Enhanced Activation of T Lymphocytes by Urease-Deficient Recombinant Bacillus Calmette-Guérin Producing Heat Shock Protein 70-Major Membrane Protein-II Fusion Protein

Mycobacterium leprae is a causative bacterium of leprosy (1, 2). Leprosy is clinically divided into two major categories, paucibacillary and multibacillary leprosy (2). In the lesion of paucibacillary leprosy, CD1a⁺ dendritic cells (DCs) are involved, and the substantially activated T cells are observed (3, 4). These observations indicate that host defense activity against M. leprae is chiefly conducted by adaptive immunities, and both IFN-γ–producing type 1 CD4⁺ T cells and CD8⁺ T cells act to inhibit the active multiplication of M. leprae. Thus, few numbers of bacilli are observed in the lesion of paucibacillary leprosy. The activation of T cells is induced by DCs loaded with bacilli or its component, which display various antigenic molecules on their surface, including the immunodominant Ags (5, 6). Previously, we identified major membrane protein (MMP)-II (gene name bfrA or ML 2058) as one of the immunodominant Ags of M. leprae (7). MMP-II ligates TLR2 and consequently activates the NF-κB pathway (7). DCs pulsed with MMP-II protein activate both naive and memory type CD4⁺ and CD8⁺ T cells to produce IFN-γ in an Ag-specific fashion (7, 8). Further, the MMP-II is supposed to be recognized by T cells in vivo of M. leprae-infected individuals, including paucibacillary leprosy patients (8).

Multidrug therapy introduced by the World Health Organization has been effective to reduce the number of leprosy patients registered. However, the drug therapy seems ineffective to reduce the number of newly developed leprosy patients; thus, the useful vaccine is essential to control the number of new patients. So far, Mycobacterium bovis bacillus Calmette-Guérin (BCG) is used as a vaccine against leprosy, although not broadly (9–11). However, nowadays, BCG is not recognized as a reliable vaccine, because an overall efficacy of BCG against leprosy is reported to be 26%, which is calculated by meta-analyses enrolling several studies and observations (12). However, BCG intrinsically possesses the ability to activate type 1 CD4⁺ T cells, although not convincingly, and may share some antigenic molecules with M. leprae (9, 10). These observations suggest that the improvement of BCG may be one of the critical ways to develop new effective vaccines against leprosy. However, BCG also has its intrinsic defect, an activity to inhibit the fusion of BCG-infected phagosomes with lysosomes (13–15). This defect seems to be a major factor associated with unconvincing activation of naive T cells. Therefore, we tried to upregulate the T cell-stimulating activity of BCG by overcoming the intrinsic defect of the bacteria. First of all, we produced recombinant BCG (rBCG) (BCG-SM) that secretes MMP-II in the infected cells (16). As expected, BCG-SM substantially activated both naive CD4⁺ and CD8⁺ T cells and consequently inhibited the growth of M. leprae to some extent, but not completely, in the footpad of C57BL/6 mice (17). It is known that parent BCG partially activates naive CD4⁺ T cells but is not efficient in stimulating naive CD8⁺ T cells quickly to produce IFN-γ (14, 16). In this respect, the fact that BCG-SM can activate DCs to produce IL-12p70 and both subsets of naive T cells to produce IFN-γ indicates that secretion of MMP-II of M. leprae presumably in phagosomes of APCs of host cells is a useful strategy to activate both APCs and T cells (16). Usefulness of the enhancement of secretion of vaccinated BCG-derived Ags is revealed in the other intracellular infection system such as Mycobacterium tuberculosis, in which the active secretion of Ag85 complex was effective in inhibiting the replication of subsequently challenged M. tuberculosis (18).

Then, we undertook two other strategies to further enhance the T cell-stimulating activity of BCG. One of them was aimed at potentiating the activation of naive CD4⁺ T cells. BCG possesses urease, which produces ammonia and inhibited the acidication of BCG-infected phagosomes to avoid the fusion with lysosomes (19, 20). To inhibit the ammonia production, we produced urease-deficient BCG (BCG-ΔUT-11-3) (15). BCG-ΔUT-11-3 was feasibly translocated into lysosomes and activated both DCs and naive CD4⁺ T cells of human (15). Further, BCG-ΔUT-11-3 efficiently produces memory type CD4⁺ T cells in mice that can recognize M. leprae-derived proteins (15). Thus, the disruption of the UreC gene of BCG is useful tool to enhance the CD4⁺ T cell-activating activity of BCG. The second strategy for potentiation of BCG activity is aimed to provide BCG the ability of activating IFN-γ–producing CD8⁺ T cells quickly and strongly. To this end, we used heat shock protein 70 (HSP70) (21–24). The gene encoding HSP70 was directly connected with that of MMP-II and was extrachromosomally transformed into BCG (production of BCG-70M). BCG-70M secreted HSP70–MMP-II fusion protein and activated not only Ag-specific naive CD8⁺ T cells polyclonally, but also naive CD4⁺ T cells and DCs (25). Thus, the production and secretion of HSP70 in phagosomes accompanied by MMP-II seems an effective strategy to activate human naive CD8⁺ T cells using BCG.

