albumin are filtered each day (44). Proximal tubule uptake then reduces excretion to <30 mg/day.

Mathematical models of sieving data (20, 24, 25, 35, 45, 46) agree that the GCW is best described as a hydrated gel that hinders entry and movement of macromolecules based on size, shape, flexibility and charge (21, 22). Gaps in the gel that allow relatively free filtration of water and small solutes are modeled as abundant small “pores” with a molecular radius cutoff in the 45–50 Å range. The models include a small number of larger “pores” to account for the transit of a small fraction of large macromolecules. It turns out that the shape and flexibility of macromolecules influence movement through the small, abundant gaps given that large, elongated uncharged carbon nanotubes seem to be filtered relatively freely (47). Recent data furthermore suggest that compression of GCW components against intact podocytes may influence the size of gaps in the gel and therefore the molecular size cutoff (48). Abundant experimental data in animals (25–27, 30, 34, 49–52) and humans (20, 29) and ensuing mathematical models (20, 22, 27, 53–56) have concluded that for molecules like albumin whose size is close to the 45–50 Å radius cutoff, negative charge also impedes movement into and through the gel, compared to the same or similar neutral molecule. Conversely, neutralization of negative charges in the GCW with cationic protamine sulfate (57, 58), hexadimethrine (34, 59) or their removal with neuraminidase/sialidase (60, 61) which strip sialic acid from the GCW, all increase albumin excretion rapidly and reversibly. However, because these interventions also cause structural changes in podocytes and glomerular EC, the cause-effect relationship specifically between the reduction in GCW negative charge density and albuminuria was not proven. Nonetheless, infusion of enzymes to destroy negatively charged glycosaminoglycans (GAGs) also increase the fractional clearance of albumin across the GCW (62, 63), even without changes in EC or podocyte ultrastructure. By contrast, in isolated GBM, no change in albumin permeability was observed when negative charges were neutralized with protamine (64), and the substantial reduction of GBM negative charges due podocyte-specific deletion of agrin ± perlecan (65), or the heparan sulfate glycosyltransferase EXT1 (66) raise albumin excretion only minimally. While a change in _gθ_Alb in the knockout mice could have been masked by proximal tubule albumin reabsorption, the results nevertheless cast some doubt on the possibility that “fixed negative charges” located in the GBM play a major role in charge selectivity.

The first detailed transmission electron microscopy (TEM) studies of glomeruli caused Farquhar (67) to rule out the glomerular endothelium as a component of the barrier because its fenestrae, lacking visible proteinaceous diaphragms, seemed simply too large to restrict anything smaller than circulating cells. Hence, the glomerular basement membrane (GBM) (68) and podocyte filtration slit diaphragms were held to be the main barrier to macromolecule filtration, with charge selectivity assigned to the GBM (46, 69, 70). This deduction was strengthened by findings of negatively charged sites within the GBM (71–73), congruent functional studies showing charge selectivity of the GCW (34, 51, 74), and the fact that disorders affecting podocytes or GBM all lead to proteinuria (75).

Nonetheless, the concept that the GBM and podocyte slit diaphragm constitute the main barrier to GCW protein flux cannot be reconciled with the fact that bulk convective transit of macromolecules through wide open glomerular endothelial fenestrae would rapidly clog the filter unless high-capacity mechanisms returned them, intact, to the circulation (76, 77). While podocytes endocytose and degrade albumin and other macromolecules (78), this mechanism does not have the capacity to deal with a daily load of albumin in the 4–7 kg range. Long albumin and immunoglobulin half-lives and a low renal albumin degradation rate (79) are also inconsistent with removal and degradation of massive quantities macromolecules by podocytes. Farquhar (67) suggested that macromolecules pass through the endothelium into the GBM and sub-podocyte space and then are swept into the mesangium. However, bulk transit of plasma proteins through the mesangium back into the circulation has never been demonstrated, and glomerular lymphatics that would be needed to clear them from the mesangium have not been found (80).

