Use of Humanized Mice to Study the Pathogenesis of Autoimmune and Inflammatory Diseases

Abstract

Animal models of disease have been used extensively by the research community for the past several decades to better understand the pathogenesis of different diseases as well as assess the efficacy and toxicity of different therapeutic agents. Retrospective analyses of numerous preclinical intervention studies using mouse models of acute and chronic inflammatory diseases reveal a generalized failure to translate promising interventions or therapeutics into clinically-effective treatments in patients. Although several possible reasons have been suggested to account for this generalized failure to translate therapeutic efficacy from the laboratory bench to the patient’s bedside, it is becoming increasingly apparent that the mouse immune system may not adequately recapitulate the immuno-pathological mechanisms observed in human diseases. Indeed, it is well-known that >80 major differences exist between mouse and human immunology; all of which contribute to significant differences in immune system development, activation and responses to challenges in innate and adaptive immunity. This inconvenient reality has prompted investigators to attempt to humanize the mouse immune system in order to address important, human-specific questions that are impossible to study in patients. The successful long-term engraftment of human hemato-lymphoid cells in mice would provide investigators with a relatively inexpensive, small animal model to study clinically-relevant mechanisms as well as facilitate the evaluation of human-specific therapies in vivo. The discovery that targeted mutation of the IL-2 receptor common gamma chain in lymphopenic mice allows for the long-term engraftment of functional human immune cells has advanced greatly our ability to humanize the mouse immune system. The objective of this review is to present a brief overview of the recent advances that have been made in the development and use of humanized mice with special emphasis on autoimmune and chronic inflammatory diseases. In addition, we discuss current challenges and possible solutions for utilizing these unique mouse models to define the human-specific immuno-pathological mechanisms responsible for the induction and perpetuation of chronic gut inflammation.

INTRODUCTION

Animal models of disease have been used extensively by the research community over the past several decades as surrogates to better understand the pathogenetic mechanisms responsible for the induction of different diseases, to assess toxicity, determine pharmacokinetics of different compounds and to evaluate the efficacy of novel therapeutic agents. Surprisingly, retrospective meta-analyses of preclinical intervention studies using animal models of a variety of different acute and chronic inflammatory diseases such as stroke, sepsis, diabetes, multiple sclerosis, and rheumatoid arthritis report that only a small fraction of promising preclinical interventions and novel therapeutics translate to clinical efficacy (1–7). A cursory search of PubMed using the keywords mouse and colitis identifies >5,800 studies that have been published using mouse models of the inflammatory bowel diseases (IBD; Crohn’s disease, ulcerative colitis). Of these, hundreds of studies report significant anti-inflammatory effects of numerous small molecules, biologics, genetic alterations or immune manipulations in these models of IBD. Yet very few of the potential “targets” or therapeutic interventions identified in this voluminous literature have been taken to the next level and evaluated in clinical studies. In fact, of the more than 50 novel small molecules, biologics and cell-based therapies that have been reported to be effective in preclinical animal studies and have been or are currently being evaluated in several hundred phase I–III clinical studies, only monoclonal antibodies directed against TNF (i.e. inflixamab, adlimumab, certolizumab, golimumab) or α₄(β₇) integrins (i.e. natalizumab, vedolizumab) have been shown to be effective in clinical studies and approved for treatment of patients with IBD (Reviewed in (8); http://wwwclinicaltrialsgov). The reasons for the disconnect between preclinical studies and therapeutic efficacy have not been clearly delineated; however several possible factors are thought to be involved including: a) the use of animal models that do not adequately mimic the chronic immunopathology of human IBD, b) the use of inbred strains of mice as surrogates for heterogeneous human populations, c) differences in intestinal microbiota d) flawed experimental design and/or data analyses and e) publication bias (1–7;9). In addition to these shortcomings in the design and evaluation of preclinical studies, a particularly troubling situation has emerged over the past few years that has garnered a great deal of attention by funding agencies and the publishing community: the inability of academic and industry investigators to reproduce published studies demonstrating therapeutic efficacy of novel small molecules and biologics in animal models of disease (2;10–15).

One potential strategy for improving the bench-to-bedside transition for promising therapeutics is to identify and utilize the most immunologically relevant mouse models of IBD and pharmacologic strategies that most closely mimic the clinical situation (1). However, even with more rigorous standardization of preclinical studies, we are faced with the reality that mice are not humans and thus the immuno-pathogenetic mechanisms observed in mouse models of chronic inflammation may not necessarily recapitulate those for human disease. It is well-known that the structure and function of the mouse immune system is, in many instances, significantly different from humans (16;17). For example, mice possess relatively few lymph nodes that are grouped together in small numbers of “chains”. Humans on the other hand, possess larger numbers of lymph nodes that are organized into complex chains that drain relatively small areas of tissue when compared to mouse (16). In addition to peripheral lymphoid structures, the structure and function of mouse bronchus-, nasal- and gut-associated lymphoid tissues may differ dramatically from those of the human depending upon its anatomic location (16). Given the structural differences in the immune systems between species, it is not surprising to learn that certain functions of the immune system are also quite different in mice when compared to humans. It is well-appreciated that >80 major differences exist between mouse and human immunology that are involved in immune system development, activation and responses to immune challenges (17). Major differences between the two species include hematopoietic activity of the spleen, circulating levels of lymphoid and myeloid leukocytes, innate immune mechanisms [e.g. Toll Like Receptors (TLRs), NOD Like Receptors (NLRs), defensins], T cell signaling pathways, B cell function, IgA production and isotype, intestinal intraepithelial and lamina propria cell composition, gut-associated dendritic cell and Natural Killer T (NKT) cell subsets and inducible nitric oxide synthase (9;17–21). Indeed, these differences may account, in part, for the lack of consensus as to whether mouse models of inflammation can be used as surrogates for human inflammatory diseases. Although one recent study suggests that genomic responses in mouse models of inflammation do not recapitulate those observed in human diseases (22), another equally compelling study concluded that gene expression patterns in mouse models closely mirror human inflammatory responses (23).

This inconvenient reality, coupled to the fact that animal models have been underwhelming in their ability to predict clinical trial outcomes in a variety of human diseases, reinforces the need to develop small animal models that more closely recapitulate human immunity. The development of mice with a fully functional human immune system via the long-term engraftment of the human hemato-lymphoid cells (e.g. hematopoietic stem cells, lymphocytes) would provide investigators a preclinical model to more accurately define disease pathogenesis as well as facilitate the evaluation of human-specific therapies in vivo. This approach has been successfully utilized for mechanistic and intervention studies in models of human-specific infectious diseases (e.g. HIV)(9;24–26). In addition, great strides have been made in using humanized mice to study the pathogenesis and human-specific therapies in different diseases such as diabetes, arthritis, graft vs. host disease, transplant rejection and cancer (25–31). A major breakthrough in the evolution of the humanization of the mouse immune system came with the discovery that targeted mutation of the IL-2 receptor common gamma chain (IL-2γ) in NOD/scid mice (termed NOD/scid-IL2rγ^−/− or NSG mice) greatly enhanced the long-term engraftment of human hemato-lymphoid cells in these severely immuno-deficient mice (32–35). The objective of this review is to present a brief overview of the recent advances that have been made in the development and use of humanized mice with special emphasis on their use in autoimmune and inflammatory diseases. In addition, we discuss the challenges and possible solutions for utilizing these unique mouse models to define the human-specific immuno-pathological mechanisms responsible for the induction and perpetuation of chronic gut inflammation.

The Use of Immuno-Deficient Mice as Recipients for Human Hemato-lymphoid Cells: A Look Back

Adoptive transfer of xenogeneic (human) hemato-lymphoid cells (e.g. hematopoietic stem cells or peripheral blood mononuclear cells) into healthy wild type mice results in rapid rejection of these cells and, depending upon the type of cells injected, may result in lethal xenogeneic graft vs. host disease. In order to limit these damaging immune responses, immunological tolerance of recipient mice for donor human cells as well as tolerance of human cells for mouse tissue antigens is required for successful engraftment and survival of human progenitor and/or immune cells (9). Thus, the ideal hosts for creating this “bidirectional tolerance” are mice devoid of their adaptive and innate immune systems and major histocompatibility (MHC) complexes I and II. Although these mice have not yet been generated, substantial progress has been made in creating severely immuno-deficient mice that support the long term engraftment of human progenitor and immune cells (9;25;26;30). These new generation immuno-deficient mice are being used as platforms to study the immunological mechanisms associated with hematopoiesis, infectious diseases, cancer, autoimmune diseases, gene therapy and regenerative medicine (9;25;26;36;37). In addition, these humanized mice are providing preclinical models to evaluate the therapeutic efficacy of new drug and vaccine therapies (9;25;26). Below, we provide a brief history of development of immuno-deficient mice and highlight those discoveries that proved to be instrumental for enhancing the long-term engraftment of human hemato-lymphoid cells.

Nude Mice

The engraftment of human cells or tissue into immuno-deficient mice began more than 50 years ago following the discovery that T cell-deficient mice failed to reject normal and malignant human tissue (38). These hairless “nude” (nu/nu) mice were originally observed in a colony of albino mice operated by Dr. N.R. Grist in 1962 at the Virus Laboratory in Ruchill Hospital in Glasgow, Scotland. Dr. Grist subsequently sent the mice to the Institute of Animal Genetics in Edinburgh where Dr. S.P. Flanagan provided a detailed description of the hairless phenotype as well as described several abnormalities associated with these mice including problems with fertility and neonatal viability, liver disease and susceptibility to systemic toxoplasmosis (38). In 1968, Pantelouris reported that nu/nu mice lacked a thymus and thus were devoid of functional T cells (39) (Table 1). A subsequent series of investigations revealed that the nu/nu phenotype arose from a mutation of the Foxn1 gene previously known as the whn gene (winged-helix nude or Hfh11^nu)(40;41). This pleotropic transcription factor is critical for both thymus development and the expression of different hair keratin genes. These T cell-deficient mice have proven to be one of the most popular mouse models used for engrafting normal or malignant human tissue. Unfortunately, nu/nu mice are of limited use for humanization of the mouse immune system as they do not support engraftment of human hemato-lymphoid cells (42). As correctly surmised by Ganick et. al., the failure of human immune cells to survive and proliferate in nu/nu mice may be due to the presence of functional murine NK cells (42)(see below).

Table 1.

