A novel role for the adaptor molecule CD2-associated protein in transforming growth factor-beta-induced apoptosis.

CD2-associated protein (CD2AP) is an adaptor molecule involved in T cell receptor signaling and podocyte homeostasis. CD2AP-deficient mice develop nephrotic syndrome and renal failure caused by glomerulosclerosis. Here we report that increased transforming growth factor-beta1 (TGF-beta1) expression and apoptosis were present in podocytes at the onset of albuminuria and were followed by depletion of podocytes associated with progressive focal-segmental glomerulosclerosis in CD2AP-/- mice. Conditionally immortalized podocytes derived from CD2AP-/- mice were more susceptible to TGF-beta-induced apoptosis compared with CD2AP+/+ podocytes. Reconstitution of CD2AP rescued CD2AP-/- podocytes from TGF-beta-induced apoptosis. CD2AP was required for early activation of anti-apoptotic phosphatidylinositol 3-kinase (PI3K)/AKT and extracellular signal-regulated kinase 1/2 by TGF-beta. In contrast, activation of pro-apoptotic p38 MAPK by TGF-beta was accelerated and enhanced in the absence of CD2AP. CD2AP was not required for PI3K/AKT activation by insulin and epidermal growth factor, indicating that CD2AP is a selective mediator of anti-apoptotic TGF-beta signaling. In summary, we identified CD2AP as a novel mediator for selective activation of survival pathways and repression of apoptosis signaling by TGF-beta in podocytes. Together, our in vitro and in vivo findings suggest that TGF-beta-induced podocyte apoptosis is an early pathomechanism in mice developing focal-segmental glomerulosclerosis associated with functional impairment of CD2AP.

The transforming growth factor-␤ (TGF-␤) 1 superfamily consists of secreted peptides, of which the three TGF-␤ isoforms (TGF-␤1-3), activins, and bone morphogenetic proteins are best known in mammalian development, homeostasis, and pathobiology. The TGF-␤ isoforms are widely expressed and act on virtually every cell type in mammals by engaging a ubiquitous intracellular signaling cascade of Smad family proteins through ligand-induced activation of heteromeric transmembrane TGF-␤ receptor kinases. Receptor-activated Smad protein complexes accumulate in the nucleus where they participate directly in transcriptional activation of target genes. In addition, TGF-␤ receptors can activate Smad-independent signaling mechanisms, including mitogen-activated protein kinases (MAPKs), and PI3K (1,2). However, molecular mechanisms of activation of Smad-independent pathways by TGF-␤ receptors remain unclear.
The mouse CD2 receptor-associated protein (CD2AP) and its human orthologue Cas ligand with multiple SH3 domains (CMS) belong to a family of ubiquitously expressed adaptor molecules that also includes the human Cbl-interacting protein of 85 kDa (CIN85) and its rat and mouse orthologue regulator of ubiquitous kinase (Ruk) and SH3 domain-containing gene expressed in tumorigenic astrocytes (SETA) (for review, see Ref. 3). These proteins are defined as cytoplasmic adaptor or scaffolding proteins by three N-terminal SH3 domains, a proline-rich domain, and a C-terminal coiled-coil domain. CIN85/ CMS family proteins have been shown to interact with prolinerich regions present in a variety of signaling proteins, including Cbl oncogene, CD2 receptor, AIP1 apoptosis-inducing protein-1, and the p85 subunit of phosphatidylinositol 3-kinase (PI3K), mediated through the SH3 domains, and with focal adhesion kinase p130Cas, Src family kinases Fyn, Src, and Yes, Grb2, and endophilins, mediated through the proline-rich region (for review, see Ref. 3). Other structural features include FXDXF motifs, mediating interactions with endocytic proteins, and actin binding motifs, mediating interactions with actin cytoskeleton. Based on these observations, CIN85/CMS proteins are thought to exert multiple and diverse signaling functions in the organization of the immunological synapse in T cells, endocytosis and signaling of receptor tyrosine kinases, and neuronal apoptosis (3).
