Selection of Stem Cells by Using Antibodies That Target Different CD34 Epitopes Yields Different Patterns of T-Cell Differentiation


  • Mario Otto,

    1. Division of Stem Cell Transplantation, Department of Hematology/Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee, USA
    Search for more papers by this author
  • Xiaohua Chen,

    1. Division of Stem Cell Transplantation, Department of Hematology/Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee, USA
    Search for more papers by this author
  • William J. Martin,

    1. Integrated Research Center, St. Jude Children's Research Hospital, Memphis, Tennessee, USA
    Search for more papers by this author
  • Wing Leung,

    1. Division of Stem Cell Transplantation, Department of Hematology/Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee, USA
    Search for more papers by this author
  • James Knowles,

    1. Division of Stem Cell Transplantation, Department of Hematology/Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee, USA
    Search for more papers by this author
  • Marti Holladay,

    1. Division of Stem Cell Transplantation, Department of Hematology/Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee, USA
    Search for more papers by this author
  • Jim Houston,

    1. Division of Stem Cell Transplantation, Department of Hematology/Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee, USA
    Search for more papers by this author
  • Rupert Handgretinger,

    1. Department of Pediatrics, University of Tübingen, Tübingen, Germany
    Search for more papers by this author
  • Raymond C. Barfield M.D., Ph.D.

    Corresponding author
    1. Division of Stem Cell Transplantation, Department of Hematology/Oncology, St. Jude Children's Research Hospital, Memphis, Tennessee, USA
    • Division of Stem Cell Transplantation, Department of Hematology/Oncology, St. Jude Children's Research Hospital, 332N Lauderdale Street, Memphis, Tennessee 38105-2794, USA. Telephone: 901-495-4720; Fax: 901-521-9005
    Search for more papers by this author


The objective of this study was to compare the patterns of T-cell differentiation from CD34+ human stem cells selected with different classes of antibody targeting the CD34 molecule. We compared signal-joint T-cell receptor excision circle (sjTREC) production in thymocytes selected with different classes of anti-CD34 antibody. Based on these results, we studied immune reconstitution in nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice using human stem cells selected with the same antibodies that yielded variation in the thymocytes. Human CD34+ stem cells were immunomagnetically selected using the class II QBEnd antibody (prevalent in clinical graft engineering) and the class III 8G12 antibody (common in diagnostic tests). Engraftment and T-cell reconstitution were examined after transplantation. Thymocytes selected with the 8G12 class III antibody have a higher TREC production than those selected with the QBEnd class II antibody. Of mice transplanted with cells selected using the 8G12 antibody, 50% had sjTREC production, compared with 14% of mice transplanted with cells selected using the clinically common antibody QBEnd. 8G12 thymic progenitors are characterized by higher quality in thymic distribution and higher activity in T-cell differentiation. Using class III antibody targeting the CD34 molecule resulted in increased T-cell reconstitution in the NOD/SCID mouse. Use of a single antibody epitope targeting the CD34 molecule may lead to loss of cells that might provide richer T-cell reconstitution. Use of different or multiple epitopes, targeting of alternate stem cell markers, or use of cell-depletion strategies might prevent this loss.


The CD34 molecule has long been recognized as a marker for hematopoietic stem cells (HSCs) and has been the primary target used to isolate these cells for clinical hematopoietic stem cell transplantation (HSCT). The complex CD34 antigen has been classified into three groups on the basis of the epitopes bound by monoclonal antibodies to CD34 glycophosphoprotein and the differential sensitivity of the CD34 epitopes to enzymatic cleavage by neuraminidase, chymopapain, and a glycoprotease from Pasteurella hemolytica [1].

