Humanized mice and tissue transplantation

Abstract

Our understanding of the molecular pathways that control immune responses, particularly immunomodulatory molecules that control the extent and duration of an immune response, have led to new approaches in the field of transplantation immunology to induce allograft survival. These molecular pathways are being defined precisely in murine models, and are now being translated into clinical practice. However, many of the newly available drugs are human-specific reagents and furthermore, there exist many species-specific differences between mouse and human immune systems. Recent advances in the development of humanized mice, i.e., immunodeficient mice engrafted with functional human immune systems, have led to the availability of a small animal model for the study of human immune responses. Humanized mice represent an important pre-clinical model system for evaluation of new drugs as well as identification of the mechanisms underlying human allograft rejection without putting patients at risk. This review highlights recent advances in the development of humanized mice and their use as pre-clinical models for the study of human allograft responses.

Introduction

Major histocompatibility complex (MHC)-mismatched grafts induce the activation of a cell-mediated immune response that leads to graft rejection in the absence of immunosuppressive therapy. As the technology for transplanting cells and tissues has improved, the major remaining limiting factor influencing the success of an organ transplant is the ability to control this immune response. Understanding the mechanisms of allograft rejection is imperative for developing new immune therapies to improve the long term success of organ transplants. Mouse models of allogeneic rejection have provided insights that have led to the development of several new immunosuppressive and immunomodulatory approaches. However, successful immunotherapies in animal models when translated into the clinic have to date produced limited success, likely in part due to the many species-specific differences between mouse and human immune responses (1). An important additional difference is the presence of memory T cells in humans that are potentially alloreactive (2–4) and are absent in naïve mice housed in microisolator cages in specific pathogen-free facilities and explicitly not exposed to viral infections.

Recent advances in the development of humanized mice have positioned them as an excellent preclinical model to investigate species-specific molecules, molecular pathways, and mechanisms underlying human allograft rejection and for the investigation and evaluation of potential therapies without putting patients at risk. This review outlines these recent advances and the use of humanized mice for the study of transplantation.

Humanized mouse strain development

The development of immunodeficient mice for engraftment with functional human immune systems over the last 25 years has been reviewed extensively (5–7). Simply put, the goal is to generate an immunodeficient mouse that can be engrafted with functional human innate and adaptive immune cells or with human hematopoietic stem cells (HSC). The type of human cells and tissues used in part determines the quality and robustness of the engrafted human immune system. The major breakthrough in humanized mice in the early 2000’s came with the addition of mutations targeting the IL-2 receptor common gamma chain (IL2rg^null) (5–7). The common gamma chain is essential for high affinity receptor signaling for IL2, IL4, IL7, IL9, IL15 and IL21. Blocking signaling for this group of cytokines severely inhibits both adaptive and innate immunity, including natural killer (NK) cell development. Crossing the IL2rg^null mutation to mice homozygous for the Prkdc^scid (scid), recombination activating gene 1 (Rag1^null) or Rag2^null mutations allowed for heightened human engraftment of both lymphoid and myeloid cells and supported the development of a more complete human immune system following transplantation with HSCs. There are three major immunodeficient mouse stocks that are widely currently used: NSG (NOD.Cg-Prkdc^scidIl2rg^tm1Wjl/Sz), NOG (NODShi.Cg-Prkdc^scidIl2rg^tm1Sug) and BRG (C;129S4-Rag2^tm1.1FlvIl2rg^tm1.1Flv) mice (see (5–7) and Table 1).

Table 1.