It turns out that under physiological conditions, endogenous albumin (81, 82), or infused gold-conjugated albumin (68), actually do NOT penetrate glomerular endothelial fenestrae, leading to the more attractive conclusion that a barrier covering the endothelium and extending into endothelial fenestrae retains all but a small fraction of albumin and other large proteins within the circulation. Indeed, disruption of glomerular EC adherens junctions by EC-specific notch1 activation or VE-cadherin deletion results in glomerular EC glycocalyx damage and significant proteinuria (83), implying that fully differentiated glomerular EC with intact adherens junctions and glycocalyx are critically important components of the GCW barrier. No doubt, as detailed by comprehensive models of GCW permselectivity (48, 84–87), one cannot consider any single GCW component in isolation (88), but the role of the glomerular endothelium in GCW permselectivity, for which data were already accumulating in the 1980's (9, 89) is only now becoming widely accepted (10, 53, 88, 90–95).

With conventional processing for transmission or scanning EM the glomerular endothelium has the appearance of a sieve, with fenestrae ~60–80 nm (600–800 Å) in diameter accounting for ~30% of the glomerular EC surface area. Glomerular EC fenestrae are plasma membrane-lined, transcellular pores that lack the proteinaceous PV-1-based diaphragms observed in most other fenestrated endothelia (97). The size and density of glomerular EC fenestrae accounts for the enormous hydraulic conductivity of the GCW (98). Any decrease in their density and/or size leads to a reduction in GFR, for instance in experimental models of uranyl nitrate (99) and gentamicin (100)-induced acute renal failure, streptozotocin induced diabetes (101), and in humans with diabetes (102) and preeclampsia (103).

The radius of glomerular EC fenestrae is much larger than the effective radius of circulating macromolecules that are not filtered, for instance orosomucoid (29 Å), albumin (36 Å), Transferrin (43 Å), IgG (55 Å), α2-macroglobulin (90 Å) and fibrinogen (108 Å) (35, 104), so they were initially assumed to allow their free convective movement into the GBM. Yet, studies in non-glomerular capillaries had suggested that EC fenestrae are impermeant to macromolecules (104), and Luft (105) found that EC do not present a “naked” surface to circulating plasma, given that perfused electron-dense ruthenium red accumulated on the EC luminal surface revealing an anionic coat. Avashi and Koshy (106) perfused kidneys with ferritin, a multimer ~120 Å nm in diameter, so much smaller than glomerular EC fenestrae. Cationic ferritin densely decorated the glomerular EC surface and the core of fenestrae and did not penetrate into the GBM. Anionic ferritin was completely excluded from the EC surface and the GBM, indicating that the EC coat excludes negatively charged macromolecules. Furthermore, adhesion of cationic ferritin was removed by neuraminidase and reduced by heparinase and hyaluronidase without change in EC or podocyte ultrastructure. The authors concluded that “glomerular endothelial fenestrae are not empty holes” but “are occupied by an anionic matrix that is visualized only following the binding of an electron-dense tracer. In this respect the matrix in the fenestrae is similar to the glycocalyx at the external surface of cells which also remains invisible in unstained preparations” (106). Rostgaard and Qvortrup (96) extended these observations using oxygen-carrying perfusion fixation and tannic acid/uranyl acetate staining. They observed “sieve plugs” in fenestrae of intestinal and peritubular capillary EC, and a similarly stained layer covering the EC. But in glomerular EC the same procedure revealed only a delicate ~300 nm thick surface coat (96). Hjalmarsson et al. (107) reported a colloidal lanthanum labeled ~60 nm thick EC glycocalyx that was revealed in oxygen-carrying perfusion fixed, tannic acid-stained tissue. They observed a thicker ~200 nm coat ascribed to glycocalyx plus ESL. In their study, cupromeronic blue stained tissue showed a semi-ordered proteoglycan network within the fenestrae (107). In glomerular EC, Hegermann (108) recently visualized an amorphous 200–300 nm thick layer with alcian blue. With cationic colloidal thorium they observed an electron-dense layer that filled the fenestrae, extended from the EC surface by 50 to 300 nm and was organized into bundles that were about 50 nm wide at the EC surface, with sub-organization into wider and wider bundles as they moved away from the surface. They concluded that the glycocalyx proper represents bundles of proteoglycans that are anchored to the EC plasma membrane and extend vertically from the cells toward the capillary lumen (Figure 1). These findings are consistent with those by Squire et al. (109) in non-glomerular EC, who reported vertically organized bundles extending from the EC surface, intertwined with horizontal strands forming a lattice with gaps that could account for size-selectivity. Indeed, Fan et al. (110) were able to visualize hyaluronan (HA) and heparan sulfate (HS) at the single molecule level in cultured EC, using stochastic optical reconstruction microscopy (STORM), i.e. a super-resolution imaging technique with a resolution of 20 × 50 nm. They reported that HS bundles extend vertically from the EC surface and are intertwined with horizontally arranged, long HA strands to form an organized lattice-like network on the EC surface (110, 111).

