TISSUE-SPECIFIC STEM CELLS

TISSUE-SPECIFIC STEM CELLS Functional Characterization of TPO-Expanded CD34 1 Cord Blood Cells Identifies CD34 2 CD61 2 Cells as Platelet-Producing Cells Early After Transplantation in NOD/SCID Mice and rcd34 1 Cells as CAFC Colony-Forming Cells LAURUS F. SCHIPPER, a ANNEKE BRAND, a,b WILLEM E. FIBBE, b YVETTE VAN HENSBERGEN a a Sanquin Blood Supply Foundation, Division of Research, Department of Transfusion Medicine, Leiden, The Netherlands; b Leiden University Medical Center, Department of Immunohematology and Blood Transfusion, Leiden, The Netherlands Key Words. Cord blood Expansion Thrombopoietin Megakaryocyte CD34 1 ABSTRACT Transplantation of thrombopoietin (TPO)-expanded cord blood CD34 1 cells accelerates human platelet recovery in NOD/SCID mice. It is unknown which subpopulations of the TPO-expanded cells mediate accelerated platelet recovery and bone marrow (BM) engraftment. In this study, the contribution of these subpopulations to human platelet appearance in the blood and BM engraftment was studied in NOD/SCID mice. Following transplantation of CD34 2 / CD61 2 /lineage 2 cells (Lin 2 ), human platelets were detected in the blood of recipient mice from day 4. Both time to platelet recovery and blood platelet counts at 6 weeks after transplantation showed Lin 2 dose dependence. The Lin 2 population was virtually negative for lineage marker expression and lacked CD42b expression but was heterogeneous with regard to CD36 and CD38 expression, reflecting a population in transit but not fully committed toward the megakaryocyte (MK) lineage. Although no definitive pheno- Disclosure of potential conflicts of interest is found at the end of this article. type could be established of the cells generating prompt platelet production and cells generating platelets 6 weeks after transplantation, this relatively heterogeneous Lin 2 population is prerequisite to accelerate platelet recovery in vivo. The interval to platelet recovery after transplantation of the CD34 1 cells remaining after expansion (rcd34 1 )wassimilar to mice transplanted with nonexpanded CD34 1 cells, although the total platelet counts and the engraftment levels in the BM were lower. Cobblestone area-forming cell colony-forming cells resided mostly in the rcd34 1 population. The pro-mk CD61 1 cells did not contribute to human platelet recovery or engraftment in the BM. Our study shows that not all expanded cells appear critical for transplantation. These data support that functional characterization of the expanded cell populations is warranted to make future expansion protocols suitable for clinical application. STEM CELLS 2012;30:988 996 INTRODUCTION Umbilical cord blood (UCB) is an alternative hematopoietic stem cell source that is increasingly used for stem cell transplantation. A major disadvantage of transplantation with UCB is the delayed engraftment, in particular time to platelet recovery, that is related to the small number of stem and progenitor cells transplanted. Several approaches have been investigated or are used to overcome this drawback. Double UCB transplantation is recently implemented in clinical research settings, but also the role and use of accessory stromal or expanded mesenchymal cells is under clinical investigation [1 6]. Ex vivo expansion of progenitor and stem cells has also been intensively studied [7, 8]. Central concern with ex vivo expansion is the risk of exhaustion of true stem cells, which may compromise long-term engraftment. Thus far, transplantation in patients has mainly been performed with a combination of expanded cells and nonexpanded cells, and it provided insufficient information on the role of the expanded cells in hematopoietic repopulation [9, 10]. Recently, a study by Delaney et al. showed rapid although temporary engraftment of Notch-mediated expanded human UCB progenitor cells in patients, confirming loss of true stem cell capacity after expansion in most cases [11]. Platelet recovery after UCB transplantation is consistently delayed. In previous studies, others and we have shown that ex vivo expansion focused on partial differentiation toward megakaryocyte (MK) progenitor cells improves platelet recovery [12 14]. In contrast to most expansion studies that use combinations of cytokine cocktails, our strategy was to expand the UCB CD34 þ cells with modest proliferation and to differentiate toward predominantly the megakaryocytic Author contributions: L.F.S. and Y.v.H.: designed research, performed research, analyzed the data, and wrote the paper; A.B. and W.E.F.: wrote the paper. Correspondence: Yvette Van Hensbergen, Ph.D., Sanquin Blood Supply Foundation, Division of Research, Department of Transfusion Medicine, Plesmanlaan 1A, 2333 BZ, Leiden, The Netherlands. Telephone: þ31-71-5685608; Fax: þ31-71-5685191; e-mail: y.vanhensbergen@sanquin. nl Received June 15, 2011; accepted for publication February 6, 2012; first published online in STEM CELLS EXPRESS February 29, 2012. VC AlphaMed Press 1066-5099/2012/$30.00/0 doi: 10.1002/stem.1071 STEM CELLS 2012;30:988 996 www.stemcells.com

Schipper, Brand, Fibbe et al. 989 lineage, thus limit multilineage differentiation. Indeed, with CD34 þ UCB cells cultured for 10 days with thrombopoietin (TPO) as a single growth factor, we could achieve accelerated recovery of circulating human platelets in the blood of NOD/ SCID mice without loosing engraftment and multilineage hematopoiesis in the bone marrow (BM) [13]. These results were confirmed by Mattia et al., who showed improved thrombopoiesis and overall engraftment in NOD/SCID mice with cells expanded with TPO alone, in contrast to cells expanded with cytokine cocktails [14]. To date, most culture protocols are designed to reach optimal expansion numbers of the stem cells, whereas less effort is taken to investigate the contribution of the individual cell populations present in the expanded graft to hematopoietic recovery as well as to engraftment. Performing such functional studies is important with regard to optimizing the expansion protocols for clinical use. Limited information is available on the subpopulations that are important for accelerated platelet recovery. Although a correlation between time to platelet recovery and the number of CD34 