Optimizing Escherichia coli as a protein expression platform to produce Mycobacterium tuberculosis immunogenic proteins

Production and purification of Trx-TB10.4 and TB10.4

The synthetic cDNA coding for TB10.4 (GenBank Accession no. CAA17363.1) was optimized to match the codon usage for E. coli expression (see Additional file 1: Figures S1 and S2). Codon usage optimization was essential since E. coli is a low G + C content (~ 50%) Gram-negative bacterium while M. tuberculosis is a high G + C (> 65%) Gram-positive actinobacterium. The cDNA was subcloned into pET32b plasmid in frame with nucleotidic sequence coding for thioredoxin (Trx), with a His₆-tag sequence and with a sequence recognized by enterokinase (EK) – these sequences are located at the N-terminus of TB10.4 (Figure 1). Trx is a 12 kDa protein that remarkably increases the solubility of fusion proteins [13]. The chimeric Trx-TB10.4 protein (261 amino acids, molecular mass 28.1 kDa, Table 1 and Figure 1) was produced in E. coli using BL21(DE3) strains. Basal expression was performed in LB medium, adding 0.1 mM IPTG when OD_600nm reached 0.6-0.8 and collecting cells after additional 4 hours of growth at 18°C: using these conditions, a fairly good amount of chimeric protein was produced as soluble form and about 2 mg/L of > 85% pure Trx-TB10.4 was purified after a single-step purification on nickel-affinity chelating column. Optimization of the expression conditions (i.e. growing cells for 16 hours at 18°C after IPTG addition) increased significantly the production of the soluble form of the chimeric protein: 12 mg/L of > 95% pure Trx-TB10.4 was purified by metal-chelating chromatography (Table 2, see Additional file 2: Figures S3A, B and C). Mass spectrometry (MS) spectrum of the fusion protein reported in Figure 2A confirmed the identity and purity of the fused protein, indicating a molecular mass corresponding to the isoform lacking methionine at the N-terminal.

Complete cleavage of Trx-TB10.4 fusion protein was obtained using 3 U of recombinant enterokinase (rEK) per mg at 20°C for 16 hours. The cleaved products (Trx and TB10.4) were separated by metal-chelating chromatography (see Additional file 2: Figure S3D and E). The expected mature recombinant TB10.4 is a 103 amino acid long protein (Table 1), bearing a seven amino acid residue pre-sequence (AMAISDP) at the N-terminal region originating from the cloning procedure (Figure 1). MS analysis of TB10.4 preparation (Figure 2B) showed the presence of the full-length TB10.4, with an average mass of 11076.0 Da (nominal mass 11076.36 Da), and an additional product (accounting for ca. 40% of the total protein) with a calculated average mass of 10776.0 Da. This mass shift (-300 Da) corresponds to the lack of three C-terminal amino acids (WGG) from the mature TB10.4. This secondary cleavage might be due to the AEAAK sequence – resembling the EK consensus sequence (i.e. DDDDK) – that is located just before the three C-terminal amino acids. When the TB10.4 preparation was further incubated for 24 hours at 37°C, additional degradation products appeared, whose formation was instead avoided adding a specific EK inhibitor: this result indicates a persisting proteolytic activity by residual EK not (completely) removed by the chromatography on HiTrap chelating column following the cleavage step (see above). Formation of the truncated form was avoided using an optimized rEK (Tag · Off High Activity rEK) – 2 U per mg of substrate protein, 6 hours of incubation at 25°C – and removing rEK using EKapture Agarose (Figure 2C). Pure mature TB10.4 was recovered (Table 2) and its stability at room temperature was checked by MS analysis: following the optimized cleavage procedure, no secondary (truncated) forms were apparent. When the same preparation was further incubated for 24 hours at 37°C in the absence of an EK inhibitor ca. 74% of the mature TB10.4 form was still present (Figure 2D).

