Molecular and Clinical Hematology Branch, National Institutes of Diabetes and Digestive and Kidney Disorders, Maryland, USA
National Heart, Lung, and Blood Institute, NIH, Bethesda, Maryland, USA
Molecular and Clinical Hematology Branch, National Institutes of Diabetes and Digestive and Kidney Disorders and National Heart, Lung, and Blood Institute, National Institutes of Health, 9000 Rockville Pike, Building 10, Room 9N 116, Bethesda, Maryland 20892, USA
Author contributions: J.H.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; M.M.H.: conception and design, administrative support, manuscript writing, data analysis and interpretation; N.U.: data analysis and interpretation; O.P.: collection and assembly of data; J.F.T.: conception and design, financial support, administrative support, final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSExpress October 16, 2008.
Xenografting immunodeficient mice after low-dose irradiation has been used as a surrogate human hematopoietic stem cell (HSC) assay; however, irradiation requires strict and meticulous animal support and can produce significant mortality rates, limiting the usefulness of this model. In this work, we examined the use of parenteral busulfan as an alternative conditioning agent. Busulfan led to dose-dependent human HSC engraftment in NOD/LtSz-scid/IL2Rγnull mice, with marked improvement in survival rates. Terminally differentiated B and T lymphocytes made up most of the human CD45+ cells observed during the initial 5 weeks post-transplant when unselected cord blood (CB) products were infused, suggesting derivation from existing mature elements rather than HSCs. Beyond 5 weeks, CD34+-enriched products produced and sustained superior engraftment rates compared with unselected grafts (CB CD34+, 65.8% ± 5.35%, vs. whole CB, 4.27% ± 0.67%, at 24 weeks). CB CD34+ group achieved significantly higher levels of engraftment than mobilized CD34+-enriched peripheral blood (PB CD34+). At 8 weeks, all leukocyte subsets were detected, yet human red blood cells (RBCs) were not observed. Transfused human red cells persisted in the chimeric mice for up to 3 days; an accompanying rise in total bilirubin suggested hemolysis as a contributing factor to their clearance. Recipient mouse-derived human HSCs had the capacity to form erythroid colonies in vitro at various time points post-transplant in the presence of human transferrin (Tf). When human Tf was administered singly or in combination with anti-CD122 antibody and human cytokines, up to 0.1% human RBCs were detectable in the peripheral blood. This long evasive model should prove valuable for the study of human erythroid cells. STEM CELLS2009;27:175–182
Despite the development of sophisticated in vitro assays for primitive hematopoietic stem cells (HSCs), transplantation assays remain the gold standard for HSC enumeration. Large animals model human HSCs activity better, yet they are cumbersome, costly, and outbred, yielding high variability. The creation of humanized mice that carry partial or complete human physiological systems may overcome some of these obstacles. Transplantation of human HSCs into immunodeficient mice has been used as a surrogate HSC assay. HSC xenotransplantation in NOD/SCID mice has proven a useful tool in gaining a better understanding of HSC behavior in vivo [1–3]. NOD/SCID mice have long been the standard for xenografting human HSCs, but the remaining natural killer (NK) cell activity interfered with efficient human cell engraftment. To reduce residual NK cell activity in recipient xenograft mice, a new strain, the NOD/SCID/β2mnull mouse, was established. In this strain, efficient human cell engraftment was observed; however, human T cells were not detected in high numbers [4, 5]. Third-generation NOD/LtSz-scid/IL2Rγnull (NOD/SCID/IL2Rγnull) mice were subsequently established to overcome these issues . Another similar strain of mice, NOD/Shi-scid/IL2rγnull, with a truncated or complete null γC mutation, has also been reported [7, 8]. These mice have no lymphocyte or NK cell activity and demonstrate impaired dendritic cell function. When human HSCs are transplanted into these strains after low-dose total body irradiation (TBI), T, B, and NK cells of human origin are easily detectable in the circulation [6–9]. To further develop this model for gene therapy applications, several remaining issues require further study, including optimal conditioning, HSC source, cell dose, graft composition, and the lack of significant human erythroid output observed to date.
