Tissue Antigens ISSN 0001-2815

Production and characterization of murine monoclonalantibodies against human podocalyxinR. B. Rodrıguez, N. Butta, S. Larrucea, S. Alonso & R. Parrilla

Department of Physiopathology and Human Molecular Genetics, Centro de Investigaciones Biologicas (CSIC), 28040-Madrid, Spain

Key words

anti-podocalyxin antibodies; expression

pattern; HEK293; K562; Meg01; Tera-1 cells

Correspondence

Roberto Parrilla

Centro de Investigaciones Biologicas (CSIC)

Ramiro de Maeztu 9

Madrid-28040

Spain

Tel/Fax: 34 91 5349154

e-mail: rparrilla@cib.csic.es

Received 28 April 2006; revised 27 June 2006;

re-revised 04 August 2006; accepted 11

September 2006

doi: 10.1111/j.1399-0039.2006.00692.x

Abstract

Podocalyxin (podxl) is a protein with a peptide bone of �55.5 kDa that undergoes

a post-translational glycosylation, yielding a final molecular mass from �145 to

�200 kDa.This protein is normally found covering the vascular side of the epithelial

glomerular cells, the podocytes, and its presence is essential to maintain a normal

renal function. It has also been reported in other cells and tissues although its

function has not been yet clarified. The carboxy-terminal intracellular domain of

podxl is nearly 100% identical in most species; however, the ectodomain shows

considerable variations although the cysteine residues are conserved. Detection of

this protein is elusive, most likely due to differences in post-translational

modifications. We aimed at producing murine monoclonal antibodies against

human podxl. Immunization with Chinese hamster ovarian -hpodxl-green

fluorescence protein live cells yielded five different monoclonal antibodies that

were characterized by enzyme-linked immunosorbent assay, sodium dodecyl

sulfate–polyacrylamide gel electrophoresis/western blot, flow cytometry, immu-

nohistochemistry, and immunoprecipitation. The different behavior of these

antibodies suggests that some of them may react against epitopes masked by

different glycosylated protein moieties.

Introduction

Michael et al. (1) reported that the luminal side of the

epithelial cells of the renal glomerulii, the podocytes, was

covered with a material named epithelial polyanion due to

its affinity for cationic stains, which could be digested with

neuraminidase. Kerjaschki et al. (2) isolated for the first

time a �140-kDa glomerular sialoprotein that could be

detected by cationic stains like the Alcian Blue, digested by

neuraminidase and capable of binding to wheat germ

agglutinin. These authors concluded that this protein was

identical to the glomerular polyanion identified histochem-

ically and named it podocalyxin (podxl) based on their

exclusive localization in the glycocalyx of the podocytes.

Further work identified the presence of a high content of

sulfate and sialic acid as the responsible for the negative

charge of the protein (3). The glomerular podxl is essential

to maintain the structural organization of the podocytes,

preventing the occlusion of the epithelial slits and urinary

spaces (1, 2–4). In human pathological states or in

experimental nephrosis, distortion of the glomerular

podocytes, obliteration of the epithelial slits, and decreased

glomerular content of sialic acid are found, suggesting the

functional importance of podxl (5, 6). Finally, podxl null

mice showed profound defects on kidney development and

die within hours of birth with anuric renal failure (7).

In addition to its location in the renal glomerulus, podxl

can be found covering non-uniformly the luminal side of the

vascular endothelium of many tissues (8). More recently, its

presence has also been reported in cells of hematopoietic

lineage from chicken and rat (9, 10).

Antibodies against podxl are uncommon, and their

specificity for the protein in various tissues varies. However,

an antigenic variability is expected considering the heavy

glycosylation of this protein. The aim of this study is to

produce specific murine monoclonal antibodies directed

against human podxl and to test their ability to detect the

protein in a variety of cell lines. Five monoclonal antibodies

with distinct properties have been obtained.

ª 2006 The AuthorsJournal compilation 68 (407–417) ª 2006 Blackwell Munksgaard 407

Materials and methods

Animals and reagents

Balb/c mice, 6–8 weeks old, of either sex, were originally

purchased from Charles River (Barcelona, Spain) and

maintained in our institution. Electrophoretic reagents

were from Bio-Rad Laboratories, S.A., (Madrid, Spain),

and most other reagents were from Sigma (Madrid, Spain).

Antibodies and cell lines

Mouse monoclonal antibody 3D3 against human podxl was

a gift from Dr D. Kershaw (University of Michigan, USA).

Polyclonal rabbit anti-green fluorescence protein (GFP), and

anti-Wilm’s Tumor (WT)-1 antibodies were from Santa Cruz

Biotechnologies (SantaCruz,CA).Fluorescein isothiocyanate

(FITC)-labeled (Fab#)2 anti-mouse immunoglobulin G (IgG)

antibody was from DAKO A/S (Denmark), and Alexa 546

fluor-labeled anti-rabbit IgG was from Molecular Probes

Europe BV (Leiden, The Netherlands). Hybridoma 187.1

producing an IgG2c directed against the k light chain of

amouse Igwasagift fromDrJ.M.Rojo (CIB,Madrid,Spain).

Rabbit antisera against CD31 (11) was a gift from Dr C.

Bernabeu (CIB).

Chinese hamster ovarian (CHO) cells, human kidney

carcinoma cells HEK293, human hematopoietic K562 and

Meg01 cells lines and NS-1 myeloma cells were obtained

from the American Type Culture Collection repository.

Tera-1 cells (12), isolated from a human testicular carci-

noma, were from Dr W.M. Schopperle (University of

Michigan, Ann Arbor, MI).

