Immunoproteomic Identification of Human T Cell Antigens of Mycobacterium tuberculosis That Differentiate Healthy Contacts from Tuberculosis Patients*

Identification of Mycobacterium tuberculosis antigens inducing cellular immune responses is required to improve the diagnosis of and vaccine development against tuberculosis. To identify the antigens of M. tuberculosis that differentiated between tuberculosis (TB) patients and healthy contacts based on T cell reactivity, the culture filtrate of in vitro grown M. tuberculosis was fractionated by two-dimensional liquid phase electrophoresis and tested for the ability to stimulate T cells in a whole blood assay. This approach separated the culture filtrate into 350 fractions with sufficient protein quantity (at least 200 μg of protein) for mass spectrometry and immunological analyses. High levels of interferon-γ (IFN-γ) secretion were induced by 105 fractions in healthy contacts compared with TB patients (p < 0.05). Most interesting was the identification of 10 fractions that specifically induced strong IFN-γ production in the healthy contact population but not in TB patients. Other immunological measurements showed 42 fractions that induced significant lymphocyte proliferative responses in the healthy contact group compared with the TB patients. The tumor necrosis factor-α response for most of the fractions did not significantly differ in the tested groups, and the interleukin-4 response was below the detectable range for all fractions and both study groups. Proteomic characterization of the 105 fractions that induced a significant IFN-γ response in the healthy contacts compared with the TB patients led to the identification of 59 proteins of which 24 represented potentially novel T cell antigens. Likewise, the protein identification in the 10 healthy “contact-specific fractions” revealed 16 proteins that are key candidates as vaccine or diagnostic targets.

Tuberculosis (TB) 1 is a major health problem throughout the world. A recent World Health Organization report shows that TB has been increasing at a rate of 1% per year, and an estimated 9.2 million new cases arise each year (1). Although TB is preventable, there has been an increase in its incidence in recent years. Re-emergence of TB is mainly due to its association with human immunodeficiency virus infection (2) and also due to the occurrence of multidrugresistant strains of the causative agent, Mycobacterium tuberculosis (3).
Vaccination of general population is cost effective and represents one of the best biological measures for disease control. The current vaccine against tuberculosis, Bacille Calmette-Gué rin (BCG), has been administered to more people than any other vaccine. The side effects of BCG are tolerable, and it prevents miliary and meningeal tuberculosis in young children. In striking contrast, it affords limited and highly variable protection (0 -80%) against pulmonary TB (4). Thus, BCG does not seem to be a satisfactory vaccine (5,6) and necessitates exploration of newer strategies to improve BCG or to develop a more effective vaccine.
One of the potential strategies for the development of an improved TB vaccine involves the use of the proteins secreted by M. tuberculosis during growth. There is evidence that proteins actively secreted by M. tuberculosis during growth induce cell-mediated immune responses by causing expansion of specific interferon-␥ (IFN-␥)-producing T lymphocytes that are capable of recognizing and exerting antimicrobial effects against infected macrophages (7). The importance of IFN-␥ pathways in host defense against M. tuberculosis was clarified by experimental studies on IFN-␥ knock-out mice as well as the identification and characterization of humans with mutations in IFN-␥ receptor (8,9).
Several studies have been carried out to define the secreted proteome of M. tuberculosis. The earliest study aimed at the identification of mycobacterial culture filtrate proteins, using chromatography and N-terminal sequencing to identify eight culture filtrate proteins (10). Later, many studies used twodimensional (2D) PAGE combined with sensitive mass spectrometric methods for identification of proteins. The above mentioned approaches have identified nearly 300 culture filtrate proteins (11)(12)(13).
Identification of T cell antigens in a complex mixture was first done by a T cell Western blot method (14). Later, twodimensional separation methods were used that involved protein separation by either IEF (15) or chromatography (16) in the first dimension and preparative SDS-PAGE followed by whole gel elution (17) in the second dimension. Mouse T cell antigens of M. tuberculosis were identified using this method (15). Mycobacterial antigens that induce an immune response in healthy household contacts and treated TB patients were also mapped using this approach (16).
