Humanized mouse models: Application to human diseases

1.1 Production of humanized mice

Various humanized mouse models, which are engrafted with human transplants, including peripheral blood mononuclear cells (PBMCs), a combination of bone marrow, liver, and thymus (BLT), and hematopoietic stem cells (HSCs), are widely used for the biomedical research. In human PBMC-transferred mice, which are classically termed hu-scid mice (Mosier, Gulizia, Baird, & Wilson, 1988), human T cells were highly engrafted and expanded, and the mice developed severe graft-versus-host disease (GVHD) (Ito et al., 2009; King et al., 2009; van Rijn et al., 2003). PBMC-transferred models are used predominantly to investigate the mechanisms of GVHD, graft rejection, or T-cell suppression, and for screening of therapeutic drugs. In the BLT model, a fragment of human fetal liver (FL) and thymus are implanted under the kidney capsule of the mice, and CD34⁺ HSC from the same FL are intravenously injected (Brainard et al., 2009; Denton et al., 2008). Although BLT mice are highly susceptible to human immunodeficiency virus (HIV) infection because the human fetal thymus provides a suitable environment for T-cell differentiation (Denton et al., 2008), technical and ethical issues have hampered use of these mice. Human CD34⁺ HSC-transferred mice are the most commonly used humanized mice worldwide. HSCs can be obtained from umbilical cord blood (CB), bone marrow (BM), FL, and granulocyte colony-stimulating factor (G-CSF)-mobilized peripheral blood (MPB). Previously, we and another group compared the efficacy of the human cell engraftment using HSCs from CB, BM, FL, and MPB (Lepus et al., 2009; Matsumura et al., 2003). Transfer of CB or FL-derived HSCs resulted in significantly enhanced engraftment of CD45⁺ cells compared to BM or MPB-derived HSCs, so the authors concluded that CB or FL are appropriate sources of HSCs. With regard to the age of the recipient of humanized mice, newborn mice are superior to adult mice in terms of the engraftment and function of human immune cells. Traggiai et al. (2004) and Ishikawa et al. (2005) reported that humanized mice created using newborn recipients showed better reconstitution of human immune cells and produced an antigen-specific antibody. Our previous report also demonstrated that newborn recipients showed a slightly higher engraftment ratio of human CD45⁺ cells compared to adult recipients (Katano et al., 2014). Unlike in human, however, IgM, not IgG was the predominant type of antibody produced; therefore, the class-switch mechanism of human B cells may be non-functional in a murine environment (Watanabe et al., 2009).

1.2 Next-generation humanized mice

Novel, genetically modified immunodeficient mice, termed next-generation humanized mice, have been generated to improve human hematopoiesis. In conventional humanized mice transferred with HSCs, although human lymphoid-lineage cells (including human T and B cells) undergo differentiation, human myeloid-lineage cells do not. To overcome this, new humanized mouse platforms with well-developed human myeloid lineage cells have been created (Billerbeck et al., 2011; Ito et al., 2013; Rathinam et al., 2011; Rongvaux et al., 2014, 2011; Willinger et al., 2011). Functional human macrophages differentiate in these humanized mice, possibly due to the effect of human cytokines such as M-CSF, GM-CSF, and IL-6, together with TLR stimulation (Rongvaux et al., 2014). Because human mast cells and granulocytes differentiated in humanized mice expressing human IL-3 and GM-CSF, these models can be used to investigate human allergic diseases (Bryce et al., 2016; Ito et al., 2013). Development of functional human T cells to recapitulate acquired immunity is another focus of research. To improve adaptive immune responses, several strains of HLA-Tg mice have been established (Danner et al., 2011; Shultz et al., 2010; Suzuki et al., 2012). Because antigen-specific humoral and/or cytotoxic responses mediated by human T cells are enhanced in these models, they can be applied to vaccine development and models of immunological diseases.