Molecular cloning and bacterial expression of the

human and mouse podxl ectodomains

A DNA fragment comprising the entire putative ectodo-

main (341–1537 bp) of human podxl (r-hpodxlD) (Gene-

Bank NM-005397) was amplified by polymerase chain

reaction (PCR) using the following primers: 5#-GTCGCC

GAATTCGTCGCCCTCCCAGAAT-3# (sense) and 5#-GATGAGAAGCTTGCTGAAGCGGTCCTCG-3# (anti-sense), containing EcoRI (sense primer, underlined) and

HindIII (antisense primer, underlined) restriction sites to

allow in-frame subcloning in the same sites of the bacterial

expression plasmid pET-24b(1) (Novagen, Nottingham,

UK). The resulting plasmid was transformed into competent

DH5a cells and sequenced. The plasmid was then trans-

formed into competent BL21 cells. When expressed in BL21

cells, this construction yields a recombinant podxlD fragment

of 429 amino acids (aa), with a calculated molecular mass of

44.62 kDa. It contains 16 aa at the amino end and 13 aa

including a 6xHis-tag at the carboxy end from the expression

vector. Putative signal peptide and transmembrane domain,

as determined from the hydrophobicity profile (Kyte–

Doolittle method), are not included in the fragment.

To determine the specificity of antibodies against hpodxl,

we also have generated the whole putative ectodomain of

mouse podxl (r-mpodxlD). A DNA fragment (216–1385 bp,

GeneBank AB028048) was amplified by PCR using the

following primers: 5#-CTGCTGCTGAATTCGTCGCC

TGCA-3# (sense) and 5#-ATGATGAGCTCGAGGCTG

AAGCGG-3# (antisense), containing EcoRI (sense primer,

underlined) andXhoI (antisense primer, underlined) restric-

tion sites. The same cloning procedure and induction

protocol as for r-hpodxlD were then followed. This

construction codes for an r-mpodxlD fragment of 414 aa

with a calculated molecular mass of 43.4 kDa. It contains

16 aa at the amino end and 7 aa at the carboxy end includ-

ing the 6x His-tag, coded from the expression vector. The

core fragment of r-mpodxlD goes from serine-16 to leucine-

405 of the translated protein. Putative signal peptide and

transmembrane domain were determined by the hydropho-

bicity profile of the peptide (Kyte–Doolittle method).

Expression of r-h/mpodxlD was induced by adding

isopropyl b D-thiogalactopyranoside (IPTG) to a final con-

centration of 1 mM to exponentially growing (A600 ¼ 0.4–

0.5) 10-ml cultures of BL21 cells, using the empty vector as

a control and incubating at 37�C under constant shaking.

The protein expression profile was analyzed by solubilizing

samples taken at time intervals into sodiumdodecyl sulfate–

polyacrylamide gel electrophoresis (SDS-PAGE) sample

buffer under reducing conditions, run on a 10% T (5% C)

polyacrylamide gel and stained using Coomassie Blue.

Insoluble material was solubilized in urea–Tris buffer and

treated similarly. Volumes loaded in the gel were equivalent

to the same number of cells in all cases. Expressed proteins

were found only in the soluble fraction.

For preparative purposes, cultureswere scaled up to 500ml

and induced with IPTG for 4 h. Portions of 50-ml culture

were then centrifuged and washed in phosphate-buffered

saline (PBS), and the bacterial pellets were kept at 280�C.For processing, one portionwas thawed in 100ml ofHis-tag

binding buffer (0.5 M NaCl, 20 mM Tris–HCl, pH 8.0),

1 mMphenylmethylsulfonyl fluoride (PMSF) and 0.2 mg/ml

lysozyme and incubated for 20 min at room temperature

with shaking. The suspension was then sonicated on ice (10

bursts of 10 s at the maximum power output) and

centrifuged for 40 min at 36,600 g in an SS34 rotor at 4�C,and the supernatant was filtered through a 0.22-mm filter

and saved. The recombinant human or mouse podxlD was

purified from the bacterial lysate by immobilized metal-

chelating affinity chromatography. Lysate supernatant

was passed through a column with 5 ml of Chelating

Sepharose FF gel (Pharmacia Biotech, Uppsala, Sweden),

previously loaded with SO4Ni as recommended by the

manufacturer and equilibrated in His-tag binding buffer.

The column was washed with His-tag binding buffer and

washed again with 30 mM imidazole in His-tag binding

buffer with reversed flow, and the bound proteins were

Anti-podocalyxin antibodies R. B. Rodrıguez et al.

408ª 2006 The Authors

Journal compilation 68 (407–417) ª 2006 Blackwell Munksgaard

eluted in 2 ml fractions with 0.5 M imidazole in His-tag

binding buffer. Fractions were analyzed by SDS-PAGE on

a 10%(T) 5%(C) polyacrylamide gel and Coomassie Blue

staining. Fractions containing the protein band of interest

were pooled, and the proteins precipitated overnight at 4�Cin (NH4)2SO4 at 50% of saturation. The precipitate was

spun at 36,600 g for 20 min at 4�C, then solubilized in 5 ml

of PBS, dialyzed against PBS, and kept frozen at 280�C.

Production of mammalian cells expressing

recombinant podxl

Human or mouse podxl complementary DNA (cDNA)

were obtained by PCR from genomic DNA as previously

described(13). The expression plasmid (pCDNA3) carrying

either humanormouse podxl full-length cDNA3# linked in-frame to GFP (h/mpodxl–GFP) was stably transfected into

CHO cells by the calcium phosphate precipitation pro-

cedure. The transfected cells were selected by their resistance

to G418 (Gibco-BRL, Madrid, Spain) and further on by

cell sorting.

Production of murine monoclonal antibodies

Three 2-month–old, female Balb/c mice were repeatedly

immunized (four intraperitoneal injections at the intervals

of 3, 2, and 1 weeks) with 1.0 � 106 to 1.5 � 106 CHO–

hpodxl–GFP live cells resuspended in 0.5 ml of sterile PBS.

