Expression of blood group genes by mesenchymal stem cells


Willy A. Flegel, MD, Department of Transfusion Medicine, Clinical Center, National Institutes of Health, Building 10, Room 1C711, 10 Center Drive, Bethesda, MD 20892-1184, USA. E-mail:


Incompatible blood group antigens are highly immunogenic and can cause graft rejections. Focusing on distinct carbohydrate- and protein-based membrane structures, defined by blood group antigens, we investigated human bone marrow-derived mesenchymal stem cells (MSCs) cultured in human serum. The presence of H (CD173), ABO, RhD, RhCE, RhAG, Kell, urea transporter type B (SLC14A1, previously known as JK), and Duffy antigen receptor of chemokines (DARC) was evaluated at the levels of genome, transcriptome and antigen. Fucosyltransferase-1 (FUT1), RHCE, KEL, SLC14A1 (JK) and DARC mRNA were transcribed in MSCs. FUT1 mRNA transcription was lost during differentiation. The mRNA transcription of SLC14A1 (JK) decreased during chondrogenic differentiation, while that of DARC increased during adipogenic differentiation. All MSCs synthesized SLC14A1 (JK) but no DARC protein. However, none of the protein antigens tested occurred on the surface, indicating a lack of associated protein function in the membrane. As A and B antigens are neither expressed nor adsorbed, concerns of ABO compatibility with human serum supplements during culture are alleviated. The H antigen expression by GD2dim+ MSCs identified two distinct MSC subpopulations and enabled their isolation. We hypothesize that GD2dim+ H+ MSCs retain a better ‘stemness’. Because immunogenic blood group antigens are lacking, they cannot affect MSC engraftment in vivo, which is promising for clinical applications.

The bone marrow (BM) harbours a variety of adult stem cells, among which mesenchymal stem cells (MSCs), also called mesenchymal stromal cells, constitute <0·01% of all BM-derived mononuclear cells (Phinney & Prockop, 2007). Clinically useful for immunomodulation, multipotent differentiation and paracrine effector secretion (Dominici et al, 2001; Herzog et al, 2003; Satija et al, 2009; Schaefer et al, 2009; Siegel et al, 2009), MSCs are utilized in immunomodulatory therapies (Le Blanc et al, 2004, 2008; Uccelli et al, 2006; Ball et al, 2007) and regenerative medicine (Horwitz et al, 1999, 2001). Characterization of MSC subpopulations and their involvement in graft compatibility and immunogenicity may be instrumental in better defining their therapeutic potential in clinical trials.

Several distinct subpopulations sharing various degrees of MSC-like properties are collectively labelled MSCs. Molecules encoding surface structures like CD271, CD49a, W7C5, W8B2, C15, CDCP1, CD340, CD349 and SSEA1/4 may be used to define such subpopulations, while the neural ganglioside GD2, expressed exclusively on MSCs in the BM, is particularly promising as a unifying marker (Quirici et al, 2002; Giesert et al, 2003; Buhring et al, 2004; Martinez et al, 2007; Phinney & Prockop, 2007).

Membrane structures of blood group antigens, like the Duffy (FY) antigen receptor of chemokines (DARC) (Hadley & Peiper, 1997; Swardson-Olver et al, 2002), the ammonia transporter RhAG (Marini et al, 2000) or the urea transporter type B (SLC14A1, previously termed JK) (Shayakul & Hediger, 2004; Wester et al, 2008), and the RhBG and RhCG glycoproteins (Zidi-Yahiaoui et al, 2005; Han et al, 2006), function as receptors or transporters. Investigating these structures may allow classification of MSCs by morphological or genetic criteria, possibly implying membrane functionality. Because alloantibodies play a pivotal role in organ-allograft rejection (Colvin & Smith, 2005; Terasaki & Cai, 2008; Tobian et al, 2008), immunization induced by MSC antigens may result in the alteration or even rejection of a cellular graft is of particular concern for promising MSC applications in regenerative medicine without immunosuppressive therapy.

Focusing on carbohydrate- and protein-based membrane structures that are defined as blood group antigens, we investigated the characteristics of undifferentiated and differentiated MSCs derived from BM and cultured in medium free of animal serum.

Materials and methods

Isolation and differentiation of human MSCs

Bone marrow was obtained under sterile conditions from patients without metabolic or neoplastic diseases during orthopaedic operations. All patients gave written informed consent. The study was approved by the Institutional Review Board (IRB) of the University Hospital Tübingen (IRB no. 95/2005V).

