Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Tuberculosis is spread from person to person by aerosolized Mycobacterium tuberculosis. The majority of infected individuals readily control primary infection and do not progress to clinical tuberculosis (1). Cell-mediated immunity in the lung, mediated by T cells and macrophages, controls infection, and granulomas are the pathological hallmark for this protective immune response. As a result, tubercle bacilli are contained within developing granulomas and although most bacteria are destroyed, a small number of bacilli persist and may cause reactivation disease. Little is known about the initiation and expression of protective T-cell responses in the lung during early and quiescent mycobacterial infection.

Although the lung is the portal of entry, it is also uniquely susceptible to M. tuberculosis infection. In mice, lethal infection with aerosolized M. tuberculosis requires a 100-fold smaller inoculum than does intravenous infection (2). M. tuberculosis infection is readily controlled in liver and spleen, in contrast to lung wheredespite receiving only 1% of an intravenous inoculumthe infection rapidly becomes uncontrolled (3). Like M. bovis-Calmette-Guerin bacillus (BCG) infection, T-cell responses develop and contain M. tuberculosis growth but are unable to reduce the numbers of M. tuberculosis. Apart from the inhibitory capacity of alveolar macrophages for naive T-cell activation, little is known about the cellular mechanisms responsible for the lung's unique susceptibility to tuberculosis (6, 7).

Early growth or elimination of M. bovis-BCG Montreal in various inbred strains of mice has been used to classify mice as susceptible (BCG^s) or resistant (BCG^r) before the development of antigen-specific immunity (8). A genetic locus expressed exclusively in macrophages, designated NRAMP for BCG (lsh for Leishmania donovani and Ity for Salmonella typhimurium) has been identified, cloned and partially characterized (11, 12). For M. bovis- BCG Montreal, susceptible and resistant phenotypes have been best defined using mycobacterial growth in spleen after intravenous infection. However, BCG^r mice may appear susceptible when the intravenous mycobacterial inoculum size is increased, and BCG^s mice appear resistant when infected with less-virulent M. bovis-BCG strains Japan and Prague (13, 14). Further, BCG^r and BCG^s mice appear equally susceptible to growth of virulent M. tuberculosis Erdman in the spleen (15). Expression of the BCG^r and BCG^s phenotypes is much less clear for mycobacterial growth in the lung after aerogenic or intravenous infection. Both BCG-susceptible and -resistant strains of mice acquire specific T-cell immunity during mycobacterial infection in the lung (15).

Recruitment and activation of T cells are critical for protective immunity to M. tuberculosis. Studies in humans and animal models have established that CD4⁺, CD8⁺, and T-cell receptor-positive (TCR⁺) T-cell subsets contribute to the cellular response to M. tuberculosis (16- 22). Few studies have specifically addressed T-cell recruitment and activation during primary and quiescent phases of mycobacterial infection in different lung compartments. A better understanding of the protective cellular immune mechanisms in the lung microenvironment are necessary not only to understand the basic biology of host defenses to M. tuberculosis but also for vaccine development, immunotherapy, and efforts to identify surrogate markers for protective immunity.

Studies of primary pulmonary infection require animal models because access to human lung tissue during primary M. tuberculosis infection is exceedingly difficult. Primary infection of mice with M. bovis-BCG, the vaccine strain used in humans, results initially in mycobacterial growth followed by control and near complete clearance of organisms from lung (23). This model mimics the control of primary M. tuberculosis infection in humans. In contrast to M. bovis-BCG, infection of BCG-resistant or -susceptible mice with virulent M. tuberculosis results in rapid growth in lungs followed by control without clearance of bacteria (24). High bacterial burdens persist in the lung (10⁶ to 10⁸ colony-forming units [CFUs]/lung) associated with marked cellular inflammation. Bacterial growth resumes after 6 to 7 mo with animals succumbing to progressive lung infection (2, 22). Thus primary infection with M. bovis-BCG represents a good model in which to analyze the development of protective immunity in the lung. This study addresses T-cell subset expansion and cytokine expression in bronchoalveolar spaces, lung parenchyma, and mediastinal lymph nodes of mice infected intratracheally with M. bovis-BCG to determine whether compartmentalization of immune responses occurred with the lung. Further, this study determined changes in histopathology and bacterial burden that coincided with early and resolving stages of T-cell activation. We found that early T-cell activation is initiated in and restricted primarily to lung parenchyma compared with regional and distant lymphoid organs.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Mice

Pathogen-free, female C57Bl/6 mice (H-2^b- and BCG-susceptible) were purchased from Charles River Laboratories (Wilmington, MA) and used at between 9 and 12 wk of age. Infected and control mice were housed in microisolator cages (Lab Products, Inc., Maywood, NJ) and fed a standard rodent diet and water ad libitum.

Bacteria

M. bovis-BCG Connaught (ATCC #35745) was grown for 18 to 21 d in Middlebrook 7H9 broth supplemented with 10% albumin, dextrose, and NaCl (ADC) and 0.2% glycerol, per manufacturer's instructions (Difco Laboratories; Detroit, MI). M. bovis-BCG was harvested at mid-log phase by pelleting at 14,000 × g for 10 min at 4°C and resuspended in fresh, cold 7H9 broth. The bacterial suspension was vortexed vigorously with 5 cc of sterile glass beads for 5 min. To further disrupt clumps, M. bovis-BCG was sonicated (90 W; 20 kHz; Heat Systems-Ultrasonics; Farmingdale, NY) on ice for 1 min (five times). Glass beads and remaining BCG aggregates were pelleted at 200 × g for 5 min at 4°C. The supernatant was collected, sonicated an additional 2 min on ice, aliquotted, and stored at 70°C.

Antibodies

The following phycoerythrin (PE)-conjugated or fluoroscein isothiocyanate (FITC)-conjugated monoclonal antibodies were purchased from PharMingen (San Diego, CA): PE hamster immunoglobulin (Ig)G anti-tri-nitrophenol (TNP) (#11095A), PE hamster antimouse CD3 (#01085A), FITC rat IgG2a, (#11024C), FITC hamster IgG anti-TNP (#11094), FITC rat antimouse CD4 (#01064A), FITC rat antimouse CD8a (#01044A), and FITC hamster antimouse TCR (#01413D).