The progress of immunization was monitored by testing the

serum against r-hpodxlD by enzyme-linked immunosorbent

assay (ELISA) and western blot. A fifth injection was given

3–4 days before the fusion. Hybridomas were produced by

the fusion of splenocytes and Sp2/O-Ag14 mouse myeloma

cells. Fused cells were plated onto five 96-well plates and

incubated with Dulbecco�s modified Eagle�s minimal

essential medium (DMEM; Gibco) supplemented with

penicillin, streptomycin, 20% fetal calf serum (FCS), 1�hypoxanthine/azaserine (Sigma, Madrid, Spain) and 1�OPI media supplement (Sigma). Fresh medium was added

when needed. The antibody production of hybridomas was

checked by indirect ELISA of the supernatants after 10–14

daysof culture. Positive cloneswere further grown inDMEM

supplemented with 10% FCS and penicillin/streptomycin

and tested by western blot against r-hpodxlD and CHO–

hpodxl–GFPandTera-1 cell extracts. Five clones positive in

at least one of these assays were frozen in liquid nitrogen,

cloned by limiting dilution and frozen again.

Indirect ELISA of hybridoma supernatants

One hundred micrograms of r-hpodxlD was immobilized

on microtiter 96-well ELISA plates (Nunc, Wiesbaden,

Germany) in Na2CO3 buffer 4 h at room temperature or

overnight at 4�C. Wells were washed with PBS, blocked for

2 h at room temperature with 3% non-fat powdered milk in

PBS, and kept frozen at 280�C in the same buffer until

needed. For the assay, plates were washedwith PBS and 100

ml of hybridoma supernatant or the appropriate controls

were added and incubated for 2 h. Wells were washed and

incubated for at least 1 h with 100 ml of a goat anti-mouse

IgG (heavy and light chains (H 1 L)) conjugated with

peroxidase (Bio-Rad Laboratories S.A) diluted 1/3000 in

1% non-fat powdered milk in PBS. The detection step was

performed using O-Phenylenediamine dihydrochloride as

a substrate, and the color read at 495 nm.

SDS-PAGE and western blot

Cells were routinely lysed in non-denaturing cell lysis buffer

(1%TritonX-100, 0.05%Tween-20, 0.3MNaCl and 1mM

PMSF in PBS) and a cocktail of protease inhibitors

(Complete Mini, EDTA free; Roche, Madrid, Spain) at a

concentration of 20� 106 or 40� 106 cells/ml. Samples were

made 1� in Laemmli SDS-PAGE loading buffer, with or

without a reducing agent, and run on a 7.5%T, 3%C

polyacrylamide gel. The separated proteins were transferred

to a 0.2-mmnitrocellulosemembrane, blockedwithTBS-3%

fat-free powderedmilk, and incubated with antibody (0.5 or

1.0 mg/ml) in 0.05% Tween-20, 1% fat-free powdered milk

inTBS, or culture supernatant diluted 1/5 to 1/10 in the same

buffer, or the appropriate controls, for 2 h at room tem-

perature or for overnight at 4�C. Bound antibodies were de-

tectedwith aperoxidase-labeled goatanti-mouse IgG (H1L)

antibody, and peroxidase activity was visualized using

luminol as a substrate. Multiple blotting was performed

using a multiscreen apparatus (Bio-Rad Laboratories S.A).

Immunoprecipitation

Protein G-sepharose beads (Pharmacia) as needed for each

experiment were blocked for 1–2 h at room temperature

under rotation with 1 ml of hybridoma culture medium.

Beads were aliquoted in 15-ml portions and incubated for

4–6 h at room temperature with 1 ml of hybridoma

supernatant or the appropriate controls. When an immu-

noglobulin M (IgM) was used, 1 ml of a saturated

supernatant of hybridoma 187.1 was also added. Beads

were washed once with hybridoma medium and twice with

cell lysis buffer.Onemilliliter of cell lysate at a concentration

of 1.0 � 106 to 1.5 � 106 cells/ml or cell lysis buffer was

added to the beads and incubated overnight at 4�C under

rotation. After washing three times with cell lysis buffer, the

immunoprecipitated proteins were eluted from the beads

with reducing SDS-PAGE sample loading buffer and

analyzed by western blot as described above.

Fluorescent immunostaining

For fluorescent immunostaining, hybridoma culture

medium diluted 1/2 in PBS was used as blocking buffer

ª 2006 The AuthorsJournal compilation 68 (407–417) ª 2006 Blackwell Munksgaard 409

R. B. Rodrıguez et al. Anti-podocalyxin antibodies

and diluting medium for purified control and antibodies

labeled with FITC and Alexa 546. Tissue was fixed with

freshly made 4% paraformaldehyde in PBS at room

temperature. Hybridoma supernatants were diluted 1/2

with PBS before the assay.

Frozen human kidney of 10-mm sections kept at 280�Cwere air-dried for 1 h, fixed for 10 min, washed twice with

PBS, and blocked for 1 h. Sections were then incubated 1–2 h

with a 1/2 dilution of hybridoma supernatant with either

anti-WT-1 antibody (2 mg/ml) or anti-CD-31 antisera (1/100

final dilution). Irrelevant mouse IgG or nonimmune rabbit

sera were used as controls at the same concentrations. After

washing with PBS for 2 � 5 min and 2 � 10 min, FITC-

labeled rabbit (Fab#)2 anti-mouse antibody and Alexa 546-

labeled anti-rabbit IgG at a final 1/400 dilution, each was

added and incubated in the dark for 1 h. Sections were

washed, mounted, and kept in the dark until examined

under a Leica confocal microscopy. All procedure was

performed at room temperature.

