The Glomerular Basement Membrane

Abstract

The kidney’s glomerular filtration barrier consists of two cells—podocytes and endothelial cells—and the glomerular basement membrane (GBM), a specialized extracellular matrix that lies between them. Like all basement membranes, the GBM consists mainly of laminin, type IV collagen, nidogen, and heparan sulfate proteoglycan. However, the GBM is unusually thick and contains particular members of these general protein families, including laminin-521, collagen α3α4α5(IV), and agrin. Knockout studies in mice and genetic findings in humans shows that the laminin and type IV collagen components are particularly important for GBM structure and function, as laminin or collagen IV gene mutations cause filtration defects and renal disease of varying severities, depending on the nature of the mutations. These studies suggest that the GBM plays a crucial role in establishing and maintaining the glomerular filtration barrier.

Introduction

The glomerular basement membrane (GBM) is the extracellular matrix component of the selectively permeable glomerular filtration barrier (GFB) that separates the vasculature from the urinary space. The GBM lies between, and is initially synthesized by, the glomerular endothelial cells that line the glomerular capillaries and the podocytes (also called visceral epithelial cells) that sit on the opposite side of the GBM within the urinary space (Fig. 1). This three-layered structure—endothelium, GBM, and podocyte—facilitates the flow of plasma water and small solutes while restricting the flow of large plasma proteins such as albumin. The presence of high levels of albumin in the urine is considered to be indicative of a defect in at least one of the layers of the GFB.

Figure 1.

Ultrastructure of a typical glomerular capillary loop. A red blood cell (RBC) is present in the capillary lumen, which is lined by an endothelial cell with fenestrations (black arrowheads). The glomerular basement membrane (GBM) is a ribbon-like extracellular matrix that lies between the endothelium and the podocyte foot processes (FPs). The mesangium contains mesangial cells and their associated matrix. A parietal epithelial cell (PEC) is visible lining Bowman’s capsule.

The glomerular endothelial cells and podocytes are highly specialized cells. The endothelium bears fenestrations (transcellular “windows”) that presumably allow the plasma flowing through the capillaries to reach the GBM, although there is evidence that the fenestrations are plugged by a glycocalyx-like material that imparts barrier properties [1]. The podocytes send out a multitude of extensions called foot processes that interdigitate with those of adjacent podocytes and cover the outer aspect of the glomerular capillary (Fig. 1). The foot processes are connected by a unique cell-cell junction called the slit diaphragm that maintains defined spacing between the processes and at the same time permits the efficient flow of water and small solutes across the filtration barrier [2].

Because a major physiological property of the GFB is its ability to restrict the flow of plasma proteins such as albumin into the urinary space, there has been intense interest in defining the specific structure that imparts the GFB with this ability. Since the GFB was first visualized by electron microscopy in the 1950’s, the GBM and the slit diaphragm have been the subjects of intense debate among nephrologists, anatomists, and pathologists regarding which one constitutes the major barrier to albumin [3]. Although the debate has not been settled, what is clear is that defects in either the GBM or in the slit diaphragm can cause leakage of albumin into the urine (albuminuria) and the nephrotic syndrome, characterized by high albuminuria, low plasma albumin, high plasmid lipid, and edema. More recently the glomerular endothelium and its surface glycocalyx have entered into this debate, as it is clear that injury to the endothelium, as occurs in preeclampsia or as a side effect of certain drugs that affect the vascular endothelial growth factor signaling axis, can also cause proteinuria [4].

The focus of this review is the GBM. Like all basement membranes, the GBM is a sheet-like extracellular matrix composed of four major macromolecules: laminin, type IV collagen, nidogen, and heparan sulfate proteoglycan, the major one in the GBM being agrin (Fig. 2) [5]. Studies over the past two and a half decades have shown that 1) the GBM contains specific basement membrane protein isoforms; 2) some of these isoforms are crucial for glomerular development, morphology, and function; and 3) mutations in four of the nine known genes that encode the GBM’s components cause human kidney disease. The purpose of this review is to summarize the major findings regarding these aspects of the GBM, focusing on the specific basement membrane protein isoforms present within it.

Figure 2.

