(Received for publication, March 4, 1997, and in revised form, April 18, 1997)
From the Departments of Pediatrics and
¶ Internal Medicine, University of Michigan,
Ann Arbor, Michigan 48109
Human renal cortex and heart cDNA libraries were screened for a human homolog of rabbit PCLP1 using the rabbit PCLP1 cDNA as a probe. Clones spanning 5869 base pairs with an open reading frame coding for a 528-amino acid peptide were obtained. The putative peptide contains a potential signal peptide and a single membrane-spanning region. The extracellular domain contains multiple potential sites for N- and O-linked glycosylation and 4 cysteines for potential disulfide bonding similar to rabbit PCLP1. On Northern blot a major transcript is seen at 5.9 kilobases. Antibodies to this protein show a doublet at 160/165 kDa on Western blots of human glomerular extract and a pattern of intense glomerular staining and vascular endothelial staining on immunofluorescence of human kidney sections. Comparison of the rabbit and human peptide sequences shows a high degree of identity in the transmembrane and intracellular domains (96%) with a lower degree of identity in the extracellular domain (36%). An antibody to the intracellular domain reacted across species (human, rabbit, and rat) and recognized both rabbit PCLP1 and rat podocalyxin. An interspecies Southern blot probed with a cDNA coding for the intracellular domain showed strong hybridization to all vertebrates tested in a pattern suggesting a single copy gene. We conclude that this cDNA and putative peptide represent the human homolog of rabbit PCLP1 and rat podocalyxin.
The glomerular epithelial cell (podocyte) is a highly differentiated cell with characteristic interdigitating foot processes covering the outer aspect of the glomerular basement membrane. The space between these foot processes is spanned by a modified tight junction (slit diaphragm) and provides the large surface area for filtration. The foot processes are covered on their non-sole (apical) surface with an anionic glycocalyx. A major component of this glycocalyx is thought to be podocalyxin, a sialoglycoprotein described by Kerjaschki et al. in rat (1).
The potential importance of the podocyte's anionic glycocalyx is well established. In children with minimal change disease the podocyte polyanion as visualized by histochemical staining is markedly reduced (2). In experimental models neutralization of the glomerular polyanionic charge with polycations or desialylation with neuraminidase is associated with proteinuria (3-5). Sialylation of podocalyxin decreases in the puromycin aminonucleoside model of nephrosis in the rat (6).
Efforts to define the podocyte polyanion further have included the description of a major sialoglycoprotein of the human podocyte by Kerjaschki and colleagues (7). Both the lectin binding properties of this protein and its distribution on the surface of podocyte foot processes and the luminal surface of vascular endothelial cells are similar to rat podocalyxin. However, this sialoglycoprotein differs from rat podocalyxin in its apparent molecular mass on SDS-PAGE1 (a 165/170-kDa doublet in contrast to a 140-kDa band for rat podocalyxin) and in its peptide digest pattern. Antibodies to this molecule and those to rat podocalyxin have been reported not to react across species (7), and neither of these molecules has been cloned to date.
We reported recently the cloning and characterization of a rabbit sialoglycoprotein with a size, staining characteristics, and tissue distribution similar to those of rat podocalyxin. We named this protein rabbit podocalyxin-like protein 1 (PCLP1) (8). Using the rabbit PCLP1 cDNA as a probe we have now cloned a human podocalyxin-like protein (PCLP). In this report we characterize the molecular structure of human PCLP and define its relationship with rabbit PCLP1 and rat podocalyxin.
