Enhanced podocalyxin expression alters the structure of podocyte basal surface

Podocytes have a unique and complex cellular organization. With respect to their cytoarchitecture, podocytes consist of three different segments: the cell body, major processes, and characteristic interdigitating foot processes. The foot processes are separated by filtration slits, bridged by the slit diaphragm, a meshwork of proteins anchored at the sides of the foot processes. The slit diaphragm complex contains the transmembrane proteins P-cadherin and nephrin, the intracellular ZO-1, α-, β-, γ-catenin and CD2AP (Somlo and Mundel, 2000). The external surface of podocytes is covered with a sialic acid-rich glycocalyx known as the glomerular epithelial polyanion (Michael et al., 1970). Podocalyxin (PC), the major sialoprotein of podocyte glycocalyx was initially identified in the rat (Kerjaschki et al., 1984; Horvat et al., 1986). The human counterpart of PC was recently cloned (Kerjaschki et al., 1986; Kershaw et al., 1997). The high net negative charge of PC in the foot processes, together with the meshwork of the slit diaphragm, constitute a barrier to glomerular permeability (Kerjaschki et al., 1984; Kerjaschki, 1994; Mundel and Kriz, 1995).

The basal surface of podocyte foot processes is connected to the glomerular basement membrane (GBM) at focal contacts mainly by α3β1 integrin heterodimers (Adler, 1992; Baraldi et al., 1994). GBM consists of a cross-linked meshwork of collagen IV, laminin, heparan sulfate proteoglycan, fibronectin, and other molecules (Miner, 1999).

Podocytes are the target of injury in many glomerular diseases affecting the shape of foot processes. In minimal-change disease or diabetic nephropathy, broadening of foot processes is accompanied by podocyte dysfunction (Kerjaschki, 1994; Pagtalunan et al., 1997). Over the past decade the molecular components affecting podocyte morphology with respect to foot process and slit diaphragm formation have been studied extensively. For example, the expression of PC and nephrin, two tissue-specific proteins, is related to podocyte differentiation (Schnabel et al., 1989; Kawachi et al., 2002). Deletion of the PC gene in the mouse resulted in failure of foot process and filtration slit formation as well as postnatal lethality in the homozygote (Doyonnas et al., 2001). Foot processes were also reported to be underdeveloped in the nephrin knockout mouse model (Putaala et al., 2001). Moreover, GBM components such as laminin also participate in the formation of foot processes (Noakes et al., 1995).

Although a functional role of PC in maintaining podocyte morphology has been established, little is known about other functional properties of this molecule. Biochemical and sequence analysis in different organisms revealed that PC is a 150-165 kDa transmembrane glycoprotein (Kershaw et al., 1995; Kershaw et al. 1997; McNagny et al., 1997). Its extracellular domain, rich in serine, threonine or proline residues, provides many potential O-glycosylation sites, which are sialylated. Although it was first described as the sialoprotein of podocytes, PC has also been detected in high endothelial venules, platelets, hematopoietic precursor cells, megakaryocytes and thrombocytes (McNagny et al., 1997; Sasseti et al., 1998; Miettinen et al., 1999). Based on its structural properties and localization, PC was grouped into the family of cell membrane-associated, mucin-like glycoproteins. Epithelial mucins prevent cell-cell or cell-matrix adhesion (Hilkens et al., 1992; Shimizu and Shaw, 1993; Simmons et al., 2001). It was recently reported that the cell-cell antiadhesive properties of PC are essential in maintaining the filtration slits between foot processes (Takeda et al., 2000).

In the present study, we have used T-SV40 immortalized human glomerular epithelial cells (HGEC) to examine the expression and additional functional properties of proteins related with podocyte morphology. These cells exhibit a regular cobblestone appearance in culture. The phenotype of HGEC fully agrees with parental podocytes. These cells express visceral glomerular epithelial cell markers such as PC, CALLA, cytokeratin, α3β1 integrin, α1 and α5 chains of collagen IV and WT1. Moreover HGEC have already been used for expression and functional studies (Delarue et al., 1991; Krishnamurti et al., 1996; Krishnamurti et al., 1997; Krishnamurti et al., 2001; Kitsiou et al., 2003). In this report we demonstrated that these cells also express nephrin, and that EHS-derived laminin, or intact human GBM enhanced both PC and nephrin expression, and additionally induced structural changes of HGEC. We also provide evidence that PC expressed by HGEC, inhibited cell binding to the matrix by competing with β1 integrin-mediated cell adhesion. Based on our data, we suggest that the conformation of the basal surface of HGEC grown on basement membrane substrates, is due at least in part, to the up-regulation of PC expression by matrix components and subsequently to the inhibition of cell binding to matrix by this sialoprotein.

T-SV40-immortalized human glomerular epithelial cells (HGEC) were cultured at 37°C in an environment of 5% CO₂ in media composed of DMEM/Ham's F-12 containing 1% FCS, 15 mM HEPES, 2 mM glutamine, ITS (5 μg/ml insulin, 5 ng/ml sodium selenite, 5 μg/ml transferrin), 50 nM dexamethasone, 100 U/ml penicillin, 100 μg/ml streptomycin, 25 μg/ml amphotericin and 5 mM D-glucose. We have also used HGEC cultured in 25 mM D-glucose for a minimum of 6 months before performing experiments. Cells were released from their tissue culture flasks, for passaging or use in experiments, by treatment with 0.05% trypsin in 1 mM EDTA at 37°C for 2 minutes. For expression experiments on basement membrane protein substrates, cells were seeded on 24-well plates containing glass coverslips, or petri dishes, either uncoated or coated with the appropriate substrate, and cultured to similar densities for approximately 48 hours.

