Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Tuberculosis is characterized histologically by granulomas at the site of disease activity. Other granulomatous diseases include those caused by diverse mycobacteria, fungi, and inert particles such as beryllium and silica. Granulomas consist of aggregations predominantly of T lymphocytes and macrophages which act to wall off and to destroy the offending agent (reviewed by Kunkel and associates [1]). Macrophages further develop into epithelioid cells and multinucleated giant cells characteristic of most granulomas. Although local proliferation of the cellular constituents of granulomas could occur, it is likely that macrophages and lymphocytes also reach the tissue from the blood. The mechanisms of recruitment of these cells into granulomas, however, are not well understood.

Chemotaxis is the directed migration of cells according to concentration gradients of substances in their environment. These substances may be one of many known chemotactic factors, including leukotriene B₄, complement component 5a, platelet activating factor, and N-formyl-peptides. Another more recently discovered group of chemotactic substances is cytokines known as chemokines (reviewed by Oppenheim and coworkers [2] and Miller and Krangel [3]). Chemokines are members of a large superfamily of low molecular-weight proteins that are structurally and functionally related. As a group, these proteins function in both the recruitment and activation of leukocytes and other cells at sites of inflammation. Each member of this superfamily displays four conserved cysteine residues in either of two patterns. In the C-X-C family of chemokines, whose genes reside on chromosome 4, the first two conserved cysteines are separated by one amino acid. In the C-C family, whose genes are on chromosome 17, the first two conserved cysteines are adjacent. Interleukin (IL)-8 is a well-described member of the C-X-C family and is chemotactic for neutrophils and lymphocytes (4- 7). Zhang and colleagues demonstrated that IL-8 is induced by infection of alveolar macrophages (AM) with Mycobacterium tuberculosis (MTB) and in lavage fluid of patients with tuberculosis (8). Regulated upon activation, normal T cell expressed and secreted (RANTES), macrophage chemotactic protein-1 (MCP-1), and macrophage inflammatory protein-1 (MIP-1) are members of the C-C family and are generally chemotactic for monocytes (MN) and lymphocytes (7, 9).

MTB and its purified protein derivative (PPD) induce the synthesis of several inflammatory cytokines by MN and AM including IL-1, tumor necrosis factor (TNF), and tumor growth factor  (14). Moreover, during tuberculosis these cytokines are elevated in in vitro cultures of MN (17, 18). In the current study, we examined the production of the C-C chemokines RANTES, MCP-1, and MIP-1 by MN and AM infected in vitro with MTB or stimulated with PPD. IL-8 was studied concurrently (as a C-X-C chemokine). Each of the chemokines studied was increased by MTB and PPD, and AM produced higher levels than did MN. Supernatants of infected cells were chemotactic for both MN and lymphocytes, and this activity was partially inhibitable with neutralizing antibodies against the measured chemokines. The bronchoalveolar lavage (BAL) fluid of patients with active pulmonary tuberculosis, furthermore, showed increased levels of RANTES, MCP-1 and IL-8, but not MIP-1.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Human Subjects

Subjects were recruited from the greater Cleveland, OH, area to undergo BAL and venipuncture. None of the subjects had had an upper respiratory tract infection or had received medication including nonsteroidal inflammatory agents for 4 wk before the study. All subjects were nonsmokers, 29 ± 9 yr of age. Four patients from Mexico City with active, sputum smear, and culture-positive pulmonary tuberculosis were also recruited for studies of chemokine levels in the BAL fluid during active disease. None of these patients had been treated for tuberculosis before the lavage and all had moderate to advanced radiographic abnormalities on chest X-ray. Patients from Mexico City were nonsmokers, 30 ± 10 yr of age and were HIV-1 seronegative. Healthy subjects (n = 6) from Mexico City were recruited to serve as healthy control subjects for the patients with tuberculosis. Healthy Mexican subjects were 31 ± 9 yr of age and were nonsmokers.