Production and purification of Trx-Ag85B and Ag85B

The synthetic cDNA coding for the mature form (285 amino acids) of Ag85B lacking the region encoding for the N-terminal 40 residues corresponding to the putative signal peptide (UniProtKB/Swiss-Prot Accession no. P0C5B9.1; PDB Accession no. 1F0N_A), also optimized for E. coli expression, was subcloned into pET32b plasmid, in frame with nucleotidic sequence coding for Trx, with a His₆-tag sequence and with a sequence recognized by EK – these sequences are located at the N-terminus of Ag85B (Figure 1). The Trx-Ag85B chimeric protein is 450 amino acid long and its molecular mass is 48.4 kDa (Table 1 and Figure 1, see also Additional file 1: Figures S1 and S2). Using the same standard conditions set up for Trx-TB10.4, a very low amount of soluble Trx-Ag85B was produced and a low amount of pure protein (< 1 mg/L) was isolated by metal chelating chromatography. In order to improve the amount of protein produced, a number of expression conditions - using different growth media (LB, TB, SB and minimal medium M9), adding 0.1 or 1.0 mM IPTG when OD_600nm reached 0.7, 2 or after overnight growth (saturation condition) - were tested. In all cases, following IPTG addition, cells were grown at 18°C and collected after 4 hours, and the expression level was assessed by Western blot using anti-His-tag antibody. The largest amount of soluble Trx-Ag85B was obtained growing cells in SB medium and inducing expression with 0.1 mM IPTG at an OD_600nm = 2 (see Additional file 2: Figure S4). Further improvement was obtained growing cells for additional 16 hours (overnight) after IPTG addition. Using these conditions, ca. 25-30% of the chimeric protein was soluble (Additional file 2: Figure S5B and C) corresponding to a productivity of 20 mg/L: 9 mg/L of Trx-Ag85B (> 95% purity degree) were isolated after a single-step purification on nickel-affinity column chromatography (Table 2, see Additional file 2: Figure S5A, B and C). MS spectrum of the fusion protein confirmed its identity and purity (Figure 3A).

Complete maturation of the fusion protein was obtained using 3 U of rEK per mg of Trx-Ag85B at 20°C for 16 hours (Table 2, see Additional file 2: Figure S5D and E). The mature Ag85B obtained by rEK cleavage is a 292 amino acid long protein (molecular mass: 31.3 kDa, see Table 1 and Figure 1), bearing at the N-terminal end a pre-sequence (AMAISDP) originating from the cloning procedure. MS analysis confirmed the identity and purity of the mature Ag85B, with a determined mass of 31346.0 Da (nominal mass 31347.9 Da, Figure 3B).

Production and purification of TB10.4-Ag85B (full) protein

The synthetic cDNA coding for the fusion protein obtained by the combination of the nucleotide sequences of TB10.4 and Ag85B (namely, full protein) was subcloned into pET32b plasmid, in frame with the sequence coding for Trx, with a His₆-tag sequence and with a sequence recognized by rEK (useful to remove Trx). The chimeric protein Trx-TB10.4-Ag85B (Trx-full) is 546 amino acid long (molecular mass is 58.8 kDa, Table 1 and Figure 1). Since the aim of this work was to produce soluble antigens, the TB10.4-Ag85B fusion protein was preferred to the Ag85B-TB10.4 version because the latter chimera was never produced in a significant amount as soluble protein [6, 12]. The solubility of Trx-TB10.4-Ag85B protein was reported to be >10-fold increased compared to that of Trx-Ag85B-TB10.4 [12].

Trials for the expression of Trx-full were conducted using the same conditions set up for the single antigenic proteins. Western blot analysis of the produced Trx-full in cell extracts showed that a consistent fraction (80-90%) of the fusion protein accumulated into the insoluble fraction, being only traces of the protein detectable in the soluble fraction (and together with proteolytic products, data not shown). Co-expression of different chaperon proteins (AraB/DnaK/DnaJ/GrpE or GroES/GroEL/TiG) did not increase the solubility of the fusion protein. Purification of soluble Trx-full yielded less than 1 mg/L, with purity inferior to 50% (data not shown).

To obtain higher amounts of the full protein as soluble form, the chimeric cDNA was inserted into pColdI plasmid, a system based on low-temperature expression gene (cold shock gene) specifically designed to improve the solubility in E. coli of heterologous proteins [14]. Two different gene constructs were prepared, named His-full 1 and His-full 2: in both cases, the full protein bears the His-tag at the N-terminus without thioredoxin (Figure 1). His-full 1 is a 414 amino acid long protein (molecular mass of 44.9 kDa, Table 1 and Figure 1, see also Additional file 1: Figure S2), still bearing at the N-terminal region the sequence recognized by EK. Expression trials for His-full 1 were conducted in BL21(DE3) and BL21(DE3)pLysS E. coli strains, growing cells in LB, TB or SB media at 37°C, adding 1.0 mM IPTG when cells reached early/middle exponential growth phase (OD_{600 nm} = 0.6-0.7 for LB and 1.2-1.5 for TB and SB) or middle/late exponential growth phase (OD_{600 nm} = 1.2-1.5 for LB and 4–5 for TB and SB). Cells were then grown for additional 20 hours at 15°C prior harvesting. Western blot analysis of cells extracts showed the major part of the expressed His-full 1 (ca. 80%) accumulated as IBs. A maximum yield of 1.3 mg/L of His-full 1 with ca. 85% purity was obtained after a single-step purification (on HiTrap Chelating chromatography) from cells grown in TB medium added with 5 g/L of NaCl and by adding further 25 g/L of NaCl at the moment of induction with IPTG, according to a protocol previously developed in our lab [15] (data not shown). Incubation of purified His-full 1 with EK in the same condition used for Trx-Ag85B (3 U/mg His-full 1, for 16 hours at 20°C) yielded a consistent amount of proteolytic products, indicating a high sensitivity to proteolysis of the chimeric protein (not shown).

