Selection Against Undifferentiated Human Embryonic Stem Cells by a Cytotoxic Antibody Recognizing Podocalyxin-Like Protein-1



Future therapeutic applications of differentiated human embryonic stem cells (hESC) carry a risk of teratoma formation by contaminating undifferentiated hESC. We generated 10 monoclonal antibodies (mAbs) against surface antigens of undifferentiated hESC, showing strong reactivity against undifferentiated, but not differentiated hESC. The mAbs did not cross react with mouse fibroblasts and showed weak to no reactivity against human embryonal carcinoma cells. Notably, one antibody (mAb 84) is cytotoxic to undifferentiated hESC and NCCIT cells in a concentration-dependent, complement-independent manner. mAb 84 induced cell death of undifferentiated, but not differentiated hESC within 30 minutes of incubation, and immunoprecipitation of the mAb-antigen complex revealed that the antigen is podocalyxin-like protein-1. Importantly, we observed absence of tumor formation when hESC and NCCIT cells were treated with mAb 84 prior to transplantation into severe combined immunodeficiency mice. Our data indicate that mAb 84 may be useful in eliminating residual hESC from differentiated cells populations for clinical applications.

Disclosure of potential conflicts of interest is found at the end of this article.


Human embryonic stem cells (hESC), derived from the inner cell mass of the blastocyst, are pluripotent stem cells that have the ability to proliferate indefinitely in vitro in the undifferentiated state. Under the appropriate conditions, hESC can also be differentiated in vitro and in vivo to cell types representative of all three germ layers: mesoderm, endoderm, and ectoderm. Morphologically, the cells have a high nuclear to cytoplasmic ratio and grow as distinct colonies. They also express high levels of alkaline phosphatase, telomerase, and the transcription factors Oct-4 and Nanog [1, [2], [3]–4]. Routinely, hESC are characterized by the expression of cell surface markers, including stage-specific embryonic antigens (SSEA)-3 and SSEA-4, TRA-1–60 and TRA-1–81. However, these surface antigens are not unique to hESC and have been previously characterized in human embryonal carcinoma (EC) cells [5, 6]. Furthermore, the monoclonal antibodies (mAbs) targeting these antigens were raised against human EC cells and mouse embryos. The generation of mAbs specific to hESC surface markers is important because it allows for the identification of novel antigens whose expression is restricted to undifferentiated hESC. Elucidating the role/mechanism of these different cell surface antigens in development and pluripotency will contribute to the understanding of stem cell regulation. Another benefit of hESC-surface specific mAbs would be the ability to incorporate them into a cell separation process before therapy. Following differentiation of hESC to cells of a specific lineage, it is crucial to eliminate residual undifferentiated hESC from the population because these cells could potentially result in the formation of teratomas in vivo after transplantation [7, [8]–9].

In this study, a panel of 10 mAbs was generated following immunization of Balb/C mice with live hESC. These mAbs showed strong reactivity to undifferentiated hESC lines, however, reactivity was reduced or absent in hESC-derived embryoid bodies, mouse embryonic stem cells, mouse feeders, human EC cells, and other human cell lines. Interestingly, one of the clones, mAb 84, which reacts with podocalyxin-like protein-1 (PODXL), not only binds but also kills undifferentiated hESC and the EC cell line NCCIT within 15–30 minutes of incubation in a concentration-dependent and complement-independent manner. Cytotoxicity was, however, restricted to the undifferentiated phenotype; thus, differentiation of hESC resulted in a reduction in killing efficiency by the mAb. In severe combined immunodeficient (SCID) mice model, hESC and NCCIT treated with mAb 84 prior to injection did not form teratomas in vivo compared to untreated cells that developed teratomas within 6–9 weeks postinjection. Because of its selectivity to undifferentiated hESC, this mAb could potentially be used to efficiently remove residual hESC prior to transplantation of the differentiated cell types. To our knowledge, this is the first report of a cytotoxic mAb specifically targeting undifferentiated hESC.

Materials and Methods

Cell Culture

Human embryonic stem cell lines, HES-2, HES-3, and HES-4, were obtained from ES Cell International (ECI, Singapore, The cells were cultured at 37°C/5% CO2 on mitomycin-C inactivated feeders (∼8 × 104 cells/cm2) in gelatin-coated organ culture dishes (cocultures) or on matrigel-coated organ culture dishes supplemented with conditioned medium from mouse feeders, ΔE-MEF (feeder-free cultures) [10]. The mediums used for culturing hESC were either HES medium or KNOCKOUT (KO) medium [11]. For cocultures, hESC colonies were passaged by mechanical dissection [2]; for feeder-free cultures, hESC were passaged following enzymatic treatment as described previously [11].

