Podocyte Differentiation and Glomerulogenesis

This review will concentrate on the assembly of the glomerulus and the differentiation of the podocyte during that process. The differentiation of the vascular component of the glomerulus will not be emphasized, as it has been the subject of several recent reviews (1–4), except in discussing the influence of the podocyte on the development of the glomerular vasculature. Glomerulogenesis Our understanding of the morphologic features of glomerular assembly, and indeed of the development of the nephron and the entire kidney, owe much to the work of Edith Potter and her colleague Vitoon Osathanondh during the mid 1960s. In a seminal series of articles that are today overly neglected, they provided much of understanding of how the kidney as an entire organ is assembled (5–10). With regard to the glomerulus, it was shown how it emerges from one end of the S-shaped body during the development of the nephron (10). In understanding this process, and indeed understanding the development of the nephron in its entirety, it is essential to convert two-dimensional pictures into a three-dimensional understanding. This is especially difficult as the nephron is not a fully symmetrical structure on all axes. With these caveats in mind, the following is a description of glomerular development. As shown in Figures 1A and 1B, at one end of the S-shaped body a layer of columnar epithelial cells is present, which represents the future visceral epithelial cells or podocytes. The basal aspect of these cells rests on the future glomerular basement membrane (GBM); on the other side of this basement membrane is a cleft between the podocytes and the cells, which will contribute to the tubular portion of the nephron. The cells that will contribute to the glomerular capillaries, i.e., endothelial and mesangial cells, migrate into this cleft. The origin of these cells is a subject of ongoing research and, as mentioned above, has been discussed in recent reviews (1–4). On the other side of this patch of future podocytes, overlying their apical surface, is a lining of thin cells that will become the parietal epithelium, also known as Bowman's capsule. Figure 1. : Development of the glomerulus. (A and B) show two different sections through an S-shaped body. In both cases, the early capillary loop is observed within the glomerular cleft and the thin lining of cells that will comprise Bowman's capsule is adjacent to the apical side of the podocytes. In panel B, the cells that will comprise the proximal tubule are also noted. Panels C and D show how the podocyte layer begins to form a "cup" around the capillary bundle and how the capillary bundle remains attached to the outside vasculature. Panel E shows a more mature glomerulus, by which time the podocytes have migrated around the capillary loops. The capillary loops are still wider than they will be in a fully mature glomerulus. Panel F is an electron micrograph showing the podocyte layer at the point where most lateral cell attachments have been lost, except at the basal end of the lateral membrane, where the slit-diaphragm complex will be assembled in mature podocytes. A thin GBM is present at this stage. P, podocyte; C, capillary loop; BC, Bowman's capsule; PT, proximal tubule.The development of the glomerulus is a dynamic process involving the expansion of the original capillary component into a plexus of six to eight individual loops, and the concomitant migration of the podocytes to be distributed around these loops (Figure 1, C and D). Although cause and effect relationships are not known, the capillary bundle displaces the layer of future podocytes as it expands. This layer of future podocytes, in turn, form a "pocket" surrounding the capillary bundle (Figure 1D), which maintains contact with the outside vasculature through an arterial and venous supply that will become part of the glomerular stalk. As the primitive podocytes form this pocket, the GBM remains a constant barrier between the epithelial and capillary components. The podocytes themselves do not remain a columnar epithelium. As they form this pocket, they begin to lose their lateral cell-cell attachments to each other (Figure 1F), but they remain attached to the GBM such that they no longer resemble a traditional epithelial cell. During this time, they also begin to migrate around the capillary loops, so that they no longer form a continuous uniform patch of cells. By this time, the glomerulus can be recognized as a discrete structure apart from the remainder of the nephron. It is during this phase of glomerular development and podocyte maturation that foot processes begin to form (Figure 1E). Early pictures of foot process assembly suggest that it begins with the selective detachment of podocytes from the GBM. However, the appearance of mature podocytes, with foot processes extending a significant distance from the main cell body, are also suggestive of a process whereby cytoplasmic extensions resembling filopodia extend themselves as a scaffolding around the capillary loops. Foot Process