Most commonly used immunodeficient strains engrafted with human hematopoietic cells

Commonly used strains	Common abbreviations	Il2rg mutation	Characteristics	Immunological Characteristics	Availability
NOD.Cg-Prkdcscid Il2rgtm1Wjl	NSG	Mutation is a complete null, is not expressed and will not bind cytokines	NOD strain. Immunodeficient and relatively radiosensitive due to a defect in DNA repair	Lacks T, B and NK cells, additional defects in innate immune cells	The Jackson Laboratory Stock: 005557
NOD.cg-Prkdcscid Il2rgtm1Sug	NOG	Lacks the intracytoplasmic domain and will bind cytokines but will not signal	NOD strain. Immunodeficient and relatively radiosensitive due to a defect in DNA repair	Lacks T, B and NK cells, additional defects in innate immune cells	Taconic Bioscience Stock: CIEA NOG mouse®
NOD.Cg-Rag1tm1Mom IL2rgtm1Wjl	NRG	Mutation is a complete null, is not expressed and will not bind cytokines	NOD strain. Immunodeficient and relatively radioresistant	Lacks T, B and NK cells, additional defects in innate immune cells	The Jackson Laboratory Stock: 007799
C.Cg-Rag2tm1Fwa Il2rgtm1Sug	BRG	Lacks the intracytoplasmic domain and will bind cytokines but will not signal	Mixed background, predominately BALB/c strain: Immunodeficient and relatively radioresistant	Lacks T, B and NK cells, remaining innate immune cells are functional	Taconic Bioscience Stock: 11503

The four strains of immunodeficient mice bearing targeted mutations in the IL2 receptor common gamma chain that have been most commonly used for humanization. An extensive list of immunodeficient mice that have been engrafted with human immune systems has been previously provided (5,8)

Engraftment Approaches

There are three standard approaches to engrafting a human immune system into immunodeficient mice. Mice can be humanized by the injection of human peripheral blood mononuclear cells (PBMC), HSC, or HSC in combination with implantation of autologous fragments of fetal thymus and liver. The injection of human PBMC into immunodeficient mice, also known as the Hu-PBL-SCID model, leads to the engraftment of primarily T cells (5–8). Hu-PBL-SCID mice develop a xenogeneic graft-versus-host like disease (GVHD) within a few weeks, but this model can be used for short term studies to examine T cell rejection of human allografts (9). The injection of CD34⁺ HSC into newborn or young mice, also known as the Hu-SRC-SCID model, allows for the differentiation and development of a more complete immune system including T cell, B cells, and innate immune cells. HSC can be sourced from human bone marrow, umbilical cord blood (UCB), fetal liver, or granulocyte-colony stimulating factor (G-CSF) mobilization of HSC into the blood. The major limitation of this model is that the human T cells are selected and educated on mouse MHC within the host thymus and thus are H2-restricted, leading to complex T cell-antigen presenting cell interactions in the murine host during the development of an immune response. To improve this model, human fetal thymus can be surgically implanted under the kidney capsule of adult conditioned mice, which are then injected with CD34⁺ HSC isolated from the autologous fetal liver (10–12). This enhances the immune system as T cells that develop are human leukocyte antigen (HLA)-restricted. This is currently the most robust human immune system engraftment protocol currently available (5–7).

Human skin allografts

Allogeneic rejection of human allografts by humanized mice has been studied for over 20 years and this literature was recently reviewed by us (13) and others (8). Early studies with the Hu-PBL-SCID model using CB17-scid mice led to inconsistent results due to poor human T cell engraftment which was improved by elimination of host NK cells using anti-monosialotetrahexosylganglioside (GM1) antibody or addition of the Lyst^bg (beige) mutation to CB17-scid strain (8,13). Using SCID/beige (CB17-Lyst^bg Prkdc^scid) mice led to the conclusion that human CD4 and CD8 T cells were both important in skin allograft rejection (14,15), which was confirmed using NOD-scid (NOD.CB17 Prkdc^scid) mice (16). Transitioning to NSG mice, which achieve high levels of human cell engraftment, it was observed that they need to be treated with mouse anti-granulocyte receptor 1 (anti-Gr1) antibody after skin transplantation for successful healing of the graft due to the infiltration of mouse innate immune cells (9). Skin grafts on NSG mice treated with anti-Gr1 mAb heal in and exhibit excellent vascularization and morphology. In the Hu-PBL-SCID model using NSG mice, 100% of allografts are rejected within 21 days providing a robust and reproducible model for study of human T cell-mediated skin allograft rejection (9).

