INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mycobacterial infections, particularly those caused by Mycobacterium tuberculosis, continue to represent a major public health problem. More than 8-million new cases of tuberculosis occur each year, and this infection causes more deaths worldwide than any other single pathogen (1, 2). Despite the manifest virulence of M. tuberculosis, less than 10% of infected individuals develop symptomatic disease. The mechanisms through which immune and inflammatory responses control mycobacterial growth in most individuals remain to be defined, and it is possible that better understanding of these processes will provide insights useful in guiding the development of new approaches to the control of mycobacterial infections.

Because mycobacteria have the ability to survive within mononuclear phagocytescells capable of killing most internalized organismsthese pathogens present a serious challenge to the immune system. In vivo studies have clearly shown that T-lymphocyte activation is indispensable for effective antimycobacterial immunity, and it is generally believed that signals delivered by T cells are responsible for improving the antimycobacterial activity of mononuclear phagocytes (3- 5). The nature of these signals remains to be clearly established. In murine models, cytokines, and particularly the combination of tumor necrosis factor- (TNF-) and interferon- (IFN-), induce strong macrophage mycobactericidal activity (6). Multiple attempts to reproduce these findings with human macrophages have failed (10), and although cytokines have been identified that limit mycobacterial growth in human macrophages (10), no cytokine or combination of cytokines has been identified that reproducibly induces the expression of strong mycobactericidal activity.

A second mechanism through which T cells can transmit signals to macrophages is in the course of cell-cell interactions. A number of different ligand-receptor pairs are expressed on both T cells and macrophages that can deliver such signals. For example, the interaction of macrophage ligands with T-cell costimulatory molecules is known to play an important role in the course of T-cell activation (11), and recent evidence indicates that these interactions can also modify the functional properties of macrophages, including their surface phenotype, release of proteases, secretion of cytokines, and expression of cytotoxic activity (12). In this regard, signaling via the macrophage CD40 molecule by cells expressing the cognate ligand has been shown to improve resistance to the intracellular pathogens Leishmania and Pneumocystis (16, 17). The potential importance of such activation signals in the development of mycobactericidal activity has received little attention.

One obstacle to the identification of signals stimulating mycobactericidal activity has been the relative difficulty of measuring survival of mycobacteria with standard techniques (e.g., quantification of colony forming units [cfu]). Pathogenic mycobacteria are slow-growing organisms, and several weeks are required to obtain results for a single experiment. Because each specimen is typically plated at several different dilutions, practical constraints limit the number of conditions that can be evaluated. In addition, mycobacteria have the strong tendency to form aggregates, and it is often difficult to determine whether techniques used to disperse samples have reduced mycobacterial viability and have successfully eliminated clumps of organisms. Recently, mycobacterial reporter strains have been developed that permit the rapid and sensitive assessment of mycobacterial number based on the detection of luciferase activity. Although these strains have been used successfully in assessing mycobacterial antibiotic resistance, mycobacterial survival in vivo, and the penetration of antibiotics into macrophages (18), their application to the evaluation of mycobactericidal activity has not been reported.

In the study reported here we developed an assay using an M. bovis Bacillus Calmette-Guerin (BCG) reporter strain to measure mycobacterial survival in human monocyte-derived macrophages. We then used this technique to compare the effect of pretreatment with a variety of cytokines and signaling through macrophage surface molecules on the ability of human macrophages to limit mycobacterial growth.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Isolation and Culture of Cells

Peripheral blood mononuclear cells (PBMC) from healthy volunteers were isolated from leukaphoresis concentrates by centrifugation on Lymphoprep (Nycomed Pharma, Oslo, Norway), and had a viability of greater than 95% in all cases. Monocytes were purified from PBMC by counterflow centrifugal elutriation, using a J2-21ME centrifuge and JE-6B rotor (Beckman Instruments, Palo Alto, CA) (22), and had a purity of greater than 90% in all cases as assessed by examination of May-Grunwald-Giemsa stained cytospin preparations.

