Molecular cloning, expression, and characterization of podocalyxin-like protein 1 from rabbit as a transmembrane protein of glomerular podocytes and vascular endothelium.

Podocytes are responsible in part for maintaining the size and charge filtration characteristics of the glomerular filter. The major sialoprotein of the podocyte foot process glycocalyx is a 140-kDa sialoprotein named podocalyxin. Monoclonal antibodies raised against isolated rabbit glomeruli that recognized a podocalyxin-like protein base upon size, Alcian blue staining, wheat germ agglutinin binding, and distribution in renal cortex were used to expression clone cDNAs from a rabbit glomerular library. On Northern blot the cDNAs hybridized to a 5.5-kilobase pair transcript predominantly present in glomerulus. The overlapping cDNAs spanned 5,313 base pairs that contained an open reading frame of 1,653 base pairs and were not homologous with a previously described sequence. The deduced 551-amino acid protein contained a putative 21-residue N-terminal signal peptide and a 26-amino acid transmembrane region. The mature protein has a calculated molecular mass of 55 kDa, an extracellular domain that contains putative sites for N- and O-linked glycosylation, and a potential glycosaminoglycan attachment sites. The intracellular domain contains potential sites for phosphorylation. Processing of the full-length coding region in COS-7 cells resulted in a 140-kDa band, suggesting that the 55-kDa core protein undergoes extensive post-translational modification. The relationship between the cloned molecule and the monoclonal antibodies used for cloning was confirmed by making a fusion protein that inhibited binding of the monoclonal antibodies to renal cortical tissue sections and then raising polyclonal antibodies against the PCLP1 fusion protein that also recognized glomerular podocytes and endothelial cells in tissue sections in a similar distribution to the monoclonal antibodies. We conclude that we have cloned and sequenced a novel transmembrane core glycoprotein from rabbit glomerulus, which has many of the characteristics of podocalyxin. We have named this protein podocalyxin-like protein 1.

Podocytes are responsible in part for maintaining the size and charge filtration characteristics of the glomerular filter. The major sialoprotein of the podocyte foot process glycocalyx is a 140-kDa sialoprotein named podocalyxin. Monoclonal antibodies raised against isolated rabbit glomeruli that recognized a podocalyxinlike protein based upon size, Alcian blue staining, wheat germ agglutinin binding, and distribution in renal cortex were used to expression clone cDNAs from a rabbit glomerular library. On Northern blot the cDNAs hybridized to a 5.5-kilobase pair transcript predominantly present in glomerulus. The overlapping cDNAs spanned 5,313 base pairs that contained an open reading frame of 1,653 base pairs and were not homologous with a previously described sequence. The deduced 551-amino acid protein contained a putative 21-residue N-terminal signal peptide and a 26-amino acid transmembrane region. The mature protein has a calculated molecular mass of 55 kDa, an extracellular domain that contains putative sites for N-and O-linked glycosylation, and a potential glycosaminoglycan attachment site. The intracellular domain contains potential sites for phosphorylation. Processing of the full-length coding region in COS-7 cells resulted in a 140-kDa band, suggesting that the 55-kDa core protein undergoes extensive post-translational modification. The relationship between the cloned molecule and the monoclonal antibodies used for cloning was confirmed by making a fusion protein that inhibited binding of the monoclonal antibodies to renal cortical tissue sections and then raising polyclonal antibodies against the PCLP1 fusion protein that also recognized glomerular podocytes and endothelial cells in tissue sections in a similar distribution to the monoclonal antibodies. We conclude that we have cloned and sequenced a novel transmembrane core glycoprotein from rabbit glomerulus, which has many of the characteristics of podocalyxin. We have named this protein podocalyxin-like protein 1.
Podocalyxin is the major sialoprotein of the glycocalyx lining the foot processes of glomerular epithelial cells (podocytes) where it is thought to maintain foot process structure and function in part by virtue of its negative charge (1). Podocalyxin was first identified by Kerjaschki,Sharkey,and Farquhar (1) as an Alcian blue staining 140-kDa sialoprotein. Subsequent studies have shown that the negative charge is contributed by sulfate as well as by sialic acid (2) and that podocalyxin is present on the surface of endothelial cells as well as glomerular epithelial cells (3,4).