Detailed isolation and characterization protocols and further Materials and Methods specifications are available online (see Supporting information online) (Pittenger et al, 1999; Colter et al, 2001; Flegel et al, 2002; Ji et al, 2004; Kern et al, 2006; Muruganandan et al, 2009).

Characterization of MSCs by surface antigens

CD antigens other than blood group antigens: MSCs were stained using monoclonal antibodies specific for 30 distinct MSC surface antigens and controls. Blood group antigens: MSCs were analysed using antibodies to H; A and B; D, C, E, c and e; K and k; Jka and Jkb; and Fya and Fyb. We evaluated patients who were known to be positive or negative for these antigens. Immunocytochemistry: to explore the expression of the H and GD2 antigens, the MSCs were cultivated and stained in chamber slides using anti-H and anti-GD2 antibodies. Fluorescence labelling was performed with Alexa 594-labelled (anti-H) and Alexa 488-labelled (anti-GD2) antibodies. Nuclei were stained with 4′,6-Diamidino-2-phenylindole (DAPI).

Western blotting for blood group proteins

The protein expression of SLC14A1 (JK) and DARC by undifferentiated MSCs and adipogenic, osteogenic, and chondrogenic differentiated MSCs was analysed by Western blot.

Molecular biology for blood group genes

Blood group genotyping: commercial blood group genotyping kits for ABO, CDE, Kell, Jk and Fy were used (Inno-Train Diagnostik, Kronberg, Germany). MSCs, peripheral blood mononuclear cells (PBMNCs), and KG-1a (myeloblast, chronic myeloid leukaemia) and SK-MEL-28 (melanoma) cell lines were genotyped for blood groups. Patient’s genotypes corresponded to their serological blood groups (data not shown). Qualitative reverse transcription polymerase chain reaction (RT-PCR): we determined the transcription of blood group genes by testing for the mRNA in MSCs using RT-PCR. Quantitative RT-PCR (qRT-PCR): to quantify the influence of differentiation on the blood group expression at the transcriptome level, the mRNA expression of SLC14A1 (JK) and DARC genes was evaluated by qRT-PCR in adipogenic, osteogenic and chondrogenic differentiated as well as in undifferentiated MSCs. The expression of peroxisome proliferator-activated receptor-γ (PPARG) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was determined as controls.


Red blood cell (RBC) blood group phenotyping of the MSC donors: RBC samples of the MSC donors were typed for antigens of the blood group systems ABO (A and B), H (H antigen), Rhesus (D, C, E, c, e), Kell (K and k), Kidd (Jka and Jkb), and Duffy (Fya and Fyb) by standard serological methods. Adsorption of H, A or B antigens to MSCs: after four passages of cultivation in AB serum or O serum the MSCs from an O donor were analysed by flow cytometry using antibodies to the H, A and B antigens.


We used the two-tailed Mann–Whitney U-test and considered < 0·05 statistically significant.


MSCs were isolated from whole BM by a density gradient technique (Colter et al, 2001) and characterized functionally by in vitro differentiation assays for their potential to differentiate into the adipogenic, osteogenic or chondrogenic mesenchymal lineage (Fig S1) (Pittenger et al, 1999). We confirmed the known surface antigen pattern by flow cytometry of undifferentiated MSCs. They were positive for CD10, CD29, CD44, CD59, CD71, CD73, CD90, CD105, CD106, CD130, CD140a, CD140b, CD146, CD166, CD 271, GD2, W8B2 and HLA class I; and negative for negative for CD14, CD15, CD31, CD34, CD43, CD45, CD56, CD93, CD117, CD133, CD243 and HLA class II (Fig S2).

Transcription of blood group mRNA in MSCs

MSCs expressed FUT1, RHCE, KEL, SLC14A1 (JK) and DARC mRNA. No mRNA for ABO, RHD and RHAG was found (Fig 1). The native, undifferentiated MSCs expressed FUT1 mRNA, whereas under adipogenic, osteogenic and chondrogenic differentiation no expression of FUT1 mRNA was detected.

Figure 1.

 Blood group mRNA expression by undifferentiated MSCs and three differentiated human MSC lineages at passages 1 and 2. RNA from peripheral blood mononuclear cells (PBMNCs), two human cell lines and kidney are tested as controls. Representative results are shown. Comparable results were obtained with five donors for the MSCs panels; two experiments for PBMNCs and the two cell lines; and more than 10 experiments with kidney RNA.