Unless otherwised noted, human monocytes were cultured at 2 × 10⁵ cells/well in 96-well flat-bottom plates with opaque sides and transparent bottoms (EG&G Wallac, Turku, Finland), containing 200 µl of complete medium (Iscove's modified Dulbecco's medium [Sigma, St. Louis, MO] supplemented with 2 mM L-glutamine, 200 U/ml penicillin G, 1 µg/ml kanamycin (Sigma), and 20% human AB serum [Institut Jacques Boy, Reims, France]). Cultures were incubated at 37° C in 95% air/5% CO₂. To evaluate the effect of pretreatment with soluble mediators on monocyte activity, 1 × 10⁸ M 1,25(OH)₂-vitamin D₃ (Calbiochem, La Jolla, CA) and/or 10 ng/ml of the following recombinant human cytokines were added during the first 3 d of culture: interleukin-1 (IL-1), IL-2, IL-3, IL-4, IL-6, IL-7, IL-8, IL-10, TNF-, lymphotoxin- (LT-), transforming growth factor- (TGF-), granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF) (Boehringer Mannheim, Mannheim, Germany), macrophage colony-stimulating factor (M-CSF), IL-5 (Genzyme, Cambridge, MA), IL-12, IL-13, IL-15, IFN- (R&D, Abingdon, UK).

To stimulate monocytes through surface molecules, avidin-induced crosslinking was used (15). Briefly, cells were cultured for 3 d in the presence or absence of cytokines, and 4 h before infection the culture medium was aspirated and adherent cells were incubated for 1 h at 4° C in the presence of complete medium containing appropriate dilutions of biotinylated monoclonal antibodies recognizing one of the following surface antigens or control antibodies: CD4 (BL4 + 13B8.2; Immunotech, Marseille, France), CD40 (EA-5; Biosource International, Camarillo, CA), CD45 (HI-30), CD86 (IT2.2), and control antibody (107.3), all from Pharmingen, San Diego, CA. The medium was then removed and cells were washed and incubated for an additional hour at 4° C in complete medium containing 40 µg/ml avidin (Sigma). Cultures were briefly warmed to 37° C and infected as described subsequently. To evaluate crosslinking of Fas (CD95) molecules, cells were incubated for 1 h at 4° C with complete medium containing 10 µg/ml monoclonal anti-Fas IgM antibody 7C11 (23) or a control IgM antibody prior to infection. The phorbol myristate acetate (PMA)- induced respiratory burst of monocytes was measured by luminol- enhanced chemiluminescence as described (15). Following addition of PMA (Sigma; final concentration 100 nM), luminescence was integrated over 30 min with a MicroLumat LB96P luminometer (EG&G Berthold, Badwildbad, Germany).

Media, buffers, and culture supernatants did not contain detectable endotoxin as assayed with the Limulus amoebocyte lysate test (Sigma).

Preparation of a Mycobacterial Reporter Strain Expressing Luciferase Activity

To permit the expression of luciferase activity by M. bovis BCG, the Luc1 gene, encoding the enzyme luciferase from Photinus pyralis (24) under control of the PAN promoter (25), was inserted into a mycobacteria/Escherichia coli shuttle plasmid containing the mycobacterial origin of replication from pAL5000 (26), the E. coli origin of replication from pUC18, and encoding kanamycin resistance (Km), producing the plasmid pPV12. Electrocompetent M. bovis BCG 1173 P2 (Pasteur Institute, Paris, France) were then electroporated in the presence of 1 µg of plasmid DNA, as previously described (27). Transformant colonies were picked after 6 wk, and expanded in 7H9 liquid medium, and glycerol stocks were prepared and stored at 80° C. All experiments were done with cultures derived from a single glycerol stock.

