The Inhibitory Action of Mycobacterium ulcerans Soluble Factor on Monocyte/T Cell Cytokine Production and NF-κB Function¹ Free

Buruli ulcer, caused by Mycobacterium ulcerans, is the third most common mycobacterial disease after tuberculosis and leprosy, in the immunocompetent host (1). M. ulcerans infection is characterized by a massive necrosis at the site of infection frequently followed by debilitating disfiguration and a paucity of inflammatory cells within the necrotic areas (2, 3, 4, 5, 6). Other than radical surgery there is no treatment for the disease, and often severe disability is a consequence. Unlike other mycobacterial diseases, histology shows that M. ulcerans remains mainly extracellular (2, 7, 8). The containment of small foci of M. ulcerans growth in large regions of necrotic tissue led Connor and Lunn (3) to suggest that M. ulcerans causes massive necrosis by producing a soluble toxic substance. Culture filtrates (CF)³ of M. ulcerans have been found to contain an active factor. The toxic effect of the CF has been demonstrated both in animal models using guinea pigs and in the murine fibroblast cell line L929 (7, 8, 9).

The persistence of the M. ulcerans infection and the ineffectiveness of antibiotics are thought at least in part to relate to a local immunosuppression at the site of infection. The relationship between M. ulcerans, and the immunosuppressive effect is unknown. Therefore, in this study we investigated the activity of a partially purified extract of M. ulcerans factor on a variety of human immune-competent cells and their products. We provide evidence of effects of the factor on T cell and monocyte functions. The factor also had a selective effect on the activity of TNF, a major cytokine in anti-microbial responses. Our data might explain the underlying immunological unresponsiveness and specifically the poor inflammatory reaction during the necrotizing stage of Buruli ulcer disease. These studies would indicate that neutralization of the toxin would relieve the immunosuppressive effect associated with the disease and may provide a new aspect to future treatment of this disabling condition.

The pathogenic strain (7634) from an original clinical isolate and the nonpathogenic strain (5113) were provided by Dr. J. L. Stanford (Middlesex Hospital, London, U.K.) and Prof. F. Portaels (Institute of Tropical Disease, Antwerp, Belgium), respectively. Mycobacteria were grown in Dubos broth base medium (Difco, Detroit, MI) as previously described (8). Pathogenic activity was determined by mouse footpad inoculation.

CF were prepared according to the method of Read et al. (9) and Hockmeyer et al. (8) from bacteria in exponential growth phase. HDL fraction was isolated from CF following KBr density gradient ultracentrifugation and dialysis. The final protein concentration (10) of HDL fraction was adjusted to 100 μg/ml, and aliquots were kept at −70°C until use; 1 L typically provided 5 ml of HDL. In this paper the acronyms aHDL and iHDL refers to preparations obtained from CF in which pathogenic and nonpathogenic M. ulcerans were grown, respectively. A chemically defined extracted factor from M. ulcerans CF termed MUPT (M. ulcerans polyketide toxin) was also used in our studies (11).

HDL was treated with cold ethanol (−20°C) at a dilution of 1/5 (v/v) overnight. The samples were centrifuged for 20 min at 13,000 × g in a microfuge at 4°C. The precipitate was washed twice with the same volume of ethanol and further centrifuged at 4°C. The precipitate was dissolved in 10 ml of PBS and then concentrated to the original volume. The ethanol extract was vacuum evaporated. The dried material was then dissolved in 10 ml of PBS and subsequently concentrated to the original volume. Samples were also heat treated in the autoclave for 20 min at 121°C (15 lb/ in²) and stored at −20°C. HDL was treated with 28 mM SDS for 2 h at room temperature. The samples were dialyzed against PBS and concentrated to the original volume by using Centricon-3 concentrator (Amicon, Danvers, MA).

