Key Ags of Mycobacterium tuberculosis initially identified in the context of host responses in healthy purified protein derivative-positive donors and infected C57BL/6 mice were prioritized for the development of a subunit vaccine against tuberculosis. Our lead construct, Mtb72F, codes for a 72-kDa polyprotein genetically linked in tandem in the linear order Mtb32C-Mtb39-Mtb32N. Immunization of C57BL/6 mice with Mtb72F DNA resulted in the generation of IFN-γ responses directed against the first two components of the polyprotein and a strong CD8+ T cell response directed exclusively against Mtb32C. In contrast, immunization of mice with Mtb72F protein formulated in the adjuvant AS02A resulted in the elicitation of a moderate IFN-γ response and a weak CD8+ T cell response to Mtb32c. However, immunization with a formulation of Mtb72F protein in AS01B adjuvant generated a comprehensive and robust immune response, resulting in the elicitation of strong IFN-γ and Ab responses encompassing all three components of the polyprotein vaccine and a strong CD8+ response directed against the same Mtb32C epitope identified by DNA immunization. All three forms of Mtb72F immunization resulted in the protection of C57BL/6 mice against aerosol challenge with a virulent strain of M. tuberculosis. Most importantly, immunization of guinea pigs with Mtb72F, delivered either as DNA or as a rAg-based vaccine, resulted in prolonged survival (>1 year) after aerosol challenge with virulent M. tuberculosis comparable to bacillus Calmette-Guérin immunization. Mtb72F in AS02A formulation is currently in phase I clinical trial, making it the first recombinant tuberculosis vaccine to be tested in humans.
Globally, approximately two billion people are infected with Mycobacterium tuberculosis, the causative agent of tuberculosis (TB),4 and an estimated three million deaths due to this disease occur annually (1, 2) Although combination chemotherapy is generally effective in the treatment of TB, the treatment is arduous and requires stringent compliance to avoid the development of multidrug resistant strains of M. tuberculosis (3, 4).
With respect to prophylaxis against TB, the attenuated strain of Mycobacterium bovis, bacillus Calmette-Guérin (BCG), is currently the only available vaccine (5, 6). However, the prophylactic use of BCG has demonstrated varying levels of efficacy in different clinical trials and geographically distinct populations (6, 7). It has been proposed that the variable efficacy in BCG vaccine trials may be due to interference caused by previous exposure to environmental mycobacteria (8, 9). A further deficiency of BCG is that it can cause disseminated disease in immunocompromised individuals (10, 11). Thus, whereas BCG has a protective effect in children, particularly against forms such as meningeal TB, it does not consistently prevent the development of pulmonary TB in adults. Consequently, there is need for the development of more effective vaccines against TB.
M. tuberculosis is an intracellular pathogen, and, as such, cell-mediated immunity plays a key role in the control of bacterial propagation and subsequent protection against TB. In animal studies, acquired resistance against TB is mediated by sensitized T lymphocytes, in particular, IFN-γ-secreting CD4+ T lymphocytes are critical in mediating protection against TB in the murine model of this disease (12, 13, 14, 15, 16). The central role of IFN-γ in the control of TB has been further demonstrated by the high susceptibility to mycobacterial infections in mice with a disrupted IFN-γ gene and in humans with a mutated IFN-γ receptor (10, 11, 17, 18, 19). In addition, MHC class I-restricted CD8+ T cells may be required for resistance to M. tuberculosis as an alternative source of IFN-γ. Recent studies have suggested that whereas CD8+ T cells do not appear to be critical during acute TB, they may play an important role in preventing the reactivation of latent TB infection (10, 11, 17, 18, 19, 20).
The identification of mycobacterial Ags that preferentially activate T cells to proliferate and secrete IFN-γ by both CD4 and CD8 T cells, is critical to the development of subunit vaccines against TB. We have identified several M. tuberculosis Ags in the context of controlled infection in humans and C57BL/6 mice characterized by their ability to elicit T cell and Ab responses (21, 22, 23, 24, 25, 26, 27, 28). Of these, two proteins, Mtb32 (generated as overlapping amino and carboxyl fragments) and Mtb39 (Rv0125 and Rv1196, respectively) (23, 26), were developed further, and their open reading frames (ORFs) were expressed as a single recombinant polyprotein with a predicted size of 72 kDa (Mtb72F). We report in this study that immunization of C57BL/6 mice with Mtb72F DNA or recombinant protein formulated in two different adjuvant systems, AS01B and AS02A, resulted in the elicitation of differential immune responses (both qualitative and quantitative) to the components of Mtb72F, with the AS01B formulation eliciting the broadest range of immune response. Despite these immunogenicity differences, all three forms of immunization protected mice against TB infection. Most significantly, immunization of guinea pigs with Mtb72F, delivered either as DNA or as a rAg-based vaccine, prolonged the survival of the animals after aerosol challenge with virulent M. tuberculosis.
Materials and Methods
Generation of a tandemly linked ORF encoding Mtb72F
Mtb72F was generated by the sequential linkage in tandem of the ORFs of the ∼14-kDa C-terminal fragment of mtb32 (26) (residues 192–323; 132 aaa) to the full-length ORF of mtb39 (23), followed at the C terminus with the ∼20-kDa N-terminal portion (residues 1–195) of mtb32. These two genes correspond to the ORFs Rv0125 and Rv1196, respectively, as defined in the TubercuList H37Rv database (http://genolist.pasteur.fr/Tuberculist/). This was accomplished using sequence-specific oligonucleotides containing unique restriction sites (EcoRI and EcoRV) and devoid of the authentic stop codons at the C-terminal ends (in the case of Mtb32-C and Mtb-39) by PCR using genomic DNA from the M. tuberculosis strain H37Rv as template. The details of the process were as follows.
