Dr Aaro Miettinen, Department of Bacteriology and Immunology, Haartman Institute, PO Box 21, University of Helsinki, 00014 Helsinki, Finland. E-mail: firstname.lastname@example.org
Podocalyxin-like protein (PCLP) is a sialomucin-type membrane protein structurally related to CD34 and endoglycan. It was first described in glomerular podocytes and endothelial cells. In mice, PCLP is present in haemangioblasts, and in both chicken and mice it is a marker of early haematopoietic stem cells and lineage-restricted haematopoietic progenitors. Its expression decreases during differentiation of haematopoietic cells. Of mature blood cells, only chicken and rat thrombocytes express PCLP protein. PCLP expression in human haematopoietic cells has not been studied. Here we demonstrate PCLP mRNA in human CD34+ cells, in lineage committed erythroid, megakaryocyte and myeloid progenitors, in K562 leukaemia cells, and in peripheral blood leucocytes. The mRNA expression level was higher in developing cells than in mature leucocytes. By Northern blotting and cDNA sequencing, the haematopoietic and renal PCLP mRNAs were identical. Of the mobilized CD34+ cells, 28% (mean; range 14–61%) expressed PCLP protein and the majority of PCLP+ cells were CD117+. Almost all of the K562 cells expressed PCLP protein. Surprisingly, PCLP protein was not detected in any mature blood cells. These results suggest that human PCLP may be a valuable marker for a subset of haematopoietic stem cells.
Podocalyxin is a heavily glycosylated transmembrane protein originally described on the apical membrane of rat podocytes and endothelial cells (Kerjaschki et al, 1984). Its homologue, podocalyxin-like protein (PCLP), with similar tissue distribution has been identified in several species. In addition to rat podocalyxin, rabbit (Kershaw et al, 1995), human (Kershaw et al, 1997), chicken (McNagny et al, 1997), and mouse (Hara et al, 1999) PCLPs have been cloned. These proteins have almost identical intracellular and transmembrane domains suggesting a conserved function. The extracellular domains show less sequence homology, but are structurally similar with four conserved cysteines in the membrane proximal part and a serine-, threonine- and proline-rich N-terminal domain typical for mucin-type glycoproteins.
Podocalyxin-like protein and podocalyxin have been shown in the haematopoietic cells of chicken, mice and rats, but its expression in these cells has been systematically studied only in chickens. In the embryo, chicken PCLP (thrombomucin) is present in the multipotent haematopoietic stem cells. In the adult animals, early multipotent stem cells as well as committed erythroid and megakaryocyte progenitors and mature thrombocytes express PCLP (McNagny et al, 1997). Rat podocalyxin is present in megakaryocytes and thrombocytes (Miettinen et al, 1999). In the aorta–gonad–mesonephric region of the mouse embryo, PCLP is present in haemangioblasts, the early progenitor cells that mature into endothelial or haematopoietic stem cells (Hara et al, 1999). In mice, PCLP expression has not been studied in the more committed haematopoietic progenitors.
Based on their shared structural features, it has been suggested that PCLP, CD34-protein and the recently found endoglycan form a sialomucin family (Sassetti et al, 2000). All three proteins are present in endothelial cells. The CD34-protein and endoglycan are also present in human haematopoietic stem cells. Human PCLP is expressed by the glomerular podocytes but its expression during human haematopoiesis has not been studied. The functions of PCLP, CD34 and endoglycan are mostly unknown. In the kidney, PCLP has an anti-adhesive function, keeping the glomerular filtration slits open. This role is due to both the repulsion forces created by PCLPs negatively charged sulphate and sialic acid residues (Dekan et al, 1991) and its carboxyterminal DTHL-sequence, the consensus binding domain for PDZ-proteins. In glomerular podocytes, podocalyxin, with this domain, is bound to Na+/H+ exchange regulatory factor-2 and further, via ezrin, to the actin cytoskeleton (Takeda et al, 2000; Li et al, 2001; Orlando et al, 2001). Another described function of PCLP is adhesive. Both PCLP and CD34 of the high endothelial venule cells express the MECA-79 epitope, a ligand for l-selectin, which is a key molecule in the homing process of leucocytes entering into the peripheral lymphoid tissues (Sassetti et al, 1998). The MECA-79 epitope has not been found in the PCLP of podocytes or other types of endothelium (Segawa et al, 1997), which suggests that PCLP has tissue specific functions.