It could still be argued that tissue processing and deposition of electron-dense material for transmission EM could produce artifact that might overestimate the dimensions of the EC glycocalyx and/or might remove the ESL. To assess the thickness of the EC glycocalyx/ESL the zone of exclusion for RBCs or fluorescently labeled tracers above the EC plasma membrane is therefore commonly determined. For instance, in hamster cremasteric muscle capillaries, the exclusion zone for dextran 70 and RBCs was found to be 400–500 nm (112), and was significantly reduced by hyaluronidase, and partially reconstituted hyaluronan or chondroitin sulfate infusion (113). In renal glomerular capillaries, the zone of exclusion for infused intralipid droplets was ~200 nm and was also significantly reduced by enzymes that cleave glycosaminoglycans (114) and by elution of ESL components with hypertonic NaCl (115). Evaluation of the EC glycocalyx/ESL thickness is now a commonly used technique in human clinical research (116) and has helped define changes in the height of the glycocalyx/ESL in disease.

Hence, all EC surfaces are covered by an organized glycocalyx and an associated ESL. These form an anionic surface that results in repulsion of anionic macromolecules as well as the anionic glycocalyx of circulating cells. The anionic EC glycocalyx extends into the fenestrae, forming a semi-permeable matrix that allows rapid filtration of water and small solutes, but not macromolecules. The lattice-like arrangement of the EC coat furthermore suggests that it participates in size-selectivity of the GCW. It is already well-established that immune-activation of EC changes its glycocalyx, breaching the normal glycocalyx/ESL (116), allowing EC interactions with circulating cells and platelets and facilitating thrombosis (117). Under those conditions it is therefore expected that permselectivity is also reduced.

The structure and function of the EC glycocalyx depends critically on its glycosaminoglycan (GAG) composition. GAGs are long, unbranched polymers of repeating disaccharides, each consisting of an amino sugar (N-acetylglucosamine or N-acetylgalactosamine) and either a galactose or uronic acid sugar (Figure 1A). Due to their high hydroxyl and sulfate content, GAGs are negatively charged; they bind large amounts of water, critical for their viscoelastic properties; they repel negatively charged molecules like albumin and they often serve as co-receptors for growth factors and cytokines. In the luminal EC glycocalyx GAGs confer anti-coagulant properties, they repel circulating cells, and they impart the charge barrier to the endothelium. Heparan sulfate (HS), chondroitin sulfate (CS)/dermatan sulfate (DS) and keratan sulfate (KS) GAGs are all assembled on core proteins of distinct proteoglycans. Hyaluronic acid (HA; aka hyaluronan) (119) is the only GAG synthesized outside the Golgi as a stand-alone polysaccharide composed of non-sulfated N-acetylglucosamine/glucuronic acid disaccharide repeats. HS (120), the most abundant GAG in the EC glycocalyx, consists of N-acetylglucosamine/uronic acid repeats, CS GAGs (121) consist of N-acetylgalactosamine/glucuronic acid and KS (122). GAGs are composed of N-acetylglucosamine/galactose disaccharide repeats (Figure 1A). Addition of GAG chains to proteoglycans in the Golgi is initiated by specific glucuronyl transferases that covalently couple a bridging tetrasaccharide through O-linkage on serine or threonine or N-linkage on asparagine, followed by elongation of the polysaccharide chain and subsequent position-specific modifications through de-acetylation/sulfation and epimerization. For instance, within the initial HS polymer, epimerization converts some of the glucuronic acids to iduronic acid (123) and de-acetylation converts some of the N-acetylglucosamine to glucosamine. Similarly, CS is converted to DS through epimerization of glucuronic to iduronic acid. For proteoglycans, structural and functional diversity is therefore not only due to their different protein cores, but also the GAG type, the number and length of their GAGs, as well as position-specific epimerization and sulfation. Since enzymes involved in GAG-core protein attachment, chain elongation and modification serve multiple proteoglycans, any mutations or deletion in the many enzymes that carry out these functions tend to have different, and often more severe phenotypes than mutations or deletion of any one proteoglycan core protein (124).