In the present study, 2D liquid phase electrophoresis (LPE) along with an in vitro IFN-␥ assay and LC-MS/MS were used to identify potential human T cell antigens. Systematic screening of the M. tuberculosis culture filtrate (CF) proteome and comparative evaluation of cellular immune responses between TB patients and healthy contacts led to the identification of 59 proteins in the most immunogenic 2D LPE fractions. Twenty-four potentially novel T cell antigens were identified, and 16 proteins were identified in 10 2D LPE fractions that differentiated healthy contacts from TB patients based on IFN-␥ responses.

Growth of M. tuberculosis and Preparation of Culture Filtrate Protein (CFP)-M. tuberculosis
H37Rv (ATCC 27294) colonies were transferred from Lowenstein-Jensen slants to 2 ml of Sauton's medium (Himedia Laboratories), and the cells were dispersed using glass beads under sterile conditions. The bacterial cell suspension was transferred to 10 ml of Sauton's medium in a McCartney bottle for incubation at 37°C for 2 weeks. The bacilli were then transferred to 200 ml of Sauton's liquid medium, grown in shaker culture for 4 weeks, transferred to a 4-liter culture flask containing 2 liters of Sauton's medium, and grown as stationary culture for 4 weeks at 37°C. The bacilli in the culture were harvested by centrifugation at 3000 rpm for 30 min. The culture supernatant was filter-sterilized using a 0.20-m filter (Pal Gelman Laboratory, Saint Germain en Laye, France), and the filtrate was concentrated using the Quixstand benchtop hollow fiber system (GE Healthcare). The protein content in the CF was estimated using a BCA assay (Pierce). The proteins were then distributed into smaller volumes along with sodium azide at a final concentration of 0.2% and stored at Ϫ80°C for later use.
Separtation of CFP by 2D LPE-CFP (300 -350 mg) was solubilized in 60 ml of a buffer containing 8 M urea, 1 mM DTT, 5% glycerol, 2% digitonin, and 2% ampholytes (pH 3.0 -10.0 and pH 4.0 -6.0 at a ratio of 1:4) (Bio-Rad). The sample was loaded onto a liquid IEF system (Rotofor, Bio-Rad) maintained at 4°C. Separation was achieved by applying constant power (12 watts) until the voltage stabilized at ϳ1400 V. Focusing was continued for an additional 30 min and terminated. The individual IEF fractions were harvested, and their pH value was determined. Each aliquot was subjected to SDS-PAGE, and the proteins were visualized by staining with silver nitrate. A total of three IEF runs was performed, and fractions having a similar pH from each run were pooled.
Preparative SDS-PAGE was performed with individual IEF fractions. Briefly, each IEF fraction (protein quantities ranged from 0.08 to 43 mg) was prepared for SDS-PAGE by the addition of 6ϫ SDS-PAGE sample buffer (18) and heating at 95°C for 5 min. The reduced samples were loaded on 16 ϫ 20-cm polyacrylamide gels comprising a 4% stack over a 12.5% resolving gel. The stacking gel contained a single 13-cm-long sample well. Electrophoresis was performed at a constant current of 50 mA/gel until the dye front was ϳ2 cm from the bottom of the gel. The gel was equilibrated in elution buffer (60 mM Tris (pH 9.4), 40 mM CAPS) for 10 min and transferred to a Whole Gel Eluter (Bio-Rad) according to the manufacturer's instructions. The proteins were eluted from the gel using a constant current of 250 mA for 1 h. Thirty protein fractions (ϳ2.5 ml each) were harvested, and the protein concentration of each fraction was determined using the BCA assay (Pierce). Prior to testing for T cell stimulation, the 2D LPE fractions were filter-sterilized using a 0.2-m filter. An aliquot (10 g) of each eluted fraction was analyzed by SDS-PAGE and silver-stained (19).
