American Journal of Transplantation 2016; 16: 389–397 © Copyright 2015 The American Society of Transplantation Wiley Periodicals Inc. and the American Society of Transplant Surgeons doi: 10.1111/ajt.13520

Humanized Mouse Models for Transplant Immunology

L. L. Kenney1, L. D. Shultz2, D. L. Greiner1,* NOD.Cg-PrkdcscidIL2rgtm1Wjl; PBMC, peripheral blood and M. A. Brehm1 mononuclear cell; PD-L1, programmed death ligand 1; Rag2, recombination activating gene 1; Rag2, recom- scid 1 bination activating gene 2; scid, Prkdc , severe Department of Molecular Medicine, Diabetes Center of combined immunodeficiency; SCID/beige, CB17- Excellence, University of Massachusetts Medical School, Lystbg Prkdcscid; TGF-b, transforming growth factor Worcester, MA b; Tregs, regulatory T cells; UCB, umbilical 2The Jackson Laboratory, Bar Harbor, ME Corresponding author: Dale L. Greiner, Received 20 July 2015, revised 02 September 2015 and [email protected] accepted for publication 04 September 2015

Our understanding of the molecular pathways that Introduction control immune responses, particularly immunomod- ulatory molecules that control the extent and duration MHC-mismatched grafts induce the activation of a cell- of an immune response, have led to new approaches in mediated immune response that leads to graft rejection in the field of transplantation immunology to induce the absence of immunosuppressive therapy. As the allograft survival. These molecular pathways are being technology for transplanting cells and tissues has improved, defined precisely in murine models and translated into the major remaining limiting factor influencing the success of clinical practice; however, many of the newly available an organ transplant is the ability to control this immune drugs are human-specific reagents. Furthermore, many response. Understanding the mechanisms of allograft species-specific differences exist between mouse and rejection is imperative for developing new immune therapies human immune systems. Recent advances in the development of humanized mice, namely, immunode- to improve the long-term success of organ transplants. ficient mice engrafted with functional human immune Mouse models of allogeneic rejection have provided insights systems, have led to the availability of a small animal that have led to the development of several new immuno- model for the study of human immune responses. suppressive and immunomodulatory approaches. Neverthe- Humanized mice represent an important preclinical less, successful immunotherapies in animal models, when model system for evaluation of new drugs and translated into the clinic, have produced limited success to identification of the mechanisms underlying human date, likely in part because of the many species-specific allograft rejection without putting patients at risk. This differences between mouse and human immune review highlights recent advances in the development responses (1). An important additional difference is the of humanized mice and their use as preclinical models presence of memory T cells in humans that are potentially for the study of human allograft responses. alloreactive (2–4) and that are absent in na€ıve mice housed in microisolator cages in specific pathogen-free facilities and Abbreviations: anti-Gr1, anti- antibody; anti-GM1, anti–monosialotetrahexosylganglioside 1; explicitly not exposed to viral infections. BAFF, activating factor; beige, Lystbg; BLT, transplantation with fetal and and Recent advances in the development of humanized mice injection of autologous liver hematopoietic stem cells; have positioned them as an excellent preclinical model to BRG, C.Cb-Rag2tm1Fva IL2rgtm1Sug; CTLA4, cytotoxic T investigate species-specific molecules, molecular path- lymphocyte associated protein 4; ESC, embryonic stem ways and mechanisms underlying human allograft rejection cell; GVHD, graft-versus-host disease; HSC, hemato- and to investigate and evaluate potential therapies without poietic stem cell; Hu-PBL-SCID, transplantation with putting patients at risk. This review outlines these recent human peripheral blood mononuclear cells; Hu-SRC- advances and the use of humanized mice for the study of SCID, transplantation with human CD34þ hematopoi- transplantation. etic stem cells; IFN-g, interferon-g; IL2rgnull, IL-2 receptor common gamma chain knockout; iPSC, induced pluripotent stem cell; MST, median survival Humanized mouse strain development time; NICC, neonatal porcine islet cell cluster; NOG, The development of immunodeficient mice for engraftment NOD.Cg-PrkdcscidIL2rgtm1Sug; NOD, nonobese diabetic; with functional human immune systems over the past NRG-Akita, NOD-Rag1tm1MomIL2gnullIns2Akita; NSG, 25 years has been reviewed extensively (5–7). Simply put,

389 Kenney et al the goal is to generate an immunodeficient mouse that can humanized by the injection of human peripheral blood be engrafted with functional human innate and adaptive mononuclear cells (PBMCs), HSCs or HSCs in combination immune cells or with human hematopoietic stem cells with implantation of autologous fragments of fetal thymus (HSCs). The type of human cells and tissues used partly and liver. The injection of human PBMCs into immunode- determines the quality and robustness of the engrafted ficient mice, also known as the Hu-PBL-SCID model, leads human . The major breakthrough in to the engraftment primarily of T cells (5–8). Hu-PBL-SCID humanized mice in the early 2000s came with the addition mice develop a xenogeneic graft-versus-host-like disease of targeting the IL-2 receptor common gamma (GVHD) within a few weeks, but this model can be used for chain (IL2rgnull) (5–7). The common gamma chain is short-term studies to examine rejection of human essential for high-affinity receptor signaling for IL-2, IL-4, allografts (9). The injection of CD34þ HSCs into newborn or IL-7, IL-9, IL-15 and IL-21. Blocking signaling for this group young mice, also known as the Hu-SRC-SCID model, of cytokines severely inhibits both adaptive and innate allows for the differentiation and development of a more immunity, including NK cell development. Crossing the complete immune system including T cells, B cells and IL2rgnull to mice homozygous for the Prkdcscid innate immune cells. HSCs can be sourced from human (severe combined immunodeficiency [scid]), recombination , umbilical cord blood (UCB), fetal liver or activating gene 1 (Rag1null) or recombination activating granulocyte colony-stimulating factor mobilization of HSCs gene 2 (Rag2null) mutation allowed for heightened into the blood. The major limitation of this model is that the human engraftment of both lymphoid and myeloid cells human T cells are selected and educated on mouse MHC and supported the development of a more complete within the host thymus and thus are H2 restricted, leading human immune system following transplantation with to complex interactions of T cells and antigen-presenting HSCs. Three major immunodeficient mouse stocks are cells in the murine host during the development of an widely used currently: NSG (nonobese diabetic strain immune response. To improve this model, human fetal [NOD]; NOD.Cg-PrkdcscidIL2rgtm1Wjl/Sz), NOG (NODShi. thymus can be surgically implanted under the kidney Cg-PrkdcscidIL2rgtm1Sug) and BRG (C.Cg-Rag2tm1Fva capsule of adult conditioned mice, which are then injected IL2rgtm1Sug) mice (5–7) (Table 1). with CD34þ HSCs isolated from the autologous fetal liver (10–12). This enhances the immune system because Engraftment approaches T cells that develop are HLA restricted. This is the most There are three standard approaches to engrafting a human robust human immune system engraftment protocol immune system into immunodeficient mice. Mice can be currently available (5–7).