Because we employed two independent strategies to overcome the intrinsic defect of BCG (inhibition of phagosome-lysosome fusion), in this study, we combined the two strategies and produced another rBCG (BCG-D70M) that is deficient in urease activity but is introduced with the gene encoding HSP70–MMP-II fusion protein and evaluated its immunostimulatory activities. The BCG-D70M showed the strongest activities in terms of activation of naive CD4⁺ and CD8⁺ T cells among the rBCGs produced by us so far.

Peripheral blood was obtained from healthy purified protein derivative-positive individuals under informed consent. In Japan, BCG vaccination is compulsory for children (0–4 y old). PBMCs were isolated using Ficoll-Paque Plus (Pharmacia, Uppsala, Sweden) and cryopreserved in liquid nitrogen until use, as previously described (26). The viability of T cells obtained from cryopreserved PBMCs was >90%, and no selection in terms of functionality was induced in both monocytes and T cells by the cryopreservation of PBMCs. For the preparation of peripheral monocytes, CD3⁺ T cells were removed from either freshly isolated heparinized blood or cryopreserved PBMCs using immunomagnetic beads coated with anti-CD3 mAb (Dynabeads 450, Dynal Biotech, Oslo, Norway). The CD3⁻ PBMC fraction was plated on collagen-coated plates, and the nonplastic adherent cells were removed by extensive washing. The remaining adherent cells were used as monocytes (27). Monocyte-derived DCs were differentiated as described previously (26, 28). Briefly, monocytes were cultured in the presence of 50 ng rGM-CSF (PeproTech EC, London, U.K.) and 10 ng/ml rIL-4 (PeproTech) (28). On day 4 of culture, immature DCs were infected with rBCG at an indicated multiplicity of infection (MOI), and, on day 6 of culture, DCs were used for further analyses of surface Ag and mixed lymphocyte assays. Macrophages were differentiated as described previously (29, 30). In brief, monocytes were cultured in the presence of 10 ng/ml rM-CSF (R&D Systems, Minneapolis, MN). On day 5 of culture, macrophages were infected with rBCG at an indicated MOI, and, on day 7 of culture, they were used for further analyses of mixed lymphocyte assay. The rMMP-II protein was produced as described previously (7, 31), and the rHSP70 Ag was purchased (HyTest, Turku, Finland).

The genomic DNAs from BCG substrain Tokyo and from M. leprae Thai-53 strain were extracted by using phenol-chloroform. The oligonucleotide primers for the hsp70 gene were FMb70Bal (5′-aaaTGGCCAtggctcgtgcggtcggg-3′; capital letters indicate a BalI site) and RMb70Eco (5′-aaaGAATTCcttggcctcccggccg-3′; capital letters indicate an EcoRI site). The primers for MMP-II sequence from M. leprae genomic DNA was amplified with FMMP Eco4 (5′-aaaGAATTCcaaggtgatccggatgt-3′; capital letters indicate an EcoRI site) and RMMP Sal (5′-tgaGTCGACttaactcggcggccggga-3′; capital letters indicate a SalI site). The amplified products were digested with appropriate restriction enzymes and cloned into BalI-SalI–digested parental pMV261 plasmid. For replacing kanamycin resistance gene to hygromycin resistance cassette, the XbaI-NheI fragment from pYUB854 (32) was cloned into SpeI-NheI fragment of each plasmid (32). The rBCG that lacks ureC gene (BCG-ΔUT-11) was produced as described previously (15). The hygromycin cassette was removed by using a plasmid encoding γδ-resolvase (γδ-tnpR) encoded in pTYUB870 (32). The unmarked BCG having the hygromycin gene was named BCG-ΔUT-11-3. The HSP70–MMP-II fusion protein expression vector was introduced into BCG-ΔUT-11-3 by electroporation method. BCG-70M was produced as described previously (25). BCG-ΔUT-11-3 containing pMV-HSP70-MMP-II as an extrachromosomal plasmid is referred to as BCG-D70M and that containing pMV-261-hygromycin is referred to as BCG-261H (BCG vector control). rBCGs were grown to log phase and stored at 10⁸ CFU/ml at −80°C. Preinfection to DCs and macrophages, BCGs were counted by colony assay method. There is no significant difference in the in vitro culture growth between BCG-261H and BCG-D70M.