1.3 Cancer

Animal models involving engrafted tumors are used in preclinical studies for predicting the efficacy and toxicity of anticancer drugs (Morton & Houghton, 2007). Nude mice have been used for >50 years and were the first animal model for engraftment of tumors (Issacson & Cattanach, 1962). Because NOG, NSG, and BRG mice are more accepting of human cell transplants (including of cancer cells) than nude or other immunodeficient mouse strains (Choi et al., 2016; Machida et al., 2009), they are used extensively in cancer research. To evaluate the safety of induced pluripotent stem (iPS) cells, in vivo screening using immunodeficient mice is necessary (Ben-David & Benvenisty, 2011; Brickman & Burdon, 2002; Menendez et al., 2012). The ability of iPS or embryonic stem (ES) cells to differentiate into multiple cell lineages has been evaluated by assessing teratoma formation in immunodeficient mice (Feraud et al., 2016; Thomson et al., 1998).

The patient-derived xenograft (PDX) model, which is generated by direct transfer of a patient-derived tumor into immunodeficient mice, is an efficient in vivo platform to investigate genomic profiling and drug efficacy studies in individual patients, and is likely to be applied to personalized therapy (Cho et al., 2016; Choi et al., 2016; Dobbin et al., 2014; Izumchenko et al., 2016). The US National Cancer Institute (NCI) has decided to preferentially use PDX samples rather than cancer cell lines for drug screening programs because the tumor microenvironment of PDX mimicks the human body better; the NCI will distribute them with data on each tumor's genome sequence and gene-expression pattern, and the treatment history of the patients (Ledford, 2016). Fujii et al. (2008) developed PDX mice using NOG mice transplanted with 326 fresh specimens of several types of tumor, 54 of which were successfully engrafted. Interestingly, the tumors retained a morphology, architecture, and molecular signature of the original cancers. Chijiwa et al. (2015) reported that the gene expression profiles and genetic alteration patterns of patient-derived tumors were preserved in mice. PDX models are also advantageous in preclinical studies of targeted agents because drug efficacy can be tested in advance of treatment. Park et al. (2015) identified drug targets by genomic alteration profiling of 103 patients with gastric cancer and validated drug efficacy using PDX models, which successfully recapitulated the tumor biology. Moreover, the combination of a standard chemotherapeutic agent and a BCL2L1 inhibitor was effective against gastric cancer with genomic mutations in PDX models. Thus, PDX models have various advantages over cell-line-derived xenograft and other models. Recapitulation of patient-derived tumors and the human immune system in humanized mice will enhance our understanding of the immune responses to tumors and enable development of novel immunotherapeutics.

Recently, we developed a new immune therapy model in which human immune-reconstituted mice were engrafted with human tumors to evaluate the antitumor effect of an immune checkpoint inhibitor, an anti-programmed death-1 (PD-1) antibody (Ashizawa et al., 2017). Treatment with the anti-PD-1 antibody suppressed growth of tumors engrafted in PBMC-transferred NOG MHC class I and II-deficient mice; the mice did not develop severe GVHD. We also confirmed a significant increase in the number of granzyme-producing activated CD8⁺ T cells infiltrating in the tumor in mice administered the PD-1 antibody. Ma et al. (2016) established an Epstein-Barr virus (EBV)-infected B cell-lymphoma model in which EBV-infected CB were transferred into NSG mice, followed by injection of anti-PD-1 and CTLA-4 antibodies. Administration of both anti-PD-1 and CTLA-4 antibodies resulted in a greater reduction in tumor size than either anti-PD-1 Ab or anti-CTLA-4 Ab alone. The resulting autologous reaction may result in activated human T cells attacking the B-cell lymphoma without allo-reactivity. Therefore, this model may recapitulate the physiological conditions in patients treated with immune-checkpoint inhibitors.

Another unique immunotherapy approach using humanized mice has recently been reported. Anti-CD19 chimeric antigen receptor (CAR) T cells with genomic disruption of PD-1 have been generated by CRISPR/Cas9-based gene editing (Rupp et al., 2017). Injection of PD-1 deficient CAR T cells resulted in significant rejection of a CD19⁺PD-L1⁺ K562 tumor in NSG mice, while injection of PD-1⁺ CAR T cells did not. These results suggest that treatment with CAR T cells and anti-PD-1 antibody enhance the ability of T cells to specifically target and kill tumors.