Degeneration of differentiated glomerular, tubular, and vascular cells in the normal nephron is a hallmark of chronic progressive kidney disease along with readily apparent renal scarring (4 -6). Sclerosing glomeruli are characterized by progressive depletion of podocytes, resulting in denuded glomerular basement membrane areas and tuft adhesions, which may be considered as initial lesions of irreversible glomerular injury (4,7). Indeed, recent studies demonstrate that podocyte numbers are reduced early in glomeruli of diabetic patients with nephropathy (8,9). We have shown that TGF-␤1 induces apoptosis in podocytes in vitro and in vivo via a p38 MAPK and caspase-dependent mechanism. We further demonstrated in a TGF-␤1 transgenic mouse model of glomerulosclerosis that podocyte apoptosis is an early glomerular phenotype leading to progressive podocyte depletion. Together, our studies suggest that TGF-␤ may induce podocyte apoptosis and depletion in glomerulosclerosis (10).
An important functional role for CMS/CD2AP in the kidney was suggested initially by phenotype features in CD2AP knockout mice (CD2APϪ/Ϫ), which manifest high grade albuminuria at a young age followed by progressive glomerulosclerosis and tubulointerstitial fibrosis and death from renal failure (11).
In the present study we describe a novel and essential role for CD2AP in TGF-␤ signaling in podocytes. We demonstrate that CD2AP is required for rapid activation of PI3K and ERK MAPK pathways by TGF-␤. Failure of TGF-␤ receptors to engage these anti-apoptotic pathways early in the absence of CD2AP is associated with hyperactivation of proapoptotic p38 MAPK and accelerated apoptosis in podocytes in vitro. Finally, rates of apoptosis and expression of TGF-␤1 were greatly increased, specifically in podocytes in the kidneys of CD2APϪ/Ϫ mice, coincident with the onset of albuminuria, and this preceded podocyte depletion and glomerulosclerosis.

EXPERIMENTAL PROCEDURES
Mice-CD2AP knockout mice in mixed C57Bl6/129J background were described previously (11). Kidney tissue was harvested and either immersion-fixed in formaldehyde for paraffin embedding or embedded in O.C.T. compound (Sakura Finetek, Torrance, CA) for frozen sections. Animal protocols and procedures were reviewed for ethical and humane standards and approved by an institutional Animal Use Committee.
Cell Culture-Conditionally immortalized podocytes were derived from CD2APϪ/Ϫ mice and CD2APϩ/ϩ littermates and cultured as previously reported (12). In brief, podocytes were grown on collagen type I (BD Biosciences) at 33°C with interferon-␥ (10 units/ml; Invitrogen) to drive expression of thermosensitive SV40 large T antigen. To induce differentiation, podocytes were maintained at 37°C without interferon-␥ for 14 days.
Immunostaining in Situ and in Vitro-For in situ detection of proteins we used the following antibodies: monoclonal anti-synaptopodin antibody (14), rabbit polyclonal anti-WT-1 antibody (Santa Cruz Biotechnology, St. Cruz, CA), and anti-active TGF-␤1 antibody (LC 1-30-1) (a gift from Anita Roberts, National Cancer Institute). The samples were evaluated on a Bio-Rad MR600 confocal microscope, and images were captured using a Kanton Electronic Prog/Res/3012 digital video camera and digitally processed using Adobe Photoshop 5.0.2 (Adobe Systems, Inc., San Jose, CA).
Scoring of Apoptotic Nuclei in Situ-To detect DNA fragmentation in situ, apoptotic nuclei were detected by a terminal dUTP nick-end labeling assay of kidney sections using peroxidase and counterstaining with hematoxylin and periodic acid-Schiff stain, as described earlier (10). The terminal dUTP nick-end labeling assay was used according to the manufacturer's protocol (Intergen Co., Purchase, NY). Positive cells were counted as podocytes when residing on the outer aspect of periodic acid-Schiff-positive basement membrane. Cells residing in areas circumscribed by periodic acid-Schiff-positive basement membrane were counted as endocapillary/mesangial cells. Cells lining the inner surface of the Bowman's capsule were counted as parietal epithelial cells.