The murine fibroblast cell line L929, murine fibrosarcoma cell line WEHI 164 clone 13, and murine macrophage cell line RAW 264.7 were maintained in DMEM (BioWhittaker, Walkersville, MD) supplemented with 10% FCS, 2 mM l-glutamine, penicillin, and streptomycin at final concentrations of 100 IU/ml and 100 μg/ml, respectively. The murine IL-2-dependent CT6 T cell line, murine pre-B cell line 70Z/3, and human T cell line Jurkat subclone 3KB5.2 (NF-κB) containing β-galactosidase reporter genes (12) were maintained in RPMI with 5% FCS and antibiotics as described above. Human PBMC were prepared from blood bags (North London Blood Transfusion Centre, Edgware, U.K.) using Ficoll (13). Human peripheral monocytes were isolated from PBMC by elutriation. Monocyte enrichment was confirmed by flow cytometry using a FACS (FACScan, Becton Dickinson, Immunocytometry Systems, San Jose, CA, USA) with forward and side scatter characteristics and staining with fluorochrome-conjugated anti-CD45 and anti-CD14 Abs (Leucogate, Becton Dickinson). The human T cell line 3895 was provided by Dr. S. Cohen (14). Monocytes were activated by 10 ng/ml LPS extracted from Salmonella abortus equi, provided by Drs. C. Galanos and M. Feudenberg (Max Plank Institute, Freiburg, Germany). PBMC or T cell lines were activated with 1 μg/ml PHA from Difco, 50 ng/ml PMA from Sigma (St. Louis, MO), and 1 μg/ml ionomycin (IMN) from Calbiochem (La Jolla, CA) or solid phase mouse anti-human CD3 mAb (10 ng/ml OKT3). This and subsequent proliferation assays were performed according to Katsikis et al. (13).

One-tenth milliliter of concentrated and filter-sterilized CF and HDL was injected intradermally into the right and left flanks of the guinea pigs. The animals were monitored up to 48 h postinjection. Skin samples from inoculated areas of the guinea pigs were removed and fixed in 4% (v/v) neutral buffered formalin. The sections were stained (hematoxylin and eosin) and examined histologically.

This was assayed using 1% nigrosin dye in an exclusion technique (9). Cells were collected, centrifuged, resuspended in DMEM, and stained by nigrosin. At least 200 cells for each sample were enumerated, and the percentage of live and dead cells was determined. Cell death was equated with the cytopathic effect (CPE). The CPE was graded from zero (indicating no CPE) to 4+ (indicating >80% cell death). Human monocytes were simultaneously treated with aHDL and LPS for 12 h. Cells were examined microscopically, and the number of cell clumps was counted. A minimum of 10 separate fields were analyzed.

Culture supernatants were assayed for IL-2 by ELISA (Biokine IL-2 Test Kit, T Cell Science, Cambridge, U.K.). TNF production was assayed by ELISA (15), and reagents were gifts from Dr. W. Buurman (Rijks Universiteit Limburg, Maastricht, The Netherlands). Recombinant human TNF, a gift from Prof. W. Stec (Center of Molecular and Macromolecular Studies, Polish Academy of Science, Lodz, Poland), was used to construct a standard curve. Alternatively, TNF was assayed by a bioassay technique using WEHI 164 (clone 13) cells (16). Reagents for the IL-10 ELISA were purchased from PharMingen (San Diego, CA) and performed as described previously (17). Recombinant human IL-10, a gift from Dr. S. Smith (Schering Plough, Kenilworth, NJ), was used to construct a standard curve.

Jurkat T cells were washed twice and resuspended at 2 × 10⁶ ml in complete RPMI without phenol red. One hundred microliters per well of cell suspension was added to a 96-well plate and treated in the presence or the absence of HDL with stimuli; TNF (10 ng/ml), PMA (50 ng/ml), or a combination of PMA (50 ng/ml) and IMN (1 μg/ml) in triplicate overnight. The cells were lysed by adding 20 μl of Triton X-100 (1% in 25 mM TBS) and then adding 25 μl of enzyme substrate, chlorophenol red-β-d-galacto-pyranoside (CPRG) (Boehringer Mannheim, Indianapolis, IN; 8 mg/ml in 25 mM TBS). The cell lysates were incubated at 37°C. The absorbance was then measured at 574 nm by a Labsystem Multiskan Bichromatic spectrophotometer (18).