þ /CD61 þ cells in the graft was observed in patients transplanted with mobilized peripheral blood stem cells [15, 16], it is not known how these results relate to UCB transplantation, since UCB contains a small percentage of CD34 þ /CD61 þ cells and this subpopulation hardly develops during expansion with TPO [13, 17, 18]. TPO culture of UCB CD34 þ cells mainly generates CD34 / CD61 þ (CD61 þ ) cells and a substantial number of CD34 / CD61 /Lin cells (Lin ), only a small amount of CD34 þ cells remain present after culture (rcd34 þ ). Both the CD61 þ and the Lin population contain a large number of MK progenitor cells [17] and may play a role in the early human thrombopoiesis observed in the blood of NOD/SCID mice transplanted with TPO-expanded cells. In this study, we assessed the role of the subpopulations that are obtained after TPO culture of UCB CD34 þ cells on peripheral blood recovery of platelets and engraftment in the NOD/SCID mouse model. The results show that not the MK precursors (CD61 þ cells) but the Lin population is responsible for accelerated human thrombopoiesis, while the Lin and the rcd34 þ population are both able to give BM engraftment at 6 weeks after transplantation. In vitro experiments show that mainly the rcd34 þ cells contain cobblestone area-forming cell (CAFC) colony-forming cells. Thus, both Lin cells and rcd34 þ cells, representing a minority of TPO-expanded UCB CD34 þ cells, and not the CD61 þ MK precursors are required for rapid platelet recovery and permanent engraftment of the BM. MATERIALS AND METHODS Collection of UCB UCB was drawn into MacoPharma collection bags containing 25 ml citrate phosphate dextrose adenine-1 (MacoPharma, Utrecht, The Netherlands) after written informed consent from the mother. The protocol was approved by the Medical Ethical Committees of the participating hospitals. The blood was stored for maximal 24 hours at 4 C until processing. CD34 1 Cell Purification Mononuclear cells were isolated from UCB using ficoll density gradient (1.079 g/cm 3, Pharmacy LUMC, Leiden, The Netherlands). The CD34 þ cell population was isolated by magnetic cell separation using the direct CD34 þ progenitor cell isolation kit (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany). The purity of the CD34 þ cell population (always www.stemcells.com >90%) was verified by flow cytometry (Beckman Coulter, Woerden, The Netherlands). Expansion of the CD34 1 Cells Isolated CD34 þ cells were cultured for 10 days in medium with TPO (50 ng/ml mpl-ligand, kind gift from KIRIN Brewery Ltd., Tokyo, Japan) toward MKs as described before [13, 17]. Expansion was calculated by [total cell number culture day 10 total number of CD34 þ cells culture day 0]. Cells were analyzed by flow cytometry using mouse-anti-human CD45-fluorescein isothiocyanate (FITC), CD61-FITC, and CD34-phycoerythrin (PE) antibodies (all Beckman Coulter). Non-MK differentiation was determined with CD14-FITC and CD15-PE. The CD34 /CD61 (Lin ) subpopulation was further characterized with CD3, CD14, CD15, CD19, and CD33 antibodies for lineage differentiation and CD36, CD38, and CD42b antibodies for maturation (all Beckman Coulter). Purification of Subpopulations Derived from Cultured CD34 1 Cells Expanded cells obtained after 10-day cultures of CD34 þ cells were depleted for CD61 þ cells using flow cytometry cell sorting or CD61 microbeads and depletion columns (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany); the method of selection of CD61 þ cells did not influence the in vivo results (data not shown). The sorted CD61 þ cells or the CD61 þ cells recovered from the depletion column were collected, resulting in a CD34 / CD61 þ population (CD61 þ ). Subsequently, the remaining CD34 þ cells were isolated using the direct CD34 þ progenitor cell isolation kit. The latter procedure resulted in the CD34 þ / CD61 (rcd34 þ ) and the CD34 /CD61 (Lin ) populations. The Lin population was subsequently depleted for CD3, CD14, CD15, CD19, and CD33 expressing cells (Fig. 1A). Transplantation in NOD/SCID Mice Female, 5 6 weeks old, NOD/SCID mice (Bomholtgard Breeding & Research Centre (Ry, Denmark) or Charles River (France)) were kept in microisolator cages in laminar flow racks in the animal facilities of the LUMC. Mice were given autoclaved acidified water containing 0.07 mg/ml Polymixin-B (Bufa B.V., Uitgeest, The Netherlands), 0.09 mg/ml ciproxin (Bayer B.V., Mijdrecht, The Netherlands), and 0.1 mg/ml Fungizone (Bristol-Myers-Squibb, Woerden, The Netherlands). The animal ethical committee of the LUMC approved all animal experiments. The animals were housed in the animal housing facilities for at least 1 week before the onset of the experiments. NOD/SCID mice were treated with total body irradiation (3.5 Gy), 4 48 hours before transplantation. Via tail injection, mice were transplanted with either 2 10 5 nonexpanded CD34 þ cells (CD34 þ control) or with all cells of the purified subpopulations (rcd34 þ cells, Lin,andCD61 þ ) obtained after expansion of 2 10 5 CD34 þ cells. In the first set of experiments, four mice per group were transplanted. Figure 1B shows the protocol followed for the experiment in which the nonexpanded and expanded cells from one UCB were compared. For this experiment, 2 10 5 CD34 þ cells were either transplanted nonexpanded or 2 10 5 CD34 þ cells were expanded for 10 days with TPO and all purified rcd34 þ cells and Lin cells obtained, respectively, were transplanted (n ¼ 3 mice per group). Human Platelet Detection in Peripheral Blood of NOD/SCID Mice Blood collection via tail incision was performed twice weekly during the first 3 weeks after transplantation and once weekly thereafter. Blood collection and human platelets measurements were performed as described previously [19]. Briefly,