Identification of Proteins-Proteins were digested in gel or in solution with modified trypsin (Roche Applied Science). For in-gel digestions, 10 g of selected 2D LPE fractions was resolved by SDS-PAGE using 4 -12% NuPAGE bis-Tris polyacrylamide gels and MES-SDS running buffer (Invitrogen). The gel was stained with Coomassie Brilliant Blue R-250 (Bio-Rad), and the resulting bands were excised from the gel. Digestion of proteins with modified trypsin was performed as described previously (20,21). For in-solution digestions, 5 g of each fraction was digested with 0.8 g of modified trypsin in 10% acetonitrile, 0.2 M ammonium bicarbonate and incubated overnight at 37°C. The reactions were terminated by the addition of 3 l of 10% TFA. The resulting peptides from the tryptic digestions were dried; suspended in 15 l of 5% acetonitrile, 0.1% acetic acid; and applied to a 0.2 ϫ 50-mm C 18 reversed phase HPLC column (Agilent Technologies, Santa Clara, CA). The peptides were eluted with an increasing gradient of acetonitrile at a flow rate of 5 l/min using an Agilent 1100 capillary HPLC solvent delivery system. The effluent was introduced directly into an linear trap quadrupole electrospray ion trap mass spectrometer (Thermo-Finnigan, San Jose, CA). The electrospray needle of the mass spectrometer was operated at 4 kV with a sheath gas flow of nitrogen at 30 p.s.i. and a heated capillary temperature of 200°C. Data-dependent tandem MS was used to generate fragment ions of individual peptides. The five most intense ions from the full MS scan were selected for fragmentation. The precursor ion was placed on the dynamic exclusion list for 1 min and Tandem MS was performed for each precursor ion, a maximum of two times.
MS/MS data were analyzed using Sequest (Thermo Fisher Scientific, San Jose, CA; Version 27, Revision 12) and X! Tandem (Global Proteome Machine Organization; Version 2006.04.01.2) software packages with data interrogation against the annotated M. tuberculosis genome (3912 entries). Fragment ion mass tolerance of 0.50 Da, parent ion tolerance of 1.5 Da, oxidation of methionine, and acrylamide adduct of cysteine were specified in Sequest and X! Tandem as part of the search criteria and variable modifications. The allowance for missed cleavages was 2. Scaffold (Version Scaffold-01-06-05; Proteome Software Inc., Portland, OR) was used to validate peptide and protein identifications using a two-peptide minimum and 95.0% peptide and protein probabilities as specified by the Peptide and Protein Prophet algorithms (22,23).The Universal Protein Resource (UniProt) database was searched to get the subcellular localization information of identified proteins.
Study Population-The study was approved by the Institutional Ethics Committee of the Tuberculosis Research Centre, and informed consent was obtained from all persons who were enrolled in this study.
Ten patients with pulmonary TB were enrolled at the Tuberculosis Research Centre clinic. The subjects of this group had not undergone antituberculosis treatment when recruited for the study. Their age ranged from 26 to 52 years, and the male to female ratio was 7:3. Two spot and one overnight sputum specimens were collected from each patient. Sputum specimens were examined for acid-fast bacteria by fluorescence microscopy (auramine O phenol staining), and the samples were examined microscopically. All the TB patients were positive by sputum smear microscopy. For culture, the sputum specimens were processed by modified Petroff's method, inoculated onto Lowenstein-Jensen medium, and incubated for up to 8 weeks at 37°C (24). Blood was collected from these subjects before treatment.
Seven individuals who shared living quarters with the tuberculosis patient agreed to join the study as healthy contacts whose age ranged from 28 to 55 years. The male to female ratio was 5:2. Three heavily exposed health care workers, working closely with pulmonary tuberculosis patients for at least 2 years at the Tuberculosis Research Centre, were also included in the study. Their age ranged from 28 to 35 years, and all were male. These individuals had no history of tuberculosis on the basis of personal history, physical examination, chest x-ray, and negative acid-fast bacilli sputum smear microscopy.
