New generation humanized mice for virus research: Comparative aspects and future prospects

Abstract

Work with human specific viruses will greatly benefit from the use of an in vivo system that provides human target cells and tissues in a physiological setting. In this regard humanized mice (hu-Mice) have played an important role in our understanding of viral pathogenesis and testing of therapeutic strategies. Limitations with earlier versions of hu-Mice that lacked a functioning human immune system are currently being overcome. The new generation hu-Mouse models are capable of multilineage human hematopoiesis and generate T cells, B cells, macrophages and dendritic cells required for an adaptive human immune response. Now any human specific pathogen that can infect humanized mice can be studied in the context of ongoing infection and immune responses. Two leading humanized mouse models are currently employed: the hu-HSC model is created by transplantation of human hematopoietic stem cells (HSC), whereas the BLT mouse model is prepared by transplantation of human fetal liver, thymus and HSC. A number of human specific viruses such as HIV-1, dengue, EBV and HCV are being studied intensively in these systems. Both models permit infection by mucosal routes with viruses such as HIV-1 thus allowing transmission prevention studies. Cellular and humoral immune responses are seen in both the models. While there is efficient antigen specific IgM production, IgG responses are suboptimal due to inefficient immunoglobulin class switching. With the maturation of T cells occurring in the autologous human thymus, BLT mice permit human HLA restricted T cell responses in contrast to hu-HSC mice. However, the strength of the immune responses needs further improvement in both models to reach the levels seen in humans. The scope of hu-Mice use is further broadened by transplantation of additional tissues like human liver thus permitting immunopathogenesis studies on hepatotropic viruses such as HCV. Numerous studies that encompass antivirals, gene therapy, viral evolution, and the generation of human monoclonal antibodies have been conducted with promising results in these mice. For further improvement of the new hu-Mouse models, ongoing work is focused on generating new strains of immunodeficient mice transgenic for human HLA molecules to strengthen immune responses and human cytokines and growth factors to improve human cell reconstitution and their homeostatic maintenance.

Introduction

Mice have long been indispensable in infectious disease research by contributing a great deal to our understanding of disease pathogenesis and in developing prophylactic and therapeutic approaches. However since they differ evolutionarily in many aspects from humans, many important knowledge gaps still exist in translating the results to humans. In this regard, use of macaques has been invaluable for deriving more relevant comparative preclinical data with human pathogens that can productively infect these animals (Voevodin and Marx, 2009). The high cost associated with using these animals limits their use in large numbers and their inability to fully mimic some disease manifestations necessitates alternative approaches. In this regard, hu-Mice that harbor functional human cells in a physiological setting offer tremendous advantages for studying various human pathogens (Shultz et al., 2007). Studies employing these novel mice are likely to bridge the gap that exists between the results obtained from macaque studies to those seen in the natural human host. Hu-Mice are created by transplanting human cells or tissues into immunodeficient mice that accept human grafts without rejection. The engrafted human cells occupy relevant physiological niches, continue to function normally and provide ideal substrates for infection and proliferation of human pathogens thus providing a more dynamic setting compared to in vitro experimentation. Following infection, the pathogen can spread in vivo to different organ systems and elicit appropriate responses both at the cellular and whole organismal levels, allowing human immune responses and associated immunopathologies to be aptly studied (Shultz et al., 2007; Berges and Rowan, 2011; Nischang et al., 2012). During the last 25 years numerous advances have been made in the creation and exploitation of humanized mice for infectious disease research (Mosier, 2000; Jamieson and Zack, 1999; Shultz et al., 2007; Berges and Rowan, 2011; Nischang et al., 2012). Many knowledge gaps are beginning to be filled with the derivation of promising data with direct relevance for clinical studies. First and foremost, hu-Mice are currently put to use to study human specific pathogens such as HIV-1 and EBV that only infect humans in a natural setting. Second, human specific immune responses enable identification of the immunogenic epitopes that trigger the human immune system versus that of murine or non-human primates. These two elements shed light on the complex picture of human pathophysiology and immunopathogenesis. This brief review is focused on presenting a summary of the recent improvements in humanized mouse models using representative virus examples and highlighting comparative aspects.

Evolution of immunodeficient mouse strains for human cell engraftment

A common feature for all humanized mouse models is transplantation of human cells/tissues by various routes into immunodeficient mice that are receptive to xenografts without graft rejection. Gradual evolution and improvements in the derivation of immunodeficient mouse strains permitted generation of new and more complex humanized mouse models (Shultz et al., 2007). Nude mice, while lacking T cells and consequently having defects in T cell responses, still harbored mouse B cells and NK cells and thus were not permissive for lasting human cell reconstitution. The availability of severe combined immunodeficiency (SCID) mice lacking both T and B cells permitted creation of hu-PBL-SCID and SCID-hu-mouse models. Later derivation of NOD-SCID mice with lower levels of NK cells and additional innate immune defects allowed higher levels of human cell engraftment but were still far from ideal. A major breakthrough in this area was the targeted disruption of the murine IL-2 receptor common gamma chain (IL2-Rγc) gene. This gene encodes the common and essential signaling component for the action of cytokines IL-2, IL-4, IL-7, IL-9, IL- 15 and IL-21 (Shultz et al., 2007; Ito et al., 2008). Disruption of IL-7 and IL-15 signaling blocks native mouse NK cell development thus permitting enhanced human cell engraftment. This IL2-Rγc mutation, when selectively bred together with SCID, NOD, RAG1 or RAG2 gene mutations in different combinations yielded a variety of more severely immunocompromised recipient mice to achieve far superior human cell engraftment (Shultz et al., 2011). Rag2^−/− γc^−/−, Rag1^−/− γc^−/− (denoted as RG, BRG, DKO, or RAG DK), NOD/Shi-scid/γc^−/− null (NOG) and NOD/SCID/γc^−/− (NSG) mice are among some of these which are currently being used. Of these, NOG and NSG mice are similar with the minor difference being the nature of the common gamma chain disruption wherein the NSG mice have a full null phenotype and the NOG mice have disruption of the cytoplasmic tail of the molecule. Further improvements are now focused on introducing human HLA Class I and II immune system and cytokine genes to generate new transgenic mice on the above genetic backgrounds to permit improved engraftment and generate more robust human cellular and humoral immune responses (see below) (Shultz et al., 2011; Willinger et al., 2011).

Old and new humanized mouse models

A variety of hu-Mice are currently employed in virus research based on the specific pathogen studied, experimental needs and ease of preparation (Berges and Rowan, 2011; Nischang et al., 2012). Various human cells and tissues are engrafted into different immunodeficient mouse strains, each with its own advantages and disadvantages. A major advancement with the new hu-Mouse models over the older versions is their capacity for generating primary human immune responses. These models are summarized in Table 1 with a broad description provided below.

Table 1.

Summary of old and new humanized mouse models.

Model	Method/Mice used	Advantages	Disadvantages	Viruses/Antigens studied	Primary immune response
hu-PBL	i/p injection of human PBMC. SCID, NOD-SCID, NSG, NOG.	Easy to prepare. Immediate use. Good T cell engraftment.	No multilineage hematopoiesis. No primary immune response. Graft versus host disease.	HIV-1, EBV, Measles	No
hu-HSC	Intrahepatic injection of CD34+ HSC into neonates. Intravenous injection of CD34+ HSC into adults. Rag1−/− γc−/−, Rag2−/− γc−/−, NSG, NOG.	Easy to prepare. Multilineage hematopoiesis. Primary immune response. Mucosal engraftment.	No human HLA restriction.	HIV-1, Dengue, EBV, HBV, HCV, Hib vaccine, HSV-2, Salmonella typhi, Tetanus toxoid	Yes: humoral and cellular. IgM, weak IgG. No human HLA T cell Restriction.
SCID-hu	Co-implantation of human fetal liver and thymic fragments under kidney capsule. SCID or NOD-SCID.	Abundant T cell lymphopoiesis.	Surgery needed. Requires human fetal tissue. No multilineage hematopoiesis. No primary immune response. Poor peripheral T cell engraftment.	HIV-1, CMV, EBV, HHV6, HSV-1, HTLV-1, KSHV, VZV	No
BLT	Co-implantation of human fetal liver and thymic fragments under kidney capsule with additional i/v injection of autologous CD34+ HSC. Rag1−/− γc−/−, Rag2−/− γc−/−, NOD-SCID, NSG.	Multilineage hematopoiesis. Primary immune response. Mucosal engraftment. Presence of human thymus. Human HLA T cell restriction.	Surgery needed. Requires human fetal tissue.	HIV-1, Dengue, EBV, TSST1, WNV envelope, DNP-KLH	Yes: humoral and cellular. IgM, weak IgG. Human HLA T cell restriction.

hu-PBL mice

These are created by reconstitution with human mature PBMCs by intraperitoneal (i/p) route into SCID mice (Mosier, 2000). Current versions use either NSG or RG mice due to more efficient engraftment. The passively transferred human immune cells persist for several weeks and are capable of effector functions to a certain extent. They can be productively infected by hematopoietic cell tropic viruses such as HIV-1. Human memory B cells continue to generate antibodies from prior antigen exposure. However, there is no de novo multilineage human hematopoiesis, and consequently, no primary immune response. Graft versus host disease (GVHD) by the injected T cells is a drawback. This model is very appropriate for studying xeno-GVHD (Shultz et al., 2011).