990 Functional Characteristics of TPO-Cultured CB Figure 1. Study design for frozen CD34 þ cord blood cells. (A): Umbilical cord blood (UCB) CD34 þ cells were ex vivo expanded for 10 days with TPO and subsequently the three subpopulations were isolated. A representative example of the rcd34 þ, Lin, and CD61 þ subpopulations after isolation is shown. (B): CD34 þ cells were isolated from UCB and frozen in two populations. Sublethally irradiated non-obese diabetic/ severe combined immune deficiency mice were injected i.v. with either 2 10 5 cells (CD34 þ control cells) or the subpopulations (Lin or rcd34 þ cells) obtained after 10 days expansion with TPO (50 ng/ml) starting with 2 10 5 nonexpanded CD34 þ cells. human platelets were stained with a non-cross-reactive mouse-anti-humancd41-pe (Beckman Coulter), and erythrocytes were lysed with IQTest3 Lysing solution (Beckman Coulter). Flow-Count fluorospheres (Beckman Coulter) were added to enable the measurement of the absolute number of circulating human platelets. Analysis was performed with flow cytometry (Coulter EPICS XL-MCL, Beckman Coulter) running EXPO32 software. Analysis of BM Engraftment Six weeks after transplantation, mice were sacrificed and the BM was obtained from the femur by flushing in Iscove s modified Dulbecco s medium (IMDM). Evaluation of human cell engraftment and the relative lineage distribution of the engrafted human cells in the BM were performed with flow cytometry analysis. The BM-WBC cells were labeled with goat-antimouse-cd45-pe (leukocyte common antigen (LCA), Ly-5, 30- F11, Pharmingen, Alphen a/d Rijn, The Netherlands), mouseanti-human CD45-FITC, CD33-FITC, CD34-PE, CD19-PE (all from Beckman Coulter), and the appropriate isotype controls. Subsequently, erythrocytes were lysed with IO Test3 lysing solution (Beckman Coulter). Analysis was performed with flow cytometry (Coulter EPICS XL-MCL) with EXPO32 software (Beckman Coulter). Hematopoietic Progenitor Cell Cultures Human progenitor cell (colony-forming unit [CFU]-granulocytes, erythrocytes, and monocytes (GEM) assays and CFU-MK assays were performed as described by the manufacturer with the BM or the isolated subpopulations. Briefly, for hematopoietic progenitor culture (HPC)-GEM, 2.7 10 4 total BM-nucleated cells, 0.32 10 4 rcd34 þ or Lin cells, or 0.32 10 6 CD61 þ cells per 1.5 ml were added to Methocult (StemCell Technologies Inc., Grenoble, France) and cultured for 14 days at 37 C, 5% CO 2 in a humidified atmosphere (>95%) after which colonies were counted. The CFU-MK assay was performed in MegaCult- C medium with cytokines (StemCell Technologies Inc.) in a concentration of 2 10 5 total BM-nucleated cells, 0.25 10 4 rcd34 þ or Lin cells or CD61 þ cells per 3.3 ml. The number of colonies was counted at day 10 after culture. All samples were tested in duplicate for each assay. In earlier studies, we showed that the colonies were of human origin only [13]. For the CAFC assay, 2.7 10 4 NIH-3T3 cells per well (irradiated with 500 rad) were plated in a gelatin precoated 96- well plate. Twenty-four hours after plating, 5,000 rcd34 þ or Lin cells were seeded on top of the NIH-3T3 cells in CAFC medium (IMDM supplemented with 3.2% inactivated fetal calf serum (FCS) (Gibco, Breda, The Netherlands), 3.2% inactivated human serum (Sanquin), 2.3 mm glutamine (Gibco), 3 10 2 U/ml penicillin (Bio-Whittaker, Verviers, Belgium) and 3 10 2 lg/ml streptomycin (Bio-Whittaker), 7.2 10 3 mm hydrocortisone (Sigma Aldrich, Zwijndrecht, The Netherlands), and 7.2 mm b-mercapto-ethanol (Sigma Aldrich, Zwijndrecht, The Netherlands)). Cells were cultured at 37 C, 5% CO 2 in a humidified atmosphere (>95%), half of the medium was refreshed once weekly. Colony growth in all individual wells was scored at 5 weeks of culture. Statistical Analysis All results are presented as mean 6 SEM. Two-sided Fisher s exact test was used to calculate the statistical difference between the number of mice with human platelets in the groups. In all other situations, the statistical differences were calculated with a nonpaired alternate two-sided t test, assuming unequal SDs. For the multilineage engraftment, this test was performed with the mean, the SD, and the number of mice per cohort. Differences were considered significant when p <.05. RESULTS Expansion and Differentiation of UCB CD34 1 Cells with TPO After a culture period of 10 days with TPO, a mean expansion of 12.5 6 5.9-fold was observed. TPO-expansion of

Schipper, Brand, Fibbe et al. 991 Table 1. Surface marker expression of the sub-populations observed after thrombopoietin (TPO) expansion of CB CD34 1 cells (mean 6 SEM) Antibody rcd34 þ Lin CD61 þ CD36 þ 15.9 6 11.0 24.9 6 12.0 60.9 6 10.3 CD38 þ 95.3 6 0.0 74.8 6 8.6 21.1 6 5.0 CD42b þ 1.0 6 4.6 0.3 6 0.2 55.6 6 9.2 UCB CD34 þ cells gave rise to three main subpopulations, a large population of (pro-) MKs CD34 /CD61 þ cells (CD61 þ ) (78.4% 6 8.4%), an intermediate population of CD34 / CD61 /lineage cells (further referred to as Lin ) (13.2% 6 3.6%), and a small population of CD34 þ cells, hereafter referred to as remaining CD34 þ cells (rcd34 þ ) (5.1% 6 2.3%). CD61 expression highly correlated with CD41 expression, no CD41 þ /CD61 population was observed, and only a minor population of the cells was CD34 þ /CD61 þ (1% 6 1%). Contamination consisted of either CD14 þ or CD15 þ cells and was always low (<5%). No other hematopoietic cell lineage markers were found on the Lin cells (CD2, CD3, CD14, CD19, CD56, and glycophorin-a) and further characterization of the three subpopulations revealed a high percentage of more mature CD34 þ CD38 þ committed hematopoietic progenitor and precursor cells (Table 1). In the Lin and CD61 þ populations, the maturation continued, as illustrated by a gradual reduction of the CD38 expression. A population of CD36 þ cells was present in all populations after expansion (rcd34 þ, Lin, and CD61 þ ), but highest in the most expanded CD61 þ population differentiating toward MKs. The MK maturation marker CD42b was expressed on the CD61 þ cells only (55.6% 6 9.2%) (Table 1). Human Platelet Recovery in Blood of NOD/SCID Mice After Transplantation of the Subpopulations After transplantation of the purified subpopulations derived after 10 days expansion of CD34 þ cells with TPO, we analyzed the appearance of human platelets in the blood of recipient mice. Nonexpanded CD34 þ was used as control. For these experiments, cells were obtained from different UCB. As expected, following transplantation of nonexpanded CD34 þ cells (0.2 10 6 cells; 93% 6 11% purity), detectable platelets (more than one human platelet per microliter recipient mouse blood) were observed from day 12 onward (Fig. 2A, circle). Human platelet recovery following transplantation