Further studies characterizing skin allograft infiltrating T cells documents the presence of human CD4 and CD8 T cells producing IL-17A in the graft dermis and epidermis (17). IL-17A has strong inflammatory properties and has been associated with psoriasis (18–20) and atopic dermatitis (21). T regulatory cells (Tregs) are a major player in immunoregulation and CD4⁺FoxP3⁺ T cells are present in the basal layer of skin grafts on SCID/beige mice after PBMC injection suggesting an attempt by the immune system to down modulate the immune inflammation (17).

Human islet allografts

Islet transplantation has been used to successfully restore glucose homeostasis in individuals with type 1 diabetes, but insulin independence is subsequently lost in most patients over time and the procedure comes with adverse side effects, predominately due to the use of immunosuppressive drugs to prevent graft rejection (22). Humanized mouse models are becoming a valuable tool to study the mechanisms of islet allograft rejection and to test potential therapeutics prior to their advancement to the clinic.

The Hu-PBL-SCID model has been used to study islet rejection, but there is a “race” between the rejection of the islets and the development GVHD making long-term monitoring for therapeutic interventions difficult (23). Although islet transplantation into humanized mice has been ongoing for ~25 years, prior to the development of NSG mice T cell engraftment and islet rejection was variable, and required complex approaches to achieve complete rejection (8,13). Simultaneous injection of allogeneic PBMC with islet grafts into NSG mice results in rapid graft rejection and return to hyperglycemia within 21 days in 100% of mice (24). When islet grafts were allowed to become established in diabetic NSG mice for 37 days prior to allogeneic PBMC injection, the return to hyperglycemia occurred in 2 out of 3 mice before the termination of the study due to the development of GVHD (24).

The Hu-SRC-SCID model develops a more complete immune system but is limited by the selection of human T cells on mouse MHC. NSG and BRG HSC-engrafted mice have shown contrasting results in when transplanted with allogeneic human islets. HSC-engrafted hyperglycemic BRG mice are not able to reject human islets and exhibit minimal T cell infiltration into the graft (25). In contrast, hyperglycemic HSC-engrafted NRG-Akita (NOD-Rag1^Tm1Mom IL2g^null Ins2^Akita) mice engrafted with 5×10⁴ umbilical cord blood (UCB)-derived CD34⁺ HSC were shown to reject ~60% of human islet allografts with evidence of strong mononuclear cell infiltration into the rejecting grafts (26). In those mice that did not revert to hyperglycemia, a mononuclear infiltrate into the graft was still observed.

In a more recent study following injection of higher numbers of UBC-derived CD34⁺ HSC (2×10⁵) into NSG mice, all of the islet grafts were rejected within 17 days after transplantation as evidenced by the return to hyperglycemia (27). Graft rejection in NSG mice was confirmed by the loss of histological staining for human insulin and was associated with infiltration of human CD4 T cells, macrophages and neutrophils. Interestingly in this model very few CD8 T cells were found to infiltrate the graft, which is in contrast to Hu-PBL-SCID models. Of particular interest in this report is that injection of 6×10⁶ ex vivo expanded human Tregs completely prevented islet graft rejection (27). A more complete comparison of these three different HSC engrafted models of islet allograft rejection is provided in Table 2. The differences in the rejection of human islets in these models may be due to differences in the host strain or in the numbers of human HSCs used for engraftment leading to differences in human immune function in these models.

Table 2.

