Summary of presentations at the NIH/NIAID New Humanized Rodent Models 2007 Workshop

It has long been recognized that a small animal model susceptible to HIV-1 infection with a functional immune system would be extremely useful in the study of HIV/AIDS pathogenesis and for the evaluation of vaccine and therapeutic strategies to combat this disease. By early 2007, a number of reports on various rodent models capable of being infected by and responding to HIV including some with a humanized immune system were published. The New Humanized Rodent Model Workshop, organized by the Division of AIDS (DAIDS), National Institute Allergy and Infection Diseases (NIAID), NIH, was held on September 24, 2007 at Bethesda for the purpose of bringing together key model developers and potential users. This report provides a synopsis of the presentations that discusses the current status of development and use of rodent models to evaluate the pathogenesis of HIV infection and to assess the efficacy of vaccine and therapeutic strategies including microbicides to prevent and/or treat HIVinfection.


Introduction
Investigation of many aspects of the in vivo behavior of HIV as well as testing of the in vivo efficacy of novel anti-HIV therapies and vaccines has been hampered by the restriction of HIV infection to humans and primates [1]. Mice cannot be infected with HIV-1, because sequence differences in mouse homologues of the human proteins required for HIV replication prevent their interaction with essential HIV proteins critical for HIV replication such as Env, Tat [2,3] and Rev [3,4], as well as prevent and potentially limit efficient assembly and budding of virus from the cell membrane. These genetic differences result in blocks at several stages of HIV replication that prevents cellular infection and efficient production of HIV-1 by mouse cells.
It has long been recognized that a small animal model with a reconstituted human immune system would be extremely useful in the study of HIV/AIDS pathogenesis and for the evaluation of vaccine and therapeutic strategies to combat this disease. By early 2007, a number of reports on rodent models with a humanized immune system capable of being infected by and responding to HIV were published. The New Humanized Rodent Model Workshop, organized by Janet Young, Paul Black, Tony Conley, Jim Turpin, Fulvia Veronese and Opendra Sharma from DAIDS, NIAID, NIH, was held on September 24, 2007 at Bethesda for the purpose of bringing together key model developers and potential users. The meeting included a discussion by a panel about the current status of the models, future plans, as well as potential use of the models for addressing critical issues in basic immune response studies, pathogenesis, therapeutics, vaccines and microbicides development. Speakers were asked to address the following questions:

HLA restriction and antibody responses
Are there human HLA-restricted CD4 or CD8 responses? Are there antigen-specific human antibody responses and how do antibody titers compare with responses in humans?
Two broad approaches have been used to circumvent the replication blocks in rodents to generate small animal models for studying HIV infection. One broad approach used transgenic techniques to generate mice or rats capable of supporting HIV replication either by introducing transgenes encoding the human proteins critical for HIV replication or an HIV provirus into the genome of rodents. At the workshop, Drs. Littman, Keppler and Goldstein discussed these approaches. A second increasingly adopted approach utilizes chimeric human/mouse models to circumvent the inability of mouse cells to support HIV replication by transplanting human hematopoietic cells and/or human thymic tissues and/or fetal liver into immunodeficient mice. A model initially described by the McCune group designated either the SCID-hu mouse [5] or the thy-liv SCID mouse was constructed by surgically implanting fragments of human fetal thymus and liver under the kidney capsule of a SCID mouse. Two to three months after implantation, a thymus-like conjoint human organ grows which supports long-term multi-lineage hematopoiesis that leads to maturation of human thymocytes [6]. If sufficient thymic tissue is implanted, human T-cells are found in the peripheral blood for over a year, but no mature B cells are generated [7,8]. Injection of HIV-1 into the implant results in the killing of human thymocytes and the severe depletion of human CD4+ cells in the implant within a few weeks as well as plasma viremia. A limitation of this model is that no humoral or cellular responses to the HIV infection, including primary immune responses, occur in these chimeric mice [6,9]. This model is also limited by its construction using implanted tissues that are of fetal origin whose response to infection may not necessarily reflect the course of HV infection in patients where HIV predominantly infects lymph nodes and the gut associated lymphoid tissues. In a presentation at the workshop Dr. Stoddart discussed the current status and uses of this model. The chimeric human/mouse approach has been expanded by the recent description of new models that take advantage of novel mouse strains, Rag2 -/-γ c -/mice and NOD/SCID/IL2Rγ null mice, that are more immunodeficient than SCID mice and support engraftment and maturation of human hematopoietic stem cells into human T cells, B cells, monocytes and dendritic cells after injection with human CD34+ hematopoietic stem cells [10][11][12][13][14][15][16].