Immuno-deficient Mice Used for Human Cell and Tissue Engraftment

Strain	Phenotype	Limitations
Nude (nu/nu)	Lack T cells	Intact innate system Fully functional myeloid and NK cells No engraftment of hemato-lymphoid cells
SCID (Severe Combined Immuno-deficient)	Lack T & B cells	Intact innate immune system Fully functional myeloid and NK cells Functional C5 complement component “Leaky”; T and B cells develop with age Develop thymic lymphoma Very low engraftment of human cells
RAG-1−/− and RAG-2−/− (Recombination Activating Gene-1 or -2 Deficient)	Lack T & B cells	Intact innate immune system Fully functional myeloid and NK cells Very low engraftment of human cells
NOD (Nonobese Diabetic)	Defective innate immune system Abnormal myeloid and NK cell development and function. Lack C5 Hemolytic Complement	Develop insulin-dependent diabetes mellitus Human cell engraftment has not been studied in the NOD mouse
NOD/SCID	Lack T & B cells Defective innate immune system Abnormal myeloid and NK cell development and function. Lack C5 Hemolytic Complement Enhanced human cell engraftment	Residual myeloid and NK cell activity innate remains Decreased lifespan; thymic lymphoma

Severe Combined Immuno-deficient and Recombination-Activating Gene-1 or -2-Deficient Mice

The severe combined immunodeficiency (scid) mutation was originally described by Bosma et. al. in C.B.-17 mice (43). It was ultimately revealed that the scid gene encodes for protein kinase, DNA activated catalytic polypeptide (Prkdc)(44;45). Mutation of the Prkdc gene prevents the expression of rearranged antigen receptors thereby preventing the development of mature T and B cells (46;47;47–51) (Table 1). Because scid mice lacked both cell-mediated and humoral immunity, they became the strain-of-choice for engrafting different populations of human hemato-lymphoid cells to study hematopoiesis, infectious disease, gene therapy protocols and tumor cell trafficking (reviewed in (52)). Several variations of the human-scid (hu-scid) chimeric mouse model were developed that used different routes of administration of the human cells (or tissue) into irradiated or un-irradiated scid mice (52). Although exciting new data were generated using the different hu-scid models, the numbers of human cells that engrafted and proliferated in these mice were very low (0.5–5% of total scid bone marrow cells)(52). Although young scid mice were devoid of both T and B cells, investigators soon found that T and B cell receptors undergo age-dependent rearrangements resulting the in formation of functional T and B cells, a phenomenon called “leakiness”(53–55). In addition, these immuno-deficient mice developed life-shortening thymic lymphomas (47;53;56). Furthermore, Bosma et. al. observed an increase in the development of thymic tumors in mice that became leaky as measured by the appearance of serum immunoglobulin (Ig⁺)(53). They observed that the incidence of thymic lymphomas increased from 8% in 3–5 month old (Ig⁻)scid mice to approximately 32% in age- matched (Ig⁺)scid mice (53). Furthermore, they found that almost 60% of the scid(Ig⁺) mice developed these malignant lymphomas at 5–9 months of age (53).

The inability of scid mice to support robust engraftment of xenogeneic immune cells was not specific for the scid strain or to their age-dependent propensity to generate functional T and B cells as human cell engraftment in recombination-activating gene-2-deficient (RAG-2^−/−) mice was also very low or absent (52;57). Both the RAG-1 and RAG-2 genes are responsible for the activation of the V(D)J recombination reaction such that deletion of either gene produces mice devoid of T and B lymphocytes (58;59) (Table 1). Unlike scid mice, RAG-1^−/− and RAG-2^−/− mice (collectively referred to as RAG^−/− mice) are not “leaky”, do not develop thymic lymphomas and are less sensitive to radiation (58;59). The reasons for the disappointingly low engraftment of human cells in scid or RAG^−/− mice were found to be due to: a) lack of cross-reactivity between human progenitor cells and essential murine growth factors and b) the presence of robust innate immune systems in the lymphopenic recipients. It was clearly demonstrated that human hematopoietic stem cells were incapable of interacting with critical mouse cytokines and growth factors that are required for progenitor cell survival and proliferation (52). Surprisingly, neither exogenous administration nor transgenic expression of human hematopoietic growth factors (e.g. IL-3, GM-CSF, SCF) increased substantially the engraftment of human cells in scid recipients (52;60). Another confounding limitation that became apparent during this time was that both scid and RAG^−/− mice maintained normal (or increased) numbers of myeloid cells (neutrophils, monocytes, macrophages) and NK cells as well as normal or increased hemolytic complement; any one of which were capable of destroying xenogeneic progenitor cells (52;55;60).

Nonobese Diabetic/scid Mice

Recognition of the critical role that innate immune cells (particularly NK cells) played in restricting the engraftment of human immune cells in scid and RAG^−/− mice prompted investigators to develop new stocks of immuno-deficient mice that would allow for greater repopulation xenogeneic immune cells. A major turning point in these efforts came in 1994 when Shultz et. al. reported that back crossing nonobese diabetic (NOD) mice onto the scid background produced offspring (NOD/scid mice) that were devoid of both T and B cells, produced little or no serum immunoglobulin, had low numbers of functional innate immune cells (e.g. NK and myeloid cells) and lacked the hemolytic complement component C5(55) (Table 1). Unlike their scid relatives, <10% of the NOD/scid mice developed functional T and B cells at 200 days of age (55). The rationale for generating NOD/scid mice was based upon previous studies that were investigating the immuno-pathological mechanisms responsible for the spontaneous development of insulin-dependent diabetes mellitus (IDDM) in NOD mice (61). These earlier studies demonstrated that while NOD mice possessed autoreactive CD4⁺ and CD8⁺ T cells, these animals displayed defective innate immune function including reduced numbers and function of NK and myeloid cells, defective dendritic cell and macrophage function and absence of the hemolytic complement C5(52;62–64) (Table 1). In addition, it was found that NOD mouse-derived macrophages expressed the sigma regulatory protein alpha (SIRPα) receptor that was able to bind to CD47 expressed on human immune cells (65;66). This receptor-ligand interaction induces an inhibitory or “do not eat me” signal within murine macrophages thereby preventing these phagocytes from engulfing and destroying human progenitor or immune cells (9;65;66). Thus, the progeny of the NOD-scid cross were not only lymphopenic but they also possessed phagocytic tolerance as well as a defective innate immune system making them more likely to support long term engraftment of xenogeneic immune cells (55;67) (Table 1).

The generation of the NOD/scid stock initiated one of the most active periods of investigations in the humanization of the mouse immune system. Indeed, for more than 10 years, the NOD/scid mouse was considered the “gold standard” for long-term repopulation of human hemato-lymphoid cells (25;26;55;68). Investigators found that the engraftment of human peripheral blood mononuclear cells (PBMCs) or hematopoietic stem cells (HSCs) was 5–10-fold greater in NOD/scid mice when compared to scid mice (25;52;55;69). During the ensuing decade of investigations using the NOD/scid stock, it soon became clear that despite its numerous advantages over scid or RAG^−/− recipients, the NOD/scid platform possessed its own set of shortcomings that limited its overall utility for humanization studies. For example, it was found that NOD/scid mice had a relatively short life span (8 months) due to the development of thymic lymphomas in approximately 70% of these mice by 40 weeks of age (52;55;70). In addition, while significantly reduced, residual NK and myeloid cell activities limited robust engraftment of human immune cells in these mice (55). These shortcomings made it clear that additional molecular and/or genetic alterations of this promising mouse strain would have to be made in order to enhance further the engraftment of human hemato-lymphoid cells.

Immuno-deficient Mice Devoid of Functional IL-2 Receptor Common γ Chain

A fundamental breakthrough that greatly enhanced the long-term engraftment of even greater numbers of human hemato-lymphoid cells came in 2002 with the development of NOD/scid mice that lacked a functional IL-2 receptor common gamma chain (IL-2rγ^−/−)(33). As early as 1995, investigators knew that IL-2rγ was a critical component of the high-affinity receptors for IL-2, IL-4, IL-7, IL-9, IL-15 and IL-21(71;72) (Figure 1). Because these cytokines are required for T, B and NK cell development, targeted mutation of IL-2rγ resulted in the loss of receptor-ligand signaling thereby creating mice devoid of T, B and NK cells (73–75). Over the next few years, several laboratories developed immunodeficient mice devoid of IL-2rγ that differed with respect to strain background and the type of IL-2rγ mutation (26;28;68) (Table 2). The nomenclature for describing and categorizing the different IL-2rγ−/− strains may appear a bit confusing for those readers who are not mouse geneticists. Therefore, we have found it helpful to group the most popular IL-2rγ^−/− strains into three major categories depending upon whether the IL-2rγ mutation has been bred onto the NOD/scid, NOD/RAG^−/− backgrounds (Table 2). For example, the major NOD/scid-IL-2rγ^−/− strains include the NOD.Cg-Prkdc^scidIl2rg^tm1Wjl (NSG) mouse and the NOD.cg-Prkdc^scidIl2rg^tm1Sug (NOG) mouse. The IL-2rγ mutation in NSG mice is a complete null mutation and thus is not expressed whereas NOG mice express a truncated/inactive form of the γ chain that can bind cytokines but does not promote cell signaling. Similarly, the NOD/RAG1^−/−IL-2rγ^−/− mouse (NOD.Cg-Rag1^tm1MomIl2rg^tm1Wjl; or NRG) contains a complete null mutation of IL-2rγ. The third major group of IL-2rγ^−/− mice currently being used for humanization studies consists of BALB/c/Rag1^−/− IL-2rγ^−/− and BALB/c/Rag2^−/− IL-2rγ^−/− mice (both are called BRG). These mice express a truncated/inactive form of IL-2rγ similar to that expressed in NOG mice. We refer readers to several recent reviews that describe the generation and unique characteristics of these different (26;28;68;69).

Figure 1. The common ƒÁ chain plays a critical role in interleukin signaling and the development of immune cells.

The common γ chain is associated with receptors for IL-2, IL-4, IL-7, IL-9, IL-15 and IL-21. These cytokines are produced by different immune, stromal cells and epithelial cells. Receptor/ligand signaling is critically important for the generation of different populations of immune cells (Modified from reference 75 with permission).

Table 2.