Recipient strain and engraftment protocols affect islet allograft survival in humanized mice

Recipient Strain	Engraftment protocol	HSC source and number injected	Time after engraftment islets transplanted	Graft outcome	Ref
BALB/c-Rag2null IL2rgnull (BRG)1	Pregnant BRG mice injected with 0.5 mg Busulfan 2– 7 days prior to delivery, BRG newborns irradiated with 550 cGy	1×105 CD34+ umbilical cord blood cells injected intrahepatically	300–500 human islets transplanted subrenal capsule 8–26 weeks post CD34 HSC engraftment into normoglycemic BRG mice	Human CD45 engraftment levels of 22.0 ±16.6% in the blood, grafts analyzed histologically day 14 or 35 after transplantation. No evidence of rejection determined by no loss of insulin staining and little CD3 or CD20 infiltration observed in the graft or loss of Human C-peptide in the serum	25
NOD.Cg- Rag1tm1Mom IL2rgtm1Wjl Ins2Akita (NRG- Akita)	1–3 day old NRG- Akita mice irradiated with 400 cGy	5×104 CD34+ umbilical cord blood cells	4000 human islets transplanted subrenal capsule into diabetic NRG-Akita mice	NSG mice with 12.9 ± 2.5% human CD45+ blood cells used in experiment. 8/13 (62%) islet grafts rejected as defined by a return to hyperglycemia and loss of insulin staining in graft. CD45 infiltration observed even in non-rejected grafts	26
NOD.Cg- Prkdcscid Il2rgtm1Wjl (NSG)	4–6 week old NSG irradiated with 240 cGy	20×104 CD34+ umbilical cord blood cells	3000–4000 human islets transplanted subrenal capsule into diabetic NSG mice, mice treated with streptozotocin to induce diabetes (timing not reported in relation to HSC engraftment or islet transplantation)	NSG mice with >15% human CD45+ blood cells used in experiment. 21/21 (100%) rejected islet grafts as defined by return to hyperglycemia and loss of insulin staining in graft. Infiltration of graft with macrophages (CD11b), neutrophils (CD66b) and CD4 T cells, but few to no CD8 T cells	27

Engraftment of immunodeficient mice bearing targeted mutations in the IL2rg gene with human CD34⁺ umbilical cord blood HSCs and human islets.

¹The derivation of the BRG strain, the origin of the Rag2^null mutation, and the origin of the IL2rg^null mutation 3ere not reported

Human cardiac tissue

The endothelial cells lining arterial grafts are immunogenic and have been shown to activate allogeneic human T cells (28). Much of the literature relating to cardiac grafts in humanized mice have been performed in humanized SCID/beige mice due to the high sensitivity of NSG mice to paralysis and/or death after human artery transplantation (29). The majority of SCID/beige mice receiving artery transplants followed by allogeneic PBMC exhibit enhanced infiltration of human cells into the graft and histological changes consistent with rejection (30). The extensive study of this model has revealed several mediators of graft injury that have possible therapeutic potential. Infiltrating T cells produce interferon gamma (IFNγ) (31) which sustains enhanced HLA-DR expression on endothelial cells lining the graft and induces neointima, or scarring of the inner most layer of the blood vessels within the human arterial graft (32). Neutralization of IFNγ blocks rejection of grafts by infiltrating T cells (31). Arterial graft pretreatment with rapamyacin reduces graft injury and the number of T cells that infiltrate into the intimal (33). This protection from graft injury is mediated by the upregulation of programmed death (PD) receptor ligands 1 and 2 and conversely, blockade of PD1 increases graft injury (33), suggesting an immunomodulatory approach to prevent arterial graft rejection may be effective.

Alternatively, blockade of transforming growth factor-β (TGFβ) enhances graft injury (34). TGFβ is produced by the arterial graft and reduces the ability of the infiltrating allogeneic T cells to produce IFNγ. Moreover, proinflammatory cytokines IL1 and IL6 produced from the arterial graft appear to promote graft injury (35,36). IL6 blocking antibody reduces graft injury, with a reduction in total vessel area and the intimal area and increased lumen area (37). A similar reduction of graft injury was found using a human-specific IL-1R antagonist accompanied by reduced infiltration of human immune cells and decreased IL17 production. These studies suggest that arterial grafts produce cytokines that both protect and enhance allogeneic immune responses. Using the Hu-SRC SCID/beige humanized mouse model, which has poor engraftment of T and B cells, human arterial grafts showed graft injury with the formation of neointimas that contain macrophages, but no CD3, CD11c or CD19 positive cells suggesting a role for macrophages in arterial graft injury (29).