Laminin and collagen IV were isolated from EHS tumor and purified using previously described techniques (Timpl et al., 1979; Kleinman et al., 1982; Yurchenco and Furtmayr, 1984). GBM was isolated from glomeruli as previously described (Anderson et al., 1994). Laminin concentration was determined by the method of Bradford using a Coomassie Plus Protein Assay Reagent Kit (Pierce). Concentration of collagen IV or sonicated GBM, was determined using the method of Waddell (Waddell, 1956).

Tissue culture dishes (60×15 mm) or glass coverslips (13 mm diameter) were coated with collagen IV, laminin or GBM according to protocols previously described (Haitoglou et al., 1992; Anderson et al., 1994; Kalfa et al., 1995). Briefly, 1 ml or 200 μl of each protein solution were added per dish or coverslip respectively and allowed to dry overnight at 29°C. Upon drying, dishes or coverslips were washed twice with PBS and the remaining active sites were subsequently blocked with 0.2% BSA in PBS for 2 hours at 37°C. Dishes or coverslips were then washed twice with PBS and sterilized by UV irradiation for 2 hours.

In this study we used 3D3 monoclonal antibody (mAb) against human PC, which recognizes part of the extracellular domain of PC (base pairs 1004-1492) (Kershaw et al., 1997). This domain corresponds at least in part to the previously described mucin-like domain of PC (Sassetti et al., 2000). Additionally, we used PI18 monoclonal antibody against rat PC, a kind gift from D. Kerjaschki. We also used 48E11 monoclonal antibody against human nephrin, a kind gift of K. Tryggvason. mAb 48E11 recognizes the fibronectin type III extracellular domain of nephrin. We also used rabbit polyclonal antibody directed against ZO-1 (Zymed Laboratories), mouse anti-human mAb P5D2 against β1 integrin subunit, suitable for inhibition of cell adhesion (Chemicon International), monoclonal antibody against β-tubulin (Sigma) and W6/32 against class I HLA (ATCC).

Cells were cultured as described, released from dishes by trypsin treatment, and lysed in PBS containing 1% TritonX-100, 1 mM PMSF, 1 mM NEM, and a cocktail of protease inhibitors (Sigma P8340) for 30 minutes on ice. Protein estimation was performed by the method of Bradford (Pierce). Equal amounts of total protein were loaded onto a 7.5% polyacrylamide gel under non-reducing conditions (monoclonal antibodies against PC or nephrin recognized the epitopes under reducing and non-reducing conditions). Proteins were then transferred to Hybond-ECL nitrocellulose membrane (Amersham). Blots were saturated with 5% non-fat milk in TBS, 0.1% Tween-20 and incubated with the appropriate dilution of monoclonal antibody in the same buffer without Tween-20 overnight at 4°C. Incubation with peroxidase-conjugated goat anti-mouse IgG (Amersham) and detection of bound peroxidase activity was carried out as described in ECL-blotting detection system (Amersham).

For immunofluorescence, cells were cultured on 24-well plates containing coverslips as described above. Cells were fixed with 2% formaldehyde in PBS for 10 minutes. For ZO-1 immunolabeling, cells were also permeabilized with 0.2% Triton X-100 at 25°C for 3 minutes. Cells were then incubated with monoclonal antibodies overnight at 4°C. Detection was performed with FITC-conjugated goat anti-mouse IgG (Cappel) for 1 hour at 25°C. Coverslips were mounted and examined under a Zeiss Axiophot photomicroscope or a Bio-Rad confocal microscope (MRC 1024 ES) equipped with Lasersharp software (BioRad) and a krypton-argon laser. The X-Y sections were generated with a motor step of 0.5 μm. Images from Axiophot or confocal microscopy were processed with Photoshop version 6 software (Adobe) or Confocal Assistant respectively.

Sprague-Dawley rats were rendered diabetic by an intravenous injection of streptozotocin (STZ), as reported previously (Wu et al., 1997). Seven months after diabetes induction, eight rats (four control, four diabetic) were sacrificed and their kidneys were processed for immunohistochemistry. Control and diabetic kidney tissues were paraffin-embedded and 4-μm-thick sections were subjected to antigen retrieval in a microwave oven for 12 minutes in 10 mM citrate buffer, pH 6.0. Endogenous peroxidase was blocked by 1% H₂O₂ in PBS. Blocking was performed in 10% rabbit serum. Sections were incubated overnight at 4°C with PI18 anti-rat PC diluted 1:250 in PBS and were then incubated for 1 hour at 25°C with biotinylated rabbit anti-mouse antibody (DAKO) diluted 1:200 in PBS. Slides were exposed to the DAKO ABC reagent for 30 minutes at 25°C, washed with PBS and stained with DAB solution (1.7 mM DAB in 10 mM Tris–HCl pH 7.6 containing 0.03% H₂O₂). Slides were analyzed using a Zeiss Axiophot photomicroscope. The intensity of PC staining in glomerular cells was evaluated blindly, by two independent observers, on at least four different optical fields per section, in four different kidney sections for normal and diabetic rats.

