Humanized NOG mice as a model for tuberculosis vaccine‐induced immunity: a comparative analysis with the mouse and guinea pig models of tuberculosis

Summary

The humanized mouse model has been developed as a model to identify and characterize human immune responses to human pathogens and has been used to better identify vaccine candidates. In the current studies, the humanized mouse was used to determine the ability of a vaccine to affect the immune response to infection with Mycobacterium tuberculosis. Both human CD4⁺ and CD8⁺ T cells responded to infection in humanized mice as a result of infection. In humanized mice vaccinated with either BCG or with CpG‐C, a liposome‐based formulation containing the M. tuberculosis antigen ESAT‐6, both CD4 and CD8 T cells secreted cytokines that are known to be required for induction of protective immunity. In comparison to the C57BL/6 mouse model and Hartley guinea pig model of tuberculosis, data obtained from humanized mice complemented the data observed in the former models and provided further evidence that a vaccine can induce a human T‐cell response. Humanized mice provide a crucial pre‐clinical platform for evaluating human T‐cell immune responses in vaccine development against M. tuberculosis.

Introduction

Mice with reconstituted human immune system components (humanized mice) are increasingly being explored for efficient and cost‐effective evaluation of disease pathogenesis, therapeutic responses in vivo, and for the development of new vaccines.1 Humanized mice are better suited to study infection with human‐restricted pathogens and the immune responses against them than other small animal models such as mice and rats, which have been traditionally used as hosts to study disease pathogenesis.2, 3 Accordingly several human‐trophic viruses including human immunodeficiency virus type‐1, Epstein–Barr virus, Kaposi's sarcoma‐associated herpesvirus, herpes simplex virus, human cytomegalovirus, hepatitis C virus and Dengue virus, as well as bacterial pathogens such as Mycobacterium tuberculosis and Salmonella typhi, have been studied in humanized mice primarily to investigate disease pathogenesis and the innate and adaptive immune responses that these pathogens evoke post infection.4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 Additionally, a variety of antigens and human vaccine preparations have been shown to induce innate and adaptive T‐cell‐mediated immune responses in humanized mice, including dinitrophenyl, keyhole limpet haemocyanin antigen, toxic shock syndrome toxin 1, tetanus toxoid, DTaP vaccine and hepatitis B virus vaccine.21, 22, 23 Therefore, vaccine development that aims to primarily evaluate antigen plus adjuvant candidates for improved T‐cell immunity can be advanced in humanized mice.

The immunopathogenesis of disease caused by M. tuberculosis has been examined in the humanized mouse model and has gained interest because the model may provide a better understanding of the infectious process and characterization of the response of the human immune system. These studies, one using Mycobacterium bovis bacillus Calmette–Guérin (BCG) and M. tuberculosis Harlingen,24 and the other using M. tuberculosis H37Rv,25 demonstrated human T cells within granulomas. Another study using BCG suggested that the T cells were anergic to M. tuberculosis antigens, but responded to mitogen stimulation.26 Interestingly, infection in all these studies resulted in relatively high numbers of mycobacteria in multiple organs regardless of the virulence of the strain. It is not clear why these mice have difficulty in limiting infection, despite the presence of effector immune cells. None of these previous studies examined the effect of vaccination on infection with M. tuberculosis.

C57BL/6 mice are used to test vaccine candidate immunogenicity and protective immunity, during early vaccine development, whereas Hartley guinea pigs provide information about a vaccine candidate to prevent disease.27 These commonly used models have provided valuable information about tuberculosis immunopathogenesis, but shortcomings have been reported in their ability to model all aspects of tuberculosis in humans.28

The objective of the current studies was to assess the lymphocyte response to vaccination against M. tuberculosis in the humanized mouse using the standard BCG vaccine and a vaccine that contained CpG‐C as a molecular adjuvant and determine if vaccination of the humanized mouse could induce a cytokine response predictive of what is seen in humans. A comparison was made with the humanized mouse model and two commonly used models for testing novel tuberculosis vaccines, the C57BL/6 mouse model and the Hartley guinea pig model.

Materials and methods

Generation of humanized mice

Humanized mice were constructed and validated as per standard operating procedures by HuMurine Technologies (La Verne, CA). Briefly, human CD34⁺ hematopoietic progenitor cells (HPCs) were isolated and enriched by using a commercially available kit (Miltenyi Biotech, San Diego, CA) as per the manufacturer's instructions. Isolated cells were cryopreserved in freezing media containing DMSO and then stored in liquid nitrogen. Newborn NOG pups (NOD.Cg‐PrkdcscidIl2rgtm1sug/JicTac; Taconic Biosciences, Hudson, NY) were irradiated using a γ irradiator within 96 hr of birth with one dose of 100 cGy and immediately after were injected intra‐hepatically with 1 × 10⁵–5 × 10⁵ thawed HPCs in 50 μl of PBS. All engrafted mice were bled at 12 weeks post‐engraftment and peripheral blood was analysed for human leucocyte reconstitution by assessment of the ratio of human CD45⁺ to mouse CD45⁺ cells. Mice with human CD45 levels > 30% in the peripheral blood were selected for experiments.