Assembly The extent to which foot process assembly reflects selective cell detachment versus cell extension and migration is not known; indeed, it is difficult to determine, given our inability to observe this process in living cells. One informative observation bearing on this issue is that adjacent foot processes are derived from different podocytes. Cells that begin as adjacent epithelial cells end up as cells with isolated cell bodies but interdigitated foot processes. It is possible that this situation arises entirely by extension of foot processes from cells that have initially dissociated from each other. However, another possibility is suggested by recent work examining cell-cell contact in keratinocytes (11). In this case, which is not generalizable to all epithelial cells, cell-cell contact is initiated through the extension of interdigitated filopodia between two cells (11). In the case of keratinocytes, these filopodia resolve into a conventional cell-cell junction mediated by cadherins. These findings in keratinocytes suggest two possible hypotheses for the assembly of foot processes. In one model, foot process assembly begins when podocytes are still maintaining cadherin-mediated adhesions along their lateral membranes (Figure 2B). In this case, adjacent cells would extend interdigitating filopodial-type extensions along their basal aspects, where they are attached to the GBM. Then, these filopodial extensions would be maintained as foot processes, while the remainder of the cell loses its cadherin-mediated adhesion such that appear as independent cell bodies. In the alternate model, podocytes first entirely dissociate from each other and then extend filopodial-like extensions that interdigitate to form foot processes (Figure 2A). Present observations appear to favor the model depicted in Figure 2B, as it is not normal to observe capillaries in developing glomeruli whose outer walls are not associated with an extension from a podocyte, implying that poodcytes or mesangial cells always fully encompass the capillary loops. However, the author is unaware of any studies that directly assess these hypotheses. A necessary component of both models is a phase in which there is partial dissociation of the initial cell body from the GBM, in places where it was originally fully attached along its entire basal membrane. Figure 2. : Two models of foot process formation. These are apical views from "on top" of the podocytes. For clarity, only two adjacent podocytes are depicted in both cases. (A) Two primitive podocytes that have separated from each other begin to extend processes that meet and become interdigitated. For simplicity, only one major extension is shown, although several would normally extend from the full circumference of the podocyte. (B) Two podocytes, which begin as columnar epithelial cells, always remain attached to each other along the basal end of their lateral membranes. Their common cell-cell junction is remodeled from a simple linear one to one that interdigitates and that contains the slit-diaphragm complex. The heavy lines represent the cell body of the podocyte, which loses attachment to other podocytes, except at the cell-cell junction between foot processes.Maturation of the GBM The maturation of the GBM has been the subject of several recent reviews (12,13) and is briefly dealt with here. The earliest epithelial cells of the nephron mainly express laminin-1. As soon as it is possible to define a nascent GBM, i.e., the basal lamina underlying the layer of future podocytes, a shift in laminin expression to isoforms containing the α4 subunit is observed (14–16). Upon further maturation of the GBM, there is a second shift to the expression of α5 and β2 subunits, which are components of laminin 11, and this continues to be the major laminin isoform in the mature GBM (14–16). There is also a shift in the expression of type IV collagen (17). The early nephron mainly expresses the α1 (IV) and α2 (IV) subunits (17); upon maturation of the GBM, there is a shift to α3, α4, and α5 (IV) subunits. Other major components of the GBM are nidogen (entactin) and heparan sulfate proteoglycans, most notably agrin (13). Slit-Diaphragm The slit-diaphragm (SD) has also been the subject of many recent reviews (18–20). This is a structure observed by electron microscopy between adjacent foot processes, which is a major component of the protein barrier between the circulation and Bowman's space. The SD has, in recent years, been the focus of intense study, as mutations in several genes encoding protein components of the SD have been found to be the cause of various forms of childhood kidney disease (21). The first of these was nephrin; mutations in the NPHS1 gene that encodes nephrin are responsible for the Finnish form of congenital nephrotic syndrome (22). Nephrin is a transmembrane protein localized to the SD (23–25) that was originally thought to provide the structural link between two foot processes, without which it is not possible to maintain the foot process structure. Recent work has suggested a more complicated