In order to eliminate the EK cleavage step and to reduce the length of the additional sequence at the N-terminal end of the chimeric protein, site-directed mutagenesis of the full cDNA introduced a Nde I restriction site allowing its cloning in pColdI plasmid using Nde I and Eco RI restriction sites (Figure 1). By this strategy, a 397 amino acid long protein (namely His-full 2, molecular mass: 43.06 kDa, Table 1 and Figure 1) was designed. His-full 2 contains 16 additional amino acids including the His₆-tag sequence (MNHKVHHHHHHIEGRH, see Additional file 2: Figure S2) at the N-terminal end vs. the 33 amino acids present in the previous TB10.4-Ag85B chimeric proteins.

Expression trials carried out on His-full 2 showed that, similarly to Trx-Ag85B (see the paragraph above), the best expression conditions were based on the use of SB medium and induction with 0.1 mM IPTG at OD_600nm = 2. An increase in the production of the recombinant protein was obtained using NaCl in the medium (10 g/L) and by adding further 25 g/L of NaCl simultaneously to IPTG (not shown). As reported in Table 2 and explained below, His-full 2 production increased significantly when scaled-up to a 3 L bioreactor. However, the major part of the fusion protein again accumulated as IBs (see Additional file 2: Figure S6B and C, that refer to cells produced at bioreactor scale). About 0.5 mg/L of His-full 2 with a purity of ca. 90% was obtained after the HiTrap Chelating chromatography performed under optimized conditions (linear gradient from 100 to 250 mM imidazole in 3-column volumes, see Additional file 2: Figure S6A and C). Despite the lower yield than for the His-full 1, this second chimeric protein had the advantage to skip the EK cleavage step. MALDI-TOF/TOF MS spectrum of the chimeric protein is given in Figure 4. The two principal ions observed correspond to the doubly charged natrium adduct (at 21557.68 m/z) and to the doubly charged natrium/synapinic acid adduct (21668.3 m/z) of the His-full 2 protein, confirming its identity.

Scaling up of protein antigen production

After optimization at the shaken flask-scale, processes and production media were scaled up in 3 L bioreactor. As shown by the parameters controlled on line (dO₂ and pH) and by densitometric analysis of cell growth during batch cultivation (Figure 5), the growth of recombinant E. coli BL21(DE3) cells holding the pColdI-His-full 2 was faster than those transformed by pET32b-Trx-TB10.4 and pET32b-Trx-Ag85B, even if a lower temperature was adopted after induction of protein expression (15°C for His-full 2 production vs. 18°C for Trx-TB10.4 and Trx-Ag85B). In 6 hours from inoculum, maximum biomass was achieved and dO₂ was completely depleted in cells producing His-full 2. In Trx-Ag85B production, oxygen depletion occurred after 8 hours from inoculum and the phase of oxygen limitation lasted for further ten hours. E. coli Trx-TB10.4-producing cells showed a similar time course of oxygen consumption, but the levels of dO₂ never decreased below 50% of saturation, indicating a minor respiratory activity; growth of the culture was slower and reached a lower level of biomass production (in wet weight: 19.5 and 8.5 g/L were recovered after 24 hours from cells producing Trx-Ag85B or Trx-TB10.4, respectively, see Table 2). For both Trx-Ag85B or Trx-TB10.4 producing cells, after an initial slight decrease, pH tended to alkalinization in the following fermentation phase. In cells producing His-full 2, a marked acidification of medium and a rapid re-increase in oxygen level after the initial sharp reduction occurred, probably reflecting a stress status arresting cell proliferation after IPTG addition and growth temperature reduction. Notwithstanding the higher OD_600nm of pColdI-His-full 2 containing cells, final biomass was only 6.5 g/L in wet weight (Table 2).

