Development of a Human Adaptive Immune System in Cord Blood Cell-Transplanted Mice

See allHide authors and affiliations

Science  02 Apr 2004:
Vol. 304, Issue 5667, pp. 104-107
DOI: 10.1126/science.1093933


Because ethical restrictions limit in vivo studies of the human hemato-lymphoid system, substitute human to small animal xenotransplantation models have been employed. Existing models, however, sustain only limited development and maintenance of human lymphoid cells and rarely produce immune responses. Here we show that intrahepatic injection of CD34+ human cord blood cells into conditioned newborn Rag2–/–γc–/– mice leads to de novo development of B, T, and dendritic cells; formation of structured primary and secondary lymphoid organs; and production of functional immune responses. This provides a valuable model to study development and function of the human adaptive immune system in vivo.

Biomedical research in humans is restricted largely to in vitro assays that lack the components and complexity of a living organism. To overcome this limitation, substitute in vivo models have been developed in which human hematopoietic cells and tissues are transplanted into mice that are compromised in their capacity to reject xenogenic grafts. Engraftment was first reported after transfer of mature human peripheral blood leukocytes in severe combined immunodeficient mice (hu PBL-SCID mice) (1) and transplantation of blood-forming fetal liver cells, fetal bone, fetal thymus, and fetal lymph nodes in SCID mice (SCID-hu mice) (2, 3). Subsequently, some level of human hematopoietic development was achieved by transplantation of blood-forming cells in NOD/SCID, NOD/SCIDβ2m–/–, or NOD/SCIDγc–/– mice (47). However, transfer of human cells in immunodeficient mice has, so far, not appeared to result in the de novo formation of a functional human adaptive immune system (1, 716).

The liver contributes to perinatal hematopoiesis, and the hemato-lymphoid system expands most significantly during the first weeks of life. Thus, we reasoned that human hematopoietic stem and progenitor cells transplanted into the liver of immunodeficient newborn mice might find better conditions to engraft, expand, and reconstitute a human immune system. We transplanted newborn Rag2–/–γc–/– mice, a mutant strain that lacks B, T, and NK cells (17, 18), intrahepatically (i.h.) with CD34+ cord blood cells (19). Mice were subsequently analyzed between weeks 4 and 26 of age, and human CD45+ hematopoietic cells were detected in all animals (Fig. 1A). An increase in splenic and thymic cellularity was detectable, and all mice beyond 8 weeks developed mesenteric lymph nodes, several within size and cellularity of wild-type controls (Fig. 1B).

Fig. 1.

Human hematopoietic cell engraftment, organ enlargement, and T cell development in transplanted animals. (A) Bar graphs represent percent of human CD45+ cells in bone marrow and spleen of consecutively analyzed mice (n = 30) at indicated weeks after transplantation. Bars are broken up into human CD45+ CD3+ T cells, CD45+CD19+ B cells, and CD45+ CD19 CD3 cells. Letters X, Y, Z indicate numbers of CD34+ cells transplanted per animal: X, 3.8 to 7 × 104; Y, 7 to 9 × 104; Z, 9 to 12 × 104. *Indicates animal of which organs are shown in (B). (B) Representative photographs of spleen, thymus, and mesenteric lymph nodes of 19-week-old BALB/c, transplanted Rag2–/–γc–/–, and Rag2–/–γc–/– control mice. (C) Contour plots show representative human thymocyte staining profiles of a young and older animal. Histogram shows TCRαβ expression on gated double-negative (DN), double-positive (DP), and single-positive CD4 (CD4 SP) and CD8 (CD8 SP) thymocytes from the same animal. (D) Representative mesenteric lymph node profile for CD4 and CD8, and CCR7 versus CD45RA, expression on gated CD3+ cells. For comparison, same staining on cord blood cells is shown.

Most CD19+ cells in bone marrow (BM) of engrafted mice were negative for surface immunoglobulin M (IgM) and CD20 expression, whereas spleen, lymph node, and blood CD19+ cells expressed these antigens (fig. S1A). This was consistent with generation of B cells in BM and subsequent migration to spleen and lymph nodes. Human IgM was detectable in serum of young transplanted animals and increased in most over time; IgG was detected in older animals, demonstrating class switching of Ig isotypes (fig. S1B). Ig-producing cells were located in BM and spleen and correlated closely with numbers of CD19+CD27+CD138+ plasma cells (fig. S1, C and D). Thus, in contrast to transplanted NOD/SCID mice where human B cells fail to produce Ig (13, 14), full B cell maturation occurred in reconstituted Rag2–/–γc–/– mice.