The interdigitating foot processes of neighboring podocytes create the huge intercellular surface area for glomerular filtration. The importance of charge for maintenance of this structure has been demonstrated by experiments where charge neutralization with polycations or desialylation with neuraminidase is associated with loss of the interdigitating foot process structure of the podocyte (5)(6)(7). Similarly, the induction of podocyte foot process effacement in rats by injection of puromycin aminonucleoside is accompanied by leakiness of the glomerular filter, podocyte foot process detachment, and a reduction in sialylation of podocalyxin (8,9).
From these studies we have the concept of podocalyxin as an intensely negatively charged molecule present in large amounts on the podocyte foot processes and lining the surface of endothelial cells. At these sites podocalyxin may function in part by charge repulsion to maintain the distance between foot processes of neighboring cells and between circulating cells and the endothelium ("anti-adhesion molecule"). We report here the molecular structure of a transmembrane glycoprotein that we have cloned and sequenced as part of an effort to understand in molecular terms how the glomerular filter works and how it becomes dysfunctional in children and adults with the nephrotic syndrome. We have named this protein podocalyxinlike protein 1 (PCLP1). 1

MATERIALS AND METHODS
Preparation of Monoclonal Antibodies-mAbs 5F7 and 4B3 were produced from BALB/C mice immunized with isolated rabbit glomeruli (10,000/immunization) by standard methods as described previously (10). The resulting hybridomas grown out in 96-well plates were selected and subcloned based on immunofluorescence pattern assayed on cryostat sections of rabbit renal cortex.
Glomerular Isolation, Protein Extraction, and Western Blots-Rabbit glomeruli were isolated from New Zealand White rabbits (2.0 -2.5 kg) by iron oxide magnetization as described previously (11). For glomerular extraction, 5 ϫ 10 4 glomeruli were suspended in 1 ml of PBS containing 1% Triton X-100, 0.1% SDS, 2 mM phenylmethylsulfonyl fluoride, 5 mM N-ethylmaleimide, 2 mM EDTA, and 8 M urea and sonicated in six short bursts of 10 s as described previously (10). Iron oxide and debris were removed by centrifugation. The supernatant was stored at Ϫ70°C. Glomerular extracts were analyzed by SDS-PAGE. For Alcian blue staining, gels were fixed in 25% isopropanol overnight, stained with 0.1% Alcian blue in 2% acetic acid, and destained with 1% acetic acid. Blots (model SBD-1000 polyblot, American Bionetics, Hayward, CA) were performed as described previously (12). After transfer, lanes were excised and nitrocellulose strips were incubated with 125 Iwheat germ agglutinin (WGA) (ICN, Costa Mesa, CA). Some strips were preincubated with neuraminidase (0.05 units/ml) (Sigma) in 50 mM sodium acetate at 37°C for 24 h and then incubated with 125 I-WGA. Western blots were incubated with the primary antibody 4B3, 5F7, or a control mAb (BB5) that does not recognize rabbit tissues. All mAbs were subclass IgG1. Alternatively blots were incubated with polyclonal guinea pig anti-PCLP1 fusion protein antibody at a dilution of 1:1000 in Tris-buffered saline with 1% bovine serum albumin (Sigma). Preimmune serum from the guinea pig at the same dilution was used as a control. After a wash step, blots were incubated with 125 I-anti-mouse IgG (DuPont NEN) or 125 I-protein A (ICN) and then exposed for autoradiography.