To achieve high-level human cell engraftment in the xenograft model, TBI as preconditioning has been nearly exclusively used; however, the administration of TBI in our facility results in substantial mortality. Moreover, TBI requires strict facility regulation to allow the use of irradiators, often located in areas remote from animal housing. Care of animals following TBI requires autoclaved cages and chow, antibiotic administration, HEPA-filtered airflow, and pathogen-free conditions to handle the animals. Clinically, TBI has been progressively replaced by the use of a combination of chemotherapeutic agents such as busulfan [10–12]. Thus a simple, convenient, effective, and less toxic conditioning regimen remains desirable.
Busulfan (Bu; 1,4-butanediol dimethanesulfonate) is an alkylating agent with predominantly myelosuppressive and minimally immunosuppressive properties that has long been used in human HSC transplantation . Indeed, busulfan has been extensively used in experimental animal models [14–16], but no reports have demonstrated its use in the NOD/SCID/IL2Rγnull strain. Because NOD/SCID/IL2Rγnull mice have a more immature immune system, we hypothesized that only myelosuppression was necessary to achieve human cell engraftment, allowing significant human cell engraftment with busulfan conditioning alone. In this report, we describe our experience with busulfan conditioning in NOD/SCID/IL2Rγnull mice, comparing HSC sources, graft composition, and conditions for the elusive human erythroid output.
MATERIALS AND METHODS
Male NOD.Cg-Prkdcscid IL2rgtm1Wjl/Sz/J (NOD/LtSz-scid/IL2Rγnull, NOD/SCID/IL2Rγnull) mice were purchased from Jackson Laboratory (Bar Harbor, ME, http://www.jax.org) and have been described elsewhere . Mice were 7–10 weeks old at the time of transplant. Mice were housed in a pathogen-free facility and handled according to an animal care and use protocol approved by the Animal Care and Use Committee at the NIH.
Conditioning of NOD/LtSz-Scid IL2RγNull Mice for Transplantation
Various doses of busulfan (Busulfex; PDL BioPharma, Redwood City, CA, http://www.pdl.com) were diluted with phosphate-buffered saline (PBS; Quality Biological, Gaithersburg, MD, http://www.qualitybiological.com) to a final volume of 500 μl and injected into recipient mice intraperitoneally. Doses of 10 and 25 mg/kg busulfan were injected 24 hours before infusion of human cells; 50 mg/kg doses were injected in two equal doses of 25 mg/kg 48 and 24 hours prior. Some mice were injected with 100 μg of anti-CD122 antibody generated from TM-β1 hybridoma cell lines (Becton, Dickinson and Company, Mountain View, CA, http://www.bd.com), as previously reported . One group of control mice received low-dose (100–300 cGy) total body radiation from a cesium source at 91.7 cGy/minute (MDS Nordion, Ottawa, http://www.mds.nordion.com).
Cell Preparation and Infusion
Cord blood (CB) cells were obtained from eight different normal volunteers during deliveries after written informed consent using a protocol approved by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Institutional Review Board (IRB) at the NIH. Mononuclear cells were obtained after density-gradient centrifugation using Ficoll-Paque Plus (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com), and red blood cells (RBCs) were removed using ammonium chloride potassium lysis buffer (Quality Biological). CD34 or CD3 enrichment was performed using anti-CD34 or anti-CD3 biotinylated antibody and streptavidin microbeads (Miltenyi Biotec, Auburn, MA, http://www.miltenyibiotec.com). Cells were passed through the positive selection column twice, and purity was assessed by flow cytometry using phycoerythrin (PE)-conjugated anti-human CD34 antibody (clone 563) or anti-human CD3 antibody on a FACSCalibur (all BD Biosciences, San Jose, CA, http://www.bdbiosciences.com).