Infection of Monocytes

Seven days before each experiment, a 1-ml aliquot of the glycerol stock was rapidly thawed and suspended in 24 ml of 7H9 medium (Difco, Detroit, MI) supplemented with Middlebrook ADC enrichment mixture (Difco), 20 µg/ml kanamycin (Sigma), and 0.05% Tween-80 (Sigma). Cultures were maintained in 12-well flat-bottom plates (Costar, Cambridge, MA) at 37° C in 5% CO₂ without shaking. This culture system reduced aggregation of mycobacteria and thereby facilitated the preparation of single-cell suspensions.

On the day of infection of cultured monocytes, mycobacteria were pelleted by centrifugation at 1,400 × g for 15 min and resuspended in 45 ml of complete medium by repeated pipetting. The suspension was then centrifuged at 150 × g for 10 min, and the upper 30 ml of supernatant was removed for use. Microscopic evaluation confirmed that these suspensions contained more than 80% single organisms. Organisms were > 95% viable as assessed with the live/dead viability stain (Molecular Probes, Eugene, OR). Luciferase activity was measured as described subsequently, and the suspension was diluted with complete medium to 10 to 20 × 10³ relative light units (RLU)/25 µl. To infect monocyte cultures, the medium was aspirated and replaced with 200 µl of fresh complete medium containing 25 µl of the mycobacterial suspension. After varying periods of culture, 175 µl of medium was removed from each well and the microplate was frozen at 20° C. A comparison of quantification of organisms by measurement of luciferase activity and numbers of cfu (see the following discussion) demonstrated that 1 cfu = 0.64 ± 0.18 RLU (mean ± SEM for five different suspensions).

Quantification of Mycobacterial Survival

Measurement of luciferase activity. To evaluate the number of organisms remaining after varying periods of culture, the luciferase activity expressed by the viable mycobacteria was measured. Ninety-six-well plates were thawed, and 75 µl of 25 mM Tris-HCl (pH 7.8) containing 2 mM dithiothreitol (DTT), 2 mM diaminocyclohexane tetraacetic acid, 1% Triton X-100, 10% glycerol, 2.5 mg/ml bovine serum albumin, and 1.25 mg/ml lysozyme (luciferase assay lysis reagent; Promega, Madison, WI) was added to each well and incubated at 25° C for 30 min to lyse the cells and bacteria. One hundred microliters of a solution containing 0.47 mM luciferin and 0.53 mM adenosine triphosphate (ATP) (luciferase assay reagent; Promega) was automatically injected into each well, and after a 1.6-s delay, luminescence was measured over a 50-s interval, using an EG&G Berthold MicroLumat LB96P luminometer. Results are expressed as RLU, and are reported as the mean ± SD of triplicate determinations.

Evaluation of cfu. To evaluate cfu, 96-well plates were thawed, 0.1 ml of lysis solution (0.016% digitonin [Sigma], and 0.1% Tween-80 [Sigma] in phosphate-buffered saline [PBS]) was added to each well and the plates were incubated at 37° C for 15 min. Each lysate was vigorously pipetted and transferred to a conical tube, and each well was washed with two additional 0.1-ml aliquots of lysis solution. Pooled lysates were vortexed and diluted serially in 7H9 medium, and 0.1-ml aliquots were plated on 7H10 medium supplemented with Middlebrook OADC enrichment mixture (Difco) and 20 µg/ml kanamycin, and were incubated at 37° C in 5% CO₂ for 21 d before colonies were counted.

Microscopic Evaluation

To evaluate the distribution of mycobacteria in the cultures, and to permit electron microscopic examination, we used a protocol that permitted the recovery of nonadherent cells and extracellular bacteria. Briefly, 50 µl of medium was removed and replaced with 50 µl of 8% albumin, after which the culture was left at room temperature for 15 min. Following this, an aliquot of 50 µl was again removed and replaced with 50 µl of PBS containing 5% (wt/vol) each of glutaraldehyde and paraformaldehyde. After 30 min at room temperature, a gelatinous layer of precipitated albumin had formed at the bottom of each well, and had captured all mycobacteria and cells. Each well was then washed with PBS, postfixed in 1% osmic acid for 30 min, dehydrated, and embedded in Epon. Sections of 5 µm thickness were stained with methylene blue and Azur II to allow counting of infected cells with light microscopy. Corresponding ultrathin sections were examined with a Philips EM 410 electron microscope (Einhoven, The Netherlands).