Table 1: Most commonly used immunodeficient strains engrafted with human hematopoietic cells Commonly Common Immunological used strains abbreviations IL2rg mutation Characteristics characteristics Availability

NOD.Cg-Prkdcscid NSG Mutation is a NOD strain; Lacks T, B and The Jackson IL2rgtm1Wjl complete null, is not immunodeficient and NK cells, Laboratory, expressed and will relatively radiosensitive additional defects stock 005557 not bind cytokines due to a defect in DNA in innate immune repair cells NOD.cg-Prkdcscid NOG Lacks the NOD strain; Lacks T, B and Taconic IL2rgtm1Sug intracytoplasmic immunodeficient and NK cells, Biosciences, domain and will relatively radiosensitive additional defects stock CIEA bind cytokines but due to a defect in DNA in innate immune NOG mouse will not signal repair cells NOD.Cg-Rag1tm1Mom NRG Mutation is a NOD strain; Lacks T, B and The Jackson IL2rgtm1Wjl complete null, is not immunodeficient and NK cells, Laboratory, expressed and will relatively radioresistant additional defects stock 007799 not bind cytokines in innate immune cells C.Cg-Rag2tm1Fwa BRG Lacks the Mixed background, Lacks T, B and Taconic IL2rgtm1Sug intracytoplasmic predominately BALB/c NK cells, Biosciences, domain and will bind strain; immunodeficient remaining innate stock 11503 cytokines but will not and relatively immune cells are signal radioresistant functional The four strains of immunodeficient mice bearing targeted mutations in the IL-2 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 was provided previously (5,8). NOD, nonobese diabetic. Taconic Biosciences, Hudson, NY.

390 American Journal of Transplantation 2016; 16: 389–397 Humanized Mice in Transplant Immunology

Human skin allografts rejection (8,13). Simultaneous injection of allogeneic Allogeneic rejection of human allografts by humanized mice PBMCs with islet grafts into NSG mice results in rapid has been studied for >20 years, and this literature was graft rejection and return to hyperglycemia within 21 days in reviewed recently by us (13) and others (8). Early studies 100% of mice (24). When islet grafts were allowed to with the Hu-PBL-SCID model using CB17-scid mice led to become established in diabetic NSG mice for 37 days prior inconsistent results due to poor human T cell engraftment, to allogeneic PBMC injection, the return to hyperglycemia which was improved by elimination of host NK cells using occurred in two of three mice before the termination of the anti–monosialotetrahexosylganglioside 1 (anti-GM1) anti- study, due to the development of GVHD (24). body or addition of the Lystbg (beige) mutation to the CB17-scid strain (8,13). Using CB17-Lystbg Prkdcscid (SCID/ The Hu-SRC-SCID model develops a more complete beige) mice led to the conclusion that human CD4 and CD8 immune system but is limited by the selection of human T cells were both important in skin allograft rejection (14,15), T cells on mouse MHC. NSG and BRG HSC-engrafted mice and that was confirmed using NOD.CB17 Prkdcscid b2mnull have shown contrasting results when transplanted with (NOD-scid b2mnull) mice (16). Transitioning to NSG mice, allogeneic human islets. HSC-engrafted hyperglycemic which achieve high levels of human cell engraftment, it was BRG mice are not able to reject human islets and exhibit observed that they needed to be treated with rat anti– minimal T cell infiltration into the graft (25). In contrast, mouse granulocyte (anti-Gr1) antibody after skin transplan- hyperglycemic HSC-engrafted NOD-Rag1tm1Mom IL2gnull tation for successful healing of the graft due to the Ins2Akita (NRG-Akita) mice engrafted with 5 104 UCB– infiltration of mouse innate immune cells (9). Skin grafts derived CD34þ HSCs were shown to reject 60% of on NSG mice treated with anti-Gr1 mAb heal and exhibit human islet allografts, with evidence of strong mononucle- excellent vascularization and morphology. In the Hu-PBL- ar cell infiltration into the rejecting grafts (26). In those mice SCID model using NSG mice, 100% of allografts are that did not revert to hyperglycemia, a mononuclear rejected within 21 days, providing a robust and reproducible infiltrate into the graft was still observed. model for study of human T cell–mediated skin allograft rejection (9). In a more recent study following injection of higher numbers of UCB-derived CD34þ HSCs (2 105) into NSG Further studies characterizing skin allograft–infiltrating mice, all of the islet grafts were rejected within 17 days T cells documents the presence of human CD4 and CD8 after transplantation, as shown by the return to hypergly- T cells producing IL-17A in the graft dermis and epi- cemia (27). Graft rejection in NSG mice was confirmed by dermis (17). IL-17A has strong inflammatory properties the loss of histological staining for human insulin and was and has been associated with psoriasis (18–20) and atopic associated with infiltration of human CD4 T cells, macro- dermatitis (21). T regulatory cells (Tregs) are a major phages and neutrophils. Interestingly, in this model, very player in immunoregulation, and CD4þFoxP3þ T cells are few CD8 T cells were found to infiltrate the graft, which is in present in the basal layer of skin grafts on SCID/beige contrast to Hu-PBL-SCID models. Of particular interest in mice after PBMC injection, suggesting an attempt by this report is that injection of 6 106 ex vivo expanded theimmunesystemtodownmodulatetheimmune human Tregs completely prevented islet graft rejection (27). inflammation (17). 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 Human islet allografts these models may be due to differences in the host strain or Islet transplantation has been used to successfully restore in the numbers of human HSCs used for engraftment, glucose homeostasis in patients with type 1 diabetes, but leading to differences in human immune function in these most patients subsequently lose insulin independence over models. time, and the procedure comes with adverse side effects, predominantly due to the use of immunosuppressive drugs to prevent graft rejection (22). Humanized mouse models Human cardiac tissue are becoming a valuable tool to study the mechanisms of The endothelial cells lining arterial grafts are immunogenic islet allograft rejection and to test potential therapeutics and have been shown to activate allogeneic human T prior to their advancement to the clinic. cells (28). Many of the studies relating to cardiac grafts in humanized mice were performed in humanized SCID/beige The Hu-PBL-SCID model has been used to study islet mice due to the high sensitivity of NSG mice to paralysis rejection, but there is a ‘‘race’’ between the rejection of the and/or death after human artery transplantation (29). The islets and the development of GVHD, making long-term majority of SCID/beige mice receiving artery transplants monitoring for therapeutic interventions difficult (23). Islet followed by allogeneic PBMCs exhibit enhanced infiltration transplantation into humanized mice has been ongoing for of human cells into the graft and histological changes 25 years; however, prior to the development of NSG consistent with rejection (30). The extensive study of this mice, T cell engraftment and islet rejection were variable model has revealed several mediators of graft injury that and required complex approaches to achieve complete have possible therapeutic potential. Infiltrating T cells