Cleavage by enzymes that are GAG-specific, namely hyaluronidases, heparinase and chondroitinase have been used extensively to define the functional role of GAGs in the GCW and EC glycocalyx, and shedding of HA and EC cell-surface proteoglycans (125) into the circulation due to endogenous enzymes is used as an indicator of glycocalyx damage (126, 127).

In vertebrates HA is produced by HA synthases encoded by three distinct genes (HAS1-3). The HA synthases are integral plasma membrane proteins with multiple membrane-spanning domains (Figure 1C). The catalytic site within their hydrophilic core acts as a polymerase, converting soluble intracellular UDP-GLcNAc and UDP-GLcA to polymeric HA, simultaneously extruding the growing polymer it into the extracellular space (128). The length of the HA polymer varies from about 1,000 to 10,000 kDa. The rate of HA synthesis depends on the availability of substrate sugars (129), and conversely, high rates of HA synthesis are associated with a shift of cellular metabolism to glycolysis (130–132). HA binds to cell-surface receptors CD44 (133), RHAMM (receptor hyaluronan mediated motility) (134) and the lymphatic EC receptor LYVE-1. The cytoplasmic domain of CD44, a single pass membrane-spanning receptor, is coupled to cortical actin by ERM (ezrin radixin moesin) proteins (135). The interaction of CD44 with HA enhances EC barrier function (136) and is necessary for transmission of shear force signals that cause Rac1-dependent EC re-orientation (137), enhanced nitric oxide synthesis (138) and increased HAS2 expression (129, 139). In keeping with luminal HA/CD44 interactions, HA loss from the glycocalyx profoundly reduces shear-force induced NO synthesis (140).

HA turnover is rapid and regulated, in part, through HA degradation by hyal-1 and−2 (Hyaluronidase-1 and−2) and by Cemip-1 and−2 (Cell Migration-Inducing hyaluronidase-1 and−2; the latter also known as transmembrane protein 2/TMEM2) (131, 141). Hyal-2 is a GPI-anchored plasma membrane-associated enzyme that cleaves CD44-bound HA. The fragments are then internalized by the GAG scavenger receptor HARE (142). The cemid1 (143) and−2 (144) hyaluronidases are single-pass plasma membrane-spanning proteins that degrade extracellular HA into small, bioactive extracellular fragments (oHA). These oligosaccharides modify VEGF signaling in EC (144).

HA polymers are hydrated with 15 H₂O molecules per disaccharide unit (145). At the apical/luminal EC surface HA interweaves with other components of the glycocalyx/ESL (110, 134) and shear stress (129, 146) and inflammatory stimuli (147) augment HA accumulation in the EC glycocalyx. HA binds proteoglycans in the hyalectin family (see below) forming large, patterned aggregates. A HA-versican lattice may in fact account, at least in part, for the semi-ordered appearance of the apical EC glycocalyx (108, 109), and perhaps also contribute to size-selectivity of the GCW.

In systemic microvessels, destruction of HA by hyaluronidase markedly reduces the height of the EC glycocalyx and its macromolecular barrier function (113). Similarly, in glomerular capillaries the height of the EC glycocalyx is reduced by hyaluronidase infusion along with an increase in the fractional excretion of albumin (114). EC-specific, conditional HAS2 deletion in mice reduced glomerular EC HA and cationic ferritin labeling, along with progressive proteinuria, glomerular EC ultrastructure changes and capillary involution (148). Conversely, in the mouse streptozotocin model of diabetes deletion of the HYAL1 gene reduced hyaluronidase activity, preserved the EC glycocalyx and was associated with less glomerular barrier disruption than in wild-type mice (149). These findings are consistent with those in diabetic patients, where higher levels of circulating HA and hyaluronidase were found to be associated with the development of microalbuminuria (150, 151), and where endothelial glycocalyx disruption was associated with a substantial reduction in glomerular endothelial HA content (148).