BAL

BAL was performed as described previously (19). Briefly, the naso-oropharynx was anesthetized with 2% lidocaine. An Olympus flexible fiberoptic bronchoscope type BF P30 (Olympus Corp. of America, New Hyde Park, NY) was introduced through the upper airways, using 1% lidocaine to further anesthetize the subglottic area. The bronchoscope was wedged into two segments of the right middle lobe in healthy control subjects or into a segment of the radiographically affected and a segment of the unaffected lung of patients with pulmonary tuberculosis. To obtain bronchoalveolar cells (BAC), sterile 0.9% NaCl was instilled in 30-cc aliquots and aspirated, using a total of 180 ml in each of the two segments. The average of instilled saline that was retrieved was 85% both from patients with tuberculosis and from healthy control subjects.

Preparation of Cells

BAL fluid was centrifuged at 350 × g for 15 min at 4°C to obtain BAC. BAC were adjusted to 10⁶ cells/ml in RPMI 1640 culture medium (Bio-Whittaker, Walkersville, MD) supplemented with 2 mM L-glutamine (Gibco BRL, Gaithersburg, MD), 50 U/ml penicillin (Squibb-Marsam, Cherry Hill, NJ), 5 µg/ml gentamicin (Bio-Whittaker) (complete RPMI), and 10% heat-inactivated pooled human serum (PHS). BAC were 90 to 95% nonspecific esterase positive, < 1% peroxidase positive, and < 1% granulocytes as assessed by Wright's stain, and therefore will be referred to as AM.

To obtain MN, peripheral blood mononuclear cells (PBMC) were prepared by centrifugation of whole heparinized blood over Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden) using previously described standard techniques (20). PBMC were adjusted to 10⁷ cells/ml in complete RPMI containing 10% PHS and incubated for 1 h at 37°C on 100-mm plastic Petri dishes (Falcon Plastics, Oxnard, CA) which had been precoated with approximately 1 ml of PHS. Nonadherent cells were removed by washing the plates with warm complete RPMI (37°C) containing 10% fetal calf serum (FCS), sedimenting them then resuspending in complete RPMI. Morphologically, these nonadherent cells were predominantly lymphocytes, as determined by Wright's stain, and were < 5% nonspecific esterase or peroxidase positive. The nonadherent cells were used as the lymphocyte responder population in migration assays. The remaining monolayer of adherent cells was covered with Hanks' balanced salt solution without magnesium or calcium (Gibco BRL, Laboratories, Grand Island, NY) and incubated at 4°C for 30 min. Adherent cells were removed by scraping the plates with a plastic scraper. Recovered adherent cells were resuspended in complete RPMI at a concentration of 10⁶ cells/ ml. The adherent cells were 90 to 94% peroxidase positive, 70 to 80% nonspecific esterase positive, and < 1% granulocytes by Wright's stain. The peroxidase-positive cells showed the morphology of MN and adherent cells therefore will be referred as to MN. Viability of AM, MN, and lymphocytes was > 95% as determined by exclusion of 2% trypan blue.

Infection of Cells with MTB

The avirulent strain of MTB H37Ra was grown in Middlebrook 7H9 broth (Difco Laboratories, Detroit, MI) at 37°C in 5% CO₂ in air until bacterial clumps were clearly discernible on visual assessment. MTB H37Ra was aliquotted and stored at 70°C. The concentration of MTB in the stock preparations was determined by culturing several dilutions of the stock on 7H10 agar for 3 wk at 37°C. Before use, the stock was sonicated for 10 s in an ultrasonic disrupter (Heatsystems Ultrasonics, Inc., Farmingdale, NY) and diluted to 10⁷ colony-forming units/ml in RPMI containing L-glutamine without antibiotics and 15% PHS that had not been heat-inactivated. MN and AM were plated onto 24-well plates at 10⁶ cells/ml in 0.5 ml for 1 h. Cells then were infected for 1 h with MTB using ratios of 0.5:1 and 5:1 (MTB:cells) and washed. Medium containing 2% PHS was added for most cultures. Replicate wells to which medium containing 2% FCS was added, however, were established for measurement of RANTES because PHS contains RANTES. MN and AM also were stimulated with PPD (100 µg/ml) as a soluble mycobacterial preparation (Lederle, Cambridge, MA). Lipopolysaccharide (LPS) from Escherichia coli serotype 0127:BB (Sigma, St. Louis, MO) was used as a positive control stimulus. Preliminary experiments showed that 10 µg/ml LPS induced peak levels of chemokines and this concentration was used in subsequent experiments. Supernatants of MTB-infected, PPD- or LPS-stimulated, or medium controls (with and without cells) were collected at 24, 48, and 72 h after stimulation and stored at 70°C until assay. In selected experiments, the virulent strain of MTB H37Rv was used which was grown in 7H9 broth and prepared similarly to H37Ra stocks, except that H37Rv was treated with glass beads to decrease clumping, as described by Schlesinger (21).