To verify the expression of MMP-II and HSP70 in rBCGs, we prepared cell lysates from BCG-70M and BCG-D70M as described previously (16). Briefly, the protein fraction of the rBCGs was prepared as follows: harvested cells were washed with PBS and sonicated. Disrupted cells were centrifuged at 10,000 × g at 4°C, and the supernatant was taken as the cell lysate. SDS-PAGE and electroblotting were carried out using standard methods. Western blotting was performed as follows: a polyvinylidene difluoride membrane having the transferred protein was blocked in 5% skim milk and then incubated with anti–MMP-II mAb 202-3 (IgG2a) or anti-mycobacterial HSP70 mAb (HyTest), which is not cross-reactive to mammalian HSP70 homologs. Anti-Ag85B rabbit polyclonal Ab was used as an internal control. An alkaline phosphatase-conjugated anti-mouse IgG Ab (Biosource International, Camarillo, CA) or alkaline phosphatase-conjugated anti-rabbit IgG Ab (Tago, Burlingame, CA) was used as the secondary Ab. Color development was performed using NBT/BCIP detection reagent (Calbiochem, San Diego, CA).

The expression of cell surface Ag on DCs and lymphocytes was analyzed using FACSCalibur (BD Biosciences). Dead cells were eliminated from the analysis by staining with propidium iodide (Sigma-Aldrich, St. Louis, MO), and 1 × 10⁴ live cells were analyzed. For the analysis of the cell surface Ag, the following mAbs were used: FITC-conjugated mAb against HLA-ABC (G46-2.6, BD Biosciences, San Jose, CA), HLA-DR (L243, BD Biosciences), CD86 (FUN-1, BD Biosciences), CD83 (HB15a, Immunotech, Marseille, France), CD1a (NA1/34, DakoCytomation, Glostrup, Denmark), CD62L (Dreg 56, BD Biosciences), CCR7 (clone 150503, R&D Systems), CD27 (M-T271, BD Bioscience), and PE-conjugated mAb to CD162 (TB5, Exbio, Praha, Czech), CD8 (RPA-T8, BD Biosciences), and CD4 (RPA-T4, BD Biosciences).

The expression of MMP-II on rBCG-infected DCs was determined using the mAb against MMP-II (M270-13, IgM, κ), which probably detects MMP-II complexed with MHC molecules on the surface of DC (8), followed by FITC-conjugated anti-mouse Igs Ab (Tago-immunologicals, Camarillo, CA). For the inhibition of the intracellular processing of phagocytosed bacteria, DCs were treated with 50 μM chloroquine (Sigma-Aldrich) for 2 h, washed, and subsequently infected with rBCG and subjected to analyses of MMP-II surface expression. The intracellular production of perforin was assessed as follows: naive CD8⁺ T cells were stimulated with rBCG-infected DCs for 5 d in the presence of naive CD4⁺ T cells, and CD8⁺ T cells were surface stained with PE-labeled mAb to CD8 and were fixed in 2% formaldehyde. Subsequently, the cells were permeabilized using Permeabilizing solution (BD Biosciences) and stained with FITC-conjugated mAb to perforin (δG9, BD Biosciences) or FITC-labeled isotype control.

The ability of BCG-infected DCs and macrophages to stimulate T cells was assessed using an autologous APC–T cell coculture as previously described (6, 28). Purification of CD4⁺ and CD8⁺ T cells was conducted by using negative-isolation kits (Dynabeads 450, Dynal Biotech) (28). The purity of the CD4⁺ and CD8⁺ T cells was >95% when assessed using an FACSCalibur (BD Biosciences). Naive CD4⁺ and CD8⁺ T cells were produced by further treatment of these T cells with mAb to CD45RO, which were followed by beads coated with mAb to goat anti-mouse IgGs (Dynal Biotech). The purity of both subsets of naive T cells was >97%. However, there was no contamination of memory type T cells in the naive T cell preparations. More than 98% of CD45RA⁺ T cells were positive in the expression of CCR7 molecule. Memory type T cells were similarly produced by the treatment of cells with mAb to CD45RA Ag. The purified responder cells (1 × 10⁵/well) were plated in 96-well round-bottom tissue-culture plates, and DCs or macrophages infected with rBCG were added to give the indicated APC/T cell ratio. Supernatants of APC–T cell cocultures were collected on day 4, and the cytokine levels were determined. In some cases, rBCG-infected DCs and macrophages were treated with mAb to HLA-ABC (W6/32, mouse IgG2a, κ), HLA-DR (L243, mouse IgG2a, κ), CD86 (IT2.2, mouse IgG2b, κ, BD Biosciences), MMP-II (M270-13), or normal mouse IgG. The optimal concentration was determined in advance. Also, in some cases, immature DCs and macrophages were treated with the indicated dose of chloroquine, brefeldin A (Sigma-Aldrich), or lactacystin (Sigma-Aldrich) and subsequently infected with BCG-D70M. The optimal dose of these reagents was determined in advance.

