Long-term human immune system reconstitution in non-obese diabetic (NOD)-Rag (–)-γ chain (–) (NRG) mice is similar but not identical to the original stem cell donor

Abstract

The murine immune system is not necessarily identical to it human counterpart, which has led to the construction of humanized mice. The current study analysed whether or not a human immune system contained within the non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mouse model was an accurate representation of the original stem cell donor and if multiple mice constructed from the same donor were similar to one another. To that end, lightly irradiated NRG mice were injected intrahepatically on day 1 of life with purified cord blood-derived CD34⁺ stem and progenitor cells. Multiple mice were constructed from each cord blood donor. Mice were analysed quarterly for changes in the immune system, and followed for periods up to 12 months post-transplant. Mice from the same donor were compared directly with each other as well as with the original donor. Analyses were performed for immune reconstitution, including flow cytometry, T cell receptor (TCR) and B cell receptor (BCR) spectratyping. It was observed that NRG mice could be ‘humanized’ long-term using cord blood stem cells, and that animals constructed from the same cord blood donor were nearly identical to one another, but quite different from the original stem cell donor immune system.

Introduction

Mice, the most common research model in immunology, although inexpensive, do not necessarily reflect the human immune system in terms of its interaction with infectious agents of human origin or environmental factors. Therefore, a major challenge for the field is to develop a model that functionally mimics human immune system development and function, is relatively inexpensive and can be manipulated to explore the intricacies of the human immune response. The model analysed in this study built upon work reported in the literature developing humanized mice [1–3], and expanded these studies to include long-term reconstituted humanized animals. During the past several years a number of novel humanized mouse strains have been developed in an attempt to provide model systems to address some of these issues, although there is some debate as to whether one model is better than another. Many of these models have utilized purified umbilical cord blood CD34⁺ stem cells injected into newborn (triply) immunodeficient mice which then reconstitute the animal with some or all parts of the human innate and adaptive immune systems. There have been questions as to whether a truly functional (as opposed to phenotypic) human immune system develops, whether it represents a diverse and properly restricted adaptive immune system [focused upon recognition of human leucocyte antigen (HLA) molecules], and if all parts of the innate immune system are intact. Data have been published to support and refute each of these claims [4–9].

The intention of the current study was to determine if the humanized immune system contained within non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice constructed from a single stem cell donor, as measured by flow cytometry, T cell receptor (TCR) spectratyping and B cell receptor (BCR) analyses, was an accurate representation of the immune system found in the original stem cell donor, and how different transplant recipients compared to each other. We sampled the same mice serially over a time-period of 1 year or longer. It was observed that NRG mice could be ‘humanized’ long term using cord blood stem cells, and that animals constructed from the same cord blood donor were nearly identical to one another but quite different from the original stem cell donor's immune system.

Materials and methods

Mice

Humanized mice were created using a variation of the NRG model as described in the literature . Briefly, female mice homozygous for both the Rag1^null and interleukin (IL)-2rγ^null (common gamma chain null) mutations were bred with male mice homozygous for the Rag1 knock-out mutation and hemizygous for the X-linked IL-2rγ^null mutation, as obtained from Jackson Laboratories (Bar Harbor, ME, USA). Mice were bred at the University of Arizona animal facilities, an AAALAC (Association for Assessment and Accreditation of Laboratory Animal Care)-approved institution.

Cord blood collections and CD34 purifications

Cord blood (CB) samples were collected, processed and cryopreserved as described previously [11] under an Institutional Review Board (IRB)-approved protocol. CD34⁺ CB stem cells were positively enriched using magnetic bead separation (StemCell Technologies, Vancouver, BC, Canada), as per the manufacturer's instructions, to 95% or greater purity.

