Glomerular Filtration Barrier

The glomerular filtration barrier is structurally composed of three layers, the capillary endothelial cells, the glomerular basement membrane (GBM) and the podocyte filtration slit membrane ( Fig. 73.2 ). The barrier is freely permeable to water, solutes and small molecules however, increasing size of macromolecules causes increasing restriction to filtration as do negative charge.

Electron micrograph of the glomerular filtration barrier. Filtration takes place from the glomerular capillaries (Cap) through the pores (arrowheads) of the endothelial cells (Endo), the glomerular basement membrane (GBM) consisting of the three layers, lamina rara interna (LRI), lamina densa (LD) and lamina rara externa (LRE) and finally through the filtration slit membrane (arrows) between the foot processes (Fp) of the podocytes (Pod) into the urinary space (Us).

The fenestrated endothelium ( Fig. 73.2 ) is unusual since the fenestrae generally are not closed by diaphragms, except as demonstrated in rat where the capillaries which are direct tributaries to the efferent arteriole do indeed have diaphragms closing the fenestrae. The pores in the endothelium appear not to be fully open holes. By special fixation procedures it has been demonstrated that the pores are filled with glycoproteins forming “sieve plugs” probably contributing to the endothelial part of the filtration barrier. The endothelial cells have a thick glycocalyx and an even thicker endothelial cell surface coat, which are believed by many authors to contribute significantly to the charge selectivity of the barrier (for a recent review see Haraldsson et al. ).

The basement membrane, which in man is about 300 nm thick, consists of three layers, a lamina densa, located between a lamina rara interna facing the endothelial cell and a lamina rara externa, facing the podocyte ( Fig. 73.2 ). In the 1970s the GBM was considered the major contributor to the charge selectivity of the filtration barrier (see also Kanwar et al. for references). In vitro studies, however, on isolated GBM showed no charge selectivity and removal of charged components of the GBM in mouse knock out studies in general did not change charge selectivity, for discussions see. A large variety of both genetic and acquired, albuminuric diseases affect the GBM, e.g., Alport syndrome and diabetes mellitus (see also later).

The third component of the barrier, the podocyte filtration slit membrane ( Fig. 73.2 ), has attracted great interest as a key part of the filtration barrier, especially since the findings that, for example, gene defects of nephrin ( NPHS1 ) induce the congenital nephrotic syndrome of the Finnish type and gene defects of podocin (NPHS2) induce nephrotic syndrome of the non-Finnish type. The porous structure of the filtration slit membrane was first described by Rodewald and Karnowsky measuring the dimensions of the pores to be 4×14 nm. In a recent publication the mean radius of the observed irregular circular or elipsoid pores were measured to be 12.1 nm. In proteinuric rats additional large pores were observed and suggested to contribute to the increased filtration of protein/albumin in pathologic conditions, a finding which awaits confirmation.

An elegant model for the charge restriction has recently been put forward, based on micropuncture experiments in glomeruli of Necturus maculosus . The authors identified a filtration dependant negative electrical charge in the Bowman’s space compared to the capillary lumen, a charge that would allow negatively charged proteins like albumin to be electrophoresed back to the blood and the opposite for positively charged proteins.

There is no doubt that all three structural components of the glomerular filter are necessary for maintaining the barrier, illustrated by the observations that damage to any part eventually leads to albuminuria, that the GBM is synthesized from both the epithelial- and endothelial cells and that vascular endothelial growth factor (VEGF)-A produced by the podocytes influences development and maintenance of the endothelial cells which possesses receptors for VEGF-A, VEGFR-1 and 2.

Glomerular Filtration of Albumin

The amount of albumin normally filtered in the glomeruli has been estimated using various techniques, including micropuncture of rats and dogs, estimating the concentration of albumin in the ultrafiltrate between 1 and 50 µg/ml. This corresponds to a filtered load of albumin between 170 mg and 9 g per 24 h in normal humans. Inhibition of tubular albumin uptake in humans by lysine suggested filtration of at least 281 µg/min, corresponding to ~400 mg/24 h. Similar studies in lysine treated rats resulted in the excretion of 2.5 mg to 25 mg/24 h corresponding to 0.7–7 g/24 h in humans. In rat the filtration fraction of albumin was estimated to 0.0006 by micropuncture studies in good agreement with the results mentioned above. This figure was, however, challenged by Comper and colleagues who estimated the filtration fraction by two-photon microscopy to be 0.034. As calculated by Gekle this implies a filtration of 225 g/24 h of albumin in humans and the results were immediately questioned by several groups. Subsequently, three studies have seriously questioned the technical approach applied by Russo et al. Thus, the notion of normal glomerular filtration of such large amounts of albumin remains highly controversial. For an excellent review comparing glomerular permselectivity of ficoll, dextran and globular proteins, see Venturoli and Rippe.