Quantitative SLC14A1 (JK) and DARC mRNA expression in MSCs

Because the largest mRNA amounts were detected for SLC14A1 (JK) and DARC (Fig 1), we quantified these mRNAs by qRT-PCR. SLC14A1 (JK) mRNA was reduced in chondrogenic differentiated MSCs relative to undifferentiated MSCs (Table I). Increased DARC mRNA was detected in adipogenic differentiated MSCs; some osteogenic differentiated MSCs had also hugely increased DARC mRNA, although there was much variation among the donors. For comparison, PPARG mRNA was increased, particularly in adipogenic differentiated MSCs. The overall SLC14A1 (JK) mRNA amounts exceeded that of PPARG mRNA, while DARC mRNA amounts were low.

Table I.   Quantitative expression of SLC14A1 (JK), DARC and PPARG mRNA. Thumbnail image of

Intracellular SLC14A1 (JK) and DARC proteins in MSCs

Following the detection of small quantities of SLC14A1 (JK) and DARC mRNA by qRT-PCR, we explored the presence of these proteins by Western blot. Some SLC14A1 (JK) protein was found in all MSC populations, while no DARC protein was detectable (Fig 2).

Figure 2.

 Expression of SLC14A1 (JK) and DARC protein in MSCs. Western blots for SLC14A1 (JK) and DARC proteins are depicted along with controls for the housekeeping protein GADPH. Representative results are shown. Comparable results were obtained in experiments with MSCs from three different donors.

Blood group antigens on the MSC surface

We examined the expression of blood group antigens during several passages. The antigens A, B, D, C, E, c, e, K, k, Jka, and Jkb, as well as the DARC protein could not be detected by flow cytometry (Fig 3A), an established sensitive technique for RBCs (Flegel et al, 2002).

Figure 3.

 Flow cytometry analyses of blood group antigen expression on MSCs. Expression of all 12 blood group antigens tested was lacking from the surface of MSCs (panel A). Representative results are shown for undifferentiated MSCs. Comparable results were obtained with three donors for up to six passages each (P1–P6). H antigen expression is detected in a small subpopulation of MSCs (panel B, red dots in the lower right quadrant of the flow cytometry dot plot), which is much more prominent in KG-1a cells tested for comparison. A representative result from more than 10 experiments with undifferentiated MSCs from different donors is shown. MSCs from a blood group O donor were cultivated in AB serum for four passages and tested for absorption of A, B and H substances (panel C); MSCs with O serum are shown as a control. Comparable results were obtained with MSCs from a blood group A donor. All histogram plots depict the specific antibody profile (solid) and the isotype control staining (line).

Because FUT1 mRNA, encoding the fucosyltransferase enzyme that synthesizes the H antigen, was found in native MSCs (Fig 1), we explored H antigen expression by flow cytometry. Whereas the histogram plots were not indicative, the flow cytometric dot plots revealed a small subpopulation of native MSCs that was positive for the H antigen (anti-CD173; Fig 3B, red dots in the lower right quadrant). The myeloblast KG-1a cells showed a similar wide variability in H antigen expression, although the fraction of strongly H positive cells was much larger than in MSCs. Thus, the flowcytometric dot plots showed a cell population stained by anti-H representing a distinct subpopulation of viable H antigen positive MSCs and KG-1a cells.

MSCs might adsorb A and B antigens to their cell surfaces from the standard culture media containing human serum with soluble A and B substances, as with AB serum. However, A and B antigens were not detected on MSCs after cultivation in AB serum after four passages, indicating no adsorption of the respective blood group substances to the surface of the MSCs (Fig 3C).

Antigen H is expressed on the surface of GD2dim+ MSCs

Immunocytochemical staining of undifferentiated MSCs identified MSC subpopulations differing by GD2 expression, which were dubbed GD2bright+ and GD2dim+ (Fig 4). To corroborate our results with flow cytometry (Fig 3B), we detected some MSCs that expressed the H antigen (CD173). Double staining of GD2 antigen by green fluorescence and the H antigen by red fluorescence revealed that the GD2bright+ cells expressed no H antigen, although a GD2dim+ H+ MSC subpopulation was identified.

Figure 4.