Levels of the following cytokines were measured: IFN-γ produced by CD4⁺ and CD8⁺ T cells, and IL-12p70, TNF-α, and IL-1β produced by DCs or macrophages stimulated for 24 or 48 h with rBCGs. The concentrations of these cytokines were quantified using the enzyme assay kit Opt EIA Human ELISA Set (BD Biosciences).

For inoculation into mice, rBCGs were cultured in Middlebrook 7H9 medium to log phase and stored at 10⁸ CFU/ml at −80°C. Before the aliquots were used for inoculation, the concentration of viable bacilli was determined by plating on Middlebrook 7H10 agar plate. Three 5-wk-old C57BL/6J mice (Clea Japan, Tokyo, Japan) per group were inoculated s.c. with 0.1 ml PBS or PBS containing 1 × 10³ rBCGs. The animals were kept in specific pathogen-free conditions and supplied with sterilized food and water. Four or 12 wk postinoculation, the spleens were removed, and the splenocytes were suspended at a concentration of 2 × 10⁶ cells/ml in culture medium. The splenocytes were stimulated with an indicated concentration of rMMP-II or rHSP70 (HyTest) in triplicate in 96-well round-bottom microplates (15, 16). The individual culture supernatants were collected 3 to 4 d poststimulation, and IFN-γ was measured using the Opt EIA Mouse ELISA Set (BD Biosciences). Five C57BL/6 mice per group were vaccinated with 1 × 10³ CFU/mouse either BCG-261H or BCG-D70M for 4 wk and were challenged with 5 × 10³/mouse of M. leprae in footpad. Thirty-two wk later, the footpad was removed. The number of M. leprae grown in the footpad was enumerated by Shepard method (33). Animal studies were reviewed and approved by the Animal Research Committee of Experimental Animals of the National Institute of Infectious Diseases and were conducted according to their guidelines.

Student t test was applied to determine the statistical differences.

Because BCG-D70M possess two modified characteristics, 1) a lack of urease activity and 2) an expression of HSP70 and MMP-II fusion protein, we assessed the expression level of HSP70 and MMP-II proteins in BCG strains: BCG-70M and BCG-D70M. Both BCG-70M and BCG-D70M equivalently expressed both HSP70 and MMP-II molecules as examined by Western blot analyses using Ag85B as an internal control (Fig. 1). Further, the T cell activation activity of BCG-D70M was evaluated by using not only vector control BCG (BCG-261H), but also rBCGs that lacks urease activity (urease-deficient BCG-ΔUT-11-3 and BCG-70M that secretes HSP70–MMP-II fusion protein) as control BCG (Fig. 2A). When these rBCGs were infected to DCs to use as a stimulator of naive CD4⁺ T cells, both BCG-ΔUT-11-3 and BCG-70M showed higher T cell-stimulating activity than vector control BCG. However, BCG-D70M showed the highest T cell-stimulating activity among these rBCGs at both MOIs: 0.125 and 0.25. More than 350 pg/ml IFN-γ was produced by stimulation with BCG-D70M (MOI: 0.25; T/DC ratio: 40:1). Also, at different T/DC ratios, BCG-D70M exhibited the highest activity (not shown). On addition to IFN-γ, TNF-α was also efficiently produced by BCG-D70M stimulation (not shown). So far, when macrophages were used as APCs, it was difficult to activate CD4⁺ T cells to produce IFN-γ significantly. In contrast to the activity of rBCGs, such as BCG-261H, BCG-ΔUT-11-3, and BCG-70M (15, 25), newly constructed BCG-D70M efficiently stimulated CD4⁺ T cells to produce the cytokine through macrophages at various conditions, although the T cell-stimulating activity of macrophages was much less efficient comparing to that of DCs (Fig. 2B). More than 100 pg/ml IFN-γ was produced from responder CD4⁺ T cells; however, BCG-D70M–infected macrophages failed to induce the production of significant dose of IFN-γ from naive CD4⁺ T cells (not shown). Also, BCG-D70M did not induce IFN-γ production from DCs or macrophages (not shown). Although normal murine IgG did not affect the T cell-stimulating activity of both BCG-D70M–infected DC and the BCG-D70M–infected macrophages, the treatment of these APCs with either anti–HLA-DR mAb, anti-CD86 mAb, or anti–MMP-II mAb significantly inhibited the activation of naive CD4⁺ T cells and CD4⁺ T cells, respectively (Fig. 2C). More than 90% of IFN-γ production was inhibited by the treatment of APCs when mAb to HLA-DR or CD86 was used. Furthermore, when naive CD8⁺ T cells were stimulated with DC infected with various rBCGs, BCG-D70M induced the strongest activation of naive CD8⁺ T cells (Fig. 3A). Both BCG-70M and BCG-D70M induced significant IFN-γ production, but BCG-D70M activated the T cells more strongly than BCG-70M. More than 400 pg/ml IFN-γ can be produced from naive CD8⁺ T cells. These phenomena were observed consistently at various conditions including the different MOIs and T cell/DC ratios, although high doses of BCG-D70M and high doses of BCG-D70M–infected DCs were required to induce the production of abundant IFN-γ from naive CD8⁺ T cells compared with the dose required to stimulate naive CD4⁺ T cells. Again, when BCG-D70M–infected DCs were treated with mAb to either HLA-ABC or CD86, the IFN-γ production from naive CD8⁺ T cells was significantly inhibited, whereas normal murine IgG did not affect the activation of the responder T cells (Fig. 3B).

