The glomerular capillary wall, also described in the literature as the filtration unit of the glomerulus, fractionates the blood into an ultrafiltrate that is devoid of plasma proteins and cellular elements that are retained within the capillary lumina.¹¹ The ultrafiltration unit is a stratified structure made up of GBM flanked on the inner and outer aspects by a highly attenuated fenestrated endothelium and interdigitating foot processes of the podocytes, respectively (Figure 1B). The endothelial fenestrae are quite large (~100 nm in diameter) and are devoid of diaphragm such that a bulk flow of plasma water solute can be channeled toward the GBM. On the other hand, the spaces between the interdigitating foot processes are spanned by a thin membrane, known as the slit diaphragm, that has a well-defined structure with 4 × 14 nm rectangular pores,¹² and it apparently also restricts the transcapillary passage of macromolecules. Conceivably, the GBM, an amorphous extracellular matrix scaffold of 300-nm width, serves as the major barrier for the passage of large macromolecules and is further stratified into a central dense layer known as lamina densa that is flanked on either side by relatively loose electron lucent layers termed as lamina rara interna and externa. The GBM is made up of high-molecular weight proteins, including type IV collagen, laminins, entactin/nidogen, and sulfated PGs.^13,14,15,16 The latter include, agrin, perlecan, and collagen XVIII, and all these PGs are chimeric molecules with a wide variety of potential functional domains.^14,15,16 In addition, glycosaminoglycans, made up of heparan (HS) and chondroitin sulfate chains, are attached to their respective core peptides of the PGs. These chains by virtue of having sulfate radicals endow the GBM with charge-selective properties besides size-selectivity as with laminin or type-IV collagen, concentrated in the lamina densa. On the other hand, the electronegative charge is mainly clustered in the lamina rarae, where its localization has been elucidated by various cationic probes, including polyethylenimine, Alcian blue, ruthenium red, and cationic ferritin (Figure 1C).^8,11 At present, it is unclear as to the amount of electrical charge contributed individually by each of these PGs, besides by other acidic proteins of the GBM.

Figure 1 Electron micrograph of the ultrafiltration unit. This unit (B) is made up of GBM, which is stratified into lamina densa (LD) and lamina rara interna (LRI) and externa (LRE). The GBM is lined by the fenestrated (fn) endothelium (En) from inside and with (more ...)

The charge selectivity is also contributed by the sialic acid-rich glycocalyx coating the plasmalemma of the attenuated glomerular endothelium and podocyte foot processes (Figure 1A). Other plasmalemmal glycoproteins of the endothelium include podoendin, podocalyxin, and syndecans, the latter being cell surface-associated sulfated PGs, which may also contribute to the electrical charge of the capillary. The biochemical composition of the slit diaphragm, a prototype of shallow adherens junctions, is quite complex and has been extensively investigated in recent years. Its major components include nephrin, NEPH1, NEPH2, and podocin, in addition to junctional proteins like cadherins.¹⁷ The latter interact with actin cytoskeleton via linker proteins, such as ZO-1, CD2AP, CASK, and MAGI-2. Other plasmalemmal proteins that mediate interactions with the actin cytoskeleton include integrin α6β1 and α and β subunits of dystroglycan along with sarcospan and sarcoglycans. This entire complex of plasmalemmal proteins interacts with E3 and E8 domains of laminin-1 and anchors the sole of the foot processes with the GBM and thus conceivably modulates the hydraulic flux across the filtration barrier.¹⁷ Interestingly, adhesion of foot processes is also mediated by sialic acid-rich glycocalyx because its enzymatic elimination leads to the detachment of the podocytes from the GBM, thereby resulting in the disruption of the hydraulic flux and consequential proteinuria.¹¹

The seminal studies of the 1960s by Farquhar and Palade addressed this issue by using electron-dense tracer ferritin with biophysical properties similar to albumin.^2,3 The ferritin introduced into the circulation was found restricted to the endothelial fenestrae and inner layers of the GBM.^2,3 Similar localization was observed in in situ experiments where perfusate containing ferritin was directly introduced into the renal artery (Figure 2A), indicating the GBM as the primary filtration barrier. Subsequent work by Karnovsky, Venkatachalam, and colleagues^4,5 with peroxidatic tracers having varying isoelectric points and molecular weight revealed that some of them accumulated underneath the slit diaphragms following their introduction into the blood circulation. This meant that perhaps the slit diaphragm may be the ultimate filtration barrier. To resolve this issue, Caulfield and Farquhar used graded series of dextrans for further in vivo tracer studies in the early 1970s. The dextrans less than the molecular weight of albumin appeared in the urinary spaces, whereas the larger ones were retained within the capillary lumina. At no time point were the dextrans observed within the basement membrane; thus, the pendulum swung back to the notion that GBM is the primary filtration barrier. In support of this notion is also a recent elegant gene disruption study with laminin β2-deficient mice that develop proteinuria during the neonatal period associated with disorganization of the GBM components, especially the anionic sites.¹⁹

Figure 2 Electron micrographs of the capillary loops of kidneys perfused with native (pI ~4.9, A and B) and cationic (pI >7.5, C and D) ferritins. B and D: Kidneys treated with heparitinase and then perfused with ferritins. Increased permeation (more ...)