Mouse embryonic stem cell lines (mESC), CS-1, and E14 were gifts from Dr. Chyuan-Sheng Lin (College of Physicians and Surgeons, Columbia University [12]) and Dr. Bing Lim (Genome Institute of Singapore [13]), respectively and were cultured as described previously [14]. Human EC cell line, 2,102 Ep was a gift from Prof. Peter Andrews (University of Sheffield) and was cultured as described before [15]. Human EC and cancer cell lines NTERA-2 cL.D1 (CRL-1973), NCCIT (CRL-2073), and HeLa (CCL-2) were purchased from the American Type Culture Collection and cultured according to suppliers instructions. Human embryonic kidney cell line, 293-HEK (Invitrogen, Carlsbad, CA, was cultured according to the protocol provided.

To induce hESC differentiation in vitro, HES-3 cells were harvested as clumps and cultured as embryoid bodies (EB) for 8 days in EB-medium (80% KO-DMEM, 20% FCS, 25 U/ml penicillin, 25 μg/ml streptomycin, 2 mM L-glutamine, 0.1 mM NEAA, and 0.1 mM 2-mercaptoethanol) on nonadherent suspension culture dishes (Corning Life Sciences, Acton, MA, Subsequently, the EB were dissociated with trypsin and plated on gelatinized culture dishes in EB-medium for an additional 14 days [16].

Generation of Monoclonal Antibodies

Two 6-week old female BalbC mice received five consecutive weekly immunizations of 5 × 106 HES-3 cells/mice suspended in phosphate-buffered saline (PBS+) or monophosphoryl-lipid A + trehalose dicorynomycolate (MPL+TDM) adjuvant (Sigma-Genosys, Cambridge, U.K., in the intraperitoneal cavity. Splenocytes from the mice were fused with SP2/0 mouse myeloma cells using the ClonalCell-HY Hybridoma Cloning Kit (Stem Cell Technologies Inc., Vancouver, BC, Canada, Hybridomas were isolated 10–14 days after plating and cultured in 96-well followed by 24-well tissue culture plates containing Medium E. Culture supernatant from each hybridoma was collected and reactivity to hESC assessed by flow cytometry. Isotyping was performed using the ClonalCell-InstantCHEK Isotyping Kit (Stem Cell Technologies). Animal experiments were performed in accordance with NIH and The National Advisory Committee for Laboratory Animal Research guidelines (The National University of Singapore Institutional Review Board protocol 05-122 and Biopolis Institutional Animal Care and Use Committee approval 050,049).

Flow Cytometry Analysis

Cells were harvested as single cell suspensions using trypsin, resuspended at 2 × 105 cells per 10 μl volume in 1% bovine serum albumin (BSA)/PBS and incubated for 45 min with each mAb clone (150 μl culture supernatant or 5 μg purified mAb in 200 μl 1% BSA/PBS) or mAb to SSEA-4 (neat, Developmental Studies Hybridomas Bank, MC-813–70,, TRA-1–60, TRA-1–81, 2.5 μg in 200 μl 1% BSA/PBS, (Chemicon, Temecula, CA,, human podocalyxin (PODXL; R&D systems, MAB1658) and polyclonal antibody (pAb) to human PODXL (5 μg in 200 μl 1% BSA/PBS, R&D Systems Inc., Minneapolis, AF1658). Cells were then washed with cold 1% BSA/PBS, and further incubated for 15 minutes with a 1:500 dilution of goat α-mouse antibody fluorescein isothiocyanate (FITC)-conjugated (DAKO, Glostrup, Denmark, After incubation, the cells were again washed and resuspended in 1% BSA/PBS and 1.25 mg/ml propidium iodide (PI) for analysis on a FACScan (Becton, Dickinson and Company, Franklin Lakes, NJ, All incubations were performed at 4°C unless otherwise indicated. As a negative control, cells were stained with either IgM (SSEA-1) or mouse IgG2b (BD Pharmingen, San Diego, isotype controls.


Cells were fixed in 4% paraformaldehyde at room temperature for 45 minutes and incubated with culture supernatant from each mAb clone at room temperature for 1 hour. Localization of antibodies was visualized using goat α-mouse antibody conjugated with either FITC or phycoerythrin (1:500 dilution; DAKO).

