J Am Soc Nephrol 12: 2060–2071, 2001 Podocyte Bridges between the Tuft and Bowman’s Capsule: An Early Event in Experimental Crescentic Glomerulonephritis

MICHEL LE HIR,* CORNELIA KELLER,* VALE´ RIE ESCHMANN,* BRUNHILDE HA¨ HNEL,† HILTRAUDE HOSSER,† and WILHELM KRIZ† *Institute of Anatomy, University of Zurich, Switzerland; and †Institute of Anatomy, University of Heidelberg, Germany.

Abstract. Although experimental crescentic glomerulonephritis and prominent microvillous transformation, and some cres- starts with an endocapillary inflammation, the crescents them- cents were found. On day 10, crescents were found in 40% of selves seem to originate from the proliferation of parietal glomeruli. The most surprising finding was podocytes that epithelial cells (PEC). In this study, an attempt was made to adhered to both the glomerular basement membrane and the disclose a link between the two processes by a morphologic parietal basement membrane, thus forming bridges between the analysis of early stages of the disease. Mice were immunized tuft and Bowman’s capsule. Those podocyte bridges were with rabbit IgG in complete Freund’s adjuvant on day Ϫ6. At sparse on day 3 but were regularly encountered on days 6 and day 0, they received an intravenous injection of a rabbit anti- 10 in glomeruli without crescents and also as a component of glomerular basement membrane serum. On days 3, 6, and 10, crescents. They were interposed between PEC and later be- the kidneys were fixed by vascular perfusion for examination tween the cells of a crescent without formation of junctional by light and electron microscopy. On day 3, morphologic connection with these cells. It is proposed that the spreading of alterations affected mainly the endocapillary compartment; podocytes on the parietal basement membrane represents a most podocytes appeared to be intact. On day 6, alterations of lesion of the parietal epithelium and that this process initiates podocytes were widespread, including foot process effacement the proliferation of PEC to form a crescent.

A cellular crescent consists of a multilayered accumulation of the disease. Indeed, Boucher and colleagues (5) found a pre- cells that limit a diseased glomerulus in the place of the simple dominance of PEC in crescents in patients who showed an epithelium of Bowman‘s capsule. Concepts about the cellular intact parietal basement membrane (PBM). In contrast, mac- composition of crescents have evolved along with the available rophages were the major cell population of crescents in patients techniques. Three periods may be identified, of which the two in which ruptures of the PBM were detected, which possibly first have been reviewed by Jennette and Hipp (1). In early represents a later stage of disease (5). In contrast to clinical studies, it was postulated on the basis of structural studies that studies, animal models of anti–glomerular basement mem- crescents originated from the proliferation of epithelial cells. In brane (GBM) glomerulonephritis offer the possibility to ana- a second period, as antibodies against leukocytes became lyze the first days of crescent formation. In rats, crescents broadly available, many studies demonstrated the presence of consisted of epithelial cells as long as the PBM was intact (7). macrophages in crescents, and this cell type was for many Rupture of the PBM initiated a fibrocellular organization of years the focus of interest. In a third period, it became possible crescents, with a high incidence of macrophages (7). In rabbits, to identify parietal epithelial cells (PEC) by use of antibodies macrophages were rare in crescents, in spite of their abundance against cytokeratin (1–4) or other antigens (5,6). It then be- in the endocapillary compartment (8). In mice, the model used came evident that, in most patients, PEC are the predominant in this study, crescents were composed of PEC, whereas mac- cell type in crescents. The fact that, in some patients, PEC rophages were lacking (6). Increased proliferation of parietal represented only a small proportion of cells in crescents does epithelial cells has been measured in anti-GBM glomerulone- not necessarily reflect a heterogeneous origin of crescents. phritis in the three species (8–11). Rather, the cellular composition changes with the evolution of The events that trigger the proliferation of PEC in crescentic glomerulonephritis remain obscure. PEC are not in direct con- tact with the endocapillary compartment in which the inflam- Received November 6, 2000. Accepted April 12, 2001. matory process is initiated. Furthermore, it is not known how Correspondence to Michel Le Hir, Institute of Anatomy, University of Zu¨rich, the simple squamous parietal epithelium acquires the specific 8057 Zu¨rich, Switzerland. Phone: ϩ41 1 635 53 16; Fax: ϩ41 1 635 57 02; E-mail: [email protected] organization of a cellular crescent. Clearly, that process impli- cates more than just cell proliferation. Therefore, in this study, 1046-6673/1210-2060 Journal of the American Society of Nephrology we focused on the morphologic alterations of the parietal Copyright © 2001 by the American Society of Nephrology epithelium in the early stages of crescent formation. To this J Am Soc Nephrol 12: 2060–2071, 2001 Podocyte Bridges in Glomerulonephritis 2061