Albumin Uptake in Glomerular Cells

Albumin uptake has been demonstrated in vivo in podocytes from human, rat and mouse and in vitro in mouse and human podocytes. Accumulation of endocytosed protein in podocytes is also indicated by podocyte vacuolization in proteinuric patients and experimentally, endocytic uptake of tracer proteins in podocytes have been demonstrated in vivo . The albumin binding receptor megalin (see below) has been identified in rat podocytes and very recently also on human podocytes providing a mechanism for the endocytic uptake of albumin and other proteins. It has been proposed that unless removed, filtered proteins would clog the glomerular filter due to the podocyte slit membrane. Such a theoretical clogging of the slit diaphragm may be attenuated by megalin mediated, podocyte endocytosis of trapped protein, including albumin. It should be emphasized that the endocytic uptake of albumin in the podocytes is minimal compared to the subsequent uptake in the proximal tubule (see below).

Tubular Albumin Uptake

Renal Expression, Structure and Sorting of Megalin, Cubilin and AMN

Megalin is heavily expressed in the renal proximal tubule brush border, the endocytic compartments, and membrane recycling system. Megalin is also detectable in lysosomes in small quantities, but after endocytosis the majority of megalin is recycled to the apical membrane via dense apical tubules. Cubilin co-localizes with megalin in the renal proximal tubule and AMN co-localizes with cubilin. In rats and humans additional expression of megalin in the podocytes has been identified. For more information of the expression of the receptors in other epithelia see.

Megalin ( Fig. 73.3 ) was identified nearly 30 years ago as a large glycosylated protein (600 kDa) belonging to the low density lipoprotein (LDL) receptor family. Megalin is a multiligand receptor with four binding clusters in the extracellular domain. The binding clusters each contain 7 to 11 complement-type repeats. Structural studies of the LDL receptor and the twelfth repeat in megalin have revealed that each consists of approximately 40 amino acids with 6 cysteines forming 3 disulfide bridges and a C-terminal calcium cage. Separation of the binding clusters is obtained by 16 growth factor repeats separated by 8 YWTD spacer regions, which are involved in pH dependent release of ligands. Finally the extracellular domain contains one epidermal growth factor-like repeat which is situated next to the membrane spanning area. The four binding clusters are responsible for the binding of albumin; one of the more than 50 ligands which so far has been reported for megalin.

Schematic illustration of megalin, cubilin and amnionless (AMN). The three receptors co-operate in the proximal tubular uptake of filtered albumin. Cubilin binds albumin, but is dependent on complex formation with megalin and/or AMN to sustain endocytosis. Megalin and AMN contain NPXY motifs necessary for recruitment of the endocytic machinery and initiation of endocytosis. AMN is furthermore involved in translocation of cubilin from the RER to the plasma membrane.

Figure modified from.

The receptor spans the membrane once and is equipped with a short intracellular tail (209 amino acids) which contains 2 endocytic motifs (NPXY) driving clustering into coated pits and a NPXY-like motif (NQNY) involved in apical sorting of the receptor. Apical sorting is also dependent on the receptor associated protein (RAP), acting as a chaperone and phosphorylation of a PPSP motif in the cytoplasmic tail.

Cubilin ( Fig. 73.3 ) is identical to the intestinal intrinsic factor-B12 receptor and is a 460 kDa, peripheral membrane protein. Its structure is dominated by 27 ekstracellular CUB domains (complement c1r/C1s, Uegf (epidermal growth factor-related sea urchin protein)) and bone morphogenic protein 1. The CUB domains are involved in ligand binding, but despite the numerous domains only around 15 ligands have been identified including albumin. The N-terminus cubilin contains 8 epidermal growth factor (EGF) repeats and a 110 amino acid stretch. No transmembrane domain has been identified and sorting and anchorage of cubilin to the membrane is accomplished through physical interaction with AMN and megalin. In vitro studies showed that sorting relies on interaction of the EGF domains in cubilin with AMN. This cooperation between cubilin and AMN is supported by intracellular retention of cubilin in dogs suffering from AMN mutations causing Imerslund-Gräsbeck’s syndrome and in AMN deficient mice. The apical plasma membrane localization of AMN is in the same manner dependent on intact cubilin as AMN is retained intracellularly both in cubilin deficient mice and in humans with no cubilin expression.

