Pulmonary Mononuclear Cell Responses to Antigens of Mycobacterium tuberculosis in Healthy Household Contacts of Patients with Active Tuberculosis and Healthy Controls from the Community1 2

Protective immunity against Mycobacterium tuberculosis requires CD4+ lymphocyte-mediated immune responses and IFN-γ activity. As the primary portal of entry of M. tuberculosis is the lung, pulmonary immune responses against multiple M. tuberculosis Ags were compared between both M. tuberculosis-exposed tuberculin skin test-positive healthy household contacts (HHC) of patients with active sputum smear and culture-positive tuberculosis and tuberculin skin test-positive healthy control individuals from the community (CC). Frequencies of M. tuberculosis Ag-specific IFN-γ-producing cells, IFN-γ concentrations in culture supernatants, and DNA synthesis in bronchoalveolar cells (BAC) and PBMC were studied in HHC (n = 10) and CC (n = 15). Using enzyme-linked immunospot assay we found higher frequencies of IFN-γ-producing cells with specificity to M. tuberculosis-secreted Ag 85 (Ag 85) in BAC from HHC than in BAC from CC (p < 0.022) and relative to autologous PBMC, indicating compartmentalization of Ag 85-specific cells to the lungs. Further, IFN-γ-producing cells with specificity to components A and B of Ag 85 were specifically compartmentalized to the lungs in HHC (p < 0.05). IFN-γ concentrations in culture supernatants of BAC and Ag-specific DNA synthesis were low and comparable in the two subject groups. Increased immune responses to Ag 85 at the site of repeated exposure to M. tuberculosis (the lung) may represent an important component of protective immunity against M. tuberculosis. Correlates of protective immunity against M. tuberculosis are required for assessment of the efficiency of anti-tuberculous vaccines.

T here is abundant clinical and epidemiological evidence that supports the existence of protective immunity against Mycobacterium tuberculosis in humans. Approximately 90 -95% of M. tuberculosis-infected humans successfully contain their primary M. tuberculosis infection and never develop tuberculosis. In fact, tuberculosis develops in only 5-10% of M. tuberculosis-infected individuals.
Active immune surveillance is required to maintain the latency of quiescent M. tuberculosis foci, and CD4 ϩ T cells are critical to cell-mediated anti-tuberculosis immunity (1,2), presumably be-cause they are the primary source of the protective cytokine IFN-␥ (3,4). The importance of IFN-␥ responses in mycobacterial infection is underscored by the observation that children with hereditary IFN-␥ receptor 1 deficiency are prone to infection with ubiquitous mycobacteria or dissemination of Calmette-Guérin bacillus after vaccination (5)(6)(7). Further, IFN-␥ was successfully used as adjunctive therapy in patients with refractory nontuberculosis mycobacterial infection (8) and has shown promising results in patients with multidrug-resistant pulmonary tuberculosis (9).
In patients with pulmonary tuberculosis, however, M. tuberculosis Ag-specific IFN-␥-producing cells (10) and spontaneous IFN-␥ production (11) are compartmentalized to the site of infection (the lung) despite ongoing disease. Tuberculosis patients represent subjects who are at one end of the spectrum of M. tuberculosis infection, i.e., individuals who have lost immunological control, reactivated their infectious focus, and developed disease. Tuberculosis patients, thus, may not allow the study of protective immunity operative in the majority of M. tuberculosis-infected individuals.
Healthy household contacts of patients with active tuberculosis (HHC) 5 are repeatedly exposed to M. tuberculosis aerogenically. Under situations of intense and prolonged exposure, the proportion of contacts found to be infected with M. tuberculosis at the time of diagnosis of sputum smear-positive tuberculosis index cases is as high as 30 -40% (12); however, infection rates can be even higher with extended exposure to untreated patients with tuberculosis (up to 80%) (13). The immunological study of M. tuberculosis-reactive HHC therefore may allow elucidation of protective immunity against M. tuberculosis. Because the lung is the primary portal of entry for infection with M. tuberculosis in humans, in the present study pulmonary immune responses of tuberculin skin test-positive HHC to M. tuberculosis Ags were compared with autologous systemic (blood) immune responses and with pulmonary and systemic immune responses of tuberculin skin test-positive healthy control individuals from the community (CC). We found that IFN-␥-producing cells with specificity for Ag 85 complex (Ag 85co), which is considered to be a protective Ag of M. tuberculosis, were compartmentalized to the lungs of HHC, and that such cells were significantly more frequent in bronchoalveolar cells (BAC) from HHC than in those from CC.

