Development of functional human blood and immune systems in NOD/SCID/IL2 receptor γ chain^null mice

To analyze human immune and hematopoietic development and function in vivo, a number of studies have been tried to reproduce human hematopoiesis in small animal xenotransplantation models.¹  Successful transplantation of human hematopoietic tissues in immune-compromised mice was first reported in late 1980s by using homozygous severe combined immunodeficient (C.B.17-SCID) mice. In the first model of a humanized lymphoid system in a SCID mouse (SCID-hu model), McCune et al simultaneously transplanted human fetal tissues, including fetal liver hematopoietic cells, thymus, and lymph nodes, into SCID mice and induced mature human T- and B-cell development.²  Mosier et al successfully reconstituted human T and B cells by transferring human blood mononuclear cells into SCID mice.³  These initial studies suggested the usefulness of immunodeficient mice for reconstitution of the human lymphoid system from human bone marrow hematopoietic stem cells (HSCs).

After these initial reports, a number of modified SCID models have been proposed to try to reconstitute human immunity.⁴  In addition, recombination activating gene (RAG)–deficient strains have been used as recipients in xenotransplantation: T- and B-cell–deficient Prkdc^scid, Rag1^–/–, or Rag2^–/– mutant mice^5-7  were capable of supporting engraftment of human cells. The engraftment levels in these models, however, were still low, presumably due to the remaining innate immunity of host animals.¹  Nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice have been shown to support higher levels of human progenitor cell engraftment than BALB/c/SCID or C.B.17/SCID mice.⁸  Levels of human cell engraftment were further improved by treating NOD/SCID mice with anti-asialo GM1 (ganglioside-monosialic acid) antibodies⁹  that can abrogate natural killer (NK) cell activity. Recently, NOD/SCID mice harboring either a null allele at the β₂-microglobulin gene (NOD/SCID/β2m^–/–)¹⁰  or a truncated common cytokine receptor γ chain (γc) mutant lacking its cytoplasmic region (NOD/SCID/γ_c^–/–)^11,12  were developed. In these mice, NK- as well as T- and B-cell development and functions are disrupted, because β2m is necessary for major histocompatibility complex (MHC) class I–mediated innate immunity, and because γc (originally called IL-2Rγ chain) is an indispensable component of receptor heterodimers for many lymphoid-related cytokines (ie, IL-2, IL-7, IL-9, IL-12, IL-15, and IL-21).¹³  Injection of human bone marrow or cord blood (CB) cells into these mice resulted in successful generation of human T and B cells. In our hands, efficiencies of CB cell engraftment represented by percentages of circulating human (h) CD45⁺ cells were significantly (2- to 5-fold) higher in NOD/SCID/β2m^–/– newborns than those in adults (F.I., M.H., and L.D.S., unpublished data, April 2003). More recently, transplantation of hCD34⁺ CB cells into Rag2^–/– γc^–/– newborns regenerated adaptive immunity mediated by functional T and B cells,¹⁴  suggesting heightened support for xenogeneic transplants especially in the neonatal period. Efficiency of reconstitution of human hematopoiesis may be, however, still suboptimal in these models because chimerisms of human cells are not stable in each experiment.^11,12,14  Furthermore, there is little information regarding reconstitution of human myeloerythroid components in these xenogeneic models.

Two types of mouse lines with truncated or complete null γc mutant^15-17  have been reported. NOD/SCID/γ_c^–/– and Rag2^–/– γc^–/– mouse strains harbor a truncated γc mutant lacking the intracellular domain,¹⁵  and therefore, binding of γc-related cytokines to each receptor should normally occur in these models.¹⁸  For example, IL-2R with the null γc mutations would be an αβ heterodimer complex with an affinity approximately 10 times lower than that of the high affinity αβγ heterotrimer complex in mice with the truncated γc mutant.¹⁹  γc has also been shown to dramatically increase the affinity to its ligands through the receptors for IL-4, IL-7, and IL-15.^20-23  Previous studies suggested that γc-related receptors including IL-2Rβ chain and IL-4Rα chain could activate janus-activated kinases (JAKs) to some extent in the presence of the extracellular domain of γc, independent of the cytoplasmic domain of γc.^24,25  Thus, in order to block the signaling through γc-related cytokine receptors more completely, we made NOD/SCID mice harboring complete null mutation of γc¹⁶  (the NOD/SCID/IL2rγ^null strain). By using NOD/SCID/IL2rγ^null newborns, we successfully reconstituted myeloerythroid as well as lymphoid maturation by injecting human CB or highly-enriched CB HSCs at a high efficiency. Reconstitution of human hematopoiesis persisted for a long term. The developing lymphoid cells were functional for immunoglobulin (Ig) production and human leukocyte antigen (HLA)–dependent cytotoxic activity. Our data show that the NOD/SCID/IL2rγ^null newborn system provides a valuable tool to reproduce human hemato-lymphoid development.

