CD2AP is expressed with nephrin in developing podocytes and is found widely in mature kidney and elsewhere

Cong Li, Vesa Ruotsalainen, Karl Tryggvason, Andrey S. Shaw, Jeffrey H. Miner


CD2-associated protein (CD2AP) is an adapter molecule that can bind to the cytoplasmic domain of nephrin, a component of the glomerular slit diaphragm. Mice lacking CD2AP exhibit a congenital nephrotic syndrome characterized by extensive foot process effacement, suggesting that CD2AP-nephrin interactions are critical to maintaining slit diaphragm function. We have examined the patterns of expression of both CD2AP and nephrin in developing mouse and human kidney. Both proteins were first detected in developing podocytes at the capillary loop stage of glomerulogenesis and eventually became concentrated near the glomerular basement membrane. CD2AP was also observed diffusely in collecting duct and apically in many cells of proximal and distal tubule. Kidneys fromCd2ap −/− mice initially exhibited normal nephrin localization, but as the mice aged and foot processes became effaced, nephrin disappeared. In laminin-β2 mutant mice exhibiting nephrotic syndrome, CD2AP in glomeruli was aberrantly localized in a primarily punctate pattern. Extensive extrarenal expression of CD2AP was observed in endothelial and epithelial cells, in many cases with a specific subcellular localization. Together, these results suggest that CD2AP is not only involved in maintaining the slit diaphragm but may also have a general role in maintaining specialized subcellular architecture. The severity of kidney disease in Cd2ap mutant mice may have eclipsed manifestation of defects in other tissues.

  • CD2-associated protein
  • renal development
  • nephrotic syndrome
  • glomerulus
  • nephron
  • epithelia

CD2-associated protein (CD2AP) is an 80-kDa Src homology 3 domain-containing protein. It was first identified in a yeast two-hybrid screen as binding to the cytoplasmic domain of CD2 (2), a transmembrane protein of the immunoglobulin superfamily expressed on T cells and natural killer cells. Binding of CD2AP to CD2 is thought to facilitate binding of T cells to antigen-presenting cells. CD2AP was independently identified (as CMS) in a different yeast two-hybrid screen as binding to p130Cas (7), a cytoplasmic protein that is involved in the formation of focal adhesions induced by integrin signaling (13).

CD2AP mRNA is expressed widely in both mouse and human tissues (2, 7). However, the most significant defect detected in mutant mice lacking CD2AP is in the kidney (16).Cd2ap −/− mice are a model for congenital nephrotic syndrome. They die at 6–7 wk of age from proteinuria and renal failure. Histologically, there is effacement of podocyte foot processes beginning at 1 wk of age, followed by excessive deposition of extracellular matrix by mesangial cells. This leads to a greatly expanded mesangium, distension of the glomerular capillary loops, and eventual blockage of the capillaries by matrix material. Immunohistochemical localization of CD2AP in normal kidney demonstrated that, in the glomerulus, it is restricted to podocytes in a pattern consistent with concentration in foot processes, suggesting that the mesangial expansion is secondary to a podocyte foot process defect.

Another molecule found in podocytes and associated with congenital nephrotic syndrome is nephrin. Nephrin, like CD2, is a transmembrane protein of the immunoglobulin superfamily. It is encoded by the geneNPHS1, which is mutated in congenital nephrotic syndrome of the Finnish type (6, 20). Nephrin has been localized to the glomerular slit diaphragm (3, 4, 14), suggesting that, together with the genetic data, nephrin is an essential component of this specialized structure. Analogous to its association with CD2, CD2AP can bind to the cytoplasmic tail of nephrin, suggesting that the foot process effacement in the Cd2ap −/− kidney is related to the absence of CD2AP/nephrin interactions, which may be necessary to stabilize the slit diaphragm (16).