Chemokine Immunoassay

The concentrations of the chemokines IL-8, RANTES, MCP-1, and MIP-1 were assayed by enzyme-linked immunosorbent assay (ELISA) in supernatants using commercially available kits (R&D Systems, Minneapolis, MN). The sensitivity of the IL-8 ELISA was 3.0 pg/ml; RANTES, 2.5, pg/ml; MCP-1, 5.0 pg/ml; and MIP-1, 2.0 pg/ml.

Migration Assay

To measure migratory activity, a chamber (MBA 96 4 042) manufactured by Neuroprobe Inc. (Bethesda, MD) was used. The Neuroprobe chamber consisted of 96-well upper and lower chambers and a 5-µm polycarbonate membrane filter between the two chambers. Cells (MN or lymphocytes) were placed in the upper chamber using 0.75 × 10⁶ cells/ml in 395 µl of RPMI. In preliminary experiments, this number of cells was found to be optimum for migration. Supernatant samples to be tested were placed in the lower chamber using 390 µl/well. To establish a linear curve of cell number as a standard, known numbers of blood MN or lymphocytes between 2 × 10⁶ and 10³ also were placed in the lower chamber in each experiment. The chamber was incubated at 37°C for 90 min. Cells from the upper chamber were then aspirated and 20 µl of 0.5 M ethylenediamine tetraacetic acid in 10% phosphate-buffered saline was added to the upper chamber for 20 to 30 min at 4°C to dislodge cells from the filter. A total of 50 µg of 3-(4,5 dimeth-ylthiazol-2-yl)-2,5-diphenol tetrazolium bromide (MTT; Sigma) was added to each well of the lower plate and the plate was incubated for 4 h at 37°C. Dehydrogenase in the mitochondria of viable cells reduces the water-soluble yellow MTT dye to insoluble dark purple crystals. The MTT reduction is directly proportional to the number of metabolically active cells that have migrated through the polycarbonate membrane (22). After incubation, the plate was centrifuged at 350 × g for 10 min to allow insoluble crystals to settle. The crystals were carefully separated by aspiration of the lower-chamber fluid. Crystals were then dissolved in 100 µl of acid isopropanol (2 mM HCl) and read colorimetrically at 490 nm. The migration index was calculated as the number of cells migrating to sample supernatants divided by the number of cells migrating to cell-free medium alone.

Neutralization of Chemotactic Activity

To establish the concentration of neutralizing antibodies necessary to block the migratory activity of sample supernatants, first the migratory activity of various concentrations of recombinant human proteins IL-8, RANTES, MCP-1, and MIP-1 (R&D) was determined using the Neuroprobe chamber. Next the concentration of human neutralizing antibodies (R&D) that could neutralize the predetermined migratory activity of each of these chemokines was established in preliminary dose-response experiments by preincubating the appropriate concentration of recombinant proteins with various concentrations of neutralizing antibody at 37°C for 1 h before the migration assay. The concentration of neutralizing antibody with maximum inhibition of migration of recombinant proteins was used to neutralize the migratory activity in sample supernatants.