Study subjects
Subjects identified as HHC and CC who were willing to volunteer to undergo bronchoalveolar lavage, venipuncture, and HIV-1 serology were screened for participation in this study between 1996 and 1999 at the National Institute of Respiratory Diseases (Mexico City, Mexico). Approval to perform bronchoalveolar lavages, venipunctures, and HIV-1 serologies on consenting healthy individuals was given by the institutional review boards of National Institute of Respiratory Diseases and University Hospitals of Cleveland.
For recruitment of HHC, households of sputum smear-positive (grade 3) patients with active tuberculosis who had been receiving antimycobacterial chemotherapy no longer than 1 wk were chosen. This was done under the assumption that shedding of M. tuberculosis into the room air, and thus exposure of HHC to M. tuberculosis, would be greater in these households than in households of tuberculosis patients with lesser grades of smear positivity (14). All HHC of the household who were willing and met the following eligibility requirements were enrolled: 1) living with the tuberculosis patient for a duration of Ն3 mo, 2) apparent clinical health, 3) normal chest radiograph, 4) tuberculin skin test induration size Ͼ10 mm at 48 -72 h upon injection of 5 tuberculin units (TU) of purified protein derivative of M. tuberculosis (PPD) by the Mantoux technique, 5) age 18 -65 years, and 6) HIV-1 seronegativity.
CC were enrolled from staff at the National Institute of Respiratory Diseases and their relatives; none had any involvement in health care or known contact with patients with active tuberculosis. Eligibility requirements for the CC group were 1) apparent clinical health, 2) normal chest radiograph, 3) tuberculin skin test induration size Ͼ10 mm at 48 -72 h upon injection of 5 TU PPD by the Mantoux technique, 4) age 18 -65 years, and 5) HIV-1 seronegativity.
Exclusion criteria for HHC and CC were 1) blood hemoglobin concentration Ͻ10 g/dl, 2) any chronic systemic (diabetes, cancer) or pulmonary (asthma) disease, 3) upper respiratory tract infection within 1 mo before the study, and 4) immunosuppressive medications or immunosuppressive diseases.
After obtaining written informed consent from eligible HHC and CC, bronchoalveolar lavage (BAL) and venipuncture were performed in the morning on the day of each experiment.

Preparation of BAC
BAL was performed using an Olympus flexible fiberoptic bronchoscope as previously described (15). Briefly, after local anesthesia of the upper airways with 2-4% lidocaine, 150 ml of sterile isotonic saline was instilled into each of two segments of the right middle lobe. BAL fluid was centrifuged at 300 ϫ g for 15 min at 4°C. BAC were counted and resuspended in complete medium (RPMI 1640 (BioWhittaker, Walkersville, MD) with 50 g/ml gentamicin sulfate, 200 mM L-glutamine, and 10% heat-inactivated pooled human serum).

Preparation of PBMC
Between 40 and 60 ml of whole heparinized venous blood were obtained from each study subject. PBMC were prepared by centrifugation of the blood over Ficoll-Paque gradients (16) and were counted and resuspended in complete medium.