model. First, mice containing a targeted mutation in the nephrin gene were surprisingly capable of assembling foot processes, although the SD structure itself was not apparent (26–28). These mice suffered from heavy proteinuria during the neonatal period and died within a few days, with foot process effacement observed before death. These results confirm that the SD itself is a major component of the protein barrier and that the GBM by itself is not sufficient. Second, they indicate that there must be other structural molecules besides nephrin responsible for the initial assembly of foot processes. Recently, additional members of the nephrin family, neph1 to neph3, have been identified (29,30). Neph1-deficient mice also develop heavy proteinuria (30), indicating that neph1 is also an important component of the protein barrier. Two other proteins have also been identified as important components of the SD: podocin and CD2-AP (18,31–34). Podocin, the product of the NPHS2 locus, was identified as the product of a gene involved in autosomal recessive steroid-resistant nephrotic syndrome (35). Podocin is a transmembrane protein that interacts with nephrin and neph1 (29,31). CD2-AP, originally identified as an important protein in lymphocytes, was found to also be important in the kidney when CD2-AP knockout mice developed glomerular disease (36). CD2AP is also associated with the SD. Podocin, CD2AP, and nephrin all appear to be present in a complex within lipid rafts (31,37), which may also be associated with the cytoskeleton (38). Foot processes can be assembled in the absence of nephrin; it is therefore necessary to formulate new hypotheses about the role of the SD complex in foot process assembly. As alluded to above, it is possible that Neph1 is partially redundant with nephrin. Alternatively, it is possible that the driving forces responsible for the cytoskeletal reorganization involved in foot process assembly are entirely anchored at the adhesion apparatus linking the podocyte to the GBM. Finally, it is possible that there are additional molecules at the SD involved in cytoskeletal reorganization. Two additional adhesion molecules have been described at the SD: P-cadherin and a large proto-cadherin named FAT (39,40). Finding P-cadherin at the SD suggests that this structure or some subset of its components may bear some relationship to adherens junctions found in more traditional epithelial cells. However, the importance of P-cadherin at the SD is obscured by the failure to find significant renal dysfunction in P-cadherin knockout mice, although their kidneys have not been closely examined (41). FAT is a large proto-cadherin, containing a much larger extracellular domain than traditional cadherins (42,43). It was first defined as a tumor suppressor gene in drosophila (44). Its function in mammalian cells is unknown. ZO-1 is a protein most commonly associated with tight junctions, but it is also sometimes associated with adherens junctions (45,46). ZO-1 is found at the cytoplasmic face of the SD (47), providing another indication that the SD complex may functionally resemble an epithelial cell-cell junction. The observation that cell junctional proteins are found at the SD may have implications for the assembly of foot processes. Podocytes begin as columnar epithelial cells, and adjacent foot processes are always from different podocytes; it is possible that the SD complex begins as a cell-cell junction between adjacent epithelial cells and is conserved as these cells otherwise dissociate from each other. However, this model does not explain how the extensive interdigitation would occur to achieve mature foot processes. Podocyte Differentiation Podocytes are distinguished as a separate population from other cells in the developing nephron during the transition from the S-shaped body to the distinct glomerulus. At this time, they are expressing podocyte-specific markers such as WT1 and nephrin (22,48,49), which are not expressed in other cells of the nephron (in the case of nephrin) or whose expression pattern becomes restricted to podocytes (e.g., WT1). Little is known about the molecular genetic basis for segmentation of the nephron, particularly which genes act to define the podocyte versus tubular lineages. There is speculation that genes generally involved in segmentation, such as Hox genes or members of the Notch family, may be involved, but there is no published evidence of a role for any of these genes in segmentation of the mammalian nephron. However, in the development of the frog pronephros, inhibition of the Notch signaling pathway expands the domain of Wt1 expression (50), the latter usually associated with development of the glomus. Thus, this provides support for the involvement of the Notch signaling pathway in segmentation of the nephron. It should be noted that because of the extraordinary convolution of the mammalian nephron, analysis of segmentation will be a difficult proposition, as it is not straightforward to recognize the various nephronic segments within the nephrogenic zone of a developing kidney. It