Quantification of Apoptotic Nuclei in Vitro-Podocytes were seeded on glass coverslips and cultured at 33°C. After 24 h of serum starvation in 0.2% fetal bovine serum, cells were pretreated with chemical inhibitors as indicated and treated with or without 5 ng/ml TGF-␤ for 24 or 48 h as indicated before fixation in 2% paraformaldehyde. Nuclei were stained with 4Ј,6-diamidino-2-phenylindole (DAPI) (Sigma), and coverslips were mounted on slides. Apoptotic nuclei were counted in 20 fields per sample by a person blinded to the experimental conditions. Numbers were normalized for the total cell number per high power field. For reconstitution of CD2AP, cells were cotransfected with pcDNA3.1-myc-CD2AP or empty control vector pcDNA3.1 together with pEGFP (Clontech Laboratories Inc., Palo Alto, CA) in a ratio of 5:1 using Effectene transfection reagent (Qiagen Inc., Valencia, CA). Transfection efficiency was verified after 24 h by exam-ination of enhanced GFP expression. Cells were serum-starved for 12 h before stimulation with or without TGF-␤1 as indicated.
Cell Death ⌬405 Enzyme-linked Immunosorbent Assay (ELISA)-Apoptotic cell death was also quantified in vitro using the Cell Death ⌬405 ELISA (Roche Diagnostics) according to the manufacturer's protocol. In brief, 10 4 cells were lysed in the provided lysis buffer, and lysates were placed in streptavidin-coated microtiter plates. Anti-histone biotin and anti-DNA peroxidase-labeled antibodies were added. The anti-histone antibody binds to the histone components of the nucleosomes and captures the immunocomplex to streptavidin via its biotinylation. The anti-DNA peroxidase reacts with the DNA components of the nucleosomes. After incubation and washing, color was developed using the provided 2,2Ј-azino-bis(3-ethylbenzthiazoline-6sulfonic acid) substrate solution to the wells. Absorbance was measured at 405 nm.
Western Blotting-To detect levels of total protein and phosphorylated proteins, cells were lysed in radioimmune precipitation assay buffer containing protease inhibitors and phosphatase inhibitors as previously described (15). Lysates were subjected to 10% SDS-PAGE before transfer to polyvinylidene difluoride membranes (PerkinElmer Life Sciences). Western blotting was performed as previously described, and the following primary antibodies were used for antigen detection: mouse monoclonal anti-total AKT1 (Santa Cruz Biotechnology), rabbit polyclonal anti-S(P) 475 AKT, rabbit anti-phospho-Smad2, mouse antiphospho-Erk1/2, rabbit anti-total ERK1/2, rabbit anti-phospho-p38, and mouse anti-total p38 (all from Cell Signaling Inc., Beverly, MA).

Increased Expression of TGF-␤1 in Podocytes in CD2APϪ/Ϫ Mice Is Associated with Onset of Albuminuria and FSGS-We
previously reported that Smad7 and TGF-␤1 induce apoptosis in podocytes in vitro and in vivo (10). To determine whether manifestations of FSGS typically observed in CD2APϪ/Ϫ mice were associated with TGF-␤ expression profiles in glomeruli, we examined functional and histopathological manifestations of FSGS in CD2APϪ/Ϫ and CD2APϩ/ϩ littermates at 1, 2, 3, and 4 weeks of age. Immunoperoxidase labeling of kidney sections from 2-, 3-, and 4-week-old mice for TGF-␤1 protein demonstrated that podocytes in 3-week-old (n ϭ 5) (Fig. 1d) and 4-week-old CD2APϪ/Ϫ mice (n ϭ 5) (Fig. 1f) stained strongly positive for TGF-␤1, whereas 3-and 4-week-old CD2APϩ/ϩ controls (Fig. 1, c and e) and 2-week-old mice of either genotype showed no or minimal TGF-␤1 staining in podocytes ( Fig. 1, a and b). The strong increase in TGF-␤1 protein synthesis in podocytes in this model is similar to the increase in Smad7 protein in podocytes that we observed previously in CD2APϪ/Ϫ mice (15).