Nuclear proteins were extracted and assayed for NF-κB DNA binding activity using an oligonucleotide encoding the NF-κB binding sequence according to Clarke et al., (18).

Cells were lysed in buffer containing 1% Triton X-100, 10% glycerol, 20 mM HEPES (pH 7.4), 150 mM NaCl, 2 mM EGTA, 50 mM β-glycerophosphate supplemented with 1 mM sodium orthovanadate, 1 mM PMSF, 1 mM DTT, 3 μg/ml aprotinin, 5 μg/ml_, leupeptin, and 10 mM NaF and left on ice for10 min. The cell suspension was spun down in a microfuge (10 min, 13,000 × g, 4°C), and the supernatant was collected followed by determination of the protein concentrations in samples using a bicinchoninic acid protein assay kit (Pierce, Rockford, IL). Equal protein samples were resolved by SDS-PAGE followed by transfer to nitrocellulose or polyvinylidene difluoride membranes. Immunoblotting was performed using rabbit Abs to IκBα (Ab C-15, Santa Cruz Biotechnology, Santa Cruz, CA) and phospho-p38 MAPK (New England Biolabs, Beverley, MA). Blots were developed with pig anti-rabbit HRP conjugate Ab (Dako, Carpinteria, CA) in conjunction with the ECL Western blotting system (Amersham Life Science, Arlington Heights, IL) and further exposure to Hyperfilm MP (Amersham Life Science, Buckingham Shire, UK) at room temperature (18).

The labeling protocol of DNA for flow cytometric analysis has been previously described (19, 20). Briefly, the cell pellet was gently resuspended in 1.5 ml of hypotonic fluorochrome solution (apoptosis study) or the isotonic solution (viability study). The fluorochrome used contained 50 μg/ml propidium iodide (PI) and 0.1% sodium citrate in PBS (viability). Alternatively, the PI and citrate were dissolved in 0.1% (v/v) Triton X-100 (apoptosis). The PI fluorescence of individual nuclei was measured using a FACScan flow cytometer (FACScan, Becton Dickinson, Immunocytometry Systems, San Jose, CA, USA)

The concept of an exotoxin causing Buruli ulcer when first proposed (3, 7, 9) employed both animal and tissue culture systems. Originally, murine fibroblast L929 was the cell line used to assess the biological activity of M. ulcerans factor, and it has remained the cornerstone for all further studies. L929 cells were used in a 48-h assay, and the CPE of the exotoxin was graded from 1+ to 4+ by eye. Using this basic assay we partially purified the exotoxin by KBr density gradient centrifugation and ultrafiltration. The exotoxin also caused lesions, reminiscent of Buruli ulcer (Fig. 1,A) when injected intradermally in guinea pigs; the extracts from the nonpathogenic strain had no effect (results not shown). Lack of accurate quantitative analysis led to a search for a more appropriate assay. Different cell lines either of human or murine origin were tested (data not shown). Murine 70Z/3 was the most sensitive cell line in a 24-h, [³H]thymidine/proliferation assay. There is a marked difference in the activity of extracts on 70Z/3 cells (Fig. 1,B) obtained from the culture filtrates of pathogenic and nonpathogenic M. ulcerans (referred to aHDL and iHDL, respectively). The aHDL had 50% activity at 1/1280, whereas the iHDL was >1/40. In contrast, aHDL showed no activity on Jurkat T cells in the same assay (Fig. 1,C). We used the 70Z/3 assay to determine the physicochemical properties of aHLD. The activity was totally extractable in organic and polar solvents and was resistant to extremes of pH and moderate heating (Table I). However, autoclave conditions halved activity (Fig. 1,B and Table I), and the inhibitory effect was destroyed by SDS treatment.