To stimulate responder T cells efficiently, APCs susceptible to BCG infection should be adequately activated. We assessed the activation of DCs from the aspects of cytokine production and phenotypic changes (Fig. 4). When DCs were stimulated with various rBCGs including BCG-261H, BCG-ΔUT-11-3, BCG-70M, and BCG-D70M, BCG-D70M stimulated DCs to produce IL-12p70 most efficiently at both MOIs: 0.25 and 0.5 (Fig. 4A). Further, BCG-D70M induced significantly higher dose of IL-1β and TNF-α production from DCs and also induced a higher dose of TNF-α from macrophages compared with BCG-261H (Fig. 4B). To assess the phenotypic changes induced by BCG-D70M infection, we assessed the expression of MHC, CD86, CD83, and CD1a molecules on DCs (Fig. 4C). Infection with BCG-D70M induced significantly higher level of expression of HLA-ABC, HLA-DR, CD86, and CD83 Ags compared with BCG-261H infection. The expression of CD1a was significantly downregulated. We used various dose of rBCGs for the assessment, and the similar changes were observed (not shown). These results indicated that BCG-D70M more efficiently activated DCs than BCG-261H.

Previously, we reported that BCG-70M, which was genetically manipulated to produce HSP70–MMP-II fusion protein, induced expression of MMP-II on the surface of BCG-70M–infected DCs (25). We then analyzed the BCG-D70M–infected DCs in terms of MMP-II expression (Fig. 5A). Whereas DCs uninfected or infected with BCG-261H did not express MMP-II significantly, BCG-D70M induced significant expression of MMP-II. Further, when immature DCs were treated with chloroquine, an inhibitor of phagosomal acidification, in advance to the infection with BCG-D70M, the MMP-II expression was significantly inhibited. In addition to the MMP-II expression, the chloroquine treatment on DCs affects the activation of responder T cell by BCG-D70M (Fig. 5B). IFN-γ production from naive CD4⁺ T cells by stimulation with BCG-D70M (MOI: 0.125 and 0.25) was significantly inhibited by chloroquine. Also, on treatment of macrophages with chloroquine, ∼90% of IFN-γ produced from CD4⁺ T cells by BCG-D70M stimulation (MOI: 0.25 and 0.50) was inhibited. Similarly, the production of IFN-γ from naive CD8⁺ T cells was significantly inhibited by the chloroquine treatment of DCs, which were subsequently infected with BCG-D70M (MOI: 0.25 and 0.50). We further confirmed that BCG-D70M secreted 92 kDa protein (molecular mass: MMP-II 22 kDa and HSP70 70 kDa) in vitro (not shown). These results suggest the possibility that the secreted HSP70–MMP-II fusion protein is one of the responsible elements for the activation of both CD4⁺ T cells and CD8⁺ T cells as observed in BCG-70M. Further, we previously reported that BCG-70M stimulated naive CD8⁺ T cells through TAP- and proteosome-dependent cytosolic cross-presentation pathway, because the T cell activation was inhibited by the pretreatment of DCs with brefeldin A and lactacystin (25). In this study, we assessed whether BCG-D70M activates naive CD8⁺ T cells by using the similar cross-presentation pathway (Fig. 5C). When immature DCs were treated with either brefeldin A or lactacystin and were subsequently infected with BCG-D70M at MOI 0.25 or 0.50, the IFN-γ production from naive CD8⁺ T cells was significantly inhibited.

Because BCG-D70M activated both naive CD4⁺ T cells and naive CD8⁺ T cells, we assessed the influence of the presence of CD4⁺ T cells on the activation of naive CD8⁺ T cells (Fig. 6A). The naive unseparated T cell population was stimulated with DCs infected with BCG-261H or BCG-D70M, and CD8⁺ T cells were gated and analyzed by FACS. Compared to CD8⁺ T cells stimulated with BCG-261H, the BCG-D70M–stimulated naive CD8⁺ T cells produced significantly higher number of perforin-producing CD8⁺ T cells and CD62L^low CD8⁺ T cells. Further, CCR7^low CD8⁺ T cells and CD27^low CD8⁺ T cells were more efficiently produced by BCG-D70M stimulation (Fig. 6A). The efficient production of these CD8⁺ T cells was observed with different doses of BCG; however, in the absence of naive CD4⁺ T cells, these changes were not induced (not shown). Also, CD4⁺ T cells producing intracellular perforin was not produced from naive CD4⁺ T cells by the stimulation with BCG-D70M–infected DCs. These results indicate that BCG-D70M may produce effector CD8⁺ T cells having killing activity and memory type CD8⁺ T cells efficiently from naive population. When BCG is used as a vaccine, it is required to produce memory T cells having a high migratory function (34). To examine this point, we assessed the expression of CD162 molecules on both CD4⁺ T cells and CD8⁺ T cells, which were stimulated with DCs infected with BCG-D70M (Fig. 6B). On BCG-D70M stimulation, both CD4⁺ T cells and CD8⁺ T cells that express high levels of CD162 Ag were produced more efficiently than by the stimulation with BCG-261H. A similar difference between BCG-261H and BCG-D70M was induced when different doses of BCG were used (not shown).