Cytotoxicity Assays

Cytotoxicity of mAb 84 on cells was evaluated using propidium iodide (PI) exclusion assays and flow cytometry. As described above, single cell suspensions at 2 × 105 cells per 10 μl volume in 1% BSA/PBS were incubated with mAb 84 (150 μl culture supernatant or 5 μg purified mAb in 200 μl 1% BSA/PBS), mAb to human PODXL or pAb to human PODXL (5 μg in 200 μl 1% BSA/PBS, R&D systems) for 45 minutes. After which, cells were washed and resuspended in 1% BSA/PBS and 1.25 mg/ml PI for analysis by Facial Action Coding System FACS. For dosage studies, HES-3 cells were incubated with 0.1, 0.5, 1, 5 and 15 μg purified mAb 84 in 200 μl 1% BSA/PBS. For time course studies, HES-3 cells were incubated with 5 μg purified mAb 84 in 200 μl 1% BSA/PBS and harvested for analysis at 15, 30, and 45 minutes after addition of the mAb. For hypercross-linking experiments, HES-3 cells after primary mAb incubation were washed and further incubated with a goat α-mouse secondary antibody (5 μg in 200 μl 1% BSA/PBS, DAKO) for 45 minutes. As a negative control, cells were incubated with the isotype control, mAb 85. All incubations were performed at 4°C unless otherwise indicated. To validate the results obtained using PI exclusion assays, viability for each sample was also determined using trypan blue exclusion.


In order to identify the antigen target for mAb 84, feeder-free cultures of hESC were grown to confluence in 6 cm Falcon Petri dishes (BD Biosciences) and lysed by scraping in 2% Triton/PBS+. Cell lysate was clarified by centrifugation and used immediately for immunoprecipitation (IP). IP of the antigen was carried out using the automated Phynexus MEA system (Phynexus, Inc., San Jose, CA, Briefly, mAb 84 (∼100 μg) was directly captured onto Protein A PhyTip columns (5 μl resin bed). After washing away unbound proteins with Wash Buffer I (10 mM NaH2PO4/140 mM NaCl pH 7.4), clarified cell lysate from 5 × 106 cells were passed through the column functionalized with mAb 84. The column was further washed with Wash Buffer II (140 mM NaCl pH 7.4), and bound proteins were eluted at low pH with Elution Buffer (200 mM NaH2PO4/140 mM NaCl pH 2.5) and neutralized immediately with 1 M Tris-Cl pH 9.0. The eluate was stored at 4°C for further analysis.

SDS–PAGE and Western Blot Analysis

SDS–PAGE and Western blotting were performed by the methods of Laemmli [17] and Towbin [18] respectively. Briefly, eluates from IP were separated by SDS–PAGE (NuPAGE 4–12% gradient gel, Invitrogen) under reducing conditions followed by either Western blotting or silver staining. For Western blotting, resolved proteins were transferred onto polyvinylidene fluoride membrane (Millipore, Billerica, MA, at 100 V for 2 hours and immunoblotted with either mAb 84 culture supernatant (diluted 1:1 with 1% BSA/PBS/0.1% Tween-20), mouse mAb to human PODXL or goat pAb to human PODXL (200 ng/ml, R&D Systems) followed by goat α-mouse or rabbit α-goat antibodies horseradish peroxidase-conjugated (1:10000 dilution, DAKO and Pierce (Rockford, IL,, respectively). Binding of HRP-conjugated secondary antibodies were visualized by ECL detection (GE Healthcare, Uppsala, Sweden, Silver staining was performed using SilverQuest silver staining kit (Invitrogen, Carlsbad, CA, according to the manufacturer's protocol and the protein band corresponding to the band on the Western Blot was manually excised for Mass Spectrometry Analysis (see online supplementary methods) [19].

SCID Tumor Model with hESC and NCCIT

To test the tumor-forming potential of hESC and NCCIT in vivo, cells were harvested by PBS and trypsin, respectively, and treated with antibodies as described above. The single-cell suspension from hESC (4 × 106 cells/per animal in a 50 μl volume) was injected directly into the quadriceps of the right hind leg of a male SCID mouse. For NCCIT, the cells were subsequently mixed with CCD919 feeder fibroblasts (105 NCCIT + 106 CCD919 in 50 μl per animal) and injected as described above.

Animal experiments were performed in accordance with NIH and NACLAR guidelines (NUS IRB protocol 05–020, Biopolis IACUC approval 050,008). Tumor formation was monitored visually using a simple grading system that was confirmed by caliper measurements: grade 0= no teratoma visible (6.32 mm average maximal hind leg diameter, n= 10), grade 1= teratoma just detectable (10.55 mm average), grade 2= teratoma obvious (13.2 mm average), and grade 3= teratoma impedes locomotion (14.52 mm average).

Statistical Analysis

All experiments were performed minimally in duplicates and were representative of at least duplicate experiments under different culture conditions. Error bars represent mean ± SEM.