Human haematopoiesis is not fully understood and more markers for cells at different stages of development are needed. In experimental animal models, PCLP is expressed by haematopoietic stem cells and mature thrombocytes, but the expression of human PCLP has not been studied. Here we show that PCLP is expressed by human CD34-positive haematopoietic stem cells. In addition we show that PCLP expression decreases during the maturation of haematopoietic cells.
Materials and methods
Cells and tissues
CD34+ cells were collected from mobilized peripheral blood mononuclear cells obtained from autologous transplants from patients with multiple myeloma or chronic lymphoid leukaemia. Purification of the CD34+ cells was performed according to the standard methods of CLINIMacs (mean purity 93·5%) and by using anti-CD34 magnetic beads (Miltenyi Biotech GmbH, Bergisch Gladbach, Germany).
The cells from Granulocyte–Macrophage Colony Forming Unit (CFU-GM), Granulocyte-CFU (CFU-G), Macrophage-CFU (CFU-M), and erythroid-burst forming units (BFU-E) colonies were obtained from the in vitro cultures of normal human bone marrow mononuclear cells (MNC) cultured for routine purposes as previously described (Juvonen et al, 1991). The CFU-GM, CFU-G, CFU-M or BFU-E colonies were identified by inversion microscopy, based on the typical morphology or red colour of the cells, respectively. BFU-E, CFU-G, CFU-M, CFU-GM and mixed type colonies (CFU-GM cells mixed with some CFU-G and CFU-M colony cells) were picked up separately with a thin glass pipette under microscopic observation for further studies after 14 d of culture (Fig 1).
For the CFU-Meg (megakaryocyte) cultures, mobilized peripheral blood CD34+-cells (1 × 104 / ml) were cultured for 7 d in six-well culture plates (CellStar, Greiner, Germany) in 5 ml medium (StemSpan SFEM, StemCell Technologies, Vancouver, Canada) supplemented with 10 ng/ml of human interleukin (IL)-3 and IL-6 and 50 ng/ml of thrombopoietin (StemCell Technologies). At day 7, cells expressing the megakaryocyte marker CD41+ were isolated using fluorescein isothiocyanate (FITC)-conjugated anti-CD41 antibodies (1 μg/1×106 cells; Caltag Laboratories, Burlingame, CA, USA) and anti-FITC magnetic beads with miniMacs purification columns (Miltenyi). The isolated cells were still mononuclear and small in size and megakaryocytes could not be identified (Fig 1).
For the reverse transcription-polymerase chain reaction (RT-PCR) and Western blot experiments, peripheral blood leucocyte populations were isolated from buffy coat samples obtained from the Finnish Red Cross Central Laboratory (Helsinki). First the cells were separated into MNC and granulocytes by gradient centrifugation (Ficoll-paque, Pharmacia Biothec, Uppsala, Sweden) and then isolated further by using the following FITC-conjugated antibodies: anti-CD3 for T-lymphocytes, anti-CD19 for B-lymphocytes, anti-CD14 for monocytes (Becton Dickinson), anti-FITC magnetic beads, and miniMacs purification columns (Miltenyi). The Ficoll-separated granulocytes were incubated with the same antibodies, and collected by negative selection with the magnetic beads. The purity (mean 96%) of the isolated populations was checked with flow cytometry. Erythrocytes and thrombocytes were isolated from whole blood donations. After Ficoll-paque gradient centrifugation, thrombocytes were obtained by centrifuging (1300 g, 15 min) the platelet rich plasma fraction into a pellet. Erythrocytes were isolated from the lowest fraction that passed through the ficoll-paque reagent. In order to eliminate the remaining leucocytes in the thrombocyte and erythrocyte populations, the cells were stained with FITC-conjugated anti-CD45 antibodies (Becton Dickinson), and after miniMacs purification, negative fractions were collected. The purity of the isolates was checked with May Grunwald Giemsa (MGG)-staining.
Cells from a chronic myeloid leukaemia cell line K562 [Andersson et al, 1979, American type cell culture (ATCC) number CCL 243], were cultured in Roswell Park Memorial Institute (RPMI) medium supplemented with 10% fetal calf serum (FCS), l-glutamine 300 μg/ml, penicillin 100 U/ml, and streptomycin 100 μg/ml (EuroClone, Wetherby West Yorkshire, UK). Human umbilical vein endothelial cell line (HU-VEC-C, ATCC number CRL 1730) cells were cultured in the same medium. The kidney tissues for immunofluorescence staining experiments and preparation of glomerular extracts were obtained from the normal parts of kidneys that had been removed because of renal cancer, as described earlier (Holthofer et al, 1994). The Ethical Committees of the Helsinki University Hospital and the Finnish Red Cross Central Laboratory approved the study.