Immunoelectron Microscopy-Gold-labeled immunoelectron microscopy was performed on paraformaldehyde-lysine periodate fixed, methanol-dehydrated, and Lowicryl K4M-embedded thin rabbit kidney cortex sections on formvar-coated nickel grids as described previously (13). mAbs 4B3 or BB5 were used as primary antibodies, followed by a polyclonal rabbit anti-mouse antibody (Jackson Immunoresearch, Philadelphia, PA) and then gold conjugated polyclonal goat anti-rabbit antibody (Calbiochem, San Diego, CA) (14). Before each antibody incubation, the sections were washed and incubated for 5 min in 1.0% skim milk (Life Technologies, Inc.) in 0.01 M PBS with 0.02% polyethylene glycol 6000 and 0.02% Tween 80. The final rinse was with distilled water, and all incubations and washes were done at room temperature. Sections were counterstained with uranyl acetate and examined using a Hitachi H-7000 transmission microscope (13).
Immunoperoxidase and Immunofluorescence Studies-Kidney segments (1 mm) were microwaved in 0.5% glutaraldehyde according to the protocol of Login et al. (15). Sections (2 m) were cut on a cryostat for subsequent analysis using the immunoperoxidase methodology according to Vector Laboratories (Burlingame, CA). For immunofluorescence, cryostat sections of rabbit renal cortex were used. Immunoperoxidase and indirect immunofluorescence was performed using the primary antibodies described for Western blot.
cDNA Library, Screening, and Sequence Analysis-Total RNA was prepared from isolated glomeruli and renal cortex by modification of the CsCl/guanidine isocyanate method of Chirgwin et al. (16) as described previously (17). Polyadenylated RNA isolated from these preparations was used to construct random primed directional glomerular and cortical libraries in Uni-ZAP XR and Lambda ZAP II vectors through the custom library services of Stratagene, Inc. (La Jolla, CA). These libraries were screened using the 4B3 and 5F7 mAbs by the method of Young and Davis (18). The cDNA inserts were rescued in Bluescript SKphagemid by the in vivo excision protocol of Stratagene or by excision and ligation into Bluescript SK-plasmid. Subsequent screens were performed using cDNA inserts as probes (19). Double-stranded DNA sequencing was done by the dideoxy chain termination method of Sanger et al. (20) using the Sequenase kit (U. S. Biochemical Corp.). For regions of high secondary structure dimethyl sulfoxide 10% (v/v) was added to the reactions (21). All clones shown were sequenced in both directions. 5Ј rapid amplification of cDNA ends was performed using 2 g of glomerular RNA and a kit from Life Technologies, Inc. according to the manufacturer's protocol. The PCR product was ligated into the pAMP vector (Life Technologies Inc.) and used to transform DH5␣competent cells. Data base management and sequence analysis was done with version 8.0 of the Wisconsin Sequence Analysis Package (Genetics Computer Group, Madison, WI). Data base searches were performed using the Blast Network Service from the National Center for Biotechnology Information on the "nonredundant" data base from the Brookhaven Protein Data Bank, Genbank, EMBL, PIR, and SwissProt data bases (22).
Construction and Purification of Fusion Proteins-The fusion protein derived from the extracellular domain was prepared as follows. A region of the PCLP1 extracellular domain (bases 490-1002) was PCR-amplified using the primers TTTGAATTCGGGCGTCAGTGTCGAAGGCTT and TTTGGATCCAACACTACACCCATGACGACG. The expression vector pGEX-KT and the PCR product were digested with EcoRI and BamHI, purified, and ligated. The ligation mixture was used to transform competent Escherichia coli TG1. The reading frame was confirmed by DNA sequencing. Fusion protein expression was performed as described by Smith and Johnson (23). Fusion protein purification was performed as described by Guan and Dixon (24).
Immunoadsorption Studies-mAbs 4B3 and 5F7 culture supernatants were assayed by dilution using immunofluorescence intensity as an end point. In the experiment shown, mAb 4B3 (1/25 dilution in PBS containing 1% bovine serum albumin) was incubated with the purified PCLP1-GST fusion protein (20 g) or a fusion protein derived from the extracellular domain of GLEPP1 (20 g) constructed in the same fusion protein system (12).
Northern Blot Analysis-cDNA fragments were labeled with [ 32 P]dCTP using a random prime DNA labeling kit (Boehringer Mannheim). Prehybridization, hybridization, and washings were carried out as described by Sambrook et al. (19). Conditions for the final wash were 0.2 ϫ SSC at 60°C.