Transplantation

Newborn mice were irradiated in a 3–4-h interval (2 × 2 Gy) using a Caesium or Cobalt source. At 4–12 h post-radiation, mice were transplanted intrahepatically with CD34⁺ (> 95% purity) human cord blood stem cells [30 μl in phosphate-buffered saline (PBS), 0·5–4 × 10⁵ cells/mouse] collected from healthy, disease-free mothers and infants. The mice were anaesthetized, the liver was visualized through the skin, and cord blood stem cells were injected directly into the organ. This procedure was performed by the Experimental Mouse Shared Service (EMSS) of the Arizona Cancer Center at the University of Arizona. Mononuclear cells from the cord blood collection used for transplantation were frozen and stored for later use in future experiments. Engraftment of human cells was ascertained by flow cytometric [fluorescence activated cell sorter (FACS)] analyses of peripheral blood obtained from each transplanted mouse.

FACS analysis

The humanized mice were characterized 90–120 days post-transplant for the presence of human immune cells, as determined by flow cytometry, using heparinized blood obtained from the submandibular pouch (i.e. presence of CD45⁺, CD14⁺, CD3⁺, CD19⁺, CD16⁺, CD11c⁺ populations, etc.) of each animal. Samples were analysed either with a FACScan flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA), with single laser emitting at 488 nm, an LSRII using lasers emitting at 488 nm and 532 nm. Data were analysed and displayed with FacsDiva software or Cytlogic. A minimum of 10 000–20 000 gated events were analysed for each sample. Antibodies were obtained from Biolegend (San Diego, CA, USA) and BD Biosciences.

Quantitative human T and B cell repertoire analysis

Human T and B cell repertoires of donor cells and reconstituted human cells obtained from mouse bone marrow and spleen were analysed quantitatively for T cell receptor beta chain (TCR-BV) and immunoglobulin (Ig)M-heavy chain (IgM-VH) gene expression profiles by spectratyping. For quantitative T cell receptor repertoire analysis, the samples were processed and analysed based on the method described earlier [12]. Briefly, complementary DNA of the total RNA isolated from cells were prepared and the copy numbers of each BV gene families were assessed by real-time polymerase chain reaction (PCR) using BV gene-specific forward and beta-constant gene-specific reverse primers along with constant gene-specific 5′-cyanin 5 (Cy5)-labelled TaqMan probe. Amplified PCR products that contained different lengths for each BV family were resolved through capillary electrophoresis. The intensities measured for each fragment length within the BV family and absolute copy number measured for that family were used to calculate the relative concentration of BV peaks across all 24 TCR-BV families. Similarly for BCR analysis, in particular IgM-VH family genes, quantitative spectratyping was performed and measured and the relative concentration of each VH family genes and the CDR3-length fragment distribution (peak distribution) was analysed for TCR-BV spectratyping [12]. The forward primers of eight different VH gene families and μ-chain-specific reverse primer and probe were adopted from Lim et al. [13], with minor modifications.

Results

Characterization of long-term human engraftment

NRG mice were transplanted with CB CD34⁺ cells on day 1 of life, as described in Materials and methods. At 90–120 days post-transplant mice were analysed for the presence of human cells in the peripheral blood. As shown in Fig. 1, high levels of human CD3⁺ T cells and CD19⁺ B cells were present at day 100 post-transplant and thereafter. These mice were ones that had survived for at least 1 year or longer post-transplant, and were analysed further. The mice were engrafted from a single cord blood stem cell donor, allowing for analysis of variability in the creation of these ‘syngeneic’ animals, with all animals being derived from a single cord blood donor. T cell numbers increased over time while B cells decreased, with both populations reaching plateau levels of 80 and 10% of peripheral blood mononuclear cells, respectively. There did not appear to be any significant phenotypic differences between mice reconstituted with stem cells from the same donor or from different donors.

Fig. 1.

T and B cell engraftment versus time. Non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice were transplanted on day 1 of life, as described in Materials and methods. Animals were analysed by flow cytometry monthly for the presence of T (CD3⁺) and B (CD19⁺) cells in peripheral blood. Each line represents an independent animal. The blue bar represents antigen levels observed in the stem cell donor. Each line represents an individual animal constructed from a single cord blood donor.