Pluripotent stem cells and stem cell-derived cell populations

A new area of transplantation biology is the use of mature cell populations differentiated from pluripotent stem cells. The major question involved in these transplantations is whether the pluripotent stem cells and their differentiated progeny are susceptible to autologous or allogeneic immune responses. In the murine system, it has been reported that differentiated cells from induced pluripotent stem cells (iPSC) elicit immune responses when transplanted into syngeneic hosts (38). In contrast, a recent report suggests that differentiated cells derived from iPSCs fail to elicit an immune response in syngeneic murine hosts (39). Understanding the immunogenicity of human cells and tissues derived from iPSCs or embryonic stem cells (ESCs) will be an important consideration as this cellular therapy progresses towards clinical trials.

Using the NSG mouse engrafted with human fetal liver and thymus and injected with autologous liver CD34+ HSC (BLT model), it was shown that allogeneic teratomas derived from human ESCs are heavily infiltrated with immune cells, regress in size, and after 6 weeks consist mostly of liquid cysts suggesting they are being rejected (38). Similarly, hESCs differentiated into fibroblasts or cardiomyocytes and transplanted subcutaneously or intra-muscularly, respectively, into NSG-BLT mice became necrotic and infiltrated with human immune cells. To prevent injury of transplanted human ESCs and their differentiated fibroblasts or cardiomyocytes, the human ESCs were genetically engineered to express cytotoxic T lymphocyte associated protein 4 (CTLA4)-Ig and PD-L1, which modulate T cell costimulatory pathways. ESC teratomas, and differentiated fibroblasts and cardiomyocytes expressing these molecules were effectively protected from human immune infiltration in NSG-BLT mice (40). However, it must be cautioned that relying solely on the presence or absence of human T cell infiltration is not a sufficient measure of graft rejection as infiltrates may in fact be suppressing the immune response (39).

The ability of co-stimulation blockade to protect iPSC-derived beta cells from both xenogeneic and allogeneic rejection has been reported (41). Human iPSC-derived beta cells are rapidly rejected when transplanted into immunocompetent mice. Similarly, human iPSC-derived beta cells are also rejected in NSG mice injected with 15×10⁶ allogeneic human PBMC (41). In contrast, both xenograft and allogeneic rejection are prevented when the mice are treated with the co-stimulation blockade agents CTLA4-Ig plus anti-CD154 (anti-CD40L) antibody (41). These data combined with previous reports suggest that iPSC-derived cells and tissues can be rejected by allogeneic human immune systems, but that this rejection can be prevented by co-stimulation blockade.

Human Tregs and MSCs as regulatory cells

The ability to evaluate the in vivo efficacy of human regulatory populations to prevent graft rejection in humanized mice clearly has advantages for testing their functional capability as well as a model that allows the optimization of the source and ex vivo expansion protocols for human regulatory cells. However, as shown by the approaches described below, the choice of the immunodeficient strain, allograft target, human immune engraftment approach, definition of successful immune “engraftment” and evaluation of immune rejection, and the efficacy of the regulatory population in preventing rejection of allografts are not well defined, and are “model and system” dependent. Depending on the recipient strain and model system, 5–300×10⁶ effector PBMCs are required for infiltration and damage to the target graft.