SCID-hu mice

Surgical co-implantation of human fetal thymus and liver (containing hematopoietic stem cells) fragments under the SCID mouse renal capsule is performed to generate mice that harbor a functional human thymus (denoted as thy/liv organoid) (McCune, 1996). There is vigorous thymopoiesis with generation of human thymocytes and naive T cells. However, mature T cells are confined predominantly to the thy/liv organoid with poor peripheral circulation. These mice are unable to generate a human immune response due to the lack of a full spectrum of immune cells. Nevertheless, these models have been instrumental in the study of key aspects of viral pathogenesis with viruses such as HIV-1 and HTLV-1 and have laid the foundation for the generation and continued improvements of current humanized mouse models (Jamieson and Zack, 1999).

hu-HSC mice

Transplantation of hematopoietic stem cells (HSC) into a variety of immunodeficient mice by various routes is employed to generate hu-HSC mice. This model has evolved substantially over the years (Legrand et al., 2009). Early versions involved injection of hematopoietic progenitor cells (CD34+HSC, also termed SCID repopulating cells or SRC) into conditioned adult NOD-SCID mice by intravenous (i/v) or intra-femoral (i/f) routes. While there is de novo lymphopoiesis, T cell development is poor. Use of new generation IL-2 γc^−/− mice such as RG, NOG or NSG mice led to better human cell engraftment. Two versions of hu-HSC mice with distinct preparations currently exist, though with important differences. The first version calls for injection of HSC into adult irradiated NSG/NOG mice. Although cells of multiple hematopoietic lineages are generated, the yield of T cells is poor. The second, improved version is intrahepatic injection of HSC into conditioned newborn RG, NSG or NOG mice resulting in far superior human cell engraftment with generation of a full complement of T cells, B cells, macrophages, NK cells and dendritic cells (Berges and Rowan, 2011; Shultz et al., 2011). Infection of either version of these mice with different pathogens or immunization with different antigens gives rise to human immune responses (see below). There is also good mucosal human cell engraftment in neonatally transplanted mice permitting HIV-1 infection by mucosal routes (Berges et al., 2008).

BLT mice

This model is a slight modification of the earlier SCID-hu mouse model and a notable improvement. The acronym derives from transplantation of bone marrow, liver and thymus (BLT), the chief difference with the SCID-hu mouse model being the additional reconstitution with autologous HSC purified from the same fetal liver source (Lan et al., 2006; Melkus et al., 2006). The original BLT version used NOD-SCID mice while newer improved versions use NSG, NOG or RG mice (Biswas et al., 2011; Stoddart et al., 2011). Human cell engraftment with multilineage generation of T cells, B cells, macrophages, NK cells and dendritic cells is seen, as well as appropriate human T cell education and restriction, due to the presence of an autologous human thymus.

hu-Liver-HSC mice

The above hu-Mouse models are restricted to human immune cell/tissue reconstitution with de novo generation or maintenance of human immune cell subsets. Hu-Mice bearing other human tissues can also be generated to permit infection with other human specific pathogens that have a predilection to infect organ systems such as the liver (uPA/SCID and RG Fah^−/− mice) (Meuleman and Leroux-Roels, 2008; Zhou et al., 2012). This further broadened their application in infectious disease research. Novel transgenic mice that simultaneously permit human hepatocyte and HSC engraftment were recently developed to generate RG AFC8-hu HSC/Hephu mice. These mice are susceptible to infection with HCV and give rise to human specific immune response (Washburn et al., 2011).

hu-Mice preparation, infection and immunization: An overview

A general outline for creating hu-HSC and BLT hu-Mice is presented in Fig. 1. The following description summarizes some important aspects of the preparation of various versions of hu-Mice and their application for viral infection and immunization studies. The easiest version of hu-Mice to generate is hu-PBL mice. Adult mice, such as NSG mice, are injected i/p with 2.0 × 10⁷ PBMC enriched from human blood (Mosier, 2000). The mice can be used immediately for various purposes that include viral infections and newer protocols evaluating the effector functions of designer T cells with engineered T cell receptors. Preparation of hu-HSC mice involves injection of human CD34+ hematopoietic progenitor cells (a heterogeneous population of which approximately 1% are believed to be HSC) derived from different sources such as cord blood, bone marrow, cytokine mobilized peripheral blood or human fetal liver. Of these, the efficiency and duration of engraftment appear to be better with the fetal liver derived CD34+ cells due to their more primitive lineage development status.

Fig. 1.

Schematic for generation of hu-HSC RG (Rag-hu) and BLT mice.

Compared to SCID-hu mice and BLT mice, preparation of hu-HSC mice is not technically intensive, as no surgery is involved. Immunodeficient mouse strains, namely RG, NOG or NSG, are commonly used (Traggiai et al., 2004; Berges et al., 2006; Becker et al., 2010; McDermott et al., 2010; Akkina et al., 2011; Shultz et al., 2011). Injection of HSC into neonatal mouse livers within 3 days of birth versus adult mice gives far superior engraftment, with generation of human immune competent mice that harbor the four essential immune cell subsets, namely T cells, B cells, macrophages and dendritic cells. The following protocol is routinely used in our laboratory and yields well engrafted mice. Fragments of human fetal liver (16–20 weeks gestation) are subjected to enzymatic digestion with collagenase, DNase and hyaluronidase to prepare a single cell suspension. The cells are incubated with anti-CD34 antibody and enriched by immunomagnetic bead-based positive selection. The purity of CD34+ cells generally ranges between 90 and 99% after two successive cycles of selection. The purified cells are cultured overnight in a human cytokine media mix containing IL-3, IL-6 and SCF. Freshly purified cells are preferred although previously frozen cells can also be used. The neonatal mice are conditioned by irradiation at 350 rads 2–24 h prior to cell injections. We routinely use 5 × 10⁵ human fetal liver derived CD34+ cells per mouse pup to ensure consistent engraftment, although fewer cells can be used, in which case the duration of engraftment may be shorter due to lower numbers of true hematopoietic stem cells. A 30 µL volume of cells is injected intrahepatically by visualizing the dark area occupied by the liver under the relatively transparent skin (Fig. 2). Post-manipulation, pups are returned to their mothers in BSL-2 conditions and weaned 3 weeks later. The engrafted mice are screened to determine the levels of human CD45+ cells in peripheral blood at around 12 weeks of age. On average we obtain mice with 40–90% human cell engraftment. Human cell engraftment is seen in primary and secondary lymphoid organs, as well as mucosal engraftment in the female reproductive tract and the gut, permitting HIV-1 mucosal transmission (Berges et al., 2008). In general 20–30 hu-HSC mice can be made with a typical batch of fetal liver derived CD34+ cells.

Fig. 2.

Intra-hepatic injection of CD34+ hematopoietic stem cells (HSC) into newborn RG mouse.

With regard to the preparation of BLT mice, original reports used NOD-SCID mice (Lan et al., 2006; Melkus et al., 2006; Denton and Garcia, 2011). More recent protocols use NSG mice due to far superior engraftment (Biswas et al., 2011; Denton et al., 2012; Marsden et al., 2012). Adult mice are conditioned by sub-lethal whole body irradiation at 325 rads prior to human tissue transplantation. Human fetal thymic and liver tissues (16–22 weeks gestation) are dissected into 1 mm fragments and introduced together by the use of a trochar under the left kidney capsule of anesthetized mice. Each of the mice is later injected (i/v, tail vein) with 2.5 × 10⁵ autologous CD34+ HSC purified from the remaining fetal liver. The transplanted mice are evaluated for human cell engraftment at 7–12 weeks before their use for various experiments. In practice, 15–20 mice can be made with a typical set of fetal tissues. Preparation of SCID-hu mice is similar to that of BLT mice with the exception that no CD34+ cells are injected after implantation of thymus and liver tissues and the mice are not irradiated.