To understand more about the biology of CD2AP and nephrin, we have determined and compared their patterns of expression in the developing mouse and human kidney. In addition, we have examined CD2AP localization in a mouse model of nephrotic syndrome and have defined sites of CD2AP expression in other tissues. Immunohistochemical localization of CD2AP to podocytes was an important step for understanding the nature of the kidney defect in Cd2apmutant mice (16), and by analogy we presume that defining its localization elsewhere may provide clues regarding other functions of this molecule. The expression pattern of CD2AP suggests that it may have a general role in the formation or maintenance of specialized subcellular architecture in diverse cell types, but particularly in epithelial cells.



Cd2ap +/− (control) and −/− tissues were obtained from mice that have been described previously (16). Fetal human kidney at 83 days gestation was obtained from the University of Washington School of Medicine, Central Laboratory for Human Embryology (Seattle, WA). Tissues were frozen in OCT compound and sectioned at 7 μm on a cryostat. Sections were fixed in 2 or 4% paraformaldehyde for 10 min, rinsed in PBS, and blocked for 30 min in 1% goat serum, 1% BSA in PBS. Primary antibody diluted in 1% BSA in PBS was applied for 1 h. Sections were rinsed in PBS, and then fluorophore-conjugated second antibody was applied for 1 h. After rinsing in PBS, sections were mounted in 90% glycerol/0.1× PBS/1 mg/ml p-phenylenediamine. Sections were viewed under epifluorescent illumination on a Nikon Eclipse 800 microscope. Images were captured with a Spot 2 cooled color digital camera (Diagnostic Instruments, Sterling Heights, MI) using Spot Software Version 2.1. Images were imported into Adobe Photoshop 5.0.2 for final processing and layout.

Antibodies and lectins.

Rabbit anti-CD2AP antiserum has been previously described (2). Rabbit anti-nephrin antiserum was generated by immunization with a bacterial fusion protein containing portions of the human nephrin cytoplasmic tail (amino acids 1084–1241 in Ref.6) and mouse dihydrofolate reductase (J. Patrakka, V. Ruotsalainen, P. Reponen, I. Ketola, C. Holmberg, M. Heikinheimo, K. Tryggvason, and H. Jalanko., unpublished observations). Mouse anti-synaptopodin monoclonal antibody (9) was a gift from Peter Mundel (Albert Einstein College of Medicine, Bronx, NY). Rat anti-mouse laminin-α1, clone 8B3 (1), was a gift from Dale Abrahamson (Univ. of Kansas Medical Center, Kansas City, KS). Rat anti-mouse laminin-γ1 (mAb-1914) and -β1(mAb-1928) were from Chemicon (Temecula, CA). Rat anti-integrin-α6, clone GoH3, and rat anti-platelet endothelial cell adhesion molecule were from Pharmingen (San Diego, CA). Rabbit anti-Tamm-Horsfall protein was from Biomedical Technologies (Stoughton, MA). FITC and Cy3-conjugated second antibodies were from ICN/Cappel (Costa Mesa, CA). FITC-conjugated Dolichos biflorus agglutinin and Lotus tetragonolobus lectin were from Vector Laboratories (Burlingame, CA).


CD2AP and nephrin expression in developing glomeruli.

The expression of both CD2AP and nephrin has been shown to be restricted to podocytes in mature mouse glomeruli. Because these molecules have been proposed to interact to stabilize the slit diaphragm, we immunostained developing kidneys to determine exactly when during development CD2AP and nephrin are first expressed in podocytes. Because both the CD2AP and the nephrin antisera we used were developed in rabbit, we could not double label the same section with the two antisera. Instead, we stained consecutive sections individually with the two antisera and doubly labeled each with a monoclonal antibody to laminin-γ1 to define the basic structure of the developing glomeruli. Some sections were costained with a monoclonal antibody to synaptopodin, an actin-associated protein that in the kidney is specific to podocytes (10). In addition, as negative controls for CD2AP staining we used age-matchedCd2ap −/− kidneys, which never showed specific reactivity with the CD2AP antiserum.