Flow cytometry

Cells were washed in PBS and resuspended in hybridoma

culture medium at a concentration of 6 � 106 to 10 � 106

cells/ml. Fifty microliters of cell suspension was incubated

with 100 ml of hybridoma supernatant or 5 mg of control

antibody in hybridoma culture medium for 15 min at room

temperature. Cells were spun for 3 min at 800 � g and

resuspended in 100 ml of hybridoma medium. Twenty

microliters of a 1/10 dilution of the same FITC-labeled

antibody used for immunostaining was added, and the cell

suspension incubated in the dark for 20 min at room

temperature. Cells were spun off as above, resuspended in

0.3 ml of freshly made 1% paraformaldehyde in PBS, and

kept in the dark at room temperature for immediate analysis

or overnight at 4�C when analyzed on the next day.

Fluorescence was analyzed by flow cytometry.

Results

Expression of r-hpodxlD and r-mpodxlD in

bacterial cells

IPTG-induced BL21 cells harboring the construct for

r-hpodxlD overexpressed a protein that could be retained

and purified in a Ni21 affinity column. This protein showed

an apparent molecular mass of about 51 kDa, as determined

bySDS-PAGE,which is aboveof the theoretical value of 44.6

kDa predicted from the aa sequence of the protein. However,

by matrix-assisted laser desorption ionization time-of-flight

(MALDI-TOF) analysis of the purified protein, protein

of molecular mass of 44,436.9 Da was revealed. This protein

is recognized by anti-podxl antibody 3D3 by western blot

(Figure 1). Similarly, r-mpodxlD was also obtained and

purified from IPTG-induced BL21 cells. With a theoretical

molecular mass of 43.4 kDa, the r-mpodxlD migrates on

a SDS-PAGE gel as a protein of molecular mass of 53 kDa.

However, byMALDI-TOF analysis, a protein of molecular

mass of 43,241.3 Da was revealed. Thus, we assume that

this protein is r-mpodxlD.

Expression of hpodxl–GFP ormpodxl–GFP in CHO cells

CHO cells were stably transfected with expression plasmids

encoding GFP, hpodxl–GFP or mpodxl–GFP. In extracts

from CHO–GFP cells, only a band corresponding to GFP

was detected by SDS-PAGE and western blot using anti-

GFP antibody (data not shown). CHO–hpodxl–GFP cells

Figure 1 Characterization of the recombinant human podocalyxin

(hpodxl) ectodomain and podxl–green fluorescence protein (GFP) fusion

protein by sodium dodecyl sulfate (SDS)–polyacrylamide gel electropho-

resis/western blot. Proteins (1.5 mg) or Chinese hamster ovarian–hpodxl–

GFP cell extracts (10 ml of a 20 � 106 cells/ml lysate) in 200 ml of loading

buffer were loaded on a single well containing 7.5%(T), 3%(C) SDS-

polyacrylamide minigel under reducing conditions, electrophoresed, and

transferred onto a nitrocellulose membrane. After blocking, membranes

were sequentially incubated with the indicated antibodies using a multi-

screen apparatus and with a secondary horseradish peroxidase-labeled

anti-mouse (heavy and light chains (H 1 L)) antibody, and peroxidase

activity was visualized using luminol as substrate. mpodxl, mouse podxl.

410ª 2006 The Authors

Journal compilation 68 (407–417) ª 2006 Blackwell Munksgaard

Anti-podocalyxin antibodies R. B. Rodrıguez et al.

expressed amajor protein of�190 kDa that is recognized by

both 3D3 and anti-GFP antibodies (Figure 1). The reaction

of the 190-kDa band with the 3D3 antibody confirmed that

it was indeed hpodxl. CHO–mpodxl–GFP cells expressed a

protein with a slightly lower size than its human counter-

part, detectable with anti-GFP antibody (data not shown).

Characterization of the antibodies

We have isolated five different mouse hybridomas pro-

ducing monoclonal antibodies against the podxl cluster:

1401F9.15,B24B3.8,B31A2.7,B32E3, andB34D1.3.Here,we

have determined their class and characterized their immuno-

reactivity by ELISA, SDS-PAGE/western blot, immuno-

precipitation, immunohistochemistry, and flow cytometry.

Class determination of antibodies and characterization

by ELISA

Antibody B31A2.7 is an IgM, as indicated by the size of its

heavy chain after protein G co-immunoprecipitation with

antibody 187.1 and reducing SDS-PAGE and Coomassie

Blue staining.All other four antibodies are IgGs, as deduced

by protein G immunoprecipitation and by determination of

the size of their heavy chain in a similar way. All antibodies

reacted in ELISA with r-hpodxlD (data not shown).

SDS-PAGE/western blots of r-h/mpodxlD and

cells extracts

All results are shown in Figure 2.

Binding to r-h/mpodxlD

All antibodies reacted with the reduced form of r-hpodxlD.Only B24B3.8 antibody failed to recognize the non-reduced

r-hpodxlD.Antibody B31A2.7 detects both the reduced and

the non-reduced forms of r-mpodxlD.

Binding to CHO–h/mpodxl–GFP cells

Antibodies 1401F9.15, B32E3, B31A2.7, and B34D1.3

detected a major protein band of 180–190 kDa in non-

reduced CHO–hpodxl–GFP cell extract; minor bands in the

range 150–160 kDa were also detected. In reduced extracts,

protein bands in the same size rangewere also observedwith

these antibodies, but B32E3 only detected a weak protein

band of about 110 kDa. Antibody B24B3.8 failed to

recognize hpodxl–GFP. No protein bands were detected

in either CHO cells or CHO cells expressing mpodxl–GFP.

Binding to Meg01 cells

Monoclonal antibody B34D1.3 binds to protein band of

approximately 160 kDa from both non-reduced and

reduced Meg01 cells extract, which is also recognized by

the anti-podxl antibody 3D3 (data not shown).Monoclonal

antibody B32E3 detects this band only in reduced extract

and with less intensity. Antibody B31A2.7 detects a weak

band of 62 kDa in the reduced extract. All other antibodies

did not react.