Expression Construct, in Vitro Transcription/Translation, and Transfection Experiments-A PCLP1 cDNA from bases 1-2943 was ligated into the mammalian expression vector pcDNA3 (Invitrogen, San Diego, CA). This construct (0.5 g) or pcDNA3 vector without insert (0.5 g) was used for [ 35 S]methionine-labeled in vitro transcription and translation using the TnT T7 coupled reticulocyte lysate (Promega, Madison, WI) system with and without 1.5 l of canine pancreatic microsomal membranes (Promega) in 25-l reactions per the manufacturers protocol.
For transfection experiments COS-7 cells were plated at 2.7 ϫ 10 5 cells/60-mm dish overnight in DMEM (BioWhittaker, Walkersville, MD) with 10% newborn calf serum. Cells were washed once with serumfree DMEM before DMEM with Lipofectamine (Life Technologies, Inc.) 6 l/ml and 2 g of either the PCLP1 mammalian expression construct or pcDNA3 vector. After 6 h of incubation, one volume of DMEM with 10% fetal calf serum and 10% newborn calf serum was added. Cells were lysed and extracted at 24 h after transfection in 200 l of PBS containing 1% Triton X-100, 0.1% SDS, 2 mM phenylmethylsulfonyl fluoride, 5 mM N-ethylmaleimide, 2 mM EDTA, and 8 M urea. Protein content of the extracts were analyzed by a modified Bradford technique (Bio-Rad, Richmond, CA).
Partial Purification of PCLP1 and Digestion with Glycosidases-Rabbit glomerular extract from 450,000 glomeruli was prepared in 9 ml of glomerular extraction buffer (PBS containing 1% Brij 35, 2 mM EDTA, leupeptin (2 g/ml), and aprotinin (2 g/ml)). The extract was centrifuged at 100,000 ϫ g for 1 h at 4°C, and the supernatant was allowed to bind to 5 ml of wheat germ agglutinin agarose (Pharmacia Biotech Inc.) equilibrated with glomerular extraction buffer and transferred to a column, washed with 10 column volumes of PBS containing 0.1% Triton X-100, and eluted with glomerular extraction buffer containing 120 mM N-acetyl glucosamine. The first three 5-ml fractions containing the PCLP1 were dialyzed against four changes of 10 mM Tris-HCl, pH 7.5, and finally against distilled water. These fractions were then lyophilized, and the dry powder was dissolved in a minimum volume of 20 mM Tris-HCl, pH 7.5.
The wheat germ agglutinin purified fraction of PCLP1 (8 l containing 16 g of total protein) was mixed with 1 l of 2% SDS and 1 l of 1 M 2-mercaptoethanol and kept in boiling water bath for 5 min. The mixture was diluted to 200 l in 20 mM Tris-HCl, pH 7.5. 20-l aliquots of the above preparation were digested with heparinases I and II, heparitinase mix (0.5 units each), chondroitinase ABC (0.5 units), N-glycosidase F (0.4 units), O-glycosidase (2 milliunits), and neuraminidase (0.1 unit) by incubating at room temperature overnight followed by an additional 3 h at 37°C (endo-␤-galactosidase and N-and O-glycosidase were obtained from Boehringer Mannheim; the other enzymes were obtained from Sigma). Multiple enzyme digestions were done similarly by adding the enzymes simultaneously to the aliquot of denatured PCLP1 buffer mix. For endo-␤-galactosidase digestions, 8 l of wheat germ agglutinin fraction was boiled as described above but diluted with sterile distilled water. Endo-␤-galactosidase digestion was performed in 50 mM sodium acetate buffer, pH 5.8, with 5 milliunits of the enzyme. Aliquots without the addition of any enzyme incubated under identical conditions served as controls. After incubation an equal volume of 2 ϫ SDS gel loading buffer with 2-mercaptoethanol was added, and the reaction mixtures were kept in boiling water bath for 5 min, separated on SDS-PAGE, and immunoblotted with 4B3 and BB5 mAbs as described above.