Cubilin also associates with megalin through CUB domain 12–17 and 22–27 as well as through the N-terminus and megalin appears to be important for the stability of cubilin in the membrane as suggested by decreased levels of cubilin in in vitro studies and in rabbits producing megalin antibodies as well as by immunocytochemistry in megalin knockout mice.

AMN ( Fig. 73.3 ) has a molecular weight ranging from 38–50 kDa and was initially recognized as being involved in normal development of the middle portion of the primitive streak in mice. Mice which lack AMN fail to develop amnion. It has a cytoplasmic tail with the endocytic motif NPXY, a transmembrane domain and N-terminal extracellular domain encompassing approximately 70 amino acids with a cysteine rich region as the only characteristic domain. It should be noted that the above described interdependence of the receptors for normal sorting complicates the endocytic analysis of each receptor separately.

Endocytosis of Albumin Receptors

Endocytosis of megalin is driven by endocytic motifs in the cytoplasmic tail (see above) that mediate the assembly of clathrin and adaptor proteins such as AP-2, Dab2 and ARH. The significance of Dab2 to the endocytosis of megalin is supported by mutual dependence of megalin and Dab2 for normal localization and the mild, but significant proteinuria observed in Dab2 knockout mice. Dab2 further associates megalin with nonmuscle myosin heavy chain IIA and actin, whereas the adaptor protein GAIP interacting protein C-terminus (GIPC) attaches myosin VI to the complex. These interactions are suggested to be important for the post endocytic transport of the complex corroborated by the albuminuria observed in GIPC- and myosin VI knockout mice.

Interaction of megalin and cubilin together with their strict co-localization in the proximal tubule suggests that megalin mediates the internalization of cubilin and its ligands. Several in vitro studies strengthened this concept such as decreased cubilin endocytosis in megalin compromised cells. AMN holds a NPXY motif and it is apparent that in cell cultures lacking megalin, cubilin can work in conjunction with AMN. In vivo in the kidney, however, this co-operation is not able to sustain normal endocytosis. The concept of megalin being the motor for albumin endocytosis is supported by a recent study showing that an SNP in the megalin binding area of cubilin is associated with microalbuminuria in the normal population and in diabetics. It has been shown recently, by analysis of the crystal structure of cubilin, that changes in a residue that does not directly interact with intrinsic factor B12 result in Imerslund-Gräsbeck’s disease. Thus, a SNP in the megalin binding region might similarily interfere with cubilin association to megalin hindering endocytosis of cubilin-bound albumin.

Regulation of Albumin Receptors

It has generally been assumed that under normal physiological conditions proximal tubule endocytic uptake of albumin is a constitutive process determined only by the receptor expression. Little is known, however, about the regulation of receptor expression in normal physiology. Studies in disease and transgene mice have suggested, that megalin expression may be under influence of the renin-angiotensin system. Angiotensin II decreases megalin expression in a long term proteinuric mouse model induced by renin overexpression. Furthermore, angiotensin II has recently been shown to stimulate trafficking of several sodium transporters and megalin into microvilli. However, in this latter study no apparent decrease in megalin levels was observed. An intrarenal proximal tubular renin-angiotensin system has been suggested and the molecular actions of this system are complex. The role of these components in the direct or indirect regulation of megalin is at present very difficult to deduce. Furthermore, an additional link between angiotensin II and megalin expression has evolved. Transforming growth factor (TGF)-β reduces megalin expression in vitro and in a proteinuric hypertensive mouse model the action of TGF-β has been shown to be modulated by angiotensin II receptor antagonists at doses that do not decrease aterial blood pressure. This is believed to be mediated by reestablishment of the equilibrium between TGF-β and bone morphogenic protein 7 (BMP7)/BMP7-inhibitors. Thus, treatment with angiotensin II antagonists might elevate megalin expression through a decrease in BMP7-inhibitors resulting in BMP7 mediated inhibiton of TGF-β. Obviously the effect of angiotensin II on glomerular function in addition to the proposed effect on tubular megalin expression may contribute to the changes in urinary albumin excretion observed with modulation of this system. In a recently published study it is shown that megalin contains 3 PPAR (peroxisome proliferator-activated receptors) response elements in its promotor region and that megalin expression increases in response to PPAR ligands in mouse kidneys suggesting a direct regulation of megalin expression by PPAR. The physiological significance of this regulation remains to be established, however, PPAR agonists have in a number of studies been shown to be renoprotective in diabetic patients including an amelioration of albuminuria.