Infection

For each experiment, fresh M. bovis-BCG was thawed, pelleted (14,000 × g for 15 min), resuspended in phosphate-buffered saline (PBS), sonicated (1 min), and diluted to 10⁷ CFUs/ml in PBS. Mice were anesthetized intraperitoneally with tribromoethanol solution (TBE) (180 mg/kg body weight). The trachea was exposed by incising the skin and fascia in the midline. A total of 10⁵ CFUs in 10 µl PBS was injected using a 25-µl microsyringe (Fisher, Pittsburgh, PA). The skin was closed with 6-0 absorbable Vicryl (Ethicon, Somerville, NJ). Control mice were injected with 10 µl of PBS and housed separately from infected mice. There were no recorded deaths due to surgery, wound infection, or M. bovis-BCG (monitored 8 to 9 mo) infection.

Bronchoalveolar Lavage

At designated intervals after inoculation, control and infected mice were anesthetized with a lethal dose of TBE (240 mg/kg). The abdominal cavity was incised and each mouse exsanguinated by splenectomy and transection of the left renal artery and vein. Before removal of the lungs, the trachea was exposed and the chest decompressed through a transverse incision at the level of the xiphoid. Bronchoalveolar lavage (BAL) was performed by inserting an 18-gauge Angiocath intravenous catheter (Becton Dickinson, Sandy, UT) through a 1-mm incision at the first tracheal ring. A total of 1.2 ml PBS was instilled in three separate 0.4-ml lavages. BAL cells were centrifuged at 2,000 × g for 10 min at 4°C, resuspended in RPMI-1640 (BioWhittaker, Walkersville, MD), and counted in 0.4% Trypan Blue. Serial dilutions were also prepared and plated on 7H10 agar to determine the number of CFUs associated with BAL cells. CFUs were not detected in BAL supernatants. Cytospin slides of 2.5 × 10⁴ cells/BAL/mouse were prepared using a Cytospin 3 centrifuge (600 rpm for 8 min; Shandon, Pittsburgh, PA) and stained with either Diff-Quik (Dade Diagnostics, Aguada, PR) for differential cell counts or Ziehl-Neelson for the detection of acid-fast bacilli (AFB).

Homogenization of Lung, Lymph Node, and Spleen

Lung-cell homogenates were prepared using modifications of previously described techniques (25). Briefly, after BAL, lungs were perfused by gently infusing 10 ml PBS into the right ventricle of the heart to minimize contamination with blood. Perfused lungs were dissected from the mainstem bronchi, minced with scalpels in 4 ml of RPMI, and digested with 500 U of type IV collagenase (C5138; Sigma Chemical Co., St. Louis, MO). After 1.5 h incubation at 37°C, the tissue was further disrupted by sequential passage through 18G and 21G needles. The homogenate was clarified by filtration through sterile cotton and centrifuged (800 × g for 15 min). The cells were resuspended in red blood cell (RBC) lysis buffer (containing 10 mM Tris-HCl and 0.83% ammonium chloride), incubated at room temperature for 5 min, and pelleted (800 × g for 15 min). The lung cells were resuspended in PBS and held on ice.

Mediastinal lymph nodes were crushed in RPMI with sterile syringe plungers, triturated though a 21G needle, pelleted (800 × g for 15 min), and resuspended in RPMI. Spleens were first crushed in RBC lysis buffer, centrifuged (800 × g for 15 min), and resuspended in RPMI on ice. Viable lung, spleen, and lymph node cells were counted by Trypan Blue exclusion.

CFU Determinations

Aliquots of BAL cells and tissue homogenates were serially diluted in PBS (containing 0.05% Tween 20), plated on Middlebrook 7H10 agar enriched with 10% o-ADC (Difco) and 0.5% glycerol, and incubated at 37°C for 2 to 3 wk. Colonies were counted and total tissue CFUs calculated based on the volume of homogenate obtained from each organ/mouse.

Cell Staining and Flow Cytometry

A total of 5 × 10⁵ lung, spleen, or lymph node cells were resuspended in 5% bovine serum albumin in PBS and incubated for 10 min on ice. After blocking, cells were resuspended in 60 µl of cold PBS containing either PE- and FITC-conjugated isotypic control antibodies or PE-conjugated anti-CD3 and FITC-conjugated anti-CD4, CD8, or TCR. Cells were stained on ice for 45 min, washed in PBS, fixed in 1% paraformaldehyde in PBS (pH 7.4), and stored at 4°C before fluorescence-activated cell sorter (FACS) analysis.

Cells were analyzed using a Becton Dickinson FACScan flow cytometer (Becton Dickinson, San Jose, CA). Forward light scatter, side light scatter, FL1 (FITC), and FL2 (PE) settings were determined empirically to maximize the difference between isotype staining (negative) and positive CD3⁺ staining. A total of 10,000 ungated events (i.e., cells) were recorded and the percentage of CD3⁺-CD4⁺ and CD3⁺-CD8⁺ within the lymphocyte gate measured. CD3⁺-⁺ T cells were identified outside the typical lymphocyte gate consistent with their larger size and more complex light scattering properties (26). The percentage CD3⁺-⁺ T cells was determined directly from 10,000 ungated events. Dot-plot and histogram analyses were performed using PC-LYSIS (Becton Dickinson). At each time point, total organ (lung, node, and spleen) T-cell number was calculated by multiplying the percent T-cell subtype by the total number of cells isolated from lung, lymph node, and spleen from individual mice.