Reactivity of mAbs with hESC and Other Cells Lines In Vitro

In order to raise a panel of mAbs to cell surface markers on undifferentiated hESC, viable HES-3 cells were used to immunize Balb/C mice either in PBS or in MPL+TDM adjuvant. From a total of 1,114 hybridomas selected (114 and 1,000 hybridomas obtained using the respective adjuvants), 10 clones were found to produce mAbs that had reactivity to surface antigens on HES-3 cells following primary screening by flow cytometry (Fig. 1A). Furthermore, binding of the mAbs to HES-3 colonies was confirmed by immunocytochemistry (Fig. 1B and online supplemental Fig. 1). By isotyping, it was found that nine of the mAbs are immunoglobulin M while the remaining one is an IgG2a (mAb 8). Screening with other hESC lines, HES-2 and HES-4 revealed that the mAb reactivity was not only limited to HES-3 (Table 1). Furthermore, antibody binding was reduced after the cells were induced to form EBs, suggesting a downregulation of antigen expression during differentiation.

Figure Figure 1..

Analysis of mAb binding to undifferentiated hESC lines by flow cytometry and immunocytochemistry. (A): Human embryonic stem cell line, HES-3 was stained with different mAb clones raised against cell surface antigens on human embryonic stem cell (hESC) or mAb to TRA-1–60. Antibodies bound to cells were detected with a fluorescein isothiocyanate (FITC)-conjugated anti-mouse antibody. The shaded histogram represents staining with the negative control and open histograms represent staining with primary antibodies. (B): Immunostaining of HES-3 colonies cultured on feeders with different mAb clones raised against cell surface antigens on hESC. Colonies were subsequently stained with α-mouse detection antibody conjugated with either FITC (mAb 5, 14, 63, 85 and 95) or PE (mAb 84). Negative staining was observed with the isotype control, SSEA-1 (data not shown). Scale bar: 200 μm. Abbreviations: mAb, monoclonal antibody.

Table Table 1.. Summary of mAb Reactivity to Different Cell Lines
original image

Secondary screening was performed for these 10 clones to determine the cross reactivity of the mAbs with other cell lines, namely mouse feeders (ΔE-MEF), mESC, human EC cells, such as NCCIT, and other human cell lines (HEK-293, HeLa) (Table 1). It was observed that there was no mAb reactivity with the mouse feeders that the hESC were cultured on prior to immunization. In addition, mAb reactivity was restricted to hESC and not to either of the mESC lines tested (except mAb14 on CS-1). When we compared the reactivity of TRA-1–60, TRA-1–81, and SSEA-4 with our panel of mAbs on human EC cells, as expected, strong reactivity was observed for all three antibodies (TRA-1–60/81 and SSEA-4) on the EC cell lines tested. In contrast, most of our mAb panel had weak reactivity with at least two out of the three EC cell lines tested. Interestingly, mAb 84, 95, 375, 432 and 529 had no or weak reactivity with 2102 Ep, NTERA or NCCIT. This result is indicative that there are differences in antigen expression between hESC and EC cells, and even between different EC cell lines. Furthermore, when screened against other human cell lines, seven of the mAb clones do not bind to HEK-293 or HeLa cells. However, for mAb 5 and mAb 63, the reactivity with these two cell lines increased compared to hESC. This result suggests that the expression of the antigen targets for these two mAbs clones may be higher in terminally differentiated cells compared to ES cells. In 2-color flow cytometry analysis, we observed that >95% of mAb-positive HES-3 cells were also positive for Oct-4 suggesting a strong correlation between the expression of the target antigen and Oct-4 (see online supplemental Fig. 2).

Characterization of the Cytotoxic Antibody mAb 84

During screening of the mAb panel, it was observed that HES-3 cells incubated with mAb 84 had a significant decrease in viability compared to cells incubated with other IgM mAbs (e.g., mAb 85) (Fig. 2A, column 1). Based on PI exclusion assays, only 7% of cells remained viable after incubation with mAb 84 (45 minutes at 4°C) compared to mAb 85. This cytotoxic effect was not restricted to HES-3 but was also present for two other hESC lines, HES-2 and HES-4, and the EC cell line, NCCIT (Fig. 2A, columns 2–4), with a viability of 8%, 15%, and 9%, respectively, after incubation with mAb 84. In contrast, another EC cell line, 2,102 Ep maintained a viability of 94% after incubation (Fig. 2A, column 5). When the cells were visualized under phase contrast microscopy, it was apparent that HES-3 and NCCIT cells incubated with mAb 84 showed significant clumping compared to NTERA and 2,102 Ep cells incubated with mAb 84 or HES-3 and NCCIT cells incubated with mAb 85 (Fig. 2B). Taking together binding and cytotoxicity data for mAb 84 (Table 1 and Fig. 2), it can be inferred that the cytotoxic effect of mAb 84 is correlated with the binding of the mAb to the cell lines, as mAb 84 does not exert a cytotoxic effect on cells that do not bind the mAb.

Figure Figure 2..