Among many glycosylated proteins, proteoglycans are distinguished by their very long, unbranched, sulfated GAG sidechains usually accounting for at least 60% of their molecular mass, the exception being perlecan, where the GAG chains are a minor component (see below). Classification of distinct proteoglycans is based on the structure of their protein core, the type and number of associated GAG chains, and their molecular interaction profile. Some proteoglycans are integral membrane-spanning proteins, some are covalently bound to the outer leaflet of the plasma membrane by GPI (glycosylphosphatidylinositol) anchors, and others are secreted. The principal structural glycocalyx proteoglycans in EC are membrane-spanning syndecans, GPI-anchored glypicans and secreted perlecan and versican. Other small, secreted proteoglycans are produced by EC and participate in defining the dynamic EC phenotype. In addition to their GAG chains, proteoglycans can also be modified by branched oligosaccharide side chains, some terminated by sialic acid. There currently is a paucity of data on potential effects of such modifications on the properties of the glycocalyx/ESL. Thus, future studies are needed to better understand their impact on glomerular permselectivity.

Sialomucins in the EC glycocalyx are integral plasma membrane glycoproteins each with a single membrane-spanning domain, a large extracellular “mucin” domain and a cytoplasmic domain that interacts with cortical actin via ERM (ezrin radixin moesin) proteins. Mucin domains are ser/thr/pro-rich regions densely decorated by O-glycans initiated by core 1 β1,3 galatosyltransferase (215) and containing terminal sialic acids. Silomucins largely accounting for the high sialic acid content of the EC glycocalyx. Several sialomucins, including podocalyxin (216, 217), endoglycan (218) (aka podocalyxin 2), CD34 (219), and endomucin (220, 221) are expressed by EC, while podoplanin is restricted to lymphatic EC (215). Sialomucins are sorted to the apical/luminal surface (216, 222) of EC where they play a repulsive role during embryonic vascular lumen formation (222, 223) and they repel circulating cells by virtue of their negative charge (219). EC sialomucins (219) play a role in hematopoietic precursor trafficking (219) and as counter-receptors for L-selectin, though this latter function requires modification of the O-glycan by carbohydrate 6-O-sulfotransferase restricted to high endothelial venules (224). The potential role of sialomucins the glomerular EC barrier to macromolecule flux has only been studied indirectly, through infusion of neuraminidase (60, 61, 95, 106, 225–227), which removes sialic acid from the GCW and consistently results in albuminuria. However, since podocytes also express the sialomucins podocalyxin and podoplanin, it is possible that the neuraminidase-induced GCW barrier results from stripping of sialic acid from both, EC and podocyte sialomucins. Even so, in cultured EC, podocalyxin knock-down markedly reduces the trans-endothelial resistance. EC-specific podocalyxin deletion in mice alters EC structure and reduces the EC barrier function in lung and brain in the presence of pro-inflammatory stimuli (228–230). Global podocalxyin deletion in mice is lethal due to a major podocyte defect, though in these mice glomerular EC are also thickened and lack fenestrae (231). Conditional deletion of the core 1 β1,3 galatosyltransferase, critical for sialylation of all sialomucins, results in marked albuminuria (232). Finally, in children with streptococcus pneumoniae associated HUS, neuraminidase-mediated removal of sialic acid from sialoglycoproteins in the EC glycocalyx likely plays a significant role in triggering intravascular coagulation, hemolysis, and acute renal failure accompanied by proteinuria (233).