1.4 Graft versus host disease (GVHD)

GVHD is a severe adverse event with a high mortality rate in patients undergoing allogeneic BM transplantation for acute/chronic leukemia, aplastic anemia, or congenital immunodeficiency (Bleakley & Riddell, 2004; Goker, Haznedaroglu, & Chao, 2001; Shlomchik, 2007). Although rodent models have been used to investigate pathologic and physiologic aspects of GVHD, few in vivo studies have used human cells. Immunodeficient mice are amenable to engraftment of human PBMCs and enable analysis of the physiological mechanisms of GVHD (Ito et al., 2009; King et al., 2009; van Rijn et al., 2003). Xeno-GVHD models have been used to develop a biomaterial-based therapeutic strategy for preventing T-cell activation. Mesenchymal stem cells (MSCs) have been used to treat steroid-resistant GVHD (Baron et al., 2010; Kebriaei et al., 2009; Ringden et al., 2006). Moreover, MSC-infused therapeutic models for GVHD using human PBMC-transferred humanized mice have been developed (Gregoire-Gauthier et al., 2012; Maitra et al., 2004; Tisato, Naresh, Girdlestone, Navarrete, & Dazzi, 2007). Gregoire-Gauthier et al. (2012) investigated the ability of human MSCs to inhibit GVHD in human PBMC-transferred NSG mice. MSC injection ameliorated GVHD mortality, irradiation-induced weight loss, and T-cell proliferation, which is important for protecting against GVHD. Regulatory T (Treg) cell-dependent suppression is another strategy for prevention of GVHD (Beres & Drobyski, 2013; Komanduri & Champlin, 2011). Mutis et al. (2006) demonstrated that depletion of human Treg cells significantly exacerbated the symptoms and lethality of GVHD in PBMC-transferred BRG mice, and that administration of Treg cells reduced the mortality rate of GVHD. Although these approaches are clinically useful, the mechanisms are unclear because of our limited understanding of the pathogenesis of GVHD.

We developed a novel xeno-GVHD model in which isolated human CD4⁺ or CD8⁺ T cells were separately transferred into NOG mice (Ito et al., 2017). CD4⁺ T cell-transferred mice developed severe skin inflammation, which was mediated by production of IL-17 and IFNγ by pathogenic Th17 cells. Human IL-17 activates murine skin epithelial cells to produce a neutrophil chemoattractant, resulting in inflammation. The skin symptoms were significantly suppressed by an anti-IL-17 antibody (Secukinumab). In contrast, transfer of an identical number of CD8⁺ T cells did not induce GVHD. However, CD8⁺ T cells acquired effector functions and induced severe GVHD symptoms after transplantation into NOG human IL-2 transgenic mice. These two models will facilitate analysis of the cellular and molecular roles of human CD4⁺ and CD8⁺ T cells in the pathogenesis of GVHD.

Further improvement of the human PBMC-transferred model has been achieved by using HLA transgenic humanized mice that can mimic the HLA-mismatched allogenic or HLA-matched syngeneic conditions. Covassin et al. (2011) developed a human allo-GVHD model by transplanting human CD4⁺ T cells into NSG-HLA-DR4 transgenic (Tg) mice. A human allo-GVH reaction occurred in the mice, and transplanted human CD4⁺ T cells recognized allo-HLA class II molecules. In contrast, HLA-matched syngeneic models recapitulate the human acquired immune system without GVHD. Zeng et al. (2017) generated HLA-A2 and -DR1 transgenic Rag2^−/−/IL2rγ^−/−/Perforin^−/− mice that lack murine MHC class I and II. This strain manifested no GVHD symptoms and had a longer survival duration than previous other models despite expansion of human T cells following transfer of HLA class I and II-matched human PBMCs. Additionally, IgG against hepatitis B virus was detected in the serum of the mice. This mouse strain will facilitate investigation of human infectious and autoimmune diseases, and development of novel vaccines.

1.5 Allergy

Conventional humanized mice, in which human HSCs are transplanted into immunodeficient mice, do not mimic human allergic responses due to incomplete differentiation of myeloid lineage cells, including mast cells and granulocytes. These cells induce allergic symptoms by releasing intracellular granules following IgE crosslinking. Several strains of humanized mice with human mast cells and granulocytes have been established. NOG mice systemically expressing human IL-3 and GM-CSF (NOG IL-3/GM-CSF Tg) were the first humanized mouse model of an allergic disease (Ito et al., 2013). IL-3 and GM-CSF induce differentiation of human HSCs into myeloid-lineage cells; therefore, human monocytes/macrophages, dendritic cells, granulocytes, and mast cells were differentiated in the NOG IL-3/GM-CSF Tg mice. Compared to conventional NOG (non-Tg) mice, the human mast cells and basophils in IL-3/GM-CSF Tg mice strongly expressed Fc-epsilon receptor (FcϵR), which transduces the degranulation signal after binding of the IgE-allergen complex. In fact, co-administration of serum from patients with cedar pollinosis and cedar pollen as a specific antigen induced a human mast cell-mediated passive cutaneous anaphylaxis (PCA) reaction in Tg mice. Using this strain, type-2 cytokine-induced asthmatic airway inflammation can be recapitulated by intratracheal administration of human IL-33 (Ito et al.,submitted).