Characterization of mAbs 5F7 and 4B3
Recognizing PCLP1 mAbs prepared from splenocytes of mice immunized against whole rabbit glomeruli were selected based upon immunofluorescent patterns on cryostat sections of rabbit renal cortex. mAbs 5F7 and 4B3 showed the following properties.
Immunostaining Pattern on Cryostat Sections of Renal Cortex-mAbs 5F7 and 4B3 produced an intense reaction product in glomeruli and also stained endothelium of arteries, veins, and peritubular capillaries (Fig. 1a).
Western Blotting-Blots of extracts from isolated glomeruli developed with mAbs 5F7 and 4B3 showed a major band at 140 kDa under reducing and nonreducing conditions (Fig. 1b).
Alcian Blue Binding-A major band at 140 kDa was also recognized by Alcian blue (Fig. 1b). 125 I-WGA Binding-A major band at 140 kDa was also bound by 125 I-WGA on a Western blot of rabbit renal glomerular extract.
Abolition of 125 I-WGA Binding by Prior Neuraminidase Treatment-125 I-WGA binding was abolished by prior neuraminidase treatment of the blot (Fig. 1b). In all reduced gels in this experiment, a second band at lower molecular mass was recognized by each detection system, thereby providing further data that the same target molecule was being recognized by the mAb, Alcian blue, and WGA.
Ultrastructural Localization of mAb Binding to Podocyte Foot Processes and Endothelium- Fig. 1c shows immunogold labeling of PCLP1 localized to the nonbasement membrane surface of the epithelial foot processes and to a lesser extent the endothelial plasma membrane. No immunogold staining was observed in the absence of primary antibodies or the presence of nonspecific anti-rabbit IgG. These characteristics are identical to those defined by Kerjaschki, Sharkey, and Farquhar (1) for rat podocalyxin. We therefore elected to call the molecule recognized by these mAbs podocalyxin-like protein 1.

Cloning and Sequencing of PCLP1 cDNAs
A rabbit glomerular cDNA library was constructed in Uni-Zap XR vector and screened with mAbs 5F7 and 4B3 in a bacterial expression system. Six positive cDNA clones were isolated with overlapping nucleotide sequence. 42 additional cDNAs were obtained by using these cDNAs as probes and by anchored PCR strategies. Fig. 2 (upper panel) shows in diagrammatic form nine selected overlapping cDNAs used to construct the nucleotide sequence.
Northern blot analysis was performed to determine the approximate length of the transcript and to show the pattern of mRNA tissue expression. Northern blots performed using a 3.5-kb PCLP1 clone showed a major band at approximately 5.5 kb with minor bands at 7.1 and 4.4 kb (Fig. 3). Screening of different tissues shows the relatively much greater amount of mRNA in glomerulus as compared with other tissues (Fig. 3). This relative distribution of PCLP1 mRNA is similar to PCLP1 protein distribution as assessed by immunofluorescence using the 5F7 and 4B3 mAbs.
The full length of the nucleotides sequenced from the overlapping cDNA clones spans 5313 base pairs (Figs. 2 and 4). An initiator methionine (base pairs 304 -306) was identified by the following criteria, (a) The amino acid sequence was consistent with Kozak's consensus sequence (first methionine in the open reading frame, purine in position Ϫ3) (25); (b) this methionine was followed by a 21-amino acid putative signal peptide containing 12 consecutive hydrophobic amino acids (ALA-LAALLLLLL) (Fig. 4) (26); (c) the presence of numerous CpGrich "islands" is compatible with this region being 5Ј-untranslated sequence (underlined in Fig. 4) (27).
A stop codon was found at base pairs 1957-1959, indicating the end of the open reading frame (Fig. 4). This would correspond to an open reading frame of 1653 base pairs or a total of 551 amino acids. If the putative signal peptide (21 amino acids) is cleaved off in post-translational processing, then the mature protein will be 530 amino acids long, and the N-terminal amino acid would be glutamine. This conclusion has not been confirmed by N-terminal sequencing. The putative PCLP1 protein has a calculated molecular mass of 54.9 kDa and a calculated isoelectric point of 4.8. A Blast search of the available data bases showed no significant similarities to published nucleotide or protein sequences.