Antigens
PPD was purchased (Statens Seruminstitut, Copenhagen, Denmark). Mannose-capped lipoarabinomannan (manLAM; purified from M. tuberculosis strain H37Rv culture filtrate; endotoxin contamination, 3.46 ng/mg man-LAM, determined by Limulus amebocyte assay) and Ag 85co purified from culture filtrate from H37Rv and its three closely related components, Ag 85A, Ag 85B, and Ag 85C (17), were provided by Dr. John T. Belisle (Department of Microbiology, Colorado State University, Fort Collins, CO). Ag 85co is a major secreted protein of metabolically active M. tuberculosis, is fibronectin binding (18,19), and functions as a mycolyltransferase that is involved in the final stages of mycobacterial cell wall assembly (20,21). Secreted protein from short term culture filtrate of M. tuberculosis, ESAT-6 (endotoxin contamination, Ͻ0.05 ng/g protein) (22), and short term culture filtrate (ST-CF; endotoxin contamination Ͻ0.05 ng/g protein) (23) were additional M. tuberculosis Ags used in this study. A nonmycobacterial Ag prepared from Candida albicans (Candida; Greer Laboratories, Lenoir, NC) was used to assess the specificity of responses. All Ags were used in concentrations that resulted in peak stimulation, as determined in previous studies (10). Dose responses were determined for Ag 85A, Ag 85B, Ag 85C, ESAT-6, and ST-CF for optimal induction of DNA synthesis, and performance of IFN-␥ and IL-10 enzymelinked immunospot (ELISPOT) assays.

DNA synthesis
PBMC and BAC were added to round-bottom 96-well plates (Nunc, Copenhagen, Denmark) at 5 ϫ 10 4 cells/well. Cells were stimulated with mycobacterial and nonmycobacterial Ags and the mitogen PHA (Sigma, St. Louis, MO). Complete medium was used as a negative control stimulus. Cultures were pulsed with 1 Ci/well [ 3 H]thymidine (Amersham, Aylesbury, U.K.; sp. act., 5 Ci/mmol) on day 4, and cells were harvested 24 h later. Incorporation of [ 3 H]thymidine was determined with a scintillation counter and expressed as counts per minute.

ELISAs
Twenty-four-hour supernatants from short term cultures of PBMC and BAC that were stimulated with mycobacterial and nonmycobacterial Ags were assessed by commercially available ELISA kits for their content of IFN-␥ (Endogen, Woburn, MA), and TNF-␣ (R&D Systems, Minneapolis, MN).

ELISPOT assays
Because cytokine levels in cell culture supernatants reflect production, degradation, and consumption by cells, ELISPOT analysis was performed to determine the frequency of cytokine-producing cells and to assess cytokine production on a single-cell basis. ELISPOT analysis is 10-to 200-fold more sensitive than ELISA (24). Frequencies of cytokine-producing cells in PBMC and BAC were determined by ELISPOT assay as previously described (25). Briefly, T spot 96-well assay plates (Whatman, Clifton, NJ) were coated overnight at 4°C with primary Abs against either IFN-␥ (Endogen) or IL-10 (PharMingen, San Diego, CA) at 2-10 g/ml. Plates were washed extensively and blocked with 1-2% BSA (Sigma). Then, PBMC and BAC were added to the Ab-coated wells at a concentration of 10 5 cells/well and stimulated with mycobacterial (PPD, manLAM, Ag 85co, Ag 85A, Ag 85B, Ag 85C, ESAT-6, and ST-CF) and nonmycobacterial (Candida) Ags and PHA. Complete medium was used as a negative control stimulus. Cultures were incubated at 37°C, and cells were removed after 24 h (for IFN-␥) and 72 h (for IL-10) by washing the wells three times with PBS. Biotinylated secondary Abs to IFN-␥ (Endogen, Woburn, MA) and IL-10 (PharMingen) were then added to the wells, and plates were incubated at 4°C overnight. Peroxidase-conjugated streptavidin (Dako, Glostrup, Denmark; 1/2000) was added, and spots were visualized with 1% 3-amino-9-ethylcarbazole (Pierce, Rockford, IL).
The frequencies of IFN-␥-and IL-10-producing cells in each well were determined using a computerized series 1 ImmunoSpot Image Analyzer (Cellular Technology, Cleveland, OH). The average number of cytokine spots in duplicate or triplicate wells was then calculated.

Statistical analysis
Comparisons between the HHC and CC groups were made using nonparametric methods: Wilcoxon's two-sample rank-sum test for comparisons of distributions or Fisher's exact test when comparing proportions. In comparisons within groups examining paired samples of lung vs blood, Wilcoxon's signed-rank test (for nonparametric data) was used. All statistical tests were conducted using SigmaStat (version 2.0, Jandel, Chicago, IL) and SPSS for Windows (version 8.0, SPSS, Chicago, IL) statistical software. Statistical significance was set at p Յ 0.05.