Total
RNA from was prepared from renal cortex by modification of the
CsCl/guanidine isocyanate method of Chirgwin et al. (9) as
described previously (10). The kidneys used for RNA preparation were
from a cadaver organ donor whose kidney could not be used for
transplantation and a patient with congenital nephrotic syndrome (Finnish type) undergoing a pretransplant nephrectomy. Libraries were
produced from these preparations by the custom library services of
Stratagene, Inc. (La Jolla, CA). In addition, a commercial human heart
cDNA library was used (Stratagene). These libraries were initially
screened using rabbit PCLP1 cDNA as probes (11). Sequencing was
done by the method of Sanger et al. (12) using the Sequenase
kit (U. S. Biochemical Corp.) with modifications described previously
(8, 13). Additional automated sequencing was performed by the
sequencing core at the University of Michigan on a fee-for-service
basis. All clones shown were sequenced in both directions. 5-Rapid
amplification of cDNA ends (RACE) was performed using 1 µg of
renal cortical RNA isolated from a normal kidney or a kidney of a
patient with congenital nephrotic syndrome and a kit from Life
Technologies Inc. according to the manufacturer's protocol with
dimethyl sulfoxide 10% (v/v) added to the PCRs. The PCR product was
ligated into the pCR 2.1 vector (Invitrogen, San Diego) and used to
transform INV
F
competent cells. Data base management, sequence
analysis, and comparison were 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
"non-redundant" data base from the Brookhaven Protein Data Bank,
GenBank, EMBL, PIR, and SwissProt data bases (14).
The following primers were used to PCR amplify regions of rabbit PCLP1 and human PCLP. To make the fusion protein used to raise antibodies, a portion of the human extracellular, transmembrane, and intracellular domain (base pairs 1004-1835) was PCR amplified using the primers TTTGGATCCCAGATGCCAGCCAGCTCTACG and TTTGAATTCTTAGAGGTGCGTGTCTTCCTC. A portion of the human extracellular domain (base pairs 1004-1492) was PCR amplified using the primers TTTGGATCCCAGATGCCAGCCAGCTCTACG and TTTGAATTCCTTCATGTCACTGACCCCTGC. A region of the rabbit PCLP1 extracellular domain (bases 490-1002) was PCR amplified using the primers TTTGAATTCGGGCGTCAGTGTCGAAGGCTT and TTTGGATCCAACACTACACCCATGACGACG. A region of the rabbit PCLP1 intracellular domain (bases 1726-2912) was PCR amplified using the primers TTTGAATTCAAGTCCCTGAGTTCTCTATGC and TTTGGATCCTGCTGCCACGAGCGCCTCTCC. The expression vector pGEX-KT and the PCR products were digested with EcoRI and BamHI, purified, and ligated. Fusion protein expression was performed as described by Smith and Johnson (15). Fusion protein purification was performed as described by Guan and Dixon (16).
Northern and Southern Blot AnalysisA human multiple tissue
Northern blot (CLONTECH Laboratories, Palo Alto, CA) containing 2 µg
of poly(A)+ RNA/lane was probed with
[32P]dCTP-labeled human podocalyxin cDNA or -actin
cDNA. Prehybridization, hybridization, and washings were carried
out per the ExpressHyb protocol (CLONTECH). The probes for this
analysis were the human PCLP cDNA from base pair 1004 to 2029 which
was PCR amplified with [32P]dCTP using the primers
TTTGGATCCCAGATGCCAGCCAGCTCTACG and ACAAGAGGAATCTGGACA and a random
[32P]dCTP-labeled
-actin cDNA as a RNA loading
control. Conditions for the final wash were 0.1 × SSC and 0.1%
SDS at 50 °C.
For Southern blot a portion of the human PCLP cDNA (base pairs 1603-1835) was PCR amplified using the primers TTTGAATTCAAGTCCCTGAGTTCTCTATGC and TTTGAATTCTTAGAGGTGCGTGTCTTCCTC with [32P]dCTP. A commercial Interspecies Zoo-Blot (CLONTECH Laboratories) containing 5 µg of genomic DNA/lane was probed with the human PCLP cDNA probe as described for Northern blot. Conditions for the final wash were 0.1 × SSC and 0.1% SDS at 50 °C.