Spontaneous Cytokine Production and Interferon- Measurement

A total of 1 × 10⁶ lung, spleen, or lymph node cells were washed in RPMI and pelleted for 15 min (800 × g). Cells were resuspended in 0.5 ml RPMI-1640 medium containing 10% heat-inactivated fetal calf serum (Hyclone, Logan, UT), 50 µM 2-mercaptoethanol, 20 mM N-2-hydroxyethylpiperazine-N'-ethane sulfonic acid, 2 mM L-glutamine, 1 mM sodium pyruvate, and 100 mM nonessential amino acids and penicillin-streptomycin (100 IU/ml and 100 µg/ ml, respectively), and were cultured (2 × 10⁶/ml) for 72 h at 37°C in 5% CO₂. Supernatants were harvested, filtered, and stored at 70°C.

Recall immune responses were evaluated using purified CD4⁺ and CD8⁺ T cells isolated from lung and lymph node cell homogenates. CD4⁺ and CD8⁺ T cells were isolated (> 90% purity) using the MiniMACS magnetic microbead system according to the manufacturer's directions (Miltenyi Biotec, Inc., Sunnyvale, CA). Naive mouse spleen cells were isolated as described, cultured overnight in the presence of interferon (IFN)- (45 U/ml), irradiated (1,500 to 2,000 rad), and used as antigen-presenting cells. A total of 2 × 10⁵ irradiated spleen cells were washed three times and incubated for 72 h with 75,000 purified T cells and viable M. bovis-BCG (2 × 10⁵ ml) in 200 µl of culture medium in round-bottom 96-well plates. For negative and positive controls, T cells and spleen cells were incubated with medium alone or concanavalin A (Sigma) at a concentration of 1 µg/ml. Supernatants were removed after 72 h incubation and frozen at 70°C.

To measure IFN-, Immulon 4 plates (Dynatech Laboratories, Chantilly, VA) were coated overnight at room temperature with XMG antimouse-IFN- (MM-700; Endogen, Woburn, MA) diluted in PBS (2 µg/ml) and washed twice with 50 mM Tris (0.2% Tween-20, pH 8.0). After blocking with Superblock (Pierce, Rockford, IL), plates were washed three times and 50 µl of standard IFN- or culture supernatant was added to each well for 1 h. Without washing, 100 µl of biotinylated-XMG (MM-700B; Endogen) was added and the plates were incubated for an additional hour at room temperature. After three washes, 100 µl streptavidin-peroxidase (200 ng/ml) was added and the mixture was incubated for 30 min at room temperature and washed three times. The peroxidase substrate (TMB, Sigma T-4305) was added and incubated in the dark for 30 min. After stopping the reaction with 3 N HCl, absorbance was measured at 450 nm and IFN- levels were estimated by comparison with a standard curve. The lower limit of detection was 31.25 pg/ml.

Semiquantitative Reverse Transcriptase/Polymerase Chain Reaction

Total cellular RNA was isolated from 2 × 10⁶ lung cells. Briefly, lung cells were homogenized in 0.4 ml of Tri- reagent (Molecular Research Center, Cincinnati, OH) and extracted according to the manufacturer's instructions. The RNA was quantitated spectrophotometrically and 1.0 µg of intact, total cellular RNA was reversed transcribed as described previously (27). Semiquantitative conditions for amplification were determined experimentally for each primer pair. We determined the optimal number of cycles whereby the polymerase chain reaction (PCR) product was detectable in an amount proportional to the input quantity of complementary DNA (cDNA). Aliquots of cDNA, 5 µl (diluted 1:4), were PCR-amplified in a 25-µl reaction volume for hypoxanthine phosphoribosyltransferase (HPRT), IFN-, interleukin (IL)-4, IL-5, IL-10, IL-12 p40, IL-12 p35, and tumor necrosis factor (TNF)-. All samples were first denatured at 95°C for 30 s and cycled with nonsaturating, log-linear conditions as follows: (1) for HPRT, IFN-, IL-10, and TNF-, respectively, 24, 24, 29, and 29 cycles of denaturation (95°C for 30 s), annealing (60°C for 30 s), and extension (72°C for 2 min); (2) for IL-12 p40, IL-12 p35, and IL-4, respectively, 34, 32, and 36 cycles of denaturation (95°C for 30 s), annealing (60°C for 1 min), and extension (72°C for 2 min); and (3) for IL-5, 32 cycles of denaturation (95°C for 30 s), annealing (60°C for 1 min), and extension (72°C for 2 min). All samples were incubated at 72°C for 10 min to complete extension. The primer sequences were as follows: for HPRT, 5'-GTT GGA TAC AGG CCA GAC TTT GTT G-3' (sense) and 5'-GAT TCA ACT TGC GCT CAT CTT AGG C-3' (antisense); for IFN-, 5'-TGT TAC TGC CAC GGC ACA GTC ATT-3' (antisense) and 5'-GTG GAC CAC TCG GAT GAG CTC ATT-3'; for IL-12 p40, 5'-AGA GGT GGA CTG GAC TCC CGA-3' (sense) and 5'-CCT GAT GAA GAA GCT GGT GCT-3' (antisense); for IL-12 p35, 5'-GTG CCT TGG TAG CAT CTA TGA-3' (sense) and 5'-AGA GAA GCG ATG GAG GGC ACC-3' (antisense); for IL-5, 5'-GTG AAA GAG ACC TTG ACA CAG CTG-3' (sense) and 5'-CAC ACC AAG GAA CTC TTG CAG GTA-3' (antisense); for IL-10, 5'-TCA AAC AAA GGA CCA GCT GGA CAA-3' (sense) and 5'-ATC AGA TTT AGA GAG CTC TGT CTA-3' (antisense); for IL-4, 5'-CTA GTT GTC ATC CTG CTC TTC TTT-3' (sense) and 5'-CTT TAG GCT TTC CAG GAA GTC TTT-3'; for TNF-, 5'-GAT CTC AAA GAC AAC CAA CTA GTG-3' (sense) and 5'-CTC CAG CTG GAA GAC TCC TCC CAG-3' (antisense). Aliquots, 20 µl, of PCR products were electrophoresed in 1.5% agarose and visualized using ethidium bromide staining as described. All gels were photographed similarly using an f5.6 aperture at 89 cm × 2 s. The photographs were scanned (Deskscan II; Hewlett-Packard, Palo Alto, CA) and analyzed using the NIH Image program (National Institutes of Health at http://rsb.info.nih.gov/nih-image/).