aim, we used a murine model of crescentic glomerulonephritis Immunohistochemistry in which the disease is induced by intravenous injection of an Paraffin sections of 3-␮m thickness were rehydrated in alcohol anti-GBM serum in preimmunized mice. The pathology relies series and finally in phosphate-buffered saline (PBS). For detection of on a cellular immune response, with virtually no role of autol- macrophages, we used rat anti-mouse antibodies: anti-MHC class II ogous antibodies (12,13). We observed that formation of podo- (clone M5/114.15.2) and anti-F4/80 (clone CI:A3-1). Podocytes were cyte bridges between the GBM and the PBM is an early event identified with rabbit polyclonal antibodies against WT-1, CD2AP (both Santa Cruz Biotechnology, Santa Cruz, CA), ␣ integrin in this model of crescentic glomerulonephritis. 3 (Chemicon International Inc., Temecula, CA), synaptopodin, and podocin (both gifts from Dr. P. Mundel, Albert Einstein College of Materials and Methods Medicine, New York; a manuscript concerning the characterization of Animals podocin, by Schwarz K. et al., has been submitted for publication). Female C57BL/6 mice were obtained, specified pathogen-free, and Before the rabbit antibodies were used, the sections were submitted to kept in sterilized filter-top cages during the experimental period. 5 ϫ 5 min microwaving at 600 W. Detection was carried out with Manipulation of the animals was carried out on a sterile bench. The Vectastain ABC kits (Vector Laboratories, Burlingame, CA) with the experimental protocol was approved by the Cantonal Veterinary Of- use of peroxidase as label and diaminobenzidine as substrate. fice of Zurich. Results Induction of Anti-GBM Glomerulonephritis As found previously in this model (13,14), proteinuria Female mice, 13 wk old, were immunized by subcutaneous injec- started on day 2 after injection of the anti-GBM serum. On day tion of 0.2 mg of rabbit IgG (Jackson ImmunoResearch Laboratories, 3, many glomerular capillary loops displayed marked alter- West Grove, PA) in 0.2 ml of a 1:1 emulsion with complete Freund’s ations in the endocapillary compartment. On day 6, a broad adjuvant (Sigma, St. Louis, MO). Six days later (day 0), glomerulo- array of histologic alterations, ranging from almost normal nephritis was induced by intravenous injection of 0.4 ml of a 1:10 histology to fully developed cellular crescents, were seen dilution of rabbit anti-mouse GBM serum (14). Urinary protein con- among glomeruli of the same kidney. On day 10 after the centration was evaluated daily by use of dipsticks (Albustix; Miles, application of the serum, profound alterations were seen in West Haven, CT). virtually all glomeruli, and about 40% of them exhibited crescents. Fixation and Tissue Processing Mice were anesthetized with 17 mg/kg body weight xylasine hy- Histopathology of the Glomerular Tuft drochloride and 50 mg/kg body weight ketamine hydrochloride, in- A vivid endocapillary inflammation was encountered in traperitoneally. Kidneys were fixed by vascular perfusion via the abdominal aorta. The fixative consisted of 3% paraformaldehyde many glomeruli on day 3, but, at that time, the majority of (PFA) and 0.05% picric acid. It was dissolved in a 3:2 mixture of 0.1 capillary loops still appeared healthy. On day 6 endocapillary M cacodylate buffer (pH 7.4, adjusted to 300 mOsm with sucrose) and injuries were more widespread, and on day 10 only few loops 10% hydroxyethyl starch in saline (HAES; Fresenius AG, Germany). displayed a normal structure. Qualitatively however, morpho- After 5 min of fixation in situ, the kidneys were removed and cut into logic alterations of the tuft remained similar from day 3 to day coronal slices. Some slices were embedded in paraffin. The remaining 10. tissue was immersed for at least 24 h in the 3% PFA fixative solution, Because of widespread mesangiolysis, mesangial spaces to which 0.5% glutardialdehyde was added. Thereafter, the tissue was were dramatically expanded. Subendothelial spaces created by postfixed in 1% OsO4 and embedded in epoxy resin. detachment of endothelium from the GBM contained variable amounts of a proteinaceous material (hyaline), cell processes Light Microscopy and Transmission of mesangial cells (which frequently had lost contact with the Electron Microscopy GBM), and inflammatory cells (Figures 1 through 3). In TEM, Light microscopical investigations were carried out on series of sparse filaments of fibrin could be detected amid of hyaline 100 sections of 1-␮m thickness, cut from epoxy resin-embedded material; conspicuous accumulation of fibrin was found in only tissue and stained with azure II–methylene blue. Additionally, series few capillary loops in TEM as well as by immunofluorescence ␮ of 16 sections of 5- m thickness were cut from paraffin-embedded (not shown). Adhering mononuclear cells and neutrophils were tissue and stained with silver-metheneamine for examination of the seen in the capillary lumina (Figures 1 and 2). Although the basement membranes. For transmission electron microscopy (TEM), GBM was often attenuated over extended areas, we encoun- ultrathin sections from epoxy resin-embedded tissue were contrasted with uranyl acetate and lead citrate. tered sites of complete dissolution in only two instances. Some glomeruli displayed an endocapillary necrosis of most or even all loops. The most striking alteration of podocytes was the Immunofluorescence Microscopy formation of villi and clublike protrusions of the luminal cell Collagen type IV was detected using a rabbit polyclonal antibody domain (microvillous transformation) (Figures 1, 3, and 4). At (Progen Biotechnik, Heidelberg, Germany) on 1 ␮m cryosections. The second antibody was goat anti-rabbit Cy3 (Rockland, Gilberts- day 10, the vast majority of podocytes were affected. Simpli- ville, PA). For fibrin detection a goat anti-mouse fibrinogen antibody fication of the foot process pattern up to foot process efface- was used (Nordic Immunologicals, Tilburg, The Netherlands), fol- ment was also widespread (Figures 1 through 3); the remaining lowed by donkey anti-goat FITC (Jackson Immunoresearch Labora- interdigitating podocyte cell processes were connected either tories, West Grove, Pennsylvania). through slit membranes or, more frequently, by tight junction- 2062 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 2060–2071, 2001 like structures. In addition, podocytes frequently contained rarely found between these bridging cells and PEC; in most large intracytoplasmic vacuoles; some microcysts were also cases gaps were seen between them, which indicates the observed. In heavily damaged loops, circumscribed detach- absence of junctional belts (Figure 2). The processes of the ment of podocytes from the GBM also occurred. The sub- bridging podocytes often covered large areas of the PBM epithelial space created by detachment appeared to be empty (Figure 2). Frequently, they extended between parietal cells or it contained basement membrane–like material (Figure and the PBM (Figure 4A). 1). Of note, the continuity of the podocyte layer along the The bridging cells were often assembled in groups (Figure GBM was preserved even in the most heavily damaged 4E), enclosing extracellular spaces that were filled with base- glomeruli. ment membrane–like material (Figure 3), which showed im- munoreactivity for collagen type IV (Figure 5). Because these Cell Bridges between the Tuft and Bowman’s Capsule spaces began at the GBM and extended to the PBM, they might in Glomeruli without Crescents represent extracellular routes for diffusion of substances from In the 1-␮m section series, we traced glomeruli with the capillaries to the PBM. respect to alterations that might represent early stages of By means of immunocytochemistry, we could show that the crescent formation. Thereby, we observed, as the earliest structural change, already found on day 3, cellular bridges cells forming the bridges between the tuft and Bowman’s ␣ between the tuft and Bowman’s capsule. These bridges capsule were positive for WT1 (Figure 6A), integrin 3 (Fig- consisted of one or a few epithelial cells that were attached ure 6B), synaptopodin (Figure 6C), podocin (Figure 6D), and to both the GBM and the PBM (Figures 2 and 3). From their CD2AP (Figure 6E), as were the podocytes of the tuft; parietal structure, most notably from the presence of foot processes cells were negative for these markers. at the tuft side, these bridging cells appeared to be podo- In several glomeruli that presented podocyte bridges, we cytes. At the capsular side, their cell processes, displaying encountered cells within the parietal epithelium that contained manifold elaboration, were interposed between PEC and mitotic figures (Figure 7). Mitotic figures in PEC are extremely attached to the PBM (Figure 2). Cell-to-cell contacts were rare in healthy kidneys.