Preparation of Monoclonal AntibodiesMonoclonal antibodies
(mAbs) 3D3, 4F10 and 2A4 were produced from BALB/C mice immunized with
purified human podocalyxin-glutathione S-transferase (GST)
fusion protein containing the intracellular, transmembrane, and a part
of the extracellular domain of human PCLP by standard methods as
described previously (17). The resulting hybridomas grown out in
96-well plates were selected and subcloned based on immunofluorescence
pattern assayed on cryostat sections of human renal cortex. The
anti--galactosidase (clone Gal-40) used as a control IgM antibody
was obtained from Sigma. The VWM control IgM antibody was provided by
the Hybridoma Core Facility at the University of Michigan. The
monoclonal antibody 5A (anti-rat podocalyxin) was kindly provided by
Robert Orlando of the University of California, San Diego. Other
antibodies used were as described previously (8, 18, 19). All mAbs were
IgG except antibodies 2A4, 4F10, VWM, and anti-
galactosidase, which
were IgM.
The human kidneys used were as described for RNA preparation. Glomerular isolation and extraction were performed at 4 °C by differential sieving as described previously (18). Rabbit glomeruli were isolated from New Zealand White rabbits (2.0-2.5 kg) by iron oxide magnetization as described previously (20). Rat glomeruli were isolated from Harlan Sprague Dawley rats by progressive sieving using 180-, 106-, and 75-µm sieves as described by Salant et al. (21). For glomerular extraction, 5 × 104 glomeruli were suspended in 1 ml of phosphate-buffered saline 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 (19). Glomerular extracts were analyzed by SDS-PAGE. Blots (model SBD-1000 Polyblot, American Bionetics, Hayward, CA) were performed as described previously (19). Western blots were developed using the ECL reagent (Amersham Corp.).
Immunoprecipitation StudiesImmunoprecipitations were
carried out using a modification of published protocols (22). For
immunoprecipitation experiments rat glomerular extract was preabsorbed
with anti-mouse IgM (µ chain-specific)-agarose beads (Sigma). The
mAbs (2A4 and 4F10 or control IgMs anti--galactosidase and VWM) were
incubated with anti-mouse IgM agarose beads, washed four times with
Tris-buffered saline, and incubated with the preabsorbed rat glomerular
extract for 20 min at room temperature and then overnight at 4 °C on
a rotor. Beads were washed six times with Tris-buffered saline
containing protease inhibitors (CompleteTM protease inhibitor,
Boehringer Mannheim). Samples were prepared as described for Western
blot.
Kidney segments were cut on a cryostat for subsequent analysis by indirect immunofluorescence performed using the primary antibodies described for Western blot as described previously (23). For the blocking experiment shown the primary antibody was preincubated with 20 µg of the immunizing fusion protein.
Initial human PCLP
clones were obtained by screening with rabbit PCLP1 cDNA. These
positive clones were used for subsequent screening of the three human
cDNA libraries used (one heart and two renal cortical libraries).
Thirty-four clones were obtained by this method, and two additional
clones were produced by an anchored PCR strategy using RNA from a
normal kidney (clone RACEN5) and the kidney from a patient undergoing a
pretransplant nephrectomy for congenital nephrotic syndrome (clone
RACEC12). Fig. 1A shows six clones that were
used to assemble the nucleotide sequence.
The cDNAs spanned 5869 base pairs (Figs. 1 and 2).
An initiator methionine (base pairs 251-253) was identified using the
following criteria. (a) The sequence was consistent with
Kozak's consensus sequence (first methionine in the open reading
frame, purine in position 3) (24). (b) A likely site for
signal peptidase cleavage at amino acid 21 is preceded by hydrophobic
amino acids in 15 out of 20 positions (25). (c) The 250 base
pairs upstream of the putative start methionine is highly GC-rich
(78%) and contains numerous CpG "islands" compatible with this
region being a 5
-untranslated region (26).
The first stop codon in the open reading frame occurred at base pairs 1835-1837 (Fig. 2). This would correspond to a 1584-base pair or 528-amino acid open reading frame. After removal of the 21-amino acid signal peptide the peptide is calculated to have a molecular mass of 54 kDa. A Blast search of the non-redundant data bases showed significant similarities only to rabbit PCLP1 (8).
