hUCB was collected from the John Radcliffe Hospital, Oxford, UK or provided via the NHS Cord Blood Bank, London and used with informed, written pre-consent and ethical approval from the South Central Oxford C and Berkshire Ethical Committees (# 15/SC/0027) and the Oxfordshire Research Ethics Committee B (#07/H0605/130), in accordance with the Helsinki Declaration of 1975, as revised in 2008.

Mononuclear cells (MNCs; density <1.077g/ml) were isolated by density gradient centrifugation no more than 24 hours after hUCB collection. Human CD133⁺/hCD34⁺ hematopoietic stem and progenitor cells (HSPCs) were enriched by magnetic bead selection using the human CD133/hCD34 direct microbead kits (MACS, Miltenyi Biotec GmbH) and cryopreserved until use (23, 24). Purity was routinely assessed by flow cytometry and only cell isolates with >90% hCD133⁺ or hCD34⁺ cell purity were used for experiments. PBMCs were isolated from leucocyte cones obtained from healthy donors (NHS Blood and Transplant [NHSBT] UK) by LSM1077 (PAA) gradient centrifugation.

NOD,B6.SCID Il2rγ ^-/- Kit^W41/W41 (NBSGW) mice were obtained from the Jackson Laboratory and then bred and housed in the Biomedical Services Unit of the John Radcliffe Hospital (Oxford, UK) in individually ventilated cages. All experiments were performed using protocols approved by the Animal Care and Ethical Review Committee at the University of Oxford, in accordance with the UK Animals (Scientific Procedures) Act 1986 and under the PPL P8869535A. Cells were quantified for injection using Countbright absolute counting beads (Molecular Probes) by flow cytometry (23). HSPCs were injected intravenously at doses of between 1,000-250,000 CD133⁺ cells, suspended in 200μl of IMDM/1%HSA (human serum albumin) into non-irradiated adult (5-16 weeks of age) recipient male or female mice. Mice were regularly bled for human leucocyte reconstitution assessment. 20-22 weeks after HSPC injection, peripheral blood, spleen, thymus and bone marrow were harvested. Engraftment was defined as >0.1% hCD45⁺ cell chimaerism in bone marrow at final harvest (25). A separate group of 8-10 week old male or female mice were injected intravenously with 5x10⁶ human PBMCs suspended in 200μl of RPMI 1650. 4-6 weeks after PBMC injection peripheral blood, spleen and thymus were harvested.

Blood, spleen, both femurs and thymus were harvested. Spleen and thymus were mashed and filtered using a 70µm nylon filter, washed with MACS buffer, and red blood cells lysed (BD Pharmlyse), followed by repeated filtering. Femurs were cleaned, crushed with the back end of a 10ml syringe plunger and filtered through a 70µm nylon filter with MACS buffer to wash out bone marrow cells. 1/50^th of a femur was taken for RBC staining and remaining bone marrow was lysed with RBC lysis buffer (BD Pharmlyse). Cells were then washed and filtered again. 10µl of peripheral blood was taken for RBC staining (no RBC lysis) and 50µl of blood was taken for each of the remaining stains after undergoing RBC lysis with BD Pharmlyse buffer, followed by washing with MACS buffer.

Bone marrow, spleen, thymus and blood single cell suspensions prepared as described above were used for flow cytometry. The antibodies against human and mouse cell surface antigens are detailed in Supplementary Table 1 . Briefly, cells were incubated with a mixture of fluorescently labeled antibodies diluted in FACS buffer for 30 minutes on ice. Cells were washed once, resuspended in FACS buffer and acquired immediately on a BD FACSCanto flow cytometer (BD Biosciences). 7-AAD (eBioscience) was added to the antibody mix to distinguish live and dead cells. Data were analyzed using FACSDiva (BD) and FlowJo software (TreeStar Inc.).