NOD.Cg-Prkdc^scidIL2rg^tmlWjl/Sz (NOD/SCID/IL2rγ^null) and NOD/LtSz-Prkdc^scid/B2m^null (NOD/SCID/β2m^null) mice were developed at the Jackson Laboratory (Bar Harbor, ME). The NOD/SCID/IL2rγ^null strain was established by backcrossing a complete null mutation at γc locus¹⁶  onto the NOD.Cg-Prkdc^scid strain. The establishment of this mouse line has been reported elsewhere.²⁶  All experiments were performed according to the guideline in the Institutional Animal Committee of Kyushu University.

CB cells were obtained from Fukuoka Red Cross Blood Center (Japan). CB cells were harvested after written informed consent. Mononuclear cells were depleted of Lin⁺ cells using mouse anti-hCD3, anti-hCD4, anti-hCD8, anti-hCD11b, anti-hCD19, anti-hCD20, anti-hCD56, and anti–human glycophorin A (hGPA) monoclonal antibodies (BD Immunocytometry, San Jose, CA). Samples were enriched for hCD34⁺ cells by using anti-hCD34 microbeads (Miltenyi Biotec, Auburn, CA). These cells were further stained with anti-hCD34 and hCD38 antibodies (BD Immunocytometry), and were purified for Lin^–CD34⁺CD38^– HSCs by a FACSVantage (Becton Dickinson, San Jose, CA). Lin^– hCD34⁺ cells (10⁵) or 2 × 10⁴ Lin^–hCD34⁺hCD38^– cells were transplanted into irradiated (100 cGy) NOD/SCID/IL2rγ^null or NOD/SCID/β2m^null newborns via a facial vein²⁷  within 48 hours of birth.

At 3 months after transplantation, samples of peripheral blood, bone marrow, spleen, and thymus were harvested from recipient mice. Human common lymphoid progenitors were analyzed based on the expression of hCD127 (IL-7 receptor α chain) and hCD10 in Lin (hCD3, hCD4, hCD8, hCD11b, hCD19, hCD20, hCD56, and hGPA)^– hCD34⁺hCD38⁺ fraction.^28,29  Human myeloid progenitors were analyzed based on the expressions of hCD45RA and hCD123 (IL-3 receptor α chain) in Lin^-CD10^–CD34⁺CD38⁺ fractions. For the analysis of megakaryocyte/erythroid (MegE) lineages, anti-hCD41a (HIP8), anti-hGPA (GAR-2), anti-mCD41a (MW Reg30), and anti-mTer119 (Ter-119) antibodies were used. Samples were treated with ammonium chloride to eliminate mature erythrocytes, and were analyzed by setting nucleated cell scatter gates. For the analysis of circulating erythrocytes and platelets, untreated blood samples were analyzed by setting scatter gates specific for each cell fraction. Human B lymphoid progenitors were evaluated according to the criteria proposed by LeBien.³⁰ 

Bone marrow cells of recipient mice were stained with anti-hCD34, hCD38, hCD45RO, hCD123, and lineage antibodies. Human HSCs, CMPs, GMPs, and MEPs were purified according to the phenotypic definition^28,29  by using a FACSVantage (Becton Dickinson). One hundred cells of each population were cultured in methylcellulose media (Stem Cell Technologies, Vancouver, BC, Canada) supplemented with 10% bovine serum albumin (BSA), 20 μg/mL steel factor, 20 ng/mL IL-3, 20 ng/mL IL-11, 20 ng/mL Fms-like tyrosine kinase 3 (Flt3) ligand, 50 ng/mL granulocyte-macrophage colony-stimulating factor (GM-CSF), 4 U/ml erythropoietin (Epo), and 50 ng/mL thrombopoietin (Tpo). Colony numbers were enumerated on day 14 of culture.