of rcd34 þ cells (0.15 10 6 cells; >95% purity) was similar to the platelet recovery observed with nonexpanded CD34 þ cells. Detectable platelets were observed at day 12 (Fig. 2A, triangle); however, at 6 weeks after transplantation, the platelet counts in the blood were lower as compared to mice transplanted with nonexpanded CD34 þ cells (39 6 27 platelet per microliter vs. 299 6 55 platelet per microliter; p ¼.01). After transplantation of the Lin cell population (2.6 10 6 cells; 86.3% 6 3.5% purity; contaminated with 2% rcd34 þ and 11.7% CD61 þ cells), human platelets were detected in the blood as early as day 4 after transplantation (Fig. 2A, diamonds). The human platelets persisted in the blood for at least 6 weeks after transplantation, with a transient drop in platelet count between day 20 and day 35. After transplantation of the CD61 þ population (10 10 6 cells; 96.7% 6 1.5% purity), human platelets were observed already at 1 hour after administration but in contrast to transplantation of the other subpopulations, the human platelets disappeared www.stemcells.com Figure 2. Platelet recovery in blood after transplantation of expanded subpopulations. (A): Platelet concentrations in the blood of non-obese diabetic/severe combined immune deficiency (NOD/SCID) mice after transplantation of rcd34 þ cells, Lin cells, or CD61 þ cells obtained from TPO-expanded CD34 þ umbilical cord blood (UCB) cells. Transplantation of nonexpanded CD34 þ cells was used as control. For all groups, cells from different UCB were used. The results are given as mean 6 SEM (n ¼ 4). (B): Platelet concentrations observed in the blood of NOD/SCID mice after transplantation of 2 10 5 nonexpanded CD34 þ cells, transplantation of all Lin cells obtained after 10-day TPO-expansion of 2 10 5 CD34 þ cells or transplantation of all rcd34 þ cells obtained after 10-day TPO-expansion of 2 10 5 CD34 þ cells. For all groups, cells from the same UCB were used. The results are given as mean 6 SEM (n ¼ 3). quickly from the blood during the following days (Fig. 2A, square). Because of differences between individual UCB [20], we compared platelet recovery following transplantation of the cells derived from a single UCB. Either 2 10 5 CD34 þ cells per mouse were directly transplanted or 2 10 5 CD34 þ cells per mouse were expanded for 10 days with TPO and subsequently, all the purified Lin and the rcd34 þ cells isolated after expansion were transplanted (Fig. 1B). In this experiment, there was 8.3-fold expansion of cells, of which 4.9% was rcd34 þ (0.8 10 5 cells per mouse), and 12.8% was Lin (2.1 10 5 cells per mouse). Since in earlier experiments nonengraftment of the CD61 þ was consistently observed, the CD61 þ cells (70.2%; 11.6 10 5 cells) were not transplanted. Transplantation of Lin and rcd34 þ cells in the NOD/SCID mice resulted in the appearance of human platelets in the blood of all recipient mice. Human platelets were observed after 4 days in mice transplanted with the Lin population (Fig. 2B) and 15 days after transplantation with the rcd34 þ cells. Following transplantation of the Lin population, the highest concentration of human platelets was obtained at day 21, and a substantial number of platelets were still present at 6 weeks after transplantation. Transplantation of rcd34 þ cells gave rise to a maximum number of platelets at 6 weeks after transplantation, but at this interval, the

992 Functional Characteristics of TPO-Cultured CB Figure 3. Engraftment and lineage differentiation of human cells in mouse BM after transplantation. Mice were transplanted with nonexpanded CD34 þ cells, Lin cells, or rcd34 þ cells from the same umbilical cord blood. Six weeks after transplantation, the mouse BM was analyzed by flow cytometry for engraftment with human cells by analysis of the percentage of human CD45 cells in the total BM- WBC population and by determining the lineage distribution of the human CD45 þ cells. Results are given as mean (n ¼ 3). Abbreviation: BM, bone marrow. Table 2. MK and myeloid colony formation of the human cells engrafted in mouse BM (mean 6 SEM) Per 10 4 total BM-nucleated cells Per 10 3 hcd45 þ in mouse BM CFU-MK HPC-GEM CFU-MK HPC-GEM Nonexpanded 3.6 6 1.8** 17.9 6 12.4 4.8 23.9 CD34 þ Lin 1.8 6 0.7* 18.9 6 16.7 2.5 25.9 rcd34 þ 0.3 6 0.4** 7.0 6 4.6 2.5 58.3 *p ¼.053 with nonexpanded CD34 þ. **p <.01 with nonexpanded CD34 þ and Lin. Six weeks after transplantation, the BM of transplanted mice was harvested and cultured in CFU-MK and HPC-GEM assays. In addition, the engraftment expressed as percentage of hcd45 cells compared to mcd45 cells was determined. The total number of colonies was expressed per 10 4 total BM-nucleated cells cultured and per 10 3 human cells cultured. Abbreviations: BM, bone marrow; CFU, colony-forming unit; GEM, granulocytes, erythrocytes, and monocytes; HPC, hematopoietic progenitor culture; MK, megakaryocyte. highest platelet concentration was observed in mice transplanted with nonexpanded CD34 þ cells (169 6 138 human platelets per microliter), as compared to the mice transplanted with Lin cells (40 6 30 human platelets per microliter) or rcd34 þ cells (31 6 21 human platelets per microliter). Engraftment of the Expanded Subpopulations in the BM of NOD/SCID Mice Detection of human CD45 þ cells in BM of the mice transplanted with rcd34 þ and Lin cells was performed at 6 weeks after transplantation. Since the percentage of engrafting human CD45 þ cells in the BM varies widely between individual UCB, we compared the engraftment potential of expanded and nonexpanded cells of the same UCB. A similar percentage of human CD45 þ cells was observed in the BM of mice transplanted with nonexpanded CD34 þ cells (7.5% 6 4.9%) and in mice transplanted with the Lin subpopulation (7.3% 6 4.2%), while a lower engraftment level was observed after transplantation of the rcd34 þ subpopulation (1.2% 6 0.4%) (Fig. 3). A dissimilar differentiation pattern of the human CD45 þ cells into myeloid and lymphoid cells was observed between mice transplanted with nonexpanded CD34 þ cells, Lin cells, and rcd34 þ cells. Myeloid cells were more abundant than lymphoid cells after transplantation with the nonexpanded CD34 þ cells or the rcd34 þ cells, while this was reverse after transplantation of