All engrafted animals were kept in microisolator cages under specific pathogen-free conditions and fed sterilized food and water. Therefore, we expected to find that all human T cells present in the animals were of the naive subset (i.e. expressing CD45RA), which is exactly what was observed for the first 4–6 months following engraftment. Subsequently, as shown in Fig. 2, T cells expressing the memory phenotype (i.e. CD45RO⁺) dominated in the peripheral blood of the humanized mice. It remains to be determined if these are truly memory T cells (and if they are central or effector memory cells), or if the phenotypic change is a result of homeostatic proliferation. If this observation was due to the latter possibility, why it occurred so late in time after engraftment is unknown, as cellularity was generally constant over this time-period. The change in phenotype was seen in both CD4⁺ and CD8⁺ T cells.

Fig. 2.

T cell subsets. Non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice were transplanted on day 1 of life, as described in Materials and methods. Animals were analysed by flow cytometry monthly for the expression of CD45RA and CD45RO on helper (CD4⁺) and cytotoxic (CD8⁺) T cells in peripheral blood. Each line represents an independent animal. The blue bar represents antigen levels observed in the stem cell donor. Each line represents an individual animal constructed from a single cord blood donor.

Immune repertoire analyses

The diversity of the TCR and BCR repertoires was investigated (Figs 5). It was observed that all humanized NRG mice expressed diverse BCR and TCR repertoires. In addition, mice constructed with stem cells from the same donor expressed similar if not identical antigen receptor spectratypes. However, the repertoires were different from those observed in the cells of the cord blood donor, particularly for the T cells.

Fig. 5.

B cell receptor (BCR) analyses. Non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice were transplanted on day 1 of life, as described in Materials and methods. Animals were analysed for expression of BCR families as described using both bone marrow and spleen tissues, and compared to the original cord blood stem cell donor. Additional spectratyping analyses were also performed (b). Results from a single representative mouse are shown at the time of killing (day 328).

It was observed that the TCR BV family gene usage of the donor (CD34-negative cells, Fig. 3a) was similar to the TCR BV gene usage observed in healthy peripheral blood mononuclear cells (PBMCs), as reported previously [12]. Human T cell reconstitution after cord blood CD34⁺ stem cell transplantation into the mice showed similar levels of TCR BV family gene usage to those found in CD34-negative donor cells, although certain TCR BV families (BV04, BV06a, BV10, BV12 and BV28) were preferentially over-represented after transplantation. Reconstituted human T cells after transplantation in mice immune compartments (i.e. bone marrow and spleen) were similar in magnitude of TCR BV gene usage (Fig. 3b) and the BV-family peak distributions (Fig. 4b,c).

Fig. 3.

T cell receptor (TCR) analyses. Non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice were transplanted on day 1 of life, as described in Materials and methods. Animals were analysed for expression of TCR families as described using both bone marrow and spleen tissues, and compared to the original cord blood stem cell donor. Results from a single cord blood donor transplanted into three individual mice are shown at the time of killing (day 328).

Fig. 4.

T cell receptor (TCR) spectratyping. Non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice were transplanted on day 1 of life, as described in Materials and methods. T cell receptor (TCR) analyses were performed by spectratyping as described using both bone marrow and spleen tissues, and compared to the original cord blood stem cell donor. Results from a single cord blood donor transplanted into three individual mice are shown at the time of killing (day 328).

Reconstitution of the B cell repertoires as assessed by quantitative spectratyping of immunoglobulin heavy chains of the IgM class revealed a similar pattern to that observed in CD34-negative cord blood donor cells. However, the ratio of the VH03b family was higher in the human cord blood transplanted mice compared to the ratio observed in the CD34-negative cord blood donor cells (Fig. 5a). Also, the VH-chain length peak distribution was skewed in the reconstituted B cells in the mice compared to the donor cells (Fig. 5b).