Cells were cultured as described, released from their dishes by trypsin treatment, washed in PBS and resuspended in FACS buffer (2% FCS, 0.02% sodium azide in PBS). Cells were incubated with 3D3 anti-PC antibody for 45 minutes on ice, and washed with FACS buffer. Cells were then incubated with anti-mouse IgG-FITC conjugated for 45 minutes on ice, washed with FACS buffer and fixed with 1% formaldehyde in PBS. Analysis was performed using CELL QUEST software on a FACScan (Becton Dickinson). Fluorescence was determined on a four-decade log scale and fluorescence intensity (log F1) was expressed as the mean channel number of 10,000 cells. The mean of fluorescence and the percentage of positive cells was calculated in the histogram section selected by the marker (M1), in order to subtract the fluorescence of the negative control (cells incubated only with fluorescein-conjugated anti-mouse IgG). mAbs were used at saturating concentration for FACS analysis.

In this study we used the cDNA fragment of human PC (293 bp, nucleotides 1004-1296) cloned into the Srf I site of pCR-Script-Amp SK⁺ (Stratagene). We also used the 600-bp EcoRI/BamHI fragment from GAPDH cDNA, which was cloned into pBSK plasmid. Inserts were labeled with [³²P]dCTP using HexaLabel DNA Labeling kit (Fermentas).

Total cellular RNA was isolated using Total RNA Isolation Reagent (Gibco BRL) according to the manufacturer's instructions. The concentration and the purity of each sample were determined spectrophotometrically. Total RNA from each sample (10 μg) was denatured in 1× MOPS buffer containing 6% formaldehyde and 50% formamide by heating at 65°C for 15 minutes. Samples were electrophoresed in 1% agarose-formaldehyde gel in 1× MOPS running buffer containing 6% formaldehyde. Subsequently RNA was transferred overnight on to Hybond-N⁺ nylon membrane (Amersham). Hybridizations were carried out at 65°C overnight with [³²P]dCTP-labeled PC cDNA in hybridization solution (50 mM phosphate buffer pH 7, 0.3 M NaCl, 5 mM EDTA, 10% dextran sulfate, 1% SDS, 5 mg/ml yeast RNA, 12.5× Denhardt's). After hybridization, filters were washed sequentially twice for 15 minutes in 2× SSC, 0.1% SDS, twice for 15 minutes in 0.5× SSC, 0.1% SDS and 0.2× SSC, 0.1% SDS at 65°C and exposed for 10 days to X-ray film at –70°C.

96-well plates (Microlon 600, Greiner) were coated with 50 μl of 5 μg/ml collagen IV or 25 μg/ml laminin according to protocols previously described (Haitoglou et al., 1992; Anderson et al., 1994; Kalfa et al., 1995). HGEC grown in either 5 mM or 25 mM glucose until 70-80% confluent, were metabolically labeled for 18 hours with 0.15 mCi [³⁵S]-methionine (Amersham) per T-25 flask. Labeled cells were detached from their culture flasks with trypsin treatment as described, washed twice with DMEM, counted and resuspended in binding buffer (DMEM, 25 mM HEPES, 2 mg/ml BSA). 50 μl of cell suspension containing 5000 cells were seeded to each well already containing the appropriate dilution of monoclonal antibody. The plates were incubated at 37°C for 40 minutes and then washed three times with binding buffer to remove non-adherent cells. Adherent cells were lysed with 100 μl 0.5 N NaOH, 1% SDS and the lysates were counted in a scintillation counter. For competition assays a mixture of anti-β1 integrin and 3D3 anti-PC antibodies was added at the appropriate dilution. The lowest dilution of antibodies used in the adhesion assays was the saturating dilution, as determined by flow cytometry. Adhesion experiments were performed at least three times in hexaplicates.

96-well plates were coated with substrates as already described. Detached cells from their culture flasks were pre-incubated with 3D3 anti-PC or anti-HLA antibody for 30 minutes at 37°C and were allowed to spread on coated wells for 50 minutes at 37°C. At the end of the incubation period non-adherent cells were aspirated. The wells were gently washed twice with PBS, and cells fixed with 2% formaldehyde in PBS for 5 minutes. All antibodies were used at saturating concentrations as determined by flow cytometry. Random fields from hexaplicate samples were photographed, and the surface area of 100 cells from each experiment, randomly selected from these fields, was measured using Image Pro software. Spreading assays were performed at least three times.