Lysosomal Degradation of Albumin

Following proximal tubule endocytic uptake, albumin accumulates within the lysosomal compartment where it is degraded. In rats, albumin is reabsorbed within all three segments of the proximal tubule, however, catabolism appears to be slower in the S3 segment. This process is dependent on ATP-mediated acidification of the lysosomal compartment. Interference with lysosomal function from increased filtration of albumin has been suggested by experimental induction of albuminuria. This resulted in functional and structural changes in lysosomes and increased activity of lysosomal enzymes such as cathepsin suggesting that albumin overload changes the protein catabolic function of the proximal tubule cell.

It is generally believed that albumin is fully degraded within the lysosomal compartment and that amino acids are returned to the circulation through a basolateral exit mechanism. Based on the identification of albumin fragments in normal human and animal urine it has been proposed that albumin fragments resulting from the proximal tubule degradation of proteins are normally excreted at the luminal cell surface. Most of these fragments are suggested to be very small with MW <500 Da and apparently undetectable by conventional immune or dye-based assays, however, may be detected either by size exclusion chromatography of injected and processed labeled albumin or by the Biuret protein assay. Although cells grown in vitro may secrete fragments so far no direct evidence for the tubular generation and luminal secretion of albumin fragments in vivo has been published. In vitro microperfusion of proximal tubules suggested peritubular, but not luminal release of degradation products and knockout of megalin and cubilin in mice did not change the excretion of urinary albumin fragments showing that receptor mediated endocytosis is not involved in the generation of these. Furthermore, other studies have not confirmed the presence of significant amount of albumin fragments in the urine. Thus, the amount, origin and pathophysiologic implication of potential urinary albumin fragments remain to be established.

Transtubular Recovery of Albumin

In addition to the well established endocytic pathway for albumin uptake, the existence of a high capacity pathway for the transtubular transport of very large amounts of intact albumin has been suggested. Such a pathway would be essential for the recovery of ~225 g/24 h of albumin suggested to be filtered in the human kidney if, as proposed by Comper and colleagues, the filtration fraction for albumin is much higher than previously believed.

The transtubular recovery of intact albumin was proposed after the identification of a small, second peak of labeled albumin appearing in the renal vein following bolus injection of labeled albumin into the isolated perfused kidney. Later studies by electron microscopy of endogenous and fluorescent labeled albumin injected into normal rats showing albumin labeled vesicles in the basolateral aspect of the proximal tubule cell and apparent fusion with the basolateral membrane were interpreted as evidence of vesicular transcytosis. So far there has been no published, direct evidence for a specific, high capacity, transcellular transport of albumin. In fact, when a high capacity, low affinity mechanism for albumin uptake was identified using isolated, perfused proximal tubules only negligible transtubular transport was found and the uptake was believed to represent non-specific, fluid phase endocytosis.

The proposed high capacity, transtubular albumin retrieval pathway does not involve megalin and cubilin as demonstrated by the fact that inactivation of these receptors in mice leads to albuminuria in the order of 20–40 µg/h which is much less than the proposed amount of transcellular albumin transport. Furthermore, inactivation of megalin and cubilin leads to total abolishment of the normal tubular vesicular labeling ( Fig. 73.4 ) disputing the existence of an alternative and significant pathway of vesicular albumin uptake.