Northern Blot Analysis

Total RNA was extracted from 5 × 10⁶ AM by the guanidinium cesium chloride method. Northern blot analysis was performed as previously described (23) using DNA probes (R&D) that were ³²P-labeled (24). The DNA probes were single-stranded, antisense oligonucleotides that consisted of an equimolar mixture of several region-specific probes (RANTES, catalog no. BPR246; MCP-1, catalog no. BPR250; the MIP-1 probe was a gift from Dr. J. Wallace, Imperial College, London, UK), and were used as directed by the manufacturer.

Statistical Analysis

Data were analyzed using the paired t test. Values were considered significant at P < 0.05. Data shown represent the means ± SE of separate experiments.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Induction of Chemokines by MTB

The capacity of MTB infection of macrophages to induce the production of the chemokines RANTES, MCP-1, and MIP-1 was studied. These chemokines were chosen because they have chemotactic activity for lymphocytes (RANTES) and MN (MCP-1, MIP-1) in in vitro systems. IL-8, known to be produced by MTB-infected macrophages (8), was measured concurrently to place production of C-C chemokines (RANTES, MCP-1, and MIP-1) into perspective with a representative C-X-C chemokine (IL-8) with respect to MTB infection. MN or AM from healthy subjects were infected with or without MTB H37Ra at ratios of 0.5:1 and 5:1 MTB:cells. Cells also were stimulated with LPS as a positive control. Preliminary kinetic experiments were performed in which supernatants from infected cells were collected at 24, 48, and 72 h after infection. Chemokine levels were elevated in MTB-infected cell supernatants at 24 h but peaked at 72 h for each of the chemokines studied (data not shown).

Figure 1 shows the concentration of chemokines in infected cell supernatants at 72 h after infection. Supernatants of MTB-infected MN and AM at ratios of 0.5:1 or 5:1 MTB:cells contained significantly higher levels of RANTES (Figure 1A), MCP-1 (Figure 1B), and MIP-1 (Figure 1C), than found in unstimulated cells. Furthermore, at the higher inoculum (5:1), MTB-infected AM produced significantly higher levels than MN of RANTES (by 2.1-fold), MCP-1 (by 6.9-fold), and MIP-1 (by 5.5-fold) (P < 0.05 for each). MTB-infected AM also produced IL-8 (150 ± 55 ng/ml at an inoculum of 5:1, not shown in figure), and this level was higher than RANTES and MCP-1 (P < 0.01) and comparable to MIP-1.

View larger version (13K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   Figure 1.   Induction of chemokines by infection of MN and AM with MTB. MN and AM were unstimulated (0), infected with MTB at a ratio of 0.5:1 or 5:1 MTB:cells, or stimulated with 10 µg/ ml LPS. RANTES (A), MCP-1 (B), and MIP-1 (C) were measured by ELISA in supernatants collected at 72 h after infection, which was the peak of production of these chemokines determined in preliminary kinetic experiments. Data shown are the means ± SE of the results obtained from 10 subjects. Asterisk indicates P < 0.05 for MTB-infected or LPS-stimulated cells versus unstimulated cells.

To determine whether the increased levels of chemokines in supernatants of MTB-infected cells could be attributed to release of preformed proteins, cell lysates were assayed for chemokines. Lysates of freshly isolated MN and AM from healthy subjects were obtained by freeze-thawing. Lysates contained no detectable levels of IL-8, MCP-1, or MIP-1 (< 32 pg/ml, n = 3). Lysates of infected MN had 411 pg/ml RANTES; AM, 636 pg/ml. Therefore, with the exception of RANTES, the chemokines measured in the supernatants of infected cells likely reflected release of newly produced proteins. RANTES, however is preformed in MN and AM, and infection with MTB induces the release of this protein.

LPS contamination in the MTB preparations was less than 1 ng/ml, as determined by the limulus amoebocyte lysate assay (Bio-Whittaker), and LPS (E. coli) at  1 ng/ml did not stimulate production of chemokines by MN and AM (data not shown). Furthermore, preincubation of MTB with polymyxin B (12.5 µg/ml) did not decrease the concentration of chemokines measured in supernatants of MTB-infected cells (n = 5, data not shown). Therefore, it is unlikely that the chemokines measured in supernatants of MTB-infected cells were induced by LPS contamination of the MTB stock.

