Dynamics of Mycobacterium tuberculosis Ag85B Revealed by a Sensitive Enzyme-Linked Immunosorbent Assay

Bacterial protein secretion contributes to host-pathogen interactions, yet the process and consequences of bacterial protein secretion during infection are poorly understood. We developed a sensitive ELISA to quantitate a protein (termed Ag85B) secreted by M. tuberculosis and used it to find that Ag85B secretion occurs with slower kinetics than for proteins secreted by Gram-positive and Gram-negative bacteria and that accumulation of Ag85B in the lungs is markedly regulated as a function of the bacterial population density. Our results demonstrate that quantitation of bacterial proteins during infection can reveal novel insights into host-pathogen interactions.

Due to their ability to induce adaptive CD4 and CD8 T lymphocyte responses in a broad range of vertebrate hosts, Ag85A and Ag85B have been investigated as antigens for tuberculosis vaccines and are prominent components of at least seven candidate vaccines in various stages of development (http://www.aeras.org). Quantitative assays of mRNA have revealed that the genes encoding Ag85A and Ag85B (fbpA and fbpB, respectively) are expressed at high levels by bacteria in the lungs early after aerosol infection of mice, but their mRNA expression decreases markedly after the recruitment of antigen-specific effector T cells to the lungs (12)(13)(14). Consistent with the results of bacterial RNA quantitation, CD4 T cells specific for Ag85B are activated in the lungs between 2 and 3 weeks after infection of mice, but their activation markedly decreases concurrent with decreased bacterial expression of the fbpB gene (12,15).
Despite considerable knowledge of the properties of the fbpA and fbpB genes and the antigenicity of their products, there is less information on the secretion, in vivo expression, and trafficking of the Ag85A or Ag85B proteins. Because of interest in Ag85B as a vaccine and/or diagnostic antigen, we generated new monoclonal antibodies (mAbs) to Ag85B and used them to establish a highly sensitive and specific enzyme-linked immunosorbent assay (ELISA). We then employed the ELISA in studies of secretion and trafficking of the Ag85B protein in vitro and in vivo.

RESULTS
Generation and characterization of monoclonal antibodies to Ag85B. Monoclonal antibodies were generated, using standard methods (16), from mice immunized with purified recombinant M. tuberculosis Ag85B (rAg85B) expressed in Escherichia coli. Three mAbs, termed 710, 711, and 712, were selected for characterization. When examined by a direct ELISA using wells coated with purified rAg85A, rAg85B, or rAg85C and various concentrations of antibody, all three mAbs recognized Ag85B and yielded equivalent signals (Fig. 1A). The three mAbs also recognized Ag85A in the direct ELISA, although recognition of Ag85A required higher antibody concentrations and reached a lower maximum signal intensity than with Ag85B. mAb 710 generated a higher signal intensity and at lower antibody concentrations than with mAbs 711 and 712; the latter two mAbs also exhibited detectable binding to Ag85A when used at concentrations of Ն1 g/ml (Fig. 1B). In contrast, mAbs 711 and 712 did not bind to Ag85C in the direct ELISA at any antibody concentration examined, while mAb 710 bound Ag85C at antibody concentrations as low as 0.01 g/ml (Fig. 1C). Testing of the mAbs at a fixed concentration of 1 g/ml on a dilution series of recombinant protein in the direct ELISA revealed that mAb 710 bound to Ag85A, Ag85B, and Ag85C at lower antigen concentrations than did mAb 711 or 712; mAbs 711 and 712 were indistinguishable in this assay. All three mAbs bound to Ag85B at lower antigen concentrations than required for binding to Ag85A or Ag85C ( Fig. 1D and E). Together, these results indicate that mAbs 710, 711, and 712 preferentially recognize Ag85B, although each of the mAbs also binds Ag85A and Ab85C when these antigens are present at high concentrations.