Immuno-Deficient Mice Lacking Functional IL-2rγ

Group	Strain	Name	Characteristics
NOD/scid-IL-2rγ−/−	NOD.Cg-PrkdcscidIl2rgtm1Wjl	NSG	The Il2rγ mutation is a complete null such that IL-2rγ is not expressed.
	NOD.cg-PrkdcscidIl2rgtm1Sug	NOG	The Il2rγ mutation encodes a truncated form of the γ chain such that it lacks the intracytoplasmic domain; it can bind cytokines but will not promote cell signaling
NOD/RAG-1−/−IL-2rγ−/−	NOD.Cg-Rag1tm1MomIl2rgtm1Wjl	NRG	The Il2rγ mutation is a complete null such that IL-2rγ is not expressed.
BALB/c/RAG−/−IL-2rγ−/−	C.Cg-Rag1tm1MomIl2rgtm1Wjl	BRG	This strain is generated on a BALB/c background that has the RAG-1−/− mutation. The IL-2rγ chain lacks the intracytoplasmic domain thereby preventing intracellular signaling.
	C.Cg-Rag2tm1FwaIl2rgtm1Sug	BRG	This strain is generated on a BALB/c background that has the RAG-2−/− mutation. The IL-2rγ chain lacks the intracytoplasmic domain thereby preventing intracellular signaling.

The development of these immuno-deficient IL-2rγ^−/− strains was a defining moment in the quest to humanize the mouse immune system as each of these stocks were shown to engraft human hematopoietic stem cells (HSCs) or peripheral blood mononuclear cells (PBMCs) to a much greater extent than had been observed for any of the previous immuno-deficient strains (26;28;68;69). In addition, these new-generation IL-2rγ^−/− mice do not develop thymic lymphomas and live a normal lifespan of 2 years. These novel platform strains have been particularly useful for studies involved in human immunity, hematopoiesis, infectious disease, autoimmune disease, regenerative medicine and cancer (9;26;69).

Humanized Mouse Models

There are currently three major humanized mouse models that utilize the immuno-deficient IL-2rγ^−/− strains described above (Table 2). As with any animal model, each of these models has advantages and limitations. These humanized mouse models have been shown to possess different capabilities of engrafting human hemato-lymphoid cells. For example, investigators have determined that NSG and NRG mice engraft human HSCs and promote T cell development to a greater extent than do BRG mice (76) whereas NSG mice support engraftment of larger numbers of human HSCs than do NOG mice (77). Therefore, the choice of a specific humanized model will depend upon the specific questions being posed by the investigator. Theoretically, it should be possible to model most if not all of the human autoimmune and chronic inflammatory diseases using one or more of the immuno-deficient IL-2rγ^−/− strains. This section describes the generation, uses and limitations of the 3 major humanized mouse models

Engraftment of Human Peripheral Blood Mononuclear Cells into Lymphopenic IL-2rγ^−/− Mice

In 1988, Mosier et. al. demonstrated that intra-peritoneal (i.p.) injection of human peripheral blood leukocytes (PBLs) into scid mice (termed Hu-PBL-scid mice) resulted in the stable reconstitution of a functional human immune system (78). Although all populations of circulating leukocytes were injected in the Hu-PBL-scid model, the large majority of stably engrafted cells were activated T cells with only very small numbers of human B cells, myeloid cells or other immune cells present in the scid recipients. As discussed in the preceding section, the total numbers of human T cells that engraft and proliferate in scid mice are actually quite low due to the age-related development of mouse T and B cells and the presence of an active innate immune system that together destroy xenogeneic cells (52;55;60). In addition, scid mice developed life-shortening thymic lymphomas thereby limiting the usefulness of this model (43;53;56). We would like to caution the reader that the use of the term Hu-PBL-scid has been used and continues to be used to describe any humanized mouse model where human PBMCs are injected into lymphopenic recipients such as a scid, NOD/scid or the newer immuno-deficient IL-2rγ^−/− strains. In order to avoid any confusion, we have chosen to focus primarily on the use of lymphopenic IL-2rγ^−/− strains as recipients for human PBMCs in this and the following sections (Table 3). Indeed, the generation of these immuno-deficient IL-2rγ^−/− strains greatly advanced the field as investigators observed much greater levels of engraftment of T cells than had been documented in other lymphopenic recipients. In addition, lymphopenic IL-2rγ^−/− strains possessed a much greater lifespan due to the lack of thymic lymphoma development (26;68;69). Intravenous injection of human PBMCs into irradiated (conditioned) or naive IL-2rγ^−/− strains produces the most robust engraftment of human T cells than in any previous immuno-deficient recipient (79). The Hu-PBMC-IL-2rγ^−/− mouse model is currently used to study human immunity, autoimmunity, infectious diseases, regenerative medicine, allograft rejection and cancer (9;26;28;68;80) (Table 3). In addition, this model has been used to assess novel cell-based therapies (36). Although this model is well-suited for investigating T cell-mediated immune responses of patients with immunologic disorders as well as autoimmune and chronic inflammatory diseases, the “window of opportunity” for performing these types of studies is limited by the development of lethal xenogeneic graft-versus-host disease (GVHD) within 3–4 weeks following T cell transfer (81). King and coworkers have demonstrated that xenogeneic GVHD results from the ability of human T cells to recognize and mount aggressive immune responses towards murine MHC class I and II (81). These investigators have reported a significant delay in xenogenic GVHD when human PBMCs are engrafted into NSG mice devoid of MHC class I and II (81). Despite this limitation, the Hu-PBL-IL-2rγ^−/− model has been used successfully to explore immuno-pathological mechanisms involved in different autoimmune and infectious diseases.

Table 3.

Models of Human Hemato-Lymphoid Cell Engraftment into Lymphopenic IL-2rγ^−/− Mice

Model Name	Description	Advantages	Applications	Limitations
Hu-PBMC-IL-2rγ−/− (Human Peripheral Blood Mononuclear Cells engrafted into lymphopenic IL-2rγ−/− mice)	Lymphopenic IL-2rγ−/− mice are injected with human peripheral mononuclear cells	A technically simple model to establish Good engraftment of effector and memory T cells.	Induced models of inflammation (e.g. sepsis, arthritis, colitis). Model of xenogeneic GVHD. Assess effector functions of T cells obtained from patients with RA, lupus, MS, diabetes or IBD Human-specific infectious diseases (HIV) Allograft tissue rejection	Only activated/memory T cells are present Lack of mature human B cells, myeloid cells, DCs, platelets and erythrocytes. Xenogeneic GVHD develops after 4–5 weeks due to human T cell reactivity against mouse MHC molecules. Limited primary immune responses.
Hu-HSC-IL-2rγ−/− (Human Hematopoietic Stem Cells engrafted into lymphopenic IL-2rγ−/− mice)	Lymphopenic IL-2rγ−/− mice are injected with CD34+ HSCs derived from fetal liver, cord blood, bone marrow, or from peripheral blood following G-CSF-mediated mobilization	Generates a naïve human immune system Development of T and B cells, APCs, myeloid cells and NK cells	Human hematopoiesis Induced models of inflammation (e.g. sepsis, arthritis, colitis) Engraft human HSCs from patients with autoimmune or chronic inflammatory diseases Human-specific infectious diseases Transplantation biology	Low rate of T cell engraftment Impaired T and B functions No mucosal immune system Lack of expression of human HLA within the thymus prevents education and development of HLA-restricted CD4+ and CD8+ T cells (DTH response is suboptimal) Only small numbers of PMNs, RBCs and megakaryocytes are present in the blood
BLT-IL-2rγ−/− (Bone marrow, Liver, Thymus engrafted into lymphopenic IL-2rγ−/− mice)	Lymphopenic IL-2rγ−/− mice are implanted with small pieces of human fetal liver and autologous thymus under the renal capsule; the mice are then injected with human CD34+ HSCs purified from the same fetal liver sample.	Human immune system engraftment is much more robust than in the Hu-SRC-NSG model. Sustained high level of T cell development; T cells are educated by the human thymus and are HLA restricted Produces human mucosal immune system	Human hematopoiesis Engraft human HSCs from patients with autoimmune or chronic inflammatory diseases Human-specific infectious diseases Transplantation biology	Surgical expertise and fetal tissue are required Responses to vaccination protocols are limited to IgM antibody production. A delayed xenogeneic GVHD (>4 months) occurs in these mice that result from the lack of negative selection against murine antigens in human thymus and/or to lack of peripheral regulation

Engraftment of Human Hematopoietic Stem Cells into Lymphopenic IL-2rγ^−/− Mice

Following the early successes of engrafting PBMCs into scid recipients, investigators initiated studies to ascertain the ability of scid mice to support the engraftment and differentiation of human CD34⁺ hematopoietic stem cells (HSCs) derived from human fetal tissue or bone marrow (82;83). Lapidot and coworkers reported that adoptive transfer of human CD34⁺ HSCs into scid recipients resulted in the generation of small numbers of immature lymphocytes and myeloid cells (83)(Table 3). This model, called the human scid repopulating cell scid or Hu-SRC-scid model, has been used and continues to be used for exploring different immunological aspects of human hematopoiesis, infectious diseases, and adaptive immunity. However, the utility of this model was limited by disappointingly low levels of HSC engraftment and lack of formation of functional T and B lymphocytes, myeloid cells, and dendritic cells. The development of immuno-deficient IL-2rγ^−/− mice provided a much needed catalyst for developing a more robust humanized immune system in mice (32–35). A series of studies demonstrated that HSCs derived from fetal liver, umbilical cord blood (UC), bone marrow, and G-CSF mobilized peripheral blood could be used to engraft different lymphopenic IL-2rγ^−/− mice resulting in dramatic increases in engraftment and differentiation of HSCs into different immune cell subsets (76;84;85). Brehm and coworkers recently compared the engraftment efficiency of CD34⁺ HSCs obtained from cord blood in three different immuno-deficient IL-2rγ^−/− mouse strains at different ages (76). They found that engraftment was more robust in irradiated newborns compared to irradiated adults and that T and B cell numbers were significantly greater in NSG and NRG mice compared to BRG recipients. Other investigators have found that intrahepatic injection of HSCs also provides an effective mode of delivery of human HSCs for humanizing the immune system (86;87). Indeed, Misharin et al reported that NSG mice engrafted via intrahepatic injection of human CD34⁺ HSCs spontaneously produced human immunoglobulins as well as generated specific IgG following immunization (87). In addition to T and B cells, adoptive transfer of human HSCs into IL-2rγ^−/− recipients promotes the formation of small but significant numbers of myeloid cells and DCs (32–34;88–90).

A major advantage of Hu-HSC-IL-2rγ^−/− models is the absence of xenogeneic GVHD (83). Because human T cells undergo positive and negative selection within the mouse thymus, these lymphocytes are restricted to mouse MHC and thus, do not react to highly antigenic murine proteins. However, this advantage is also its primary drawback i.e. human T cells are not human leukocyte antigen (HLA) restricted. This means that T cells will not recognize nor react to antigen presented by human antigen presenting cells that develop within tissue (91). In addition, HSC engraftment in immuno-deficient IL-2rγ^−/− mice have poor T cell development most likely due to inefficient human thymopoeisis in the mouse thymus (89). Furthermore, very low numbers of circulating platelets, erythrocytes and polymorphonuclear neutrophils (PMNs) are produced in following transfer of human HSCs (26;80;91;91). Despite these shortcomings, the Hu-HSC-IL-2rγ^−/− models have provided invaluable information on many aspects of human immuno-biology.