American Journal of Transplantation 2016; 16: 389–397 391 Kenney et al

Table 2: Recipient strain and engraftment protocols affect islet allograft survival in humanized mice Time after Engraftment HSC source and engraftment islets Recipient strain protocol number injected transplanted Graft outcome Reference

BALB/c-Rag2null Pregnant BRG mice 1 105 CD34þ UCB 300–500 human islets Human CD45 engraftment levels of 25 IL2rgnull injected with cells injected transplanted in the 22.0 16.6% in the blood, grafts (BRG) 0.5 mg busulfan intrahepatically subrenal capsule analyzed histologically at day 14 2–7 days prior to 8–26 weeks after or 35 after transplantation; no delivery, BRG CD34 HSC evidence of rejection determined newborns engraftment into by no loss of insulin staining, and irradiated with normoglycemic little CD3 or CD20 infiltration 550 cGy BRG mice observed in the graft or loss of human C-peptide in the serum NOD.Cg- 1- to 3-day-old NRG- 5 104 CD34þ 4000 human islets NSG mice with 12.9 2.5% human 26 Rag1tm1Mom Akita mice UCB cells transplanted in the CD45þ blood cells used in IL2rgtm1Wjl irradiated with subrenal capsule experiment; 8 of 13 (62%) islet Ins2Akita 400 cGy into diabetic NRG- grafts were rejected, as defined (NRG-Akita) Akita mice by a return to hyperglycemia and loss of insulin staining in graft; CD45 infiltration observed even in nonrejected grafts NOD.Cg- 4–6 week old NSG 20 104 CD34þ 3000–4000 human NSG mice with >15% human 27 Prkdcscid irradiated with UCB cells islets transplanted CD45þ blood cells used in IL2rgtm1Wjl 240 cGy in the subrenal experiment; 21 of 21 (100%) (NSG) capsule into diabetic rejected islet grafts, as defined by NSG mice, mice return to hyperglycemia and loss treated with of insulin staining in graft; streptozotocin to infiltration of graft with induce diabetes (CD11b), (timing not reported neutrophils (CD66b) and CD4 T in relation to HSC cells but few to no CD8 T cells engraftment or islet transplantation) Engraftment of immunodeficient mice bearing targeted mutations in the IL2rg gene with human CD34þ umbilical cord blood HSCs and human islets. HSC, ; UCB, umbilical cord blood. The derivation of the BRG strain, the origin of the Rag2null mutation and the origin of the IL2rgnull mutation were not reported. produce interferon-g (IFN-g) (31), which sustains enhanced area (37). A similar reduction of graft injury was found HLA-DR expression on endothelial cells lining the graft and using a human-specific IL-1R antagonist, accompanied by induces neointima, or scarring of the inner most layer of the reduced infiltration of human immune cells and decreased blood vessels within the human arterial graft (32). Neutrali- IL-17 production. These studies suggest that arterial zation of IFN-g blocks rejection of grafts by infiltrating grafts produce cytokines that both protect and enhance T cells (31). Arterial graft pretreatment with rapamyacin allogeneic immune responses. Using the Hu-SRC SCID/ reduces graft injury and the number of T cells that infiltrate beige humanized mouse model, which has poor engraft- into the intima (33). This protection from graft injury is ment of T and B cells, human arterial grafts showed graft mediated by the upregulation of programmed death ligand injury with the formation of neointimas that contained 1 (PD-L1) and 2; conversely, blockade of programmed cell macrophages but no CD3-, CD11c-, or CD19-positive death 1 increases graft injury (33), suggesting that an cells, suggesting a role for macrophages in arterial graft immunomodulatory approach to prevent arterial graft injury (29). rejection may be effective. Pluripotent stem cells and stem cell–derived cell Alternatively, blockade of transforming growth factor b populations (TGF-b) enhances graft injury (34). TGF-b is produced by A new area of transplantation biology is the use of mature the arterial graft and reduces the ability of the infiltrating cell populations differentiated from pluripotent stem cells. allogeneic T cells to produce IFN-g. Moreover, proin- The major question involved in these transplantations flammatory cytokines IL-1 and IL-6 produced from the is whether the pluripotent stem cells and their differen- arterial graft appear to promote graft injury (35,36). IL-6– tiated progeny are susceptible to autologous or allogeneic blocking antibody reduces graft injury, with a reduction in immune responses. In the murine system, it has total vessel area and the intimal area and increased lumen been reported that differentiated cells from induced