Statistical Analysis

Results are expressed as mean ± SD unless otherwise stated. Statistical comparisons were made by analysis of variance (ANOVA). A value of p < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Evaluation of Mycobacterial Number by Measurement of Luciferase Activity

To facilitate the evaluation of mycobacterial survival in vitro, we inserted the gene encoding firefly luciferase, under control of the mycobacterial PAN promoter, into a shuttle plasmid and used it to transform M. bovis BCG. To evaluate survival, mycobacteria containing the plasmid were cultured in the presence or absence of human monocyte-derived macrophages, and after varying periods of culture the medium was removed, the cells and mycobacteria were lysed, and luciferase activity present in the lysate was determined by luminometry.

Initial experiments demonstrated a linear relationship, extending over at least a 2 log range, between luciferase activity and the number of mycobacteria present in the culture (Figure 1A). The amount of light emitted was similar whether the measurements were made on cultures containing mycobacteria only or on cultures containing the same number of mycobacteria in the presence of varying numbers of macrophages, at up to 5 × 10⁵ cells/well. RLU values as low as 2,000 RLU/ well were clearly distinguishable from background luminescence. Mycobacterial killing resulted in the rapid loss of luciferase activity. Thus, when 10 µg/ml gentamicin was added to cultures containing mycobacteria, luciferase activity was reduced to near background levels within 24 h (Figure 1B).

View larger version (16K): [in this window] [in a new window]   Figure 1.   Evaluation of mycobacterial number by measurement of luciferase activity. (A) The indicated number of mycobacteria was added to 200 µl (final volume) of complete medium in the absence of monocytes (asterisks) or in the presence of 2 × 105 (open inverted triangles) or 5 × 105 (open triangles) cells. After 1 h in culture, 175 µl of medium was removed, the microtiter plate was frozen, and luciferase activity was subsequently measured as described in METHODS. Mycobacterial number was determined by determining the number of cfu in parallel cultures. The dashed line indicates the mean value for background luminescence (730 ± 14 RLU/well). Results are the mean for triplicate values from one of four experiments that gave similar results. For most dilutions, error bars are too small to be visible on the graph. (B) Mycobacteria were cultured in complete medium in the absence (solid triangles) or presence (open triangles) of 10 µg/ml gentamicin for the indicated times before measurement of luciferase activity. For comparison, the intracellular growth of mycobacteria cultured with 2 × 105 human monocytes is also shown (solid squares). The dashed line indicates the mean value for background luminescence (968 ± 24 RLU/well). Results are the mean ± SD for one of four experiments that gave similar results. Error bars for mycobacteria cultured in the presence of gentamicin are too small to be visible on the graph. (C ) Aliquots of 2 × 105 human monocytes were cultured for 3 d in 200 µl complete medium and infected with the M. bovis BCG reporter strain, as described in METHODS. After the indicated periods of culture, the number of mycobacteria was evaluated with the luciferase assay (RLU, open symbols) or through evaluating cfu (filled symbols). Results are the mean ± SD for six independent experiments done with cells from different individuals.

The kinetics of mycobacterial growth in human monocyte-derived macrophages are shown in Figure 1C. As assessed through luciferase activity (RLU), the number of mycobacteria increased during the first 48 h of culture and then reached a plateau. Similar kinetics were observed when mycobacterial number was measured by evaluation of cfu.

Under the culture conditions used in the study, mycobacteria were essentially all found within macrophages. Microscopic evaluation of semithin sections, prepared through a protocol in which extracellular mycobacteria are retained, indicated that > 99% of mycobacteria in these specimens were associated with macrophages. Electron microscopic evaluation confirmed that these bacteria were present within macrophage phagosomes (Figure 2). In accord with these observations, luciferase activity could not be detected in culture supernatants.

