The major objective of this study was to evaluate humanized NOD-scid IL2rγnull mice as a small animal model for the study of human DENV infection. Using a low dose of a laboratory strain of DENV, our results indicate that NOD-scid IL2rγnull mice engrafted with human hematopoietic stem cells are permissive to DENV infection and generate DENV-specific human immune responses. We identified human CD45+ cells that were targets of DENV infection following in vitro and in vivo infection. To our knowledge, we are the first to demonstrate infection of human cells and assess the generation of DENV-specific human T cell responses in a humanized mouse model.
Our in vitro studies with clinical and laboratory passaged DENV strains demonstrate that human CD45+
cells in the bone marrow of engrafted NOD-scid IL2rγnull
mice are differentially infected with multiple serotype-2 DENV strains. We and others have observed varying degrees of infectivity following in vitro infection of human primary cells and cell lines with clinical and laboratory passaged strains of DENV-2 
. The vast majority of infected cells detected were human CD45+
cells and preliminary data suggest that CD11c+
dendritic cell precursors in the bone marrow of NOD-scid IL2rγnull
mice may harbor DENV antigen (data not shown). Although some studies have identified B cells as an important target for DENV infection in humans, most data have pointed to infection of monocytes and possibly dendritic cells in vivo 
. Further studies using different serotypes and strains of DENV are required to identify predominant targets of infection after in vitro infection.
Bente and Mota et al recently reported that NOD-scid
mice engrafted with human CD34+
stem cells were susceptible to infection with a low passage clinical isolate of DENV 
. Infected mice showed typical signs of dengue infection, including fever, rash, weight loss and thrombocytopenia. We did not consistently see any of these signs in our model using the laboratory passaged DENV-2 NGC, although several infected mice did show significant weight loss and had ruffled fur. We performed limited studies with the same Thai isolate of DENV-2 used by Bente et al and obtained similar results to those seen with DENV-2 NGC (data not shown). A possible explanation for differences between our studies and others may include varying strains of humanized mice used for the experiments. The NOD-scid
mice used by Bente et al do not generate human T or B cell responses which may have influenced the ability of DENV to replicate in target tissues. Since cord blood samples from several different donors were used for reconstitution it may account for differing susceptibilities of mature human cells to DENV infection in NOD-scid IL2rγnull
mice, which could contribute to the lack of consistent findings of dengue disease in our mice.
We performed most of our experiments using a single immunization with a prototype laboratory adapted strain DENV-2 NGC. The NGC strain of DENV was originally serially passaged in mouse brains and later extensively passaged in cell culture for use by many laboratories 
. We have not detected any neurological symptoms in engrafted mice after infection with DENV-2 NGC. Additionally, no virus was detected in any organs of infected non-engrafted mice suggesting that the engrafted human cells were necessary for in vivo replication of this mouse adapted neurovirulent strain of virus and that the symptoms of dengue disease that we have observed in humanized mice is not likely due to replication in neuronal sites.
Detection of DENV RNA at multiple late time points post infection in organs known to be targets of infection in humans is encouraging and suggests that DENV-2 replicates within human cells of infected mice. Detection of the NS1 protein in predominantly hCD45+ cells using the mAb 7E11 suggests that the viral RNA detected by RT-PCR is due to replication of DENV within cells in humanized NOD-scid IL2rγnull mice, consistent with productive infection. Confirmation of the ability of infectious virus in liver lysates from infected mice to infect C6/36 cells further validates our PCR and flow cytometry results. Further studies will be needed to compare DENV tropism in this mouse model to those in humans.
Similar to results by Kuruvilla et al, we have detected virus in the sera (data not shown) and livers of mice that persisted for >20 days in our initial studies in a subset of NOD-scid IL2rγnull mice. Mice with high viremia lost weight and appeared sick while mice with low level or intermittent viremia had less appreciable weight loss. We are still unclear why some mice harbor DENV several weeks after infection. We speculate that the low IgG responses and inefficient human T cells generated in NOD-scid IL2rγnull mice may contribute to inefficient viral clearance in this model. In our more recent studies with NOD-scid IL2rγnull Tg(HLA-A2/Huβ2M) mice however, we have not detected any virus in the sera of mice at 2 or more weeks post infection. The results with humanized A2 transgenic NOD-scid IL2rγnull mice are highly encouraging and must be considered when evaluating varying humanized models as suitable animal models to assess dengue infection and immunity since infectious virus is cleared in approximately 2 weeks after natural infection in humans.