Histopathology

At each time point, five to seven infected or three to five control mice were killed for pathology studies. After BAL, all five lobes were removed in toto without perfusion of the capillary bed. The lungs were placed in 10% buffered formalin (Fisher) and stored at 4°C. The entire left and right upper lobes were dehydrated, paraffin-embedded, and sectioned in 5-micron increments starting at the pleural surface. Four adjacent sections were stained with standard hematoxylin and eosin and examined. Representative sections were also stained for AFB using the standard Ziehl-Neelson technique.

Statistical Analysis

Results were analyzed using the Wilcoxon rank sum test to compare M. bovis-BCG-infected and mock-infected mice.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Growth and Clearance of M. bovis-BCG in BAL, Lung Parenchyma, and Mediastinal Lymph Node

To establish the pattern of M. bovis-BCG infection in the lung, four cohorts of C57Bl/6 mice were infected intratracheally with 0.5 to 1.0 × 10⁵ CFUs of M. bovis-BCG. At weekly intervals, mice were killed and CFUs within the distal airways (BAL), lung parenchyma (lung), mediastinal lymph nodes, and spleen enumerated. Growth and clearance of M. bovis-BCG in BAL, lung, lymph nodes, and spleen from four experiments are shown in Figure 1.

View larger version (25K): [in this window] [in a new window]   Figure 1.   Growth and clearance of M. bovis-BCG in the lung (closed squares), BAL (closed circles), mediastinal lymph node (open triangles), and spleen (open circles). Four cohorts of C57Bl/6 mice (10 wk of age) were inoculated intratracheally with 0.5 to 1.0 × 105 CFUs of M. bovis-BCG. At designated times, individual mice were exsanguinated and lungs lavaged with 1.2 ml of cold PBS. BAL cells were pelleted and resuspended in PBS. Single-cell suspensions of lung, lymph node, and spleen were prepared as described in MATERIALS AND METHODS. Samples were serially diluted and plated onto 7H10 agar, and mycobacterial colonies were counted after 2 wk. Whole organ CFUs for each mouse were calculated and data expressed as mean CFUs ± standard error of the mean (SEM) (per organ/mouse) for six to 12 mice per time point.

Within bronchoalveolar spaces M. bovis-BCG was always recovered with BAL cells and not BAL supernatant (data not shown). Greater than 90% of M. bovis-BCG was cleared from the bronchoalveolar space during the first 3 wk. A slower elimination phase cleared M. bovis- BCG completely from the distal airspaces by 12 wk.

In lung parenchyma, maximal growth of M. bovis-BCG occurred after 3 wk (P < 0.005 compared with Day 1) consistent with the BCG^s phenotype (9, 13, 15). After 4 wk, lung CFUs progressively decreased (P < 0.05 Day 28 versus Days 35 to 98), resulting in a 3-log reduction in CFUs by Week 8. A bacterial load of 100 to 500 CFUs/whole lung persisted for at least 14 wk (P > 0.05 comparing Week 8 with Weeks 9 to 14) with bacilli still detectable at 8 mo (102.0 ± 93.0 CFUs/lung; n = 5).

Dissemination of M. bovis-BCG to mediastinal lymph nodes and spleen was detected within the first week, peaked by Week 3, and decreased to a static level approximately 8 wk after infection. During late infection (> 8 wk), lung and lymph node had similar bacterial burdens (102 ± 93 versus 197 ± 78; n = 5) which were not statistically different when compared at Weeks 8, 10, and 12 (P > 0.05). Slightly more M. bovis-BCG CFUs persisted in spleen (3,780 ± 1,435; n = 5; P < 0.05 compared with lung or lymph node).

Histopathologic Staging of Granulomatous Inflammation in the Lung

Bacterial growth and clearance correlated with four general histopathologic stages. Lung sections were classified according to predominant histologic findings, although there was some overlap between pathologic stages.

Normal lung parenchyma. Lung parenchyma was defined as all tissues distal to and including small bronchi and their accompanying small arteries (bronchovascular bundles). Larger airways, arteries, and septae with large veins and lymphatics were excluded. Tissues distal to small bronchovascular bundles include respiratory bronchioles, alveolar ducts, terminal alveoli, and their associated vasculature (capillaries, small veins, and venules). Figure 2A shows a representative section illustrating typical lung architecture. It should be noted that bronchial-associated lymphoid tissue was not seen in the pulmonary parenchyma of control mice, as noted in some studies (28).

View larger version (139K): [in this window] [in a new window]   Figure 2.   Pathologic stages of M. bovis-BCG infection in the mouse lung. (A) Normal, mock-infected lung showing a small bronchovascular bundle (lower left to upper right) consisting of a small branching bronchus (arrow), an accompanying small artery (a), and a small amount of accompanying loose fibrous connective tissue. The remainder of the field shows more distal alveoli (Bar = 1,000 microns). (B and C) During early infection (less than 2 wk), subbronchial inflammatory cell aggregates (arrows) are seen, composed of small lymphocytes, lobulated macrophages, and occasional neutrophils. The bar in B represents 500 microns and also applies to D, F, and H, which were similarly magnified (×26). Likewise, the bar in C represents 100 microns and applies to E, G, and I, which were similarly magnified (×103). (D and E) During peak infection (3 to 5 wk), small bronchi (b) and adjacent arteries show an intense circumferential lymphoid infiltrate (D, arrow). The alveolar infiltrate is characterized by macrophages with enlarged nuclei and abundant eosinophilic cytoplasm suggestive of cellular activation; multinucleate giant cells (E, arrowhead) and lymphoblasts show irregular, hyperchromatic nuclei and clear cytoplasm. (F and G) During resolving infection (6 to 8 wk), small bronchi (b) show a patchy infiltrate of small lymphocytes (F). The alveoli are filled by foamy macrophages (F, arrow) that have small, inconspicuous nuclei and abundant foamy cytoplasm consistent with decreased cellular activation (G, arrowhead). (H and I) After 8 wk, quiescent infection is characterized by persistent perivasculitis and focal peribronchitis. Lymphocytic aggregates (H, arrow) are composed of small lymphocytes with a nondescript nuclear chromatin pattern and scant cytoplasm suggestive of decreased cellular activation (I).