View larger version (142K): [in this window] [in a new window]   Figure 2.   Electron micrograph of monocyte-derived macrophages 96 h after infection with M. bovis BCG. Note that all organisms are contained within typical phagolysosomes. Original magnification: ×20,000.

Effect of Cytokines on Survival of M. bovis BCG in Human Macrophages

To identify cytokines that influence the ability of macrophages to limit mycobacterial growth, monocytes were precultured with various cytokines for 3 d before being infected, and mycobacterial number was evaluated 2 d and 4 d later. Of the 19 cytokines tested, only two, IL-3 and GM-CSF, were found to significantly reduce the growth of M. bovis BCG (Table 1). In no case did the addition of cytokines to cultures containing mycobacteria in the absence of macrophages modify the survival of these organisms (data not shown). To confirm the findings for IL-3 and GM-CSF, and to evaluate whether pretreatment with these cytokines prior to infection was required for their effect, we compared the bactericidal activity of cells exposed to cytokines at the time of infection with that of cells pretreated with cytokines in a second series of experiments. As shown in Figure 3, mycobacterial survival at 96 h was again reduced in monocytes pretreated for 3 d with GM-CSF or IL-3, but was not different from that of control cells when cytokines were added at the time of infection.

View larger version (25K): [in this window] [in a new window]   Figure 3.   The effect of IL-3 and GM-CSF on mycobacterial survival in human macrophages. Aliquots of 2 × 105 human monocytes were cultured in the presence of GM-CSF or IL-3 (10 ng/ml) for 3 d (hatched bars) or in complete medium alone (open bars), and were then infected with the M. bovis BCG reporter strain. GM-CSF or IL-3 (10 ng/ml) were added to the indicated unpretreated cultures on the day of infection. After 4 d of culture, 175 µl of medium was removed, microtiter plates were frozen, and luciferase activity was subsequently measured. Results are expressed as the percentage of values obtained for unpretreated cells, and represent the mean ± SD for three independent experiments (p < 0.01 for unpretreated cells versus cells pretreated with IL-3 or GM-CSF).

Both GM-CSF and IL-3 have been reported to induce proliferation of monocytes, but several experiments in our study demonstrated that the effect of these cytokines on mycobacterial survival did not result from changes in cell number. First, measurement of DNA content showed that the two cytokines had only a modest effect on cell number (22% and 27% increase, respectively, for GM-CSF and IL-3 at 48 h). As described earlier (Figure 1), the measurement of luciferase activity was insensitive to changes in cell number in this range. Furthermore, doubling the number of unpretreated monocytes in the cultures (e.g., 4 × 10⁵ cells) did not significantly modify mycobacterial survival at either 2 d or 4 d. Increasing the concentration of IL-3 and GM-CSF from 10 to 20 ng/ml did not further reduce mycobacterial survival, and preincubation of monocytes in the presence of both cytokines was no more effective than pretreatment with a single cytokine (data not shown). It should be emphasized that macrophages pretreated with either GM-CSF or IL-3 were at best bacteriostatic, and in no case were cells able to reduce mycobacterial number at 4 d below its initial value.

Mycobacterial survival was approximately 20% greater in macrophages pretreated with IL-10 and IFN- than in control cells at 4 d (Table 1). These differences did not, however, achieve statistical significance during the initial screening.

Effect of Stimulation via Costimulatory Molecules on Mycobacterial Survival in Macrophages