Tuberculin skin test using 2 tuberculin units of purified protein derivative (PPD) (Statens Serum Institut, Copenhagen, Denmark) was performed on all subjects, and an induration of 15 mm or more after 48 h was considered positive. All 10 healthy contacts enrolled in this study were PPD skin test-positive. Of the 10 TB patients, seven were skin test-negative, and three were positive. All subjects were human immunodeficiency virus-negative as determined by Tridot (J. Mitra and Co.) and Retroquic (Qualprodiagnostics) assays with serum. Heparinized blood (21 ml) was collected from all the subjects for the immunological assays presented in this study.
Lymphocyte Proliferation Assay (LPA)-An LPA was performed by diluting whole blood 1:10 in RPMI 1640 medium (Sigma) supplemented with glutamine (0.29 g/liter), penicillin (100 IU/ml), streptomycin (0.1 mg/ml), and amphotericin B (5 mg/ml) and culturing in 96-well flat bottom tissue culture plates. These cultures were stimulated with the 2D LPE fractions and with PPD and phytohemagglutinin as Positive controls. Cells cultured under similar conditions without any stimulation served as the negative control. Each antigen or fraction was added in triplicate wells to a final concentration of 5 g/ml. The antigen-stimulated cells were cultured for 6 days at 37°C in a 5% CO 2 atmosphere (Hera Cell, Kendro Laboratories). Sixteen hours before the termination of the cultures, 1 Ci of tritiated thymidine ([ 3 H]thymidine) (Board of Radiation and Isotope Technology, Mumbai, India) was added to each well. The cells were harvested onto glass fiber filters using a cell harvester (PHD, Cambridge Technology Ltd., Watertown, MA), and the filter discs were dried overnight. An aliquot (2 ml) of scintillation fluid (0.05 mg/ml 1,4-bis[2-(5-phenyloxazolyl)]benzene and 4 mg/ml 2,5-diphenyloxazole in toluene) was added to each filter disc and counted using a liquid scintillation ␤ counter (Wallac oy, Torku, Finland). The proliferation was measured as uptake of [ 3 H]thymidine by the cells and expressed as cpm. The mean cpm value of the triplicate culture was calculated. Proliferation was expressed as the stimulation index (SI): SI ϭ mean cpm with antigen/ mean cpm without antigen. Two analyses were performed with the lymphocyte proliferation data. In the first analysis, the mean cpm value obtained for the 10 healthy contacts was compared with that of the 10 TB patients for each of the fractions P values were calculated using the Mann-Whitney U test (Graphpad Software, San Diego, CA), and p values Ͻ0.05 were considered significant. Based on the p value, the fractions were grouped as "very highly significant" (p Ͻ 0.0005), "highly significant" (p Ͻ 0.005), "significant" (p Ͻ 0.05), and "non-significant" (p Ͼ 0.05). In the second analysis, a subject showing a cutoff value of Ն3 SI was classified as a responder, and the number of responders in each group was counted.
Cytokine Measurements-For quantification of cytokines (IFN-␥, TNF-␣, and IL-4), cell-free culture supernatants were harvested after 6 days of in vitro stimulation by fractions and stored immediately (at Ϫ80°C) until assayed. Cytokine production was determined by a standard ELISA technique using commercially available BD OptEIA kits (BD Biosciences) according to the manufacturer's instructions.
The OD values were read at 450 nm using an ELISA reader (Molecular Devices, Sunnyvale, CA).