Table 2 shows that recombinant E. coli cells produced more biomass in the stirred bioreactors than in the shaken flasks, due to the better oxygen distribution in the larger scale system which sustains higher density cultivation. Application of the same protocols of metal-chelating chromatography described above to purify Trx-Ag85B, Trx-TB10.4 or His-full 2 gave a significantly increased purification yield from biomass generated in bioreactor than from shaken flasks: 9.4 mg pure Trx-TB10.4, 1.5 mg pure Trx-Ag85B and 0.38 mg pure His-full 2 per gram of biomass were achieved at bioreactor scale compared to figures of 2.5 mg Trx-TB10.4, 0.85 mg Trx-Ag85B and 0.076 mg His-full 2 per gram of biomass obtained at flask-scale. These data indicate a beneficial effect of scaling up on both the volumetric productivity (mg of protein per liter of cultivation broth) and the specific productivity (mg of protein per gram of biomass). In the case of His-full 2, protein purification was feasible only at the bioreactor scale, due to low specific productivity of cells grown in flasks: the soluble form of His-full 2 increased from 5 to ca. 15% of the total expressed chimeric protein shifting from flasks to bioreactor. In the case of Trx-TB10.4 and Trx-Ag85B, the step of EK digestion and mature protein recovery was not affected by the scaling up, being of about one third for TB10.4 and 50% for Ag85B independently if biomass was generated in flasks or in bioreactors. Identity of the fused and mature antigens was confirmed by MS analyses (spectra overlap with those reported in Figure 2A and C, Figure 3A and B, and Figure 4).

Biochemical and immunological characterization of the protein antigens

The quality of the recombinant antigenic proteins was assessed in terms of purity by SDS-PAGE (Additional file 2: Figure S7) and MS analyses (see above) and in terms of protein conformation by circular dichroism (CD) spectroscopy. The secondary structure of pure Ag85B (as determined by the far-UV CD spectra reported in Figure 6A) is composed of both α-helices and β-sheets (ca. 26 and 23% as estimated by k2D2 software vs. 37 and 20% from the crystal structure, pdb code 1F0N), while for TB10.4 the signal for α-helices only is apparent, in good agreement with the known 3D structure (55% α-helix content, pdb code 1F0N) and previous analyses [12]. Similarly, the signal for the tertiary structure significantly differs for the two M. tuberculosis antigens, see Figure 6B. The near- and far-UV spectra of His-full 2 resemble those of Ag85B, according to the fact that Ag85B molecular mass is 3-fold higher than that of TB10.4 (Table 1). All together, spectral analyses indicate the acquisition of a well defined protein conformation for all the recombinant soluble antigenic proteins.

Immunological activity of the recombinant proteins was then assessed by the capability of the proteins to be recognised as antigen by sera (ELISA) or T cells (ELISPOT) of active-TB patients. In a first set of ELISA tests, variable concentrations of the antigens (100, 200 and 400 ng) were adsorbed and tested with TB sera at 1:100 dilution. For the Ag85B and the His-full 2 proteins, a dose dependent reactivity of TB sera vs. the concentration of the antigen adsorbed on the ELISA plate was observed (Figure 7). On the contrary, almost no reactivity was observed for the TB10.4 protein, in agreement with previous studies [6–8]. Furthermore, in order to assess the specificity of the observed reactivity against the M. tuberculosis recombinant Ag85B and His-full 2 protein antigens, antibody titre was determined for each control and active-TB serum by end final point dilution test. As shown in Additional file 3: Figure S8, TB sera present a significant higher antibody titre (median 600) than controls (median 100, p < 0.001) for the Ag85B protein antigen. Similar results were obtained with the chimeric His-full 2 protein (TB patients median titre 1200, controls median titre 200, p < 0.001). These data were further confirmed by analysing the single control and active-TB sera reactivity at 1:200 dilution, as reported in Additional file 3: Figure S9.

Figure 8 shows the results of the T-cell reactivity observed in active TB-patients and healthy controls. Control and active-TB subjects presented a similar T-cell response capability for the Staphylococcus aureus Enterotoxin B (SEB) superantigen used as positive control. For all of the M. tuberculosis recombinant proteins tested, a significant higher number of IFN-γ producing effector T-cells were observed in active-TB patients compared to MTB-unexposed controls -Ag85B: controls 312 ± 363 (mean ± std) spots/million peripheral blood mononuclear cells (PBMCs), active-TB 2361 ± 1863; p < 0.0001; TB10.4: controls 184 ± 209, active-TB 2009 ± 1901; p < 0.0001; His-full 2: controls 701 ± 859, active-TB 22183 ± 1173; p = 0.0381 - confirming that the cell mediated immune response is effectively triggered by the antigens produced during this study.