A large number of human mast cells develop in a triple transgenic NSG strain expressing human IL-3, GM-CSF, and SCF (NSG-SGM3) (Billerbeck et al., 2011) following transplantation of human BM, liver, and thymus (BLT), in addition to HSCs. These mast cells undergo degranulation in an IgE-dependent manner by intradermal priming with a chimeric IgE containing human Fc regions, resulting in the development of a robust passive cutaneous anaphylaxis response (Bryce et al., 2016). Moreover, they recapitulated an antigen-dependent passive systemic anaphylaxis response mediated by human mast cells.

NSG mice expressing the membrane-bound form of human stem-cell-factor (SCF) (NSG hSCF-Tg) also produced human mast cells after HSC transplantation (Takagi et al., 2012). Burton et al. (2017) demonstrated that oral feeding of peanut butter induces anaphylactic responses in humanized hSCF-Tg mice. Surprisingly, the anaphylaxis was mediated by antigen-specific human IgE, and inhibition of IgE by omalizumab suppressed the symptoms. Current understanding of the human immunology in humanized mouse models suggests that antigen-specific responses are not feasible because human T cells are educated by murine MHC in the thymus, and do not recognize HLA and antigen complexes in the periphery (Ito, Takahashi, Katano, & Ito, 2012; Theocharides, Rongvaux, Fritsch, Flavell, & Manz, 2016; Watanabe et al., 2009). Hence, new mouse models expressing HLA class I or II in the thymus have been developed to restore the antigen specificity of human T and B cells (Danner et al., 2011; Shultz et al., 2010; Suzuki et al., 2012). hSCF-Tg mice probably possess acquired immune responses, but it is necessary to unravel the mechanisms underlying HLA restriction in them.

Therefore, some allergic mechanisms can be recapitulated using NOG-IL-3/GM-CSF Tg, NSG-hSCF-Tg, and NSG-SGM3 mice, which will enable elucidation of the physiological mechanisms of human allergic diseases and development of novel therapeutics.

Regarding the limitations of HSC-transferred humanized mice as models of human allergic diseases, human type 2 innate lymphoid cells (ILC2), which contribute to the initiation and exacerbation of allergic inflammation, such as in asthmatic airway diseases, may not develop, despite the presence of ILC1 and ILC3 (Legrand et al., 2011; Zhang et al., 2015). ILC2 produce T helper type-2 (Th2) cytokines (such as IL-4, IL-5, and IL-13) when stimulated with the epithelial cell-derived Th2-promoting cytokines IL-33, TSLP, and IL-25 and following exposure to allergens (Kabata et al., 2013; Oboki, Nakae, Matsumoto, & Saito, 2011; Saluja, Khan, Church, & Maurer, 2015; Yao, Sun, Wang, & Sun, 2016). Rigas et al. (2017) reported NSG mice with adoptive transfer of cultured human ILC2 isolated from PBMCs. They intratracheally administered human IL-33 into ILC2-reconstituted mice, which resulted in several symptoms of asthmatic inflammation, such as airway hyper responsiveness and murine eosinophilic infiltration. Furthermore, Lim et al. (2017) discovered Lin⁻CD7⁺CD127⁺CD117⁺ cells which is the progenitor population of all human ILC subsets, including ILC2, in cord and adult blood, the fetal liver, and several adult tissues (Lim et al., 2017). They demonstrated that this progenitor could be engrafted in BRG Sirpa^NOD (BRGS) mice and differentiated into all the subsets of mature ILCs in humanized mice after transplantation. These studies will enable the development of models of human allergic diseases that feature complete reconstitution of human immune cells, including ILC2.