The major components of basement membranes: laminin, type IV collagen, nidogen, and heparan sulfate proteoglycan (agrin is shown due to its prevalence in the GBM, though perlecan is more widely found in basement membranes). Collagen IV is a triple helical protein with C-terminal noncollagenous domains (NC1) and N-terminal 7S domains; these are important in network formation. Laminin α, β, and γ chains assemble with each other via the laminin coiled-coil (LCC) domain. Laminin N-terminal (LN) domains are involved in polymerization of trimers, which initiates basement membrane formation. The C-terminal laminin globular (LG) domain contains binding sites for cell surface receptors. Agrin, a modular protein containing glycosaminoglycan (GAG) side chains, binds to the laminin long arm via the γ1 chain, whereas nidogen binds to the short arm of laminin γ1 as well as to collagen IV.

Laminin

Laminin is a ubiquitous basement membrane component that actually describes a family of several different isoforms. All laminins are secreted as αβγ heterotrimers, the structures of which are stabilized by limited interchain disulfide bonding. There are five α, four β, and three γ chains that assemble with each other nonrandomly to form at least 15 different heterotrimers. Laminin heterotrimers are named based upon the specific αβγ chain composition; for example, laminin α2β2γ1 is referred to as laminin-221 or LM-221 [6]. An excellent discussion of laminin chain diversity and evolution was recently published [7].

Laminin chains are evolutionarily related to each other and therefore share a number of structural features. The typical laminin heterotrimer is schematized as a cross-shaped structure in which the three chains wrap around each other via their laminin coiled-coil (LCC) domains to form the lower long arm of the cross (Fig. 2). Extending past this arm is a large (approximately 900 amino acid) laminin globular (LG) domain that is found at the COOH termini of all five α chains. The LG domains contain five tandem subdomains that bear binding sites for cell surface receptors such as integrins and dystroglycan. The three other arms of the cross, termed short arms, are formed by alternating globular and rod-like domains (Fig. 2). Of note, the globular laminin N-terminal (LN) domain is responsible for mediating the trimer-trimer interactions in the extracellular matrix that lead to formation of the laminin polymer and initiation of basement membrane assembly [8]. The rod-like domains are composed of laminin-type epidermal growth factor-like (LE) repeats whose secondary structure depends upon extensive disulfide bonding.

As far as the mature GBM is concerned, the major laminin is LM-521. However, during the processes of GBM formation and maturation that occur during glomerulogenesis, there are developmental transitions in laminin trimer deposition. These transitions can be briefly summarized as LM-111 to LM-511 to LM-521 [9-11].

The importance of these laminin transitions is revealed by the effects of mutations that prevent them from occurring. For example, a null mutation of laminin α5 (Lama5) in mice prevents the LM-111 to LM-511 transition and results in breakdown of the GBM and subsequent failure of glomerular vascularization [12]. GBM breakdown presumably occurs due to the lack of a sufficient concentration of polymerized laminin trimers, as the laminin network is required to maintain basement membrane integrity [13]. In support of this, a hypomorphic mutation in Lama5 caused by insertion of a neo cassette that impairs splicing of the laminin α5 mRNA results in reduced levels of laminin α5 in the GBM as well as in tubular basement membranes, but there appear to be sufficient levels of laminin α5-containing trimers to maintain GBM integrity. However, the reduction from the normal laminin level causes glomerular proteinuria and hematuria and, unexpectedly, polycystic kidneys and renal failure within a month [14]. Although this provided long sought support for the hypothesis that impaired tubular epithelial cell-matrix interactions might be involved in cystogenesis, subsequent studies showed the cystic phenotype to be due solely to the defective GBM [14].

Mutation of laminin β2 (Lamb2) either in mice or humans results in a congenital nephrotic syndrome with variable ocular and neurological manifestations, which in humans is called Pierson syndrome [15,16]. Our data suggests that the reduced LM-521 results in ectopic deposition of other laminin trimers that cannot make a properly permselective GBM, leading to leakage of albumin across the glomerular filtration barrier and nephrotic syndrome [17]. We recently tested the hypothesis that it is not the lack of laminin β2 per se but rather the lack of a sufficient concentration of laminin α5-containing trimers that causes nephrotic syndrome. Indeed, by overexpressing laminin β1 in podocytes on the Lamb2−/− background we could successfully prevent the nephrotic syndrome, and this correlated with high level deposition of LM-511 in the GBM [18].

Collagen IV

Like most collagens, type IV collagen is a trimeric extracellular matrix protein consisting of α chains that are rich in Gly-X-Y amino acid triplet repeats. Three α chains wind around one another to form the collagen triple helix (Fig. 2); the Gly at every third position is necessary because it is the only amino acid with a side chain small enough to fit at the center of the helix. But unlike most other collagen types, type IV collagens have interruptions of the Gly-X-Y repeats, presumably to make a more flexible trimer and a more flexible network that imparts basement membranes with the flexibility that is likely important for their function.