Human tissue samples were obtained with full informed written consent and with ethical approval from the Oxfordshire Research Ethics Committee (REC B), under study number 07/H0605/130. In 2 separate experiments, 14 non-irradiated adult NBSGW mice were injected with CD133⁺ hUCB cells suspended in 200μl of IMDM/1%HSA and 11-13 weeks after humanization, received a 1cm x 1cm allogeneic split-thickness human skin graft harvested from excess abdominal tissue used for reconstructive surgery (as previously described (26)). 6 additional mice, which did not receive skin transplants were observed to 100 days to compare incidence of adverse effects. At the time of skin transplantation, multilineage leucocyte reconstitution can be observed in the peripheral blood (data not shown). Grafts were monitored for clinical features of rejection and mice were sacrificed when these were observed. In the absence of rejection grafts were harvested 100 days after transplantation. Spleens were harvested concurrently and splenocytes, processed (as above) and analyzed by FACS for the frequency of class-switched memory B cells. Additionally, blood was harvested and serum separated from cells following centrifugation at 1300rpm for 10 minutes. IgD and IgG expression was assessed by cytometric bead array (Biolegend Legendplex Human Ig Isotyping Panel (8-plex)) according to the manufacturer’s instructions. Data were analyzed using FACSDiva (BD) and FlowJo software (TreeStar Inc.).

For the phagocyte depletion assays, mice were injected with 50,000 CD34⁺ hUCB HSPCs. On week 10, mice were checked for human leucocyte reconstitution and allocated to the clodronate and PBS-treated groups to provide equal distribution of human leucocyte reconstitution levels. Mice received either 200μl clodronate liposomes (at 5mg clodronate per ml; Liposoma B.V.) or 200μl PBS intraperitoneally. Five (5) days after injection, mice were analyzed for mouse macrophage depletion and human RBC levels in the peripheral blood (27). The bone marrow was then harvested at day 6 post-clodronate injection.

Statistical analyses were performed using GraphPad Prism 5 software using the Mann Whitney test and paired t-test as described in each figure legend. A p value of <0.05 was considered statistically significant. Median values are shown, unless stated otherwise.

We first examined human hematopoietic chimaerism in the peripheral blood following transplantation of hUCB CD133⁺ HSCs (≥1x10³ CD133⁺ cells injected) into non-irradiated NBSGW mice (referred to herein as HSPC-NBSGW mice). Following isolation, >96% of CD133⁺ cells were strongly positive for CD34, and included a small number of the rare but highly potent CD133⁺CD34^- LT-HSCs ( Supplementary Figure 1a ). Successful humanization was demonstrated by a dose-dependent increase in human CD45⁺ cell chimaerism in the bone marrow, spleen and peripheral blood by 20-22 weeks ( Figures 1A, B ; Supplementary Tables 2 and 3 , Supplementary Figure 1B ). Engraftment was greatest in the bone marrow (mean ± SD from 35.0±36.2 to 97.4±1.9%) and spleen (from 32.0±39.4 to 98.5±0.9%) ( Supplementary Table 2 ). Robust engraftment was also present in the peripheral blood (from 12.1±24.7 to 80.0±14.2%), with higher chimaerism than described in comparable irradiated HSPC mice similarly humanized intravenously with hUCB HSCs (28). At all doses administered, reconstitution in the blood demonstrated no signs of declining at 20-22 weeks ( Figure 1B ). Recipient sex did not significantly affect chimaerism in the bone marrow, spleen or peripheral blood ( Figures 1C–G , Supplementary Figure 1C ), however we identified a trend consistent with other immunodeficient strains including NSG mice in which lower levels of human CD45⁺ cell engraftment are seen in the bone marrow of male recipients. Engraftment of a sizeable proportion of Lin^-CD34⁺ and Lin^-CD34⁺CD38^lo/- cells in the bone marrow was seen and suggests a pool of stable, self-renewing HSPCs is maintained for at least 20 weeks ( Figures 2A–D ). Relative engraftment of these cells was greater in the bone marrow of female as compared to male recipients when 5-10x10³ HSPCs were transplanted ( Figures 2C, D ). We previously demonstrated functional self-renewal through engraftment of human leukocytes in NBSGW mice secondarily transplanted with bone marrow cells from HSPC-NBSGW mice (29),