Viral pathogens such as HIV-1, dengue and EBV have been more commonly studied in these new hu-Mouse models (Van Duyne et al., 2009; Berges and Rowan, 2011; Ramer et al., 2011). The most common route of experimental infection with HIV-1 is via the i/p route, which typically involves use of either the CCR5-tropic viral strain BaL or the CXCR4-tropic strain NL4-3 (1 × 10⁵ i.u) (Berges et al., 2006). Viremia is seen within one week and infection can persist life-long based on the maintenance of human cell engraftment, which in turn depends on the quality of the HSCs injected (Berges et al., 2010). Viral inocula (1 × 10⁶ i.u.) are delivered by i/p, subcutaneous (s/c) or intradermal (i/d) routes for dengue infection (Kuruvilla et al., 2007; Jaiswal et al., 2012). Mice develop acute viremia generally lasting for three weeks. With EBV, 1 × 10⁵–1 × 10⁶ RIU are injected i/p (White et al., 2012). In addition to infection with live viruses, a variety of antigens have been used for experimental immunizations that include tetanus toxoid (TT), HBV, HIV-1 and West Nile virus envelop antigens, among others (Traggiai et al., 2004; Berges and Rowan, 2011; Becker et al., 2010; Biswas et al., 2011). The immune response onset is slow in these mice and takes longer to peak, generally around 4–10 weeks.

Humoral immune responses in hu-Mice

Several reports documented antigen-specific human antibody responses in hu-HSC mice (Matsumura et al., 2003; Traggiai et al., 2004; Ishikawa et al., 2005; Gorantla et al., 2007; Kuruvilla et al., 2007; Watanabe et al., 2007; Shultz et al., 2010; Garcia and Freitas, 2012). While both IgM and IgG responses were reported by different investigators, in general, IgG responses were found to be somewhat weak. Antibody repertoire in hu-HSC RG mice by analysis of the length of the CDR3 hypervariable regions revealed that the human IgM B cell repertoire was akin to that of normal healthy individuals thus indicating no obvious limitations to generate human antibodies of various specificities (Becker et al., 2010). However, TT and HBV vaccine immunizations of hu-HSC RG mice gave a predominantly IgM response with limited antigen specific IgG production, thus indicating a general failure to class switch. Given that these mice do accumulate total serum IgG efficiently, the paucity of antigen specific IgG is puzzling. One study employed PBMC transfected with a human TCR specific for influenza HA peptide and bearing a matched HLA with that of Hu-Mice prepared with matched HSC. When these T cells were passively transferred into the respective mice and antigen challenged, there was an increase in total IgG indicating that the isotype switch deficiency is not due to an intrinsic defect in the B cells but rather an impairment in T cell cooperation (Watanabe et al., 2009). This can be attributed to human T cell restriction by murine MHC in addition to a potential T cell dysfunction. When IgD+ CD19+ naive B cells from hu-HSC mice were treated in vitro with anti-CD40 antibodies, IL-2, and IL-21 in the presence of antigen they became activated and secreted IgG, again confirming the functionality of B cells in these hu-HSC mice. In substantiating human HLA restricted T cell help, another study using HLA-DR4 (MHC Class II) transgenic mice and reconstitution with matching HSC reported improved immune responses with higher levels of IgG production with efficient class switching (Danner et al., 2011). A more recent phenotypic analysis of B cells derived from hu-HSC mice generated using adult NSG mice revealed a normal B cell developmental pathway (Chang et al., 2012). Molecular analysis of single B cells however indicated that while the overall distribution of Vh genes reflected a normal human antibody repertoire, mature B cell subsets showed autoimmune characteristics (Chang et al., 2012). The wide variations seen in hu-HSC mice with regard to antibody production and class switching could be attributed to a number of factors, including a lack of proper human T cell restriction and help, as well as differences in protocols, i.e., utilizing neonatal mice versus adult mice for HSC transplantation or using different HSC sources, namely cord blood and fetal liver. Nevertheless, additional improvements of hu-HSC are necessary as discussed below. Conventional SCID-hu mice have been modified to generate BLT mice and thus provide a more appropriate thymic microenvironment for human T cell development and improved T/B cell cooperation (Lan et al., 2006; Melkus et al., 2006). This resulted in a more robust T cell development in addition to the generation of B cells, macrophages and dendritic cells. Positive and negative selections of T cells are expected to occur in the autologous human thymus during their maturation. Mature T cells in these mice were shown to generate MHC class I and II restricted human immune responses and offer T cell help to the antigen stimulated B cells (Melkus et al., 2006; Tonomura et al., 2008). A number of early studies in BLT mice have shown both IgM and IgG antigen specific responses, albeit with varied robustness (Melkus et al., 2006; Berges and Rowan, 2011). However a subsequent study failed to show antigen specific IgG responses despite repeated booster immunizations (Rajesh et al., 2010). This is attributed to a lack of optimal conditions for germinal center formation and Ig class switching. A recent study evaluated BLT mice more in depth to determine the antigen specific antibody responses by immunization with adjuvanted HIV-1 and WNV envelop antigens (Biswas et al., 2011). Marked differences were noted both in terms of B cell composition and antibody responses compared to that of a healthy human. Even repeated booster immunizations did not result in secondary responses characterized by the production of IgG. In contrast to the human, an abundance of a “B-1 like” B cell population (CD19+CD5+) was noted. The predominant IgM antibody response and lack of IgG is attributed to the CD5+ B cell subset believed to be responsible for production of ‘natural antibody‘ using a T cell independent pathway.

hu-Mice for generation of human monoclonal antibodies

The new hu-Mouse models can be exploited to generate a broad spectrum of antibodies, both neutralizing and non-neutralizing since they are shown to harbor a normal human antibody repertoire. The broad scheme would involve either infection or vaccination with a desired antigen with or without an adjuvant. Splenocytes containing the antigen specific B cells can be harvested at the peak time of immune response for later selection, expansion and immortalization. In an alternative approach, antigen specific individual B cells can be sorted and their antibody genes cloned for incorporation into an ectopic expression system. Based on the desired application, specific antibody gene sequences can also be class switched for therapeutic purposes using molecular techniques. Following these lines one study achieved generation of human monoclonal antibodies employing hu-HSC RG mice (Becker et al., 2010). Following immunization with commercially available TT and HBC vaccines, memory B cells expressing surface immunoglobulins were FACS sorted from splenic and mesenteric lymph node single cell suspensions. A retroviral vector encoding BCL6 and BCL-XL genes was used to immortalize the sorted B cells which were cultured in the presence of CD40L and IL-2. Culture supernatants were tested by ELISA to identify specific antibody producing wells followed by limiting dilution culturing to obtain monoclonal B cell lines. Only IgM antibody-secreting B cell clones could be obtained, although IgG responses were seen in the immunized mice albeit at lower levels. Further improvements in hu-Mice should lead to more efficient generation of all immunoglobulin classes.

Human T cell immune responses and HLA restriction in hu-Mice

Early reports showed induction of antigen specific T cell responses in the new generation hu-HSC mice against various human pathogens (Hiramatsu et al., 2003; Strowig et al., 2009; Berges and Rowan, 2011). Depletion of CD8+ T cells abrogated the immune control of HIV-1 and EBV infection in hu-HSC mice thus providing additional evidence for their role in protection (Yajima et al., 2009; Gorantla et al., 2010a, 2010b). However, other studies reported deficiencies in T cell responses (Baenziger et al., 2006; An et al., 2007; Watanabe et al., 2009). Polyclonal stimulation of hu-HSC mouse splenocytes by PHA, anti-CD3/anti-CD28 antibodies and PMA/ionomycin, while leading to cell proliferation and cytokine secretion, revealed 10-fold lower response than was seen with human PBMC, suggesting a functional defect. Also, human T cells responded poorly to in vivo immunizations as shown by the lack of IFN-γ or IL-4 secretion after specific antigen re-stimulation ex vivo. Specific CD4 and CD8 responses were measured by cytokine secretion assays, cell proliferation assays or cytotoxic assays in vitro by restimulation of cells from mice infected with either HIV-1 or EBV. Even in reports showing immune activity, the responses are low. These suboptimal responses are attributed to several factors. First is the overall low level of T cells in these mice, which is believed to result from a potential lymphopenia-induced T cell activation among other causes (Garcia and Freitas, 2012). Second, lower levels of T cells in hu-HSC mice is also thought to be due to a lack of human HLA restriction, since T cell selection is happening in the xenogenic mouse thymic environment in a H-2 restricted fashion, which is likely not efficient for human cells. Third, T cells generated also exhibit poor survival in the periphery as a result of their less than optimal interactions with mouse APCs and weak signaling (Garcia and Freitas, 2012). In support of these possibilities, it was recently reported that when RG mice transgenic for HLA-DR4 were reconstituted with matching HLA-DR4 CD34+ cord blood cells, there was a drastic elevation in the numbers of thymic and peripheral T cells (Danner et al., 2011). Shultz et al. generated Class I HLA-A2 transgenic mice and reconstituted them with matched HSC (Shultz et al., 2010). These mice were shown to be capable of HLA restricted cellular immune responses to EBV. Based on these findings it is now evident that forced expression of HLA-A2 and HLA-DR4 enables HLA restricted T cell functions correlating with improved cytokine secretion and IgG production. T cell responses in BLT mice are found to be qualitatively normal due to reconstitution with autologous human thymus and HSC thus permitting HLA restriction of T cell responses and more efficient T and B cell interactions. In a recent study HIV-1 infection in BLT mice gave rise to epitope specific anti-viral responses in a class I restricted manner (Dudek et al., 2012). Antigen specific CD8+ T cell responses were found to mimic those in humans in terms of their specificity, kinetics and immunodominance. It was also found that mice expressing the particular HLA class allele HLA-B 57 exhibited enhanced control of HIV-1 infection as seen in humans that bear the same allele.