In clinical transplantation, the QBEnd antibody to class II CD34 antigen has been used most commonly to isolate HSCs. However, other classes of anti-CD34 antibodies used in diagnostics identify populations of cells that are CD34+ but are not identical to those selected using the QBEnd antibody [2, [3], [4]–5]. Several groups have found different distributions of CD34 epitopes on HSCs. Class III epitopes have a broader distribution on both normal HSC and leukemic blast cells than do class I and II epitopes [2, 6]. It was recently reported that high numbers of CD34+ cells in human thymus positively correlate with the thymic activity [7]. Because rapid T-cell reconstitution after stem cell transplantation is a crucial step for the success of this therapy modality, we were intrigued by the possibility that the use of QBEnd-CD34 antibody for stem cell enrichment might exclude important CD34+ subpopulations, especially CD34+ T-cell precursors that could otherwise be identified and selected by using a different antibody, such as 8G12-CD34.

In this study, we began by looking at human thymocytes selected with antibodies targeting several different classes of CD34 antigen. Based upon these observations, we asked whether engraftment and reconstitution patterns would differ in nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice that received grafts of human HSCs selected with QBEnd antibodies versus 8G12 antibodies. If engraftment, especially T-cell reconstitution, differs when different epitopes are targeted to select CD34+ cells, the dominant use of the QBEnd epitope in the clinical setting should perhaps be re-evaluated: a single antibody might not only reduce the total number of stem cells collected but also miss subtypes of progenitor cells with potential for speeding immune recovery. To our knowledge, there have been no studies assessing engraftment with different classes of CD34+ stem cells transplanted into experimental animals.

Materials and Methods

Acquisition and Processing of Thymocytes

Whole human thymuses discarded in the normal course of cardiac surgery on children 2 months to 10 years of age were obtained with Institutional Review Board approval (kindly provided by the Department of Surgery, LeBonheur Children's Medical Center, Memphis, TN). The tissue was placed in sterile RPMI-1640 medium (Invitrogen Corporation, Carlsbad, CA,, cut into small pieces, and further processed by using a CELLECTOR Tissue Sieve System (E-C Apparatus Corporation, Holbrook, NY). The cell-rich suspension was strained through 40-μm filters. The resulting product was separated by centrifugation on a Ficoll density gradient (Ficoll-PaquePlus; GE Healthcare, Little Chalfont, Buckinghamshire, U.K., The mononuclear cell layer was removed, washed twice with phosphate-buffered saline (PBS) (Cambrex Bio Science Walkersville, Inc., Walkersville, MD,, and divided into two aliquots. For QBEnd selection, cells were labeled with anti-CD34 microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany, and subsequently positively enriched on an AutoMACS device (Miltenyi Biotec) in accordance with the manufacturer's instructions. For 8G12 selection, mononuclear cells were incubated for 20 minutes with 4 μl of phycoerythrin (PE)-conjugated anti-CD34 antibody (BD Biosciences, San Jose, CA, per 1 × 106 cells at a final concentration of 1 × 108 per milliliter in labeling buffer containing PBS (Cambrex Bio Science Walkersville, Inc.) supplemented with 0.5% human serum albumin (Bayer Corp., Emeryville, CA, and 2 mM EDTA (Cambrex Bio Science Rockland, Inc., Rockland, ME, After the incubation, the cell suspension was washed twice with labeling buffer, incubated with anti-PE microbeads (Miltenyi Biotec), and subsequently positively enriched on the AutoMACS device in accordance with the manufacturer's instructions. For some experiments, cells were further sorted by flow cytometry on a FACSVantage DiVa (BD Biosciences) to obtain the purest T-cell subset samples attainable by current methods.

Detection and Quantification of Human Signal-Joint T-Cell Receptor Excision Circle in Human and Mouse Thymus

CD34+ thymocytes from six donors were enriched with either anti-CD34 (8G12) or QBEnd antibody, as described above, and tested for signal-joint T-cell receptor excision circle (sjTREC) production. Isolated CD34+ thymocytes from two donors were further sorted into CD4+ and CD4 subpopulations on a flow cytometric sorting device (FACS Vantage DiVa; BD Biosciences) and subjected to sjTREC analysis. The sjTREC levels were quantified by real-time polymerase chain reaction (PCR) as described previously [8, 9]. In brief, the DNA was purified by using the QIAmp Blood Kit (Qiagen Inc., Valencia, CA, The real-time PCR was performed in a volume of 25 μl, with a final concentration of 1× Taqman Universal PCR Master Mix (Applied Biosystems, Foster City, CA,, 250 nM forward primer and reverse primer, 200 nM probe, and 100 ng of DNA. The real-time PCR was performed on an ABI 7700 instrument (Applied Biosystems). All of the samples were measured in duplicate PCR reactions. The Cα constant region was used as an internal control and to normalize for input DNA. The standard was created by cloning the sjTREC fragment in PCR 2.1-TOPO vector using the TOPO TA Cloning Kit (Invitrogen Corporation). The standard curve and TREC values were analyzed by using ABI 7700 software. The number of TREC molecules in the sample was calculated as copies per 105 cells.