Drs. Akkina, Luban, Su, and Speck discussed their experiences with models that use Rag2 -/-γ c -/mice and Dr. Shultz has included his experience with a model that uses NOD/ SCID/IL2Rγ null mice. Another chimeric human/mouse model was discussed by Dr. Martinez that combines human thymic implantation and transplantation with human HSC by implanting NOD/SCID mice with fetal liver and thymus and then transplanting them with syngeneic human CD34+ hematopoietic stem cells [17]. A synopsis of these presentations and Tables summarizing the techniques used to construct these models, the major features of the different models and the specific experiences of individual investigators using these models for HIV-related studies are presented below.

Transgenic rodent models
Dr. Littman (Howard Hughes Research Institute, New York University Medical Center) has been working on utilizing transgenic approaches to overcome the replication barriers that prevent HIV infection of murine T cells (Table 1). These include the inability of HIV to enter mouse cells, the subsequent inefficient support of Tatmediated trans-activation, the aberrant processing of HIV-Gag protein and the defective virion budding in mouse cells. To overcome the entry block, Dr. Littman used transcriptional regulatory sequences from mouse and human CD4 genes to construct transgenic mice expressing human CD4 and CCR5 in mouse CD4+ T cells, myeloid cells, dendritic cells and microglia. Using a Vpr-β-lactamase assay, his group demonstrated that HIV could efficiently enter into activated CD4+ T cells from these hCD4/CCR5 transgenic mice. After entry, he demonstrated that RT was functional in primary mouse T cells as indicated by the efficient generation of nuclear 2-LTR circles. The block due to inefficient Tat-mediated trans-activation is related to structural differences between mouse and human cyclin T1 (hCyclin T1), a protein which is required for Tat function and efficient HIV replication. A single amino acid difference at position 261 in mouse cyclin T1 (mCyclin T1) compared to hCyclin T1 prevents mCyclin T1 from binding of HIV Tat. The Littman group constructed mice transgenic for hCyclin T1 under the control of the CD4 promoter and crossed them with hCD4/CCR5 mice. Although expression of hCyclin T1 was associated with a several-fold increase in the production of HIV by mouse cells also expressing CD4 and CCR5, HIV RNA levels in the infected hCycT1 mouse T cells were still 10-fold lower than human cells. Efficient infection of mouse T cells required continued activation of the TCR with anti-CD3/ CD28, particularly for the 12-20 hour period after infection. HIV production by mouse cells is also limited by a processing defect in the conversion of the gag p55 precursor to p24, leading to decreased production of p24 antigen which is required for construction of the viral capsids. Furthermore, the HIV produced by the mouse cells was less infectious than HIV produced by human cells. Electron microscopy demonstrated the abnormal budding of HIV in infected mouse T cells into the nuclear envelope and not the cell membrane. Murine Apobec3 cannot interact with HIV Vif, and hence can also inhibit HIV production by mouse cells, but this has not yet been fully assessed in murine T cells. This transgenic mouse model therefore does not support sufficient levels of HIV replication for pathogenesis, drug or vaccine studies.
These transgenic mice were developed to also study the pathogenesis of HIV-1 infection. Early in the course of HIV infection, CCR5+CD4+ T cells are depleted from the lamina propria in HIV-infected individuals which is not reversed despite treatment with HAART. Critical questions that need to be addressed are the mechanism for this selective depletion of mucosal CD4 T cells, what interventions can reverse this depletion, and the role of dendritic cells and TH17 cells in this process. Development of a mouse model infectible with HIV would greatly support investigation of these critical questions. Future studies of the Littman group will focus on identifying barriers to gag processing in mouse cells, how to regenerate the CD4+ T cell population in the mucosal associated lymphoid tissues (MALT) after HIV infection and the role of dendritic cells in HIV infection of MALT.