Animals

Pathogen‐free, female, 6‐ to 8‐week‐old C57BL/6 mice (Jackson Laboratory, Bar Harbor, ME) and out‐bred Hartley guinea pigs, weighing 450–500 g (Charles River Laboratory, Wilmington, MA) were maintained in the Animal Biosafety Level 3 facility at Colorado State University with sterile chow and water ad libitum. The pathogen‐free nature of the mice was determined by routine screening of sentinel mice. Colorado State University Animal Care and Use Committee approved all experimental procedures.

Reagents

1,2‐Dioleoyl‐3‐trimethylammonium‐propane (chloride salt) (DOTAP; Avanti Polar Lipid Inc, Alabaster, AL) was used to prepare liposomes for the vaccine formulation. The Toll‐like receptor 9 agonist, CpG‐C DNA (ODN 2395), Human/Mouse (Hycult Biotech Inc., Plymouth Meeting, PA) was also used for vaccine formulation. Recombinant ESAT‐6 (6 kDa early secretory antigenic target of M. tuberculosis) was obtained from BEI Resources (Manassas, MD).

Mycobacteria and infection

Mycobacterium tuberculosis H37Rv (TMCC#102) was grown as a pellicle on Proskauer and Beck (P&B) medium then passaged three times in P&B medium containing 0·05% Tween‐80 to mid‐log phase and vials of working stocks were frozen at −80° until used. Mycobacterium bovis BCG Pasteur (TMCC#1011) was grown in P&B medium with 0·01% Tween‐80 to mid‐log phase. Aliquots were stored at −80° and thawed before use. Mice were infected with virulent M. tuberculosis H37Rv through the aerosol route using the Middlebrook Aerosol Exposure Chamber (Glas‐Col, Terre Haute, IN) using the standard exposure protocol to deliver approximately 100 CFU of bacilli per mouse.29 Guinea pigs were exposed through the respiratory route to 10–20 CFU of virulent M. tuberculosis H37Rv using a Madison Aerosol Chamber (Madison, WI).29 Post‐infection, guinea pigs were monitored daily for body temperature, health conditions and weighed weekly until euthanasia was required due to disease progression. Humane end‐point criteria, specified and approved by the Institutional Animal Care and Use Committee, were applied to determine the time of euthanasia.

Determination of colony‐forming units

Lungs from infected mice were homogenized in sterile saline and plated in 10‐fold serial dilutions onto 7H11 agar to determine the colony‐forming units (CFU). Plates were incubated at 37° for 21 days, after which colonies were counted. In mice vaccinated with BCG and then infected with M. tuberculosis, organ homogenates were also plated on 7H11 agar supplemented with thiophene‐2‐carboxylic acid hydrazide to inhibit the growth of BCG. When compared with organ homogenates plated on 7H11 agar, there were no differences in the CFU, indicating that only M. tuberculosis was being detected.

Vaccination protocols

Mice were inoculated subcutaneously with 5 × 10⁴ CFU M. bovis BCG and guinea pigs intradermally with 10³ CFU M. bovis BCG. For intranasal inoculations, a formulation consisting of 1 mm DOTAP, liposome, 20 μg CpG‐C ODN and 2 μg recombinant ESAT‐6 protein in a total volume of 0·02 ml was administered three times at 2‐week intervals.

Primary cell isolation and flow cytometry

Single‐cell suspensions from lungs and spleens of mice were obtained as previously described.30 For surface staining, cells were resuspended to 10⁷ cells/ml in FACS buffer and incubated with primary antibody at the manufacturer's recommended concentration. Monoclonal antibodies, conjugated to their respective fluorochromes were purchased from BD Biosciences (San Jose, CA) and eBioscience (San Diego, CA). For surface phenotype characterization, cells were incubated with CD3‐phycoerythrin (PE) (Clone: HIT3a), CD4‐FITC (Clone: SK3), CD8‐V500 (Clone: RPT‐T8), CD45RA‐PE (Clone: HI100), CD45RO‐allophycocyanin (APC) (Clone: UCLH1). For intracellular cytokine staining interferon‐γ (IFN‐γ) ‐APC PE‐Cyanine7 (Clone: 4S.B3), interleukin‐2 (IL‐2) ‐APC (Clone: MQ1‐17H12), tumour necrosis factor‐α (TNF‐α) ‐FITC (Clone: MAb11), Granzyme B‐V450 (Clone: GB11), granulysin‐Alexa Fluor^® 488 (Clone: RB1) and perforin‐FITC (Clone: dG9) were used. Cells were analysed on a FACSCanto II flow cytometer (BD Biosciences). Data were analysed using flowjo software (TreeStar, Ashland, OR). Analysis and presentation of distributions of CD4⁺ and CD8⁺ T cells for cytokine expression were performed using spice version 5·1, downloaded from http://exon.niaid.nih.gov.31