RNA isolation and RT-reaction
The RNAs for the RT-PCR samples were purified from the haematopoietic cells by using Trizol (Life Technologies, Gibco BRL Inc., Gaithersburg, NJ, USA). The isolated RNA was DNAse-treated (DNAse RQ1, Promega, Madison, WI, USA) and reverse transcribed using random hexanucleotide primers (Boehringer Mannheim, Germany) and Moloney-murine leukaemia virus (MMLV) reverse transscriptase (Promega). For the RT-negative controls (RT-), no MMLV was added. The amount of the isolated RNA was measured with a RiboGreen RNA quantification Reagent Kit (Molecular Probes, Leiden, The Netherlands).
The cDNAs were amplified (total volume 25 μl) in the presence of 0·5 μmol/l primers, 0·2 mmol/l dNTP (Finnzymes, Espoo, Finland), 10x PCR buffer with 15 mmol/l MgCl2 (Perkin-Elmer, Cetus, CT, USA) and 0·6 U AmpliTaq polymerase (Perkin-Elmer). The following program was used: 95°C for 5 min followed by 30 cycles of 30 s at 94°C, 30 s at 55°C for first round, 57°C for nested primers, and 72°C for 30 s, and then finally 72°C for 10 min. The PCR product was analysed by agarose gel electrophoresis and ethidium bromide staining, then purified (QIAquick Gel Extraction Kit, Qiagen GmbH, Hilden, Germany), and sequenced directly using gene-specific primers (ABIPrism 310; Perkin Elmer Applied Biosystems, Foster City, CA, USA). The following primers were designed for the nested RT-PCR of human PCLP: For the first round, sense primer 5′-CCG TGG TCG TCA AAG AAA TC, corresponding to nucleotides 1380–1399, and antisense primer 5′-GTC GTC CTT GGT CAG GTT GT (1788–1807) were used. For the second round, sense 5′-GAC ATG AAG CTA GGG GAC CA (1484–1503), and antisense 5′-TTG AGG CTG ACC ACC TTC TT (1733–1752), were used; numbering is according to GenBankTMaccession number NM 005397. The specificity of the primers to PCLP was confirmed by using the Basic Local Alignment Search Tool (BLAST) protein database. The β-actin gene was used as a control.
Real-time quantitative PCR
For the real time quantitative RT-PCR experiments, carboxyfluorescein (FAM)-labelled probes with primers were used for both the human PCLP gene (Taqman, Assays-on-Demand Hs00193638 Applied Biosystems) and the E2 human ubiquitin conjugating enzyme (UbcH5B gene, GenBankTM Accession number U39317, Taqman, Assays-on-demand Hs00366152). UbcH5B was used as the endogenous control to normalize the amounts of PCLP PCR products (Hamalainen et al, 2001). The experiments with normal leucocytes were performed with samples from four different individuals, and the samples of mobilized CD34+ cells were from two patients. The K562 RNA was isolated from two different cultures and the RNAs from CFU-Meg cells, HU-VEC-C cells, and kidney cortex were isolated from single samples. All samples were analysed at least twice with two duplicates of which the mean values and SD were calculated. The ABI Prism 7700 sequence detector (Applied Biosystems), which was used for the signal detection, was programmed to an initial step of 2 min at 50°C and 10 min at 95 °C, followed by 40 thermal cycles of 15 s at 95°C and 1 min at 60 °C.
The mRNAs were isolated from CD34+ cells, renal cortex and HU-VEC-C endothelial cells by using μMACS isolation kit (Miltenyi). mRNA (4–12 μg per each sample) was separated by electrophoresis in 1% agarose-formaldehyde gel and transferred to a Magna Nylon transfer membrane (MSI Separations Inc., Westborough, MA, USA). The cDNA probe corresponding to nucleotides 1380–1807 (integrated into and amplified in a pGEM®-T Vector, Promega) was labelled with [α-32 P]-dCTP (Amersham, Little Chalfont, Buckinghamshire, UK) by using the random labelling method (Prime a gene, Promega). The filter was prehybridized and hybridized at 42°C over night under denaturing conditions [50% formamide, 5x SSC, 5x Denhartz, 50 mmol/l sodium phosphate pH 6·5, 0·1% sodium dodecyl sulphate (SDS), SScDNA herring sperm 2·5 mg/ml] and washed twice with 2x SSC, 0·1% SDS at 42°C for 30 min, then twice with 1x SSC, 0·1% SDS, at 55°C for 30 min. The results were visualized by a phosphoimager (Fuji Films, Tokyo, Japan).