Immunohistochemical analysis of albumin reabsorption in a mosaic megalin- and a mosaic cubilin knockout mouse showing that both receptors are required for albumin endocytosis. Several proximal tubules (PT) from these mice contain both receptor deficient – and wildtype proximal tubule cells. A. Representative micrograph of a megalin knockout mouse with wildtype cells containing megalin (green) dispersed among megalin deficient cells. All cells contain cubilin (not labeled). Albumin (red) is almost exclusively detected in megalin containing cells indicating that cubilin is unable to sustain endocytosis of albumin without megalin. B. Representative micrograph of a cubilin knockout mouse with wildtype cells containing cubilin (green) dispersed among cubilin deficient cells. All cells contain megalin (not labeled). Albumin (red) is almost exclusively detected in cubilin containing cells suggesting that cubilin is the receptor responsible for binding albumin.

It has been hypothesized that the transcellular transport of intact albumin may involve the major histocompatibility complex-related Fc receptor (FcRn). FcRn is responsible for the transcytosis of proteins such as IgG across the placental barrier and possibly podocytes. It has been identified in the luminal part of the kidney proximal tubule cells and binds albumin. FcRn knockout mice are hypoalbuminemic and the half live of circulating albumin is reduced. When kidneys from FcRn deficient mice are transplanted into wild type mice the mice become hypoalbuminemic. Although this points to a role for the FcRn in albumin metabolism, no information on the absolute albumin excretion rate in the urine of these mice was given. Furthermore, when FcRn is expressed in Madin-Darby canine kidney (MDCK) cells transcytosis of IgG but not albumin was observed. Finally, albumin binds FcRn preferable at pH 6.5 suggesting that the affinity for albumin at the luminal membranes of the proximal tubule is low. In conclusion, although the possibility of transtubular transport of intact albumin cannot be excluded the evidence for the transport of large amount of albumin so far has been indirect. The implication of the FcRn as an additional albumin receptor within the proximal tubule is intriguing, however, needs confirmation.

Albumin Induced Renal Tubular and Interstitial Damage

The concept of proteinuria-induced renal tubular and interstitial damage has been propagated as a major factor causing the progressive loss of renal function in CKD. Albuminuria has been shown to be a major risk-factor for the progression of renal disease and the protective effect of anti-proteinuric treatment has been shown to be correlated to the reduction in proteinuria. Experimental studies have demonstrated the ability of albumin, or albumin bound substances, to induce phenotypic changes in tubular cells and animal models have demonstrated albumin overload induced tubular and interstitial damage ( Fig. 73.5 ). In a model based on the amphibian kidney in which some nephrons open into the peritoneal cavity and therefore are selectively exposed to serum or albumin when this is injected intraperitoneally, fibrosis was observed around the exposed tubules. Most of the evidence for the direct effects of albumin on renal tubular cells originates from in vitro studies in which various renal cell lines are exposed to albumin or derivatives and it is not clear whether the tubular effects of albumin involve the megalin/cubilin receptor complex and whether it is dependent on endocytosis. In fact, studies in analbuminemic rats suggested that albumin itself was not important for the progression of proteinuric, renal disease. Interestingly albumin at low or moderate concentrations may promote in vitro cell growth possibly by a phosphatidylinositide 3-kinase dependent pathway whereas high concentrations of albumin activates pathways that induce apoptosis, endoplasmic reticulum (ER)-stress, cytokine production and phenotypic changes. Ultimately these changes can cause cell death, local inflammation and fibrosis. Although it is important to recognize that the phenotype of renal cell lines in culture, in particular immortalized cell lines, may not resemble the in vivo phenotype, similar changes have been observed in experimental models and human diseases characterized by albuminuria as discussed below.

Effects of albumin on tubular cells. Filtered albumin, whether native or modified, as well as albumin bound substance may affect tubular function and phenotype. Albumin in vitro activates various signaling pathways initiating inflammation fibrosis, apoptosis, and possibly EMT (see text for details). The exact role of the balance between the many pathways in normal physiology is not established, however, experimental studies associated with increased filtration of albumin have demonstrated activation of both apoptosis, inflammation and fibrosis suggesting these mechanism to be involved in the development and progression of renal insufficiency. It is not fully established to which extent all or some of these pathways are dependent on albumin binding to the megalin/cubilin-receptor complex, on endocytosis, and/or on the activation of other potential signaling receptors.