We examined the effect of aHDL on cell cycle progression of 70Z/3 cells. After 24-h aHDL (1/400) treatment there was inhibition of cell cycle progression, as shown by a decrease in the G₂/M peak (Fig. 2,A). This was dose dependent. Also, a broad sub-G₁ PI fluorescence indicative of apoptotic DNA fragmentation was observed (Fig. 2,A). This was not seen in untreated cells or in cells exposed to iHDL. At 48 h, the apoptotic effect was more pronounced, affecting the majority (1 in 400) of cells. Again, this was dose dependent. The iHDL had no effect on either cell cycle progression or apoptosis. Analysis of cell viability by PI exclusion flow cytometry supported the above data and showed an increase in viability with lower dose (Fig. 2 B).

The failure of the host to mount an immune challenge to M. ulcerans infection (21, 22) suggested that the infection is associated with an immunosuppressed state. A previous report showed an inhibitory effect of crude culture filtrate on the Con A-stimulated proliferation of murine lymphocytes (23). This suggested that M. ulcerans factor may mediate immunosuppressive activity. Our own results with human cells showed a similar inhibition by aHDL (but not iHDL) on PHA- or Ag (tetanus toxoid)-induced proliferation of human PBMC/T cells (data not shown). The study was extended by examining the effect of aHDL on IL-2 production by T cells. This was investigated in three systems, an anti-CD3-activated T cell line (Fig. 3,A), PMA/ionomycin-stimulated PBMC (Fig. 3,B), and PMA/ionomycin-stimulated Jurkat T cells (Fig. 3 C). In each case aHDL abolished the production of IL-2, whereas iHDL had no effect. The aHDL also appears to have an effect on T cell growth directly, as the addition of exogenous IL-2 to cultures failed to restore PHA-induced PBMC proliferation (results not shown).

Monocytes/macrophages and the monokine TNF are key elements in anti-mycobacterial responses. LPS-induced TNF production from purified human peripheral blood monocytes was markedly inhibited by aHDL, whereas iHDL had no effect (Fig. 4,A). The effect was dose dependent with an IC₅₀ of 1/1600 (Fig. 5,B), comparable to that of the 70Z/3 assay. The suppression of TNF production was not due to increased induction of the potent anti-inflammatory cytokine IL-10, because the LPS-induced expression of IL-10 was also inhibited by aHDL (Fig. 4,C). As a consequence of measuring TNF activity by bioassay using WEHI 164 as well as by ELISA, we assayed the effect of aHDL on TNF-induced apoptosis (24). aHDL had no effect on TNF activity in this assay, although aHDL alone had a marginal, but not significant, effect on the cell line itself (Fig. 4,D). These data indicate that the effects of aHDL are not indiscriminate. Regardless of the profound inhibitory effect of aHDL on cytokine production by monocytes, there was no loss of viability as measured by dye exclusion (data not shown); however, we observed that the factor prevented monocytes from adhering to plastic and forming clumps in a dose-dependent manner following LPS activation (Fig. 5).