Human pathogens and other antigens studied in hu-Mice

A variety of human pathogens, particularly viruses, have been studied in new generation hu-Mice (Table 2). Of these HIV-1, EBV, dengue and HCV are by far the most widely studied and are discussed in more detail in sections below. Hu-Mice have also been used to study HTLV-1 proviral integration and induction of T cell lymphomas (Banerjee et al., 2010; Yamamoto et al., 2010; Villaudy et al., 2011). A number of studies focused on dengue viral infection and showed viremia with concomitant humoral and cellular responses (Kuruvilla et al., 2007; Cox et al., 2012; Jaiswal et al., 2012). Hu-Mice with human hepatocyte reconstitution allowed infection with hepatotropic viruses such as HCV and HBV inducing pathologies and immune responses (He et al., 2010; Washburn et al., 2011). A variety of human herpes viruses have also been studied. These reports documented HLA-restricted adaptive T cell immune responses to EBV, CMV reactivation from latency, protective innate and adaptive immune responses against intravaginal HSV-2, and generating an anti-KSHV-antibody immune response (Melkus et al., 2006; Parsons et al., 2006; Kwant-Mitchell et al., 2009; Strowig et al., 2009; Smith et al., 2010). More recent studies have expanded the use of hu-Mice to other non-viral pathogens. These encompass work with drug resistant Salmonella typhi (Libby et al., 2010; Firoz Mian et al., 2011), persistent infection with the malaria parasite P. falciparum (Arnold et al., 2011) and detection of hu-Mouse adaptive and innate immune responses against Leishmania (Wege et al., 2012). In addition to live pathogens, a number of antigens and human vaccine preparations have also been tested to evaluate Hu-Mice human immune responses. Immunization with DNP(23)-KLH antigen generated human T cell proliferation and human IgG responses (Tonomura et al., 2008). Toxic shock syndrome toxin 1 caused an expansion of human T cells and activation of human dendritic cells (Melkus et al., 2006). Administration of a variety of vaccines demonstrated adaptive immune responses including influenza-specific human CD8+ T cells, human IgM antibody responses to tetanus toxoid and HBV (Yu et al., 2008; Becker et al., 2010). This list is only a partial representation of that reported in the current literature which is expected to grow as the use of these mouse models becomes more wide spread.

Table 2.

Examples of viruses studied in new Hu-Mouse models

Virus	Research area	Topic	Model	References
HIV	CD4 T cell loss/Immune response	CD4 depletion/Cellular	NOG	(Nie et al., 2009)
		CD4 depletion/Cellular	RG	(Berges et al., 2006; An et al., 2007; Zhang et al., 2007; Baenziger et al., 2006; Berges et al., 2010)
		Cellular (CD8 depletion)	NSG	(Gorantla et al., 2010b))
		Cellular	NOK (HLA)	(Sato et al., 2012)
		Humoral	RG	(Gorantla et al., 2007)
		Humoral	NOG	(Watanabe et al., 2007)
		Cellular/Humoral	NSG	(Singh et al., 2012)
		Cellular/Humoral	NSG-BLT, NOD/SCID-BLT	(Brainard et al., 2009)
		Cellular/Humoral	NOG	(Sato et al., 2010)
		Humoral	NSG	(Chang et al., 2012)
		Nef and replication	BLT	(Zou et al., 2012)
		Vpu and replication	NOG	(Sato et al., 2012)
	Immunopathogenesis	Atherosclerosis	NSG	(Dubrovsky et al., 2012)
		Bacterial translocation	RG	(Hofer et al., 2010)
		IFNα	NSG-BLT	(Long and Stoddart 2012)
		pDC	RG	(Zhang et al., 2011)
		Treg	RG	(Jiang et al., 2008)
		Viral dissemination	BLT	(Murooka et al., 2012)
	Neuropathogenesis	Neuroinflammation	NSG	(Gorantla et al., 2010a)
		Neuronal integrity	NSG	(Dash et al., 2011)
	Latency	Latency/Drug reactivation	NSG-BLT	(Marsden et al., 2012)
		Latency	NSG-BLT	(Denton et al., 2012)
		Latency	RG	(Choudhary et al., 2012)
	Evolution/Immune escape	APOBEC3-mediated mutagenesis	NOG	(Sato et al., 2010)
		Cellular immunity escape	BLT	(Dudek et al., 2012)
		Env mutations	RG	(Ince et al., 2010)
		gp41 mutation	NSG-BLT	(Garg et al., 2011)
	Transmission/Prevention	Vectored broadly neutralizing abs	NSG, RG	(Balazs et al., 2012)
		Broadly neutralizing antibody vaginal microbicides/VRC01	RG	(Veselinovic et al., 2012)
		Antiretrovirals/Oral	BLT	(Wahl et al., 2012)
		Antiretrovirals/Rectal and vaginal	NOD/SCID-BLT	(Denton et al., 2008; Denton et al., 2010)
		Antiretrovirals/Vaginal	NSG-BLT	(Denton et al., 2011)
		Antiretrovirals/Vaginal	RG	(Neff et al., 2010; Neff et al., 2011)
		Aptamer-siRNA	NSG-BLT	(Wheeler et al., 2011)
		Rectal transmission	NOD/SCID-BLT	(Sun et al., 2007)
		Vaginal transmission	NSG-BLT	(Stoddart et al., 2011)
		Vaginal and rectal transmission	RG	(Berges et al., 2008; Hofer et al., 2008; Akkina et al., 2011)
	Antiretrovirals	Nano-formulated drugs	NSG	(Dash et al., in press)
		HAART	RG	(Choudhary et al., 2009; Sango et al., 2010)
		Long acting drugs	NOG	(Nischang et al., 2012)
	Novel therapeutics	Aptamer, aptamer-siRNA chimera	RG	(Neff et al., 2011; Zhou et al., in press)
		Dendrimer formulated-siRNA	RG	(Zhou et al., 2011)
		siRNA	RG	(ter Brake et al., 2009)
		siRNA	NSG	(Kumar et al., 2008)
		siRNA-nanoparticles	NSG-BLT	(Kim et al., 2010)
		Tat peptides	RG	(Van Duyne et al., 2008)
	Gene therapy	Broadly neutralizing 2G12 Ab	RG	(Luo et al., 2010)
		Anti-CCR5 zinc finger nuclease	NSG	(Holt et al., 2010)
		Anti-CXCR4 zinc finger nuclease	NSG	(Wilen et al., 2011; Yuan et al., 2012)
		Anti-gag TCR	NSG-BLT	(Kitchen et al., 2012)
		Antisense Env	NSG	(Mukherjee et al., 2010)
		shRNA	NOD/SCID-BLT	(Shimizu et al., 2010)
		TRIM5α isoform, CCR5 shRNA, Tar decoy	RG	(Walker et al., 2012)
		Broadly neutralizing 2G12 Ab	NSG	(Joseph et al., 2010)
CMV	Immune response	Humoral	NOD-SCID β2 M	(Aspord et al., 2011)
	Latency	Latency activation	NSG	(Smith et al., 2010)
Dengue	Infection/Immune response	Humoral	RG	(Kuruvilla et al., 2007)
		Cellular/Humoral	NSG-BLT	(Jaiswal et al., 2012)
	Transmission	Mosquito	NSG	(Cox et al., 2012)
	Immunopathology	Clinical signs	NOD-SCID	(Bente et al., 2005)
		Clinical signs	NSG	(Mota and Rico-Hesse 2011)
EBV	Infection/Immune response	Cellular	RG	(Traggiai et al., 2004)
		Tumorigenesis	NSG	(White et al., 2012)
		Cellular	NOG	(Yajima et al., 2009)
	Antivirals	Protease inhibitor	NSG	(Dewan et al., 2009)
HBV	Transmission	Cross-species	uPA/SCID	(Sa-Nguanmoo et al., 2011)
	Immunopathology	cccDNA loss	uPA/SCID	(Lutgehetmann et al., 2010)
	Antivirals	IFNα, Ribavirin, Cyclophilin inhibitor	RG Fah−/−	(Bissig et al., 2010)
		Capsid destabilizer	uPA/SCID	(Brezillon et al., 2011)
HCV	Immune Response	Cellular	RG/AFC8	(Washburn et al., 2011)
	Transmission	Envelope glycoprotein fitness	uPA/SCID	(Brown et al., 2012)
	Antivirals	IFNα, Ribavirin, Cyclophilin inhibitor	RG Fah−/−	(Bissig, Wieland et al., 2010)
HSV-2	Infection/Immune response	Cellular/Humoral	RG	(Kwant-Mitchell et al., 2009)
HTLV-1	Infection/Immune response	Cellular	NOG	(Takajo et al., 2007)
	Latency	Proviral integration	NOG	(Yamamoto et al., 2010)
	Tumorigenesis	T cell development	RG	(Villaudy et al., 2011)
Influenza	Immune response	Vaccine	NOD-SCID β2 M	(Yu et al., 2008)
	Antivirals	Aminobisphosphonate	RG	(Tu et al., 2011)
KSHV	Infection/Immune response	Humoral	NOD-SCID	(Parsons et al., 2006)
	Immunopathology	Splenic B cell expansion	NSG	(Boss et al., 2011)