In BRG mice transplanted with human aortic grafts and injected with 10×10⁶ PBMC with or without 1–2×10⁶ Treg cells, it was observed that the addition of Tregs could prevent the development of transplant arteriosclerosis when studied histologically 30 days later (42). Graft preservation was associated with a decrease in the secretion of IFNγ. In BRG mice engrafted with human skin for 35 days, injection of 5×10⁶ human PBMC (engraftment of human PBMC defined as >1% human CD45+ cells in the spleen) led to rejection with a median survival time (MST) of 34–40 days (43). Interestingly, in this report they also showed that skin grafts survived for >100 days following injection of PBMC autologous to the skin graft (i.e., from the same donor), highlighting that the rejection appeared to be an allogeneic and not a non-specific immune response. When allogeneic PBMC was mixed with ex vivo expanded Tregs at a 1:1 ratio prior to injection, no skin graft rejection was observed up to 100 days, the duration of the experiment (43). Recently the ability of Tregs to prevent allograft rejection was extended to islet allografts using similar approaches (44). In this study, a high number of human islets (8000 IEQ) were transplanted into BRG mice for 14 days followed by injection of 40×10⁶ PBMC with or without an equal number of ex vivo expanded Tregs. Islet allografts receiving only allogeneic PBMC were rejected with a MST of 23 days, but when an equal number of Tregs were co-injected, MST was >45 days with only 2 of the 13 mice receiving Tregs rejecting the grafts (44).

In a similar study using NSG or BRG mice engrafted with human skin for 4–6 weeks and injected with 5×10⁶ PBMC with or without co-injection of 1×10⁶ Tregs (engraftment defined as >0.5% human CD45+ splenic cells) allo-specific Tregs were found to be more potent inhibitors of skin graft rejection than polyclonal expanded Tregs (45). In an extension of this study, a method to generate clinical grade allo-specific Tregs was described, and again, these allo-specific Tregs transplanted into BRG mice with human skin grafts were found to be more potent inhibitors of graft injury than polyclonal expanded Tregs using very similar conditions but with increased numbers of PBMC (10×10⁶) and Tregs (2×10⁶) (46). Graft injury in both studies was evaluated at 4 weeks after PBMC injection and consisted of analyses of human CD45 infiltration, Ki67 and involucrin staining, expression of TUNEL⁺ nuclei in dermal infiltrates, preservation of human CD31 staining, and detection of CD4⁺FoxP3⁺ Tregs. Overall graft survival was not reported in either study (45,46).

In a variation of this approach, CB17-scid mice were transplanted with human skin grafts, and 6 weeks later, anti-asialo GM1 antibody was administered to deplete murine host NK cells followed by injection of 300×10⁶ allogeneic PBMC. Complete rejection of the graft was not observed, but addition of bone marrow or adipose-derived mesenchymal stem cells (MSC) as an immunoregulatory population to the allogeneic PBMC led to decreased leukocyte infiltration into the graft and reduced expression of inflammatory cytokines such as IFNγ, tumor necrosis factor α (TNFα), IL-1β and IL6 (47). These studies suggest that humanized mice can be used to evaluate the ability of human regulatory cells to modulate allograft rejection, but the results show that variations in the approach may influence the data generated.

Xenografts

Xenograft transplantation using porcine grafts provides a potentially unlimited source for donor tissues if the xenograft rejection process could be overcome. Humanized mice have been used to begin to address the issue of xenograft rejection by human immune systems. In the Hu-PBL-SCID model, diabetic NSG mice were transplanted with 5000 neonatal porcine islet cell clusters (NICC), which restored normoglycemia (48). When 1–10×10⁶ human PBMC were injected, the NICC grafts were destroyed in a PBMC number-dependent manner. Graft infiltration could be blocked by the co-injection of ex vivo expanded human Tregs that was mediated in part by Tregs production of IL-10 (49).

Extending these observations to investigate the ability of central tolerance in the thymus to induce xenograft tolerance, NSG mice were transplanted with a porcine thymus and injected with human CD34⁺ HSC (50). In this system, human T cells were generated, and they were specifically non-reactive to the murine host, the human HSC donor, and to the MHC of the porcine donors. Skin grafts from porcine donors sharing the MHC of the thymus were not rejected, while skin grafts from other non-MHC-matched porcine donors were rapidly rejected (50). These various xenograft models will be valuable tools for the study of human xenograft responses and accelerate efforts to identify approaches for the induction of xeno-tolerance and permanent xenograft survival.