The observation that production of the cytokine was inhibited but TNF-mediated cytotoxicity was not led to examination of another aspect of this cytokine’s activity, the activation of NF-κB. The importance of NF-κB in the transcription of a number of proinflammatory genes make this transcription factor key to the immune inflammatory response (reviewed in Ref. 25). We investigated the effect of aHDL on TNF-induced NF-κB activation in the human Jurkat T cell subline 3KB5.2 that contains an NF-κB-linked β-galactosidase reporter gene. TNF-induced activation of the reporter gene was inhibited by aHDL, but not by iHDL (Fig. 6,A). In contrast, aHDL had a synergistic effect on NF-κB activation when the cells were stimulated with PMA (Fig. 6,B). No consistent effect on NF-κB activity of aHDL alone was observed (Fig. 6, A and B). Recently, George et al. (11, 26) also isolated a purified active component, termed mycolactone, from M. ulcerans by alternative means. We examined this material for its effects on both TNF- and PMA-induced activation of NF-κB. The purified material, like aHDL, also reproduced the inhibition of TNF-induced NF-κB activation and the enhancement of PMA signal (Fig. 6, C and D). The fact that both preparations, aHDL and mycolactone, behaved similarly and had an identical band on TLC (data not shown) implies that both activities are exhibited by one molecule. We also examined the effect of aHDL on NF-κB activation using EMSA (Fig. 7,A). As in the reporter gene assay, aHDL inhibited NF-κB DNA binding activity. In contrast, aHDL slightly enhanced (∼10%) PMA-induced DNA binding activity (Fig. 7,A, detected by phosphorimager analysis). We also examined a proximal event of NF-κB activation, the degradation of IκBα. Despite the marked inhibition of nuclear NF-κB DNA binding activity, there was no inhibitory effect on IκBα degradation (Fig. 7,B). These data would suggest that aHDL inhibits NF-κB activity at some undefined stage after nuclear translocation. Unexpectedly, aHDL does not prevent the resynthesis of IκBα, although it is generally thought that this requires NF-κB. Previous studies have suggested (27) that p38 MAPK may have a role in NF-κB activation subsequent to IκBα degradation. As shown in Fig. 7,C, aHDL had no effect on the TNF-induced stimulation of this kinase, as measured by Western blot with an Ab specific for the phosphorylated tyrosine and threonine residues required for kinase activation. The inhibition of NF-κB activation would provide a broad mechanism by which M. ulcerans could effectively paralyze the immune response, as many proinflammatory mediators, including TNF, have an NF-κB site in their promoters. We therefore investigated the effect of aHDL on LPS- and IL-1-induced NF-κB activity. The aHDL had no effect on nuclear NF-κB EMSA activity induced by LPS or IL-1 (Fig. 8). This would suggest that the effects of aHDL are targeted to elements of the TNF signaling pathways to NF-κB that are distinct from pathways used by IL-1 and LPS (see Discussion below). We did note that unlike Jurkat and HeLa cells, aHDL induced some NF-κB activity in RAW 246.9 cells. This finding was variable from experiment to experiment.

M. ulcerans produces an indolent, necrotizing skin disease (Buruli ulcer). Uniquely for a mycobacterium, this is thought to involve elaboration of a soluble factor/toxin. The disease is associated with the lack of an immune response, and there is no effective treatment except radical surgery. We attempted a purification of the factor to determine whether it provided the underlying cause of the immunosuppression. We observed that the partially pure factor had profound inhibitory effects on the elaboration of TNF and IL-2 by monocytes and T lymphocytes, respectively. The factor would therefore be expected to have potent inhibitory effect on mounting an immune-inflammatory response to these mycobacteria.

Previous attempts to elucidate the role of the M. ulcerans toxin have focused solely on the cytotoxic activity described in the L929 murine fibroblast line. We used this as an assay for identifying the active extract during purification. Analysis of the activity of our product, aHDL, showed that the cytotoxic effect was variable between cell lines. in a pre-B cell line, 70Z/3, the most sensitive line we examined, aHDL had a profound inhibitory effect on proliferation, as measured by thymidine incorporation after 24 h. Cell cycle analysis at this time point showed no apparent arrest, but evidence of apoptosis. By 48 h apoptosis was complete, and no viable cells were detected. It is unclear whether aHDL induces apoptosis directly or whether cell death follows inhibition of an undefined earlier event in the cell cycle. This differs from the finding in a recent study, in which L929 cells arrested in G₀/G₁ (11). However, recent studies (K. George and P. Small, manuscript in preparation) have also reported that longer exposure (72 h) of L929 cells to M. ulcerans factor results in apoptosis. These differences in timing are most likely related to the use of alternative transformed cell lines. We used the exquisite sensitivity of the 70Z/3 assay to examine some of the basic physicochemical properties of aHDL. The heat insensitivity, solubility in organic solvents, and resistance to a range of pH indicated a nonproteinaceous organic-like molecule. This has been confirmed by the very recent structural elucidation of the factor (26).