For scanning electron microscopy (SEM), cells were grown on 24-well plates containing glass coverslips as described, fixed in 50 mM sodium cacodylate buffer pH 7.4 containing 2.5% glutaraldehyde for 30 minutes at 32°C and then postfixed in 1% OsO₄ for 30 minutes at 4°C. Cells were dehydrated through a graded series of ethanol (5 minutes each in 25%, 50%, 70%, 95% and twice in 100%) containing 3% uranyl acetate followed by dehydration in a graded mixture of ethanol with amyl-ester (1:2, 1:1). The cells were subsequently dried using hexamethyldisilazane. SEM was performed on a JEOL JSM 5600 microscope, and cells were examined at 5-10 kV at an angle of 45° to enable visualization of the underlying surface close to the cell edges. For transmission electron microscopy (TEM), the cells were cultured on dishes as described, and then fixed for 30 minutes at 32°C in PBS pH 7.4 containing 2% glutaraldehyde. Cultures were post-fixed in 1% OsO₄ for 15 minutes at 4°C and dehydrated by treatment with a graded series of ethanol (5 minutes each in 50%, 70%, 95%, and twice in 100%). In sequence, the cells were detached with the aid of a rubber pipette bulb in the presence of propylene oxide and embedded in Epon/Araldite according to standard procedures. Sections were cut using an Ultracut R ultratome (Leica) and counterstained with 7% uranyl acetate and lead citrate (Reynolds, 1963). TEM was performed using a Zeiss 900 microscope operating at 80 kV. Micrographs were recorded on Kodak electron image plates (SO-163, ESTAR thick base).

Images of western and northern blots were analyzed using image processing software (Bioprofil Vilber Loumart).

Mean values were derived from experiments performed at least three times as described above. Single factor ANOVA was used to evaluate the results of western blotting, northern blotting or spreading assays on basement membrane substrates in the presence of monoclonal antibodies. Additionally, post-hoc testing, using the Newman-Keuls (SNK) test was used to compare the differences between the selected pairs of means. For adhesion experiments on basement membrane substrates in the presence of monoclonal antibodies, mean values were derived from hexaplicates within each experiment. ANOVA and SNK multiple range tests were used to compare the differences between the selected pairs of means within each experiment. In all instances P<0.05 was considered statistically significant.

We first examined whether culturing HGEC on increasing concentrations of basement membrane protein substrates affected the expression of foot process-related PC. Western blotting demonstrated that in HGEC cultured on increasing concentrations of laminin, PC-protein levels increased in a dose dependent manner (22.5-65%), compared to control cells which were cultured in the absence of substrate (Fig. 1B,E). In addition, when HGEC were cultured on increasing concentrations of GBM, a ∼30% increase in the expression of PC was observed at all concentrations of GBM used (Fig. 1C,F). However, culturing cells on increasing concentrations of collagen IV, did not affect PC-protein levels, even at the highest concentration of collagen IV used (Fig. 1A,D).

Similar results were obtained when the cells were synchronized at G_o, in which they presumably occur in situ, and cultured either on tissue culture plastic, or on basement membrane substrates (data not shown). Basement membrane substrates also increased PC mRNA levels, as demonstrated by northern blot analysis (Fig. 2). When HGEC were cultured on the highest concentration of laminin or GBM (20 μg/cm²), the observed increase in mRNA levels was ∼80% and ∼40% respectively compared to control cells (Fig. 2A, lanes 3, 4; Fig 2B). However, culture of cells on 20 μg/cm² of collagen IV had no effect on PC-mRNA levels (Fig. 2A, lane 2; Fig. 2B).

The increase in expression of PC with laminin or GBM, was also confirmed by flow cytometric analysis. In HGEC cultured on plastic approximately 40% of cells expressed PC. However, when cells were grown on the highest concentration (20 μg/cm²) of laminin or GBM substrate, an increase in the percentage of cells expressing PC (∼60% and ∼30% respectively), as well as cell surface-associated PC (∼65% and ∼35% respectively) was observed compared to control cells (Fig. 3b,c). Collagen IV did not significantly alter the surface expression of PC (Fig. 3a).

Confocal microscopy was then used to monitor cell-surface distribution of PC in non-permeabilized HGEC cultured in the absence or presence of basement membrane substrates. As shown in Fig. 4, PC was distributed over the entire cell surface in all cases. PC staining had a pattern of continuous rimming around the cells and a punctate pattern at the basal surface. When HGEC were cultured on GBM or laminin substrates, cell borders had increased expression of PC, compared to control cells. Additionally, in HGEC cultured on GBM or laminin, the basal punctate pattern was more prominent and occurred several planes away from the base (2-3 motor steps) in these instances. In control cells the punctate pattern was faint and appeared at the basal surface (first motor step from base) (Fig. 4, panels a-c).

We also examined the effect of GBM or laminin substrates on the expression of nephrin, an additional differentiation marker for podocytes. Nephrin was detected by western blotting in HGEC cultured on plastic. When cells were cultured on 20 μg/cm² of laminin or GBM, there was a significant increase in nephrin levels by ∼55% and ∼37% respectively compared to control cells (Fig. 5A). Confocal microscopic analysis in HGEC revealed a mainly basolateral distribution of nephrin. Moreover, there was an increased cell-surface nephrin expression in cells grown on laminin or GBM substrates compared to controls (Fig. 5B).

We investigated the appearance of junctions in HGEC cultured on GBM by immunolabeling the junction-related ZO-1 protein. Confocal microscopy revealed that ZO-1 was present mainly on the lateral cell surface (Fig. 6). However ZO-1 failed to form a complete rimming around the cell border, instead it appeared in a dotted or discontinuous pattern. The same pattern albeit more faint, was also observed in the basal or apical cell surface.

