CD8⁺ Cell Depletion Accelerates HIV-1 Immunopathology in Humanized Mice Free

In the past two decades, our laboratories have both developed and characterized small animal models for studies of HIV-1 infection reflective of human disease (1–4). Recently, NOD/scid-γ_c^null (NOD/Shi-scid, NOD/Shi-scid-IL-2Rγcnull [NOG], or NOD/LtSz-scid, NOD/LtSz-scid-IL-2Rγcnull [NSG]) mice transplanted with human CD34⁺ hematopoietic stem cells (HSCs), alone or in combination with fetal liver/thymus implant (bone marrow–liver–thymus [BLT] mice), have become promising models to study HIV-1 infection. This is due to graft longevity and the establishment of chronic viral infection in the mice (5–8). SCs, when transplanted into immunodeficient mice, on either NSG/NOG or BALB/c-Rag2^−/−γ_c^−/− backgrounds, develop a functional human immune system. These chimeric mice are susceptible to HIV-1 infection and demonstrate many components of natural human disease (4, 9–16). However, neither BALB/c-Rag2^−/−γ_c^−/− nor NSG/NOG humanized mice transplanted with HSCs alone have shown temporal control of viral replication, robust humoral, and cellular virus-specific adaptive responses [as was found in BLT animals (8)] or establishment of a stable virologic set point, as might be attributed to CTL-mediated control of viral replication.

Control of HIV-1 replication is dependent on human and viral genetics and innate and adaptive (humoral and cellular) immune responses (reviewed in Refs. 17, 18). Levels of HIV-1 replication in an infected human host markedly decline after an initial viremia and establish a stable set point. The temporal relationship between the decrease in viral load and the appearance of HIV-specific CD8⁺ CTL responses suggests that the latter may regulate viral presence (19). CD8⁺ CTL control in treatment-naive patients was determined by limited dilution functional cytotoxic assay (20) or tetramer staining (21) combined with intracellular cytokine profiles of CD8⁺ cells (22, 23). The administration of CD8-specific Abs to macaques that had been infected with SIV or SIV/HIV has been shown to abrogate the decline in viremia from its peak level, result in increased peripheral viral load, and accelerate CD4⁺ cell destruction and disease progression (24–35). The most efficient depletion of CD8⁺ cells in monkeys (lasting up to 6 wk, with near total depletion of CD8⁺ cells from blood and lymph nodes) was achieved by using cM-T807 chimeric Abs in which the heavy and L chain V region genes were isolated from the murine M-T807 hybridoma and ligated to the human γ1 H chain and κ L chain genes, respectively. Complement-independent mechanisms have been shown to be primarily responsible for cM-T807–induced CD8⁺ lymphocyte depletion in vivo, although long-term use of these Abs resulted in the development of humoral immune responses in macaques, due to xenoreactivity (27).

We now posit that further manipulations of the human immune system can be achieved in this rodent model (NSG/hCD34) of HIV-1 infection, which affects the course of disease. Herein, we demonstrate that NSG/hCD34 mice mount an HIV-specific cellular immune response following viral infection. This was shown by detecting IFN-γ and IL-2 cytokine production in response to HIV-1–derived peptide pools by human CD8 and CD4 T cells collected at 5 wk postinfection. CD8⁺ cell depletion strategies in virus-infected chimeric mice were then applied. Acceleration of HIV-1 replication was observed when CD8⁺ cell depletion was done 2 wk after viral infection. The viral load was also increased, but to a lesser extent, when depletion was conducted at 5–7 wk after viral infection. After CD8⁺ cell removal, preservation of T cell development in the thymus with the presence of CD4/CD8 double-positive cells was observed, and reappearance of human CD8⁺ cell in circulation was seen as early as 2–3 wk after depletion. Our findings underscore the importance of CD8⁺ T cell-mediated control of HIV-1 infection, are reflective of viral and CD4⁺ T cell dynamics seen previously for SIV-infected monkeys, and support the importance of this rodent model for the studies of HIV-1 immunobiology.

NOD/scid–IL-2Rγ_c^null mice were obtained from The Jackson Laboratory (Bar Harbor, ME) and bred under specific pathogen-free conditions in accordance with ethical guidelines for care of laboratory animals at the University of Nebraska Medical Center, as set forth by the National Institutes of Health.