Generation of Mtb32c construct devoid of a stop codon
The 5′and 3′ oligonucleotides to the C-terminal portion of mtb32 (mtb32c) were designed as follows: 5′ (5′-CAA TTA CAT ATG CAT CAC CAT CAC CAT CAC ACG GCC GCG TCC GAT AAC TTC-3′) and 3′ (5′-CTA ATC GAA TCC GGC CGG GGG TCC CTC GGC CAA-3′). The 5′ oligonucleotide contains an NdeI restriction site (underlined) preceding an ATG initiation codon, followed by nucleotide sequences encoding six histidine residues (italics) and sequences derived from the first seven amino acid residues of Mtb32c (bold). The 3′ oligonucleotide contains an EcoRI restriction site (underlined), followed immediately by sequences comprising the last seven amino acid residues (bold) and devoid of the termination codon. These oligos were used to amplify Mtb32c, the carboxyl 396-nt portion (aa residues 192–323; a 14-kDa 132 aa residues) of Mtb32 and the resulting PCR-amplified product digested with NdeI and EcoRI, followed by subcloning into the pET17b expression vector similarly digested with NdeI and EcoRI. Ligated products were then transformed into Escherichia coli, and transformants with the correct insert were identified by restriction digest and verified by DNA sequencing. The mtb32c-pET plasmid was subsequently linearized by digestion with EcoRI and EcoRV. The latter cuts within the polylinker sequence of the pET vector that is located downstream of the EcoRI site.
PCR amplification of the full-length coding sequences of Mtb39 and subcloning into the Mtb32c-pET plasmid
The 5′and 3′ oligos of Mtb39 contain the following sequences: 5′ (5′-CTA ATC GAA TTC ATG GTG GAT TTC GGG GCG TTA-3′) and 3′ (5′-CTA ATC GAT ATC GCC GGC TGC CGG AGA ATG CGG-3′). The 5′ oligonucleotide contains an EcoRI restriction site (underlined) preceding the first seven amino acid residues of Mtb39 (bold). The 3′ oligonucleotide contains an EcoRV restriction site (underlined), followed immediately by sequences comprising the last seven amino acid residues (bold) and devoid of the termination codon. These were used to amplify the full-length coding sequence of Mtb39 (1173 bp; a 391-aa stretch with a predicted size of ∼39 kDa), and the resulting PCR-amplified product was digested with EcoRI and EcoRV, followed by subcloning in-frame with the predigested Mtb32c-pET plasmid. The ligated products were then transformed into E. coli, and transformants with the correct insert were identified by restriction digest and verified by DNA sequencing. For expression of the recombinant Mtb32c-Mtb39, the pET plasmid construct was transformed into the bacterial host (BL-21; pLysE), and expression of the protein resulted in a single recombinant.
Cloning of the N-terminal 195-aa sequence of mtb32 into the mtb32c-mtb39 pET construct
The 5′ and 3′ oligonucleotides of the N-terminal fragment of Mtb32 were designed as follows: 5′ (5′-CTA ATC GAT ATC GCC CCG CCG GCC TTG TCG CAG GAC-3′) and 3′ (5′-CTA ATC GAT ATC CTA GGA CGC GGC CGT GTT CAT AC-3′). Both sets of oligonucleotides contain an EcoRV restriction site (underlined) preceding the first eight amino acid residues of Mtb32 (bold) and immediately following the sequences upstream of the stop codon (italics). The 3′ oligonucleotide also includes sequences comprising the last 20 nt (bold) of Mtb32n. They were designed to amplify the N-terminal 585-bp (195-aa residue) portion of Mtb32. The resulting PCR-amplified product was digested with EcoRV, followed by subcloning into the Mtb32c-Mtb39 fusion pET plasmid (similarly digested with EcoRV). Ligated products were then transformed into E. coli, and transformants with the correct insert and orientation were identified by restriction digest and verified by DNA sequencing. The final construct, a 72-kDa polyprotein (Mtb72F), comprises a single ORF organized in the linear order, Mtb32C-Mtb39-Mtb32N.
Recombinant protein expression and purification of Mtb72F
The recombinant (His-Tag) Ag was purified from the insoluble inclusion body of 500 ml of isopropyl-β-d-thiogalactoside (IPTG)-induced batch cultures by affinity chromatography using the one-step QIAexpress Ni-NTA-agarose matrix (Qiagen, Chatsworth, CA) in the presence of 8 M urea. Briefly, 20 ml of an overnight-saturated culture of BL21 containing the pET construct was added to 500 ml of 2× YT medium containing 50 μg/ml ampicillin and 34 μg/ml chloramphenicol and grown at 37°C with shaking. The bacterial cultures were induced with 2 mM IPTG at an OD 560 of 0.3 and grown for an additional 3 h (OD, 1.3–1.9). Cells were harvested from 500-ml batch cultures by centrifugation and resuspended in 20 ml of binding buffer (0.1 M sodium phosphate (pH 8.0) and 10 mM Tris-HCl (pH 8.0)) containing 2 mM PMSF and 20 μg/ml leupeptin. E. coli was lysed by adding 15 mg of lysozyme and rocking for 30 min at 4°C after sonication (four times, 30 s each time), then spun at 12,000 rpm for 30 min to pellet the inclusion bodies.