Sandwich ELISA for quantitation of Ag85B. When the three mAbs were used in pairwise combinations, using one mAb for capture and another mAb conjugated to horseradish peroxidase (HRP) for detection of bound antigen, the highest sensitivity for detection of Ag85B was obtained when mAb 710 was used as the capture antibody ( Fig. 2A). Although the differences were slight, sensitivity appeared to be greater when mAb 711 rather than mAb 712 was used as the detection antibody. Compared with Ag85B capture by mAb 710, mAb 711 capture allowed detection of bound Ag85B by mAb 710 but not by mAb 712. When mAb 712 was used for capture, mAb 710 was able to bind Ag85B, but mAb 711 was not. Together, these data indicate that the epitopes recognized by mAbs 711 and 712 overlap or may even be identical, while the epitope recognized by mAb 710 is distinct from those of mAbs 711 and 712. Despite detectable binding of all three mAbs to Ag85A in the direct ELISA (Fig. 1B), none of the combinations of capture or detecting antibodies yielded a signal when Ag85A was used as the antigen in the sandwich ELISA (Fig. 2B). Likewise, despite binding of mAb 710 to Ag85C by the direct ELISA, none of the capture or detection antibody combinations resulted in a detectable signal in the sandwich ELISA when Ag85C was used as the antigen (Fig. 2C). Together, these results suggest that the epitope recognized by mAb 710 is at least partially shared by Ag85A, Ag85B, and Ag85C, while the epitope(s) recognized by mAbs 711 and 712 is more specific to Ag85B.
To confirm the specificity of the ELISA with the combination of mAbs 710 and 711 using native M. tuberculosis proteins, we examined culture filtrates of wild-type H37Rv and of our previously characterized Ag85B-deficient (fbpB-null) mutant strain of H37Rv (12). This yielded no detectable signal in the ELISA when culture filtrates of the Ag85B-deficient bacteria were examined, despite the detection of an abundant signal when wild-type culture filtrates were examined (Fig. 2D). Since the Ag85B-null bacteria retain the capacity to synthesize and secrete Ag85A and Ag85C (17), these results provide further evidence that the sandwich ELISA with mAbs 710 and 711 is highly specific for Ag85B.
The assay in this form, with mAb 710 as the capture antibody and HRP-coupled mAb 711 as the detection antibody, has been run in our laboratory Ͼ30 times to determine Ag85B content in various samples. Each assay comprised at least one standard curve with rAg85B per ELISA plate, starting at 10 ng/ml or above and including 2-fold dilutions to loss of signal (generally, Յ20 pg/ml). Three representative assay standard curves are shown in Fig. 2E. Ag85B in samples was quantitated based on either a dose-response sigmoidal regression curve or a linear regression in the linear Ag85B Secretion and Trafficking ® proportion of the curve (0 ng/ml to 2.5 ng/ml). With both methods, regression curves had R 2 values of Ͼ0.99. When necessary, samples were diluted for the ELISA to be in the linear range of the standard curve, and at least two technical replicates or biological replicates (when available) were run. Carbon source effects, strain-dependent variation, and kinetics of Ag85B secretion. We used the ELISA to examine whether Ag85B secretion is governed by bacterial growth conditions. When grown in rich broth (Middlebrook 7H9 medium with 10% albumin-dextrose-catalase [ADC]), a 10-fold difference in the Tween 80 concentration (0.5% versus 0.05%) did not affect the amount of Ag85B secreted by M. tuberculosis H37Rv over a 24-h period (Fig. 3A). When H37Rv was grown in minimal defined medium (Sauton's medium with 0.05% Tween 80, 0.5% bovine serum albumin [BSA], and 0.05% tyloxapol) or 7H9 broth with the addition of 0.2% acetate, dextrose, or glycerol as the carbon source, the addition of acetate or glycerol did not alter Ag85B secretion, whereas 0.2% dextrose increased Ag85B secretion approximately 2-fold (Fig. 3A). These results indicate that bacterial metabolism as dictated by alternative carbon sources can affect the rate and/or amount of Ag85B secretion but that the concentration of Tween 80 had little measurable effect.
We previously reported that Mycobacterium africanum expresses lower levels of Ag85B than M. tuberculosis H37Rv, as detected by immunoblotting and by the magnitude of in vivo antigen-specific CD4 T cell responses (18). To verify this result and determine whether other strain-dependent variation in Ag85B secretion exists, we used the Ag85B-specific ELISA to assay culture filtrates from phylogenetically distinct bac- terial isolates. This confirmed that M. africanum secretes significantly smaller quantities of Ag85B than M. tuberculosis H37Rv (Fig. 3B). We also found that two distinct isolates of M. tuberculosis from lineage 2 (which includes the Beijing family) secrete significantly larger quantities of Ag85B than H37Rv (Fig. 3B). These results indicate substantial variation in the secretion of Ag85B, according to the bacterial strain, possibly in a lineage-dependent manner.
To further characterize the properties of Ag85B, we determined the kinetics of its secretion by M. tuberculosis H37Rv. Washed mid-log-phase bacteria were suspended in fresh 7H9 medium at 10 8 CFU/ml, and culture filtrates were harvested at multiple intervals and assayed by an ELISA. The rate of accumulation of Ag85B in culture filtrates was linear (r 2 ϭ 0.93) during the 24-h period of sampling, at 1.8 Ϯ 0.2 ng/ml/h (Fig. 3C). Since the mature Ag85B protein has a molecular weight of 34,580, this is equivalent to 52 fmol per 10 8 CFU per h, or approximately 300 molecules secreted per bacterial cell per h.