Foot processes were retracted in diabetic nephropathy as a result of increased glucose concentrations (Kanwar et al., 1996; Kanwar et al., 1997). As podocalyxin has been linked to the formation/maintenance of foot processes (Doyonnas et al., 2001), we examined the expression of PC in HGEC grown in the presence of 25 mM glucose. Immunoblotting, northern blot analysis and flow cytometry, demonstrated that PC expression was almost totally suppressed in HGEC cultured in 25 mM glucose, compared to cells grown in the presence of 5 mM glucose (Fig. 7A-C). Moreover, the presence of laminin or GBM did not reverse the dramatic downregulation provoked by increased glucose concentrations (Fig. 7D). Decreased PC (∼45%) was also observed in glomeruli of streptozotocin (STZ)-diabetic rats, compared to control glomeruli (Fig. 7E).

Scanning and transmission electron microscopy were used in order to investigate the morphology of HGEC grown either in the absence or presence of laminin or GBM substrate. SEM demonstrated that cells cultured on uncoated coverslips were flattened and did not extend processes (Fig. 8A, panel a), whereas the same cells cultured on laminin had ruffled lateral processes (Fig. 8A, panel b). In HGEC cultured on GBM, this effect was more pronounced, and several elongated processes were observed in the main body, extending from the central area towards the periphery (Fig. 8A, panel c). Under high magnification, the edges of control cells were flat and adherent (Fig. 8B, panel a). In the presence of laminin or GBM, cellular edges were irregular with a ruffled appearance, prominent numerous end processes, and spike-like formations best visualized at an angle of 45° suggesting that they represent parts of the basal or basolateral cell surface (Fig. 8B, panels b,c). TEM revealed that in HGEC cultured on laminin or GBM substrates, locally raised areas appeared in the basal surface often in areas of cell-cell contact around 1 μm from the basal surface, compared to control cells, which were totally flattened (Fig. 9). In addition, in certain instances, the basal cell surface was locally lifted from the substrate (Fig. 9c, arrowhead).

We also examined the morphology of HGEC cultured in 25 mM glucose and additionally grown in the absence or presence of 20 μg/cm² laminin or GBM substrate. As it has been shown, under high glucose conditions, PC protein levels were minimal at all instances (see Fig. 7D). SEM in HGEC grown in 25 mM glucose in the absence of substrate demonstrated that cells were more spread out than HGEC grown in 5 mM glucose (Fig. 10a,b). Moreover laminin and GBM induced the formation of body processes by HGEC grown in increased glucose concentrations, resembling the main processes formed by cells grown in 5 mM glucose in the presence of laminin or GBM substrates (Fig. 10c,d). However laminin or GBM did not promote any specialized modifications of the basal or basolateral cell surface, such as ruffles or areas locally lifted from the substrate similar to those observed in control cells (5 mM glucose). High-glucose cells were flattened and cellular edges remained adjacent to the substrate, whereas occasionally fine end processes were observed (Fig. 10e-g).

Finally, we examined the functional role of PC in cell-matrix interactions. We performed cell adhesion assays on collagen IV or laminin, in the presence or absence of anti-PC monoclonal antibody using HGEC grown in either 5 or 25 mM glucose. For these experiments we selected substrate concentrations that promoted similar levels of cell adhesion. Thus, the concentrations corresponded to approximately half of the maximal cell adhesion in the case of collagen IV, and maximal adhesion in the case of laminin (data not shown). As shown in Fig. 11A, HGEC cultured in 5 mM glucose adhered to a greater extent to both collagen IV and laminin in the presence of anti-PC antibody, compared to control cells (adhesion in the absence of antibody). This increase in adhesion depended on the concentration of anti-PC antibody and the observed changes were statistically significant in antibody dilutions 1:50-1:200. Maximal increase in adhesion was observed with the lowest antibody dilution, and corresponded to ∼40% increase in the case of collagen IV, and ∼60% increase in the case of laminin. The presence of anti-HLA antibody did not affect the binding of cells to collagen IV or laminin (data not shown). In addition, HGEC grown in 25 mM glucose expressing minimal levels of PC were allowed to adhere to collagen IV and laminin in the presence or absence of anti-PC antibody. As shown in Fig. 11B, the presence of anti-PC antibody had no effect on the extent of cell adhesion to either substrate.

We examined the effect of anti-PC antibody on integrin-mediated cell adhesion to collagen IV. We selected the blocking antibody P5D2 directed against the β1 integrin subunit; β1 integrin is present at high density on the surface of 97% HGEC (Kitsiou et al., 2003). Anti-β1 mAb resulted in inhibition of cell binding in a dose-dependent manner. An almost complete inhibition of adhesion (∼95%) was achieved at the lowest antibody dilution (1:400) whereas ∼65% inhibition of adhesion was observed when anti-β1 mAb was diluted 1:1500. The above-mentioned inhibition of adhesion was observed in both HGEC expressing minimal (grown in 25 mM glucose) or higher levels of PC (cultured in 5 mM glucose) (Fig. 12). In the presence of saturating concentrations of anti-β1 antibody (1:400) in combination with anti-PC antibody, there was no change in the extent of cell adhesion, which was minimal in HGEC expressing PC (Fig. 12A). However, when anti-β1 mAb was used at a dilution of 1:1500 in combination with serial dilutions of anti-PC, a significant reversal of the inhibitory effect of anti-β1 mAb was observed, resulting in increased binding. The observed increase in cell binding was concentration dependent, reaching ∼90% at the lowest anti-PC dilution used (1:50), when compared to cell adhesion in the presence of anti-β1 integrin mAb alone (Fig. 12A). In HGEC expressing minimal levels of PC, the presence of anti-β1 monoclonal antibody in combination with anti-PC antibody did not reverse the inhibitory effect of anti-β1 mAb (Fig. 12B).