392 American Journal of Transplantation 2016; 16: 389–397 Humanized Mice in Transplant Immunology

pluripotent stem cells (iPSCs) elicit immune responses approach, definition of successful immune ‘‘engraftment’’ when transplanted into syngeneic hosts (38). In contrast, a and evaluation of immune rejection, and the efficacy of the recent report suggested that differentiated cells derived regulatory population in preventing rejection of allografts from iPSCs fail to elicit an immune response in syngeneic are not well defined and are ‘‘model and system’’ murine hosts (39). Understanding the immunogenicity of dependent. Depending on the recipient strain and model human cells and tissues derived from iPSCs or embryonic system, 5–300 106 effector PBMCs are required for stem cells (ESCs) will be an important consideration as this infiltration and damage to the target graft. cellular therapy progresses toward clinical trials. In BRG mice transplanted with human aortic grafts and Using the NSG mouse engrafted with human fetal liver and injected with 10 106 PBMCs with or without 1–2 106 thymus and injected with autologous liver CD34þ HSCs Tregs, it was observed that the addition of Tregs could (BLT model), it was shown that allogeneic teratomas prevent the development of transplant arteriosclerosis derived from human ESCs are heavily infiltrated with when studied histologically 30 days later (42). Graft immune cells; regress in size; and, after 6 weeks, preservation was associated with a decrease in the consist mostly of liquid cysts, suggesting they are being secretion of IFN-g. In BRG mice engrafted with human rejected (38). Similarly, human ESCs differentiated into skin for 35 days, injection of 5 106 human PBMCs fibroblasts or cardiomyocytes and transplanted subcutane- (engraftment of human PBMCs defined as >1% human ously or intramuscularly, respectively, into NSG-BLT mice CD45þ cells in the spleen) led to rejection, with a median became necrotic and infiltrated with human immune cells. survival time (MST) of 34–40 days (43). Interestingly, To prevent injury of transplanted human ESCs and their in this report, the authors also showed that skin grafts differentiated fibroblasts or cardiomyocytes, the human survived for >100 days following injection of PBMCs ESCs were genetically engineered to express cytotoxic autologous to the skin graft (i.e. from the same donor), T lymphocyte associated protein 4 (CTLA4) Ig and PD-L1, highlighting that the rejection appeared to be an allogeneic which modulate T cell costimulatory pathways. ESC and not a nonspecific immune response. When allogeneic teratomas and differentiated fibroblasts and cardiomyo- PBMCs were mixed with ex vivo expanded Tregs at a 1:1 cytes expressing these molecules were effectively pro- ratio prior to injection, no skin graft rejection was observed tected from human immune infiltration in NSG-BLT up to 100 days (the duration of the experiment) (43). mice (40). It must be cautioned, however, that relying Recently, the ability of Tregs to prevent allograft rejection solely on the presence or absence of human T cell was extended to islet allografts using similar ap- infiltration is not a sufficient measure of graft rejection proaches (44). In this study, a high number of human because infiltrates, in fact, may be suppressing the immune islets (8000 islet equivalents) were transplanted into BRG response (39). mice for 14 days, followed by injection of 40 106 PBMCs with or without an equal number of ex vivo expanded The ability of costimulation blockade to protect iPSC- Tregs. Islet allografts receiving only allogeneic PBMCs derived beta cells from both xenogeneic and allogeneic were rejected with an MST of 23 days, but when an equal rejection has been reported (41). Human iPSC-derived beta number of Tregs were coinjected, MST was >45 days, cells are rapidly rejected when transplanted into immuno- with only two of the 13 mice receiving Tregs rejecting the competent mice. Similarly, human iPSC-derived beta cells grafts (44). are also rejected in NSG mice injected with 15 106 allogeneic human PBMCs (41). In contrast, both xenograft In a similar study using NSG or BRG mice engrafted with and allogeneic rejection are prevented when the mice are human skin for 4–6 weeks and injected with 5 106 treated with the costimulation blockade agents CTLA4 Ig PBMCs with or without coinjection of 1 106 Tregs plus anti-CD154 (anti-CD40L) antibody (41). These data (engraftment defined as >0.5% human CD45þ splenic combined with previous reports suggest that iPSC-derived cells), allospecific Tregs were found to be more potent cells and tissues can be rejected by allogeneic human inhibitors of skin graft rejection than polyclonal expanded immune systems but that this rejection can be prevented Tregs (45). In an extension of this study, a method to by costimulation blockade. generate clinical-grade allospecific Tregs was described, and again, these allospecific Tregs transplanted into BRG Human Tregs and mesenchymal stem cells as mice with human skin grafts were found to be more potent regulatory cells inhibitors of graft injury than polyclonal expanded Tregs The ability to evaluate the in vivo efficacy of human under very similar conditions but with increased numbers of regulatory populations to prevent graft rejection in human- PBMCs (10 106) and Tregs (2 106) (46). Graft injury in ized mice clearly has advantages as a test for their both studies was evaluated at 4 weeks after PBMC functional capability and as a model that allows the injection and consisted of analyses of human CD45 optimization of the source and ex vivo expansion protocols infiltration, Ki67 and involucrin staining, expression of for human regulatory cells. Nevertheless, as shown by the terminal transferase-mediated dUTP nick end labeling– approaches described below, the choice of the immunode- positive nuclei in dermal infiltrates, preservation of human ficient strain, allograft target, human immune engraftment CD31 staining, and detection of CD4þFoxP3þ Tregs.