Human cord blood was obtained, with parental written informed consent, from healthy full-term newborns (Department of Gynecology and Obstetrics, University of Nebraska Medical Center). After density gradient centrifugation, CD34⁺ cells were enriched using immunomagnetic beads according to the manufacturer's instructions (CD34⁺ selection kit; Miltenyi Biotec, Auburn, CA). Purity of CD34⁺ cells isolated was evaluated by flow cytometry and was >90%. Cells were either frozen or immediately transplanted into newborn mice at 10⁵ cells per mouse intrahepatically in 20 μl PBS using a 30-gauge needle. Before cell transplantation, newborn pups were irradiated at 1 Gy using a C9 [⁶⁰Co] source (Picker Corporation, Cleveland, OH). Cells were transplanted between 4 and 12 h after irradiation. Ranging from two to seven, littermates were reconstituted with one cord blood sample derived from one donor. The number of animals reconstituted was dependent on the number of CD34⁺ cells isolated from cord blood. Mice were weaned at 3 wk of age and randomly distributed between different experimental groups.

The CCR5 coreceptor-utilizing HIV-1_ADA strain was propagated in human monocyte-derived macrophages (36). Viral preparations were screened and found to be negative for endotoxin (<10 pg/ml) (Associates of Cape Cod, Woods Hole, MA) and Mycoplasma (Gen-Probe II; Gen-Probe, San Diego, CA). The viral titers were assayed on human monocyte-derived macrophages and determined to be 10⁵ tissue culture-infectious dose₅₀ (TCID₅₀) per milliliter.

HIV-1_ADA was injected i.p. at 10⁴ TCID₅₀. The levels of viral RNA copies per milliliter in plasma were analyzed using the automated COBAS Amplicor System (Roche Molecular Diagnostics, Basel, Switzerland) with a detection limit of 50 viral RNA copies per milliliter. Mouse plasma samples (20 μl each) were diluted up to the volume of 700 μl with normal human serum for assay use, which made the detection limit 1750 copies per milliliter. HIV-1 infection was confirmed by virologic and histologic examinations. Reconstituted noninfected animals of similar age served as controls.

The chimeric cM-T807 mAb was obtained from the National Center for Research Resources, National Institutes of Health (Dr. Keith Reimann). Mice were administered 10 mg/kg mAb s.c. (and 5 mg/kg i.p. in a 3-d interval). The doses of cM-T807 were selected to create a sufficient and sustained concentration of Ab for elimination of CD8⁺ cells due to the C5a component of complement deficiency and to prevent rapid clearance of chimeric Abs in immune-deficient mice (37). Two different time points were used for depletion: 2 and 5–7 wk postinfection. Information for all of the animals used in the study (age, period of infection, time after depletion, and levels of IgM/IgG) is shown in Table I and in Supplemental Table I.

Peripheral blood samples were collected into EDTA-coated tubes from the facial vein by using lancets (MEDIpoint, Mineola, NY) or by cardiocentesis at the end of observation. Four- or seven-color flow cytometric analyses of whole blood samples were performed to monitor changes in T cell populations. In brief, 100 μl aliquots of whole blood were incubated with respective Abs for 30 min at 4°C, and RBCs were first lysed with FACS Lysing Solution (BD Biosciences, San Jose, CA) and then washed twice with PBS containing 2% FBS. Spleen was divided into two halves; one was used for flow cytometry, and the other for immunohistochemistry. Blood leukocytes and splenocytes were tested for human pan-CD45, CD3, CD4, CD8, CD11c, CD14, CD19, and HLA-DR markers as four- or seven-color combinations. Abs and isotype controls were obtained from BD Pharmingen (San Diego, CA), and staining was analyzed with a FACS DIVA (BD Immunocytometry Systems, Mountain View, CA). To reproducibly resolve the depletion of CD8⁺ cells in the presence of cM-T807, anti–CD8-RPE clone DK25 (DakoCytomation, Carpinteria, CA) was used. Depletion was confirmed by showing an absence of CaD3⁺CD8⁺ and CD3⁻CD8⁺ lymphocytes. Results are expressed as percentages of total number of gated lymphocytes. The gating strategy was human CD45 → CD3 → CD4/CD8, CD45 → CD19, and CD45 → CD14.