To investigate the functional role of PC in cell binding to matrix components further, we examined HGEC spreading on both laminin and collagen IV substrates, in the presence or absence of anti-PC antibody. As shown in Fig. 13A, most of the cells grown in 5 mM glucose, incubated in the absence of anti-PC antibody (control), and bound to collagen IV or laminin, remained rounded with a surface area of <200 μm² (75% and 88% respectively). Only a small percentage of adherent cells (25% and 12% respectively) were flattened with a large surface area (>200 μm²). In the presence of anti-PC antibody, the percentage of cells with a surface area >200 μm² was increased by ∼1.5-fold on both collagen IV and laminin compared to the control. Anti-HLA antibodies did not have any effect on cell spreading. HGEC expressing minimal levels of PC (cultured in 25 mM glucose), spread more effectively and appeared more flattened with increased surface area on both substrates, compared to HGEC grown in 5 mM glucose, which express higher PC levels (Fig. 13B). In HGEC cultured in 5 mM glucose, the percentage of flattened cells on collagen IV or laminin with a surface area >200 μm² was 25% and 12% respectively. The corresponding percentage of flattened HGEC cultured in 25 mM glucose, was 45.5% and 23% respectively. In addition, in HGEC cultured in 25 mM glucose, the presence of anti-PC antibody did not affect cell spreading (Fig. 13B).

Podocyte differentiation is an elaborate process. During the early stages of glomerular development (S-shaped body, comma shaped body) the GBM contains laminin 1 (α1β1γ1), and collagen IV with the composition of (α1)₂(α2). With further development there is a switch in GBM composition, so that mature GBM is enriched in laminin 10/11 (α5β1γ1/α5β2γ1) whereas collagen IV chains α1/α2 and α3-α5 occupy the subendothelial and central parts of this basement membrane, respectively (Kashtan and Michael, 1996; Miner, 1999). In mice lacking laminin β2 chain, foot processes appear abnormal and fused (Noakes et al., 1995). In contrast, in mice lacking α3 chain of collagen IV, there are no changes in podocyte morphology (Miner and Sanes, 1996).

PC and nephrin are differentiation markers of podocytes, and their expression and distribution has been associated with podocyte development. These proteins initially appear on the apical or basolateral surface of podocytes respectively, during the stage of the S-shaped body. As the visceral epithelial cells mature, PC migrates towards the lateral cell surface along with junctional complexes. Migration of PC coincides with foot process formation whereas nephrin becomes concentrated in slit diaphragms (Schnabel et al., 1989; Kawachi et al., 2002). When PC expression was suppressed, foot processes did not form, whereas the corresponding nephrin knock-out mice had poorly developed processes (Doyonnas et al., 2001; Putaala et al., 2001). Foot process interdigitation was also suggested to be involved in the expression of other podocyte components such as α3-integrin, CD2AP, nephrin, α-actinin or podocin (Shirato, 2002; Aaltonen et al., 2001; Luimula et al., 2002).

The above-mentioned studies suggest that: (1) GBM components including laminin and podocyte components such as PC, nephrin, α3 integrin, podocin, and CD2AP, play a morphogenetic role in foot process formation; (2) There is a correlation between foot process formation/maintenance and the expression of the related podocyte proteins; (3) As PC and nephrin were initially detected during the stage of the S-shaped body in glomerular development, when the GBM contained laminin 1 and collagen IV (α1)₂(α2), these isoforms of laminin and/or collagen IV should be involved in the expression of PC and nephrin.

In this report we have used immortalized human podocytes (HGEC), which express several podocyte markers including nephrin, in order to study the effects of basement membrane components on the expression of PC and nephrin. Cells were cultured on several substrates including EHS-derived intact laminin (corresponding to laminin 1), intact collagen IV (composed of (α1)₂(α2) chains) or intact human GBM. It was previously reported that these substrates promote cell adhesion, differentiation, or gene expression in cultured podocytes (Krishnamurti et al., 1996; Krishnamurti et al., 1997; Krishnamurti et al., 2001; Kobayashi et al., 2001; Kitsiou et al., 2003) and other epithelial cells of the kidney (Chen et al., 1996; Karamessinis et al., 2002). We showed that EHS-derived laminin (laminin 1), up-regulated the expression of PC by HGEC. In response to laminin substrates, the increase of PC expression was dose-dependent, whereas GBM containing laminin, increased the expression of PC to a similar extent at all concentrations used. However, EHS-derived collagen IV did not have any effect on PC expression. It remains to be substantiated whether laminin 10/11 or α3-α6 chains of collagen IV, components of the mature GBM, have similar effects with their EHS-derived counterparts, on the expression of PC. The ability of laminin to up-regulate the expression of PC in contrast to collagen IV may be due to the different functional roles that these components play in the glomerulus. It has been suggested that collagen IV is prerequisite for the structural integrity of the GBM, whereas laminin is important for podocyte maturation (Miner and Sanes, 1996; Müller and Brandli, 1999). Laminin and GBM also increased the expression of nephrin. Our data provide evidence that laminin increased the expression of foot process-related PC and slit pore-related nephrin, thus confirming further an important role of laminin in foot process formation.