is therefore likely that our understanding of segmentation will require further studies in frogs and possibly zebrafish before this can be adequately approached in mammalian nephrons. Transcription Factors in Podocytes WT1. The transcription factors best studied in podocytes include WT1, Lmx1b, Pod1, and Kreisler. WT1 is a zinc finger transcription factor and RNA binding protein first identified as a tumor suppressor gene for Wilms tumor, a tumor of the kidney observed in young children (reviewed in reference 51). WT1 is expressed in the metanephric mesenchyme at the beginning of kidney development, but as the S-shaped tubules appear, expression of WT1 becomes restricted to the podocyte lineage, where it is maintained throughout adult life, albeit at lower levels than in developing glomeruli (48,52). Mouse embryos deficient in WT1 suffer from complete renal agenesis (53); thus the Wt1 knockout is not informative about a role for WT1 in podocyte differentiation or function. It must also be stated that, although there is copious literature on the molecular function of WT1, including the identification of many genes that are putative regulatory targets (51), much of this work has not been very informative of a role for WT1 in podocytes. Fortunately, recent studies have begun to shed more light on the role of WT1 in podocytes. The WT1 gene has two alternatively spliced exons, producing four major splice forms of WT1 mRNA (54). There are several alternative translational start sites, such that at least 16 distinct peptides are possible (55), although there is little information on distinct functions of peptides with different start sites. The first alternatively spliced exon, exon 5, inserts 17 amino acids in between the proline-rich amino terminal domain of WT1 and the four zinc finger domains (54). Mice unable to express WT1 containing exon 5 are normal (56), suggesting that isoforms containing exon 5 do not have a major role in kidney development or function. The other alternative splice site is not a separate exon, but rather features two distinct splice donor sites at the end of exon 9 such that three amino acids, lysine-threonine-serine (KTS), can be alternatively spliced in between exons 9 and 10 (54). The KTS sequence changes the structure of the third zinc finger, altering its DNA binding capabilities (57). Denys-Drash Syndrome (DDS) results from mutations that affect the DNA-binding ability of the WT1 zinc finger domains (58). DDS involves a diffuse mesangial sclerosis within the glomeruli. Because the mutant forms of WT1 found in DDS presumably maintain the ability to interact with putative WT1-interacting proteins through the proline-rich amino terminal domain, it is hypothesized that the pathologic sequelae associated with DDS are due to a dominant-negative phenotype (59). A recent study from our laboratory involved the derivation of transgenic mice that expressed a DDS truncation mutation of WT1 specifically in podocytes (60). Podocyte differentiation, as judged from the expression of GBM, SD, and cytoskeletal proteins, appeared unaffected by this mutation, but there was abnormal development of the glomerular capillaries and decreased expression of PECAM on capillary endothelial cells (60). These results suggested that WT1 might mainly be responsible for directing expression of growth factors that regulate development of the glomerular vasculature, rather than regulating the intrinsic differentiation of the podocyte itself. A contrasting conclusion was obtained in a recent study using an immortalized kidney cell line, which suggested that WT1 may regulate expression of podocalyxin, a highly charged podocyte cell surface protein that is thought to be involved in maintaining separation between podocytes due to electrostatic charge repulsion (61). In the aforementioned transgenic study, podocalyxin expression appeared unchanged in glomeruli of transgenic mice (60). A recent study by Hammes et al. (62) was more pertinent to Frasier syndrome, which also involves mutations in WT1, which lead to glomerulosclerosis. Frasier syndrome results from dominant splice donor site mutations that render an allele unable to express the +KTS form of WT1, thus altering the splice form ratio of WT1. Hammes et al. (62) targeted mutations to the murine Wt1 gene that resulted in the inability to express either the −KTS or +KTS forms of WT1. Both mutations resulted in abnormal glomerular development, although mice only able to express the +KTS isoform had a more severe phenotype, suggesting that there may be an earlier or broader requirement for the −KTS isoform during glomerular development (62). Mice containing one normal and one −KTS only allele reproduced the Frasier phenotype, with glomerulosclerosis developing several weeks after birth. Interestingly, the expression of nephrin, a component of the slit diaphragm, was decreased in these knockout mice (62), again in contrast to studies with the WT1 transgene (60). Together, these studies emphasize the complexity of the Wt1 gene, and the difficulty in making progress in achieving a full