Tissue samples were fixed with 4% paraformaldehyde and dehydrated with graded alcohol. After treatment with heated citrate buffer for antigen retrieval, paraformaldehyde-fixed paraffin-embedded sections were immunostained with mouse anti-hCD19, anti–human IgA, anti-hCD3, anti-hCD4, anti-hCD8, and anti-hCD11c antibodies (Dako Cytomation, Carpinteria, CA). Stained specimens were observed by confocal microscopy (LSM510 META microscope; Carl Zeiss, Oberkochen, Germany). Image acquisition and data analysis were performed by using LSM5 software. Numerical aperture of the objective lens (PlanApochromat ×63) used was 1.4.

Human Ig concentration in recipient sera was measured by using a human immunoglobulin assay kit (Bethyl, Montgomery, IL). For detection of ovalbumin (OVA)–specific human IgM and IgG antibodies, 5 recipient mice were immunized twice every 2 weeks with 100 μg of OVA (Sigma, St Louis, MO) that were emulsified in aluminum hydroxide (Sigma). Sera from OVA-treated mice were harvested 2 weeks after the second immunization. OVA was plated at s concentration of 20 μg/mL on 96-microtiter wells at 4°C overnight. After washing and blocking with bovine serum albumin, serum samples were incubated in the plate for 1 hour. Antibodies binding OVA were then measured by a standard enzyme-linked immunosorbent assay (ELISA).

Alloantigen-specific human CD4⁺ and CD8⁺ T-cell lines were established according to the method as reported.³¹  After stimulation with an Epstein Barr virus–transformed B lymphoblastoid cell line (TAK-LCL) established from a healthy individual (TAK-LCL) for 6 days, 100 hCD4⁺ T cells or hCD8⁺ T cells were plated with 3 × 10⁴ TAK-LCL cells in the presence of 10 U/mL human IL-2 (Genzyme, Boston, MA), and were subjected to a chromium 51 (⁵¹Cr) release assay. A Limiting number of effector cells and 10^{4 51}Cr-labeled allogeneic target cells were incubated. KIN-LCLs that do not share HLA with effector cells or TAK-LCL were used as negative controls. Cytotoxic activity was tested in the presence or absence of anti-HLA class I or anti–HLA-DR monoclonal antibodies.

NOD/SCID/IL2rγ^null mice lacked mature murine T or B cells evaluated by fluorescence-activated cell sorting (FACS), and displayed extremely low levels of NK cell activity.³¹  This mouse line can survive more than 15 months³¹  since it does not develop thymic lymphoma, usually a fatal disease in the immune-compromised mice with NOD background.³² 

Lin^–hCD34⁺ CB cells contain HSCs, and myeloid and lymphoid progenitors.^28,29  We and others have reported that engraftment of human CB cells, which contain hematopoietic stem and progenitor cells, was efficient in NOD/SCID/β2m^–/– and Rag2^–/–/γ_c^–/– mice, especially when cells were transplanted during the neonatal period.^14,33  We therefore transplanted purified Lin^–hCD34⁺ CB cells into sublethally irradiated NOD/SCID/IL2rγ^null newborns via a facial vein.²⁷ 

We first transplanted 10⁵ Lin^–hCD34⁺ CB cells from 3 independent donors into 5 NOD/SCID/IL2rγ^null newborns, and found that the NOD/SCID/IL2rγ^null newborn system is very efficient for supporting engraftment of human hematopoietic progenitor cells. Table 1 shows percentages of hCD45⁺ cells in these mice 3 months after transplantation. Strikingly, the average engraftment levels were approximately 70% in both the bone marrow and the peripheral blood. Compared with 4 control NOD/SCID/β2m^–/– recipient mice given transplants from the same donors, engraftment levels of hCD45⁺ cells in NOD/SCID/IL2rγ^null mice were significantly higher (Table 1).