 Identification of MSC subpopulations by GD2 and H antigen expression. Two immunocytochemistry slides from different donors (upper and lower panels) are stained for GD2 (green fluorescence) and H antigens (red fluorescence). The fluorescence for GD2 (left panels); GD2 and H antigens (middle panels); and H (right panels) are shown. Nuclei are stained blue (DAPI). Comparable results were obtained with MSCs from two additional donors.

Phenotyping of RBCs by flow cytometry

As a control for antigen detection by flow cytometry, blood group antigens on RBCs were confirmed using standard serological techniques (Fig S3 and S4).


The application of BM-derived MSCs, utilized with immunomodulating therapies and for repair of the musculoskeletal system, may provide a means for avoiding the use of immunosuppressive drugs that is often necessary with other cellular therapies. However, allogeneic MSCs can induce alloantibody production, as shown in an animal model (Beggs et al, 2006). Blood group antigens are potent activators of the host immune system and can lead to antibody-mediated destruction of not only RBCs, but also nucleated cells and organs (Colvin & Smith, 2005).

We investigated the expression of clinically important blood group antigens by MSCs at the levels of transcription, intracellular proteins and presence of antigens on the surface (Table II). MSCs were isolated and cultivated using pooled allogeneic human serum, employed in several studies evaluating the clinical utility of MSCs (Bieback et al, 2009, 2010; Turnovcova et al, 2009). Under Good Manufacturing Practice conditions, the ex vivo cultivation and expansion of MSCs should be free of animal serum, avoiding the immunization to fetal calf serum (FCS) known to occur in patients receiving MSCs cultured in FCS-containing media (Sundin et al, 2007).

Table II.   Summary of mRNA and protein expression by clinically important blood group genes in human bone marrow-derived mesenchymal stem cells (MSCs).
Blood group systemParameter of MSCs
GeneMajor antigensTranscriptome mRNA in cellsProteome/antigen on cell surface
ABOA and B
RHCEC, E, c, and e+
KELK and k+
SLC14A1 (JK)Jka and Jkb+
DARCFya and Fyb+

Blood group status can be determined at the genomic level, a technology increasingly performed in the clinical setting using blood group genotyping platforms. Our serological and genotyping results were equivalent indicating blood group genotyping is reliable. Because most antigens are not expressed on MSCs, blood group genotyping is currently not needed to characterize undifferentiated BM–MSCs for clinical use. While we detected mRNA transcription of the FUT1, RHCE, KEL, SLC14A1 (JK) and DARC genes, no ABO, RHD or RHAG mRNA occurred in MSCs. mRNA transcription varied among subpopulations of MSCs that differentiated into adipogenic, osteogenic and chondrogenic lines.

Trypsin affects the detection of few clinically relevant blood group proteins. However, Rhesus, Kell, Kidd and Duffy, and the carbohydrate-based antigens are known to be resistant to trypsin treatment. Therefore, we used trypsin to detach the MSCs from the culture flasks in our study.

MSCs did not express ABO mRNA and consequently neither A nor B antigen was detectable on their surface. Both results indicate that ABO compatibility is not required for the human serum used in MSC culture. If supplies of AB serum become short with anticipated increases in clinical stem cell culture and its applications increase as anticipated, it can be replaced by serum of other blood groups. In addition, because the soluble A and B antigens occurring in AB serum were also not adsorbed onto MSCs, use of AB serum for culture is unlikely to interfere with the ABO compatibility of MSCs in vivo.

The key enzyme for synthesizing the H antigen is the β-d-galactoside 2-α-l-fucosyltransferase (FUT1), which is in turn a prerequisite for the generation of the A and B antigens. FUT1 and H antigen are expressed by KG-1a cells and immature CD34+ haematopoietic progenitor cells but absent in mature lymphocytes (Cao et al, 2001). We detected the H antigen on a GD2dim+ subpopulation of undifferentiated MSCs, whereas no FUT1 mRNA was found in differentiated MSCs. H is a strong sugar-based antigen and avid poly- and monoclonal antibodies are available, to facilitate the identification as well as the purification and clinical testing of the newly defined GD2dim+ H+ and GD2bright+ H cells occurring among undifferentiated MSCs (Fig 4). Based on this technical feature, the H antigen may become instrumental for the identification of immature MSCs, which may differ in ‘stemness’ from the remaining MSCs in the BM. FUT1 is also involved in the biosynthesis of Globo H, a potential tumour-associated antigen in human breast cancer stem cells (Chang et al, 2008).