Two types of analyses were performed with the IFN-␥ results. In the first analysis, the actual amount of IFN-␥ secreted (pg/ml) in response to each fraction was calculated. The levels induced by each fraction were compared in the TB patient and healthy contact groups using the Mann-Whitney U test (Graphpad Software), and p values Ͻ0.05 were considered significant. Based on the p value, fractions were grouped as very highly significant (p Ͻ 0.0005), highly significant (p Ͻ 0.005), significant (p Ͻ 0.05) IFN-␥-inducing, and non-significant (p Ͼ 0.05).
In the second analysis, the cutoff value for IFN-␥ was fixed for each fraction by using the mean ϩ 2 S.D. of IFN-␥ levels in TB patients (susceptible population). Any subject with an IFN-␥ value above the cutoff point was classified as a responder. The number of responders to each fraction in each study group was calculated.

2D LPE Separation of CFP-
The yield of M. tuberculosis CFP after 4 weeks of culture was 50 Ϯ 5 mg/liter. A large quantity of culture filtrate protein (1 g) was used as the starting material to ensure sufficient protein for immunological analysis and molecular identification of fractions. Initial experiments showed that when quantities of CFP greater than 350 mg were applied to preparative IEF excessive precipitation resulted. Therefore, three technical replicates of preparative IEF runs were performed using 300 -350 mg of protein per run. The 20 fractions collected from each replicate were analyzed by SDS-PAGE (data not shown). Fractions from each replicate were pooled based on their pH values, resulting in a total of 20 pooled fractions, each corresponding to a specific pH range. The pH of the separated IEF fractions ranged from 2.5 to 12.9, and the protein content varied from 0.08 to 43 mg.
The second dimension preparative SDS-PAGE performed on each IEF fraction resulted in 30 subfractions for a cumulative total of 600. All the fractions were numbered by IEF fraction number first followed by whole gel elution fraction number (for e.g. 1_1, 3_4, etc.). Initially, the fractions were analyzed by analytical SDS-PAGE. On analysis, it was observed that each fraction showed one to three bands. The protein quantity of each fraction ranged from 50 g to 4 mg. A minimum of 200 g of protein was required for immunological as well as proteomic characterization. Of the 600 2D LPE fractions, 350 possessed 200 g or more of protein, and these were selected for further analyses.
Definition of Immunodominant Fractions for TB Healthy Contacts-IFN-␥, an important effector cytokine in tubercu-losis infection (7), was used as a marker of antigen-specific T cell activation in whole blood assays with the 2D LPE fractions. In general, the antigen-induced IFN-␥ levels were higher in the healthy contact group when compared with the TB patients. The level of IFN-␥ secreted by TB patients for all 2D LPE fractions ranged from 0 to 2000 pg/ml, whereas for healthy contacts, it ranged from 0 to 8000 pg/ml. Of the 350 2D LPE fractions screened, 105 induced significant IFN-␥ levels in the healthy contact group compared with the TB patient group (supplemental Table S1). Based on the IFN-␥ levels and the significance in the difference between the two study populations, the reactive fractions were subdivided into three groups as follows: (i) 32 fractions induced very highly significant levels of IFN-␥ (p Ͻ 0.0005), (ii) 34 fractions stimulated highly significant levels of IFN-␥ (p Ͻ 0.005), and (iii) 39 fractions produced "significant levels" of IFN-␥ (p Ͻ 0.05) in the healthy contacts as compared with the TB patients. There were no fractions at all that induced significantly higher levels of IFN-␥ in TB patients as compared with the healthy contacts.
Interestingly, nine of the 2D LPE fractions designated very highly significant and one designated highly significant for IFN-␥ production in the healthy contact population showed a positive IFN-␥ response in all healthy contacts (n ϭ 10) but a negative response in all TB patients (n ϭ 10) based on the mean IFN-␥ response ϩ2 SD cutoff value (Fig. 1). These "contact-specific fractions" are listed in Table I.