Stage I infection (less than 2 wk). Stage I was characterized primarily by inflammatory cells within airways consistent with BAL findings (Figure 3). The inflammatory response in lung parenchyma was limited to rare clusters of lymphocytes, monocytoid macrophages, and occasional neutrophils located immediately beneath the basement membrane of small bronchi in the vicinity of small venules (Figures 2B and 2C). Occasional intraepithelial inflammatory cells were observed (not shown). These findings are suggestive of transmigration across bronchial epithelium and consistent with the severity of the bronchoalveolar infiltrate at this stage. Alveolitis was minimal and frank perivasculitis not observed.

View larger version (28K): [in this window] [in a new window]   Figure 3.   M. bovis-BCG induces neutrophilia and lymphocytosis in the bronchoalveolar space. Mice were inoculated with 0.5 to 1.0 × 105 CFUs of M. bovis-BCG and exsanguinated at weekly intervals. For each mouse, the trachea was cannulated and the bronchoalveolar space lavaged three times with 0.4 ml of PBS. Cells were centrifuged (1,000 × g for 10 min), resuspended, and cytocentrifuged (25,000 cells) onto glass slides. Cytospins were stained with Wright-Giemsa. A total of 200 to 400 cells from each BAL were examined and the percent neutrophils and lymphocytes determined. Absolute neutrophil and lymphocyte counts were calculated by multiplying the percent times total BALF cell yield. Results are shown as mean cell numbers ± SEM for six to 12 mice per time point.

Stage II (Weeks 3 to 5). Pathologic stage II corresponded to the time of peak CFU counts and cytokine levels and showed the most severe parenchymal inflammatory response (Figures 2D and 2E). Peribronchial inflammation was circumferential, composed predominantly of lymphoblasts. Both small and large veins as well as occasional arteries showed marked lymphocytic perivasculitis. A total of 30 to 70% of air spaces showed severe alveolitis with predominance of activated epithelioid macrophages and some multinucleate giant cells (Figure 2E). Lymphoblasts comprised a distinct population of inflammatory cells within alveolar spaces at this stage, which correlated with BAL cell findings.

Stage III (Weeks 6 to 8). Pathologic stage III showed resolving inflammation (Figures 2F and 2G). The overall number of inflammatory cells was reduced dramatically. Frank necrosis and fibrosis were not observed. The peribronchial infiltrate was decreased (patchy) and predominantly consisted of small, inactive-appearing lymphocytes. Marked perivasculitis of both small veins and arteries persisted but fewer lymphoblasts were observed. Alveolitis was decreased in extent, severity, and character with a predominance of inactive-appearing foamy macrophages and only rare epithelioid forms (Figure 2G).

Stage IV (greater than 8 wk). Pathologic stage IV showed quiescent infection with residual inflammatory cells (Figures 2H and 2I). The peribronchial infiltrates of small lymphocytes were further decreased (focal). Chronic perivasculitis was mild but most blood vessels maintained a distinct circumferential collar of small lymphocytes. Alveolitis had largely subsided although occasional clusters of inactive-appearing, foamy macrophages were seen. The most distinctive feature of stage IV inflammation was the presence of lymphoid aggregates scattered throughout the parenchyma (Figure 2I).

Cellular Composition of BAL Fluid

To characterize immune cell populations during different phases of mycobacterial growth and histopathology, cells from distal airways (BAL), lung parenchyma, and regional lymph nodes (mediastinal) were harvested. M. bovis-BCG infection resulted in a 5-fold increase in total BAL cells at 3 to 4 wk (data not shown, P < 0.01 compared with control mice). Both neutrophils and lymphocytes were recruited to the air space during the first 2 to 4 wk (Figure 3), correlating with a 2-log decrease in M. bovis-BCG CFUs (Figure 1). Percentages of neutrophils (29.0 ± 2.96) and lymphocytes (30.9 ± 3.8) between 3 and 4 wk were similar (P  > 0.4 by Student's t test) in infected animals (n = 20 mice). In control BAL fluid (BALF) (n = 43 mice), lymphocytes outnumbered neutrophils 5:1 (9.6 ± 1.0% versus 1.5 ± 0.5%; P < 0.005 by Student's t test). The remainder of BAL cells included pneumocytes, alveolar macrophages, immature monocytoid cells, and rare ciliated epithelial cells. Mycobacteria were observed within neutrophils and alveolar macrophages when CFU counts were highest (data not shown). Neutrophilia and lymphocytosis increased in parallel from Week 1 to Week 4. Neutrophilia resolved by Week 6, whereas BAL lymphocytosis persisted through 10 wk when a 4-log reduction in BAL CFUs had occurred. BAL lymphocytes consisted of both CD4⁺ and CD8⁺ T cells in a 2:1 ratio by flow cytometry (n = 5; data not shown).

Recruitment and Expansion of CD4⁺, CD8⁺, and ⁺ T Cells in Lung, Mediastinal Lymph Node, and Spleen during M. bovis-BCG Infection

Phenotypes of T cells within lung, mediastinal lymph node, and spleen were determined by two-color flow cytometry during M. bovis-BCG infection. Single-cell homogenates were prepared and analyzed for CD3 and CD4, CD8, or TCR. The percentage of each T-cell subset was measured and used to calculate the total T-cell number present in each organ. The results are shown in Figure 4 and represent four cohorts of animals inoculated with M. bovis-BCG. A total of five to 10 mice/time point were evaluable for CD4⁺ and CD8⁺ T cells and three to five mice/time point for T cells.