Peripheral blood (PB) CD34+ cells were obtained from 10 different healthy adult volunteers (age 18 and older) as previously described  using a protocol approved by the NIDDK IRB. Briefly, the HSCs were mobilized with 5 days of 10 μg/kg recombinant human (rHu) granulocyte colony-stimulating factor (G-CSF; Amgen, Thousand Oaks, CA, http://www.amgen.com). On day 5 of mobilization, PB CD34+ cells were collected by one 15-l leukapheresis, and CD34-positive selection was performed using the Isolex 300i automated immunomagnetic system (Nexell, Miltenyi, Auburn, CA, http://www.miltenyibiotec.com).
Donor human cells were suspended in PBS to a final volume of 500 μl and infused intravenously via the tail vein as previously described . After infusion of donor human cells, murine blood was obtained every 6 days for complete blood counts, using a Celldyne 3500 automated cell counter (Abbott, Abbott Park, IL, http://www.abbott.com).
Flow Cytometric Analysis for Donor Chimerism and Leukocyte Subset
Bone marrow (BM) was harvested as previously described , and splenocytes were obtained by gently mashing the spleens with forceps through a nylon mesh in Dulbecco's modified Eagle's medium (Mediatech, Herndon, VA, http://www.cellgro.com) . The splenocytes were then washed and resuspended at a concentration of 2 × 106 cells per milliliter in 0.1% bovine serum albumin. BM, spleen, and PB cells were stained with fluorescein isothiocyanate (FITC)-conjugated anti-human CD45 and PE-conjugated anti-mouse CD45; red cells, anti-mouse TER119-FITC, and anti-human glycophorin A (GPA)-PE (CD235a). Leukocyte subsets were stained with FITC-conjugated anti-human CD45 and one of the following: CD14-PE, CD16-PE, CD20-PE, CD3-PE, CD56-PE, CD41-PE, CD61-PE, or CD71-PE. B-lymphocyte subsets were stained with CD19-PE and immunoglobulin IgM-FITC or IgG-FITC, and T lymphocytes were stained with CD3-FITC and CD4-PE or CD8-PE. All antibodies were purchased from BD Biosciences.
Colony-Forming Unit Assay
BM or spleen cells of chimeric NOD/SCID/IL2Rγnull mice were inoculated at 1 × 105 cells per milliliter in three different methylcellulose culture media, as previously reported: [3, 21]: MethoCult H4436 (StemCell Technologies, Vancouver, BC, Canada, http://www.stemcell.com), which contains human cytokines and transferrin (Tf); MethoCult H4230, supplemented with 5 U/ml rHu erythropoietin (Epo; Amgen), 10 ng/ml rHu granulocyte-macrophage colony-stimulating factor (GM-CSF; Sandoz, East Hanover, NJ, http://www.us.sandoz.com), 10 ng/ml rHu interleukin-3 (IL-3), and 100 ng/ml rHu stem cell factor (SCF; R&D Systems Inc., Minneapolis, http://www.rndsystems.com); and MethoCult M3434 (StemCell Technologies), which contains murine cytokines and Tf. Plated cells were incubated at 37°C with 5% CO2. Between days 10 and 14, colonies were counted, and 12–36 individual colonies were plucked, digested with 20 μg/ml proteinase K (Qiagen, Gaithersburg, MD, http://www1.qiagen.com) at 55°C for 1 hour, and incubated at 99°C for 10 minutes. The resulting DNA was analyzed for human ALU by nested polymerase chain reaction (PCR)  using the following oligonucleotide primers: ALU-5′, CACCTGTAATCCCAGTTT-3′; ALU-3′, CGCGATCTCGGCTCACTGCA. PCR cycles were set at denaturation at 95°for 10 minutes and then 20 cycles of amplification at 95°for 1 minute, 55°for 45 seconds, and 72°for 1 minute.
Red Cell Transfusion
Type O RBCs were collected from healthy human volunteers. For each mouse, 1,000 μl of human whole blood was centrifuged at 427g, serum was removed, and RBCs were washed twice and diluted with PBS to a final volume of 750 μl.