HIV pathogenesis and latency

Since HIV-1 is a human specific virus, hu-Mice that harbor HIV-1 susceptible cells in a physiological setting provide an excellent in vivo experimental system. Indeed numerous studies were conducted in the early versions that include hu-PBL and SCID-Hu mice providing valuable data (Jamieson and Zack, 1999; Mosier, 2000). However, a major limitation of these systems was the lack of a full repertoire of human immune cells and a functioning immune system, thus limiting detailed studies on immunopathogenesis which plays a central role in disease onset. The new hu-Mice have mostly rectified these deficiencies and thus permitted many new experimentations (Legrand et al., 2009; Berges and Rowan, 2011; Nischang et al., 2012). The hu-Mice used consist of RG, NOG and NSG backgrounds. HIV-1 infection is readily established by different routes that include i/v, i/p and vaginal and rectal mucosal routes. Viremia is established generally within a week. One study using hu-HSC RG Mice showed that chronic viremia can persist for as long as one year, thus permitting long term studies in this model (Berges et al., 2010). Both CCR5 and CXCR4 tropic as well as dual tropic viruses can readily establish infection and there is characteristic helper CD4+ T cell loss, which is a hallmark of HIV-1 immunosuppression. As seen in the human, CCR5 tropic viruses preferentially deplete CD45 RA CD4+ memory T cells whereas the CXCR4 viruses quickly deplete both CD45 RA+ naive and CD45 RA− memory CD4+ T cells (Nie et al., 2009; Berges and Rowan, 2011). HIV-1 infection in these mice is disseminated and infected cells are seen in both primary and secondary lymphoid organs as well as in the brain.

In addition to CD4+ T cell loss many other mechanisms have been proposed for HIV pathogenesis (Sodora and Silvestri, 2010; Zhang and Su, 2012). These include chronic immune activation due to microbial translocation as a result of loss of gut integrity, loss of regulatory T cell (Tregs), activation of pDCs and induction of type I interferon among others. Activation of pDCs leading to production of type I IFN and other cytokines was demonstrated during HIV-1 infection of hu-HSC RG mice (Zhang et al., 2011). This positively correlated with immune activation and helper CD4+ T cell depletion. Studies in NSG-BLT mice also showed elevated levels of IFNα and immune activation (Long and Stoddart, 2012). Tregs (CD4+CD25+FoxP3+) are believed to play a central role in modulating the induction and suppression of immune activation (Zhang and Su, 2012). With the predilection of HIV-1 to CD4+ T cells including Tregs, paradoxical immunological consequences are predicted. For example, during HIV-1 infection Tregs can inhibit immune activation, which is beneficial. On the other hand, their suppression of T cell responses is detrimental by preventing elimination of infected cells. Hu-HSC RG mice were shown to harbor functional human Tregs and HIV-1 infection of these mice led to preferential infection and depletion of Tregs during the acute phase (Jiang et al., 2008). T cell exhaustion characterized by PD-1 upregulation during chronic HIV-1 infection is believed to play a significant role in reducing the function of HIV-1-specific T cells. Upregulation of PD-1 was observed in HIV-1 infected hu-Mice akin to that seen in humans, thus opening new avenues for therapeutic targeting of this inhibitory receptor using this system (Brainard et al., 2009). Administration of monoclonal antibodies to PD-L1, the ligand for PD-1, to chronically HIV-1 infected hu-HSC RG mice blocked its function and resulted in decreased viral loads and increased T cell levels giving credence to this therapeutic approach (Palmer et al., in press).

Primary HIV-1 infection is predominantly established by CCR5 tropic virus. During disease progression X4 viruses arise in majority of patients although their origin is not entirely clear. Hu-Mice that support long term HIV-1 infection permit evaluation of viral evolution in vivo. A study in hu-HSC RG mice evaluated this question by looking at the evolution of R5 tropic virus JR-CSF for 44 weeks and found that the mean rate of viral env gene evolution is similar to that seen in the human (Ince et al., 2010). It is well known that HIV-1 establishes latency early on during the acute phase of infection and that these latently infected cells serve as a viral reservoir even in the presence of HAART (Marsden and Zack, 2010). The mechanisms of viral latency are beginning to be understood. Latently infected cells are unrecognizable by the immune system and the silent provirus is impervious to ART. Cessation of HAART results in viral rebound. Current strategies are aimed at eliminating the latently infected cells to achieve a complete cure (Archin et al., 2012). Therefore an in vivo experimental system that mimics HIV latency in the human will be invaluable. Towards this goal, three reports have recently demonstrated HIV-1 latency in humanized mice (Choudhary et al., 2012; Denton et al., 2012; Marsden et al., 2012). Using the RG mouse system, Choudhary et al. found that HIV-1 viral loads could be suppressed to undetectable levels by ART and cessation of treatment led to viral rebound. After ex vivo stimulation, recoverable latent virus was found in resting CD4+ T cells of which the majority were of the central memory phenotype. Similar experiments using BLT mice also demonstrated the presence of latent virus in resting CD4+ T cells (Denton et al., 2012; Marsden et al., 2012). Ex vivo treatment with phorbol esters prostratin and 12-deoxyphorbal-13-phenylacetate induced HIV-1 outgrowth from these latently infected cells (Marsden et al., 2012). The ability to suppress viral loads to undetectable levels by HAART, viral rebound after cessation of treatment and the ability to induce the latent virus towards productive infection employing the newer Hu-Mouse models now provides an ideal in vivo experimental system to test future strategies of purging latently infected cells and in achieving a complete cure for HIV disease.

HIV-1 mucosal transmission and prevention

Primary infection via vaginal and rectal mucosal routes constitutes the predominant mode of viral transmission with HIV-1 (Denton and Garcia, 2011). With no effective HIV-1 vaccine on the horizon, alternative preventive methods such as topically applied microbicides or orally administered prophylactic drugs are practical alternatives (Heneine and Kashuba, 2012; Shattock and Rosenberg, 2012). Until recently the monkey-SIV/SHIV model has been the gold standard for studying comparative aspects of HIV mucosal transmission and prevention (Veazey et al., 2012). However, since it does not employ HIV-1 itself and is expensive, rapid progress has not been possible in these critical areas of HIV prevention. Furthermore, it is not possible to test promising HIV-specific compounds against the various HIV-1 strains and drug resistant viruses that occur in the field. Humanized mouse models can overcome these important limitations (Denton and Garcia, 2011). In this regard, the classical SCID-hu-PBL humanized mouse model was evaluated for early microbicide testing. However, due to inadequate mucosal tissue engraftment by human cells, the infection rate was found to be low and variable, and therefore the model was not consistently reliable (D’Cruz and Uckun, 2007). The newer hu-Mouse models have overcome these limitations (Denton and Garcia, 2012). Both hu-HSC RG and BLT mouse models have human cell engraftment with HIV-1 susceptible cells being present in the female reproductive tract (Berges et al., 2008; Denton et al., 2008). Efficient and consistent HIV-1 mucosal transmission via vaginal routes was achieved in both these models (Berges et al., 2008; Denton et al., 2008). Based on reports that employed hu-HSC mice, it became clear that HSC transplantation alone was adequate to achieve human cell mucosal engraftment and that human thymus transplantation (as done with BLT mice) is not necessary (Berges and Rowan, 2011). Human cell engraftment is also shown in the intestinal and rectal tracts in both the models (Sun et al., 2007; Berges et al., 2008; Denton and Garcia, 2011), with the exception of one report using hu-HSC RG mice (Hofer et al., 2008). This discrepancy was later ascertained to be due to not culturing the HSC in human cytokine media prior to transplantation as well as insufficient time allowed before evaluation (Berges and Rowan, 2011; Nischang et al., 2012). With regard to human cell engraftment in BLT mice, it was found that NSG mice were far superior compared to NOD-SCID mice with drastic differences seen in the rates of vaginal HIV-1 transmission (Stoddart et al., 2011). However for intestinal T cell reconstitution, a recent report determined that IL-2R γc is critical since NOD-SCID mice had better human cell reconstitution in the gut compared to NSG mice that lack this receptor (Denton et al., 2012). Mucosal transmission via rectal route was achieved in both the BLT and hu-HSC RG mouse models (Sun et al., 2007; Berges et al., 2008). Rectal abrasion was necessary in BLT mice prior to HIV-1 challenge to achieve infection, whereas this was not required in the case of hu-HSC RG mice. While mucosal infection could be established with both R5 and X4 HIV-1 viral strains, R5 virus was found to be more efficient, consistent with its predominant role in primary infections in the human. With these new advances in hu-Mice, the dynamics of HIV-1 transmission via mucosal routes, early virus-cell interactions and establishment of initial infected cell foci and their later systemic dissemination can be studied in vivo. An additional advantage with the hu-Mouse models over that of macaques is that there is no requirement for hormonal treatment to induce vaginal thinning prior to viral challenge to achieve consistent infection.