View larger version (24K): [in this window] [in a new window]   Figure 4.   M. bovis-BCG infection results in early expansion of CD4+, CD8+, and + T cells in the lung and mediastinal lymph node. Mice were inoculated with 0.5 to 1.0 × 105 CFUs of M. bovis-BCG and exsanguinated at designated times. After BAL, lung, lymph node, and spleen cell homogenates were prepared as described in MATERIALS AND METHODS. For individual mice, 5 × 105 cells from each organ were stained for CD3 and either CD4, CD8, or the TCR. Nonspecific staining was determined by staining cells with isotype control antibody. Percent of CD4-, CD8-, or TCR-positive events (minus background staining with isotype control antibody) was measured and total T-cell numbers per lung, lymph node, and spleen calculated ([percent of total FACS events] × [total cell yield]) for each infected mouse. The results represent the mean numbers of CD4+, CD8+, or TCR+ T cells (± SEM) present in the lung (seven to nine mice per time point), lymph node (three to five mice per time point), and spleen (eight to 10 mice per time point) after M. bovis-BCG infection. In the lung (left panels), the black line represents the maximal number of CD4+ or CD8+ (2.5 × 106, top) or (0.60 × 105, bottom) T cells detected in lungs of mock (PBS)-infected mice (three to five mice per time point). Statistically significant differences (compared with mock-infected controls) are designated with an asterisk (*P < 0.05). In lymph nodes (middle panels), the horizontal black line represents the maximal number of CD4+ or CD8+ cells (0.54 × 106, top) or (0.50 × 104, bottom) measured in mock-infected mice (three to five mice per time point). Statistically significant differences were detected for both CD4+ and CD8+ cells (denoted with a single asterisk at each time point). In the spleen (right panels), the black and dashed lines, respectively, represent maximal numbers of CD4+ (12.3 × 106) and CD8+ (10.5 × 106) cells detected in mock-infected mice (four to six mice per time point). Statistical comparisons between infected and control mice were made using the Wilcoxon rank sum test (*P < 0.05).

In lung (Figure 4, left panels), CD4⁺ T cells increased in number 3 to 5 wk after infection compared with mock- infected mice (P < 0.05). Peak CD4⁺ T-cell expansion coincided with pathologic Stage II and the initial decline in M. bovis-BCG CFUs (Figure 1). Although the number of CD4⁺ T cells in lung decreased in parallel with decreasing CFUs, the number of CD4⁺ T cells remained statistically higher than controls until late pathologic Stage III and Stage IV. Recruitment and expansion of T cells (Figure 4, bottom left panel) paralleled changes in CD4⁺ cells although the number of T cells was 20-fold lower than CD4⁺ cells. In contrast to CD4⁺ and T cells, CD8⁺ T-cell number increased slightly (10 to 20%) above controls but this increase was not statistically significant (P > 0.10 compared with controls). CD4⁺, , and CD8⁺ T-cell infiltration of the lung correlated with BAL findings and the severe perivascular and peribronchial lymphocytic inflammation described for Stage II histopathology. CD4⁺ and T-cell numbers decreased during Stage III and returned to baseline (P > 0.05 compared with controls) by Stage IV infection when residual pulmonary inflammation was observed.

Expansion of both CD4⁺ and CD8⁺ T cells was observed in mediastinal lymph node after 3 wk (P < 0.05 compared with Days 7 and 14 and control animals) as shown in Figure 4 (middle panels). In contrast to lung parenchyma, where CD4⁺ and CD8⁺ T cell numbers declined to control levels by Week 10, CD4⁺ and CD8⁺ T-cell numbers in lymph node remained increased up to 12 wk. T cells in lymph node increased slightly and did not persist, in contrast to CD4⁺ and CD8⁺ T cells.

Although dissemination of M. bovis-BCG to spleen occurred during Week 1 and appeared static after 7 to 8 wk (see Figure 1), statistically significant increases in CD4⁺ T cells were not observed until very late (Weeks 10 to 12, P < 0.05 compared with control). Further, no significant increase in splenic CD8⁺ T cells was observed. In contrast to lymph node, splenic T cells were not expanded. Thus, during early and persistent infection, patterns of T-cell subset expansion differed between lung, lymph node, and spleen, with earliest T-cell expansion occurring primarily in lung.

Production of IFN- by Lung, Lymph Node, and Spleen Cells

To measure in vivo IFN- production (i.e., spontaneous IFN-), cell homogenates were prepared from infected tissues and cultured for 72 h ex vivo without additional antigen. Figure 5 demonstrates that peak spontaneous IFN- in lung correlated with peak M. bovis-BCG CFU and T-cell expansion (Figure 4) during histopathologic Stages II and III (Figure 2), spanning a period of 3 to 9 wk. Statistically significant increases in lung cell IFN- levels were detected between Week 1 and Weeks 3, 4 and 7 (P < 0.05) correlating with the lymphocytic infiltration of lung. During early and maximal recruitment and expansion, lung cell homogenates contained 25 to 40% CD3⁺ T cells (data not shown). In addition, lung levels of IFN- in infected mice were significantly greater than in controls at all time points (P < 0.05 compared with a control mean of 114.4 ± 42.3 pg/ml representing 12 PBS-infected mice examined between Weeks 4 and 15). Because IFN- expression was detected before significant T-cell infiltration (Figures 4 and 5) we cannot exclude the possibility that some IFN- is produced by natural killer cells. In addition, both CD4⁺ and CD8⁺ T cells purified from lung parenchyma during early and late-stage infection produced IFN- when restimulated with M. bovis-BCG in vitro (data not shown). Thus, even though expansion of CD4⁺ T cells in the lung was greater than CD8⁺ during early infection, antigen-specific sensitization of both CD4⁺ and CD8⁺ T cells had occurred. Peak IFN- levels in BAL supernatants also were measured after 3 to 4 wk in parallel with spontaneous production by lung cells (data not shown). Persistent spontaneous IFN- production was measured during Stage IV in lung associated with bacterial persistence and despite downregulation of T-cell subset expansion.