Analysis of the derived amino acid sequence showed a single 26-amino acid hydrophobic sequence compatible with a single transmembrane region (Figs. 2 and 4). Immediately C-terminal to the putative transmembrane region are positively charged amino acids (HQRLSHRK) as is typically found at the cytoplasmic side of a transmembrane region (28). This orientation relative to the cell membrane is also supported by the fact that the clone Jo3 isolated using the mAbs codes for the region of the molecule N-terminal to the transmembrane region and that these two mAbs also bind to nonpermeabilized isolated rabbit glomeruli as assessed by immunofluorescence (data not shown). Thus the region of the molecule N-terminal to the putative transmembrane domain must be extracellular.

Analysis of the Extracellular Domain
As indicated above, the N-terminal 21 amino acids of the PCLP1 open reading frame are typical of a signal peptide sequence. The remaining 429 amino acids of the extracellular region were analyzed for the presence of potential O-and N-linked glycosylation sites. Analysis of the putative PCLP1 amino acid sequence using the Motifs program of the Genetics Computer Group sequence software package indicates there are three potential sites for N-linked glycosylation (Fig. 4). One of these potential N-linked sites (amino acid 333) is flanked by leucines as has been described in chick lumican (29). The putative PCLP1 amino acid sequence has five serine-threonine clusters providing potential acceptor sites for O-linked oligosaccharides (Fig. 4). Rat podocalyxin has been suggested to contain at least two N-linked oligosaccharide groups (8) and O-linked oligosaccharide groups (1, 2). There is one potential glycosaminoglycan attachment site (amino acids 215-218) that contains the consensus sequence (SGXG) found in small proteoglycans (30). There are four other glycine-serine or serineglycine groups that could also potentially serve as attachment sites for glycosaminoglycan chains, although they lack associated acidic residues that are frequently but not always found adjacent to glycosaminoglycan attachment sites (31,32).
There are several other features of the extracellular domain that appear to be noteworthy. A 24-amino acid span (amino acids 245-268) is very rich in serine and proline (21 of 24 amino acids) and lies in the middle of the extracellular domain (Figs.  2 and 4). There are 4 cysteines available in the extracellular domain for potential disulfide linkage to form two loop structures (Figs. 2 and 4). One clone (GN2) from the glomerular library contained an additional 60-nucleotide insert (1324 -1384) coding for 20 additional amino acids (Figs. 2 and 4). This putative alternately spliced region contains an unusual series of 3 alternating glutamines followed by 3 alternating glutamic acids (QRQSQGEGETE). The extracellular domain has 5 doublets of acidic amino acids (281-282, 366 -367, 430 -431, 445-446, and 448 -449) similar to clusters of acidic amino acids thought to mediate calcium binding in other proteins (33,34).

Analysis of the Intracellular Domain
The intracellular domain contains 75 amino acids (amino acids 477-551). Overall this region is highly acidic (pI ϭ 4.3). There are two potential casein kinase II phosphorylation sites (amino acids 511 and 539) and one potential protein kinase C phosphorylation site (amino acid 481). At the C-terminal end FIG. 2. Diagrammatic illustration of PCLP1 cDNA and derived protein structure. Top, diagrammatic representation of the cDNAs used to derive the PCLP1 nucleotide structure. Clone Jo3 is one of the cDNAs cloned with the mAbs 5F7 and 4B3. Clone RACE 10 was cloned by PCR techniques. The remainder of the cDNAs shown were obtained using labeled cDNAs as probes. All clones were sequenced in both directions. Clone GN2 contains a 60-nucleotide amino acid putative alternative splice region in the coding region (see below). Bottom, diagrammatic representation of PCLP1 protein structure derived from the nucleotide sequence and aligned with a Kyle-Doolittle amino acid hydropathy plot. A single putative 26-amino acid transmembrane region is shown (solid black box). The N-terminal domain contains a hydrophobic 21-residue putative signal peptide (horizontal striped box). In addition are shown a serine-proline-rich region (diagonal striped box), a putative alternate splice region (vertical striped box), three potential sites for N-linked glycosylation (arrows), a potential glycosaminoglycan attachment site (V), cysteines for possible disulfide linkage (C), acidic areas (lightly shaded boxes), and a highly acidic C-terminal region (darkly shaded box). The binding region for the mAbs is between amino acids 63 and 246 and is based on the distribution of the cDNAs identified with the mAbs. This region is known to be extracellular because the mAbs 4B3 and 5F7 bind to nonpermeabilized glomeruli. there is a highly acidic 10-amino acid region containing 4 aspartic acid and 3 glutamic acid residues (pI ϭ 3.5).