Engraftment of Human Hematopoietic Stem Cells into Lymphopenic IL-2rγ^−/− Mice Implanted with Human Fetal Thymic and Liver Tissues

To create a more robust humanized environment that would allow for more complete differentiation of engrafted human HSCs, the Bone marrow-Liver-Thymus (BLT) humanized mouse model was developed (Table 3). In this model, animals are first conditioned using sublethal irradiation and then are implanted with small fragments (~1 mm³) of autologous human fetal liver and thymus under the kidney capsules (9;26;92;93). Recipients of the human tissues are then injected (i.v.) with 1–5 ×10⁵ autologous human HSCs obtained from the fetal human liver implanted into the recipients (94). The BLT model, originally developed in NOD/scid mice, convincingly showed long-term engraftment of multi-lineage human cells including T and B cells, DCs, monocytes, macrophages, erythrocytes and platelets (9;69;92;93;95). In addition, these studies demonstrated that human T cells were HLA restricted and were capable, along with human B cells, of mounting adaptive immune responses in vivo (9;69;92;93;95). Furthermore, BLT mice developed secondary lymphoid tissue (e.g. mesenteric lymph nodes) and human mucosal immune systems thereby making this the model of choice for HIV investigators (9;69;92;93;95). Following the development of immuno-deficient IL-2rγ^−/− mice, investigators compared the relative engraftment efficiencies of human HSCs in the NOD/scid-BLT vs. NSG-BLT model with varying results. Although some studies reported similar levels of reconstitution of human HSCs in blood, spleen, liver and lung in the two BLT models (77;96;97), other studies reported superior human immune cell engraftment in NSG-BLT vs. NOD/scid-BLT mice (32;33;98;99). One important difference between the two BLT models reported by Denton and coworkers was the relative lack of intraepithelial and lamina propria T cells within the intestines of NSG-BLT mice when compared to NOD/scid BLT mice (97). In a more recent study, Nochi et. al. demonstrated that cryptopatches and gut associated lymphoid tissue (GALT) failed to develop in NSG-BLT mice whereas intestinal T cells and cryptopatches were readily observed in NOD/scid-BLT mice (100). Furthermore, these investigators demonstrated that cryptopach development was required for the generation of a functional GALT. Taken together, these two studies highlight the importance of the IL-2rγ chain in IL-7/IL-7 receptor signaling for normal GALT development (100).

Despite model variability, the NOD/scid-BLT and NSG-BLT models are considered two of the best humanized mouse models for studies involved in human hematopoiesis, infectious diseases, acute and chronic diseases and therapeutic/vaccine development. Despite the fact that BLT mice have a “blended” immune system composed of murine myeloid cells co-existing with a human adaptive immune system, these humanized animals are capable of mounting both direct and indirect delayed type hypersensitivity (DTH) responses to two different antigenic challenges (101). Because DTH responses are considered the most stringent test for a fully functional and well-integrated immune system, this model is suitable for studies evaluating T cell- dependent immune responses. As with the other humanized mouse models, the BLT model has a few significant drawbacks including the requirement for small animal surgical skills, the use of human fetal tissue, generation of relatively low levels of mature/functional human granulocytes, myeloid cells, B cells and a delayed xenogeneic GVHD (94;102–105). It is not clear why GVHD develops by 4–5 months post-reconstitution however, a loss of tolerance of human immune cells to murine MHC antigens is considered a distinct possibility. A recent study demonstrates that deletion of either MHC class-I or -II fails to delay the development of GVHD in the NSG-BLT model (94). Interestingly, Lavender and coworkers have demonstrated that deletion of CD47 in C57Bl/6-RAG-2^−/−IL-2rγ^−/− mice renders these triple knock out (TKO; CD47^−/−RAG-2^−/−IL-2rγ^−/−) animals resistant to GVHD in the BLT model (106;107). Based upon previous work by Wang et. al.(108), the authors hypothesize that during development, phagocytes (i.e. macrophages, DCs) within TKO mice become tolerized to xenogeneic HSCs (106). Because the expression of SIRPα on macrophages derived from CD47^−/− mice is very similar, if not identical to that observed on WT macrophages, tolerance observed in TKO-BLT mice is unlikely to be caused by SIRPα-CD47 signaling (106;108). The major advantages of using this C57Bl/6-based, TKO-BLT model vs. the NOD/scid-BLT or NSG-BLT model are: a) C57Bl/6-TKO mice maintain an intact complement system, b) C57Bl/6-TKO mice are more radiation resistant allowing for more efficient conditioning prior to human cell engraftment and c) genetic manipulations (e.g. gene inactivation; introduction of transgenes) are more easily accomplished due to the ready availability of a variety of different mutant mice on the C57Bl/6 background (106).

Current Strategies for Improving Engraftment of Human Hemato-lymphoid Cells in Immuno-deficient Mice

The advances that have been made in humanizing the immune system of immunodeficient mice over the past 20 years has been truly remarkable. Nevertheless, a number of limitations remain that prevent the development of a fully functional human immune system. For example, virtually all humanized mouse models possess only small numbers of human granulocytes (e.g. PMNs, eosinophils), myeloid cells (e.g. monocytes, macrophages), erythrocytes and platelets as well as reduced numbers of fully functional T and B cells (9;28;80). Many of these deficiencies are most likely due to the lack of cross reactivity of certain murine cytokines and/or growth factors with human progenitor cell receptors that are known to be important for hematopoiesis and immune system development (9). However, poor development of peripheral lymph nodes, GALT and mucosal immune systems undoubtedly contribute to an underdeveloped immune system (28;80;97;100). The following section discusses some of the new approaches that are currently being used to improve engraftment and differentiation of human HSCs in immuno-deficient mice.

Strategies for enhancing the human innate immune system

The development, numbers and function of human PMNs, monocytes, macrophages and NK cells are generally quite low in the currently used models of humanized mice. For example, human myeloid cells have been found to be less than 5–10% of what is found in humans[reviewed in (9)]. Although some studies report that these myeloid cells are reasonably functional (90;109), other studies have described these myeloid cells as phenotypically immature and functionally defective (110). Furthermore, human NK cell maturation, function and homeostasis are also defective in humanized mice owing in part, to NK cell susceptibility to phagocytosis by murine macrophages (98) (discussed below) and lack of trans-presentation hIL15-hIL15Rα(111). Many of the cytokines and growth factors required for innate immune system development and function are produced by non-hematopoietic cells and thus are of murine origin. A recent review by Rongvaux et. al. presents an extensive list of these cytokines and factors and highlights the significant differences in sequence homologies between the mouse and human proteins (9). The lack of effective receptor-ligand interactions between human progenitor cells and mouse-derived cytokines and growth factors would be expected to greatly diminish hemato-lymphoid cell development in immuno-deficient mice (9).

In order to improve innate immune cell development and function, several genetic and technological approaches are currently being employed to promote and/or enhance expression of crucial human cytokines and growth factors that are known to be important in innate immune system development and function including: a) use of cDNA constructs expressing human genes that can be controlled by ubiquitous or tissue-specific promoters; b) use of bacteria artificial chromosomes (BACs) that have regulatory elements for gene expression; and c) use of knock-in technology to replace mouse genes with specific human genes. The cDNA construct approach can lead to transgene expression at non-physiological levels or without temporal and tissue specific regulation. BACs often include regulatory elements for control of gene expression facilitating gene expression at physiologic levels and at the appropriate time and location. It should be noted that both the cDNA and BAC approaches result in expression of both the mouse and human gene(s) of interest because the homologous mouse gene is left intact. However, knock-in technology replaces the mouse gene with the equivalent human gene resulting in human gene expression without expression of the mouse protein. Indeed, knock-in technology has been used to increase the numbers of myeloid cells in humanized mice expressing human interleukin 3 (IL3) and granulocyte-macrophage colony-stimulating factor (GM-CSF)(112), thrombopoietin (113) or macrophage colony-stimulating factor (M-CSF)(114). In a more recent study, Rongvaux et. al.(115) report the generation of mice expressing human genes for M-CSF, human IL-3, human GM-CSF and human thrombopoietin that were knocked-in to their corresponding mouse loci in immunodeficient RAG-2^−/− IL-2rγ ^−/− mice (termed MITRG mice). They also generated MITRG mice that also express human SIRPα encoded in a BAC (termed MISTRG). Engraftment of human fetal liver-derived CD34⁺ HSCs into the MITRG, MISTRG, RAG-2^−/− IL-2rγ ^−/− and NSG mice resulted in significantly greater myeloid cells in the blood and BM and nearly 10-fold increases in myeloid cells in the lung, liver and colon of MITRG and MISTRG mice compared to engrafted RAG-2^−/− IL-2rγ ^−/− and NSG mice (115). Human peripheral blood myeloid cell populations in of MITRG and MISTRG mice were composed predominantly of monocytes indicating that the mouse environment remained insufficient for the development and/or survival of human PMNs. Human NK cells were also detected in tissues at numbers approximately 10-fold higher in MISTRG engrafted mice compared to NSG engrafted mice. These exciting results were tempered by the poor development and survival of human RBCs in MITRG and MISTRG mice resulting in progressive anemia that developed due to poor mouse erythropoesis after irradiation preconditioning. Both the MITRG and MISTRG mice are currently commercially available.

Hydrodynamic delivery (via i.v. injection) of plasmids encoding different human cytokines has also been used in humanized mouse models to transiently increase expression of important cytokines and factors known to be important for hemato-lymphoid system development and function. Using this delivery system, Chen et. al. demonstrated that transient expression of human IL-15 and Flt-3/Flk-2 ligand in NSG mice increased the numbers of human NK cells for more than 30 days following HSC engraftment (116). These investigators also demonstrated that transient expression of GM-CSF and IL-4, macrophage colony stimulating factor (M-CSF) or erythropoietin and IL-3 significantly enhanced the generation of DCs, monocytes/macrophages or erythrocytes, respectively in Hu-HSC-NSG mice (116). In more recent studies, Li et. al. reported that hydrodynamic delivery of a plasmid encoding for human M-CSF enhanced the generation of mature and functional monocytes and tissue-associated macrophages in Hu-HSC-NSG mice (109). The use of transgenic technology is another approach that is currently being used to enhance expression of human cytokines and growth factors in immuno-deficient mice. For example, Billerbeck and coworkers have reported that transgenic expression of human stem cell factor (SCF), GM-CSF and IL-3 in NSG mice greatly enhanced the generation of myeloid cells following engraftment with CD34⁺ HSCs (117). Surprisingly, these investigators also observed an increase in the numbers of functional CD4⁺Foxp3⁺ regulatory T cells (Tregs) in peripheral tissues of these transgenic mice when compared to humanized NSG animals. Another group of investigators have shown accelerated differentiation of human granulocytes as well as generation of myeloid cells in NSG mice expressing the human transgene encoding SCF (118).