the Lin cells (Fig. 3). The percentage of human CD34 þ cells in the human CD45 þ population present in the BM of recipient mice was comparable after transplantation of nonexpanded CD34 þ cells, Lin cells, or rcd34 þ cells (Fig. 3). As described above and shown in Figure 2A, at 6 weeks after transplantation, no human platelets could be detected in the blood of recipient mice transplanted with the CD61 þ subpopulation. In addition, no human CD45 þ cells were observed in the recipient mouse BM indicating that the CD61 þ cells were not able to engraft in the mouse BM. MK and Myeloid Colony Formation of the Human Cells in Mouse BM BM analysis of recipient mice at 6 weeks after transplantation demonstrated that the frequencies of MK precursors (CFU- MK) per 10 4 total BM-nucleated cells were higher in the marrow of mice transplanted with nonexpanded CD34 þ cells (3.6 6 1.8 CFU-MK per 10 4 total BM-nucleated cells) compared to mice transplanted with Lin cells (1.8 6 0.7 CFU-MK per 10 4 total BM-nucleated cells; p ¼.053). In the mice transplanted with rcd34 þ cells, only 0.3 6 0.4 CFU-MK per 10 4 total BM-nucleated cells were observed in the marrow, which was significantly lower as compared to the marrow of mice transplanted with nonexpanded CD34 þ or Lin cells (p <.004 and p <.006, respectively) (Table 2). The frequency of myeloid precursors per 10 4 total BMnucleated cells was similar in BM of mice transplanted with the nonexpanded CD34 þ cells (17.9 6 12.4 colonies per 10 4 total BM-nucleated cells) or with the Lin cells (18.9 6 16.7 colonies per 10 4 total BM-nucleated cells). The frequency of myeloid precursors was lower in the mice transplanted with the rcd34 þ population (7.0 6 4.6 colonies per 10 4 total BMnucleated cells; ns) (Table 2). In addition, the frequency of MK and myeloid precursor was calculated per 1 10 3 human CD45 þ cells. Transplantation of nonexpanded CD34 þ cells gave rise to 4.8 CFU-MK colonies per 1 10 3 human CD45 þ cells. A lower frequency of CFU-MK precursors as compared to transplantation of nonexpanded CD34 þ cells was observed after transplantation of rcd34 þ cells or Lin cells (2.5 colonies per 1 10 3 human CD45 þ cells for both subpopulations). Notably, the frequency of myeloid precursors per 1 10 3 human CD45 þ cells was higher in the BM of mice transplanted with rcd34 þ cells (58.3 colonies per 1 10 3 human CD45 þ cells) as compared to the BM of mice transplanted with nonexpanded CD34 þ cells or Lin cells (23.9 vs. 25.9 colonies per 1 10 3 human CD45 þ cells, respectively) (Table 2). CD34 1, rcd34 1, Lin 2, and CD61 1 Cells Have Different Long-Term and Short-Term Hematopoietic Progenitor and Precursor Cell Culture Characteristics CD34 þ cells from four different UCB were cultured for 10 days with TPO and subsequently the rcd34 þ cells, Lin cells, and CD61 þ cells were isolated. The purity of the subpopulations was high for the rcd34 þ cells (96.6% 6 1.7%) and the CD61 þ cells (95.0% 6 4.5%) and lower for the Lin cells (58.4% 6 12.3%) that still contained rcd34 þ cells (18.0% 6 6.2%) and CD61 þ cells (23.7% 6 8.8%).

Schipper, Brand, Fibbe et al. 993 Figure 4. Growth capacity of the rcd34 þ, Lin, and CD61 þ subpopulations in hematopoietic progenitor culture (HPC)-GEM and cobblestone area-forming cell (CAFC) cultures. CD34 þ umbilical cord blood cells were expanded for 10 days with TPO and subsequently rcd34 þ cells, Lin cells, and CD61 þ cells were isolated. (A): HPC-GEM growth capacity of the cell populations was determined in standard myeloid (HPC- GEM) and MK progenitor (CFU-MK) growth cultures. For each subpopulation, the number of colonies per 10 2 seeded cells was calculated. In the HPC-GEM cultures also, the differentiation pattern toward erythrocytes (BFU-E), CFU-GM, or a combination of both (CFU-GEM) was determined (mean 6 SEM; n ¼ 3). (B): The capacity of the expanded cell populations to form cell colonies was determined in CAFC cultures on NIH-3T3 cells. A representative example of a CAFC colony from rcd34 þ cells and Lin cells is shown; colonies are indicated by the white arrow. Abbreviations: BFU, burst forming unit; CFU, colony-forming unit; GEM, granulocytes, erythrocytes, and monocytes; GM, monocytes/ granulocytes; MK, megakaryocyte. Short-term HPCs were performed with these four different UCB (Fig. 4A). A significantly higher number of colonies were observed in the HPC of the rcd34 þ cells as comparedtothelin cells; in total 44.5 6 8.6 colonies per 10 2 rcd34 þ cells were detected, compared to 23.4 6 5.0 colonies per 10 2 Lin cells (p ¼.025). Moreover, the differentiation of the colonies toward CFU-monocytes/granulocytes (GM) or BFU-E was significantly different between the rcd34 þ and Lin cells. The rcd34 þ cells differentiated predominantly toward CFU-GM colonies (81.2% 6 5.6% of the colonies), whereas the Lin cells differentiated equally toward all lineages (CFU-GM: 37.2% 6 4.3%;BFU-E:35.4% 6 14.4%; CFU-GEM: 27.4% 6 16.6%) (Fig. 4A; p <.05 with rcd34 þ for all colony types). MK progenitor growth was analyzed in the CFU-MK assay. The Lin subpopulation gave rise to a higher number of CFU-MK colonies (14.8 6 1.7/10 2 Lin cells) as compared to the rcd34 þ subpopulation (6.4 6 0.8/10 2 rcd34 þ cells; p ¼.0042). Hardly any colony formation in the HPC and CFU-MK assay was observed when culturing the CD61 þ subpopulation (0.07 6 0.04 HPC-GEM colonies and 0.7 6 0.1 CFU-MK colonies per 10 2 CD61 þ cells). Long-term hematopoietic progenitor growth of three independent UCB was analyzed in a CAFC assay. Since only few www.stemcells.com short-term hematopoietic progenitor and precursor cells were found in the CD61 þ subpopulation, these cells were not included in the CAFC assay. For all three UCB, seeding of the rcd34 þ cells always resulted in CAFC colony growth. On average, 93.3% 6 8.9% of the wells seeded with rcd34 þ was positive for CAFC colonies after 5 weeks culture. Seeding of the Lin cells gave rise to CAFC colony growth for two of the three UCB, and CAFC colonies were observed in 40% of the wells (p ¼.01 compared to rcd34 þ ). In general, the CAFC colonies that were derived from rcd34 þ cells were larger as compared to those that were derived from Lin cells (Fig. 4B). DISCUSSION Ex vivo expansion of UCB CD34 þ cells is an extensively studied approach to improve the delayed engraftment that is observed after UCB transplantation. Previously, we and others have repeatedly shown that transplantation in NOD/SCID mice of short-term (7 10 days) TPO-expanded UCB cells gives rise to earlier human platelet detection in the blood of recipient mice [13, 14]. To date, most studies focused on cytokine