understanding of its function during podocyte differentiation. LMX1B. LMX1B is a LIM-homeodomain transcription factor, mutations of which are responsible for nail-patella syndrome, a condition that may involve glomerular disease (63–66). LMX1B is expressed in podocytes, and recent studies involving LMX1B knockout mice demonstrated reduced assembly of podocyte foot processes and reduced expression of CD2AP, podocin, and the α3 and α4 chains of type IV collagen in LMX1B-deficient glomeruli (65,66). LMX1B binding sites have been found in the upstream regions of these target genes (65,66), indicating a major role for LMX1B in regulating podocyte-specific gene expression. Pod1. Pod1 (also known as epicardin and capsulin) is a basic helix-loop-helix (bHLH) transcription factor expressed early in kidney development and subsequently in podocytes of S-shaped bodies (67,68). Expression persists in podocytes of adult kidneys. In Pod1 knockout mice, glomerular development appears arrested at the single capillary loop stage (68). In most Pod1-deficient glomeruli, the podocytes remain as columnar-shaped cells that have lost their lateral cell-cell attachments but remain fully adhered to the GBM without any foot processes. Thus the requirement for Pod1 in podocyte differentiation and glomerular development is slightly before the time when podocytes would normally begin migrating around the capillary loops and assembling foot processes. At present, there is no information on possible Pod1 target genes in podocytes. Kreisler. Kreisler (MafB) is a basic domain leucine zipper (bZip) transcription factor of the Maf subfamily, expressed in podocytes of capillary loop stage glomeruli (69). It has previously been shown to have an important role in segmentation of the hindbrain. Expression is maintained in mature podocytes. There is a failure of foot process formation in Kreisler-deficient podocytes, though similarly to Pod1 mutants, podocytes of Kreisler mutants have lost their lateral cell-cell attachments (69). Pod1 is expressed in Kreisler mutant podocytes, indicating that Kreisler is more likely acting either downstream or in a separate pathway from Pod1 (69). There is a small decrease in expression of podocin and nephrin in Kreisler mutants, although this probably does not, by itself, account for the glomerular phenotype. Mf2. Mf2 (mesoderm/mesenchyme forkhead 2) encodes a forkhead/winged helix transcription factor expressed in the developing kidney, among other tissues in the developing embryo (70). It is expressed in the condensed mesenchyme; as the nephron matures, MF2 expression appears to be restricted to the podocytes, similarly to Wt1. Mf2-deficient mice are generally normal, but some experience renal hypoplasia (70). The morphology of podocytes in these hypoplastic kidneys has not been reported. Podocyte Adhesion Molecules Two major adhesion complexes are involved in interactions between the podocyte and the GBM, α3β1 integrin, and the α,β dystroglycan complex. α3β1 is the major integrin expressed by podocytes (71), although low levels of α6β4 are also detected (72). The importance of α3β1 integrin in podocyte differentiation first became apparent when the knockout of the α3 integrin gene resulted in an inability to assemble foot processes, and α3β1 integrin neonatal mice do not survive more than a few hours after birth, most likely due to glomerular dysfunction (73). The GBM itself is fragmented, suggesting a role for α3β1 integrin in maintaining the structural organization of the GBM. Interestingly, the portion of the α3β1 integrin-deficient podocyte in contact with the GBM appears fully adhered (73). Thus a simple role of an adhesion receptor is not appropriate for α3β1, as this might suggest that podocyte should adhere more loosely to an adjacent extracellular matrix. This is consistent with findings in other cells which suggest a regulatory role for α3β1 integrin (74). α3β1 integrin is a receptor for laminin-5 and laminin-11 (75,76), the latter of which is a major component of the GBM (16). Older studies on α3β1 had also ascribed roles as a receptor for laminin-1, collagen, and fibronectin, but it is currently thought that α3β1 acts mainly as a receptor for these other isoforms of laminin (reviewed in reference 77). α3β1 is expressed early in podocyte differentiation, probably before the transition from laminin-1 to laminin-11 (71). It is not known with what ligand α3β1 integrin interacts before expression of laminin-11; it may interact weakly with laminin-1, or perhaps with small amounts of laminin-11 that are beginning to accumulate in the GBM. It should also be noted that the preferential binding of α3β1 to laminin-11 is described in tissue culture studies; it is possible that the preference for laminin-11 is not as dramatic in the in vivo situation. CD151, a member of the tetraspanin family, is often if not always co-expressed with α3β1 integrin, including in podocytes (78,79). A tight interaction between α3β1 and CD151 has been defined, which also appears to recruit certain isoforms of PKC to a site on CD151 (79,80). In