Report

CD2-Associated Protein Haploinsufficiency Is Linked to Glomerular Disease Susceptibility

See allHide authors and affiliations

Science  23 May 2003:
Vol. 300, Issue 5623, pp. 1298-1300
DOI: 10.1126/science.1081068

Abstract

Loss of CD2-associated protein (CD2AP), a component of the filtration complex in the kidney, causes death in mice at 6 weeks of age. Mice with CD2AP haploinsufficiency developed glomerular changes at 9 months of age and had increased susceptibility to glomerular injury by nephrotoxic antibodies or immune complexes. Electron microscopic analysis of podocytes revealed defects in the formation of multivesicular bodies, suggesting an impairment of the intracellular degradation pathway. Two human patients with focal segmental glomerulosclerosis had a mutation predicted to ablate expression of one CD2AP allele, implicating CD2AP as a determinant of human susceptibility to glomerular disease.

CD2-associated protein (CD2AP) is an Src homology 3 (SH3) domain–containing protein expressed mainly in epithelial and lymphoid cells. Mice completely lacking CD2AP die of massive proteinuria shortly after birth, suggesting an important role for CD2AP in the filtration apparatus of the kidney (1). Because CD2AP heterozygous (+/–) mice exhibit reduced CD2AP protein levels (1), they were further analyzed for kidney defects. Twenty CD2AP+/– mice and 17 wild-type littermates were examined for proteinuria for 1 year. Mice were killed for histological analysis at 6, 9, and 12 months of age. Although none of the mice exhibited proteinuria during the course of the study (2), kidneys from CD2AP+/– mice at 9 months of age displayed glomerular abnormalities, including mesangial expansion and hypercellularity that were not apparent in 6-month-old animals (Fig. 1). In contrast to glomeruli from wild-type littermates (Fig. 1A), glomerular lesions from heterozygous mice showed varying degrees of abnormality, from a mild, stalklike mesangial expansion (Fig. 1C) to an expansion that impinged upon the lumen of the capillaries (Fig. 1D). The pathology was similar to that seen in CD2AP-null mice before the development of severe kidney disease at ∼3 to 4 weeks of age. Some lesions (Fig. 1B) were reminiscent of the human glomerular disease, focal segmental glomerulosclerosis (FSGS). Infiltration of inflammatory leukocytes was also detected in some affected glomeruli (Fig. 1, C and D).

Fig. 1.

Glomerular pathology in CD2AP+/– mice. Sections of formalin-fixed kidney cortex from 9-month-old CD2AP+/– mice and wild-type littermates were stained with hematoxylin and eosin and with periodic acid–Schiff. (A) Representative glomerulus from a wild-type control. (B) Glomerulus from a CD2AP+/– mouse, showing changes similar to FSGS. (C and D) Representative glomeruli from a CD2AP+/– mouse, showing increased mesangial cellularity and expansion of the mesangium.

Immunoglobulin deposits contribute to the pathology of many human kidney diseases. Such deposits were observed in either punctate or linear patterns at the glomerular basement membrane (GBM) in 4 out of 20 CD2AP+/– mice (Fig. 2, A, B, and C). Electron microscopic (EM) examination of CD2AP+/– kidney tissue also revealed electron-dense mesangial deposits that occluded the capillary lumen (Fig. 2E) and subepithelial and subendothelial deposits along the GBM (Fig. 2, F and G) that were not detected in glomeruli from wild-type mice (Fig. 2D). In some older animals (12 to 18 months of age), highly organized microtubular deposits in the mesangium were detected (Fig. 2, H and I). These deposits stained negative by Congo Red staining and are diagnostic for immunotactoid glomerulopathy (ITG) in humans (3). To our knowledge, this phenotype has not been reported before in mice.

Fig. 2.

Immunofluorescence and EM analysis of CD2AP+/– kidney tissue. (A to C) Staining of fresh frozen kidney from a 9-month-old CD2AP+/– mouse, showing punctate patterns of (A) immunoglobulin M and (B) C3 deposition in the GBM, or a linear pattern of (C) immunoglobulin G1 deposition in the GBM. (D) EM view of the normal capillary loop from a wild-type mouse glomerulus at 9 months of age, showing normal podocyte foot processes (P) and capillary lumens (CL). (E) EM view of the capillary loop from a CD2AP+/– mouse glomerulus at 9 months of age, showing mesangial expansion (M) and encroachment of the mesangium into the capillary lumen. (F and G) Electron-dense deposits (arrows) identified on the (F) subepithelial and (G) subendothelial face of the GBM from CD2AP+/– kidney at 9 months of age. (H) Immunotactoid deposits detected in the mesangium (M). (I) Magnification of the boxed region in (H), showing the microtubular morphology of deposits.