Cytotoxicity of mAb 84 on human embryonic stem cell (hESC) and embryonal carcinoma (EC) cells. (A): Cell were incubated with 5 μg mAb 84 or mAb 85 (Isotype control) at 4°C for 45 minute, after which cells were harvested for analysis by propidium iodide exclusion assay on the flow cytometer. Gated region in the scatter plot represents the viable cell population. (B): Microscopic analysis of hESC and EC cells incubated with mAb 84. Similar to (A), cells were incubated with mAb 84 and mAb 85 at 4°C for 45 min in 24-well plates and analyzed microscopically. Scale bar: 100 μm.

In time course studies, HES-3 cells (2 × 105) were incubated with 5 μg mAb 84 or mAb 85, and the cells were harvested every 15 minute for analysis by PI exclusion and trypan blue exclusion assays (Fig. 3A and 3B, respectively). By PI exclusion assay, the cytotoxic effect of mAb 84 on HES-3 cells was observed as quickly as 15 minutes after incubation with the viability dropping to 33%, with a further decrease in viability to 20% after 45 minutes. These results were confirmed by trypan blue exclusion. Interestingly, the decrease in viability based on this assay occurred between 15–30 minutes after incubation, however, the final viability after 45 minutes of incubation also corresponded to ∼20%. When the concentration of mAb 84 was titrated over the range of 0.1–15 μg, it was found that the cytotoxic effect of mAb 84 on HES-3 cells was concentration-dependent (Fig. 3C). Approximately 1 μg (1 pmol) of purified mAb 84 was sufficient to cause a >70% decrease in hESC viability. Until this stage, cytotoxicity assays had been evaluated at 4°C to minimize internalization of the antigen-antibody complex into the cells. To investigate the effect of temperature on cytotoxicity, hESC was incubated with both purified and non-purified (culture supernatant) mAb 84 at 4°C and 37°C (Fig. 3D). By PI exclusion assays, it was found that temperature did not affect the cytotoxicity of mAb on hESC (>75% killing with non-purified mAb 84). Furthermore, mAb 84 was equally cytotoxic to hESC after purification by protein A (>77% killing was observed for both purified and non-purified mAb 84 at 4°C incubation). This result suggests that mAb 84-induced toxicity on hESC was not complement-mediated because cell-killing efficiency was comparable in the presence or absence of fetal bovine serum in the medium. We previously observed that mAb 84 binding to hESC was down-regulated in 8-day-old embryoid bodies (Table 1). In order to determine if cytotoxicity of mAb 84 was specific to the undifferentiated phenotype, hESC was induced to differentiate either by depriving the cultures of fibroblast growth factor 2 (FGF2) or by EB formation (Fig. 4A). Differentiation was assessed based on the expression of the pluripotent marker, TRA-1–60. After 12 days of FGF2 withdrawal, partial differentiation of hESC was observed, with only 49% of the cell population still expressing TRA-1–60 compared to the undifferentiated hESC culture (>95% TRA-1–60 +ve). Differentiation via the EB route yielded >99% of TRA-1–60 -ve cells. When cells from these three conditions were incubated with mAb 84, the efficiency of cell killing corresponded closely with the percentage of TRA-1–60 +ve cells (Fig. 4B). For undifferentiated hESC, only ∼1% of cells remained viable after incubation with mAb 84. This percentage increased to 69% and 99% for FGF2-starved and EB cultures, respectively. In addition, using 2-color flow cytometry to analyze the colocalization of TRA-1–60 and mAb 84, we observed a decrease in TRA-1–60+/mAb 84+ population from >82% in undifferentiated hESC to less that 26% after 7 days of differentiation as embryoid bodies. By day 10 of EB formation, only 2% of the cells were TRA-1–60+/mAb 84+ (see supplemental online Fig. 3).

Figure Figure 3..

Effect of time, dosage and temperature on mAb 84-mediated killing of hESC. (A and B) HES-3 cells were incubated with 5 μg of mAb 84 (▴) and mAb 85 (▪, Isotype control) at 4°C. Cells were harvested at 15, 30 and 45 min after the addition of the mAb and cell viability was analyzed by PI exclusion (A) or trypan blue exclusion assays (B) (n= 3). (C): HES-3 cells were incubated with 0.1–15 μg of mAb 84 (▴) and mAb 85 (▪) at 4°C. After 45 minutes, cells were harvested and viability was assessed by PI exclusion assay (n= 3). (D): HES-3 cells were incubated with purified (equation image) and non-purified (culture supernatant) (▪) mAb 84 at 4°C and 37°C for 45 minutes. The cells were then harvested for analysis by PI exclusion assay on the flow cytometer. Incubation with mAb 85 served as an isotype control (□) (n= 3). All error bars represent mean ± SEM. Abbreviation: mAb, monoclonal antibody.

Figure Figure 4..