The inclusion bodies were washed three times in 1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate in 10 mM Tris-HCl (pH 8.0). This step greatly reduced the level of contaminating LPS. The inclusion body was finally solubilized in 20 ml of binding buffer containing 8 M urea, or 8 M urea was added directly to the soluble supernatant. Recombinant Ags with His-Tag residues were batch-bound to Ni-NTA-agarose resin (5 ml of resin/500 ml of inductions) by rocking at room temperature for 1 h, and the complex was passed over a column. The flow through was passed twice over the same column and the column washed three times with 30 ml each of wash buffer (0.1 M sodium phosphate and 10 mM Tris-HCl (pH 6.3)) also containing 8 M urea. Bound protein was eluted with 30 ml of 100 mM imidazole in wash buffer, and 5-ml fractions were collected. Fractions containing the rAg were pooled, dialyzed against 10 mM Tris-HCl (pH 8.0), bound one more time to the Ni-NTA matrix, eluted, and dialyzed in 10 mM Tris-HCl (pH 7.8). The yield of purified recombinant protein varied from 50–75 mg/l induced bacterial culture with >98% purity. Endotoxin levels were typically <10 endotoxin U/mg protein (i.e., <1 ng of LPS/mg).
Mice, immunizations, and cytokine assays
Female C57/BL6 mice were obtained from Charles River and age-matched (4–6 wk) within each experiment. Mice were immunized three times (3 wk apart) with 8 μg of rMtb72F formulated with the adjuvants AS02A or AS01B or with a 100-μg dose of Mtb72F-DNA. For protein formulations, the required immunization dose of rMtb72F (typically 8 μg) was brought up to either 50 or 43 μl with 1× PBS (pH 6.8) and mixed with 50 μl of AS01B or 57 μl of AS02A, respectively. Mice were injected with a total volume of 100 μl/mouse via the i.m. (tibialis) route with 50 μl/leg. Three weeks after the last boost, animals designated for immunogenicity studies were killed, and spleen cells were obtained by conventional procedures. Mononuclear cells were cultured at 37°C in 5% CO2 in the presence of either medium (containing 10% FBS, 50 μM 2β-ME, and 50 μg/ml gentamicin), or medium plus rAg. For cytokine analysis, spleen cells were plated in duplicate 96-well tissue culture plates at 2.5 × 105 cells/well and cultured with or without Ags for 72 h. Supernatants were harvested and analyzed for IFN-γ by a double-sandwich ELISA using specific mAb (BD PharMingen, San Diego, CA) as previously described (22, 27).
An ELISPOT assay was used to determine the relative number of IFN-γ-expressing cells in the single-cell spleen suspensions. A MultiScreen 96-well filtration plate (Millipore, Bedford, MA) was coated with 10 μg/ml rat anti-mouse IFN-γ capture Ab (BD PharMingen) and incubated overnight at 4°C. Plates were washed with PBS, blocked with RPMI 1640 and 10% FBS for at least 1 h at room temperature, and washed again. Spleen cells were plated in duplicate at 1 × 105 cells/well in 100 μl and stimulated with the specific rAg at a 10 μg/ml dose mixed with 0.2 ng/ml IL-2 for 48 h at 37°C. The plates were subsequently washed with PBS and 0.1% Tween and incubated overnight at 4°C with a biotin-conjugated, rat anti-mouse IFN-γ secondary Ab (BD PharMingen) at 5 μg/ml in PBS, 0.5% BSA, and 0.1% Tween. The filters were developed using the Vectastain ABC avidin peroxidase conjugate and Vectastain AEC substrate kits (Vector Laboratories, Burlingame, CA) according to the manufacturer’s protocol. The reaction was stopped by washing the plates with deionized water, plates were dried in the dark, and spots were counted.
IgG isotype ELISA
Mice were bled 3 wk after the last immunization, and sera were stored at −20°C until use. The specific serum IgG isotype Ab response was measured by conventional ELISA. Recombinant Ags were coated onto 96-well Immulon-4 ELISA plates at a concentration of 100 ng/well and incubated overnight at 4°C. The plates were blocked with 200 μl/well PBS containing 0.05% Tween 20 and 1% BSA for 2 h at room temperature and washed with PBS-0.05% Tween, followed by the addition of 50 μl/well of diluent (PBS-0.05% Tween and 0.1% BSA). Sera were added at serial 2-fold dilutions (beginning at a 1/200 dilution) incubated in a humidified box on a rocking platform for 2 h at room temperature and washed (as described above), followed by the addition of 100 μl/well of biotinylated isotype-specific secondary Abs (rat anti-mouse IgG1 or IgG2a; Southern Biotechnology Associates, Birmingham, CA) diluted at 1/4000 in PBST-0.1% BSA. Plates were incubated for 1 h at room temperature, washed, and developed with for ∼5 min with 100 μl/well of tetramethylbenzidine peroxidase substrate (Kirkegaarde & Perry Laboratories, Keene, NH) mixed 1/1 with peroxidase solution B. Reactions were stopped by the addition of 50 μl/well of 1 N H2SO4 and were read on an ELISA plate reader (Dynatech Laboratories, Chantilly, VA) at 450 nm with 570 nm as the reference wavelength.
DNA vaccine and retroviral constructs
The full-length coding sequence of Mtb72F was PCR amplified from the protein expression vector (pET-72F) using primer-specific pairs. Except for the N-terminal six histidine residues, the sequence of the entire ORFs of Mtb72F was identical in the protein and DNA constructs. The 5′ primer was designed to contain a HindIII recognition site and a Kozak sequence upstream of the initiator ATG codon. The resultant PCR product was digested subcloned into the eukaryotic expression vector pJA4304 (gift from J. I. Mullins and J. Arthors, University of Washington School of Medicine, Seattle, WA).
In addition, the three subcomponents of Mtb72F (mtb32-C, Mtb39, and Mtb32-N) were subcloned into the retroviral vector pBIB-X, a retroviral expression vector that contains a selectable marker (bsr) under translation control of an intraribosomal entry site sequence. This vector is under the control of the murine leukemia virus long terminal repeat promoter. The sequences of all three genes were obtained by PCR amplification using 5′ oligonucleotides designed with the initiating methionine. The 3′ oligonucleotide included the stop codon. The 5′ primers also included a Kozak consensus sequence (GCCGCCACC) upstream of the initiation codon to ensure efficient translational initiation in the pBIB-X vector.