During the first 2 weeks, kidney structure and function were not significantly different in CD2APϪ/Ϫ and control littermates, respectively. However all CD2APϪ/Ϫ animals had developed high grade albuminuria when examined at 3 weeks of age; significant glomerular histopathology characterized by mesangial proliferation and early FSGS lesions was detectable at 3 weeks and progressed to sclerosis in more than 50% of examined glomeruli in CD2APϪ/Ϫ mice at 4 weeks (data not shown). Thus, the onset of glomerular disease in this model with high grade proteinuria is rapid and occurs between 2 and 3 weeks of age.
Next, we isolated glomeruli from CD2APϩ/ϩ and CD2APϪ/Ϫ kidneys at 4 weeks of age to quantitate mRNA expression of key components of TGF-␤ signaling directly in primary glomerular fractions by quantitative real-time PCR. Transcript levels for TGF-␤1, type II TGF-␤ receptor (TGF-␤RII), and Smad7 were all significantly increased in glomeruli obtained from CD2APϪ/Ϫ compared with glomeruli from CD2APϩ/ϩ mouse kidneys (Fig. 1g). In contrast, type I TGF-␤ receptor mRNA was not significantly different (Fig. 1g). Together, these in situ protein and mRNA expression data indicate that increased expression of TGF-␤1 and Smad7 by differentiated podocytes in CD2APϪ/Ϫ mice is associated with albuminuria and mesangial activation at early stages of glomerulosclerosis.

CD2AP Deficiency Predisposes Podocytes to Undergo Apoptosis Induced by Autocrine and Paracrine TGF-␤ in Vitro-
After we determined that increased expression of mediators of TGF-␤ signaling in podocytes of CD2APϪ/Ϫ mice coincided with onset of albuminuria, we wished to examine whether TGF-␤ signaling and apoptosis were affected by CD2AP deficiency in vitro. Conditionally immortalized CD2APϪ/Ϫ and control CD2APϩ/ϩ podocyte cell lines were established from littermate mice as described previously (12). After 14 days of culture in non-permissive conditions, morphology and expression patterns of podocyte markers such as WT-1, synaptopodin, and podocin was indistinguishable in CD2APϪ/Ϫ and control CD2APϩ/ϩ podocyte cell lines (data not shown). In addition, distribution patterns of ZO-1 protein indicated the formation of characteristically interdigitated cell extensions in cultured CD2APϪ/Ϫ and CD2APϩ/ϩ podocytes (data not shown). Three independent cell clones for each genotype were examined. Quantitation of apoptotic nuclei showed that base-line (untreated) rates of apoptosis were significantly higher in CD2APϪ/Ϫ compared with CD2APϩ/ϩ (2.4 Ϯ 0.9 versus 0.7 Ϯ 0.3%, respectively) (Fig. 2a). TGF-␤ treatment for 24 h increased apoptosis rates by 2.5 Ϯ 0.5-fold over base line in CD2APϩ/ϩ cells and 4.8 Ϯ 1.7-fold over base line in CD2APϪ/Ϫ podocytes (Fig. 2a). Thus, the apoptosis-inducing effect of TGF-␤ was significantly enhanced in CD2APϪ/Ϫ compared with CD2APϩ/ϩ podocytes (4.8 Ϯ 1.7 versus 2.5 Ϯ 0.5fold induction; p Ͻ 0.03). These findings suggest that CD2AP exerted an anti-apoptotic function in podocytes both at base line and in particular when exposed to exogenous (paracrine) TGF-␤1.
Because we have found that podocytes also secrete latent and active autocrine TGF-␤2, 2 we examined base-line apoptosis rates in the absence or presence of pan-neutralizing anti-TGF-␤ antibody 2G7. Neutralizing antibody (2G7) significantly reduced base-line apoptosis in CD2APϪ/Ϫ podocytes by 49% compared with control IgG (1.09 Ϯ 0.39 versus 2.13 Ϯ 0.22%, p Ͻ 0.004) (Fig. 2b). Thus, increased base-line apoptosis observed in CD2APϪ/Ϫ podocytes is largely, but not exclusively, dependent on autocrine TGF-␤ signaling, indicating that TGF-␤ receptor activity and survival pathway signaling are coupled in podocytes by the CD2AP adaptor molecule. We conclude that CD2AP deficiency predisposes podocytes to apoptosis induced by both exogenous (paracrine) and autocrine TGF-␤.