Table 2 shows the analysis of human hematopoietic cell progeny in mice that received transplants of human Lin^–hCD34⁺ CB cells. In the peripheral blood, hCD45⁺ cells included hCD33⁺ myeloid, hCD19⁺ B cells, and hCD3⁺ T cells in all mice analyzed (Figure 1A and Table 2). We then analyzed the reconstitution of erythropoiesis and thrombopoiesis in these mice. Anti–human glycophorin A (hGPA) antibodies recognized human erythrocytes, while mTer119 antibodies³⁴  recognized GPA-associated protein on murine erythrocytes, respectively (Figure 1B). Human and murine platelets could also be stained with anti–human and anti–murine CD41a, respectively (Figure 1B). Circulating hGPA⁺ erythrocytes and hCD41a⁺ platelets were detected in all 3 mice analyzed (Figure 1B, right panels). hGPA⁺ erythroblasts and hCD41a⁺ megakaryocytes were detected as 9.5% ± 6.2% (n = 5) and 1.64% ± 0.42% (n = 5) of nucleated bone marrow cells, respectively. Thus, transplanted human Lin^–hCD34⁺ CB cells differentiated into mature erythrocytes and platelets in NOD/SCID/IL2rγ^null recipients.

In all engrafted mice, the bone marrow and the spleen contained significant numbers of hCD11c⁺ dendritic cells as well as hCD33⁺ myeloid cells, hCD19⁺ B cells, and hCD3⁺ T cells (Table 2 and Figure 1C). hCD11c⁺ dendritic cells coexpressed HLA-DR that is essential for antigen presentation to T cells (Figure 1D). In contrast, in the thymus, the majority of cells were composed of hCD3⁺ T cells and rare hCD19⁺ B cells (Table 2).

Figure 2A shows the change in the percentage of circulating hCD45⁺ cells in another set of NOD/SCID/IL2rγ^null newborns injected with 2 × 10⁴ Lin^–hCD34⁺hCD38^– CB cells. Surprisingly, the level of hCD45⁺ cells in the blood was unchanged, and was maintained at a high level even 24 weeks after transplantation. Mice did not develop lymphoid malignancies or other complications. Furthermore, we tested the retransplantability of human HSCs in primary recipients. We killed mice at 24 weeks after the primary transplantation of hCD34⁺ cells, purified 1 to 5 × 10⁴ hCD34⁺ cells from primary recipient bone marrow cells, and retransplanted them into NOD/SCID/IL2rγ^null newborns. In all 3 experiments, secondary recipients successfully reconstituted human hematopoiesis at least until 12 weeks after transplantation, when we killed mice for the bone marrow analysis (Figure 2B). Thus, the NOD/SCID/IL2rγ^null newborn system can support human hematopoiesis for the long term.

The Lin^–hCD34⁺ CB fraction contains early myeloid and lymphoid progenitors as well as HSCs.²⁸  To verify that differentiation into all hematopoietic cells can be initiated from human HSCs in the NOD/SCID/IL2rγ^null newborn system, we transplanted Lin^–hCD34⁺hCD38^-CB cells that contain the counterpart population of murine long-term HSCs,³⁵  and are highly enriched for human HSCs.^36,37  hCD34⁺ CB cells (15%-20%) were hCD38^– (data not shown). Mice given transplants of 2 × 10⁴ Lin^–hCD34⁺hCD38^– cells displayed successful reconstitution of similar proportion of human cells compared with mice reconstituted with 1 × 10⁵ Lin^–hCD34⁺ cells at 12 weeks after transplantation (Table 2). In another experiment, mice injected with 2 × 10⁴ Lin^–hCD34⁺hCD38^– cells exhibited the high chimerism (> 50%) of circulating human blood cells even 24 weeks after transplantation (not shown), suggesting the long-term engraftment of self-renewing human HSCs.