Target cells were EL-4 cells retrovirally transduced with mtb32-C, mtb39, and Mtb32N essentially as previously described (27). Briefly, the retroviral constructs (described above) were used in transfections of Phoenix-Ampho, an amphotropic retroviral packaging line. Approximately 48 h post-transfection, supernatants containing recombinant virus were harvested and used to transduce EL-4 cells. Transduction efficiency was measured by flow cytometry using EL-4 transduced with pBIB-enhanced green fluorescent protein viral supernatants as a positive control. All transductants were selected with blastocidin-S (Calbiochem, San Diego, CA) at a concentration of 10 μg/ml. These cells were then used as targets in standard 51Cr release CTL assays. For analysis of effector CTL, spleen cells from immunized mice were cultured in 24-well Costar (Corning Glass, Corning, NY) tissue culture plates at a density of 5 × 106/well and stimulated with 2.5 × 105/well of EL4 cells expressing Mtb-specific Ags in complete medium for 6 days, harvested, and tested in a standard 5-h 51Cr release assay as previously described (27).
The Mtb32c-specific CTL line was generated from spleen cells from mice that were immunized with Mtb72F DNA and restimulated in vitro with EL4 cells transduced with mtb32c. The CTL line was maintained by weekly stimulation of 4 × 105 CTL with 2 × 105 irradiated EL4 transductants and 2.5 × 106 irradiated syngeneic spleen cells in 2 ml of medium containing 1 ng/ml IL-2 in 24-well plates.
CTL epitope mapping and MHC restriction analysis
Peptides used in this study were synthesized using a Symphony system (Rainin Instrument, Woburn, MA) according to the guidelines of the manufacturer. P815 (H-2d) cells and H-2Db and H-2Kb transfectants of P815 were gifts from M. Bevan (University of Washington, Seattle, WA). Either EL4 (H-2b) or P815 (parental or H-2Db or H-2Kb transfected) stimulator cells were pulsed with peptides at various concentrations for 1.5 h at 37°C and then incubated with Mtb32C-specific CTL. After 48 h, culture supernatants were assayed for IFN-γ by ELISA.
Mice were immunized i.m., three times, 3 wk apart, with 8.0 μg of the rAg formulated in the indicated adjuvant or 100 μg of plasmid DNA containing the gene of interest. Positive control mice were immunized with BCG (5 × 104 CFU) in the base of the tail (once), and negative control animals were injected with saline, adjuvant alone, or DNA vector. Thirty days after the last immunization, mice were challenged by low dose aerosol exposure with M. tuberculosis H37Rv strain (ATCC 35718; American Type Culture Collection, Manassas, VA) using a Glas-Col (Terre Haute, IN) aerosol generator calibrated to deliver 50–100 bacteria into the lungs. Four weeks later, mice were euthanized, and lung and spleen homogenates were prepared in PBS/Tween 80 (0.05%). Bacterial counts were determine by plating serial dilutions of individual whole organs on nutrient Middlebrook 7H11 Bacto Agar (BD Biosciences, Cockeysville, MD) and counting bacterial colony formation after 21-day incubation at 37°C in humidified air.
Guinea pigs were similarly immunized via the i.m route, three times, 3 wk apart, with either a 200-μg dose of Mtb72F-DNA (brought up to 250 μl with 1× PBS, pH 7.0) or a 20-μg immunization dose of rMtb72F (or a mixture of the three components on a molar basis) formulated in AS02A in a final volume of 250 μl. The protocol for rMtb72F formulation in AS02A (Ag:adjuvant) was identical with those described for the mouse studies, but was adjusted for a 20-μg immunization dose in a final volume of 250 μl. Animals were immunized with 125 μl of the final formulation per leg. BCG (a single dose of 103 CFU) was used as the positive control and administered via the intradermal route. Negative control groups include adjuvant and saline alone groups. Thirteen weeks after the third immunization, the animals were challenged with the virulent H37Rv strain via the aerosol route by calibrating the nebulizer compartment of the Middlebrook airborne-infection apparatus (using the same Glas-Col aerosol generator used in the mouse experiments) to deliver ∼20–50 bacteria into the lungs. All challenge studies reported in this paper were performed in the same laboratory. Animals were killed at the indicated time. At necropsy, all lung lobes were removed from the thorax individually to enable separate manipulations with each lobe. The number of viable bacteria in the lungs was determined by plating serial 10-fold dilutions of right cranial lung lobe homogenates onto nutrient Middlebrook 7H11 agar, and bacterial colony formations were counted after 21 days of incubation at 37°C under 5% CO2. Data are expressed as log10 of the mean number of bacteria recovered.
Generation, expression, and purification of recombinant Mtb72F
Two Mtb Ags, Mtb32 and Mtb39 (Rv0125 and Rv1196, respectively), previously identified by serological and T cell expression cloning and demonstrated to stimulate CD4 and CD8 responses in PBMC of healthy, purified protein derivative-positive (PPD+) donors and in infected or immunized mice (23, 27), were selected for the development of a TB vaccine. An initial, straightforward approach was to link in tandem the ORFs of both genes such that the final construct would result in the generation of a single rAg comprising both Ags. However, attempts to express and purify Mtb32 in E. coli (either as the full-length or mature form) or as a fusion with Mtb39 (at either the N or C end) were unsuccessful. This was attributed to the inherent serine protease activity of Mtb32 probably leading to toxicity of the expressing host cell (26). In contrast, it was relatively easy to express and purify high levels of two separate fragments of Mtb32 (N- and C-terminal portions). Therefore, we reasoned that it should be feasible to generate a chimeric construct by fusing at either end of Mtb39, the N- and C-terminal portions of Mtb32 (designated Mtb32N and Mtb32C). Our initial experience with the expression level and purification profile of the C-terminal portion of Mtb32 (Mtb32C) revealed that when expressed in E. coli, the recombinant protein partitions into the soluble fraction. In addition, Mtb32C has been demonstrated to drive the high level expression of heterologous proteins in E. coli when fused to their N-terminal end (29). We therefore engineered for expression in E. coli a polyprotein construct organized in the linear order Mtb32C-Mtb39-Mtb32N coding for an ORF with a predicted molecular mass of 72 kDa (Mtb72F).