Figure 1. Injured capillary loops (day 10) (transmission electron microscopy [TEM]). In the two loops on the right, the endothelium has detached from the glomerular basement membrane (GBM); an inflammatory cell is visible in the subendothelial space. The third loop is necrotic and filled with inflammatory cells (૽). Flattened foot processes of podocytes show densely packed microfilaments (*), and the apical cell portions display microvillous transformation (3). c, capillary lumen; PE, parietal epithelium. J Am Soc Nephrol 12: 2060–2071, 2001 Podocyte Bridges in Glomerulonephritis 2063

Figure 2. Podocytes bridging the GBM and the parietal basement membrane (PBM) (day 10) (TEM). The structures that belong to podocytes on the one hand, and that are seen on the micrograph to bridge both basement membranes on the other hand, are highlighted in yellow. Tight junctions cannot be detected between podocytes and parietal epithelial cells. (a) A broad area of the PBM is covered by two podocytes. A gap is visible between a podocyte and the neighboring parietal epithelial cell (insert). (a1,a2) Closer views on different sections show the formation (ء) of foot processes on the GBM by the podocyte seen at the bottom of A. (b) The podocyte has produced basement membrane–like material that seems to be continuous with the PBM. A gap is visible between the podocyte and a parietal cell. (c) The necrotic capillary loop is filled with hyaline. Gaps are visible between the process of podocyte adhering to the PBM and the adjacent parietal epithelial cells (PEC).

Structure of Crescents serial sections stained with silver-methenamine. In our hun- Crescents were generally separated from the interstitium by dreds of TEM pictures, we encountered in only one instance a a continuous PBM. At day 10, a rupture in the PBM was found rupture of the PBM. in only 5 of 136 crescentic glomeruli examined on 5-␮m-thick The morphology of most cells in the crescents was compat- 2064 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 2060–2071, 2001

Figure 3. Podocyte bridges that involve two podocytes (highlighted). The spaces between the podocytes are filled with extracellular matrix (* in b) and cell processes. Intercellular junctions can be detected between cell processes that undoubtedly belong to podocytes (&U279E; in a), whereas they are lacking between podocytes and cell processes of PEC (• in a), as well as between the cell processes of PEC (➣ in a). In b, a continuous matrix (*) route exists between the endocapillary compartment limited by a bare area of GBM (➙) and the periglomerular interstitium. BS: Bowman’s space.