Design, synthesis and cloning of cDNA coding for antigenic proteins

Synthetic cDNAs coding for TB10.4 and Ag85B antigenic proteins were designed in silico by back translation of the amino acid sequences reported in the database (TB10.4: GenBank Accession no. CAA17363.1; Ag85B: UniProtKB/Swiss-Prot Accession no. P0C5B9.1). A third cDNA, in which the nucleotide sequence coding for TB10.4 is upstream to the sequence coding for Ag85B (full cDNA) was also designed. All synthetic cDNAs were produced by Eurofins Medigenomix GmbH (Ebersberg, Germany), after optimization of the nucleotidic sequences for expression in E. coli (see Additional file 2: Figure S1). The three synthetic genes were cloned into pET32b using Bam HI and Eco RI restriction sites. By this cloning strategy, the genes were fused with thioredoxin (Trx) encoding gene and tagged by an His₆-tag. Two additional constructs of the full cDNA were also produced. To prepare His-full 1 form, the full cDNA was extracted from pET32b plasmid using Kpn I and Eco RI restriction sites and cloned into pColdI plasmid (TaKaRa Bio Inc., Otsu, Japan). For the production of His-full 2 form, site-directed mutagenesis was carried out on the full cDNA using the QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA, USA), to insert a Nde I restriction site at the 5′ of the full cDNA, including the triplet coding for the initial methionine. The following mutagenesis primers were used: 5′GCCATGAGCTCGGATCAT ATGAGCCAGATCATG3′; 5′CATGATCTGGCTCATAT GATCCGAGCTCATGGC3′ (mutated bases are in italics). The presence of the desired mutation was confirmed by automated DNA sequencing. The mutated cDNA was then cloned into pColdI plasmid using Nde I and Eco RI restriction sites. The gene sequences are reported in Figure 1.

Strain and growth conditions

For protein expression, all plasmids were transferred to the E. coli hosts BL21(DE3) and BL21(DE3)pLysS. The following media were used: M9/glucose (6.78 g/L Na₂HPO₄, 3 g/L KH₂PO₄, 1 g/L NH₄Cl, 0.5 g/L NaCl, 0.1 mM CaCl₂, 2 mM MgSO₄, 10 g/L glucose); Luria-Bertani (LB, 10 g/L bacto-tryptone, 5 g/L yeast extract, 5 g/L NaCl); Terrific Broth (TB, 12 g/L bacto-tryptone, 24 g/L yeast extract, 8 mL/L glycerol, 2.2 g/L KH₂PO₄, 9.4 g/L K₂HPO₄); Terrific Broth/NaCl (TB, 12 g/L bacto-tryptone, 24 g/L yeast extract, 8 mL/L glycerol, 2.2 g/L KH₂PO₄, 9.4 g/L K₂HPO₄, 5 g/L NaCl); Super Broth (SB, 32 g/L bacto-tryptone, 20 g/L yeast extract, 5 g/L NaCl); Super Broth/NaCl (SB/NaCl, 32 g/L bacto-tryptone, 20 g/L yeast extract, 10 g/L NaCl) [15, 20]. Starter cultures were prepared from a single colony of E. coli carrying the recombinant plasmid, employing the same medium used for protein expression containing ampicillin (100 μg/mL, final concentration) and, in the case of BL21(DE3)pLysS strain, also chloramphenicol (34 μg/mL, final concentration). Cultures were grown overnight under vigorous shaking at 37°C. For chaperone co-expression, BL21(DE3) E. coli cells were co-transformed with pKJE7 (coding for AraB, DnaK, DnaJ and GrpE) or with pG-Tf2 (coding for GroES, GroEL and TiG) plasmids (Chaperone Plasmid Set, TaKaRa Bio Inc., Otsu, Japan), adding chloramphenicol (30 μg/mL, final concentration) to the growth medium. Expression of the chaperone proteins was induced by adding 1 mg/mL L-arabinose (for pKJE7) or with 2 ng/mL tetracyclin (for pG-Tf2), according to the manufacturer’s instructions. Protein production trials were carried out in 500 mL or 2 L baffled Erlenmeyer flasks containing 80 mL or 500 mL of each medium inoculated with the starter cultures (initial OD_600nm = 0.1). Cells were grown at 37°C with shaking (150 rpm) until protein expression was induced by adding IPTG: cells were then grown at 15 or 18°C until harvesting. Collected cells were washed with STE buffer and stored at -20°C.