Mobilization and Collection of Peripheral Blood Stem Cells

Peripheral blood stem cells were mobilized in five healthy volunteer donors by administering granulocyte cell-stimulating factor at 10 μg/kg per dose (maximum dose 480 μg/day) (Neupogen; Amgen, Thousand Oaks, CA, A single leukapheresis was performed on day 5 by using a Cobe Spectra instrument (Cobe, Lakewood, CO, All donors gave written informed consent, and the study was approved by the St. Jude Children's Research Hospital Institutional Review Board.

Stem Cell Enrichment from Leukapheresis Product

To most efficiently eliminate mature T and B cells from the graft, we performed large-scale T- and B-cell depletion before stem cell selection as previously described by our group [10]. After T- and B-cell depletion, the mean percentage of residual T cells was 0.02% (range 0.01%–0.04%), with a mean log10 depletion of 3.4 (range 3–3.8). The mean percentage of contaminating B cells was 0.1% (range 0.01%–0.4%), with a mean log10 depletion of 2.2 (range 1.4–3). 8G12 and QBEnd CD34+ progenitor cells were subsequently isolated using the procedure described for the thymic CD34+ progenitors.

Flow Cytometry

Flow cytometric analysis and data analysis were performed on a BD LSR flow cytometer (BD Biosciences). Expression of human leukocyte surface markers was determined using antibodies labeled with fluorescein isothiocyanate, PE, PE-cyanine 7 (PE-Cy7), PE-Cy5, allophycocyanin, or PE-Texas Red targeting the following antigens: CD3, CD4, CD13, CD19, and CD34-QBEnd10 (Dako North America, Inc., Carpinteria, CA,; CD20, CD34-581, and CD45-streptavidin (Beckman Coulter, Fullerton, CA,; pan-T-cell receptor (TCR)-γδ and CD34-8G12 (BD Biosciences); CD7 pan-TCR-αβ (Caltag Laboratories, now part of Invitrogen Corporation); CD34-AC136 (Miltenyi Biotec); and CD34-12.8 (Baxter, Deerfield, IL, We stained for the mouse leukocyte common antigen with anti-mouse CD45 (BD Biosciences). Cells were stained according to the manufacturers' protocols. When necessary, red blood cells were lysed with 1× ammonium chloride solution. Viability was assessed by propidium iodide staining (Roche Diagnostics, Manheim, Germany,

Transplantation of NOD/SCID Mice

NOD/SCID mice (NOD/LtSz-Prkdcscid/J) cared for according to institutional guidelines were sublethally irradiated with 325 cGy by using a 137Cs source (J L Shepherd and Associates, San Fernando, CA, Two groups of mice received transplants of 5 × 105 HSC selected with either QBEnd (n = 69) or 8G12 (n = 44). In addition to the mice transplanted with the stem cell groups of interest, we transplanted mice with nonselected cells from the T/B-depleted product, targeting the same number of CD34 stem cells in the grafts in order to compare engraftment and sjTREC production (n = 47). The cells were injected into the lateral tail vein. Half of the mice were humanely sacrificed after 10–12 weeks and half after 6 months. Bone marrow was harvested by flushing the femora and tibiae. Spleen and thymus tissues were processed to obtain single-cell suspensions.

Statistical Analyses

For statistical evaluation, the paired Student's t test was used. A p value less than .05 was considered to indicate statistical significance.