Intratransplant variability

All mice humanized from a single cord blood donor seemed to reconstitute the T and B lymphocyte populations similarly, with similar changes observed over time. It was of interest to determine if similar results would be observed within the immune repertoires. To that end, three individual mice humanized with single cord blood donor stem cells were compared for TCR and BCR repertoires via spectratyping, as described above. As shown in Figs 6 and 7, the TCR repertoire was almost identical between the mice, albeit quite different from the stem cell donor used to create the mice. This observation was true whether T cells were obtained from the bone marrow (data not shown) or spleen of the animals. The conclusions reached above held true for each of the humanized mice.

Fig. 6.

T cell receptor (TCR) analysis of multiple mice humanized with a single cord blood donor. Non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice were transplanted on day 1 of life, as described in Materials and methods. Animals were analysed for expression of TCR families as described using both bone marrow and spleen tissues, and compared to the original cord blood stem cell donor. Results from three individual mice are shown at the time of killing (day 328).

Fig. 7.

T cell receptor (TCR) spectratype analysis of multiple mice constructed with a single cord blood donor. Non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice were transplanted on day 1 of life, as described in Materials and methods. TCR spectratype analysis was performed as described using spleen tissues, and compared to the original cord blood stem cell donor. Results from three individual mice are shown at the time of killing (day 328). Similar results were obtained when bone marrow was analysed.

Similarly, as shown in Figs 8 and 9, the BCR repertoires were very similar, although not identical, between mice displaying more intra-animal variability than that seen with the TCR analyses. However, each of the humanized mice was quite different from the cord blood donor repertoire. The same conclusions reached above for the individual mouse held true for each of the other mice constructed with this stem cell donor. Similar results were obtained for B cells found in the spleen and bone marrow (data not shown).

Fig. 8.

B cell receptor (BCR) analysis of multiple mice humanized with a single cord blood donor. Non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice were transplanted on day 1 of life, as described in Materials and methods. Animals were analysed for expression of BCR families as described using both bone marrow and spleen tissues, and compared to the original cord blood stem cell donor. Results from three individual mice are shown at the time of killing (day 328).

Fig. 9.

B cell receptor (BCR) spectratype analysis of multiple mice constructed with a single cord blood donor. Non-obese diabetic (NOD)-Rag1^null-γ chain^null (NRG) mice were transplanted on day 1 of life, as described in Materials and methods. BCR spectratype analysis was performed as described using spleen tissues, and compared to the original cord blood stem cell donor. Results from three individual mice are shown at the time of killing (day 328). Similar results were obtained when bone marrow was analysed.

Discussion

Recently, based on the original work by Traggiai et al. [1], considerable effort has been invested in the construction of a variety of humanized mouse strains to provide model systems to interrogate the human immune system. The NRG mouse was described originally by Pearson et al. [14] as a radioresistant version of the previously derived NSG mouse model [15]. These investigators reported that human immune cell engraftment was similar to that reported with the NSG mouse model [2] at early time-periods (12–16 weeks). Their work delineated the human engraftment of adult mice after 12 weeks after injection, demonstrating T cell, B cell and myeloid reconstitution in various lymphoid organs. However, these investigators did not follow these mice long term, and no information is available regarding stable haematopoiesis.

Further, the studies above have generally not compared the humanized blood and immune systems to the original donor(s) used to construct the strains to determine if, indeed, a faithful recapitulation had occurred. Finally, if multiple animals were constructed from a single donor, no reports have been published comparing these ‘syngeneic’ animals to each other to determine whether (or not) each of the animals was essentially identical to the other. In this study each of these possibilities was analysed. The goal of this study was to address several of these overarching questions. First, is haematopoietic reconstitution from the same haematopoietic stem cell (HSC) donor similar in multiple recipients? Secondly, is reconstitution from different HSC donors also similar; and thirdly, does reconstitution faithfully recapitulate the immune system observed in the HSC donor? Answering these questions might allow for the identification of molecular mechanisms important in these processes, and determine how accurately these humanized models reflect the human situation.