In all mice injected with Lin^–hCD34⁺hCD38^– cells, hGPA⁺ erythroid cells and hCD41a⁺ megakaryocytes were present (not shown). We then tested whether differentiation of Lin^–hCD34⁺hCD38^– HSCs in the NOD/SCID/IL2rγ^null mouse microenvironment can recapitulate normal developmental processes in the human bone marrow. We and others have reported that phenotypically separable myeloid and lymphoid progenitors are present in the steady-state normal bone marrow in both mice^38,39  and humans.^28,29 Figure 2C shows the representative FACS analysis data of recipient's bone marrow cells. In all 3 mice tested, the bone marrow contained the hCD34⁺hCD38^– HSC^36,37  and the hCD34⁺hCD38⁺ progenitor fractions.²⁸  The hCD34⁺hCD38⁺hCD10⁺hCD127 (IL-7Rα)⁺ common lymphoid progenitor (CLP) population²⁹  was detected (Figure 2C, top panels). According to the phenotypic definition of human myeloid progenitors,²⁸  the hCD34⁺hCD38⁺ progenitor fraction was subfractionated into hCD45RA^–hCD123 (IL-3Rα)^lo common myeloid progenitor (CMP), hCD45RA^–hCD123^– megakaryocyte/erythrocyte progenitor (MEP), and hCD45RA⁺hCD123^lo granulocyte/monocyte progenitor (GMP) populations (Figure 2C, bottom panels). We then purified these myeloid progenitors, and tested their differentiation potential. As shown in Figure 2D, purified GMPs and MEPs generated granulocyte/monocyte (GM)– and megakaryocyte/erythrocyte (MegE)–related colonies, respectively, while CMPs as well as HSCs generated mixed colonies in addition to GM and MegE colonies. These data strongly suggest that hCD34⁺hCD38^– human HSCs differentiate into all myeloid and lymphoid lineages tracking normal developmental steps of the steady-state human hematopoiesis within the NOD/SCID/IL2rγ^null mouse bone marrow.

We further evaluated development of the human immune system in NOD/SCID/IL2rγ^null recipients. In the thymus, thymocytes were mostly consisted of hCD3⁺ T cells with scattered hCD19⁺ B cells (Figure 3A-B). This is reasonable since the normal murine thymus contain a small number of B cells in addition to T cells.⁴⁰  Thymocytes consisted of immature hCD4⁺hCD8⁺ double-positive (DP) T cells (Figure 3C) as well as small numbers of hCD4⁺ or hCD8⁺ single-positive (SP) mature T cells (Figure 4A, top panel), while hCD3⁺ human T cells in spleen were mainly constituted of either hCD4⁺ or hCD8⁺ single positive T cells (Figure 4A, bottom panel). These data suggest that normal selection processes of T-cell development may occur in the recipients' thymi.

In the spleen, lymphoid follicle-like structures were seen (Figure 3D-E), where predominant hCD19⁺ B cells were associated with surrounding scattered hCD3⁺ T cells (Figure 3F). Development of mesenteric lymph nodes was also observed, where the similar follicle-like structures consisted of human B and T cells were present (not shown). In the bone marrow and the spleen, nucleated cells in each organ contained hCD34⁺hCD19⁺ pro-B cells, hCD10⁺hCD19⁺ immature B cells, and hCD19⁺hCD20⁺ mature B cells (Figure 4B). Figure 4C shows the expression of human immunoglobulins on hCD19⁺ B cells. A significant fraction of hCD19⁺ B cells expressed human IgM on their surface. A fraction of cells expressing IgD were also observed in the blood and the spleen, suggesting that class switching occurred in these developing B cells. As reported in the Rag2^–/– γc^–/– mouse models,^10,12,14  hCD19⁺IgA⁺ B cells were detected in the bone marrow and the spleen in NOD/SCID/IL2rγ^null recipients. We then evaluated concentrations of human immunoglobulins in sera of mice given transplants of human Lin^–hCD34⁺ CB cells by ELISA. In all sera from 3 NOD/SCID/IL2rγ^null recipients, a significant amount of IgG (257 ± 76 μg/mL) and IgM (600 ± 197 μg/mL) were detectable, whereas sera from the control NOD/SCID/β2m^–/– mice contained lower levels of IgM (76 ± 41 μg/mL) and little or no IgG (Table S1, available on the Blood website; see the Supplemental Table link at the top of the online article). These data collectively suggest that class-switching can effectively occur in NOD/SCID/IL2rγ^null mice.

The intestinal tract is one of the major sites for supporting host defense against exogenous antigens. Since bone marrow and spleen hCD19⁺ B cells contained a significant fraction of cells expressing IgA, we tested whether reconstitution of mucosal immunity could be achieved in the NOD/SCID/IL2rγ^null recipients. Immunohistologic analyses demonstrated that the intestinal tract of recipient mice contained significant numbers of cells expressing human IgA in addition to hCD3⁺ T cells (Figure 3G-I). Thus, human CB cells could reconstitute cells responsible for both systemic and mucosal immunity in the NOD/SCID/IL2rγ^null newborn system.