Distribution of Class II and Class III Epitopes of CD34 Antigen in Human Thymus

To understand the distribution of CD34 class III epitope in thymus, we collected 13 thymuses and examined the proportion of class III and class II epitopes on thymocytes by flow cytometry. We found that both QBEnd-class II and 8G12-class III epitopes are expressed on thymocytes. The average percentage for 8G12-class III population was higher (p = .0015) than for QBEnd-class II population. (Table 1; Fig. 1).

Table Table 1.. Flow cytometric analysis of thymocytes from 13 different human thymuses
original image
Figure Figure 1..

Variations in expression of CD34+ antigen in human thymocytes. The x-axis shows fluorescence density of CD7. The y-axis represents the fluorescence density for each CD34 antibody, including (A) CD34-class III-12.8, (B) CD34-class II-QBEnd, (C) CD34-class III-8G12, and (D) CD34-class III-581. Abbreviations: APC, allophycocyanin; Cy5, cyanine 5; ECD, phycoerythrin-Texas Red; FITC, fluorescein isothiocyanate; SA ECD, streptavidin phycoerythrin-Texas Red.

sjTREC Production in Class II or Class III Progenitor Cells in Thymus

To test T-cell differentiation of class II and class III progenitor cells in thymus, we examined sjTREC production in these two populations of thymocytes. The mean sjTREC production in 8G12-class III thymocytes was 1,161 ± 774 (± SD) and was 514 ± 297 in QBEnd-class II thymocytes (p = .028; Table 2). In two thymuses, we separated the thymocytes into CD34+CD4+ and CD34+CD4 in each CD34 epitope. The sjTREC level was very low in both CD34+CD4 populations. However, sjTREC production in 8G12-CD34+CD4+ cells was 2.5-fold of that in QBEnd-CD34+CD4+ cells in both donors (Table 3).

Table Table 2.. sjTREC production in thymocytes
original image
Table Table 3.. sjTREC production in thymocytes sorted by flow cytometry
original image

Engraftment of Human CD34+ Cells in NOD/SCID Mice

To estimate the engraftment of human CD34 cells in the mice, we used anti-human CD45 antibody to recognize human CD34 cells and lymphocytes in the mice at 10 weeks, 3 months, and 6 months after the transplantation. We collected the cells from three locations, including spleen, bone marrow, and residual thymus. We found no statistically significant difference in human CD45 engraftment from any compartment (Fig. 2, showing human CD45 engraftment from bone marrow). Mean engraftment for the CD34 QBEnd group was 18% at 3 months (n = 30) and 5.4% at 6 months (n = 28). Mean engraftment for the CD34 8G12 group was 9.7% at 3 months (n = 29) and 3.66% at 6 months (n = 18). Mean engraftment for the T/B-depleted group with no stem cell enrichment was 32% at 3 months (n = 20) and 24% at 6 months (n = 6). Using flow cytometry, we found no differences in human CD19+ B-cell development or in human CD56+ natural killer (NK) cell engraftment (data not shown). There were no mature T cells detectable by flow cytometry using anti-CD3 antibodies.

Figure Figure 2..

Human CD45+ engraftment in the bone marrow of nonobese diabetic/severe combined immunodeficient mice. The y-axis shows the percentage of human CD45+ cells in the mice transplanted with T/B dep cells, and cells selected with either the QBEnd antibody or the 8G12 antibody at 3 and 6 months after transplantation are described on the x-axis. • represents the samples tested. The mean and median CD45 percentages for each group are represented as thick bars or thin bars, respectively. Abbreviations: mo, months; T/B dep, T/B-depleted.