We observed multilineage human leucocyte engraftment in the bone marrow, spleen and peripheral blood ( Figure 3 ; Supplementary Figure 2 ). B cells engrafted with the greatest frequency ( Figures 3A, B ; Supplementary Figures 2A–C , followed by myeloid and T cells ( Supplementary Tables 2 and 3 ). In mice receiving 1x10³ hUCB HSPCs, hCD19⁺ B cell engraftment was higher in females ( Supplementary Figure 2d ). No other sex-specific differences in leucocyte subset engraftment were observed ( Supplementary Figures 2D–F ). We also identified human natural killer cells, which were most populous in the peripheral blood. ( Supplementary Figures 2G, H )

B cell hematopoiesis was assessed by examining the developmental stages of B cells in the bone marrow, spleen and peripheral blood 20-22 weeks after humanization ( Figure 3B ). Within the bone marrow, immature B cells expressing CD10 were most populous (30) ( Figures 3C, D ), contrasting with increasing frequencies of larger CD10^- mature B cells in the spleen and blood ( Figures 3D, E ), supporting early observations of the memory phenotype being associated with greater-sized B cells (31). This correlates with human bone marrow, which contains more CD10⁺ cells (with higher CD10 density) than both hUCB CD34⁺ cells and G-CSF-mobilized peripheral blood (30). Next, we analyzed mature B cell phenotypes. As expected, only a small proportion of CD10^- B cells populated the bone marrow and the majority were naïve, lacking CD27 expression ( Figures 3F–G ). Of these, almost half were CD27^-IgD^- double-negative B cells ( Figure 3G ), a heterogenous population described as early-stage bone marrow cells (32), and more recently as memory precursor, extracellular antibody secreting cell precursor, and atypical/tissue based memory cells, with activated phenotypes in inflammatory diseases (33). As in humans, antigen-inexperienced CD27^-IgD⁺ B cells formed the other major mature B cell population in the bone marrow (34) ( Figure 3G ). Within the spleen a greater proportion of mature B cells was observed and among them both unswitched (CD27⁺IgD⁺) and class-switched memory B cells (CD27⁺IgD^-) ( Figures 3F–G ). This reflected the notion that class switch recombination (CSR) occurs predominantly (although not exclusively) in germinal center (secondary lymphoid organ) B cells, initiated by activation-induced cytidine deaminase (35). To assess the capacity for isotype switching to be stimulated in vivo, we introduced an antigen challenge by transplanting allogeneic human skin onto HSPC-NBSGW mice 11-13 weeks after humanization. At the point of skin allograft rejection (or after 100 days), we analyzed the phenotypes of human leucocytes, splenic B cells and immunoglobulin subtypes in peripheral serum ( Figures 3H–K ; Supplementary Figures 2i–k ). We found a significantly higher frequency of CD27⁺IgD^- memory B cells ( Figure 3I ), together with lower serum IgD and higher IgG1 levels ( Figures 3J, K ) in mice exposed to allogeneic skin transplants compared with mice not exposed. This indicates the potential for specific human B cell responses to be mounted in response to antigen challenge within this model.

Altogether, these findings demonstrate that following transplantation of hUCB CD133⁺ HSPCs a continuous process of human B cell development and functional maturation occurs within the primary and secondary lymphoid organs of HSPC-NBSGW mice.

Having identified long-term engraftment of T cells ( Figure 3A ; Supplementary Figure 2 ), we sought to investigate whether persistence of these cells is also supported by continuous hematopoiesis and thymic development. We first assessed T cell subset profiles within the bone marrow, spleen and blood. Transplantation of high doses (≥50x10³) of HSPCs produced robust engraftment of both CD8⁺ and CD4⁺ T cells, predominantly in the spleen and peripheral blood ( Figures 4A, B ). At these doses we identified the major T cell subtypes within both CD8⁺ and CD4⁺ T cell populations: 1) central memory (Tcm), 2) effector memory (Tem), 3) T effector memory re-expressing CD45RA (TemRA) and 4) naïve (Tn) ( Figures 4C, D ; Supplementary Table 4 ). Tem and Tcm cells were the most populous subtypes, however a significant proportion of CD45RA-expressing cells also persisted long-term ( Figures 4C–F ; Supplementary Figure 3D , Supplementary Table 4 ). We found T cell reconstitution in HSPC-NBSGW mice to more accurately reproduce the average human peripheral blood profiles than in NBSGW mice humanized with peripheral blood mononuclear cells (PBMCs), which fail to reconstitute CD45RA-expressing T cells and are instead composed entirely of Tem and Tcm ( Supplementary Figures 3A–D ).