Cell-free Ag85B during infection in vivo and in vitro. Despite their biological activity and roles in pathogenesis and immune responses, little is known of the in vivo fate or distribution of secreted M. tuberculosis proteins, especially during infection. Therefore, we examined supernatants of lung homogenates obtained at various intervals after infecting C57BL/6 mice with M. tuberculosis H37Rv by aerosol. We found that Ag85B was detectable in lung homogenate supernatants in some, but not all, mice as early as 4 to 8 days postinfection, followed by a progressive increase between 14 and 21 days postinfection (Fig. 4A). Since the samples for the assay were taken during the progressive growth phase of the bacteria in the lungs, the quantity of Ag85B detected is a function of the amount secreted by each bacterium and the number of bacteria, which increases progressively until approximately 21 days postinfection (19). Therefore, we normalized the concentration of Ag85B in lung homogenates by the number of bacteria present in the lungs at each time point sampled. This revealed a progressive decrease in the amount of Ag85B relative to the bacterial population size, commencing at between 8 and 11 days postinfection (Fig. 4B). This result is consistent with results of Ag85B RNA quantitation in bacterial populations in the lungs of immunocompetent mice (12)(13)(14) and with results of studies indicating that activation of Ag85B-specific CD4 There are several potential sources of free Ag85B in the lungs. First is that some of the bacteria may be extracellular and may secrete Ag85B directly to the intercellular spaces in lung tissues. Second is that, as we have recently reported, Ag85B and other secreted mycobacterial proteins can be exported from infected cells by a vesicular transport pathway (20,21). A third potential mechanism is the release of Ag85B associated with bacterial membrane vesicles (22) and/or exosomes shed by infected cells (23,24). A fourth possible mechanism is that Ag85B synthesized by intracellular bacteria may be released from dying infected cells. We investigated the latter possibility using bone marrow-derived dendritic cells (BMDC) infected with M. tuberculosis H37Rv. Under conditions of the multiplicities of infection, bacterial strain, and time points used, we observed a range of loss of cell viability as reflected by a luminescence assay of ATP in cell lysates, after harvesting conditioned medium for the assay of Ag85B by an ELISA. This revealed a correlation between the loss of cell viability and the quantity of Ag85B in conditioned medium (r 2 ϭ 0.5777; P Ͻ 0.0001) (Fig. 5), indicating that Ag85B can be released from dead and/or dying infected cells in a form that remains detectable by the ELISA. This suggests that death of infected cells can be a source of the Ag85B detected in cell-free homogenates of the lungs of infected mice, as shown in Fig. 4.

DISCUSSION
In this work, we developed three new murine monoclonal antibodies by immunizing mice with recombinant M. tuberculosis Ag85B and used them to develop a sensitive and   (25)(26)(27)(28), the results suggest that differences in Ag85B secretion may contribute to strain-dependent differences in the proinflammatory properties of distinct strains and lineages in the M. tuberculosis complex (29). Strain variability in secretion may also influence the frequency and magnitude of T cell responses to Ag85B, which may provide an explanation for the finding that few human subjects exhibit detectable responses to Ag85B compared with other antigens such as ESAT-6 and/or CFP-10 (30,31). Likewise, bacterial strain-dependent variation in the secretion of Ag85B may influence the protective efficacy of TB subunit vaccines that include Ag85B: even if a vaccine generates immune responses to Ag85B, vaccination may provide little protection against infection with mycobacterial strains that express and secrete small quantities of Ag85B. In studies of subunit vaccines that include Ag85B, it may be informative to characterize Ag85B expression and secretion in isolates of M. tuberculosis from subjects who develop breakthrough infections despite apparently appropriate immune responses to the vaccine antigen.
We also used the ELISA to determine that the rate of secretion by M. tuberculosis H37Rv in broth culture is approximately 300 molecules of Ag85B per bacterial cell per h. This is 10-to 100-fold lower than the rate of translocation of E. coli proOmpA (4.5 mol/min) (32) or the secretion of staphylococcal toxic shock toxin (1.1 ϫ 10 4 molecules/CFU/h) (33). A full understanding of the basis for this difference will require determination of the rates of protein synthesis in M. tuberculosis compared with those in other bacteria and characterization of the factors and mechanisms that determine the rate of protein secretion by distinct secretion systems in bacteria. Given the important roles of secreted protein virulence factors in M. tuberculosis and other bacterial pathogens, a better understanding of protein secretion may reveal new targets for therapeutic modulation and reduction of disease and pathogen transmission.