Cloning of PCLP into pGex-6p-2 vector
The 5′-sequence of PCLP was divided into three parts. The extracellular parts corresponding to nucleotides 234–928 and 920–1588 (with some transmembrane region in the end) and the intracellular part 1542–1884 (with some transmembrane region in the beginning) were amplified with primers having restriction sites for BamHI (sense) and EcoRI (anti-sense). The PCR products obtained from the glomerular, K562 or mixed CFU-GM and BFU-E colony cell cDNAs were inserted into pGEX-6p-2 vectors (Amersham Pharmacia Biotech, Uppsala, Sweden), respectively, and the constructs were transformed into competent E. coli strain DH5α grown in Luria medium. The plasmids were then isolated (Plasmid maxi kit, Qiagen) and the DNA was sequenced using both Glutathione-S-transferase and PCLP specific primers (ABIPrism 310; Perkin Elmer Applied Biosystems).
Anti-human PCLP antibody
Two rabbits were immunized with multiple subcutaneous injections of 1 mg (in 0·5 ml phosphate buffered saline, PBS) of lysine-branched 19 amino acid peptide (vpldnltkddldeeedthl), corresponding to the intracellular carboxyl terminus of PCLP (GenBankTM accession number NM 005397), mixed in 0·5 ml Freund's complete adjuvant (Difco Laboratories, Detroit, MI, USA). The rabbits were boosted twice with 1 mg of the same peptide, mixed in 0·5 ml of Freund's incomplete adjuvant, at 4 and 7 weeks after the initial immunization. Sera were collected 2 weeks after the second boost. Affinity purified anti-human PCLP antibodies (#014) were obtained from the serum by using an affinity column of CNBr-activated Sepharose 4B (Amersham Pharmacia Biotech) coupled with the linear form of the immunization peptide (2 mg/2ml). The antibody was concentrated to 0·38 mg of IgG/ml. Some of anti-PCLP IgG was labelled with FITC (Sigma-Aldrich Finland, Helsinki, Finland), as described earlier (Goding, 1976). The molar FITC/protein ratio of the conjugate (#015) was 2·5.
The specificity of the affinity-purified anti-PCLP antibodies was tested by immunofluorescence on frozen sections of human kidney and on cultured HU-VEC-C cells, and by Western blot analysis of human glomerular or HU-VEC-C cell lysates. The previously described monoclonal anti-PCLP antibody PHM5 (a kind gift from Dr Robert C. Atkins, Monash Medical Centre, Clayton, Australia) was used as a positive control. In order to verify the recognition epitope of anti-PCLP, 1 mg/ml of the original immunizing peptide was mixed in the antibody solution to block the immunofluorescence and Western blot reactions. The recognition site of anti-PCLP was also confirmed with epitope mapping (PEPSpot filter) as described earlier (Laune et al, 2002).
The PB samples were stained at 4°C for 30 min with anti-CD45 antibodies coupled with peridinin chlorophyll, and with phycoerythrin (PE)-coupled anti-CD3, -CD19 or -CD14 antibodies (Becton Dickinson), 10 μl of each antibody/100 μl blood, washed and fixed in 3% paraformaldehyde (PF) at room temperature for 15 min. For staining experiments with the anti-PCLP antibodies the cells were further permeabilized in PBS into which 0·05% saponin and 5% normal rabbit serum was added at 4°C for 10 min. Then the cells corresponding to 100 μl of original blood were stained with 2·5 μg of FITC-conjugated anti-PCLP (#015) or with affinity purified control conjugates including FITC-conjugated rabbit anti-chicken IgG, rabbit anti-mouse IgG (Jackson ImmunoResearch, West Grove, PA, USA) or rabbit anti-goat IgG (Cappel, Costa Mesa, CA, USA), referred to in the figures as C1, C2, and C3, respectively. The cells were incubated at 4°C for 30 min, washed in the presence of saponin and preserved in 1% PF-PBS until they were analysed by flow cytometry using a FACScan flow cytometer (Becton Dickinson). For the CD34+, K562 and HU-VEC-C cells, the staining procedure was similar to that of the leucocytes. For the characterization of the PCLP-positive population of the mobilized CD34+ cells, the cells were double stained with the anti-PCLP antibodies and with PE-conjugated anti-CD117 (C-kit, stem cell factor receptor) or anti-common myeloid progenitor marker CD33 (Becton Dickinson) antibodies, respectively. For each cell population, 10 000 cells were analysed by using a live gate-counting system.