Characteristics of study groups and cells
Ten HIV-1-seronegative HHC (7 men and 3 women) and 15 HIV-1-seronegative CC (11 men and 4 women) underwent BAL and venipuncture to obtain BAC and PBMC. The median age of the HHC group (23.5 years) was not different from that of the CC group (24 years). An equal number (n ϭ 4) of participants in each group were smokers. HHC were family members (spouse (n ϭ 2), son (n ϭ 2), daughter (n ϭ 4), sister (n ϭ 1), life partner (n ϭ 1)) who lived in the same house (or the same room) with the index patient for a duration of at least 3 mo before the BAL.
The following median parameters were comparable between HHC and CC: body weight (62 and 67 kg), serum albumin (4.6 and 4.6 g/dl), peripheral white blood cell count (6.1 and 6.5 ϫ 10 3 /l), and hemoglobin (16.3 and 15.9 g/dl), respectively. Median proportions of neutrophils were slightly higher in HHC than CC (63.8 and 56.1%, respectively; p ϭ 0.045). In four of six HHC and seven of seven CC examined, scars indicating preceding vaccination with Calmette-Guérin bacillus were found. Tuberculin skin test indurations to 5 TU of PPD were comparable (20 and 18 mm in HHC and CC, respectively).
Retrieved proportions of BAL fluid, yield of BAC, and cellular profiles of BAC, including proportions of peroxidase-positive cells (median, 2 and 1.5%) were comparable in HHC and CC, respectively. Proportions of alveolar neutrophils were Յ1% in BAC of both HHC and CC. Median proportions of alveolar lymphocytes were 14.1 and 14.0% and of alveolar macrophages were 85.8 and 83.9%, in HHC and CC, respectively, similar to those in CC in previous studies (10,26).

DNA synthesis of BAC and PBMC in response to M. tuberculosis Ags
DNA synthesis (incorporation of [ 3 H]thymidine) was assessed in PBMC (Fig. 1, A and C) and BAC (Fig. 1, B and D) from HHC (A and B) and CC (C and D) to optimal concentrations of Candida (10 g/ml), PPD (10 g/ml), Ag 85co (5 g/ml), ESAT-6 (2 g/ml), ST-CF (1 g/ml), manLAM (10 g/ml), and PHA (1 g/ml). As expected from the cell composition of BAC in HHC and CC, DNA synthesis in response to all stimuli was low in the two groups. DNA synthesis in response to PPD, Ag 85co, ST-CF, Candida, and PHA was significantly lower ( p Ͻ 0.05) in BAC than in PBMC from both HHC and CC. Interestingly, DNA synthesis in PBMC from HHC in response to PPD, Ag 85co, and PHA was significantly lower than that in CC ( p Ͻ 0.05).

Frequencies of M. tuberculosis Ag-specific PBMC and BAC
The concentrations of Ags used for optimal induction of IFN-␥ spots were identical with those used for induction of DNA synthesis (see above). PBMC and BAC from HHC ( Fig. 2A) and CC (Fig. 2B) were stimulated with culture medium, PPD, and Ag 85co. Fig. 2 depicts the frequencies of Ag-specific IFN-␥-producing cells (i.e., IFN-␥ spots) in pairs of autologous PBMC and BAC of HHC and CC. Due to the limited availability of lung cells, not all Ags could be tested in all individuals.
The median frequency of Ag 85co-specific IFN-␥-producing cells was significantly higher in BAC from HHC (n ϭ 10) than in BAC from CC (n ϭ 13; p ϭ 0.022). The frequencies of PPDspecific IFN-␥-producing cells were not statistically different in BAC of HHC compared with those of CC. Further, the frequencies of PPD-and Ag 85co-specific IFN-␥-producing cells were higher in BAC than in autologous PBMC from six and seven HHC, respectively. In contrast, in CC, the frequencies of PPD-and Ag 85co-specific IFN-␥-producing cells were lower in BAC than in autologous PBMC in 7 of 15 CC tested for PPD and in 7 of 13 CC tested for Ag 85co. When a cutoff of 3 was chosen for the fold increase in frequencies of IFN-␥-producing cells in BAC compared with that in autologous PBMC, the frequencies of IFN-␥producing cells in BAC of HHC (n ϭ 10) were 3-fold higher than those in autologous PBMC in 40% to PPD and in 50% to Ag 85co. Using the same criteria, the frequencies of IFN-␥-producing cells in BAC of CC were higher than those in autologous PBMC in 7% to PPD (1 of 15) and in 12% to Ag 85co (1 of 13; p ϭ 0.05 for Ag 85co, HHC vs CC). Thus, M. tuberculosis-specific IFN-␥-producing cells appeared to be enriched in the lungs (compared with the blood) of HHC and were present at higher frequencies in BAC from HHC than in those from CC. In only a limited number of CC (n ϭ 2) were the frequencies of PPD-and Ag 85co-specific IFN-␥-producing BAC higher than those in PBMC, indicating possible exposure to M. tuberculosis within the community.

