Exchanging ESAT6 with TB10.4 in an Ag85B Fusion Molecule-Based Tuberculosis Subunit Vaccine: Efficient Protection and ESAT6-Based Sensitive Monitoring of Vaccine Efficacy ¹

Tuberculosis (TB) ³ is a re-emerging disease that remains one of the leading causes of morbidity and mortality in humans, and it represents a major public health problem in many countries (1). The current vaccine against Mycobacterium tuberculosis, Mycobacterium bovis bacillus Calmette-Guérin (BCG), has been extensively evaluated and demonstrated variable protective efficacies ranging from 0 to 85% in different field trials (2, 3, 4). A major contributor to this is thought to be the fact that the protective efficacy of BCG wanes significantly over a period of 10–15 years (5). Therefore, although BCG is efficient against severe forms of childhood tuberculosis (6, 7, 8), it is of limited use against adult pulmonary disease. An improved second-generation vaccine is therefore urgently needed, that can act as an efficient prophylactic vaccine and/or a vaccine that can boost immunity in BCG-vaccinated individuals. Alternative strategies in TB vaccine development, such as subunit vaccines (9, 10, 11, 12), genetic immunization (13, 14), and attenuated strains of M. tuberculosis (15), are currently being explored in many laboratories.

M. tuberculosis expresses and secretes three closely related mycolyl transferases of 30–32 kDa mass, also known as the Ag85 protein complex (Ag85A, -85B, and -85C). Both Ag85A and Ag85B have been shown to be among the most potent Ag species yet identified; they are major targets of human T cell responses to M. tuberculosis and leading vaccine candidates (13, 16, 17, 18, 19, 20, 21). Ag85B has been shown to induce partial protection in murine models of infection (13, 20). In guinea pigs vaccination with purified Ag85B protein also induces substantial protective immunity against aerosol challenge with M. tuberculosis (11), and a rBCG vaccine expressing and secreting the Ag85B protein (rBCG30) induced stronger protective immunity against aerosol challenge with M. tuberculosis than conventional BCG vaccine (22). In addition, a vaccine based on recombinant modified vaccinia virus Ankara expressing Ag85A (MVA85A) was shown to significantly boost BCG-primed and naturally acquired antimycobacterial immunity in humans (23).

Due to the complexity of the host immune response against tuberculosis and the genetic restriction imposed by MHC molecules, it has become clear that an effective subunit vaccine containing multiple epitopes may be required to ensure a broad coverage of a genetically heterogeneous population. Recently, we showed that vaccination with a fusion protein consisting of Ag85B and ESAT6 (Hybrid1) promoted a strong immune response, which is highly protective against TB in the mouse, guinea pig, and non-human primate models (10, 11, 24). This fusion Ag is also effective if delivered in a viral vector or as a DNA vaccine (25). Importantly, Hybrid1 was more protective in both mouse and guinea pig animal models than either of the single components (25). However, because the strongly immunodominant ESAT-6 Ag is an extremely valuable diagnostic reagent and the basis of a number of commercial diagnostic tests (26, 27, 28), finding a vaccine as effective as Hybrid1 that does not contain ESAT-6 is a high priority.

The ESAT-6 family is comprised of 22 low molecular mass proteins, some of which can be divided further into subfamilies due to high sequence relatedness of the individual genes and many of which are strongly immunogenic. We recently identified one such subfamily containing the proteins TB10.4, TB10.3, and TB12.9 (29, 30). Interestingly, we found that TB10.4 is strongly recognized by BCG-vaccinated donors, and in TB patients TB10.4 was even more strongly recognized than ESAT-6 (30), suggesting TB10.4 may be an ideal candidate to replace ESAT-6.

The purpose of the present study was to evaluate the potential of TB10.4 as a substitute for ESAT-6 in the fusion protein, Ag85B-ESAT-6. We show that TB10.4 is strongly immunogenic and induces partial protection against TB. More importantly, when fused to Ag85B, the resulting construct induced protective immunity at the same level as BCG or Hybrid1 vaccination and at the same time allowed the use of ESAT-6, not only as a diagnostic reagent, but also as a reagent for monitoring vaccine efficacy.

Studies were performed with 8- to 12-wk-old female C57BL/6 × BALB/c F₁ mice, purchased from Taconic Farms. Infected animals were housed in cages contained within laminar flow safety enclosures in a BSL-3 facility. The use of mice was conducted in accordance with the regulations set forward by the Danish Ministry of Justice and animal protection committees and in compliance with European Community Directive 86/609 and the U.S. Association for Laboratory Animal Care recommendations for the care and use of laboratory animals.

M. tuberculosis H37Rv and Erdman were both grown at 37°C on Löwenstein-Jensen medium or in suspension in Sauton medium enriched with 0.5% sodium pyruvate and 0.5% glucose.

Mice were immunized three times at 2-wk intervals s.c. on the back with experimental vaccines containing 5 μg of Ag85B, ESAT-6, TB10.4, Ag85B-ESAT-6, or Ag85B-TB10.4/dose, emulsified in dimethyl dioctadecyl ammonium bromide (DDA; 250 μg/dose; Eastman Kodak) with monophosphoryl lipid A (MPL; 25 μg/dose; Avanti Polar Lipids) in a volume of 0.2 ml. The adjuvans was prepared as follows. MPL was mixed into sterile water containing 0.2% triethylamine. The mixture was heated in a 70°C water bath for 30 s and then sonicated for 30 s. The heating and sonicating procedure was repeated twice. The MPL was mixed with DDA immediately before use.

At the time of the first subunit vaccination, one group of mice received a single dose of BCG Danish 1331 (2.5 × 10⁵ CFU) injected s.c. at the base of the tail. Mice were challenged 10 wk after the first vaccination.

When challenged by the aerosol route, the animals were infected with ∼100 CFU of M. tuberculosis Erdman/mouse. These mice were killed 6 wk after challenge. Numbers of bacteria in the liver, spleen, or lung were determined by serial 3-fold dilutions of individual whole-organ homogenates in duplicate on 7H11 medium. Organs from the BCG-vaccinated animals were grown on medium supplemented with 2 μg of 2-thiophene-carboxylic acid hydrazide/ml to selectively inhibit the growth of the residual BCG bacteria in the test organs. Colonies were counted after 2–3 wk of incubation at 37°C. Protective efficacies are expressed as log₁₀ bacterial counts in immunized mice compared with bacterial counts in the adjuvant controls. All results are based on groups of five animals.

The plasmid encoding Ag85B-TB10.4 was generated by linkage of the coding regions of the ∼10-kDa TB10.4 polypeptide to the COOH-terminal end of the ∼30-kDa mature Ag85B polypeptide and fusing this directly to a translational initial Met-Lys peptide in the context of the expression vector pQE60 (Qiagen). These two genes correspond to the coding regions of Rv1886c (Ag85B) and Rv0288 (TB10.4), respectively (31), except that the 51-bp sequence encoding the NH₂-terminal 17 aa residues of Ag85B have been changed without changing the identity of the translated product to destabilize the secondary structure of the RNA transcript for improved translation initiation in Escherichia coli (32). Initially, the open reading frame of Ag85B was generated by PCR amplification from a plasmid DNA encoding Ag85B/ESAT6 (10), the 5′-specific primer (CTCTCATGAAATTCTCACGTCCAGGACTTCCTGTCGAATACCTCCAAGTGCCATCTCCATCAATGGGCCGCGACATCAAGGTTCAGTTCC), and the 3′-specific primer (TATAAGGATCCTATGCGAACATCCCAGTGACGTTGCC). The fragment was directionally cloned into the NcoI-BamHI sites of pQE60 as a BspHI-BamHI fragment. The open reading frame of TB10.4 was generated by PCR amplification from M. tuberculosis H37Rv chromosomal DNA and the 5′-specific primer (GCATGGCGCCGGCATGTCGCAAATCATGTACAACTACC) and the 3′-specific primer (GCATAAGCTTCTAGCCGCCCCATTTGGCGGCTTCGGCCG). This fragment was cloned into the unique NarI-HindIII sites of the previously constructed pQE60/Ag85B plasmid. The NarI site is situated just upstream the 3′ end of the Ag85B open reading frame. The final plasmid construct was transformed into the E. coli strain NF1830 for production of Ag85B-TB10.4. The genotype of NF1830 is galUK ΔlacX74 rpsL thi recA1 araD139Δ (araABOIC-leu)7679 F′ proAB⁺ lacIq1 lacZ::Tn5 lacY⁺. The sequence of the Ag85B-TB10.4 DNA insert was verified by sequencing using the dideoxy chain termination method. Sequences were analyzed with Vector NTI Suite software package (InfoMax).

E. coli NF1830 expressing the recombinant expression plasmid was grown in culture flasks to an OD of ∼0.5 at 600 nm before induction with 1 mM isopropyl β-d-thiogalactopyranoside. After induction, growth was continued for 4 h at 37°C. Cells were harvested by centrifugation at 15,000 × g for 10 min at 4°C and were stored at −20°C.

Frozen cells were suspended in a 2.5 ml/g cell paste in buffer A (25 mM Tris-HCl (pH 7.75) and 4 M urea) and subjected to cell disruption for 3 min at 0°C using a BeadBeater (BioSpec Products) and 106-μm glass beads (Sigma-Aldrich) according to the manufacturer’s instructions. The homogenized mixture was diluted in 5 vol of buffer A, and the insoluble material containing the Ag85B-TB10.4 aggregated in inclusion bodies was washed three times for 1 h at 4°C. The insoluble material was finally precipitated by centrifugation at 15,000 × g for 10 min at 4°C, and Ag85B-TB10.4 was solubilized and extracted for 2 h at 4°C in 1 vol of buffer B (25 mM Tris-HCl (pH 7.75) and 8 M guanidium chloride). Two successive chromatographic purification steps were involved in purification of Ag85B-TB10.4. The initial gel filtration was performed on a Sephacryl S-300 (Pharmacia Biotech) column using buffer B as eluent. Fractions containing Ag85B-TB10.4 were pooled and further purified by an immobilized metal affinity chromatography purification step using HiTrap Chelating HP column (Amersham Biosciences) loaded with Cu²⁺. Ag85B-TB10.4 was eluted with elution buffer (25 mM Tris-HCl (pH 7.75), 6 M guanidium chloride, and 1 M NH₄Cl) using a linear gradient. Fractions containing Ag85B-TB10.4 were identified by SDS-PAGE, pooled, and dialyzed to the final storage buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 40% glycerol). The concentration of the purified Ag85B-TB10.4 was determined by bicinchoninic acid test using micro bicinchoninic acid protein assay reagent kit (Pierce).

Lymphocytes from spleens were obtained as described previously (33). Blood lymphocytes (PBMCs) were purified on a density gradient. Cells pooled from three to five mice in each experiment were cultured in microtiter wells (96-well plates; Nunc) containing 2 × 10⁵ cells in a volume of 200 μl of RPMI 1640 supplemented with 5 × 10⁻⁵ M 2-ME, 1% (v/v) premixed penicillin-streptomycin solution (Invitrogen Life Technologies), 1 mM glutamine, and 5% (v/v) FCS. Based on previous dose-response investigations, the mycobacterial Ags were all used at 5 μg/ml, whereas Con A was used at a concentration of 1 μg/ml as a positive control for cell viability. All preparations were tested in cell cultures and found to be nontoxic at the concentrations used in the present study. Supernatants were harvested from cultures after 72 h of incubation for the investigation of IFN-γ.

As previously described (30), synthetic overlapping peptides (16- or 18-mer) covering the complete primary structure of the three proteins were synthesized by standard solid phase methods on an ABIMED peptide synthesizer at the Department of Infectious Diseases and Immunohematology/Bloodbank C5-P, Leiden University Medical Center (TB10.4), or at Schafer-N. The peptides were purified by reverse phase HPLC. Purified peptides were lyophilized and stored dry until reconstitution in PBS.

Microtiter plates (96 well; Maxisorb; Nunc) were coated with monoclonal hamster anti-murine IFN-γ (Genzyme) in PBS at 4°C. Free binding sites were blocked with 1% (w/v) BSA-0.05% Tween 20. Culture supernatants were tested in triplicate, and IFN-γ was detected with a biotin-labeled rat anti-murine mAb (clone XMG1.2; BD Pharmingen). rIFN-γ (BD Pharmingen) was used as a standard.

The ELISPOT technique has been described previously (34). Briefly, 96-well microtiter plates (Maxisorp; Nunc) were coated with 2.5 μg of monoclonal hamster anti-murine IFN-γ/well (Genzyme). Free binding sites were blocked with BSA, followed by washing with PBS-0.05% Tween 20. Analyses were always conducted on cells pooled from three to five mice. Cells were stimulated with 5 μg of ESAT-6 or TB10.4/ml in modified RPMI 1640 for 48 h. The cells were removed by washing, and the site of cytokine secretion was detected with a biotin-labeled rat anti-murine IFN-γ mAb (clone XMG1.2; BD Pharmingen) and phosphatase-conjugated streptavidin (Jackson ImmunoResearch Laboratories). The enzyme reaction was developed with 5-bromo-4-chloro-3-indolylphosphate (Sigma-Aldrich). Blue spots were counted microscopically. The correlation between the number of cells per well, and the number of spots was linear at concentrations of 2 × 10⁵ to 2.5 × 10³ cells/well. Wells with <10 spots were not used for calculations.

Cells were isolated, as described above, from the blood and spleen of mice. Cells (2 × 10⁵) were stimulated overnight with 2 μg/ml TB10.4 or 10 μg/ml peptide and subsequently incubated for 6 h with 10 μg/ml brefeldin A (Sigma-Aldrich). Thereafter, cells were incubated with Fc-block (BD Pharmingen), washed in FACS buffer (PBS containing 0.1% sodium azide and 1% FCS), and stained for surface markers using CD4-PerCP and CD8-allophycocyanin (BD Pharmingen). Cells were then washed in PBS, permeabilized using the Cytofix/Cytoperm kit (BD Pharmingen), and stained intracellularly with PE-labeled anti-IFN-γ mAb. After washing, cells were finally resuspended in FACS buffer and analyzed by FACS (BD Biosciences).

Assessment of experiments was conducted using ANOVA. Differences between means were assessed by Tukey’s test. A value of p < 0.05 was considered significant. The computer program PRISM (GraphPad) was used for these calculations.

Both TB10.4 and ESAT-6 are part of the same gene family (29), but only the gene for TB10.4 is present in BCG. The genome region containing ESAT-6 was lost as one of the first deletion mutations during the attenuation of BCG (35). We therefore initially analyzed the recognition of TB10.4 and ESAT-6 in M. tuberculosis-infected and BCG-vaccinated mice. The latter was important, because the ability to boost BCG-generated immunity is considered a desirable function for any vaccine to be used against TB. C57BL/6 × BALB/c mice were vaccinated with BCG or challenged with 100 CFU of M. tuberculosis via the aerosol route. At the indicated time points after vaccination or infection, lymphocytes obtained from spleens of infected or BCG-vaccinated mice were cultured 3 days in the presence of recombinant TB10.4 or ESAT-6 and analyzed by ELISPOT for the presence of IFN-γ-producing cells. The results showed that TB10.4 and ESAT-6 were equally well recognized in M. tuberculosis-infected mice (Fig. 1,A). However, in BCG-vaccinated mice, ESAT-6 (as expected) was not recognized, although stimulation with TB10.4 induced a strong response (Fig. 1 B).

We next measured the recognition of TB10.4 and ESAT-6 after immunizing with these Ags. TB10.4 or ESAT-6 was emulsified in DDA and MPL, an adjuvant that has previously been used successfully to induce a highly efficient Th1 response protective against TB (33, 36). The vaccine was given three times at 2-wk intervals, and the immune response induced in the spleen (or blood, data not shown) was investigated 2 wk after the last booster injection. TB10.4 and ESAT-6 (in DDA/MPL) were both highly immunogenic and strongly recognized in spleen and blood (Fig. 1 C and data not shown).

To further characterize the response against TB10.4 after immunization, we next analyzed the pattern of epitope recognition after immunization with the soluble protein. To achieve this, overlapping 16- to 18-aa peptides (10-aa overlap) covering the entire TB10.4 molecule were synthesized (30). Mice were immunized with TB10.4 mixed with the adjuvant DDA/MPL, and lymphocytes from immunized mice were cultured 3 days in the presence of either medium or each of the peptides covering the entire TB10.4 molecule and analyzed by FACS for the presence of IFN-γ-producing cells. These results clearly showed that P3, P7, and P8 primarily stimulated CD4 cells to express IFN-γ (Fig. 2 A).

Analysis of IFN-γ secretion by ELISA confirmed these results (data not shown). Furthermore, the same appeared to be true for cell activation, because CD4 cells stimulated with these three peptides actively up-regulated the activation marker CD69 whereas CD8 cells did not (Fig. 2 B and data not shown).

We next analyzed the protective efficacy of a vaccine based on TB10.4 compared with one containing ESAT-6. Because the vaccines BCG and Hybrid1 (Ag85B-ESAT-6) have demonstrated high efficacy against TB infection in animal models (37), these were included in the experiment as a gold standard against which efficacy can be assessed (25). Mice were vaccinated with 5 μg of TB10.4, ESAT-6, or Hybrid1 in DDA/MPL or with BCG. Ten weeks after the first vaccination, the mice were challenged by the aerosol route with virulent M. tuberculosis. Six weeks after challenge, the mice were killed, and the bacterial burden (CFU) was measured in the lung (Fig. 3). The results showed that vaccination with TB10.4 reduced the bacterial burden of M. tuberculosis by 0.6 log₁₀± 0.15 CFU compared with naive infected mice (p < 0.001), an amount that was not significantly different from the protection induced by ESAT-6 (p > 0.05). Vaccinating with Hybrid1 or BCG reduced the bacterial numbers (CFU) in the lung by ∼1 log₁₀, in agreement with previous results (37) (Fig. 3). In both the H1 and BCG groups, the protection differed significantly (p > 0.05) from that in groups vaccinated with TB10.4 or ESAT-6). Thus, TB10.4 equals ESAT-6 in terms of both immunogenicity and protective efficacy.

Because TB10.4 proved equal to ESAT-6 in all the above experiments, it constituted an interesting potential substitution candidate for ESAT-6. We therefore produced TB10.4 as a recombinant fusion protein with Ag85B and analyzed the protective efficacy of this new vaccine, Ag85B-TB10.4 (Fig. 4,A). Ag85B-TB10.4 was recombinantly produced in E. coli. The fusion protein was purified by a combination of gel filtration and immobilized metal affinity chromatography. Ag85B-TB10.4 was purified to homogeneity and finally analyzed by SDS-PAGE together with the individual proteins, Ag85B and TB10.4 (Fig. 4, B and C). To analyze the immunogenicity of Ag85B-TB10.4 and to clarify whether both components of the fusion protein were recognized by the immune system after processing, groups of mice were immunized with the fusion protein or with the single components, Ag85B or TB10.4, emulsified in MPL and DDA. As negative control, a group of mice received the adjuvant combination alone (data not shown). One week after the last injection, mice were bled, and IFN-γ release was evaluated after in vitro stimulation of purified PBMCs with different concentrations of Ag85B and TB10.4 (5, 1.25, and 0.25 μg/ml; Fig. 5). Immunization with either Ag85B-TB10.4 or Ag85B induced a strong IFN-γ response upon restimulation with Ag85B, indicating that the fusion to TB10.4 did not significantly affect the recognition of Ag85B by the immune system (Fig. 5). Interestingly, when TB10.4 was fused to Ag85B, it was even more immunogenic than the TB10.4 protein alone (Fig. 5).

We next compared the recognition of TB10.4 and ESAT-6 when these two molecules were fused to Ag85B. Mice were immunized with either Ag85B-TB10.4 or Hybrid1. and 1 or 2 wk after the last immunization, cells from blood or spleen were cultured in the presence of TB10.4 or ESAT-6 and analyzed by ELISA for IFN-γ secretion. The results showed that TB10.4 is more strongly recognized than ESAT-6 when fused to Ag85B, even though the two Ags displayed similar levels of immunogenicity when given as separate proteins in DDA/MPL (Fig. 6 and data not shown). A comparison of the immunogenicity of Ag85B in Ag85B-TB10.4 or Hybrid1 showed that Ag85B was strongly recognized after immunization with both fusion proteins (Fig. 6).

In conclusion, the fusion between Ag85B and TB10.4 did not reduce the response against either of the proteins. In fact, immunizing with Ag85B-TB10.4 may increase the number of TB10.4-specific T cells compared with immunizing with only TB10.4.

One of the requirements for any new vaccine is that it should be at least equal in terms of protection to the best candidates already identified. We therefore compared Ag85B-TB10.4 to Hybrid1 in terms of protective efficacy. Additionally, the protective efficacy of the fusion protein Ag85B-TB10.4 was compared with that of a simple mix of Ag85B and TB10.4 or the single components given separately. The molar concentrations of Ag85B and TB10.4 in the mixture were adjusted to be the same level as the concentrations of the two components in the fusion protein. A group of mice receiving the adjuvant combination alone and a group of naive mice were included as controls, and as a positive control for protection, mice were immunized once with BCG. Ten weeks after the first vaccination, the mice were challenged by the aerosol route with virulent M. tuberculosis. Six weeks after challenge, the mice were killed, and the bacterial numbers were determined in the lungs. All vaccinated groups showed a protection significantly different from that of the naive group (p < 0.001). Ag85B-TB10.4 induced high levels of protection (1.06 ± 0.02 log₁₀ CFU reduction in the lung), comparable to that induced by BCG or Hybrid1 (Fig. 7) and significantly higher than that induced by Ag85B (p < 0.05) or TB10.4 (p < 0.001). Although not statistically significant in this experiment, it was generally observed that the mixture of Ag85B and TB10.4 induced slightly lower protection than Ag85B-TB10.4. Thus, the fusion of TB10.4 and Ag85B constitutes a very effective prophylactic vaccine against infection with M. tuberculosis.

ESAT6 is a well-known diagnostic marker of ongoing infection, and recent work in both clinical settings and cattle indicated a correlation between the magnitude of ESAT-6 responses and the extent of the disease (38, 39). We therefore investigated the magnitude of ESAT-6 responses after infection in animals vaccinated with vaccines providing different levels of protection against M. tuberculosis. This experiment took advantage of the fact that through the development of Ag85B-TB10.4 we could reserve ESAT6 as a diagnostic/infection marker. Ten weeks after the first vaccination, the mice were challenged by the aerosol route with virulent M. tuberculosis. Twenty days after challenge, mice were bled, and IFN-γ release was evaluated after in vitro stimulation of purified PBMCs with ESAT-6. Six weeks after challenge, mice were killed, and bacterial numbers (log₁₀ CFU) were determined in the lungs. The results showed that the response against ESAT-6 was indeed highest in infected nonvaccinated mice (Fig. 8,A). In mice vaccinated with Ag85B or TB10.4, we observed a decreased ESAT-6 response after challenge, which declined even further in mice vaccinated with BCG and Ag85B-TB10.4. Importantly, by comparing the ESAT-6 response with the subsequent protection against TB (Fig. 8), a strong correlation (correlation coefficient = 0.83) was observed between the postchallenge ESAT-6 response and the subsequent outcome of the disease, and thereby an inverse relationship to the protective efficacy of the vaccines. Thus, the magnitude of the response against ESAT-6 appears to be an accurate correlate of disease development/vaccine efficacy.

The availability of the M. tuberculosis genome sequence and the current efforts to sequence a large number of additional mycobacterial genomes have set the stage for postgenomic approaches to the identification of novel Ags. One such novel Ag is the recently identified TB10.4 (29), a protein of unknown function. The lack of diversity in the TB10.4 sequence originating from 13 clinical isolates of M. tuberculosis obtained from different geographical locations (30) suggests that it has an important, but yet unidentified, biological function. The expression of TB10.4 is not restricted to M. tuberculosis, because it is recognized by T cells from BCG-vaccinated individuals as well as TB patients (30).

In the present study we have evaluated the potential of TB10.4 as a subunit vaccine against TB. Initially, we showed that T cells specific for TB10.4 were generated after the infection of mice with M. tuberculosis or BCG (Fig. 1). Immunizing with TB10.4 induced a strong CD4 T cell response against epitopes contained within the peptides P3, P7, and P8 (Fig. 2). Moreover, the protective efficacy of TB10.4 (in DDA/MPL) was comparable to that of ESAT-6 (Fig. 3). We have previously shown that the protective efficacy of a subunit vaccine based on the fusion of ESAT-6 and Ag85B was significantly greater than that of a vaccine based solely on ESAT-6 or Ag85B (10, 25). Not all Ags tested show such additive effects, and it is for this reason that ESAT-6 was used in the Hybrid1 vaccine despite its well-known utility as a diagnostic reagent (26, 27, 28). Because there is obviously value in reserving ESAT-6 for diagnostic purposes and based on the encouraging preliminary results in mice (Fig. 1) and humans (30), we evaluated TB10.4 as a substitute for ESAT-6 as a fusion partner for Ag85B (Fig. 4).

The subunit vaccine composed of Ag85B-TB10.4 and DDA/MPL did generate a strong, specific immune response in mice. Importantly, both components of Ag85B-TB10.4 were recognized to at least the same degree as after vaccination based on the single components (Fig. 5). In fact, careful comparative studies showed that TB10.4 was more strongly recognized after immunization with Ag85B-TB10.4 than with TB10.4 alone. This is in contrast to results obtained with ESAT-6, which appears to be subdominant when fused to Ag85B (Fig. 6). Vaccinating mice with Ag85B-TB10.4 induced a level of protection against TB comparable to that produced by BCG and better than that achieved by vaccinating with either of the single proteins. Ag85B-ESAT6 also induced a protection comparable to that of Ag85B-TB10.4 and better than that with either Ag85B or ESAT-6 (Fig. 7) (11). Taken together, the fusion of TB10.4 and Ag85B did not decrease the immunogenicity of either TB10.4 or Ag85B, and linking to Ag85B may, in fact, have a beneficial effect on the immunogenicity of TB10.4. In addition, Ag85B-TB10.4 formulated in DDA/MPL constituted an efficient vaccine against TB.

Using a multicomponent vaccine may generate a broad, strongly recognized response that would be beneficial for vaccination of genetically diverse human populations. Even though TB10.4 is recognized by cells from most TB patients (30), there is still the possibility that a vaccine based solely on TB10.4 (or any other single, nonessential protein, for that matter) may select for M. tuberculosis bacteria that do not express TB10.4. In line with this, it could be speculated that the bacteria may be able to compensate for the apparent lack of TB10.4 sequence polymorphism by having duplicated the genes encoding major T cell Ags, leading to several copies (homologues) of proteins that can substitute each other functionally, but differ in their immunodominant epitopes. Tightly controlled expression of these homologues may result in Ag variation and immune evasion. The probability of such an evasion mechanism is probably decreased when using a multicomponent vaccine. Our study clearly demonstrates that a subunit vaccine based on a fusion protein between Ag85B and TB10.4 is able to induce efficient protection against TB. Moreover, because both Ag85B and TB10.4 are expressed by BCG, it could be speculated that Ag85B-TB10.4 is an ideal candidate to boost BCG-generated immunity and thus increase protection against TB compared with BCG alone. Experiments are presently ongoing in our laboratory by which we aim to examine the potential of Ag85B-TB10.4 to be used as a BCG booster vaccine.

Another interesting point in the present work was the inverse relationship between the postchallenge ESAT-6 response and the protective efficacy of the vaccines tested. Recent work in a model of TB in cattle (38) or in nonhuman primates (24) demonstrates that the magnitude of the in vitro response against the M. tuberculosis virulence factor ESAT-6 after infection correlated very well with bacterial load and also with disease-associated pathology. This suggests that a most accurate correlate of vaccine efficacy is not necessarily the immune response after vaccination, but the response after infection, and this is supported by a recent study (25). Data from human studies are consistent with this; individuals recently exposed to M. tuberculosis who subsequently made strong responses against ESAT-6 were almost 10 times more likely to progress rapidly to clinical tuberculosis than were low responders (39). All these data can be explained by hypothesizing that the ESAT-6 response after infection correlates with Ag load and therefore, ultimately, with bacterial load. Not surprisingly, if bacterial replication is restricted by an effective memory immune response, it would then be predicted that the response to ESAT-6 postinfection would be reduced. The data shown in Fig. 8 confirm this quite strikingly. Because none of the vaccines shown in this experiment contains ESAT-6, the lymphocytes responsive to this Ag must result from M. tuberculosis infection (and, indeed, uninfected controls do not respond to ESAT-6; data not shown). There is a perfect correlation between restriction of bacterial growth (protection) and diminished ESAT-6 response postinfection.

If this finding proves to be robust, it opens the way for monitoring vaccine efficacy in clinical trials. One of the greatest challenges facing phase III clinical trials is the long incubation period between infection and development of overt disease and the fact that the majority of infected individuals develop latent infections (40). Because the clinical end point used has traditionally been overt, symptomatic TB, this has meant that clinical trials have needed very large cohorts and extended periods of follow-up, rendering them almost prohibitively expensive. If it proves feasible to measure the rate of productive infection, rather than the rate of subsequent disease, this would dramatically improve our ability to conduct clinical trials in a timely fashion.

The technical help of Charlotte Fjordager, Lene Rasmussen, Bente Sølvig, and Karina Fürst Andersen is gratefully acknowledged. We thank Dr. Niels Fiil (Institute of Microbiology, Copenhagen, Denmark) for use of the E. coli production strain NF1830. We thank Charlotte Green Jensen for construction of the Ag85B-TB10.4 production vector.

The authors have no financial conflict of interest.