Other cytokines such as TNF-␣ and IL-4 are known to influence the outcome of TB (7,25). Thus, these two cytokines were also measured in the whole blood assays with the 2D LPE fractions to determine whether there were other immunological markers that could differentiate the healthy contacts and TB patients. All the 2D LPE fractions induced a TNF-␣ response, but this response was not significantly different between the two study populations. It was noted that the spontaneous as well as antigen-induced TNF-␣ level was found to be higher in TB patients when compared with healthy contacts for all the fractions of which three fractions (6_7, 6_8, and 9_9) showed a statistically significant difference (data not shown). In contrast to the TNF-␣ response, the IL-4 response was found to be below the detectable range in all the cell culture supernatants from both groups (data not shown).
A fourth immunological parameter evaluated was induction by the 2D LPE fractions of lymphocyte proliferation (LP) in the whole blood. Of the 350 fractions screened, 42 (12%) induced a significant proliferative response in the healthy contacts as compared with the TB patients (supplemental table S2). Eight of these fractions were designated highly significant (p Ͻ 0.005), and 34 were designated significant (p Ͻ 0.05). All of the highly significant LP-inducing fractions except 8_30 showed positive LP responses in all 10 healthy contacts, and LP positivity in TB patients ranged from four to eight individuals depending on the 2D LPE fraction (supplemental Table  S2). The significant LPA fractions gave a positive LPA stimu-lation index for seven to 10 individuals in the healthy contact group, and in the TB patient group, four to 10 individuals were LPA-positive for the same fractions. Thus, unlike the IFN-␥ assay, no 2D LPE fractions were identified that could differentiate all healthy contacts from TB patients by the LPA. There was also variability in the correlation between the IFN-␥ and LP responses. All but three of the 2D LPE fractions inducing significant LP responses also induced significant IFN-␥ secretion (supplemental table S2). However, several of the fractions not inducing a significant LP response did induce significant IFN-␥ responses (data not shown). The variability in these two immunological measurements carried over to the 10 contact-specific fractions defined through the IFN-␥ response. Specifically, five of the contact-specific fractions (8_2, 8_29, 8_30, 9_26, and 12_21) induced highly significant or significant LP responses in healthy contacts as compared with the TB patients (Fig. 2), However, the remaining five contact-specific fractions did not result in significant differences in the LP responses between healthy contacts and TB patients (Table I).
Molecular Characterization of Immunodominant 2D LPE Fractions-Given that current in vitro T cell-based diagnostic assays for TB are based on the measurement of IFN-␥ responses (26) and the majority of anti-TB vaccine candidates were selected based on their ability to induce an antigen-specific IFN-␥ response (27), we focused protein identification efforts on those 2D LPE fractions that resulted in a significant IFN-␥ response in the healthy contact population. Specifically, the IFN-␥-inducing 2D LPE fractions were analyzed by in-gel and in-solution proteolytic digestions followed by LC-MS/MS, and data interrogation via Sequest and collation via Scaffold were used to identify the dominant proteins in these immunologically reactive fractions (supplemental Tables S3, S4, and S5).  When all of the data were combined, a total of 59 proteins were identified for the 105 2D LPE fractions that had significant IFN-␥ responses for the healthy contact population (Table II). For those proteins in the 10 -30-kDa molecular mass range, most were identified with one to seven peptides with a maximum of seven unique peptide sequences found for the protein Rv2626c followed by Cfp-17 (Rv1827) with six unique peptides. Proteins in the molecular mass range of 31-120 kDa were identified by matching 1-12 distinct peptides with 12 distinct peptide sequences found to match KatG (Rv1908c) in fraction 13_12 (supplemental Tables S6, S7, S8, and S9 contain the detailed proteomics data for each LPE fraction inducing a significant IFN-␥ response). Among the 59 identified novel T cell antigens, the UniPort database showed that 12 were secretory, four proteins localized to the cell membrane, 15 proteins localized to the cytoplasm, and one protein had an integral membrane location. Subcellular localization information for 27 proteins was not annotated in the UniPort database (Table II).