ELISpot assay

To determine the frequency of human granzyme B secreting cells, splenocytes from infected mice were cultured at 5 × 10⁵/ml in RPMI‐1640 containing 10% fetal bovine serum (Atlas Biologicals, Fort Collins, CO) and penicillin and streptomycin with antigen for 24 hr at 37° in 5% CO₂. The human granzyme B ELISpot assay (BD Biosciences) was used according to the manufacturer's protocol. For IFN‐γ and IL‐2 in the C57BL/6 mouse model, splenocytes were isolated as described above and the number of spot‐forming units was detected using the IFN‐ γ and IL‐2 ELISpot assays (eBioscience) according to the manufacturer's protocol. The number of spot‐forming cells was determined using a Series 5 UV‐Immunospot Analyzer (Cellular Technology, Ltd., Shaker Heights, OH).

Analysis of data

The log₁₀‐transformed data were used to determine significant differences between groups. Data were assessed for normal distribution before analysis. For analysis between multiple groups a one‐way analysis of variance with a Bonferroni t‐test was used for all pairwise multiple comparisons.

Results

Development and validation of the humanized immune system mouse

HuMurine's humanized NOG (Hu‐M™) mice are individually constructed by injecting pre‐qualified human CD34⁺ HPCs into newborn non‐obese diabetic NOD/SCID/γ _c R^−/− (NOG) mice resulting in stable multi‐lineage human haematopoiesis at 12–32 weeks post engraftment (Fig. 1a). The Hu‐M™ mouse develops and sustains humanization levels of > 50%, as determined by the ratio of human lymphocytes to murine leucocytes in the peripheral blood at 12 and 32 weeks’ post engraftment. These mice also display human T cells (CD3⁺) and B cells (CD19⁺) (Fig. 1a), as well as lower levels of human natural killer cells (CD56⁺) and macrophages (CD14⁺) (data not shown) in the peripheral blood and can sustain enhanced differentiation and maturation of human CD4⁺ (T helper type 1 cells), CD8⁺ T cytotoxic T cells for up to 32 weeks post engraftment (Fig. 1b). These mice can generate CD4⁺ and CD8⁺ naive (CD45RA⁺) and memory (CD45RO⁺) T cells, both of which are thought to be major drivers of immunity against tuberculosis through the production of effector molecules such as IFN‐γ, TNF‐α and interleukin IL‐2 (Fig. 1c).32, 33 Given the emerging role of the B‐cell‐mediated immune response against tuberculosis, we also established the presence of various B‐cell subsets in these mice including immature (CD38⁺), memory (CD27⁺), plasma (CD138⁺) and IgM⁺ cells (Fig. 1d).34 As the Hu‐M™ mouse supports the development and maturation of the immune cells implicated to play an important role in immunity against tuberculosis, we decided to implement this platform to understand the immunological parameters that determine long‐term protective immunity against tuberculosis.

Figure 1.

Profile of human lymphoid cell reconstitution in peripheral blood and spleen of the Hu‐M™. At 12 and 31–32 weeks, post‐engraftment, peripheral blood (PB) was collected, processed and examined for human CD45, CD3, CD4, CD8 and CD19 expression (a and b). All events were first gated on human CD45 expression and subsequently examined for T‐ and B‐cell‐specific markers (n = 11). Representative dot plot for the expression of (c) human T‐cell subsets including CD4, CD8, CD45RA and CD45RO (d) and human B‐cell subsets including CD27, CD38, CD138 and IgM in the spleen of Hu‐M™ mice at 25 weeks’ post engraftment.

Cellular phenotype analysis after low‐dose aerosol infection with M. tuberculosis

To determine the human T‐cell response to infection, we examined the distribution of CD45RA (naive) and CD45RO (memory) expressing CD4⁺ and CD8⁺ T cells in the lungs and spleens of Hu‐M™ mice at days 7, 22 and 34 post‐infection. The phenotype of these cells was identified based on the gating strategy outlined in Fig. 2(a). At day 0 post‐infection the lungs of naive mice contained a significantly greater number of human CD4⁺ CD45RA⁻ CD45RO⁺ T cells compared with CD4⁺ CD45RA⁺ CD45RO⁻ T cells, suggesting the presence of activated memory T cells before infection (Fig. 2b). No difference was observed between cell phenotypes at days 7 and 22 post‐infection until day 34, at which point there was a significant increase in CD4⁺ CD45RA⁻ CD45RO⁺ T cells, coupled with a decline in CD4⁺ CD45RA⁺ CD45RO⁻ T cells (Fig. 2b). In the spleen, the number of CD4⁺ CD45RA⁺ CD45RO⁻ and CD4⁺ CD45RA⁻ CD45RO⁺ T cells declined from the onset of infection until day 22, after which at day 34 there was a significantly greater number of activated CD4⁺ CD45RA⁻ CD45RO⁺ T cells. The response profile observed with the human CD8⁺ T cells reflected the CD4⁺ T‐cell response observed in the lungs and spleens (Fig. 2b), suggesting a multi‐subpopulation T‐cell response including higher levels of effector and memory cells in Hu‐M™ mice infected with M. tuberculosis.