Strategies for enhancing the human adaptive immune system

The currently used humanized mouse models generate, to varying degrees, all lineages of human immune cells. The Hu-HSC models have been found to possess relatively weak primary and secondary immune responses including lack of immunoglobulin (Ig) class switching whereas NOD/scid-BLT and NSG-BLT models demonstrate more robust adaptive immune responses that include the generation of both IgM and IgG (9;26;80;92;93;95). As pointed out above, factors that limit the development of a well-integrated and fully functional adaptive immune system include defective cytokine and growth factor-mediated signaling in human progenitor cells, underdeveloped peripheral and mucosal lymphoid tissues and susceptibility of human lymphocytes to phagocytosis by mouse macrophages (9;26). A few studies have demonstrated that humanized mice develop weak antigen-specific T cell responses (26;119). Human T cells isolated from the spleens of NSG-BLT mice reveal functional deficiencies characterized by loss of expression of certain co-stimulatory molecules such as CD27(120). This specific deficiency could be partially reversed in vitro by addition of human IL-2 and IL-7. However, in general, development, maintenance and function of human T cells in current humanized mice are suboptimal (9). A few studies have shown, with varying degrees of success, that the development and maintenance of human T cells may be enhanced using delivery of human recombinant IL-7 or IL-15(34;121;122). Another factor that is critical for T cell development and function is the interaction between T cell receptors (TCR) and MHC class I and II. Several groups of investigators have recently reported the generation of transgenic IL-2rγ^−/− mice (e.g. NSG, NRG, NOG) expressing human MHC I or II molecules termed human leukocyte antigen (HLA)-I or II, respectively (9;28;98;123–129). Transgenic expression of HLA-I or -II increased the numbers and function of human T cells as well as enhanced antigen-specific immune responses following viral infection or immunization. The future development of transgenic IL-2rγ^−/− mice expressing both HLA-I and II but devoid of both murine MHC class I and II will allow human T cells to develop in a HLA- restricted manner (28).

In contrast to the relative paucity of functional T cells in Hu-HSC-IL-2rγ^−/− mouse models, human B cell numbers are actually quite reasonable. However, a number of different investigators have reported that B cell development is incomplete and antigen specific B cell responses are significantly reduced in this model compared to human responses (96;120;130–132). Investigators have observed a skewing of B cell development towards a more immature phenotype with steady state levels of human IgG at ~1% of that observed in humans (96;131;132). Although some investigators have shown that immunization or infection of Hu-HSC or BLT mice may increase production of human IgM and IgG, the results have been variable (132). For example, IgG formation following immunization of humanized mice with tetanus toxoid or viral antigens has not been consistently observed whereas immunization of BLT mice with 2,4-dinitrophyl-hapten-keyhole limpet hemocyanin produced antigen-specific generation of IgG1 and IgG2 (95;132). Because B cells isolated from Hu-HSC-IL-2rγ^−/− mice exhibit a highly diverse immunoglobulin repertoire that is very similar to that found in human B cells, it has been hypothesized that humanized mice have the “genetic potential” to mount antibody responses similar to humans (132). Studies to enhance human B cell maturation in humanized mice are currently being explored by different laboratories.

The lack of fully mature B cells in humanized mice can be explained in part by the apparent inability of murine B lymphocyte stimulating factor (BLyS or BAFF) to interact with human progenitor cells. Administration of human recombinant BLyS to NOD RAG-2^−/− Prf1^−/− mice engrafted with human PBMCs increased both human B and T cell engraftment (133). Treatment with human BLyS for 14 days resulted in 40 fold greater IgM and a 10 fold greater IgG production than controls and mounted 10-fold higher IgM and IgG de novo antibody response to pneumococcal thymus-independent antigens. Knock-in of human BLyS into immuno-deficient mice should be beneficial to enhance humoral and cell mediated immune responses in humanized mice in Hu-PBMC-IL-2rγ^−/−, Hu-HSC-IL-2rγ^−/− and BLT mice. Hydrodynamic injection of plasmids encoding for human GM-CSF and IL-4 has been reported to increase production of antigen-specific IgG following immunization in Hu-HSC mice (134). Another reason for reduced humoral responses in Hu-HSC-IL-2rγ^−/− mice may be due to defective T cell help since human T cells are selected within the mouse thymus based upon murine MHC-II rather than HLA-2 as described above. In support of the importance of HLA-II restricted T cells in humoral responses is a recent study reporting that engraftment of human HSCs into immuno-deficient IL-2rγ^−/− mice expressing the human HLA-2 (HLA-DR4) transgene results in the generation of IgM, IgG (all four subclasses), IgA and IgE following immunization (123).

Another important determinant for the development of a fully functional adaptive immune system is the presence of peripheral and mucosal lymphoid tissue. While immuno-deficient mice devoid of functional IL2rγ allow for greater engraftment of human hemato-lymphoid cells, there are few if any peripheral lymph node (PLN) or gut-associated lymphoid tissue (GALT) development in these animals due to impaired IL-7/IL-7R signaling (9;28;97;100). Absence of this important signaling pathway prevents the generation of lymphoid tissue inducer (LTi) cells that are required for PLN and GALT development. Transgenic overexpression of thymic stromal lymphopoietin (TSLP), a protein that has functional overlapping activity with IL-7, increased LTi numbers and restored PLN development in IL-7^−/− and RAG-2^−/−IL-2rγ^−/− mice (135). Thus, treatment with or overexpression of TSLP may be used to restore PLN and GALT development in humanized NSG mice. Although the effect of this approach on PLN and GALT development in immuno-deficient IL2rγ^−/− mice reconstituted with human hemato-lymphoid cells has not been reported, hTSLP transgenic NSG mice have been developed and reported in abstract form (136).

An under-investigated aspect of humanized mouse models is the role that intestinal microbiota play in immune system development. Chung and coworkers elegantly demonstrated that the immune system in germ-free mice fully develops in the presence of mouse but not rat or human microbiota (137). Germ-free mice colonized with human microbiota had markedly reduced CD4⁺ and CD8⁺ T cells in the intestine along with fewer proliferating T cells, dendritic cells, and low antimicrobial peptide expression compared to those colonized with mouse microbiota. Colonization of the mice with rat microbiota also failed to fully expand the numbers of intestinal T cells. However, mouse-segmented filamentous bacteria partially restored T cell numbers in the human microbiota colonized mice. Thus, human-specific microbiota is likely required for healthy human immune system development. Whether microbial-host interaction is required between the microbial cells and immune cells or microbial cells and host non-hematopoietic cells or both will need to be determined and humanized mice should be an excellent system to define these interactions.

Eradication of residual murine innate immune cells

Residual mouse innate immune function in immuno-deficient IL2Rγ^−/− mice continues to limit robust hemato-lymphoid cell engraftment. Although these mice have a complete absence of NK cells due to the deletion of the IL2rγ chain that is necessary for IL15R signaling, it is clear that their absence is not sufficient for full human immune cell engraftment.(9;80). Most human cells (especially lymphocytes and erythrocytes) are particularly susceptible to phagocytosis and destruction by murine macrophages and possibly other phagocytic leukocytes (e.g. PMNs). A major component of the innate immune system that prevents phagocytic destruction of its hemato-lymphoid cells is the interaction between the CD47 ligand and SIRPα rector. As briefly described previously, the SIRPα receptor is highly expressed on macrophages, dendritic cells and PMNs whereas its ligand CD47 is expressed on virtually all other cell types. Interaction of CD47-expressing cells with phagocyte-associated SIRPα initiates anti-phagocytosis signaling thereby preventing the inadvertent auto-destruction of nonphagocytic cells (9;65;66). Because murine SIRPα does not recognize human CD47, murine phagocytes readily engulf and destroy human hemato-lymphoid cells. The NOD strain contains a SIRPα polymorphism making it functionally similar to human protein that partially accounts for the improved engraftment in the NOD/scid and NSG strains (65;66). Genetically-modified strains have been developed with transgenic overexpression of human SIRPα (98) or that express murine CD47 in the human xenograft (138) that have enhanced engraftment and immune system development. As discussed previously, Lavender and coworkers have demonstrated that CD47^−/−RAG-2^−/−IL-2rγ^−/−-BLT mice are resistant to GVHD (106;107).

A number of other innovative approaches have been used in an attempt to reduce or eliminate macrophage-mediated destruction of engrafted human cells. Clodronate liposomes have been used to kill macrophages in vivo in NOD/scid or NSG mice receiving human CD34⁺ fetal liver cells resulting in enhanced survival of human red cells (139). However, clodronate also kills human macrophages which will limit it’s use in studies of immune function with its long-term use limited by toxicity. Brehm, Shultz and colleagues are evaluating CD11b or CD11c promoter-driven diphtheria toxin receptor expression as a means to selectively eliminate murine myeloid or dendritic cell populations in NSG mice (28;80). Murine macrophages and PMNs can also be depleted with antibodies such as RB6–8C5 (specific for macrophages and PMNs) or 1A8 (specific for PMNs alone)(140;141).

The ability to effectively study innate immune responses in autoimmunity or infectious diseases is complicated by the presence of murine Toll-like receptor (TLR) expression as well as reactive oxygen production. In order to address this, NSG mice devoid of TLR4, MYD88 cytosolic adapter protein, and type 1 interferon (IFN) receptors are currently being generated (80); however, this approach may have unintended consequences due to disruption of physiologic homeostatic and inflammatory roles for TLR expression in a variety of host cells including intestinal epithelial cells and myofibroblast (142;143). Additionally, NSG mice deficient for neutrophil cytosolic factor 1 (Ncf1), a gene that controls macrophage and PMN superoxide production (141;144), are currently being generated to reduce the cytotoxic activity of murine phagocytic leukocytes (80).