Injection of Human Cytokines for Erythropoiesis
We transplanted additional mice with CB CD34+ cells and treated them using a variety of cytokine cocktails to mimic in vitro culture conditions: (a) human Tf alone; (b) colony-forming unit (CFU) conditions: human Tf, SCF, Epo, IL-3, and GM-CSF; (c) CFU cytokines and anti-human CD122 antibody; (d) erythroid culture conditions: human Tf, SCF, dexamethasone, Epo, and transforming growth factor-β (TGF-β); (e) erythroid culture conditions and anti-human CD122.
Sixteen milligrams of human holo-Tf per mouse (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) was administered three times a week for 3–5 weeks beginning at day 0 (n = 3–4, repeated four times), 60 (n = 3), 120 (n = 3), 150 (n = 3), or 180 (n = 3) post-transplant. This dose was calculated on the basis of the upper limit of normal human serum Tf level in the clinical chemistry laboratory, and it was confirmed by human Tf enzyme-linked immunosorbent assay kit (Bethyl Inc., Montgomery, TX, http://www.bethyl.com). In the day 0 group only, 100 μg of human anti-CD122 (BD Biosciences) was administered once at the day of transplant; 10−6 M dexamethasone (Sigma-Aldrich) and cytokines (20 U of rHu Epo, 100 ng of SCF, 10 ng of IL-3, 100 ng of TGF-β, 10 ng of GM-CSF) were all administered at days 0, 7, and 14.
Mass Spectrometry Analysis
Murine whole blood (500 μl) was collected from mice that received transferrin, stained with human GPA and Ter119 antibody, and sorted using a FACSAria (BD Biosciences). The GPA-positive cells were pelleted, washed twice, lysed with water, frozen at −20°C for 5 minutes, and thawed at 37°C for 5 minutes. Proteins were visualized using a colloidal Coomassie Blue staining kit (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) and analyzed using tandem high-performance liquid chromatography (HPLC)-mass spectrometry. Masses from eluted peptides were queried in the Mascot database and searched for human α- and β-globin chains.
All measures of variance are presented as SEs. Statistical analyses were performed using Student's t test, with a p value <.05 deemed significant.
Busulfan and PB CD34+ Dose in Xenograft Transplantation
We initially transplanted mice with PB CD34+ cells (2 × 106 cells per mouse) after irradiation with 1 (n = 10), 1.5 (n = 7), 2 (n = 8), and 3 Gy (n = 9), but they experienced high mortality (Fig. 1A). Mortality estimates of 100% were seen at all doses higher than 1 Gy. Human cell engraftment levels after 1 Gy were maximal at 8 weeks and averaged 16.0% ± 1.07% (data not shown).
NOD/SCID/IL2Rγnull mice were then treated with escalating doses of busulfan from 10 to 50 mg/kg prior to infusion of 2 × 106 PB CD34+ cells. The mean percentage of human CD45+ cells detected at 24 weeks was 2.5% ± 0.5% in the 25 mg/kg group and 14.1% ± 0.8% in the 50 mg/kg group (Fig. 1B). In surviving mice with follow-up for up to 12 months, human cell engraftment remained persistent and stable (data not shown). Mortality estimates were more favorable after busulfan treatment and infusion of PB CD34+ (2 × 106 cells per mouse), with the majority of mice surviving long-term (Fig. 1C).
We then sought to determine whether increasing PB CD34+ cell number could improve engraftment following 50 mg/kg busulfan. The percentage of human CD45+ cells in the recipient mice rose in a dose-dependent manner: 6.1% ± 0.9% in the 5 × 105 group (n = 4), 13.7% ± 2.5% in the 2 × 106 group (n = 6), and 18.3% ± 1.6% in the 4 × 106 group (n = 7). The p value for the comparison between the 5 × 105 and 2 × 106 groups was 0.03, and the p value between the 2 × 106 and 4 × 106 groups was 0.156 (Fig. 1D). With higher PB CD34+ or busulfan doses, the percentage of human chimerism appeared to be the highest at 8 weeks; it then decreased, and it plateaued at 20–24 weeks.