Figure 2.

Human CD4⁺ and CD8⁺ T‐cell phenotypes in the lungs and spleen of Hu‐M™ mice after infection with a low‐dose aerosol of Mycobacterium tuberculosis H37Rv at 7, 22 and 34 days of infection. (a) Cells were analysed, initially by gating for the lymphocyte population using forward scatter (FSC) versus side scatter (SSC) parameters, then gating for CD3⁺ T cells and then for the CD4⁺ or CD8⁺ subsets, and then for phenotypic expression of CD45RA and CD45RO (representative gating strategy). (b) The lungs and spleens from infected animals were excised and examined for the presence of naive (CD45RA) and effector (CD45RO) T cells by flow cytometry after pulmonary infection. ***P<0·001.

Post‐infection cytokine expression

Cytokines such as IFN‐γ, IL‐2, TNF‐α, IL‐12 and chemokines are important for recruitment, differentiation, proliferation and activation of cells within the granuloma.35, 36, 37, 38, 39 Of interest in tuberculosis are the expression of IFN‐γ, TNF‐α and IL‐2, which have all been shown to be necessary for adaptive immunity to intracellular pathogens.32, 33 Therefore, we wanted to examine the expression of IL‐2, TNF‐α and IFN‐γ by human CD4⁺ and CD8⁺ T cells in response to infection. In the lungs, IL‐2 secretion was not detected for either CD4⁺ or CD8⁺ T cells at day 0 (Fig. 3a), despite the presence of elevated numbers of CD45RA⁻ CD45RO⁺ suggesting the possibility that these cells were a pool of activated T cells resulting from engraftment, that had stopped proliferating at the time of infection. By day 7, the number of CD4⁺ IL‐2⁺ T cells increased significantly in this organ, when compared with day 0 (P < 0·001), but waned at day 22, only to increase again at day 34 (P < 0·001), suggesting waves of proliferation in response to infection. Similar kinetics of expression for CD4⁺ IL‐2⁺ T cells was observed in the spleen, with a peak at day 7 (Fig. 3b). CD4⁺ IFN‐γ ⁺ and CD4⁺ TNF‐α ⁺ T cells increased significantly at day 34 in the lungs, compared with all other time‐points (P < 0·05). Splenic CD4⁺ IFN‐γ ⁺ and CD4⁺ TNF‐α ⁺ T‐cell numbers were reduced at day 21 compared with all other time‐points (P < 0·01), before a significant increase at day 34 (P < 0·01).

Figure 3.

Human CD4⁺ and CD8⁺ T‐cell responses in the lungs and spleens of mice to pulmonary infection after low‐dose aerosol infection with Mycobacterium tuberculosis H37Rv. Lung (a) and spleen (b) CD4⁺ and CD8⁺ T cells were analysed for expression of interferon‐γ (IFN‐γ), interleukin‐2 (IL‐2) and tumour necrosis factor‐α (TNF‐α), and for the expression of cytolytic proteins in CD8⁺ T cells at days 7, 22 and 34 post infection (c). The flow cytometric gating strategy represented in (d) was used to obtained cell numbers. n = 5 or n = 6 mice per group per time‐point. *P<0·05, **P<0·01, ***P<0·001.

CD8⁺ IL‐2⁺ T‐cell numbers increased significantly in the lungs (Fig. 3a) at all time‐points after infection (P < 0·001), but only at day 7 in the spleen (P < 0·05) (Fig. 3b). CD8⁺ IFN‐γ ⁺ and CD8⁺ TNF‐α ⁺ T cells were elevated significantly in the lungs at day 34 post‐infection compared with all other time‐points (P < 0·001), suggesting the recruitment of effector CD8⁺ T cells to the lung in the latter stages of infection. In the spleen, the number of CD8⁺ IFN‐γ ⁺ T cells was significantly greater at day 34 compared with days 0 and 22 (P < 0·01) and at day 7 compared with day 22 (P < 0·05). The number of CD8⁺ TNF‐α ⁺ T cells was elevated at day 0 and subsequently declined as infection progressed in the spleen. Analysis of CD8⁺ T‐cell cytolytic proteins in the lung and spleen showed elevated number of cells expressing Granzyme B from days 7 to 34 of infection, whereas expression of perforin and granulysin increased at day 34 (Fig. 3c). These data suggest that there was a significant human T‐cell response during pulmonary infection with M. tuberculosis. Furthermore, these cells were able to undergo differentiation to an effector state and produce cytokines that are induced during the human infection.