Although the etiology of ITG is unknown, it can be associated with monoclonal antibody production, and mesangial deposits are thought to be composed of immunoglobulin (4, 5). We therefore analyzed B cell function in CD2AP+/– mice. Histological analysis did not indicate B cell hyperplasia or malignancy (2). Basal immunoglobulin isotype levels were indistinguishable from those of wild-type controls (2). Serum cryoglobulins were also not detected. Equivalent levels of antigen-specific antibodies were generated in immunized heterozygous and wild-type littermates (fig. S1). No autoantibodies were detected in heterozygous animals as normal kidney sections did not react with sera from 9- to 12-month-old heterozygous mice (2). Therefore, increased deposition of immunoglobulin in CD2AP+/– mice could reflect an intrinsic glomerular defect.

CD2AP and its homolog Cin-85 are implicated in endocytosis and lysosomal sorting (6, 7). Podocytes can endocytose proteins, including components of the complement pathway, lipoproteins (8, 9), and experimental agents such as ferritin and horseradish peroxidase (1012). To determine whether podocytes from heterozygous mice exhibit defects in endocytosis or vesicle trafficking, cationized ferritin was injected intravenously into 6-week-old wild-type and heterozygous mice. Animals were killed at 30, 60, 90, and 120 min after injection, a time period in which ferritin progressively moves through the GBM into podocytes (11). In wild-type mice, ferritin had accumulated within the GBM at 30 min and was also trapped in the slit diaphragms (Fig. 3A). At 60 min, ferritin was visible within endocytic vesicles and at the apical plasma membrane, suggesting that ferritin had traversed the slit diaphragm (Fig. 3B). Ferritin-containing coated pits were visible on both apical and basal membranes, indicating that endocytosis occurred on both surfaces of the podocyte (Fig. 3C) (13). At 90 to 120 min, less iron was visible in podocytes, suggesting that degradation of ferritin and removal of iron from the podocyte had occurred and that ferritin was sorted to the degradative pathway. An increased number of multivesicular bodies (MVBs), which represent prelysosomal compartments, were also observed (Fig. 3D) (14).

Fig. 3.

The time course of ferritin uptake by podocytes in vivo. Wild-type mice were injected intravenously with cationized ferritin and killed at (A) 30 min, (B and C) 60 min, and (D) 90 min. Sections of kidney cortex were analyzed by electron microscopy. (A) At 30 min, ferritin accumulated at the slit diaphragm (arrowheads) and the subendothelial side of the GBM (arrows). (B) By 60 min, ferritin was detected inside the podocyte in small internalized vesicles (arrows). (C) Membrane invaginations (arrows) suggest that endocytosis occurs from both the apical and basal surfaces of the foot process. (D) Endocytosis is associated with the formation of MVBs (arrows).

To determine whether there were differences in endocytosis or sorting between wild-type and heterozygous animals, endocytic vesicles and MVBs were quantitated by electron microscopy (in number per 100 μm of GBM) before and after ferritin injection. At early time points, no differences in endosome number were detected between wild-type and heterozygous kidneys (2). Whereas ferritin injection induced a twofold increase in MVBs in wild-type mice (from 8.6 ± 2.8 to 17.5 ± 5, n = 7 mice), there was no induction detected in heterozygous kidneys (8.5 ± 1.6 to 8.3 ± 2.6, n = 7 mice). This suggests that CD2AP is involved in MVB formation.

To determine whether these defects in the intracellular degradation pathway increase susceptibility to glomerular injury, wild-type and heterozygous mice were injected with a low dose of a nephrotoxic antibody (15). Although most wild-type mice (25 out of 27) did not develop proteinuria at this dose, proteinuria was detected in more than half of the heterozygous mice (24 out of 36, P value < 0.0001) (fig. S2). In addition, whereas proteinuria in the wild-type mice was mild and resolved within 2 weeks, proteinuria that developed in heterozygous mice was more severe, more persistent, and resulted in the death of some of the animals (4 out of 24). The data suggest that CD2AP+/– mice have a decreased capacity to sustain glomerular damage.