The collagen IV family consists of six genetically distinct α chains that trimerize with each other in specific stoichiometries to make three different types of network-forming building blocks called protomers. These include the (α1)₂α2, α3α4α5, and (α5)₂α6 protomers. Protomers are secreted into the extracellular matrix, where they self-polymerize via their NH2- and COOH-terminal domains to make a network within the plane of the basement membrane that becomes crosslinked due to the activity of several secreted enzymes. Although basement membranes can form in the absence of type IV collagen, the collagen IV network is crucial for basement membrane stability [19].

Like GBM laminin developmental transitions, there are also type IV collagen transitions during glomerulogenesis. Initially the nascent GBM contains the α1/α2 network, but as the glomerular capillaries begin to form and function the podocytes (but not the endothelial cells) begin to secrete α3α4α5 trimers. These then polymerize to form what will become the mature GBM collagen IV network, and the α1/α2 network becomes a minor component [11,20].

Mutations that affect the genes encoding any one of the collagen IV α3, α4, or α5 chains can cause defects in the GBM. The defects can be mild, as in thin basement membrane disease, or severe, as in Alport syndrome. Thin basement membrane disease, also called benign familial hematuria, shows autosomal dominant inheritance and has been found in 40 to 50% of patients to result from heterozygous null mutations in COL4A3 or COL4A4, which encode the α3 and α4 chains of type IV collagen, respectively. As the alternate disease names imply, affected individuals exhibit thinning of the GBM and blood in the urine but do not usually progress to overt renal disease requiring treatment. However, these same COL4A3 and COL4A4 mutations in the homozygous state cause autosomal recessive Alport syndrome, a basement membrane disease leading to eventual kidney failure that is associated with deafness and ocular abnormalities. The most common version of Alport syndrome is the X-linked form, which is caused by mutations in COL4A5. Both the autosomal and X-linked forms of Alport syndrome share the same glomerular histopathology; there is both thinning of the GBM and variable segmental thickening of the GBM that by electron microscopy imparts a basket weave appearance rather than the typical ribbon-like morphology (Fig. 3).

Figure 3.

GBM lesions in a mouse model of Alport syndrome (hereditary nephritis). The mouse is homozygous for a null mutation in Col4a3 and therefore lacks the collagen α3α4α5(IV) network. Note the outpocketings of the GBM (asterisks) that project toward the podocyte, which has lost much of the normal foot process architecture.

Why do these COL4 mutations cause these different diseases? Logically, because the great majority of GBM protomers are obligate α3α4α5(IV) heterotrimers, reduced levels or any one of these three chains (as in thin basement membrane disease) could reduce the quantity of protomer secreted, and this could theoretically impact the density of the collagen IV network and therefore the architecture of the GBM. On the other hand, complete absence of α3α4α5(IV) network due to the total lack of one of these chains, as in Alport syndrome, is expected to cause more severe GBM abnormalities. In the absence of this network, there seems to be increased accumulation of the (α1)₂α2 network as an attempt to compensate, but the severe ultrastructural abnormalities and eventual progressive scarring of glomeruli that impairs kidney function indicates that compensation is not complete. It has been proposed that the α3α4α5(IV) network is more resistant to proteases and can be more highly crosslinked compared to the (α1)₂α2 network, resulting in superior stability that can maintain GBM architecture.

As logical as these zygosity-phenotype correlations seem to be for Alport syndrome and thin basement membrane disease, a less clear issue is why mutations in COL4A1 can cause kidney disease and systemic extrarenal manifestations, including brain structural defects. Hereditary angiopathy with nephropathy, aneurysms, and muscle cramps (HANAC) syndrome is a recently described disease entity associated with hematuria and bilateral large cysts [21]. The syndrome is caused by heterozygous glycine substitutions in a particular region of COL4A1, the CB3[IV] domain [21,22]. This domain is particularly important for integrin binding to collagen IV [23], so the mutations may impact the avidity of cell/matrix interactions at the affected sites. Heterozygous mutations in other regions of COL4A1 have been shown to cause stroke, porencephaly, and small vessel disease in both humans and mice [24], and kidney basement membrane defects have been observed in mutant mice [25,26]. Additional studies are necessary to determine the mechanism of disease pathogenesis in these patients and in the corresponding mutant mice.