Next, induction of chemokines by other stimuli was investigated. MN or AM were infected with H37Rv (virulent strain of MTB), or were exposed to 2 µm latex beads (which is the approximate length of the bacilli of MTB). Alternatively, cells were stimulated with soluble PPD of MTB H37Rv. As shown in Table 1, H37Rv and its PPD stimulated one or more chemokines by MN and AM to levels comparable to that induced by H37Ra. Latex beads, however, did not stimulate chemokines. Therefore, chemokine induction is not specific to MTB H37Ra but does not extend to a phagocytic inert stimulus such as latex beads.

To determine whether infection of mononuclear phagocytes with MTB induces expression of messenger RNA (mRNA) for chemokines, AM were infected with H37Ra (5:1). LPS was used as a positive control. Infected and uninfected cells were lysed at 24 h. RNA was extracted and Northern blot analysis performed. As shown in Figure 2, although constitutive expression of mRNA was noted particularly for MIP-1, MTB increased steady-state expression of MIP-1, RANTES, and MCP-1 mRNA by AM. LPS stimulated low levels of expression of mRNA for these chemokines by AM at 24 h but it is possible that the peak of expression occurred much earlier.

View larger version (52K): [in this window] [in a new window]   Figure 2.   Chemokine mRNA expression in mononuclear phagocytes infected with MTB. AM were infected with H37Ra (5:1) and cells lysed after 24 h. Northern blot analysis was performed.

Migratory Activity in Supernatants of MTB-Infected Cells

IL-8, RANTES, MCP-1, and MIP-1 are known chemotactic factors for MN and/or lymphocytes. The migratory activity for MN and lymphocytes in supernatants of MTB-infected MN and AM containing demonstrable chemokines therefore was examined. Figure 3 shows that migratory activity was higher in supernatants of LPS-stimulated as compared with unstimulated cells, as predicted. Supernatants of unstimulated MN and AM induced the migration of both lymphocytes and MN (compared with medium controls). There was, however, increased migratory activity in supernatants of MTB-infected MN for both lymphocytes (by 1.6 ± 0.2-fold) and MN (3.3 ± 1.9-fold) (n = 3, P < 0.05). Likewise, supernatants of MTB-infected AM had increased migratory activity for lymphocytes (by 4 ± 1.4-fold) and MN (by 1.4 ± 0.9-fold) (n = 3, P < 0.05). Although each of the starting populations (MN, lymphocytes, and AM) contained approximately 5 to 10% other cells, it is unlikely that the only cells migrating through would be these few contaminating cells because that few cells would have been difficult to detect in the MTT assay.

View larger version (16K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   Figure 3.   Migratory activity in supernatants of MN and AM for lymphocytes. MN and AM were either unstimulated (0), infected with MTB at a ratio of 5:1 MTB:cells, or stimulated with 10 µg/ml LPS. Supernatants of cells and medium controls were collected at 72 h and assayed for migratory activity for lymphocytes (A) and MN (B). Data shown are representative of three experiments.

These data demonstrate that MTB infection of mononuclear cells can induce the migration of lymphocytes and MN, but do not discriminate between a chemotactic and a chemokinetic response. In chemotactic reactions, the direction of cells is determined by substances in the environment, whereas in chemokinetic responses the rate of movement of cells is determined by environmental substances. Checkerboard analyses were performed to distingush whether the migratory response was chemotactic or chemokinetic. Table 2 shows that migration of lymphocytes and MN was found only when a concentration gradient of supernatants of MTB-infected cells was established across the membrane between upper and lower chambers. Little migration was observed when MTB-stimulated samples were present only in the upper chamber or when MTB-stimulated samples were present in both the upper and lower chambers. Thus, the migration of lymphocytes and MN to supernatants of MTB-infected cells is a chemotactic response. Interestingly, chemotactic activity in supernatants of uninfected cells also required a concentration gradient.