Splenocytes were collected and enriched for human CD45-positive cells using magnetic bead separation (Miltenyi Biotec). Enriched cells were cultured in RPMI 1640 supplemented with 10% FCS and penicillin/streptomycin, at 2 million cells per milliliter, in the presence of either HIV-1 gag or envelope peptide pools (AIDS Research and Reference Reagent Program, Division of AIDS, National Institutes of Health). Peptide pools were used at a concentrations of 1 μg/ml each. Cells were incubated at 37°C for 18 h with brefeldin A (100 ng/ml) added during last 6 h. Cells incubated in the presence of peptide diluents served as a nonstimulated control. At the end of incubation, cells were collected and stained for surface Ags—human CD45, CD3, CD4, and CD8—followed by intracellular staining with Abs to human IFN-γ and IL-2. All of the Abs and reagents for intracellular staining were obtained from eBioscience (San Diego, CA). Stained cells were analyzed using a LSR II (BD Biosciences) and gated for human CD45 → CD3 → CD4 or CD8 → IFN-γ or IL-2.

Mouse plasma samples were diluted five times. The concentrations of human cytokines IFN-γ, IL-2, IL-4, IL-6, IL-10, and TNF-α were determined by FACSArray according to the manufacturer’s instructions (BD Biosciences).

Spleen and lymph node samples were fixed with 4% paraformaldehyde overnight and embedded in paraffin. Five-micrometer-thick sections were stained with mouse HLA-DR (clone CR3/43, 1:100), CD8 (clone 144, 1:50), CD20 (clone L26, 1:50), CD68 (clone KP-1, 1:50), or HIV-1 p24 (clone Kal-1, 1:10); all were obtained from DakoCytomation. Mouse mAbs to Ki-67 (clone DVB-2, 1:50), CD14 (clone 7, 1:50), and CD4 (clone 1F6, 1:40) were purchased from Biocare Medical (Concord, CA) and Novocastra (Norwell, MA), respectively. The polymer-based HRP-conjugated anti-mouse EnVision systems (DakoCytomation) are used as secondary detection reagents, and 3,3′-diaminobenzidine (DakoCytomation) was used as the chromogen. All of the paraffin-embedded sections were counterstained with Mayer’s hematoxylin. Deletion of primary Ab or mouse IgG served as controls. Images were obtained by Optronics (Goleta, CA) digital camera fixed to an Eclipse E800 (Nikon Instruments, Melville, NY) using MagnaFire 2.0 software.

The plasma levels of IgM and IgG were determined by ELISA (Bethyl Laboratories, Montgomery, TX).

Plasma collected at different time points were analyzed for HIV-1–specific human Igs by using Con A-based ELISA, as previously described (38). HIV-1_ADA envelope protein was expressed in a baculoviral system from a plasmid obtained from Dr. Lee Ratner (39). Purified envelope protein was used to coat the ELISA plate at 0.5 mg per well. HRP-conjugated secondary Abs to either human IgM or IgG were used (Bethyl Laboratories) and finally developed with 3,3′,5,5′-tetramethylbenzidine substrate. Plasma samples from uninfected humanized mice were used as negative controls, and HIV-1–seropositive human serum was used at a 1:10,000 dilution as a positive control.

Data were analyzed using Prism (GraphPad, San Diego, CA) and Excel software; statistical tests employed were nonparametric Mann-Whitney U test, Wilcoxon, test and one-way ANOVA for comparisons of multiple groups. A p value <0.05 was considered statistically significant. For statistical analyses, viral load values of <1750 and >35,000,000 were set to 1750 and 35,000,000 (which represent the lower and upper thresholds for the assay). Because a Wilcoxon test was used to compare the distributions, setting these interval values to their lower and upper limits, respectively, had no effect on the results.