Nidogen

There are two nidogen proteins, nidogen-1 and nidogen-2 (previously also known as entactin-1 and entactin-2). Both are dumbbell-shaped (Fig. 2), virtually ubiquitous basement membrane proteins. Because of its earlier discovery, much more is known about the biochemistry and function of nidogen-1. Nidogen-1 binds both laminin and type IV collagen, and because of this it was originally proposed to link the separate laminin and collagen IV networks [27] and to therefore be crucial for basement membrane formation. However, the results from generating nidogen-1 and nidogen-2 mutant mice did not fully agree with this assessment. The individual mutants are actually viable and fertile with apparently normal basement membranes [28,29], which was somewhat surprising until one considers that the two proteins are at least partially redundant in their abilities to bind both laminin and type IV collagen [30]. Consistent with this, deletion of both nidogen genes results in perinatal lethality [31]. However, another surprise was that basement membranes could form in the absence of both nidogens, and most organogenetic programs proceeded normally. The major exceptions were defects in the late stages of lung development and in maintenance/integrity of cardiac muscle that were associated with basement membrane alterations. Although most kidneys were fully formed, there was occasional renal dysgenesis or agenesis, or hydronephrosis or dilation of the ureteric bud derivatives, but the GBM had a normal appearance [31].

Agrin

Agrin is the major heparan sulfate proteoglycan of the GBM [32]. There are multiple splice forms of agrin, one of which is present in the basement membrane of the synaptic cleft at the neuromuscular junction, where it has been shown to be critical for localization and differentiation of neuromuscular synapses [33]. In addition, although the secreted form of agrin is present in the GBM, some neuronal cells express a transmembrane form of agrin through the use of an alternative promoter [34]. This transmembrane form seems to be important for inducing the formation of neuronal processes [35] but is not involved in organizing neuromuscular junctions [34].

As a heparan sulfate proteoglycan, agrin has a high net negative charge due to the presence of sulfated glycosaminoglycan (GAG) side chains (Fig. 2). Because all basement membranes, and the GBM in particular, exhibit a net negative charge, heparan sulfate proteoglycans such as perlecan and agrin are presumed to be important contributors towards this negative charge [36]. It has long been accepted dogma in the field of renal physiology that the net negative charge of the GBM is a crucial component of the glomerular capillary wall’s filtration barrier to plasma albumin, which is also negatively charged and should therefore be repelled by the GBM. Indeed, classic studies showed that for tracer molecules presumed to be of similar size, positively charged ones cross the filtration barrier more readily than neutral ones, which cross more readily than negatively charged ones [37]. However, the concept of charge selectivity has recently been called into question by different types of studies. For example, when we removed agrin from the GBM by podocyte-selective knockout of Agrn, we could detect a dramatic reduction in GBM anionic charge (Fig. 4) yet no effect on the glomerular filtration barrier either to albumin or to a negatively charged tracer [38]. And although perlecan is at most a minor component of the mature GBM, we went on to show that removal of both heparan sulfate-linked perlecan and agrin from the GBM was also compatible with a normal filtration barrier [39]. Furthermore, reduction of GBM anionic charge in vivo by infusion of heparanase, which removes heparan sulfate side chains from proteoglycans, does not lead to proteinuria [40]. In a different type of approach, Rippe and colleagues used charged and neutral fluorescein isothiocyanate (FITC)-labeled Ficolls of varying sizes as tracers in anesthetized rats and found that there was a noticeable charge effect (inhibition of diffusion) for molecules of 20-35Å, but not for larger molecules of 35 to 80Å [41]. They concluded that GBM charge plays a minor role, primarily for “small pores”, in imparting the glomerular filter with charge selectivity and suggested that the endothelial glycocalyx may be more important [42], but that overall the charge selective properties of the glomerular filter are weaker than previously envisioned [41]. Recently, an exciting new model to explain the charge-selective nature of the glomerular filtration barrier was presented; in this model, the charge barrier is formed by glomerular filtration itself, and albumin is essentially “electrophoresed” away from the urinary space and towards the capillary lumen [43].

Figure 4.

Detection of a GBM charge defect in a podocyte-specific agrin mutant mouse using the cationic probe polyethyleneimine to detect anionic sites. In the control, discrete subepithelial anionic sites are revealed (A), but these are significantly reduced in the mutant (B). Despite this reduction in anionic charge, mutants displayed no detectable filtration barrier defects.

Conclusions

Studies of the glomerular basement membrane by anatomists, biochemists, geneticists, and cell and developmental biologists over the past few decades have led to a basic understanding of its composition, structure, genesis, and function. Although there is still much to be learned about GBM and what it really does, the ability to manipulate its composition or dynamics in the future could have important implications for the treatment of glomerular disease.