We also examined PC and nephrin distribution in HGEC using confocal microscopy. PC was abundant on the entire cell surface with a continuous rimming around the cells and a punctate pattern on the basal surface. In control cells this pattern was very faint. In the presence of laminin and GBM substrates, the punctate pattern became substantially more prominent indicating increased PC expression in these areas. Moreover in cells grown on laminin or GBM substrates, the basal punctate pattern occurred several planes away from the base. HGEC grown on laminin or GBM substrates also had increased expression of nephrin, mainly at their basal or lateral cell surfaces compared to control cells.

In situ, PC occurs only above the filtration slits of podocytes, and nephrin is expressed in slit diaphragms (Schnabel et al., 1989; Somlo and Mundel, 2000). The observed non-polarized occurrence of PC and nephrin in cultured HGEC could be attributed to leaky or incomplete junctional complexes, that were reported to occur in these cells (Delarue et al., 1991). This was demonstrated by the distribution of ZO-1-containing junctional areas in HGEC cultured on GBM. Confocal microscopic analysis of ZO-1 revealed a dotted pattern mainly along the lateral surface confirming that junctional complexes between these cells were discontinuous. The non-polarized occurrence of nephrin in cultured podocytes is also confirmed by other studies (Saleem et al., 2002).

In diabetes, foot processes are effaced, as a result of high glucose levels (Kanwar et al., 1996; Pagtalunan et al., 1997; Kanwar et al., 1997). As there is an obvious relevance between foot process formation/maintenance and the expression of related proteins (Aaltonen et al., 2001; Luimula et al., 2002), we used increased glucose concentrations in order to examine PC expression. In HGEC cultured in 25 mM glucose, on plastic, laminin or GBM substrates, PC expression was dramatically down-regulated. Glucose-induced decrease in PC expression was also confirmed by immunohistochemistry in glomeruli of STZ-diabetic rats, a well-established experimental model for studying diabetic nephropathy (Velasquez et al., 1990). PC reduction was also reported to occur in kidney biopsies from diabetic patients (Koop et al., 2003).

We also examined the importance of laminin and GBM in podocyte morphology using electron microscopy. Cells cultured on laminin formed lateral processes, which appeared more elongated and better differentiated in the presence of GBM substrates. It was previously reported that both EHS-laminin and laminin 10/11 induced process formation in murine podocytes, whereas collagen IV did not result in modifications of podocyte morphology (Kobayashi et al., 2001). According to our results, laminin contributed to process formation by HGEC, but the presence of GBM was required to induce changes resembling the in situ appearance of podocytes, possibly due to a more native conformation of the whole GBM structure or because it contained all the components of the mature native basement membrane.

We also examined the nature of the punctate pattern of PC expression, which was observed by confocal microscopy in HGEC grown on GBM and laminin. In SEM images, under high magnification, control cells were totally flattened, whereas laminin or GBM induced characteristic modifications; cellular edges were locally raised with ruffled surfaces and thin, elongated end processes as well as spike-like formations appeared in basal and/or basolateral cell surfaces. TEM images indicated that segments of the basal cell surface abutting the substrates alternated with segments raised away from the substrate, resulting in a ruffled configuration. The areas raised away from the substrate occurred at approximately the same distance from the basal area as the punctate pattern of cells, which was observed by confocal microscopy (∼1 μm). Thus, confocal microscopy indicated that PC was also expressed in segments of the basal area, which did not abut the substrate.

As laminin and GBM induced ruffling of the basal/basolateral cell surfaces in PC-expressing HGEC, we examined whether these substrates had a similar effect in PC-deficient cells. Low magnification SEM images revealed that PC-deficient cells (cells cultured in high glucose), were more spread out compared to their control counterparts (cells grown in 5 mM glucose). In addition, laminin or GBM induced lateral processes in PC-deficient HGEC, which were similar to those induced by the same substrates in PC-expressing HGEC. However, high magnification SEM images indicated that PC-deficient HGEC cultured either in the absence or presence of GBM or laminin remained flattened. Therefore, PC-deficient HGEC spread more effectively and did not have areas locally raised from the substrate in their basal or basolateral surfaces. These observations suggest that the presence of PC plays a role in the differentiated conformation of the basal surface of HGEC.

In order to investigate the possible role of PC in the formation of locally raised areas in HGEC grown on laminin or GBM substrates, cell adhesion and spreading assays were performed on EHS-derived laminin or collagen IV, in the absence or presence of anti-PC monoclonal antibody. These experiments were carried out in PC-expressing HGEC (grown in 5 mM glucose) or PC-deficient cells (grown in 25 mM glucose). In the absence of antibody, cells grown in 5 mM glucose adhered to a greater extent to laminin or collagen IV compared to cells cultured in 25 mM glucose. We have previously reported that high level glucose-induced changes in adhesion of HGEC to basement membrane substrates were related to alterations in integrin expression (Kitsiou et al., 2003). In the present report, the spreading of HGEC on laminin and collagen IV (in the absence of anti-PC antibodies) was increased in PC-deficient cells compared to HGEC expressing PC, thus confirming the observations revealed by SEM. Additionally, HGEC expressing PC, adhered and spread to a greater extent to both collagen IV and laminin in the presence of anti-PC antibody, compared to control cells (absence of antibody). As expected, anti-PC antibody had no effect on PC-deficient cell binding or spreading. These observations strongly suggest that the presence of PC inhibits cell-matrix interactions in vitro.