The ability of BCG-D70M to produce T cells highly responsive to the secondary in vitro stimulation was examined by in vivo functional studies (Fig. 7). C57BL/6 mice were s.c. inoculated with 1 × 10³/mouse of rBCGs 4 wk prerestimulation in vitro. Both MMP-II and HSP70 proteins were used as a restimulator. These proteins induced IFN-γ production from T cells in all infected or uninfected mice because they have high immunogenicity, and BCG-Tokyo, a parental strain of all rBCGs, has the gene encoding BCG-derived MMP-II. However, splenic T cells from BCG-D70M–infected mice respond most vigorously to the stimulators and produced higher doses of IFN-γ (Fig. 7A) and IL-2 (not shown) than those from mice uninfected or infected with control rBCGs including BCG-261H, BCG-ΔUT-11-3, and BCG-70M. To examine the long-term effect of the inoculation of rBCGs on the production of such responsive T cells, C57BL/6 mice were s.c. inoculated with 1 × 10³/mouse of rBCGs 12 wk before the restimulation. Again, a significantly higher dose of IFN-γ (Fig. 7B) was produced from splenic T cells obtained from mice inoculated with BCG-D70M by the stimulation with MMP-II and HSP70 than those from mice uninfected or infected with control rBCGs.

C57BL/6 mice vaccinated with either BCG-261H or BCG-D70M (1 × 10³ CFU/mouse) for 4 wk were challenged with 5 × 10³ M. leprae in the footpad. Thirty-two weeks later, the footpad was removed, and the M. leprae recovered from the footpad was enumerated (Fig. 8). A total of 2 × 10⁵ M. leprae were recovered from mice inoculated with PBS and challenged with M. leprae. Although the mice vaccinated with BCG-261H inhibited the multiplication of M. leprae significantly, the BCG-D70M vaccination significantly and more efficiently inhibited the M. leprae multiplication than BCG-261H. A similar difference was observed when 1 × 10² CFU/mouse rBCG was inoculated for the inhibition of M. leprae.

M. leprae is well-known as a representative slow-growing Mycobacterium. Usually, M. leprae needs 12–14 d for one division and at least 2–5 y for the manifestation of the disease. In vivo studies using the immunodeficient nude mouse indicate that adaptive immunities play an important role in inhibiting the multiplication of M. leprae, and the activation of both CD4⁺ T cells and CD8⁺ T cells is an essential element for controlling M. leprae infection (5, 6, 35). Although CD4⁺ T cells chiefly act at the initial phase of infection, the contribution of CD8⁺ T cells in terms of IFN-γ production and killing of mycobacteria-infected host cells is necessary in the chronic phase of the infection (36). BCG was used so far as vaccine against leprosy; however, its efficacy is nowadays considered not as convincing as expected (12). The reason for why BCG cannot prevent the leprosy manifestation convincingly may be due to its inadequate ability to stimulate T cells. The poor T cell-stimulating activity seems to be based on the intrinsic defect of BCG not being able to enter the lysosome feasibly. Also, poor stimulation of T cells would result in the meager production of competent memory T cells, including both CD4⁺ and CD8⁺ subsets, capable of convincingly responding to mycobacterial Ags. Especially, BCG cannot activate naive CD8⁺ T cells adequately in the absence of CD4⁺ T cell-derived help (14), so that BCG may poorly control the disease in the chronic phase or in the inhibition of disease manifestation for a long time postinfection (14). This fact is important when the growth rate of M. leprae is taken into account.