Among the three Rh-associated ammonia transporters, RhBG and RhCG are expressed on many nucleated cells, but not on RBCs (Handlogten et al, 2005). The third, RhAG, recently recognized as blood group system no. 30, is found on RBC and required for membrane integration of the two proper Rhesus proteins, RhD and RhCE (Huang, 1998; Mouro-Chanteloup et al, 2002; Daniels et al, 2009). Because RHAG is not transcribed in MSCs, the lack of the RhCE protein on the surface, despite some RHCE transcription, can be explained by the known posttranslational dependence of RhCE on RhAG. Considering the erythroid-restricted expression of RhAG (Zidi-Yahiaoui et al, 2005), our data confirm the non-erythroid character of MSCs.

Prichett et al (2000) detected the human urea transporter HUT11 (SLC14A1; JK), the protein carrying the Jk antigen, in ‘explant cultures of human bone’ containing osteoblasts and an unknown fraction of MSCs. We confirmed a reduced SLC14A1 (JK) transcription in differentiating MSCs. However, there was no good correlation with PPARG transcription, which is a marker for adipogenic differentiation (Wu et al, 2010): SLC14A1 (JK) was least transcribed in chondrogenic MSCs, while PPARG transcription in chondrogenic MSCs equaled undifferentiated MSCs. In conclusion, PPARG and SLC14A1 (JK) seem to be regulated independently in MSCs. DARC transcription is increased under adipogenic and apparently osteogenic differentiation. It occurs however in such low copy numbers that, not surprisingly, no DARC protein was found on the MSC surface. The Kell blood group protein, a zinc endopeptidase, is expressed on erythroid and many non-erythroid cells (Russo et al, 2000). We found some KEL transcription in MSCs but no KEL protein on the MSC surfaces, which is consistent with the immature stem cell character of MSCs.

Allogeneic MSCs can be immunogenic and contribute to rejection of haematopoietic stem cells, inducing a memory T-cell response in the host, as shown in a non-myeloablative murine transplantation setting (Nauta et al, 2006). Interferon-γ induced human leucocyte antigen (HLA) class II antigen expression may contribute to alloimmunization by MSCs (Chan et al, 2006), which may warrant further investigation. However, we could not detect such HLA class II antigen expression on undifferentiated MSCs during our culture with human serum. Among the clinically most relevant and potentially immunizing blood group proteins, only the SLC14A1 (JK) protein could be detected in the cytoplasm, but notably not on the surface of MSCs. Our data suggest that, even in the non-immunosuppressed graft recipient, a rejection or malfunction of the cellular graft due to an immunization against these blood group proteins and their antigens may not be expected. This lack of immunogenic blood group antigens on MSCs is a plus for promising clinical applications.

In further characterizing human MSCs, the data presented here suggests not only practical aspects of current GMP production, but also compatibility of these cells in the clinical setting of stem cell transplantation and regenerative medicine.


We gratefully acknowledge Martina Wölfle, née Maisel, Department of Neurodegenerative Diseases, Hertie-Institute for Clinical Brain Research, University of Tübingen for the RHAG primer design; Rainer Kehlbach, Department of Diagnostic Radiology, University Hospital Tübingen for discussing flow cytometry data; Elizabeth J. Furlong, Dept. Transfusion Medicine, Clinical Center, NIH for excellent editing of the manuscript; and the medical technologists at IKET, University Hospital Tübingen for general technical support.

Disclosure of conflict of interest

The authors declare no competing interests relevant to this article.

Authorship contributions

R.S. initiated the study and supervised the performance of the experiments; R.S., Ma.S. and W.A.F. provided the study concept and design; R.S., R.A.K., G.S. and Mi.S. collected samples; T.K. provided study material; C.G. designed primers; R.S., G.S., M.A.R., U.H.-K., M.A. and M.B. collected data; R.S., T.K., Mi.S. assembled data; L.D. performed statistical analyses; R.S., Ma.S., R.A.K., G.S., C.G. and W.A.F. interpreted data; R.S., H.N. and W.A.F. discussed data; C.G., L.D. and H.N. contributed to manuscript writing; R.S. wrote drafts and W.A.F. the final version of the manuscript.


The views expressed do not necessarily represent the view of the National Institutes of Health, the Department of Health and Human Services, or the US Federal Government.