Next, the effect of neutralizing antibodies against chemokines on the migratory activity in supernatants of MTB-stimulated cells was examined (Table 3). The control antibody IgG had no effect on migratory activity. Antibodies against IL-8, RANTES, MCP-1, and MIP-1 partially decreased migratory activity for lymphocytes and MN. The level of inhibition by these antibodies was variable among three experiments but ranged from 40 to 85% for supernatants of MTB-infected AM and 25 to 88% in supernatants of infected MN (P < 0.05 for each chemokine). Thus, these chemokines released by MTB-infected cells have demonstrable migratory activity for lymphocytes and MN.

Chemokines in BAL Fluid during Tuberculosis

To assess whether the chemokines induced in vitro by MTB-infected macrophages might also be elevated during tuberculosis, the concentration of RANTES, MIP-1, and MCP-1 was determined in the cell-free BAL fluid of patients with active pulmonary tuberculosis versus healthy subjects (Figure 4). (The phenotype of the BAC from the affected lungs of these patients was approximately 70% AM, 20% immature macrophages, and 25% alveolar lymphocytes, as determined by morphology and cytochemistry and as described previously [25]. Neutrophils also were elevated to 11% in one of the patients.) Total protein levels were higher in the BAL fluid of tuberculous patients (411.7 ± 7.2 µg/ml) as compared with healthy individuals (224.3 ± 14.6 µg/ml). Therefore the chemokine data were normalized to total protein levels in the BAL fluid of each individual. All of the chemokines assayed were detected in the BAL fluid from healthy control subjects, but levels were low. Among patients with tuberculosis, MIP-1 levels were not higher than that in healthy subjects. RANTES, however, was increased by 8-fold and MCP-1 by 2.7-fold (P < 0.05). Not shown in the figure, IL-8 was also increased in the BAL fluid of tuberculosis patients by 8.9-fold, confirming the results of Zhang and colleagues (8).

View larger version (12K): [in this window] [in a new window]   Figure 4.   Chemokines in the BAL fluid during tuberculosis. Cell-free BAL fluid from patients with active tuberculosis (n = 6) and healthy control subjects (n = 6) was concentrated 10-fold by centrifugation over a 3 kD (Concentrator Model Centriprep-3; Amicon, Inc., Beverly, MA). The concentrated fluid was assayed for RANTES, MCP-1, and MIP-1 by ELISA and expressed as picograms per microgram of total protein (Bradford assay).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This study demonstrates that infection of mononuclear phagocytes with MTB stimulates production of MCP-1 and MIP-1 and release of RANTES. Steady-state levels of mRNA for MCP-1, MIP-1, and, to a lesser extent, RANTES were also increased in MTB-infected AM. AM, however, produced increased amounts of these chemokines as compared with MN. This study also confirmed that MTB induces IL-8 production by AM. Supernatants of MTB-infected cells demonstrated chemotactic activity for both monocytes and lymphocytes partially inhibitable by neutralizing antibodies against the chemokines studied (RANTES, MCP-1, MIP-1, and IL-8). Since the four anti-chemokine antibodies individually inhibited migratory activity by at least 50%, we speculate that several chemokines must be concurrently released for maximal migration to occur in this system. It was beyond the scope of this study, however, to determine the relative contribution of these chemokines to chemotactic activity in culture supernatants of MTB-infected cells. BAL fluid from patients with active pulmonary tuberculosis contained increased levels of RANTES, MCP-1, and IL-8, but not MIP-1, as compared with healthy control subjects, suggesting a role for these chemokines in host defense during active disease.

Supernatants of uninfected cells showed low and variable levels of chemokines, indicating constitutive production or in vitro activation, perhaps by adherence to plastic. Nevertheless, culture supernatants of MTB-infected cells always had significantly higher levels of chemokines than did those of uninfected cells. Serum (2%) was used routinely in cultures of MTB-infected cells because the viability of MN/AM decreases substantially in the absence of serum and because the presence of serum is more physiologic than its absence. Since serum was present in cultures, we cannot exclude the possibility that serum factors such as LPS-binding protein or other factors might contribute to the induction of chemokines in MTB-infected cells.