Dr. Goldstein discussed an alternative approach used in his laboratory to construct mice that are transgenic for a provirus encoding a full length primary R5-tropic isolate, HIV-1 JR-CSF , capable of producing HIV proteins and infectious virus, JR-CSF mice (Table 1) [18]. To circumvent the restricted trans-activating function of the Tat protein in mice and to specifically target HIV replication to CD4expressing cells, Dr. Goldstein crossed the JR-CSF mice with transgenic mice that carry a transgene of hu-CycT1 under the control of the CD4 promoter and express hu-CycT1 in CD4 T cells, monocytes/macrophage dendritic cells and microglia to yield JR-CSF/hu-CycT1 mice [19]. As a consequence of being able to support Tat-mediated transactivation in CD4-expressing cells, HIV production is markedly increased in the JR-CSF/hu-CycT1 mouse CD4 T cells, monocytes and microglia. Stimulated JR-CSF/hu-CycT1 mouse CD4 T cells produced between 1-to 10% of the quantity of HIV produced by activated JR-CSF/hu-cycT1 mouse monocytes, indicating that mouse T cells have a specific block in post-HIV replication that is absent in mouse monocytes. While the population of peripheral CD4 T lymphocytes in the peripheral blood of JR-CSF mice remained stable over time, the peripheral CD4 T cells population in the JR-CSF/hu-cycT1 mice became gradually depleted so that by one year of age the CD4 to CD8 T cell ratio in the peripheral blood of the JR-CSF/hu-cycT1 mice had reversed to less than one, similar to the temporal course in HIV infected individuals that develop AIDS. In addition to being useful for studying the pathogenesis of HIV-mediated depletion of CD4 T lymphocytes in lymphoid tissues, these mice can also be used to investigate the in vivo effects of HIV infection on other organs, including the brain. The Goldstein group demonstrated that microglia and astrocytes from the JR-CSF/hu-cycT1 mice are more sensitive to in vivo activation by inflammatory stimuli such as LPS than are microglia from JR-CSF mice or wild-type littermates that is manifested by more extensive phenotypic changes and increased production of chemokines including of MCP-1. These mice provide a useful model for investigating the direct and indirect longterm effects of HIV-infection on cellular and organ function.
Dr. Keppler is pursuing the goal of humanizing rats to generate an immunocompetent multi-transgenic rat model of HIV-1 infection (Table 1). While cells from native rodents do not or only inefficiently support distinct steps of the HIV replication cycle, rats appear to be intrinsically more permissive than mice for supporting HIV replication. Of conceptual importance, the barriers to HIV replication in rat cells identified thus far appear to result from the inability of individual rat proteins to support HIV-1 replication rather than from the action of speciesspecific restriction factors. To circumvent these barriers, the Keppler group has pursued a block-by-block approach to humanize Sprague Dawley rats by the introduction of human transgenes that encode proteins that are required to overcome these barriers. Transgenic rats that express the HIV receptor complex hCD4 and hCCR5 on CD4 T-cells, macrophages and microglia (hCD4/hCCR5 rats) can be infected systemically with HIV [20,21]. Following intravenous challenge with HIV-1, lymphatic organs from hCD4/ hCCR5 rats contained HIV cDNAs and early viral proteins, demonstrating successful in vivo infection. Furthermore, hCD4/hCCR5 rats infected with HIV YU2 displayed lowlevel plasma viremia (~150 copies/ml) for up to 7 weeks post-challenge as well as episomal HIV cDNA species in splenocytes and thymocytes 6 months post-infection. A recent proof-of-principle study showed the suitability of these double-transgenic animals for the rapid preclinical evaluation of the inhibitory potency and of pharmacokinetic properties of antiviral drugs targeting HIV entry or reverse transcription [22]. Prophylactic administration of Sustiva (efavirenz) or Fuzeon (enfuvirtide, T20) markedly inhibited the level of HIV infection measured several days after in vivo challenge with HIV. Additional novel drugs, including an integrase inhibitor, are currently being tested. In contrast, administration of a semen-derived fibril-forming peptide that has been shown by the Kirchhoff group to promote in vitro HIV infection increased the splenic HIV cDNA load in hCD4/hCCR5 rats after in vivo HIV challenge by 4.5 fold [23]. In their attempts to further enhance the HIV susceptibility of transgenic rats, the limited support of HIV replication at the transcriptional level that leads to reduced early HIV gene expression in rat Tcells was largely surmounted by the transgenic expression of a third human transgene, the Tat-interacting protein hCyclin T1, a component of the P-TEFb transcription complex [20]. T-cells from triple-transgenic rats produced 3-fold higher levels of HIV early gene products than rats transgenic only for hCD4 and hCCR5. However, robust replication is still precluded, most probably due to a disproportional representation of Rev-dependent HIV RNAs and viral proteins. The current work of the Keppler group focuses on the identification of a relevant factor that may overcome this third and possibly final barrier to HIV replication in primary target cells in rats. As a complementary approach, the Keppler group is pursuing strategies to adapt HIV to replicate in primary T-cells from transgenic rats.