For immmunofluorescence staining experiments we used HU-VEC-C cells grown on glass cover slips and cryostat sections of human kidney. The cells and cryostat sections were fixed in −20°C acetone or in 3% PF-PBS at room temperature for 10 min, and stained with the primary anti-PCLP antibodies, washed, and incubated with the secondary swine anti-rabbit IgG antibodies conjugated with tetramethylrhodamine isothiocyanate (TRITC; R0156, DAKO), as described (Miettinen et al, 1990, 1999). An Olympus BX50 fluorescence microscope equipped with an Orca IIIm CCD-camera (Hamamatsu Photonics, Hamamatsu City, Japan) was used for analysis and documentation of the results.
Western blot analysis
For SDS polyacrylamide gel electrophoresis (SDS-PAGE), the isolated cell populations or human glomeruli were lysed in radioimmunoprecipitation assay (RIPA)-lysis buffer [Tris-buffered saline (TBS), 1% nonidet P-40 (NP-40), 0·5% Sodium deoxycholate, 0·1% SDS] supplemented with a cocktail of enzyme inhibitors (CompleteTM, Boehringer Mannheim), suspended in 4x reducing Laemmli sample buffer (RLSB), boiled for 10 min, and run under reducing conditions using 7·5% gels and a protein Mini-gel electrophoresis system (Bio-Rad Laboratories, Hercules, CA, USA). For Western blot analysis, the separated proteins were transferred to nitrocellulose sheets, reacted with the primary antibodies, washed and incubated with horseradish peroxidase (HRP)-coupled anti-mouse or anti-rabbit IgG (P0260 or P0217, DAKO). An equal amount of protein lysate was loaded in all lanes, which was checked by Coomassie staining of test samples. After washing, the bound antibodies were visualized with the enhance chemiluminescence (ECLTM ) technique (Amersham Life Science, Buckinghamshire, UK).
PCLP mRNA in haematopoietic cells
PCLP mRNA expression was first studied in isolated peripheral blood cell populations. By nested RT-PCR the RNAs isolated from CD19+ cells (B-lymphocytes), CD3+ cells (T-lymphocytes), CD14+ cells (monocytes), and granulocytes were positive, but no PCLP expression was detected in erythrocytes or thrombocytes (Fig 2A).
To study the erythroid and granulocyte-macrophage progenitor cells, colony forming cell cultures of isolated bone marrow MNC were used. After 14 d of culture, the colonies were identified by their typical morphology and collected for RNA isolation (Fig 1). Mobilized peripheral blood CD34+ cells were used for CFU-Meg cultures. After 7 d of culture, the CD41+ cells (Fig 1) were isolated and their RNA was extracted. RNA was isolated also from the mobilized CD34+ cells and from cultured K562 leukaemia cells. K562 cells can be induced to differentiate into myeloid, erythroid or megakaryoid directions depending on the culture conditions (Andersson et al, 1979; Alitalo et al, 1988). The cells were cultured in RPMI medium supplemented with FCS without lineage-inducing reagents. In these culture conditions the cells were mostly of the myeloid phenotype: 82% (70–87%, n = 4) were positive for the myeloid marker CD33, which is expressed by myeloid, erythroid and megakaryocyte CFU-cells and later by myeloblasts and myelocytes (see flow cytometry results below). The RNA extracted from the cells of the CFU-G, CFU-M, CFU-GM and BFU-E colonies, from the CFU-Meg cultures, and from the mobilized CD34+ cells were all PCLP-positive. In addition, the RNA extracted from the K562 cells was positive (Fig 2B). All RT− controls were negative (results not shown).