Fig. 1⇓A is a diagrammatic representation of Mtb72F showing the physical organization and the restriction enzyme sites used to link the three subunits. The length of the predicted ORF of Mtb72F is 2187 nt, coding for a 729-aa polypeptide with a predicted molecular mass of ∼72 kDa, an isoelectric point of 4.98, and a net charge of −15.88 at pH 7.0. Because of the way the construct was generated, the addition of the hinge sequences (EcoRI and EcoRV) resulted in the introduction of six nucleotides (two amino acid residues) at each of the junction sites. In addition, the construct was designed such that the N terminus contained six histidine residues for ease of purification by affinity chromatography over Ni-NTA matrix. After expression in E. coli, the recombinant protein was purified from inclusion bodies with yields ranging from 50–75 mg of purified protein/l induced culture. Fig. 1⇓B shows a Coomassie Blue-stained SDS-PAGE gel of the E. coli culture before and after induction along with the final purified recombinant Mtb72F (migrating at its predicted size of ∼72 kDa).
Immune responses in mice immunized with Mtb72F DNA
To investigate the specificity and breadth of the immune response to Mtb72F in C57BL/6 mice, we initially used the naked DNA approach because this form of immunization is known to stimulate both CD4+ and CD8+ T cell responses. The Mtb72F ORF was subcloned into the eukaryotic expression vector pJA4304, which is under control of the CMV promoter. Plasmid DNA was prepared with >95% migrating as a supercoiled form, as visualized by ethidium bromide staining (not shown). Mice (typically eight to 10/group) were immunized with three doses of 100 μg of Mtb72F-DNA via the i.m. route at 3-wk intervals. Three weeks after the third immunization, three mice from each group were killed for evaluation of anti-Mtb72F Ab and T cell responses (CD4+ and CD8+). IgG1 and IgG2a Ab responses were evaluated by ELISA using specific mouse IgG isotype antibodies. The results revealed that mice immunized with DNA developed an Mtb72F Ab response of the IgG2a, but not the IgG1, subclass, and the response was exclusively directed against the N-terminal (Mtb32C) portion of the molecule (Fig. 2⇓A). No Mtb72F-specific Ab was detected in sera of mice immunized with the empty vector or injected with saline (data not shown).
CD4+ T cell responses were evaluated by stimulating spleen cells in vitro with either the full-length rMtb72F protein or with each of the three separate protein components, rMtb32C, rMtb39, and rMtb32N, at 10, 2, and 0.4 μg/ml Ag. Supernatants were harvested and assayed for IFN-γ at 72 h poststimulation. The results revealed that immunization of C57BL/6 mice with Mtb72F DNA (but not the vector or saline control groups) induced the production of high amounts of IFN-γ (>20 ng/ml) after in vitro stimulation with rMtb72F (Fig. 2⇑B). Stimulation of the same splenocyte culture with each of the three components of Mtb72F revealed that the IFN-γ response is elicited predominantly by sequences comprising the first two components (Mtb32c and Mtb39) of the polyprotein construct, with little or no response directed against the C-terminal portion (Mtb32N) of the molecule. Finally, PPD (at a 10 μg/ml dose) also stimulated the splenocyte cultures from Mtb72F DNA-immunized mice to produce IFN-γ.
Because DNA immunization has been shown to be an effective method for the induction of a CD8+ T cell response, we evaluated whether immunization with Mtb72F DNA could result in the generation of CTL responses. Splenocytes from immunized mice were stimulated for 6 days with EL4 cells that had been transduced with Mtb32C, Mtb32N, or Mtb39. The cells were subsequently washed and evaluated for cytotoxicity against specific targets or EL-4 cells transduced with enhanced green fluorescent protein as a negative control. Fig. 2⇑C shows that immunization of mice with Mtb72F-DNA induced the generation of Mtb72F-specific CTL, and the CTL response was directed exclusively against the N-terminal portion (Mtb32C) of the polyprotein. The lysis of Mtb32C targets was specific to immunized mice, because no lysis was observed in control groups immunized with either saline or the empty vector (data not shown).
Mapping of the CD8 T cell epitope of Mtb32C
We next analyzed the CTL response to Mtb32C in more detail. First, we generated an Mtb32C-specific CTL line from spleen cells of mice immunized with Mtb32C DNA and stimulated in vitro with EL-4 cells retrovirally transduced with Mtb32C. Epitope-mapping studies were performed by stimulation of the Mtb32C-specific CTL with 23 peptides (15-mer overlapping by 10) spanning the entire length of Mtb32C (Fig. 3⇓A). This analysis revealed that the Mtb32c-specific CTL made a strong IFN-γ response to peptide 18 (residues 92-DGAPINSATAMADAL-106) even at the low stimulation dose of 1 ng/ml (Fig. 3⇓A). The adjacent sequence (peptide 17, residues 87-VITAVDGAPINSATA-101) read out at the high dose, but rapidly titrated out at 0.1 μg/ml. Thus, the sequence of overlap between the two positive peptides was identified as DGAPINSATAM. Given these results, the deduced sequence of the CD8 T cell epitope was further dissected into overlapping 9- and 10-mer peptides. In addition, we wanted to determine which MHC allele was responsible for presentation of the peptide to the Mtb32c-specific CD8 T cells. Therefore, we used P815 (H-2d-restricted) cells that had been transfected with either H-2Kb or H-2Db as APCs (Fig. 3⇓, B–D). The data showed that only the peptide-pulsed H-2Db, not H-2Kb-transduced, P815 cells stimulated strong responses by the Mtb32C-specific CTL line. The two peptides that stimulated the strongest IFN-γ responses were the 10-mer (93-GAPINSATAM-102) and 9-mer (94-APINSATAM-102) sequences (Fig. 3⇓C). In agreement with these data, both peptides are predicted to bind with high affinity to Db, as determined by class I MHC prediction algorithms.