ible with an epithelial origin; however, junctional complexes were rare in Bowman’s space. It must be stressed that immu- could not be detected. Crescents often contained accumulations noreactive cells in the extracapillary compartment are not of cell debris, generally in the vicinity of the PBM. Fibrin was necessarily macrophages, given that podocytes in injured glo- not detected in crescents, neither by TEM nor by immunoflu- meruli may express macrophage markers (15). In contrast to orescence. The cells were separated from each other by abun- bridging cells and podocytes on the tuft, the vast majority of dant basement membrane–like material that mostly was in cells within a crescent were immunonegative for WT-1, syn- ␣ continuity with the PBM (Figure 8) and showed immunoreac- aptopodin, CD2AP, 3 integrin (not shown), and podocin (Fig- tivity for collagen type IV (not shown). Because the cells of the ure 6D), which indicates that they were not podocytes. innermost layer of a crescent were also not connected by junctions, the intercellular spaces of a crescent communicated Quantification of Cellular Bridges with Bowman’s space (Figure 8). Serial sections across complete glomeruli were carried out, Immunostaining with anti-MHC class II and F4/80 antibod- to obtain insights in the frequency of cellular connections ies revealed that macrophages represented only a small fraction between the tuft and Bowman’s capsule. Fifty-nine glomeruli of cells in a crescent. Many crescents showed no immunore- were followed in three series of semithin (1 ␮m) sections, each activity at all (Figure 6F). Likewise, immunoreactive cells series derived from a different animal killed at day 10. The J Am Soc Nephrol 12: 2060–2071, 2001 Podocyte Bridges in Glomerulonephritis 2065

Figure 5. Immunofluorescence for collagen type IV (day 6). (a and b) Fluorescence and interference contrast images, respectively, of the same area. Collagen type IV is present in two connections of the tuft with the capsule (➙). There are no discontinuities in the GBM.

results (Table 1) show that cellular bridges are common before crescents have developed and are a consistent attribute of later stages, when crescents are present. Twenty-eight glomeruli in section series from two control mice did not reveal any con- nection between the tuft and Bowman’s capsule.

Periglomerular Interstitial Response Glomeruli with an apparently intact parietal epithelium, i.e., glomeruli without tuft adhesion or crescent, generally dis- played no or little periglomerular inflammation. In contrast, crescents were associated with abundant periglomerular in- flammatory cells (Figures 6F and 8). There was often a striking spatial association between a crescent and an infiltrate (Figure 6F). Focal inflammatory reactions were also often seen in the Figure 4. Podocyte bridges in series of semithin sections (day 10) periglomerular interstitium in the vicinity of podocyte bridges. ϩ (light microscopy). (a through d) Twenty-eight 1-␮m sections sepa- Infiltrates comprised macrophages (Figure 6F), T cells (CD3 , rate a from d. The arrows show cell processes of podocytes adhering not shown), and few neutrophils. In infiltrated areas, curtain- to the PBM. All capillary loops are necrotic. One loop on the right like cytoplasmic processes of fibroblasts covered the lesion side in a is filled with hyaline. (e) Bridge composed of several from outside. The crescent and the periglomerular reaction neighboring podocytes. A total of seven adjacent bridging podocytes were usually separated from each other by an intact PBM; as were found in the section series; five are seen on the micrograph. described above, disruptions of the PBM were rare. 2066 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 2060–2071, 2001

Figure 6. Immunohistochemistry for podocyte markers (a through e) and MHC class II (f) (day 6). Immunoreactive cells stain brown. (a through e) Podocytes on the tuft and the cells that connect the tuft with Bowman’s capsule express WT-1 (a), ␣3 integrin (b), synaptopodin (c), podocin (d), and CD2-AP (e), whereas parietal cell are negative. In d, an early crescent is visible. (f) The interstitium facing the crescent displays a focus of inflammation with numerous MHC class II–positive cells. The crescent itself does not show immunoreactive cells. Counterstaining with hematoxilin (b through e) or periodic acid–Schiff with hematoxilin (a and f).

Discussion junctions. As was shown previously in experimental models, This study dealt with the early stages of crescent formation crescentic glomerulonephritis starts as an endocapillary inflam- in a murine model of crescentic glomerulonephritis. The orga- mation (8,9,14,21). In this study, the inflammation was vividly nization of crescents was similar to that reported for other flourishing 3 d after the initiation of the disease. Crescents models of the crescentic glomerulonephritis in mice (6,16,17), were encountered after 6 d, and after 10 d they had developed as well as for early stages of the crescentic glomerulonephritis in about 40% of glomeruli. It is an open question how these in humans (1,5,18,19). Most of the cells in early crescents are two processes, endocapillary inflammation and crescent for- likely of epithelial origin, derived from proliferation of PEC mation, are related to each other. (2,3,8–11,20). They are embedded in a basement membrane– The essential new finding of this study was that podocytes like material without being joined to neighboring cells by frequently form bridges between the tuft and Bowman’s cap- J Am Soc Nephrol 12: 2060–2071, 2001 Podocyte Bridges in Glomerulonephritis 2067