Because the dextrans are also well suited for in vivo physiological studies, Brenner and his colleagues⁷ measured renal clearances of dextrans of varying size and charge in the mid-1970s. The dextrans with similar molecular weight but with basic charge had higher clearance rate, establishing that the glomerular capillary behaves as the charge- as well as size-selective barrier. The charge selectivity was later found to be attributed to the GBM anionic sites enriched with heparan sulfate proteoglycans, because their in situ enzymatic removal led to an increased permeation of the native ferritin (pI ~4.9) within GBM (Figure 2B).⁹ Moreover, cationic ferritin (pI >7.5), which normally binds to the anionic sites in the lamina rarae (Figure 2C),^9,11 formed a gradient within the basement membrane on the PGs’ enzymatic degradation, with relatively high concentration of the tracer in inner layers of the basement membrane (Figure 2D), thus reinforcing the notion of the GBM being the primary filtration barrier with respect to its charge selectivity. Intriguingly enough, the cationic ferritin given intravenously was localized underneath the slit diaphragm within 1 to 2 hours (Figure 3A) and could be seen for up to 24 hours (Figure 3B). These findings meant that all of the strata of the capillary wall serve as charge- as well as size-selective barrier to varying degrees during the various stages of transglomerular passage of macromolecules, with the slit diaphragm acting as the final fence before the ultrafiltrate is formed in the urinary space. Conceivably, the disruption of the slit diaphragm would be expected to induce proteinuria, and that seems to be the case with nephrin-deficient mice and in patients with congenital nephrotic syndrome of Finnish type, in which mutations in the nephrin gene have been observed.²⁰ Such results can be reciprocated by the administration of nephrin monoclonal antibody to rats, which as a result develop a proteinuric state.²⁰ Thus, it seems that the biology of the slit diaphragm and/or of the podocytes are intricately linked to the pathogenesis of proteinuria as is further evidenced by the phenotypic changes seen in the Cd2ap-deficient mouse with widespread foot process effacement and albuminuria.¹⁹

Figure 3 Electron micrographs of glomerular capillary loops of a rat that received an intravenous injection of cationic ferritin (pI >7.5) after 2 (A) and 24 hours (B). Initially, the cationic ferritin is seen traversing the GBM and accumulating underneath (more ...)

As indicated earlier, there are three major extracellular matrix-sulfated PGs, collagen XVIII, perlecan, and agrin, that have variable degree of expression within the GBM and mesangial matrix of the mammalian glomerulus.^14,15,16 These PGs, besides being chimeric molecules, include both the HS and chondroitin sulfate glycosaminoglycan chains that impart variable degree of charge density to the GBM individually. The removal of PGs’ chains with use of broad-spectrum glycosaminoglycan-degrading enzymes leads to an expected facilitated permeation of ferritin within the GBM (vide supra) and also increased fractional clearances (θ) of fluorescein isothiocyanate-Ficoll and albumin.²¹ Likewise, intravenous administration of a nonspecific monoclonal antibody to HS chains has been shown to induce selective proteinuria in rats.²² The fact that there is a generalized reduced expression of HS chains involving various pathogenetic mechanisms, eg, depolymerization and damage by reactive oxygen species, in a wide variety of renal glomerular diseases leads one to ponder which of the three PGs are lost in a given disease process and whether or not the disruption of an individual PG leads to an altered permeability of the glomerular capillary wall.²³ These issues were addressed by recent gene disruption studies in mice. The collagen XVIII-deficient mice revealed expanded basement membrane matrices and elevation of serum creatinine, suggesting altered GBM composition conceivably due to the defective interactions with other glycoproteins.²⁴ Unfortunately, permeability studies related to the charge-selective properties were not performed. The knockin perlecan mice Hsph^2Δ3/Δ3 lack HS chains but did not reveal any obvious ultrastructural abnormalities and had significantly increased urinary protein excretion following a protein overload.²⁵ The anionic sites were evaluated only in the lamina rara externa; thus, a detailed view of the renal phenotype in terms of charge characteristics of the glomerular capillary is lacking in this mouse model as well. Likewise, the current study by Harvey et al,¹⁰ although it may not be comprehensive, is noteworthy because it at least documents the absence of proteinuria in agrin-deficient mice, suggesting that the gene disruption of one class of sulfated PGs is not sufficient enough to alter the permeability of glomerular capillary wall. This may be because agrin is not highly glycosylated/sulfated to impart sufficient electronegativity, and moreover, the subendothelial anionic sites remained intact in neonatal agrin-deficient mice, suggesting a normal maturation of the GBM as far as the contribution of the endothelium is concerned. Another factor that should be borne in mind is that the disruption of one class of PGs may lead to minor increases in the fractional clearance of macromolecules that are insufficient to overcome tubular maximum. As a result, the proteins cannot be detected in the urine, which means the warranting of the essential tracer studies using electron microscopy as pioneered by Farquhar and Palade^2,3 decades ago.

In reconciliation with past tracer permeability experiments and current understanding of pathogenesis of proteinuria from knockout and knockin mice, it seems obligatory to conclude that the integrated functions of all strata of the glomerular capillary wall are essential to maintain its permeability characteristics. With the disruption of any component, either of slit diaphragm or GBM, one would anticipate a compromise in the barrier functions of the capillary wall. Moreover, a structural change in a given component at times induces conformational change in other components of the capillary wall. This is highlighted by the experiments in which reactive oxygen species-mediated disruption of the α-dystroglycan/matrix transmembrane complex led to a disordered organization of the fibrillar components of lamina rara externa as well as foot process effacement, which would conceivably result in detachment of the podocytes and ensuing proteinuria.²⁶ Such may be the case since neuraminidase-induced detachment in situ leads to increased permeability of ferritin (Figure 3B).²⁷ This notion is further supported by the fact that its intravenous administration results in an increased urinary albumin excretion.²⁸ The reactive oxygen species that are generated in a wide variety of renal disease states do induce damage to the GBM, including the PGs, besides the structural perturbations in the slit diaphragm and podocytes.^29,30 All in all, it means that all components of the capillary wall operate in synchrony to maintain normal homeostasis and integrated functions of the renal glomerulus.