To identify whether de novo T cell development from transplanted HSPCs occurs in this model, we analyzed human and mouse leucocyte populations in the thymi of recipient mice humanized with HSPCs or PBMCs ( Figure 5 ). The majority of thymic cells were human CD45⁺ leucocytes ( Figures 5A, B ) expressing CD3, together with a small population of CD19⁺ cells ( Supplementary Figure 4A ). The majority of CD3⁺ cells were CD4⁺CD8⁺ double-positive (DP) thymocytes ( Figures 5C, D ) with an average frequency equivalent to those seen in thymus biopsies from human infants (36). In contrast, following humanization with mature PBMCs, no double-positive T cells were identified within the thymus ( Supplementary Figures 4B–C ).

Developing humanized mouse models capable of reconstituting cells of the innate and adaptive immune systems and long-term survival is difficult to achieve, especially in the absence of irradiation and engineered or exogenous cytokine expression. Having identified engraftment of human CD33⁺ myeloid cells ( Figure 3A ; Supplementary Table 2 ; Supplementary Figure 2 ), we assessed subtype reconstitution ( Figure 6 ). At all doses of HSPCs transplanted, myeloid cells engrafted with the greatest frequency in the bone marrow ( Figure 6A ; Supplementary Table 2 ). Within the bone marrow and spleen we identified populations of HLA-DR^hi, HLA-DR^int and HLA-DR^- cells ( Figures 6B, C ). The CD33⁺HLA-DR^hi subset, which includes CD11c⁺ conventional dendritic cells (cDCs), formed the most common subtype in the bone marrow ( Figures 6B–G ; Supplementary Figures 5A, B ). As similarly described in human bone marrow, these cDCs expressed CD33, high levels of CD11c and HLA-DR, and were CD14^- (37) and CD11b^lo/- aiding the distinction from monocyte/macrophage lineage cells ( Figures 6D–F ). The HLA-DR^int fraction, which includes CD14-expressing mature monocytes/macrophages is found predominantly in the blood where it forms approximately half of all myeloid cells ( Figures 6B–F and H ). A less mature CD14⁺HLA-DR^- population, which may include myeloid-derived suppressor cells is also present, especially in the spleen ( Figure 6C ). Together, these data demonstrate successful engraftment and reconstitution of phenotypically distinct subsets of innate myeloid cells, which express molecules of antigen presentation.

Next we investigated engraftment of erythrocytes. In HSPC-NBSGW mice, Glycophorin A analysis (CD235a) demonstrated dose-dependent engraftment of erythroid cells within the bone marrow, however only a small, transient population of RBCs were found in the peripheral blood ( Figures 7A, B , Supplementary Figure 6A ). To determine whether absence from the periphery results from defective erythroid lineage differentiation, we assessed erythroid cell maturity in the bone marrow ( Figures 7C–F ). In addition to a small population of mature erythrocytes ( Figures 7C, D ; Supplementary Figure 6A ), dose-dependent frequencies of erythroid precursors at various stages of development were identified, with no significant discrepancies noted between male and female mice ( Figures 7E, F ; Supplementary Figures 6B–D ). To identify whether the lack of peripheral reconstitution results instead from phagocytosis of human erythroid cells by mouse phagocytes, we assessed survival following phagocyte depletion ( Figures 8A–G ). Using a previously reported technique (27), clodronate liposomes (CloLip) were injected intravenously 11 weeks after humanization, successfully reducing the frequencies of monocytes in the peripheral blood ( Figure 8A ). This was associated with increased numbers of mature erythrocytes in the blood (168-fold compared with control; 335 ± 237 vs 2 ± 1.6 per μl of blood) ( Figures 8B–D ), suggesting that phagocytosis contributes at least in part to the impaired reconstitution of human erythroid cells in this model. While there was no significant difference in the frequencies of early erythroid precursors in the bone marrow following phagocyte clearance, as expected, an increase in reticulocytes and mature erythrocytes was seen (19% (16-27.7%) vs 31.5% (5.2-36.9%) as median (with range), p<0.05) ( Figures 8E–G ).