taneously in inflamed glomeruli, given that we never found bare areas of PBM. The rupture of junctions in the parietal epithelium at the site of intrusion by podocytes might play a central role in the proliferation of PEC and the organization of the crescent. Molecular links between junctional complexes and the pattern of growth in epithelia have been recently disclosed. Oncogene Raf-1 induces the loss of both tight junctions and contact inhibition of growth in a parotid epithelial cell line. Introduc- tion of exogenous occludin in Raf-1–activated cells not only restores the tight junctions in monolayers but also suppresses growth in soft agar (29). Functional inhibition of the E-cad- herin/catenin complex suppresses contact inhibition of motility in a mammary carcinoma cell line (30). In Drosophila ovary epithelium, a tight molecular link has been disclosed among Figure 7. Mitotic parietal epithelial cell (day 10) (light microscopy). The mitotic figure is adjacent to a bridge that contains a total of four junctional belts, cell polarity, and cell proliferation (31). PEC bridging podocytes as identified in the series of semithin sections. adjacent to podocytes irrupting into the parietal epithelium are in a situation similar to cells at the margin of the lesion of a monolayer in in vitro models of wound healing. Their edge that faces the lesion has lost junctional contacts. This induces the sule—or, more precisely, between the GBM and the PBM. At formation of lamellipodia and a crawling behavior (32). The the PBM, the adherent podocyte cell portion is interposed decreased stability of microtubules at the noncontacting edges between PEC. Podocyte bridges were found early in the dis- probably contributes to this phenomenon (33). The cells behind ease, when crescents were not yet established, and in later the first row also participate in the generation of force for stages they were found within crescents. They may consist of closure of the defect (32), and they are induced to proliferate single cells, but frequently they were arranged in small groups. (34). In crescentic glomerulonephritis, restoration of the integ- Podocyte bridges were a very regular finding: in total, we rity of the parietal epithelium cannot be achieved, because documented more than 80 cases by photography at the light- processes of podocytes are interposed between PEC. The com- microscopic or TEM level, and we could easily have collected bined stimulation of proliferation and migration may, instead, more. A different type of bridging podocytes, so-called parietal end in crescent formation. podocytes, may be found around the vascular pole in normal Proteins required for the adhesion to basal membranes are human kidneys (22). They probably represent a displacement normally localized in the basolateral membrane domain (35). of the transition from the visceral to the parietal epithelium into Thus, the adherence of podocytes with part of their luminal Bowman’s capsule. In this study, parietal podocytes were membrane to the PBM quite obviously reflects a loss of cell ␣ observed neither in control mice nor in experimental animals. polarity. The positive staining of those cells with 3 integrin Taking podocyte bridges as essential features that underlie alongside the PBM clearly shows the erroneous sorting of a the development of crescents, a hypothetical sequence of basolateral membrane protein to a portion of the former lumi- events that leads from the endocapillary inflammation to a nal membrane. The loss of cell polarity may also explain the crescent is proposed in Figure 9. Each drawing in Figure 9 secretion of extracellular matrix material by apical membrane illustrates a situation that has been repeatedly encountered and domains of podocytes; such material is consistently encoun- documented. The endocapillary inflammation quite obviously tered in the niches between bridging podocytes. Moreover, the affects podocytes. In addition to foot process effacement fact that bridging podocytes do not establish junctions to neigh- (which is the most common change seen in podocytes in boring PEC suggests that parietal cells at those sites have also response to any challenge), podocytes exhibit “microvillous lost polarity. transformation” (23–26) to a striking degree. Even if the rele- A spatial association between a crescent and a periglomeru- vance of this type of alteration is not clear, it quite obviously lar inflammation has been observed previously (36), and it was represents extensive membrane dynamics. It is tempting to confirmed in this study. This suggests that inflammatory me- speculate that this activity enables podocytes to disjoin junc- diators from inside the glomerulus have diffused across the tions between PEC, which permits them to fix to the PBM. It PBM, triggering the interstitial response. The open intercellular is certainly possible that junctional instability provoked by spaces between bridging podocytes and PEC provide continu- inflammatory mediators facilitates that process. For instance, ous diffusional pathways from Bowman’s capsule to the peri- the proinflammatory cytokine tumor necrosis factor–␣ in- glomerular interstitium. In addition, matrix spaces between creases the paracellular permeability of epithelia and decreases bridging podocytes, starting at the GBM and extending into the the number of strands in tight junctions (27,28). Interestingly, extracellular spaces within a crescent, might represent a direct tumor necrosis factor–␣ is important for the formation of diffusion pathway from the endocapillary compartment to the crescents in anti-GBM glomerulonephritis in mice (14). It is glomerular interstitium. not likely that disruption of junctional complexes occurs spon- The effects of manipulating the activities of proteins of the 2068 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 2060–2071, 2001

Figure 8. Ultrastructure of crescentic glomeruli (day 10) (TEM). (a) Three adhesions of the tuft to the crescent are visible (circles). Aside from the adhesions, the urinary space is open, although very narrow. There is a conspicuous interstitial inflammation all around the glomerulus. (b) A podocyte is partly embedded in the crescent and thus mediates a tuft adhesion. The cells in the crescents are not linked by junctions. The extracellular matrix consists of sheaths of basement membrane–like material (*). It communicates with Bowman’s space (➙) and with the PBM (•). a, afferent arteriole; Po, podocyte.