Fusion Protein and Immunoadsorption Studies
To demonstrate that the cDNA sequence obtained codes for a protein recognized by the mAbs, a PCLP1 fusion protein was prepared. The region of the cDNA corresponding to amino acids 63-233 (Fig. 4) was amplified by PCR, ligated into the expres-sion vector pGEX-KT, and expressed in frame with GST in E. coli TG1. After purification by glutathione affinity chromatography, the fusion protein was used to demonstrate the capacity to immunoadsorb the binding of the mAbs 5F7 and 4B3 to kidney tissue (Fig. 5A). The fusion protein of another podocyte protein, GLEPP1 (12), did not immunoadsorb the binding of mAbs 5F7 and 4B3. Thus, this region of the cDNA (bases 490-1002) codes for an epitope recognized by mAbs 5F7 and 4B3. The PCLP1-GST fusion protein was also used to immunize guinea pigs to raise polyclonal antibodies. On Western blot both the polyclonal antiserum and mAb 4B3 recognized a major band at approximately 140 kDa under both reducing and nonreducing conditions (Fig. 5B). The fusion protein was able to immunoadsorb the binding of both polyclonal and monoclonal antibodies to PCLP1 as assessed by immunofluorescence and Western blot (Fig. 5, A and B). Furthermore both the polyclonal anti-PCLP1 fusion protein serum and mAb 4B3 showed identical staining on immunofluorescence of the glomerulus and peritubular capillaries to that seen with the monoclonal antibodies (Fig. 5C). We conclude that the cDNAs cloned code for a molecule with the same size and distribution as that recognized by the monoclonal antibodies.

The Effect of Processing on the Apparent Size of PCLP1 Protein
The calculated molecular mass of the PCLP1 protein core from the derived amino acid sequence is about 55 kDa. In vitro transcription/translation of PCLP1 cDNA (bases 1-2943) in a T7 coupled reticulocyte lysate system shows the cDNA to encode a protein with an apparent molecular mass of 70 kDa (Fig.  6, left panel). Thus the unprocessed PCLP1 molecule migrates more slowly on SDS-PAGE than expected from its calculated size. Similar differences in calculated and apparent molecular mass have been reported for chromogranin A, an acidic protein similar to PCLP1 (35), and for the core proteins of proteoglycans such as syndecan (36) and neurocan (37). After processing by canine pancreatic microsomal membranes, the apparent molecular mass increases to about 80 kDa, representing minor processing of the PCLP1 core protein (Fig. 6, left panel). Expression of the PCLP1 protein in a mammalian epithelial cell line (COS-7 cells) resulted in production of a form of PCLP1 with an apparent molecular mass of about 140 kDa on Western blot (Fig. 6, right panel). No bands were seen with COS-7 cells alone or with COS-7 cells transfected with pcDNA3 vector (Fig.  6). Western blot with control mAb BB5 (Fig. 6, lane I) of an extract of COS-7 cells transfected the PCLP1 expression construct shows no band. This experiment clearly demonstrates the PCLP1 cDNA codes for a 140-kDa protein when fully processed. The PCLP1 from glomerular extract appeared to be slightly smaller than the PCLP1 from transfected COS-7 cells ( Fig. 6, lane H); however, this apparent difference was eliminated when the extracts were run in buffers with identical urea concentrations in subsequent experiments (data not shown). Thus the processed PCLP1 molecule from transfected COS-7 cells and from glomerular extract have the same apparent molecular mass of 140 kDa.