As localized immunosuppression is key to M. ulcerans infection, we examined whether aHDL could be responsible. The only previous study (23) showed that there was an inhibition of Con A-induced proliferation of murine T cells by crude culture filtrate. Using aHDL, we observed a similar effect on anti-CD3- and PHA-stimulated human T lymphocytes. Further studies showed that the factor had a profound inhibitory effect on IL-2 production from primary T cells and the Jurkat T cell line activated with a range of stimuli. There was also a marked inhibition of proliferation induced by IL-2 and other T cell growth factors. By inhibiting both the production of IL-2 and the growth-inducing effect of this cytokine, the M. ulcerans factor would have a potent inhibitory effect on the adaptive arm of the immune system. The mechanism(s) by which aHDL inhibits the production of IL-2 by T cells is the subject of ongoing work.

Our studies have begun to examine the aspects of innate immune response, which is key to any anti-microbial challenge. The aHDL potently inhibited LPS-induced TNF production in human monocytes. The factor also inhibited the production of IL-10 by LPS-activated monocytes, showing that suppression of TNF is not secondary to the production of this potent anti-inflammatory cytokine. Abrogation of TNF production in the presence of M. ulcerans may go some way toward explaining why there is little or no immune response in the necrotizing stage of disease, a finding that distinguishes M. ulcerans infections from other mycobacterial diseases. TNF is usually referred to as the inflammatory cytokine, emphasizing its central role in initiating (along with IL-1) the cytokine cascade and the release of other factors that make up the immune system’s response to infection (28, 29). Such a response is critical in the successful resolution of chronic infectious diseases. In tuberculosis, TNF synergy with IFN-γ increases the production of nitric oxide metabolites (30), which contributes to mycobacterial killing (31). The potential to inhibit TNF expression may explain why in M. ulcerans infections the patient’s general health is described as normal (3, 32), whereas during episodes of tuberculosis, the systemic TNF generated results in general manifestations, such as fever and cachexia (33). We also noted that although aHDL did not appear to have any cytopathic effect on monocytes, exposure to the factor did block the ability of the cells to adhere. The reason for this is unknown, but is likely to relate to the inhibition of expression/function of adhesion molecules. Further studies are needed to confirm this.