Vaccination alters the immune response to infection in Hu‐M™ mice

To determine if the cellular response to infection was influenced by previous vaccination, lung CD4⁺ and CD8⁺ T cells from humanized mice vaccinated with either BCG or CpG‐C/liposome/ESAT‐6 and subsequently infected with M. tuberculosis were assessed for expression of IFN‐γ, IL‐2 and TNF‐α. For vaccination, BCG was used for one group and a second group received intranasal CpG‐C formulated with liposomes and ESAT‐6, and the response was monitored at day 30 post‐infection. Day 30 post‐infection was chosen for analysis for two reasons: first, because based on our initial infection, the day 34 time‐point demonstrated a significant difference between infected and non‐infected mice with regards to the presence of effector T cells (Fig. 2) and second, because day 30 coincided with the C57BL/6 mouse model for testing vaccine candidates, in which mice are analysed for CFU and immunological status.27 It has been our experience that at this time‐point a significant difference between vaccinated and non‐vaccinated groups can be observed. Studies with this vaccine formulation and regimen had previously shown that it was effective in significantly reducing the mycobacterial burden in M. tuberculosis‐infected mice and limiting disease in infected guinea pigs (c.f. Fig. 7). Therefore, it was chosen for the Hu‐M™ mouse study to compare responses in the different models. Analysis of cytokine expression in CD4⁺ and CD8⁺ T cells in the lung (Fig. 4) and spleen (see Supplementary material, Fig. S1) by flow cytometry, based on the gating strategy in Fig. 4(a), showed that the quality of immune response varied between the treatment groups in response to pulmonary infection, suggesting that vaccination as well as the type of vaccine, both affected the phenotype of the immune response. All groups responded to infection by induction of CD4⁺ IFN‐γ ⁺ IL‐2⁺ TNF‐α ⁺ T cells, although infection alone (Dextrose 5% in Water (D5W)‐treated mice) resulted in a significantly greater number of these cells compared with the vaccinated groups (P < 0·05). The diversity in the CD4⁺ T‐cell response to infection was greater for the D5W group (diluent used to prepare the CpG‐C‐formulated vaccine), whereas BCG vaccination caused the response to infection to be focused, with fewer numbers of CD4⁺ IFN‐γ ⁺ IL‐2⁻ TNF‐α ⁺ and CD4⁺ IFN‐γ ⁺ IL‐2⁻ TNF‐α ⁻ T cells when compared with the D5W group, although this was not statistically significant. For CD4⁺ IFN‐γ ⁻ IL‐2⁺ TNF‐α ⁻ and CD4⁺ IFN‐γ ⁻ IL‐2⁺ TNF‐α ⁺ T cells, BCG vaccination caused a significant reduction in cell numbers after infection (P < 0·05) indicating the possibility of decline in the expansion of antigen‐specific T cells and moving into the effector phase, with cells producing macrophage‐activating factors. There was no statistical difference in CD4⁺ IFN‐γ ⁻ IL‐2⁻ TNF‐α ⁺ T cells between groups. In the spleen, the CD4⁺ T‐cell response to infection was focused predominantly on cells that were IFN‐γ ⁺ IL‐2⁻ TNF‐α ⁻ (see Supplementary material, Fig. S1b) and may reflect the diverse requirement for effector T‐cell function in different organs. The breadth of the response in lungs of infected mice was broader for CD8⁺ T cells than for CD4⁺ T cells (Fig. 4b). Although CD8⁺ IFN‐γ ⁺ IL‐2⁺ TNF‐α ⁺ T cells were observed in all groups, the predominant response consisted of CD8⁺ IFN‐γ ⁺ IL‐2⁺ TNF‐α ^– T cells and CD8⁺ IFN‐γ ^– IL‐2⁺ TNF‐α ^– T cells, although this was not statistically different (Fig. 4b). CD8⁺ T cells were also examined for their ability to produce granzyme and perforin in response to infection (Fig. 4c). A significant reduction in CD8⁺ granyme⁺ perforin⁺ T cells was observed in BCG‐vaccinated mice when compared with D5W‐treated mice (P < 0·05), possibly indicating the skewing of the CD8⁺ T‐cell response to cytokine production, rather than cytolytic activity. In the spleen, CD8⁺ T cells displayed a greater diversity in their response to infection, when compared with the lung, without significant differences among groups, except for the BCG group, in which there were significantly fewer IFN‐γ ⁺ IL‐2^‐ TNF‐α ⁺‐expressing cells in comparison with D5W‐treated mice (see Supplementary material, Fig. S1b, P < 0·01).

Figure 7.

(a) C57BL/6 mouse model, CFU in the lungs of mice and interferon‐γ (IFN‐γ) and interleukin‐2 (IL‐2) response in spleens of infected mice. Mice were vaccinated, rested for 30 days and then infected with a low‐dose aerosol of Mycobacterium tuberculosis H37Rv and CFU and ELISpot performed at day 30 post‐infection (n = 4 or n = 5 mice per group). (b) Hartley guinea pigs were vaccinated, rested for 10 weeks and then infected with a low‐dose aerosol of M. tuberculosis H37Rv and monitored for disease development (n = 10 guinea pigs per group). *P < 0·05, ***P < 0·001, when compared with the saline‐treated group.

Figure 4.