Analysis of the derived amino acid sequence shows a single 26-amino acid hydrophobic region compatible with a single membrane spanning domain (Fig. 1B and Fig. 2) similar to rabbit PCLP1 (Fig. 1C) (8). The region COOH-terminal to the hydrophobic putative transmembrane domain contains positively charged amino acids as is typically described for the cytoplasmic side of transmembrane proteins (27). We have shown previously that rabbit PCLP1 is a transmembrane protein with an NH2-terminal extracellular domain, and we have aligned human PCLP in this fashion.
Analysis of the Intracellular DomainSimilar to rabbit PCLP1 the intracellular domain contains 75 amino acids, which contained one potential protein kinase C site (amino acid 457) and two potential casein kinase II phosphorylation sites (amino acids 488 and 516). Overall this region was highly acidic (pI = 4.2) with the final 10 amino acids containing 4 aspartic acid and 3 glutamic acid residues (Fig. 1, B and C and Fig. 2) as described previously for rabbit PCLP1 (8).
Analysis of the Extracellular DomainThe 406 amino acids of
the extracellular domain (after signal peptide cleavage) were analyzed
for potential structural features and sequence motifs using the
Peptidestructure and Motifs programs. There were five potential sites
for N-linked glycosylation. In the extracellular region from
amino acids 22 to 295 the high serine and threonine content (39%)
provides numerous sites for potential O-linked glycosylation
some with proline in the 1 and +3 positions, which is seen in many
O-glycosylation sites (28). There are three serine-glycine
sites and one serine-glycine-glycine site for potential
glycosaminoglycan attachment (Fig. 3), but it should be
noted that these sites lack acidic residues 2-3 amino acids amino-terminal to the serine, which has been shown to increase the
acceptor activity for glycosaminoglycans (29, 30). Four cysteines for
potential disulfide linkage were present in the extracellular domain
(Fig. 1B and Fig. 2) as described previously for rabbit
PCLP1 (8).
In the extracellular domain there were three differences in nucleotide sequence of the clones which resulted in amino acid changes. The clone RACEN5 derived from RNA from a patient with congenital nephrotic syndrome had as nucleotide 435 a G, making amino acid 62 an arginine instead of a threonine, and nucleotides 315-320 were missing deleting amino acids 23 and 24. Both clones in this region were obtained by the RACE methodology so cloning artifacts cannot be excluded. Nucleotide 837 was a C in clone NP3, making amino acid 196 a serine instead of a leucine.
Northern Blot AnalysisNorthern blot analysis was performed on RNA from multiple tissues to determine transcript size and the tissue distribution of human PCLP mRNA expression (Fig. 3). A 32P-labeled PCR product from base 1004 to base 2029 was used as probe. A major band was seen at approximately 5.9 kilobases with minor bands at 9.6 and 4.4 kilobases. The mRNA transcript expression was highest in the kidney, pancreas, and heart. Lesser amounts were present in the placenta, lung, and skeletal muscle; a low but detectable signal was present in brain and liver. This tissue distribution is similar to that seen with rabbit PCLP1 on Northern blot (8).
Characterization of Antibodies to Human PCLP Fusion ProteinsThe cDNA coding for base pairs 1004-1836 of the
human PCLP protein was PCR amplified and ligated into the expression
vector PGEX-KT. The fusion protein was purified with
glutathione-agarose affinity chromatography and used to immunize mice.
Three monoclonal antibodies 3D3, 4F10, and 2A4 appeared to recognize a
protein in human renal cortical sections on immunofluorescence (Fig.
4, A-C, antibodies 4F10 and 3D3 shown) with
a glomerular epithelial cell and vascular endothelial cell distribution
similar to rabbit PCLP1 (8) and rat podocalyxin (1). The signal was
abolished when the fusion protein used to raise the mAbs was
preincubated with the antibodies (Fig. 4D, mAb 3D3
shown).