Relationship between hESC pluripotency and killing efficiency by mAb 84. Single cell suspensions of undifferentiated and differentiated HES-3 cells were: (A): stained with mAb to TRA-1–60. Antibodies bound to cells were detected with a FITC-conjugated anti-mouse antibody. The shaded histogram represents staining with the negative control and open histograms represent staining with anti-TRA-1–60 mAb. (B): incubation with 5 μg mAb 84 at 4°C for 45 minutes. After which, cells were harvested for analysis by PI exclusion assay on the flow cytometer. Gated region in the scatter plot represents the viable cell population. Abbreviation: mAb, monoclonal antibody; PI, propidium iodide.

Identification and Validation of mAb 84 Antigen Target on hESC

To identify the target antigen on hESC responsible for the cytotoxic effect of mAb 84, immunoprecipitation experiments were performed. Whole cell lysate was passed through a PhyTip column (Phynexus) containing protein A resin and mAb 84. Proteins that were captured by affinity interaction were resolved on protein gels and probed with mAb 84. Based on molecular weight markers, an antigen band of <190 kDa was detected (Fig. 5A, lane 1). The lower band at ∼25 kDa detected by the secondary antibody was identified as the light chain of mAb 84 after reduction. The corresponding band on a silver-stained gel was isolated and identified by mass spectrometry. From protein database search with the peptides obtained, the antigen band was identified as podocalyxin-like protein-1 precursor (PCLP1 or PODXL; Accession No O00592). In order to validate that the antigen target is PODXL, immunoprecipitation with mAb 84 was repeated and the eluate from the column was probed with commercially-available antibodies to PODXL (Fig. 5A lanes 2 and 3). From the Western blots, a band of comparable molecular weight was detected in all three lanes thus confirming the identity of PODXL. By RT-PCR, the two variants of PODXL (Accession Nos: NP_001018121 and O00592 for variant 1 and 2, respectively) were also found to be transcribed in hESC (data not shown).

Figure Figure 5..

Identification and characterization of PODXL on hESC. (A): Western blot analysis of target antigen immunoprecipitated by mAb 84. Affinity purified antigen from hESC lysate using PhyTip columns containing protein A resin and mAb 84 was resolved on SDS-PAGE, subjected to Western blot and probed with mAb 84 (Lane 1), mAb to human PODXL (mAb-PODXL, Lane two) and pAb to human PODXL (pAb-PODXL, Lane 3). (B): Cytotoxicity of commercially-available anti-PODXL antibodies on hESC compared to mAb 84. HES-3 cells were incubated with 5 μg of mAb 84, mAb-PODXL pAb-PODXL or mAb 85 (Isotype control) at 4°C for 45 min. In some experiments, hESC were further hypercross-linked with goat-anti mouse antibodies. After which, cells were harvested for analysis by propidium iodide exclusion assay on the flow cytometer. Gated region in the scatter plot represents the viable cell population. All error bars represent mean ± SEM. Abbreviation: mAb, monoclonal antibody.

Having identified the antigen target of mAb 84, we proceeded with investigating whether commercially-available antibodies to PODXL exerted a similar cytotoxic effect on hESC. From Figure 5A, it is apparent that though the three sources of antibodies (mAb and pAb) were specific to human PODXL, cytotoxicity was only observed for mAb 84 and not the two commercially available anti-PODXL antibodies (Fig. 5B). Several groups have previously reported that apoptosis can be induced by hypercross-linking of primary antibodies bound to antigens on cells, such as CD19, 20, and 22 on cells [20, 21]. Since mAb 84 is an IgM (pentameric) while mAb-PODXL and pAb-PODXL are both IgG (bivalent), we decided to investigate if hypercross-linking of mAb-PODXL or pAb-PODXL with goat-anti mouse (GAM) antibodies would mimic mAb 84-mediated killing of hESC. Incubation of hESC with primary antibodies followed by GAM antibodies failed to induce a similar cytotoxic effect as mAb 84 (Fig. 5B).

Transplantation of mAb84-Treated Cells into a SCID Mouse Tumor Model

Finally, we investigated whether the cytotoxic effect of mAb 84 could be used to prevent tumor formation induced by a cell line that is responsive to mAb 84 killing. Single-cell suspensions of hESC and NCCIT were either treated with mAb 84 or left untreated under identical conditions. When injected into the hind leg of SCID mice, all untreated transplants generated tumors (Fig. 6), which started to form palpable and visible tumors between 6–9 weeks' postinjection and grew to full-sized tumors within 12 weeks. In contrast, when cells were treated with mAb 84 prior to injection, none of the injected animals developed tumors after 18–24 weeks. We conclude from these preliminary data that mAb 84 has the potential to be used for the eradication of contaminating, tumor-generating hESC in preparations used for future therapeutic applications.