Newborn NSG mice were reconstituted with human HSCs (hCD34⁺) isolated from umbilical cord blood. Phenotypic development of human lymphoid tissue was evaluated by flow cytometry analyses of human cells in peripheral blood (CD45, CD3, CD4, CD8, CD19, and CD14) to determine the relative abundance of immune cell groups. Presence of human Abs in peripheral blood was analyzed by anti-human IgG/IgM ELISA (Table I, Supplemental Table I). At the age of ∼5 mo, chimeric mice showed partial maturation of a reconstituted human immune system. The number of human cells in circulation became stable at 5–7 mo of age, with the majority being CD3⁺ cells (Fig. 1). At this stage, humanized mice were infected with HIV-1_ADA (a macrophage-tropic CCR5-utilizing virus) by administering 10⁴ TCID₅₀ i.p. Progression of viral infection was determined by measuring plasma viral load at different time points. Virus was rarely detected in the serum as early as 2 wk postinfection, but by 5–6 wk all of the tested animals were infected. Fig. 1 demonstrates the percentages of various phenotypes of human immune cells in spleens from uninfected and HIV-1–infected animals, as analyzed by flow cytometry. The total number of human cells (CD45⁺) in the spleen was unaffected by viral infection. However, in HIV-1–infected animals, CD4⁺ T cell numbers declined, and a parallel reduction of the CD4/CD8 ratio was noted when compared with that in uninfected animals (n = 10 per group; p = 0.032 and p = 0.026, respectively). Peripheral viral load and number of CD4⁺ and CD8⁺ human cells in circulation were analyzed in three animals biweekly (Fig. 2) and for the rest of animals at weeks 4–8 or at time of sacrifice (weeks 8–13).

In prospectively studied animals, the peak of infection occurred at 5–7 wk following viral inoculation, when it reached 5.5 (#149) and 6.5 log₁₀ (#159) copies of viral RNA, and the viral load declined by 1.5 (#149) and 0.5 log₁₀ (#159) at 9–10 wk. With the selected infectious dose of 10⁴ TCID₅₀, which provides successful infection of ∼50% of animals, the peak of viral load was observed at 5–6 wk postinfection, and the median reached 5.31 log₁₀ with a range of 4.72–6.00 range (n = 6). By 8–10 wk postinfection, viral load declined to 4.80 log₁₀ with a range of 4.12–5.39 (p = 0.023 by Mann-Whitney U test, n = 6).

To confirm CD8⁺ cell control of HIV-1 replication in this model and to characterize the cellular immune antiviral responses, we infected and sacrificed five additional humanized animals at 5 wk postinfection. To detect cellular immune responses, splenocytes were enriched for human CD45⁺ cells by magnetic bead positive selection for in vitro stimulation with HIV-1 Ags. Three uninfected animals similar in levels of reconstitution served as controls. To evaluate the presence of cellular responses and their functional properties, we used intracellular cytokine staining (ICS). The number of IFN-γ, IL-2, or dual cytokine-producing cells was analyzed following stimulation with HIV-1 gag and envelope peptide pools. Representative FACS plots for three infected mice and one uninfected mouse are shown in Fig 3. Four of five infected animals efficiently responded to HIV-1 gag peptide pool stimulation compared with HIV-1 envelope peptide pool stimulation and had detectable levels of CD8⁺IFN-γ⁺ cells, and 30–87% of these cells were double-positive for IFN-γ⁺ and IL-2⁺. All five mice showed a virus-specific CD4⁺ T cell response, as demonstrated by CD4⁺IL-2⁺ cells; a subset of these CD4⁺ cells was double-positive for both IFN-γ and IL-2. These data confirm that HIV-1 infection elicited Ag-specific human cellular immune responses.

To understand what would happen if we depleted the CTL control of viral replication in this humanized mouse model of HIV-1 infection and to provide a proof-of-concept that this animal model can be used for such immune manipulations, we used CD8 depletion strategies. CD8⁺ cell depletion was achieved by using two sequential injections of cM-T807 Abs (s.c. and i.p. within 3 d). Two schemes of depletion were used: 1) 5–7 wk postinfection, as a model of established HIV-1 infection, and 2) 2 wk postinfection to prevent the development of CD8⁺ cell-mediated antiviral responses and to accelerate the development of neuropathology (S. Gorantla, E. Makarov, J. Finke-Dwyer, A. Castanedo, A. Holguin, C.L. Gebhart, H.E. Gendelman, and L. Poluektova, submitted for publication). To assure that CD8⁺ cells were not detected in the presence of the chimeric anti-CD8 Ab anti–CD8-RPE clone DK25 Ab was used (27).