Macrophages express numerous surface receptors capable of transmitting signals to the cell. To begin to evaluate the potential role of such molecules in regulating mycobactericidal activity, we stimulated macrophages via molecules known to participate in macrophage/lymphocyte interactions (CD4, CD40, CD45, B7-2 [CD86], and Fas [CD95]). To crosslink the receptors, cells were incubated at 4° C in the presence of biotinylated monoclonal antibodies recognizing the surface molecules, followed by incubation with avidin. Cells were then warmed to 37° C and infected, and mycobacterial survival was followed. As shown in Figure 4A, stimulation through these surface receptors had no significant effect on mycobacterial survival measured at 2 d and 4 d after infection. These negative results could not be explained by the absence of these molecules on the cell surface or by failure of crosslinking to stimulate the cells. Cytofluorometric studies confirmed expression of the molecules on the monocyte-derived macrophages on the day of stimulation. In addition, crosslinking of biotinylated CD45 on these cells resulted in a 60% increase in their PMA-induced respiratory burst as measured with luminol- enhanced chemiluminescence, and the Fas antibody at the concentration used in these studies induced apoptosis in > 95% of Fas-sensitive Jurkat cells (data not shown). Moreover, in no case did crosslinking of these molecules, including Fas, modify the viability of monocyte-derived macrophages.

View larger version (46K): [in this window] [in a new window]   Figure 4.   Effect of stimulation through costimulatory molecules on the survival of M. bovis BCG in human macrophages. Aliquots of 2 × 105 human monocytes were cultured for 3 d in the presence of complete medium (A) or medium containing 10 ng/ml IFN- (B). Cells were then incubated at 4° C with biotinylated monoclonal antibodies recognizing CD4, CD40, CD45, or CD86, washed once, and incubated with 40 µg/ml avidin. Cells were then warmed to 37° C and infected with M. bovis BCG for 48 h (open bars) or 96 h (hatched bars) prior to evaluation of mycobacterial survival with the luciferase assay. For stimulation via CD95, cells were incubated in the continuous presence of a monoclonal anti-Fas IgM antibody. Results are expressed as the percentage of values obtained for unpretreated cells reacted with a control monoclonal antibody, and represent the mean ± SD. CD4, CD40, CD45 (n = 8); CD86, CD95 (n = 4). In no case were results significantly different from those obtained for cells reacted with the control antibody.

Interferon- and Mycobacterial Survival

Recent in vivo studies have underscored the importance of IFN- for control of mycobacterial infection in humans (28, 29). Because pretreatment of macrophages with IFN- alone appeared to have a slight deleterious effect on mycobactericidal activity, we made attempts to uncover positive effects by combining this cytokine with other stimuli. Potent mycobactericidal activity has been reported for human macrophages exposed to IFN- and TNF- in the presence of 1,25(OH)₂D (30). No evidence for enhanced control of mycobacterial infection was observed in our system, however, for macrophages treated with either of these two cytokines alone, with 1,25(OH)₂D alone, or with the three mediators together (Figure 5). Indeed, pretreatment with the combination of IFN- and TNF- led to substantial increase in mycobacterial growth in two of five experiments, whether or not cells were pretreated with 1,25(OH)₂D. Similarly, pretreatment of macrophages with IFN- in the presence of GM-CSF or IL-3 did not lead to improved mycobactericidal activity compared with that of cells not receiving IFN-. Nor did the pretreatment of cells with IFN- prior to stimulation via costimulatory molecules unmask any latent effect of such signaling on mycobacterial survival (Figure 4B). In each of the studies described here for evaluating the effect of IFN-, this cytokine tended to impair mycobactericidal activity, but because of considerable variability in the responses, in no case were significant differences observed. When all of these experiments were considered together in a paired analysis, however, pretreatment with IFN- was found to significantly increase mycobacterial survival at both 48 h and 96 h (p < 0.01 for both comparisons).

View larger version (31K): [in this window] [in a new window]   Figure 5.   Effect of IFN-, TNF-, 1,25(OH)2D3, and the combination of these mediators on the survival of M. bovis BCG in human macrophages. Aliquots of 10 ng/ml TNF-, 1 × 108 M 1,25(OH)2D3, or combinations of these mediators, as indicated, were used to treat human monocytes, which were then infected with M. bovis BCG for 48 h (open bars) or 96 h (hatched bars) before mycobacterial survival was evaluated with the luciferase assay. Results are expressed as the percentage of values obtained for unpretreated cells, and represent the mean ± SD for five independent experiments.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