Figure Figure 6..

Prevention of tumor formation by mAb 84 in a SCID mouse tumor model. (A): Single-cell suspension of hESC (4 × 106 cells/animal) or (B) NCCIT (105 cells/animal) were incubated with mAb 84 at 4°C for 45 minutes and then injected into the right hind leg muscle of SCID mice (n= 4/group and 3/group respectively). Tumor formation was evaluated with a grading method defined in Materials and Methods. Abbreviation: mAb, monoclonal antibody.


The identification of cell surface antigens is important to hESC research because it is an invaluable tool for monitoring pluripotency and the development of specific cell populations during differentiation. Furthermore, because the respective methods are noninvasive, antibodies specific to cell surface antigens can be used to purify subsets of cells within a heterogeneous pool for detailed analysis or cell transplantation. Several of such cell surface antigens routinely used to characterize pluripotent hESC are SSEA-3, SSEA-4, TRA-1–60, and TRA-1–81. These antigens are also present on human EC cells and a recent study by Draper et al. [22] found that the changes in expression of these antigens during differentiation in hESC are very similar to human EC cells. Thus, the benefit of using hESC as an immunogen is to raise antibodies that are unique to undifferentiated hESC that do not bind to its differentiated progenies.

In this study, live hESC were used for immunization of mice and after primary screening of the hybridomas for mAbs that bind to hESC surface markers, a panel of 10 mAbs were identified. Unlike SSEA-4 and TRA-1–60/81 which reacted strongly to both hESC and human EC cells, five of our antibodies (mAb 84, 95, 375, 432, 529) reacted strongly only with hESC and were negative or weakly reacting to human EC cells. Furthermore, antibody binding correlated with Oct-4 expression and was down-regulated as the hESC differentiate to form EB. These data strongly support the presence of antigens that are differentially expressed on undifferentiated hESC. Moreover, when the entire mAb panel was screened against the H1 hESC line (data not shown), we found that the reactivity profile was similar to that of HES-2, -3 and -4 suggesting that the mAbs bind to antigens that are conserved across the different hESC lines. Uniquely, mAb 84 not only bound to hESC but was also cytotoxic to the cells within 15–30 minutes of incubation. Unlike other cytotoxic mAbs that may require either the activation of complement or hypercross-linking to induce cell death [21, 23], mAb 84 mediated-killing of hESC was found to be independent of both mechanisms. By IP and mass spectrometry analysis, PODXL was identified as the target antigen for mAb 84 on hESC.

PODXL is a heavily glycosylated type-I transmembrane protein belonging to the CD34 family of sialomucins, which include CD34 and endoglycan [24, 25]. PODXL was originally described as the major sialoprotein on podocytes of the kidney glomerulus [26], but was later found to be expressed on vascular endothelial cells and early hematopoietic progenitors [27, 28]. More recently, PODXL has been implicated as an indicator of tumor aggressiveness in breast, liver, and prostate cancers [29, [30]–31]. Human PODXL is located on chromosome 7q32-q33 and encodes for a protein of 528 amino acids [32]. However, because the extracellular domain of PODXL is extensively glycosylated with sialylated O-linked carbohydrates and five potential sites for N-linked glycosylation, the approximate molecular weight of PODXL is 160–165 kDa [33].

Functionally, PODXL has been reported to have quite diverse roles depending on the cell type. In podocytes, PODXL acts as an anti-adhesion molecule that maintain the filtration slits open between podocyte foot processes by charge repulsion [34]. However, in high endothelial venules, PODXL acts as an adhesion molecule binding to L-selectin and mediating the tethering and rolling of lymphocytes [24]. In hESC, PODXL was identified transcriptionally to be highly expressed in undifferentiated hESC [35, 36]. By expressed sequence tag frequency analysis, the level of PODXL expression was down-regulated by almost 2.5-fold in 7–8-day EB and approximately seven and 12 fold in neuroectoderm-like cells and hepatocyte-like cells respectively [35]. This result was supported by immunohistochemistry of hESC and 8-day EB where staining was significantly reduced in the latter [37]. In a separate study by Wei et al. comparing the transcriptome profile of hESC and mESC, they observed that the expression of PODXL was not detected by MPSS in mESC line, E-14 compared to hESC [38]. At the protein level, Schopperle and DeWolf [39] reported that PODXL underwent post-translational glycosylation changes after the exposure of 2 embryonal carcinoma lines to retinoic acid (reduction in MW from 200 kDa to 170 kDa). The failure of anti-TRA-1–60/81 antibodies to bind to the modified PODXL prompted them to suggest the presence of a Stem Cell PODXL (SC-PODXL) on pluripotent stem cells. Taken together, these independent reports of PODXL expression in ESC corresponds closely with our observations of mAb 84 by flow cytometry where binding reactivity was reduced in day 8 EB compared to undifferentiated hESC and absent in mESC. Concomitantly, the decrease or loss in mAb 84-mediated killing on FGF2-starved hESC and day 22 EB respectively can be attributed to the down-regulation of PODXL expression upon differentiation. Furthermore, the simultaneous decrease in mAb 84 and TRA-1–60 binding to hESC during EB formation may implicate the loss of SC-PODXL during differentiation.