Both an avirulent (H37Ra) and a virulent (H37Rv) strain stimulated induction of chemokines, as did the PPD of H37Rv. Whether other types of mycobacteria or other microorganisms could induce these chemokines by mononuclear phagocytes was not studied, but it is not unlikely that other intracellular organisms also can stimulate chemokine expression in infected cells. Latex beads, however, did not induce the chemokines measured (IL-8 and RANTES), indicating that phagocytosis per se is insufficient for their induction.

Inflammatory cytokines such as IL-1 and TNF directly stimulate expression by MN and AM of certain chemokines including IL-8 and MCP-1 (6, 26). Thus it is possible that the known induction of inflammatory cytokines by MTB in mononucelar phagocytes (14, 15) stimulates chemokine expression through an autocrine fashion. Zhang and coworkers indeed demonstrated that neutralizing antibodies against IL-1 and TNF- significantly inhibits IL-8 production by MTB-infected AM (8). Brieland and colleagues found that IL-1 and TNF- induce MCP-1 production by rat AM (27), although Streiter and associates did not observe this effect in human AM, perhaps due to differences in culture conditions (28). Human AM produce significantly higher levels of TNF- than IL-1 when stimulated with LPS, in contrast to LPS-stimulated MN which produce more IL-1 (29). It is possible that differences in pattern of expression of inflammatory cytokines by MTB-infected AM and MN contribute to the observed increased levels of chemokines induced in AM. Several components of MTB and PPD elicit production of IL-1 and TNF-, such as the 30-kD secreted antigen believed to be protective (30), a 58-kD protein (31), and lipoarabinomannan (32). Current studies are directed at which constituents of MTB and PPD induce chemokines by mononuclear phagocytes and at the role of concurrent stimulation of inflammatory cytokines in activation of chemokine expression in these cells.

Both IL-8 and MCP-1 have been examined for their roles in pathogensis of tuberculosis. Wilkinson and Newman found that PPD stimulates PBMC to release IL-8, and that IL-8 is chemoattractant only for activated T lymphocytes of the CD45RO memory-cell phenotype (33). Our study, however, shows that resting blood lymphocytes migrate toward supernatants of MTB-infected mononuclear phagocytes and that anti-IL-8 partially neutralizes that movement. Whether activated and/or memory T cells would be selectively recruited by MTB-infected cells was not investigated. Monocyte migration also was inhibitable by anti-IL-8 but to a lesser extent than was migration of lymphocytes.

IL-8 is not elevated in the pleural fluid of patients with tuberculosis (6). Our finding of increased IL-8 in lavage fluid of patients with active tuberculosis, however, confirms the findings of Zhang and colleagues, who found that MTB and several of its components stimulate IL-8 production and IL-8 mRNA expression by AM (8). Our study extends their findings with respect to IL-8 by showing that MTB induces chemotactic activity for lymphocytes, in part, attributable to IL-8. To varying extents, both lymphocytes and neutrophils are elevated in the BAL fluid of patients with active tuberculosis (8, 25). It is possible, therefore, that IL-8 induced in response to MTB and/or its products may be involved in recruitment of these cells to the lung during tuberculosis. Other chronic lung diseases are also associated with increased levels of IL-8 in the lavage fluid and/or lung tissue, including sarcoidosis and hypersensitivity pneumonitis (which are granulomatous lung diseases) and idiopathic pulmonary fibrosis and the adult respiratory distress syndrome (which are not granulomatous) (6, 34, 35). Thus, IL-8 may be involved in the pathogenesis of a variety of lung diseases.