However, BCG has also intrinsic benefit, because it activates human naive CD4⁺ T cells to produce IFN-γ to some extent. Therefore, we tried to improve the potency of BCG, especially with regard to immunostimulatory activities. We chiefly focused on overcoming the defect of BCG—that is, the ability to avoid the fusion of BCG-infected phagosomes with lysosomes. One of the approaches we carried out previously is the production of UreC gene-deficient rBCG (BCG-ΔUT-11-3), which successfully produces acidic phagosomes and facilitates them to fuse with lysosomes (15). In fact, BCG-ΔUT-11-3 efficiently colocalizes with lysosomes and preferentially and effectively stimulates human naive CD4⁺ T cells (15). Therefore, the disruption of the UreC gene of BCG seems to be a useful strategy to translocate BCG to lysosomes. However, unfortunately, BCG-ΔUT-11-3 did not activate naive CD8⁺ T cells effectively. Then, the second approach for overcoming the lack of phagosome–lysosome fusion was carried out—that is, to induce the secretion of immunodominant Ag into phagosome. In this study, we used MMP-II as the immunodominant Ag of M. leprae (7). In one case, sole MMP-II protein (production of BCG-SM) and in the other case HSP70–MMP-II fusion protein (production of BCG-70M) secreting BCG was constructed (16, 25). Both BCGs were quite efficient in the induction of activation of not only naive CD8⁺ T cells, but also naive CD4⁺ T cells. However, BCG-70M was superior to BCG-SM in activating both subsets of T cells, especially naive CD8⁺ T cells (not shown). The activation of naive CD8⁺ T cells by BCG-70M is highly dependent on the secretion of HSP70–MMP-II fusion protein, because the activation seems to be induced by TAP- and proteosome-dependent cross-presentation of the secreted protein (24, 25, 37). Therefore, the secretion of MMP-II in the combination with HSP70 seems to be an efficient strategy to overcome the intrinsic defect of BCG.

Because the two independent strategies for overcoming the intrinsic defect of BCG were useful, we tried to combine both strategies and produced new rBCG (BCG-D70M), in which BCG-ΔUT-11-3 was integrated with gene encoding HSP70–MMP-II fusion protein. As previously reported (25), BCG-70M secreted 92 kDa HSP70–MMP-II fusion protein after being phagocytozed by APCs, and the secreted protein was transported to functional lysosomes. In the phagolysosomes, some portions of HSP70–MMP-II fusion protein could be degraded, but rest of the protein may be sequestrated into the cytosol, where they could be degraded and used for cross-priming CD8⁺ T cells. In this respect, when immature DCs were pretreated with chloroquine and subsequently infected with newly produced BCG-D70M, both the expression of MMP-II and the activation of naive CD4⁺ and CD8⁺ T cells by the rBCG were inhibited. Thus, protein secreted from BCG-D70M seems to be responsible for the activation of naive T cells. Further, the activation of naive CD8⁺ T cells by BCG-D70M was also abolished by pretreatment of immature DCs with lactacystin, a proteosomal protein degradation blocker and brefeldin A that is an inhibitor of antegrade Golgi transportation and of TAP-dependent transportation. Therefore, it is highly likely that the 92-kDa fusion protein secreted from BCG-D70M could be sequestrated into cytosol from lysosome, degraded in proteosome, and used for loading on MHC class I molecules through the TAP-dependent pathway. Thus, similar to BCG-70M, BCG-D70M also used the cytosolic pathway, which is known as the most effective cross-presenting pathway (38), to cross-prime CD8⁺ T cells. In this respect, it is known that HSP plays an important role in the induction of the cytosolic cross-presentation pathway (39, 40). HSP70 secreted as a part of the fusion protein seems to be closely associated with the cross-priming CD8⁺ T cells. The activation of both naive CD8⁺ T cells and naive CD4⁺ T cells by BCG-D70M was induced in an Ag-specific fashion, because treatment of BCG-D70M–infected DCs with mAb to MHC molecules or CD86 Ag inhibited the IFN-γ production from naive T cells. However, the naive CD4⁺ T cells seemed to be polyclonally activated by the stimulation, because the treatment of DCs with mAb to MMP-II partially, but significantly, inhibited the activation (Fig. 2C). In C57BL/6 mice, a single injection of BCG-D70M produced T cells capable of responding to both MMP-II and HSP70 several weeks postinoculation. Therefore, the HSP70–MMP-II fusion protein activated both APCs and T cells by the similar mechanisms as observed in in vitro experiments and was probably used as antigenic molecules in vivo. Because M. leprae-infected DCs expressed MMP-II–derived antigenic determinants on their surface (7, 16), the production of T cells responsive to MMP-II in vivo may be useful to prevent the disease manifestation. This speculation seems to be supported by the present observation that the vaccination with BCG-D70M more efficiently inhibited the multiplication of M. leprae in vivo than that with vector control BCG.