To define further the epitopes recognized by these antibodies the
immunizing fusion protein containing the intracellular, transmembrane
and a portion of the extracellular domain of human PCLP (base pairs
1004-1836), a fusion protein containing a portion of the extracellular
domain of human PCLP (base pairs 1004-1492), and a fusion protein
containing the intracellular domain of rabbit PCLP1 (bases 1726-2912)
(Fig. 5C) were analyzed by Western blot (Fig.
5, A and B). The mAb 3D3 recognized the fusion
protein used to raise the antibodies (Fig. 5B, lane
C) and the extracellular domain fusion protein, indicating that
the epitope recognized by 3D3 is coded for by this region of the
extracellular domain (Fig. 5B, lane B). In
contrast, mAbs 2A4 and 4F10 recognized the fusion protein used to raise
the antibodies and the rabbit intracellular domain fusion protein but
did not recognize the extracellular domain of the human PCLP fusion
protein (Fig. 5A, mAb 2A4 shown), indicating that the
epitopes recognized by 2A4 and 4F10 are coded for by the intracellular
domain in both rabbit PCLP1 and human PCLP.
Comparison of Human PCLP and Rabbit PCLP1
We have reported
previously the cloning and molecular characterization of a rabbit
protein with size, staining, and tissue distribution similar to those
of rat podocalyxin (8), and we have called this protein rabbit
podocalyxin-like protein 1. On immunofluorescent staining of kidney
sections antibodies to rabbit PCLP1 and human PCLP showed a strong
signal in the glomerulus and on the endothelium of the blood vessels in
their respective species. Overall the cDNA nucleotide sequences for
rabbit PCLP1 and human PCLP showed 72% identity. Both sequences were
highly GC-rich in the 5-untranslated region. Rabbit PCLP1 has an open reading frame of 531 amino acids excluding an alternative splice, and
human PCLP had an open reading frame of 528 amino acids. The peptide
cores of rabbit PCLP1 and human PCLP were calculated at 55 and 54 kDa,
respectively. Both molecules had a 21-amino acid putative signal
peptide, multiple sites for potential O-linked and
N-linked glycosylation, along with 4 cysteines for potential disulfide interactions in their extracellular domains. The position of
3 of the potential N-linked glycosylation sites was well
conserved and the position of the 4 cysteines was identical in both
species relative to the transmembrane region (Figs. 1 and 2). In the
intracellular domains both rabbit PCLP1 and human PCLP had 2 potential
casein kinase II and a protein kinase C phosphorylation sites at
identical positions.
Despite the similarities between rabbit PCLP1 and human PCLP the
overall amino acid identity between rabbit PCLP1 and human PCLP was
only 48%, whereas transmembrane and intracellular domains had 96%
amino acid identity. The extracellular NH2-terminal regions showed a low degree of identity (36% identity) except for the putative
signal peptide regions (75% identical). This is shown graphically in
the similarity plot in Fig. 6. This degree of
dissimilarity is reported for the murine/human homologs of CD28 (68%
identical) (31) and the murine/human homologs of the CD28 ligand B7
(44% identical) (32). Similarly the human mucosal addressin cell adhesion molecule (MAdCAM-1) shows 39% identity to its murine homolog,
but both bind specifically to their
4
7-integrin ligand (33).
Relationship of Rat Podocalyxin, Rabbit PCLP1, and Human PCLP
On Western blot of human glomerular extract both mAbs 2A4
and 3D3 recognized a 160/165-kDa doublet similar to that described previously by Kerjaschki et al. (7) for the protein
recognized by mAb PHM5 (Fig. 7 A and
C). The anti-PCLP mAbs (2A4 and 4F10), which recognized the
conserved intracellular domain of rabbit PCLP1 and human PCLP (Fig. 5),
also reacted across species. This is shown for mAb 2A4, which
recognized bands on blots of rabbit and rat glomerular extract with the
same approximate molecular mass as PCLP1 detected with mAb 4B3 in
rabbit and podocalyxin detected with mAb 5A in rat (Fig. 7,
A and C). In contrast the mAb 3D3 did not reacted
across species.