Modification of Partially Purified PCLP1 from Isolated Glomeruli by Glycosidase Digestion
Podocalyxin has been previously shown to be susceptible to enzymatic digestion with N-glycanase and O-glycanase and to contain both O-and N-linked oligosaccharide groups (2). Digestion with WGA affinity purified PCLP1 with N-glycosidase (Fig. 7, lane F) and neuraminidase (Fig. 7, lane I) shows shifts in molecular mass as described for podocalyxin (2). No shift was seen on digestion with O-glycosidase (Fig. 7, lane G).
That PCLP1 binds Alcian blue and has potential glycosaminoglycan attachment site(s) (Figs. 1B and 4) suggests it may be a proteoglycan. However, digestion with heparinases (Fig. 7, lane B), chondroitinase ABC (Fig. 7, lane C), heparinase and chondroitinase ABC (Fig. 7, lane D), and endo-␤-galactosidase (Fig. 7, lane L) yielded no significant change in apparent molecular mass. Similar results were obtained for digestions performed in the absence of SDS or 2-mercaptoethanol (data not shown). As a positive control blots of the same digests were probed with mAb 4C3, which we have recently reported and used to clone the podocyte membrane protein tyrosine phosphatase GLEPP1 (12). These experiments clearly showed changes in molecular mass of GLEPP1 on incubation with chondroitinase ABC, N-glycosidase F, O-glycosidase, and neuraminidase (data not shown). Thus the conditions used allowed digestion in the same incubation mix of another glomerular epithelial glycoprotein. This result suggests that the glycosidic linkages of PCLP1 are not accessible to the glycosidic enzymes, that PCLP1 might be an unusual glycoconjugate not susceptible to the enzymes used, or that the post-translational increase in apparent molecular mass is not due to glycosylation.

Conclusions About the Glycosylation of PCLP1
The difference in the apparent molecular mass of the PCLP1 from in vitro transcription/translation system (70 kDa) and transfected cells (140 kDa) reflects a high degree of post-translational modification. However, as noted above, commonly used glycan linkages were not detected by digestion experiments described above. Farquhar and colleagues have determined that podocalyxin is highly glycosylated with 20% hexose, 4.5% sialic acid (1), and additional N-acetylglucosamine (2). Endothelial podocalyxin has been suggested to contain mixed glycans or an unusual oligosaccharide structure (38). PCLP1, with only about 50% of its apparent molecular mass due to its peptide core, also appears to be heavily glycosylated, is likely sulfated based on its Alcian blue binding, and may contain an unusual oligosaccharide/glycosaminoglycan structure based on its resistance to glycosidase digestion.

The Relationship of PCLP1 to Podocalyxin
We cannot conclude from these studies that PCLP1 (characterized in rabbit) is identical to podocalyxin (characterized in rat) because anti-podocalyxin (rat) IgG is reported not to react across species (39). The antibodies we have identified and developed also do not react across species. However, PCLP1 does appear to be similar to podocalyxin in all respects examined including molecular mass, sialylation, Alcian blue binding, and tissue distribution. Definitive confirmation will be available once rat podocalyxin is cloned.
The identification of the nucleotide sequence for this major protein distributed on glomerular epithelial cell and vascular endothelial cell surfaces is an important step toward understanding the role this molecule may play in glomerular and endothelial cell biology and pathology. Note Added in Proof-During the review of this manuscript, the following short human sequence submissions to GenBank™ were found to be homologous to the 3Ј-untranslated region of PCLP1: T87928, R99975, R99976, H65205, T87719, and H64714.  1 unit); lane L, Endo-␤-galactosidase (5 milliunits). As positive controls done in the same incubation mixture, Western blots showed that the GLEPP1 protein (which is copurified with PCLP1 from isolated glomeruli by WGA affinity chromatography) showed digestion with N-glycodidase, O-glycosidase, neuraminidase, and chondroitinase ABC, thereby confirming that there were not inhibitors for these enzymes present in the incubation mixture and that these enzymes were active under the conditions used (data not shown).