Immune responses to Mtb72F protein formulated in two adjuvant systems, AS02A and AS01B
The immune response to Mtb72F protein were subsequently evaluated in two adjuvants, AS01B and AS02A (GlaxoSmithKline Biologicals, Rixensart, Belgium). AS01B contains monophosphoryl lipid A (MPL) and QS-21 in a liposomal formulation, whereas AS02A has the same components formulated in an oil-in-water emulsion. Mice were immunized three times, 3 wk apart, with 8 μg of rMtb72F formulated in AS01B or AS02A via the i.m. route. Three weeks after the third immunization, three animals from each group were killed, and anti-Mtb72F Ab and T cell responses were evaluated. With regard to Ab responses, mice immunized with rMtb72F formulated in either AS01B or AS02A mounted strong and comparable IgG1 and IgG2a responses against Mtb72F (Fig. 4⇓A) and to each of the three components comprising the construct (data not shown). No Mtb72F-specific responses were detected in the adjuvant-alone control groups.
CD4+ T cell responses were evaluated by IFN-γ ELISA of the supernatant culture or by the more sensitive IFN-γ ELISPOT assay. Immunization of mice with Mtb72F formulated in AS02A resulted in the production of a strong IFN-γ response after in vitro stimulation of splenocyte cultures with rMtb72F or rMtb39, a relatively weaker response to Mtb32C, and a low to undetectable response to Mtb32N (Fig. 4⇑B). Immunization of mice with rMtb72F formulated in AS01B stimulated a robust production of IFN-γ (5- to 10-fold higher) against all three components of the polyprotein. Of particular interest, the adjuvant AS01B induced a strong IFN-γ response specific for the Mtb32N portion of the molecule, a response not readily observed after immunization with rMtb72F formulated in AS02A or with Mtb72F-DNA. Qualitatively, the corresponding IFN-γ ELISPOT is in agreement with the ELISA data, in that AS01B stood out as inducing the broadest and most robust T cell response to all three components of Mtb72F (Fig. 4⇑C).
Formulation of Mtb72F in AS01B elicits a robust CD8+ (CTL and IFN-γ) T cell response in C57BL/6 immunized mice
CD8+ T cell responses are typically associated with DNA or viral delivery approaches. In contrast, rAg based formulations do not readily induce such responses. Given that the quality and strength of the immune response to rMtb72F were remarkably affected by the adjuvant system (AS01B or AS02A) used in the formulation, we extended our study to include CD8+ T cell responses to rMtb72F in immunized C57BL/6 mice. Fig. 5⇓A shows that immunization of C57BL/6 mice with rMtb72F formulated in AS01B resulted in a strong CTL response directed exclusively against Mtb32C. As with Mtb72F-DNA immunization, this CTL response was directed against the same CD8 10-mer peptide (residues 92-GAPINSATAM-102; data not shown). In contrast, mice immunized with a formulation of Mtb72F in AS02A mounted a relatively weaker CTL response (Fig. 5⇓B). Spleen cells from rMtb72F-immunized mice were also stimulated with the CTL peptide derived above and assayed for IFN-γ by ELISPOT. In agreement with the cytotoxicity data, strong IFN-γ responses to peptideGAPINSATAM were observed in mice immunized with rMtb72F formulated in AS01B (Fig. 4⇑C). Responses in mice immunized with rMtb72F in AS02A were weaker, but stronger than those in the nonstimulated cultures.
Mtb72F protects C57BL/6 mice against M. tuberculosis infection
Given the results of the above immunogenicity experiments, we sought to determine whether the way in which Mtb72F is delivered (naked DNA or recombinant protein formulated in AS01B or AS02A) impacts the outcome of protection against infection of C57BL/6 mice with M. tuberculosis. Briefly, C57BL/6 mice were immunized three times i.m (at 3-wk intervals) with Mtb72F-DNA (100 μg/dose) or with three doses (8 μg/dose) of recombinant protein formulated in AS01B or AS02A. As controls, groups of mice were immunized with saline, adjuvant, or BCG. Four weeks after the last immunization, mice were challenged via the aerosol route with ∼100 CFU of the virulent M. tuberculosis strain H37Rv. Bacteriological burden (CFU) was measured in the lungs of mice at 4 wk postchallenge. The results from three independent challenge experiments revealed that immunization of C57BL/6 mice with Mtb72F-DNA consistently led to a 0.7- to 1.0-log reduction in bacterial burden approaching the protective efficacy observed with BCG (Fig. 6⇓A). We also compared the protective efficacy of rMtb72F formulated in AS02A with the DNA delivery approach in the same experiment (Fig. 6⇓A). Mtb72F in AS02A has been evaluated exhaustively, and in six independent experiments we found that this immunization with this formulation resulted in a 0.4- to 0.6-log reduction of bacterial burden in the lung. This level of protection was comparable to that seen after immunization with a formulation comprising a mixture of the three components of Mtb72F (data not shown). Interestingly, despite the qualitative and quantitative differences observed in the immune responses to rMtb72F formulated in AS01B Vs AS02A, immunization of mice with three doses of 8 μg of rMtb72F in either of the two adjuvant systems resulted in a comparable reduction in bacterial burden in the lung (∼0.6 log; Fig. 6⇓B).