The PCLP mRNA expression levels of the different cell populations that were positive by RT-PCR were compared more accurately by real time RT-PCR. On average, the CD34+ cells had 11, eight, six, and three times more PCLP mRNA than the isolated granulocytes, T-lymphocytes, monocytes, and B-lymphocytes, respectively. The PCLP-expression level of CD34+-cells was two or 3·5-fold that of the CFU-GM or BFU-E colony cells, respectively, and was equal to that of the CFU-Meg cells. K562 cells contained three times more PCLP mRNA than the CD34+ cells. In all haematopoietic cell populations, the PCLP mRNA levels were lower than in the kidney cortex or HU-VEC-C cells, which expressed approximately 30 times more PCLP mRNA than the CD34+ cells and 10 times more than the K562 cells (Fig 3).
To characterize the haematopoietic PCLP mRNA, the PCLP cDNA from the K562 cells and renal cortex were cloned in three segments into the pGEX-6p-2 vectors. The intracellular domain of CFU-GM PCLP was also cloned. The cloned cDNAs were sequenced and compared with each other and to published sequences. All sequences were identical with the GenBank sequence XM 004727.5/gi 18566712, which is based on expressed sequence tag (EST) sequences. The original sequence (Kershaw et al, 1997, GenBank accession number NM 005397.1) differs by one nucleotide from our results (nucleotide 1570 is adenine instead of guanidine), but both sequences code for alanine. Northern blotting showed that the dominant PCLP mRNA in the mobilized peripheral blood CD34+ cells, HU-VEC-C cells, and renal cortex was approximately 5·4 kb. Minor products of approximately 8·0 and 4·2 kb were also seen in all tissues. In line with the quantitative RT-PCR results, the mRNA signal of the CD34+ cells was weak as compared with the kidney or HU-VEC-C cell samples (Fig 4). Only a very weak signal of 5·4 kb was detected in the mRNA samples of peripheral blood MNC, and no signal could be detected in the mRNA of granulocytes (results not shown).
PCLP protein in haematopoietic cells
To analyse the expression of PCLP protein, affinity purified antibody against the intracellular domain of the human PCLP was used. The anti-PCLP antibody (#014) was specific for PCLP, as shown by its typical immunofluorescent staining of renal glomeruli and endothelial cells [Figs 5A and B (Kerjaschki et al, 1986; Miettinen et al, 1990)], and by the Western blot results on glomerular extracts (Fig 5C). In the PEPSpot filter binding experiments, the antibodies recognized only those PCLP peptides that shared sequences with the 19 amino acid carboxyterminal peptide, which was used as the immunogen (results not shown).
None of the mature leucocytes, thrombocytes or erythrocytes were PCLP-positive by Western blot (Fig 6) or by flow cytometry (Fig 7A). CD34 and K562 cells both expressed PCLP of the same size as in the glomerulus (165 kDa, Fig 6). Of the mobilized CD34+ cells, 28% (mean; range 14–61%, n = 7; Fig 5A) were positive when stained with the FITC-conjugated anti-PCLP antibody (Fig 7B). Also, 90% (mean; range 78–97%, n = 5; Fig 5B) of the K562 cells were positive (Fig 7B). As a positive control, we used detached HU-VEC-C endothelial cells, which were all (n = 4) PCLP-positive (Fig 7B).
The mobilized peripheral blood CD34+ cells are a heterogeneous population containing both fully committed progenitors as well as cells at various stages of differentiation. Only about 1% of the cells are primitive pluripotent haematopoietic stem cells (Osawa et al, 1996). To further characterize the phenotype of the PCLP-positive CD34+ cells, double staining experiments using antibodies against the stem cell marker C-kit receptor (CD117) and the myeloid progenitor marker CD33 were performed. Of the CD34+ cells, 22% and 85% (n = 2) were CD33-positive. In both cases, almost all PCLP positive cells were in the CD33+ population, as shown in Fig 7C. Very few PCLP-positive cells were seen in the CD33-negative fraction. Of the CD34+ cells, 31% (mean; range 10–47%, n = 3) were CD117+. The majority of the PCLP-positive cells was found in the CD117-positive fraction (Fig 7D). We also stained K562 cells with CD117 and CD33 in order to further characterize the nature of these cells when cultured in serum without lineage specific cytokines. Most of the K562 cells, 70–87% (mean 82%, n = 4), were positive for CD33 and 26–47% (mean 34%, n = 4) were positive for CD117 (results not shown).
Our results show that PCLP is expressed in human haematopoietic progenitor cells although at a lower level than in glomerular podocytes or endothelial cells. PCLP mRNA, but no PCLP protein, was found in the mature peripheral blood leucocytes (Figs 2, 3, 6 and 7). The mRNA levels were higher in the more immature haematopoietic progenitor cells as well as in the K562 leukaemia cells (Fig. 3), thus in the cells that also expressed the PCLP protein (Figs 6 and 7). This is the first report demonstrating PCLP in developing human blood cells.