CD2AP Is Required for Rapid Activation of PI3K and ERK1/2 by TGF-␤ but Not by EGF and Insulin-We reasoned that the anti-apoptotic effect of CD2AP on TGF-␤-induced apoptosis could be due to essential survival signaling and/or inhibition of pro-apoptotic pathways or both. Because PI3K/AKT functions as a primary cell survival pathway in many cell types (16), we examined activation profiles of PI3K/AKT in response to TGF-␤1 in three independent lines of CD2APϪ/Ϫ and CD2APϩ/ϩ podocytes, respectively. In CD2APϩ/ϩ podocyte lines, TGF-␤1 induced a strong and rapid (minutes) phosphorylation of AKT on serine 473 (Fig. 3a). Serine 473 phosphorylation peaked between 1 and 2 h and remained elevated for up to 4 h (Fig. 3b). In contrast, S(P) 473 -AKT was not detectable until 1 h of TGF-␤ treatment in CD2APϪ/Ϫ podocyte lines (Fig.  3a), and peak phosphorylation was delayed at 4 h of TGF-␤1 treatment (Fig. 3b). Among MAPKs, the ERK1/2 mediates primarily mitogenic and/or anti-apoptotic signaling (17,18). Interestingly, the profile of ERK1/2 activation in control CD2APϩ/ϩ podocytes treated with TGF-␤1 was very similar to the S(P) 473 -AKT profile, including a strong and rapid activation detectable at 5 min (Fig. 3a) followed by peak phosphorylation at 2 h (Fig. 3b). Similar to S(P) 473 -AKT, phosphorylation of ERK1/2 was not detectable during the initial 45 min (Fig. 3a), and peak phosphorylation was delayed between 2 and 4 h in CD2APϪ/Ϫ podocytes (Fig. 3b). Thus, the anti-apoptotic mediators PI3K/AKT and ERK1/2 are activated by TGF-␤ in podocytes with similar kinetic profiles, consistent with previous reports that ERK1/2 and PI3K are synergistically activated to mediate anti-apoptotic pathways (19,20). Our results presented here demonstrate for the first time that early activation of PI3K/AKT and ERK1/2 MAPK by TGF-␤ in podocytes requires the CD2AP adaptor protein.
To examine whether CD2AP and PI3K function in a linear pathway, we quantitated and directly compared the extent of TGF-␤-stimulated apoptosis as assayed by DAPI count and cell death ELISA (⌬405) in CD2APϪ/Ϫ podocytes and CD2APϩ/ϩ podocytes pretreated with wortmannin, a chemical inhibitor of PI3K function. Pretreatment with wortmannin resulted in a significant increase of the proapoptotic TGF-␤ effect in CD2APϩ/ϩ podocytes, but not in CD2APϪ/Ϫ podocytes, as determined by DAPI (Fig. 4a) and cell death assay (Fig. 4b).
Similar results were obtained when the AKT-specific inhibitor SH-5 was used in place of wortmannin (data not shown). These findings demonstrate that both CD2AP-deficiency or functional inactivation of PI3K were associated with enhanced TGF-␤induced apoptosis signaling in a non-additive, quantitatively comparable manner, indicating that CD2AP and PI3K function in a linear pathway downstream of TGF-␤ receptors.
To confirm this hypothesis at a molecular level by direct reconstitution experiment, we transfected CD2APϪ/Ϫ podocytes with CD2AP expression plasmid. Expression of epitopetagged, reconstituted CD2AP in transfected CD2APϪ/Ϫ cells was validated by Western blot analysis (data not shown). Rates of apoptotic nuclei were significantly reduced in CD2APϪ/Ϫ cells transfected with CD2AP expression plasmid compared with empty vector (control)-transfected cells (Fig. 4c). Together these findings demonstrate that PI3K/AKT mediate survival signals in podocytes exposed to TGF-␤1 and show that CD2AP is required for activation of this survival pathway by TGF-␤.