Mtb72F protects guinea pig against aerosol challenge with virulent M. tuberculosis
Having demonstrated that in the mouse model immunization with Mtb72F-DNA or rMtb72F formulated in AS02A protected against TB, we next evaluated whether these vaccination approaches would also protect against TB infection in the guinea pig model. Groups of five guinea pigs were immunized with three doses each of 200 μg of Mtb72F-DNA or 20 μg of rMtb72F formulated in AS02A. Control groups were immunized with AS02A adjuvant alone or saline via the i.m. route or with a standard dose of 103 CFU BCG administered via the i.d. route. Thirteen weeks after the third immunization, the animals were aerosol-challenged with 20–50 CFU of the virulent Mtb strain H37Rv. Protection was monitored by outward signs of infection (difficulty in breathing and weight loss), with survival as an end point. One of the five animals in the BCG control group died at ∼7 wk postchallenge, but of causes not related to TB infection. Therefore, in the case of the BCG control group, the percent survival was based on n = 4. The results from this study revealed that at 30 wk postchallenge (Fig. 7⇓) although all animals in the saline and adjuvant control groups were moribund and had to be euthanized, three of four (75%) of the guinea pigs immunized with BCG, four of five (80%) of those immunized with rMtb72F, and three of five (60%) of the animals immunized with Mtb72F-DNA were still alive. At 40 wk (∼3 mo after all animals in the control groups succumbed), three of five (60%) of the animals in each of the Mtb72F-vaccinated groups were still alive. By 70 wk (∼15 mo) postchallenge, two of five (40%) animals in the DNA-immunized groups and one of four (25%) animals in the BCG group were still alive. Taken together, the results revealed that in the guinea pig model, Mtb72F, delivered either as naked DNA or recombinant protein formulated in AS02A, protected guinea pigs against virulent TB challenge to an extent comparable to that seen with BCG and for periods lasting >1 year.
The development of an efficacious subunit-based recombinant vaccine against TB would require a multivalent mixture of Ags for a broad coverage of a heterogeneous MHC population. Despite the fact that a single Ag could by itself induce protection in inbred strains of mice, a mixture comprising several Ags is conceivably a better vaccine for applications in humans because it is less likely to suffer from MHC-related unresponsiveness in a heterogeneous population. In fact, using in vitro responses (proliferative and IFN-γ production) of PBMC from healthy PPD+ donors as a measure of the extent of coverage of infected responders to a panel of defined TB Ags, a broad-range coverage could only be accomplished with Ag combinations. Therefore, although the murine model is an important first step to determine the nature of the immune responses as well as the protective capacity of Ags, this model does not necessarily predict the outcome in the context of human MHC restriction.
Although the nature of an effective immune response to TB is incompletely understood, particularly in humans, the most effective vaccination strategies in animal models are those that stimulate T cell responses, both CD4+ and CD8+, to produce Th1-associated cytokines. Therefore, formulations that induce the production of enduring Th1 responses are desirable and probably are an essential element of a successful vaccine. Of about a dozen T cell Ags initially identified in the context of host response to infection in infected donors and C57BL/6 mice (21, 22, 23, 24, 25, 26, 27, 28), we prioritized Mtb32 (Rv0125) and Mtb39 (Rv1196) as our lead candidates for the development of a TB vaccine. Their potential was further corroborated in animal protection studies suggesting that the combination of Ags was more effective at protecting mice and guinea pigs than the individual subunits (not shown). However, from a practical standpoint for the development of a TB vaccine for developing countries, a vaccine consisting of multiple recombinant proteins would be too expensive to manufacture and formulate. For this reason, together with the difficulties encountered in the expression of a stable form of the full-length secreted version of Mtb32, we engineered for expression in E. coli a genetic fusion construct encoding a 72-kDa polyprotein in a contiguous ORF organized in the linear order Mtb32C-Mtb39-Mtb32N (Mtb72F). In addition to simplifying the manufacturing process, we reasoned that immunization with a single construct may ensure equivalent uptake of the components by APCs and, in turn, generate an immune response that is broadly specific.
Because genetic delivery approaches of vaccine candidates are efficient methods for the elicitation of both CD4+ and CD8+ T cell responses and, as well, an effective vaccination protocol for conferring protection against several infectious disease targets, we used the naked DNA approach to initially determine the nature of the immune responses and the protective efficacy of Mtb72F in C57BL/6 mice. Indeed, we found that immunization of C57BL/6 mice with Mtb72F-DNA elicited both CD4+ and CD8+ IFN-γ responses. The CD4+ T cell response was predominantly directed against Mtb39, whereas the CD8+ T cell response (as assessed by both IFN-γ and CTL) was directed against a single epitope within Mtb32C. We next evaluated the immune responses and protective efficacy of rMtb72F formulated in two novel adjuvant systems, AS01B and AS02A (30, 31, 32, 33, 34).