Produced by the liver at a rate of ~10 g/day, albumin is the most abundant circulating protein, with normal plasma concentrations of 35–50 g/L and a half-life of 19–20 days. Encoded by a single gene, human albumin is secreted as a monomeric non-glycosylated polypeptide consisting of 585 amino acids (MW ~66.5). A relatively high content of acidic amino acids and fatty acid binding result in an estimated isoelectric point of 4.7–5.8 (234). Hence, in physiologic solutions albumin is negatively charged. Structural analyses (235–238) show that albumin is not a simple sphere, but that it consists of 3 major domains, each containing subdomains, with 17 intramolecular disulfide bonds contributing to 3D folding. Normally, albumin assumes a heart-shaped triangular structure (237) with a hydrodynamic radius of 36.2 Å, though it can assume other conformations depending on pH, including an expanded cigar-like shape with a hydrodynamic radius of 61.5 Å (238). Were it not for its negative charge, the structure of albumin and its ability to take on different conformations suggest it could penetrate a meshwork with mean pore radii in the range of 40–60 Å, like the glomerular capillary wall (20). The albumin monomer contains hydrophobic pockets that bind many lipophilic substances, among them endogenous fatty acids, steroid hormones, thyroid hormone, bilirubin, vitamins, and phytochemicals. Its binding affinity for many drugs and its potential as drug carrier have been extensively investigated. Non-enzymatic glycation of albumin results in conformational changes that alter its interaction with endogenous substances and drugs, increase its half-life and reduce formation of albumin aggregates (38, 239).

While albumin flux across the endothelial glycocalyx and ESL is highly restricted (81), albumin also associates with the ESL and alters the endothelial barrier function. In vitro NMR studies show interactions between albumin and hyaluronan resulting in albumin/hyaluronan complexes that hinder the mobility of albumin in solution (240). Albumin binds to immobilized artificial glycocalyx composed of hyaluronan, heparan sulfate and chondroitin sulfate GAGS though its binding affinity is low (241). In cultured EC, albumin similarly associates with the EC cell surface in a reversible fashion (242), and in perfused frog mesenteric microvessels (243) endogenous albumin was observed in a ~200 nm thick layer covering the EC surface (243). Likewise, albumin associates with lung EC glycocalyx; Lowering perfusate plasma protein/albumin content significantly increased penetration of endogenous, negatively charged ferritin into the vessel wall (244). Similarly, in isolated dog glomeruli, lowering perfusate albumin concentrations raised GFR not only due to a reduction in the colloid osmotic pressure, but also due to an increase in the hydraulic conductivity of the glomerular capillary wall (245). A similar effect of albumin on the hydraulic conductivity was also reported for non-glomerular vessels in frog (225) and rabbit (246). Finally, in the analbuminemic Nagase rats, the negative charge density of the glomerular EC coat was reduced, with enhanced penetration by macromolecules in the 60–90 kDa range both corrected by albumin infusion (247). In tracer studies, enhanced flux of glycated albumin across the EC layer has been reported (248), though by two-photon microscopy its GCW sieving coefficient was not different than that of native albumin (38). Instead, reduced uptake of glycated albumin by the neonatal Fc receptors (FcRn) in proximal tubule cells enhanced its renal excretion (38). Thus, in vitro and in vivo studies all indicate that albumin associates with the EC coat, reducing the filtration coefficient and the trans-endothelial flux of macromolecules (179). The relatively low affinity of albumin for glycocalyx/ESL components furthermore suggests that bound albumin is constantly exchanged with circulating albumin. Given that albumin binds the bioactive lipid S1P (178, 249), and that S1P protects the EC glycocalyx (250), it is likely that albumin not only changes the function of the endothelial glycocalyx/ESL through physical binding, but that it also delivers mediators to the EC that alter glycocalyx/ESL synthesis and degradation.