American Journal of Transplantation 2016; 16: 389–397 393 Kenney et al

Overall graft survival was not reported in either that CD4 effector/memory T cells but not CD4 central study (45,46). memory T cells targeted endothelial cells in skin grafts, and in vitro analyses indicated that these cells mediated In a variation of this approach, CB17-scid mice were their effect in part by secretion of IFN-g (52). Immunode- transplanted with human skin grafts, and 6 weeks later, ficient mice bearing human skin grafts injected with the anti-asialo GM1 antibody was administered to deplete anti-HLA rat antibody W6/32 were shown to have activated murine host NK cells, followed by injection of 300 106 endothelial cell exocytosis and leukocyte trafficking in allogeneic PBMCs. Complete rejection of the graft was not the graft, suggesting that alloantibodies may have a role observed, but addition of bone marrow or adipose-derived in transplant rejection (53). This study was extended mesenchymal stem cells as an immunoregulatory popula- and showed that, combined with >6 hours of cold ische- tion to the allogeneic PBMCs led to decreased leukocyte mia, antibody administration—regardless of specificity— infiltration into the graft and reduced expression of promoted transplant vasculopathy, as demonstrated by inflammatory cytokines such as IFN-g, tumor necrosis intimal expansion in human vessels transplanted into BRG factor a, IL-1b and IL-6 (47). These studies suggest that mice (54). Using an in vitro model system, the binding of humanized mice can be used to evaluate the ability of anti-HLA antibodies to human aortic, venous and microvas- human regulatory cells to modulate allograft rejection, but cular endothelial cells induced endothelial P-selectin and the results show that variations in the approach may increased adherence of monocytes, implying a role of influence the data generated. alloantibodies in recruitment of monocytes that participate in antibody-mediated rejection of human allografts (55). Xenografts This suggestion was confirmed in vivo using a mouse/ Xenograft transplantation using porcine grafts provides a mouse transplant model of C57BL/6-Rag1null recipients of potentially unlimited source for donor tissues if the BALB/c cardiac allografts that were passively transferred xenograft rejection process could be overcome. Human- with donor-specific MHC class I antibodies (56). ized 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 Remaining limitations and future opportunities with 5000 neonatal porcine islet cell clusters (NICCs), which Although humanized mice represent exciting new models restored normoglycemia (48). When 1–10 106 human that permit the direct study of human transplantation in a PBMCs were injected, the NICC grafts were destroyed in a small animal model without putting patients at risk, PBMC number-dependent manner. Graft infiltration could limitations remain for the available models. be blocked by the coinjection of ex vivo expanded human Tregs, but that was mediated in part by Treg production of The development of GVHD in almost all of the model IL-10 (49). systems represents a constraint on the experimental window available for the study of graft survival, particularly Extending these observations to investigate the ability of in the Hu-PBL-SCID model system (57); however, much of central tolerance in the thymus to induce xenograft the xenoreactivity of human immune systems to murine tolerance, NSG mice were transplanted with a porcine antigens is to the MHC class I and II molecules (57). In the thymus and injected with human CD34þ HSCs (50). In this NSG-BLT model, CD8 T cells, which are reactive to murine system, human T cells were generated, and they were MHC class I, appear to have a central role in the specifically nonreactive to the murine host, the human HSC development of GVHD symptoms (58). A recently de- donor and the MHC of the porcine donors. Skin grafts from scribed model based on the development of B6.129- porcine donors sharing the MHC of the thymus were not Rag2tm1Fwa CD47 tm1fpl IL2rgtm1cgn mice suggests that the rejected, whereas skin grafts from other non–MHC- lack of murine CD47 decreases the development of GVHD matched porcine donors were rapidly rejected (50). These symptoms in BLT humanized mice (59). Development of various xenograft models will be valuable tools for the study immunodeficient hosts deficient in murine MHC and/or of human xenograft responses and will accelerate efforts to CD47 may decrease the development of GVHD symptoms identify approaches for the induction of xenotolerance and and extend the experimental window available for the study permanent xenograft survival. of transplantation.

Naıve,€ effector, and memory T cells and Additional opportunities to improve humanized mouse alloantibodies in graft survival models include optimizing models for enhanced humoral Humanized mice have also been used to investigate the responses, memory T cell formation, lymphoid structures role of T cells involved in allograft rejection and the role of and germinal center formation, which were recently alloantibodies in graft survival. In early studies in humanized reviewed extensively (5,60). Due to the lack of reactivity mice, it was shown that memory but not na€ıve T cells were between many murine cytokines and human cells, additional the principal mediators of skin graft injury and that this improvements in the murine host—focusing on the provision destruction could be blocked using a 4-1BB ligand, an ICOS of species-specific human cytokines important in human ligand or an OX40 ligand (51). This was extended to show immune development—are under way in a number of

394 American Journal of Transplantation 2016; 16: 389–397 Humanized Mice in Transplant Immunology

laboratories (5–7). Mouse B cell activating factor (BAFF; also community. Improvements in immune system engraft- termed B lymphocyte stimulator), for example, binds to but ment, immune function and the potential for development does not signal human B cells (61), and efforts are under way human myeloid cells, an important component of allograft to generate human BAFF transgenic mice in our laboratory to rejection, are under development. The use of humanized determine whether this will improve human B cell responses mice as a model for the study of human immune rejection of to immunization. Nevertheless, recent reports on transgenic allografts and xenografts may provide novel insights into generation of various human species-specific cytokines the mechanisms responsible for graft rejection and permit suggest caution because these factors are expressed in the rapid evaluation of new approaches to prevent graft loss in murine host. Murine IL-2, for example, does not cross-react transplant recipients. with human T cells, and transgenic expression of human IL-2 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, Disclosure IL-3, granulocyte colony-stimulating factor, macrophage colony-stimulating factor and signal-regulatory The authors of this manuscript have conflicts of interest to protein a leads to robust human innate immune develop- disclose as described by the American Journal of Trans- ment, but the experimental window of this strain is relatively plantation. Drs. Michael Brehm and Dale Greiner are limited (63). Additional modifications of the murine host to consultants for the Jackson Laboratory. The other authors reduce macrophages and granulocyte populations will likely have no conflicts of interest to disclose. be important to achieve circulation of human red blood cells, and platelets, which circulate at only low levels in the currently available models of humanized mice (5–7). References