Use of Humanized Mice to Study Autoimmune and Inflammatory Diseases

The National Institutes of Health (NIH) estimates that ~24 million Americans suffer from one of the major autoimmune diseases such as Type 1 diabetes, lupus, rheumatoid arthritis, multiple sclerosis or the inflammatory bowel diseases. In contrast, the American Autoimmune Related Diseases Association (AARDA) reports that the actual number may be closer to 50 million individuals because the NIH only tracks and reports only those 24 autoimmune diseases with well-documented etiology studies (http://www.aarda.org/autoimmune-information/autoimmune-statistics/). In fact, the AARDA states that there are greater than 100 different autoimmune diseases that contribute to direct health care costs of more than $100 billion per year. Thus, there is an urgent need to develop new and more potent therapeutic agents to treat these lifelong diseases. The development of the new generation humanized mouse models has produced a great deal of excitement within the autoimmunity research community as in theory, it should be possible to model many of the major autoimmune and inflammatory diseases. The following section describes some of the more recent studies evaluating the pathogenesis and new drug therapies in humanized mouse models of autoimmune and chronic inflammatory diseases.

Graft-Versus-Host and Host-Versus-Graft Disease

Allogeneic bone marrow transplantation (BMT) may, in some instances, cure certain types of blood cancers and genetic disorders. Despite the long-term success of BMT, the development of graft-versus-host disease (GVHD) remains a major cause of morbidity and mortality in these recipients (68;145;146). Although GVHD is known to result from donor innate and adaptive immune cells responding to recipient antigens (ie MHC complexes), donor T cells are thought to be the most important disease-producing effector cells. Two types of GVHD have been described in humans and mice including acute GVHD (aGVHD) and chronic GVHD (cGVHD) (68;145;146). Development of aGVHD in mouse models is characterized by systemic inflammatory tissue injury of the intestine, liver, lungs and skin whereas cGVHD involves oral and salivary tissue, intestine, joints, liver, lungs and connective tissue (68;145;146). Several of the newer immuno-deficient IL-2rγ^−/− mice have been used to examine the immuno-pathogeneis of GVHD and evaluate new drug therapies to treat this systemic inflammatory disorders. In general, the Hu-PBMC-IL-2rγ^−/− mouse models appear to model a Th1/Th17-driven aGVHD (81;103;147–149). One concern with the use of conventional Hu-PBMC-IL-2rγ^−/− model is that it is, in actuality, a model of xenogeneic not allogenic GVHD. Covassin et. al. recently reported the generation of a novel humanized mouse model of allogeneic GVHD (150). Using NSG mice devoid of murine MHC class II but expressing the human transgene HLA DRB1-0401(called NSG-Ab° DR4 mice), these investigators found that injection of human DR4-positive CD4⁺ T cells failed to induce GVHD in NSG-Ab° DR4 mice whereas injection of CD4+ T cells obtained from human DR4-negative donors induced allogeneic GVHD (150). Although the majority of Hu-PBMC-IL-2rγ^−/− mice do not develop auto-antibody-mediated cGVHD, a few studies have reported cGVHD-like immunopathology (with fibrosis) in the BLT-IL2rγ^−/− model (68;94;102–105;151). These models have been particularly useful in testing a variety of different human-specific therapies such as anti-TNF antibodies, thymus-derived or induced regulatory T cells (Tregs) or mesenchymal stem cells (81;152–155). A more recent study demonstrates that administration of an antibody directed against HLA-A*02:01 to NSG mice injected with PBMCs expressing this human antigen, enhanced survival of these recipients (156). These exciting new data may provide a new approach to treating GVHD in HLA mismatched BMT.

Transplantation of allogeneic organs or tissues also induces potent immune responses in which the host immune system reacts vigorously to the engrafted tissue creating host versus graft disease, more commonly referred to as allograft rejection. (68;157). Despite decades of immunosuppressive therapy, allograft rejection remains a major hurdle for the long-term survival of the transplanted tissue. The use of humanized mice to study the immunopathology of xenograft or allograft rejection has been driven by the realization that the immune responses to human allografts in these animals differs substantially to those observed in the same species (mouse into mouse) allograft rejection models[reviewed in (68)]. However, the inability to transplant entire human organs into the mouse has limited allograft rejection studies to using human blood vessels, skin or islet cells. It should be noted that although a number of important studies have been performed using Balb/c-scid beige (SCID/beige) recipients, relatively few studies have utilized the newer immuno-deficient IL-2rγ^−/− platforms such as the NSG, NOG, NRG or BRG mice (Table 2). When human vascular allograft rejection has been studied in a Hu-PBMC-IL-2rγ^−/− model, investigators have found that engraftment of human arterial vessels and human PBMCs were significantly better than what occurs in the well-characterized Hu-PMC-SCID/beige model (158–160). The use of Hu-PBMC-IL-2rγ^−/− models has allowed investigators to evaluate new, human-specific cell-based therapies in treating vascular allograft rejection (159;160). In addition to vascular allograft rejection, investigators have utilized the newer humanized mouse models to explore the immuno-pathogenesis of skin allograft rejection. Transplantation of human skin in mice is technically easy and once engrafted, the allograft exhibits viable epidermis and dermis as well as a functioning vasculature. In addition, skin is a highly immunogenic tissue that elicits vigorous immune responses by the recipient (157;161). Racki et. al. have reported an important study in which they utilized NSG mice to examine human skin allograft rejection in absence or presence of human PBMCs (162). Surprisingly, they found that transplantation of human skin into NSG recipients elicited extensive perivascular infiltration of large numbers of mouse Gr1⁺ granulocytes that ultimately lead to the allograft injury and rejection (162). Treatment of NSG-engrafted mice with an antibody specific for Gr1 remarkably reduced the leukocyte infiltration as well as promoted wound healing and graft remodeling. Using this modification to allow allograft survival, the authors were able to demonstrate human PBMC-mediated allograft rejection following human PBMC administration (162). These studies have allowed investigators to evaluate the therapeutic efficacy of human specific therapies such as the use of human Tregs (163;164). The new and exciting work being performed on allogeneic islet cell transplantation is discussed in the following section.

Type 1 Diabetes

Type 1 diabetes (T1D) is a multifactorial autoimmune disease that arises from a complex interaction among genetics, the immune system and environment. As with other autoimmune and chronic inflammatory diseases, the incidence and prevalence of T1D has increased dramatically over the past 50–60 years especially in “modernized” societies (165). These data coupled with those demonstrating that concordance rates of <50% in genetically identical twins suggest that environmental factors (e.g. intestinal microbiota) may play an important role in the pathogenesis of T1D in genetically-susceptible individuals (165;166). Numerous animal and human studies suggest that immune-mediated destruction of pancreatic β cells involves both CD4⁺ and CD8⁺ T cells (165;166). A variety of different rat and mouse models of T1D have been used over the past several decades to investigate the immuno-pathogenesis and treatment of this autoimmune disease. Because a T1D-like disease was found to spontaneously develop in NOD mice without the need for addition of toxic chemicals or infectious agents, it became and continues to be the most popular animal model for T1D preclinical studies (19;167;168). Despite the wealth of information that has been generated related to the immuno-pathogenesis and treatment of T1D using this mouse model, no therapy has been shown to prevent or cure patients with T1D. The differences between mouse and human immune systems as well as differences in islet cell composition and function may be major contributors to this lack of bench to bedside transition (19;168). The development of humanized mouse models based upon the different immuno-deficient IL2rγ^−/− stocks provides investigators with immunologically more relevant animal models for investigating immuno-pathogenesis and treatment of T1D. King and coworkers developed a chemically-induced model of T1D in NSG mice in which mice are rendered hyperglycemic via injection (i.p) of the β cell toxin streptozotocin (STZ). These investigators found that transplant of human islet cells into STZ-treated NSG mice reduced the hyperglycemia created in these mice (79;169). Injection of human allogeneic PBMCs into NSG engrafted with islet cells abrogated the protective effect resulting in allograft rejection and return to hyperglycemia. Unfortunately, the use of STZ to induce β cell injury is problematic given that the response to STZ can be quite variable and mouse islets/β cells are known recover from the chemical injury (79;157;169). In an interesting variation of this model, Zhao et. al. demonstrates that adoptive transfer of irradiated splenic mononuclear cells (SMCs) derived from NOD mice with active diabetes into NSG mice prior to engraftment of human T cells induces a mild insulitis that acts to trigger severe insulitis and β cell destruction following engraftment of human PBMCs derived from T1D patients (170). These investigators show that human T cells migrate to pancreatic islets via the interaction between chemokine stromal cell-derived factor 1 and chemokine receptor CXCR-4. Severe insulitis and β cell destruction correlated with the infiltration of large numbers of human CD4⁺ and CD8⁺ T cells into the pancreatic islets (170). Engraftment of irradiated, NOD-derived PBMCs or human T1D PBMCs alone did not induce diabetes in the NSG recipients. Taken together, these data suggest that this approach may provide a more immunologically-relevant mouse model to study the role of T cells in the pathogenesis of T1D. However, it must be remembered that the Hu-PBMC-IL-2rγ^−/− models utilized to investigate the immunological mechanisms responsible for induction of T1D as well as islet allograft rejection restrict investigator to studying only the of role of human T cells in these processes.

A major step forward in generating mouse models to study islet allograft rejection in mice with a fully humanized immune system has been described by Brehm and colleagues (171). This model is based upon the discovery that mice possessing a spontaneous point mutation in the insulin 2 gene spontaneously develop diabetes by 7 weeks of age (172;173). Investigators determined that disease was due to insulin 2 protein misfolding resulting in induction of the unfolded protein response followed ultimately by β cell apoptosis (174;175). The diabetes/hyperglycemia that develops in these mice (called Ins2^Akita mice) are not associated with obesity or insulitis. Backcrossing Ins2^Akita mice with NRG mice produced immuno-deficient offspring that develop spontaneous hyperglycemia by 3–6 weeks of age (19;157;171). Investigators demonstrated that normoglycemia could be achieved by transplantation of human or mouse islet cells (171). They also found that within one month following engraftment of allogenic human HSCs into normoglycemic, human islet cell-containing NRG-Ins2^Akita mice, the majority (60%) of mice became hyperglycemic following rejection of allogeneic islet cells (171). Thus, this modification of the Hu-HSC-IL-2rγ^−/− model may offer new insights into the treatment and immuno-pathogenesis of allograft rejection. It should be noted that these data contrast with those reported by Jacobson and coworkers who found that human islet cells were not rejected by BRG mice engrafted with allogeneic HSCs (176). The reasons for these differences are not readily apparent at the current time. With refinement of these current humanized models and development of new models that more accurately mimic the rejection processes, investigators will be in a position to evaluate novel, human-specific therapies to reverse or even prevent allograft rejection. New cell-based therapies that include administration of ex vivo-expanded Tregs or bone-marrow-derived mesenchymal stem cells are showing great promise in prolonging islet allograft survival in diabetic mice (177–179).