Human sjTREC Production in Mice

SjTREC is an episomal excision product formed at the point of TCR-δ deletion in the early stages of TCR-αβ formation in thymus. Because the environment of mice thymus may not be suitable for the formation of complete human TCR-CD3 complex (targeted by the antibodies used in flow cytometry to detect CD3+ T cells), we used sjTREC as a marker for detection of the early stages of human T-cell formation. We examined human sjTREC production in mouse bone marrow, spleen, and thymus at 10 weeks, 3 months, and 6 months post-transplantation. There was no detectable sjTREC in the spleen and bone marrow at any time points for the mice transplanted either by 8G12 or QBEnd progenitor cells. There was also no detectable sjTREC in their thymus at 10 weeks after the graft. However, at 3 months, 3 of 27 (11%) mice from both groups had sjTREC detectable in the thymus. At 6 months, 5 of 10 mice (50%) in the 8G12 group and 3 of 22 (14%) in the QBEnd group had detectable sjTREC in the thymus. In the mice transplanted with grafts that were T- and B-cell-depleted, but not stem cell-selected, TRECs were detected as early as 10 weeks in 3 of 11 mice (27%). The production declined over time to 10% by 6 months (Table 4).

Table Table 4.. Signal-joint T-cell receptor excision circle production in the thymuses of nonobese diabetic/severe combined immunodeficient mice
original image


Since the 1980s, the CD34 molecule has become very important in clinical stem cell transplantation [11, [12], [13], [14]–15]. Although the function of CD34 has largely remained elusive, CD34+ stem cells have the proven ability to fully reconstitute the human immune system in the setting of clinical transplantation [11]. This molecule is highly glycosylated, and over the past decades, various antibodies have been generated which have been shown to target different epitopes of the molecule. Interestingly, there seems to be only partial cross-reactivity, and some of the antibodies do not react with certain CD34+ subpopulations that antibodies targeting other epitopes do recognize [2, 6, 16].

We began this study because of our interest in improving thymus-dependent T-cell reconstitution in patients transplanted with CD34-selected stem cell grafts. When immunomagnetic beads are used for cell selection, the antigen specificity of the antibody that is used will affect cell selection. Given this and given the variation in the CD34 molecule, we looked at human thymocytes and evaluated the CD34 expression using different classes of anti-CD34 antibodies, namely 8G12 (class III) and QBEnd (class II). We found a clear and statistically significant difference in the ability of 8G12 antibody to identify CD34 expression in human thymocytes in comparison with QBEnd (p = .001; Table 1). This finding suggests that the 8G12 population may contain quantitatively more progenitors in thymus.

Sovalat et al. reported that the CD34hiCD38dim population, which represents very immature progenitors, seems to be preferentially recognized by the class III antibody 8G12 [16]. Notch-1, signaling a critical differentiation pathway toward T-cell differentiation at the expense of B-cell differentiation, is also highly expressed in CD34hiCD38dim cells. In addition, it is the Notch ligand Delta-1 that has been shown specifically to promote the generation of thymus repopulating T-cell precursors in vitro and in vivo [17]. Indeed, when we isolated human CD34+ thymocytes using QBEnd and 8G12 and examined sjTREC production as a marker for TCR rearrangement and T-cell differentiation in the two different populations, we found a significant difference. Mean sjTREC production in the 8G12 group was approximately twofold higher than in the QBEnd group (p = .028; Table 2). However, because CD34 is expressed in a heterogeneous population of cells in different early stages of lineage commitment and differentiation, we wanted to ascertain whether the populations we were examining in the thymus truly were T-cell progenitors. Therefore, we isolated CD34+ cells from two thymuses with either QBEnd or 8G12 and sorted these populations for positivity of CD4, a very early surface marker for distinct T-cell differentiation [18, 19]. Both 8G12- and QBEnd-isolated CD34+CD4 cells exhibited either no detectable or extremely low numbers of sjTREC copies, whereas the CD34+CD4+ populations, as expected, clearly showed sjTREC production (Table 3), indicating that the isolated CD34+ cells indeed contained early T-cell progenitors. 8G12-selected T-cell progenitors have higher activity toward T-cell differentiation.