To examine whether CD2AP is a general or pathway-selective mediator of PI3K/AKT activation in podocytes, we treated CD2APϪ/Ϫ and control CD2APϩ/ϩ podocytes with insulin and EGF, two well characterized extracellular stimulators of PI3K/ AKT-dependent anti-apoptotic signaling (21,22). In contrast with TGF-␤ stimulation, insulin and EGF induced early AKT phosphorylation rapidly and to a similar extent in both CD2APϩ/ϩ and CD2APϪ/Ϫ podocytes, respectively (Fig. 4d), suggesting that CD2AP is required for activation of PI3K/AKT pathway by TGF-␤ signaling through receptor serine/threonine kinases but not by insulin or EGF signaling through receptor tyrosine kinases. We reasoned next that if CD2AP is not re-2 D. Wu and E. P. Böttinger, unpublished observations.
Taken together, the kinetic profiles of PI3K/AKT and MAPK activation by TGF-␤ in podocytes suggest that ligand-activated TGF-␤ receptors control anti-apoptotic PI3K/AKT and ERK1/2 activation possibly through common upstream mediators, which include CD2AP as a novel essential signaling component between TGF-␤ receptor and PI3K. Moreover, TGF-␤ receptors control pro-apoptotic p38 activation through a pathway that is distinct from PI3K/AKT and ERK1/2. CD2AP-dependent signals negatively trans-modulate pro-apoptotic MAPKs. Thus, the net effect of loss of CD2AP function in TGF-␤ signaling is enhanced proapoptotic p38 signaling activity and diminished anti-apoptotic PI3K/AKT and ERK1/2 pathway activity, con-sistent with our finding of increased sensitivity to TGF-␤1induced apoptosis in CD2APϪ/Ϫ podocytes.
Podocyte Apoptosis Is an Initial Manifestation in Albuminuric CD2APϪ/Ϫ Mice Developing Podocyte Depletion and Glomerulosclerosis-Because CD2AP deficiency in podocytes in vitro caused increased susceptibility to TGF-␤-induced apoptosis, we examined whether apoptosis was increased in glomerular cells in CD2APϪ/Ϫ mice in vivo. Glomerular cell turnover was examined by terminal dUTP nick-end labeling immunoperoxidase staining in combination with periodic acid-Schiff staining to demarcate glomerular basement membrane. We observed significantly increased apoptosis in 3-week-old CD2APϪ/Ϫ mice compared with age-matched control mice in podocytes and parietal cells, with a further increase at 4 weeks of age (Fig. 6, a and b, respectively). Interestingly, few apoptotic nuclei were already detectable in some 2-week-old CD2APϪ/Ϫ mice. In contrast, in the endocapillary/mesangial compartment increased apoptosis was detectable only at 4 weeks (Fig. 6c). Thus, glomerular epithelial cells and podocytes are characterized by apoptosis coincident with onset of albuminuria at the early stages of FSGS in CD2APϪ/Ϫ mice.
To examine whether increased podocyte apoptosis was associated with depletion of podocytes from glomeruli during progression of glomerulosclerosis in CD2APϪ/Ϫ mice, we determined podocyte counts per glomerular section based on immunofluorescence double-labeling using anti-WT1 and antisynaptopodin antibodies (Fig. 7, a-f). Double-positive cells per glomerular section were counted in 30 glomeruli per animal. The average counts of podocytes in 30 glomeruli per animal and 5 animals per experimental group were not significantly different between CD2APϩ/ϩ and CD2APϪ/Ϫ animals at 2 and 3 weeks of age (Fig. 7g). In contrast, WT1/synaptopodin doublepositive cells per glomerular section were significantly (p Ͻ 0.01) fewer in the kidneys of 4-week-old CD2APϪ/Ϫ compared with CD2APϩ/ϩ mice (4.7 Ϯ 1.5 versus 11.3 Ϯ 0.2 podocytes per glomerular section, respectively) (Fig. 7g). These findings demonstrate that with the onset of albuminuria in CD2APϪ/Ϫ mice an increase in podocyte apoptosis occurs that leads to subsequent loss of podocytes associated with progression of FSGS. Thus, CD2AP deficiency in podocytes caused increased susceptibility to undergo apoptosis associated with TGF-␤ signaling in vitro and in vivo.