As discussed above, T1D is a polygenic disease that appears to be mediated by both CD4⁺ and CD8⁺ T cells (165;166). Recent genome-wide association studies have confirmed earlier reports demonstrating that the greatest genetic susceptibility to T1D is localized to the class II region of the HLA complex on chromosome 6(165;166). Indeed, >90% of all pediatric patients with T1D cases possess HLA-DR3,DQB1*0201 (called DR3-DQ2) or HLA-DR4,DQB1*0302 (called DR4-DQ8)(165;166). There is good evidence to suggest that priming and activation of CD4⁺ T cells is critical to providing help to CD8⁺ cytotoxic T cells that ultimately destroy islet β cells (reviewed in (165;166)). Activation of CD4⁺ T cells occurs via the interaction between the CD4⁺ T cell receptor (TCR) and autoantigens presented by the HLA-DR or HLA-DQ complexes residing on antigen presenting cells (APCs) in the draining lymph nodes and/or within the pancreatic islets. Examples of known T1D-related autoantigens include preproinsulin (PPI), insulinoma-associated antigen-2, glutamic acid decarboxylase, zinc transporter and islet-specific glucose-6-phosphatase catalytic subunit (IGRP) to name just a few (165;180).

The generation of transgenic IL-2rγ^−/− mice expressing HLA class I or II has provided investigators the tools to evaluate the roles of different T cell subsets in autoantigen-driven insulitis and diabetes in vivo. Recent work by Viehman-Milan et. al. shows that engraftment of HLA-matched human PBMCs (DRB1*0401) that were pulsed (ex vivo) with autoantigen-derived peptides (PPI or IGRP) into NSG-DR tg mice induced the infiltration of T cells into mouse islets (180). Although T cell infiltration was observed following transfer of healthy, autoantigen-pulsed PBMCs, insulitis was found to be more severe following engraftment of autoantigen-pulsed PBMCs obtained from diabetic donors. Furthermore, the authors observed decreased insulin staining and β cell injury in NSG-DR tg mice engrafted with diabetic vs. healthy PBMCs. Although overt diabetes was not observed in this model, this is a very good mouse model for investigating the mechanisms responsible for the selective trafficking of autoreactive T cells to pancreatic islets as well as the mechanisms by which they damage the tissue. Other investigators have used a similar approach to ascertain the role that CD8⁺ T cells play in the induction of insulitis in the humanized IL-2rγ^−/− mouse models. Whitfield-Larry et. al. demonstrated that transfer of HLA-A2–matched PBMCs from patients with T1D but not from healthy donors into NSG-HLA-A2 transgenic mice induced selective and robust islet inflammation (181). The inflammatory infiltrate was composed of both CD4⁺ and CD8⁺ T cells. Importantly, the authors showed that invading CD8⁺ T-cells were autoantigen-specific that could be stimulated to produce large amounts of interferon-γ following ex vivo stimulation with different T1D-associated autoantigens (181). Similar to the model described above, these mice did not develop overt diabetes. Nevertheless, this model represents another important step forward that will be useful for defining the cellular and immunological mechanisms responsible for islet-specific lymphocyte trafficking and HLA-restricted antigen responses. In a third study, Unger and coworkers reported that injection of a IGRP-specific CD8⁺ T-cell clone generated from diabetic donor blood into NSG-HLA-A2 tg mice induced severe insulitis and islet cell damage (182). These data clearly showed that autoreactive CD8⁺ T-cell clones are capable of trafficking specifically to pancreatic islets where they mediated injury to HLA-expressing beta-cells. The fact that none of the three NSG-HLA transgenic mouse models described above develop hyperglycemia/overt diabetes suggest that engraftment with both CD4⁺ and CD8⁺ T cells may be required for full expression of disease. Transgenic expression of both HLA class I and II molecules may help to promote development of diabetes provided that disease can develop within 4–5 weeks i.e. before the onset of GVHD. Furthermore, the use of HLA tg versions of the Hu-HSC-NSG or BLT models may provide the necessary innate and adaptive immune cells required for full expression of disease in the absence of GVHD.

Multiple Sclerosis (MS)

MS is an autoimmune disease characterized by chronic inflammation and demyelination of the central nervous system (183). Although the precise etiology of MS remains to be defined, there is accumulating evidence that the inflammation and consequent neuronal demyelination occurs from aberrant T cell responses to myelin-derived antigens, such as proteolipid protein (PLP) and myelin basic protein (MBP)(183–185). The strongest genetic factor associated with the development of MS is expression of certain HLA class II molecules such as the HLA-DRB1*1501 allele which confers a ~3 fold increase in risk (186). A large number of studies, using different mouse models of MS have been used over the past 30 years that have greatly advanced our understanding of the immuno-pathological mechanisms responsible for CNS inflammation and injury (183). The most popular mouse models have included induced models based upon autoimmune encephalomyelitis (EAE) studies, spontaneous models of MS and more recently, transgenic mice expressing different human HLA alleles (183). Although these models have been instrumental in revealing important cellular and molecular pathways responsible for induction of CNS inflammation, they did not recapitulate human MS. An obvious reason for the differences is that CNS inflammation in mice is mediated by “mouse-specific” effector cell mechanisms. More recently, investigators have utilized the Hu-PBL-NSG model to produce a subclinical autoimmune model of encephalomyelitis (187). For this model, investigators co-transferred myelin oligodendrocyte glycoprotein (MOG)-pulsed, human DCs together with healthy human PBMCs obtained from the same donor as DCs into NSG mice. Control mice were injected with both human PBMCs and DCs that were not pulsed with antigen or pulsed with irrelevant peptides. Following the initial injections, the humanized mice received a “booster” injection (s.c.) of immature/nonpulsed DCs followed 12 hrs later with a subcutaneous injection of MOG (or irrelevant peptide) suspended in Freund’s complete adjuvant (FCA) to induce DC maturation. The authors found that at 24 days following engraftment of MOG-pulsed PBMCs, significant CNS inflammation could be observed. The CD45⁺ inflammatory infiltrate was composed primarily of human CD4⁺ and CD8⁺ T cells. Although human T cells obtained from spleens of humanized NSG mice were shown to mount robust MOG-specific immune responses, the classic signs of EAE (e.g. paralysis and paresis) were not observed in this novel model (187). Nevertheless, this new model of myelin-induced CNS inflammation will be a valuable tool to assess new drug therapies targeted at T cell trafficking and inflammatory cytokine production in a humanized system in vivo.

Autoimmune and Inflammatory Rheumatic Diseases

Autoimmune and inflammatory rheumatic diseases represents the largest subset of autoimmune diseases affecting over 45 million individuals in the US (188–190). These chronic inflammatory diseases are characterized by painful swelling of the joints, muscle and/or tissue. The most prevalent of the rheumatic diseases include rheumatoid arthritis (juvenile and adult), systemic lupus erythematosus, Sjogren’s syndrome, the spondylarthritides (e.g. ankylosing spondylitis, reactive arthritis, psoriatic arthritis, enteropathic arthritis) and systemic sclerosis (190). Although a number of different mouse models have been developed to act as surrogates for some of the major rheumatic diseases, the use of humanized mice to study human-specific immune mechanisms and treatment modalities in these disease models have only recently been attempted. Below we describe the use of immuno-deficient IL-2rγ^−/− stocks to generate mouse models of some the major rheumatic diseases.

Rheumatoid Arthritis

A variety of mouse models have been used to study various aspects of induced or spontaneous arthritis; however none of these models completely recapitulate the immuno-pathogenesis of human rheumatoid arthritis (RA)(191). Similar to diabetes, the susceptibility to RA is associated with certain HLA class II alleles. Previous studies have demonstrated that immunization of transgenic mice expressing the human HLA-DR4 or -DR1 allele with type II collagen (CII) induces chronic joint inflammation. As with other mouse models of autoimmunity, interpretation of data generated using these tg models is limited by the fact that joint inflammation is driven entirely by the mouse immune system (192;193). Recent studies have attempted to utilize the new-generation, humanized mouse models to examine arthritis pathogenesis and treatment. Based upon intriguing clinical observations demonstrating that Epstein-Barr virus (EBV) infection is associated with the development of RA in humans (194;195), Kuwana et. al. have developed an interesting model of erosive arthritis based upon the Hu-HSC-NOG mouse model (196). These investigators observed that 65% of the NOG mice that were engrafted with human HSCs and then infected with EBV exhibited extensive arthritis in knee and ankle joints beginning at 26 days post-infection. No joint inflammation was observed in the absence of EBV infection. Histopathological analysis revealed remarkable synovial proliferation as well as extensive bone destruction and edema all of which appear very similar to those observed in human RA (196). In addition, they found that the inflamed joint and synovial tissue contained large numbers of human CD4⁺ and CD8⁺ T cells, B cells and macrophages. These interesting results demonstrate that EBV infection is capable of inducing human immune cell-mediated erosive arthritis that is quite similar to human RA. In another study, Misharin and coworkers report the development of a humanized mouse model of acute arthritis (87). These investigators used a similar humanized mouse model in which NSG mice were engrafted with human CD34⁺ HSCs and then subjected to intra-articular injection of FCA 16 weeks following HSC engraftment to induce acute arthritis. At 7 days post FCA injection, mice developed histological evidence of arthritis that was characterized by erythema, edema, pannus formation and infiltration of human lymphocytes as well as human and mouse PMNs and macrophages. Furthermore, these investigators demonstrated that FCA-induced arthritis could be attenuated by administration of the TNF inhibitor Etanercept (87).