Binding to HEK293 cells

All monoclonal antibodies bind to non-reduced HEK293

cells extract. Antibodies 1401F9.15, B24B3.8, B31A2.7, and

B32E3 detect a group of protein bands ranging from 100 to

more than 300 kDa, with the two most intense bands of 280

and 180 kDa common to all of them. Monoclonal antibody

B34D1.3 detects a single 200-kDa band that is also

recognized with less intensity by the other antibodies. In

reduced HEK293 cells extract, only antibodies B32E3 and

B34D1.3 detect protein bands within the same size range as

in non-reduced extract, 155 kDa for B32E3, which is also

detected in the non-reduced extract, and 180 kDa for

B34D1.3. Monoclonal antibodies 1401F9.15 and B31A2.7

detect a single band of 51 kDa.

Binding to K562 cells

In reducedK562 cells extract, antibodyB32E3detects a 155-

kDa band and monoclonal antibody B31A2.7 detects two

protein bands of 60 and 62 kDa.Nobandsweremade visible

in reduced or non-reduced extracts with all the other

monoclonal antibodies.

Binding to Tera-1 cells

Monoclonal antibodies 1401F9.15 and B34D1.3 bind to

protein bands in the range 200–250 kDa in non-reduced

Tera-1 cells extract, but onlymonoclonal antibodyB34D1.3

also detects these bands under reducing conditions. These

protein bands are also recognized by the anti-podxl

antibody 3D3 (see total lysate lane in Figure 3, panel A).

No other antibodies were reactive.

Immunoprecipitation of podxl from Tera-1 and

CHO–hpodxl–GFP cells

Figure 3 (panel A) shows the results obtained by the

immunoprecipitation of podxl from Tera-1 cells extract

with our set of antibodies blotting with the anti-podxl

monoclonal antibody 3D3. Monoclonal antibodies

1401F9.15 and B24B3.8 immunoprecipitated a 300-kDa

protein band that can also be seen in the total cell lysate

loaded into the gel. However, the 223-kDa protein present

in the total cell lysate is not immunoprecipitated by the

antibodies. A 100-kDaprotein bandpresent in the cell lysate

is not detected by immunoprecipitation with monoclonal

antibodies B32E3 and B34D1.3. Antibody B31A2.7 did not

immunoprecipitate any protein from the cell lysate.

We also immunoprecipitated hpodxl–GFP from CHO–

hpodxl–GFP cells extracts. The results are shown in

ª 2006 The AuthorsJournal compilation 68 (407–417) ª 2006 Blackwell Munksgaard 411

R. B. Rodrıguez et al. Anti-podocalyxin antibodies

412ª 2006 The Authors

Journal compilation 68 (407–417) ª 2006 Blackwell Munksgaard

Anti-podocalyxin antibodies R. B. Rodrıguez et al.

Figure 3 (panel B). Monoclonal antibodies 1401F9.15,

B34D1.3, and B24B3.8 immunoprecipitated protein bands

of 190, 130, 119, and 107 kDa that are detected in the total

cell lysate. TheB32E3 immunoprecipitated a single 119-kDa

protein that is hardly detected. Antibody B31A2.7 did not

immunoprecipitate any protein from the cell lysate.

Immunohistochemistry of human kidney

Weassayed the ability of our antibodies to recognize podxl in

human kidney cortex sections. Figure 4 (upper panel) shows

the green fluorescent staining produced by the binding of our

antibodies: monoclonal antibodies 1401F9.15, B24B3.8, and

B34D1.3 bound to glomeruli and vascular endothelium,

while B31A2.7 and B32E3 were negative.

Co-staining of glomerulus and blood vessel with a repre-

sentative positive anti-podxl antibody, 1401F9.15, and

antibody anti-WT-1 or anti-CD31 is shown in Figure 4

(mid-and lowerpanels, respectively).Overlappingof thegreen

and red signals of anti-podxl and anti-WT-1 renders a yellow

pattern in the glomerulus, indicating the co-localization of

both antigens. However, the vascular endothelium was only

stained of green by the podxl since the expression of WT-1 is

restricted to renal epithelial cells. Similarly, co-staining with

the same anti-podxl antibody and anti-CD31 (a marker for

endothelial cells) reveals apartial overlappingofbothantigens

in the glomerulus and vascular vessels.

Flow cytometry

Antibodies did not bind to CHO and K562 cells (data not

shown). As shown in Figure 5, monoclonal antibodies

1401F9.15, B24B3.8, and B34D1.3 bound to HEK293,

Meg01, and Tera-1 cells; B31A2.7 bound to HEK293 and

Tera-1 cells; and monoclonal antibody B32E3 did not bind

to any of these cell types. All positive antibodies produce

higher mean fluorescence values in HEK293 and Tera-1

cells than in Meg01 cells, suggesting a higher concentration

of surface antigen in those cell lines.

Table 1 summarizes the binding characteristics of our

antibodies to fully glycosylated podxl in the human cell lines

tested, determined by flow cytometry and western blot.

K562 cells are not included since in this cell line fully

glycosylated podxl is undetectable by these methods.

Discussion

The recombinant h/mpodxlD and the full-length h/mpodxl–

GFP proteins, produced either in bacteria or in mammalian

cells, were indeed podxl, as indicated by their molecular

mass and/or by their reactivity to 3D3 or anti-GFP

antibody, or both.

We have generated a set of monoclonal antibodies di-

rected against human podxl. We initially attempted to use

the ectodomain of hpodxl expressed in bacteria as im-

munogen; however, after multiple trials, we obtained very

poor producers of IgM-secreting hybridomas, whose

activity was lost after a short number of passes. Instead,

Figure 2 Characterization of anti-human-podocalyxin (podxl) antibodies by sodium dodecyl sulfate (SDS)–polyacrylamide gel electrophoresis/western

blot. Proteins or cells extracts were loaded on a 7.5%(T), 3%(C) SDS-polyacrylamide gel under non-reducing (left panels) or reducing (right panels)

conditions, electrophoresed, and transferred onto a nitrocellulose membrane. After blocking, membranes were sequentially incubated with the indicated

antibodiesandwith a secondaryhorseradish peroxidase-labeled anti-mouse (H1 L) antibody, andperoxidaseactivity wasvisualized using luminol as substrate.