MCP-1 is primarily chemotactic for MN and produced by the same cells. Snyderman and coworkers first described an MN chemotactic factor in PPD-stimulated PBMC now recognized as MCP-1 (36). Pleural effusions from patients with tuberculosis but not malignancy contain high levels of MCP-1 (6). Our study further demonstrates increased levels of MCP-1 in BAL fluid of tuberculous patients as compared with healthy subjects. We recently found an increase in the number of immature macrophages that were cytochemically like blood MN (peroxidase-positive) among BAC from patients with active pulmonary tuberculosis (25). Whether the observed increase in MCP-1 and other chemokines contributes to recruitment of MN to the lungs needs to be investigated.

The role of MIP-1 in the pathogenesis of tuberculosis has not been studied. MIP-1, however, is elevated in the BAL and/or lung tissue of several lung diseases, including hypersensitivity pneumonitis (34), sarcoidosis, and idiopathic pulmonary fibrosis (37). MIP-1 is also present in AM, interstitial macrophages, and fibroblasts in bleomycin models of lung injury in rats (38). In experimental models of schistosomiasis, MIP-1 is expressed in both early and later stages of granuloma formation, but levels decrease significantly in the later stages (39). Our failure to find elevation in MIP-1 levels in the BAL fluid of patients with tuberculosis could therefore relate to decreased production of this chemokine after granulomas are well formed. MIP-1 is chemotactic for MN and for lymphocytes with a predilection for CD8 T cells (40) and MN are the major known source of MIP-1 (2). VanOtteren and colleagues reported that murine AM and peritoneal macrophages produce MIP-1 in response to LPS (41). The current study further shows that human AM produce MIP-1 in response to LPS, PPD, and MTB infection.

RANTES is preferentially chemotactic for CD4 T cells and for memory (CD45RO) T cells as well as MN (9, 10). T cells are the principle source of RANTES, and its gene was originally characterized by subtractive hybridization techniques that selected for genes expressed by T-cell lines but not by B-cell lines (9). Phytohemagglutinin-stimulated PBMC also produce and express RANTES, although the relative contribution by MN and lymphocytes within PBMC for production of RANTES is not clear (4).

RANTES was primarily released, and not produced, in response to MTB infection of mononuclear phagocytes, and mRNA was only weakly upregulated by MTB infection. Further, levels of RANTES in MTB-infected MN and AM were at least a log lower (picogram quantities) than the other chemokines induced (nanogram quantities). Although adherent cells contained few lymphocytes (< 5%), we cannot exclude that contaminating T cells within the macrophage populations produced RANTES in response to MTB antigens. Alternatively, macrophage cytokines such as IL-1 induced by MTB infection, in turn, could stimulate lymphocytes to produce RANTES. RANTES was elevated in the BAL fluid of patients with tuberculosis in picogram quantities similar to IL-8 and MCP-1. Since T lymphocytes can produce RANTES, alveolar T lymphocytes which are increased during active tuberculosis could be a source of RANTES in the tuberculous lung.

In summary, MTB infection of mononuclear phagocytes increased production and/or release of four chemokines with chemoattractant activity for MN and lymphocytes and there was elevation of three of these in the BAL fluid of patients with active pulmonary tuberculosis. Kurashima and associates recently reported also that chemokines are increased in alveolar spaces during active tuberculosis (42). Whether the increased levels of chemokines in alveolar spaces during tuberculosis are a consequence of MTB infection of cells cannot be concluded from our study. The results suggest, however, that MTB infection of cells may directly contribute to the rise in chemokine levels. Rhoades and colleagues demonstrated in a mouse model that chemokines are induced by various strains of MTB both in vitro and in vivo, but growth characteristics of MTB in the model suggested that chemokines may not control the protective granulomatous response (43). Nevertheless, in humans, monunuclear cell recruitment by chemokines could be a pivotal factor in granuloma formation. The release of multiple chemokines at a single site may be a biologic strategy that is the rule rather than the exception, as suggested by studies of other inflammatory conditions such as rheumatoid arthritis and atherosclerosis (3). Such redundancy of chemotactic molecules likely assures that the appropriate cells at the appropriate times are recruited to the site of disease activity.