The active ingredients of both adjuvants are 3-deacylated MPL (3D-MPL) (35, 36, 37, 38), a nontoxic derivative of LPS and QS21 (a triterpene glycoside purified from the bark of Quillaja saponaria, (39, 40, 41), and both components have a good clinical safety record (31, 32, 42, 43). The biological properties of MPL are attributed to its immunostimulatory effects on the innate immune system (via activation of the Toll-like receptor 4) and the direct activation of APCs resulting in enhanced phagocytosis and microbicidal activities as a consequence of the production of IL-12, TNF-α, GM-CSF, and IFN-γ (36, 38, 44, 45). The immunostimulant activity of QS21 extract resides in the saponic fraction and acylation appears to be critical for adjuvant activity (46). QS21 promotes both humoral and cell-mediated immunity when added to parenteral or mucosal vaccine formulations (46). Analysis of cytokine secretion by Ag-specific T cells demonstrated that QS21 augmented Th1 and Th2 responses, whereas addition of 3D-MPL resulted in preferential induction of type 1 T cells (46, 47, 48, 49). The distinction between AS01B and AS02A formulations resides in their liposome or oil-in-water emulsion properties, respectively. Both adjuvants as well as their components are currently under clinical evaluation for various vaccines and has been tested in thousands of patients in several clinical trials, including infectious disease vaccines such as malaria (31, 33), hepatitis B (50, 51), and allergy desensitization (52, 53, 54). Furthermore, the use of MPL-stable emulsion as an alternate adjuvant to IL-12, known for its Th1-inducing properties, in conjunction with a polyprotein Ag was recently demonstrated as a safe and effective vaccine against Leishmania infection (55).
The immunogenicity studies revealed that immunization of C57BL/6 mice with rMtb72F formulated in AS01B elicited an immune response profile that is stronger and broader than those observed with AS02A formulation or with naked DNA delivery. AS01B stood out at eliciting a robust and comprehensive (CD4+,CD8+ T cells and Ab) immune response profile encompassing all three components of the vaccine. Therefore, with an appropriate adjuvant formulation, it is, in fact, possible to generate a CD8+ T cell response (typically associated with DNA or recombinant viral delivery systems) using a recombinant-based vaccine. In addition, with AS01B, a relatively robust CD4+ T cell IFN-γ response was generated against the N-terminal portion of Mtb72F (Mtb32C). This was not detected after immunization with rMtb72F formulated in AS02A or with naked DNA or a recombinant Mtb72F adenovirus. Given this unique aspect of AS01B, experiments are currently underway to determine the strength of AS01B by varying the Ag dose and coupling the immunological readout with protection against TB challenge.
There is a general consensus that the guinea pig is currently the best animal model of human TB, and for this reason, it has been widely used as a model to evaluate new vaccines and vaccine delivery approaches. The central reason for this assumption is the fact that guinea pigs develop granulomas similar to those seen in humans with active TB. Guinea pigs can be used for both short term protection studies (CFU) or as longer term disease models (56, 57, 58, 59). As lung tissue necrosis progresses, guinea pigs, like humans, begin to undergo weight loss and eventually die from TB. Successful immunization leads to reduced necrosis, with small lesions characterized by infiltrating lymphocytes, decreased weight loss, and prolonged survival (56, 60). In this model, BCG protects guinea pigs from disease and death for periods of >1 year compared with nonvaccinated animals (56, 58, 60, 61). Thus, more recent and rigorous vaccine evaluation in guinea pigs has focused on using disease and survival as end points instead of only short term bacterial counts, which are not always predictive of long term disease outcome. Of importance to the development of Mtb72F as a vaccine against TB, we report that in this model the delivery of Mtb72F as recombinant protein or DNA protected guinea pigs from death against infection with TB. The extent of protection with Mtb72F represents the longest documented survival end point reported to date for any defined subunit vaccine. In addition, given that the animals in this study were aerosol-challenged with TB at ∼13 wk after the third immunization (instead of the standard 4–6 wk after the third immunization), the protective outcome confirms that the immune responses elicited by Mtb72F, delivered as either rAg formulated in AS02A or naked DNA, are long lasting. The protection with Mtb72F DNA immunization was comparable to that with BCG, whereas the efficacy of rMtb72F formulated in AS02A approached that of BCG. At the time these experiments were initiated, the adjuvant AS01B was not available for evaluation. Based upon more recent data obtained with AS01B, experiments are currently underway in the guinea pig model aimed at comparing the protective efficacy of rMtb72F formulated in either AS01B or AS02A.
Even though DNA and viral delivery approaches are good and effective first methods for the screening of vaccine candidates, particularly in small animal models, the apparent ineffectiveness of most DNA vaccines in humans and nonhuman primates and, from a regulatory standpoint, the lack of a sufficient safety record on DNA vaccines make the recombinant subunit-plus-adjuvant approach more desirable. Further, there is sufficient proof of concept for the latter approach given that it has already been demonstrated to be safe and efficacious in several human clinical trials. Therefore, the use of AS01B and AS02A in conjunction with the feasibility of manufacturing Mtb72F polyprotein under GMP conditions represent major developments toward the realization of an affordable and safe vaccine against TB. The rMtb72F formulated in AS02A adjuvant is currently in phase I clinical trial in the U.S., making it the first recombinant TB protein vaccine ever to be tested in humans.
We thank Patrick McGowan for his critical review of this manuscript.
↵1 This work was supported in part by National Institutes of Health Grants AI49505, AI44373, AI43528, AI75320, AI43528, and AI45707.
↵2 Address correspondence and reprint requests to Dr. Yasir A. W. Skeiky at the current address: Aeras Global TB Vaccine Foundation, 7500 Old Georgetown Road, Suite 800, Bethesda, MD 20814. E-mail address:
↵3 Current address: The Forsyth Institute, Boston, MA 02115.
↵4 Abbreviations used in this paper: TB, tuberculosis; BCG, bacillus Calmette-Guérin; 3D-MPL, 3-deacylated monophosphoryl lipid A; IPTG, isopropyl-β-d-thiogalactoside; ORF, open reading frame; PPD, purified protein derivative.
- Received January 9, 2004.
- Accepted April 13, 2003.
- Copyright © 2004 by The American Association of Immunologists