Immunoprecipitations of rat glomerular extracts were performed with the anti-PCLP mAbs 2A4 and 4F10 or control IgM antibodies to define further the protein recognized by the anti-PCLP mAbs in the rat. The anti-podocalyxin mAb 5A recognized a 140-kDa band on Western blot of rat glomerular extract immunoprecipitated with anti-PCLP mAbs 2A4 and 4F10 (Fig. 7E). The mAb 5A also recognized a 140-kDa band (rat podocalyxin) in the glomerular extract lane. The anti-human PCLP mAb 2A4 also recognized a 140-kDa band on Western blot of rat glomerular extract (Fig. 7, C and E). No 140-kDa band was seen in the control IgM immunoprecipitation lane or on Western blot developed with a control antibody (Fig. 7, E and F). These results indicate that mAbs 2A4 and 4F10 together recognized and immunoprecipitated rat podocalyxin. We conclude that rat podocalyxin contains the conserved intracellular epitopes present in human PCLP and rabbit PCLP1 which are recognized by mAbs 2A4 and 4F10.
Conservation of the PCLP Intracellular Domain between SpeciesSince the intracellular domain of PCLP appears to be
highly conserved, we have used this region of the human PCLP cDNA
to probe for sequences homologous to this domain in the genomic DNAs of
a wide range of species (Fig. 8). Under high stringency
conditions one restriction fragment (two for monkey) of genomic DNA was
found to hybridize to this probe for all eukaryotes tested except
yeast. This observation supports the conclusion that the intracellular domain of PCLP is highly conserved in vertebrates and suggests that
podocalyxin-like protein is a single copy gene.
Kerjaschki and colleagues have described previously a human protein similar to rat podocalyxin which has a size and tissue distribution similar to those of the protein characterized in this report (7). However this molecule recognized by mAb PMH5 differs from rat podocalyxin in apparent molecular mass on SDS-PAGE gels and the sizes of proteolytic fragments seen on peptide digests. Kerjaschki and colleagues therefore suggested that this human sialoprotein may be evolutionary distinct but have a function similar to podocalyxin (rat). Our findings would support the contention that the podocalyxin-like protein from man would have a different pattern from rat podocalyxin on proteolytic peptide maps because of a poorly conserved amino acid sequence in the extracellular domain, and it would have a different apparent molecular mass on SDS-PAGE. We conclude that these proteins (rat podocalyxin, human PCLP, and rabbit PCLP1) are derived from a single related gene in rat, rabbit, and human which codes for a molecule with a highly conserved intracellular and transmembrane and a variable extracellular domain.
The calculated putative size of human PCLP and rabbit PCLP1 peptide is similar, but the apparent mass on SDS-PAGE differs considerably suggesting that post-translational modification differs considerably among species. We reported previously for rabbit PCLP1 that the discrepancy between the calculated size of the peptide and the observed mass on SDS-PAGE is accounted for by post-translational modifications, most likely glycosylation. The interspecies differences in apparent molecular mass are likely to be due to differences in glycosylation as well. We speculate that the major function of the PCLP extracellular domain is to support large negatively charged carbohydrate residues which contribute to the podocyte's anionic glycocalyx. The composition of the extracellular domain peptide framework shows considerable drift between species and even within species as reported here. If this is the case we speculate that this molecule might be a target for immune recognition on the endothelial surface of transplanted organs.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U97519[GenBank].
We are grateful to Lisa Riggs for technical help, Jill Baney (University of Michigan Multipurpose Arthritis Center) for performing fusions and hybridoma production (supported by National Institutes of Health Grant P560AR20557), the General Clinical Research Center at the University of Michigan, funded by a grant (M01RR00042) from the National Center for Research Resources for supplying data base search facilities, the Tissue Procurement Core at the University of Michigan for providing human tissue used in this report, and Dr. Lawrence Holzman for advice and helpful discussions.