3 L bioreactor cultivations

SB for E. coli BL21(DE3) cells containing the pET32b-Trx-Ag85B plasmid, LB for those transformed with pET32b-Trx-TB10.4 plasmid and SB/NaCl for those transformed with pColdI-His-full 2 plasmid were used as production media in 2 L working volume P-100 Applikon 3 L glass reactor (height 25 cm, diameter 13 cm) equipped with a AD1030 Biocontroller and AD1032 motor [20, 21]. Cultivations in bioreactor were conducted in batch-mode at 37°C, 500 rpm stirring (corresponding to 1.17 m/s of tip speed) and 2 L/min aeration rate. Foam production was controlled by the addition of Hodag antifoam through an antifoam sensor. Starter cultures were grown overnight in SB, LB or SB/NaCl medium and diluted up to an initial OD_600nm of 0.1. For pET32b-Trx-Ag85B and pET32b-Trx-TB10.4 recombinant cells, after 2–4 hours of growth (corresponding to an OD_600nm of 2 for the previous and of 0.8 for the latter), 0.1 mM IPTG was added and the temperature was decreased to 18°C. pColdI-His-full 2 recombinant cells grown for 4–6 hours (corresponding to an OD_600nm of 2) were added of 0.1 mM IPTG and 25 g/L NaCl and the temperature was decreased to 15°C. After 16 hours, cells were harvested by centrifugation, washed with STE buffer and stored at -20°C.

Crude extract preparation and purification of fusion proteins

Cell paste was re-suspended in phosphate buffer saline (PBS) (8.1 mM Na₂HPO₄, 1.76 mM KH₂PO₄, 0.137 M NaCl, 2.68 mM KCl), pH 7.0, 1 mM phenylmethanesulphonylfluride (PMSF), 10 μg/mL deoxyribonuclease. Cells were disrupted by sonication (5 cycles of 30 s each, in ice, with 30-s interval). The insoluble fraction was removed by centrifugation at 39000 × g for 1 hour at 4°C. The crude extract (added of 0.9 M NaCl) was loaded on a HiTrap Chelating column (GE Healthcare, Piscataway, NJ, USA), pre-loaded with Ni²⁺ and equilibrated in PBS, 0.9 M NaCl, pH 7.0. Elution of the fusion protein was performed increasing the amount of PBS buffer containing 0.5 M imidazole, pH 7.0 [22]. The pure proteins were then equilibrated in 20 mM TrisHCl, 50 mM NaCl, pH 7.4 by size-exclusion chromatography on a PD10 desalting column or by extensive dialysis.

Enterokinase cleavage of the fusion proteins

Enterokinases (EK, Recombinant Enterokinase, and Tag · off High Activity Recombinant Enterokinase) and Recombinant Enterokinase Capture Agarose were purchased from Novagen. One EK unit is defined as the amount of enzyme that cleaves 50 μg of a control protein (supplied by the producer) in 16 hours at room temperature and in 20 mM TrisHCl, 50 mM NaCl, 2 mM CaCl₂, pH 7.4.

SDS-PAGE electrophoresis and Western blot analysis

Proteins from total cell extracts or from both soluble and insoluble cell fractions were separated by SDS-PAGE: gels were stained for proteins with Coomassie Blue R-250. For total cell extracts and for insoluble fractions after cell disruption, cell pellets were directly resuspended in an appropriate volume of Laemmli sample buffer. For Western blot analysis, upon separation by SDS-PAGE, proteins were transferred electrophoretically onto a nitrocellulose membrane. The fusion proteins were detected using anti-His-tag mouse monoclonal antibody (His-probe, Santa Cruz Biotechnology, Santa Cruz, CA, USA), rabbit anti-TB10.4 polyclonal antibodies (Antibodies-online GmbH, Aachen, Germany), or rabbit anti-Ag85B polyclonal antibody (DIATHEVA, Fano, Italy), in combination with the appropriate secondary antibody: goat anti-mouse (Santa Cruz Biotechnology, Santa Cruz, CA, USA), donkey anti-mouse and donkey anti-rabbit (Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA) IgG HRP-conjugated antibody. The immunorecognition was detected by a chemioluminescence method (ECL Plus Western Blotting Detection System, GE Healthcare, Piscataway, NJ, USA). The amount of protein was estimated by evaluating the intensity of the signal using the public domain, Java-based image processing program ImageJ (National Institutes of Health, http://rsb.info.nih.gov/ij/). His-tagged D-amino acid oxidase [23] was used as standard.