Interstitial Inflammation and Fibrosis

In vitro studies using proximal tubule cells have shown that albumin exposure induces the expression of a number of inflammatory and fibrogenic mediators, including cytokines such as regulated on activation normal T cell expressed and secreted (RANTES), monocyte chemotactic protein (MCP)-1, fractalkine and tumor necrosis factors (TNF)- α, as well as endothelin, TGF-β and collagen, and may induce changes in tubular cell expression of surface integrins. These data suggest that albumin and/or bound ligands initiate a series of events that eventually lead to interstitial fibrosis. The exact cascade has not been established, however, proposed intracellular signalling cascades include the activation of NF-κB as well as signal transducer and activator of transcription (STAT) factors, possibly through formation of reactive oxygen species, a process that may also depend on protein kinase C. There is some evidence to suggest that the process involves the initial endocytic uptake of albumin as renal cell lines with low endocytic activity do not respond to albumin by activation of intracellular pathways and collagen synthesis. Interestingly, differential effects have been observed depending on post-translational modifications of albumin. Two studies have evaluated the effect of proteinuria secondary to glomerular damage in megalin deficient mice in which albumin reabsorption is abrogated. In one study crescentic glomerulonephritis was induced by injection of anti-glomerular basement membrane serum into conditional megalin knockout mice revealing ~70% deficiency of megalin in proximal tubule cells. Eighteen days after the induction of glomerulonephritis no difference in tubular degeneration or interstitial fibrosis was observed between megalin deficient mice and controls. However, an upregulation of proinflammatory and profibrotic mediators such as inter-cellular adhesion molecule (ICAM)-1, vascular cell adhesion protein (VCAM)-1, and TGF-β was most pronounced in megalin positive tubular cells. In contrast, an increase in apoptosis markers was more pronounced in megalin deficient proximal tubule profiles. In another study heavy proteinuria was induced by injection of an immunotoxin in sensitive mice expressing human CD25 in podocytes. These mice were combined with conditional megalin knockout mice revealing ~60% deficiency of megalin in proximal tubule cells. Ten days after the induction of proteinuria a higher expression of heme-oxygenase-1 and MCP-1 was identified in megalin expressing tubular cells associated with increased apoptosis. None of these studies examined the isolated effect of albumin on proximal tubule cells nor did they assess any megalin independent effects of albumin which itself was blunted by a residual megalin expression of 30–40%. Furthermore, the studies examined the effect of deficient tubular protein uptake less than three weeks after the induction of proteinuria suggesting that they are not optimal models of CKD. Thus, further studies are needed to examine to effect of targeted disruption of the tubular, albumin receptors in animal models of kidney disease allowing for longer follow up.

Epithelial-Mesenchymal Transition (EMT)

The transition of differentiated tubular cells into myofibroblast-like cells producing α-smooth muscle actin and interstitial matrix components leading to progressive fibrosis has been implicated in the progression of kidney fibrosis. Several mediators have been implicated in this process and recently in vitro studies have suggested that albumin may induce EMT possibly through activation of the mTOR pathway. Whether this occur in vivo following renal injury and whether EMT is in fact induced by protein overload is still controversial.

Effect of Albumin Distal to the Proximal Tubule

There is some suggestion of distal tubular injury associated with albuminuria, however, whether this is a primary or secondary event has not been established. It has been suggested that the filtration of albumin in excess of the capacity for proximal tubule reabsorption, thus leading to the exposure of distal tubule cells to filtered albumin may cause distal tubular cell damage. Apoptosis as well as the increased expression of TGF-β and its receptor have been induced by albumin in vitro in renal cells assumed to have a more distal phenotype such as MDCK cells. In vitro these cells were able to endocytose albumin by a clathrin-mediated pathway. How-ever, although endocytic uptake of proteins has been demonstrated in distal tubular/collecting duct cells the mechanism of distal tubular endocytosis in vivo is unresolved as they do not express megalin or cubilin. See also Figures 73.1 and 73.6 for collecting duct and distal tubular uptake of endogenous albumin in megalin KO mice. Thus, MDCK cells may not be an optimal cell type for the study of the effect of albumin on distal tubule function.

Immunohistochemical investigation of albumin uptake in renal medulla of a megalin knockout mouse showing that when proximal tubular uptake is disrupted, albumin can be detected in more distal part of the nephron as well as in collecting ducts. A. Micrograph of distal tubule segments (thick ascending limbs, TAL) with visible albumin vesicles (red), next to collecting ducts with intercalated cells positive for H ⁺ -ATP’ase (green). B. Micrograph of collecting duct principal cells positive for albumin (red) and intercalated cell H ⁺ -ATP’ase (green).