Using the nested RT-PCR technique, we found PCLP mRNA in lymphocytes, monocytes and granulocytes, but not in erythrocytes isolated from peripheral blood. Surprisingly, unlike in chicken or rat, PCLP mRNA was not detected in human platelets. At the protein level, PCLP was not found in mature blood cells by flow cytometry or by Western blotting (Figs 6 and 7). In chicken and mice, PCLP marks the earliest detectable haematopoietic progenitors but its expression decreases as the cells differentiate. To study human haematopoietic progenitors, we made clonogeneic cultures of the mononuclear cells from normal human bone marrow and of isolated mobilized CD34+ cells. On average, isolated mobilized CD34+ cells expressed 11, eight, six or three times more PCLP mRNA than mature granulocytes, T cells, monocytes, or B cells, respectively, as detected by quantitative RT-PCR. The PCLP mRNA levels of the CFU and BFU colonies were only slightly higher than those of the mature leucocyte populations. These cells were analysed after 2 weeks of culture, at a time point when many cells in the colonies were already fully committed, as shown by their typical morphology (Fig 1), and the number of progenitors left in the colonies was probably low. The CFU-Meg cells, derived from isolated CD34+ cells, were cultured for only 7 d in the differentiating conditions before CD41+ cells of the megakaryocyte lineage were isolated for RNA extraction. The PCLP mRNA level of these cells was about the same as that of the original mobilized CD34+ cells. The mobilized CD34+ cells are a heterogenous mixture of primitive and more committed progenitors, and the relative amounts of the different cell populations may vary in isolates obtained from different individuals.
The level of the PCLP protein expression also varied from 14% to 61% (mean 28%) in the different CD34+ cell isolates. In order to further characterize those cells that expressed PCLP, we performed double staining experiments using anti-CD117 or anti-CD33 antibodies and anti-PCLP antibodies. The c-kit receptor, CD117, is expressed by approximately 50% of CD34+ early stem cells as well as by lineage committed myeloid and lymphoid progenitors (Escribano et al, 1998). CD33 is a marker for colony forming cells of the myeloid (including erythroid and megakaryocyte) lineage and is present only in myeloblasts and meylocytes at the later stages of development (de Wynter & Ploemacher, 2001). Among the CD34+ cells almost all PCLP+ cells were also found to be CD33+ and most of the PCLP+ cells were also CD117+ (Figs 7C and D). This shows that at least a subset of the myeloid progenitors express PCLP. To show that these cells include true pluripotent stem cells, antibodies that react with the extracellular domain of haematopoietic PCLP are needed to isolate CD34+ CD117+ PCLP+ cells for in vitro long-term clonogeneic cultures. Such antibodies are currently under study in our laboratory. We used the K562 leukaemia cells as they have the potential to differentiate into the myeloid, erythroid or megakaryocyte lineages (Andersson et al, 1979; Lozzio et al, 1981; Alitalo et al, 1988) and are considered as early myeloid progenitors. In our culture conditions, 78–97% (mean 90%) of the K562 cells expressed PCLP protein and their PCLP mRNA levels were also three times higher than those of the CD34+ cells (Fig 3), which was also in accordance with the protein expression levels (Figs 6 and 7). These results suggest that in man, PCLP is a protein of developing haematopoietic cells and its expression decreases or is lost during maturation into peripheral blood cells, as has been described in chicken and mice. We did not study developing lymphoid cells in this work, but the relatively high expression of PCLP mRNA in B-lymphocytes may be of functional relevance. McNagny et al (1997) did not find thrombomucin in chicken lymphocytes, but in accordance with the present study, Sassetti et al (2000) found PCLP mRNA in human tonsillar lymphocytes. Further studies concerning the presence of PCLP in lymphoid progenitors are needed.
At present we do not know whether the PCLP mRNA of mature leucocytes is functional. We could not demonstrate PCLP protein expression in any peripheral blood cells by flow cytometry or by Western blotting (Figs 6 and 7). Our antibodies recognized the last 19 amino acids of the intracellular domain of PCLP. At least in mice and in chicken, one of the alternatively spliced isoforms codes for a truncated protein that lacks most of the intracellular domain (McNagny et al, 1997; Li et al, 2001) but it is unlikely that mature leucocytes would only express this PCLP. By Northern blotting, the major and the minor PCLP mRNA isoforms of the mobilized CD34+ cells, the HU-VEC-C cells, and the renal cortex were identical, i.e. 5·4, 4·2 and 8·0 kb. In addition, the only PCLP mRNA that was faintly detected in the peripheral blood mononuclear cells was the 5·4 kb major isoform. The PCLP cDNA sequences of the K562 cells, the cells of the CFU-GM colonies, and of renal cortex were also identical.