All thymi contained double-positive, as well as CD4 and CD8 single-positive, T cells in 1:1 to 4:1 ratios, with thymi of young mice containing fewer mature thymocytes than thymi of older mice (Fig. 1C). αβ T cell receptor (TCRαβ) was upregulated normally during transition from double-negative to single-positive stages (Fig. 1C), and typical thymic cortex and medulla structures were apparent (fig. S2). Some thymi in animals beyond 25 weeks still contained >70% double-positive cells, indicating continuous T cell generation over time. Interestingly, CD25 and Foxp3 expression on some CD4+ thymocytes suggested that human regulatory T cells might also be generated via thymic development (fig. S3, A and B) (20, 21).

Mature T cells with a broad Vβ repertoire were detectable in thymus, spleen, mesenteric lymph nodes, and bone marrrow (Fig. 1D; fig. S3C). More than 40% of T cells displayed a naïve phenotype as assessed by CD45RA/CCR7 expression (22) (Fig. 1D). To test whether human T cells in mouse secondary lymphoid organs underwent similar numbers of post-thymic cell divisions as T cells in human newborn blood, TCR-rearrangement excision circles (TRECs) were measured. TREC levels were highest in mouse thymi, whereas somewhat reduced levels were detected in mouse spleen and lymph nodes, which correlated closely with TREC levels in nontransplanted human cord blood (fig. S3D). Thus, human T cells in mouse peripheral tissues appeared to have undergone similar numbers of divisions as T cells in cord blood. Together these data imply that in the thymi of reconstituted Rag2–/–γc–/– mice human T cells with a broad repertoire develop over at least 6 months and home to secondary lymphoid organs without being massively activated.

The proliferative capacity of T cells in response to mouse or human major histo-compatibility complex (MHC) antigens was next tested using mixed lymphocyte reactions (MLRs). Human T cells isolated from mouse lymph nodes and spleen proliferated vigorously when stimulated with human allogeneic dendritic cells (DCs) but weakly, or not at all, when stimulated with autologous human DCs (fig. S4A). This indicates that developing T cells had undergone some level of selection on human MHC, possibly within the mouse thymus. Reactivity to mouse DCs was, generally, low, possibly reflecting suboptimal xenogenic cell interactions (fig. S4B). However, the small difference in response to host BALB/c (Rag2–/–γc–/–) versus mismatched C57BL/6 DCs would also be consistent with the possibility that some thymic selection had occurred on mouse MHC.

Human DCs can be divided into CD11c+ DCs and CD11cCD123+ plasmacytoid, type I interferon producing pre-DCs (23, 24). Both subsets were present in bone marrow, spleen, liver, and, at low levels, thymus and lymph nodes of reconstituted mice (Fig. 2, A and B). Upon maturation ex vivo, CD11c+ cells displayed typical DC morphology and became potent stimulators of allogeneic T cells (Fig. 2, C and D; fig. S5B). CD11c cells, stimulated ex vivo with influenza virus, produced high amounts of α-interferon (IFNα) (Fig. 2E). Therefore, reconstituted animals supported development of both functional DCs and plasmacytoid pre-DCs.

Fig. 2.

Functional human dendritic cell subsets develop in transplanted animals. (A) Identification of human CD11cCD123+ plasmacytoid pre-DC and CD11c+ DC in bone marrow, spleen, and liver of an 11-week-old transplanted animal. (B) Histograms show plasmacytoid pre-DC and DC associated maker expression (solid lines) and isotype controls (dashed lines) on CD11cCD123+ and CD11c+ gated cells in bone marrow. (C) Sorted bone marrow CD11c+ cells, activated with lipopolysaccharide (LPS) and granulocyte-macrophage colony-stimulating factor (GM-CSF) show typical dendritic cell morphology. (D) CD11c+ but not CD19+ bone marrow cells induce strong proliferation of allogeneic peripheral blood T cells. Histogram shows overlay of CD3+ gated cells stimulated with CD11c+ cells (solid line, closed histogram) or CD19+ cells (dashed line). T cell proliferation was evaluated at day six of culture. One of two experiments depicted. CFSE, carboxyfluorescein diacetate succinimidyl ester. (E) CD123+BDCA-4+ bone marrow cells produce high amounts of IFNα upon viral stimulation. CD123+BDCA-4+CD45+ (1) and CD123BDCA-4CD45+ (3) cells from bone marrow of a transplanted animal, and BDCA-4+CD19CD14 cells from peripheral blood of a healthy adult (2) were sorted and stimulated overnight with influenza virus (3.5 × 104 cells each). IFNα in supernatants was evaluated by enzyme-linked immunosorbent assay (ELISA). Representative experiment of three.