Naïve, effector and memory T cells and alloantibodies in graft survival

Humanized mice have also been used to investigate the role of T cells involved in allograft rejection as well as the role of alloantibodies in graft survival. In early studies in humanized mice, it was shown that memory but not naïve T cells were the principal mediators of skin graft injury and that this destruction could be blocked using a 4-1BB ligand, ICOS ligand, or OX40 ligand (51). This was extended to show that CD4 effector/memory T cells but not CD4 central memory T cells targeted endothelial cells in skin grafts, and in vitro analyses indicated that these cells mediated their effect in part by secretion of IFNγ (52). Immunodeficient mice bearing human skin grafts injected with the anti-HLA rat antibody W6/32 were shown to have activated endothelial cell exocytosis and leukocyte trafficking in the graft, suggesting that alloantibodies may have a role in transplant rejection (53). This study was extended showing that combined with more than 6 hours of cold ischemia, antibody administration, regardless of specificity, promoted transplant vasculopathy as evidenced by intimal expansion in human vessels transplanted into BRG mice (54). Using an in vitro model system, the binding of anti-HLA antibodies to human aortic, venous, and microvascular endothelial cells induced endothelial P-selection and increased adherence of monocytes, implying a role of alloantibodies in recruitment of monocytes that participate in antibody-mediated rejection of human allografts (55). This suggestion was confirmed in vivo using a mouse/mouse transplant model of C57BL/6-Rag1^null recipients of BALB/c cardiac allografts that were passively transferred with donorspecific MHC class I antibodies (56).

Remaining limitations and future opportunities

Although humanized mice represent exciting new models that permit the direct study of human transplantation in a small animal model without putting patients at risk, there remain limitations to the available models.

The development of GVHD in almost all of the model systems represents a constraint on the experimental window available for the study of graft survival, particularly in the Hu-PBL-SCID model system (57). However, much of the xenoreactivity of human immune systems to murine antigens is to the MHC class I and class II molecules (57). In the NSG-BLT model, CD8 T cells, which are reactive to murine MHC class I, appear to have a central role in the development of GVHD-like symptoms (58). A recently described model based on the development of B6.129-Rag2^tm1Fwa CD47^tm1fpl IL2rg^tm1cgn mice suggests that the lack of murine CD47 decreases the development of GVHD-like symptoms in BLT humanized mice (59). Development of immunodeficient hosts deficient in murine MHC and/or CD47 may decrease the development of GVHD-like symptoms and extend the experimental window available for the study of transplantation.

Additional opportunities to improve humanized mouse models include optimizing models for enhanced humoral responses, memory T cell formation, lymphoid structures and germinal center formation (recently reviewed extensively in (5,60)). Due to the lack of reactivity between many murine cytokines and human cells, additional improvements in the murine host focusing on the provision of species-specific human cytokines important in human immune development are underway in a number of laboratories (5–7). For example, mouse B cell activating factor (BAFF) (also termed B lymphocyte stimulator) binds to but does not signal human B cells (61), and efforts are underway to generate human BAFF transgenic mice in our laboratory to determine if this will improve human B cell responses to immunization. However, recent reports on transgenic generation of various human species-specific cytokines suggest a note of caution as these factors are expressed in the murine host. For example, murine IL2 does not cross react with human T cells, and transgenic expression of human IL2 in NOG mice leads to robust human NK cell development at the expense of the other immune cell populations (62). In BRG mice, transgenic expression of human thrombopoietin, IL3, granulocyte macrophage colony stimulating factor (GM-CSF), macrophage colony stimulating factor (M-CSF), and signal-regulatory protein α (Sirpα) leads to robust human innate immune development, but the experimental window of this strain is relatively limited (63). Additional modifications of the murine host to reduce macrophages and granulocyte populations will likely be important to achieve circulation of human red blood cells, granulocytes, and platelets, which circulate at only low levels in the currently available models of humanized mice (5–7).