By its protein structure and genomic organization, PCLP resembles the two other sialomucins CD34 protein and endoglycan (Sassetti et al, 2000). It has been suggested that all three may share functional similarities (McNagny et al, 1997; Sassetti et al, 1998). CD34 and endoglycan are expressed by endothelial and haematopoietic cells, but their functions are mostly unknown. It has been known for some time that most of the true multipotent haematopoietic stem cells that have the ability to restart haematopoiesis are found among the CD34+ cells (Berenson et al, 1988). It has also been suggested that CD34 may play an adhesive role in the homing process to the bone marrow (Healy et al, 1995). Both the CD34 and PCLP proteins of the high endothelial venules express the GlcNac-6-sulphate structure, the MECA-79 epitope, which acts as a l-selectin ligand. The MECA-79 epitope, however, is not found in PCLPs of other tissues (Segawa et al, 1997). This suggests that tissue specific differences in glycosylation affect the functions of PCLP.
The CD34-knockout mice had delays in the kinetics of their haematopoietic recovery, but showed no total lack of haematopoietic stem cell production or defects in the homing of lymphocytes into peripheral lymph nodes (Cheng et al, 1996; Dick, 1999). On the other hand, the homozygous knockout mice for podocalyxin had no vascular or haematopoietic phenotype but had delayed closure of the abdominal cavity as well as glomerular malformations leading to anuria and neonatal death (Doyonnas et al, 2001). Lack of drastic haematopoietic or vascular defects in the CD34 or PCLP knockout mice suggests mutual functional compensation in the cells expressing both sialomucins. In cells that selectively express only PCLP, like in glomerular podocytes or peritoneal mesothelial cells, such compensation is not possible, and therefore it seems that the lack of PCLP leads to a defective phenotype. These results suggest that PCLP, CD34-protein and endoglycan may have overlapping functions in haematopoiesis. All these three proteins are lost upon the differentiation of haematopoietic cells (Katz et al, 1985; Andrews et al, 1986; Sassetti et al, 2000).
In the developing kidney and in mature podocytes the expression of podocalyxin is under the transcriptional control of the Wilms’ tumour suppressor gene WT1 (Palmer et al, 2001; Guo et al, 2002). The time courses of expression of these two proteins overlap and both become restricted to the glomerular podocytes. WT1 is also expressed in human haematopoietic progenitor cells in the bone marrow and is later found in myeloid cells expressing granulocytic/monocytic markers and also in the CD19+ B-lymphocytes, but not in erythroid and megakaryoid cells (Ellisen et al, 2001). Most interestingly WT1 is also expressed in the K562 leukaemia cell line and is down-regulated in these cells during induction of erythroid and megakaryocyte differentation (Phelan et al, 1994) Both in the kidney and haematopoietic cells the expression pattern of PCLP parallels that of WT1. Whether WT1 regulates the expression of PCLP also in the haematopoietic cells as it does in the kidney is currently not known. Similarly, we do not know whether PCLP expression is altered in human leukaemias.
Our data show that, like the other sialomucins CD34 and endoglycan, PCLP is present in human haematopoietic cells. In addition, our data show that PCLP expression is higher in immature cells, such as mobilized CD34+ cells, than in mature cells. Its presence in the CD117-positive as well as in the CD33-positive myeloid progenitor cell populations suggests that PCLP could be used as a new marker for these early myeloid progenitors. To further confirm the nature of PCLP-positive haematopoietic progenitors, anti-PCLP antibodies that are specific for the extracellular epitopes of PCLP are needed. With such antibodies PCLP-positive progenitors can be isolated and analysed in clonogeneic in vitro cultures for their haematopoietic capabilities.
This study was supported by grants from the Sigrid Juselius Foundation, Finska Läkarsällskapet, Paulo Foundation and Helsinki University Central Hospital. We thank Riitta Väisänen and Monica Shoulz for technical assistance, and Eva Åström and Marja-Liisa Solin for their technical support and advice in molecular biology.