Transplantation of CD34⁺ cord blood cells into day 1 NRG mice resulted in stable long-term engraftment of the mice with human blood and immune cells. T cells, B cells, myeloid cells and neutrophils were observed over long periods of time, easily detectable for periods of up to 600 days post-transplant. There did not seem to be significant differences between mice engrafted with CD34⁺ cells from the same donor. Mice engrafted with CD34⁺ cells from different donors showed minimal discrepancies when compared to each other (‘Long term human reconstitution and aging in NOD-Rag (–)-γ chain (–) (NRG) mice’, by Harris and Badowski, paper submitted for publication). Long-term engraftment resulted in an abundance of CD4⁺ T cells at the expense of the CD8⁺ T cell subset (in a separately submitted manuscript, CD4 : CD8 T cell ratios were 4–5:1, rather than 2:1, as would normally be expected). B cells appeared earlier than T cells and then declined, while myeloid cells were stably present at low levels at all time-points. After 20–25 weeks (140–175 days), previous investigators using the NSG model, but not the NRG model, found large numbers of naive B and myeloid cells early after reconstitution with mainly ‘central memory’ CD4⁺ T cells (CD45RO⁺, CD62L⁺, CCR7⁺) later, and few CD8 T cells late after reconstitution [16]. In terms of the CD45RO⁺ T cells observed, it would be useful to know the diversity of the CDR3 sequences to determine if the T cells underwent antigen-driven expansion or if ‘homeostatic proliferation’ had occurred. Some evidence for antigen-driven proliferation of T cells might be implied in the over-representation of TCR BV in Fig. 3 and the spectratyping results shown in Fig. 4b,c. However, we did not select for or analyse T cells based on phenotype and cannot answer this question definitively at this time.

Our analysis of the TCR repertoire in these humanized mice at the time of euthanasia, compared to the CB donor, revealed differences. This result was not unexpected, in that human T cells developing and being educated in a mouse thymus would be educated on murine H-2 molecules [and any HLA molecules displayed on infiltrating human antigen-presenting cells (APC)] rather than on only human HLA molecules (as would have occurred in the original cord blood donor). Similarly, analysis of the BCR repertoire revealed, compared to the CB donor, that it was also significantly different. As B cells are not dependent upon MHC molecules for development and selection, this difference is probably a reflection of different self- (and microbial) antigens encountered during selection in the bone marrow of mice versus humans. Other investigators have also reported a diverse TCR and IgH repertoire, similar to that seen in humans, without suggestion of lymphopenia-induced proliferation [17]: a diversified repertoire approaching 100% of human reference samples with no major alterations. Other investigators have characterized the B cell repertoire in NSG mice 8–10 months after engraftment with CB CD34⁺ stem cells and reported the BCR repertoire to be large but predisposed towards autoreactivity, indicative of defects in selection [4].

In this study we have attempted to answer a series of questions pertaining to the humanized mouse model. First, is haematopoietic reconstitution from the same stem cell donor similar in multiple recipients? The answer appears to be yes. In fact, based on detailed analyses of the T cell and B cell repertoires, these animals might even be considered ‘syngeneic’ to one another. Secondly, is reconstitution from different HSC donors also similar? Yes, in preliminary analyses immune reconstitution appears to be almost identical, although in terms of immune repertoire it appears to be similar although not identical, as might be expected from different stem cell donors. Thirdly, does reconstitution faithfully recapitulate the immune system observed in the stem cell donor? Yes, in the sense that the TCR and BCR repertoires appear to be diverse, and all aspects of the immune system seem to be represented. Thus, this model seems to be a fairly accurate reflection of the human situation. However, the immune reactivity of both the T and B cell populations would be expected to be quite different from the original stem cell donor in terms of what antigens can be responded to.

Finally, these results should have a broad impact, as this model system should allow for experimental manipulations to improve our understanding of the human immune system, as well as for investigation of novel interventions to improve the function of the immune system. Such an understanding should have significant benefits in terms of dealing with infectious disease and improving methods for vaccination.

Author contributions

M. B. performed the experimental portions of the project along with the FACS analyses. A. B. and O. O. Y. performed the TCR and BCR analyses. D. T. H. designed the experimentation, performed the overall analyses and was the primary author of the manuscript.

Disclosure

The authors have no conflicts of interest to declare.