Currently oral pre-exposure prophylaxis (PrEP) with RT inhibitors tenofovir and emtricitabine has recently been FDA approved based on successful clinical trials (Steinbrook, 2012). Similarly, vaginally applied tenofovir microbicide gel was found to be effective with an overall 39% efficacy showing the promise of this approach (Abdool Karim et al., 2012). However, protection was not complete and a later clinical trial failed to show efficacy thus warranting detailed pharmacokinetic-phamacodynamic (PK-PD) analysis to derive critical data. Additionally, it is also clear that use of a single drug as a prophylactic will not be adequate given the propensity of HIV-1 for drug resistance. These criteria form a basis for conducting rapid experimental evaluations of numerous promising anti-HIV compounds and biological molecules for their prevention potential. Indeed recent studies in Hu-Mice have begun to yield promising preclinical data. It was found that oral administration of CCR5 inhibitor maraviroc and integrase inhibitor raltegravir can fully protect against HIV-1 vaginal challenge in hu-HSC RG mice (Neff et al., 2010). Topical vaginal application of maraviroc gel was also found to confer full protection against HIV-1 vaginal transmission (Neff et al., 2011). Extending this principle to antibodies (less likely to have side effects with prolonged use) for HIV prevention, the new generation broadly neutralizing antibodies (bNAb) to HIV-1 were also tested in this system (Veselinovic et al., 2012). It was found that anti-HIV bNAb VRC01 confers significant protection against HIV-1 vaginal infection. Pointing to the benefits of a combinatorial use of different antibodies, it was shown that a four bNAb combination confers full protection against vaginal challenge. BLT mice have also been widely used to evaluate the efficacy of various microbicide and oral PrEP strategies (Denton and Garcia, 2012). The compounds tested successfully in this system for microbicide potential include tenofovir, emtricitabine, C52L, C5A, PIE12-Trimer, and TC247 (Denton et al., 2011). It was also found that a combination of TDF and FTC given orally provided full protection against rectal viral challenge (Denton et al., 2010). Current major research questions in the HIV-PrEP field are what pharmacologically effective concentrations of ARTs need to be reached in the mucosal tissue for preventing HIV-1 sexual transmission and what combination of drugs will provide full protection. In this regard Hu-Mice are likely to play an ever increasing important role in the near future. While both the hu-HSC RG and BLT mice are shown to be equally amicable for testing microbicide and oral PrEP strategies against vaginal HIV-1 transmission, the latter received more attention until recently due to its early application. However it is worth noting that the hu-HSC RG model offers several advantages due to its simplicity of preparation since no surgery is involved, longer life span, lower cost and larger number of reconstituted animals that can be generated per human tissue.

Antiretroviral therapies for HIV-1

There are many advantages to evaluating anti-HIV therapies in the new hu-Mice (Van Duyne et al., 2009). First, the chronic nature of HIV-1 infection is replicated in these mice thus permitting long-term drug evaluation and identification of drug resistant mutants that arise in vivo. Additionally, restoration of CD4+ T cell levels resulting from effective treatment can be evaluated, since there is continuous de novo hematopoiesis in these mice unlike in early hu-Mouse models.

A number of studies evaluated combinations of ARTs to suppress viral loads. Choudhary et al. tested a combination of RT inhibitors emtricitabine and TDF together with a strand transfer inhibitor in hu-HSC RG mice (Choudhary et al., 2009). There was a drastic reduction in viral loads with a concomitant rise in CD4+ T cells. Treatment interruption resulted in viral rebound and renewed the loss of helper CD4+ T cells, similar to that seen in human infection. Furthermore, failure of HAART in some mice was determined to be due to the emergence of drug resistant viruses. In this study PK-PD aspects of these drugs were also successfully evaluated, establishing the feasibility of such studies in this system. Thus many aspects of HIV-ART can be studied in this system to derive important preclinical data on new candidate drugs. Later studies that showed more complete viral suppression in the context of identifying viral latency used more intensified drug regimens that included AZT, lamivudine, indinavir (Sango et al., 2010), AZT-Indinavir-didanosine (ddI) (Marsden and Zack, 2010) and Tenofovir-Emtricitabine and raltegravir (Denton et al., 2011). An emerging trend in HIV treatment is to develop long-acting drugs so the number of doses can be reduced. A recent report evaluated the efficacy of a combination of two long acting drug formulations, TMC278-LA and TMC-181-LA, together with other formulations of 3TC and TDF (Nischang et al., 2012). Emergence of drug resistant mutants was seen in insufficiently treated mice and viral rebound was noted in treatment interrupted mice, confirming the presence of a latent reservoir.

In addition to small molecule drugs, other studies investigated therapeutic strategies using novel molecules in hu-Mice. Van Duyne et al. showed the efficacy of tat peptide analogs in hu-HSC RG Mice (Van Duyne et al., 2008). SiRNAs targeted to either viral (tat, rev among others) or cellular (CCR5 or CXCR4) molecules have been shown to have potent anti-HIV activity in many in vitro studies (Kitchen et al., 2011). However, in vivo delivery presented challenges. A number of novel siRNA delivery strategies have recently been successfully evaluated in hu-Mice, including dendrimer formulated siRNA nanoparticles for improved cellular uptake (Zhou et al., 2011), single chain antibodies binding to CD7 surface molecule on lymphocytes (Kumar et al., 2008), and immunoliposomes targeted to LFA-1 integrin present on white blood cells (Kim et al., 2010). While found to be effective in HIV-1 inhibition, these strategies are not specific for drug delivery into HIV-1 infected cells. A novel approach exploited the specificity offered by a gp120 aptamer which can bind to and be internalized by HIV-1 infected cells (Neff et al., 2011). This aptamer was conjugated to an siRNA targeted to tat-rev transcripts, creating a dual-function molecule with viral neutralization and replication inhibition capacities. When HIV-1 viremic mice were treated with this chimeric construct, there was drastic suppression of viral loads demonstrating the efficacy of this strategy. Another set of useful molecules for HIV-1 protection and therapy are broadly neutralizing anti-HIV antibodies (Joseph et al., 2010). Their efficacies have been recently validated in the hu-Mouse system by deploying antibody producing cellular ‘backpacks’ (Luo et al., 2010) or utilizing adeno-associated virus vectors (Balazs et al., 2012).

Gene therapy strategies for HIV-1

HIV-1-induced disease is an excellent target for novel gene therapy approaches (Strayer et al., 2005; Rossi et al., 2007). Unlike acute viral infections that progress rapidly, HIV/AIDS is chronic in nature thus providing a long window of opportunity for genetic manipulation of virus susceptible cells and protective effector T cells. Moreover the virus susceptible cells are regenerated from a stem cell source in the bone marrow. The critical cellular and viral molecules necessary for viral entry and replication are known, and several methods and techniques currently exist to disable their function. For example, disrupting the viral co-receptors CCR5 and CXCR4 by various methods, such as siRNAs or zinc finger nucleases, can generate virus resistant cells not permissive to viral entry (Kitchen et al., 2011). For a long lasting therapy, genetic modification of hematopoietic stem cells holds the most promise since the modified cells are expected to generate HIV-1 resistant cells for life, providing effective immune reconstitution (Kiem et al., 2012). Early evidence for the success of such an approach came from the SCID-hu mouse studies in which retrovirally gene-transduced CD34+ HSC were differentiated successfully into mature T cells expressing the vector (Akkina et al., 1994). The new generation hu-Mouse models, due to their long-lasting multilineage hematopoietic cell engraftment, should prove to be far superior to test novel gene therapy strategies. Indeed, many new approaches have been successfully tested in these systems. Current methods center on inhibiting viral entry, harnessing host restriction factors, disabling essential viral regulatory molecules and fortifying host effector T cells for improved antiviral effects (Kitchen et al., 2011; Kiem et al., 2012).