Spectral measurements

Extinction coefficients of native proteins were determined using a Jasco V-560 spectrophotometer by complete denaturation of appropriate dilutions of protein samples in 6 M urea assuming as extinction coefficients of the fully denatured proteins those calculated by ExPASy Bioinformatic Resource Portal (http://www.expasy.org/) using ProtParam tool [24].

Circular dicroism (CD) spectra were recorded on a J-815 Jasco spectropolarimeter [24]: cell path-lenth was 1 cm for measurements above 250 nm (0.4 mg protein/mL) and 0.1 cm for measurements in the 190–250 nm range (0.1 mg protein/mL). Secondary structure fractions were calculated from deconvolution of the CD spectra with the program K2D2 (http://www.ogic.ca/projects/k2d2/) [25].

Temperature ramp experiments were carried out by using the same instrumentation equipped with a software-driven Peltier-based temperature controller (temperature gradient: 0.5°C/min). All spectral experiments were carried out in 20 mM MOPS, 0.4 M NaCl, pH 7.0 or in 10 mM ammonium acetate, pH 7.0 at 15°C.

Mass spectrometry analyses

Intact MS experiments were in general carried out on a LTQ-MS (Thermo Electron, San Jose, CA, USA) with an ESI source. The protein samples were prepared in 10 mM ammonium bicarbonate (pH 7.4), 0.1% acetonitrile/trifluroacetic acid (50/50) to a final concentration of 0.3 mg/mL. Full scan intact MS experiments were carried out under the following instrumental conditions: positive ion mode, mass range 700–2000 m/z, source voltage 4.5 kV, capillary voltage 35 V, sheat gas 15, auxiliary gas 2, capillary temperature 220°C, tube lens voltage 140 V. Five spectra were acquired for each data and each spectrum was the composite of two averaged scans. Data processing was performed using Bioworks Browser (Thermo Electron, revision 3.1). Theoretical average masses were calculated by Ion Source Bioinformatic Resource Portal (http://www.ionsource.com/) using Peptide Mass Calculator tool. Identification of truncated forms of the proteins was carried out by MS-non specific database search program Protein Prospector v 5.10.4 (http://prospector.ucsf.edu).

The His-full2 protein was dissolved in acetonitrile and analyzed using a MALDI-TOF/TOF Ultraflex III spectrometer (Bruker Daltonics, Bremen, Germany) in linear positive mode. A solution of sinapinic acid (SA) 20 mg/mL solved in a 70/30 acetonitrile/trifluoroacetic acid (TFA) 0.1% solution, was used as matrix. The spectra were acquired in a mass range of 6000–52000 m/z and processed using Flex Analysis software 3.0 (Bruker Daltonics).

Immunological studies: study population

The study population included 19 subjects. Twelve patients with newly diagnosed, untreated active pulmonary TB and 7 healthy individuals without any history of TB exposure (hereafter indicated as TB unexposed controls). Study subjects were recruited at the Department of Infectious Diseases of the Fondazione IRCCS-Policlinico San Matteo of Pavia (Italy), after informed consent. In all cases, the diagnosis of active TB was confirmed by M. tuberculosis culture isolation.

Peripheral venous blood was obtained for serum samples from all the participants to the study to perform ELISA assay. Additionally, peripheral blood was collected into tubes containing heparin for ELISPOT assay. Peripheral blood mononuclear cells (PBMC) were isolated by standard density gradient centrifugation using Lymphoprep (Axis-Shield, Oslo, Norway). Isolated PBMC were cryopreserved in freezing medium: 10% v/v DMSO (Sigma-Aldrich, St. Louis, MO, USA), 25% v/v human albumin (Grifolds Biologicals, Los Angeles, CA, USA) and 65% v/v RPMI 1640 supplemented with 2 mM L-glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin (all from Euroclone, Milan, Italy) and kept in liquid nitrogen until ELISPOT analyses.