Spits et al. have shown that early in development thymic progenitor cells have the highest expression of CD34 and lack CD1a [18, 20]. An important checkpoint for T-cell development is marked by the appearance of CD1a on CD34+ thymocytes. CD34 downregulates in the subsequent population characterized by the presence of CD4. These cells, known as CD4 immature single positive (ISP) cells, lack CD8 and CD3. TCR-β gene rearrangements are more abundant in the CD4 ISP cells than in the preceding CD1a+CD4CD34+ population. No germline TCR-Á genes are observed in the CD4 ISP cells (TCR-Á gene deletion), and clearly, these cells are fully committed to the T-cell lineage. Further development is characterized by disappearance of CD34 and coexpression of CD4/CD8-α and then CD4/CD8-α/CD8-β. The findings in this study are in agreement: the sjTREC production in our study was found predominantly in the CD34+CD4+CD3 thymocytes, a population approximating the CD4 ISP cells with TCR-Á depletion. Both 8G12- and QBEnd-isolated CD34+CD4 cells exhibited either no detectable or extremely low numbers of sjTREC copies (Table 3). These findings suggest a difference between 8G12 and QBEnd progenitor cells in αβ T-cell commitment at the CD4 ISP stage.

The severe immunosuppression that is encountered directly after stem cell transplantation leads to dangerous susceptibility to life-threatening infections, so increasing the amount of thymus repopulating T-cell precursors in the graft is crucial. Hence, the difference we found between these two groups might have an impact on the recovering thymic output and, subsequently, transplant morbidity and mortality. Therefore, we asked whether the choice of antibody used to isolate human CD34+ stem cells from mobilized donors would lead to a difference in T-cell reconstitution in a mouse model.

The NOD/SCID mouse model has been widely used for immune reconstitution studies in the past. Although this model allows for evaluation of most lineages, xenografting of conventional NOD/SCID strains solely with human CD34+ stem cells does not lead to measurable peripheral CD3+ T-cell reconstitution when assessed by flow cytometric methods. This may be due to remaining mouse NK activity or the lack of an appropriate cellular and cytokine environment to support CD3 assembly and surface expression [21]. Only very recently, models have been described that allow some T-cell development to the developmental stage where CD3 surface expression on matured T cells can be assessed [22, 23]. Thus, gene-targeting experiments have been instrumental in the identification of crucial components of the pre-TCR assembly [24]. Thymic sjTREC content has been shown to reflect thymus-dependent T-cell reconstitution after HSCT in the immunocompromised mouse and therefore was used for the assessment and comparison of T-cell engraftment and early differentiation in our setting [25]. We evaluated the engraftment and differentiation at different time points after transplantation. There were no differences in total human CD45+ engraftment among the groups (Fig. 2). The same was true for the spleen and the thymus. As expected, we detected no CD3+ cells by flow cytometry. However, more 8G12-transplanted animals showed thymic sjTREC production than QBEnd-transplanted mice 6 months after transplant. sjTREC was not detected in spleen or bone marrow in either group, suggesting that the murine thymus is capable of allowing immigration of very early human T-cell progenitors and able to support some first, albeit crucial, steps in human T-cell differentiation and TCR rearrangement.

As shown earlier, 8G12 in general detected more CD34+ cells in the thymus than did QBEnd. Thus, 8G12 + thymocytes might simply contain quantitatively more T-cell progenitors. On the other hand, it might contain stem cells with a faster output of T-cell lineage progeny. Little is known about the function of the CD34 molecule. The different epitopes on the molecule might also lead to transmembrane cell signaling favoring a specific lineage differentiation over another. However, we did not address these questions and thus this remains speculation.


We have shown that the anti-CD34 class III antibody 8G12, which is used for diagnostic purposes, detects a significantly higher amount of CD34+ cells in human thymus than does QBEnd, an antibody used for clinical enrichment of HSC used in transplantation. 8G12-isolated CD34+ thymocytes show significantly higher activity of T-cell differentiation at the developmental stage approximating CD4 ISP. 8G12-selected and -transplanted human CD34+ HSC also lead to a higher thymus-dependent T-cell reconstitution in an NOD/SCID mouse model. Further studies to support our data are warranted. Such information may help the clinician to rethink strategies in the setting of stem cell transplantation to overcome the life-threatening complications of infection and relapse that result in part from T-cell deficiency after transplantation.


The authors indicate no potential conflicts of interest.


This work was supported by Cancer Center Support Grant CA 21765 and research grants from American Lebanese Syrian Associated Charities and the Assisi Foundation of Memphis. M.O. and X.C. contributed equally to the study.