Transgenic expression of human HLA class I and II 1. Mestas J, Hughes CC. Of mice and not men: Differences between molecules has also been shown to enhance development mouse and human immunology. J Immunol 2004; 172: 2731– of HLA-restricted T cell responses in the Hu-SRC-SCID 2738. model. HLA-A2 transgenic NSG mice engrafted with HLA- 2. van den Heuvel H, Heidt S, Roelen DL, Claas FH. T-cell alloreac- A2 HSCs, for example, have been shown to generate HLA- tivity and transplantation outcome: A budding role for hetero- A2–restricted Epstein–Barr virus (EBV)-specific cytotoxic T logous immunity? Curr Opin Organ Transplant 2015; 20: 454– cells in humanized mice (64,65). Extension of the use of 460. human immune engrafted HLA-Tg immunodeficient mice 3. Amir AL, D’Orsogna LJ, Roelen DL, et al. Allo-HLA reactivity of to allograft rejection has not been reported. Using EBV- virus-specific memory T cells is common. Blood 2010; 115: 3146–3157. infected mice and/or the HLA-restricted cell lines that have 4. Macedo C, Orkis EA, Popescu I, et al. Contribution of naive and been reported (66) may be an approach to study the role of memory T-cell populations to the human alloimmune response. human memory cells in allograft rejection, an area of Am J Transplant 2009; 9: 2057–2066. investigation that has not been pursued to date in 5. Shultz LD, Brehm MA, Garcia-Martinez JV, Greiner DL. Humanized humanized mice. mice for immune system investigation: Progress, promise and challenges. Nat Rev Immunol 2012; 12: 786–798. The current models of humanized mice have limited B cell 6. Ito R, Takahashi T, Katano I, Ito M. Current advances in humanized maturation with low antibody levels and deficiencies in their mouse models. Cell Mol Immunol 2012; 9: 208–214. ability to isotype switch following exposure to antigen, 7. Rongvaux A, Takizawa H, Strowig T, et al. Human hemato- resulting in few or no human IgG antibody responses lymphoid system mice: Current use and future potential for medicine. Annu Rev Immunol 2013; 31: 635–674. following immunization. Two groups, however, have 8. Hogenes M, Huibers M, Kroone C, de Weger R. Humanized mouse reported that engraftment of human HLA-DR4 transgenic models in transplantation research. Transplant Rev (Orlando) 2014; NRG/NOG mice with HSCs from HLA-DR4 donors leads to 28: 103–110. improved human IgG production following immuniza- 9. Racki WJ, Covassin L, Brehm M, et al. NOD-scid IL2rgnull (NSG) tion (67,68). Although the available models of humanized mouse model of human skin transplantation and allograft rejection. mice likely would generate only poor and predominately Transplantation 2010; 89: 527–536. IgM alloantibody responses to tissue grafts, the injection of 10. McCune JM, Namikawa R, Kaneshima H, Shultz LD, Lieberman M, human serum containing known alloantibodies into human- Weissman IL. The SCID-hu mouse: Murine model for the analysis ized mice might be an approach to study the role of of human hematolymphoid differentiation and function. Science alloantibodies in graft rejection in this model system. 1988; 241: 1632–1639. 11. Lan P, Tonomura N, Shimizu A, Wang S, Yang YG. Reconstitution of a functional human immune system in immunodeficient mice through combined human fetal thymus/liver and CD34þ cell Conclusions transplantation. Blood 2006; 108: 487–492. 12. Melkus MW, Estes JD, Padgett-Thomas A, et al. Humanized mice Exciting new models of humanized mice to study trans- mount specific adaptive and innate immune responses to EBV and plant rejection are becoming available to the research TSST-1. Nat Med 2006; 12: 1316–1322.