In the mouse, development of the definitive kidney begins on embryonicday 11 and continues for 2–3 wk after birth. Because of the nature of the process, new nephrons are induced as others are maturing, and a newborn kidney exhibits all the stages of nephrogenesis (15). In newborn control kidney, both CD2AP and nephrin were first detected in developing podocytes at the capillary loop stage (Fig. 1, A-D, arrows), a stage at which the glomerular capillaries are forming and the podocyte epithelium is beginning to constrict around them. At a later stage, after glomerular constriction had occurred, both CD2AP and nephrin were concentrated basally, but staining was also observed elsewhere in the podocytes (Fig. 1, A-D). As expected, Cd2ap −/− kidney was negative for CD2AP staining, but nephrin expression and localization were indistinguishable from that seen in wild-type kidney (Fig. 1, E-H). These results show that CD2AP and nephrin are essentially coexpressed during development but that CD2AP is not necessary for stabilization or localization of nephrin in developing podocytes.

Fig. 1.

Expression of CD2-associated protein (CD2AP) and nephrin in developing mouse glomeruli. Sections of newborn control (A-D) and Cd2ap −/− (E-H) kidneys were doubly labeled with antibodies to laminin-γ1(A,C, E, G) and either CD2AP (B, F) or nephrin (D, H). In the control kidneys, both CD2AP and nephrin are initially expressed in capillary loop stage podocytes (A-D, arrows) and are also found in maturing glomeruli (g). Note in the S-shaped structure (s) in C and D, a precursor to the capillary loop stage, is negative for nephrin. In the mutant (E-H), CD2AP is undetectable in glomeruli (E,F), whereas nephrin is expressed normally in capillary loop stage (arrowheads) and maturing glomeruli (G, H). Bar, 50 μm.

Next, we immunostained 2-wk-old mouse kidneys for CD2AP and nephrin. Double labeling with the synaptopodin antibody confirmed that both proteins were restricted to podocytes in glomeruli (data not shown). Both were detected basally adjacent to the glomerular basement membrane (GBM), but there was also a diffuse labeling of the podocytes that was somewhat more extensive for nephrin (Fig.2, A and B). In theCd2ap −/− kidney, no staining was observed with the CD2AP antibody, whereas anti-nephrin staining was similar to that seen in the control kidney (Fig. 2, C and D). In adult glomeruli (7 wk of age) CD2AP was less diffuse and more basally concentrated, whereas nephrin maintained a diffuse pattern restricted to podocytes (Fig. 2, E and F). InCd2ap −/− glomeruli, CD2AP was undetectable (Fig.2 G), and the same was now true for nephrin in the great majority of glomeruli. This is probably due to the fact that by 7 wk all glomeruli were severely damaged by mesangial matrix deposition, chronic nephrotic syndrome, and widespread foot process effacement. Segmental nephrin reactivity was, however, observed in occasional glomeruli (Fig. 2 H).

Fig. 2.

Expression of CD2AP and nephrin in 2- and 7-wk-old mouse glomeruli. Sections of 2- (A-D) and 7-wk-old (E-H) kidneys from Cd2ap +/− and −/− mice were stained with CD2AP or nephrin antisera, as indicated. CD2AP is undetectable in the −/− kidneys (C and G). In the mutant, nephrin appears to be expressed normally at 2 wk (D) but is only detected segmentally in a very few glomeruli at 7 wk (H). Arrowheads in H point to podocytes lacking nephrin. Bar, 50 μm.

To determine whether these patterns of expression are a general feature of podocyte development, we immunostained sections of fetal human kidney for CD2AP and nephrin. Double labeling with the synaptopodin antibody revealed that nephrin became detectable at a slightly earlier stage than did CD2AP; synaptopodin-positive podocytes were always nephrin positive but not always CD2AP positive. However, this difference was minimal, as both CD2AP and nephrin were first detected at the capillary loop stage (Fig. 3,A-D). In maturing glomeruli, CD2AP and nephrin were detected in podocytes both adjacent to the GBM, which was labeled with an antibody to laminin-γ1, and elsewhere in podocytes, as is evident in the merged images (Fig. 3, E-J).

Fig. 3.