Because CD2AP+/– mice exhibit a pathology similar to the human disease FSGS, we screened DNA from human FSGS patients for changes in CD2AP. FSGS is a major cause of nephrotic syndrome in both adults and children and has an increased incidence in African Americans (1618). Our study population consisted of 30 African Americans with idiopathic FSGS and 15 African Americans with human immunodeficiency virus (HIV)–associated FSGS. As a normal control group, we selected 45 African Americans who had had HIV infection for at least 8 years with normal kidney function (serum creatinine <1.5 mg/dL and urine protein/creatinine ratio <0.5). We screened for CD2AP mutations using polymerase chain reaction (PCR) and denaturing high-performance liquid chromatography (19).

We detected six distinct DNA variants in 10 out of 45 patients that were not present in any of the control subjects. One nucleotide variant, detected in two of the patients with primary FSGS, is predicted to alter the expression of CD2AP (Fig. 4A). The change, affecting the splice acceptor of exon 7 on one allele, replaces two nucleotides, GC, with CT. Isolation of individual clones bearing either wild-type or polymorphic sequences confirmed heterozygosity of the polymorphism (Fig. 4B). The guanine, the last nucleotide of intron 6, is absolutely conserved in all 3′ splice acceptor sites (20); its replacement with cytosine is expected to prevent splicing to this site. In addition, the change from cytosine to thymidine is predicted to change the encoded amino acid, proline243, to serine. To confirm that splicing does not occur at this site, we used reverse transcription (RT)–PCR to amplify mRNA that encoded exons 6 to 8 from immortalized B cells generated from these two patients. All sequenced products contained only the wild-type cytosine, with no evidence of thymidine at the sequence that corresponds to the beginning of exon 7 (2), confirming that splicing to exon 7 did not occur on the polymorphic allele. RT-PCR amplification of a larger segment of the CD2AP mRNA demonstrated a truncated message that corresponds to splicing between exons 4 and 18. Because exon 18 codes for part of the 3′ un-translated region, the predicted protein product of this variant message would lack more than 80% of the CD2AP protein (Fig. 4C).

Fig. 4.

Characterization of the CD2AP exon 7 mutation in FSGS patients A and B. (A) Direct DNA sequencing of PCR amplification of exon 7 from patient A shows mixed peaks at the last nucleotide of the splice acceptor site (boxed) and the following nucleotide. (B) Cloning of the PCR product and sequencing of individual clones demonstrates heterozygosity and the substitution of GC (rows C3 and C4) with CT (rows C1 and C2). (C) Detection of the alternatively spliced transcript in patients with the GC/CT substitution. Cloning and sequencing of the aberrant transcript showed precise splicing between exon 4 and exon 18, suggesting that the mutation results in aberrant splicing. (D) Decreased CD2AP expression in two patients (A and B) with the exon 7 splice acceptor mutation. Cell lysates from Epstein-Barr virus–immortalized B cells were immunoblotted with amino- and carboxy-terminal CD2AP antiserum and with a control antiserum to the p38α protein kinase. Wt, wild type.

Immunoblot analysis of CD2AP expression from lysates of immortalized B lymphocytes confirmed that CD2AP expression was indeed lower in these two patients, compared to a control sample with no detectable CD2AP mutations (Fig. 4D). Neither of the two antibodies used recognized other protein products, suggesting that a stable truncated CD2AP protein was not generated.

Mutations in several podocyte genes have recently been implicated in susceptibility to glomerular disease (21, 22). Here we suggest that in addition to a possible structural role in the slit diaphragm, CD2AP also targets proteins to the degradative pathway. Recent work on CD2AP and its homolog Cin-85 are consistent with this idea (6, 7, 23). The accumulation of immunoglobulin in the GBM of CD2AP+/– mice may reflect the impaired ability of the podocyte to clear plasma proteins from the GBM rather than any aberrant immunological process. Although the GBM has a barrier function, we suggest that some fraction of plasma proteins during nonpathological states becomes trapped in the GBM; such proteins may normally be undetected because of their rapid clearance and degradation by the podocyte. The pathogenesis of glomerular diseases may not only involve autoimmune antibodies or immune complexes, but protein clearance by podocytes may also be an important modulator and perhaps instigator of disease. Thus, many renal diseases might be classified as diseases of intracellular trafficking.

Supporting Online Material

www.sciencemag.org/cgi/content/full/300/5623/1298/DC1

Materials and Methods

Figs.S1 and S2

References and Notes

  • * These authors contributed equally to this work.

References and Notes

View Abstract

Navigate This Article