American Journal of Transplantation 2016; 16: 389–397 395 Kenney et al

13. Brehm MA, Shultz LD. Human allograft rejection in humanized reconstitution in severe combined immunodeficient mice. Trans- mice: A historical perspective. Cell Mol Immunol 2012; 9: 225–231. plantation 1999; 67: 897–903. 14. Murray AG, Petzelbauer P, Hughes CC, Costa J, Askenase P, 31. Wang Y, Burns WR, Tang PC, et al. Interferon-gamma plays a Pober JS. Human T-cell-mediated destruction of allogeneic dermal nonredundant role in mediating T cell-dependent outward vascular microvessels in a severe combined immunodeficient mouse. Proc remodeling of allogeneic human coronary arteries. FASEB J 2004; Natl Acad Sci U S A 1994; 91: 9146–9150. 18: 606–608. 15. Murray AG, Schechner JS, Epperson DE, et al. Dermal microvas- 32. Pober JS, Bothwell AL, Lorber MI, McNiff JM, Schechner JS, cular injury in the human peripheral blood lymphocyte reconsti- Tellides G. Immunopathology of human T cell responses to skin, tuted-severe combined immunodeficient (HuPBL-SCID) mouse/ artery and endothelial cell grafts in the human peripheral blood skin allograft model is T cell mediated and inhibited by a lymphocyte/severe combined immunodeficient mouse. Springer combination of cyclosporine and rapamycin. Am J Pathol 1998; Semin Immunopathol 2003; 25: 167–180. 153: 627–638. 33. Yi T, Fogal B, Hao Z, et al. Reperfusion injury intensifies the 16. Turgeon NA, Banuelos SJ, Shultz LD, et al. Alloimmune injury and adaptive human T cell alloresponse in a human-mouse chimeric rejection of human skin grafts on human peripheral blood artery model. Arterioscler Thromb Vasc Biol 2012; 32: 353–360. lymphocyte-reconstituted non-obese diabetic severe combined 34. Lebastchi AH, Khan SF, Qin L, et al. Transforming growth factor immunodeficient beta2-microglobulin-null mice. Exp Biol Med beta expression by human vascular cells inhibits interferon gamma (Maywood) 2003; 228: 1096–1104. production and arterial media injury by alloreactive memory T cells. 17. de Oliveira VL, Keijsers RR, van de Kerkhof PC, et al. Humanized Am J Transplant 2011; 11: 2332–2341. mouse model of skin inflammation is characterized by disturbed 35. Fogal B, Hewett SJ. -1beta: A bridge between keratinocyte differentiation and influx of IL-17A producing T cells. inflammation and excitotoxicity? J Neurochem 2008; 106: 1–23. PLoS One 2012; 7: e45509. 36. Rao DA, Eid RE, Qin L, et al. Interleukin (IL)-1 promotes allogeneic 18. Wilson NJ, Boniface K, Chan JR, et al. Development, cytokine T cell intimal infiltration and IL-17 production in a model of human profile and function of human interleukin 17-producing helper T artery rejection. J Exp Med 2008; 205: 3145–3158. cells. Nat Immunol 2007; 8: 950–957. 37. Fogal B, Yi T, Wang C, et al. Neutralizing IL-6 reduces human 19. Lowes MA, Kikuchi T, Fuentes-Duculan J, et al. Psoriasis vulgaris arterial allograft rejection by allowing emergence of CD161þ lesions contain discrete populations of Th1 and Th17 T cells. CD4þ regulatory T cells. J Immunol 2011; 187: 6268–6280. J Invest Dermatol 2008; 128: 1207–1211. 38. Zhao T, Zhang ZN, Rong Z, Xu Y. Immunogenicity of induced 20. Hueber W, Patel DD, Dryja T, et al. Effects of AIN457, a fully human pluripotent stem cells. Nature 2011; 474: 212–215. antibody to interleukin-17A, on psoriasis, rheumatoid arthritis, and 39. de Almeida PE, Meyer EH, Kooreman NG, et al. Transplanted uveitis. Sci Transl Med 2010; 2: 52ra72. terminally differentiated induced pluripotent stem cells are 21. Koga C, Kabashima K, Shiraishi N, Kobayashi M, Tokura Y. Possible accepted by immune mechanisms similar to self-tolerance. Nat pathogenic role of Th17 cells for atopic dermatitis. J Invest Commun 2014; 5: 3903. Dermatol 2008; 128: 2625–2630. 40. Rong Z, Wang M, Hu Z, et al. An effective approach to prevent 22. Vantyghem MC, Defrance F, Quintin D, et al. Treating diabetes immune rejection of human ESC-derived allografts. Cell Stem Cell with islet transplantation: Lessons from the past decade in Lille. 2014; 14: 121–130. Diabetes Metab 2014; 40: 108–119. 41. Szot GL, Yadav M, Lang J, et al. Tolerance induction and reversal 23. King M, Pearson T, Rossini AA, Shultz LD, Greiner DL. Humanized of diabetes in mice transplanted with human embryonic stem mice for the study of type 1 diabetes and beta cell function. Ann N cell-derived pancreatic endoderm. Cell Stem Cell 2015; 16: Y Acad Sci 2008; 1150: 46–53. 148–157. 24. King M, Pearson T, Shultz LD, et al. A new Hu-PBL model for the 42. Nadig SN, Wieckiewicz J, Wu DC, et al. In vivo prevention of study of human islet alloreactivity based on NOD-scid mice bearing transplant arteriosclerosis by ex vivo-expanded human regulatory a targeted mutation in the IL-2 receptor gamma chain gene. Clin T cells. Nat Med 2010; 16: 809–813. Immunol 2008; 126: 303–314. 43. Issa F, Hester J, Goto R, Nadig SN, Goodacre TE, Wood K. Ex vivo- 25. Jacobson S, Heuts F, Juarez J, et al. Alloreactivity but failure to expanded human regulatory T cells prevent the rejection of skin reject human islet transplants by humanized Balb/c/Rag2gc mice. allografts in a humanized mouse model. Transplantation 2010; 90: Scand J Immunol 2010; 71: 83–90. 1321–1327. 26. Brehm MA, Bortell R, DiIorio P, et al. Human immune system 44. Wu DC, Hester J, Nadig SN, et al. Ex vivo expanded human development and rejection of human islet allografts in spontane- regulatory T cells can prolong survival of a human islet allograft in a ously diabetic NOD-Rag1null IL2rgnull Ins2Akita mice. Diabetes 2010; humanized mouse model. Transplantation 2013; 96: 707–716. 59: 2265–2270. 45. Sagoo P, Ali N, Garg G, Nestle FO, Lechler RI, Lombardi G. Human 27. Xiao F, Ma L, Zhao M, et al. Ex vivo expanded human regulatory T regulatory T cells with alloantigen specificity are more potent cells delay islet allograft rejection via inhibiting islet-derived inhibitors of alloimmune skin graft damage than polyclonal monocyte chemoattractant protein-1 production in CD34þ stem regulatory T cells. Sci Transl Med 2011; 3: 83ra42. cells-reconstituted NOD-scid IL2rgammanull mice. PLoS One 46. Putnam AL, Safinia N, Medvec A, et al. Clinical grade manufactur- 2014; 9: e90387. ing of human alloantigen-reactive regulatory T cells for use in 28. Thomsen M, Yacoub-Youssef H, Marcheix B. Reconstitution of a transplantation. Am J Transplant 2013; 13: 3010–3020. human immune system in immunodeficient mice: Models of 47. Roemeling-van Rhijn M, Khairoun M, Korevaar SS, et al. Human human alloreaction in vivo. Tissue Antigens 2005; 66: 73–82. bone marrow- and adipose tissue-derived mesenchymal stromal 29. Kirkiles-Smith NC, Harding MJ, Shepherd BR, et al. Development cells are immunosuppressive and in a humanized allograft rejection of a humanized mouse model to study the role of macrophages in model. J Stem Cell Res Ther 2013; (Suppl 6): 20780. allograft injury. Transplantation 2009; 87: 189–197. 48. Ji M, Jin X, Phillips P, Yi S. A humanized mouse model to study 30. Lorber MI, Wilson JH, Robert ME, et al. Human allogeneic vascular human immune response in . Hepatobiliary rejection after arterial transplantation and peripheral lymphoid Pancreat Dis Int 2012; 11: 494–498.