Chimeric human/mouse models
The generation of humanized mice for HIV research has benefited from a progression of genetic modifications made possible by the occurrence of spontaneous immunological mutations, the targeting of genes required for the development of innate and adaptive immunity, and the availability of inbred mouse strains exhibiting depressed innate immunity. The first widely used model for human hematolymphoid engraftment and subse-  quent HIV infection was the CB17-Prkdc scid (abbreviated as scid) mouse. CB17-scid mice supported engraftment with human (HSC), peripheral blood mononuclear cells (PBMC), and human fetal tissues. However, levels of engraftment were limited by many factors including host natural killer (NK) cell activity, spontaneous generation of mouse lymphocytes (leakiness), and the occurrence of spontaneous thymic lymphomas [24]. The subsequent development of the NOD-scid mouse stock exhibiting depressed NK cell activity resulted in heightened support of human hematolymphoid engraftment [24,25].

Investigation of HIV Pathogenesis
Humanized mice were incrementally improved over the next decade by the targeting of genes at a number of loci including the recombination activating genes 1 and 2 (Rag1 and Rag2) and the beta 2 microglobulin (B2m) locus [26]. Mutations at the Rag-1 and Rag2 loci prevent development of mature mouse lymphoid cells but do not reduce NK cell activity. The B2m mutation prevents NK cell development. Although NOD-scid B2m null mice lack NK cell activity, a shortened lifespan due to early occurrence of thymic lymphomas and other pathologic changes limited the use of this model in HIV research [27].
A major advance in development of humanized mice was made possible by the targeting of the gene encoding the interleukin-2 receptor common gamma chain (Ilrg), abbreviated as IL2Rγ. The IL2Rγ chain is indispensable for IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21 high affinity binding and signaling [28]. The IL2Rγ mutation prevents NK cell development and causes other defects in innate immunity as well as depressed adaptive immunity. In humans, IL2Rγ deficiency causes X-linked SCID [29]. Four different groups independently targeted the mouse IL2Rγ gene [30][31][32][33]. Genetic crosses of IL2Rγ null mice with scid, Rag1 null and Rag2 null mice on a several different mouse strain backgrounds resulted in a number of new immunodeficient models that support engraftment with human HSC, PBMC, and fetal tissues [26].

NOD/SCID/IL2Rγ null mouse model
Dr. Shultz and his colleagues, Drs. Dale Greiner and Fumihiko Ishikawa generated mice chimeric for the human hematopoietic system using one of the most widely available of the IL2Rγ deficient mouse stocks, the NOD-scid Il2rg tm1wjl (NOD-scid IL2Rγ null ) model (Table 2). These mice lack mature lymphocytes and NK cells, survive beyond 16 months of age, and do not develop lymphomas [34]. The Shultz group demonstrated that newborn [12] and adult [34] NOD-scid IL2Rγ null mice support high levels of engraftment with human umbilical cord blood (UCB) HSC and mobilized HSC. The human HSC engrafted mice develop mature human lymphoid and myeloid cells and mount a humoral immune response to thymic-dependent antigens [11,12]. Engraftment of NOD-scid IL2Rγ null mice with either human committed lymphoid or myeloid progenitor cells isolated from human UCB results in development of both human conventional and plasmacytoid dendritic cells [35]. Adult NOD-scid IL2Rγ null mice also support heightened engraftment with human PBMC following intravenous, intraperitoneal, or intrasplenic injection [36]. Current ongoing genetic modifications of the NOD-scid IL2Rγ null model in Dr. Shultz's lab include further reductions of innate immunity as well as transgenic expression of human HLA molecules, cytokines, and other components needed to optimize human hematolympoid engraftment and function.