View larger version (26K): [in this window] [in a new window]   Figure 5.   IFN- production occurs early in lung and late in lymph node and spleen. Four experimental cohorts of mice were infected and single-cell suspensions of lung, lymph node, and spleen prepared. A total of 2 × 106 cells/ml were cultured in vitro and supernatants harvested after 72 h. IFN- was measured by enzyme-linked immunosorbent assay and the results expressed as means (pg/ml ± SEM) for three to five infected mice at each time point. Mean IFN- levels (± SEM) in mock (PBS)-infected mice were: 114.4 ± 42.3 (lung, n = 12); 212.6 ± 106.7 (spleen, n = 12); and 66.4 ± 43.7 (lymph node, n = 6). Statistical comparisons between infected and control mice were made using the Wilcoxon rank sum test (*P < 0.05).

After M. bovis-BCG infection, IFN- production by lymph node and spleen cell homogenates (which contain 30 to 60% CD3⁺ T lymphocytes [data not shown]) was not significantly elevated compared with control cell cultures (66.4 ± 43.7 pg/ml for lymph node and 212.6 ± 106.7 pg/ml for spleen) during Weeks 1 to 10, despite dissemination to and maximal growth of M. bovis-BCG in lymph node and spleen by Weeks 2 to 4. Significant increases in IFN- in lymph node were observed only after 14 wk correlating with Stage IV infection in lung, persistence of M. bovis- BCG, and ongoing CD4⁺ and CD8⁺ T-cell expansion in lymph node. In contrast, significant increases in IFN- expression were not detected in the spleen even when bacterial burdens were maximal and similar to lung.

Expression of Proinflammatory Cytokine Messenger RNA in Lung Cells after Intratracheal Inoculation of M. bovis-BCG

Temporal expression of other cytokines in addition to IFN- in lung was measured by reverse transcriptase- PCR. By densitometric analysis (Figure 6), the signal ratio of BCG-infected (cytokine/HPRT) divided by control (cytokine/HPRT) showed that IFN- and IL-12 p40 messenger RNA (mRNA) were upregulated during early and late infection. In a representative cohort of mice, peak induction occurred after 3 to 4 wk and approached control levels by 10 wk after inoculation. Constitutive expression of IL-12 p35 mRNA was also noted (data not shown). IL-12 and IFN- mRNA expression correlated with macrophage and lymphocyte infiltration characteristic of histopathologic Stages I to III. During Stage IV histopathology, IFN- and IL-12 p40 mRNA densitometric ratios approached baseline consistent with resolving inflammation. Enhanced expression of IL-4, IL-5 (not shown), IL-10, and TNF- mRNA was not observed in lung cells of infected compared with control mice (Figure 6).

View larger version (20K): [in this window] [in a new window]   Figure 6.   M. bovis-BCG enhances expression of IFN- and IL-12 mRNA in lung. (IL-4, lightly shaded bars; TNF-, darker bars; IFN-, blackened bars; IL-10, widely hatched bars; and IL-12, tightly hatched bars). Mice were infected with M. bovis-BCG and exsanguinated, and lungs were perfused and lavaged at weekly intervals. After collagenase digestion, 2 × 106 lung cells were processed for cytokine mRNA as described in MATERIALS AND METHODS. PCR products were separated in a 1.5% agarose gel and visualized with ethidium bromide staining. Polaroid photographs were taken under identical exposure conditions, scanned, and analyzed using NIH image software. Cytokine/HPRT ratios between infected and control mice were compared. An experiment representative of four is shown.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Acquired immunity is required for control of M. tuberculosis infection of the lung. Studies of human immune responses in primary pulmonary M. tuberculosis infection are hampered by our inability to reliably detect infection and by difficulty in obtaining serial samples from lung. Animal models provide a means to study primary mycobacterial infection in the lung, to determine mechanisms of protection, and to study the unique susceptibility of the lung.

We chose M. bovis-BCG infection of the lung in mice as a model because this model mimics infection in humans where primary M. tuberculosis infection is controlled successfully in most individuals, resulting in persistence of small numbers of organisms. In the mouse, the ability to control initial growth is partially controlled by genetic factors, however bacterial clearance depends upon acquired cellular immunity (9, 15). Both BCG-resistant and -susceptible mice generate protective immunity following low-dose M. bovis-BCG infection. We used a BCG-susceptible strain to maximize the role of acquired immunity during control of BCG infection in the lung and to determine how protective immunity is generated in the lung microenvironment. We used intratracheal M. bovis-BCG infection to analyze T-cell recruitment and cytokine production in three pulmonary compartments (bronchoalveolar spaces, lung parenchyma, and mediastinal lymph nodes) and to compare these immune parameters with changes in mycobacterial growth and histopathology.

These studies revealed that M. bovis-BCG was cleared from bronchoalveolar spaces by 3 mo without evidence of persistence. In contrast, organisms confined to lung parenchyma grew during the first 4 wk and were gradually cleared, with fewer than 1% of the original inoculum persisting for more than 8 mo. Persistence of M. bovis-BCG in the lung has been noted in other studies, however bacterial growth and persistence were not correlated with specific immune function and histopathology (23, 24). Although immunodeficiency results in accelerated M. bovis-BCG growth (4, 16), it is not known whether persistent M. bovis-BCG can be reactivated. Histopathologic changes induced by M. bovis-BCG and host immune responses evolved through four distinct phases. Stage I was notable for peribronchitis with minimal alveolitis. As the bacterial burden increased (Stage II), diffuse alveolitis with aggregates of epithelioid cells and occasional giant cells was observed. During Stage III, inflammation started resolving. Granulomas, defined as aggregates of epithelioid macrophages admixed with lymphocytes and occasional giant cells, were seen during Stages II and III. These changes correlated with a 3-log reduction in bacterial burden. Our Stages I to III correlated with early pathologic stages (Stages I and II) reported by Rhoades and colleagues for murine lungs infected with small numbers of virulent M. tuberculosis (29).