Unselected CB Graft Contributed Mostly T Cells Detectable as Early as 6 Days Post-Transplant
We then chose 50 mg/kg busulfan to compare engraftment of PB CD34+, unselected cord blood mononuclear cells (unselected CB), and CD34+-selected cord blood cells (CB CD34+). The busulfan dose was lowered to 25 mg/kg in the unselected CB group because in an initial experiment, all unselected CB recipient animals died from pancytopenia after receiving 50 mg/kg busulfan (n = 8; data not shown). To evaluate the depth and duration of busulfan effects on hematopoiesis in this third-generation NOD/SCID mouse and to allow comparison with previously published results, we monitored the blood count recovery pattern and human cell engraftment in the first 50 days after the transplantation in four groups: unselected CB with 25 mg/kg busulfan (n = 7), CB CD34+ with 50 mg/kg busulfan (n = 8), PB CD34+ with 50 mg/kg busulfan (n = 5), and PB CD34+ with 1 Gy of radiation (n = 5). With the exception of the unselected CB group, the counts for white blood cells, hemoglobin, and platelets in busulfan-treated mice reached their nadir near day 18 and approached baseline values by day 42 (Fig. 2A–2C).
We then assessed the percentage of human CD45+ cells in these mice during the same early engraftment period (Fig. 2D). In these recipients, 36% ± 2.4% human cells were detected as early as day 6 in the unselected CB mice, and percent of human cells remained higher (42.3%–56.7%) than the other three groups (1.8%–33.5%) for the first 30 days. The unselected CB grafts contain mature leukocytes, namely CD3+ cells. After infusion, this likely led to the observed higher percentages of human cells from days 6–30 (Fig. 2A). In the other three groups, human cells were not detected until day 18. In addition, cord blood grafts consistently led to higher percentages of human CD45+ cells detected in the CB CD34+ group (61.2% ± 4.0%; n = 8) versus the PB CD34+ group (19.9% ± 1.5%; n = 5; p < .001).
CD34+-Selected CB Cells Gave Rise to All Human White Cell Subsets at 8 Weeks Post-Transplant
Since infusion of unselected CB cells resulted in the highest percentage of engraftment in the first 4 weeks, we sought to determine to what extent human cell “engraftment” represents mature blood elements from the graft assayed in the circulation and, furthermore, how long such mature elements persist. Three groups of mice were compared: 25 mg/kg busulfan and 1 × 108 unselected CB cells, 50 mg/kg busulfan and 2 × 106 CB CD34+ cells, and 50 mg/kg busulfan and 1 × 107 CB CD3+ cells. At day 6, the average percentages of human cells were 27.3% ± 2.7% in the unselected CB group, 0.17% ± 0.04% in the CB CD34+ group, and 6.51% ± 2.1% in the CB CD3+ group. One representative mouse from the unselected CB and CB CD34+ groups was sacrificed, and the percentages of human cells at day 6 in the spleen and bone marrow were as follows: unselected CB group, 45.1% and 1.1%; CB CD34+ group, 0.1% and 0%, respectively. These results confirmed that during the first week post-transplant, most of the circulating human leukocytes and splenocytes were T cells from the infused unselected CB cells. At week 3, the average percentage of human cells in peripheral blood was 49.7% ± 2.4% in unselected CB, 25.9% ± 1.7% in CB CD34+, and 8.8% ± 2.3% in CB CD3+.
We then compared transplant outcome in various leukocyte lineages in unselected CB and CB CD34+ mice at 8 weeks. A representative flow panel (Fig. 3) showed that CB CD34+ mice have a higher percentage of human myeloid (CD14, CD16) and B (CD20) cells, a lower percentage of T (CD3) cells, and similar percentages of NK (CD56) and platelet progenitor (CD41) cells compared with unselected CB mice. The B cells stained positively for human IgM or IgG, and the T cells stained positively for CD4 or CD8. When BM and spleen cells were assayed, the white cell subsets were similar to the observed percentages in peripheral blood (data not shown). This pattern of delayed CD3 engraftment was also seen in mice that received PB CD34+ cells (data not shown).