ELISA assay

ELISA assay was performed with the method of [26] with minor modifications. Microplates 96-wells high-binding capacity flat-bottom (Greiner-bio-one GmbH, Frickenhausen Germany) were coated with different concentrations (400, 200, 100 ng/well) of Ag85B, TB10.4, His-full 2 in 50 mM bicarbonate buffer (pH 9.6) and incubated at room temperature overnight. Plates were washed three times with 0.05% (v/v) Tween-20 in phosphate buffered saline (PBS-T, Sigma-Aldrich, St. Louis, MO, USA) and incubated with 200 μL/well of blocking buffer (PBS with 1% BSA) at room temperature for at least 1 hour and then washed three times with PBS-T. Serum samples were diluted with dilution buffer (PBS-T with 1% BSA) at scalar dilution of 1:50 to 1:800. One-hundred microliter of the diluted sera was added to each well and incubated for 1 hour at 37°C. Plates were then washed four times with PBS-T. Anti-human specific IgG antibody labeled with horseradish peroxidase (MP Biomedicals, LLC Santa Ana, CA, USA) was added to each well and incubated at 37°C for 1 hour. Plates were again washed four times with PBS-T and 100 μL of o-phenyl-diamine/H₂O₂ substrate (Sigma-Aldrich, St. Louis, MO, USA) was added for 30 minutes. The reaction was stopped by adding 50 μL of 0.5 M H₂SO₄. The absorbance at 492 nm was measured with an ELISA plate spectrophotometer (Titertek Plus MS212 ICN, Eschwege, Germany).

Evaluation of peptide specific single T-cell IFN-γ release

The enumeration of peptide specific T-cell producing IFN-γ was performed at the single cell level by using ELISPOT assay, as previously described, with minor modifications [27].

In detail, PBMC were thawed, washed and re-suspended in the culture medium: RPMI 1640 medium supplemented with 2 mM L-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin and 10% heat-inactivated fetal calf serum (Euroclone, Milan, Italy). Cells were kept overnight at 37°C in a humidified 5% CO₂ atmosphere. Cells were then transferred to a 24-well flat-bottom plate (5 × 10⁵ cells/mL per well), stimulated with 1 μg/mL of each protein (one protein per well) or culture medium only or phytohemagglutinin (PHA) (5 μg/mL; Sigma-Aldrich, St. Louis, MO, USA), and cultured at 37°C in a humidified 5% CO₂ atmosphere for 10 days. On days 3 and 7, half of the supernatant from each well was removed and replaced with wet culture medium supplemented with 20 IU/mL recombinant human IL-2 (Peprotech, London, UK). After 10 days, cells from each well were harvested, washed three times with culture medium and re-suspended at a concentration of 4 × 10⁵ cells/mL before their use in ELISPOT assay. Un-stimulated PBMC were included as negative controls.

In ELISPOT assay, multiScreen-IP 96-well plates (Merck Millipore, Darmstadt, Germany) were coated with IFN-γ capture antibody and incubated overnight at 4°C. Plates were then washed 5 times with PBS and blocked with culture medium for 2 hours at room temperature. Antigen-stimulated cells for 10 days were added in duplicate (4 × 10⁴ cells/well) and re-stimulated with each protein used for stimulation during the 10-day period. Cells stimulated with PHA during the 10-day period were re-stimulated with Staphylococcus aureus Enterotoxin B (SEB) superantigen (2 μg/mL; Sigma-Aldrich, St. Louis, MO, USA). After incubation at 37°C in a humidified 5% CO₂ atmosphere for 24 hours, plates were washed five times with PBS-T (Sigma-Aldrich, St. Louis, MO, USA). Biotinylated detection antibody for IFN-γ was added and incubated overnight at 4°C. After five washes with PBS-T, streptavidin-alkaline phosphatase conjugate was added and plates were incubated at 37°C, 5% CO₂ atmosphere for 1 hour. Plates were then washed, 5-bromo-4-chloro-3-indolyl-phosphate/nitro blue tetrazolium (BCIP/NBT) was added and incubated at room temperature for 20 minutes. Wells were then washed several times under running water and air-dried. Spot-forming cells (SFCs) were counted in duplicate wells using an automated ELISPOT reader system (Autoimmun Diagnostika GmbH, Strassberg, Germany). The mean value of the SFCs was calculated, and the value of the control subtracted to the SFCs of the stimuli: values were referred to million PBMC.

Statistical analysis

Data are expressed using mean and standard deviation of the mean or median and percentiles as appropriate. Comparison between groups have been made using Mann–Whitney and χ² tests. A p value below 0.05 was considered significant. All tests were performed using the GraphPad Prism 4.0 (Graphpad software, San Diego, CA, USA) software package.