Five NOD/SCID/IL2rγ^null mice reconstituted with 3 independent human CB samples were immunized twice with ovalbumin (OVA) at 3 months after transplantation. Two weeks after immunization, sera were collected from these immunized mice, and were subjected to ELISA to quantify OVA-specific human IgG and IgM. As shown in Figure 5A, significant levels of OVA-specific human IgM and IgG were detected in all serum samples from immunized mice, but not in samples from nonimmunized engrafted mice. Thus, the adaptive human immune system properly functioned in the NOD/SCID/IL2rγ^null strain to produce antigen-specific human IgM and IgG antibodies.

We next tested the alloantigen-specific cytotoxic function of human T cells developed in NOD/SCID/IL2rγ^null recipients. hCD3⁺ T cells isolated from the spleen of NOD/SCID/IL2rγ^null recipients were cultured with allogeneic B-LCL (TAK-LCL). We established 8 hCD4⁺ and 10 hCD8⁺ T-cell clones responding LCL-specific allogeneic antigens. We then estimated cytotoxic activity of these T-cell clones in the presence or absence of anti–HLA-DR and anti–HLA class I antibodies. We randomly chose 3 each of CD4 and CD8 clones for further analysis (Figure 5B). A ⁵¹Cr release assay revealed that both hCD4⁺ and hCD8⁺ T cell clones exhibited cytotoxic activity against allogeneic TAK-LCL, whereas they showed no cytotoxicity against KIN-LCL, a cell line not sharing HLA classes I or II with TAK-LCL. Cytotoxic activity of hCD4⁺ and hCD8⁺ T cell clones was significantly inhibited by the addition of anti–HLA-DR and anti–HLA class I antibodies, respectively. These data clearly demonstrate that human CB-derived T cells can exhibit cytotoxic activity in an HLA-restricted manner.

Xenogeneic transplantation models have been extensively used to study human hematopoiesis in vivo.^1,41,42  In the present study, we describe a new xenogeneic transplantation system that effectively supports human hemato-lymphoid development of all lineages for the long term.

NOD/SCID/IL2rγ^null newborns exhibited very efficient reconstitution of human hematopoietic and immune systems after intravenous injection of a relatively small number of CB cells. In our hands, NOD/SCID/IL2rγ^null newborns displayed a significantly higher chimerism of human blood cells compared with NOD/SCID/β2m^–/– newborns under an identical transplantation setting (Table 1). This result directly shows that the IL2rγ^null mutation has a merit on human cell engraftment over the β2m^–/– mutation.

One of the critical problems in the NOD/SCID strain for the use of recipients is that this mouse line possesses a predisposition to thymic lymphoma due to an endogenous ectropic provirus (Emv-30).³²  Because of this, NOD/SCID and NOD/SCID/β2m^–/– mice have the short mean lifespan of 8.5 and 6 months, respectively, while NOD/SCID/IL2rγ^null mice did not develop thymic lymphoma surviving more than 15 months,³¹  which allows a long-term experimentation.

In our study, NOD/SCID/IL2rγ^null newborns injected with 10⁵ hCD34⁺ CB cells via a facial vein consistently displayed high levels of chimerism of human hematopoiesis (50%-80%; Table 1). This model is comparable to, or may be more efficient than the Rag2^–/– γc^–/– newborn model where intrahepatic injection of 0.4 to 1.2 × 10⁵ hCD34⁺ CB cells generated variable levels of chimerism of human cells (5%-65%).¹⁴  This slight difference of engraftment efficiency, however, could reflect the homing efficiency of HSCs by each injection route. The NOD/SCID/IL2rγ^null newborn model might be more efficient than the NOD/SCID/γ_c^–/– adult model in which the majority of recipients showed approximately 30% chimerism of human cells after transplantation of 5 × 10⁴ hCD34⁺ CB cells.¹¹  Although we did not test NOD/SCID/γ_c^–/– newborns side by side in this study, we have found that the engraftment level of hCD34⁺ cells of human acute myelogenous leukemia is approximately 3-fold higher in newborns than adults in the NOD/SCID/IL2rγ^null strain (F.I., T.M., S.Y., M.Y., M.H., K.A., and L.D.S., manuscript in preparation). Therefore, it remains unclear whether the IL2rγ^null mutation has a significant advantage over the truncated γc mutation¹¹  for human cell engraftment. It is still possible that the improved engraftment efficiencies in the NOD/SCID/IL2rγ^null newborn system as compared to those in the NOD/SCID/γ_c^–/– adult system reflect the age-dependent maturation of the xenogeneic barrier.