Transgenic expression of human HLA class I and II molecules has also been shown to enhance development of HLA-restricted T cell responses in the Hu-SRC-SCID model. For example, HLA-A2 transgenic NSG mice engrafted with HLA-A2 HSCs have been shown to generate HLA-A2 restricted EBV-specific cytotoxic T cells in humanized mice (64,65). Extension of the use of human immune engrafted HLA-Tg immunodeficient mice to that of allograft rejection has not been reported. Using EBV-infected mice and/or the HLA-restricted cell lines that have been reported (66) may be an approach to study the role of human memory cells in allograft rejection, an area of investigation that has not been pursued to date in humanized mice.

The current models of humanized mice have limited B cell maturation with low antibody levels and deficiencies in their ability to isotype switch following exposure to antigen, resulting in little to no human IgG antibody responses following immunization. However, it has been reported by two groups that engraftment of human HLA-DR4 transgenic NRG/NOG mice with HSCs from HLA-DR4 donors leads to improved human IgG production following immunization (67,68). Although the available models of humanized mice likely would generate only poor and predominately IgM alloantibody responses to tissue grafts, the injection of human serum containing known alloantibodies into humanized mice might be an approach to study the role of alloantibodies in graft rejection in this model system.

Conclusions

Exciting new models of humanized mice to study transplant rejection are becoming available to the research community. Improvements in immune system engraftment, immune function, the potential for development human myeloid cells, an important component of allograft rejection, are under development. The use of humanized mice as a model for the study of human immune rejection of allografts and xenografts may provide novel insights into the mechanisms responsible for graft rejection and permit rapid evaluation of new approaches to prevent graft loss in transplant recipients.

Abbreviations

BAFF

B cell activating factor

Beige

Lyst^bg

BLT

transplantation with fetal liver and thymus and injection of autologous liver HSC

BRG

C; 129S4- Rag2^tm1FlvIl2rg^tm1Flv

CTLA4

cytotoxic T lymphocyte associated protein 4

ESC

embryonic stem cells

G-CSF

granulocyte-colony stimulating factor

GM1

monosialotetrahexosylganglioside 1

GM-CSF

granulocyte macrophage-colony stimulating factor

Gr1

granulocyte receptor 1

GVHD

graft-versus-host like disease

HLA

human leukocyte antigen

Hu-PBL-SCID

transplantation with human PBMC

Hu-SRC-SCID

transplantation with human CD34+ HSC

HSC

hematopoietic stem cells

IFNγ

interferon gamma

IL2rg^null

IL-2 receptor common gamma chain knockout

iPSC

induced pluripotent stem cells

M-CSF

macrophage colony stimulating factor

MHC

major histocompatibility complex

MSC

mesenchymal stem cells

MST

median survival time

NICC

neonatal porcine islet cell clusters

NK cell

natural killer cell

NOG

NODShi.Cg-Prkdc^scidIl2rg^tm1Sug

NOD

non-obese diabetic

NRG-Akita

NOD-Rag1^Tm1MomIL2g^nullIns2^Akita (NRG-Akita)

NSG

NOD.Cg-Prkdc^scidIl2rg^tm1Wjl

PBMC

peripheral blood mononuclear cells

PBL

peripheral blood lymphocyte

PBMC

peripheral blood mononuclear cells

PD1

programmed death receptor 1

Rag1/2

recombination activating genes 1 and 2

scid

Prkdc^scid

SCID/beige

CB17-Lyst^bg Prkdc^scid

Sirpα

signal-regulatory protein α

TGFβ

transforming growth factor-β

TNFα

tumor necrosis factor α

Tregs

regulatory T cells

UBC

umbilical cord blood