Table 1. Frequency of cellular connections of the tuft with of fibrin deposition (39,40), the role of fibrin itself in glomer- either BC or a crescenta ulonephritis has still to be defined. In animal models (8,9,37,41,42) and in human crescentic glomerulonephritis Glomeruli (n ϭ 59) Frequency (%) (18,43), necrotizing glomerular lesions, together with breaks in With crescent 22 (37) the GBM and exocapillary fibrin deposits, have been observed with connection between tuft and 22 (100) early in the disease course. In contrast, in this study and a crescent previous study of murine crescentic glomerulonephritis (6), with cell bridge between tuft and BC 20 (90) fibrin was found neither in Bowman’s space nor in crescents outside crescent but was confined to the endocapillary compartment. This dis- Without crescent 37 (62.7) crepancy may be due to the fact that, in this model, the disease with cell bridge between tuft and BC 18 (48.6) advanced slowly (proteinuria did not start before day 2), with without cell bridge 19 (51.4) a low incidence of GBM breaks and an absence of loop necrosis; thus, the conditions for a leakage of fibrinogen/fibrin a ␮ Three series of 100 sections (1- m epon sections), from the into Bowman’s space were simply not established. Neverthe- three animals killed at day 10, have been evaluated. These series contained 59 complete glomeruli. In addition, a large number of less, crescents did develop in these mice, which suggests that cell bridges between the tuft and Bowman’s capsule (BC) or a fibrin deposition in Bowman’s space is not necessarily a pre- crescent were encountered by light and transmission electron requisite for their development. In this respect, the rather slow microscopy in glomeruli of all experimental animals. and mild evolution of the disease with preservation of the glomerular architecture in early stages of crescent formation provides an opportunity to dissociate the relevance of the coagulation system in experimental anti-GBM glomerulone- various exocapillary factors associated with crescent phritis have been considered an evidence for a stimulatory role formation. of fibrin in the formation of crescents (37,38). However, be- In two previous studies on anti-GBM glomerulonephritis in cause those proteins can promote inflammation independently mice, macrophages could not be identified in crescents (6,36). J Am Soc Nephrol 12: 2060–2071, 2001 Podocyte Bridges in Glomerulonephritis 2069

Figure 9. Development of a crescent shown in schematic drawings. In (a through d), the right half of the glomerulus shows the morphology of a healthy glomerulus; the left half shows pathological alterations observed in anti-GBM glomerulonephritis. Only alterations shown in a were found on day 3. All alterations shown in Panels a through d could be found on days 6 and 10. (a) Inflammatory cells (violet) infiltrate the tuft. The endothelium (purple) in two capillary loops has detached from the GBM. A cell process of a podocyte has intruded between PEC and established a bridge between the GBM and the PBM (basement membranes appear as black lines). A diffusion pathway exists from Bowman’s space toward the periglomerular interstitium, where an interstitial infiltration (green) composed of fibroblasts and inflammatory cells starts to develop. (b) The podocyte bridge involves two cells. The periglomerular inflammation is conspicuous. A mitotic figure is seen in a PEC. (c) A continuous network of basement membrane–like extracellular matrix (yellow) fills the space between bridging podocytes, between bridging podocytes and PEC, and between processes of PEC that have lost contact the PBM. The rupture of junctions between foot processes opens a matrix route from the endocapillary compartment toward the periglomerular interstitium. (d) The lesion displays the characteristic aspect of an early crescent.