We detected DENV-2-specific IgM antibodies in the sera of mice 5–14 days after infection indicating that these mice are able to generate antibody responses against DENV. However, as reported by others 
, we detected little or no DENV-specific IgG (data not shown) suggesting that very low levels of class switching occurred in these mice. Variable and low IgG responses in these mice might be due to a lack of species cross-reactive cytokines in the xenogenic environment 
. We treated a small group of mice with B Lymphocyte Stimulatory Factor (BLyS), a factor known to promote human B cell survival 
and B cell engraftment in NOD-scid IL2rγnull
mice (unpublished observations). Sera from BLyS-treated DENV-infected mice did not show substantially higher DENV-specific IgM or IgG responses (data not shown). Kuruvilla et al demonstrated high levels of DENV-specific IgM and IgG in stem cell engrafted RAG-hu mice several weeks post infection 
. However, only RAG-hu mice that were immunized with a pool of four different strains of DENV-2 generated a strong antibody response while mice infected with individual strains of DENV-2 had very low levels of dengue-specific antibodies. The basis for the difference in response to the mixture of viruses was not determined. Antibody responses in our model might be enhanced by using a pool of different DENV-2 strains as done by Kuruvilla et al or by increasing the viral inoculum. However neither strategy mimics the natural course of DENV infection.
One major difference between our study and previous studies of DENV infection in humanized mouse models is our demonstration of IFN-γ production by human T cells in response to DENV-2 antigen and virus-specific peptides. Most of the mice immunized by either the i.p. or s.c. route had significant DENV-specific IFN-γ responses to the inactivated antigen with low intensity staining at 1–2 weeks post infection. Several studies have reported the important role of IFN-γ in the clearance of DENV infection 
. It remains unclear how the human T cells generated in NOD-scid IL2rγnull
mice are effectively educated. While mature human T cells could develop by extrathymic education and selection 
, presumably most of their education occurs in the murine thymus. Melkus et al were able to overcome this problem by using NOD-scid
mice implanted with human fetal liver and thymic tissues to measure human EBV-specific immune responses 
Our data using NOD-scid IL2rγnull
Tg(HLA–A2/Huβ2M) mice are very encouraging and indicate that expression of human HLA molecules allow for better education of dengue-specific human T cells in humanized mice. As far as we know, we are the first to demonstrate A2 restricted peptide-specific responses in humanized mice. The ability of T cells to respond to ex vivo peptide stimulation by secreting multiple cytokines including IFN-γ, TNF-α and IL-2 demonstrate the polyfunctionality of T cells in NOD-scid IL2rγnull
Tg(HLA-A2/Huβ2M) mice. The frequency of cells that respond to the 3 A2 restricted peptides by secreting IFN-γ in splenocytes of humanized mice (0.1–2.8% of total CD3+
T cells) is in line with frequencies detected in human PBMC of DENV immune donors (0.1–0.68% of total CD3+
T cells) 
. These preliminary studies suggest that NOD-scid IL2rγnull
Tg(HLA-A2/Huβ2M)mice can be used effectively to assess T cell responses to virus-specific peptides identified in humans. Further experiments will need to be performed to assess the frequency and function of dengue-specific T cells during acute infection, in memory as well as following sequential heterologous dengue virus infections.
There are still some limitations in immunodeficient IL2rγnull
humanized mouse models including the requirement of human specific molecules for optimal function of the human immune system and issues of remaining innate immunity that present obstacles to human hematopoietic stem cell engraftment 
. In addition, optimal conditions to ensure reliable human T and B cell engraftment will likely involve for example the use of neonatal rather than adult mice for engraftment and transgenic expression of select cytokines. Overcoming these limitations should be achievable in the future and will allow investigators to better recapitulate immune responses observed in immunocompetent individuals in humanized mouse models.
In summary, we have characterized a humanized mouse model with functional DENV-specific adaptive immune responses. The virologic and immunologic data presented set the stage for investigating both host and virus-specific mechanisms that control primary and sequential DENV infections. Since prior immunity is a major risk factor to developing DHF, these mice could potentially be used to study the role of cross-reactive sub-neutralizing antibodies and T cells during sequential DENV infections. Understanding the contribution of host components to severe dengue disease will lead to the development of effective therapeutics and vaccines.