Systemic Lupus Erythematosus

Systemic lupus erythematosus (SLE) is a multi-organ, autoimmune disease that affects primarily the joints, kidneys and skin but may also involve the lungs and central nervous system. This rheumatic disease is characterized by production various autoantibodies and deposition of immune complexes into the different tissues. Although investigators believe that one or more of the autoantibodies generated during the development of disease is/are important in disease pathogenesis, the precise etiology of SLE has yet to be defined. As with RA and other autoimmune diseases, induced and spontaneous mouse models of SLE have greatly advanced our understanding of those immune cells and mediators that promote disease; however, development of disease in many of the mice is strain-specific and do not recapitulate certain immuno-pathological aspects of human SLE (197;198). In an attempt to model SLE in mice with a partially humanized immune system, Andrade et. al. have utilized the Hu-PBMC-BRG model to create mice that develop most or all of the pathological hallmarks observed in human SLE (197). These investigators found that human PBMCs obtained from either SLE or healthy donors (HD) engrafted very well in NRG recipients at 3–4 weeks post transfer. In addition, the circulating numbers of CD4⁺ T cells was remarkably reduced in SLE- vs. HD-engrafted mice whereas CD8⁺ levels were substantially greater in SLE vs. HD engrafted animals (197). The authors found no significant differences in human IgG and anti-dsDNA antibody levels in SLE vs. HD-engrafted mice. However, if mice were injected with PBMCs obtained from a SLE donor that was expressing high levels of anti-cardiolipin antibodies (aCL), all engrafted mice expressed 2–3 times more aCL than did the other SLE or HD engrafted mice. At 4–8 weeks post engraftment, SLE-NRG mice were found to develop proteinuria and human IgG immune complex deposition within the kidney glomeruli. Microthrombi as well as CD8⁺ T cell and B cell infiltration were observed in NRG mice engrafted with PBMCs obtained from SLE donors that expressed high titers of aCL. Taken together, this novel model appears to recapitulate some of the pathological features observed in human SLE.

Sjogren’s Syndrome

Sjogren’s syndrome (SS) is one of the most prevalent autoimmune diseases that affect ~3% of the population. Individuals who suffer from SS experience systemic inflammation however disease is usually localized to salivary and lacriminal glands (199). Histopathological analyses reveal the infiltration of large numbers of CD4⁺ T cells and the upregulation of different pro-inflammatory cytokines (199). Because the pathology observed in these mouse models of SS do not reproduce some of the key alterations described in human disease, investigators are just beginning to explore the use of humanized mice to study the immuno-pathological mechanisms responsible for induction of this prevalent disease. In the first and only study reporting the use of humanized mice in SS investigations, Andrade and coworkers showed that adoptive transfer PBMCs obtained from SS donors into NSG mice resulted in increased serum levels of a variety of human inflammatory cytokines including IFN-γ, IL-17, Il-6 and TNFα when compared to NSG engrafted with healthy donor (HD) PBMCs (200). In addition, these investigators observed the infiltration of significantly larger numbers of human CD4⁺ T cells into the salivary and lacriminal glands of the SS- vs HD-engrafted NSG mice. Although significantly larger numbers of human CD8⁺ T cells and B cells were observed in both glands in SS- vs. HD-engrafted animals, the absolute numbers of these lymphocytes were much smaller than those observed for CD4⁺ T cells (200). This inflammatory infiltrate correlated well with loss of target organ function as saliva production was significantly decreased by 36% at 4 weeks post transfer of SS-derived PBMCs. This exciting new model will likely provide investigators the unique opportunities to not only explore T cell-dependent mechanisms of tissue injury but will also provide an important platform to assess new therapeutic agents to treat this rheumatic disease.

Inflammatory Bowel Diseases

Crohn’s disease (CD) and ulcerative colitis (UC) are two of the best-characterized inflammatory bowel diseases. They are chronic and unrelenting inflammatory disorders of the small bowel and/or colon that affect approximately 1.5 million people in the US with a calculated annual cost for both medical expenses and work loss of almost $4 billion dollars (201). Although the etiologies of CD and UC have yet to be fully elucidated, there is growing experimental and clinical evidence to suggest that chronic gut inflammation results from a dysregulated immune response to components of the normal gut flora in genetically-susceptible individuals. This paradigm is based upon both animal and human studies demonstrating that: a) intestinal bacteria are required for the development of chronic gut inflammation in genetically-susceptible mice; b) increased numbers of epithelial-associated bacteria as well as enhanced bacterial translocation occur during chronic gut inflammation; c) certain subsets of patients with IBD express polymorphisms in genes that are involved in intracellular processing and killing of bacteria and d) IBD is associated with major alterations in the luminal composition of the microbiota, a situation termed dysbiosis. Whether dysbiosis is a cause or consequence of IBD has yet to be determined. Neither CD nor UC are considered classical autoimmune diseases since no convincing data have been reported demonstrating that these diseases develop from auto-aggressive immune responses to intestinal tissue self-antigens. If however, we consider the trillions of intestinal microbes as a “virtual organ” that is present for our entire lifetime, we may wish to redefine our microbiota as “self”(202;203). Using this definition of self, it is quite reasonable to consider IBD as just another type of autoimmune disease. Few autoimmune or chronic inflammatory diseases have garnered more attention from their respective research communities in such a short period of time as has IBD. In just the last 20 years, thousands of studies have been published using mouse models of IBD in order to define disease pathogenesis and potential treatment modalities. However, the success of translating the hundreds of promising preclinical studies into effective new therapies to treat IBD has been unimpressive. The reasons for the lack of bench-to-bedside transition have already been described in previous sections and are beyond the scope of this overview. We do however, refer the reader to recent reviews that examine this subject (1;8).

In view of the large number of preclinical studies that have been and continue to be performed, it is rather surprisingly to find that very few studies have attempted to model IBD in humanized mice. Nolte et. al. recently described a model of chemically-induced IBD using the Hu-PBMCs-NSG mouse model (204). These investigators engrafted NSG mice with human PBMCs obtained from healthy donors (HD) or from individuals suffering from UC or atopic dermatitis (AD). At 7 days post transfer, mice were sensitized, via dermal application, to the vehicle (ethanol) or the hapten oxazalone (OXA). The mice were then challenged 24 hrs later via intrarectal administration of either 50% ethanol or OXA in 50% ethanol (204;205). Histopathological analyses of the different groups revealed that mice challenged with ethanol or OXA developed the same degree of colonic inflammation that consisted of edema, fibrosis, crypt dropout and infiltration of T cells into the lamina propria (204). No significant differences were observed in the histopathology scores among NSG mice engrafted with HD, UC or CD PBMCs and challenged with OXA. Although this study presents interesting data related to “hapten”-induced colitis in humanized mice, it illustrates a major problem associated with mouse models of chemically-induced colitis: the use of erosive/toxic chemicals such as ethanol creates severe inflammation that may mask more subtle alterations induced in humanized mice. Another limitation with the use of this type of chemical model is that lacks participation by the human innate system that will likely be important to faithfully model human IBD.

Using a similar approach as described for the OXA model, Goettel et. al. recently reported in abstract form only the use of transgenic NSG mice that expressed human HLADR1 but lacked the murine MHC class II (called NSGAb0DR1 mice) as a platform for inducing acute colitis (206). These investigators engrafted the NSGAb0DR1 mice with HLA-matched CD4⁺ T cells and then sensitized (via dermal administration) the mice with the hapten, trinitrobenzene sulfonic acid (TNBS). One week later, these mice received an intra-rectal challenge with TNBS in ethanol resulting in clinical and histologic evidence of intestinal inflammation that was not observed in the NSGAb0DR1 mice receiving TNBS in ethanol but no human T cells (206). Although promising, the TNBS model suffers from many of the same limitations as does the do other chemically-induced models of gut inflammation that use erosive/toxic substances to “break” the mucosal barrier (1). A second preliminary study reported by this same group (abstract form only) showed that engraftment of NSGAb0DR1 mice with HLA-matched CD34⁺ HSCs from a patient with immunodysregulation polyendocrinopathy enteropathy X-linked (IPEX) syndrome induced systemic and lethal autoimmunity similar to that observed in humans with IPEX and mice that lack functional Foxp3 transcription factor (207). It will be of great interest to see how this type of humanized mouse model may be used in the future to model human IBD in a well-controlled and statistically-powered study.

A novel mouse model of allergen-induced intestinal inflammation has recently been developed by Weigmann et. al.(208). These investigators engrafted PBMCs from donors with clinical allergy to grass, birch pollen or hazelnut together with their respective allergen (ip) into NSG mice. At 21 days post engraftment, mice were challenged orally or rectally with their specific allergen. Mice engrafted with allergic PBMCs developed colonic inflammation that consisted primarily of human lymphocytes and neutrophils in both the oral or rectal challenged groups but not following saline challenge. Mice engrafted with non-allergic (healthy) PBMCs did not develop colitis when challenged with the allergens (208). Furthermore, the colitis could be significantly attenuated by administration of anti-human IgE antibody but not with control human IgG. Although it was not clear whether the PMNs infiltrating the colon were of human and/or murine origin, this study clearly demonstrates that human leukocyte trafficking to the intestine is largely intact and sufficient to develop colonic inflammation. Whether trafficking is sufficient to allow the full spectrum of IBD pathology remains to be determined.

Looking Forward

The development of humanized mouse models of IBD offers investigators the ability to study the pathophysiology of CD or UC in a manner not possible in humans. Pre-clinical therapies can be tested, and the environmental influences such as diet and the gut microbiome can be tested and defined. However, a number of issues and fundamental questions remain regarding our ability to generate these humanized mouse models. It may be a bit naïve to believe that chronic gut inflammation can be generated by simply engrafting HSCs obtained from patients with CD or UC into any of the immuno-deficient IL-2rγ^−/− mice. Because genetically-identical twins express relatively low concordance rates for CD (~30–35%) or UC (~15–20%), environmental factors (e.g. intestinal bacteria, diet, housing) most likely will play important roles in model development (209–214). Although mouse and human flora are in general, fairly similar in composition, significant differences do exist (9). Indeed, the enormous impact that mouse vs. human microbiota has on murine intestinal immune system development (137) suggests that the antigenic stimuli may require colonization with human flora, disease specific flora or autologous flora specific to the PBMC or HSC donors. If donor specific flora is required for induction of disease, antibiotic treatment followed by human fecal transfer may be sufficient and alleviate the need for a gnotobiotic facility (215). Transfer of autologous PBMCs into immuno-deficient mice engrafted with HSCs from IBD patients may also prove useful in providing reactive T-cells with immunologic memory to induce disease. Further refinements could include treatment with indomethacin or piroxicam to initiate and synchronize the onset of disease, an approach that has been used in the IL10^−/− model of chronic colitis (216;217). Another concern relates to the extent of engraftment of CD34⁺-derived endothelial cells in mice (9). Although the incomplete engraftment of human endothelial cells may influence trafficking of human leukocytes to lymphoid and nonlymphoid tissue, current evidence suggests that leukocyte infiltration and intestinal inflammation do occur in the xenogeneic GVHD that develops in the Hu-PBMC-IL-2rγ^−/− and BLT-IL-2rγ^−/− mouse models (104;155). In summary, the outlook for the generation of even better humanized mouse models to study autoimmune and chronic inflammatory diseases is very bright. It will be very interesting to following the evolution of this exciting aspect of animal model development over the next few years.