More intriguingly, we are interested to elucidate the mechanism responsible for hESC-killing by mAb 84 after binding to PODXL. In a report by Zhang et al. [40], an IgM mAb that targets the cell surface receptor, Porimin (Pro-oncosis receptor inducing membrane injury), was able to induce cell death in Jurkat cells by a process called oncosis [41]. Porimin, like PODXL, is a member of the mucin family because it has multiple O- and N-linked glycosylation sites on the extracellular domain of the protein [42]. Incubation of Jurkat cells with anti-Porimin resulted in rapid cell aggregation in suspension and an increased membrane permeability in >75% of cells after only 20 minutes of incubation. Cell killing was also independent of complement and temperature. Distinct from apoptosis, no DNA fragmentation or apoptotic bodies were observed after incubation with the mAb. By scanning electron microscope, anti-Porimin treated cells were found to have increased membrane pores, blebs, and surface wrinkling. Comparing this with our data, it is surprising that mAb 84 and anti-Porimin share many similar hallmarks of cell killing. Additionally, preliminary results from our group found that mAb 84-treated cells did not exhibit elevated level of caspases, a characteristic of apoptosis (data not shown). Thus, we hypothesize that mAb 84-mediated killing of hESC may be due to oncosis, and we are currently investigating the molecular mechanism involved in this process.

The use of hESC as a starting source of material for differentiation to any cell type in the body has significant benefits to regenerative medicine. However, one of the major concerns after differentiation is the elimination of residual undifferentiated hESC prior to transplantation, because these cells are tumorigenic. Previous work has shown that as few as two mouse ESC implanted into nude mice resulted in the formation of teratomas, and grafting of in vitro differentiated ES cells did not alleviate the situation either [43, 44]. Moreover, Cooke et al. reported that the site in which hESC was grafted would influence the outcome of the teratomas formed. Large tumors of immature cells expressing the pluripotent marker SSEA-3 was predominant in grafts to the liver while smaller tumors of differentiated tissues were prevalent in subcutaneous implants [8]. Several different strategies have been developed to overcome this issue. In two separate studies, Chung et al. [7] and Fukuda et al. [45] demonstrated that recombinant mouse ESC lines carrying the sox1-GFP reporter gene could be used to purify sox1+/GFP+ differentiated neural precursors from ESC by fluorescence activated cell sorting (FACS). Transplantation of the purified cells did not result in teratoma formation while sox1/GFP cells did. In hESC, Hewitt et al. engineered a line expressing α1,3 galactosyltransferase (GalT) under the control of the hTert promoter [46]. Undifferentiated hESC will express GalT which in turn catalyzes and presents the α-gal epitope on the cell surface. The presence of the epitope will render the cells susceptible to circulating antibodies in human serum resulting in cell death in vitro. Despite the success of these strategies, they all require the generation of recombinant ESC lines carrying a selectable gene. Here, we demonstrated that non-manipulated undifferentiated hESC can be rapidly eliminated following incubation with mAb 84 in vitro. Furthermore, from in vivo studies, treatment of hESC and NCCIT with mAb 84 abolished the induction of tumor formation in a sensitive SCID mouse model. We are now expanding these animal models with the ultimate goal to show that in a mixed population of differentiated and undifferentiated hESC, the application of mAb 84 is able to eradicate the tumor-forming undifferentiated cells but would yet allow differentiated cells to engraft. Additionally, we also propose that several of the other hESC-specific mAbs in our panel be used in combination with mAb 84 to ensure the complete removal of residual hESC that may survive mAb 84 killing due to the down-regulated expression of PODXL. These mAb, though not cytotoxic, can be conjugated to magnetic beads and used to isolate the hESC from their differentiated progenies by magnetic cell separation technologies.

In conclusion, this is the first report of a cytotoxic mAb that selectively binds and kills undifferentiated hESC. Potentially, mAb 84 may be used prior to cell transplantation to eliminate residual hESC thus increasing the success and safety of the graft.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


We thank the Agency for Science Technology and Research (A*STAR) for generous funding of this project and Vanessa Ding for the culture and differentiation of hESC by FGF-2 deprivation. Special thanks to Dr. Alan Colman, Dr. Lim Sai Kiang, and Dr. Song Zhiwei for their helpful advice and critical review of the manuscript.