Up-regulation of the homophilic adhesion molecule sidekick-1 in podocytes contributes to glomerulosclerosis

Abstract

Focal segmental glomerulosclerosis (FSGS) is a leading cause of nephrotic syndrome and end-stage renal disease worldwide. Although the mechanisms underlying this important disease are poorly understood, the glomerular podocyte clearly plays a central role in disease pathogenesis. In the current work, we demonstrate that the homophilic adhesion molecule sidekick-1 (sdk-1) is up-regulated in podocytes in FSGS both in rodent models and in human kidney biopsy samples. Transgenic mice that have podocyte-specific overexpression of sdk-1 develop gradually progressive heavy proteinuria and severe FSGS. We also show that sdk-1 associates with the slit diaphragm linker protein MAGI-1, which is already known to interact with several critical podocyte proteins including synaptopodin, alpha-actinin-4, nephrin, JAM4, and beta-catenin. This interaction is mediated through a direct interaction between the carboxyl terminus of sdk-1 and specific PDZ domains of MAGI-1. In vitro expression of sdk-1 enables a dramatic recruitment of MAGI-1 to the cell membrane. Furthermore, a truncated version of sdk-1 that is unable to bind to MAGI-1 does not induce podocyte dysfunction when overexpressed. We conclude that the up-regulation of sdk-1 in podocytes is an important pathogenic factor in FSGS and that the mechanism involves disruption of the actin cytoskeleton possibly via alterations in MAGI-1 function.

FIGURE 1.

sdk-1 is overexpressed in podocytes in FSGS. A, expression of sdk-1 RNA is significantly up-regulated in podocytes in cell culture when exposed to either HIV-1 infection or PAN as determined by quantitative PCR. Massive up-regulation of sdk-1 RNA after HIV-1 infection was previously reported (16, 17). WT, wild type. Error bars indicate S.E. B, the up-regulation of sdk-1 protein after exposure to PAN was confirmed by Western blotting. C, 293T cells transfected with an sdk-1 expression construct were used as the positive control. C, sdk-1 protein expression was detected by immunofluorescence in podocytes in PAN-treated rats but not in sham-injected controls. Podocyte-specific expression was confirmed by colocalization with the podocyte-specific marker vimentin. D, upper, a novel rabbit anti-human sdk-1 antibody specifically interacts with lysate from cells transfected with an sdk-1, but not an sdk-2, expression plasmid. Lower, when preincubated with excess immunizing peptide, this antibody showed no reactivity against human specimens. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. E, by immunohistochemistry, sdk-1 protein was detected in a podocyte distribution in kidney biopsies from two patients with idiopathic FSGS but not in normal human kidney.

FIGURE 2.

De novo expression of sdk-1 specifically in podocytes in vivo leads to FSGS. A, double transgenic mice (carrying both podocin-Cre and floxed sdk-1 transgenes) express sdk-1 specifically in podocytes as determined by immunostaining. B, monthly urine dipstick measurements were recorded for double transgenic (n = 22), podocin-Cre alone (n = 5) and floxed sdk-1 alone (n = 5) mice for a total of 1 year. Double transgenic mice begin to develop overt proteinuria at an age of 5 months as compared with single transgenic littermates that did not become proteinuric. Mice were considered proteinuric for a dipstick reading of 2+ or greater that persisted for 2 consecutive months or if an animal expired spontaneously after a single positive reading before a second reading could be performed. C, at 9 months of age, double transgenic mice were sacrificed, and kidneys were harvested for routine pathology. Periodic acid-Schiff staining demonstrates protein casts and tubular protein resorption droplets (upper left, 20×), classic FSGS lesions (upper right, 60×), glomerulus with collapse and epithelial cell proliferation with pseudocrescent (lower left), and a globally sclerotic glomerulus (lower right). D, electron microscopy shows severe foot process effacement with condensation of actin filaments above the glomerular basement membrane, and focal cytoplasmic vacuolization. E, urine albumin:creatinine ratios shows massive proteinuria in double transgenic animals as compared with single transgenic littermates. Error bars indicate S.E.

FIGURE 3.

Podocyte-specific expression of sdk-1 in transgenic mice induces FSGS that has elements of both idiopathic and collapsing types. A, synaptopodin (Synpo) expression is focally lost in sclerotic glomeruli but is preserved in unaffected areas. PAS, periodic acid-Schiff. B, high levels of Ki-67 expression were seen in cells making up a pseudocrescent (upper panel) and in podocytes within the glomerular tuft (lower panel; arrowhead identifies a podocyte).

FIGURE 4.

MAGI-1 directly interacts with sdk-1. A, all splice variants of MAGI-1 were precipitated from podocyte cellular lysate using a full-length FLAG-sdk-1 protein as bait, but not with a truncated version that lacks its PDZ binding domain. IP, immunoprecipitation; IB, immunoblotting. B, MAGI-1 was endogenously precipitated from double transgenic mouse glomerular protein lysate using an anti-sdk-1 antibody but not when using control IgG. C, Myc-MAGI localizes to the cytoplasm of transfected 293T cells when expressed alone (left). However, when Myc-MAGI-1 is co-expressed with full-length sdk-1, there is a dramatic recruitment of MAGI-I protein to the membrane, where it colocalizes with sdk-1 particularly at sites of intercellular contact. However, when Myc-MAGI-1 and a truncated sdk-1 that lacks its PDZ binding domain are co-expressed, Myc-MAG-I remains primarily cytoplasmic (with clear exclusion of the nuclear compartment), whereas sdk-1 is still in the membrane. DAPI, 4′,6-diamidino-2-phenylindole. D, various Myc-tagged MAGI proteins were expressed in 293T cells. Full contains all domains of MAGI including all six PDZ domains (schematic shown in top panel); II contains PDZ0 and glucokinase (GK); III contains two WW domains and PDZ1; IV contains PDZ2 and PDZ3; and V contains PDZ4 and PDZ5 (described in Ref. 22). Lysates were incubated with FLAG beads that had been bound to either FLAG-sdk-1-Full (F) or FLAG-sdk-1-Delete (Δ) that lacks its C terminus, and pulldowns were performed with Western blotting for Myc (lower left). GST fusion proteins encoding individual PDZ domains of MAGI-1 (PDZ1, PDZ4, and PDZ5 (described in Ref. 20)) were expressed in bacteria, and lysates were bound to glutathione-Sepharose beads. Lysates containing FLAG-sdk-1-Full and FLAG-sdk-1-Delete proteins were added to the beads, and pulldown assays were performed with Western blotting for FLAG (lower right). These results suggest that sdk-1 binds to PDZ1, PDZ4, and PDZ5 of MAGI-1 but that, unlike for PDZ4 and PDZ5, PDZ1 as a single domain is not sufficient to mediate the interaction.

FIGURE 5.

A truncated version of sdk-1 that is unable to bind to MAGI-1 does not induce podocyte dysfunction when overexpressed. A, as shown in the schematic representation, sdk-1 ΔSSFV is identical to full-length sdk-1 except for a deletion of its last four amino acids. B, podocyte cell lines stably transfected with full-length sdk-1 or the truncated version express similar levels of sdk-1 protein. C, podocytes stably transfected with the full-length sdk-1 expression plasmid show dramatic loss of normal actin cytoskeletal structure characterized by a centrifugal distribution of actin filaments as determined by phalloidin staining as compared with podocytes expressing sdk-1 ΔSSFV. D, two blinded observers each counted from random fields the number of cells containing substantial stress fibers throughout the entire cell (in a non-centrifugal distribution). The percentage of positive cells was calculated per observer and then averaged (*, p < 0.01). Error bars indicate S.E.