DISCUSSION
Our findings presented in this report demonstrate a novel role for the adaptor molecule CD2AP in TGF-␤ signaling. Specifically, CD2AP is required for rapid and kinetically defined early (up to 1 h) activation of anti-apoptotic PI3K/AKT and ERK1/2 MAPK survival signaling pathways by TGF-␤ in podocytes. In addition, the activation of pro-apoptotic p38 MAPK by TGF-␤ is accelerated and enhanced in the absence of CD2AP. Interestingly, TGF-␤ treatment was associated with kinetically defined late (1-4 h) activation of both pathways in CD2APϪ/Ϫ podocytes, indicating that secondary, delayed activation mechanisms are CD2AP-independent. Nevertheless, the CD2AP-dependent early activation of anti-apoptotic signaling by TGF-␤ is important for counterregulation of the pro-apoptotic signaling activity of TGF-␤ in podocytes previously described by our group (10), because podocytes lacking CD2AP are significantly more susceptible to TGF-␤-induced apoptosis, and reconstitution of CD2AP directly reduces TGF-␤-induced apoptosis. CD2AP is not required for PI3K/AKT activation by EGF and insulin, suggesting that CD2AP functions selectively in TGF-␤ receptor-activated pathways but not in receptor-tyrosine kinase pathways. Importantly, we demonstrate in vivo that a 6-fold increase in apoptosis coincides with increased expression of TGF-␤1 in podocytes detectable at the onset of albuminuria in kidneys of CD2APϪ/Ϫ mice. Moreover, sustained apoptosis and TGF-␤ expression in podocytes is associated with a progressive loss of podocytes per glomerulus and with progressive FSGS in this model. In conclusion, we have identified a previously unrecognized molecular mechanism linking TGF-␤ expression, TGF-␤ receptor, and CD2AP signaling as the candidate pathway to specify the irreversible cell fate of podocytes in glomerular injury. The presence of CD2AP restricts podocyte apoptosis by enhancing critical PI3K/AKT and ERK1/2 survival pathways, whereas absent or reduced function of CD2AP may result in accelerated podocyte apoptosis because of insufficient activation of survival pathways and enhanced pro-apoptotic p38 activation in podocytes exposed to increased TGF-␤ in vitro and in vivo.
Our study is further significant because it provides an alternative model to the prevailing paradigm of the functional role of CD2AP in podocytes. Studies of the functional significance of CD2AP in the kidney were first prompted by the surprising phenotype of CD2APϪ/Ϫ mice (11), manifesting with high grade albuminuria, FSGS, and renal failure. Subsequent stud- FIG. 7. Analysis of podocyte counts in kidneys of CD2AP؉/؉ and CD2AP؊/؊ mice. a, indirect immunofluorescence double-labeling using anti-synaptopodin and anti-WT1. Renal cortex sections from CD2APϩ/ϩ (a-c) and CD2APϪ/Ϫ (d-f) mice sacrificed at 2 weeks of age (a and d), 3 weeks of age (b and e), and 4 weeks of age (c and f) were used, and synaptopodin staining was visualized using a fluorescein isothiocyanate-conjugated goat anti-mouse IgG (green) and WT-1 staining was visualized using a Cy3-conjugated goat anti-rabbit antibody (red). ies demonstrated that CD2AP can be localized in podocyte foot processes in vivo (23,24) and is able to interact directly with the slit-diaphragm proteins nephrin and podocin in vitro (25). Because defects in the genes encoding nephrin and podocin, NPHS1 and NPHS2, respectively, cause congenital forms of nephrotic syndrome and FSGS in humans (26) and because this was consistent with the phenotype observed in CD2APϪ/Ϫ mice, it has been proposed that CD2AP together with nephrin and podocin is important for slit diaphragm and foot process formation (27). However, the functional role of CD2AP in this context remains unclear. A recent report indicates that nephrin and/or CD2AP can interact with the p85 regulatory subunit of PI3K, associated with AKT phosphorylation (28). The authors propose that nephrin and/or CD2AP initiate signaling from the slit diaphragm (28). However, regulation and functional significance of proposed "slit diaphragm signaling" are unknown and remain to be validated in the context of endogenous CD2AP function in podocytes.