Multi‐parameter analysis of lung CD4⁺ and CD8⁺ T cells in mice vaccinated with either BCG or CpG‐C/liposome/ESAT‐6 and then infected with Mycobacterium tuberculosis H37Rv at day 30 post‐infection. (a) Selection of the final multi‐parameter populations was based on the identification of lymphocyte cellular morphology from the forward scatter (FSC) and side scatter (SSC) characteristics. The CD4⁺ and CD8⁺ T‐cell populations expressing interferon‐γ (IFN‐γ) or not was selected for further classification of the expression of tumour necrosis factor‐α (TNF‐α) and interleukin‐2 (IL‐2). (b) Pie charts showing the percentage of the relative distribution for each treatment group of the subsets of CD4⁺ and CD8⁺ T cells expressing IFN‐γ, IL‐2 and TNF‐α. (c) Lung cells were also analysed for the presence of Granzyme⁺ Perforin⁺ CD8⁺ T cell. n = 4 to n = 6 mice per group. Statistical analysis was performed for each treatment group for a cytokine combination in comparison with the D5W‐treated group. *P < 0·05.

Lung cells from vaccinated mice were examined for expression of IFN‐γ, IL‐2 and granzyme B by ELISpot assays (see Supplementary material, Fig. S2). These results supported the flow cytometry data showing that human immune cells express key anti‐mycobacterial molecules in response to infection after vaccination.

We next examined the CFU and extent of pathology in the lungs of mice after infection to determine if vaccination affected the infection outcome. Figure 5 shows data from a representative experiment of the log₁₀ CFU in the lungs of the vaccinated groups, compared with the log₁₀ CFU in the lungs of the D5W‐treated group. There was no significant difference in the log₁₀ CFU between groups, but in contrast there was a range of approximately 2 log₁₀ CFU between mice within each group. Examination of lung sections revealed that in each group, infection was focused into defined granulomatous areas. On closer inspection (Fig. 6), mice in the D5W and CpG‐C/liposome‐treated groups had multiple aggregates of cellular accumulations containing cellular debris. In contrast, mice vaccinated with BCG or CpG‐C/liposome/ESAT‐6 granulomatous lesions consisted of cellular organized aggregates. These data suggest that vaccination resulted in the accumulation of cellular aggregates in the lungs of infected mice that were able to respond to infection.

Figure 5.

Representative study of log₁₀ CFU in the lungs of Hu‐M™ mice vaccinated with either BCG or CpG‐C/liposome/ESAT‐6 compared with mice treated with D5W. CFU was determined at day 30 post low‐dose aerosol Mycobacterium tuberculosis infection. Day 30 was chosen based on the mouse model used for testing vaccine candidates, at which time the vaccine‐induced immunity will have induced an effector response. n = 4 to n = 6 mice per group.

Figure 6.

Representative photomicrographs of histopathology from lung sections taken from Hu‐M™ mice infected with a low‐dose aerosol of Mycobacterium tuberculosis H37Rv. Mice were treated with D5W (diluent for CpG‐C/liposome formulation), vaccinated with BCG, CpG‐C/liposome/ESAT‐6 or treated with CpG‐C/liposome. Lungs from day 30 post infection, were placed into formalin, embedded in paraffin and sections were stained with haematoxylin & eosin. All groups displayed focal lesions with necrosis, but vaccinated animals had organized cellular aggregates. Original magnification, 60×. Bar = 20 μm.

Comparison of Hu‐M™ mice to the C57BL/6 and guinea pig models of tuberculosis

The same vaccines used in the humanized mouse were used to inoculate C57BL/6 mice and Hartley guinea pigs. In the mouse model, the CpG‐C/liposome/ESAT‐6 formulation caused a significant reduction in CFU in the lung and spleen when compared with the saline‐treated group (Fig. 7a) that was associated with increased numbers of cells secreting IFN‐γ and IL‐2 when compared with the BCG‐vaccinated group (Fig. 7b). In the guinea pig, the vaccine formulation caused a significant increase in the survival time of guinea pigs when compared with the saline‐treated group (P < 0·05, Fig. 7c). These data suggest that in two animal models used for assessing the efficacy of vaccines, the CpG‐C‐based vaccine was able to stimulate a potent T‐cell response that caused a reduction in CFU in mice and prolonged the survival of guinea pigs after pulmonary infection. The three models provide a broad picture of the capacity of a vaccine to induce protective immunity in the standard mouse and guinea pig models and then its ability to generate human T cells that can respond to infection. In this instance, it was notable that in the humanized mouse the CpG‐G‐based vaccine induced a different type of immune response to infection from that of the attenuated vaccine strain, BCG.