FIGURE 2.
Frequencies of IFN-␥-producing cells in PBMC and BAC of HHC and CC by ELISPOT assay. Frequencies of IFN-␥-producing cells in PBMC and BAC from tuberculin skin test-positive healthy household contacts (HHC; n ϭ 10) of patients with active tuberculosis (A) and healthy tuberculin skin test-positive community controls (CC; n ϭ 15; B) were determined by ELISPOT assay and are expressed as IFN-␥ spots per 10 5 PBMC and BAC. PBMC and BAC were stimulated with PPD and Ag 85co from M. tuberculosis H37Rv culture filtrate (Ag 85co). Control cultures (background) were stimulated with medium alone (Medium). Results are presented as pairs of autologous PBMC and BAC data for each study individual. Visibility of overlapping data points was improved by separating the data points through minor horizontal shifting. F, HHC and CC individuals in whom frequencies of IFN-␥-producing cells in BAC were Ն3-fold higher than those in PBMC (A, PPD (n ϭ 4) and Ag 85co (n ϭ 5); B, PPD (n ϭ 1) and Ag 85co (n ϭ 1)). Frequencies of IFN-␥-producing BAC in response to Ag 85co were significantly higher in BAC from HHC (A) than in BAC from CC (B; p ϭ 0.022).
To estimate the output of IFN-␥ from IFN-␥-producing cells in PBMC and BAC from HHC and CC, sizes of spots were measured with the image analyzer. Proportions of IFN-␥ spots from Ag 85co-stimulated PBMC and BAC of HHC (n ϭ 8) and CC (n ϭ 8) within each size category (10 Ϫ1 -10 Ϫ3 mm 2 ) were comparable between BAC and PBMC from HHC and CC, indicating comparable outputs of IFN-␥ per cell (Fig. 3).
In a subset of study subjects (HHC, n ϭ 6; CC, n ϭ 6) frequencies of IFN-␥-producing cells were also assessed in response to the three components of Ag 85co, namely, Ag 85A (5 g/ml), Ag 85B (5 g/ml), and Ag 85C (5 g/ml; Table I). The ratios of the frequencies of Ag 85A-and Ag 85B-specific, but not of Ag 85Cspecific, IFN-␥-producing BAC and PBMC (BAC/PBMC) were significantly higher in HHC than in CC ( p ϭ 0.01).
The frequencies of PPD-and Ag 85co-specific IL-10-producing PBMC and BAC in HHC and CC, respectively, were comparable between cell populations and subject groups (data not shown).

IFN-␥and TNF-␣ concentrations by ELISA
In a subgroup of study subjects (HHC, n ϭ 6; CC, n ϭ 8), 24-h culture supernatants from in vitro Ag 85co-and PPD-stimulated PBMC and BAC were available for analysis of IFN-␥ and TNF-␣ concentrations by ELISA. IFN-␥ production in response to Ag 85co was significantly lower in BAC from HHC than in BAC from CC ( p ϭ 0.032). Thus, high frequencies of Ag 85co-specific IFN-␥-producing cells in BAC from HHC were not accompanied by a corresponding higher concentration of IFN-␥ in cell culture supernatants.