Expression of CD2AP and nephrin in developing human glomeruli. Sections of human kidney at 83 days gestation were doubly labeled with the indicated antibodies, paired as follows: Aand B; C and D; E-G;H-J. An antibody to synaptopodin (Synpo) was used to identify developing podocytes. Both CD2AP and nephrin were detected initially in capillary loop stage podocytes (arrowheads in Band D), but nephrin was detectable first (see text). In maturing glomeruli (E-J), CD2AP and nephrin were detected in podocytes near the laminin-γ1-positive glomerular basement membrane and elsewhere in podocytes (G andJ). Bar, 50 μm.

Localization of CD2AP in Lamb2 mutant glomeruli.

Because we hypothesize that CD2AP is involved in maintaining the structure of the slit diaphragm, we looked at its localization in glomeruli of Lamb2 mutant mice. These mice lack the laminin-β2 chain, a major component of the GBM and the neuromuscular synaptic basement membrane. They initially present with proteinuria at 8 days of age, exhibit extensive foot process effacement by 15 days of age, and die from renal and neuromuscular defects at 3–5 wk of age (11, 12). At 9 days of age, when there is little ultrastructural podocyte damage, no differences in CD2AP localization were noted between control and mutant mice (data not shown). However, at 21 days of age, CD2AP was found in a primarily punctate distribution in most mutant glomeruli, a pattern very different from that seen in control mice (Fig.4). Thus CD2AP is aberrantly localized in podocytes exhibiting or undergoing foot process effacement.

Fig. 4.

Aberrant localization of CD2AP in Lamb2 mutant glomeruli. Sections of 3-wk-old kidney from a control and a mutant lacking the laminin-β2 chain were stained with the CD2AP antiserum. Much of the CD2AP immunoreactivity is punctate in the mutant (B and D) but not in the control (Aand C). The mutant exhibits nephrotic syndrome and extensive foot process effacement at this age. Bar: 33 μm forA and B; 16 μm for C andD.

Extraglomerular expression of CD2AP in mouse kidney.

We previously reported that CD2AP was expressed in a subset of tubules (16). Here, we have explored this issue in more detail by using markers to identify tubular segments. These includedD. biflorus agglutinin for ureteric bud and its collecting duct derivatives, laminin-α1 antibody, and L. tetragonolobus agglutinin for proximal tubules, and Tamm-Horsfall antiserum for thick ascending limb and segments of distal tubule (Fig.5 and data not shown). At birth, CD2AP was detected strongly in the cortical ureteric bud epithelium, although expression was weak in the distal branching tips (Fig. 5, Aand B). The medullary collecting ducts also exhibited robust CD2AP expression (Fig. 5 C); this expression was consistently diffuse within these epithelial cells.

Fig. 5.

Extraglomerular expression of CD2AP in mouse kidney. Sections of newborn (A-C) and 2-wk-old (D-I) kidneys were labeled with the indicated antibodies or DBA lectin.A and B: in the nephrogenic zone, CD2AP is expressed in the Dolichos biflorus agglutinin (DBA)-positive ureteric bud, but staining is weak at the branching tip (t).C: medullary collecting ducts are strongly positive for CD2AP. D-I: in cortex, CD2AP is concentrated apically in cells of some proximal tubules (which are laminin-α1positive; arrowheads) and distal tubules (laminin-α1negative; arrows) but is mostly diffuse in collecting duct cells (*). Bar: 100 μm for C; 12.5 μm for I; 50 μm for all other panels.

In the 2-wk-old kidney, CD2AP was expressed in proximal and distal tubules where it was concentrated apically (Fig. 5, D-I). In collecting ducts, CD2AP showed a diffuse cytoplasmic localization, although there was also evidence of apical concentration in some segments (Fig. 5, D-F). The pattern at 7 wk of age did not differ significantly from that seen at 2 wk (data not shown). Thus at maturity CD2AP is highly expressed in collecting duct epithelial cells and is concentrated apically in many but not all tubular epithelial cells of the nephron.

Expression of CD2AP in other tissues.