With regard to viral entry, it has been known for some time that individuals lacking the CCR5 co-receptor due to the CCR5Δ32 mutation are relatively resistant to HIV-1 infection since nearly all primary infecting viral strains use this co-receptor. Therefore this is thought of as an attractive target for gene disruption (Anderson et al., 2007). This is further substantiated by the example of the so called ‘Berlin patient’ who has been cured from active HIV infection following a bone marrow transplant consisting of CCR5 null cells (Allers et al., 2011). Studies by Holt et al. have shown that disruption of the CCR5 gene in CD34+ HSC by zinc finger nucleases gave rise to HIV resistant cells in hu-HSC mice (Holt et al., 2010). Similarly, Shimuzu et al. reported the efficacy of siRNA targeting the CCR5 gene delivered via a lentiviral vector using the BLT mouse model (Shimizu et al., 2010). Other studies reported the efficacy of CXCR4 gene disruption in hu-Mice (Wilen et al., 2011; Yuan et al., 2012). Ter Brake et al. used an siRNA against HIV-1 nef gene delivered via lentiviral vector into CD34+ HSC for evaluation in hu-Mice (ter Brake et al., 2009). It was found that the differentiated siRNA transgenic cells showed significant protection against HIV-1 challenge ex vivo.

For long-term success of gene therapy it is necessary to employ more than one anti-HIV gene to prevent the generation of escape mutants. Using such an approach a combination of three different genes, namely a tat-rev siRNA, tar decoy and an anti-CCR5 ribozyme delivered via lentiviral vector into CD34+ cells was initially tested in vivo in hu-Mice (Anderson et al., 2007). Results from these hu-Mouse studies helped formulate a recently conducted human clinical trial (DiGiusto et al., 2010). Along these lines a combination of CCR5 siRNA, TAR decoy and a human/rhesus macaque TRIM5 alpha isoform was also successfully tested in hu-Mice (Walker et al., 2012). In a different approach, broadly neutralizing antibodies have also been tested. In the studies of Joseph et al., gene coding sequences for anti-HIV antibody 2G12 were transduced into CD34+ cells via a lentiviral vector in hu-Mice (Joseph et al., 2010). HIV-1 challenge resulted in much lower levels of viremia. Other novel strategies engineered T cells with HIV-1 gag specific TCR with promising results (Kitchen et al., 2012). As can be seen, humanized mice have permitted in vivo testing of many novel gene therapy strategies that would otherwise not be possible. Exploitation of these mice should provide additional important pre-clinical data in the future, such as addressing the number of dominant transgenic clonal cells that engraft, their persistence and expression levels of the vector in differentiated cells and finally the immune competence of the transgenic HIV-1 resistant T cells and macrophages.

HIV-1 related dementia

A high proportion of individuals with advanced immunodeficiency develop AIDS related dementia with associated mental disorders. Brain histopathology is characterized by macrophage infiltration, formation of microglial nodules and presence of multinucleated giant cells. These changes, which disrupt neuronal networks, are attributed to viral and infected cellular products. HIV-1 infected hNSG mice have been evaluated to model the CNS pathology (Gorantla et al., 2012). It was found that HIV-1 infection accelerated the entry of activated HLA-DR+ lymphocytes and macrophages into the brain. Viral spread to the brain was determined to be due to infected cell migration crossing the blood brain barrier. The pathology includes glial cell activation, meningitis and neuronal degeneration, with virus positive cells being found in the meninges and perivascular spaces. Prior CD8+ T cell depletion by antibody treatment in infected mice led to accelerated disease progression with development of meningitis and occasional meningoencephalitis, suggesting that cellular immunity has a role in modulating CNS disease. Additional evaluations by non-invasive longitudinal proton spectroscopic and diffusion tensor imaging, together with determination of peripheral viral loads and T cell counts, revealed a sequential development of neuronal abnormalities. Based on this data, hu-Mice appear to be a good model to systematically evaluate the CNS pathogenesis by HIV and to develop novel methods of treatment to prevent/control dementia. Recently in this context, the use of long acting nanoformulated ART given once a week elicited neuroprotective responses in HIV-1 infected hNSG mice, as evidenced by increases in microtubule-associated protein-2 synaptophysin and neurofilament expression (Dash et al., 2012). These findings indicate that therapies are possible and can be tested effectively using this system.

Dengue virus

Dengue (DEN) is a mosquito-borne viral disease affecting millions of people every year (Halstead, 2008; Simmons et al., 2012). The causative virus is a flavivirus with four antigenically distinct serotypes, DEN-1 through DEN-4. All dengue serotype viruses can infect humans with equal efficiency causing a range of illnesses spanning from subclinical infection to acute febrile dengue fever to the severe and often fatal vascular leakage conditions of dengue hemorrhagic fever/dengue shock syndrome (DHF/DSS). More commonly, the severe DHF/DSS forms of the disease are believed to be due to secondary infection of individuals who had previously recovered from a primary dengue infection with a different serotype virus. This is attributed to the antibody mediated immune enhancement phenomenon. There are currently no licensed vaccines or effective anti-viral treatments available for DENV, although many are in development. Until recently a major impediment to the understanding of dengue virus pathogenesis and immunity was the lack of an ideal animal model (Zompi and Harris, 2012). While primates can be infected by DENV they do not exhibit symptoms of disease. In addition they are expensive and use of large numbers of primates is prohibitive. Interferon-deficient mice with intact adaptive immunity have been useful in the study of some aspects of pathogenesis and protective immunity since they support virus replication followed by death. Similarly, NOD-SCID mice engrafted with human cord blood HSC were found to be susceptible to DENV infection and showed fever and thrombocytopenia, signs typical of dengue infection (Bente et al., 2005). However, a major deficiency of these human cell engrafted models is the lack of a human immune response, thus precluding any vaccine work or immunopathogenesis studies that address the antibody dependant enhancement of infection. The first thorough evaluation of human immune responses was conducted by Kuruvilla et al. using hu-HSC RG Mice (Kuruvilla et al., 2007). Dengue virus-infected hu-HSC RG mice showed viremia lasting up to 21 days with generation of both IgM and IgG antibodies. While all the mice produced IgM, only a subset produced IgG, consistent with inefficient antibody class switching. This was also the first study to demonstrate generation of neutralizing human antibodies against any human pathogen using hu-Mice. Later studies by Jaiswal et al. evaluated dengue infection and immune responses in HLA-A2 transgenic hNSG mice engrafted with matched cord blood. IgM antibody responses were seen in infected mice, and human T cells produced IFN-γ and TNF-α in response to stimulation with dengue peptides, thus indicating both cellular and humoral responses. The same group extended these studies by using NSG-BLT mice and found that IgM neutralizing antibodies were produced in addition to HLA-A2 restricted T cell responses, as measured by IFN-γ secretion following peptide stimulation (Jaiswal et al., 2012). Hu-HSC mice were also successfully used to evaluate dengue viral insect transmission and immune response (Cox et al., 2012). Using hu-HSC RG as well as BLT mice, we also were able to achieve insect dengue transmission and immune responses (Akkina et al., unpublished data). However, unlike in the study above both i/p route and insect-bite acquired infections resulted in antibody responses.