Discussion
This study is the first to compare the pulmonary immune responses of healthy individuals from households of patients with active smear-and culture-positive tuberculosis, and therefore with a high likelihood of exposure to M. tuberculosis, with those from healthy M. tuberculosis-sensitized individuals in a tuberculosis-endemic community. In HHC, Ag 85co-specific IFN-␥-producing BAC were compartmentalized to the alveolar spaces. Further, frequencies of Ag 85co-specific, IFN-␥-producing cells in BAC from HHC were higher than those in BAC from CC. Also, frequencies of Ag 85A-and Ag 85B-specific, but not of Ag 85C-specific, IFN-␥-producing cells were significantly higher in BAC from HHC than in those from CC. Considering that the proportions of lymphocytes in alveolar spaces of healthy subjects (i.e., both HHC and CC) are 4-to 5-fold lower than that in peripheral blood, the expansion of Ag 85co-specific IFN-␥-producing cells in BAC from HHC to proportions higher than those in PBMC is notable. These findings are consistent with aerogenic M. tuberculosis infection/ exposure in HHC of patients with tuberculosis. Thus, it appears that during the early phases of M. tuberculosis infection, Ag 85cospecific pulmonary mononuclear cells are expanded.
Ag 85co is a predominant product of live M. tuberculosis (19,27) and may be critical in the development of protective immunity in M. tuberculosis infection (28). Recently, it has been shown that vaccination with Ag 85co conferred protection against M. tuberculosis challenge in mice (29 -31) and guinea pigs (32,33). In studies of human immune responses to Ag 85, PBMC from healthy tuberculin skin test-positive subjects demonstrated strong DNA synthesis and IFN-␥ secretion in response to both Ag 85co and Ag 85A (34). Further, IFN-␥ production of Ag 85-stimulated PBMC from HHC was significantly higher than that of PBMC from patients with tuberculosis (35). In fact, DNA synthesis of PBMC in response to Ag 85co was low in 52% (33,35) to 100% of patients with active tuberculosis (36) relative to that in healthy control individuals. Further, IFN-␥ production in response to Ag 85A was  a Frequencies of Ag 85A-, Ag 85B-, and Ag 85C-specific IFN-␥ spots were measured per 10 5 of BAC and PBMC, respectively, in healthy household contacts (HHC, n ϭ 6) and healthy tuberculin skin test positive community controls (CC, n ϭ 6). Presented are medians and 25th and 75th percentiles (in brackets) of the ratios of frequencies of Ag 85A-, Ag 85B-, and Ag 85C-specific IFN-␥ spots in BAC over that in PBMC. NS, not significant between the study groups. low in patients with pulmonary tuberculosis (34,37). Overall, it appears that T cell responses to Ag 85co may be indicative of host protective immunity against M. tuberculosis in humans. Our finding of increased frequencies of Ag 85co-specific IFN-␥-producing BAC in tuberculin skin test-positive HHC may indicate the expansion of local protective immunity in these subjects who have a high probability of having been infected with M. tuberculosis.
The observed differences between frequencies of Ag 85-specific IFN-␥-producing BAC in HHC and CC are unlikely to be due to differences in the composition of blood and lung cells, as no differences in number or percentage of lymphocytes and mononuclear phagocytes were found by histochemical analysis of BAC and PBMC from HHC and CC. In particular, proportions of lymphocytes were comparable between the study groups. Additionally, frequencies of IFN-␥-producing cells that were specific for M. tuberculosis Ags other than Ag 85co (PPD, ST-CF, manLAM, ESAT-6), were comparable in BAC from both study groups. Thus, the increased proportions of Ag 85co-specific IFN-␥-producing BAC in HHC probably resulted from expansion of Ag 85co-specific BAC that occurred in response to recent aerogenic M. tuberculosis infection/exposure in the households of tuberculosis patients. This is consistent with the finding in HHC that when frequencies of Ag-specific IFN-␥-producing cells were normalized to 10,000 lymphocytes in both PBMC and BAC, numbers of PPDand Ag 85-specific IFN-␥-producing cells were significantly higher in lymphocytes in BAC than in lymphocytes in PBMC. However, because recent data indicate that M. tuberculosis-infected AM might be sources of IFN-␥ (38), normalization of IFN-␥-producing cells to numbers of lymphocytes in BAC may not reflect the true range of IFN-␥-producing cell populations in BAC. It was beyond the scope of this study to unravel the cellular source of IFN-␥ and to assess the influence of exposure to M. tuberculosis on the cell populations involved in production of IFN-␥. Elucidation of the cellular source of IFN-␥ in HHC and CC is the subject of ongoing research.