We surveyed an assortment of mouse tissues at 1 mo of age for expression of CD2AP (Fig. 6, Table1). We used tissues from aCd2ap −/− mouse as negative controls, and tissues were doubly labeled with appropriate antibodies to aid in identification of the cell types expressing CD2AP. In small intestine, CD2AP was expressed by enterocytes along villi, and the protein was concentrated basally and laterally, with some protein also detected at the apical membrane facing the intestinal lumen [Fig. 6, A andB, *]. In pancreas, CD2AP was present apically in acinar cells and diffusely and abundantly in ductal cells (Fig. 6,C and D), including those comprising the main pancreatic duct (Table 1). In submandibular salivary gland, the secretory acini of the mucous component had the highest level of expression. These exhibited both apical staining (similar to pancreatic acini) and lateral staining at epithelial cell borders (Table 1). The serous component of the gland showed only low-level diffuse staining for CD2AP.

Fig. 6.

Expression of CD2AP in various mouse tissues. Sections of 1-mo-old tissues were doubly labeled with CD2AP and the indicated antibodies;right, merged images. A and B: in small intestine, CD2AP is expressed basally, laterally, and apically by the enterocytes on the luminal surface of villi; integrin-α6 is detected basally and, to a lesser extent, laterally on enterocytes. *, Intestinal lumen. C andD: in pancreas, CD2AP is detected apically on acinar cells (arrowheads) and in cells of the pancreatic ducts (d). Integrin-α6 is detected basally on acinar cells. Eand F: in brain, CD2AP expression is associated with capillaries, which are bound by a basement membrane containing laminin-β1. G and H: in lung, CD2AP is concentrated apically on bronchial epithelial cells (be) and is associated with capillaries (arrows) that supply the smooth muscle of the pulmonary artery (pa) and are platelet endothelial cell adhesion molecule positive. Bar, 50 μm.

View this table:
Table 1.

Expression of CD2AP in adult mouse tissues

In brain and heart muscle, CD2AP was restricted to capillaries, as shown by doubly labeling with an antibody to laminin-β1in brain (Fig. 6, E and F) and to platelet endothelial cell adhesion molecule in heart (data not shown). Liver exhibited no localized CD2AP staining, only a general, diffuse signal that was somewhat brighter than that observed for the Cd2ap−/− liver. Similarly, the lung parenchyma was modestly brighter in the control compared with the mutant (Table 1) mice. In addition, the apical side of bronchial epithelial cells and capillaries within the medial layer of the pulmonary artery were strongly positive for CD2AP (Fig. 6, G and H), as were the mesothelial cells that comprise the outer lining of the lung lobes (Table 1).


Podocytes are unique cells with both epithelial and mesenchymal characteristics. They produce a specialized repertoire of proteins, which enable them to assist in establishing and maintaining the glomerular filtration barrier (17). Here, we have compared and contrasted the patterns of expression of two proteins critical to proper podocyte function: CD2AP and nephrin. These two proteins have been proposed to interact with each other within podocyte foot processes to stabilize the slit diaphragm (16). The staining patterns of both proteins are consistent with the idea that they are associated in podocytes. Both nephrin and CD2AP were concentrated along the GBM and were also present abundantly within the cytoplasm of podocytes. Furthermore, we now have evidence for a direct interaction between CD2AP and nephrin in vivo and have mapped the sites necessary for this interaction (N. Y. Shih, J. H. Miner, and A. S. Shaw, unpublished observations).

Mice lacking CD2AP exhibit apparently normal foot processes that then become effaced, suggesting that foot processes and their slit diaphragms can form via nephrin interactions independently of CD2AP. Thus we might have found the onset of CD2AP expression in podocytes to be significantly delayed relative to that of nephrin. On the contrary, we found that in both mouse and human kidney the two proteins began to accumulate at approximately the same time in developing glomeruli, at the capillary loop stage (Figs. 1 and 3). Also, first expressed at the capillary loop stage is another component of foot processes, synaptopodin (9). In addition, progression through this stage is accompanied by developmental transitions in GBM composition, with a switch from immature to mature isoforms of laminin and type IV collagen (8). It is presumed that changes in podocyte gene expression are at least partly responsible for these transitions. Thus the capillary loop stage is a critical period in podocyte development, in terms of both alterations in morphology and the onset of a new program of gene expression that leads to production of proteins necessary for glomerular maturation and function.