The most relevant group of 2D LPE fractions are those designated contact-specific fractions. Proteomic characterization of these 10 fractions demonstrated 16 proteins (Table  III), 13 of which (Cfp-10, DnaK, FpbB, Esat-6, GroEL2, GroES, HspX, KatG, PhoS1, PstS1, Rv2626c, TB8.4, and TrxC) were previously reported as human T cell antigens. Three proteins (AcpM, Adk, and Rv3716c) identified in these fractions are potentially new human T cell antigens. Proteins could not be identified in two contact-specific fractions (8_2 and 8_30) even though these fractions induced notable IFN-␥ responses. Among the 16 proteins identified in contact-specific fractions, four proteins were secretory, two proteins were cell membrane-associated, and four proteins have a cytoplasmic location. Subcellular localization of the remaining six proteins was not annotated in the UniPort database (Table III).
A comparison of identified proteins across all of the IFN-␥inducing fractions (very highly significant, highly significant, and significant) revealed only modest overlap in the proteins identified for each group. Specifically, only one protein (Rv2465c) was found in a 2D LPE fraction of each group, and a total of eight proteins was shared between two of the groups (Fig. 3). These data demonstrate that immunological responses are stratified with respect to different antigens, and it is this stratification that has allowed for the identification of the 10 contact-specific fractions and their corresponding proteins. DISCUSSION This present study was developed to identify the antigens of M. tuberculosis that have a potential use in a TB vaccine or in diagnostic development. Like many previous T cell antigen discovery efforts (15,16,28), we focused on the secreted proteins of M. tuberculosis. However, our current efforts differ from earlier studies, which used experimental animals (15), whereas our study aims at immunological responses to human TB. Moreover, the present study used the differential immune response of TB patients and healthy contacts to identify those proteins of M. tuberculosis with the greatest potential as protective or diagnostic antigens.
Although the immunological mechanisms of protection against tuberculosis are not fully understood, consistent evidence shows a dependence on antigen-specific T lymphocytes and the ability to stimulate the antimycobacterial activity of macrophages through the release of IFN-␥. The central role of IFN-␥ in the control of TB is clearly demonstrated by experiments showing that disruption of the IFN-␥ gene in mice and mutation of IFN-␥ receptor gene in humans result in increased susceptibility to TB infection (8,9). Therefore, the ability to stimulate T cell release of IFN-␥ has been used as a critical criterion for the identification of protective antigens for tuberculosis.
In our current studies, medical, paramedical, and laboratory staff and healthy household members that are in close contact with TB patients but remain healthy with no evidence of disease are viewed as the "protected" population (contacts). Multiple studies provide evidence that antigens recognized by the protected group, but not active TB patients, can be considered for vaccine development strategies by using IFN-␥ response as a protective correlate (29 -31). Although donor to donor variation exists in antigen recognition and magnitude of response, this approach is considered a highly viable method to identify the protective antigens (28). In agreement with a previous study (32), healthy contacts in our study displayed a strong IFN-␥ response to the 2D LPE fractions as compared with that of TB patients. Additionally, the stratification of the IFN-␥ response allowed for the identification of 10 2D LPE fractions possessing at least 16 proteins that are specific for the protected (contact) population versus the susceptible (TB patient) population.
The Quantiferon TB Gold assay is a well established T cell-based in vitro diagnostic assay for infection with M. tuberculosis and is a significant improvement to PPD skin testing in a non-endemic population for tuberculosis surveillance (33). However, in endemic settings, its ability to distinguish between infected and diseased individuals is severely limited (33). In fact our unpublished data with healthy controls and healthy household contacts of TB patients in Chennai, India    differentiation of infected and diseased individuals. Furthermore, the proteins of the contact-specific fractions are also potential tools to monitor infected, healthy individuals for the development of active tuberculosis.