RAG2 -/-γ c -/mouse models
Dr. Akkina generated mouse-human chimeric mice using fresh CD34+ HSC isolated from human fetal liver cultured with cytokines for 1 day at a low density and injected hepatically (250,000 CD34+ cells/mouse) into RAG2 -/-γ c -/-Balb/c mice obtained from Dr. Irving Weissman (RAG-KO mice) ( Table 3) [10]. Within the first 3 days of life, neonatal RAG-KO mice are sublethally irradiated and injected intrahepatically with CD34+ human hematopoietic stem cells to yield RAG-hu mice. After 8-12 weeks, the peripheral blood of the RAG-hu mice is populated with human T cells (CD4+, CD8+), B cells, dendritic cells, and macrophages. The fraction of CD45+ leukocytes detected in the peripheral blood of the RAG-hu mice of human origin was between is 5-80% and the Akkina group now routinely generate RAG-hu mice where the fraction of human peripheral blood lymphocytes is greater than 30% and human CD45+ leukocytes populate the mouse primary and secondary lymphoid organs. In addition, human T cells, macrophages and dendritic cells were detected in the vaginal, rectal and intestinal mucosa [37]. In the RAG-hu mice, human hematopoiesis continues for more than 1 year as evidenced by the maintenance of a stable population averaging 20-50% of human C45+ leukocytes in the peripheral blood.
The RAG-hu mice are infectible with a variety of X4 and R5 isolates, have plasma viremia and circulating human PBMC containing HIV that is detectible by PCR [10]. The level of viremia in the plasma ranged up to 165,000-12,200,000 copies/ml, but may rise and fall over time. HIV infection is detected in lymphoid tissues and CD4 depletion occurs after HIV infection, but the extent of CD4 depletion can vary widely during infection. The infection has persisted for a long time, with HIV detected in the RAG-hu bone marrow almost 1 year after inoculation. Mucosal HIV transmission occurred in RAG-hu mice as evidenced by the development of plasma viremia within 1 week after the mice were infected vaginally and rectally with an R5 HIV isolate without any prior hormone treatment or introduction of mucosal abrasion. This primary transmission was associated with the dissemination HIV infection may undermine the human immune response of RAG-hu mice. Studies using the RAG-hu mouse system by the Akkina group to model Dengue fever, for which there is currently no ideal animal model available to study viral pathogenesis and to test vaccines, may be more informative of the capacity of RAG-hu mice to generate primary human immune responses [38]. There are 4 serotypes of Dengue virus and re-infection of individuals with a second serotype virus causes worse disease than infection with the primary virus due to antibody-dependent enhancement. After challenges of RAGhu mice with Dengue virus the mice become infected and develop Dengue-specific antibody. Viremia (10 6 particles/ mL) lasts up to 2 weeks and Dengue viral replication is detected in the mouse spleens. Dengue-specific IgM and IgG responses are first detected at 2 weeks and at 6 weeks after infection, respectively. Dengue virus neutralization was detected in the sera of some mice at a titer of up to 1,000 by using a FACS-based assay. Of interest was the observation that the immune response to Dengue was much more robust than the immune response to HIV after infection. This may reflect HIV-associated compromise of the human immune system in the RAG-hu mice.
Future studies by the Akkina group using this model will include evaluating the long-term effects of microbicides, studying viruses that infect the hematolymphoid system, evaluating gene therapy strategies using vectors carrying anti-HIV genes and drug-selection makers, investigation of the mechanism of antibody-dependent enhancement during Dengue infection and the testing of Dengue vaccines.