Dynamics of CD4⁺ and CD8⁺ cells and viral load in blood from two representative uninfected and HIV-infected/CD8 cell-depleted animals engrafted with the same donor cells are shown in Fig. 4A. CD8 cell depletion at 5 wk postinfection increased viral load in circulation and reduced temporarily CD4⁺ cell number. In spleens of these animals, the percentage of CD4⁺ cells was 8.7 ± 3.9% and was not different from the number of CD4⁺ cells in the spleens of HIV-1–infected animals analyzed at 8–10 wk postinfection (11.2 ± 3.2%, n = 5). CD8⁺ cell-depleted/uninfected animals of similar age and similar total number of human CD45⁺ cells in spleen (25.2 ± 2.6%, n = 4) analyzed at 1–3 wk postdepletion did not experience loss of CD3⁺CD4⁺ cells, but a number of CD8⁺ cells were significantly reduced to 0.38 ± 2.6% (range 0–1.4%). CD8⁺ cells reappeared in circulation by 3 wk postdepletion (Fig. 4B) with appearance of a significant number of CD4⁺CD8⁺ cells.

Changes in CD4⁺ and CD8⁺ cell numbers and the viral load in peripheral blood of four HIV-infected/CD8 cell-depleted mice (at 5–7 wk postinfection and 2 wk after CD8 depletion) and five HIV-1–infected/nondepleted control mice analyzed at the first peak of viremia 5–6 wk and 2 wk later were compared (Table II). When CD8⁺ cells were depleted in animals with established infection, an increase in viral load was detected in all of the mice with Δlog₁₀ ranging from 0.57–2.3 (p = 0.008, Mann-Whitney U test). In nondepleted mice, declines could be associated with a considerable decrease in CD4⁺ T cell count in blood. However, in depleted animals, even with lowered CD4⁺ T cell numbers, a substantial increase in viral load was observed.

In contrast to the small changes in viral load that were caused by depleting CD8⁺ cells in mice with established HIV infection (5–7 wk postinfection), depletion of CD8⁺ cells at 2 wk postinfection had a more profound effect on viral load (Fig. 5B). Only one out of five mice showed significant levels of virus at 2 wk postinfection (m341 had 6.50 log₁₀ viral copies per milliliter and was excluded from statistical analysis). After CD8 depletion, all of the animals exhibited substantial levels of virus in circulation (geometrical mean 5.13 versus 3.70 for all of the nondepleted animals, p = 0.006). The summary of the peripheral viral load values in nondepleted versus CD8⁺ cell-depleted animals is shown in Fig. 5.

To investigate the other possible mechanisms in addition to CTL loss for the increase of viral replication, we evaluated the levels of human cytokines (IL-2, IL-4, IFN-γ, TNF-α, IL-6, and IL-10) in plasma after i.p. injection of Abs by cytometric bead array. Increased viral replication in HIV-1–infected/CD8⁺ cell-depleted animals could also be associated with the homeostatic proliferation of CD4⁺ cells and increased production of cytokines in the lymphoid tissue, which are capable of stimulating HIV-1 replication. We collected plasma from seven animals at 3 d before the first s.c. injection and 3 d after the second i.p. administration of anti-CD8 Abs. Only human IL-2 levels were increased from 67.2 ± 25.9 to 158.2 ± 28.1 pg/ml (p = 0.019, with exception of one mouse with a decrease of IL-2 concentration below the level of detection limit of 10 pg/ml). Three of these seven animals were infected. We noted that in two infected animals human cells produced detectable levels of IL-4 (514 and 309 pg/ml) after CD8⁺ cell depletion. In replicate experiments, four uninfected and five HIV-1–infected animals, matched in age and reconstitution levels, were treated with isotope-matched mouse–human chimeric 1129 Ab (anti-respiratory syncytial virus, Synagis, palivizumab) and was used 2 wk after viral infection. These animals were then monitored for 10 wk. Viral loads were not affected and were in the range of 4.36–5.77 with a median of 5.25 log₁₀. The chimeric 1129 Ab contains 3% of murine sequences directed to the Ag-specific epitope and “grafted” into the CDR region of human V region. It is human IgG. Anti-CD8 cM-T807 contains complete H and L chain V region genes isolated from a murine hybridoma and ligated to the human γ1 H chain and κ L chain genes and are 30% mouse in origins. We used chimeric 1129 as a control Ab; however, the level of “humanization” of these Abs is highly discordant (data not shown).