By Stage IV, differences in histopathology between M. bovis-BCG and M. tuberculosis infection became evident. By 60 d, M. bovis-BCG was controlled and lung histopathology had evolved into one of scattered lymphoid aggregates and inactive foamy macrophages without necrosis. Most of the lung was normal, and a small, stable number of persistent bacteria was detected for more than 8 mo. In contrast, the chronic phase of M. tuberculosis infection is characterized by diffuse lung involvement with a large burden of bacteria, ultimately resulting in fatal pneumonia with airway epithelial-cell erosion, macrophage karyorrhexis, diffuse consolidation, and fibrosis after 200 d (29). Thus, histopathologic changes in lung parenchyma in response to M. bovis-BCG correlated directly with changes in patterns of bacterial growth: early growth, control of bacterial replication, followed by clearance and chronic persistence. Early growth has been used to define "BCG- susceptibility phenotype" but the distinction in the lung is unclear (8, 13, 15). Further, clearance of bacteria and quiescent infection occur in both BCG^s and BCG^r mice and is not associated with chronic lung damage, analogous to humans who have successfully controlled primary infection.

Immune cell recruitment to lung differed significantly among bronchoalveolar spaces, parenchyma, and regional lymph nodes. In bronchoalveolar spaces, significant neutrophil recruitment was measured after 2 wk and peaked 1 wk before maximal lymphocyte expansion. Human neutrophils can phagocytose and kill mycobacteria (30), and are found in BAL during active tuberculosis (33, 34). In addition, both human and rodent neutrophils produce cytokines and chemokines capable of recruiting monocytes and T cells (35). Thus, neutrophils may serve as immune effector cells recruiting T cells to bronchoalveolar spaces in response to M. bovis-BCG as suggested by others (36). The 1- to 2-wk delay between inoculation and neutrophil recruitment further suggests that neutrophil migration was dependent on other signals, such as chemokines secreted by M. bovis-BCG infected macrophages or respiratory epithelial cells, and not part of a nonspecific inflammatory response (39). Although neutrophils were detectable within lung parenchyma (Stages I and II), they were not as dominant as in BALF.

T-cell expansion in BAL paralleled that observed in lung parenchyma, with peak responses observed at 4 wk. In bronchoalveolar spaces, however, lymphocyte numbers remained elevated 1 to 2 wk longer than in lung parenchyma, where CD4⁺ and CD8⁺ T-cell numbers returned to baseline by 7 to 8 wk. During primary infection, expansion in lung of all three major T-cell subsets was observed with similar kinetics for CD4⁺, CD8⁺, and T cells. T cell expansion, however, was more pronounced and prolonged in lung parenchyma compared with mediastinal lymph node, consistent with earlier reports (42). There was no evidence that T cells preceded CD4⁺ or CD8⁺ T-cell expansion, nor was there late T-cell expansion as seen in murine influenza infection or as suggested by studies in TCR gene knockout mice in which T cells were found to play a role in late granuloma formation to M. tuberculosis (26, 43, 44). Our results are more consistent with the listeria model in which T cells are part of primary immune responses and may help regulate granulomatous inflammation (45, 46).

In the lung, CD4⁺ T cells were the dominant T-cell subset expanded, supporting their primary role in protective immunity. There was minimal expansion of CD8⁺ T cells in lung compared with bronchoalveolar spaces, where both subsets also were expanded. Further, in the mediastinal lymph node CD4⁺ T cells were not dominant but expanded to the same extent as CD8⁺ T cells and remained expanded for 12 wk. In lung parenchyma, numbers of T-cell subsets returned to baseline by 7 to 8 wk. Both CD4⁺ and CD8⁺ T cells in lung and lymph node produced IFN- in response to M. bovis-BCG bacilli, indicating that both T-cell subsets were sensitized to mycobacterial antigens during primary pulmonary BCG infection.

Ongoing spontaneous IFN- production from lung cells was detected for at least 14 wk, despite the return of T-cell numbers to control levels by Week 8. Although maximal bacterial burdens were observed after 4 wk in both lymph node and spleen, minimal spontaneous IFN- production was measured only during the quiescent phase (Week 10, Stage IV). Localized production of IFN- correlated with control of BCG growth in the lung. In a contrasting model, control of intratracheal Cryptococcus neoformans infection correlated better with higher IFN- expression within regional lymph nodes compared with lung (47). Thus, the dominant protective immune response against different intracellular, pulmonary pathogens develops in distinct compartments within the lung. Additional studies will be needed to determine how recirculation of lymphocytes between lung and lymph node affects the development of a compartmentalized protective immune response.

IFN- production by lung cells during quiescent infection suggests that small numbers of persistent bacteria or bacterial antigen continued to stimulate immune responses and that continued immune surveillance was required to control these residual bacilli. Whether all IFN- was produced by T cells only or whether natural killer cells contributed in early or late stages of infection was not determined in our studies. However, peak IFN- production at Weeks 3 and 4 correlated with peak lung T-cell expansion. Similarly, reduced IFN- expression correlated with reduced T-cell recruitment. Further evidence for ongoing immune activation was provided by cytokine mRNA measurements which demonstrated increased IL-12 and IFN- mRNA expression for at least 10 wk. These results suggest that persistent mycobacteria in this model are not necessarily dormant or immunologically silent. In fact, continuous immune surveillance, while clinically silent, may be required to control persistent foci of mycobacterial infection. Thus mycobacterial "dormancy" may not be sequestration within an immunologically silent focus.

During primary mycobacterial infection of lung, alveolar macrophages may downmodulate T-cell responses resulting in persistent infection (6). In addition, respiratory epithelial cells may be susceptible to mycobacterial infection and may modulate host immune responses preventing mycobacterial eradication (39, 50). We did not find evidence for increased IL-10 or a predominance of IL-4 or IL-5 expression in the lung to explain the permissiveness and decreased resistance of lung to M. bovis-BCG. Consistent with this finding is the lack of increased susceptibility to M. bovis-BCG in IL-4, IL-5, and IL-10 knockout mice (51). Overall, our results indicate that after M. bovis- BCG infection, bacterial clearance, T-cell expansion, and IFN- production differed between bronchoalveolar spaces, lung parenchyma, and lymph node. Further studies of compartmentalized effector T-cell function in this murine model of protective immunity to M. bovis-BCG should provide insight into the lung's unique susceptibility to M. tuberculosis infection and the failure of acquired cellular immunity to completely eradicate mycobacteria from the lung.