general, tetraspanins appear to have a role in cell-cell adhesion (81), rather than in cell-matrix adhesion; thus the functional consequence of the α3β1:CD151 interaction in podocytes is not well understood. Less is known about the role of the dystroglycan complex in podocytes. The dystroglycan knockout results in early embryonic lethality (82), making it impossible to study dystroglycan-deficient podocytes. However, conditionally mutant dystroglycan mice have now been published (83) that may be crossed with mice expressing Cre recombinase in podocytes (84,85) to address this question. Other Cell Surface Proteins Podocalyxin is a highly sulfated cell surface sialomucin-charged protein found on the cell-surface of podocytes (86). It is hypothesized to have a role in maintaining podocyte cell separation through charge repulsion, and expression of podocalyxin in MDCK cells leads to cell separation, consistent with this possibility (87). Mice deficient in podocalyxin are unable to assemble foot processes or slit diaphragms and are anuric, succumbing during the first 24 h after birth (88). GLEPP1is a receptor tyrosine phosphatase present on the apical cell surface of podocytes (89). GLEPP1-deficient mice are viable but have reduced filtration and are more susceptible to hypertension in a unilateral nephrectomy model (90). Podocyte Cytoskeletal Proteins It is assumed that specialized cytoskeletal organization is the basis for foot process assembly, although our understanding of this process is at its earliest stages. Synaptopodin is an actin-associated protein found in podocytes and in a restricted population of cells in the nervous system (91). How synaptopodin might function in foot process assembly or other aspects of podocyte cytoskeletal organization is not known. α-actinin 4 is an isoform of α-actinin, an actin-binding protein, expressed in podocytes (92). α-actinin 4 first came to prominence in the field of podocyte biology when mutations in the ACTN4 gene were identified in a familial form of focal segmental glomerulosclerosis (92). It is not understood why these mutations allow normal podocyte development, but then lead to disease later in life. Podocytes and the Expression of Signaling Molecules Podocytes express growth factors that have the potential to affect glomerular development. Most notably, they express VEGF-A and angiopoietin-1, which may affect the development of the glomerular capillaries (4,93,94). VEGF is generally involved in stimulating capillary growth, and angiopoietin-1 with the remodeling and maturation of capillaries (94,95). There is a growing body of evidence suggesting that VEGF is not only involved in the formation, but also in the maintenance of glomerular capillaries throughout life, and that failure to properly maintain the capillary environment may be a major component of glomerular disease (96,97). Thus this area warrants additional study. Notch2 is a member of the Notch family family of signaling molecules that are involved in cell fate specification. Notch2 is expressed in podocytes; in Notch2-deficient mice, there is a failure to develop the normal glomerular vasculature (98). Jagged 1, a ligand of Notch2, was expressed in the adjacent endothelial and/or mesangial cells, providing additional support for the concept that glomerular vascular development depends on signals from podocytes (98). Future Directions There is much yet to be learned about podocytes and their interactions with adjacent cells. Although there are many genetic and pathologic models in which foot process assembly is diminished or prevented, our understanding of the molecular processes that establish foot processes is still in its infancy. In particular, it will be important to establish the relative roles of the SD complex and GBM adhesion apparatus in foot process assembly. It will also be important to identify additional targets of the transcription factors expressed in podocytes, as well as whether there are additional transcription factors regulating podocyte differentiation that are yet to be identified. Potential targets whose regulated expression pattern may affect podocyte differentiation include structural molecules involved in foot process formation, as well as growth factors that affect capillary development. Another area of investigation important in podocyte differentiation as it relates to the segmentation of the nephron, i.e., during the formation of the comma and S-shaped tubules, is what regulatory events determine the identity of particular glomerular cells and distinguish them from the proximal tubule, as well as those events that segment the tubule itself. The recent derivation of mouse strains that allow podocyte-specific gene targeting (84,85) and immortalized podocyte cells lines (84,85,99) will greatly aid future efforts. Work in the author's laboratory is supported by the NIH and the March of Dimes. The author thanks Thomas Natoli, Cynthia Wong, and Mary Taglienti, Helmut Rennke, and Wilhelm Kriz for helpful comments. This review is dedicated to the memory of Edith Potter M.D., Ph.D. (1902–1993).