Effect of CD8-depleting Abs on lymph node pathologies and double-positive CD4/CD8 thymocyte differentiation in thymus was studied. Two weeks after depletion, the formation of syncytia in lymph nodes was noted (Fig. 6, HIV-1 p24 staining). CD8⁺ cell depletion resulted in a significant increase of HIV-1 p24⁺ cells, changes in B cell distribution (CD79a staining), and increased activation (HLA-DR) in the lymph nodes. To test the ability of chimeric mice to restore the CD8⁺ cell pool after depletion, we analyzed thymic tissues by immunohistochemistry and flow cytometry. Thymic tissues of uninfected/nondepleted animals were populated by human CD45RO⁺CD4⁺CD8⁺ thymocytes (Fig. 7A). Cortical areas had single-positive human CD4⁺ cells and fewer CD8⁺ cells. Murine epithelial claudin-3–positive cells (data not shown) and a small number of human HLA-DR⁺ cells with dendritic cell morphology were also seen in thymi. Four thymi (two control and two HIV-1–infected mice) were analyzed by FACS for CD11c/HLA-DR expression on human CD45⁺ cells. The average numbers for single HLA-DR, CD11c, and HLA-DR/CD11c double-positive cells were 3.1 ± 1.6, 2.0 ± 0.8, and 0.5 ± 0.3%, respectively. Depletion of CD8⁺ cells significantly reduced the number of CD4/CD8 double-positive cells and increased the density of Ki-67⁺ human cells. Fig. 7A shows the immunohistology of thymi obtained from animal infected for 8 wk with CD8⁺ cell depletion for 1 wk before euthanasia. The reduction of CD4/CD8 double-positive cell numbers was also demonstrated by flow cytometry performed at 2 wk after depletion (Fig. 7B). These data confirm that CD8⁺ cell depletion did not abolish the generation of CD4/CD8 double-positive progenitor cells in thymus. Analyzed animals did not have signs of splenomegaly or aplastic anemia.

We previously demonstrated that BALB/c-Rag2^−/−γ_c^−/− mice could be repopulated with human cells within the lymphoid tissue structures following human CD34⁺ cell transplantation. These mice developed a functional human immune system that was susceptible to HIV-1 infection (4). In the current report, NSG mice that are more readily engrafted with human HSCs were used. Animals with an established human immune system were infected with a macrophage-tropic HIV-1 virus, resulting in chronic virus infection and CD4⁺ T cell depletion. Viremia reached significant levels by 5–6 wk postinfection. We observed the presence of HIV-1 gag- and envelope-specific, IFN-γ– and IL-2–producing human CD8⁺ and CD4⁺ cells in the spleens of these mice, using ICS. Most of the HIV-specific CD8⁺ cells reacted with gag, and the number of IFN-γ/IL-2 double-producers exceeded the number of single IFN-γ–producing cells. Most of the virus-specific CD4⁺ cells produced only IL-2, although some CD4⁺ cells that produced both IFN-γ and IL-2 were also detected. These results support the notion that humanized NSG mice can generate polyfunctional cellular immune responses to HIV-1, as was shown for human subjects (40–42). We did not find strong humoral immune responses in evaluated animals. In 10 evaluated animals, eight had detectable levels of HIV-1–specific IgM that declined by 10–11 wk postinfection without the development of IgG responses (data not shown).

CTL-mediated protection against HIV-1 infection at different disease stages has not been completely elucidated (43–47). The best evidence for the role of CD8⁺ CTLs in the control of immunodeficiency virus infection was obtained in SIV-infected rhesus macaques that were subjected to the immunodepletion of CD8⁺ cells. In the current model, depletion of CD8⁺ cells 2 wk postinfection led to the acceleration in viral replication during the acute phase of infection. CD8⁺ cell depletion at 5–7 wk postinfection also enhanced the viral replication and prevented the small decline observed in nondepleted mice. CD8⁺ cell depletion was associated with a significant reduction in the number of human CD4/CD8 double-positive cells in thymus. However, the pool of CD8⁺ T cells started to return at 2–3 wk postdepletion. Overall, observations made with CD8⁺ T cell depletion using cM-T807 Ab in our humanized mice partially were reminiscent of previously published data from nonhuman primate SIV or SIV/HIV model systems. The durable levels of depletion were achieved, and a strong correlation between disease outcome and viral-specific CTL restoration was observed by use of the mouse–human chimeric mAb (cM-T807) during early stages SIV infection in rhesus macaques (26). In humanized NSG mice, earlier depletion of CD8⁺ cells accelerated viral replication. A significant reduction in CD4⁺ cell number, observed in our study in blood, was also reported in the SIV model (33). Our data are also concordant with the recently published observations that CD8⁺ lymphocyte depletion can induce a transient CD4⁺ T cell increase, thus providing increased number of targets for SIV in animals during acute SIV infection without an influence on the disease outcomes (35).