To permit comparison of the effect of numerous different conditions on macrophage mycobactericidal activity, the availability of procedures for evaluating mycobacterial survival that are more rapid and more easily performed than the traditional measurement of cfu is desirable. The strategy used in the present study, based on the measurement of luciferase activity expressed by viable mycobacteria, proved to be quite useful for several reasons. First, the technique is simple, adaptable to a microplate format, and amenable to automation. Using a luminometer with automatic injection and a 96-well format, only a single day is required to evaluate in triplicate the effect of a variety of stimuli on mycobacterial survival at several different time points. Nor is the assay influenced by the number of human cells present over a wide range. Furthermore, mycobacterial killing results in a rapid loss of luciferase activity, making the assay sensitive to changes in mycobacterial viability. As previously reported by groups who have used this approach to follow mycobacterial survival under other conditions (18), we found that mycobacterial survival in macrophages was comparable when evaluated by the measurement of luciferase activity and by measurement of cfu.

The effect on mycobacterial survival of several of the cytokines evaluated in our study has not been previously reported (i.e., IL-12, IL-13, IL-15, LT-). However, previous studies have evaluated the effects of other mediators. Most prior studies have found that exposure of macrophages to GM-CSF improves their ability to control infection with M. tuberculosis and Mycobacterium avium (31), which is compatible with the findings in this study. The mechanisms responsible for this effect remain unknown. Although GM-CSF is known to be a macrophage growth factor, our studies showed that a change in cell number did not explain the observed reduction in mycobacterial survival. Prior work has indicated that the generation of reactive oxygen intermediates also does not seem to be involved in this effect (33, 35). The recent demonstration that the local production of GM-CSF correlates with the intensity of the granulomatous response in tuberculosis (36) supports the idea that this cytokine may play an important role in host defense in vivo.

In contrast, the ability of IL-3 to improve resistance to mycobacterial infection has not been previously reported. Indeed, the few studies evaluating the effect of this cytokine on mycobacterial growth in human macrophages reported either no effect (M. tuberculosis infection) or impairment (M. avium infection) of mycobactericidal activity by IL-3, and a stimulatory effect of this cytokine on the growth of M. avium in the absence of macrophages has been described (30, 37, 38). The reasons for the differences between these results and those reported here remain unknown; manifest differences in the experimental protocols used in different studies (e.g., mycobacterial species or strains, culture conditions, dose and timing of cytokine administration, multiplicity of infection) are obvious possibilities. Our findings that both GM-CSF and IL-3 improve macrophage resistance to mycobacterial infection, and that the effects of the two cytokines are not additive, are consistent with prior information about these mediators. First, both GM-CSF and IL-3 have been reported to enhance bactericidal activity against other pathogens (39). In addition, these cytokines are also known to have similar effects on a variety of other macrophage properties, including differentiation, production of inflammatory mediators, and sensitivity to cytotoxic T cells (40). Because the receptors for these two cytokines share a common -subunit and can use common second messengers (43), additive effects of the second cytokine, under conditions of maximal signaling with one cytokine, would not be expected, owing to competition for the -subunit (40, 42, 43).