In humans, orosomucoids are produced by two distinct genes, ORM1 and 2. Orosomucoids are sialylated, negatively charged circulating glycoproteins produced mainly by the liver (251) but also by EC (252). Basal plasma concentrations are in the range of 1 g/L. Orosomucoid synthesis is strongly induced by inflammatory stimuli, like lysopolysaccharide (LPS), and interleukins-1 and−6; they are therefore considered to be acute phase reactants (251). Orosomucoid core proteins (~21.5 kDa) undergo complex and variable glycosylation prior to secretion, increasing their molecular mass to ~ 44 kDa, and resulting in a high sialic acid content. Orosomucoid glycosylation is modified in response to acute inflammatory stimuli, increasing the density of sialyl-Lewis × epitopes (sLe^x) that can interact with EC surface P- and E-selectins (253, 254). In cultured EC, orosomucoid 1 binds both high affinity, relatively low capacity cell surface receptors, and lower affinity, extremely high capacity binding sites (255). The former likely represent EC P- and L-selectin binding, the latter association with the ESL, increasing the ESL negative charge density (256). Pertinent to this discussion, orosomucoid reduces the flux of albumin across rat hindlimb microvessels (257), and lactalbumin flux across frog mesenteric vessels (258) and the blood brain barrier (259). In the kidney, perfusate containing orosomucoid reduces the fractional clearance of albumin (32, 260), and administration of of orosomucoid protects rats from puromycin aminonucleoside—induced albuminuria and GFR loss (261). Hence, orosomucoid, which is not filtered but associates with the surface of EC, reduces albumin flux across EC, by increasing the ESL negative charge density. Orosomucoid-dependent modulation of inflammatory cell recruitment and EC transmigration may also contribute to the renal response to injury. For instance, urinary excretion of orosomucoid increases in patients with type 2 diabetes and may be a biomarker for EC dysfunction due to low-grade inflammation (262). In triple (ORM1-3) knockout mice (unlike humans, mice have 3 ORM genes), enhanced inflammation and a greater susceptibility to renal fibrosis in the unilateral ureteral obstruction (263) and acute ischemia-reperfusion (264) models have been reported. At this time, quantitative glomerular permselectivity studies in ORM deficient mice are lacking.

EC dysfunction, characterized by diminished flow-mediated vasodilation due to reduced endothelial NO production, signals generalized EC abnormalities in patients with cardiovascular disease, the metabolic syndrome, diabetes and chronic kidney disease (10, 174, 265, 268). Microalbuminuria is strongly associated with EC dysfunction (269), predicts cardiovascular morbidity (19) and is one of the earliest indicators of generalized, chronic EC injury (9). Note again that microalbuminuria tends to underestimate the GCW defect, due to proximal tubule reabsorption of albumin (79). EC glycocalyx disruption with a substantial reduction in glomerular EC HA content has been documented in patients with diabetic nephropathy (148). In generalized vascular disease, microalbuminuria is associated with a reduction in EC glycocalyx/ESL height and increase in circulating EC glycocalyx components, including hyaluronan (149, 270) and proteoglycans (126, 150, 151, 174). As EC glycocalyx glypican-1 is required to elicit shear-induced NO synthesis (184, 186), it seems likely that glycocalyx degradation is, in fact, the proximate cause of reduced flow-dependent NO synthesis in generalized EC dysfunction. In experimental diabetes, endomucin restored the EC glycocalyx (221, 271), and in human diabetic patients partial restoration of the EC glycocalyx with sulodexide, an orally administered mixture of GAGs, not only lowers blood pressure, but also reduces albuminuria and other diabetic complications (272–276). Hence, microalbuminuria reflects the endothelial barrier defect that accompanies glycocalyx disruption and EC dysfunction in cardiovascular disease, diabetes and chronic kidney disease. The use of GAGs to enhance EC glycocalyx function could well develop into new therapeutic approach. It is important to note that the massive increase in cardiovascular morbidity of dialysis patients is, at least in part, due to chronic EC glycocalyx/ESL dysfunction (277, 278).