It has been demonstrated that the sialic acid residues of PC are essential for its antiadhesive properties regarding cell-cell interactions, as in vitro inducible ectopic expression inhibited cell aggregation, which was reversed by treatment with sialidase. The reported cell-cell antiadhesive properties of PC could play an important role in the morphology of podocytes (Takeda et al., 2000). Antiadhesive effects have also been demonstrated for other mucins such as episialin, epiglycanin, and leukosialin. The antiadhesive properties of mucins were often revealed by the use of antibodies directed against their extracellular domain (protein core) (Ligtenberg et al., 1992; Cyster and Williams, 1992; DeSmet et al., 1993; Kemperman et al., 1994; Wesseling et al., 1995). Cyster and Williams (1992) suggested that monoclonal antibodies, which recognize the protein core of mucin leukosialin, overcame the repulsion created by the negative charges of this sialoprotein, thus allowing other adhesion molecules to come into play inducing cell-cell interactions. In our case, it is possible that anti-PC antibody had a similar effect in HGEC insofar as cell-matrix interactions were concerned. This antibody recognizes a domain of the protein core of PC (Kershaw et al., 1997) corresponding in part, to the previously described mucin-like domain of this sialoprotein (Sassetti et al., 2000).

The mechanism by which mucin-like molecules prevent the adhesion is poorly understood. It has been proposed that mucins interfere with cell-cell or cell-matrix adhesion either through steric hindrance, by masking adhesion molecules and/or by charge repulsion (Ligtenberg et al., 1992; Kemperman et al., 1994; Wesseling et al., 1995; Takeda et al., 2000). In the present report, we provide evidence that in addition to charge repulsion, antiadhesive properties of PC may compete with β1 integrin-mediated adhesion. We demonstrated that β1 integrin subunit mediates the binding of HGEC to collagen IV. In the presence of anti-β1 monoclonal antibody the process of adhesion was inhibited in a dose-dependent manner, reaching ∼95% inhibition of adhesion at the highest (saturating) antibody concentration. In the simultaneous presence of anti-PC mAb and non-saturating concentrations of anti-β1 mAb, anti-PC increased the proportion of adherent cells and thus partly reversed the inhibitory effect of anti-β1 mAb. However, when anti-β1 mAb was used at saturating concentrations, the simultaneous presence of anti-PC mAb could not overcome the effect of anti-β1. Thus, saturation of β1 integrins by anti-β1 antibody did not allow anti-PC antibody to create an adhesive effect. Taken together these data indicate that PC exerts an antiadhesive effect by competing with β1-containing integrin dimers. In contrast, in PC-deficient cells (25 mM glucose) the presence of anti-PC antibody did not reverse the inhibitory effect of anti-β1 integrin antibody, even when the latter was used at a low concentration. We conclude that the process of adhesion of HGEC to matrix is the balanced result of the pro-adhesive effects of β1-containing integrins and antiadhesive properties of PC, which takes place by a functional interplay between these two cell surface components.

In podocytes in vivo, β1 integrin subunit forms primarily a heterodimer with α3 integrin subunit (Adler, 1992; Baraldi et al., 1994). The α3β1 complex has been identified in situ in the basal and apical parts of foot processes (Regoli and Bentayan, 1997). The simultaneous presence of PC and α3β1 integrins in the same position (above the filtration slits) may imply a functional interplay of PC-β1 integrin in the process of adhesion/anti-adhesion in vivo, which is responsible, at least in part, for lifting the cell surface away from the GBM, and thus contributes to the formation and/or maintenance of foot processes.

In conclusion, our main results demonstrated that laminin and the whole GBM regulate the expression of PC by interacting with HGEC. Furthermore, confocal microscopy revealed an enhanced non-homogeneous pattern of PC distribution, at a distance from the basal cell surface, suggesting the formation of locally lifted areas in the presence of laminin or GBM substrates. SEM and TEM showed that basal and/or basolateral cell surfaces were ruffled confirming the existence of lifted areas. Furthermore, the data obtained using anti-PC monoclonal antibody, as well as PC-deficient cells, provide evidence that PC inhibits cell binding and spreading to basement membrane components, such as collagen IV and laminin. The antiadhesive effect is possibly the result of a functional interplay between β1-containing integrins and PC, which depending on the prevailing molecule results in either adhesion or lack of adhesion, and contributes to the specialized conformation of these cells.

The authors are indebted to Agnes Davidovits, Dept. of Clinical Pathology, University of Vienna, Austria for help with rat anti-PC antibody (PI 118) and K. Tryggvason, Karolinska Institute, Stockholm, Sweeden for anti-nephrin monoclonal antibody, M. Mauer, Dept. Pediatrics, Medical School, University of Minnesota, USA for providing kidneys from streptozotocin-rats, G. Drossopoulou and P. Handris for their expert technical assistance in performing immunohistochemistry and confocal microscopy respectively.