Compartmentalization of Ag 85co-specific IFN-␥-producing BAC in lungs from HHC was detectable only by a highly sensitive assay (ELISPOT) and not by DNA synthesis or IFN-␥ immunoreactivity in supernatants from cell cultures.
The implications of increased frequencies of M. tuberculosisspecific IFN-␥-producing cells in the lungs of HHC are not presently clear, as IFN-␥-producing M. tuberculosis Ag-specific BAC are also detectable in high frequencies in BAC from radiographically affected areas of lungs from tuberculosis patients (10).
Time points of conversion of tuberculin skin tests from negative to positive could not be assessed in HHC due to the cross-sectional nature of this study. Definitive diagnosis of new infection or reinfection with M. tuberculosis in HHC as a result of contact with the tuberculosis patients, therefore, was not possible. Several observations, however, might be indicative of recent exposure to or infection with M. tuberculosis in HHC in this study. First, slightly increased proportions of circulating neutrophils in the peripheral blood might indicate a systemic and recent inflammatory response. Further, TNF-␣ production by BAC from HHC in response to Ag 85co was increased (although the difference from CC did not reach statistical significance) and might indicate local inflammatory immune responses. Lastly, lower levels of DNA synthesis in PBMC of HHC compared with CC in response to PPD, Ag 85co, and PHA might suggest the activation of immunosuppressive circuits due to recent active M. tuberculosis infection, as also seen with other infections (39).
The frequencies of IFN-␥-producing ST-CF-specific cells were higher in BAC than in PBMC in the majority of HHC, but were not significantly different from those in BAC of CC. These differences may, however, reach significance if larger populations are studied. ST-CF is a crude mixture of secreted Ags and contains Ag 85co and a multitude of other Ags. Also, since the concentration of purified Ag 85co used in the current study (5 g/ml) exceeded that in crude ST-CF (at most 1 g/ml), differences in the detectable frequencies of IFN-␥-producing Ag 85co-and ST-CF-specific cells may have resulted from suboptimal stimulation of cells by the latter Ag mixture. Alternatively, ST-CF may contain components with suppressive activities for expression of IFN-␥. Demissie et al. have recently shown that ST-CF induces greater DNA synthesis and higher IFN-␥ production in PBMC from HHC compared with those from patients with tuberculosis (40). Interestingly, levels of IFN-␥ production were significantly higher in ST-CF-stimulated PBMC from HHC of patients with advanced tuberculosis compared with those from HHC of patients with minimal disease, suggesting an association with more intense exposure to M. tuberculosis.
We found that the frequencies of IFN-␥-producing BAC with specificity for the purified components of Ag 85co, namely, Ag 85A and Ag 85B, but not Ag 85C, were significantly higher in BAC of HHC than in BAC of CC, resembling the findings with the purified Ag 85co. The relative immunogenicity of the components of Ag 85co (A, B, and C) are not known to differ despite the predominance of Ag 85B in purified Ag 85co. Also, whether there are differences among Ag 85A, -B, and -C with regard to biological activity and/or induction of protective immunity is not known.
In conclusion, recent aerogenic exposure/infection with M. tuberculosis is associated with the accumulation of Ag 85co-specific IFN-␥-producing cells in the alveolar spaces of HHC. The finding of compartmentalization of Ag 85co-specific IFN-␥-producing cells to the alveolar spaces of HHC may reflect the activation of early protective immune responses in situ. The understanding of human protective immune responses to M. tuberculosis at sites of infection may facilitate the identification of correlates of protective immunity, which are urgently needed in the evaluation of the efficacy of new anti-tuberculosis vaccines.