994 Functional Characteristics of TPO-Cultured CB combinations to reach optimal expansion but have not evaluated the functional contributions of the individual cell populations in vivo. Performing such studies is however important in order to optimize the protocols for clinical use. In this study, we unravel the individual functions of the subpopulations present after TPO expansion of UCB CD34 þ cells. Expansion of UCB CD34 þ cells for 10 days with TPO generates a unique CD34 /CD61 /Lin population (Lin ), which is not observed after expansion of mobilized peripheral blood CD34 þ cells [13, 16, 17]. Transplantation with this subpopulation of Lin cells revealed that these cells were responsible for fast appearance of human platelets in the blood of recipient mice. Moreover, we found that the first day of detection of human platelets after transplantation, the persistence of the human platelet production, and the final concentration of platelets at 6 weeks after transplantation were dependent on the number of Lin cells transplanted. Altering the Lin cell dose transplanted into NOD/SCID mice demonstrated that a minimal number of Lin cells are required to accelerate platelet recovery and to achieve stable platelet production (Supporting Information S1). These results are in line with previous studies in which we have shown that the Lin population is highly enriched for immature cells committed to the MK lineage [17]. Although there is an obvious dose dependence between the number of Lin cells transplanted and speed and magnitude of human platelet recovery in the mouse circulation, the high heterogeneity that was observed between different UCBs with regard to expansion and engraftment potential combined with the variability between mice with respect to recovery potential makes setting a threshold for the minimal number of Lin cells that needs to be transplanted unable. As seen before in our previous studies, we again observed a biphasic curve with a reproducible transient decrease of platelets between the third and fourth week after transplantation [13]. Importantly, in this study, we performed a Lin doserelationship experiment that revealed that at lower dosages of Lin cells, the decrease of platelets after the third week was persistent or resulted in complete disappearance of platelets. The observation of a biphasic curve is suggestive for different populations being present within the Lin cell fraction. The Lin subpopulation is in itself a heterogeneous population, obvious in transition as shown by the miscellaneous expression of CD38 and CD36. Although the majority of the Lin cells were CD38 þ, approximately 25% of the Lin population was characterized by a CD38 phenotype. Since differentiation toward CD61 þ cells induced loss of the CD38 marker, this suggests that the CD34 /CD38 /Lin cells represent a more mature precursor phenotype as compared to the CD34 / CD38 þ /Lin cells. We therefore assume that the more mature CD34 /CD38 /Lin cells establish the early platelet production at day 4 after transplantation, while the CD34 /CD38 þ / Lin cells are responsible for platelet production at a later stage. Importantly, transplantation with this heterogeneous Lin population, for example, the CD34 /CD38 /Lin and the CD34 /CD38 þ /Lin is required to establish at one hand an accelerated platelet recovery and at the other hand stable engraftment in mice [13]. Since we had depleted the Lin cells from more committed cells expressing other CD markers, no further characterization of the cells responsible for the two waves could be established. Mononuclear cell (MNC) from adult blood contain a CD34 /CD36 þ /CD42a þ cell population that are suggested to be MK precursors [21]. Indeed, in our cultures, CD36 þ cells were present and their number increased during MK differentiation. CD42b þ cells (forming the receptor for von Willebrand factor and thrombin together with CD42a/-c/-d) was only detected in the CD61 þ positive subpopulation and absent in our culture in the Lin subset, indicating that this specific pro-mk phenotype was not present in the Lin subpopulation. The CD34 selection procedure from adult blood results in loss of erythroid progenitor cells [21, 22] and this may also hold true for MK progenitor cells. This observation addresses the question whether CD34 selection may not be the optimal starting point for expansion of UCB, and further studies may be required to find such a specific cell population for (pro-)mk-forming cells. The number of Lin cells obtained after culture and the number of Lin cells needed to achieve early platelet recovery appear dependent on the UCB CD34 þ cell donor. High variability in ex vivo expansion and engraftment in vivo is also a general concern of clinical UCB transplantation [20]. In contrast to the general idea that (pro-)mks should be transplanted, our study shows that culture protocols should aim for the less mature Lin cells. Although shorter culture periods may increase the percentage of Lin cells in the total cell population, the total expansion will be lower and thus the absolute amount of Lin cells may be equal or, dependent on UCB donor, even lower as compared to the 10-day expansion protocol (unpublished data, Y.v.H.). However, for fast expanding UCBs, shortening of the expansion time may be warranted to prevent advanced maturation toward CD61 þ cells and to optimize the production of Lin cells. An important subject of this study was to investigate whether the rcd34 þ cells behave similarly to the nonexpanded CD34 þ cells with regard to short- and long-term repopulation potential of the CD34 þ cells. To verify possible stem cell exhaustion, the engraftment capacity of the rcd34 þ cells was directly compared to those of the nonexpanded CD34 þ cells. The rcd34 þ cells were able to establish both engraftment in the BM and stable human platelet production in the blood, although platelet recovery was