Amounts loaded were 150 ng (r-hpodxlD and r-mpodxlD), 5 mg [Chinese hamster ovarian (CHO)–hpodxl–green fluorescence protein (GFP) and CHO–

mpodxl–GFP cells], or 50mg (CHO, Meg01, HEK293, K562, and Tera-1 cells). hpodxl, human podxl; mpodxl, mouse podxl. Arrowheads indicate faint bands.

Figure 3 Immunoprecipitation of podocalyxin (podxl) from Chinese

hamster ovarian (CHO)–hpodxl–green fluorescence protein (GFP) and

Tera-1 cells. Fifteen microliters of previously blocked protein G-sepharose

beads was incubated with 1 ml of saturated hybridoma culture

supernatant or the appropriate controls and 1 ml of hybridoma 187.1

saturated culture [monoclonal antibody (mAb) B31A2.7] or 1 ml of culture

medium (all other tubes) and washed as described in Materials and

methods. One milliliter of cell lysate (1.0� 106 to 1.5� 106 cells/ml) or cell

lysis buffer was added to the beads, incubated, and washed as described

in Materials and methods. Bound proteins were eluted in sodium dodecyl

sulfate–polyacrylamide gel electrophoresis reducing sample buffer,

electrophoresed, and transferred onto nitrocellulose membrane and

blotted with anti-podxl antibody 3D3 (Tera-1 cells, panel A) or anti-GFP

antibody [Chinese hamster ovarian (CHO)–hpodxl–GFP cells, panel B] as

described in Materials and methods. Control lanes were as follow: control

1, B31A2.7 alone; control 2, 187.1 alone; control 3, B31A2.7 plus 187.1 (no

cell lysate added); control 4, irrelevant mouse immunoglobulin G (5 mg);

cell lysate, 40ml of a 10� 106 cells/ml Tera-1 (panel A) or CHO–hpodxl–GFP

(panel B) cell lysate directly loaded onto the gel. hpodxl, human podxl;

mpodxl, mouse podxl. Arrowheads indicate faint bands.

ª 2006 The AuthorsJournal compilation 68 (407–417) ª 2006 Blackwell Munksgaard 413

R. B. Rodrıguez et al. Anti-podocalyxin antibodies

the immunization ofmice with live CHO–hpodxl–GFP cells

yielded five stable monoclonal antibodies: four were IgGs

and one was IgM.

The specificity of antibodies was appraised by the

immunostaining patterns of kidney glomeruli and vascular

endothelia. Moreover, all antibodies react with the bacterial-

produced recombinant ectodomain of human podxl in

ELISA and western blot, and all but B24B3.8 detect full-

length hpodxl expressed in CHO cells. Besides, monoclonal

antibodyB31A2.7 also reacts with the ectodomain ofmouse

podxl in western blot. Despite the overall differences in the

primary structure between human and mouse podxl

ectodomains, some areas of identity are found on alignment

of both the sequences (Figure 6). Thus, most likely the

epitope recognized by monoclonal antibody B31A2.7 must

be within one of these fragments, located within the

carboxy-terminal end of the extracellular domain of the

protein, a region that is poorly glycosylated. Only one of

Figure 4 Immunostaining of human kidney cor-

tex sections. Fixed 10-mm sections from frozen

human kidney specimens were blocked and

incubated as described in Materials and methods

with supernatants of the indicated hybridomas

with either anti-Wilm’s tumor (anti-WT-1) antibody

or anti-CD-31 antisera. Irrelevant mouse immu-

noglobulin G (IgG) or nonimmune rabbit sera

were used as controls at the same concentra-

tions. After washing, fluorescein isothiocyanate-

labeled rabbit (Fab#)2 anti-mouse antibody and

Alexa 546-labeled anti-rabbit IgG were added and

incubated. Sections were washed, mounted,

and examined under a Leica confocal micros-

copy. Upper panels show the green fluorescence

due to the binding of positive anti-podocalyxin

antibodies to glomeruli and blood vessels.

Middle block of panels show the fluorescence

caused by co-staining of a glomerulus (upper

row) and blood vessels (lower row) with anti-

bodies 1401F9.15 and anti-WT-1. Similarly, the

lower block of panels shows the co-staining with

antibodies 1401F9.15 and anti-CD31. Left panels

show the green fluorescence due to the binding

of 1401F9.15, center panels show the red

fluorescence caused by the binding of the co-

staining antibody, and right panels show the

superimposition of green and red fluorescences.

Bar ¼ 150 mm (small panels) or 30 mm (large

panels).

414ª 2006 The Authors

Journal compilation 68 (407–417) ª 2006 Blackwell Munksgaard

Anti-podocalyxin antibodies R. B. Rodrıguez et al.

these putative epitopes contains an asparagine residue that

is susceptible of glycosylation.

Glycosylated and partially glycosylated forms of podxl

co-exist within the cells (14, 15).Moreover, variability in the

glycosylation patterns of the native podxl may be inferred

from its different electrophoretic mobility in extracts from

the different cell sources. Some of our antibodies detect low-

molecular-weight forms of hpodxl that could be considered

unglycosylated or poorly glycosylated, like the 50-kDa

protein detected in reduced HEK293 cells extract by

monoclonal antibodies 1401F9.15 and B31A2.7, the 60-

and 62-kDa proteins detected in reduced K562 cells extract

with antibody B31A2.7, and proteins of less than 150 kDa

detected in non-reduced HEK293 cells extract or immuno-

precipitated fromTera-1 cells extract. In all cases, these low-

molecular-weight forms are quantitatively much less

important than their high-molecular-weight counterparts,

indicating that their steady-state levels should be low and

most likely lack of physiological relevance. In the high-

molecular-weight range of podxl forms detected with our

antibodies, one or two predominate, thus probably indicat-

ing that these are the mature, functionally active proteins.