In this study, the incidence of cells labeled with F4/80 or rare. Thus, accumulation of macrophages in Bowman’s space anti–MHC class II was very low in most crescents and in does not appear to be at the origin of crescents in this model. Bowman’s space. Also, we found by TEM that cells with a Because macrophages infiltrated the tuft and the periglomeru- morphology suggestive of a macrophage in crescents were lar interstitium, the GBM and the PBM appear to efficiently 2070 Journal of the American Society of Nephrology J Am Soc Nephrol 12: 2060–2071, 2001 limit the access of inflammatory cells to Bowman’s space. 9. Silva FG, Hoyer JR, Pirani CL: Sequential studies of glomerular Actually, breaks in either of the two basement membranes crescent formation in rats with antiglomerular membrane-in- were rare in this study. It is likely that high incidences of duced glomerulonephritis and the role of coagulation factors. Lab macrophages in crescents, as often reported in human crescen- Invest 51: 404–415, 1984 tic glomerulonephritis, reflects the rupture of basement mem- 10. Sterzel RB, Pabst R: The temporal relationship between glomer- branes in advanced stages of the disease (5). This may apply ular cell proliferation and monocyte infiltration in experimental glomerulonephritis. Virchovs Arch (Cell Pathol) 38: 337–350, for fibrin as well. 1982 It is obvious that the proposed role of podocytes in crescent 11. Wheeler J, Robertson H, Morley AR, Appleton DR: Anti-glo- formation is, to some extent, the reverse of what is seen in the merular basement membrane glomerulonephritis in the mouse: formation of tuft-to-capsule adhesion in classic focal segmental BrdU-labelling indices and histological damages. Int J Exp glomerulosclerosis (44,45). In focal segmental glomeruloscle- Pathol 74: 9–19, 1993 rosis, the degeneration and finally the loss of podocytes rep- 12. Holdsworth SR, Kitching AR, Tipping PG: Th1 and Th2 helper resent the decisive event that permits the affixation of parietal cell subsets affect patterns of injury and outcomes in glomeru- cells to a naked site of GBM, whereas, in crescentic glomer- lonephritis. Kidney Int 55: 1198–1216, 1999 ulonephritis, the podocytes seem to be in an hyperactive state, 13. Schatzmann U, Haas C, Le Hir M: A Th1 response is essential which allows them to break through the parietal epithelium and for induction of crescentic glomerulonephritis in mice. Kidney to attach to the parietal basement membrane. There is no Blood Press Res 22: 135–139, 1999 indication of a loss of podocytes in early stages of crescentic 14. Le Hir M, Haas C, Marino M, Ryffel B: Prevention of glomer- glomerulonephritis. ulonephritis induced by anti-glomerular membrane (GBM) anti- This study has uncovered a novel type of lesion in crescentic body in TNF deficient mice. Lab Invest 78: 1625–1631, 1998 15. Barie´ty J, Nochy D, Mandet C, Jacquot C, Glotz D, Meyrier A: glomerulonephritis. In response to an endocapillary inflamma- Podocytes undergo phenotypic changes and express macrophag- tion, podocytes appear to undergo specific cellular changes that ic-associated markers in idiopathic collapsing glomerulopathy. cause them to break through the parietal epithelium and to Kidney Int 53: 918–925, 1998 attach to the PBM. This represents a potentially important 16. Wheeler J, Morley AR, Appleton DR: Anti-glomerular mem- pathogenic step in the development of a crescent. brane (GBM) glomerulonephritis in the mouse: Development of disease and cell proliferation. J Exp Pathol 71: 411–422, 1990 Acknowledgments 17. Kim Y-G, Alpers CE, Brugarolas J, Johnson RJ, Couser WG, This study was supported by the Deutsche Forschungsgemein- Shankland SJ: The cycline kinase inhibitor p21CIP1/WAF1 lim- schaft, Grant Kr546/9 to 3 and by the Hartmann Mu¨ller Stiftung. We its glomerular epithelial cell proliferation in experimental glo- would like to thank Rolf Nonnenmacher for excellent artwork and merulonephritis. Kidney Int 55: 2349–2361, 1999 Michele Enderlin and Myriam Erni for skilled technical assistance. 18. Morita T, Suzuki Y, Churg J: Structure and development of the References glomerular crescent. Am J Pathol 72: 349–368, 1973 19. Morel-Maroger Striker L, Killen PD, Chi E, Striker GE: The 1. Jennette JC, Hipp CG: The epithelial antigen phenotype of glo- composition of glomerulosclerosis. I. Studies in focal sclerosis, merular crescent cells. Am J Clin Pathol 86: 274–280, 1986 crescentic glomerulonephritis, and membranoproliferative glo- 2. Guettier C, Nochy D, Jacquot C, Mandet C, Camilleri JP, Bariety merulonephritis. Lab Invest 51: 181–192, 1984 J: Immunohistochemical demonstration of parietal epithelial 20. Nitta K, Horita S, Honda K, Uchida K, Watanabe T, Nihei H, cells and macrophages in human proliferative extra-capillary Nagata M: Glomerular expression of cell-cycle-regulatory pro- lesions. Virchows Arch (Pathol Anat) 409: 739–748, 1986 3. Nagata M, Horita S, Shu Y, Shibata S, Hattori M, Ito K, Wa- teins in human crescentic glomerulonephritis. Virchows Arch tanabe T: Phenotypic characteristics and cyclin-dependent kinase 435: 422–427, 1999 inhibitors repression in hyperplastic epithelial pathology in idio- 21. Huang X-R, Tipping PG, Shuo L, Holdsworth SR: Th1 respon- pathic focal segmental glomerulosclerosis. Lab Invest 80: 869– siveness to nephritogenic antigens determines susceptibility to 880, 2000 crescentic glomerulonephritis. Kidney Int 51: 94–103, 1997 4. Magil AB: Histogenesis of glomerular crescents. Am J Pathol 22. Gibson IW, Downie TT, More IAR, Lindop GBM: Tuft-to- 120: 222–229, 1985 capsule adhesions and their precursors: Difference between the 5. Boucher A, Droz D, Adafer E, Noel L-H: Relationship between vascular and tubular poles of the human glomerulus. J Pathol the integrity of Bowman’s capsule and the composition of cel- 184: 430–435, 1998 lular crescents in human crescentic glomerulonephritis. Lab In- 23. Nakahama H, Obata K, Sugita M, Horio M, Oka K, Moriyama T: vest 56: 526–533, 1987 Effect of FK506 in the treatment of autoimmune glomerulone- 6. Ophascharoensuk V, Pippin J, Gordon KL, Shankland SJ, Couser phritis in Brown Norway rats. Nephron 81: 421–427, 1999 WG, Johnson RJ: Role of intrinsic versus infiltrating cells in 24. Wang JS, Yang AH, Chen SM, Young TK, Chiang H, Liu HC: glomerular crescent formation. Kidney Int 54: 416–425, 1998 Amelioration of antioxidant enzyme suppression and proteinuria 7. Lan HY, Nikolic-Paterson DJ, Atkins RC: Involvement of acti- in cyclosporin-treated puromycin nephrosis. Nephron 65: 418– vated periglomerular leukocytes in the rupture of Bowman’s 425, 1993 capsule and glomerular crescent progression in experimental 25. Mendrick DL, Rennke HG: Induction of proteinuria in the rat by glomerulonephritis. Lab Invest 67: 743–751, 1992 a monoclonal antibody against SGP-115/107. Kidney Int 33: 8. Cattell V, Jamieson SW: The origin of glomerular crescents in 818–830, 1988 experimental nephrotoxic serum nephritis in the rabbit. Lab 26. Matsui K, Breiteneder-Geleff S, Kerjaschki D: Epitope-specific Invest 39: 584–590, 1978 antibodies to the 43-kD glomerular membrane protein podopla- J Am Soc Nephrol 12: 2060–2071, 2001 Podocyte Bridges in Glomerulonephritis 2071