somewhat slower and engraftment levels were reduced as compared to nonexpanded CD34 þ cells. Possibly, the often small number of rcd34 þ cells transplanted is responsible, but it cannot be excluded that functional differences, in particular some exhaustion of the MK lineage but not of other hematopoietic lineages, have been induced during culture. Engraftment with rcd34 þ cells was established in all mice, while graft failure often occurs in part of the mice transplanted with nonexpanded CD34 þ cells [23, 24]. These results correspond to the in vitro colony formation assays that showed a higher number of colony-forming precursor cells in the BM of mice transplanted with rcd34 þ compared to mice transplanted with nonexpanded CD34 þ cells. After transplantation of the CD61 þ cells, circulating human platelets were transiently detected and no hematopoietic engraftment in the BM was observed. Interestingly, the survival kinetics of these platelets in the peripheral blood was similar to the half-life observed after transfusion of mature human platelets in NOD/SCID mice [19]. Possibly, the high circulatory forces and the high sheer stress accelerate platelet release from CD61 þ cells resulting in early but transient circulating human platelets in the mouse but not to de novo platelet production from precursor cells in the BM [25]. This results differ from observations using mobilized peripheral blood stem cell (PBSC) in humans, which show a correlation between the number of CD34 þ /CD61 þ cells in the graft and time to platelet recovery after transplantation [15, 16, 26, 27]. This emphasizes that TPO expansion of UCB CD34 þ cells and adult CD34 þ cells gives rise to different cell populations. Mattia et al. reported a correlation between the number of CD34 þ /CD61 þ cells transplanted and platelet production in the NOD/SCID mice for TPO-expanded CD34 þ UCB cells, cultured for 7 days under conditions slightly different from ours [14]. Possibly, this correlation may not be causal but

Schipper, Brand, Fibbe et al. 995 may be due to other expanded cells. When comparing, less CD34 þ /CD61 þ cells were present in our culture at day 10 and they were included in the CD61 þ subpopulation, which did not play a role in long-term platelet production. The differentiation pattern of human CD45 þ cells in the BM of mice transplanted with Lin cells was different from that of mice transplanted with nonexpanded CD34 þ or rcd34 þ cells. While the human CD45 þ cells in the BM of mice that had received Lin cells skewed toward the lymphoid lineage, myeloid cells were more pronounced in the BM human CD45 þ population after transplantation of nonexpanded CD34 þ or rcd34 þ cells. It can be speculated that in the Lin population, the cells that are primed for differentiation toward the myeloid lineage get exhausted during expansion, while the cells that are primed for lymphoid differentiation are unaffected or enhanced by the TPO expansion, enabling these cells to engraft in the mouse. Interestingly, the Lin cells differentiated less frequently to CFU-GM colonies in vitro as compared to rcd34 þ cells. This suggests intrinsic differences between Lin and CD34 þ cells induced by TPO maturation, although additional studies are needed to resolve this observation. An important concern of ex vivo expansion of CD34 þ cells is the risk of depleting the long-term engrafting stem cell population and thus exhausting the hematopoietic stem cell pool [11]. We found that mouse BM obtained 6 weeks after transplantation of the Lin and rcd34 þ cell population contains human cells that are capable of forming myeloid, erythrocyte, and MK colonies in short-term progenitor cultures in vitro, suggesting the presence of human progenitor and precursor cells in the mouse BM. CAFC colony-forming cells were mainly preserved in the rcd34 þ population, although some CAFC colonies were detected in the Lin population. The frequency of CAFC colony-forming cells in the Lin subpopulation was however significantly less as in the rcd34 þ subpopulation; moreover, the size of the colonies formed by Lin cells was smaller. Considering the presence of 18% rcd34 þ cells in the Lin purified population in these experiments, the colonies detected in the Lin CAFC cultures may have been derived from rcd34 þ cells rather than from the Lin cells itself. Nevertheless, the TPO-expansion method as used in this study does not seem to induce definitive loss of hematopoietic progenitor and precursor cells that are able to form colonies after 5 weeks in CAFC culture. Moreover, mouse studies with TPO-expanded cells have shown that human platelets can still be detected in the murine peripheral blood at 5 months after transplantation (unpublished data, L.F.S.). Altogether, it is likely that long-term repopulating stem cells are maintained during expansion of CD34 þ cells with TPO. CONCLUSIONS This study is the first to examine the specific contribution of the different cell populations present after TPO expansion of UCB CD34 þ cells, on platelet recovery and BM engraftment. One of the main findings is that not all cells obtained after expansion participate in this. In our setting, Lin cells and rcd34 þ cells differently contributed to platelet recovery and BM engraftment in mice, while the majority of expanded cells, the CD61 þ cells, appeared irrelevant. These results indicate that not only the number of cells obtained after expansion is crucial for the outcome of the transplantation but also the composition of the expanded cell population is critical. It is unknown whether the current number of Lin cells that can be obtained with this expansion protocol is sufficient to accelerate platelet recovery in patients after UCB transplantation. In further studies, the expansion protocol should be adapted such that the expanded graft contains more functional cells (Lin and rcd34 þ cells) and less ineffective cells (CD61 þ cells). In general, future expansion protocols should be modified such that they better serve the clinical aim, which can vary from transplanting more CD34 þ hematopoietic progenitor cells into the patient with preservation of the long-term repopulating cell, to accelerate the recovery of a specific subpopulation without the necessity to maintain long-term stem cell capacity. In particular, in double-cord transplantation, the latter can be afforded. ACKNOWLEDGMENTS We thank H. de Boer, M.C. Slot, and B. Meijer for their help with performing the experiments. This work was supported by Grants PPOC-01-003 and PPOC-06-030 of the Sanquin Blood Supply Foundation, The Netherlands. TPO was a kind gift from KIRIN Brewery Company Ltd., Pharmaceutical Division, Tokyo, Japan. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST The authors indicate no potential conflicts of interest. REFERENCES 1 Sutherland HJ, Lansdorp PM, Henkelmann DH et al. Functional characterization of individual human hematopoietic stem cells cultured at limiting dilution on supportive marrow stromal layers. Proc Natl Acad Sci USA 1990;87:3584 3588. 2 Fei XM, Wu YJ, Chang Z et al. Co-culture of cord blood CD34(þ) cells with human BM mesenchymal stromal cells enhances short-term engraftment of cord blood cells in NOD/SCID mice. Cytotherapy 2007;9:338 347. 3 Huang GP, Pan ZJ, Jia BB et al. Ex-vivo expansion and transplantation of hematopoietic stem/progenitor cells supported by mesenchymal stem cells from umbilical cord blood. Cell Transplant 2007;16: 579 585. 4 Wang JF, Wang LJ, Wu YF et al. Mesenchymal stem/progenitor cells in human umbilical cord blood as support for ex vivo expansion of CD34þ hematopoietic stem cells and for chondrogenic differentiation. Haematologica 2004;89:837 844. 5 Noort WA, Kruisselbrink AB, in t Anker PS et al. Mesenchymal stem cells promote engraftment of human umbilical cord bloodderived CD34(þ) cells in NOD/SCID mice. Exp Hematol 2002;30: 870 878. 6 Chan SL, Choi M, Wnendt S et al. Enhanced in vivo homing of uncultured and selectively amplified cord blood CD34(þ) cells by cotransplantation with cord blood-derived unrestricted somatic stem cells. Stem Cells 2007;25:529 536. 7 Brunstein CG, Baker KS, Wagner JE. Umbilical cord blood transplantation for myeloid malignancies. Curr Opin Hematol 2007;14: 162 169. 8 Barker JN, Weisendorf DJ, Wagner JE. Creation of a double chimera after the transplantation of umbilical-cord blood from two partially matched unrelated donors. N Engl J Med 2001;344:1870 1871. 9 Boiron JM, Dazey B, Cailliot C et al. Large-scale expansion and transplantation of CD34þ hematopoietic cells: In vitro and in vivo conformation of neutropenia abrogation related to the expansion process without impairment of the long term engraftment capacity. Transfusion 2006;46:1934 1942. www.stemcells.com

996 Functional Characteristics of TPO-Cultured CB 10 Norol F, Drouet M, Mathieu J et al. Ex-vivo expanded mobilized peripheral blood CD34þ cells accelerate haematological recovery in a baboon model of autologous transplantation. Br J Haematol 2000;109: 162 172. 11 Delaney C, Heimfeld S, Brashem-Stein C et al. Notch-mediated expansion of human cord blood progenitor cells capable of rapid myeloid reconstitution. Nat Med 2010;16:232 236. 12 Bruno S, Gunetti M, Gammaitoni L et al. Fast but durable megakaryocyte repopulation and platelet production in NOD/SCID mice transplanted with ex-vivo expanded human cord blood CD34þ cells. Stem Cells 2004;22:135 143. 13 van Hensbergen Y, Schipper LF, Brand A et al. Ex vivo culture of human CD34þ cord blood cells with thrombopoietin (TPO) accelerates platelet engraftment in a NOD/SCID mouse model. Exp Hematol 2006;34:943 950. 14 Mattia G, Milazzo L, Vulcano F et al. Long-term platelet production assessed in NOD/SCID mice injected with cord blood CD34þ cells, thrombopoietin-amplified in clinical grade serum-free culture. Exp Hematol 2008;36:244 252. 15 Kamel AM, El-Sharkawy N, Mahmoud HK et al. Impact of CD34 subsets on engraftment kinetics in allogeneic peripheral blood stem cell transplantation. Bone Marrow Transplant 2005;35:129 136. 16 Meldgaard Knudsen L, Jensen L, Jarlbaek L et al. Subsets of CD34þ hematopoietic progenitors and platelet recovery after high dose chemotherapy and peripheral blood stem cell transplantation. Haematologica 1999;84:517 524. 17 Schipper LF, Brand A, Reniers N et al. Differential maturation of megakaryocyte progenitor cells from cord blood and mobilized peripheral blood. Exp Hematol 2003;31:324 330. 18 Schipper LF, Brand A, Reniers NC et al. Effects of thrombopoietin on the proliferation and differentiation of primitive and mature haemopoietic progenitor cells in cord blood. Br J Haematol 1998;101:425 435. 19 Schipper LF, van Hensbergen Y, Fibbe WE et al. A sensitive quantitative single-platform flow cytometry protocol to measure human platelets in mouse peripheral blood. Transfusion 2007;47:2305 2314. 20 Ballen KK, Valinski H, Greiner D et al. Variables to predict engraftment of umbilical cord blood into immunodeficient mice: Usefulness of the non-obese diabetic-severe combined immunodeficient assay. Br J Haematol 2001;114:211 218. 21 Migliaccio AR, Tirelli V, Masiello F et al. High levels of CD44 expression identify hematopoietic cells capable of generating great numbers of erythroid cells under HEMA conditions. Blood 2010;116: 3349a. 22 Migliaccio AR, Whitsett C, Migliaccio G. Erythroid cells in vitro: From developmental biology to blood transfusion products. Curr Opin Hematol 2009;16:259 268. 23 Hogan CJ, Shpall EJ, McNulty O et al. Engraftment and development of human CD34þ-enriched cells from umbilical cord blood in NOD/ LtSz-scid mice. Blood 1997;90:85 96. 24 Piacibello W, Sanavio F, Severino A et al. Engraftment in nonobese diabetic severe combined immunodeficient mice of human CD34þ cord blood cells after ex vivo expansion: Evidence for the amplification and self renewal of repopulating stem cells. Blood 1999;93: 3736 3749. 25 Dunois-Larde C, Capron C, Fichelson S et al. Exposure of human megakaryocytes to high shear rates accelerates platelet production. Blood 2009;114:1875 1883. 26 Dercksen MW, Weimar IS, Richel DJ et al. The value of flow cytometric analysis of platelet glycoprotein expression on CD34þ cells measured under conditions that prevent P-selectin mediated bindings of platelets. Blood 1995;86:3771 3782. 27 Feng R, Shimazaki C, Inaba T et al. CD34þ/CD41þ cells best predict platelet recovery after autologous peripheral blood stem cell transplantation. Bone Marrow Transplant 1998;21:1217 1222. See www.stemcells.com for supporting information available online.