However, in Meg01 cells, podxl is expressed as a single 155-

to 160-kDa protein, and its expression levels are reduced

compared with those of podxl in other cell types. This

finding may indicate either that a diminished podxl gene

expression in Meg01 cells or else that a peculiar conforma-

tion or glycosylation of podxlmight render it less responsive

to our antibodies. In K562 cells, antibody B31A2.7 only

detects low-molecular-weight forms of podxl.

Antibodies B32E3 and B34D1.3 immunoprecipitated

a 100-kDa form of podxl in Tera-1 cells but, similar to

3D3 and the other antibodies, failed to detect this band by

western blot. This indicates that the steady-state levels of

this form must be very low, and only the concentration

effect of immunoprecipitation allows its detection with

some but not all antibodies.

All our monoclonal antibodies seem to detect distinct

epitopes since their recognition pattern on different cell

types and conditions of the antigen differed (Table 1). These

patterns may reflect different extents of glycosylation and/

or conformation in the final processing of the mature

protein in each type of cell lines. In Meg01 and HEK293

cells, the epitope for monoclonal antibody B32E3 is

accessible in denatured podxl but not in the native form,

suggesting a conformational specificity. However, it could

be that interactions of podxl with other protein(s) could

mask this epitope in the native protein. In podocytes, an

interaction of podxl with b1 integrins has been suggested

(16). Most likely, the interactions of podxl should occur on

the extracellular domain(s) since ourmonoclonal antibodies

Figure 5 Flow cytometry analysis. Histograms of cell fluorescence

intensity of three different cell lines analyzed with our monoclonal

antibodies. Fifty microliters of a 6 � 106 to 10 � 106 cells/ml cell

suspension was incubated with 100 ml of hybridoma supernatant or 5 mg

of control antibody in 100 ml of hybridoma culture medium. Cells were

then treated with fluorescein isothiocyanate -labeled anti-mouse anti-

body. Fluorescence was determined by flow cytometry. Chinese hamster

ovarian cells used as negative control cell line and K562 cells were both

negative and are omitted in the figure for simplicity.

Table 1 Binding characteristics of antibodies to podocalyxin in human cell lines Meg01, HEK293, and Tera-1, determined by flow cytometry and

western blota

Podocalyxin

Native Denatured non-reduced Denatured reduced

1401F9.15 1 Meg01, 2; HEK293, 1; Tera-1, 1 2

B24B3.8 1 Meg01, 2; HEK293, 1; Tera-1, 2 2

B31A2.7 Meg01, 2; HEK293, 1; Tera-1, 1 Meg01, 2; HEK293, 1; Tera-1, 2 2

B32E3 2 Meg01, 2; HEK293, 1; Tera-1, 2 Meg01, 1; HEK293, 2; Tera-1, 2

B34D1.3 1 1 1

a �1� and �2� denote bound and unbound antibodies, respectively, in all or a particular cell line.

ª 2006 The AuthorsJournal compilation 68 (407–417) ª 2006 Blackwell Munksgaard 415

R. B. Rodrıguez et al. Anti-podocalyxin antibodies

have all been screened with the ectodomain of podxl.

However, it is worth to note that intracellular interactions

(17) could lead through allosteric changes to variations

in the conformation of the extracellular domain(s) with

the result of a masking of epitopes. Alternate disulfide

bridge formation could also lead to different protein

conformations.

Podxl is a sulfate- and sialic acid-rich protein located

primarily on the apical face of the urinary spaces,

hematopoietic tissues, and some tumor cells (17, 18). The

specific roles of this molecule in different cells under normal

physiological condition remain unknown. Recent work

suggests that podxl might have both adhesive and anti-

adhesive properties (15, 19). The mucin-like extracellular

domain of podxl undergoes extensive glycosylation, raising

the molecular mass of the protein from 55.5 to �165 kDa.

Monoclonal antibodies to other sialoproteins, like CD43,

specifically recognize the glycosylated portion of the

molecule (20–22). As aforementioned, this may not be our

case since all our antibodies recognize the unglycosylated

recombinant molecule. The different molecular mass of

podxl in several cell types indicates the extent and variability

of glycosylation patterns (23) that may be related, at least

in part, to functional differences. Different patterns of

glycosylation may also alter the contact of antibodies with

their epitopes. Our monoclonal antibodies offer the

possibility to indirectly detect the presence of diverse

carbohydrate moieties of podxl for the study of cell

differentiation and function.

To conclude, we present a set of anti-human podxl

monoclonal antibodies detecting the protein in a more

variable range of cell types and physical forms than other

known antibodies. Each of these monoclonal antibodies

binds to the forms of the hpodxl ectodomain that, as

a whole, represent almost every possible state in the

processing of the protein. The different in vivomass patterns

shown by podxl suggest that glycosylation may differ from

cell to cell and perhaps on different developmental and/or

functional stages of the cell. These antibodies may help

elucidate the qualitative and quantitative importance of

glycosylation of podxl in cell function.

Acknowledgments

Technical assistance given by T. Fontela is greatly

appreciated. We wish to thank Dr Rojo, Dr Kershaw, and

Dr Bernabeu for the generous gift of clone 187.1, mono-

clonal antibody 3D3, and anti-CD31, respectively. This

work has been supported in part by grants from the

Direccion General de Investigacion (SAF 2004-04345),

Fondo de Investigaciones Sanitarias (FIS-PI021263 and

PI050546), and Comunidad de Madrid (08.4/0029.1/2003).