Many proteins in the CFP have been identified as potent T cell antigens including Esat-6 (34), TB8.4 (16), HspX (35), Ag85 complex (36), and Cfp-10 (37). However, there is no direct correlation between the relative concentration of these antigens in CF and their immunological relevance. For instance, Esat-6, which is present in low amounts in culture filtrate, acts as one of the most potent T cell antigens (34). Of the 350 2D LPE fractions screened, we defined 105 fractions that induced significant levels of IFN-␥ in healthy contacts as compared with the TB patients. In these fractions, 59 different proteins were identified. Among these proteins, 35 observed as immunodominant antigens in our study were reported in earlier studies (Table II) either as T cell antigens in animal experiments or human studies or as B cell-stimulating antigens. Such concordance of our results with the pre-existing reports shows the reliability of the method used. Equally exciting is the fact that this methodology identified 24 proteins as potentially novel T cell antigens.
Among the 105 fractions inducing IFN-␥ in healthy contacts, 10 fractions containing 16 proteins were considered contactspecific. Of these, three proteins were identified as novel human T cell antigens (Adk, AcpM, and Rv3716c) along with the 13 already reported T cell antigens. Adk, reported as a mouse T cell antigen (15), was identified as a human T cell antigen in the present study. AcpM, previously identified as a weak B cell antigen, is reported here as strong T cell antigen (38). The presence of Rv3716c protein in culture supernatants was reported earlier in proteomics studies (11,39), and it was identified as an immunodominant molecule for the first time in the present study.
The presence of metabolic enzymes has been previously reported in the CF or extracellular milieu of M. tuberculosis and various pathogens (57). The extracellular localization of these proteins may be due to autolysis given the prolonged culture time (4 weeks) of M. tuberculosis. However, it is also possible they are present because of specialized secretion systems (58). The extracellular metabolic enzymes of M. tuberculosis also have been identified as virulence factors (53,59). A recent study (53) identified a list of proteins having plasminogen binding activity in M. tuberculosis. Interestingly, this included the immunodominant antigens Ag85 complex, DnaK, Mpt51, and GlnA1 as well as the potentially novel T cell antigens (LpdC, SahH, and Fba) identified in the immunodominant fractions of the current study.
Interaction between virulence factors and IFN-␥ has been clearly documented in the case of LpdC, a protein identified in this study. One of the activities of this protein is participating in the inhibition of phagolysosomal fusion by binding with coronin (a protein involved in phagolysosomal fusion). Experiments showed that IFN-␥ was able to inhibit LpdC protein association with coronin (60). In the present study, the Lpdcontaining fraction induced increased IFN-␥ secretion in the protected population compared with TB patients, thus corroborating that Lpd is produced in vivo.
It is generally believed that low molecular mass proteins of M. tuberculosis are predominantly recognized by human T cells (61,62), and most of the studies on human T cell antigens have focused on low molecular mass or selective pooling of fractions (16,28) and reported that antigens Ͻ15 kDa were highly immunogenic. In the present study, only 14.8% of the immunodominant T cell antigens were represented by antigens with a predicted mass of less than 15 kDa. A stronger immune response to the proteins below 25 kDa was observed in the present study (48.6%) in agreement with previous reports (16,63). However, a sizable number of recognized proteins (ϳ31%) were also found to be greater than 40 kDa. Earlier studies have shown that secreted and membrane-associated mycobacterial proteins were immunogenic (16,79) and can be tested as vaccine targets. In this study, we found 12 secretory and five cell membrane-associated proteins to be immunogenic (Table II). These proteins can be tested as potential vaccine targets.
Further animal and human studies are needed to explore the vaccine and diagnostic potential of the candidate antigens identified in this study. Experiments are underway to test the endemic normal population and TB patients who have successfully completed chemotherapy and remained quiescent for 1 year. These studies will help define the utility of the 16 contact-specific antigens in distinguishing between protected and "susceptible" populations and whether IFN-␥ responses to these antigens return after successful chemotherapeutic treatment of TB. § Supported by a senior research fellowship from the Council of Scientific and Industrial Research, New Delhi, India.