The Rag2^–/– γc^–/– newborn and NOD/SCID/γ_c^–/– adult models have provided definitive evidences that functional T cells, B cells, and dendritic cells can develop from hematopoietic progenitor cells in immunodeficient mice. Class-switching of immunoglobulin in CB-derived B cells properly occurred in the NOD/SCID/IL2rγ^null but not in the NOD/SCID/β2m^–/– newborns (Table S1), further confirming the advantage of the NOD/SCID/IL2rγ^null model. We also showed that, consistent with a previous report using the Rag2^–/– γc^–/– model,¹⁴  human T and B cells developed in NOD/SCID/IL2rγ^null mice are capable of mounting antigen-specific immune responses. Interestingly, human T and B cells migrated into murine lymphoid organs and into the intestinal tissues to collaborate in forming lymphoid organ structures. Furthermore, we found that IgA-secreting human B cells can develop in the murine intestine, suggesting that human mucosal immunity could be generated. Thus, the cellular interaction and the lymphocyte homing could occur at least to some extent across the xenogenic barrier in this model. It is also of interest that developing human cells in the thymus displayed normal distribution of SP and DP cells (Figure 4A), and that mature human T cells displayed cytotoxic functions in an HLA-dependent manner (Figure 5B). This suggests that positive and/or negative selection of human T cells could occur in NOD/SCID/IL2rγ^null recipients. Thymic epithelial cells in recipients reacted with anti-murine but not anti-human centromere probes in a FISH assay (F.I. and M.H., unpublished data, September 2004), confirming their recipient's origin. Thus, it remains unclear how these human T cells effectively educated and developed in murine thymus. It is also possible that human mature T cells developed by extrathymic education and selection.

Our data directly show that the most primitive hCD34⁺hCD38^– CB cells are capable of generating the human myeloerythroid system in addition to the immune system in the NOD/SCID/IL2rγ^null recipients. The emergence of circulating hCD33⁺ myelomonocytic cells after transplantation of human CB cells has been reported in the NOD/SCID/β2m^–/– newborn³³  and the NOD/SCID/γ_c^–/– adult¹¹  systems. Development of human erythropoiesis, however, has not been obtained in previous models, although it has been reported that NOD/SCID mice can support terminal maturation of hCD71⁺ erythroblasts that were induced ex vivo from human HSCs by culturing with human cytokines.⁴³  We showed for the first time that human erythropoiesis and thrombopoiesis can develop in mice from primitive hCD34⁺hCD38^– cells, as evidenced by the presence of erythroblasts and megakaryocytes in the bone marrow and of circulating erythrocytes and platelets in NOD/SCID/IL2rγ^null recipients. It is important to note that the hCD34⁺hCD38^– CB HSC population generated myeloid- and lymphoid-restricted progenitor populations such as CMPs, GMPs, MEPs, and CLPs in the bone marrow (Figure 2E-F). Thus, the NOD/SCID/IL2rγ^null microenvironment might be able to support physiological steps of myelopoiesis and lymphopoiesis initiating from the primitive HSC stage.

In summary, we show that the NOD/SCID/IL2rγ^null newborn system efficiently supports hemato-lymphoid development from primitive human HSCs, passing through physiological developmental intermediates. It also can support development of human systemic and mucosal immunity, and therefore may be useful to use human immunity to produce immunoglobulins or experimental vaccines. The NOD/SCID/IL2rγ^null newborn system might also serve as an efficient tool for understanding malignant hematopoiesis in humans, since the analysis of human leukemogenesis has mainly been dependent upon the NOD/SCID adult mouse system.^44-46  Our model might also be useful to reproduce the transforming process of human hematopoietic cells, as transplanted murine hematopoietic progenitor and stem cells can develop leukemia by transducing oncogenic fusion genes in syngeneic mouse models.^47,48  Thus, the use of this system should open a more efficient way to analyze normal and malignant human hematopoiesis.

We thank Bruce Gott and Lisa Burzenski for excellent technical assistance, and Dr Sato and other staff at Fukuoka Cord Blood Bank for preparation of CB.