A third major finding in these studies is that M. tuberculosis Ag85B can be found in homogenized lung tissue supernatants from infected mice. This finding has several potential implications. If data from studies of human subjects provide supportive data, then detecting Ag85B in respiratory secretions may provide a rapid and economical approach to diagnosing pulmonary tuberculosis and may also be useful in monitoring responses to treatment. Another is that the presence of extracellular Ag85B in lung tissue may make Ag85B available for uptake, processing, and presentation by uninfected dendritic cells and macrophages in the lungs. Since we have reported that direct recognition of infected cells is required for optimal CD4 T cell control of intracellular M. tuberculosis (34), acquisition and presentation of Ag85B by uninfected cells in the lungs may provide antigen-loaded decoys for Ag85B-specific CD4 T cells that reduce the frequency of recognition of infected cells by those T cells. This is consistent with our recent finding in mouse lungs (35) and the finding in lungs of M. tuberculosis-infected rhesus macaques (36) that only a small fraction of the T cells in the lungs are in close contact with M. tuberculosis-infected cells. Our associated finding in this work that death of M. tuberculosis-infected host cells can be associated with the release of Ag85B to the extracellular space indicates that the presence of Ag85B (and other mycobacterial proteins) in extracellular tissue compartments in the lungs may be the consequence of secretion by extracellular bacteria, vesicular transport from infected cells (21), carriage by exosomes (37,38), and release by dying or dead host cells. Since Ag85B (39,40), , and other mycobacterial proteins have been reported to act on host cells to modulate inflammation, our findings provide additional evidence for the plausibility that extracellular mycobacterial proteins contribute to the pathogenesis of TB and may be susceptible to therapeutic modulation. Ag85B Secretion and Trafficking ® MATERIALS AND METHODS Ethics statement. All animal experiments were done in accordance with procedures approved by the New York University School of Medicine Institutional Animal Care and Use Committee (laboratory animal care protocol 150502-01), which conformed to the guidelines provided by the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (42).
Recombinant Ag85B. RV1886c-ss.pET23b was transformed into E. coli BL21(DE3)/pLysS (Invitrogen) and induced with 0.8 mM isopropyl-␤-D-thiogalactopyranoside (IPTG) for 4 h at 37°C. Cultures were lysed with a solution containing 50 mM Tris (pH 7.0), 150 mM NaCl, lysozyme, Benzonase, and 1 mM phenylmethylsulfonyl fluoride (PMSF) for 30 min at 22°C on an orbital shaker. The lysate was spun at 10,000 ϫ g for 30 min, sterile filtered, and loaded onto an Akta fast protein liquid chromatography (FPLC) His-Trap column (GE). The column was washed with 50 mM imidazole, and recombinant protein was eluted with 250 mM imidazole. Purity was assessed by SDS-PAGE. mAb generation and selection. BALB/c mice were immunized with purified recombinant Ag85B (100 g/mouse for 2 injections followed by 50 g/mouse for 2 additional injections) subcutaneously in TiterMax gold (TiterMax, Norcross, GA, USA), followed by 50 g/mouse given intravenously 3 days before harvesting and using spleen cells in fusions with P3X63Ag8 myeloma cells. Hybridoma supernatants were screened for recognition of rAg85B-coated wells by an ELISA.
Direct ELISA for characterization of individual mAbs. Ag85A (catalog number NR-14871; BEI Resources), Ag85B (our purified recombinant), or Ag85C (catalog number NR-14858; BEI Resources) was used to coat wells at 0.5 g/well in phosphate-buffered saline (PBS) and incubated overnight at 4°C. The plates were washed 3 times with PBS (pH 7.4) with 0.05% Tween 20 and blocked with PBS containing 1.0% BSA for 1 h. A starting concentration of 5 g/ml of mAb 710, 711, or 712 was serially diluted, and 200 l/well was incubated at room temperature for 2 h. Plates were washed 5 times and incubated with goat anti-mouse IgG-HRP (MP Biomedicals) for 1 h at room temperature. Plates were washed 7 times and developed with the 3,3=,5,5=-tetramethylbenzidine (TMB) substrate according to the manufacturer's instructions (BD). The reaction was stopped with 2 M sulfuric acid, and the absorbance was read at 450 nm with a Synergy H1 microplate reader (BioTek).