The activities stimulating both subsets of naive T cells of BCG-D70M were strongest among the all rBCGs produced so far including BCG-70M. Although all of the rBCGs showed the dose-dependent effect in the T cell activation, BCG-D70M showed the strongest activity in terms of the T cell activation, even if an MOI 1.0 of BCG was used. Further, BCG-D70M most strongly activated DCs as revealed by IL-12p70 production from DCs. Because BCG-70M activated DCs through the binding of HSP70–MMP-II fusion protein with TLR2 (25), BCG-D70M seems to activate DCs with a similar mechanism, at least partially. However, it did not induce an apoptotic cell death of target APCs including DCs and macrophages, the in vitro growth rate of BCG-D70M was almost identical with that of BCG-261H, and further, the infectivity of these rBCGs to host cells in both in vitro and in vivo was identical (not shown). It is likely that the stronger DC- and T cell-activating ability of BCG-D70M than BCG-70M might be due to the absence of ammonia, products of UreC gene-encoding urease, in the phagosome. The urease depletion may facilitate the translocation of HSP70–MMP-II fusion protein secreted in phagosomes from BCG-D70M into lysosomes. However, another explanation could be that the absence of ammonia may facilitate the translocation of BCG-D70M itself to lysosomes, because it has previously reported that BCG-ΔUT-11-3 more efficiently entered lysosomes than parent BCGs, which possess the UreC gene (15, 19). BCG-D70M translocated into lysosomes or phagolysosomes secreted HSP70–MMP-II fusion protein. Therefore, it can be speculated that a larger dose of secreted protein that could be efficiently processed would be available in lysosomes, so that much or many types of antigenic peptides could be loaded on the MHC molecules. This speculation is important because it has recently been reported that quick activation of CD8⁺ T cells by BCG requires the high antigenic load on MHC class I molecules (41). These results indicate that the deletion of urease from BCG and integration of gene encoding fusion protein into BCG may act synergistically, although further detailed analyses is required.

The strong ability of BCG-D70M to stimulate T cells enables macrophages to activate CD4⁺ T cells. The CD4⁺ T cells stimulated by BCG-D70M through macrophages seemed to be activated in an Ag-specific manner, because the IFN-γ production from the T cells was largely blocked by the treatment of BCG-D70M–infected macrophages with mAbs to MHC class II and CD86 Ags. So far, rBCG including BCG-ΔUT-11-3 and BCG-70M did not activate CD4⁺ T cells efficiently through macrophages in the absence of costimulators such as CD40L and IFN-γ (15, 25). The definite reason for why BCG-D70M, but not BCG-70M, could activate CD4⁺ T cells through macrophages remains unanswered. However, the secreted fusion protein either in the phagosome or phagolysosome could be associated with CD4⁺ T cell activation through macrophages, because pretreatment of macrophages with chloroquine abolished their T cell-stimulating activities. BCG infects not only DCs, but also macrophages, which are highly active in phagocytosis of bacteria; thus, the successful activation of CD4⁺ T cells by macrophages upon an infection with BCG-D70M would provide many chances to heterogeneous CD4⁺ T cells to receive antigenic stimuli. The CD4⁺ T cell activation by macrophages should contribute to the efficient production of high doses of IFN-γ and to the production and maintenance of abundant memory T cells. In addition, in the presence of the help of CD4⁺ T cells, naive CD8⁺ T cells were differentiated into CCR7^lowCD8⁺ and CD27^lowCD8⁺ memory type T cells by the stimulation with BCG-D70M. Also, they produced phenotypically activated CD62L^lowCD8⁺ T cells as well as perforin-producing effector CD8⁺ T cells. Therefore, the efficient activation of naive and memory type CD4⁺ T cells may contribute to the efficient production of effector and memory CD8⁺ T cells. In our hands, we could not confirm the possibility that the functional perforin-producing CD8⁺ T cells, which are produced from naive T cells, can be further differentiated into memory subsets. If this were the case, effector CD8⁺ T cells having killing activity can be immediately and efficiently produced from such memory T cells upon an infection with M. leprae in vivo.

It has been reported that to prevent the disease manifestation induced by infection with mycobacteria, such as M. tuberculosis, by vaccination, the vaccinating agents should be able to produce memory T cells that have a high potency to migrate into the infection site (34). Thus, we evaluated whether BCG-D70M can produce T cells with a migration activity by monitoring the surface expression of CD162 molecules. The stimulation of naive T cells with BCG-D70M–infected DCs induced the expression of CD162 on both CD4⁺ T cells and CD8⁺ T cells. Therefore, it could be assumed that BCG-D70M may be a convincing stimulator of naive T cells.

Taken together, in this study, we newly constructed an rBCG that is deficient in production of urease, but instead produced HSP70–MMP-II fusion protein and is capable of effectively and strongly activating both naive CD4⁺ and CD8⁺ T cells, thus overcoming the intrinsic defect of BCG. Using the triple combination of expressing HSP70 and MMP-II protein in BCG and depletion of urease may result in sufficient production of memory T cells by activating both subsets of naive T cells in human.

We thank N. Makino for help in the preparation of the manuscript. We also thank M. Hasegawa for technical support and the Japanese Red Cross Society for kindly providing PBMCs from healthy donors.

Disclosures The authors have no financial conflicts of interest.