Discussion

The immune response in humans to infection with M. tuberculosis has been well documented and requires the concerted involvement of multiple cellular compartments and cytokines working together, resulting, in the majority of cases, in a latent state in which the organism persists for many years in an immunologically primed host.40 It is thought that a better understanding of the immune response during infection in humans may provide better strategies for vaccine development, given that the current vaccine against tuberculosis, BCG, provides at best only partial protection.41 Early studies provided evidence for the importance of T‐cell‐mediated immunity as a key factor for protective immunity,42, 43 and although this remains true, subsequent studies have focused on innate immune cells that can dictate the course of T‐cell immunity, B cells and antibodies that may be involved in clearance of organisms, similar to their role in other infectious agents such as viruses.34, 44, 45, 46, 47 In general, the relative contribution of each of these components in the immune response may be debatable, but the fact remains that there are currently no new vaccines approved for use to combat mycobacterial infection. Animal model testing is an integral part of screening new vaccines.

A better mechanistic understanding of the workings of a vaccine will provide valuable data for improved vaccine development. The pipeline of animal models to precisely define vaccine‐induced immunity has proven to be both beneficial and problematic. Our current data confirm previous studies showing that the humanized mouse can be used to investigate the immunopathogenesis of tuberculosis infection,24, 25, 26 and extends these studies to show that it can also be used in examining vaccine‐mediated immunity. However, we now introduce the concept that the model can also be used to study novel vaccines to characterize the immune response generated by human immune cells in vivo. In the current study, we used two distinct vaccine formulations, one was the attenuated strain of M. bovis, BCG, which is widely used and the second, a CpG‐C‐based subunit vaccine expressing a single M. tuberculosis antigen. The Hu‐M™ model was able to discern the differences in immune responses to infection generated by these two classes of vaccines. Mice inoculated with the CpG‐C formulations displayed a range of responses, suggesting that the interaction with TLR9 was viable and that it induced T‐cell‐mediated immunity. Hu‐M™ mice also developed effector T‐cell phenotypes similar to those reported by others in the C57BL/6 mouse,48 however, it is difficult to directly assess T‐cell function in the guinea pig model due to the limited availability of validated reagents.

Similar to previous studies in humanized mice infected with M. tuberculosis, we confirmed the induction of human CD4⁺, CD8⁺ T cells and the expression of cytolytic proteins.25 Our data differ from previous results in that inoculation of BCG into humanized mice induced an immune response to infection, but infection was not controlled. Based on our organ homogenate plating protocol, using thiophene‐2‐carboxylic acid hydrazide‐supplemented growth media there was no indication that BCG grew to the levels reported.26 This may be due to the different route of inoculation used in the current study. Our intention was to use BCG as a vaccine and therefore, as with the majority of mouse vaccine studies using BCG, it was administered subcutaneously, whereas the former study infected mice with a much higher dose via the intravenous route. Immediate dissemination of a high dose of BCG may have overwhelmed the immune system leading to higher bacterial loads in humanized mice. This outcome is reminiscent of experiments in which SCID mice infected with a high intravenous dose of BCG showed progressive growth of the organism, in the absence of adaptive immunity.49 Indeed, our observations indicated that aerosol infection with virulent M. tuberculosis resulted in relatively higher CFU when compared with the C57BL/6 mouse model. These differences may be due to multiple factors, possibly related to differences between the capacity of human and mouse cells to control pulmonary infection. In addition, there are substantial differences in the relative frequency and distribution of T and B cells between the inbred immunocompetent C57BL/6 and the humanized Hu‐M™ mice. The distribution of human T and B lymphocytes in the Hu‐M™ mouse is similar to what had been observed and reported for humanized NOG mice.13, 50 Development of the humanized Hu‐M™ mouse model to more efficiently control infection is a future goal, and will provide a better understanding of the mechanisms involved in anti‐tuberculosis protective immunity.

Humanized mouse models have been previously used to study tuberculosis pathogenesis.24, 25 The bone marrow, liver and thymus (BLT) reconstituted mice were shown to have CD4⁺ and CD8⁺ T cells with proliferative capacity to pan stimulation using anti‐CD3/CD28.25 A previous study using human CD34⁺ haematopoietic stem cells provided similar evidence of polyclonal stimulation of CD4⁺ and CD8⁺ T cells,24 and both studies demonstrated that these cells were localized to lung granulomas. An additional study demonstrated a lack of antigen‐specific human T‐cell responses,26 whereas the current study provides evidence that both CD4⁺ and CD8⁺ T cells undergo T‐cell receptor rearrangement that human T cells identify and respond to M. tuberculosis antigenic stimulation. We have for the first time demonstrated the utility of this platform for understanding the efficacy of novel vaccine candidates against tuberculosis in contrast to earlier studies using BLT and Hu‐NSG mice, which were primarily employed to recapitulate the pathogenesis of the TB‐induced lung pathology. As observed, T cells from vaccinated and infected humanized mice were able to respond to infection by expanding and producing higher levels of cytokines than non‐vaccinated infected mice when stimulated ex vivo.