Mycobacterial culture filtrates. M. tuberculosis strains H37RV, 4334, and HN878, or M. africanum strain 1182, were inoculated from a 1-ml frozen stock of approximately 3 ϫ 10 8 CFU/ml into 10 ml of Middlebrook 7H9 medium supplemented with 0.5% albumin, 0.2% dextrose, and 0.3 mg/100 ml catalase (termed 7H9 in the text) and grown to late log phase. The cultures were then passaged and grown once to mid-log phase. Cultures were then collected and spun at 150 ϫ g. The collected supernatant was then spun at 3,750 ϫ g for 5 min and washed with PBS. The pellets were then resuspended in fresh 7H9 medium and grown for 24 h. The cultures were pelleted, and the medium was sterile filtered.
Sandwich ELISA development and optimization. Individual mAbs were used at 2.5 g/ml in 50 l 0.05 M carbonate-bicarbonate buffer to coat wells overnight at 4°C. The plates were washed 3 times with PBS (pH 7.4) with 0.5% Tween 20 and blocked with PBS containing 1.0% BSA for 1 h. Antibodies for detection were conjugated to HRP according to the manufacturer's recommendations (Abcam). Briefly, 1 l of the modifier was mixed with 10 l of antibody and then added to a vial of HRP mix. The vial was incubated overnight at room temperature in the dark. After incubation, 1 l of the quencher was added and mixed. The conjugated antibodies were diluted to 0.5 g/ml in PBS with 1% BSA and stored in aliquots at Ϫ20°C until first use. Upon thawing, conjugated antibodies were stored at 4°C.
For ELISAs, samples were added to plates coated with the designated antibody and incubated at room temperature for 2 h, and plates were then washed 5 times. The labeled detection antibody was added, and the plates were incubated at room temperature for 1 h. Plates were washed 7 times and developed with the substrate according to the manufacturer's instructions (BD). The reaction was stopped with 2 M sulfuric acid, and the absorbance was read at 450 nm.
Sandwich ELISA application. The routine sandwich ELISA was performed according to the procedures described above, with mAb 710 as the capture antibody used to coat wells and HRP-conjugated mAb 711 used to detect bound antigen. Signal generation for the HRP reaction was done by using the TMB substrate (Thermo Scientific).
Effects of medium and carbon source. After washing in PBS, bacteria were resuspended in 6 ml of distilled water and split among 6 bottles. Ten milliliters of medium was added to the bottles. The medium used was Middlebrook 7H9 medium with 10% ADC and either 0.05% or 0.5% Tween 80 or Sauton's broth with 0.05% Tween 80, 0.5% BSA, 0.05% tyloxapol, and 0.2% (wt/vol) acetate, glycerol, or dextrose (adapted from reference 46), and bacteria were incubated for 24 h prior to preparation of culture filtrates.
Preparation of lung homogenate supernatants from M. tuberculosis-infected mice. C57BL/6 mice were infected by the aerosol route with ϳ60 CFU/mouse of M. tuberculosis H37Rv, lungs were harvested, and single-cell suspensions were prepared as previously described (47). After sampling the cell suspensions for determination of bacterial CFU, an aliquot of each sample was sterile filtered, and the Ag85B concentration was determined by an ELISA.
Culture, infection, and analysis of bone marrow-derived dendritic cell death and release of Ag85B. Bone marrow-derived dendritic cells (BMDC), generated as previously described (20), were seeded in 96-well tissue culture-treated plates (Corning) at 2 ϫ 10 6 cells/well, rested for 2 h, infected overnight at different multiplicities of infection (1, 2, 4, and 8) with M. tuberculosis H37Rv, treated with amikacin (200 g/ml for 40 min) in BMDC medium (RPMI 1640 supplemented with 10% heat-inactivated fetal bovine serum [FBS], 2 mM L-glutamine, 1 mM sodium pyruvate, 1ϫ ␤-mercaptoethanol, 10 mM HEPES, and 12 ng/ml recombinant mouse granulocyte-macrophage colony-stimulating factor [GM-CSF]), washed three times in PBS, and further cultured in fresh BMDC medium. Conditioned medium (CM) was harvested 16, 24, 34, and 48 h later and sterile filtered, and Ag85B in CM was quantified by a sandwich ELISA. At each harvest time point, infected BMDC were assayed for cell death by using CellTiter-Glo (Promega) according to the manufacturer's instructions, and the signal was read as luminescence with a Synergy H1 microplate reader (BioTek). For each harvest time point, the signal from uninfected cells was considered 100% viability for determination of loss of viability of infected cells.
Statistical analyses. All statistical analyses were performed using Prism 7 (GraphPad). The specific tests used for data analysis are specified in the individual figure legends.