396 American Journal of Transplantation 2016; 16: 389–397 Humanized Mice in Transplant Immunology

49. Yi S, Ji M, Wu J et al. Adoptive transfer with in vitro expanded 59. Lavender KJ, Pang WW, Messer RJ, et al. BLT-humanized C57BL/ human regulatory T cells protects against porcine islet xenograft 6 Rag2-/-gammac-/-CD47-/- mice are resistant to GVHD and rejection via interleukin-10 in humanized mice. Diabetes 2012; 61: develop B- and T-cell immunity to HIV infection. Blood 2013; 1180–1191. 122: 4013–4020. 50. Kalscheuer H, Onoe T, Dahmani A et al. Xenograft tolerance and 60. Brehm MA, Shultz LD, Luban J, Greiner DL. Overcoming current immune function of human T cells developing in pig thymus limitations in humanized mouse research. J Infect Dis 2013; 208 xenografts. J Immunol 2014; 192: 3442–3450. (Suppl 2): S125–S130. 51. Shiao SL, McNiff JM, Pober JS. Memory T cells and their 61. Schmidt MR, Appel MC, Giassi LJ, Greiner DL, Shultz LD, costimulators in human allograft injury. J Immunol 2005; 175: Woodland RT. Human BLyS facilitates engraftment of human 4886–4896. PBL derived B cells in immunodeficient mice. PLoS One 2008; 3: 52. Shiao SL, Kirkiles-Smith NC, Shepherd BR, McNiff JM, Carr EJ, e3192. Pober JS. Human effector memory CD4þ T cells directly recognize 62. Katano I, Takahashi T, Ito R, et al. Predominant development of allogeneic endothelial cells in vitro and in vivo. J Immunol 2007; mature and functional human NK cells in a novel human IL-2- 179: 4397–4404. producing transgenic NOG mouse. J Immunol 2015; 194: 53. Yamakuchi M, Kirkiles-Smith NC, Ferlito M, et al. Antibody to 3513–3525. triggers endothelial exocytosis. Proc Natl 63. Rongvaux A, Willinger T, Martinek J, et al. Development and Acad Sci U S A 2007; 104: 1301–1306. function of human innate immune cells in a humanized mouse 54. Goto R, Issa F, Heidt S, Taggart D, Wood KJ. Ischemia-reperfusion model. Nat Biotechnol 2014; 32: 364–372. injury accelerates human antibody-mediated transplant vasculop- 64. Strowig T, Gurer C, Ploss A, et al. Priming of protective T cell athy. Transplantation 2013; 96: 139–145. responses against virus-induced tumors in mice with human 55. Valenzuela NM, Mulder A, Reed EF. HLA class I antibodies immune system components. J Exp Med 2009; 206: 1423–1434. trigger increased adherence of monocytes to endothelial cells by 65. Shultz LD, Saito Y, Najima Y, et al. Generation of functional human eliciting an increase in endothelial P-selectin and, depending on T-cell subsets with HLA-restricted immune responses in HLA class subclass, by engaging FcgammaRs. J Immunol 2013; 190: I expressing NOD/SCID/IL2r gamma(null) humanized mice. Proc 6635–6650. Natl Acad Sci U S A 2010; 107: 13022–13027. 56. Valenzuela NM, Hong L, Shen XD, et al. Blockade of p-selectin is 66. Burrows SR, Silins SL, Khanna R, et al. Cross-reactive memory T sufficient to reduce MHC I antibody-elicited monocyte recruitment cells for Epstein-Barr virus augment the alloresponse to common in vitro and in vivo. Am J Transplant 2013; 13: 299–311. human leukocyte antigens: Degenerate recognition of major 57. King MA, Covassin L, Brehm MA, et al. Human peripheral blood histocompatibility complex-bound peptide by T cells and its role leucocyte non-obese diabetic-severe combined immunodeficien- in alloreactivity. Eur J Immunol 1997; 27: 1726–1736. cy interleukin-2 receptor gamma chain gene mouse model of 67. Suzuki M, Takahashi T, Katano I, et al. Induction of human humoral xenogeneic graft-versus-host-like disease and the role of host immune responses in a novel HLA-DR-expressing transgenic major histocompatibility complex. Clin Exp Immunol 2009; 157: NOD/Shi-scid/gammacnull mouse. Int Immunol 2012; 24: 104–118. 243–252. 58. Laing ST, Griffey SM, Moreno ME, Stoddart CA. CD8-positive 68. Danner R, Chaudhari SN, Rosenberger J, et al. Expression of HLA lymphocytes in graft-versus-host disease of humanized NOD. class II molecules in humanized NOD. Rag1KO.IL2RgcKO mice is Cg-Prkdc(scid) Il2rg(tm1Wjl)/SzJ mice. J Comp Pathol 2015; 152: critical for development and function of human T and B cells. PLoS 238–242. One 2011; 6: e19826.

American Journal of Transplantation 2016; 16: 389–397 397