Although effector cells were detected in the infected mice, it is difficult to make a statement about memory T cells, given the bacterial burden. In the Hu‐M™, 4 weeks after the final vaccination, T cells were able to respond to infection, suggesting the presence of memory T cells. Studies in C57BL/6 mice using CpG‐C formulated with liposomes and ESAT‐6 have demonstrated that this formulation does induce a memory T‐cell response (AAI, manuscript in preparation). Further studies in post‐vaccinated Hu‐M™ mice will need to determine whether these mice can develop memory T cells similar to those observed in the C57BL/6 mice and ultimately in humans.51

In human T‐cell responses during tuberculosis infection, other groups have demonstrated that CD4⁺ T cells producing multiple cytokines combinations are present in subjects with active tuberculosis disease, although no defined pattern was associated with protection.52 Similar observations were reported in patients from another study showing that CD4⁺ IFN‐γ ⁺ IL‐2⁺ TNF‐α ⁺ T cells were increased during disease as well as single IL‐2 and TNF‐α single‐positive CD8⁺ T cells.53 Although the previous observations were derived from human peripheral blood cells, the data from our studies using lung and spleen cells from Hu‐M™ mice demonstrated a more restricted cytokine CD4⁺ T‐cell response, such as splenic CD4⁺ T‐cell populations secreting predominantly IFN‐γ, regardless of the pre‐existing immune status, whereas CD8⁺ T cells displayed a greater diversity in cytokine secretion. In general, no dominant hierarchy in response was observed for the combination of cytokines, suggesting the fact that multiple cell secreting phenotypes may be required for combating M. tuberculosis infection. The current study also examined CD8⁺ T‐cell granzyme B/perforin expression, demonstrating the presence of elevated numbers of cells during infection, whereas studies using the BLT mouse had only examined granulysin expression.25 Given that CD8⁺ granulysin/perforin expression has been shown to be produced during chronic tuberculosis,54 it would be of great interest to determine if there was a relationship with chronic tuberculosis and granulysin in the CD34⁺ HPC model, requiring the study to be extended beyond 30 days. In addition, as in humans, where potent T‐cell responses were observed in the face of active disease, the infected Hu‐M™ mice displayed high levels of infection, possibly reflecting the complex interaction between the pathogen and human immune cells. Further studies will be required to better understand the immune response in the Hu‐M™ mouse to determine how closely the model reflects human disease. Indeed, we observed a wide response to infection, as reflected in the CFU reduction even from mice within a given group, possibly reflecting the diversity in response to vaccination, similar to that seen in humans.55

With respect to cytokine‐secreting cells, the Hu‐M™ mice vaccinated with CpG‐C/liposome/ESAT‐6 developed relatively similar numbers of CD4⁺ and CD8⁺ T cells when compared with BCG‐vaccinated animals. In contrast, in the C57BL/6 mouse model, although both vaccination protocols resulted in a significant reduction in CFU, significant differences in the number of cytokine‐secreting cells between the two groups were observed. This may be a specific phenomenon associated with the mouse model or may reflect the different immune mechanisms induced by the vaccines to induce protective immunity. In addition, the reason there are fewer IFN‐γ‐secreting and IL‐2‐secreting cells in the vaccinated groups compared with the saline and adjuvant only treated groups is that the recall immune response in the vaccinated groups has resulted in a reduction in CFU and therefore in the contraction phase. Based on the guinea pig model, CpG‐C/liposome/ESAT‐6 formulation was not as efficient as BCG in preventing disease and prolonging survival, again suggesting possible different mechanisms of protective immunity. These data are subsequently supported by the Hu‐M™ model, which showed a diversity in immune response to infection, depending on the type of vaccine employed.

The question is where does the humanized mouse model fit into the scheme of animal modelling for human tuberculosis and what value does it add to the testing of vaccines? Although no model perfectly recapitulates the human infection, the humanized mouse can help in identifying critical components in the human immune response that can guide future vaccine development, particularly those that are not represented in animals. The non‐human primate is becoming more widely used as the model of choice for testing vaccines;56, 57, 58 however, the limitations associated with purchase and care make this model suitable for end‐stage vaccine development, rather than an initial screen for new vaccines. Development of the humanized mouse for M. tuberculosis infection is required and warranted as an additional preclinical tool in studying various aspects of tuberculosis disease.59, 60, 61 Future employment of the humanized mouse model will allow us to better identify factors associated with human immunity against infection and induced by vaccines at an early stage of development. The use of this model provides important information with regards to other infectious pathogens and with further refinement of the engrafted cells can lead to a human immune response that closely reflects the human responses.

Disclosures

The authors declare that we have no conflict of interest.

Supporting information

Figure S1. Multi‐parameter analysis of spleen CD4⁺ and CD8⁺ T cells in mice vaccinated with either BCG or CpG‐C/liposome/ESAT‐6 and then infected with Mycobacterium tuberculosis H37Rv based on the gating strategy to identify CD4⁺ and CD8⁺ T cells secreting interferon‐γ, interleukin‐2 and tumour necrosis factor‐α.

Figure S2. ELISpot analysis of lung cells from vaccinated Hu‐M™ mice, infected with a low‐dose aerosol infection of Mycobacterium tuberculosis H37Rv at day 30 post infection for interferon‐γ, interleukin‐2 and granzyme B.