Spleens from animals older than 16 weeks possessed white pulp-like structures, consisting of central arterioles surrounded by human T and B cells (Fig. 3, A and B). Although B and T cell–deficient mice lack follicular dendritic cells (FDCs), immunoreconstitution with mouse BM induces their formation, likely from resident (host-derived) nonhematopoietic cells (25). This has been shown to be dependant on the presence of lymphotoxin α–expressing B cells (25). As expected, no human FDCs were detectable in reconstituted animals, although mouse FDCs (FDC-M1+) were induced (Fig. 3B). Furthermore, in some cases, typical germinal center–like structures could be observed (Fig. 3C). These findings provide strong evidence that functional interactions between immune cells were occurring, both within the engrafted human cell populations and across the species barrier.

Fig. 3.

Spleens of Rag2–/–γc–/– mice transplanted with human CD34+ cord blood cells develop structured white pulp. (A) Focal accumulation of T (CD3+) and B (CD20+) cells around arterioles. (B) Close-up of (A) showing white pulp containing predominantly B cells (CD20+) and T cells (CD3+), as well as naive B cells (CD23+) and mouse FDCs (FDC-M1+). Human FDCs (CD21+) were not detected. (C) Close-up showing a germinal center–like structure with accumulation of highly proliferating CD20+bcl6+ki67+ B cells surrounded by CD20+bcl6bcl2+ B cells and CD3+ T cells, in this case in a vaccinated animal. Representative analysis, 26 weeks [(A) and (B)] and 25 weeks (C) after reconstitution.

To more directly test the immune response after reconstitution, mice were vaccinated with tetanus toxoid (TT) or were infected with Epstein-Barr virus (EBV). In three animals vaccinated with TT at 8 weeks, no specific human IgG antibodies could be detected. It is possible that failure to respond at this stage may have been due to the immaturity of the immune system. In contrast, when TT vaccinations were started at 12 to 17 weeks of age, three of five mice produced measurable anti-TT IgG antibodies (table S1) and memory B cells could be detected in lymph nodes (table S2). Although antibody levels were lower than those achieved in human adults, anti-TT IgG to total IgG ratios compared closely (table S1).

Engrafted animals were next infected with increasing doses of EBV and were analyzed 5 to 10 weeks later (Fig. 4; table S3). In all animals, EBV was detectable by polymerase chain reaction (PCR) in BM, spleen, and lymph node B cells (26). Two animals, infected with the highest EBV doses, developed LMP1+ B cell proliferation in spleen, liver, and kidney (Fig. 4A; table S3), whereas five animals infected with lower EBV doses did not develop obvious B cell proliferation. In the latter, spleen T cells increased substantially (Fig. 4; table S3). Three of these five mice and both mice with B cell proliferation developed an inverse CD4+:CD8+ T cell ratio (Fig. 4B; table S3). Moreover, CD8+ T cells from EBV-infected animals proliferated strongly when stimulated with autologous EBV-transformed B cells in vitro (Fig. 4C). These data suggest that upon in vivo challenge with moderate infectious doses, T cell responses were initiated that could control EBV, at least for the time periods observed. It will be important to next determine the type of EBV infection, as well as the duration, epitope, and MHC specificity of the T cell response.

Fig. 4.

EBV infection of CD34+ cord blood reconstituted Rag2–/–γc–/– mice. (A) Spleen histology showing CD20+LMP1+ B cell proliferation in an animal infected with high doses of EBV five weeks earlier. (B) Representative fluorescence-activated cell sorter (FACS) analysis of spleen and lymph node T and B cells in a noninfected sibling control and in EBV infected animals that did and did not develop B cell proliferation. Plots show characteristic T cell increase and inversion of CD4+:CD8+ ratios. (C) Lymph node T cells of EBV-infected animals proliferate when stimulated ex vivo with autologous EBV-transformed B cells. Lymph node T cells from noninfected siblings do not proliferate when stimulated under the same conditions. Proliferation was measured by CFSE dilution at day five of cultures. Representative experiment of two.

Taken together, reconstitution of newborn Rag2–/–γc–/– mice with cord blood CD34+ cells was found to lead to ortotopic de novo generation and maturation of human DCs, B cells, and T cells with a broad repertoire. Although it remains to be clarified how human T cell selection on mouse thymic background occurs, the T cells generated discriminate self from allogeneic MHC. Together, human cells were able to collaborate in forming lymphoid organ structures and induce the differentiation of mouse FDCs, thus providing evidence for an unexpectedly robust cellular interaction across xenogenic barriers. These findings provide a technically straightforward in vivo model with which it will be possible to characterize development, homeostasis, and functional cooperation of the human adaptive immune system. Furthermore, the model should provide a valuable tool to study pathogens that specifically target the human immune system and test potential therapeutic interventions.

Supporting Online Material

Materials and Methods

Tables S1 to S3

Figs. S1 to S5


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article