The generation of HIV-1–specific human CTLs in “humanized” mice is influenced by issues of T cell maturation, receptor repertoire, interactions with murine stromal cells, and more. Nonetheless, virus-specific human CD8⁺ T cells have been observed in previous studies using immune-deficient mice reconstituted with human PBLs (3, 48–54). A more sophisticated BLT mouse model, where NOD/scid or NSG mice were transplanted with fetal thymus/liver tissue combined with i.v. transplantation of CD34⁺ stem cells (8, 55–57), also showed the generation of Ag-specific human immune responses. Our results do not contradict those that have been previously published by Brainard et al. (8). They showed peak viremia at 6 wk postinfection regardless of the strain and amount of inoculated virus (ADA or JR-CSF). They injected 70,000 TCID₅₀ dose of virus, which guaranteed 100% infection and peak viremia at 6 wk postinfection followed by a small spontaneous decline by 8–12 wk postinfection. Unfortunately, when they applied the lower inoculums of JR-CSF 2,000 TCID₅₀, no observation after 7 wk postinfection was shown. We used 5–7.5 times lower inoculums of HIV-1_ADA (10,000 TCID₅₀ dose, which did not guarantee 100% infection of mice). This likely prevented rapid exhaustion of CD8 cell functional properties. By 2 wk postinfection, a significant increase in program death receptor 1 Ag expression on CD8⁺ cells were found, which reflects the negative effect of higher infection levels on functional properties of CTLs. Human T cell selection may be another factor that could affect the adaptive immune responses. In our model, human T cell development and selection occur in murine thymus with possible elimination of mouse-reactive TCR clones, whereas in the BLT model the selection takes place in human thymus. However, the biology of thymic selection in the murine environment needs further investigation.

Our data show that in NSG mice reconstituted at birth with HSCs the thymus is highly populated with human CD45RO⁺ thymocytes, and we also observed that the T cell repertoire, analyzed by staining for variable β-chain of TCR, is preserved (data not shown). The presence of double-positive T cells in HIV-1–infected/CD8⁺ cell-depleted peripheral blood of mouse and reduction of viral load suggest that these cells may play a role in the restoration of the lymphocyte pool following CD8⁺ cell depletion and contribute to control the viral replication (58). We cannot exclude that this is an example of immunopathology induced by combination of significant levels of HIV-1 replication and CD8 depletion. This observation needs further investigation.

We have direct evidence that our chimeric mice developed cellular immune response against HIV-1. However, we cannot overlook the possibility that this increase in viral replication following CD8⁺ T cell depletion may be due to, at least in part, to increased production of IL-2/IL-4 and homeostatic expansion of CD4⁺ T cells (59). Nonetheless, our results clearly show that the NSG/hCD34 mouse model can be effectively employed to study the progression and pathogenesis of HIV-1 infection in a live animal host and to dissect the role of specific immune subsets in the control of infection. This is expected to enable future investigations of viral diversity, developmental therapeutics, and antiretroviral immunity.

We thank Dr. John G. Sharp (University of Nebraska Medical Center) for active discussions and critical insights into the immune reconstitution strategies, Dr. Keith A. Reimann (Harvard University Center for AIDS Research, Cambridge, MA) for helpful suggestions regarding anti-CD8 Abs application, Dr. Lee Ratner (Washington University, St. Louis, MO) for providing plasmids for the HIV-1_ADA envelope gp120 protein, Victoria Smith and Dr. Charles Kuszynski for FACS analyses (University of Nebraska Medical Center), Valerie K. Shostrom, Statistical Coordinator, College of Public Health Department of Biostatistics (University of Nebraska Medical Center), and Dr. Dawn Eggert for animal breeding (University of Nebraska Medical Center).

Disclosures The authors have no financial conflicts of interest.