The expression pattern of nephrin during development reported here for both human and mouse is quite similar to that previously reported for mouse (4). However, an earlier developmental study of a podocyte molecule, the 5-1-6 antigen (5), which was recently shown to be nephrin (19), demonstrated that the onset of nephrin expression in rat kidney occurred in presumptive podocytes at the S-shaped stage rather than at the capillary loop stage. We are unsure of the reason for this discrepancy, but it may be attributable to species-specific differences in regulation of nephrin expression or its accumulation.

In young Cd2ap mutant mice, we did not observe any alterations in the general level or localization of nephrin within podocytes (Figs. 1 and 2), suggesting that CD2AP-nephrin interactions are not necessary for accumulation of nephrin. Only in mice with severely damaged glomeruli did we see alterations in levels of nephrin: at 7 wk, all but a small fraction of Cd2ap −/− glomeruli had lost reactivity with the nephrin antibody (Fig. 3), although podocytes could still be labeled with the synaptopodin antibody (data not shown). This loss of nephrin correlates with the loss of slit diaphragms that accompanies foot process effacement.

In the Lamb2 mutant mouse model of nephrotic syndrome (12), CD2AP was found to be aberrantly clustered in podocytes at 3 wk of age (Fig. 4), when extensive foot process effacement had occurred. No such clustering was evident at 9 days of age, when foot processes appear normal (data not shown). Aside from theCd2ap mutant, this is the first reported correlation of aberrant localization of CD2AP with nephrotic syndrome and foot process effacement. This provides further support for CD2AP having a role in maintaining the slit diaphragm.

We found that CD2AP expression in the kidney was not confined to podocytes (Fig. 5). Inasmuch as fluorescence intensity is somewhat representative of level of expression, CD2AP accumulated to highest levels in a diffuse cytoplasmic pattern in epithelial cells of the collecting duct, whereas there was a lower level, primarily apical, distribution in segments of both proximal and distal tubular epithelium. Furthermore, the findings that CD2AP mRNA is present in most tissues (2, 7) were mirrored by our immunohistological data (Fig. 6, Table 1). We detected CD2AP within a diverse group of cell types in an assortment of tissues, but the majority of positive cells were either epithelial or endothelial. Moreover, in most cases CD2AP was specifically localized within cells, suggesting that, as in podocytes and T cells, it has a role in establishing or maintaining subcellular architecture in other cell types.

The notion that CD2AP plays roles in diverse tissues is somewhat at odds with the restricted phenotypes observed in Cd2ap −/− mice. Defects were only detected in glomeruli and T cells (16). However, several possibilities could explain this. First, there may be another adapter molecule in other tissues that can compensate for the absence of CD2AP. In fact, a CD2AP-related molecule was recently cloned and has a broad pattern of expression (18). Second, there may be subtle defects in other tissues that have not yet been detected due to the overwhelming nature of the kidney defects. Third, the mutant mice die at 7 wk of age, and defects in other tissues may need more time to become apparent. In support of the latter possibility, podocyte foot processes appear normal and function normally at birth; only after a few days is podocyte foot process effacement apparent, and proteinuria does not develop until 10 days after birth. It should be possible to generate conditional or tissue-restricted knockouts in which expression of CD2AP in podocytes is not adversely affected. These mice would therefore not succumb to renal defects and might reach an age at which defects are manifested in other tissues.


We thank Peter Mundel, Dale Abrahamson, Chris Sorenson, Susie Mudd, and Robert Senior for generously providing antibodies and reagents.


  • This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant R01 DK-53196 (to J. H. Miner).

  • Address for reprint requests and other correspondence: J. H. Miner, Renal Div., Box 8126, Washington Univ. School of Medicine, 660 South Euclid Ave., St. Louis, MO 63110 (E-mail:minerj{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract