Comparative Evaluation of Low-Molecular-Mass Proteins from Mycobacterium tuberculosis Identifies Members of the ESAT-6 Family as Immunodominant T-Cell Antigens

Bacterial strains.

The Escherichia coli strains used for cloning and expression were One Shot (Invitrogen, Leek, The Netherlands) and XL1 Blue (Stratagene, La Jolla, Calif.). Mycobacterial strains used for the interspecies study are listed in Table 1.

ST-CF and antibodies.

Short-term culture filtrate (ST-CF) of M. tuberculosis H37Rv was produced as described previously (3, 31).

The monoclonal antibody (MAb) PV-2 was described previously (2). Polyclonal antisera were raised against recombinant (r) TB7.3 and CFP10 as follows. Rabbits were immunized five times with 50 μg of recombinant antigen adjuvanted with incomplete Freund's adjuvant at 2-week intervals. The animals were bled, and sera were tested for reactivity against the recombinant protein and ST-CF by Western blotting.

Identification and characterization of low-mass proteins.

TB7.3 was identified from ST-CF as described by Rosenkrands et al. (27). In brief, ST-CF proteins were separated by thiophilic adsorption chromatography on an Affi-T gel column (Kem-En-Tec, Copenhagen, Denmark), and the TB7.3-containing fractions were further purified by anion-exchange chromatography (HR 5/5 Mono Q connected to a fast protein liquid chromatography system; Pharmacia, Uppsala, Sweden). Proteins were eluted by a 0 to 1 M NaCl gradient, and fractions enriched in the band representing TB7.3, when analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), were collected. For the identification of TB10.4 from ST-CF, chromatofocusing on a PBE 94 column equilibrated with 25 mM piperazine-HCl, pH 5.5, and elution with 10% PB74-HCl, pH 4.0, was applied.

The native protein preparations of TB7.3 and TB10.4 were separated on a 10 to 20% Tricine–SDS-PAGE gel and blotted onto Problott polyvinylidene difluoride membranes (Applied Biosystems, Foster City, Calif.), from which the protein bands of interest were excised and subjected to N-terminal amino acid analysis by automated Edman degradation with a Procise 494 sequencer (Applied Biosystems).

The gene encoding TB10.4 was identified by screening a λgt11 M. tuberculosis genome library constructed by R. Young et al. (36) with the MAb PV-2 as described previously (1, 2). A single positive clone (AA242) was found, and the EcoRI M. tuberculosis fragment was subcloned into the E. coli expression vector pBluescript II SK(+) (Stratagene) (AA251) for subsequent sequencing. The obtained nucleotide sequence was analyzed, and the open reading frame (ORF) (tb10.4) encoding the MAb PV-2 reactive protein (TB10.4) was identified.

Obtained sequences were used to search the SWISSPROT and the Sanger sequence databases with the BLASTP and the BLASTN algorithms. DNA star version 3.08b was used for molecular mass and pI calculations.

The proteins were mapped by two-dimensional electrophoresis (2-DE) as described previously (35).

Cloning, expression, and purification of recombinant proteins.

Primers were designed, based on the DNA sequences from the Sanger sequence database, and constructed to create unique restriction sites, up- and downstream of the start and stop codons, respectively, for use in the cloning procedure. The primers were synthesized at Statens Serum Institute by using a ABI-391 DNA synthesizer (Applied Biosystems). The primers used were as follows: tb7.3-sense, AAGAGTAGATCTATGATGGCCGAGGATGTTCGCG (creates a BglII site); tb7.3-antisense, CGGCGACGACGGATCCTACCGCGTCGG (creates a BamHI site); tb10.4-sense, GCAACACCCGGGATGTCGCAAATCATG (creates a SmaI site); and tb10.4-antisense, CTACTAAGCTTGGATCCCTAGCCGCCCCATTTGGCGG (creates a BamHI site). PCR was carried out in a thermal reactor (Rapid cycler; Idaho Technology, Salt Lake City, Utah) by using standard protocols (28). As template, M. tuberculosis H37Rv chromosomal DNA or plasmid DNA was used for the cloning of tb7.3 and tb10.4, respectively. The tb7.3 PCR product was cloned into the pCR2.1 cloning vector and transformed into One Shot cells (Invitrogen) as described by the manufacturer. Plasmid DNA (tb7.3) or PCR product (tb10.4) was digested with the appropriate restriction enzymes and cloned into either pMCT6 (16) (tb7.3) or pMST24 (32) (tb10.4) in frame with eight- or six-histidine residues, respectively. The correct insert was confirmed by sequencing of both DNA strands. DNA sequencing was performed at Statens Serum Institut by using the cycle sequencing system in combination with an automated gel reader (model 373A; Applied Biosystems).

The gene encoding CFP10 was cloned as described previously (9).

The histidine-tagged recombinant proteins (rTB7.3, rTB10.4, and rCFP10) were expressed and purified by metal affinity chromatography by using a Talon column (Clontech, Palo Alto, Calif.) in the presence of 8 M urea, essentially as described by the manufacturer. Purification of the proteins to homogeneity was done by anion chromatography (35) with 1-ml Hitrap columns (Pharmacia).

Protein concentrations were determined by the bicinchoninic acid test (Micro BCA Protein Assay Reagent kit; Pierce, Oud-Beijerland, The Netherlands). Lipopolysaccharide content in these preparations, measured by the Limulus amoebocyte lysate test (7), was always below 0.05 ng of lipopolysaccharide/μg of protein.

Southern blotting.

Genomic DNA from the different mycobacterial species listed in Table 1 was prepared as described previously (3). The Southern blotting was carried out as described elsewhere (22) with the following modifications: 2 μg of purified chromosomal DNA was digested to completion with PvuII and run on a 0.8% agarose gel. The gel was blotted onto a Hybond-N⁺ membrane (Amersham, Life Science, Buckinghamshire, Little Chalfont, England) in a vacuum transfer device (Milliblot-V system; Millipore).

The probe holding the whole ORF of the selected protein was amplified by PCR from plasmid DNA with the primers described above, and the probes were purified by a QIAquick PCR Purification kit (Qiagen, Hilden, Germany) and labeled by using the ECL direct nucleic acid labeling system (Amersham). Hybridization and detection were performed according to the instructions provided by the manufacturer.

Human lymphocyte cultures.

Seventeen Danish TB patients diagnosed and treated at the Department of Pulmonary Medicine, University Hospital of Copenhagen, Copenhagen Denmark, were asked to participate in the study. Blood samples were drawn between 0 and 6 months after diagnosis. Seven BCG-vaccinated and 7 nonvaccinated healthy individuals with no known history of exposure to patients with TB or laboratory exposure were recruited as controls. Blood samples were drawn 2 months to 40 years after BCG vaccination.

Separation, culture of peripheral blood mononuclear cells (PBMC), and measurement of IFN-γ in the supernatants were done as described previously by Ravn et al. (26). A dose-response study of the three recombinant proteins (rTB7.3, rTB10.4, and rCFP10) was carried out by using 0.3 to 10 μg of antigen/ml of culture. ST-CF was used at 5 μg/ml. The detection limit of the IFN-γ assay was 50 pg/ml. Positive responses were defined as delta values (IFN-γ release in the antigen-stimulated well minus IFN-γ release in the unstimulated well) above 200 pg/ml. IFN-γ release in unstimulated wells was generally below 100 pg/ml. All IFN-γ analyses were done in duplicates on supernatants pooled from three wells and were given as means. The variation on duplicate wells was always less than 10% of the mean. This part of the study was approved by the Local Ethical Committee for Copenhagen and Frederiksberg (RH 01-282/96 and KF 01-369/98).

Identification and recombinant expression of low-mass M. tuberculosis proteins.

Despite the high biological activity of the low-mass fraction of culture filtrate, the components in this fraction have mostly remained elusive, as no distinct spots indicating significant quantities of protein can be detected in this region by 2-DE of culture filtrate preparations (30, 35).

We employed two different purification strategies to obtain fractions enriched in low-mass culture filtrate proteins. A 7-kDa protein was isolated by thiophilic adsorption chromatography, followed by anion-exchange chromatography. The 15 N-terminal amino acids of the purified protein were determined (AEDVRAEIVASVLEV) and used to search the Sanger sequence database (http://www.sanger.ac.uk). The search identified an ORF of 216 bp which encoded a protein (TB7.3) with a theoretical molecular mass of 7.3 kDa and a pI of 3.8 (Table 2). TB7.3 was similar to the C terminus of oxaloacetate decarboxylases and biotin carboxyl carrier proteins, and in agreement with this observation, TB7.3 was found to be biotinylated (I. Rosenkrands and P. Andersen, unpublished data).

The second approach was to separate ST-CF by large-scale preparative SDS-PAGE into narrow fractions highly enriched in low-mass components. One of these fractions (4 to 8 kDa) was used to immunize mice, and a MAb (PV-2) (2) directed against a low-mass molecule was obtained. This MAb was used to screen a λgt11 M. tuberculosis genome library, and a phage clone with an insert containing an ORF of 291 bp was identified. The ORF encoded a protein of 96 amino acids (TB10.4) with a theoretical molecular mass of 10.4 kDa and a pI of 4.5. Searching the Sanger database, we identified two other deduced proteins (Rv3017c and Rv3019c) with homology to TB10.4 (Table 2). To confirm the correct identification of TB10.4, we purified the protein recognized by the MAb PV-2 from culture filtrate and obtained the N-terminal sequence (GHAGDMAGYAGTLQS). This sequence corresponds to residues 13 through 27 in TB10.4, and in addition to confirming our identification, it suggests an alternative start site (the Leu [encoded by ttg] in position 12 or the Met in position 11) or a partial cleavage of the protein in culture filtrate. The sequences of TB7.3 and TB10.4 were analyzed by using the Signal P database (http://www.cbs.dtu.dk/service/SignalP) and were not found to encode conventional signal sequences. Interestingly, the database searches identified both TB10.4 and the recently identified low-mass protein CFP10 (9) as members of the ESAT-6 family, which consists of small proteins homologous to ESAT-6 (Fig. 1) and organized in operon-like structures on the mycobacterial genome (9, 12). TB7.3, on the other hand, was found to be a low-molecular-mass antigen outside the ESAT-6 family.

The genes encoding the two newly identified proteins as well as ESAT-6 and CFP10 were expressed and purified as histidine-tagged products.

Characterization of low-mass M. tuberculosis proteins.

CFP10 has not been characterized in detail, and we therefore decided to investigate this molecule in comparison with TB7.3 and TB10.4. The molecules were mapped in a 2-DE reference map of M. tuberculosis ST-CF components (35). This was done by probing 2-DE immunoblots sequentially with MAb PV-2 and the polyclonal antisera directed against TB7.3 (polyclonal antibody [PAb] α-TB7.3) and CFP10 (PAb α-CFP10). The newly identified molecules are present in low quantities in ST-CF, and in particular TB7.3 is difficult to distinguish after silver staining. Western blotting allowed mapping of the molecules to distinct positions around ESAT-6 in the region below 10 kDa. These positions were subsequently transferred to silver-stained duplicate gels and compared to the already characterized proteins in this region (Fig. 2). No reaction was seen to higher molecular mass components in ST-CF, indicating that these molecules are mature proteins and not fragments of larger molecules.

The interspecies distribution of the molecules was evaluated by Southern blotting on genomic DNA from several different mycobacterial species (Table 1), using the coding regions of tb7.3, tb10.4, and cfp10 as probes. The data were compared to the previously published species distribution of esat-6 (15) (Table 1).

The three genes were all detected in M. tuberculosis and M. bovis. The close relationship between cfp10 and esat-6 was confirmed, as these genes were identically distributed. Both genes were absent in the two BCG strains tested and the majority of the environmental mycobacteria.

tb7.3 and tb10.4 were broadly distributed and could be detected in BCG, the Mycobacterium avium complex, and other environmental mycobacteria (Table 1).

Immunological recognition of low-mass M. tuberculosis proteins.

The immunological recognition of the purified low-mass recombinant proteins was evaluated by stimulating PBMC isolated from TB patients, BCG-vaccinated donors, and healthy nonvaccinated donors. A dose-response investigation was conducted for TB7.3, TB10.4, and CFP10 with concentrations ranging from 0.3 to 10 μg/ml (Fig. 3). Figure 3 shows the IFN-γ levels in lymphocyte cultures from two Danish TB patients and two healthy Danish BCG-vaccinated donors stimulated with the antigens. The lymphocyte response after stimulation with TB7.3 was moderate with IFN-γ releases generally below 1,000 pg/ml (Fig. 3A). Neither IFN-γ nor proliferative responses to this antigen (data not shown) reached more than 20% of the responses seen with ST-CF. For the two ESAT-6 family antigens (CFP10 and TB10.4), high levels of IFN-γ were induced with increasing antigen concentrations (Fig. 3B and C). Optimal concentrations of the antigens were between 1.25 and 10 μg/ml, and these concentrations gave responses in the range of 1,000 to 4,000 pg of IFN-γ/ml. The concentration of 5 μg of purified recombinant antigen per ml was chosen for the subsequent comparative evaluation of the three antigens and ESAT-6.

The four low-mass antigens were investigated in 13 to 17 TB patients, 4 to 7 BCG-vaccinated donors, and 7 nonvaccinated donors (Fig. 4). TB7.3 was recognized, but at low levels in both patients and BCG-vaccinated donors. Thirty-eight percent (5 of 13) of the TB patients recognized this molecule at a level significantly above background, and for these donors, the median response was 659 pg of IFN-γ/ml versus 4,024 pg of IFN-γ/ml in the same donors for ST-CF. The ESAT-6 family members were all recognized at a much higher level. TB10.4 was recognized by both BCG-vaccinated donors (71% responders; median IFN-γ = 3,968 pg/ml versus 5,335 pg/ml in the same donors for ST-CF) and TB patients (88% responders; median IFN-γ = 3,298 pg/ml versus 4,707 pg/ml in the same donors for ST-CF). In the TB patients, CFP10 induced a pronounced release of IFN-γ (median IFN-γ = 2,135 pg/ml versus 4,755 pg/ml in the same donors for ST-CF). As would be expected from the species distribution, reactivity to CFP10 and ESAT-6 was TB specific. These two antigens were recognized only by individuals infected with M. tuberculosis and not by BCG-vaccinated and unvaccinated healthy Danes (Fig. 4).

Compared to ESAT-6, TB10.4 induced significantly higher levels of IFN-γ in TB patients (P = 0.0017, Wilcoxon signed-rank test), whereas T-cell responses to CFP10 and ESAT-6 were similar (P = 0.121).

There was no correlation between the responses to the individual antigens in responsive patients, and several patients recognized only one or two of the three ESAT-6 family members.