CD34+-Selected CB Grafts Produced Superior Long-Term Engraftment
To assess the long-term (>8 weeks) contribution of human cell from the different cell sources, another set of mice was transplanted with comparable cell numbers of CB CD34+ cells (2 × 106; n = 7), unselected CB cells (1 × 108; n = 6), and CB CD3+ cells (2 × 107; n = 3) after busulfan conditioning (Fig. 4). The CB CD34+ group had an average of 25.9% human cells at 3 weeks, which increased to and was sustained at 60%–65% at 24 weeks post-transplant. Unselected CB group had the highest initial human chimerism (49.7%, similar to Fig. 2D), reached 62.5% at 5 weeks, decreased to 30.2% at 8 weeks, and decreased to 4.27% at 24 weeks. The CB CD3+ group had the lowest human chimerism throughout the entire monitoring period yet showed moderate levels early post-infusion.
After 6 months of follow-up, the recipient's BM (1 × 108 cells) from CB CD34+ mice were harvested and transplanted to secondary NOD/SCID IL2Rγnull recipients conditioned with same busulfan dose (50 mg/kg). Human CD45+ cells of multiple leukocyte lineages were detected at 4.93% ± 1.7% (n = 3) in both groups up to 24 weeks (data not shown). These results showed that both PB and CB CD34+ cell sources contributed long-term HSCs in this xenograft transplantation model.
Erythroid Colony Maturation Was Detectable In Vitro After Plating Marrow of Recipient Mice in Human Cytokines with Human Tf
Although human-derived leukocyte and megakaryocytic cells were detected in NOD/SCID/IL2Rγnull peripheral blood, no human red blood cells were detectable throughout the entire monitoring period, regardless of the type of donor human cells infused. To ensure that the human HSCs in the humanized mice have a capability to differentiate into erythroid cells in vitro, BM was harvested from recipient mice at 2 days, 7 days, 2 months, and 6 months after transplants in the CB CD34+ group. At 6 months post-transplant, 1 × 105 BM cells were inoculated in semisolid methylcellulose media supplemented with human cytokines and Tf; colonies were counted and are shown in Table 1. PCR for human ALU sequences was positive for colonies grown in human-specific media (data not shown).
Table 1. Chimeric bone marrow cells (1 × 105) were plated in semisolid media
Colonies were counted after 14 days of incubation.
Human Red Blood Cells Are Detectable In Vivo After Treatment with Human Tf
We then sought to determine whether RBC output is limited by the absence of human growth factors. We transplanted additional mice with CB CD34+ cells and treated them with five different human cytokine cocktails to mimic in vitro culture conditions: (a) Tf alone; (b) Tf and CFU cytokines; (c) Tf, CFU cytokines, and anti-CD122; (d) Tf and erythroid culture cytokines; and (e) Tf, erythroid culture cytokines, and anti-CD122. Human red cells were detectable in all five groups, but only from day 11 to day 30 post-transplant, with a maximum of up to 0.1% at day 18 (Fig. 5). Human cytokines or anti-CD122 antibody did not increase the percentage of human RBCs over Tf alone. These results suggested that human Tf played a crucial role in supporting human erythropoiesis in the chimeric mice. To confirm that the GPA+ cells were of human origin, GPA+ RBCs from day 18 were sorted and lysed, and the hemoglobin extracts were analyzed by tandem HPLC-mass spectroscopy (MS). MS revealed the presence of human α- and β-globin chains.
We then assayed the bone marrow and spleen to determine whether erythroid progenitors could be detected at the time when circulating human RBCs were present. In the bone marrow, 5.96% of the cells were GPA+, and 5.81% were doubly positive for human CD45 and CD71. When Tf injections were administered later, at 60, 120, 150, and 180 days post-transplant, BM erythroid progenitors decreased substantially (Table 2, with some data not shown), suggesting that late Tf administration cannot rescue erythroid progenitors and that there might be overwhelming signals in the chimeric bone marrow inhibiting human erythropoiesis.