As TNF is a key player in immuno-inflammatory responses, we chose to investigate the effect of aHDL on selected functions of this cytokine. Interestingly, the TNF-induced apoptosis of WEHI 164 cells was not inhibited by aHDL. In contrast, TNF-induced NF-κB activation was inhibited, indicating selectivity of the factor. The mechanism of this inhibition appears to lie in some unresolved area of TNF signal transduction. The aHDL had no effect on IκBα degradation, which is thought to be a consequence of triggering a cascade of signaling molecules involving TNF-R1-associated death domain (34), TRAF2 (35), NF-κB-inducing kinase (36), receptor-interacting protein (37), and IKKα/β (reviewed in Ref. 25) that finally results in the phosphorylation of IκBα with its subsequent destruction. However, aHDL had a profound inhibitory effect on nuclear DNA binding activity, suggesting the existence of other signaling components that control this aspect of NF-κB function. Studies by Ghosh et al. (38) have suggested that the signaling pathway of NF-κB may involve phosphorylation of the p65 subunit of NF-κB by protein kinase A. The precise details of this are still unclear, and other studies have indicated the involvement of additional kinases that affect the regulation of NF-κB post-IκBα degradation (39, 40, 41, 42, 43, 44, 45). Also, aHDL had no effect on TNF-induced activation of p38 MAPK, which had been implicated in controlling NF-κB function at the later stage of trans-activation (27). Activation of this kinase has been linked to TRAF2 (46). The above data may imply that the factor inhibits unknown components of the TNF signaling mechanism, an observation compatible with the finding that aHDL does not inhibit TNF-induced cytotoxicity, which is dependant on TNF-R1-associated death domain, Fas-associated death domain protein (35), and the caspase arm of the signaling cascade (47). This is also the subject of further investigation. It was further observed that aHDL failed to inhibit LPS- or IL-1-induced NF-κB activation. IL-1 and LPS signal via distinct receptors, IL-1R (reviewed in 60) and toll-like receptor 2/4 (48, 49), respectively. However, the cytoplasmic tails of these receptor proteins have homology; both share toll domains. Both receptors appear to use very similar signaling mechanisms to activate NF-κB involving myeloid differentiation primary response gene, IL-1R-associated kinase, and TRAF6, leading to activation of NF-κB-inducing kinase and IKKα/β (25, 50, 51, 52, 53, 54). This may explain the insensitivity of LPS and IL-1 to aHDL. It would also indicate that the inhibition of LPS-induced TNF release is by a mechanism other than the NF-κB pathway.

An intriguing observation was that rather than inhibiting, aHDL enhanced PMA activation of NF-κB. This would indicate yet another activation pathway for this transcription factor. The differential effect of the factor on PMA compared with TNF could have been due to the presence of more than one active principle in our partially pure extract. Very recently, the M. ulcerans factor was identified as a polyketide lipid toxin (mycolactone) (26). Examination of aHDL by TLC has shown that aHDL contains mycolactone as well as other lipid species (P. Small, unpublished observation). However, mycolactone in our hands gave identical results as aHDL by inhibiting TNF-induced NF-κB and enhancing PMA-induced NF-κB activation. We also observed that like aHDL mycolactone inhibited TNF production by LPS-activated monocytes (results not shown). Nevertheless, the possibility of other immunologically active species in addition to mycolactone in the aHDL preparation cannot be ruled out. The mechanism by which PMA induces NF-κB is still largely unknown, and thus, we do not know how the aHDL/mycolactone effect is produced. However, the data would indicate that the activation of NF-κB must involve multiple mechanisms. The M. ulcerans factor may therefore provide a useful tool for the future analysis of pathways that activate NF-κB.

Mycolactone is structurally related to the immunosuppressants FK506 and rapamycin as well as the antibiotic rifampicin, which has immunosuppression as a well-established side effect (55, 56, 57, 58). In the context of this study it is interesting to note that rapamycin has also been reported to inhibit CD28-induced NF-κB activation via inhibiting IκBα down-regulation (59). Whether this is related to effects on the target of rapamycin is unclear. FK506 may also, like cyclosporin A, with which it shares calcineurin as a common target, inhibit NF-κB.

In summary, using the partially purified M. ulcerans soluble factor, we demonstrated a number of potent inhibitory activities on monocytes and T cells. These actions are likely to account for the localized immunosuppressed state that permits the persistence of the infection. However, the effect of the polyketide factor is restricted. We found that the inhibition of TNF and IL-2 production, cytokines that are important to both innate and adaptive immune systems, may explain the lack of host resistance. Also, the findings of selective effects on aspects of TNF function and NF-κB activation suggest that the factor has a discrete intracellular target(s). The indication that the M. ulcerans factor is crucial to the pathogenesis of disease opens up the possibility that inactivation of the factor may provide a new therapeutic approach for this disabling condition.

We acknowledge the assistance of Dr. M. Kahan with the flow cytometry analysis. We also thank Ms. Abigail Hunt, Ms. Anne Hales, Mr. Ferdinand Lali, and Dr. Shara Cohen for their technical help.