Epstein-Barr virus

EBV is a widely spread herpesvirus that infects more than 90% of the world’s population and causes persistent infection (Cohen, 2000; Young and Rickinson, 2004). While EBV is shown to infect cotton top tamarins in an experimental setting, humans are the primary host for this virus in nature. The medical importance is due to its role in causing acute mononucleosis, lymphoproliferative disease (LPD) and in the development of a variety of malignancies that include Burkitt’s Lymphoma, nasopharyngeal carcinoma, and Hodgkin’s lymphoma. Following primary infection, as is typical with all human herpesviruses, EBV establishes life-long latency. The immune system of the majority of individuals controls the infection via effector T cells, keeping the development of malignancies in check. During lytic infection more than 80 viral genes are expressed, while during malignancy only latency-associated viral proteins are expressed (Ramer et al., 2011). Eight latency associated EBV proteins are described: six nuclear antigens (EBNAs) and two membrane proteins, LMP 1 and 2. Expression of these proteins varies depending on the type of latency. Only EBNA1 is expressed in Burkitt’s lymphoma, a type I latency. In Hodgkin’s lymphoma and nasopharyngeal carcinoma, both type II latencies, either one or both LMP1 and 2 are expressed in addition to EBNA1. In cancers arising in immunocompromised patients such as those with HIV or transplant subjects, all the eight EBV latency proteins can be found, reflecting latency type III. Thus EBV infection can vary between individuals displaying complex pathogenesis. Use of hu-Mice for EBV studies permitted evaluation of the infection and immune response events (Ramer et al., 2011). Early studies using NOD-SCID mice engrafted with human HSC and injection of EBV established infection and LPD with expression of latency proteins suggestive of type III latency (Islas-Ohlmayer et al., 2004). However, since these mice do not produce adequate numbers of human T cells, their cellular responses could not be studied. The development of hu-HSC RG mice with better human cell engraftment and T cell generation helped model EBV infection and immune responses more effectively. Infection with EBV resulted in viral amplification as well as T cell responses with an increase in CD8+ T cells (Yajima et al., 2009). Confirming in vivo priming, T cells from infected mice cultured in vitro produced IFN-γ in response to stimulation with autologous lymphoblastoid cell lines. Later studies using hNSG and BLT mice produced similar results. EBV infection of hNSG mice follows a dose dependant effect (Ramer et al., 2011). Low dose infection gave rise to persistent asymptomatic infection and detectable T cell responses whereas intermediate and high dose viral challenges gave rise to more vigorous T cell expansion. Higher doses viral challenge led to tumor generation. Expanded T cells were of memory T cell phenotype with protective capacity since depletion of T cells prior to infection led to more rapid and disseminated LPD. Based on the utility of this model for infection and immune response, more advanced studies using EBV viral mutants were conducted using hNSG mice. Infection with EBNA3B-knockout EBV mutant virus showed more aggressive proliferation of B cells, generating lymphomas resembling diffuse large B cell lymphomas (DLBCL) (White et al., 2012). It was also found that while infection with mutant EBV induced expansion of T cells, they failed to infiltrate the tumor tissues unlike a wild type virus infection. This was reasoned to be due to EBNA3B gene being a tumor suppressor and also playing a role as a factor responsible for secretion of T cell attracting chemokine CXCL10 (IP10).

Hepatitis C virus

Hepatitis C virus is another major human specific pathogen with a global public health concern with over 175 million people currently infected. While a minority of infected individuals (10–20%) spontaneously clear the virus due to an effective immune response, chronic infection generally ensues in 80% of patients often leading to hepatitis, liver fibrosis, cirrhosis and the eventual development of hepatocellular carcinoma in a subset of patients. Chronic HCV infection is associated with impaired CD4+ and CD8+ T cell functions, and therefore detailed studies of immunopathogenesis are critical to fully understand disease progression so preventive and therapeutic approaches can be developed. While chimpanzees played an important role in HCV studies, in addition to their cost and endangered status, other limitations included low chronic infection rate and the absence of hepatic fibrosis. Therefore, a cost-effective small animal model that permits HCV infection and immune responses would greatly aid future progress (Washburn et al., 2011; Dorner et al., in press). In this regard, hu-Mice that harbor functional human hepatocytes could fill this void. Indeed during the last decade a number of murine–human hepatocyte chimeric models have been developed. These include Alb-uPA-SCID mice and Fah-RG mice that permit human hepatocyte engraftment (Meuleman et al., 2005; He et al., 2010). While productive HCV infection with associated pathogenesis could be seen in these mice, lack of a functioning human immune system precluded studies on immunopathogenesis which plays a central role in progressive disease in chronically infected individuals. This setback was recently rectified by the generation of immunocompetent humanized mice that also harbor a functional human hepatic system. This was achieved with mice on a RG background by generating transgenic mice (AFC8 mice) that express active Caspase 8 fused to FK506 binding domain (FKBP) with inducible suicidal activity in murine hepatocytes under the control of the albumin promoter. Human hepatocyte progenitor cells, together with human HSC, were co-transplanted into these mice which were then treated with FKBP dimeriser. Destruction of native murine hepatocytes post-treatment allowed selective outgrowth of human hepatocytes, while human HSC engraftment permitted the development of human immune cell subsets giving rise to a AFC8-hu HSC/Hep mouse (simply termed as Hu-Liver-HSC mouse here) (Washburn et al., 2011). HCV challenge of these mice supported productive viral infection in engrafted human hepatocytes and generated viral specific T cell responses. Hepatitis and fibrotic lesions were seen correlating with activation of stellate cells and human fibrogenic gene expression. Whereas HCV viremia could be seen in the previous uPA and Fah mouse models (supporting > 50% hepatocyte engraftment) it was not the case in this system due to lower levels of (~15%) hepatocyte engraftment thus suggesting further improvements are necessary. Nevertheless, the new chimeric APC8-hu HSC/Hep mouse system with a capacity to support productive infection and immune response is an important step forward in this area.

Advantages and disadvantages of different hu-Mouse models

Based on the experimental needs and viruses used there are advantages and disadvantages with the currently used new hu-HSC and BLT mouse models. From a practical standpoint, the hu-HSC model is relatively easy to create since only a quick, non-surgical intrahepatic injection of HSC is needed. Human HSC from easily procurable sources such as cord blood can be used and larger cohorts of mice from a single donor can be made. Therefore hu-HSC mice can be employed for large scale drug evaluations and PrEP studies without much difficulty. A disadvantage with this model however is the lack of proper human T cell restriction due to the absence of a human thymus. Co-expression of HLA class I and II genes by transgenesis will overcome this deficiency. A particular advantage with the BLT mouse model is the presence of transplanted human thymus, permitting proper T cell education and restriction and facilitating T/B cell cooperation. Thus these mice are preferable for vaccine and immunity studies. Disadvantages include the need for complicated surgery to implant human fetal tissues under the kidney capsule, and that the number of mice that can be generated from a single fetal donor tissue is limited. Scale-up of their production for large scale experimentation and testing also poses challenges due to the requirement for fetal tissues that are often difficult to procure in sufficient quantities.

Current limitations, advances and future prospects

Current human immunocompetent hu-Mouse models have come a long way since the original description of human–mouse chimeras. However, there are still several limitations that need to be overcome (Shultz et al., 2011). These include residual innate immunity in the immunodeficient mouse strains requiring irradiation and prior conditioning, less than ideal T cell numbers and sub-optimal maturation of B cells, absence of human HLA class I and class II restriction in hu-HSC mouse models and lack of adequate levels of HLA APCs in the BLT mouse model, deficiencies in T and B cell cooperation resulting in low levels of antibody responses and inefficient immunoglobulin class switch, and poorly cross-reactive native murine growth factors and cytokines.

These deficiencies are currently being addressed and rectified. In addition to T cells, B cell and NK cells, mouse macrophages are also found to contribute to xenograft rejection in hu-Mice. Mouse macrophages expressing native SIRPα receptor clear out xenografted human cells that do not express the cognate ligand CD47 (‘don‘t eat me’ marker of self) (Ide et al., 2007; van den Berg and van der Schoot, 2008). Supporting their role in graft rejection, it has been recently shown that human SIRPα transgenic mice exhibit improved human cell reconstitution (Strowig et al., 2011). In a reverse approach, human HSC stably transduced with murine CD47 ligand by lentiviral vectors also showed increased engraftment (Legrand et al., 2011). Rectifying CD47-SIRPα interactions in hu-Mice will lead to more robust and lasting human cell engraftment. As pointed out above, transgenic expression of human HLA-DR4 also led to increased T cell numbers in reconstituted mice. Thus co-expression of HLA class I and Class II in doubly transgenic mice is likely to further improve human T cell reconstitution, T/B cell cooperation/help and mediate HLA restricted immune responses (Garcia and Freitas, 2012). Instead of standard transgenesis, the knock-in of the human HLA Class I and Class II genes in mouse MHC loci will also preclude the unwanted mouse H-2 restricted human cell immune responses. Since many murine cytokines and growth factors are poorly cross-reactive with their human counterparts thus resulting in sub-optimal human cell development and maintenance, supplying these in trans either by injection or by transgenesis can overcome these deficiencies. Among these cytokines are GM-CSF, IL-4, M-CSF for monocyte/macrophage, IL-7 for T cells, IL-15 for NK cells and EPO for erythrocytes (Willinger et al., 2011). Knock-in replacement of mouse cytokine genes with their human equivalents in appropriate loci provides an additional advantage of their constitutive expression operated by the mouse regulatory elements. Along these lines, expression of human thrombopoietin resulted in higher human cell engraftment and better HSC maintenance (Willinger et al., 2011) whereas transgenic mice with human IL-3, GM-CSF (Willinger et al., 2011) and M-CSF (Rongvaux et al., 2011) knock-in genes exhibited improved myeloid differentiation and function, thus demonstrating the benefits of these strategies. Clearly, ongoing intensive work has identified several areas for improving the existing human immunocompetent hu-Mouse models. However, these different strategies addressing cytokines, growth factors and HLA molecules need to converge together to derive a composite recipient mouse strain incorporating all the desirable attributes. Given the recent accelerated progress, it is not too far in the future that superior hu-Mouse models will become available.