Table 2. Percentage of human erythroid cells by flow cytometry
Human Red Cells Are Quickly Cleared After Transfusion in Mice
Next, to confirm whether the NOD/SCID/IL2Rγnull mice can tolerate human red cells, mice were injected with 750 μl of type O and Rh-negative human RBCs, which should increase human red cells to 10%–20% in mice. The actual percentage of human RBCs in peripheral blood was only 2%–3% at 2 hours, 1%–1.5% at 72 hours, and 0% at 1 week. Pretreatment with busulfan, anti-CD-122, or splenectomy did not prolong the survival of circulating human RBCs in mice. Serum levels of total bilirubin were elevated in the mice 72 hours after the infusion (2.13 ± 0.29 vs. 0.267 ± 0.03 mg/dl; n = 3); p < .01). Other blood types (O-positive and A-positive) were infused, and similar results were obtained (data not shown). These results indicated that human HSCs in NOD/SCID/IL2Rγnull mice can be cultured to grow human RBCs ex vivo, transfused human RBCs can transiently persist in vivo, and human RBCs were cleared, at least in part, by extravascular hemolysis.
We have previously shown that low-dose busulfan alone provides dose-dependent engraftment in congenic normal C57BL6 mice  and is sufficient preconditioning for gene therapy applications in the rhesus macaque model [22, 23]. We sought to develop a practical, humanized mouse model that allows reliable, high-level human cell engraftment to reduce the need for the more cumbersome and expensive nonhuman primate model. Here we report busulfan as a desirable alternative conditioning agent in the NOD/SCID/IL2Rγnull mice. Busulfan at doses of 25–50 mg/kg yielded 15%–20% human chimerism from G-CSF mobilized peripheral blood CD34+ cell sources, and 60%–70% human chimerism from CB CD34+ cell sources. This high level of human chimerism was achieved after busulfan conditioning, with low mortality rates. In addition, ease of administration and the lack of the more cumbersome administrative oversight required of an irradiation source make this a desirable conditioning agent for human HSC xenografting. The dose of 50 mg/kg may be approaching the lethal dosage for these mice, as none of the recipients of 1 × 108 unselected CB cells at this busulfan dose survived the early post-transplant period, although this mortality was not seen with other cell sources. Interestingly, high-level chimerism, in excess of 90%, was seen in these mice, raising the possibility that such high levels of human chimerism are detrimental in this setting. Blood counts after busulfan dosing decreased precipitously during the first week, reached a nadir in the second to third week, and began to recover at the fourth week. This pattern of hematologic toxicity is similar to that reported in congenic mouse models, monkeys, and humans [12, 23].
Our transplantation studies with CB CD34+-selected cells in 8–10-week-old recipients resulted in significant levels of human chimerism at 60%–65%. These percentages are similar to the observed 73% in newborn NOD/SCID/IL2Rγnull after 100 cGy  and 30%–40% in 8–12-week-old mice after 240 cGy [8, 24] yet significantly higher than the 5%–10% observed in mice of similar ages (8–10 weeks old) after 300 cGy . Our higher cell dose may in part explain our better results, and it further supports a dose-dependent engraftment in this xenograft model. The lower percentage of human cells following a PB stem cell source in our study has also been observed previously using 7 × 105 PB CD34+ cells after 325 cGy of irradiation . These results are in agreement with previous studies showing that CB CD34+ cells have a higher engrafting potential than PB CD34+ cells.
Regardless of the source, engrafted human HSCs differentiated into all leukocyte lineages in PB, BM, and spleen. The unselected CB group showed a 90% human T-cell chimerism at 8 weeks, which resulted predominantly from the mature T cells in the infused graft. As these mature human T cells senesced, the percentage of human chimerism decreased sharply. In contrast, the CB CD34+ group achieved 85% human B cell chimerism at 8 weeks, and the T-cell engraftment was lower and gradual. The average time to reach maximal human total leukocyte chimerism depended on the cell dose: 5 weeks in our study with 2 × 106 CB CD34+ cells and 20 weeks with 5 × 104 cells [8, 24] (Fig. 4). These engraftment kinetics suggest that inferring human HSC activity at early time points (i.e., before 12 weeks post-transplant) may be misleading. Our observations from the infusion of CD3-selected CB cells strengthens this point, showing that mature elements may persist in this model for some time post-infusion.