Dr. Luban reported on the system developed by Markus Manz at his Institute, of injecting human CD34+ HSC intrahepatically into newborn Balb/c RAG2 -/-γ c -/mice (Table  3) [15]. These mice were obtained from Dr. Weissman, who originally got them form Dr. Mamoru Ito in Japan. Strain-specific factors contributed to the degree of reconstitution. Mice carrying the same RAG2 and γ c deletions on the C57BL/6 background did not become reconstituted with human leukocytes. In contrast, the lymphoid tissues of the Balb/c RAG2 -/-γ c -/mice display reconstitution with human B cells and T cells and population of the thymus with human T cells. No significant population of human leukocytes was detected in the mouse mucosa or brain. After intra-peritoneal injection of either the R5 or X4 strains of HIV, YU2 and NL4-3, respectively, the mice developed systemic infection with sustained plasma viremia of up to 10 6 HIV RNA copies/ml [39].
The γ c -/mice used in these studies have a partial deletion of the common gamma chain receptor gene with expression of a truncated common gamma chain receptor that binds the appropriate cytokine, but lacks the intracellular signaling region. It is unclear if this truncated receptor has any functional activity, but mice having complete deletion of the common gamma chain receptor are also available. The litter size of the Balb/c RAG2 -/-γ c -/mice ranges from 3 to 11 mice, with an average of about 6 mice. Their group obtains sufficient human CD34+ HSC from each cord blood donor to inject an average of 4-6 mice. After reconstitution of the mice, analysis of whole blood after RBC lysis, demonstrated that the peripheral blood of 90%   Advantages of this model are that it uses no human fetal tissues, requires no surgery, and displays no global activation of the human leukocytes populating the mouse lymphoid tissue. The mice are hardy, breed well and develop no tumors in the thymus. These mice can be used to evaluate therapeutics, but their use for this purpose is limited by the modest throughput. They can be used to study HIV pathogenesis including which isolates infect brain and other cell types, mechanisms of cell to cell spread, restriction factor biology, and potentially in vivo imaging. They also can be used to study the impact of viral genetic and host genetic factors on HIV replication. This relatively facile reconsitution model may thus be an ideal system in which to test antiviral gene therapy.
The human T cells are likely to mature in the mouse thymus and interact with human dendritic cells that are present in the mouse thymic epithelial tissues. Positive selection is indicated by the presence of mature human T cells in the periphery and negative selection is indicated by the absence of graft vs. host disease [40].
After infection with HIV, low liters of HIV-specific antibody are detected in only 1 in 25 mice [39]. To use these mice as a model to study vaccines needs improvement. The low number of CD34+ HSC that can be isolated from the cord blood limits the number of mice that could be generated from isogenic CD34+ cells. Over several months, the levels of human CD45+ cell declined, and human CD45+ leukocytes were not detected in the mucosa or lungs of the mice. The capacity of human leu-kocytes to mature, differentiate and localize to the appropriate lymphoid tissue may be limited by the inability of some mouse molecules to exert their functional activity on human cells. The current mouse model permits introduction of gene therapy vectors into human HSC, prior to injection into the mice, including genes that could protect mature human CD4 T cells from HIV infection. For this purpose the Luban group is developing lentiviral vectors.
To circumvent the limited availability human HSC derived from cord blood, Dr. Luban is attempting to generate human HSC from human ES cells.
Dr. Su uses the same Balb/c RAG2 -/-γ c -/mouse (DKO mouse) but circumvents the limitation of the low number of CD34+ HSC obtainable from cord blood by using human fetal liver as a source of human HSC (Table 3). After intrahepatic injection of neonatal DKO mice with human CD34+ cells (5 × 10 5 cells/mouse), the periphery of the mice (hu-DKO mice) become populated with human T cells, B cells and dendritic cells, including mDc and pDC. The human leukocytes populate the mouse spleen to about 1/3 the size of the normal mouse spleen and the mouse thymus to about 20% of the size of the normal mouse thymus. The human T cells undergo positive and negative selection during maturation as indicated by the observation that the human cell-tropic EBV infection leads to effective anti-EBV T cell responses and circulating human T cells (or splenocytes) do not generate a mixed lymphocyte reaction (MLR) against human leukocytes from another hu-DKO mouse transplanted with human CD34+ HSC from the same donor but do generate an MLR against human