Preeclampsia affects 3–5% of all pregnant women and is associated with substantial risk to baby and mother. Albuminuria and hypertension are the earliest manifestations of preeclampsia. Marked glomerular EC swelling along with loss of glomerular EC fenestrae, also referred to as “glomerular endotheliosis” has long been recognized as the key glomerular abnormality in preeclampsia (279–281). EC abnormalities in preeclampsia, are not restricted to the glomerular endothelium, often involving the choroid plexus as well, and preeclampsia can progress to the full-blown thrombotic microangiopathy of pregnancy (282), the HELLP (hemolysis, elevated liver function tests, low platelets) syndrome. Even so, proteinuria is the main indicator of EC dysfunction in these patients. Glomerular EC differentiation and fenestration depend critically on podocyte-derived VEGF (283) and endotheliosis lesions are observed in mice with podocyte-specific VEGF haploinsufficiency (284). Also, bone morphogenetic protein-9 (BMP-9) signaling via the endothelial-specific ALK-1/endoglin receptor complex signals EC differentiation (285, 286). In patients with preeclampsia, placenta-derived, circulating soluble VEGF receptor and soluble endoglin inhibit VEGF- and BMP-9 signaling pathways leading to glomerular EC de-differentiation (3, 287, 288). Along with the ultrastructural EC changes, reduced EC glycocalyx/ESL height and shedding of glycocalyx components into the circulation have been documented in preeclampsia (171, 174, 202, 289). The use of VEGF inhibitors to reduce tumor angiogenesis (290) and macular degeneration (5) can evoke a similar syndrome of albuminuria, sometimes in the nephrotic range, and hypertension. It turns out that the human diacylglycerol kinase epsilon (DGKE) mutation (291), a cause of the hemolytic uremic syndrome, also reflects inhibition of VEGF signaling and consequent de-differentiation of glomerular EC (292). Hence, albuminuria in preeclampsia, and in patients treated with VEGF inhibitors, reflects EC de-differentiation resulting in a breach of the normal glomerular EC barrier to macromolecules.

Characterized by a vicious cycle of intracapillary thrombus formation, platelet consumption and microangiopathic hemolytic anemia, the thrombotic microangiopathies all involve EC activation (293), whether by Shiga toxin (294), COVID-19 (295, 296), thrombin, complement components (297) and/or inflammatory cytokines (296). Normally, endogenous inhibitors prevent activation of the coagulation and complement cascades at the EC surface and soluble EC-derived mediators like NO and prostacyclin block platelet activation. As part of the luminal EC glycocalyx, integral membrane-spanning thrombomodulin binds and inhibits thrombin, and stimulates protein C, which actively cleaves components of the coagulation cascade (298), by binding to the EC protein C receptor. Tissue factor pathway inhibitor (TFPI) (299) and complement factor H (CFH) (300) both bind HS GAGs in the EC glycocalyx/ESL, inhibiting local thrombin and complement activation, respectively. EC activation also results in reduced NO and prostacyclin production, de novo expression of membrane-anchored tissue factor, release of TFPI and CFH from the EC surface by heparinases (300), and mobilization of P-selectin and von Willebrand Factor (vWF) to capture platelets (301). It follows that even minor causes of EC activation, for instance a viral infection, can trigger run-away intravascular thrombosis in patients with genetic mutations or neutralizing antibodies to thrombomodulin (298), ADAMTS13 (297, 302), or to complement inhibitors (293). Loss of sialic acid EC glycocalyx by pneumococcal derived neuraminidase can also trigger the hemolytic uremic syndrome in children (233), as can reduced VEGF signaling due to loss of function mutations in diacylglycerol kinase (291, 292). While albuminuria is common in patients with these disorders, end-organ damage due to microvascular thrombosis are clinically more significant.

Trauma, cardiovascular surgery, septic shock (303) and more recently in critical illness due to COVID-19 (7, 295), all are associated with generalized EC activation and EC glycocalyx disruption. While proteinuria is common in critically ill patients (6, 8), pulmonary and brain EC barrier disruption tend to have greater relevance for outcomes and therapy in these patients. Endothelial cell activation by inflammatory mediators, among them TNF-α and Il1-β, results in shedding of EC glycocalyx components exposing cell-surface adhesion molecules that enable the initial capture and rolling of leukocytes on the endothelium and integrin-dependent leukocyte transmigration (304). EC glycocalyx disruption also promotes platelet adhesion and reduces the anti-coagulant and fibrinolytic activity of the EC surface (305). Even so, a recent metanalysis concluded that while EC glycocalyx shedding is common in critically ill patients, it does not distinguish between various causes and is not consistently associated with “vascular leak” (116). Similarly, albuminuria in this setting is a non-specific marker of EC glycocalyx dysfunction.