References

1. Michael AF, Blau E, Vernier RL. Glomerular polyanion:

alteration in aminonucleoside nephrosis. Lab Invest 1970: 23:

649–57.

2. Kerjaschki D, Sharkey DJ, Farquhar MG. Identification and

characterization of podocalyxin, the major sialoglycoprotein

of the renal glomerular epithelial cells. J Cell Biol 1984: 98:

1630–36.

3. Dekan G, Farquhar MG. Sulfate contributes to the negative

charge of podocalyxin, the major sialoglycoprotein of the

glomerular filtration slits. Proc Natl Acad Sci U S A 1991: 88:

5398–402.

4. Charest PM, Roth J. Localization of sialic acid in kidney

glomeruli: regionalization in the podocyte plasma membrane

and loss in experimental nephrosis. Proc Natl Acad Sci U S A

1985: 82: 8509–12.

5. Caulfield JP, Reid JJ, Farquhar MG. Alteration of the

glomerular epithelium in acute aminonucleoside nephrosis:

evidence for formation of occluding junctions and epithelial

cell detachment. Lab Invest 1976: 34: 43–59.

6. Seiler MW, Rennke HG, Venkatachalem MA, Cotran RS.

Pathogenesis of polycation-induced alteration (‘‘fusion’’) of

glomerular epithelium. Lab Invest 1977: 36: 48–61.

7. Doyonnas R, Kershaw DB, Duhme C et al. Anuria,

omphalocele, and perinatal lethality in mice lacking the

Figure 6 Alignment of mouse and human carboxy-terminal ends of podocalyxin (mpodxl and hpdoxl, respectively) ectodomain. CLUSTALX 1.83 was used

to perform the sequence alignment. �*� indicates a fully conserved residue, �:� indicates a conserved residue belonging to one of the �strong� groups, and �.�

indicates a conserved residue belonging to one of the �weaker� groups.

416ª 2006 The Authors

Journal compilation 68 (407–417) ª 2006 Blackwell Munksgaard

Anti-podocalyxin antibodies R. B. Rodrıguez et al.

CD34-related protein podocalyxin. J Exp Med 2001: 194:

13–27.

8. Horvat R, Hovorka A, Dekan G et al. Endothelial cell

membranes contain podocalyxin – the major sialoprotein of

visceral glomerular epithelial cells. J Cell Biol 1986: 102:

484–91.

9. McMagny KE, Pettersson I, Rossi F et al. Thrombomucin,

a novel cell surface protein that defines thrombocytes and

multipotent hematopoietic progenitors. J Cell Biol 1997:138:

1395–407.

10. Miettinen A, Solin ML, Reivinen J et al. Podocalyxin in rat

platelets and megakaryocytes. Am J Pathol 1999: 154: 813–22.

11. Lastres P, Almendro N, Bellon T, Lopez-Guerrero JA, Eritja

R, Bernabeu C. Functional regulation of platelet/endothelial

cell adhesion. J Immunol 1994: 153: 4206–18.

12. Schopperle WM, Kershaw DB, DeWolf WC. Human embryo-

nal carcinoma tumor antigen,Gp200/GCTM-2, is podocalyxin.

Biochem Biophys Res Commun 2003: 300: 285–90.

13. Butta N, Larrucea S, Alonso S et al. Molecular cloning and

functional features of the 5#-regulatory region of the human

podocalyxin gene. BMC Mol Biol 2006: 7: 17.

14. Kershaw DB, Thomas PE, Bryan L et al. Molecular cloning,

expression, and characterization of podocalyxin-like protein 1

from rabbit as a transmembrane protein of glomerular

podocytes and vascular endothelium. J Biol Med 1995: 270:

29439–46.

15. Takeda T, Go WY., Orlando RA, Farquhar MG. Expression

of podocalyxin inhibits cell-cell adhesion and modifies

junctional properties inMadin-Darby canine kidney cells.Mol

Biol Cell 2000: 11: 3219–32.

16. Economou CG, Kitsiou PV, Tzinia AK et al. Enhanced

podocalyxin expression alters the structure of podocyte basal

surface. J Cell Sci 2004: 117: 3281–94.

17. Tan PC, Furness SG, Merkens H et al. NHERF-1 is an

hematopoietic ligand for a subset of the CD34 family of stem

cell surface proteins. Stem Cells 2006: 24: 1150–61.

18. Somasiri A, Nielsen JS, Makretsov N. Overexpression of

the anti-adhesin podocalyxin is an independent predictor

of breast cancer progression. Cancer Res 2004: 64:

5068–73.

19. Sassetti C, Tangemann K, Singer MS et al. Identification of

podocalyxin-like protein as a high endothelial venule ligand

for L-selectin: parallels to CD34. J Exp Med 1998: 187:

1965–75.

20. FoxRI, HuenikenM, Fong S et al. A novel cell surface antigen

(T305) found in increased frequency on acute leukemia cells

and in autoimmune disease states. J Immunol 1983: 131:

762–67.

21. Carlow DA, Ardman B, Ziltener HJ. A novel CD8

T cell-restricted CD45RB epitope shared by CD43 is

differentially affected by glycosylation. J Immunol 1999:

163: 1441–8.

22. Tkaczuk J, Al ST, Escargueil-Blanc I et al. The CBF.78

monoclonal antibody to human sialophorin has

distinct properties giving new insights into the CD43

marker and its activation pathway. Tissue Antigens 1999:

54: 1–15.

23. Daniels MA, Hogquist KA, Jameson SC. Sweet �n� sour: theimpact of differential glycosylation on T cell responses. Nat

Immun 2002: 3: 903–10.

ª 2006 The AuthorsJournal compilation 68 (407–417) ª 2006 Blackwell Munksgaard 417

R. B. Rodrıguez et al. Anti-podocalyxin antibodies