leukocytes isolated from a hu-DKO mouse transplanted with CD34+ HSC from a different donor. Although mesenteric nodes draining the intestines of the hu-DKO mice contained human T cells and B cells, they did not detect either Peyer's patches or significant numbers of human CD45+ leukocytes in the lamina propria of the gut. Immunization of the mice with an HBV vaccine generated germinal centers in the mouse lymph nodes. Although the mouse B cells do not express IgG,  they can be driven in vitro to produce IgG if incubated with T cells stimulated with anti-CD3 and anti-CD28. The population of human T cells is not increased phenotypically in number by implantation of human thymic tissues from the same donor, but the implanted mice subsequently developed rashes and became sick. Dr. Speck presented his group's experience using hu-DKO mice generated from Balb/c RAG2 -/-γ c -/mice (DKO mice) injected with human CD34+ HSC isolated from cord blood either fresh or frozen and stored at -80°C (Table 3). Close to 100% of the lymphocytes present in the lymph node and thymus of the time were human. There was also a large quantity of human dendritic cells in the liver. After inoculation with HIV strains YU2 or NL4-3, the mice developed sustained high levels of viremia, and displayed CD4 depletion that was more extensive after infection with NL4-3 than with YU2 [39]. In the thymus, extensive infection with NL4-3 was detected, but almost no infection with YU2 was observed. About 1 in 25 mice generated HIV-specific antibodies. While construction of the hu-DKO mice did not require the use of fetal tissue and the hu-DKO mice developed sustained and disseminated HIV infection, the infected hu-DKO mice generated almost no HIV-specific immune responses. The mice displayed variable rates of engraftment and the function of the engrafted leukocytes may vary depending on unique genetic factors associated with the specific donor of the CD34+ HSC. Use of these mice requires a BSL2/3 facility and access to cord blood. Dr. Speck's group can generate about 50 mice/month. Engraftment depends on the quality of the mice and for optimal engraftment new breeding pairs are used every 3-4 months. Engraftment with human leukocytes decreases over time and the percentage of circulating human T cells needs to be above 1% to get successful infection after intraperitoneal injection. However, the level of engraftment with human CD45+ leukocytes does not predict the level of HIV replication after infection.
Despite the generation of Ag-specific immune response to model antigens, the hu-DKO mice displayed no or very weak adaptive immune responses to HIV. They also did not display hypergammaglobinemia, a hallmark of early HIV infection. Only scattered human CD45+ leukocytes were detected in the gut mucosa. After treatment of the mice with progesterone, the mice exhibited thinning of the vaginal epithelia and could be infected by the vaginal route with HIV-infected PBMC but not with cell-free virus.
No human T cells were detected in the lamina propria and human CD45+ leukocytes detected in the perianal tissues were likely to be human macrophages.

NOD/SCID-based mouse models
Dr. Victor Garcia presented his results using a different model, BLT mice (Table 2) [17,41] where human fetal liver and thymus are divided into pieces that are surgically implanted under the kidney capsule of NOD/SCID mice; one fraction of the liver is used for isolation of CD34+ HSC that are frozen and three weeks after surgery, the mice are sublethally irritated and intravenously injected with the freshly thawed syngeneic CD34 HSC (0.25-2.5 × 10 6 cells/mouse). One-to-two months after injection, 20-50% of the circulating leukocytes in the peripheral blood of the mice are human, and include human T cells,

SCID Mouse-based mouse models
Dr. Stoddart presented her experience using the SCID-hu Thy/Liv-mouse model ( Dr. Stoddart demonstrated that the dose response and antiviral activities of licensed ART in the SCID-hu model was comparable to that observed during treatment of HIVinfected individuals. A major advantage of the SCID-hu model is that large cohorts of mice can be constructed from the same donor removing a potential confounding variable in comparative drug studies of the differential response of tissues from different donors. This also permits the production of mice at a low cost per mouse. The SCID-hu mice can be infected with wild type or mutant isolates of HIV-1 for the evaluation of anti-HIV therapy against wild-type or drug-resistant mutants or investigation of the impact of mutations in HIV genes on in vivo replicative capacity. Use of the SCID-hu Thy/Liv model is limited by the requirement for fetal tissue, the need for surgical construction of each mouse, the lack of immune responses generated by the human T cells, and the limitation of HIV infection to the human thymic implant.