In contrast, our results support an alternate model where CD2AP functions as mediator of ligand-inducible, transmembrane TGF-␤ receptor-regulated anti-apoptotic signaling, which may be independent of slit diaphragm proteins. First, phosphorylation of AKT and ERK1/2 occurs with very rapid kinetics, suggestive of a protein synthesis-independent activation cascade. Because this rapid activation cascade is interrupted in the absence of CD2AP, we propose that CD2AP may provide a molecular linker between TGF-␤ receptor protein complexes and the p85 subunit of PI3K and/or Ras. Second, CD2AP is not required for the direct phosphorylation of Smad2 by TGF-␤ receptors, and receptor-regulated Smads are known to bind directly with TGF-␤ type 1 receptors (29). Third, TGF-␤ receptors and CD2AP are both present in lipid rafts, as shown by Schwarz et al. (25) for CD2AP and our group for TGF-␤ receptor complexes (30). Finally, the adaptor protein Dab2, which contains proline-rich domains similar to CD2AP, has been shown to interact with TGF-␤ receptor complexes including type I and type II receptors and facilitates activation of Smad2 and Smad3 (31). Thus, these intriguing findings are consistent with a model where CD2AP may interact with TGF-␤ receptors to enable activation of PI3K/AKT and/or Ras/ ERK1/2 signaling. Further studies to test this hypothesis seem warranted and are under way in our laboratory.
Glomerulosclerosis in animal models and humans is characterized by depletion of podocytes, and denuded glomerular basement membrane is thought to form aberrant adhesions and synechiae with Bowman's capsule (7,32,33). If tuft adhesions and synechiae are associated with collapse of appendant glomerular capillaries, irreversible glomerulosclerosis becomes established (34). Because podocyte depletion may be a key initiating lesion in this process, it is considered important to determine the underlying molecular and cellular mechanisms. Several possibilities need to be considered in this context. It is intriguing to note that experimental reduction of podocytes per glomerulus using puromycin amino nucleoside cytotoxicity in rats (34) was correlated directly with the extent and severity of glomerulosclerosis. Second, a recent report proposed mechanical detachment of podocytes from glomerular basement membranes and loss in urinary space as a candidate mechanism for podocyte depletion, based on the identification of viable cells expressing podocyte proteins in urine collected from humans (35). Interestingly, puromycin injury of podocytes resulted in up-regulation and activation of integrin-linked kinase ILK, which is known to activate AKT, and stable expression of active ILK was associated with decreased adhesiveness of podocytes in vitro (36). Third, it is thought that podocytes are unable to proliferate and to regenerate depleted podocytes in most forms of glomerular injury, leading to a state of "regenerative podocyte insufficiency" (7,(37)(38)(39).
In vivo and in vitro studies from our laboratory provide compelling evidence for injury-dependent activation of TGF-␤ signaling leading to apoptosis of susceptible podocytes (40). For example, the rate of podocyte apoptosis is increased 6 -20-fold and peaks already at the onset of albuminuria in three models of progressive glomerulopathy, including CD2APϪ/Ϫ mice (this report), TGF-␤1 transgenic mice (10), and diabetic db/db mice. 3 Increased expression of TGF-␤1 and Smad7 in podocytes coincides with the onset of apoptosis and albuminuria in every model examined (10,15) (this report). Finally, we have identified a molecular network of pro-and anti-apoptotic signaling pathways that can be activated by TGF-␤ in podocytes and may determine survival or apoptosis of injured podocytes, depending on their relative strengths. In this model, we describe CD2AP as a Smad-independent mediator for selective activation of survival pathways and repression of apoptosis signaling by TGF-␤. Thus, susceptibility to undergo podocyte apoptosis may be increased in individuals exposed to glomerular injury where increased TGF-␤ signaling activity coincides with genetic variants of CD2AP associated with loss of survival signaling (41).