nin causes proteinuria and rapid flattening of podocytes. JAm tissue and in experimental glomerulonephritis. Am J Pathol 156: Soc Nephrol 9: 2013–2026, 1998 1749–1765, 2000 27. Schmitz H, Fromm M, Bentzel CJ, Scholz P, Detjen K, Mankertz 36. Robertson H, Wheeler J, Morley AR: Anti-glomerular basement J, Bode H, Epple H-J, Riecken E-O, Schulzke J-D: Tumor membrane glomerulonephritis in the mouse: The role of macro- necrosis factor-alpha regulates the epithelial barrier in the human phages. Int J Exp Pathol 76: 157–162, 1995 epithelial cell line HT-29/B6. J Cell Sci 112: 137–146, 1999 37. Vassali P, McCluskey RT: The pathogenic role of the coagula- 28. Marano CW, Lewis SA, Garulacan LA, Soler AP, Mullin JM: tion process in rabbit Masugi nephritis. Am J Pathol 45: 653– Tumor necrosis factor-alpha increases sodium and chloride con- 677, 1964 ductance across the tight junction of CACO-2 BBE, a human 38. Kitching AR, Holdsworth SR, PloPlis VA, Plow EF, Collen D, intestinal cell line. J Membr Biol 161: 263–274, 1995 Carmeliet P, Tipping P: Plasminogen and plasminogen activators 29. Li D, Mrsny RJ: Oncogenic Raf-1 disrupts epithelial tight junc- protect against renal injury in crescentic glomerulonephritis. tions via downregulation of occludin. J Cell Biol 148: 791–800, J Exp Med 185: 963–968, 1997 2000 39. Cunningham MA, Rondeau E, Chen X, Coughlin SR, Hold- 30. Bracke ME, Depypere H, Labit C, Van Mark V, Vennekens K, sworth SR, Tipping PG: Protease-activated receptor 1 mediates Vermeulen SJ, Maelfait I, Philippe J, Serreyn R, Mareel MM: thrombin-dependent, cell-mediated renal inflammation in cres- Functional downregulation of E-caderin/catenin complex leads centic glomerulonephritis. J Exp Med 191: 455–461, 2000 to loss of contact inhibition of motility and of mitochondrial 40. Chambers RC, Leoni P, Blanc-Brude OP, Wembridge DE, Lau- activity, but not of growth in confluent epithelial cell cultures. rent GJ: Thrombin is a potent inducer of connective tissue Eur J Cell Biol 74: 342–349, 1997 growth factor via proteolytic activation of protease-activated 31. Bilder D, Li Min, Perrimon N: Cooperative regulation of cell receptor-1. J Biol Chem 45: 35584–35591, 2000 polarity and growth by drosophila tumor suppressors. Science 41. Tam FWK, Smith J, Morel D, Karkar AM, Thompson EM, Cook 289: 113–116, 2000 T, Pusey CD: Development of scarring and renal failure in a rat 32. Fenteany G, Janmey PA, Stossel TP: Signaling pathways and cell model of crescentic glomerulonephritis. Nephrol Dial Transplant mechanics involved in wound closure by epithelial cell sheets. 14: 1658–1666, 1999 Curr Biol 10: 831–838, 2000 42. Horio T: Development and fate of crescentic and granulomatous 33. Waterman-Storer CM, Salmon W, Salmon ED: Feedback inter- lesions in rat Masugi nephritis. Pathol Int 51: 72–81, 2001 actions between cell-cell adherens junctions and cytoskeletal 43. Min KW, Gyo¨rkey F, Gyo¨rgey P, Yium JJ, Eknoyan G: The dynamics in newt lung epithelial cells. Mol Biol Cell 11: 2471– morphogenesis of glomerular crescents in rapidly progressing 2483, 2000 glomerulonephritis. Kidney Int 5: 47–56, 1974 34. Providence KM, Kutz SM, Staiano-Coico L, Higgins P: PAI-1 44. Kriz W, Elger M, Nagata M, Kretzler M, Uiker S, Koeppen- Gene expression is regionally induced in wounded epithelial cell Hagemann I, Tenschert S, Lemley KV: The role of podocytes in monolayers and required for injury repair. J Cell Physiol 182: the development of glomerular sclerosis. Kidney Int 45 supple- 269–280, 2000 ment 42: S-64–S-72, 1994 35. Raats CJI, van den Born J, Bakker MAH, Oppers-Walgreen B, 45. Kriz W, Gretz N, Lemley KV: Progression of glomerular Pisa BJM, Dijkman HBPM, Assman KJM, Berden JHM: Ex- diseases: Is the podocyte the culprit? Kidney Int 54: 687– 697, pression of agrin, dystroglycan, and utrophin in normal renal 1998

Access to UpToDate on-line is available for additional clinical information at http://www.jasn.org/