Perspective and Recommendations
Dr. KewalRamani summarized the advantages and disadvantages of studying HIV infection using rodent-based models compared to macaque-based models. Advantages of the human chimeric mouse models are that they are populated with human lymphoid and myeloid cells, infectible with a broad range of HIV isolates, amenable to genetic manipulation of the mouse recipient, able to recapitulate variation among donor recipients and are relatively inexpensive. Disadvantages of the human chimeric models are their inability to generate a robust HIV-specific immune response which precludes their use for testing HIV vaccines, their requirement for human HSC and the lack of regional centers. Dr. Lifson discussed the selective infection of macaques with SIV and not HIV. Although the pathogenic behavior of various strains of SIV isolates closely resembles that of HIV including the induction of CD4 depletion, development of CNS disease and persistence of HIV in anatomical reservoirs during ART, experimental results using SIV-infected macaques may not correlate with the pathogenesis of HIV infection in humans. While experiments using SIV-infected macaques are expensive, the presence of regional primate centers provides investigators with access to this primate model.
A panel of investigators discussed the following Questions during Panel discussion: What are the hindrances to widespread use of the models for pathogenesis studies?
What is needed to determine the feasibility of the models for vaccine and therapeutic testing, including testing of microbicides?
What can be done to overcome some of the current blocks for the models to support quality assays? Some of these include CD34 cell suppliers, mice, how cells are treated, virus, others.
What can the NIH do to facilitate solutions to the above questions?
A limiting factor discussed by panel members was the variable availability of these newer immunodeficient mice to investigators. This could be circumvented by having NIAID set up a centralized repository in an existing animal facility to permit establishing and replenishing mouse colonies for investigators. Another bottleneck is the availability of human HSC. Three sources of human HSC are available. The first is mobilized CD34+ HSC obtained from GM-CSF-stimulated individuals, a modality used to obtain HSC autologous bone marrow for transplant, particularly in individuals with leukemia. However, Dr. Su stated that in his lab, mobilized human HSC did not effectively engraft the DKO mice with human hematopoietic cells. The second source of human HSC is cord blood. While the number of HSC obtained from a cord blood donor is only sufficient for reconstituting 4-to 6 mice, cord blood is readily available and an NIH-sponsored repository could be set up that provide cord blood HSC to investigators. Furthermore, pooled CD34+ HSC from several cord blood donors could be used for some experiments requiring larger numbers of mice. The third human HSC source is human fetal liver, which provides sufficient numbers of CD34+ HSC to populate large quantities of mice. However, because the source of this tissue is from fetal abortuses, access to this tissue is restricted and subject to additional review by oversight committees.
The panel highlighted several areas that should be the subject of future scientific investigation to advance these model systems, particularly identification of factors that would increase the development of functional human immune response in these mice. Although population of the hu-DKO mouse thymus with human double and single positive T cells was suggestive of it being the location of human thymopoiesis, double positive T cells were also detected in lymph nodes. Therefore, further dissection of the mechanism of human T cell maturation in this system could provide new methods to increase the qualitative function of T cells that mature in the human mouse chimeric model. A critical technical issue was raised regarding the importance of standardizing metrics for measuring the population of human leukocytes in the peripheral blood to permit the repopulation results of different groups to be compared. To compensate for variation in gating for live cells or lymphocytes, one approach would be to report the ungated percentage of human CD45 cells after lysis of whole blood while another approach would be to report the population as the absolute number of human leukocytes/ml of blood. Another area to investigate is the mechanism of localization of mouse lymphoid and myeloid cells to the mouse mucosal tissue. While Dr. Akkina reported that the gut mucosal tissue of hu-DKO mice constructed by his group were populated with human leukocytes, relatively more extensive human leukocyte engraftment in the mucosal tissue has been documented for the humanized BLT mouse model [17,41]. Delineation of the basis for the differential capacity of human leukocytes to populate the mucosal associated lymphoid tissues of mice may provide insights into the mechanism for lymphocyte homing to the gut and other mucosal areas. For example, is it a consequence of the different location of maturation of the human leukocytes in the mouse models, human thymic implant vs. mouse thymus or due to the different immunodeficient mice used, NOD/SCID vs Rag -/-γ c -/-? The consensus of the investigators was that there is no clear-cut best rodent model that is applicable to all studies and specific models may be more suited for investigating different aspects of HIV pathogenesis and therapeutic efficacy. For this purpose it is advisable to continue to fund the development of new rodent model systems.