Ishikawa et al. previously showed that 3 months after transplant in newborn mice, a small (unquantified) percentage of human red cells were in peripheral circulation, and 9.5% of human erythroid progenitors were present in the bone marrow . When bone marrow cells were cultured by CFU, there were few pure red-cell colonies, and white cell colonies dominated. Mazurier et al. detected human glycophorin A-positive cells in xenograft recipients in the bone marrow and peripheral blood beginning 2 weeks after direct intrafemoral injection, but the human erythrocytes decreased to 5% or less at 6 weeks . Our study used busulfan and adult mice, and human erythroid colonies could be grown using human cytokines from 2 days to 6 months after transplant. However, we observed only 0.1% human red cells in recipient peripheral blood on day 18 after transplant and 6% GPA+ cells in the bone marrow. These collective observations indicate that the bone marrow often contains much higher human erythroid content than the peripheral blood compartment. Furthermore, we observed human RBCs only with the use of human holo- or apo-Tf, suggesting that the human iron transport protein is important in erythropoiesis in this chimeric model. Tf is a glycoprotein with an approximate molecular mass of 76,500 kDa, produced in liver, and it binds and transports iron to the bone marrow . There is only 78.05% nucleic acid and 74.1% amino acid homology between human and murine Tf using the HomoloGene database.
The presence of human erythroid progenitors in BM and the much lower readout of mature human RBCs in circulation suggest there may be an inhibitory cytokine milieu suppressing human erythropoiesis. This can be inferred from Hexner et al., where adding ex vivo-expanded CD3/CD28-costimulated CB T cells, in addition to CB CD34+ cells, increased GPA+ cells in the bone marrow from an average of 9.3% to 36.4% . When we administered Tf at 2 or 6 months post-transplant, ≤0.01% human RBCs could be detected, suggesting that late Tf and human cytokine administration cannot rescue erythroid progenitors and that there might be overwhelming inhibitory signals of human erythropoiesis in the chimeric bone marrow.
When we infused mature cross-matched human RBCs in transplanted mice, only 1%–3% (compared with the calculated 10%) of recipient blood was human. These human RBCs circulated for up to 3 days; their disappearance corresponded with an increase in serum total bilirubin. The anti-CD122 antibody directs against the IL-2Rβ chain and targets NK cells and macrophages . Murine NK cells were previously shown to prevent engraftment of human cells in other NOD/SCID strains [29, 30]. We used this antibody as a strategy to deplete functional macrophages post-transplant and minimize erythrophagocytosis in the bone marrow or spleen. However, the addition of anti-CD122 antibody or pretreating the mice with splenectomy did not prolong the life span of human red cells. These results are consistent with those of Fujimi et al.  and suggest that the liver is the major organ of RBC sequestration and that extravascular hemolysis further contributes to the lack of human red cells. Taken together, several reasons explain why human RBCs are detectable at only low levels in these NOD/SCID/IL2Rγnull mice despite leukocyte engraftment that is robust: inhibitory cytokines or bone marrow environment that suppress human erythropoiesis, lack of specific human erythroid growth factors, peripheral clearance, and sequestration.
Our results demonstrate that busulfan preconditioning is sufficient to produce stable and high-level engraftment of human cells in NOD/SCID/IL2Rγnull mice. CD34+-selected cord blood grafts provide superior engrafting potential compared with mobilized peripheral blood stem cells from healthy adults. Importantly, this high-level engraftment can be achieved with low mortality, substantially reducing the number of animals required for experiments requiring long-term follow-up. Finally, supplementation with human Tf allows the detection of human erythrocytes in this chimeric mouse assay, providing a potential model for the study of disorders affecting human red blood cells.
This research was supported by the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases and the National Heart, Lung, and Blood Institute at the NIH. We thank Elizabeth Joyal for collection of the cord blood samples, Martha Kirby for cell sorting, and D. Eric Anderson for mass spectroscopy. This work was supported by the Intramural Research Program of the NIH, National Institute of Diabetes and Digestive and Kidney Diseases.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.