IFN- and TNF-, and particularly the combination of these two cytokines, reproducibly induce strong mycobactericidal activity by murine macrophages (6, 7, 44). Numerous attempts to demonstrate similar effects of these cytokines on human macrophages have generally failed. Thus, treatment with IFN- has been reported to result in impairment (33, 44- 48), no effect (30, 37, 38, 49), or only modest improvement (35, 52) in mycobacteriostatic activity. The findings in the present study (overall slight impairment), are entirely compatible with these results. More controversial have been reports that pretreatment of human macrophages with the combination of 1,25(OH)₂D, IFN-, and TNF- can induce strong mycobactericidal activity in these cells (30, 53). Others, however, have reported either no effect or an increase in mycobacterial proliferation in monocytes treated with these mediators (55, 58), and it has been suggested that the apparent strong mycobactericidal activity observed in some studies could reflect inadvertent loss of mycobacteria prior to their enumeration (58). Our study, using techniques in which loss of viable mycobacteria does not occur, also failed to show any evidence of mycobactericidal activity of monocytes pretreated with 1,25- (OH)₂D, IFN- and TNF-, and therefore strongly supports the negative findings in other studies. Because of the irrefutable evidence that IFN- is essential for effective antimycobacterial immunity in humans (28, 29), the failure to demonstrate a potent effect of this cytokine in vitro is enigmatic. Prior studies suggested that the bacteriostatic activity of IFN- could be improved if the effects of TGF- or prostaglandins were inhibited, although the overall activity observed in these studies was generally modest (37, 38, 50, 54, 56). We evaluated the effect of IFN- in association with a variety of other cytokines and stimulation through costimulatory molecules, but found no evidence for the expression of striking bactericidal activity by macrophages. Further studies will be needed to determine whether IFN-, working in concert with other, as yet unidentified signals, is directly involved in inducing strong mycobactericidal activity in human macrophages, or is required for other aspects of a successful immune response (e.g., induction of cytokine release from other immune/inflammatory cells, stimulation of granuloma formation, modulation of T-cell immunity).

In this study, crosslinking of several different costimulatory molecules did not lead to improved mycobactericidal activity. Cytofluorometic studies confirmed that these molecules were expressed by macrophages at the time of activation. Furthermore, pretreatment of cells with IFN-, which increased the expression of CD40 and CD86, and to a lesser extent Fas, did not modify these results. In accordance with our findings is the previously reported absence of an effect of Fas crosslinking on survival of M. avium in human macrophages (59). Stimulation of CD45 was shown to increase the production of reactive oxygen intermediates, and an anti-Fas antibody was effective in inducing apoptosis in Jurkat cells, demonstrating that the antibodies used in both cases were capable of signaling. Because the antibodies were not continuously present in the cultures, however, such signaling was transient. Thus, continuous stimulation obtained with soluble ligands might produce effects not observed in our study. In view of the numerous effects of signaling through these molecules on macrophage phenotype and function in vitro, and the evidence for their participation in vivo in responses against intracellular pathogens, further studies are clearly warranted to evaluate the potential of these molecules in modifying macrophage mycobactericidal activity. In this regard, recently presented data have suggested that lymphocyte/monocyte interactions improve mycobactericidal activity, although the putative surface molecules involved were not identified (60, 61). By analogy with lymphocyte activation, in which simultaneous interaction of at least two such receptor-ligand pairs is known to be required for effective stimulation, combinations of signals delivered via cell-cell contact and/or soluble mediators may be required.

In the experiments conducted in our study, the density of monocytes used was sufficient to form complete monolayers in the microplate wells, insuring that the mycobacteria, which settle to the bottom, would contact these cells. No evidence for extracellular mycobacteria was found in morphologic studies, and luciferase activity could not be detected in the supernatants. Although all mycobacteria observed by electron micrography were present within macrophages, these findings do not eliminate the possibility that a small proportion of mycobacteria remained associated with the cell surface but were not internalized. Two observations suggest, however, that any differences in the internalization of mycobacteria due to pretreatment with cytokines would have little influence on the results obtained. First, the growth of extracellular organisms was only modestly reduced as compared with that of intracellular bacteria; second, the presence of the cytokines evaluated in our study did not influence the growth of extracellular organisms. It should also be emphasized that human monocytes can produce some of the cytokines evaluated in our study, and it is possible that autocrine production of these cytokines could in some cases have masked the detection of positive effects of added recombinant cytokines. Nevertheless, a beneficial effect of exogenous GM-CSF was observed, despite the prior finding that human monocytes produce GM-CSF after mycobacterial infection (34). Further use of the techniques described here for evaluating the survival of mycobacteria should help in evaluating of the potential role of autocrine stimulation in this system, and in the identification of other signals required for optimal expression of mycobactericidal activity by mononuclear phagocytes.